Research Handbook on Digital Trade 180088494X, 9781800884946

Table of contents :
Front Matter
Copyright
Contents
Contributors
Acknowledgements
1. Introduction to Research Handbook on Digital Trade
PART I TRADE AGREEMENTS AND DIGITAL TRADE
2. Digital trade rulemaking in free trade agreements
3. Three generations of digital trade provisions in preferential trade agreements
4. E-commerce or digital trade? Why the difference should matter to trade lawyers
5. WTO law and cross-border data flows for digital trade
6. The Digital Economy Partnership Agreement (DEPA): accession to the digital-only regime
7. Placing gender equality at the centre of global digital trade policy
PART II REGIONAL APPROACHES TO DIGITAL TRADE: NORTH AMERICA AND EUROPE
8. The EU as a digital trade actor
9. Federalism and digital trade
PART III REGIONAL APPROACHES TO DIGITAL TRADE: ASIA
10. China and WTO e-commerce negotiations
11. Facilitating digital trade in Hong Kong
12. Digital trade agreements and digital policy space in the Republic of Korea
13. Towards digital special economic zones: new technology, digitalization and transformation
PART IV REGIONAL APPROACHES TO DIGITAL TRADE: AFRICA, THE MIDDLE EAST AND THE AMERICAS
14. Streamlining the law on digital trade: prospects for the African Continental Free Trade Area (AfCFTA) negotiations
15. Legal approaches to the regulation of digital trade by Middle Eastern countries
16. Latin America: the leap from the single window to the Single Submission Portal as a way of internalizing SMEs
PART V DIGITAL TRADE LEGAL AND POLICY CHALLENGES: FINANCE, INVESTMENT AND TAXES
17. The compatibility of digital services taxes with World Trade Organization law
18. The feasibility and desirability of applying international investment agreements to digital assets
19. Fintech: technology-enabled financial innovation for digital trade
20. The digital nomad and the emergence of global labour mobility
PART VI DIGITAL TRADE LEGAL AND POLICY CHALLENGES: ARTIFICIAL INTELLIGENCE
21. The principles of algorithmic justice in the digital market
22. Artificial intelligence and law: emerging divergent national regulatory approaches in a changing landscape of fast-evolving AI technologies
23. The role of artificial intelligence in international commercial arbitration in the post-Covid era
PART VII DIGITAL TRADE LEGAL AND POLICY CHALLENGES: DATA, SECURITY AND IP
24. Digital economy agreements: where do we stand and where are we going?
25. Digital trade and economic security: considerations on the Japan-US Digital Trade Agreement
26. Digital trade and intellectual property
Index

Citation preview

RESEARCH HANDBOOK ON DIGITAL TRADE

Research Handbook on Digital Trade Edited by

David Collins Professor of International Economic Law, The City Law School of City, University of London, UK

Michael Geist Canada Research Chair in Internet and E-commerce Law, University of Ottawa, Canada

Cheltenham, UK • Northampton, MA, USA

© The Editors and Contributors Severally 2023

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical or photocopying, recording, or otherwise without the prior permission of the publisher. Published by Edward Elgar Publishing Limited The Lypiatts 15 Lansdown Road Cheltenham Glos GL50 2JA UK Edward Elgar Publishing, Inc. William Pratt House 9 Dewey Court Northampton Massachusetts 01060 USA A catalogue record for this book is available from the British Library Library of Congress Control Number: 2023943159 This book is available electronically in the Law subject collection http://dx.doi.org/10.4337/9781800884953

ISBN 978 1 80088 494 6 (cased) ISBN 978 1 80088 495 3 (eBook)

EEP BoX

Contents

List of contributorsviii Acknowledgementsx 1

Introduction to Research Handbook on Digital Trade1 David Collins and Michael Geist

PART I

TRADE AGREEMENTS AND DIGITAL TRADE

2

Digital trade rulemaking in free trade agreements Mira Burri

3

Three generations of digital trade provisions in preferential trade agreements Rodrigo Polanco

4

E-commerce or digital trade? Why the difference should matter to trade lawyers54 Wolfgang Alschner

5

WTO law and cross-border data flows for digital trade Rolf H. Weber

6

The Digital Economy Partnership Agreement (DEPA): accession to the digital-only regime Joo Hyoung Lee and David Collins

7

Placing gender equality at the centre of global digital trade policy Michael Geist

PART II

9 28

73

90 102

REGIONAL APPROACHES TO DIGITAL TRADE: NORTH AMERICA AND EUROPE

8

The EU as a digital trade actor Elaine Fahey

118

9

Federalism and digital trade Patrick Leblond

133

PART III REGIONAL APPROACHES TO DIGITAL TRADE: ASIA 10

China and WTO e-commerce negotiations Henry Gao

148

11

Facilitating digital trade in Hong Kong Bryan Mercurio

168

v

vi  Research handbook on digital trade 12

Digital trade agreements and digital policy space in the Republic of Korea Tae Jung Park and Joo Hyun Park

13

Towards digital special economic zones: new technology, digitalization and transformation Julien Chaisse

178

199

PART IV REGIONAL APPROACHES TO DIGITAL TRADE: AFRICA, THE MIDDLE EAST AND THE AMERICAS 14

Streamlining the law on digital trade: prospects for the African Continental Free Trade Area (AfCFTA) negotiations Cheryl Dine

15

Legal approaches to the regulation of digital trade by Middle Eastern countries 233 Bashar Malkawi

16

Latin America: the leap from the single window to the Single Submission Portal as a way of internalizing SMEs Manuel Quindimil

PART V

218

252

DIGITAL TRADE LEGAL AND POLICY CHALLENGES: FINANCE, INVESTMENT AND TAXES

17

The compatibility of digital services taxes with World Trade Organization law 270 David Collins

18

The feasibility and desirability of applying international investment agreements to digital assets Nicolette Butler and Jasem Tarawneh

19

Fintech: technology-enabled financial innovation for digital trade Lerong Lu

306

20

The digital nomad and the emergence of global labour mobility David Collins and Ben Robinson

329

288

PART VI DIGITAL TRADE LEGAL AND POLICY CHALLENGES: ARTIFICIAL INTELLIGENCE 21

The principles of algorithmic justice in the digital market Maciej Hulicki

345

22

Artificial intelligence and law: emerging divergent national regulatory approaches in a changing landscape of fast-evolving AI technologies Lijun Zhao

369

23

The role of artificial intelligence in international commercial arbitration in the post-Covid era Muhammad Hassan Idrees and Annabelle Onyefulu

400

Contents  vii PART VII DIGITAL TRADE LEGAL AND POLICY CHALLENGES: DATA, SECURITY AND IP 24

Digital economy agreements: where do we stand and where are we going? Dan Ciuriak

25

Digital trade and economic security: considerations on the Japan–US Digital Trade Agreement Rikako Watai

26

Digital trade and intellectual property Marc D. Mimler

416

439 455

Index474

Contributors

Wolfgang Alschner, University of Ottawa Mira Burri, University of Lucerne Nicolette Butler, University of Manchester Julien Chaisse, City University of Hong Kong David Collins, City, University of London Dan Ciuriak, Centre for International Governance Innovation Cheryl Dine, City, University of London Elaine Fahey, City, University of London Henry Gao, Singapore Management University Michael Geist, University of Ottawa Maciej Hulicki, Cardinal Stefan Wyszyński University Muhammad Hassan Idrees, University of Bedfordshire Patrick Leblond, University of Ottawa Joo Hyoung Lee, Kim & Chang Lerong Lu, King’s College London Bashar Malkawi, Government of Dubai, Legal Affairs Department Bryan Mercurio, Chinese University of Hong Kong Marc D. Mimler, City, University of London Annabelle Onyefulu, University of Bedfordshire Joo Hyun Park, Yulchon LLC Tae Jung Park, KAIST (Moon Soul Graduate School of Future Strategy) Rodrigo Polanco, World Trade Institute Manuel Quindimil, Pontifical Catholic University of Peru Ben Robinson, Geeks Ltd Jasem Tarawneh, Queen Mary University of London Rikako Watai, Keio University

viii

Contributors  ix Rolf H. Weber, University of Zurich Lijun Zhao, City, University of London

Acknowledgements

Publishing a peer-reviewed handbook with numerous contributors on a rapidly evolving, high-profile issue requires a special group of contributors, publishers, editors and reviewers. Our thanks are due first and foremost to the contributors. Each was presented with challenging timelines to complete their submissions and each embraced this project with enthusiasm and professionalism. Thanks to Edward Elgar Publishing, who offered tremendous support for this book project. Thanks as well to the Social Sciences and Humanities Research Council of Canada and the Canada Research Chair programme as this book benefited from their financial support. Thanks also to our colleagues and family for their support throughout this project. It is a great honour to work with such an exceptional, supportive group of colleagues at City, University of London and the University of Ottawa. David Collins Michael Geist January 2023

x

1. Introduction to Research Handbook on Digital Trade David Collins and Michael Geist

Digital trade has emerged in recent years as a critical part of commercial activity. From a policy perspective, it is now firmly part of the international trade world, with negotiations in multilateral and bilateral trade agreements as well as at the World Trade Organization (WTO) designed to facilitate increased cross-border digital trade activity. The resulting provisions could have significant implications for commerce around the globe in the years ahead. Digital trade – often referred to interchangeably as e-commerce – enables businesses to operate entirely or partially online, thereby reducing costs associated with physical stores and other physical infrastructure and enabling access to international markets. With these incentives, more vendors enter the market, leading to greater competition. Global e-commerce growth is rapidly transforming commercial activity around the world. For example, UNCTAD estimates that global e-commerce sales in 2020 were valued at $26.7 trillion and represented 19 per cent of total retail sales in 20201 – a dramatic increase on previous numbers that partly resulted from lockdowns due to the COVID-19 pandemic, whereby many consumers sought to adjust their purchasing behaviour by shifting to online stores and shopping.2 One of the chief advantages of digital trade is its ability to connect consumers with businesses in other countries, effectively expanding the commercial reach of even micro, small and medium enterprises (MSMEs) around the globe, and treating consumers to an unprecedented degree of choice. However, to do this effectively, cross-border digital trade must be subject to laws and regulations, many of which are contained in or shaped by international trade agreements. From a policy perspective, digital trade is now a key component of international trade, with negotiations in multilateral and bilateral trade agreements designed to facilitate increased cross-border digital trade activity. The resulting provisions will have significant implications for commerce around the world in the years ahead. Moreover, they are increasingly cropping up in regional and bilateral agreements, as well as being the subject of ongoing debate within the WTO. Indeed, a growing number of regional trade agreements that have incorporated specific provisions related to digital trade or e-commerce. Many of these provisions are largely replicated across agreements and have emerged as a template for new agreements. In fact, data reveals that almost all new trade agreements now include digital trade or e-commerce provisions. Several of the most common provisions have potential implications for a variety of industries and interests. These key issues include data localization, data transfer restrictions, general

1 UNCTAD (3 May 2021). ‘Global e-commerce jumps to $26.7 trillion, COVID-19 boosts online sales.’ See https://​unctad​.org/​news/​global​-e​-commerce​-jumps​-267​-trillion​-covid​-19​-boosts​-online​-sales. 2 Ismail, Y., and Hiral Hirani (2021). Addressing the Gender Dimension of E-commerce: Towards a Holistic Analytical and Policy Framework, CUTS International, Geneva, p.11.

1

2  Research handbook on digital trade privacy standards, the validity of electronic contracting and the free flow of digital services. This book seeks to unpack many of the resulting legal and policy issues, examining both substantive challenges as well as developments at the WTO and within regional trade agreements.

PART I: TRADE AGREEMENTS AND DIGITAL TRADE The first part of the handbook explores the evolution of digital trade within the international trade system, assessing the inclusion of digital trade within the WTO and regional agreements as well as the graduation towards digital-only agreements. Mira Burri’s contribution provides a contextual overview of the dynamic process of international rulemaking in the sphere of digital trade, analysing trends observed in modern FTAs which have incorporated digital trade commitments. Examining FTAs such as the Comprehensive and Progressive Agreement for the Trans Pacific Partnership (CPTPP), the Regional Comprehensive Economic Partnership (RCEP) and the United States–Mexico–Canada Agreement (USMCA), her chapter identifies points of convergence as well as divergence in the FTA landscape, points out emerging trends and offers an outlook on future developments. Rodrigo Polanco’s chapter comprises a helpful history of digital trade provisions in preferential trade agreements. Beginning with the first e-commerce provision in 2000 and proceeding up to the present day, the chapter traces the evolution of digital trade chapters (and, later, standalone agreements) through three distinct ‘generations’. The first, Polanco observes, was largely concerned with e-commerce rules, and touched on the issues of customs duties, electronic authentication, paperless trading, consumer protection and unsolicited commercial electronic messages. The second generation strove to regulate data flows, and these provisions included rules on data flows, data localization, privacy and data protection. Finally, the third and most recent generation is advancing rules for the data-driven economy (often doing so in standalone digital economy agreements), and speak to digital identities, competition, digital inclusion and the digital divide, artificial intelligence and open government data. Understanding the topics that constitute these three generations is important as many have been flagged as areas of possible agreement at the WTO negotiations. The e-commerce and digital trade chapters of contemporary preferential trade agreements contain provisions that can generally be divided into two categories: those extending traditional trade concepts to facilitate digitally enabled trade, and those governing trade in new digital assets. Wolfgang Alschner argues that these categories have different norms, political economies, and policy implications, and it would behoove the parties to international trade negotiations to approach these issues as conceptually distinct. A bifurcated approach allows us to integrate e-commerce rules within existing trade regimes, while more troublesome trade in data rules could be integrated with other regimes and alternative governance structures, and be subject to their own dedicated agreements. Using an original dataset, Alschner demonstrates that PTAs are becoming increasingly focused on trade in data, exposing thorny issues and illustrating how this analytical separation could benefit global trade in the future. Rolf Weber canvasses WTO law as it relates to digital trade issues. The existing WTO framework – premised on GATS and GATT – is a remnant of the pre-digital legal order that has yet to catch up with the rapidly developing technologies that continue to change the manner and execution of global trade. These disruptive challenges demand regulatory solutions. Digital data flowing across international borders, in particular, poses significant issues,

Introduction  3 and the chapter examines the future of data-driven trade law. Ultimately, the WTO trade law framework requires a major overhaul. Weber identifies areas that may present ‘low-hanging fruit’ and provides an outlook for future negotiations. Digital-only trade agreements are assessed in Joo Hyoung Lee and David Collins’ chapter on the Digital Economy Partnership Agreement (DEPA), a standalone digital trade treaty signed by New Zealand, Chile and Singapore. The authors consider the benefits and drawbacks of DEPA, noting its emphasis on cooperation rather than binding commitments. The chapter focuses also DEPA’s status as a ‘living agreement’, looking in particular at the expected accession of Canada and the Republic of Korea respectively. Michael Geist’s chapter brings the issue of gender equity to the digital trade arena. Digital trade offers both benefits and disadvantages to the development of gender equality, yet including gender-equality concerns and principles in digital trade or e-commerce trade agreements remains a relatively novel concept that continues to challenge the notion that digital trade is a gender-neutral platform. While many trade agreements reference gender equity or equality, the trade commitments are typically modest and rarely address the many challenges identified in this chapter. This chapter examines both the benefits and challenges faced by women in digital trade and the role those issues have played in trade agreements and work at the WTO. It identifies several models that could be emulated within agreements and points to potential policy initiatives that could better address the role of women within digital trade.

PART II: REGIONAL APPROACHES TO DIGITAL TRADE – NORTH AMERICA AND EUROPE Part II of the handbook is the first of several sections on regional and national approaches to digital trade. The first section explores North American and Europe. Elaine Fahey’s chapter examines the development of the European Union as a digital trade actor, investigating how it has generated significant developments in digital trade through championing best practice and high standards as to privacy, often resting between US and Chinese developments thereon as a ‘middle-ground’ actor. Fahey argues that the EU appears increasingly to move beyond a ‘middle-ground actor’ position in digital trade, aiming for higher standards of data privacy yet also advocating an ambitious digital trade agenda. It also aims for binding legal solutions in an era of soft law proliferating in international economic law and, interestingly, has begun to establish a complex agenda of digital sovereignty. Patrick Leblond’s chapter explores a heretofore unacknowledged dynamic: the relationship between federalism and digital trade. By comparing the systems of Canada and the United States, as well as the relevant provisions of the USMCA, Leblond discusses the implications of subnational policies on digital trade. Leblond notes that Canada’s passive, reactive approach to digital trade leads provincial policies to have greater impacts – potentially detrimental impacts – on digital trade than the policies of American states, which are subject to stronger centralization powers. The USMCA’s digital trade provisions largely do not constrain the subnational governments of either country when it comes to policy measures detrimental to digital trade. In order to ensure that businesses are encouraged to engage in e-commerce, both countries should foster coordination between their various levels of government to reduce compliance costs.

4  Research handbook on digital trade

PART III: REGIONAL APPROACHES TO DIGITAL TRADE – ASIA Part III of the handbook continues the regional exploration of digital trade with several chapters on Asia. In his chapter, Henry Gao describes China’s role in the WTO e-commerce negotiations, explaining how the industrial giant that once resisted the e-commerce initiative has nonetheless proven itself to be one of the most active negotiators in that space. Despite its troubled history at the WTO with respect to data regulation, China has emerged the world’s largest e-commerce market. Given its interest in e-commerce, China has made three submissions to the negotiations, focusing on scope, trade facilitation and cybersecurity, among other issues, and Gao links each of these issues to Chinese policy objectives. Finally, the chapter discusses China’s problematic ‘obsession with cybersecurity’, and sketches some routes to compromise on this and related issues. The digital trade strategy of Hong Kong is the focus of Bryan Mercurio’s chapter, observing the city state’s decline during the COVID-19 pandemic and commensurate need to tap into its digital strengths to regain its economic footing. The chapter argues that Hong Kong needs to aggressive embrace digital commerce by enacting policies, including international agreements, helping to ensure that it remains a vibrant digital services hub. Tae Jung Park and Joo Hyun Park’s contribution considers the approach towards digital trade adopted by the Republic of Korea with a view to embracing the global trend towards digitalization, intensified by the COVID-19 crisis. The authors explain how Korea’s domestic laws on digitalization, termed the Digital New Deal, comply with its trade obligations including the Regional Comprehensive Economic Partnership (RCEP) and the Comprehensive Progressive Trans Pacific Partnership (CPTPP). The chapter contends that some of Korea’s laws, such as forced data-localization facilities in certain sectors, risk transgressing international commitments, compelling the republic to seek public policy exceptions contained variously in these agreements. Julien Chaisse’s chapter offers important insight into the modernization of special economic zones, as employed throughout the world, to suit the digital economic landscape. The chapter focuses on the cross-border e-commerce elements of the world’s first Digital Free Trade Zone, located in Malaysia, outlining the benefits these technologically advanced regimes can bring to trade in traditional goods as well as incentivizing investment.

PART IV: REGIONAL APPROACHES TO DIGITAL TRADE – AFRICA, THE MIDDLE EAST, AND THE AMERICAS Part IV of the handbook concludes the global tour of regional approaches to digital trade with three chapters drawn from Africa, the Middle East and the Americas. Africa’s engagement with digital trade is evaluated in Cheryl Dine’s chapter which looks at the African Continental Free Trade Area (AfCFTA)’s Protocol on Digital Trade from the perspective of the ‘digital divide’ between developing and developed countries. Noting that fewer than one-third of FTAs between developing states contain digital trade provisions, Dine suggests that it is access to technology rather than legal innovation which has impeded the progress of Africa towards fuller realization of the gains from digital trade. Bashar Malkawi’s chapter examines the growth of digital trade in the Middle East, critically reviewing the different legal approaches to digital trade regulation in the United Arab

Introduction  5 Emirates, Egypt, Jordan and Saudi Arabia. Noting the Middle East’s desire to expand their digital economies with a view to economic diversification, the chapter reviews domestic e-commerce laws, data protection regulations and free trade agreements across the region which engage with digital issues. Manuel Quindimil looks at the challenges faced by Latin America in taking full advantage of the digitalization of the global economy. Drawing on the difficulties faced in particular by SME’s seeking to expand their international profile through digital trade, Quindimil’s chapter assesses the region’s capacity to deliver digital trade facilitation through single submission portals, as implemented successfully in other parts of the world.

PART V: DIGITAL TRADE LEGAL AND POLICY CHALLENGES – FINANCE, INVESTMENT AND TAXES Having examined digital trade at the international, regional and national levels, the handbook shifts towards a consideration of specific digital trade legal and policy challenges, starting with finance, investment and tax-related issues. In his chapter on digital services taxes, David Collins explores the integration of these taxes, as planned variously by many developed and developing countries, into WTO rules. The chapter evaluates whether digital services taxes, many of which are designed to target the largest US-based technology companies, are compatible with GATS prohibitions on discrimination and whether exceptions under that agreement could operate to justify them were such pieces of legislation ever challenged through the WTO’s Dispute Settlement System. Nicolette Butler and Jasem Tarawneh turn their attention to the issue of investment protection in an era of digital commerce in their chapter, asking to what extent digital assets conform to the understanding of investment as contained in modern international investment agreements, enabling protection under these regimes. The authors contend that the lack of investment arbitration disputes, the relative novelty of the topic and the vagueness of the concepts which underpin it mean that the digital sector and its interaction with the international investment law regime remains largely unexplored in academic literature. Another vital area of policy concern in digital trade relates to the rapidly advancing field of fintech. Lerong Lu’s chapter accordingly discusses the global rise of financial technology fintech, analysing the role that it has played in the digital economy and how it has leveraged technologies such as blockchain, cloud computing and big data. The chapter goes on to examine the business models of fintech, BigTech and Metaverse corporations to understand how they create novel financial services and products to suit the needs of financial customers. Lu concludes with an examination of two case studies of two areas of fintech innovation: alternative finance platforms and digital money. David Collins and Ben Robinson explore the phenomenon of Digital Nomads, the self-employed earning their living online and lacking in a permanent base in one particular jurisdiction. Acknowledging that Digital Nomads, who have come to prominence during the COVID-19 pandemic, raise many legal issues, including tax and employment matters, the contribution focuses on the extent to which FTAs may accommodate the needs of these workers by analogizing them to SMEs.

6  Research handbook on digital trade

PART VI: DIGITAL TRADE LEGAL AND POLICY CHALLENGES – ARTIFICIAL INTELLIGENCE Artificial intelligence has emerged as an exceptionally important disruptive technology that has the potential to alter virtually every aspect of commercial activity. This part of the handbook examines the digital trade legal and policy challenges associated with AI. Maciej Hulicki examines the use of algorithms in various commercial contexts, notably in relation to the protection of privacy, technical safety and the preservation of transparency and accountability, which the author collectively terms ‘Algorithmic Justice’. The chapter explores the EU’s proposed Artificial Intelligence Act, which, while appearing to address many of the objectives engaged by Algorithmic Justice, may pose compliance challenges for businesses. Taking a broader view of the controversial issue of legal approaches to artificial intelligence, Lijun Zhao argues in her chapter that policy-makers at both national and global levels are facing the ‘pacing problem’ – technology is developing faster than the policy-makers’ ability to keep up. The chapter examines how various countries which are leading the way in the development of artificial intelligence, notably the US and China have adopted divergent regulatory responses to handle this problem while at the international level, global trade law has so far largely sidestepped this important sphere of digital trade. Muhammad Hassan Idrees and Annabelle Onyefulu examine AI through the specific use case of international arbitration in the post-COVID era. They note that AI may promote efficiency, cost and speed on the one hand, while on the other it may violate the current regulatory frameworks in international arbitration. This chapter, therefore, engages in analysing the future of international arbitration vis-à-vis the introduction of new technologies. In order to do this, the chapter aims at evaluating the possibility of replacing human arbitrators with AI arbitrators, the public policy requirements, and the ethical issues that may arise therefrom.

PART VII: DIGITAL TRADE LEGAL AND POLICY CHALLENGES – DATA, SECURITY AND IP At the very heart of digital trade are issues related to information: data protection, security and intellectual property among them. Dan Ciuriak traces the development of trade-related aspects of the data-driven economy, including the governing provisions in trade agreements. The chapter reviews the technological changes that characterize the data-driven economy and the state of digital economy agreements, and identifies the challenges facing a new WTO agreement. Modern technologies have complicated the distinction between services and goods; the value of data is distributed asymmetrically; intellectual property protections have unintended consequences; machine knowledge capital promises to disrupt income distribution across the world; and the digital transformation has exposed new vulnerabilities with respect to democratic processes and national security. Today, we find ourselves in a technological and political environment decidedly different from that in which the governing rules-based trade system developed. Our multilateral trade system will need to adapt to fit present and future circumstances. Rekato Watai considers the national security implications of digital trade in his chapter, examining the policy initiatives of Japan, as captured in the Japan–United States Digital Trade Agreement. The chapter illustrates how the security-oriented provisions of that agreement,

Introduction  7 which embody some of the domestic initiatives of Japan, are expected to become accepted norms which will be built into the digital chapters of future trade agreements. The interaction between intellectual property and digital trade is the subject of Marc Mimler’s contribution. He analyses the regulatory dilemma which arises from the fact that strong intellectual property protection, both of physical and virtual goods, often acts as a barrier to trade, particularly when intellectual property enforcement and infringement can take place in various jurisdictions spread across the internet. The chapter discusses current and emerging issues in intellectual property protection in the digital sphere such as digital exhaustion and intermediary liability, also touching upon potential impacts of new technologies such as blockchain and 3D printing.

PART I TRADE AGREEMENTS AND DIGITAL TRADE

2. Digital trade rulemaking in free trade agreements Mira Burri1

1. INTRODUCTION ‘Electronic commerce’ or ‘digital trade’,2 as it is now commonly referred to, is not an entirely new topic in the domain of international economic law. Indeed, as early as 1998 the membership of the World Trade Organization (WTO) initiated already in 1998 a Work Programme3 to address the implications of the Internet that could potentially lead to adjustments in the existing rules for trade in goods, trade in services and intellectual property (IP) protection. Yet, this effort was largely unsuccessful for two decades and in the meantime, digital trade as practice and as a subject of regulation has entered into an entirely new phase. On the one hand, this has been spurred by the progressively advancing digitization of economies and societies as a whole, as well as by the more recently emergent importance of data;4 on the other hand, the surge in digital trade rulemaking, which in this chapter covers both hard and soft rules creation,5 can be linked to the multiple new issues that the data-driven economy has raised – some of which, such as those in the area of personal data protection, demand urgent regulatory responses.6

1 The support of the European Research Council under Consolidator Grant 101003216 is gratefully acknowledged. 2 The OECD has pointed out that, while there is no single recognized and accepted definition of digital trade, there is a growing consensus that it encompasses digitally enabled transactions of trade in goods and services that can either be digitally or physically delivered, and that involve consumers, firms, and governments. Critical is that the movement of data underpins contemporary digital trade and can also itself be traded as an asset and a means through which global value chains are organized and services delivered. See Javier López González and Marie-Agnes Jouanjean, ‘Digital Trade: Developing a Framework for Analysis’, OECD Trade Policy Papers 205 (2017). 3 WTO, Work Programme on Electronic Commerce, WT/L/274, 30 September 1998. 4 See e.g. James Manyika et al., Big Data: The Next Frontier for Innovation, Competition, and Productivity (Washington, DC: McKinsey Global Institute, 2011); Viktor Mayer-Schönberger and Kenneth Cukier, Big Data: A Revolution That Will Transform How We Live, Work, and Think (New York: Eamon Dolan/Houghton Mifflin Harcourt, 2013); Nicolaus Henke et al., The Age of Analytics: Competing in a Data-Driven World (Washington, DC: McKinsey Global Institute 2016); WTO, World Trade Report 2018: The Future of World Trade: How Digital Technologies Are Transforming Global Commerce (Geneva: World Trade Organization, 2018). 5 See e.g. Kenneth W. Abbott and Duncan Snidal, ‘Hard and Soft Law in International Governance’, International Organization 54:3 (2000), 421–56; Gregory C. Shaffer and Mark A. Pollack, ‘Hard vs. Soft Law: Alternatives, Complements, and Antagonists in International Governance’, Minnesota Law Review 94 (2010), 706–99. 6 See e.g. Mira Burri, ‘Interfacing Privacy and Trade’, Case Western Journal of International Law 53 (2021), 35–88; Anupam Chander and Paul M. Schwartz, ‘Privacy and/or Trade’, University of Chicago Law Review 90:1 (2023) 49–135.

9

10  Research handbook on digital trade The chapter is set against this background and seeks to provide a better understanding and contextualization of the highly dynamic field of digital trade rulemaking driven by free trade agreements (FTAs). The new rules found in bilateral and regional FTAs not only compensate for the lack of developments in the multilateral forum of the WTO (at least so far)7 but effectively create a comprehensive, albeit fragmented, governance framework for the data-driven economy. The chapter’s analytical lens is directed in particular towards the more recent and advanced models of digital trade rulemaking, such as those under the Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) and the United States–Mexico– Canada Agreement (USMCA), as well as those endorsed by dedicated digital economy agreements (DEAs). The chapter then covers the European Union’s (EU) new generation of trade deals, in particular the post-Brexit agreement with the United Kingdom (UK), the agreement with New Zealand and the currently negotiated deals with Australia and Tunisia, and looks at the Regional Comprehensive Economic Partnership (RCEP), as the first agreement with digital trade provisions that includes China, so as to give a sense of the dynamic governance environment on issues of digital trade. Subsequently, the chapter identifies points of convergence and divergence in the FTA landscape of digital trade rules, sketches some emerging trends and, in concluding, provides an outlook on future developments.

2.

THE FTA LANDSCAPE OF DIGITAL TRADE RULEMAKING

2.1 Overview The regulatory environment for digital trade has been shaped by FTAs. Out of the 384 FTAs entered into between 2000 and 2022, 203 contain provisions relevant for digital trade and 109 have dedicated electronic commerce chapters.8 Although the pertinent rules remain highly heterogeneous and differ as to issues covered, the level of commitments and their binding nature, it is overall evident that the trend towards more and more detailed provisions on digital trade has intensified significantly over the years.9 This regulatory push in the domain of digital trade can be explained by the increased importance of the issue over the years, but also by the role played by the United States (US).10

7 For an analysis of the WTO relevance to digital trade, see Mira Burri and Thomas Cottier (eds), Trade Governance in the Digital Age (Cambridge: Cambridge University Press, 2012); Mira Burri, ‘The International Economic Law Framework for Digital Trade’, Zeitschrift für Schweizerisches Recht 135 (2015), 10–72. 8 This analysis is based on a dataset of all digital trade relevant norms in trade agreements (TAPED). See Mira Burri and Rodrigo Polanco, ‘Digital Trade Provisions in Preferential Trade Agreements: Introducing a New Dataset’, Journal of International Economic Law 23 (2020), 187–220. For all data, as well as updates of the dataset, see https://​unilu​.ch/​taped. 9 For an overview of the FTA developments, see Mira Burri, ‘Data Flows and Global Trade Law’, in Mira Burri (ed), Big Data and Global Trade Law (Cambridge: Cambridge University Press, 2021), 11–41. 10 See Manfred Elsig and Sebastian Klotz, ‘Data Flow-Related Provisions in Preferential Trade Agreements: Trends and Patterns of Diffusion’, in Mira Burri (ed), Big Data and Global Trade Law (Cambridge: Cambridge University Press, 2021), 42–62.

Digital trade rulemaking in free trade agreements  11 The US has over the years endorsed its ‘Digital Agenda’11 through the FTA channel. The agreements reached since 2002 with Australia, Bahrain, Chile, Morocco, Oman, Peru, Singapore, the Central American countries, Panama, Colombia and South Korea, all contain critical, albeit with different depths of commitment, provisions in the broader field of digital trade. The diffusion of the US template is not however limited to US agreements12 and has been replicated in a number of other FTAs as well, such as Singapore–Australia, Thailand– Australia, New Zealand–Singapore, Japan–Singapore and South Korea–Singapore. Many smaller states, such as Chile, have also become active in the area of data governance; at the same time, many other countries, such as those parties to the European Free Trade Area (EFTA),13 have not yet implemented distinct digital trade strategies.14 While the EU is to be reckoned with as a major actor in international economic law and policy, it has also been rather a late-comer in the digital trade rulemaking domain, as this chapter reveals below. The relevant aspects of digital trade governance can be found in: (1) the FTA chapters specifically dedicated to electronic commerce; (2) the chapters on cross-border supply of services (with particular relevance to the telecommunications, computer and related, audiovisual and financial services sectors); and (3) the IP chapters.15 This chapter focuses exclusively on the electronic commerce/digital trade chapters, as well as on the specific digital economy agreements, which have become the new and critical source of new rulemaking in the area of digital trade. The electronic commerce chapters play a dual role in the landscape of trade rules in the digital era. On the one hand, they represent an attempt to compensate for the lack of progress in the WTO and address many of the questions of the WTO Electronic Commerce Programme that have been discussed but not resolved.16 For instance, a majority of the chapters recognize the applicability of WTO rules to electronic commerce17 and establish an express and permanent duty-free moratorium on electronic transmissions.18 In most of the templates tailored along the US model, the chapters also include a definition of ‘digital products’, which treats products delivered offline equally to those delivered online,19 so that technological neutrality is ensured and some of the classification dilemmas under the General Agreement on Trade in Services (GATS) are cast aside (in particular when combined with negative committing for services20). The electronic commerce chapters increasingly cover also regulatory questions

US Congress, Bipartisan Trade Promotion Authority Act of 2001, H.R. 3005, 3 October 2001; Sacha Wunsch-Vincent, ‘The Digital Trade Agenda of the US’, Aussenwirtschaft 1 (2003), 7–46; also Henry Gao, ‘Regulation of Digital Trade in US Free Trade Agreements: From Trade Regulation to Digital Regulation’, Legal Issues of Economic Integration 45 (2018), 47–70. 12 Elsig and Klotz, supra note 10. 13 The EFTA Members comprise Iceland, Lichtenstein, Norway and Switzerland. 14 It should be noted in this context that the EFTA countries have now adopted a model electronic commerce chapter but it is yet to be implemented in a treaty text. 15 For analysis of all relevant chapters, see Mira Burri, ‘The Regulation of Data Flows in Trade Agreements’, Georgetown Journal of International Law 48 (2017), 408–48. 16 Sacha Wunsch-Vincent, The WTO, the Internet and Digital Products: EC and US Perspectives (Oxford: Hart, 2006). 17 See e.g. US–Singapore FTA, Article 14.1; US–Australia FTA, Article 16.1. 18 See e.g. US–Singapore FTA, Article 14.3, para. 1; US–Chile FTA, Article 15.3. For a discussion of the variety of rules on the moratorium, see Burri and Polanco, supra note 8. 19 See e.g. US–Singapore FTA, Article 14.3; US–Australia FTA, Article 16.4. 20 See e.g. Burri (2017), supra note 15. 11

12  Research handbook on digital trade that have not been treated in the WTO context – the so-called WTO-extra issues. One can group these rules into two broader categories: (1) rules that seek to enable digital trade in general, by tackling distinct issues, such as paperless trading and electronic authentication; and (2) rules that address cross-border data, new digital trade barriers and novel issues, which can encompass questions ranging from cybersecurity to open government data. It should be noted that as to the first cluster of issues on the facilitation of digital trade, the number of FTAs that contain such rules is substantial and one can observe convergence; still, only a few agreements include rules on data.21 2.2

Emerging Templates for Digital Trade and Stakeholder Positioning

In the following sections, the chapter looks at the new rules created in recent agreements through a detailed analysis of the most advanced electronic commerce chapters thus far – those of the CPTPP, the USMCA and the dedicated digital economy agreements. We complement this analysis with an enquiry into the EU treaties and the EU’s repositioning on data flows in particular, and into the RCEP as the first agreement with digital trade provisions to include China. The purpose is two-pronged – on the one hand, to highlight legal innovation in these treaties; on the other, to give a sense of the positions of the major stakeholders. 2.2.1 The Comprehensive and Progressive Agreement for Transpacific Partnership The Comprehensive and Progressive Agreement for Transpacific Partnership was agreed upon in 2017 between 11 countries in the Pacific Rim22 and entered into force on 30 December 2018. Beyond the economic significance of the agreement,23 the CPTPP chapter on electronic commerce created the most comprehensive template in the landscape of FTAs. It should be noted that despite the US having dropped out of the planned Transpacific Partnership Agreement (TPP) with the start of the Trump administration, the CPTPP chapter reflects the US efforts under its updated ‘Digital 2 Dozen’ agenda24 to secure obligations on digital trade25 and is a verbatim reiteration of the TPP chapter. A closer look at the CPTPP electronic commerce chapter is therefore well deserved. In the first part, and not unusually for US-led and other FTAs, the CPTPP electronic commerce chapter clarifies that it applies ‘to measures adopted or maintained by a Party that affect trade by electronic means’26 but excludes from this broad scope (1) government procurement and (2) information held or processed by or on behalf of a party, or measures related to such

See Burri and Polanco, supra note 8; also Mira Burri, ‘The Governance of Data and Data Flows in Trade Agreements: The Pitfalls of Legal Adaptation’, UC Davies Law Review 51 (2017), 65–132. 22 Australia, Brunei, Canada, Chile, Japan, Malaysia, Mexico, New Zealand, Peru, Singapore and Vietnam. 23 See e.g. Zachary Torrey, ‘TPP 2.0: The Deal Without the US: What’s New about the CPTPP and What Do the Changes Mean?’ The Diplomat, 3 February 2018. 24 See https://​ustr​.gov/​about​-us/​policy​-offices/​press​-office/​reports​-and​-publications/​2016/​digital​-2​ -dozen. 25 See also in this sense New Zealand’s Waitangi Tribunal, Report on the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (November 2021), at 72 and passim. 26 Article 14.2(2) CPTPP. 21

Digital trade rulemaking in free trade agreements  13 information, including measures related to its collection.27 For greater certainty, measures affecting the supply of a service delivered or performed electronically are subject to the obligations contained in the relevant provisions on investment and services;28 some additional exceptions are also specified.29 The following provisions address, again as customarily, some of the leftovers of the WTO Electronic Commerce Programme and provide for the facilitation of online commerce. In this sense, Article 14.3 CPTPP bans the imposition of customs duties on electronic transmissions, including content transmitted electronically, and Article 14.4 endorses the non-discriminatory treatment of digital products,30 which are defined broadly pursuant to Article 14.1.31 Article 14.5 CPTPP is meant to shape the domestic electronic transactions framework by including binding obligations for the parties to follow the principles of the UNCITRAL Model Law on Electronic Commerce 1996 or the UN Convention on the Use of Electronic Communications in International Contracts. Parties must endeavour to (1) avoid any unnecessary regulatory burden on electronic transactions; and (2) facilitate input by interested persons in the development of its legal framework for electronic transactions.32 The provisions on paperless trading and on electronic authentication and electronic signatures complement this by securing equivalence of electronic and physical forms. With regard to paperless trading, it is clarified that parties shall endeavour to make trade administration documents available to the public in electronic form and accept trade administration documents submitted electronically as the legal equivalent of the paper version.33 The norm on electronic signatures is more binding and provides that parties shall not deny the legal validity of a signature solely on the basis that the signature is in electronic form,34 nor shall they adopt or maintain measures that prohibit parties to an electronic transaction from mutually determining the appropriate authentication methods for that transaction; or prevent such parties from having the opportunity to establish before judicial or administrative authorities that their transaction complies with legal requirements with respect to authentication.35

27 Article 14.2(3) CPTPP. For the lack of guidance and the potential contentions around the scope of this exception, see the different experts’ opinions in New Zealand’s Waitangi Tribunal, Report on the Comprehensive and Progressive Agreement for Trans-Pacific Partnership, supra note 25, at 81–3. 28 Article 14.2(4) CPTPP. 29 Article 14.2(5) and (6) CPTPP. 30 The obligation does not apply to subsidies or grants, including government-supported loans, guarantees and insurance, nor to broadcasting. It can also be limited through the rights and obligations specified in the IP chapter. Article 14.2(3) CPTPP. 31 Digital product means a computer programme, text, video, image, sound recording or other product that is digitally encoded, that is produced for commercial sale or distribution, and that can be transmitted electronically. Two specifications in the footnotes apply: (1) digital product does not include a digitized representation of a financial instrument, including money; and (2) the definition of digital product should not be understood to reflect a party’s view on whether trade in digital products through electronic transmission should be categorized as trade in services or trade in goods. Article 14(1) includes two footnotes clarifying that: ‘For greater certainty, digital product does not include a digitised representation of a financial instrument, including money’ (footnote 2) and ‘The definition of digital product should not be understood to reflect a Party’s view on whether trade in digital products through electronic transmission should be categorised as trade in services or trade in goods’ (footnote 3). 32 Article 14.5(2) CPTPP. 33 Article 14.9 CPTPP. 34 Article 14.6(1) CPTPP. 35 Article 14.6(2) CPTPP.

14  Research handbook on digital trade The remainder of the provisions found in the CPTPP electronic commerce chapter can be said to belong to the second and more innovative category of rulemaking that tackles the emergent issues of the data-driven economy. Most importantly, the CPTPP explicitly seeks to curb data protectionism. First, it does so through an explicit ban on the use of data localization measures. Article 14.13(2) prohibits the parties from requiring a ‘covered person to use or locate computing facilities in that Party’s territory as a condition for conducting business in that territory’. Second, the CPTPP replaces the soft language from the US–South Korea FTA on free data flows and frames it as a hard rule: ‘[e]ach Party shall allow the cross-border transfer of information by electronic means, including personal information, when this activity is for the conduct of the business of a covered person.’36 The rule has a broad scope and most data transferred over the internet is likely to be covered. Measures restricting digital flows or implementing localization requirements are permitted only if they do not amount to ‘arbitrary or unjustifiable discrimination or a disguised restriction on trade’ and do not ‘impose restrictions on transfers of information greater than are required to achieve the objective’.37 These non-discriminatory conditions are similar to the strict test formulated by Article XIV GATS and Article XX GATT 1994 – a test that is supposed to balance trade and non-trade interests by ‘excusing’ certain violations but that is also extremely hard to pass, as the WTO jurisprudence has thus far revealed.38 The CPTPP test differs from the WTO norms in two significant elements: (1) while there is a list of public policy objectives in the GATT 1994 and the GATS, the CPTPP provides no such enumeration and simply refers to a ‘legitimate public policy objective’;39 (2) in the chapeau-like reiteration of ‘arbitrary or unjustifiable discrimination’, there is no GATT or GATS-like qualification of ‘between countries where like conditions prevail’. The scope of the exception is thus unclear – it can be linked to legal uncertainty, as well as to potentially unworkable safeguards for domestic constituencies.40 Lastly, it should be noted that the ban on localization measures is softened on financial services and institutions.41 An annex to the Financial Services chapter has a separate data transfer requirement, whereby certain restrictions on data flows may apply for the protection of privacy or confidentiality of individual records, or for prudential reasons.42 Government procurement is also excluded.43 The CPTPP addresses other novel issues as well – one of them is source code. Pursuant to Article 14.17, a CPTPP Member may not require the transfer of, or access to, source code of software owned by a person of another party as a condition for the import, distribution, sale or use of such software, or of products containing such software, in its territory. The prohibition Article 14.11(2) CPTPP. Article 14.11(3) CPTPP. 38 See e.g. Henrik Andersen, ‘Protection of Non-Trade Values in WTO Appellate Body Jurisprudence: Exceptions, Economic Arguments, and Eluding Questions’, Journal of International Economic Law 18 (2015), 383–405. 39 Article 14.11(3) CPTPP. 40 See e.g. in this sense New Zealand’s Waitangi Tribunal, Report on the Comprehensive and Progressive Agreement for Trans-Pacific Partnership, supra note 25, in particular at 132–42. 41 See the definition of ‘a covered person’ (Article 14.1 CPTPP), which excludes a ‘financial institution’ and a ‘cross-border financial service supplier’. 42 The provision reads: ‘Each Party shall allow a financial institution of another Party to transfer information in electronic or other form, into and out of its territory, for data processing if such processing is required in the institution’s ordinary course of business’. 43 Article 14.8(3) CPTPP. 36 37

Digital trade rulemaking in free trade agreements  15 applies only to mass-market software or products containing such software.44 This means that tailor-made products are excluded, as well as software used for critical infrastructure and those in commercially negotiated contracts.45 The aim of this provision is to protect software companies and address their concerns about loss of IP or cracks in the security of their proprietary code; it may also be interpreted as a reaction to China’s demands for access to source code from software producers selling in its market.46 These provisions illustrate an important development in the FTA rulemaking in that they do not merely seek the reduction of trade barriers, but effectively shape the regulatory space domestically. Particularly critical in this context are also the rules in the area of data protection. Article 14.8(2) requires every CPTPP party to ‘adopt or maintain a legal framework that provides for the protection of the personal information of the users of electronic commerce’. Yet, there are no standards or benchmarks for the legal framework specified, except for a general requirement that CPTPP parties ‘take into account principles or guidelines of relevant international bodies’.47 A footnote provides some clarification in saying that ‘a Party may comply with the obligation in this paragraph by adopting or maintaining measures such as a comprehensive privacy, personal information or personal data protection laws, sector-specific laws covering privacy, or laws that provide for the enforcement of voluntary undertakings by enterprises relating to privacy’.48 Parties are also invited to promote compatibility between their data protection regimes, by essentially treating lower standards as equivalent.49 The goal of these norms can be interpreted as a prioritization of trade over privacy rights. This was pushed by the US during the TPP negotiations, as the US subscribes to a relatively weak and patchy protection of privacy.50 Timewise, this push came also at a phase when the US was wary that it could lose the privilege of transatlantic data transfer, as a consequence of the judgment of the Court of Justice of European Union (CJEU) that struck down the EU–US Safe Harbour Agreement.51

Article 14.17(2) CPTPP. Ibid. On the possible interpretations of the provision and difference to including algorithms, see New Zealand’s Waitangi Tribunal, Report on the Comprehensive and Progressive Agreement for Trans-Pacific Partnership, supra note 25, at 104–12. 46 See e.g. Joint Statement on Trilateral Meeting of the Trade Ministers of the United States, Japan, and the European Union, Washington, DC, 14 January 2020, available at: https://​trade​.ec​.europa​.eu/​ doclib/​docs/​2020/​january/​tradoc​_158567​.pdf. 47 Article 14.8(2) CPTPP. 48 Ibid, at footnote 6. 49 Article 14.8(5) CPTPP. 50 See e.g. James Q. Whitman, ‘The Two Western Cultures of Privacy: Dignity versus Liberty’, The Yale Law Journal 113 (2004), 1151–1221; Paul M. Schwartz and Daniel J. Solove, ‘Reconciling Personal Information in the United States and European Union’, California Law Review 102 (2014), 877–916; also Burri (2021), supra note 6. 51 Case C-362/14 Schrems, judgment of 6 October 2015, EU:C:2015:650. Maximillian Schrems is an Austrian citizen, who filed a suit against the Irish supervisory authority, after it rejected his complaint over Facebook’s practice of storing user data in the US. The plaintiff claimed that his data was not adequately protected in light of the NSA revelations and this, despite the existing agreement between the EU and the US – the so-called safe harbour scheme. The later EU–US Privacy Shield arrangement has been also rendered invalid by a judgment in 2020: Case C-311/18, Data Protection Commissioner v. Facebook Ireland Limited, Maximillian Schrems (Schrems II), judgment of 16 July 2020, ECLI:EU:C:2020:559. A political solution for transatlantic data flows has only been recently found in March 2022 with the legal texts still pending. 44 45

16  Research handbook on digital trade Next to these important data protection provisions, the CPTPP also includes norms on consumer protection52 and spam control,53 as well as, for the first time, rules on cybersecurity. Article 14.16 is however non-binding and identifies a limited scope of activities for cooperation, in situations of ‘malicious intrusions’ or ‘dissemination of malicious code’, and capacity-building of governmental bodies dealing with cybersecurity incidents. Net neutrality is another important digital economy topic that has been given specific attention in the CPTPP, although the rules so created are of non-binding nature.54 The norm comes with a number of exceptions from the domestic laws of the CPTPP parties and permits deviations from undefined situations that call for ‘reasonable network management’ or exclusive services.55 As the obligations are unlinked to remedies for situations, such as blocking, throttling, discriminating or filtering content, it is unlikely that the CPTPP would lead to uniform approach with regard to net neutrality across the CPTPP countries. The approval for the UK to accede to the CPTPP56 and recent requests for accession by China and Taiwan57 potentially expand the commercial reach and geopolitical dimension of this agreement. Next to these possibilities for an enlarged CPTPP membership, it should also be pointed out that the CPTPP model has diffused in a substantial number of other agreements, such as the 2016 Chile–Uruguay FTA, the 2016 updated Singapore–Australia FTA (SAFTA), the 2017 Argentina–Chile FTA, the 2018 Singapore–Sri Lanka FTA, the 2018 Australia–Peru FTA, the 2019 Brazil–Chile FTA, the 2019 Australia–Indonesia FTA, the 2018 USMCA and the 2019 Japan–US Digital Trade Agreement, as well as in a number of DEAs. The chapter first discusses the USMCA, and then looks at selected DEAs. 2.2.2 The United States–Mexico–Canada Agreement After the United States’ withdrawal from the TPP, there was some uncertainty as to the direction the US would take in its trade deals in general, and on matters of digital trade in particular. The renegotiated NAFTA, which is now referred to as the ‘United States–Mexico–Canada Agreement’ (USMCA), provides a useful confirmation of the US approach. The USMCA has a comprehensive electronic commerce chapter, which is now also properly titled ‘Digital Trade’, follows all critical lines of the CPTPP and creates an even more ambitious template. With regard to replicating the CPTPP model the USMCA follows the same broad scope of application,58 bans customs duties on electronic transmissions59 and binds the parties for non-discriminatory treatment of digital products.60 Furthermore, it provides for a domestic Article 14.17 CPTPP. Article 14.14 CPTPP. 54 Article 14.10 CPTPP. 55 Article 14.10(a) CPTPP. Footnote 6 to this paragraph specifies: ‘The Parties recognise that an Internet access service supplier that offers its subscribers certain content on an exclusive basis would not be acting contrary to this principle.’ 56 On 1 February 2021, the UK formally requested to join the CPTPP and on 2 June 2021, the CPTPP commission agreed to start negotiations. For details on the UK’s goals, see UK Department for International Trade, UK Accession to CPTPP: The UK’s Strategic Approach (London: Department for International Trade, 2021). 57 US Congressional Research Service, ‘China and Taiwan Both Seek to Join the CPTPP’, 24 September 2021, at https://​crsreports​.congress​.gov/​product/​pdf/​IN/​IN11760. 58 Article 19.2 USMCA. 59 Article 19.3 USMCA. 60 Article 19.4 USMCA. 52 53

Digital trade rulemaking in free trade agreements  17 regulatory framework that facilitates online trade by enabling electronic contracts,61 electronic authentication and signatures,62 and paperless trading.63 The USMCA follows the CPTPP model also with regard to data issues and ensures the free flow of data through a clear ban on data localization64 and a hard rule on free information flows.65 Article 19.11 specifies further that parties can adopt or maintain a measure inconsistent with the free flow of data provision, if this is necessary to achieve a legitimate public policy objective, provided that there is no arbitrary or unjustifiable discrimination nor a disguised restriction on trade, and the restrictions on transfers of information are not greater than necessary to achieve the objective.66 Beyond these similarities, the USMCA introduces some novelties. The first is that the USMCA departs from the standard US approach and signals abiding to some data protection principles and guidelines of relevant international bodies. After recognizing ‘the economic and social benefits of protecting the personal information of users of digital trade and the contribution that this makes to enhancing consumer confidence in digital trade’,67 Article 19.8 USMCA requires that the parties adopt or maintain a legal framework that provides for the protection of the personal information of the users of digital trade. In the development of its legal framework for the protection of personal information, each Party should take into account principles and guidelines of relevant international bodies, such as the APEC Privacy Framework and the OECD Recommendation of the Council concerning Guidelines governing the Protection of Privacy and Transborder Flows of Personal Data (2013).68

The parties also recognize key principles of data protection, which include limitation on collection, choice, data quality, purpose specification, use limitation, security safeguards, transparency, individual participation and accountability,69 and aim to provide remedies for any violations.70 This is interesting because it may go beyond what the US has in its national laws on data protection (at least so far71) and also because it reflects some of the principles the EU has advocated for in the domain of privacy protection, not only within the boundaries of the Union but also under the Council of Europe. One can of course wonder whether this is a development caused by the so-called Brussels effect, whereby the EU ‘exports’ its own

Article 19.5 USMCA. Article 19.6 USMCA. 63 Article 19.9 USMCA. 64 Article 19.12 USMCA. 65 Article 19.11 USMCA. 66 Article 19.11(2) USMCA. There is a footnote attached, which clarifies: ‘A measure does not meet the conditions of this paragraph if it accords different treatment to data transfers solely on the basis that they are cross-border in a manner that modifies the conditions of competition to the detriment of service suppliers of another Party.’ The footnote does not appear in the CPTPP treaty text. 67 Article 19.8(1) USMCA. 68 Article 19.8(2) USMCA. 69 Article 19.8(3) USMCA. 70 Article 19.8(4) and (5) USMCA. 71 Chander and Schwarz, supra note 6. 61 62

18  Research handbook on digital trade domestic standards and they become global,72 or whether we are seeing a shift in US privacy protection regimes as well.73 Beyond data protection, three further innovations of the USMCA may be mentioned. The first refers to the inclusion of ‘algorithms’, the meaning of which is ‘a defined sequence of steps, taken to solve a problem or obtain a result’74 and has become part of the ban on requirements for the transfer or access to source code in Article 19.16.75 The second novum refers to the recognition of ‘interactive computer services’ as particularly vital to the growth of digital trade. Parties pledge in this sense not to adopt or maintain measures that treat a supplier or user of an interactive computer service as an information content provider in determining liability for harms related to information stored, processed, transmitted, distributed, or made available by the service, except to the extent the supplier or user has, in whole or in part, created, or developed the information.76

This provision is important, as it seeks to clarify the liability of intermediaries and delineate it from the liability of host providers with regard to IP rights’ infringement. It also secures the application of Section 230 of the US Communications Decency Act,77 which insulates platforms from liability78 but has been recently under attack in many jurisdictions in the face of fake news and other negative developments related to platforms’ power.79 The third and rather liberal commitment of the USMCA parties is with regard to open government data. This is truly innovative and very relevant in the domain of domestic regimes for data governance. In Article 19.18, the parties recognize that facilitating public access to 72 Anu Bradford, ‘The Brussels Effect’, Northwestern University Law Review 107 (2012), 1−68; Anu Bradford, The Brussels Effect: How the European Union Rules the World (Oxford: Oxford University Press, 2020). 73 See Anupam Chander, Margot E. Kaminski and William McGeveran, ‘Catalyzing Privacy Law’, Minnesota Law Review 105 (2021), 1733–1802. 74 Article 19.1 USMCA. 75 On the expansion of the scope of the source code provision, see New Zealand’s Waitangi Tribunal, supra note 25, at 104–12. 76 Article 19.17(2) USMCA. Annex 19-A creates specific rules with the regard to the application of Article 19.17 for Mexico, in essence postponing its implementation for three years. There is also a footnote to the provision, which specifies that a party may comply through ‘application of existing legal doctrines as applied through judicial decisions’. For the argument that Canada’s policy space has remained intact, see Robert Wolfe, ‘Learning about Digital Trade: Privacy and E-Commerce in CETA and TPP’, World Trade Review 18 (2019), s63–s84, at s78. 77 Section 230 reads: ‘No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider’ and in essence protects online intermediaries that host or republish speech. 78 See e.g. Eric Goldman, ‘Why Section 230 Is Better than the First Amendment’, Notre Dame Law Review Reflection 95 (2019), 33–46; Eric Goldman, ‘An Overview of the United States’ Section 230 Internet Immunity’, in Giancarlo Frosio (ed), The Oxford Handbook of Online Intermediary Liability (Oxford: Oxford University Press, 2020), 155–71; Tanner Bone, ‘How Content Moderation May Expose Social Media Companies to Greater Defamation Liability’, Washington University Law Review 98 (2021), 937–63. 79 See e.g. Lauren Feine, ‘Big Tech’s Favorite Law Is Under Fire’, CNBC, 19 February 2020. For an analysis of the free speech implications of digital platforms and literature review, see Mira Burri, ‘Fake News in Times of Pandemic and Beyond: An Enquiry into the Rationales for Regulating Information Platforms’, in Klaus Mathis and Avishalom Tor (eds), Law and Economics of the Coronavirus Crisis (Berlin: Springer, 2022), 31–58.

Digital trade rulemaking in free trade agreements  19 and use of government information fosters economic and social development, competitiveness and innovation. ‘To the extent that a Party chooses to make government information, including data, available to the public, it shall endeavour to ensure that the information is in a machine-readable and open format and can be searched, retrieved, used, reused, and redistributed.’80 There is in addition an endeavour to cooperate, so as to ‘expand access to and use of government information, including data, that the Party has made public, with a view to enhancing and generating business opportunities, especially for small and medium-sized enterprises’.81 Finally, it can be mentioned that the cooperation provision of the USMCA goes beyond the CPTPP82 and envisages an institutional setting to enable this cooperation, ‘or any other matter pertaining to the operation of this chapter’.83 The US approach towards digital trade issues has been confirmed by the recent US–Japan DTA, signed on 7 October 2019, alongside the US–Japan Trade Agreement.84 The US–Japan DTA replicates almost all provisions of the USMCA and the CPTPP,85 including the rules on open government data,86 source code87 and interactive computer services,88 but notably covering also financial and insurance services as part of the scope of agreement. In the current WTO negotiations on electronic commerce, the US has endorsed an ambitious template which is essentially a compilation of the USMCA and the DTA.89 2.2.3 Digital economy agreements Policymakers’ increased preoccupation with digital trade issues can be perhaps best exemplified by the new generation of the so-called digital economy agreements (DEAs). This is a relatively new phenomenon in the trade rulemaking landscape and so far only six such treaties have been adopted – the aforementioned US–Japan Digital Trade Agreement; the 2019 ASEAN Agreement on Electronic Commerce (within the context of ASEAN); the 2020

Article 19.18(2) USMCA. Article 19.8(3) USMCA. 82 The provision envisages, among other things linked to enabling global digital trade, exchange of information and experience on personal information protection, particularly with the view to strengthening existing international mechanisms for cooperation in the enforcement of laws protecting privacy; and cooperation on the promotion and development of mechanisms, including the APEC Cross-Border Privacy Rules, that further global interoperability of privacy regimes. See Article 19.14(1) USMCA, at paras (a)(i) and (b) respectively. 83 Article 19.14(2) USMCA. 84 For the text of the agreements, see: https://​ustr​.gov/​countries​-regions/​japan​-korea​-apec/​japan/​us​ -japan​-trade​-agreement​-negotiations/​us​-japan​-digital​-trade​-agreement​-text. 85 Article 7: Customs Duties; Article 8: Non-Discriminatory Treatment of Digital Products; Article 9: Domestic Electronic Transactions Framework; Article 10: Electronic Authentication and Electronic Signatures; Article 14: Online Consumer Protection; Article 11: Cross-Border Transfer of Information; Article 12: Location of Computing Facilities; Article 16: Unsolicited Commercial Electronic Messages; Article 19: Cybersecurity US–Japan DTA. 86 Article 20 US–Japan DTA. 87 Article 17 US–Japan DTA. 88 Article 18 US–Japan DTA. A side letter recognizes the differences between the US and Japan’s systems governing the liability of interactive computer services suppliers and parties agree that Japan need not change its existing legal system to comply with Article 18. 89 WTO, Joint Statement on Electronic Commerce, Communication from the United States, INF/ ECOM/5, 25 March 2019; WTO, Joint Statement on Electronic Commerce, Communication from the United States, INF/ECOM/23, 26 April 2019. 80 81

20  Research handbook on digital trade Singapore–Australia Digital Economy Agreement (ASDEA); the 2020 Digital Economy Partnership Agreement (DEPA) between Chile, New Zealand, Singapore; the 2021 Korea– Singapore DEA and the 2022 UK–Singapore DEA. What is key to mention at the outset is that these agreements can be adopted as stand-alone initiatives, such as the DEPA, or as part of existing or new trade agreements, such as the ones between Japan and the US, Singapore and Australia, and the UK and Singapore. The DEAs may also differ in scope and the extent to which they include new items on the regulation of the data-driven economy. So, while for instance the US–Japan DTA still very much resembles a conventional, albeit extended, digital trade chapter, the ASDEA, the DEPA and the UK–Singapore DEA go beyond this and engage in entirely new areas of regulatory cooperation, including a mixed set of hard and soft law provisions. This section looks more closely at the DEPA as representative of this latter category and as a model of innovative digital trade rulemaking. The 2020 DEPA between Chile, New Zealand, and Singapore90 all of which are also parties to the CPTPP, is, as earlier noted, conceptualized not as a purely trade agreement but as one that is meant to address the broader issues of the digital economy. In this sense, its scope is wide, open and flexible and covers several emergent issues, such as those in the areas of artificial intelligence (AI) and digital inclusion. The agreement is also not a closed deal but one that is open to other countries,91 and the DEPA is meant to complement the WTO negotiations on electronic commerce and build upon the digital economy work under way within APEC, the OECD and other international forums. To enable flexibility and cover a wide range of issues, the DEPA follows a modular approach that provides countries with more options to pick and choose and is very different from the ‘all-or-nothing’ approach of conventional trade treaties.92 After Module 1, specifying general definitions and initial provisions, Module 2 focuses on ‘Business and Trade Facilitation’; Module 3 ‘Treatment of Digital Products and Related Issues’; Module 4 ‘Data Issues’; Module 5 ‘Wider Trust Environment’; Module 6 ‘Business and Consumer Trust’; Module 7 ‘Digital Identities’; Module 8 ‘Emerging Trends and Technologies’; Module 9 ‘Innovation and the Digital Economy’; Module 10 ‘Small and Medium Enterprises Cooperation’; and Module 11 ‘Digital Inclusion’. The rest of the modules deal with the operationalization and implementation of the DEPA and cover common institutions (Module 12); exceptions (Module 13); transparency (Module 14); dispute settlement (Module 15); and some final provisions on amendments, entry into force, accession and withdrawal (Module 16). The types of rules vary across the different modules. On the one hand, all rules of the CPTPP are replicated; some of the USMCA rules, such as the one on open government data93 (but not source code), and some of the US–Japan DTA provisions, such as the one on ICT goods using cryptography,94 have been included too. On the other hand, there are many other rules – so far unknown to trade agreements – that try to facilitate the functioning of the digital economy and enhance cooperation on key issues. So, for instance, Module 2 on business 90 For details and the text of the DEPA, see: www​.mfat​.govt​.nz/​en/​trade/​free​-trade​-agreements/​free​ -trade​-agreements​-concluded​-but​-not​-in​-force/​digital​-economy​-partnership​-agreement/​. 91 Article 16.2 DEPA. 92 James Bacchus, The Digital Decide: How to Agree on WTO Rules for Digital Trade, Special Report (Waterloo, ON: Centre for International Governance Innovation, 2021), at 8. 93 Article 9.4 DEPA. 94 Article 3.4 DEPA. The article also provides detailed definitions of cryptography, encryption and cryptographic algorithm and cipher.

Digital trade rulemaking in free trade agreements  21 and trade facilitation includes next to the standard CPTPP-like norms95 additional efforts ‘to establish or maintain a seamless, trusted, high-availability and secure interconnection of each Party’s single window to facilitate the exchange of data relating to trade administration documents, which may include: (a) sanitary and phytosanitary certificates and (b) import and export data’.96 Parties have also touched upon other important issues around digital trade facilitation, such as electronic invoicing (Article 2.5); express shipments and clearance times (Article 2.6); logistics (Article 2.4) and electronic payments (Article 2.7). Module 8 on emerging trends and technologies is also particularly interesting to mention, as it highlights a range of key topics that demand attention by policymakers, such as in the areas of fintech and AI. In the latter domain, the parties agree to promote the adoption of ethical and governance frameworks that support the trusted, safe and responsible use of AI technologies, and in adopting these AI Governance Frameworks parties would seek to follow internationally recognized principles or guidelines, including explainability, transparency, fairness and human-centred values.97 The DEPA parties also recognize the interfaces between the digital economy and government procurement and broader competition policy and agree to actively cooperate on these issues.98 Along this line of covering broader policy matters in order to create an enabling environment that is also not solely focused on and driven by economic interests, DEPA deals with the importance of a rich and accessible public domain99 and digital inclusion, which can cover enhancing cultural and people-to-people links, including between Indigenous Peoples, as well as improving access for women, rural populations and low socio-economic groups.100 Overall, the DEPA is a unique project101 that covers well the broad range of issues upon which the digital economy impinges and offers a good basis for harmonization and interoperability of domestic frameworks and international cooperation that adequately takes into account the complex challenges of contemporary data governance that has essential trade but also non-trade elements. Its appeal as a form of enhanced, but also flexible, cooperation on issues of the data-driven economy has been confirmed by Canada’s102 and South Korea’s103 interest in joining it, as well as by the similar DEAs that have followed, such as the ones between the UK and Singapore and between Australia and Singapore.

Article 2.2: Paperless Trading; Article 2.3: Domestic Electronic Transactions Framework. Article 2.2(5) DEPA. ‘Single window’ is defined as a facility that allows parties involved in a trade transaction to electronically lodge data and documents with a single-entry point to fulfil all import, export and transit regulatory requirements (Article 2.1 DEPA). 97 Article 8.2(2) and (3) DEPA. 98 Articles 8.3 and 8.4 DEPA. 99 Article 9.2 DEPA. 100 Article 11.2 DEPA. 101 For a comparison of the DEPA with existing FTAs, see Marta Soprana, ‘The Digital Economy Partnership Agreement (DEPA): Assessing the Significance of the New Trade Agreement on the Block’, Trade, Law and Development 13 (2021), 143–69. 102 Government of Canada, Global Affairs, Background: Canada’s Possible Accession to the Digital Economy Partnership Agreement, 18 March 2021, available at: https://​www​.international​.gc​.ca/​trade​ -commerce/​consultations/​depa​-apen/​background​-information​.aspx​?lang​=​eng. 103 ‘South Korea Starts Process to Join DEPA’, 6 October 2021, available at: https://​en​.yna​.co​.kr/​ view/​PYH20211006124000325. 95 96

22  Research handbook on digital trade 2.2.4 EU’s approach to digital trade The EU has been a relatively late mover on digital trade issues and for a long time had not developed a distinct strategy. Although the EU’s FTAs did include provisions on electronic commerce, such as the 2002 agreement with Chile, the language tended to be cautious, with commitments not exceeding GATS levels, and limited to soft cooperation pledges in the services chapter104 and in the fields of information technology, information society and telecommunications.105 In more recent agreements, such as the EU–South Korea FTA (signed in 2009), the language is more concrete and binding, imitating some of the US template provisions – for instance, by confirming the applicability of the WTO Agreements to measures affecting electronic commerce and subscribing to a permanent duty-free moratorium on electronic transmissions. Cooperation is also increasingly framed in more concrete terms and includes mutual recognition of electronic signature certificates, coordination on internet service providers’ liability, consumer protection and paperless trading.106 The EU, particularly insistent on data protection policies, has also sought commitments from its FTA partners as to compatibility with the international standards of data protection.107 The 2016 EU agreement with Canada – the Comprehensive Economic and Trade Agreement (CETA) – goes a step further. The CETA provisions concern commitments ensuring (a) clarity, transparency and predictability in their domestic regulatory frameworks and (b) interoperability, innovation and competition in facilitating electronic commerce, as well as (c) facilitating the use of electronic commerce by small and medium sized enterprises.108 The EU has succeeded in deepening the privacy commitments and the CETA has a specific norm on trust and confidence in electronic commerce, which obliges the parties to adopt or maintain laws, regulations or administrative measures for the protection of personal information of users engaged in electronic commerce in consideration of international data protection standards.109 Yet, there are no deep commitments on digital trade; nor there are any rules on data and data flows.110 It is only very recently that the EU took up a more modern, CPTPP-comparable, approach towards the regulation of digital trade. Some indications for this turn were given by the 2018 EU–Japan Economic Partnership Agreement (EPA)111 and the modernization of the trade part of the EU–Mexico Global Agreement, where for the first time data flows were mentioned but still cautiously, as the Parties only committed to ‘reassess’ within three years of the entry into force of the agreement, the need for actually including provisions on the free flow of data. The new EU approach towards the issue of cross-border data is now fully endorsed in the EU’s currently negotiated deals with Australia and Tunisia, and the 2022 agreement with New Zealand. These FTAs’ digital trade chapters include norms on the free flow of data and data

104 Article 102 EU–Chile FTA. The agreement states that ‘[t]he inclusion of this provision in this Chapter is made without prejudice of the Chilean position on the question of whether or not electronic commerce should be considered as a supply of services’. 105 Article 37 EU–Chile FTA. 106 Article 7.49 EU–South Korea FTA. 107 Article 7.48 EU–South Korea FTA. 108 Article 16.5 CETA. 109 Article 16.4 CETA. 110 See e.g. Wolfe, supra note 75. 111 Article 8.81 EU–Japan EPA.

Digital trade rulemaking in free trade agreements  23 localization bans. This repositioning and newer commitments are however also linked with high levels of data protection.112 The EU wishes to permit data flows only if coupled with the high standards of its General Data Protection Regulation (GDPR)113 and endorses a distinct model of privacy as a fundamental right. While the EU and its partners seek to permit the flow of data, these commitments are conditioned: first by a dedicated article on data protection, which clearly states that: ‘Each Party recognises that the protection of personal data and privacy is a fundamental right and that high standards in this regard contribute to trust in the digital economy and to the development of trade’,114 which is followed by a paragraph on data sovereignty: Each Party may adopt and maintain the safeguards it deems appropriate to ensure the protection of personal data and privacy, including through the adoption and application of rules for the cross-border transfer of personal data. Nothing in this agreement shall affect the protection of personal data and privacy afforded by the Parties’ respective safeguards.115

The EU also wishes to retain the right to see how implementation of the provisions on data flows impacts the conditions of privacy protection, so there is a review possibility within three years of the entry into force of the agreement, and parties remain free to propose review of the list of restrictions at any time.116 In addition, there is a broad carve-out, in the sense that The Parties reaffirm the right to regulate within their territories to achieve legitimate policy objectives, such as the protection of public health, social services, public education, safety, the environment including climate change, public morals, social or consumer protection, privacy and data protection, or the promotion and protection of cultural diversity.117

The EU thus reserves ample regulatory leeway for its current and future data protection (and other) measures. The exception is also fundamentally different than the objective necessity test under the CPTPP and the USMCA, or that under WTO law, because it is subjective and safeguards the EU’s right to regulate.118

See European Commission, Horizontal Provisions for Cross-Border Data Flows and for Personal Data Protection in EU Trade and Investment Agreements, February 2018, available at: https://​trade​.ec​ .europa​.eu/​doclib/​docs/​2018/​may/​tradoc​_156884​.pdf. 113 Regulation 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation), OJ 2016 L 119/1 [hereinafter GDPR]. 114 See e.g. Article 6(1) draft EU–Australia FTA (emphasis added). The same wording is found in the EU–New Zealand and the draft EU–Tunisia FTA. 115 See e.g. Article 6(2) draft EU–Australia FTA. The same wording is found in the EU–New Zealand and the draft EU–Tunisia FTA. 116 See e.g. Article 5(2) draft EU–Australia FTA. The same wording is found in the EU–New Zealand and the draft EU–Tunisia FTA. 117 See e.g. Article 2 draft EU–Australia FTA. The same wording is found in the EU–New Zealand and the draft EU–Tunisia FTA. 118 Svetlana Yakovleva, ‘Privacy Protection(ism): The Latest Wave of Trade Constraints on Regulatory Autonomy’, University of Miami Law Review 74 (2020), 416–519, at 496. 112

24  Research handbook on digital trade The new EU approach has been confirmed by the post-Brexit Trade and Cooperation Agreement (TCA) with the United Kingdom,119 which replicates all the above provisions, except for the explicit mentioning of data protection as a fundamental right – which can be however presumed, since the UK incorporates the European Convention on Human Rights (ECHR) through the Human Rights Act of 1998 into its domestic law.120 Yet, as the UK seems to be moving away from the EU FTA model as well as from the GDPR standards domestically, this presumption may be somewhat questioned. Beyond the topic of data flows and its interface with data protection, it should be noted that the rest of the EU digital trade template includes the issues covered by the CPTPP/USMCA model, such as software source code,121 facilitation of electronic commerce,122 online consumer protection,123 spam124 and open government data;125 not including, however, a provision on non-discrimination of digital products, and excluding audiovisual services from the scope of the application of the digital trade chapter.126 This TCA template is also the one that the EU has endorsed under the WTO electronic commerce negotiations.127 2.2.5 The Regional Comprehensive Economic Partnership An interesting and much anticipated development against the backdrop of the diverging, at least on data flows, EU and US positions was the recent Regional Comprehensive Economic Partnership (RCEP) signed on 15 November 2020 between the ASEAN Members,128 China, Japan, South Korea, Australia and New Zealand, and in force since 1 January 2022.129 Chapter 12 of the RCEP includes the relevant electronic commerce rules. In a similar fashion to the 119 Trade and Cooperation Agreement between the European Union and the European Atomic Energy Community, of the one part, and the United Kingdom of Great Britain and Northern Ireland, of the other part, OJ L [2020] 444/14. 120 See e.g. Kristina Irion and Mira Burri, ‘Digitaler Handel (Commentary of the Digital Trade Title of the EU-UK Trade and Cooperation Agreement)’, in Gesa Kübek et al. (eds) Handels- und Kooperationsvertrag EU/GB Handbuch (Baden-Baden: Nomos, 2022), 343–68. 121 Article 207 TCA. Again with notable safeguards, specified in paras 2 and 3 of Article 207, including the general exceptions, security exceptions and prudential carve-out in the context of a certification procedure; voluntary transfer of source code on a commercial basis, a requirement by a court or administrative tribunal, or a requirement by a competition authority pursuant to a party’s competition law to prevent or remedy a restriction or a distortion of competition; a requirement by a regulatory body pursuant to a party’s laws or regulations related to the protection of public safety with regard to users online; the protection and enforcement of IP; and government procurement-related measures. 122 Articles 205 and 206 TCA. 123 Article 208 TCA. 124 Article 209 TCA. 125 Article 210 TCA. 126 Article 197(2) TCA. 127 WTO, Joint Statement on Electronic Commerce, EU Proposal for WTO Disciplines and Commitments Relating to Electronic Commerce, Communication from the European Union, INF/ ECOM/22, 26 April 2019; WTO, Joint Statement on Electronic Commerce, Communication from the European Union, INF/ECOM/13, 25 March 2019. 128 Brunei, Cambodia, Indonesia, Laos, Malaysia, Myanmar, Philippines, Singapore, Thailand and Vietnam. 129 RCEP entered into force on 1 January 2022 for ten original parties: Australia, New Zealand, Brunei Darussalam, Cambodia, China, Japan, Laos, Singapore, Thailand and Vietnam. RCEP entered into force for the Republic of Korea on 1 February 2022 and for Malaysia on 18 March 2022. For the details and the text of RCEP, see https://​rcepsec​.org/​legal​-text/​.

Digital trade rulemaking in free trade agreements  25 CPTPP, it clarifies its application ‘to measures adopted or maintained by a Party that affect trade by electronic means’ but excludes from this broad scope (1) government procurement and (2) information held or processed by or on behalf of a Party, or measures related to such information, including measures related to its collection. In addition, key provisions on the location of computing facilities and the cross-border transfer of information by electronic means apply in conformity with obligations established in the chapters on trade in services (Chapter 8) and on investment (Chapter 10). The RCEP electronic commerce chapter rules are grouped into four areas: (1) trade facilitation; (2) creation of a conducive environment for electronic commerce; (3) promotion of cross-border electronic commerce; and (4) others. With regard to trade facilitation, RCEP includes provisions on paperless trading130 on and electronic authentication and electronic signatures.131 On paperless trading, the RCEP Members avoid entering into binding commitments. They, instead, commit to terms such as ‘work toward’, ‘endeavour’ or ‘cooperate’.132 The norms on accepting the validity of electronic signatures are more binding but, in contrast to the CPTPP and USMCA, permit domestic laws and regulations to provide otherwise and to prevail in case of inconsistency. Regarding commitments to create a conductive environment for electronic commerce, the inclusion of provisions on online personal information protection133 and cybersecurity134 is remarkable. On the former, RCEP Members establish that they shall adopt or maintain a legal framework, which ensures the protection of personal information. Unsurprisingly, RCEP is not prescriptive as to how parties may comply with this obligation. As for the latter aspect on cybersecurity, the parties do not establish a binding provision but recognize the importance of building capabilities and using existing collaboration mechanisms to cooperate. The RCEP Members also commit to adopt or maintain laws or regulations regarding online consumer protection135 and unsolicited commercial electronic messages136 and a framework governing electronic transactions that takes into account international instruments,137 as well as committing to transparency.138 The next grouping of RCEP provisions is critical, as it deals with cross-border data flows. In essence, and actually similarly to the EU, the RCEP provides only for conditional data flows, while preserving room for domestic policies, which well may be of a data-protectionist nature. So, while the RCEP electronic commerce chapter includes a ban on localization measures139 as well as a commitment to free data flows,140 there are clarifications that give RCEP Members a lot of policy space and essentially undermine the impact of the commitments made. Along this lines, there is a possible exception for legitimate public policies and a footnote to Article 12.14.3(a) says: ‘For the purposes of this subparagraph, the Parties affirm that the necessity behind the implementation of such legitimate public policy shall be decided by the implement-

132 133 134 135 136 137 138 139 140 130 131

Article 12.5 RCEP. Article 12.6 RCEP. Article 12.5 RCEP. Article 12.8 RCEP. Article 12.13 RCEP. Article 12.7 RCEP. Article 12.9 RCEP. Article 12.10 RCEP. Article 12.12 RCEP. Article 12.14 RCEP. Article 12.15 RCEP.

26  Research handbook on digital trade ing Party.’ This essentially goes against any exceptions assessment as we know it under WTO law, and triggers a self-judging mechanism. In addition, subparagraph (b) of Article 12.14.3 says that the provision does not prevent a party from taking ‘any measure that it considers necessary for the protection of its essential security interests. Such measures shall not be disputed by other Parties.’141 Article 12.15 on cross-border transfer of information follows the same language and thus secures plenty of policy space, for countries such as China or Vietnam, to control data flows without further justification. Other provisions contained in the RCEP electronic commerce chapter include the establishment of a dialogue on electronic commerce142 and a provision on dispute settlement,143 which is separate from general RCEP dispute settlement.144 Noteworthy are also some things missing from the RCEP: in comparison to the CPTPP, RCEP does not include provisions on custom duties, non-discriminatory treatment of digital products, source code, principles on access to and use of the internet for electronic commerce and internet interconnection charge sharing. It is finally interesting to observe that the RCEP does not necessarily reflect China’s position in the WTO negotiations, where China has been more cautious and somewhat fuzzy in its demands – for instance, by subscribing to a very narrow definition of digital trade arguing that the negotiations should focus on the discussion of cross-border trade in goods enabled by the internet, together with relevant payment and logistics services, while paying attention to the digitization trend of trade in services,145 and not engaging in commitments on data flows. It is also interesting to contemplate how China’s wish to join the CPTPP would impact on the WTO negotiations as well as largely on the landscape of digital trade rulemaking – perhaps tilting it towards national security and other carve-outs that seriously diminish the value of the commitments made and prejudice legal certainty.

3.

CONCLUDING REMARKS AND OUTLOOK

The above analysis of the developments in FTAs reveals the critical importance of digital trade as a negotiation topic and the substantial efforts made, in particular in recent years, to address it and create an adequate rule framework. The achievements made in some FTAs and the dedicated digital economy agreements are quite impressive and there is a strand of legal innovation that seeks to tackle not only the ‘old’ issues raised under the 1998 WTO Electronic Commerce Programme but also the contemporary issues in the context of a global data-driven economy. The regulatory environment is however still highly fluid. Despite convergence on certain issues of digital trade facilitation, there are many points of divergence among the major stakeholders, in particular with regard to permitting cross-border data flows and the interface between economic and non-economic issues, as the latter effectively determine the digital sovereignty of states and their ability to protect the interests of their citizenry. One can also 141 Emphasis added. The ‘essential security interest’ language has been endorsed by China also in the framework of the WTO electronic commerce negotiations. 142 Article 12.16 RCEP. 143 Article 12.17 RCEP. 144 Chapter 12 RCEP. There is a possibility for this to change after a review of the chapter (Article 12.17(3) RCEP). 145 WTO, Joint Statement on Electronic Commerce, Communication from China, INF/ECOM/19, 24 April 2019, at section 3 (China Communication 1), at para. 2.5.

Digital trade rulemaking in free trade agreements  27 observe two important trends in this context. First, it is clear that the FTA efforts serve as regulatory laboratories and, while reflecting the positions of the key stakeholders, do matter for the multilateral endeavours and for what the shape and substance of any agreement on electronic commerce under the umbrella of the WTO will be.146 The second trend has to do with the forming of geopolitical blocks with overlaps that may lead to potential contestations, as well as uncertainties as to the impact of the agreements on the ground – in this context, we see for instance that New Zealand is a member of the CPTPP, the RCEP and the DEPA and also has an agreement with the EU; similarly the UK has a deal with the EU, while also entering into ambitious digital trade commitments under DEAs and the CPTPP. Overall, the regulatory landscape of digital trade rulemaking is likely to remain dynamic, as technological advances would demand new regulatory responses (for instance with regard to AI) and as countries continue to position themselves, either by starting to actively participate in new rulemaking (like many Central and Latin American countries), by forming new geo-blocks or by becoming legal entrepreneurs in departure from older stances (as with the United Kingdom in its new generation of FTAs moving away from EU positions). The coming years will also test the willingness for international cooperation in the domain of digital trade regulation, and to what extent achievements made in the FTA venues can be multilateralized and brought back to the forum of the WTO.147

Mira Burri, ‘A WTO Agreement on Electronic Commerce: An Enquiry into Its Substance and Viability’, Georgetown Journal of International Law 53 (2023) 565–625. 147 Burri, ibid. 146

3. Three generations of digital trade provisions in preferential trade agreements Rodrigo Polanco

1. INTRODUCTION Digital trade is one of the key topics in international trade policy today. In recent years it has received increased attention and study in the field of international economic law and moved up the agenda of States and international organisations. In the past two decades, the development of the internet allowed the growth of electronic commerce (‘e-commerce’). With new digital technologies and the Internet of Things, the flow of data – both commercial and personal – has increased to new levels. Therefore, regulating electronic commerce and data flows today has become one of the most critical topics in international law and policy. Today rules concerning digital trade are complex and spread across diverse fora on several different issues, including domestic laws, bilateral and regional international instruments, as well as plurilateral attempts at the World Trade Organization (WTO).1 Traditional trade rules could serve as a starting point to deal with these issues, but clearly, they were not enough. The ‘modern’ trade law rules were not designed considering the current characteristics of electronic commerce and data flows. To provide some context, in 1994 – when the WTO and its agreements were established – neither Google, nor Amazon, nor Facebook existed. Mosaic was the most used web browser (Netscape Navigator was created in 1994, and Internet Explorer was only released in 1995).2 Efforts to deal with these issues at a multilateral level started in 1998 when the WTO established a work programme on electronic commerce. At the Ministerial Conference that same year, members agreed on a temporary duty-free moratorium on all electronic transactions – a practice that has since been renewed at each WTO Ministerial Conference.3 Further development has been slow-paced; we are still far from achieving consensus on this topic, and any agreement would be, at best, plurilateral. In December 2017, 44 WTO members made a joint declaration to initiate exploratory work together towards future negotiations on trade-related aspects of electronic commerce.4 In July 2023, 86 WTO members of this joint statement ini-

1 Taku Nemoto and Javier López González, ‘Digital Trade Inventory. Rules Standards and Principles’ (2021) 251 OECD Trade Policy Paper 156. 2 Aaron Schwabach, Internet and the Law: Technology, Society, and Compromises, 2nd Edition: Technology, Society, and Compromises (ABC-CLIO 2014) xxi. 3 WTO, ‘Work Programme on Electronic Commerce. Adopted by the General Council on 25 September 1998. WT/L/274’ (30 September 1998) www​.wto​.org/​english/​tratop​_e/​ecom​_e/​wkprog​_e​ .htm accessed 9 June 2022. 4 WTO, ‘Joint Ministerial Statement on Electronic Commerce WT/MIN(17)/65. WT/L/1032’ (World Trade Organization (WTO), 13 December 2017) www​.wto​.org/​english/​thewto​_e/​minist​_e/​mc11​ _e/​documents​_e​.htm accessed 5 July 2018.

28

Three generations of digital trade provisions  29 tiative (JSI) set the goal of securing convergence on most issues by the end of 2023.5 As yet, countries have not been able to agree on a multilateral regime for the treatment of e-commerce and data flows.6 But the lack of consensus at the multilateral level does not mean that rules for digital trade are not being created elsewhere. In fact, since the beginning of the twenty-first century, certain countries have been including provisions or chapters on electronic commerce and rules on data flows in preferential trade agreements (PTAs), free trade agreements (FTAs) being the most common type of treaty to include such provisions. This chapter provides an inventory of existing rules, standards and principles related to digital trade found in preferential trade agreements (PTAs) and traces its evolution in three distinctive periods: a first focused on electronic commerce rules; a second where data flows rules were developed; and a third that has seen the appearance of ‘pure’ digital trade agreements and new data economy rules. We rely on the information available in the Trade Agreements Provisions on Electronic Commerce and Data (TAPED) dataset, which includes a detailed mapping and coding of all PTAs chapters, provisions, annexes and side documents that directly or indirectly regulate digital trade.7

2. OVERVIEW The advent of agreements including provisions on electronic commerce started in the new century, kicked off by European Union (EU) and United States (US) initiatives in this regard such as the EU’s 2000 Directive on e-commerce8 and the US Digital Agenda (with the Bipartisan Trade Promotion Authority Act of 2002).9 From 645 PTAs concluded between 2000 and 2022, we find 384 agreements with provisions related to digital trade. The most significant number of them is found in e-commerce; overall, the provisions remain, however, highly heterogeneous. The first e-commerce provision is found in the Jordan–US FTA (2000), although a joint statement on electronic commerce by the same parties actually predates that treaty by some months.10 Three years later, the Australia–Singapore FTA (2003)11 was the first treaty to have a dedicated chapter on e-commerce. Currently, specific provisions applicable to e-commerce can be found in 167 PTAs, mostly in dedicated chapters (109 in the TAPED dataset). WTO, ‘Co-convenors of E-commerce Negotiations Review Progress, Reflect on Way Forward’ (6 July 2023) https://​www​.wto​.org/​english/​news​_e/​news23​_e/​jsec​_06jul23​_e​.htm accessed 24 July 2023. 6 Mira Burri and Thomas Cottier (eds), Trade Governance in the Digital Age: World Trade Forum (Cambridge University Press 2014); Sacha Wunsch-Vincent, The WTO, the Internet and Trade in Digital Products: EC–US Perspectives (Hart Publishing 2006). 7 Mira Burri and Rodrigo Polanco, ‘Digital Trade Provisions in Preferential Trade Agreements: Introducing a New Dataset’ (2020) 23 Journal of International Economic Law 187. 8 Directive 2000/31/EC of the European Parliament and of the Council of 8 June 2000 on certain legal aspects of information society services, in particular electronic commerce, in the Internal Market, OJ L 178 (2000). 9 ‘Bipartisan Trade Promotion Authority Act of 2002. 19 USC Ch. 24’ (2002) https://​uscode​.house​ .gov/​view​.xhtml​?path​=/​prelim@​title19/​chapter24​&​edition​=​prelim accessed 8 June 2022. 10 Jordan-US FTA, Art. 7 (concluded on 24 October 2000, in force 17 December 2001). The Joint Statement on E-Commerce was signed on 7 June 2000. 11 Australia–Singapore FTA, Ch. 14. 5

30  Research handbook on digital trade Although the number of PTAs incorporating specific digital trade provisions remains limited, the past ten years have witnessed a significant increase in the number of agreements with such provisions. As shown in Figure 3.1, digital trade provisions are, on average, included in more than 40 per cent of all PTAs concluded between 2011 and 2021. The correlation is even closer if we consider the trade agreements that have been notified to the WTO, being included in around 90 per cent of them in the same period. According to Willemyns, two-thirds of the WTO Members are Party to a PTA with e-commerce-related provisions.12 See Figure 3.1.

Source: Based on data from the TAPED database, www.unilu.ch/en/faculties/faculty-of-law/professorships/ burri-mira/research/taped/

Figure 3.1

Evolution of PTAs with digital trade provisions

In total, 73 countries and 8 regional economic integration organisations (REIOs) have concluded at least one agreement with digital trade provisions. Singapore is the country with more of these agreements (24 PTAs), followed by Australia (17 PTA), Chile (16 PTAs), the US (15 PTAs), Korea (13 PTAs), Canada (12 PTAs), and Colombia, Japan and New Zealand (with 11 PTAs each). This shows the relevance of the Asia-Pacific region in the treaty-making of such provisions. From the REIOs, the EU leads by far with 24 PTAs, followed by EFTA (5 PTAs) and ASEAN (4 PTAs). Few African countries have concluded PTAs with digital trade Ines Willemyns, Digital Services in International Trade Law (Cambridge University Press 2021) 311 www​.cambridge​.org/​core/​books/​digital​-services​-in​-international​-trade​-law/​77​3170A6CFFF​ 36919550F7A2D1017E26 accessed 8 June 2022. 12

Three generations of digital trade provisions  31 provisions (Côte d’Ivoire and Mauritius, with two agreements each), and no intra-African agreement has been negotiated on this topic. This rise in PTAs with digital trade provisions is driven mainly by bilateral agreements. Of these agreements concluded since 2000, 85 per cent have only two parties (178 PTAs); among these treaties, 61 per cent include digital trade provisions (109 PTAs), as depicted in Figure 3.2. Most of these agreements are intercontinental (81 PTAs). At an intra-regional level, 25 of such PTAs have been concluded in the Americas, 22 PTAs in Asia and 5 in Europe.

Source: Based on data from the TAPED database, www.unilu.ch/en/faculties/faculty-of-law/professorships/ burri-mira/research/taped/

Figure 3.2

Type of PTA with digital trade provisions

The rise of PTAs having digital trade provisions involves both developed and developing countries. Following the country classification of the UN World Economic Situation and Prospects report,13 as of November 2022, 57 per cent of the PTAs with digital trade provisions were negotiated between developed and developing countries (76 PTAs) and 35 per cent between developing countries (47 PTAs). Developed countries only concluded 8 per cent (10 PTAs) of the agreements with e-commerce provisions. Among the PTAs, the number of agreements with digital trade clauses has increased significantly over the years. As of November 2022, eight is the average number of provisions found in e-commerce chapters of PTAs, and if we consider the past five years, that average goes up to 13. Likewise, the level of detail has also increased significantly over time. The average

United Nations, Department of Economic and Social Affairs, World Economic Situation and Prospects 2022 (2022) www​.un​.org/​development/​desa/​dpad/​publication/​world​-economic​-situation​-and​ -prospects​-2022/​accessed 8 June 2022. 13

32  Research handbook on digital trade number of words found in e-commerce chapters of PTAs is 1189, and if we examine the past five years that average goes up to 1879, as shown in Figure 2.3. At the moment of writing, the Digital Economy Partnership Agreement (DEPA) concluded between Chile, New Zealand and Singapore on 12 June 2020, and in force since 8 December of the same year, is the agreement with the highest number of provisions on digital trade (65 in total), with 10887 words.

Source: Based on data from the TAPED database, www.unilu.ch/en/faculties/faculty-of-law/professorships/ burri-mira/research/taped/

Figure 3.3

Average number of words in PTAs with digital trade provisions

One of the most common objectives of PTAs with digital trade chapters is to minimise the regulatory burden on electronic commerce (84 PTAs), usually found under provisions on the domestic electronic transaction framework. Similarly, 63 PTAs include provisions to promote and facilitate e-commerce. However, both types of clauses are largely non-binding, or part of cooperation activities. Some formulations include creating a favourable environment for global electronic commerce with clear, transparent and predictable domestic regulation.14

14 Brazil–Chile FTA, Art. 10.2.5(a); Argentina–Chile FTA, Art. 11.2.5(a); Chile–Uruguay FTA, Art. 8.2.5(a); Canada–Korea FTA, Art. 13.2.2(a); Mexico–Panama FTA, Art. 14.3.2(a); PAAP, Art. 13.3(a) Canada–Honduras FTA, Art. 16.2(a); Colombia–Costa Rica FTA, Art. 16.2.2(a); Canada–Panama FTA, Art. 15.03.2(a); Canada–Colombia FTA, Art. 1502.2(a); Canada–Peru FTA, Art. 1502.2(a).

Three generations of digital trade provisions  33 Several agreements foster, among other things, an environment of trust and confidence for e-commerce consumers or users,15 or promote interoperability, innovation and competition.16 Very few trade agreements contain references to specific international standards. When this happens, the most referenced is the UNCITRAL Model Law on Electronic Commerce, which is mentioned in 26 PTAs. Although these provisions have several variations, they are mainly binding and stipulate that each Party shall ‘maintain’ or ‘adopt’ laws and regulations governing electronic transactions based on the 1996 Model Law.17 Only in recent years, few agreements (10 PTAs) include provisions on the consistency of the domestic legal framework with the United Nations Convention on the Use of Electronic Communications in International Contracts (UNECC), which are also primarily binding.18 It is also striking how few of these agreements refer to WTO rules when regulating digital trade. Most PTAs do not include a provision referring to the applicability of WTO rules to e-commerce. From the 133 PTAs that have e-commerce provisions, only 58 include such reference, with significant differences across treaties. Some explicitly recognise the applicability of the WTO rules to e-commerce but do not specify which the relevant rules are. Such provisions are found mainly in some of the PTAs concluded by the EU,19 Australia, Canada, China, South Korea and Singapore.20 In a similar provision, found mainly in agreements concluded by Colombia and the US, the Parties recognise that WTO rules apply when they affect electronic commerce.21 In other softer variations, countries merely reaffirm commitments under the

15 USMCA, Art. 19.2.1; Australia–Singapore FTA (2017), Ch. 14, Art. 2.1; Japan–Mongolia FTA, Art. 9.1.2; Mexico–Panama FTA, Art. 14.3.2(f); Central America–EFTA FTA, Annex II, Art. 1(c). 16 Brazil–Chile FTA, Art. 10.2.5(c); Argentina–Chile FTA, Art. 11.2.5(c); CETA, Art. 16.5(b); Chile–Uruguay FTA, Art. 8.2.5(c); Canada–Korea FTA, Art. 13.2.2(c); Canada–Panama FTA, Art. 15.03.2.c; Canada–Peru FTA, Art. 1502.2(c); Canada–Peru FTA, Art. 1502.2(b). 17 Australia–Indonesia CEPA, Art. 13.9.1; USMCA, Art. 19.5.1; TPP/CPTPP, Art. 14.5.1; Australia– Peru FTA, Art. 13.5.1; Australia–Singapore FTA (2016), Ch. 14, Art. 6.1; Australia–China FTA, Art. 12.5.1; Australia–Malaysia FTA, Art. 15.5.2(a); New Zealand–Thailand FTA, Art. 10.3.1; Australia– Thailand FTA, Art. 1103.1; Australia–Singapore FTA (2003), Ch. 14, Art. 4.1. 18 See e.g. Singapore–Turkey FTA, Art. 9.5.1; TPP/CPTPP, Art. 14.5.1; RCEP, Art. 12.10.1; Australia–UK FTA, Art. 14.4.1. 19 EU–Singapore FTA, Art. 8.57.1; CETA, Art. 16.2.1; EC–Ukraine, Art. 85.1; EC–Georgia FTA, Art. 76.1. 20 Israel–Korea FTA, Art. 13.1; Central America–Korea FTA, Art. 14.1.1; Australia–China FTA, Art. 12.1; China–Georgia FTA, Art. 12.2.1; China–Korea FTA, Art. 13.1; Canada–Korea FTA, Art. 13.2.1; Colombia–Panama FTA, Art. 19.2.1; Canada–Panama FTA, Art. 15.03.1; Canada–Colombia FTA, Art. 1502.1; Canada–Peru FTA, Art. 1502.1; Panama–Singapore FTA, Art. 13.1; Korea–Singapore, Art. 14.2; India–Singapore ECA, Art. 10.1; New Zealand–Thailand FTA, Art. 10.1; Australia–Thailand FTA, Art. 1101.1; Singapore–US FTA, Art. 14.1; Australia–Singapore FTA (2003), preamble. 21 Canada–Honduras FTA, Art. 16.2: Australia–Korea FTA, Art. 15.1; Colombia–Costa Rica FTA, Art. 16.2.1; Colombia–Korea FTA, Art. 12.1.1; Australia–Malaysia, Art. 15.1; Central America– Mexico, Art. 15.2.1; Korea–Peru FTA, Art. 14.1.1; EC–Korea FTA, Art. 7.48.1; Costa Rica–Singapore Art. 12.1.1; Colombia–Northern Triangle FTA, Art. 14.2.1; Korea–US FTA, Art. 15.1; Panama–US TPA, Art. 14.1.1; Colombia–US, Art. 15.1.1; Nicaragua–Taiwan FTA, Art. 14.01.1; Peru–US, Art. 15.1.1; Oman–US FTA, Art. 14.1; Bahrain–US FTA, Art. 13.1; CAFTA–Dominican Republic–United States (CAFTA–DR–US), Art. 14.1.1; Morocco–US FTA, Art. 14.1; Australia–US FTA, Art. 16.1.

34  Research handbook on digital trade scope of WTO Agreements,22 or endeavour to develop their e-commerce legal framework by applying decisions taken within the WTO.23

3.

FIRST GENERATION: ELECTRONIC COMMERCE RULES

The first generation of trade agreements with provisions concerning digital trade are pretty much anchored to the early days of internet commerce, focused on commercial transactions on a digital market, of physical products. When included, digital products are usually understood as ‘computer programs, text, video, images, sound recordings and other products that are digitally encoded, regardless of whether they are fixed on a carrier medium or transmitted electronically’.24 Five types of clauses stand as particularly relevant in this regard: a ban on customs duties, provisions on electronic authentication, paperless trading, unsolicited commercial electronic messages and consumer protection. Other clauses concerning e-commerce are less prominent or even absent in PTAs, like those related to non-discrimination and delivery postal costs. The number of trade agreements, including explicit commitments on the non-discriminatory treatment of digital products, is relatively small. In all, 42 PTAs include national treatment obligations, and only 36 PTAs include most-favoured-nation treatment regarding electronic commerce. It is noteworthy that most of these provisions are binding.25 Interestingly, one key element for e-commerce of physical products was (and still is) left entirely outside of the scope of trade agreements: the postal costs of delivery of parcels, which is overseen by the Universal Postal Union (UPU), through a set of various international agreements, and are constantly reviewed at the UPU’s quadrennial Universal Postal Congress. When a customer sends a postal item abroad, delivery costs must be settled between the sending and the delivering post. In principle, the sending country assumes the costs of collecting and transporting the item, and the sending country should only reimburse the destination country for the costs incurred in providing the delivery service. UPU member countries agree on such remuneration rates. Currently, there are separate remuneration systems for parcel and Express Mail Service items.26 However, agreements on such remunerations are not always an easy affair. In 2018 the US announced that it would exit the UPU if the members did not agree to reform the remuneration rates, complaining that the system allowed countries such as

22 EU–Armenia CEPA, Art. 141.1; Colombia–EU–Peru FTA, Art. 107.1; EC–Moldova FTA, Art. 202.1; Jordan–US FTA, Joint Statement on Global Electronic Commerce. 23 EAEU–Vietnam FTA, Art. 13.7.a). 24 Australia–US FTA, Art. 16.3; Korea–US FTA, Art. 15.3; Canada–Peru FTA, Art. 1503; Central America–Korea FTA, Art. 14.3.1–2. 25 Only in the recent Brazil–Chile FTA (2018) is there a softer approach, whereby the parties recognize the existence of a debate in international fora (such as the WTO) on the application of non-discriminatory treatment in commerce carried out by electronic means and undertake to jointly evaluate the results of these discussions, so as to decide on the possible incorporation of norms of non-discrimination of electronically transmitted content (Art. 10.4). 26 Universal Postal Union (UPU), ‘Remuneration’, www​.upu​.int/​en/​Universal​-Postal​-Union/​ Activities/​Physical​-Services/​Remuneration accessed 9 June 2022.

Three generations of digital trade provisions  35 China to pay heavily subsidised rates due to being considered a developing country. Although an agreement was reached in September 2019, this situation evidenced the system’s fragility.27 3.1

Ban on Custom Duties

WTO members have continued the moratorium on imposition of customs duties on electronic transmissions at all Ministerial Conferences since the start of the WTO E-Commerce Work Programme in 1998, but have failed to make it permanent. In fact, in 2019, some countries, such as India and South Africa, argued that the e-commerce moratorium in the WTO led to a loss of revenue as it gave such transmissions immunity from taxation, and initially opposed its renewal.28 This situation probably explains why one of the most common provisions in agreements with digital trade rules is a ban on customs duties (97 PTAs), starting with the Jordan–US FTA (2000). Despite being commonplace, these commitments have significant differences in wording. The earliest type of these provisions merely recognises the practice of not imposing customs duties on electronic transmissions between the Parties.29 In some instances, the agreements refer to the WTO Work Programme on e-commerce, and the Parties pledge to cooperate to make this practice binding in the WTO framework.30 Yet, the most common approach is a provision for a permanent moratorium on duty-free treatment, meaning that no customs duties should be imposed on electronic transmissions and digital products. Here again, there are several variations. Some PTAs plainly stipulate that a Party may not apply customs duties on digital products of the other Party,31 or in more binding terms ‘shall’ not impose such duties,32 or generally agree not to apply them, as well as any fees or charges on import or export of digital products by electronic means.33 In certain agreements, mostly concluded by the EU, the Parties agree that electronic transmission or delivery shall be considered as the provision of services and, therefore, cannot be subject to customs duties.34 Another group of PTAs extends the moratorium only to customs duties, fees or other charges, on or in connection with the importation or exportation of digital products, 27 Logan Nagle, ‘Going Postal: President Trump and the United States’ Tumultuous Current Relationship with the Universal Postal Union, and Its Effect on the International Shipment of Opioids’ (2020) 8 Penn State Journal of Law and International Affairs 264. 28 Kirtika Suneja, ‘Setback for India as WTO Extends Nil Tax on E-Transmissions’ The Economic Times (11 December 2019) https://​economictimes​.indiatimes​.com/​news/​economy/​foreign​-trade/​setback​ -for​-india​-as​-wto​-extends​-nil​-tax​-on​-e​-transmissions/​articleshow/​72465653​.cms. 29 Jordan–US FTA, Art. 7.1(a); Australia–Singapore FTA, Ch. 14, Art. 3; Jordan–Singapore FTA, Art. 5.1(a); Australia–Thailand FTA, Art. 1102; New Zealand–Thailand FTA, Art. 10.2; New Zealand– Taiwan FTA, Ch. 9, Art. 4. 30 Japan–Switzerland FTA, Art. 76. Similar provisions can be found in the 2013 Central America– EFTA, Annex II, Art. 2; the 2015 Australia–China FTA, Art. 12.3; the 2018 Singapore–Sri Lanka FTA, Art. 9.3 and the 2015 Korea–Vietnam FTA, Art. 10.2.1. 31 Chile–US FTA, Art. 15.3. 32 Australia–Chile FTA, Art. 16.4; Australia–Korea FTA, Art. 15.3; EU–Singapore FTA, Art. 8.58; EU–Vietnam FTA, Art. 8.51; EU–Japan EPA, Art. 8.72. 33 Mexico–Panama FTA, Art. 14.4. 34 EU–Moldova AA, Art. 254.3; EU–CARIFORUM ECA, Art. 119.3; Colombia–EU–Peru FTA, Art. 162.3; Colombia–Israel FTA, Annex B, Art. 1.3; EU–Georgia FTA, Art. 127.3; EU–Ukraine FTA, Art. 139.3; EU–Armenia CEPA, Art. 193.3; EU–Korea FTA, Art. 7.48.3; Central America–EU FTA, Art. 201.3.

36  Research handbook on digital trade and not to internal taxes, fees or additional charges, provided that they are imposed in a manner consistent with the agreement.35 3.2

Electronic Authentication

An important number of trade agreements (90 PTAs) include provisions on electronic authentication, which typically allow technologies of mutual recognition of digital certificates and signatures. These provisions have evolved from soft commitments to more binding and mandatory clauses in recent agreements. The first treaty with such clauses, the Jordan–US FTA (2000), had only best effort commitments to the recognition and enforcement of electronic transactions and electronic authentication methods, including electronic signatures.36 Later agreements included cooperation commitments on electronic authentication, such as sharing information and experience on laws, regulations and programs on electronic signatures37 or secure electronic authentication.38 Other cooperation actions include research and training activities on electronic signatures and authentication,39 and in general activities to encourage their use,40 as well as seminars and expert meetings on security and interoperability.41 Most existing agreements follow this soft approach (41 PTAs in TAPED). More binding commitments on authentication and digital certificates have appeared since the Australia–US FTA (2004), establishing restrictions on legislation about electronic authentication using the models of both negative and positive obligations. Usually, under the negative obligation model, countries may not adopt or maintain laws that prevent or prohibit parties to an electronic transaction: (i) from mutually determining appropriate authentication methods;42 or (ii) from having the opportunity to prove in court that their electronic transaction complies with any legal requirements with respect to authentication.43 Further commitments limit the country’s ability to adopt or maintain laws preventing parties from choosing the tribunal to which they bring a dispute concerning an electronic transaction,44 or denying the legal validity of a signature solely on the basis that the signature is in electronic form.45 Under the positive

See e.g. New Zealand–UK FTA, Art. 15.4.2; India–United Arab Emirates CEPA, Art. 9.15.2. Jordan–US FTA, Joint Statement on Electronic Commerce. 37 See, e.g., Australia–Korea FTA, Art. 13.10.2; Central America–Mexico FTA, Art. 15.5(b); Hong Kong–New Zealand FTA, Ch. 10, Art. 2.1(f)(ii); Australia–Chile FTA, Art. 16.10.1; Colombia–Northern Triangle FTA, Art. 14.8(b). 38 See, e.g., Australia–Indonesia FTA, Art. 13.3.1(b)(v); USMCA, Art. 19.14(a)(iii); Australia–Peru FTA, Art. 13.14(b)(v); Singapore–Sri Lanka FTA, Art. 9.12(c)(iii); Argentina–Chile FTA, Art. 11.9(b); TPP/CPTPP, Art. 14.15(b)(v). 39 Korea–Vietnam FTA, Art. 10.8.1(a). 40 Singapore–Taiwan FTA, Art. 11.7(b); New Zealand–Taiwan FTA, Ch. 9, Art. 2.1(c)(ii)(iii). 41 EC–Singapore FTA, Art. 8.60.2(b). 42 See e.g. Singapore–Sri Lanka FTA, Art. 9.6.2; Singapore–Turkey FTA, Art. 9.6.2; Japan– Mongolia FTA, Art. 9.5.1; Australia–Japan FTA, Art. 13.6.1; Singapore–Taiwan FTA, Art. 11.5.1(a)(b); Japan–Switzerland FTA, Art. 78.1(a)(b). 43 Mexico–Panama FTA, Art. 14.9.1; PAAP, Art. 13.10.1; Colombia–Northern Triangle FTA, Art. 14.7. 44 Japan–Switzerland FTA, Art. 78.1(c). 45 See e.g. Singapore–Turkey FTA, Art. 9.6.3; Brazil–Chile FTA, Art. 10.6.2; USMCA, Art. 19.6.2; EU–Japan EPA, Art. 8.77.2; Australia–Peru FTA, Art. 13.6.2; Argentina–Chile FTA, Art. 11.3.2; 35 36

Three generations of digital trade provisions  37 obligation model, countries should apply domestic legislation for electronic authentication allowing parties to electronic transactions to: (i) determine the appropriate technologies and implementation models, without limiting their recognition; and (ii) prove in court that their electronic transactions comply with any legal requirements.46 However, many agreements include exceptions to these commitments − for instance, through a requirement of performance standards for sensitive transactions or communications.47 One example where such a higher degree of reliability and security is required is that of electronic financial transactions.48 Other exceptions require certification by an authority, or a supplier of such services accredited under a country’s regulations.49 Parties may endeavour to facilitate the accreditation or recognition of suppliers of certification services.50 Concerning public authentication, some agreements consider recognition of digital certificates or signatures at the governmental level,51 based on internationally accepted standards,52 cooperation mechanisms between national authorities for the accreditation of electronic transactions53 or mutual recognition agreements.54 On private authentication, some PTAs encourage the use of interoperable electronic authentication,55 electronic trust services56 or the mutual recognition of digital certificates and electronic signatures,57 which may include a trusted third-party service,58 and advanced or qualified certificates.59 Related to electronic authentication, more recent agreements include provisions on electronic invoicing, meaning the automated creation, exchange and processing of a request for payments between suppliers and buyers using a structured digital format.60

Australia–Singapore FTA (2016), Ch. 14, Art. 7.2; Chile–Uruguay FTA, Art. 8.5.2; TPP/CPTPP, Art. 14.6.2. 46 See, e.g. Australia–Indonesia FTA, Art. 13.5.2; Australia–China FTA, Art. 12.6.1; China–Korea FTA, Art. 13.4.2; Australia–Korea FTA, Art. 15.5.1; Chile–Thailand FTA, Art. 11.7(e). 47 Singapore–Turkey FTA, Art. 9.6.3. 48 Australia–Japan FTA, Art. 13.6.2; Australia–Korea FTA, Art. 15.5.2. 49 See e.g. Australia–Indonesia FTA, Art. 13.5.3; Brazil–Chile FTA, Art. 10.6.3; USMCA, Art. 19.6.3; EU–Japan EPA, Art. 8.77.3; Australia–Peru FTA, Art. 13.6.3; Singapore–Sri Lanka FTA, Art. 9.6.3; Argentina–Chile FTA, Art. 11.3.3. 50 PAAP, Art. 13.10.2; Central America–Mexico FTA, Art. 15.5(c); Japan–Switzerland FTA, Art. 78.4. 51 Korea–Peru FTA, Art. 14.8.2; Australia–US FTA, Art. 16.5.2. 52 See e.g., Korea–Vietnam FTA, Art. 10.3.2; Mexico–Panama FTA, Art. 14.9.2; PAAP, Art. 13.10.2; Chile–Thailand FTA, Art. 11.7(e); Australia–Malaysia FTA, Art. 15.6.2; Australia–Chile FTA, Art. 16.6.2. 53 Korea–Peru FTA, Art. 14.8.3. 54 EC–Singapore FTA, Art. 8.60.1; Argentina–Chile FTA, Art. 11.3.5. 55 See e.g., Australia–Indonesia FTA, Art. 13.5.4; Brazil–Chile FTA, Art. 10.6.4; USMCA, Art. 19.6.4; Australia–Peru FTA, Art. 13.6.4; Singapore–Sri Lanka FTA, Art. 9.6.4; Argentina–Chile FTA, Art. 11.3.4. 56 EU–Mexico Modernised Global Agreement, Digital Trade Chapter, Art. 6.4. 57 See, e.g., Singapore–Turkey FTA, Art. 9.6.4; Australia–China FTA, Art. 12.6.2–3; China–Korea FTA, Art. 13.4.3–4; Korea–Vietnam FTA, Art. 10.3.3; Australia–Korea FTA, Art. 15.5.3–4; Chile– Thailand FTA, Art. 11.7(e). 58 EAEU–Vietnam FTA, Art. 13.3. 59 Mexico–Panama FTA, Art. 14.9.2; PAAP, Art. 13.10.2. 60 E.g. ASEAN Agreement on Electronic Commerce, Art. 6(f); India–United Arab Emirates CEPA, Art. 9.16; UK–Singapore DEA, Art. 8.61–A; New Zealand–UK FTA, Art. 15.9.

38  Research handbook on digital trade 3.3

Paperless Trading

Starting with the New Zealand–Singapore Closer Economic Partnership Agreement (CEPA) (2000), a significant number of treaties (71 PTAs) have provisions on paperless trading. These clauses generally contribute to trade promotion between countries, enhancing their efficiency by reducing cost and time.61 Paperless trading could be applicable between States, between a State and a private entity, or between private entities. These provisions are found in different parts of the treaty, usually in the trade in goods chapter, but also as part of general cooperation activities or in an e-commerce chapter. Some promote paperless trading between the Parties as one of the overall objectives of the agreement.62 However, most of these provisions follow a soft approach and are not necessarily binding. Typically under provisions found in the trade in goods chapter, customs administrations shall consider the methodologies agreed by international organisations, such as the World Customs Organization (WCO)63 or the Asia-Pacific Economic Cooperation (APEC) forum.64 Few agreements also consider the development of a single governmental customs window following relevant international standards, but recognise that each Party shall have its own requirements and conditions.65 Some PTAs add commitments to exchange views and information between customs administrations and their trading communities,66 to employ information technology to the greatest extent possible, and to provide electronic means for customs reporting requirements as soon as is practicable.67 A second group of PTAs includes paperless trade provisions as cooperation commitments. While some merely mention it as one of many collaboration activities,68 others have cooperation at the international level to enhance the acceptance or use of paperless trading,69 or take into account the methods agreed upon by international organisations.70

61 UNECE and World Economic Forum, ‘Paperless Trading: How Does It Impact the Trade System’, White Paper, 2017, at 4. 62 Japan–Thailand FTA, Art, 1(b), Ch. 5, Arts 57–61; Japan–Singapore FTA, Art. 1(a)(iii); Ch. 5, Arts 40–44. 63 See e.g., Australia–Indonesia CEPA, Art. 5.13.1; Australia–Singapore FTA (2016), Ch. 4, Art. 4.1; Australia–China, Art. 4.6.1; China–Korea FTA, Art. 4.12; Australia–Korea, Art. 4.4.1. 64 New Zealand–Singapore CEPA, Art. 12; ASEAN–Australia–New Zealand FTA, Ch. 10, Art. 8.3; GCC–Singapore FTA, Art. 4.5.3; Peru–Singapore FTA, Art. 5.3; Trans Pacific Strategic EPA (P4), Art. 5.10. 65 Chile–China FTA (2018), Art. 56.3; Australia–Chile FTA, Art. 16.9.3. 66 Singapore–Taiwan FTA, Art. 5.4; GCC–Singapore FTA, Art. 4.5.2; Panama–Singapore FTA, Art. 4.5. 67 Australia–Singapore FTA (2016), Ch. 4, Art. 4.2–3; Chile–Thailand FTA, Art. 5.12.3–4; Australia–Chile FTA, Art. 5.11; Australia–Thailand, Art. 309. 68 Colombia–Israel FTA, Art. 4.6 and Annex B, Art. 2.1(f); Colombia–EU–Peru FTA, Art. 163.1(f); EC–Korea FTA, Art. 7.49.1(e). 69 Australia–Indonesia CEPA, Art. 13.4.3; Australia–Singapore FTA (2016), Ch. 14, Art. 10.3; Australia–China FTA, Art. 12.9.2; Japan–Mongolia FTA, Art. 9.8.3; Australia–Japan FTA, Art. 13.9.4; Australia–Malaysia FTA, Art. 15.9.2; ASEAN–Australia–New Zealand FTA, Ch. 10, Art. 8.2; Japan– Switzerland FTA, Art. 79.3; Australia–Singapore FTA (2003), Ch. 14, Art. 8.3. 70 Australia–China FTA, Art. 12.9.3; Australia–Japan FTA, Art. 13.9.3; Australia–Korea FTA, Art. 15.7; Australia–Malaysia FTA, Art. 15.9.3; Australia–Thailand, Art. 1107.2.

Three generations of digital trade provisions  39 A third type of commitment about paperless trade, usually found in e-commerce chapters, accepts electronic trade administration documents as the legal equivalent of their paper version. Although such provisions are often formulated as best efforts,71 sometimes they can be mandatory.72 A group of agreements also considers making available trade administration documents to the public in electronic form.73 However, some PTAs include exceptions to these commitments − for example, if there is a domestic or international legal requirement to the contrary, or if doing so would reduce the effectiveness of the trade administration process.74 3.4

Consumer Protection

An important number of agreements include clauses on consumer protection or ‘consumer confidence’, explicitly applicable to e-commerce or digital trade (99 PTAs), starting with the Canada–Costa Rica FTA (2001). However, these provisions are primarily non-binding. Several agreements merely consider international cooperation on consumer protection, such as sharing information and experiences on regulations, laws and programs;75 on means 71 Brazil–Chile FTA, Art. 10.9(b); USMCA, Art. 19.9; Central America–Korea FTA, Art. 14.6.2; Australia–Peru FTA, Art. 13.9(b); Singapore–Sri Lanka FTA, Art. 9.8(b); Chile–Uruguay FTA, Art. 8.8(b); TPP/CPTPP, Art. 14.9(b); Singapore–Turkey FTA, Art. 9.8(b); China–Korea FTA, Art. 13.6.2; Korea–Vietnam FTA, Art. 10.7.2; Japan–Mongolia FTA, Art. 9.8.2; Canada–Korea, Art. 13.5.2; Australia–Japan FTA, Art. 13.9.2; Australia–Korea FTA, Art. 15.7; Mexico–Panama FTA, Art. 14.7.2; PAAP, Art. 13.7.2; Singapore–Taiwan FTA, Art. 11.6.2; Colombia–Israel FTA, Annex B, Art. 4.2; Colombia–Panama, Art. 19.5.2; Colombia–Costa Rica FTA, Art. 16.5.2; Colombia–Korea FTA, Art. 12.4.2; Colombia–EU–Peru FTA, Art. 165(b); Korea–Peru FTA, Art. 14.6.2; Japan–Switzerland FTA, Art. 79.2; Canada–Colombia FTA, Art. 413.2 and Art. 1505.2; Canada–Peru FTA, Art. 413.3 and Art. 1506.2; Korea–US FTA, Art. 15.6.2; Chile–Colombia FTA, Art. 5.5.2; Colombia–US TPA, Art. 15.7.2; Peru–US TPA, Art. 15.7.2; Chile–China FTA (2018), Art. 56.1; Australia–US FTA, Art. 16.7; New Zealand–Thailand FTA, Art. 10.6.2. 72 Australia–Indonesia CEPA, Art. 13.4.2; Australia–Singapore FTA (2016), Ch. 14, Art. 10.2; Australia–China FTA, Art. 12.9.1; Chile–Thailand FTA, Art. 11.7.1(f); Australia–Malaysia FTA, Art. 15.9.1; Australia–Singapore FTA (2003), Ch. 14, Art. 8.1. 73 Australia–Indonesia CEPA, Art. 13.4.1; Brazil–Chile FTA, Art. 10.9(a); Central America–Korea FTA, Art. 14.6.1; Australia–Peru FTA, Art. 13.9(a); Singapore–Sri Lanka FTA, Art. 9.8(a); Australia– Singapore FTA (2016), Ch. 14, Art. 10.1; Chile–Uruguay FTA, Art. 8.8(a); TPP/CPTPP, Art. 14.9(a); Singapore–Turkey FTA, Art. 9.8(a); Australia–China FTA, Art. 12.9.4; China–Korea FTA, Art. 13.6.1; Korea–Vietnam FTA, Art. 10.7.1; Japan–Mongolia FTA, Art. 9.8.1; Canada–Korea, Art. 13.5.1; Australia–Japan FTA, Art. 13.9.1; Australia–Korea FTA, Art. 15.7; Mexico–Panama FTA, Art. 14.7.1; PAAP, Art. 13.7.1; Singapore–Taiwan FTA, Art. 11.6.1; Colombia–Israel FTA, Annex B, Art. 4.1; Colombia–Panama, Art. 19.5.1; Colombia–Costa Rica FTA, Art. 16.5.1; Colombia–Korea FTA, Art. 12.4.1; Colombia–EU–Peru FTA, Art. 165(a); Australia–Malaysia FTA, Art. 15.9.4; Korea–Peru FTA, Art. 14.6.1; Japan–Switzerland FTA, Art. 79.1; Canada–Colombia FTA, Art. 413.1 and Art. 1505.1; Australia–Chile FTA, Art. 16.9.1; Canada–Peru FTA, Art. 413.1, Art. 1506.1; Korea–US FTA, Art. 15.6.1; Chile–Colombia FTA, Art. 5.5.1; Colombia–US TPA, Art. 15.7.1; Peru–US TPA, Art. 15.7.1; Australia–Singapore FTA (2003), Ch. 14, Art. 8.1; Australia–US FTA, Art. 16.7. 74 Australia–Indonesia CEPA, Art. 13.4.2; Australia–Singapore FTA (2016), Ch. 14, Art. 10.2; Australia–China FTA, Art. 12.9.1; Chile–China FTA (2018), Art. 56.1; Australia–Malaysia FTA, Art. 15.9.1; Australia–Chile FTA, Art. 16.9.1; Australia–Thailand, Art. 1107.1; Australia–Singapore FTA (2003), Ch. 14, Art. 8.2. The 2015 EAEU–Vietnam FTA contains similar obligations but in negative terms (Art. 13.4). 75 Brazil–Chile FTA, Art. 10.15(b); Argentina–Chile FTA, Art. 11.9(b); Chile–Uruguay FTA, Art. 8.13(b); Mexico–Panama FTA, Art. 14.11(b); PAAP, Art. 13.12(b); Colombia–Korea FTA, Art. 12.6.1(f); CAFTA–DR–US, Art. 14.5(b); Chile–US FTA, Art. 15.5(b).

40  Research handbook on digital trade for consumer redress;76 or in confidence in e-commerce.77 Other activities include exchanging best practices, information or views on online protection,78 or access to products and services offered online;79 and maintaining dialogue/consultations80 about the protection in the ambit of electronic commerce,81 especially from fraudulent and misleading commercial practices in the cross-border context.82 Certain treaties also recognise the importance of cooperation between the respective national consumer protection agencies on activities related to cross-border electronic commerce,83 or exchanging information and experiences to enhance consumer protection.84 A second group of treaties recognises that consumers who participate in electronic commerce should be afforded transparent and effective protection that is not less than the level of protection afforded in other forms of commerce.85 In the Pacific Alliance Additional Protocol (2014), the Parties agree to several additional commitments. These include cooperation agreements for cross-border protection of consumer rights; exchanging information about suppliers sanctioned for infringement of those rights; prevention measures and training initiatives on the protection of consumer rights in e-commerce; standardising information that is provided to consumers in this environment; and encouraging e-commerce suppliers to comply with consumer protection regulations in the territory in which the consumer is located.86 Another group of PTAs also deal with consumer protection concerning the adoption of domestic standards, but mainly in a non-binding fashion, ‘recognising the importance of transparent and effective measures to protect consumers from fraudulent and deceptive commercial practices when engaging in e-commerce’.87 Only a handful of agreements have commitments to adopt or maintain consumer protection laws to prescribe these practices when they cause

Australia–Peru FTA, Art. 13.14(b)(ii); TPP/CPTPP, Art. 14.15(b)(ii); Chile–Thailand FTA, Art. 11.7.1(b). 77 Mexico–Panama FTA, Art. 14.3.2(f); Canada–Honduras FTA, Art. 16.5(b); Central America– Mexico FTA, Art. 15.5(b); Canada–Colombia FTA, Art. 1507.1(b); Canada–Peru FTA, Art. 1508(b); Colombia–Northern Triangle FTA, Art. 14.8(b); Panama–US FTA, Art. 14.5(b); Chile–Colombia FTA, Art. 12.5(b); Nicaragua–Taiwan FTA, Art. 14.05(b); Panama–Singapore FTA, Art. 13.4(b). 78 Chile–China FTA (2018), Art. 57.3(b). 79 Brazil–Chile FTA, Art. 10.15(c); Argentina–Chile FTA, Art. 11.9(c); Chile–Uruguay FTA, Art. 8.13(c); TPP/CPTPP, Art. 14.15(c). 80 EU–Mexico Modernised Global Agreement, Digital Trade Chapter, Art. 11.1(d); EC–Central America FTA, Art. 202(c); CARIFORUM–EC EPA, Art. 120.1(d); Australia–Chile FTA, Art. 16.10.1. 81 Central America–Korea FTA, Art. 14.7.1(d); Colombia–Panama FTA, Art. 19.7.1(f). 82 Colombia–Israel FTA, Annex–B, Art. 2(d). 83 See e.g. USMCA, Art. 19.7.3; EU–Japan FTA, Art. 8.78.2; EU–Mexico Modernised Global Agreement, Digital Trade Chapter, Art. 7.3; Central America–Korea FTA, Art. 14.4.2; Australia–Peru FTA, Art. 13.7.3; Argentina–Chile FTA, Art. 11.4.4; Chile–Uruguay FTA, Art. 8.6.4; TPP/CPTPP, Art, 14.7.3; Colombia–Israel FTA, Annex–B, Art. 5.2. 84 Brazil–Chile FTA, Art. 10.7.4; PAAP, Art. 13.6.2; Canada–Honduras FTA, Art. 16.4.2; Korea– Peru FTA, Art. 14.4.2; Canada–Colombia FTA, Art. 1504.2; Canada–Peru FTA, Art. 1505.2. 85 Canada–Costa Rica FTA, Joint Statement on Global Electronic Commerce. 86 PAAP, Art. 13.6.4–5. 87 See e.g., Brazil–Chile FTA, Art. 10.7.1; USMCA, Art. 19.7.1; EU–Mexico Modernised Global Agreement, Digital Trade Chapter, Art. 7.1; Central America–Korea FTA, Art. 14.4.1; Australia–Peru FTA, Art. 13.7.1; Argentina–Chile FTA, Art. 11.4.1; Chile–Uruguay FTA, Art. 8.6.1; TPP/CPTPP, Art, 14.7.1; Mexico–Panama FTA, Art. 14.6.1; PAAP, Art. 13.6.1. 76

Three generations of digital trade provisions  41 harm or potential harm to consumers.88 Moreover, few agreements consider that the Parties may evaluate the use of alternative dispute resolution mechanisms,89 or even online dispute settlement for the protection of the consumer, if feasible.90 Finally, a fourth group of treaties use more binding options for consumer protection. Some establish a criterion of ‘equivalence’ providing protection to consumers using e-commerce at least equivalent to that provided for consumers and other forms of commerce, under their respective domestic laws, regulations and policies.91 Others consider specific businesses’ obligations to protect consumers in e-commerce, including acting in accordance with fair business, advertising and marketing practices, such as providing accurate, clear and easily accessible information about goods or services offered; avoiding ambiguity on intent to make a purchase; and providing an easy-to-use, secure payment mechanism and information on the level of security such mechanisms afford.92 3.5

Unsolicited Commercial Electronic Messages

At least 63 agreements include measures on unsolicited commercial electronic messages (commonly known as ‘spam’), starting with the Caribbean Forum (CARIFORUM)–European Community EPA (2008). Around half of these provisions are non-binding. A significant part of the agreements with such provisions considers the treatment of spam as part of cooperation activities,93 such as information sharing on technical, educational and policy approaches,94 or within an overall dialogue on regulatory issues raised by electronic commerce in both bilateral and international fora.95 Some agreements subject cooperation activities to each country’s respective laws and regulations.96 A second group of treaties include ‘best efforts’ commitments to adopt or maintain measures to regulate unsolicited commercial electronic messages to minimise unsolicited spam or telemarketing.97 A third type of PTA has binding obligations, under which each Party shall adopt or maintain measures to minimise unsolicited commercial electronic messages.98 Following the TPP model (2016), some agreements consider more detailed obligations such as: (a) requiring suppliers of unsolicited commercial electronic messages to facilitate the

88 Brazil–Chile FTA, Art. 10.7.2; USMCA, Art. 19.7.2; EU–Mexico Modernised Global Agreement, Digital Trade Chapter, Art. 7.2; Australia–Peru FTA, Art. 13.7.2; Argentina–Chile FTA, Art. 11.4.2; Chile–Uruguay FTA, Art. 8.6.2; TPP/CPTPP, Art. 14.7.2. 89 Mexico–Panama FTA, Art. 14.6.2; PAAP, Art. 13.6.3; Colombia–Costa Rica FTA, Art. 16.4.3. 90 Brazil–Chile FTA, Art. 10.7.4. 91 Chile–Thailand FTA, Art. 11.7(k); Chile–China FTA (2018), Art. 54; Australia–Chile FTA, Art. 16.7.1–2(a). 92 Australia–Chile FTA, Art. 16.7.2(b) and Art. 16.7.3. 93 Japan–Switzerland FTA, Art. 82.2(b); Colombia–Panama FTA, Art. 19.7(d). 94 ASEAN–Australia–New Zealand FTA, Ch. 10, Art. 9.1(c); Chile–Thailand FTA, Art. 11.7.1(b) (iii); Australia–China FTA, Art. 12.10.3(b). 95 CARIFORUM – EC EPA, Art. 120.1(c); EU–Korea FTA, Art. 7.49.1(c); Colombia–Ecuador– EU–Peru FTA, Art. 163.1(c); CETA, Art. 16.6.1(c). 96 Australia–Malaysia FTA, Art. 15.10.2; Australia–Korea FTA, Art. 15.9.2. 97 Australia–Korea FTA, Art. 15.9.1; Japan–Mongolia FTA, Art. 97; India–Saudi Arabia CEPA, Art. 9.9. 98 Australia–Malaysia FTA, Art. 15.10.1; PAAP, Art. 13.9; Australia–Japan FTA, Art. 10.20; Chile– Uruguay FTA, Art. 8.12; Argentina–Chile FTA, Art. 11.8.

42  Research handbook on digital trade ability of recipients to prevent ongoing reception of those messages; (b) requiring the consent of recipients to receive commercial electronic messages, as specified according to the laws and regulations of each country; and (c) providing recourse against suppliers of unsolicited commercial electronic messages that do not comply with such measures.99

4.

SECOND GENERATION: DATA FLOWS RULES

Building on the first generation, a second generation of PTAs with digital trade provisions focuses particularly on data flows regulation. These rules either refer to the cross-border flow of data, its relationship with privacy and data protection, or banning or limiting data localisation requirements.100 The Korea–US FTA (2007) is the first PTA to include such provisions in general terms.101 See Figure 3.4.

Source: Based on data from the TAPED database, www.unilu.ch/en/faculties/faculty-of-law/professorships/ burri-mira/research/taped/

Figure 3.4

Data flows provisions in PTAs

See e.g., TPP/CPTPP, Art. 14.14; Australia–Peru FTA, Art. 13.13; EU–Japan EPA, Art. 8.79; Brazil–Chile FTA, Art. 10.14; USMCA, Art. 19.13; Australia–Indonesia CEPA, Art. 13.8; Chile– Ecuador ECA, Art. 10.13; RCEP, Art. 12.9; New Zealand – UK FTA, Art. 15.11. 100 Burri and Polanco (n 7). 101 Korea–US FTA, Art. 15.8. 99

Three generations of digital trade provisions  43 However, provisions on the cross-border flow of data can also be found in chapters dealing with specific services sectors where data flows are crucial or inherent to the very definition of those services, especially in telecommunications and financial services. Some agreements also include these provisions in computer and related services and audio-visual sectors.102 These sectoral provisions appeared before the general clauses on data flows. The New Zealand–Singapore CEPA (2000) already included a provision considering the transfer of financial information, as well as data processing, as part of the financial services commitments.103 Since then, banking and other financial services have commonly included the provision and transfer of financial information, financial data processing and related software by suppliers of other financial services.104 Some recent agreements even delve into more detail on the use of technology to improve and automate the delivery and use of financial services (‘FinTech’), promoting the use of information technology to improve and manage compliance with regulatory processes (‘RegTech’).105 Likewise, telecommunication services in PTAs have long been defined as including, inter alia, data transmission typically involving the real-time transmission of customer-supplied information between two or more points without any end-to-end change in the form or content of the customer’s information, or simply including the transfer of data by electronic means.106 4.1

Rules on Data Flows

We find general provisions on data flows in at least 39 PTAs. But looking at the evolution of such clauses in PTAs, there has been a substantial change over the years. Non-binding provisions on data flows appeared quite early. The Joint Statement on Electronic Commerce of the Jordan–US FTA (2000) highlighted the ‘need to continue the free flow of information’, but fell short of including an explicit provision in this regard. The first PTA having such a provision is the Taiwan–Nicaragua FTA (2006), but it is only part of the cooperation activities, affirming the importance of working to maintain cross-border flows of information as an essential element to promote a dynamic environment for e-commerce.107 Similar wording is used in subsequent agreements such as the Peruvian FTA with Canada (2008) and Korea (2011),108 the Central America–Mexico FTA (2011),109 the Colombia–Costa Rica FTA (2013),110 the Canadian FTAs with Honduras and Korea,111 and the Japan–Mongolia FTA (2015).112

Mira Burri, ‘Trade in Services Regulation in the Data-Driven Economy’ (2020) 12 Trade Law & Development 208. 103 New Zealand–Singapore CEPA, Annex 2.1, Schedule of Commitments of New Zealand. 104 Singapore–US FTA, Art. 10.20, Annex 10-A; Chile–EC AA, Art. 117.9; Japan–Singapore FTA, Annex IV-A; New Zealand–Singapore CEPA, Annex 2.1. 105 Australia–Singapore DEA, Ch. 14, Art. 1 and 32; Singapore–UK DEA, Art. 8.53. 106 Singapore–US FTA, Art. 9.16.18; Japan–Singapore FTA, Annex IV–B. 107 Nicaragua–Taiwan FTA, Art. 14.05(c). 108 Canada–Peru FTA, Art. 1508(c); Korea–Peru FTA, Art. 14.9(c). 109 Central America–Mexico FTA, Art. 15.5(d). 110 Colombia–Costa Rica FTA, Art. 16.7(c). 111 Canada–Honduras FTA, Art. 16.5(c); Canada–Korea FTA, Art. 13.7(c). 112 Japan–Mongolia FTA, Art. 9.12.5. 102

44  Research handbook on digital trade An intermediate type of provision between ‘hard’ and ‘soft’ commitments is found in the South Korea–US FTA (2007). Parties declare that they ‘shall endeavour to refrain from imposing or maintaining unnecessary barriers to electronic information flows across borders’,113 after recognising its importance in facilitating trade and acknowledging the importance of protecting personal information. The Mexico–Panama FTA (2014) is the first treaty having a binding provision on cross-border information flows. According to this agreement, each Party ‘shall allow its persons and the persons of the other Party to transmit electronic information, from and to its territory, when required by said person, in accordance with the applicable legislation on the protection of personal data and taking into consideration international practices’.114 A much more detailed provision in this regard is found in the 2015 amended version of the Pacific Alliance Additional Protocol (PAAP) (2016),115 the TPP (2016) and the CPTPP (2018). After recognising that each Party may have its own regulatory requirements concerning the transfer of information by electronic means, these treaties stipulate that each Party ‘shall allow the cross-border transfer of information by electronic means, including personal information’, when it is for the conduct of the business of a covered person. However, this shall not prevent a Party from ‘adopting or maintaining measures to achieve a legitimate public policy objective, provided that the measure is not applied in a manner which would constitute a means of arbitrary or unjustifiable discrimination or a disguised restriction on trade; and does not impose restrictions on transfers of information greater than are required to achieve the objective’. A similar hard rule on data flows is found in subsequent PTAs concluded between 2016 and 2019, mainly following a similar wording.116 In the USMCA (2018), a footnote clarifies that a measure on the restriction of data flows is not considered to achieve a legitimate public policy objective, if ‘it accords different treatment of data transfers solely on the basis that they are cross-border in a manner that modifies the conditions of competition to the detriment of the service suppliers of the other Party’.117 More recent agreements add another exclusion to commitments about the free flow of data: adopting or maintaining any measure that a Party considers necessary for the protection of its essential security interests.118 The Japan–UK FTA excludes from the same obligation government procurement or information held or processed by or on behalf of a Party, or measures by a Party related to that information, including measures related to its collection.119 One exception to this trend of binding provisions is in the recent India–Saudi Arabia CEPA (2022), where the Parties merely recognise the importance of the flow of information in facilitating trade, and acknowledge the importance of protecting personal data. As such, the Parties

Korea–US FTA, Art. 15.8 Mexico–Panama FTA, Art. 14.10. 115 PAAP (2015), Art. 13.11. 116 See e.g., Chile–Uruguay FTA, Art. 8.10; Australia–Singapore FTA, Ch. 14, Art. 13; Argentina– Chile FTA, Art. 11.6; Singapore–Sri Lanka FTA, Art. 9.9; Australia–Peru FTA, Art. 13.11; Brazil–Chile FTA, Art. 10.12; Japan–US DTA, Art. 11; Australia–Singapore DEA, Art. 23; DEPA, Art. 4.3; Chile– Ecuador ECA, Art. 10.11; Chile–Paraguay FTA, Art. 7.11; Australia–UK FTA, Art. 14.10; Singapore– UK DEA, Art. 8.61–F; New Zealand–UK FTA, Art. 15.14. 117 USMCA, Art. 19.11, fn. 6. 118 Australia–Indonesia CEPA, Art. 13.11; RCEP, Art. 12.15. 119 Japan–UK CEPA, Art. 8.84. 113 114

Three generations of digital trade provisions  45 shall endeavour to promote electronic information flows across borders subject to their laws and regulatory frameworks.120 4.2

Data Localisation

In recent years, some PTAs have started to include provisions on data localisation, either banning or limiting requirements on the location or use of data. Today, at least 23 PTA include such provisions, and almost all of these clauses are binding. The first agreement including such provisions is the Japan–Mongolia FTA (2015). According to it, neither Party ‘shall require a service supplier of the other Party, an investor of the other Party, or an investment of an investor of the other Party in the area of the former Party, to use or locate computing facilities in that area as a condition for conducting its business’.121 The same year, the amended version of the PAAP included an analogous provision on the use and location of computer facilities. Furthermore, it added that nothing shall prevent a Party from adopting or maintaining measures to achieve a legitimate public policy objective, provided that they are not applied in a manner that constitutes a means of arbitrary or unjustifiable discrimination, or a disguised restriction to trade.122 The TPP (2016) also included a similar provision. After recognising that each Party may have its own regulatory requirements regarding the use of computing facilities (such as those to ensure the security and confidentiality of communications), the treaty stipulates that a covered person shall not be required to use or locate computing facilities in a Party’s territory as a condition for conducting business in that country. Additionally, the TPP requires that such measure does not impose restrictions on the use or location of computing facilities greater than are required to achieve the objective.123 Similar binding commitments are found in subsequent agreements concluded since 2016, which largely follow the same wording.124 A variation is found in at least four agreements concluded in 2018, where there is no least restrictive measure requirement, thus coming closer to the PAAP wording than the TPP/CPTPP template.125 The USMCA (also from 2018) includes another deviation stipulating that ‘no Party shall require a covered person to use or locate computing facilities in that Party’s territory as a condition for conducting business in that territory’.126 Furthermore, the Australia–Indonesia FTA (2019) stipulates that nothing in the agreement shall prevent a Party from adopting or maintaining any measure that it considers necessary for the protection of its ‘essential security interests’.127 The Australia–Singapore

India–Saudi Arabia CEPA, Art. 9.11. Japan–Mongolia FTA, Art. 9.10. A similar provision is also included in the Japan–US DTA, Art.

120 121

12.

PAAP (2015), Art. 13.11bis. TPP/CPTPP Art. 14.13. 124 Chile–Uruguay FTA, Art. 8.11; Australia–Singapore FTA, Ch. 14, Art. 15; Australia–Singapore DEA, Art. 23; Chile–Ecuador, Art. 10.12; Chile–Paraguay FTA, Art. 7.12; Australia–UK FTA, Art. 14.11; Singapore–UK DEA, Art. 8.61–G; New Zealand–UK FTA, Art. 15.15. 125 Singapore–Sri Lanka FTA, Art. 9.10; Australia–Peru FTA Art. 13.12; Brazil–Chile FTA, Art. 10.13; DEPA, Art. 4.4. 126 USMCA, Art. 19.12. 127 Australia–Indonesia CEPA, Art. 13.12.3(b). 122 123

46  Research handbook on digital trade DEA (2020) excludes financial services from this limitation on measures on data localisation.128 RCEP added another exclusion: adopting or maintaining any measure that a Party considers necessary for the protection of its essential security interests.129 Most recent treaties add more detail to this limitation, establishing that cross-border data flows shall not be restricted between the Parties by a Party: (a) requiring the use of computing facilities or network elements in the Party’s territory for processing, including by imposing the use of computing facilities or network elements that are certified or approved in the territory of a Party; (b) requiring the localisation of data in the Party’s territory for storage or processing; (c) prohibiting the storage or processing in the territory of the other Party; or (d) making the cross-border transfer of data contingent upon use of computing facilities or network elements in the Parties’ territory or upon localisation requirements in the Parties’ territory.130 One of the few provisions on data localisation that are not directly binding is found in the Argentina–Chile FTA (2017). Under this treaty, the Parties just ‘recognise the importance’ of not requiring a person of the other Party to use or locate the computer facilities in the territory of that Party, as a condition for conducting business in that territory, and pledge to undertake to exchange good practices and current regulatory frameworks regarding the location of servers.131 4.3

Privacy and Data Protection

Today, at least 112 PTAs include provisions on privacy, usually under the concept of ‘data protection’. Yet, the way this data is protected varies considerably and can consist of a mixed bag of binding and non-binding provisions. This is symptomatic of the different positions adopted by prominent actors (the EU and the US) and the inherent tensions between the regulatory goals of data innovation and data protection.132 Earlier agreements included only non-binding declarations on privacy issues of mere programmatic nature, like the Joint Statement on Electronic Commerce in the Jordan–US FTA (2000) and the Joint Statement on Global Electronic Commerce of the Canada–Costa Rica FTA (2001). Later agreements included cooperation activities on privacy protection in electronic communications and avoiding obstacles to trade that requires personal data transfers.133 These include sharing information and experiences on regulations, laws and programs on data protection,134 or on the overall domestic regime for the protection of personal information;135 technical

Australia–Singapore DEA, Art. 24. RCEP, Art. 12.14. 130 EU–UK TCA, Art. 201. 131 Argentina–Chile FTA, Art. 11.7. 132 See e.g., Paul M Schwartz, ‘The EU–US Privacy Collision: A Turn to Institutions and Procedures’, Harvard Law Review 126 (2013) 1966; Paul M Schwartz and Daniel J Solove, ‘Reconciling Personal Information in the United States and European Union’, California Law Review 102 (2014) 877. 133 EC–Moldova AA, Art. 13.1 and Art. 99(d). 134 See e.g., Brazil–Chile FTA, Art. 10.8.5 and Art. 10.15(b); Central America–Korea FTA, Art. 14.5.2; Argentina–Chile FTA, Art. 11.5.5 and Art. 11.9(b); Chile–Uruguay FTA, Art. 8.7.4 and Art. 8.13(b); EAEU–Vietnam FTA, Art. 13.6.1. 135 Australia–Indonesia FTA, Art. 13.3.1(b)(i); USMCA, Art. 19.14.1(a)(i); Australia–Peru FTA, Art. 13.14(b)(i); Singapore–Sri Lanka FTA, Art. 9.12(c)(i); Singapore–Turkey FTA, Art. 9.9(c); China– Korea FTA, Art. 13.5; Colombia–Costa Rica FTA, Art. 16.6.2; Canada–Colombia FTA, Art. 1506.2. 128 129

Three generations of digital trade provisions  47 assistance in the form of exchange of information and experts,136 research and training activities;137 the establishment of joint programmes and projects;138 a ‘dialogue’139 or ‘consultations’ on matters of data protection;140 or, in general, other mechanisms to ensure the protection of personal data.141 In some treaties, Parties commit to publishing information on personal data protection,142 including how individuals can pursue remedies and how businesses can comply with any legal requirements in this regard.143 Many agreements (60 PTAs) have also dealt with personal data protection with reference to the adoption of domestic standards. In several treaties, Parties specifically commit to adopt or maintain legislation or regulations that protect the personal data or privacy of users.144 These measures are usually in relation to the processing and dissemination of data,145 which may also include administrative measures146 or the adoption of non-discriminatory practices.147 Some agreements consider qualifications to this commitment, which may include taking measures deemed ‘appropriate and necessary considering the differences in existing systems for personal data protection’,148 or that the Parties have ‘the right to define or regulate their own levels of protection of personal data in pursuit or furtherance of public policy objectives’, and shall not be required to disclose confidential or sensitive information or data.149 Several treaties (51 PTAs) add that in the development of online personal data protection standards, each Party shall take into account existing international standards,150 as well as

Chile–EC AA, Art. 30. Korea–Vietnam FTA, Art. 10.8.1(b). 138 Chile–EC AA, Art. 30. 139 Colombia–EU–Peru FTA, Art. 163.1(e). 140 Australia–Chile FTA, Art. 16.10.1. 141 Central America–Korea FTA, Art. 14.7.1(a); Colombia–Israel FTA, Annex–B, Art. 2(e); Colombia–Panama FTA, Art. 19.7.1(b); Colombia–Korea FTA, Art. 12.6.1(c). Some agreements aiming to develop such protection that ‘at a higher level’, but without giving any further indication of what this entails (Armenia–EU CEPA, Art. 13; EC–Ukraine AA, Art. 15; EC–Georgia AA, Art. 14). 142 Brazil–Chile FTA, Art. 10.8.4. 143 See, e.g., India–Saudi Arabia CEPA. Art. 9.10; USMCA, Art. 19.8.5; Australia–Peru FTA, Art. 13.8.4; Singapore–Sri Lanka FTA, Art. 9.7.3; Australia–Singapore FTA (2016), Ch. 14, Art. 9.4; Chile– Uruguay FTA, Art. 8.7.3; TPP/CPTPP, Art. 14.8.4. 144 See e.g., Australia–Indonesia FTA, Art. 13.7.2; Brazil–Chile FTA, Art. 10.8.2; USMCA, Art. 19.8.1–2; Australia–Peru FTA, Art. 13.8.1–2; Singapore–Sri Lanka FTA, Art. 9.7.1–2; Argentina–Chile FTA, Art. 11.5.1–2. 145 Central America–EFTA, Annex II, Art. 1(c)(i); EFTA–GCC FTA, Annex XVI, Art. 1(c)(i); EFTA–Colombia FTA, Annex I, Art. 1(c)(i); EFTA–Peru FTA, Annex I, Art. 1(c)(i). 146 Colombia–Costa Rica FTA, Art. 16.6.1; Korea–Peru FTA, Art. 14.7; Hong Kong–New Zealand FTA, Ch. 10, Art. 2.1(f); ASEAN–Australia–New Zealand FTA, Ch. 10, Art. 7.1–2; Australia–Chile FTA, Art. 16.8; Canada–Peru FTA, Art. 1507. 147 Australia–Indonesia FTA, Art. 13.6.3; Brazil–Chile FTA, Art. 10.8.3; USMCA, Art. 19.8.4; Australia–Peru FTA, Art. 13.8.3; Australia–Chile FTA, Art. 11.5.3; Australia–Singapore FTA (2016), Ch. 14, Art. 9.3; TPP/CPTPP, Art. 14.8.3. 148 Australia–China FTA, Art. 12.8.1; Chile–Thailand FTA, Art. 11.7.1(j); Australia–Singapore FTA (2003), Ch. 14, Art. 7.1. 149 EU–Japan EPA, Art. 18.1.2(h) and Art. 18.16.7. 150 See e.g. EC–Singapore FTA, Art. 8.57.4; Argentina–Chile FTA, Art. 11.5.1–2; Chile–Uruguay FTA, Art. 8.7.2; Armenia–EU CEPA, Art. 197.2; Colombia–EU–Peru FTA, Art. 162.2. 136 137

48  Research handbook on digital trade criteria or guidelines of relevant international organisations or bodies151 – such as the APEC Privacy Framework and the OECD Recommendation of the Council concerning Guidelines Governing the Protection of Privacy and Transborder Flows of Personal Data (2013).152 In recognition that the Parties may take different legal approaches to protect personal information, some agreements declare that each Party should encourage the development of mechanisms to promote compatibility between different regimes. These may include recognising regulatory outcomes, whether accorded autonomously or by mutual arrangement, or through broader international frameworks. To this end, the Parties shall endeavour to exchange information on any such mechanisms applied in their jurisdictions and explore ways to extend these or other suitable arrangements to promote compatibility between them.153 For example, the 2018 USMCA explicitly recognises that the APEC Cross-Border Privacy Rules system is a valid mechanism to facilitate cross-border information transfers while protecting personal information.154 Countries have also employed more binding tools to protect personal information online. A first option considers the protection of privacy in the processing and dissemination of personal data and the protection of confidentiality of individual records and accounts, as an exception in specific chapters of the agreement − such as for trade in services,155 investment or establishment,156 movement of persons,157 telecommunications158 and financial services.159 Certain agreements (mainly mostly EU-led), even have special chapters on the protection of personal data, including discrete principles (23 PTAs). Such principles include: purpose limitation, data quality and proportionality, transparency, security, right to access, rectification and opposition, restrictions on onward transfers, protection of sensitive data, provisions on enforcement mechanisms, coherence with international commitments and cooperation between the Parties to ensure an adequate level of protection of personal data.160 Other agreements recognise principles for collecting, processing and storing personal data, such as prior consent, legitimacy, purpose, proportionality, quality, safety, responsibility and information.161 The USMCA was the first US-led PTA to include such a provision. It also adds that any

151 See e.g., Israel–South Korea FTA, Art. 13.7; Australia–Indonesia FTA, Art. 13.7.3; Australia– Peru FTA, Art. 13.8.2; CETA, Art. 16.4; Australia–Singapore FTA (2016), Ch. 14, Art. 9.2; TPP/ CPTPP, Art. 14.8.2. 152 USMCA, Art. 19.8.2. 153 Australia–Indonesia FTA, Art. 13.7.4; Australia–Peru FTA, Art. 13.8.5; Australia–Singapore FTA (2016), Ch. 14, Art. 9.5; TPP/CPTPP, Art. 14.8.5. 154 USMCA, Art. 19.8.6. 155 Japan–Singapore FTA, Arts. 69.1(c). 156 Chile–EC AA, Art. 135.1(e)(ii); Japan–Singapore FTA, Arts. 83.1(c)(ii). 157 Japan–Singapore FTA, Arts. 95.1(c)(ii). 158 See, e.g., USMCA, Art. 18.3.4; EU–Japan EPA, Art. 8.44.4; Australia–Peru FTA, Art. 12.4.4; Singapore–Sri Lanka FTA, Art. 8.3.4; Argentina–Chile FTA, Art. 10.3.4; Australia–Singapore FTA (2016), Art. 10.3.4. 159 See e.g., USMCA, Annex 17–A; EU–Japan EPA, Art. 8.63; EC–Vietnam FTA, Art. 8.45; EC– Singapore FTA, Art. 8.54.2; Australia–Peru FTA, Art. 10.21; Armenia–EU CEPA, Art. 185. 160 Cameroon–EC Interim EPA, Ch. 6, Arts. 61–65; CARIFORUM–EC EPA, Ch. 6, Arts. 197–201. 161 Australia–UK FTA, Art. 14.12; Chile–Paraguay. FTA, Art. 7.8; Argentina–Chile FTA, Art. 11.2.5(f) fn 1; Chile–Uruguay FTA, Art. 8.2.5(f), fn 3.

Three generations of digital trade provisions  49 restrictions on cross-border flows of personal information are necessary and proportionate to the risks presented.162 A second option allows countries to adopt appropriate measures to ensure privacy protection while allowing the free movement of data, establishing a criterion of ‘equivalence’. It means that countries agree that personal data may be exchanged only where the receiving Party undertakes to protect such data in at least an equivalent, similar or adequate way to the one applicable to that particular case in the Party that supplies it. This has been largely the EU approach, and to that end, Parties commit to inform each other of their applicable rules and negotiate reciprocal, general or specific agreements.163 In the same line, under the Argentina– Chile FTA, Parties undertake to apply to data from the other Party a level of protection that is ‘at least similar to that applicable to the Party from which the data originates’, through mutual, general or specific agreements.164 A third but less used option leaves the development of data protection rules to a treaty body. The Colombia–Ecuador–EU–Peru FTA (2012) allows the Trade Committee to establish a working group with the task of proposing guidelines and strategies enabling the signatory Andean countries to become a safe harbour for the protection of personal data. To this end, the working group shall adopt a cooperation agenda ‘that defines priority aspects for accomplishing that purpose, especially regarding the respective homologation processes of data protection systems’.165

5.

THIRD GENERATION: DIGITAL TRADE AGREEMENTS AND NEW DATA-DRIVEN ECONOMY

In recent years we can discern a new wave of trade agreements advancing rules for the new data-driven economy and, in some cases, doing so in exclusive digital economy agreements (DEAs), no longer as part of FTAs or other types of ‘traditional’ PTAs. A digital trade agreement was concluded between Japan and the US on 7 October 2019, but it was signed in parallel with the FTA concluded by both countries on the same date. It was followed by the Australia–Singapore DEA signed on 23 March 2020, which amends the Australia–Singapore FTA, replacing its previous electronic commerce chapter with a new digital economy chapter, along with other modifications. The first standalone DEA without a direct connection with a ‘traditional’ PTA is the Digital Economy Partnership Agreement (DEPA), signed on 12 June 2020 by Chile, New Zealand and Singapore. Following DEPA, Singapore also concluded a DEA with the United Kingdom (UK), signed on 28 February 2022, forming an integral part of the FTA concluded between both parties on 10 December 2020. DEAs, as well as recent PTAs with digital trade provisions, have started to include clauses beyond e-commerce and the regulation of data flows and have ventured into topics

USMCA, Art. 19.8.3. The same principles are recognised in the New Zealand–UK FTA, Art. 15.13; UK–Singapore DEA, Art. 8.61–E. 163 See e.g., EU–UK TCA, Art. 202; EC–Singapore FTA, Art. 8.54.2; Understanding 3 Additional Customs–Related Provisions, Arts 9.2 and 11.1; EC–Ghana EPA, Protocol on Mutual Administrative Assistance on Custom Matters, Art. 10. 164 Argentina–Chile FTA, Art. 11.6.2 and 11.5.7. 165 Colombia–EU–Peru FTA, Art. 109(b). 162

50  Research handbook on digital trade such as e-government, competition policy, digital identities, digital inclusion and artificial intelligence, among others.166 Although such provisions are not numerous and are largely non-binding, they show a clear path for possible future negotiations of DEAs and PTAs with digital trade provisions. 5.1

Digital Identities

Some of these treaties include best efforts provisions on digital identities, mainly focusing on the development of mechanisms to promote compatibility between their respective digital identity regimes for the purposes of increasing regional and global connectivity, but also recognising that each country may take different legal and technical approaches to digital identities. These activities may include: (a) developing or maintaining appropriate frameworks or common standards to foster technical interoperability between each Party’s implementation of digital identities; and (b) supporting the development of international frameworks on digital identity regimes.167 Some agreements also add as possible activities implementing use cases for the mutual recognition of digital identities and exchanging knowledge and expertise on best practices relating to digital identity policies and regulations, technical implementation, security standards and the promotion of the use of digital identities.168 Finally, certain treaties go one step further, considering: developing comparable protection of digital identities under each Party’s respective legal frameworks, or the recognition of their legal effects, whether accorded autonomously or by mutual agreement.169 5.2

Competition and the Digital Economy

A few agreements have provisions about cooperation on competition law and policy to address the challenges that arise from the digital economy.170 Some include specific activities, such as exchanging information and experiences, sharing best practices on the enforcement of competition law and the promotion of competition in digital markets and providing advice or training, including through the exchange of officials, to assist a Party in building necessary capacities to strengthen competition policy development and competition law enforcement in digital markets.171 5.3

Digital Inclusion

Certain treaties recognise the importance of digital inclusion, understood as the possibility that all people and businesses can participate in, contribute to and benefit from the digital economy. This includes cooperation to remove barriers to participation for groups that disproportionately Mira Burri, Rodrigo Polanco and María Vásquez Callo–Müller, ‘TAPED’ (University of Lucerne, January 2022) www​.unilu​.ch/​en/​faculties/​faculty–of–law/​professorships/​burri–mira/​research/​taped/​ accessed 10 June 2022. 167 India–United Arab Emirates CEPA, Arts 9.7 and 9.20(i). 168 Australia–UK, Art. 14.7; New Zealand–UK FTA, Art. 15.8. 169 Australia–Singapore DEA, Art. 29; DEPA, Art. 7.1; Singapore–UK DEA, Arts. 8.61–S and 8.61–W.2(a)(viii). 170 Australia–Singapore FTA, Art. 14.16.3. 171 Australia–Singapore DEA, Art. 16; DEPA, Art. 8.4; Singapore–UK DEA, Art. 8.61–U. 166

Three generations of digital trade provisions  51 face such barriers (like Indigenous peoples, women and rural and low-socioeconomic groups) through the coordination, as appropriate, of the Parties’ respective agencies and stakeholders. Such activities include: (a) sharing experiences and best practices, including the exchange of experts, with respect to digital inclusion; (b) identifying and addressing barriers in accessing digital trade opportunities; (c) sharing methods and procedures for developing datasets and conducting analysis in relation to the participation in digital trade by groups that may disproportionately face barriers to participation in the digital economy; (d) improving digital skills and access to online business tools. 172 The Singapore–UK DEA also adds a labour angle, promoting protection and decent conditions for workers engaged in or supporting digital trade.173 5.4

Digital Divide

Other agreements focus on adopting government processes, services and policies with digital inclusivity in mind, aiming to overcome digital divides.174 On the latter, some treaties address the digital divide between countries, and the role of digital trade in promoting economic development and poverty reduction. Such a role considers the use of mechanisms to foster the participation in digital trade of countries that face barriers to such participation. These include sharing best practices, collaborating on capacity-building initiatives, active engagement in international fora (including the WTO) and promoting countries’ participation in, and contribution to, the global development of rules on digital trade.175 5.5

Artificial Intelligence

Recent agreements also include provisions about artificial intelligence (AI). Some treaties consider AI-related cooperation activities in general terms,176 or maintaining a dialogue on AI with a view to sharing information and experience, as appropriate, including on related laws, regulations and their implementation, and best practices.177 Others include more detailed cooperation activities such as: (a) sharing research and industry practices related to AI technologies and their governance; (b) promoting and sustaining the responsible use and adoption of AI technologies by businesses and across the community; and (c) encouraging commercialisation opportunities and collaboration between researchers, academics and industry.178 Additionally, certain agreements recognise the need to promote internationally aligned frameworks, which could include ethical and governance frameworks for the trusted, safe and responsible use of AI technologies.179 In this regard, one treaty explicitly references principles and guidelines on AI of relevant international bodies, such as the OECD and the Global Partnership on Artificial Intelligence.180 Another calls for the further use of risk-based DEPA, Art. 11.1; Chile–Paraguay FTA, Art. 7.22; New Zealand–UK FTA, Art. 15.20. Singapore–UK DEA, Art. 8.61–P. 174 India–United Arab Emirates CEPA, Art. 9.13. 175 DEPA, Art. 11.1; Singapore–UK DEA, Art. 8.61–P; New Zealand–UK FTA, Art. 15.20.4. 176 China–Mauritius FTA, Art. 12.10.2(c); Israel–Korea FTA, Art. 17.3.1(j); EU–Organisation of African, Caribbean and Pacific States, Art. 46.5. 177 Japan–UK CEPA, Art. 8.83.2(i). 178 Australia–Singapore DEA, Ch. 14, Art. 31; Australia–UK FTA, Art. 20.4. 179 DEPA, Art. 4.2. 180 New Zealand–UK FTA, Art. 15.19. 172 173

52  Research handbook on digital trade approaches to regulation based on industry-led standards and risk management best practices, and to have regard to the principles of technological interoperability and technological neutrality.181 Up to now, only one treaty has contemplated leveraging AI and other emerging technologies in government for the efficient planning, delivery and monitoring of public policies.182 5.6

Open Government Data

Along the same lines, some of these agreements also include provisions on ‘open government data’, which usually recognises that access and use of government information foster economic and social development, competitiveness and innovation. Certain of these treaties define ‘government information’ as non-proprietary information, including data, held by the central level of government.183 USMCA (2018) was the first agreement where such provisions were included. It established that if a country decides to make government information, including data, available to the public, it shall endeavour to ensure that the information is in a machine-readable and open format and can be searched, retrieved, used, reused and redistributed. Likewise, countries commit to cooperate to identify ways in which they can expand access to and use of public government information and data, to enhance and generate business opportunities, especially for small and medium-sized enterprises (SMEs).184 A similar provision was included in subsequent agreements.185 Other treaties added productivity improvement as another benefit of open government data and included best efforts commitments to have the information appropriately anonymised and containing descriptive metadata,186 and for data to be in a spatially enabled format, made available via reliable, user-friendly, and freely available Application Programming Interfaces (APIs), regularly updated, not subject to use conditions that are discriminatory or that unnecessarily restrict reuse, and made available for reuse in full compliance with the Parties’ respective personal data protection rules.187 Some agreements focus on cooperation efforts to identify ways countries can expand access to and use of open data for business opportunities. Such activities include (a) jointly identifying sectors where open datasets, particularly those with global value, can be used to facilitate technology transfer, talent formation and innovation, among other things; (b) encouraging the development of new products and services based on open datasets; and (c) fostering the use and development of open data licensing models in the form of standardised public licences available online, which will allow open data to be freely accessed, used, modified and shared by anyone for any purpose permitted by the Parties’ respective laws and regulations, and which rely on open data formats.188

UK–Singapore DEA, Art. 8.61–R. This agreement also refers to emerging technologies, including distributed ledger technologies, digital twins, immersive technologies and the Internet of Things. New Zealand–UK FTA, Art. 15.19. 182 India–United Arab Emirates CEPA, Art. 9.13.2(f). 183 Australia–Singapore DEPA, Ch. 14, Art. 27; Australia–UK, Art. 14.13.1. 184 USMCA, Art. 19.18. 185 Japan–US DTA, Art. 20; Japan–UK CEPA, Art. 8.82; India–United Arab Emirates, Art. 9.12; New Zealand–UK FTA, Art. 15.17; UK–Singapore DEA, Art. 8.61–H. 186 Australia–Singapore DEPA, Ch. 14, Art. 27; Australia–UK, Art. 14.13. 187 EU–UK Trade Cooperation Agreement, Art. 210. 188 DEPA, Art. 9.5; Chile–Paraguay FTA, Art. 7.15. 181

Three generations of digital trade provisions  53

6. CONCLUSION The first provisions on electronic commerce included in preferential trade agreements date back more than 20 years. Such clauses appeared in a context where the multilateral trade rules negotiated in the Uruguay Round that gave birth to the WTO seemed at least partially insufficient to respond to the challenges posed by the emergence of electronic commerce. Over the past two decades, such clauses have evolved in at least three distinct generations. The first one, mainly centred on e-commerce rules, shows an important level of regulatory convergence on specific objectives and principles (such as facilitating and promoting e-commerce, or avoiding unnecessary barriers), as well as on particular commitments such as a moratorium on custom duties, electronic authentication, paperless trading, consumer protection and unsolicited commercial electronic messages. The second generation of these treaties has been strongly focused on data flows rules. Such provisions were included first in discrete services sectors (mainly financial services and telecommunications, but also computer and related services and the audio-visual sector), but also in general terms, through rules on the cross-border flow of data and restrictions to data localisation. In that context, privacy has become an essential trade topic, with PTAs striving to interface with domestic regulatory regimes which can diverge significantly across jurisdictions. Notably, the EU and the US have played a significant role in pushing for the adoption of different regulatory standards in this regard, but that has also shifted over time, with the US subscribing to certain international standards and principles of data collection and processing, and the EU agreeing on the inclusion of data flows language.189 Finally, the third generation of such agreements has given birth to the first ‘pure’ digital economy agreements, which include novel clauses. Even though these provisions are primarily non-binding, they will probably charter the negotiation of future DEAs and PTAs with digital trade provisions. They could also influence the JSI negotiations currently taking place at the WTO. In fact, an important number of the topics that have been described in the three generations of agreements are signalled as key areas where convergence between member States is deemed possible.190

Burri and Polanco (n 7). These topics are consumer protection, spam, e-signatures and electronic authentication, paperless trading digital trade facilitation, source code, open government data, market access, customs duties on electronic transmissions and open internet access. 189 190

4. E-commerce or digital trade? Why the difference should matter to trade lawyers Wolfgang Alschner

1. INTRODUCTION For two decades, preferential trade agreements (PTAs) have included dedicated chapters on “electronic commerce” (or e-commerce) to account for the greater use of information technology in trade.1 The United States–Mexico–Canada Agreement (USMCA), concluded in 2019, broke with that nomenclature with a chapter on “digital trade.” This change in label from e-commerce to digital trade, echoed in other recent trade agreements,2 academic writing,3 and the wider policy debate, is not mere semantics.4 It instead represents an expanding focus of trade rules, from digitally enabled trade in analog assets to trade in digital assets. This chapter argues that we should think about these two activities—digitally enabled or electronic commerce, versus trade in data and other digital assets—as distinct spaces at the intersection of information technology and trade that come with very different norms, political economies, and policy implications. E-commerce, understood as the digitally enabled trade of analog goods and services (for example, ordering a book online from abroad) or their digital avatars (for example, ordering an online book online from abroad), is not much different from the trade that the rules of the World Trade Organization (WTO) and PTAs have governed for decades.5 In contrast, fully digital trade, understood here as the exchanges of data or other digital-only assets, is often not “trade” at all in the traditional sense, because much of it is the gratuitous byproduct of online interactions. Furthermore, transborder flows of digital assets raise a host of novel economic, political, regulatory, and normative concerns, which range from protecting personal privacy to safeguarding deliberative democracy and differ in magnitude and nature from long-debated “trade and …” concerns.6 Unfortunately, norms on digitally enabled trade and trade in digital assets are today typically lumped together in trade agreements. This is analytically unhelpful and potentially 1 The Australia–Singapore Free Trade Agreement (FTA) was the first to include a dedicated e-commerce chapter. 2 Consider, for example, the US–Japan Digital Trade Agreement (2019), the Australia–United Kingdom FTA (2021) or the New Zealand–United Kingdom FTA (2022). 3 Mira Burri and Rodrigo Polanco, “Digital Trade Provisions in Preferential Trade Agreements: Introducing a New Dataset” (2020) 23 Journal of International Economic Law 187. 4 Gregory Shaffer, “Trade Law in a Data-Driven Economy: The Need for Modesty and Resilience” (2021) 20 World Trade Review 259, 269 (noting that this changing terminology is accompanied by changes in the scope of PTAs). 5 Nivedita Sen, “Understanding the Role of the WTO in International Data Flows: Taking the Liberalization or the Regulatory Autonomy Path?” (2018) 21 Journal of International Economic Law 323, 324 (apart from it being digitally enabled, “all other aspects [of e-commerce] are similar to other conventional trade in goods and services”). 6 See generally, Shaffer (n 4).

54

E-commerce or digital trade?  55 dangerous. The more trade agreements expand from digitally enabled trade to trade in data, the less established trade law paradigms can be presumed to apply. Moreover, what is terra nullius for trade lawyers is the long-trodden turf of national data protection officers, regulators, and internet governance stakeholders. Trade law’s intrusion into these spaces risks colonizing native governance structures and upsetting existing balances. What is needed, instead, is a drawing of boundaries. Digitally enabled trade is part of the bread and butter of trade lawyers and it belongs in dedicated e-commerce chapters. Trade in data, in contrast, is still foreign to trade law(yers), may not yet be “treaty-ready,”7 and is best dealt with alongside other data governance issues. Separating digitally enabled trade and trade in digital assets comes with numerous advantages. It fosters norm consolidation and convergence around digitally enabled trade that PTAs have been regulating for 20 years, while allowing norm development and experimentation when it comes to much less mature trade-in-data rules. Keeping the two worlds more clearly apart also allows integrating e-commerce rules more tightly within the existing trade regime, including via their multilateralization in ongoing e-commerce talks at the WTO. In contrast, trade in data rules would benefit from greater integration with other regimes and alternative, multi-stakeholder governance structures and could form part of dedicated agreements alongside data protection and privacy issues, as done in the 2020 Digital Economy Partnership Agreement (DEPA) between Chile, New Zealand, and Singapore. In short, the normative and analytical separation of e-commerce and digital trade rules is vital for their respective future development. This chapter is structured as follows. Section 2 attempts to draw a line between digitally enabled trade and trade in digital assets to counter existing trends to conflate the two spaces. Sections 3 and 4 introduce the dataset and methodology and show empirically how rulemaking in PTAs has increasingly moved from one to the other. Section 5 argues that both spaces are underpinned by very different goals, economics, and regulatory concerns. Section 6 discusses the advantages of their analytical separation, identifies risks of their conflation, and discusses ongoing rulemaking efforts that promise to draw clearer boundaries between the two spaces.

2.

DRAWING A LINE: DIGITALLY ENABLED TRADE VERSUS TRADE IN DATA

Definitions in the field of digital trade are notoriously fuzzy and contested.8 In part, that is because technological change disrupts not only existing business models; it also upsets existing legal definitions and categories.9 This section highlights why definitions matter, how current categorization tends to be fuzzy, before making the case for distinguishing digitally enabled trade from trade in digital assets.

Dan Ciuriak, “Digital Trade: Is Data Treaty-Ready?” CIGI Paper No 162. Waterloo: Centre for International Governance Innovation, 21 February 2018, 24. 8 Most academic work in the area starts with a section on definitions. 9 See generally RS Neeraj, “Trade Rules for the Digital Economy: Charting New Waters at the WTO” (2019) 18 World Trade Review S121. 7

56  Research handbook on digital trade 2.1

Why Categories Matter

Is the streaming of a movie from a foreign website a trade in a (digital) product, like the purchase of the movie’s DVD? Or is it the consumption of a service? Or is it something altogether different, like a trade in information? These categories matter because law is about putting real life into legal boxes to define legal consequences. The WTO Agreements, for example, create different rules for trade in goods versus trade in services, hence categorizing the streaming of a movie as a good or a service comes with varying normative implications.10 Indeed, the absence of consensus on whether digital trade involves the consumption of a digital good or a digital service (or something else) has hampered negotiations of e-commerce rules at the WTO for decades.11 Aside from triggering varying legal consequences, categorization is also a pre-condition for rulemaking. Categories set focal points that shape joint expectations and shared understandings.12 As Robert Wolfe puts it, “if you name something, it becomes more tractable for disciplines.”13 This is particularly important for new forms of transactions that do not fit existing trade paradigms. As the OECD notes, novel commercial activities, such as social networking—where platforms provide free services to collect user data, which is then mined to sell targeted advertising services to other businesses—appear to escape existing distinctions and raise new policy concerns.14 Before they can be placed into a new or existing legal architecture, these new transactions have to be defined and distinguished from existing forms of trade. 2.2

Lacking Appetite for Categorization

Trade lawyers have struggled to classify the multi-dimensional impact of information technology on trade. The default has instead been to fall back on general principles. The first is the technological neutrality of trade rules.15 In US–Gambling, the WTO Appellate Body clarified that the same market access and national treatment privileges have to be afforded to the deliv-

Sen (n 5) 330. Even if it were clear which agreement applies, a host of other issues ensued on how to fit twenty-first-century transaction into schedules of tariffs or commitments negotiated in the pre-internet era. See ibid 327–35. 11 Mira Burri, “A WTO Agreement on Electronic Commerce: An Enquiry into Its Legal Substance and Viability” (Social Science Research Network 2021) SSRN Scholarly Paper 3976133 4 https://​papers​ .ssrn​.com/​abstract​=​3976133 accessed 20 April 2022. New technologies such as 3D printing further blur the boundary between these categories. 12 Geoffrey Garrett and Barry R Weingast, “Ideas, Interests, and Institutions: Constructing the European Community’s Internal Market” in Robert O Keohane and Judith Goldstein (eds), Ideas and Foreign Policy: Beliefs, Institutions, and Political Change (Cornell University Press 1993). 13 Robert Wolfe, “Learning about Digital Trade: Privacy and E-Commerce in CETA and TPP” (2019) 18 World Trade Review S63, 78. 14 M Jouanjean and López González, “Digital Trade: Developing a Framework for Analysis,” vol 205 (2017) OECD Trade Policy Papers 205 13 www​.oecd​-ilibrary​.org/​trade/​digital​-trade​_524c8c83​-en accessed April 20, 2022. 15 Andrew D Mitchell, “Towards Compatibility: The Future of Electronic Commerce within the Global Trading System” (2001) 4 Journal of International Economic Law 683, 688; Merit E Janow and Petros C Mavroidis, “Digital Trade, E-Commerce, the WTO and Regional Frameworks” (2019) 18 World Trade Review S1, 2. 10

E-commerce or digital trade?  57 ery of gambling services whether they are provided in direct interaction or through the medium of the internet.16 GATS rules are thus a priori blind as to whether trade occurs in cyberspace or in the real world.17 Technological neutrality together with the belief that trade law’s broad, general principles, such as non-discrimination, can accommodate all forms of commerce have taken some of the urgency out of the debate as digital trade becomes just another form of trade. In practice, however, this offers only limited guidance given that, as further discussed below, it is not clear whether parts of the digital economy involve trade at all, and how far trade rules should go in regulating these transactions. Partly cognizant of these limitations, the WTO launched its Work Program on e-commerce in 1998. The definition of e-commerce adopted was extremely broad, encompassing “the production, distribution, marketing, sale or delivery of goods and services by electronic means” to allow discussions that would extend to “all trade-related issues relating to global electronic commerce.”18 By implication, as Andrew Mitchell notes, e-commerce, in the jargon of trade negotiators, covers all business conducted online,19 which by today’s standards would turn most trade into acts of e-commerce. The label of e-commerce is thus potentially extremely broad. The more recent category of “digital trade” has further muddied the waters. It is often used as a synonym for e-commerce. For example, the chapter on “digital trade” in the USMCA and the chapter on “e-commerce” in the CPTPP cover broadly similar issues despite their different names.20 Moreover, in scholarly writing and policy documents the two terms are often used synonymously.21 The fuzziness of these notions makes it more challenging for the trade regime to make sense of the interaction of trade and information technology. First, negotiators risk talking past each other without clear and shared categories and problem definitions, and this prevents assessment of the scope of existing rules and the need for new rulemaking. Second, trade lawyers may inadvertently underestimate the challenges created by new technology because it is unclear whether they present “more of the same” or give rise to completely new types of questions. Third, trade lawyers may unwittingly encroach on other fields, such as privacy and data governance, that have emerged in response to unique concerns raised by the availability and commercial use of personal data. A clearer categorization of the distinct issues arising from the impact of information technology on trade would help mitigate these concerns.

Appellate Body Report, US–Gambling, WT/DS285/AB/R, adopted April 7, 2005. In contrast, the GATT may be require the existence of a physical product and may thus be “technologically biased.” See Joost Pauwelyn, “Squaring Free Trade in Culture with Chinese Censorship : The WTO Appellate Body Report on China: Audiovisuals” (2008) 11 Melbourne Journal of International Law 119, 132. 18 WTO, Work Programme on Electronic Commerce, WTO Doc WT/L/274 (1998). 19 Mitchell (n 15) 686. 20 The two chapters share more than 50 percent of their text. W Alschner and R Panford-Walsh, “How Much of the Transpacific Partnership Is in the United States–Mexico–Canada Agreement?” (2020) 17 Transnational Dispute Management (TDM) www​.transnational​-dispute​-management​.com/​ article​.asp​?key​=​2734 accessed May 31, 2022. 21 Burri (n 11). Drawing a distinction between the two, Susan Ariel Aaronson and Patrick Leblond, “Another Digital Divide: The Rise of Data Realms and Its Implications for the WTO” (2018) 21 Journal of International Economic Law 245. 16 17

58  Research handbook on digital trade 2.3

Separating Digitally Enabled Trade and Trade in Digital Assets

Unlike trade lawyers, trade economists have attempted to disentangle the various types of transactions generally regrouped under the label of e-commerce and digital trade. Although different typologies have emerged, depending on whether emphasis is on placed on the type of asset traded, the parties to the transaction, or the delivery method, a commonality of these typologies is the distinction between digitally enabled trade and trade in data.22 Digitally enabled trade is closely associated with the original understanding of e-commerce as the sale of goods and services online.23 The term e-commerce rose to prominence with the growth of analog trade via the internet in the late 1990s and early 2000s. Think of eBay, created in 1995—an electronic peer-to-peer marketplace where goods from shoes to cars were exchanged. What was new here was not what was traded, but how it was traded: old-fashioned trade was enabled and facilitated through an information technology infrastructure. As the OECD notes, much digitally enabled trade continues to involve the physical delivery of goods and services.24 While the sharing of data and information is crucial to support such digitally enabled trade, they are not its focus.25 In contrast, trade in digital assets is about transferring data and information as an independently traded asset. It is closely associated with the term “digital trade,” which became commonly used only in the 2010s with the rise of the data economy.26 Trade in data has since emerged as the fifth global flow alongside goods, services, capital, and labor.27 And it will likely dwarf its peers: digital assets—data, information, algorithms, and models—are “the new oil” that has fueled much of the recent economic growth and that accounts for a growing share in the creation of new wealth.28 In practice, digitally enabled trade and trade in digital assets are often intertwined. Data is traded in its own right, but it also acts as input for the production of goods and services and as facilitator of trade transactions.29 Moreover, many online transactions constitute both the delivery of a good or service and the generation and transfer of data.30 The point of distinguishing between digitally enabled trade and trade in digital assets is then not about disentangling their role in individual transactions, but about highlighting the different policy concerns they raise. Digitally enabled trade is about allowing old-fashioned trade to be conducted virtually. The main challenge it creates is norm-equivalence: how can policymakers ensure that same conventions governing traditional trade also apply to trade enabled through technology? How can traders engage in e-commerce and how can businesses and consumers trust in and rely on such 22 See e.g. Dan Ciuriak and Maria Ptashkina, “Toward a Robust Architecture for the Regulation of Data and Digital Trade” 24; Jouanjean and López González (n 14) 13. 23 Aaronson and Leblond (n 21) 248. 24 Jouanjean and López González (n 14) 13. 25 Internet (openness) has been recognized as pre-condition for all trade. See Neha Mishra, “Building Bridges: International Trade Law, Internet Governance, and the Regulation of Data Flows” (2019) 52 Vanderbilt Journal of Transnational Law 49, 501. 26 Google N-gram viewer was used to trace the usage of the term in published works. See also Ciuriak (n 7). 27 ibid 9. 28 United Nations, “Data Economy—Radical Transformation or Dystopia?” [2019] UN Frontier Technology Quarterly; Sen (n 5) 323. 29 Jouanjean and López González (n 14). 30 Mishra (n 25) 472.

E-commerce or digital trade?  59 electronic contracting? In contrast, trade in digital assets, as will be further illustrated below, is an altogether novel phenomenon. The main challenge it creates is norm-creation: how to regulate the newly emerging practice of trade in data? This difference between digitally enabled commerce and trade in digital assets is well reflected in domestic law. As Aaronson reports, most states in the world have e-commerce laws governing digitally enabled trade.31 These tend to be highly similar, which has been helped by the UNCITRAL Model Law on Electronic Commerce, created in 1996 as a common template for the domestic regulation of commerce using electronic means, which has since been supplemented by a model text on electronic signatures (2001).32 PTAs routinely include provisions endorsing or calling for the adoption of these model laws. The result is a relative convergence when it comes to e-commerce rulemaking at the national level. In contrast, trade in digital assets is just beginning to be regulated and few domestic jurisdictions have digital governance rules.33 Moreover, those states that have forged ahead with such rules have approached data governance from widely different angles. Whereas the European Union’s approach is shaped by an understanding of personal data protection as fundamental individual right, the United States’ perspective is influenced by its technology giants whose commercial success depends on data access, transfer, and analysis.34 Meanwhile China approaches data flows from a national security perspective to facilitate surveillance and promote the growth of national data champions.35 Hence, in contrast to the global convergence of e-commerce norms, there is a global divergence in digital trade rules that divide the world into distinct “data realms.”36 In part, these differences have to do with the differing policy challenges. Extending trading rules from the analog world to cyberspace (e-commerce or digitally enabled trade) is much easier than creating new rules that regulate altogether novel transactions (trade in digital assets). Yet, it is also the nature of trade in digital assets that turns their regulation into a uniquely challenging governance problem for trade lawyers. First, it is unclear whether trade in data is trade at all. Three-quarters of existing digital data is created by internet users.37 While this data is transferred across borders and clearly valuable, even vital, for data-driven businesses, much of it is not “traded” in a commercial sense but rather the “free” byproduct (or “data exhaust”) of other online activities.38 Furthermore, a lot of the information crossing borders is not imported or exported in a traditional sense. Most international data flows are the result of efficiently rerouted internet traffic.39 Moreover, data sharing involves virtually no transaction costs and data itself is a nonrival good that does not

Susan Ariel Aaronson, “What Are We Talking about When We Talk about Digital Protectionism?” (2019) 18 World Trade Review 541, 547. 32 https://​uncitral​.un​.org/​en/​texts/​ecommerce/​modellaw/​electronic​_commerce. 33 Aaronson (n 31) 547. 34 Svetlana Yakovleva, “Privacy Protection(Ism): The Latest Wave of Trade Constraints on Regulatory Autonomy” (2019) 74 University of Miami Law Review 416, 464–81. 35 Aaronson and Leblond (n 21). 36 ibid. 37 Mishra (n 25) 473. 38 Jouanjean and López González (n 14) 13. 39 Mishra (n 25) 473. 31

60  Research handbook on digital trade become less valuable when shared.40 Therefore it can easily be transferred and exist in many places simultaneously. Hence, digital trade challenges existing notions of how trade works. Second, the economics and modus operandi of digital trade are different. Ciuriak and Ptashkina call trade centered on the flow and capitalization of data “mode 5,” to denote its novelty and distinctiveness from other uses of data in trade.41As Shoshana Zuboff has shown masterfully, the data-driven economy is based on “surveillance capitalism.”42 Personal data is mined from free online activities. Once patterns are extracted, the resulting insights about personal behavior can be resold on markets in future behavior or packaged into predictive algorithms, for targeted advertisement or similar. This business model requires and gives rise to economies of scale, leading to competition-limiting internet behemoths.43 Although different types of data are being traded,44 personal data is currently the most abundant and is, indirectly, embedded in other types of data, such as social or business data.45 Since most of the data economy thus involves mining personal data, “surveillance capitalism” creates a fundamental challenge for individual privacy, but also for the functioning of society and democracy at large. Third, because data is not only an asset or form of capital, but also a carrier of personal information, an act of speech, or a transmitter of business or government secrets, its trade triggers a range of non-trade concerns. These range from the aforementioned risks to privacy and democracy, but also link to debates on misinformation all the way to (cyber)security and risks of a new-technology arms race.46 Admittedly, trade lawyers are used to dealing with what has become known as “trade and …” issues.47 Trade in cigarettes conflicts with public health objectives and environmental protection measures may be challenged as disguised protectionism. However, trade in digital assets adds a new magnitude to these challenges, raises new questions about the nature of digital trade barriers versus legitimate domestic data regulation, and forces trade lawyers to engage with unfamiliar fields such as data protection. In short, trade in digital assets is different and comes with core concerns that featured at most peripherally in trade policy circles until recently.48 Against that background of core differences between the policy implications of traditional e-commerce and new digital trade, it comes as a surprise that trade agreements lump together norms on digitally enabled trade and trade in data. As the next section shows, highly technical clauses on the acceptance of electronic signatures that seek to facilitate electronic contracting exist alongside data localization provisions that deeply affect how states regulate the flow and protection of personal information.

40 Charles I Jones and Christopher Tonetti, “Nonrivalry and the Economics of Data” (2020) 110 American Economic Review 2819. 41 Ciuriak and Ptashkina (n 22). 42 Shoshana Zuboff, The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power (Illustrated edition, PublicAffairs 2020). 43 Ciuriak and Ptashkina (n 22). 44 For a typology, see Sen (n 5) 343–5; Aaronson and Leblond (n 21) 250. 45 Mishra (n 25) 473. 46 Shaffer (n 4). 47 JP Trachtman, “Institutional Linkage: Transcending ‘Trade and...’” [2002] American Journal of International Law 77. 48 Shaffer (n 4).

E-commerce or digital trade?  61 The rest of this chapter argues that we should treat e-commerce, narrowly defined as digitally enabled trade of goods and services, and digital trade, narrowly defined as trade in digital assets, as separate spaces. Digitally enabled transactions often have components or a pendant in the real economy and can be seamlessly captured by extending or adapting existing trade law paradigms. In contrast, where technology has given rise to altogether novel forms of transactions such as trade in data, it is much less clear what role trade law has to play. Because labels matter and categorizing has legal and policy implications, norms on digitally enabled trade and trade in digital assets need to be teased apart.

3.

DATASET, METHODOLOGY, AND TAXONOMY

To tease rules on digitally enabled trade apart from those regulating trade in digital assets, we conducted an original empirical mapping of e-commerce and digital trade rules in PTAs.49 While extensive databases on digital trade norms already exist,50 they tend to lack the finegrained connection between text and coded variables that is vital for a comprehensive and in-depth content analysis of PTAs. 3.1 Dataset Our starting point was the Text of Trade Agreement (ToTA) full-text dataset of 449 PTAs notified to the WTO signed between 1948 and 2015.51 We extracted 47 chapters that refer to e-commerce, the first being the 2003 Australia–Singapore PTA. We then updated the corpus with 13 PTAs signed between 2015 and 2019 that contained an e-commerce or digital trade chapter. We also identified four treaties that contain extensive e-commerce sections within a larger chapter, which we added to our database. We did not include e-commerce-related clauses from other chapters or sections or PTAs that were not available in English. Our final dataset comprised 64 treaties, broken down into 725 articles, representing 94 countries and 57 percent of the WTO membership. However, as can be seen in Figure 4.1, some states, especially developed countries, have signed many treaties while other regions, in particular Africa, are not part of any PTA with an e-commerce chapter. 3.2

Methodology and Taxonomy

We then classified the 725 unique articles into subject matter categories. To do so, we consulted other mapping efforts of PTA content and complemented this with an inductive investigation of the texts to build our own categories. The result was a classification scheme of 20 categories and a total of 63 sub-categories. To easier navigate the data, we built an interactive dashboard using Shiny in the programming language R.

I am grateful for the excellent research assistance of Sarah Crothers and Tom Walker. Burri and Polanco (n 3). 51 Wolfgang Alschner, Julia Seiermann, and Dmitriy Skougarevskiy, “Text of Trade Agreements (ToTA)—A Structured Corpus for the Text-as-Data Analysis of Preferential Trade Agreements” (2018) 15 Journal of Empirical Legal Studies 648. 49 50

62  Research handbook on digital trade

Figure 4.1

Distribution of PTAs with e-commerce chapters or sections

For this chapter, we further aggregated our categories based on whether they primarily govern digitally enabled trade or concern trade in digital assets. Many of our categories deal with treaty administration and do not fit neatly in either box. They include articles on chapter scope, definitions, inconsistencies with other chapters, cooperation, or dispute settlement. We excluded them for this analysis. Other categories, however, can be grouped meaningfully. Articles on the treatment of digital products, paperless trading, electronic authentication, and transparency are all concerned with facilitating and enabling trade. In contrast, provisions on data flows, location of computing facilities and the treatment of source code are directed at trade in digital assets. Finally, rules on consumer protection, the protection of personal information, and unsolicited messages are relevant to both spheres (albeit to varying degrees). Figure 4.1 summarizes the categories of clauses and the typology used in this chapter.

4.

SHIFTING NORMS IN PTAS

The content of PTAs has shifted dramatically since the first chapter on e-commerce was included in the 2003 Australia–Singapore FTA.52 Most substantive articles still concern the facilitation of digitally enabled trade, however, especially since 2015 PTAs have include new provisions that focus on trade in digital assets. Indeed, most recently concluded PTAs cover both dimensions. Figure 4.3 illustrates this change. This section unpacks this change and shows that norms on digitally enabled trade essentially aim to extend existing trade rules to digital transactions to achieve technological neutrality. In contrast, norms on trade in digital assets go beyond existing trade paradigms and create novel governance structures.

For an in-depth assessment of this shift, see Rodrigo Polanco’s chapter in this volume.

52

E-commerce or digital trade?  63

Figure 4.2 4.1

Taxonomy of PTA provisions

From Facilitating Digitally Enabled Trade …

The bulk of e-commerce provisions focuses on facilitating digitally enabled trade. This is done in two main forms that together aim to establish an equivalence between electronic and analog commerce. A first set of articles extends core trade principles to e-commerce transactions. This reinforces the idea of the technological neutrality of trade norms and ensures that similar rules apply to analog and digital transactions. A second set of clauses translates digital practices into analog terms, for example through rules on the authentication of digital signatures. The basic architecture of international trade rules at the WTO is simple. States can impose market access restrictions, which take the form of tariffs when it comes to trade in goods, and quantitative or qualitative restrictions when it comes to trade in services. Behind the border, states must not discriminate against foreign goods and services and need to maintain transparency on laws and regulations affecting them. Otherwise, states are largely free to regulate goods and services in their territories (subject to more specialized WTO Agreements). E-commerce chapters in PTAs extend that architecture to digitally enabled trade. First, half of all PTAs in our dataset contain general principles that place electronic commerce at par with non-electronic commerce. 50 percent of PTAs affirm that general WTO rules apply to electronic commerce.53 Many of these agreements also call on parties “to ensure that bilateral trade through electronic commerce is no more restricted than other forms of trade” underscoring the idea of equivalence.54 Finally, 30 percent of PTAs have rules on transparency

53 54

See e.g. Australia–China FTA (2015) Art. 12.1, Canada–Colombia FTA (2008) Art. 1502. ibid.

64  Research handbook on digital trade

Figure 4.3

Changing content of e-commerce chapters in PTA

requiring parties to make e-commerce rules promptly available, mirroring similar obligations in the GATT and GATS.55 Second, when it comes to specific rules, most PTAs mirror the basic architecture of the GATT. On border measures, the most common rule across all treaties in our dataset relates to a moratorium on custom duties for digital products. In all, 92 percent of agreements, equivalent to 50 percent of WTO members, have agreed to such clauses. They thereby commit to liberalize market access for digital products. This large consensus is partially explained by an early moratorium on custom duties on digital products and services that WTO members agreed to in 1998 and have extended ever since.56 When it comes to behind-the-border measures, 44 percent of PTAs in our dataset mandate national treatment of foreign and domestic electronically traded products. Further, 42 percent of treaties also clarify that states remain otherwise free to impose behind-the-border taxes and charges, with some treaties calling on parties to establish regulatory frameworks based on international standards and to minimize regulatory burden. Finally, 33 percent of treaties subject e-commerce to general exceptions on public policy or essential security. At the same time, digitally enabled trade also creates conversion challenges on how actions in cyberspace relate to real-world transactions. A large part of e-commerce chapters is therefore dedicated to translating digital acts into real world acts to facilitate digitally enabled commerce. First, 52 percent of PTAs in our dataset comprise clauses on paperless trading. These clauses mandate that electronic versions of trade documents are to be accepted as legal equivalent to paper documents. Second, 44 percent of PTAs contain provisions on the validity

See e.g. ASEAN–Australia–New Zealand FTA (2009) Art. 10.3. Burri (n 11) 4, 6.

55 56

E-commerce or digital trade?  65 of electronic signatures and other authentication methods. The idea is again to treat digital acts like their analog counterparts. The only aspect where e-commerce chapters went beyond the regulatory architecture of the WTO from the outset concerns consumer protection. In recognition of the vulnerability of users and their data in electronically enabled transactions, 53 percent of PTAs contain rules on the protection of consumer and 48 percent of them have obligations relating to the protection of personal data. Again, at least initially, the focus of these rules was to create equivalence of protection as compared to non-electronic transactions: the Australia–Singapore PTA (2003), for example, states in Art. 6 of its e-commerce chapter that “[e]ach Party shall […] provide protection for consumers using electronic commerce that is at least equivalent to that provided for consumers of other forms of commerce under their respective domestic laws.”57 In short, norms on digitally enabled trade in PTAs seek to extend WTO principles to the digital realm and to benchmark e-commerce practices against analog trade practices. In addition, these norms have two elements in common. First, they have been part of e-commerce regulation from the start. Hence, they were shaped at a time when electronic commerce was principally about enabling analog trade through digital means and they predate the rise of the data economy in the 2010s. Second, in part because they have been around for some time, they have been adopted around the globe and come as close as it gets to a common nucleus of global e-commerce rules. 4.2

… to Regulating Trade in Digital Assets

This contrasts starkly to newly emerging norms governing digital assets. Particularly since 2015, our mapping of PTAs shows a marked shift, with the inclusion of new provisions that go beyond digitally enabled trade to regulating trade in data. Recent e-commerce chapters consider data and information as assets in their own right. The free flow of information has long been an area identified for cooperation in PTAs58 or was framed in best-efforts terms.59 It was only after 2015 that obligations mandating data flow liberalization, including of personal information, became more widespread.60 Article 1411 of the Transpacific Partnership (TPP) states that “[e]ach Party shall allow the cross-border transfer of information by electronic means, including personal information, when this activity is for the conduct of the business of a covered person.” Similar clauses have been included in 11 percent of PTAs in the dataset, regrouping primarily states around the Pacific. The obligation is accompanied by an exception allowing data flow restrictions to the extent they are necessary to achieve a legitimate public policy objective, but the scope of such exception varies.61 A small number of recent agreements also calls on states to make government data accessible, open, and machine readable.62 Emphasis added. See e.g. Canada–Colombia FTA (2008), Art. 1507. 59 South Korea–USA FTA (2007), Art. 15.8. 60 TPP (2016), Art. 1411. 61 Manfred Elsig and Sebastian Klotz, “Data Flow-Related Provisions in Preferential Trade Agreements: Trends and Patterns of Diffusion” in Mira Burri (ed), Big Data and Global Trade Law (1st edn, Cambridge University Press 2021) www​.cambridge​.org/​core/​product/​identifier/​9781108919234​ %23CN​-bp​-2/​type/​book​_part accessed April 20, 2022. 62 See e.g. USMCA Art. 19.18 “Open Government Data.” 57 58

66  Research handbook on digital trade Accompanying this shift toward trade-in-data rules are new obligations on the infrastructure necessary for mining and extracting value from data. In all, 17 percent of PTAs prevent states from conditioning market access on the transfer or access of source code or algorithms, while 12 percent of PTAs prohibit requirements to use local computing facilities. Both clauses are accompanied by exceptions that vary in strength from giving states considerable leeway to impose restrictions (such as RCEP) to those severely curtailing policy space (such as USMCA). Intermediary platforms and computer services form another important pillar of the data economy. A small number of PTAs (4 percent) have begun to regulate their role by generally shielding intermediaries from liability in relation to content that is processed or displayed under conditions and subject to exceptions. Finally, the move toward trade in digital assets has prompted the inclusion of new cyber-governance issues and the expansion of consumer protection norms: 38 percent of PTAs include cooperation provisions on cybersecurity matters, while 16 percent of PTAs contain clauses protecting consumers from unsolicited commercial electronic messages. The main takeaway from this section is thus that a new generation of norms and principles are emerging that regulate trade in data and its infrastructure. Whereas traditional e-commerce rules sought to bring digitally enabled trade under the principles governing analog trade, the rules on trade in digital assets seek to erect a new governance architecture only loosely inspired by rules on analog trade.

5.

SHIFTING PARADIGMS: WHY RULES ON TRADE IN DATA ARE DIFFERENT

This shift in the legal landscape of PTAs is significant because it brings trade rules to bear on altogether new issues. Digitally enabled trade takes familiar forms. It involves the trade of goods and services while leveraging today’s information technology architecture. Digitally enabled trade may change how we trade, but does not fundamentally change what we trade. In contrast, trade in digital assets enlarges the very definition of trade. Think of a free search engine collecting data on requests from users across the globe for use in a machine learning algorithm to continuously improve the said free search engine. What is traded is not a good or a service, but data and information. Until recently, such data and information would have been considered a worthless byproduct—mere “data exhaust”63—and it is still seen as such by the billions of users who part with their data freely every day without considering it as trade at all. And yet, that simple exchange of information powers companies like Google, worth $1 trillion US dollars. If these exchanges are considered an international trade, then a new and increasingly important dimension of virtual human interaction falls under the ambit and the governance regime of international trade. Although it seems natural for trade agreements to govern and liberalize these new, at least partially commercial, transborder flows, this extension of trade rules to a new asset class is less than straightforward. After all, distinct asset classes tend to be governed by distinct international norms because they raise varying policy concerns. Differing rules apply to trade in goods than to trade in services, and capital and labor are regulated outside the WTO entirely.

Zuboff (n 42).

63

E-commerce or digital trade?  67 Three reasons in particular warrant caution when it comes to an expansion of existing trade paradigms to trade in purely digital assets: (1) regulating trade in data is less about liberalization than a market intervention that creates protection for new asset classes; (2) the political economy of trade in data differs from trade in goods and services; and (3) it raises distinct “trade and …” issues. 5.1

Asset Creation and Protection

While a car is valuable per se, the idea of a car is worth nothing. It takes an intellectual property rights regime to take ideas, inventions, and creations and turn them into assets. Although trade agreements used to be about liberalizing the exchange of existing assets, they have become, as Ciuriak and others have pointed out, “asset value protection” agreements: they are less about removing trade barriers and more about the “protection of the value of corporate assets,” including through ambitious property rights and investment protection obligations.64 This, Ciuriak argues, necessitates a re-evaluation of traditional liberal economic accounts on the winners and losers from trade agreements. Owners of protected assets (and states hosting them) are poised to gain more from such trade agreements, while innovation and competition is stifled and smaller, open, rule-taking economies suffer. Similarly, trade agreements that protect data and information, in fact, create value and redistribute it by offering specific protections to digital assets. As noted above, other than the delivery of paid online services, it is far from clear whether the bulk of cross-border exchanges of data, from online searches to social networking, amount to trade at all. By including these exchanges under the label of trade, their commercial aspect is emphasized and they are placed into the legal boxes of trade law, triggering normative consequences such as obligations to ensure data liberalization. Irrespective of whether this characterization of data exchanges as trade is appropriate, it involves a degree of “conversion.” It turns a transaction into a trade that has not traditionally been viewed as such. This, in turn, has distributional consequences. It is unclear in most domestic jurisdictions who owns the data that is being traded. In the EU, where privacy is a fundamental right, users can request the deletion or transfer of “their” data, including part of their online history. In most jurisdictions, however, users have little control over data they generate online, which is then de facto or de jure owned and controlled by the company that harvested it. Similarly unclear are the dividing lines between company property and governmental access and regulation. Trade agreements with data provisions intervene in this emerging ecosystem by empowering those who hold and control digital assets. Consider algorithms and source code, whose forced transfer or access is curtailed in many recent PTAs. As Słok-Wodkowska and Mazur, argue this “secrecy by default” approach bestows source code with greater protection from regulatory intervention than is enjoyed by existing intellectual property rights.65 Not only do PTAs thereby make certain data assets (here source code) more valuable; they also limit the states’ scope for domestic regulation on these

Dan Ciuriak, Nobel Laureate, and Michael Spence, “A New Name for Modern Trade Deals: Asset Value Protection Agreements” (Centre for International Governance Innovation 2017) www​.jstor​.org/​ stable/​resrep17312​.7 accessed April 27, 2022. 65 Magdalena Słok-Wódkowska and Joanna Mazur, “Secrecy by Default: How Regional Trade Agreements Reshape Protection of Source Code” (2022) 25 Journal of International Economic Law 91. 64

68  Research handbook on digital trade assets. As the authors point out, the secrecy of source code, for example, conflicts with regulatory efforts to make data-driven applications more transparent and accountable, including via access to source code.66 In short, trade in data provisions create and protect a new asset class with distributional consequences. This is controversial because it occurs at a time when questions on the nature of data, its ownership and control, and its governance infrastructure are only beginning to be resolved, with diverse states providing diverse answers. According to Robert Wolfe, the US is in that context using PTAs to “externalize its regulatory preferences” that give its companies a competitive edge.67 In short, trade rules on digital assets warrant special consideration because they do not merely liberalize a widely traded asset, but they turn information into assets in the first place, amid a regulatory landscape that is yet to decide whether data is a private property or a commercial asset. 5.2

Different Political Economy

A second reason why an extension of trade rules to govern trade in digital assets raises special concerns is the different economic and political economy structures of trade in data as compared to digitally enabled analog trade. To be sure, digitally enabled trade has created disruption across industries. Books are sold online rather than in brick-and-mortar stores. Yet, the nature of trade and the drivers of liberalization and protectionism have remained largely unchanged. Trade in digital assets, or “surveillance capitalism” as Shoshanna Zuboff calls it, however, works by a different economic logic.68 Data is accumulated en masse, mined for patterns, and packaged into analytics that can be sold on behavioral future markets. Monopolies, aggressive expansion, and elimination of competition all evolve naturally from that business model. While their market power and informational impact is immense, these companies are lightweight when it comes to brick-and-mortar assets, which makes them harder to tax and regulate. The interaction between the non-monetary trade in data via free user searches or social networking and the indirect monetization of that data through advertisement create additional regulatory and conceptual difficulties. In short, trade in digital assets is different. As Zuboff puts it, the economic model of surveillance capitalism is “unprecedented.”69 All that leads to different political economy dynamics than those trade lawyers are used to. There is often no import-competing industry as few firms capture global markets, which are mostly based in the United States. This challenges the mercantilist logic of existing trade negotiations that leverages domestic export interests to trade off reciprocal market access concessions. Negotiations thereby move away from trade. Between the EU and the US, for example, market access rights to US internet giants appear to be bargained against higher European taxation and regulation. Moreover, given the sensitivity of data, including personal information, and its widespread non-commercial uses, non-economic considerations play

ibid. Wolfe (n 13) 77. 68 Zuboff (n 42). 69 ibid 13. 66 67

E-commerce or digital trade?  69 a greater role in evaluating the pros and cons of data liberalization.70 In short, faced with the unprecedented nature of surveillance capitalism, it is unclear whether and how traditional paradigms can guide rulemaking on the trade of digital assets. 5.3

Thorny “Trade and …” Issues

Although trade lawyers are used to thinking about the interface of trade and other normative concerns, from promoting public health to protecting the environment, trade in digital assets lifts this interaction to an entirely new level. More than any other asset or means of production, data and information have crucial non-economic components. Flows of data are intimately connected to a range of policy concerns including misinformation, freedom of speech, democracy, privacy, and national security. Add the economic policy considerations around data governance, including issues relating to corporate taxation and competition.71 While none of these issues are trade issues per se, they are impacted by trade rulemaking on trade in data. In addition, although these issues are new for trade lawyers, domestic regulators and other stakeholders have been grappling with these policy concerns for some time. Take data privacy. Robust and rapidly evolving national data privacy frameworks have evolved that differ starkly between states, mirroring varying societal concerns. However, data protection has also become a trade topic and, potentially, a trade barrier.72 The expansion of trade norms thus clashes with existing regimes and established norms, raising questions about how ensuing value conflicts are resolved and by whom. The point is then not that trade rules have no role to play when it comes to exchanges of data and other digital assets. Rather, the main takeaway from this scoping of issues is that rules on trade in digital assets are sufficiently distinct from digitally enabled trade and trade in analog assets to warrant special considerations. Hence, the need for greater analytical separation between digital trade issues in trade agreements.

6.

THE WAY FORWARD: THE CASE FOR ANALYTICAL SEPARATION

If we accept the argument that rather than dealing with one set of “e-commerce” or “digital trade” issues, states are grappling with two very different spheres of economic activities of digitally enabled trade and trade in digital assets respectively, then a central policy consequence flows from this: states and stakeholders should start separating the two in rulemaking and policymaking. Such analytical separation should start with varying labels and a deeper understanding of their differences. On that basis, it can lead to differing lawmaking and different lawmakers in the two spheres. Indeed, even at the WTO, there is a long tradition of providing separate rules

70 Yakovleva (n 34) 516 (“when a policy conversation, such as the one on cross-border flows of personal data, involves non-economic spill-over effects to individual rights, such conversation should not be confined within the straightjacket of trade economics, but rather placed in a broader normative perspective”). 71 Neeraj (n 9). 72 Burri (n 11) 4.

70  Research handbook on digital trade to specific product or service types that raise special regulatory or political economy concerns. For example, telecommunication and financial services are dealt with through bespoke annexes to the GATS and typically come in separate chapters in PTAs. In addition, some international economic law issues, from labor to taxation and finance, are regulated outside of trade agreements entirely. Since trade in digital assets is special, too, it should be dealt with in distinct PTA data governance chapters, or, as further argued below, distinct agreements. This section summarizes advantages of treating digitally enabled trade and trade in data as distinct policy spaces. 6.1

Sharpening Normative Development in PTAs

To begin with, a separation of norms on digitally enabled trade and those on trade in data would sharpen normative developments in both fields. Clauses on digitally enabled trade have proliferated and matured considerably since their first appearance in 2003. In PTA negotiations, the challenge lies not so much in agreeing to first principles; as noted above, some “classic” e-commerce provisions, like a moratorium of custom duties on digital products, are almost ubiquitous in existing PTAs. The challenge instead lies in agreeing on specific wording. For example, in our dataset, paperless trading clauses in 33 PTAs share on average only 43 percent of the same wording. Although there are pockets of highly converging language,73 there are also significant differences between some treaties. For illustration, Figure 4.2 compares the wording of the Australia–Thailand PTA (2004) with the TPP (2016) in relation to their respective paperless trading obligations. They only share 20 percent of their text and are normatively distinct.

Figure 4.4

Core e-commerce clauses, for example on paperless trading, vary starkly

These variations across core e-commerce principles are particularly challenging for norm-taking developing states. When they sign up to similar but not identical terms, it creates an additional burden on treaty administration and compliance as compared to a network of internally consistent commitments.74 Convergence on specific language would help states to comply with their international obligations, which can be facilitated when states negotiate dedicated chapters on digitally enabled trade.

For example, the US–Colombia PTA and the Singapore–Taiwan PTA have identical clauses. Most non-compliance with international law is unintentional. See Abram Chayes and Antonia Handler Chayes Chayes, The New Sovereignty: Compliance with International Regulatory Agreements (Harvard University Press 1995). 73 74

E-commerce or digital trade?  71 A separation would also allow negotiators to focus on controversial issues and harvest low-hanging fruit more quickly.75 Since most of the norms on trade in digital assets are new and, as noted above, raise particularly controversial issues, they risk dragging down progress even when consensus is reached on digitally enabled trade. Moreover, separating the two may also help distill consensus candidates within trade in digital assets chapters. For example, rules on source code are adopted more widely than liberalization of data flows or requirements on the use of computing facilities. In short, a separation would aid the normative development in both fields. 6.2

Facilitate Norm Convergence on the Multilateral Level

As of this writing, multilateral negotiations over e-commerce are ongoing at the WTO. At the Buenos Aires Ministerial Conference in 2017 WTO Members committed to starting negotiations on trade-related aspects of electronic commerce. After running into roadblocks in 2019, 76 WTO members sought to reinvigorate negotiations by launching a joint statement initiative on electronic commerce. Although the statement called for a “high standard outcome,” the proposals and draft texts circulated since point to a more modest outcome centered on digitally enabled trade.76 According to Mira Burri, a future WTO electronic commerce agreement is therefore likely to focus on facilitating e-commerce and excluding contentious issues on data flows and governance. 77 Rather than lamenting such a “thin” deal, it should be celebrated for focusing rulemaking on digitally enabled trade and for leaving trade on digital assets aside (for now). As discussed above, norm convergence is much more advanced on digitally enabled trade and its implications are better understood. This makes PTA rules on digitally enabled trade low-hanging fruit for multilateralization. In contrast, a multilateral agreement on trade in digital assets is hampered by disagreement among states on the role and degree of data liberalization.78 Moreover, the data economy remains poorly understood, which has prompted commentators such as Ciuriak to observe that data is not treaty-ready.79 Conversely, as Mira Burri notes, even a thin deal would create renewed confidence in the WTO’s lawmaking capabilities and establish normative focal points with global reach.80 Most importantly, however, it would give negotiators and policymakers time to reflect on what future governance of trade in data looks like, and what role trade rules and the WTO play in that space. 6.3

Prevent Norm Colonization

Finally, an analytical separation between digitally enabled trade and trade in digital assets would channel the energy of trade lawyers into the former space where they can feel bullish but exhort modesty when they enter the latter. 75 See generally, Ines Willemyns, “Agreement Forthcoming? A Comparison of EU, US, and Chinese RTAs in Times of Plurilateral E-Commerce Negotiations” (2020) 23 Journal of International Economic Law 221. 76 Burri (n 11). 77 ibid. 78 Willemyns (n 75). 79 Ciuriak (n 7). 80 Burri (n 11).

72  Research handbook on digital trade An aggressive extension of trade into the trade of digital assets risks norm colonization as trade norms and paradigm clash with the existing rules on data protection, data governance, and the broader internet architecture. As Svetlana Yakovleva points out, trade paradigms can shift defaults and perspectives.81 When free trade in data becomes the norm and data privacy the exception, legitimate data protection risks being reconceived as illegitimate data protectionism.82 Moreover, the data governance space has native regulatory structures. Internet governance, for example, is based on established norms (such as internet openness, security, and privacy) and multi-stakeholder governance mechanisms, in part via soft law and extralegal conventions.83 A growing number of trade law scholars have therefore argued for modesty and respect for diversity and existing norms to give trade lawyers time to contemplate how to make sense of trade in data.84 Finally, an analytical separation can fuel the development of a proper data governance regime that sees trade in data as just one of a panoply of policy issues created by the emergence of the data economy. Early signs suggest that this may be emerging. The Digital Economy Partnership Agreement (DEPA) between Chile, New Zealand, and Singapore, but open for signature by other states, was concluded in 2020. It addresses data governance holistically, including trade in data norms, as also found in recent PTAs; however, these provisions are embedded into a broader array of data economy concerns that stretch all the way to frameworks for ethical artificial intelligence.

7. CONCLUSION Today’s e-commerce and digital trade chapters in PTAs, in fact, contain two sets of norms: those extending traditional trade concepts to facilitate digitally enabled trade and those setting up novel norms to govern trade in new digital assets. It would help the normative development of both spaces to separate them more clearly and to acknowledge the unique issues created by trade in data.

Yakovleva (n 34). ibid 496 (“by labelling certain domestic policies such as restrictions on cross-border data flows and data localization measures as digital protectionism, it is much easier to critique them, reject them, and put competing policy interests such as privacy, data protection, or industrial policy in a subordinate position”). 83 Mishra (n 25) 478–97, 507–9. 84 Shaffer (n 4); Mishra (n 25). 81 82

5. WTO law and cross-border data flows for digital trade Rolf H. Weber

1. INTRODUCTION Technological advances and the increasing importance of data have substantially impacted the regulatory environment of cross-border data flows and international trade. The already complex normative framework is confronted with the fact that global data access and the delivery of digital assets challenge the existing normative regime. Many questions can be raised, such as:1 How do existing trade rules apply to data flows? Should data be classified as a good or a service (and in such case, as part of which services sector)? How can new trade barriers, such as localization measures, be addressed? How should the free flow of data be reconciled with privacy, national security and other public interest concerns in national legislations? Which are the appropriate forum and the decision-making processes for moving the global data economy agenda ahead? During the twentieth century, trade was largely composed of exchanges of physical goods, later increasingly accompanied by trade in services. Now, trade is substantially determined by cross-border data exchanges.2 The global flow of data fundamentally alters the ways the world economy works: digitalization is not only changing how trade is executed, it is also changing who trades.3 Data is rapidly becoming the cornerstone of the global economy, representing a new type of economic asset, commercial tool and social value.4 Yet, digital trade is not defined in the WTO trade rules and a definition of digital trade that is recognized and accepted worldwide does not exist; equally, no specific multilateral rules are available.5 According to the OECD, there is a growing consensus that digital trade encompasses digitally enabled transactions of trade in goods and services that can be digitally or physically delivered.6 Therefore, if the WTO trade rules are interpreted in a technologically neutral way, the mere fact that trade is conducted other than by means of a physical transaction

Mira Burri, Data Flows and Global Trade Law, in Mira Burri (ed.), Big Data and Global Trade Law (Cambridge University Press, Cambridge 2021), 11, 15. 2 James Bacchus, Trade Links. New Rules for a New World (Cambridge University Press, Cambridge 2022), 137. 3 Bacchus (supra n 2), 138. 4 UN Secretary-General António Guterres’ foreword to the United Nations Conference on Trade and Development’s report 2021 on the digital economy: Digital Economy Report 2021, Cross-border Data Flows and Development: For Whom the Data Flow, https://​unctad​.org/​webflyer/​digital​-economy​ -report​-2021; see also below section 4.1. 5 Bacchus (supra n 2), 141; see also Gregory Shaffer, Trade Law in a Data-Driven Economy, in Shin-yi Peng, Ching-Fu Lin and Thomas Streinz (eds), Artificial Intelligence and International Economic Law (Cambridge University Press, Cambridge 2021), 29, 30­–3. 6 OECD, The Impact of Digitalization on Trade. 1

73

74  Research handbook on digital trade should not change WTO obligations, notwithstanding the need for certain adaptations of the available provisions. This chapter addresses the existing WTO law as a pre-digital legal order by outlining the weaknesses of the normative framework, in section 2. Thereafter, the evolutions in the digital data and trade law are discussed (section 3). The main sections of this contribution look at the new technological challenges (chapter 4), the specific problems of cross-border data flows (chapter 5) and the potential new data-driven trade law architecture (chapter 6). The chapter concludes with an outlook for the future (chapter 7).

2.

WTO LAW AS PRE-DIGITAL LEGAL ORDER

The WTO Agreements, constituting the basis of international trade law, were negotiated during the Uruguay Round (1986–94) and entered into force in 1995.7 At that time, the advent of the internet, and particularly widespread digitalization, was hardly foreseeable. Despite some adjustments to new technologies – for example by the Information Technology Agreement (ITA), concluded in 19968 and updated in 2015, as well as the Trade Facilitation Agreement (TFA)9 – the original WTO law has not been fundamentally changed.10 2.1

Existing Normative Framework

As mentioned, the WTO law basically remained a pre-digital legal order. During the Uruguay Round, e-commerce was still in its infancy; consequently, neither the General Agreement on Tariffs and Trade (GATT) nor the General Agreement on Trade in Services (GATS) provide a legal definition of digitally oriented terms such as ‘cross-border data flows’, ‘e-commerce’ or ‘digital trade’.11 2.1.1 Classification issues Cross-border data flows and digital trade mostly lack physical attributes and are offered on a temporary basis. In practice, cross-border data flows and trade frequently cover items having similarities to the rendering of services. Therefore, at first glance, digital trade appears to be classifiable as ‘trade in services’.12 The term ‘services’ itself, however, is relatively vague and the GATS does not clarify whether this term also encompasses electronically supplied services. The definition in article I:3(b) GATS only states that the notion of ‘services’ includes ‘any services except services supplied in the exercise of governmental authority’. The services

General Agreement on Tariffs and Trade (GATT) 1994, 1867 U.N.T.S. 187; I.L.M. 1153 (1994); General Agreement on Trade in Services (GATS) 1994, 1869 U.N.T.S. 183; 33 I.L.M. 1167 (1994). 8 WTO, Ministerial Declaration on Trade in Information Technology Products, WT/MIN(96)/16 (1996). 9 WTO, Protocol, Decision of 27 November 2014, WT/L/940 (2014). 10 See also Mira Burri, The Governance of Data and Data Flows in Trade Agreements: The Pitfalls of Legal Adaptation, 51 UC Davies Law Review 2017, 65–132. 11 Rolf H. Weber, A New International Trade Framework for Digital Assets, in Meredith Kolsky Lewis et al. (eds), A Post-WTO International Legal Order (Springer, Cham 2021), 277, 281. 12 See also Rolf H. Weber, Digital Trade and E-commerce: Challenges and Opportunities of the Asia-Pacific Regionalism, 10 Asian J WTO Int Health Law 2015, 321, 324 with further references. 7

WTO law and cross-border data flows for digital trade  75 sectoral classification list (W/120) containing a comprehensive list of all services (sub-)sectors covered by the GATS does not make any exact mention of digital trade and data flows;13 furthermore, it provides only little guidance for the treatment of e-commerce under the GATS.14 The same appreciation holds true for the term ‘digital assets’ being related to the distributed ledger technology (blockchain) infrastructure. Digital assets are usually expressed in the technological form of a so-called non-physical token, this being a data package which is encrypted and contains the relevant information about the assets. The contents of tokens can be manifold, from services to securities, commodities and minerals. Therefore, different asset classes exist that do not necessarily have the same regulatory treatment.15 2.1.2 E-Commerce Work Programme In 1998, the WTO Council adopted the ‘Work Programme on E-Commerce’;16 this document describes the term e-commerce as ‘the production, distribution, marketing, sale or delivery of goods and services by electronic means’. The main achievements of the Work Programme are the following: (i) it introduced a definition of e-commerce in para. 1.3 (though quite limited); and (ii) it highlighted taking action in different WTO law areas in paras 2–4. However, the Work Programme does not state ‘hard law’ as the provisions of the GATT and the GATS, but only expresses the guidelines to be followed by national legislators.17 Many attempts have been undertaken to further develop the Work Programme, which is often called an ‘unfinished’ agenda. However, success has been relatively limited over the years; for example, there is still no agreement in respect of a precise digital trade agenda and a permanent moratorium on customs duties for electronic data transmissions.18 Whether the most recent general negotiations about cross-border data flows and digital trade will be more successful remains to be seen.19 2.2

Outdated Goods and Services Classification

Despite the fact that the WTO law is in many respects, both in substance and in procedure, relatively flexible and resilient, the existing normative framework does not sufficiently cope with new technologies. 2.2.1 Normative analysis The GATS is based on the so-called positive-list approach, meaning that only those services appearing on the applicable lists are subject to GATS commitments. Thus, new services are not automatically covered by the GATS; moreover, Member States have to actively agree on an obligation and notify the respective commitments. As a consequence of the ‘invention’ of many new services, the WTO legal framework is not applicable to a broad scope of services, Weber (supra n 11), 281. See below chapter 2.2.1. 15 Weber (supra n 11), 281. 16 WTO, General Council, Work Programme on Electronic Commerce, 25 September 1998, WT/L/274, para. 1.3. 17 See Rolf H. Weber, Digital Trade and E-Commerce: Challenges and Opportunities of the Asia-Pacific Regionalism, 10 Asian J WTO Int Health Law 2015, 321, 327. 18 See also Burri (supra n 1), 19. 19 See below chapter 2.2.2. 13 14

76  Research handbook on digital trade except if specific digital services fall into an existing category such as ‘telecommunications services’ (for example, email or online information).20 Trade in services under the GATS is classified into four categories. Commercial presence (mode 3) and presence of natural persons (mode 4) are not of crucial importance for the cross-border data flows. However, the question of whether to classify a certain digital service as cross-border supply (mode 1) or as consumption abroad (mode 2) has often proven difficult. Case law (for example US–Gambling and China Audiovisual Services) suggests that cross-border data flows and digital trade fall within the commitments of GATS mode 1.21 Digitalization is also transforming goods. Goods can be stored, used and distributed in a digital form; these so-called digital assets are intangible goods (for example, downloaded ebooks or streamed music plays, e-tickets or software). Whereas it appears to be uncontested that products purchased online and delivered physically fall into the scope of the GATT, the legal situation is less clear if the end-product itself is delivered electronically:22 is a film downloaded online ‘like’ a film purchased as a DVD? What about a film streamed online? The classification systems of the GATT Harmonized System (HS Code) and the GATS list (W/120) do not provide any clear guidance. Digital assets need to be processed on some kind of carrier media that is classified as tangible goods and listed in the HS Code. Thus, the legal qualification of the carrier media is quite clear, but the challenge concerns the delineation between the end of the carrier media and the beginning of the digital service: If a video game console provides for online games, audiovisual content and computer services, which legal framework applies? The qualification as good might be confronted with the reservation of a WTO member that audiovisual content can be politically sensitive.23 So far, WTO members have not reached consensus on the appropriate classification of data flows and digital services. Some voices seem to favour a classification under the GATT in case of a durable ‘good’; other voices propose a classification under the GATS arguing that digital services and data delivered electronically cannot be considered ‘like’ those products physically delivered. The WTO itself in the World Trade Report 2018 pleads for a technology-induced reshaping of the regulatory environment.24 2.2.2 Ongoing revision efforts Challenges to the normative framework by new technological developments can be overcome by applying fresh interpretation methods or by changing the regulatory regime. In the context of digital services, some voices favour the interpretative concept of the ‘predominant purpose’ test and assess the characteristics of digital services such as intangible/

Weber (supra n 11), 282; see also Bacchus (supra n 2), 132 et seq. US–Gambling: Appellate Body Report, United States – Measures Affecting the Cross Border Supply of Gambling and Betting Services, WT/DS285/AB/R (7 April 2005); China Audiovisual Services: China–Measures Affecting Trading Rights and Distribution Services for Certain Publications and Audiovisual Entertainment Products, WT/DS363/AB/R (21 December 2009). 22 Weber (supra n 11), 282. 23 As discussed in the case China Audiovisual Services (supra n 21). 24 WTO, World Trade Report 2018, The Future of World Trade: How Digital Technologies Are Transforming Global Commerce, Geneva 2018. 20 21

WTO law and cross-border data flows for digital trade  77 heterogeneous/inseparable/perishable (non-storable).25 However, even with these additional characteristics, the dilemma of classification cannot be easily overcome. Moreover, such an approach makes taxonomy even more complicated and does not really eliminate the existing tensions. Other voices plead for the realization of a hybrid regulatory approach (involving governments, multistakeholder organizations and the private sector). The WTO rules typically do not refer to private standards or industry best practices, although these are commonplace in the digital world. Improved cooperation between stakeholders would be a first step in the direction of a broader concept of rule-making.26 The WTO itself is also aware of the existing problems; therefore, efforts have been initiated to remedy the situation. In the context of the 11th Ministerial Conference (December 2017, Buenos Aires), a short ‘Joint Statement on Electronic Commerce’ (JSEC) has been adopted inviting the participating WTO members to further work on digital trade.27 Most developed countries have agreed on the JSEC; however, the negotiations did not advance substantially. More recently, a reinvigoration of some e-commerce-related negotiations among about 80 countries, based on the 2017 JSEC,28 has been agreed upon in the context of the World Economic Forum 2019. The joining countries said they sought ‘to achieve a high standard outcome that builds on existing WTO agreements and frameworks with the participation of as many WTO Members as possible’.29 Since then, numerous negotiations were conducted, including exchanges during the COVID-19 pandemic in structured virtual sessions. The negotiators have reported progress in the small groups on certain digital trade issues (for example, e-signatures and authentication, paperless trading, customs duties on electronic transmissions). The consolidated negotiating text is divided into sections on (i) facilitating electronic and other digital transactions and their logistics, (ii) openness and transparency in e-commerce, (iii) trust in e-commerce, (iv) cross-cutting issues (such as privacy and cybersecurity), (v) telecommunications, (vi) market access and (vii) general provisions.30 Nevertheless, many disruptive challenges in the context of the data-driven economy still require regulatory solutions.31

3.

EVOLUTION IN THE DIGITAL DATA AND TRADE LAW

More adaptations to the new technological needs have been realized on different regional and sub-regional levels, particularly in the form of preferential trade agreements (PTAs), some-

See Shin-yi Peng, A New Trade Regime for the Servitization of Manufacturing: Rethinking the Goods-Services Dichotomy, 54 Journal of World Trade 2020, 669–726. 26 Weber (supra n 11), 284. 27 WTO, WT/MIN(17)/60 of 13 December 2017. 28 WTO, Joint Statement on Electronic Commerce, WT/L/1056, 25 January 2019. 29 WTO, Joint Statement (supra n 28); see also James Bacchus, The Digital Decide. How to Agree on WTO Rules for Digital Trade, Centre for International Governance Innovation, Waterloo 2021, 5–6. 30 For further details see Bacchus (supra n 29), 9. 31 For further details see Mira Burri, Towards a New Treaty on Digital Trade, 55 Journal of World Trade 2021, 77–100, and Mira Burri, Trade Law 4.0: Are We There Yet? 26 J Int Econ Law 2023, 90–100. 25

78  Research handbook on digital trade times also called regional trade agreements (RTAs). Since these agreements are discussed in a separate chapter,32 the following remarks remain short. 3.1

Preferential Trade Agreements

Between 2000 and 2019, an impressive 347 PTAs were negotiated and agreed upon by a large number of states. Not less than 184 PTAs contain provisions related to digital services and data flows. At the forefront can be found provisions on e-commerce and intellectual property rights. Nevertheless, a thorough academic review of the respective chapters has shown that the concerned provisions remain highly heterogenous, addressing various issues ranging from customs duties and paperless trading to personal data protection and cybersecurity.33 Furthermore, the depth of the commitments and the extent of their binding nature varies very significantly. For example, the European Union (EU) and Singapore have gone quite far in their respective PTAs (newly called Digital Economy Partnership Agreements, DEPA) with the e-commerce commitments, whereas other countries restrict market access in digital services to a broader extent. An overview of data-related rules in PTAs shows that more attention is paid to the issues of prohibited data localization and cybersecurity challenges than to data privacy standards, which remain relatively low in the mix.34 3.2

Regional Accords

The most important regional accords containing digital services and data flow provisions are the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP), the Regional Comprehensive Economic Partnership (RCEP), and the United States-Mexico-Canada Agreement (USMCA). 3.2.1 East Asia and Pacific: CPTPP and RCEP The CPTPP was agreed upon in 2017 among 11 countries in the Pacific Rim and entered into force on 30 September 2018. The CPTPP represents 13.4 per cent of the global cross-domestic products. Cross-border data transfers and digital trade are addressed; however, the level of harmonization is not very high.35 Recently, the United Kingdom and China expressed interest becoming members of the CPTPP; positive developments here would see significant geographic extension. As a second regional agreement, the Regional Comprehensive Economic Partnership (RCEP) was signed in November 2020 and came into effect on 1 January 2022; it is composed of China (not yet a member of CPTPP), Japan, South Korea, Australia, New Zealand and 10 ASEAN members. In particular, article 15 of chapter 12 on electronic commerce deals with the ‘cross-border transmission of information by electronic means’. See Mira Burri, Chapter 2 in this book. For more details see Burri (supra n 1), 20–33; Manfred Elsig and Sebastian Klotz, Data Flow-Related Provisions in Preferential Trade Agreements: Trends and Patterns of Diffusion, in Mira Burri (ed.), Big Data and Global Trade Law (Cambridge University Press, Cambridge 2021), 42–62; Ines Willemyns, Agreement Forthcoming? A Comparison of EU, US, and Chinese RTAs in Times of Plurilateral E-Commerce Negotiations, 23 J Int Econ Law 2020, 221–44. 34 See also Bacchus (supra n 29), 6–7. 35 Burri (supra n 1), 34–6; Bacchus (supra n 2), 145–6. 32 33

WTO law and cross-border data flows for digital trade  79 3.2.2 North America: USMCA The USMCA was adopted in November 2018 (in force since 1 July 2020) and contains a comprehensive electronic commerce chapter which is also properly titled ‘digital trade’. Some provisions relate to data protection and to the use of algorithms; however, exception clauses for the pursuit of non-economic objectives are also contained.36

4.

NEW TECHNOLOGICAL CHALLENGES

The most important technological advances concern infrastructure with distributed ledger technology (DLT, also called blockchain) and software programming with artificial intelligence (AI) elements. 4.1

Blockchain and Trade

The new DLT not only offers an alternative infrastructure but also allows the digitization of physical goods in the form of ‘data packages’, that is, so-called tokens. The data flow crosses the border with the objective to enable digital trade. 4.1.1 New trade infrastructure The new DLT infrastructure could help to make the execution of international trade transactions more efficient. This aspect is relevant in connection with the recently adopted Trade Facilitation Agreement (TFA) by way of improving customs clearing processes and reducing the need for manual verification.37 Requests for advanced rulings (article 3 TFA), if submitted through a DLT platform, would be securely stored on the blockchain, in a permissioned ledger, and would remain accessible at all times for authorized stakeholders. The sharing of required data on the ledger in real time could also facilitate pre-arrival processing and the expedited release of goods (article 7 TFA).38 Blockchain could equally be a valuable tool for traceability and transparency features in the context of the Technical Barriers to Trade (TBT) and Sanitary and Phytosanitary (SPS) agreements, as well as in the intellectual property rights environment.39 Fostering transparency of trade regimes lies at the heart of the WTO work; the monitoring of legislative developments related to DLT could be performed as part of the WTO Trade Policy Review process to keep track of the evolution of the DLT regulatory environment.40 4.1.2 Treatment of digitized assets The WTO legal framework does not regulate aspects of ownership and property (with the exception of acknowledging intellectual property rights); its focus is on transactional issues.

See Burri (supra n 1), 37–9. Emmanuelle Ganne, Blockchain’s Practical and Legal Implications for Global Trade and Global Trade Law, in Mira Burri (ed.), Big Data and Global Trade Law (Cambridge University Press, Cambridge 2021), 128, 137. 38 Ganne (supra n 37), 137. 39 For further details see Ganne (supra n 37), 139–40. 40 Ganne (supra n 37), 156–7 with further references. 36 37

80  Research handbook on digital trade The challenge with digital assets consists in the fact that data and tokens are not tangible and, as a consequence, national property laws do not easily apply. A new digital assets regime should more rely on ‘factual control’ than on a traditional property understanding. Thereby, a taxonomy of asset classes needs to be developed that makes sense for international trade.41 Relevant regulatory topics for the design of a legally stable environment regarding digital assets are, for example:42 (i) Definition of a digital asset: The regulatory framework should contain a description of the term ‘digital asset’ encompassing the relevant technological elements (such as hash, encryption, and so on); (ii) Conditions for the establishment of a registry: Digital assets can only be transferred and traded in a structured way if their roots are reflected on a registry; conditions for how the registrar is implementing the registry must be set (for example secure infrastructure, access to data, and so on); (iii) Effects of registration: The entry of a digital asset into the registry needs to have certain legal effects; relevant aspects are the allocation of ownership, the protection of trust of persons in the correctness of the registration and the environment for a valid transfer of a digital asset; (iv) Transfer requirements: The regulatory framework must state the requirements for the transfer of digital assets (for example form of transfer, delisting and relisting of the asset); (v) Nullification of a digital asset: As with physical goods, the ‘loss’ of a digital good cannot be excluded: for example, the entitled person loses his/her private key or the successor of a diseased person is not aware of it, making it necessary to implement nullification proceedings; (vi) Liability rules: The provider of the registry and possibly also the provider of the technological infrastructure need to be obliged to comply with reasonable diligence standards; non-compliance therewith must have liability consequences. A potential roadmap should encompass the definition of the relevant issues, the involvement of appropriate organizations and the development of substantive standards. 4.2

Artificial Intelligence and Digital Trade

There has been intensive discussion around the question of whether the WTO Agreements can be applied to artificial intelligence (AI) decision-making. In principle, the GATS is focusing on measures regulating services without regard to the technologies by which those services are provided.43 This kind of technology neutrality has also been confirmed by the Court of Appeals in the China–Audiovisual Services case:44 by interpreting an electronic version of the service as part of the traditional services commitments and by applying treaty commitments in a dynamic form, the GATS can take account of ‘changing technologies’. However, several challenges remain. For example, concerns relate to the underlying processes of how a decision is made with respect to any service. Furthermore, the regulatory capacity of the ‘deciding machine’ could be challenged. In addition, the regulation of AI should not be used to create another behind-the-border trade barrier.45 Therefore, the benefits of AI in the digital trade environment should be assessed with some caution.

Weber (supra n 11), 288. Weber (supra n 11), 289–90. 43 Anupram Chander, Artificial Intelligence and Trade, in Mira Burri (ed.), Big Data and Global Trade Law (Cambridge University Press, Cambridge 2021), 115–27. 44 See supra n 21. 45 Chander (supra n 43), 126–7. 41 42

WTO law and cross-border data flows for digital trade  81

5.

PARTICULAR ISSUES OF CROSS-BORDER DATA FLOWS

The intensively discussed cross-border data flow issues are indirectly related to trade of digital goods or services, as global data transfers are the basis for digital trade. In addition, the source of many innovations in the digital economy is the free flow of data, and the free movement of data underpins digital trade.46 As already mentioned, data is rapidly becoming the lifeblood of the global economy, representing a key new type of economic asset.47 This assessment is particularly true in respect of the digital transitions towards the Internet of Things (IoT), artificial intelligence, virtual reality and autonomous vehicles. But cross-border data flows are ‘an unfinished agenda’ of WTO law;48 in addition, the differing data-governance paradigms of major (trading) countries cause the risk that a new ‘digital divide’ is occurring.49 5.1

Main Regulatory Issues

Cross-border data flows should be legal, free and secure. The respective main regulatory issues concern topics that have the objective to protect natural persons or public interests such as privacy and cybersecurity. In this connection, the term data sovereignty gained importance; its interpretation, however, is quite contested, as China’s new ‘Data Security Law’ and ‘Cybersecurity Law’ show. 5.1.1 Privacy Privacy concerns have gained importance over the past ten years; a majority of countries around the world now have data protection laws and specifically regulate cross-border data flows. The most well-known example is the General Data Protection Regulation (GDPR) of the European Union. According to most national laws, various conditions for cross-border data transfers are applicable, in particular the principle that transfers are only allowed to those countries having an equivalent level of data protection. Potential mechanisms for the cross-border data flows are the standard contractual clauses and the binding corporate rules that provide for compliance with specific conditions if the equivalent data protection level is not given.50 However, no general international consensus exists on the best means to achieve online privacy. The core problem lies in the fact that the socio-cultural environment on privacy across countries is different; partly, the protection of the personal integrity is at the forefront, partly, the aspect of information security prevails; even consumer rights can play a role.51 Some

Bacchus (supra n 29), 26. See above section 1 and European Union, https://​euagenda​.eu/​publications/​enter​-the​-data​ -economy​-eu​-policies​-for​-a​-thriving​-data​-ecosystem. 48 This expression is used as a subtitle in Andrew D. Mitchell and Neha Mishra, WTO Law and Cross-Border Data Flows, An Unfinished Agenda, in Mira Burri (ed.), Big Data and Global Trade Law (Cambridge University Press, Cambridge 2021), 83. 49 See Susan A. Aaronson and Patrick Leblond, Another Digital Divide: The Rise of Data Realms and Its Implications for the WTO, 21 J Int Econ Law 2018, 245 et seqq. 50 Bacchus (supra n 2), 31; Mitchell and Mishra (supra n 48), 86–7 and 96. 51 Christopher Kuner, Transborder Data Flows and Data Privacy Law (Oxford University Press, Oxford 2013), 33–4. 46 47

82  Research handbook on digital trade regional agreements (CPTPP, USMCA) have now introduced a number of generally observable standards trying to achieve minimum harmonization.52 The GATS is acknowledging the importance of privacy protection as part of its exception clauses; according to article XIV(c)(ii) national restrictions are acceptable to the extent of being necessary to secure compliance with laws or regulations addressing the ‘protection of the privacy of individuals in relation to the processing and dissemination of personal data and the protection of confidentiality of individual records and accounts’. Similarly, the GATS Telecommunications Annex states that any WTO member may take such measures as are necessary to ensure the security and confidentiality of messages (para. 5[d]).53 Nevertheless, this exception clause in the GATS appears to be insufficient as it pressurizes WTO powers to adjudicate on sensitive privacy issues being particularly difficult given that ‘data-source countries’ are unlikely to accept ‘one-sided limits on their right to protect privacy’.54 In order to deal with uncertainty and distrust in privacy matters, increasingly countries seek data protection rules enabling cross-border data flows while ensuring privacy protection as an important societal objective. As a consequence, reconciliation models mapping data flows and data protection interests must be developed based on a better understanding and contextualization of their complex interfaces; thereby, the pros and cons of the available legal solutions for reconciling trade and privacy protection need to be adequately balanced.55 In addition, internationally recognized standards and guidelines, such as the OECD Privacy Framework, provide a basis for aligning privacy laws across countries.56 5.1.2 Cybersecurity Cybersecurity recently became a more debated topic since security threats have increased and international efforts to reach a mutual agreement on the UN level were not successful.57 Divergent cybersecurity laws and technical standards around the globe make the delivery of digital products more difficult; in addition, data localization on grounds of security increases costs for companies in replicating their systems across different countries.58 Like privacy protection, cybersecurity may be covered under article XIV GATS; cybersecurity measures could be necessary to maintain public order (lit. a) or to prevent deceptive and fraudulent practices or safety (lit. c) (i) and (iii).59 Furthermore, since article XIV(c) GATS is not exhaustive, it would be possible to argue that data-restrictive measures are necessary to ensure compliance with domestic cybersecurity laws.60 Article XIVbis GATS stipulates three general security exceptions, namely the information supply contrary to essential security inter See above section 3.2. Mitchell and Mishra (supra n 48), 93–4. 54 Aaditya Mattoo and Joshua P. Meltzer, International Data Flows and Privacy: The Conflict and Its Resolution, 21 J Int Econ Law 2018, 769, 789. 55 For a detailed analysis of the available reconciliation models see Mira Burri, Data Flows versus Data Protection: Mapping Existing Reconciliation Models in Global Trade Law, in Klaus Mathis and Avishalom Tor (eds), Law and Economics of Regulation (Springer, Cham 2021), 129-158. 56 See also Mitchell and Mishra (supra n 48), 105 with further references. 57 See Rolf H. Weber, Cybersecurity Governance – International Law as Policy Driver, Weblaw IT Jusletter 27 May 2021, nos. 8 et seqq. with further references. 58 Mitchell and Mishra (supra n 48), 88 with further references. 59 Rolf H. Weber and Rainer Baisch, Revisiting the Public Moral Order and the Security Exceptions under the GATS, 13 Asian J WTO Int Health Law 2018, 375, 379 et seq. 60 Mitchell and Mishra (supra n 48), 95-96. 52 53

WTO law and cross-border data flows for digital trade  83 ests (lit. a), trade restrictions which are aimed at the protection of essential national security interests (lit. b) and measures adopted in wartime or in other emergency situations concerning international relations (lit. c).61 Obviously, cybersecurity was not an issue at the time of inception of article XIVbis GATS. The specific security reasons are also not designed in a way that would easily match the cybersecurity challenges. The WTO Panel in the Russia–Measures Concerning Traffic in Transit case asserted that interests relating to the quintessential functions of the state, namely the protection of its territory and its population from external threats and the maintenance of law and public policy order internally, would be relevant if taken in good faith.62 In legal doctrine, some reluctance can be seen in respect of qualifying cybersecurity as national security in trade agreements since cybersecurity-related laws often do not draw a rational distinction between military and social/economic security and due to a lack of international consensus on cybersecurity governance.63 Indeed, complex cybersecurity measures risk burdening trade dispute settlement bodies with the adjudication of technical and political questions. An important aspect of cybersecurity is the international cooperation related to the concerned issues. Even if these issues are not in the core of the WTO scope, experience has shown that a ‘trade+’-approach is a possible concept to find a way ahead (as, for example, in case of the trade-related environmental issues).64 In addition, WTO members should be encouraged to give preference to internationally recognized standards and best practices in cybersecurity; such an approach has been chosen in the negotiations related to a few PTAs.65 However, the legal development in the cybersecurity context remains burdensome, particularly in times of increased international political tensions. 5.1.3 Data access and data-restrictive measures (data localization) The imposition of data-restrictive measures is often driven by competition considerations between governments in order to achieve more intelligence by controlling data flows.66 In particular, different governments have adopted measures to increase regulatory control by implementing data-localization laws.67 Around the globe, such requirements are designed in a variety of forms. Information and communications technology companies may be obligated to host all subscriber and consumer data locally within the country. Data localization reduces access to data and digital technologies and may also be counterproductive. Furthermore, such provisions raise the costs of access to, and use of, data, thereby reducing gains from the digital trade. Already a quite extensive examination of GATS rules applying to data-restrictive measures is available. For example, legal obligations on national treatment would be relevant if the measure favoured domestic services and service suppliers or imposed unreasonable com Weber and Baisch (supra n 59), 382. WTO Panel Report, Russia – Measures Concerning Traffic in Transit, WT/D512/7, 29 April 2019, no. 7.130 and no. 7.133. 63 Neha Mishra, The Trade: (Cyber)Security Dilemma and Its Impact on Global Security Governance, 54 Journal of World Trade 2020, 567–90. 64 Mitchell and Mishra (supra n 48), 103–4. 65 See also Bacchus (supra n 29), 29–30. 66 John Selby, Data Localization Laws: Trade Barriers or Legitimate Responses to Cybersecurity Risks or Both?, 25 Int. J.L.& Inf. Tech. 2017, 213, 231–2. 67 Mitchell and Mishra (supra n 48), 90; Bacchus (supra n 29), 26–7. 61 62

84  Research handbook on digital trade pliance requirements on foreign services and service suppliers. Nevertheless, applying the pre-digital era GATS disciplines to data-related disputes is challenging.68 Not least, the proximity of services suppliers and consumers in the digital supply chain leads to highly intrusive (and sometimes inefficient) data-restrictive measures that are also trade-inhibiting.69 5.2

Designing a Cross-border Data and Trade Governance Framework

A new data-governance framework for global trade is needed; relevant aspects are transparency and interoperability as well as digital trust. 5.2.1 Transparency, accountability and interoperability A fundamental requirement that must be addressed in WTO law is the transparency of data regulations. Practice has shown that WTO members can adopt ambiguously worded data-restrictive measures causing considerable uncertainty; the general transparency principle of article III GATS states that this should be avoided. Therefore, more mechanisms must be designed to increase governmental transparency, accountability and interoperability of data regulations.70 WTO disciplines should ensure interoperability and transparency of the internet regulations/ regulatory frameworks to facilitate data flows and greater accountability in digital networks. Indeed, varying standards of privacy and security across countries create impediments to cross-border data flows. Article VII GATS states that mutual recognition of ‘standards or criteria for the authorization, licensing or certification of services suppliers’ should be achieved. Although non-binding, article VII:5 GATS recognizes the need for more international coordination between WTO members on domestic regulations pertaining to services suppliers standards.71 Consequently, the culture of transparency, enshrining the disclosure of data logics and access to data sets and applied algorithms must be promoted. But transparency does not necessarily require more regulation; moreover, robust and general rules are desirable. The improvement of transparency does not mean a quantitative increase in information, but rather ‘more’ in terms of higher information quality.72 Accountability helps to ensure the implementation of an environment in which individuals and enterprises assume their respective responsibilities. Accountability encompasses the obligation of one person to another, according to which the former must give an account of, explain, and justify his/her actions and decisions against criteria of the same kind. The development of these concepts in public institutions and private enterprises requires accessible accounts as a precondition for a sustainable society.

Mitchell and Mishra (supra n 48), 93 and 101–2. Susan A. Aaronson, What Are We Talking about When We Talk about Digital Protectionism, World Trade Review 2018, 1, 11. 70 Weber (supra n 11), 286. 71 Mitchell and Mishra (supra n 48), 99. 72 See Rolf H. Weber, From Disclosure to Transparency in Consumer Law, in Klaus Mathis and Avishalom Tor (eds), Consumer Law and Economics (Springer, Cham 2020), 73, 79 et seqq. 68 69

WTO law and cross-border data flows for digital trade  85 Another important topic is ‘legal interoperability’,73 which addresses the processes of creating legal rules that cooperate across jurisdictions. The respective objective can be realized in a matter of degrees, as many options exist between a full harmonization of normative rules and a complete fragmentation of legal systems. Striking the correct balance is of utmost importance: while an excessively high level of interoperability could cause difficulties in the management of the harmonized rules and fail to acknowledge social and cultural differences, a level too low could present challenges to smooth social interaction. As a consequence of legal interoperability, network neutrality is a desirable objective: all market participants (providers of goods and services as well as consumers) should have unfettered access to the digital infrastructure. Since the factual control of data is key in a digitized world, misuse of access to the infrastructure is to be avoided; otherwise, the ‘controller’ can become a ‘data demagogue’, contradicting the basic principles of an appropriate international data and trade regime.74 A further challenge, more due to efforts of governments than private enterprises, concerns data-localization requirements. Around the globe, such requirements are implemented in a variety of forms. Information and communications technology companies may be obligated to host all subscriber and consumer data locally within the country. In some instances, only information covering certain substantive areas (for example health) must be stored in the country. Data localization reduces access to data and digital technologies and may also be counterproductive.75 Requiring data localization in relation to cybersecurity increases data vulnerability in a single jurisdiction, making it easier to target and possibly prevent data backups in globally distributed data centres. 5.2.2 Digital trust and promotion of business WTO disciplines should enable digital trust, in turn requiring the preservation of user privacy and consumer protection as well as protection against fraudulent transactions and cybersecurity attacks.76 The GATS only indirectly, and in an insufficient manner, contributes to digital trust as it permits WTO members to adopt certain domestic measures.77 This situation underestimates the fact that an improvement of business trust could be achieved without insurmountable efforts. In particular, the development of appropriate digital standards and of codes of conduct would be desirable.78 Other institutions, such as the Internet Engineering Task Force (IETF), the World Wide Web Consortium (W3C), the Institute of Electrical and Electronics Engineers (IEEE) and the International Telecommunications Union (ITU), play a more vital role in the debates about digital trust and their standards are suitable to serve as a good example.

73 See John Palfrey and Urs Gasser, Interop: The Promise and Perils of Highly Interconnected Systems (Basic Books, New York 2012), 132–3 and 178–9. 74 See Rolf H. Weber, Global Law in the Face of Datafication and Artificial Intelligence, in Shin-yi Peng, Ching-Fu Lin and Thomas Streinz (eds), Artificial Intelligence and International Economic Law (Cambridge Univ. Press, Cambridge 2021), 53, 63–4. 75 To data localization see above section 4.1.3 and Weber (supra n 74), 65. 76 Mitchell and Mishra (supra n 48), 98. 77 See Andrew D. Mitchell and Neha Mishra, Regulating Cross-Border Data Flows in a Data-Driven World: How WTO Can Contribute, 22 J Int Econ Law 2019, 389, 403. 78 Bacchus (supra n 29), 32–3; Mitchell and Mishra (supra n 48), 107–8.

86  Research handbook on digital trade To contribute to digital trust, the WTO framework should (i) facilitate increased transnational dialogues as well as international regulatory coordination and cooperation on relevant issues, such as data flows, cybersecurity and privacy; and (ii) safeguard policy space necessary for countries to enable and maintain the trust in domestic cyberspace.79 Furthermore, the WTO bodies should more closely cooperate with other international bodies trying to implement more robust standards of digital trust.80

6.

DATA-DRIVEN TRADE LAW ARCHITECTURE

A new data-driven trade law architecture must encompass the implementation of adequate governance principles and adjusted digital trade rules. 6.1

New Governance Principles

The term ‘governance’, which stems from the Greek word ‘kybernetes’ and the Latin word ‘gubernator’, means steersman. The function of a steersman is of utmost importance when manoeuvring a vessel, and equally so in the cross-border data flow environment. Governance can be addressed in global trade from the perspective of different disciplines; nevertheless, at whatever level of social organization it may take place, governance refers to the appropriate business conduct of a private or a public body.81 The development of a broader rule-making approach encompassing the needs of the cross-border data flows regime and the digital environment of today’s societies appears to be unavoidable and equally justified. Relevant elements are transparency, accountability, interoperability, and digital trust, which – as mentioned82 – can be based on manifold governance guidelines. Consequently, international data and trade law must be adapted by exploring innovative and inclusive approaches in digital trade and by establishing data regulation that considers the multistakeholder nature of the internet governance regime, including the role of the private sector in ensuring openness and security of data flows.83 Different approaches have been developed. Some voices argue that the regulation of digital technologies requires polycentric governance, including a self-regulatory approach in highly technical areas; the significance of private sector instruments is obvious and reference can be made to the regulation of data flows in a ‘form of legal pluralism’ with no single authoritative framework. Other voices argue that data regulation requires a hybrid approach involving a mixture of prescriptive and self-regulatory approaches.84

Mitchell and Mishra (supra n 48), 98. See also Mitchell and Mishra (supra n 77), 412. 81 See Rolf H. Weber, Shaping Internet Governance. Regulatory Challenges (Schulthess, Zurich 2009), 2. 82 See above section 5.2. 83 For a detailed analysis of the multistakeholder approach see Rolf H. Weber, Internet Governance at the Point of No Return (EIZ Publishing, Zurich 2021), 44 et seqq. 84 For the regulatory discussions in detail see Weber (supra n 83), 27 et seqq. 79 80

WTO law and cross-border data flows for digital trade  87 6.2

Data-oriented Digital Trade Rules

6.2.1 Notion and scope of digital trade A first step in developing a data-driven trade law architecture consists in the notion and scope of digital trade. Looking at the continuing and accelerating technological evolution of digital trade, involving ever more varied dimensions of constantly transforming international commerce, digital trade should have a broad notion and a wide scope that allows potential extensions.85 Furthermore, a distinction between ‘digital trade’ and ‘e-commerce’ seems no longer justified, since the two terms are often used interchangeably. 6.2.2 Clarification of the application of existing WTO rules to digital trade Notwithstanding the fact that the existing WTO framework does not directly address digital trade, a certain number of provisions can be applied to digital trade in a comparable way. In particular, the distinction between the provisions governing the sale of goods and the provisions governing the rendering of services is problematic.86 Indeed there is no economic or legal logic for making this distinction, it is simply a consequence of a political compromise made during the Uruguay Round.87 The historical burden should now be overcome and new approaches are to be implemented. 6.2.3 Development of specific modules for a digital trade agreement Possible ideas helping to develop specific modules for a digital trade agreement can be drawn from the various DEPA; their negotiations in bilateral or plurilateral contexts have led to manifold experiences and learnings which are also valuable in the WTO environment. According to Bacchus, the following modules merit consideration:88 (1) initial provisions and general definitions; (2) business and trade facilitation; (3) treatment of digital products and related issues; (4) data issues; (5) wider trust environment; (6) business and consumer trust; (7) digital identities; (8) emerging trends and technologies; (9) innovation and the digital economy; (10) small and medium enterprises’ cooperation; (11) digital inclusion; (12) joint committee and contact points; (13) transparency; (14) dispute settlement; (15) exceptions; and (16) final provisions. Based on these modules developed in the DEPA context,89 it would be possible to introduce modular obligations in a WTO digital trade agreement.90 6.2.4 Trade facilitation Trade facilitation becomes possible if the new technologies such as blockchain are properly implemented (for example through the improvement of customs clearing processes).91 Furthermore, facilitated digital trade is achievable through the realization of so-called low-hanging fruit, that is, through easily implementable normative adaptations. The respective WTO digital modules could include, for example, (i) trade facilitation through electronic 85 Bacchus (supra n 29), 10; a broad analysis is now contained in Mira Burri, The Impact of Digitalization on Global Trade Law, 24 German Law Journal 2023, 551–573. 86 See above section 2.2. 87 Bacchus (supra n 29), 14. 88 Bacchus (supra n 29), 8. 89 See above section 3.1. 90 Bacchus (supra n 29), 16. 91 See above section 4.1.1.

88  Research handbook on digital trade transaction frameworks, (ii) transparency, (iii) non-discriminatory treatment of digital goods, (iv) electronic signature (invoicing, payments, paperless trading), (v) logistics services, (vi) digital identity and cryptography, (vii) cooperation in online consumer protection and cybersecurity, (viii) interoperability of data protection regimes, (ix) unsolicited electronic messages (spam), (x) data innovation, (xi) artificial intelligence and machine learning, (xii) prudential measures, (xiii) digital inclusion, and (xiv) dispute settlement.92 Furthermore, new business models must become part of the trade rule framework. For example, digital platforms should be able to overcome barriers that have prevented small companies from participating in international cross-border data flows and digital trade as well as from facilitating the building of trust in global transactions. Regulatory support for the implementation of digital platforms that are stable and trustworthy is an objective that must be pursued.93 6.3

International Regulatory Cooperation

Cross-border data flows and digital trade are in the process of transforming the global legal framework. Nevertheless, recent experience has shown that many governments have become less inclined to agree on new, internationally accepted rules and are increasingly restricting cross-border data flows. Data-localization requirements can be seen as a new form of digital protectionism that extends beyond justified objectives and leads to a fragmentation of international trade.94 In view of these (undesirable) developments, efforts must be undertaken to strengthen international regulatory cooperation. As a consequence, the interplay between global needs and national interests should be better analysed, and should also lead to appropriate design of such cooperation.95 Legal interoperability (for example through neutral recognition understandings) must become a regulatory objective; furthermore, convergence towards principles and standards in areas such as privacy and cybersecurity is increasingly imperative.96 Such developments can be successful if international regulatory cooperation between active public and private cross-border organizations is improved (in the form of interagency coordination and compliance with management measures). Obviously, in the light of growing nationalism and deepened sovereignty interests, international regulatory cooperation is difficult. But coordination is always better than confrontation.

7. OUTLOOK Cross-border data flows and digital trade are causing legal challenges, since the normative WTO framework dates from the pre-internet era. The respective regulatory problems must be overcome. Experiences in the light of the recent strengthening of national sovereignty by

For further details see Bacchus (supra n 29), 19–23. For a general overview see Luca Belli and Nicolo Zingales (eds), Platform Regulations – How Platforms are Regulated and How They Regulate Us (FGV Direito, Rio de Janeiro 2018). 94 See above section 5.1.3. 95 Weber (supra n 74), 68. 96 Weber (supra n 74), 68–9. 92 93

WTO law and cross-border data flows for digital trade  89 (autocratic) states show with increasing emphasis that changes to the existing legal GATT and GATS provisions are not easily reachable. Nevertheless, certain options could be pursued: (i)

(ii)

(iii)

(iv) (v)

(vi)

The existing WTO rules on electronic commerce could and should be reviewed, at least to the extent of collecting low-hanging fruit. For example, under article XVIII GATS, WTO members would have the possibility to adopt additional commitments on data flows, akin to the Telecommunications Reference Paper.97 A further measure consists in the possibility to introduce dedicated domestic regulations on digital trade based on article VI GATS.98 The negotiations following the adoption of the Joint Initiative, engaging about 80 WTO members, have the potential to lead to a better normative framework, even if their outcome is difficult to forecast and the plurilateral nature of a respective agreement would not cover the whole globe.99 More efforts should be undertaken with the objective of developing a set of binding international rules on data/privacy protection, cybersecurity and online consumer protection; appropriate means would be international standards and codes of conduct leading to a minimum harmonization of the applicable provisions.100 Special emphasis must be paid to the situation of the developing countries that can less profit from the benefits of digital trade. Further research is necessary to understand the development implications of data-driven growth.101 In view of the ‘flourishing’ preferential (regional) trade agreements102 it is important that the different plurilateral understandings are coordinated, at least to a certain extent (minimum harmonization). Thus, the WTO should play a more central role in designing trade rules for the digital age. Otherwise, global legal interoperability is at risk and many fragmented mosaic pieces with liberalized rules will accrue in the future. In general, regulation should become an enabler of digital innovation and should not limit business activities in an undue way. Overarching key elements of global law in the context of cross-border data flows and digital trade regimes are transparency, accountability, interoperability and trust.103 Even if an enhancement of the present regulatory frameworks entails complex and uncertain policy changes, it is important to balance the different regulatory concerns and to implement a stable normative environment for data flows.104 An optimal design for a balanced policy ecosystem must also consider further aspects such as risk assessment and ethical considerations, thereby strengthening the confidence of all involved stakeholders in the global legal framework.105

97 Andrew D. Mitchell and Neha Mishra, Data at the Docks: Modernizing International Trade Law for the Digital Economy, 20 Vanderbilt Journal of Entertainment and Technology Law 2018, 1073, 1127. 98 Mitchell and Mishra (supra n 48), 110. 99 See above section 2.2.2. 100 Weber (supra n 57), nos 50 et seqq. 101 Mitchell and Mishra (supra n 48), 111. 102 See above section 3. 103 Weber (supra n 74), 69. 104 Mitchell and Mishra (supra n 77), 416. 105 Weber (supra n 74), 59.

6. The Digital Economy Partnership Agreement (DEPA): accession to the digital-only regime Joo Hyoung Lee and David Collins

1. INTRODUCTION Amid the Covid-19 pandemic, in June of 2020, Chile, New Zealand and Singapore established the Digital Economic Partnership Agreement (DEPA), a ground-breaking free trade agreement designed to take advantage of the growing digitalization of the global economy. Since then, several other countries have expressed interest in joining this novel pact, including notably the developed, Pacific Rim economies of Canada and the Republic of Korea. DEPA, which entered into force in January 2021, is a first-of-its-kind agreement because it is exclusively focused on digital trade. It sets out new approaches on collaboration in digital trade issues, promotes interoperability between different regimes and addresses several new issues arising from digitalization including privacy, electronic contracting and artificial intelligence (AI). Digital-only trade agreements are a novel approach to encouraging digital trade, uncoupling digital from traditional elements of trade liberalization, such as tariffs and market access, found in conventional FTAs. The intention is that DEPA will complement the WTO negotiations on e-commerce (the Joint Statement Initiative) and build on the digital economy work underway within APEC and the OECD. As the WTO JSI on E-Commerce negotiations are still ongoing, DEPA can be seen as an important step towards a broader multilateral agreement.1 The parties, which collectively account for 15 per cent of the world’s GDP, further intend that this agreement will generate new ideas and that can be used by members in the WTO negotiations on e-commerce and by other countries negotiating FTAs.2 DEPA is a stand-alone, open plurilateral agreement to which other WTO members are eligible to join through a self-contained accession process. It is constructed as a ‘living agreement’ which allows for continual updates and modernization as required – essential for an instrument dealing with a rapidly changed field where technological innovation often precedes legal understanding. The disciplines contained in DEPA may have been significantly influenced by earlier agreements of which Chile, New Zealand and Singapore are also members.3 DEPA can continuously be updated and modernized as needed. The original DEPA signatories – Chile, New Zealand and Singapore – already enjoy market access to one another’s economies through their membership in the Comprehensive and Progressive Agreement for Transpacific Government of Canada, ‘Background: Canada’s Possible Accession to the Digital Economy Partnership Agreement’ www​.international​.gc​.ca/​trade​-commerce/​consultations/​depa​-apen/​background​ -information​.aspx​?lang​=​eng. 2 Overview, www​.mfat​.govt​.nz/​en/​trade/​free​-trade​-agreements/​free​-trade​-agreements​-in​-force/​ digital​-economy​-partnership​-agreement​-depa/​overview/​#bookmark1. 3 M Soprana, ‘The Digital Economy Partnership Agreement (DEPA): Assessing the Significance of the New Trade Agreement on the Block’ 13(1) Trade Law & Development 143 (2021) at 151. 1

90

The Digital Economy Partnership Agreement (DEPA)  91 Partnership (CPTPP). DEPA may consequently be thought of as a side agreement to CPTPP that builds on that agreements e-commerce chapter while also exploring new areas such as e-invoicing and emerging technologies such as AI and fintech. Intended to help contribute towards the establishment of international standards for dynamic and rapidly evolving digital economy, as well as to facilitate digital trade between signatories, DEPA has the potential to become a benchmark agreement for many countries in the coming decade. DEPA has already inspired other FTAs, notably the UK–Singapore Digital Economy Agreement (DEA), which entered into force in June 2022, and which is substantially similar in content to DEPA. This chapter will examine the main features of DEPA, noting stand-out provisions and their implications. It will then consider the potential accession of two new signatories, the developed countries of Canada and Korea.

2. FEATURES This section will provide an overview of the main features of DEPA.4 The chapters of DEPA (which it terms ‘modules’) of the agreement refine existing rules, particularly those of the CPTPP. DEPA also includes two e-commerce areas not covered by the CPTPP: e-invoicing and e-payments, found in Module 2. Although DEPA does not go as far as prescribing regulations for its signatories, the agreement acknowledges the growing significance of these features of the global economy and establishes interoperable standards through collaboration, building on a common set of e-commerce principles. The specific ‘modular’ elements of DEPA include: Business and Trade Facilitation; Treatment of Digital Products and Related Issues; Data Issues; Wider Trust Environment; Business and Consumer Trust; Digital Identities; Emerging Trends and Technologies; Innovation and the Digital Economy; Small and Medium Enterprises Cooperation; Digital Inclusion; Transparency and Dispute Settlement. The use of the term ‘modular’ in DEPA is an unusual one. It conveys the sense that the agreement is a sum of component parts rather than an integrated whole and, by implication, that some of the various elements may be disassociated from the text, perhaps with a view to maximizing policy flexibility of future accessors. Many elements of DEPA are ‘soft’ law – comprising best-efforts obligations rather than binding ones which lead to enforceable rights. These include ‘new initiatives which promote data exchange systems’ in Article 2.2.9 and the requirement that parties ‘Endeavour to take into account electronic payment systems’. There is also a commitment that parties will ‘Endeavour to promote innovation and competition’. The agreement addresses a range of emerging digital economy issues, including artificial intelligence (AI), digital identities and digital inclusion. The preamble discloses the progressive, public policy aspect of the agreement that stands out among many modern FTAs. Parties ‘REAFFIRM the importance of promoting corporate social responsibility, cultural identity and diversity, environmental protection and conservation, gender equality, indigenous rights, labour rights, inclusive trade, sustainable development and traditional knowledge, as well as the importance of preserving their right to regulate in the public interest’.

Ibid.

4

92  Research handbook on digital trade Inclusion of the Right to Regulate is becoming standard in FTAs, indicative of the shift towards recognition of signatory states’ sovereignty over important matters of public interest where in the past this may have been subordinated to the needs of multinational enterprises.5 It reflects the importance of addressing the needs of a range of stakeholders, rather than merely the commercial entity operating in an international environment. DEPA contains extensive material on electronic trade facilitation. Indeed, it is important to recognize that the agreement is not only about services – it is very much a goods agreement too. Indeed, digital trade does not exclusively concern services – it also embraces the application of digital technology to trade in goods. So-called smart borders will only become more important in the coming years, particularly where borders are politically sensitive.6 Module 2 on business and trade facilitation has some useful elements, including allowing trade documentation to be paperless and establishing legal frameworks at the domestic level to allow electronic submissions of documents like certificates of origin, import and export documentation as well as sanitary and phytosanitary (SPS) certificates (2.2). Electronic invoicing is to be encouraged and made interoperable (2.5). The module also includes specific rules on express shipments and clearance times (2.6). It further commits parties to deeper cooperation on other new approaches for logistics (2.4). There are commitments on electronic payments (2.7). These are especially important because it is exceedingly difficult for firms to engage in cross-border trade if they cannot be paid electronically over the internet. Again, it is noteworthy that some of the current language in this module, as with other parts of DEPA, is hortatory rather than binding (best efforts). This embeds flexibility into the text of DEPA, potentially attracting additional signatories concerned about the effects of binding international commitments on economic sovereignty as well as social matters such as data privacy. As with many digital trade chapters of modern FTAs, like the CPTPP and USMCA, DEPA’s Article 3.2 prohibits customs duties on digital transmissions. Interestingly, though, internal taxes on digital transmission are expressly permitted. This would appear to allow for the imposition of digital services taxes. There is also a national treatment provision (3.3). Both of these should help ensure competitive opportunities for firms entering digital marketplaces. The rules on non-discrimination of digital products (3.3) could become critically important as the digital economy continues to expand into areas that have not yet been conceived. The Agreement also establishes rules of non-discrimination for digital products in Module 3. This applies primarily to newer categories of digital products set out in the Agreement as ‘computer programme, text, video, image, sound recording or other product that is digitally encoded, produced for commercial sale or distribution, and that can be transmitted electronically’. There is a key exception for the broadcasting industry, reflecting the historic sensitivity of cultural industries. Module 4 on data issues builds on the commitments of the CPTPP, which prohibits data localization (that is, restricting the collection, processing and/or storage of data within a country) requirements to ensure free data flows. On the controversial topic of data protection, Article 4.2 of DEPA clarifies that parties must have a data protection framework, but the specific nature of that framework is not specified. This grants fairly wide latitude for countries to 5 This trend is perhaps more well associated with international investment law: see C Titi, The Right to Regulate under International Investment Law (Hart, 2014). 6 As in the highly controversial case of Northern Ireland and the Republic of Ireland. See also the UK’s adoption of the Electronic Trade Documents Bill, currently before the House of Lords (Bill H 57).

The Digital Economy Partnership Agreement (DEPA)  93 apply their own data protection rules rather than being bound by strict rules such as the EU’s General Data Protection Regulation (GDPR). Under Article 4.4 data localization is prohibited, but there are ‘standard’ public policy exceptions, such as those listed in the CPTPP relating to privacy and the protection of personal data. DEPA establishes that its signatories will cooperate on cybersecurity and online safety (Article 5) and on consumer protection. This is coupled with a requirement to have consumer protection laws, although again no particular regime is prescribed. As suggested earlier, this enables regulatory flexibility among signatories, both past and present. The module on business trust contains two features drawn from CPTPP, including the promise to stop unsolicited messages or ‘spam’ (6.2) and to provide online consumer protection (6.3). These commitments are binding, reflecting the fundamental importance of these issues as non-negotiable elements of a functioning digital economy. Parties promise to cooperate on Digital Identities via Article 7. Digital Identities has emerged as a somewhat controversial topic in recent years given the potential threat to civil liberties associated with ‘vaccine passports’ which led to political tensions in a number of countries during the Covid-19 pandemic. On the other hand, digital identities enable so-called verifiable credentials, enabling individuals to satisfy regulatory authentication and authorization for various purposes, eliminating many of the burdens associated with carrying sensitive personal identification in the form of documents or electronic copies thereof. They are seen by some as ‘convenient, secure, and privacy-oriented alternative to current physical and digital identity management systems’.7 This has important implications for the streamlining of identification for the purposes of digital trade, including payment authorization. Module 8 on emerging trends and technologies and Module 9 on innovation and the digital economy go further than any other trade agreement. While these sections also lack binding rules, they do create important nodes of collaboration in emerging areas where the needs of the private sector are bound to change. Module 8 sets out several topics that are poised to become even more important in terms of underpinning economic activity generally in the coming years, including fintech and AI. Signatories not only set out standards but commit to actively ‘promote development of fintech solutions’ and ‘promote cooperation between firms’ in the sector, including with start-ups, a business entity that is rarely referenced explicitly in trade agreements. On AI, a topic also typically not referenced in FTAs, even those with digital chapters,8 cooperation is also required using accepted principles of AI governance. This acknowledges that, with AI having become widespread, efforts to mitigate the potential risks arising from its use on a larger scale may have a bearing on international trade as well as economic activity generally.9 International alignment of AI governance frameworks is crucial in enabling the adoption of AI technologies across different jurisdictions.10 Yet, since Article 8.2 consists of best- efforts language, it is better described as a signalling tool.11 It sets out the intention of the parties in terms of a direction of regulatory travel, rather than applying rigid rules, which ultimately could end up stifling rather than stimulating regulation. There is also

7 J Sedlmeir, R Smethurst, A Rieger and G Fridgen, ‘Digital Identities and Verifiable Credentials’ 63 Business & Information Systems Engineering 603–13 (2021). 8 Exceptions include the UK–Australia FTA Chapter 20 (on Innovation). 9 See further S Chesterman, We the Robots (Cambridge University Press, 2021) Chapter 7. 10 Soprana above n 3 at 161. 11 Ibid.

94  Research handbook on digital trade a requirement for the parties to cooperate in relation to Innovation (Article 9) generally with a view to promoting creativity and cooperation. The Innovation module highlights areas that require ongoing cooperation in the future to foster greater innovation, such as the use of public domains and open government data. There is a commitment to data innovation through open government data and regulatory ‘sandboxes’ (something not typically found in FTAs), implying a willingness to entertain creative approaches to regulation, as first tested in low-risk environments before broader application. Such commitments to enable updates and modernization to the agreement as its membership grows and technology changes are precisely what makes DEPA a living agreement. The generality of this language reveals a broad sense of openness to ideas rather than a specific path which must be followed. Parties further pledge to cooperate in relation to SMEs. The inclusion of material on SMEs has become commonplace in recent FTAs, recognizing the importance of these firms in the global economy, especially where digitization tends to enable firms of smaller size to enter foreign markets in a manner that would have previously been impossible. This is a regulatory agenda that is increasingly attracting the attention of global trade governance.12 The Digital Inclusion material found in Art. 11 is highly innovative, addressing the needs of women and indigenous people. This commitment works towards fulfilling the UN’s Sustainable Development Goals.13 Disciplines on digital inclusion, which are also framed in hortatory language, may have also been designed to serve a signalling purpose, much as the preamble in a conventional FTA. Parties should aim to cooperate to remove barriers to the participation in the digital economy of women, rural populations, low socioeconomic groups and indigenous peoples; promote inclusive and sustainable economic growth; share experiences and best practices on digital inclusion; and develop programmes to promote participation of all groups in the digital economy. This aspirational, broadly framed material reflects DEPA parties’ concerns that the advantages of the digital economy need to be available to all.14 The requirement of access to technology, including internet connectivity, to derive benefits from digital trade underscores the reality that efforts will be needed to ensure that historically deprived groups are not marginalized. Turning to procedural matters, Art. 14 sets out the dispute settlement rules of DEPA. It is fairly well detailed, as is common in the dispute settlement chapters of most FTAs. As expected, there is an emphasis on mediation and consultation. Arbitration will be conducted by three person panels. Experts can be used, non-parties can be involved and there are strict time frames. Interestingly, on choice of forum, DEPA specifies that the WTO dispute settlement system can be used by the parties where obligations under the treaty overlap with those of the WTO, but once the parties choose a forum, they are committed to it.15 Like the WTO, the primary remedy for breach of DEPA is withdrawal of the unlawful measure rather than monetary damages, but there is an alternative remedy of suspension of other benefits under the treaty or compensation, meaning the enlargement of benefits in other areas. Many of the working elements of the dispute settlement mechanism remain to be developed or, if completed, are included in annex texts which have not yet been released (15.3).

See further D Collins, ‘Digital Trade and the WTO: Opportunities and Challenges for SMEs’ in J Chaisse ed. Elgar Companion to WTO (Edward Elgar Publishing, 2023). 13 In particular Goal #5 on Gender Equality. 14 Soprana, above n 3 at 162. 15 Art 14.7. 12

The Digital Economy Partnership Agreement (DEPA)  95 Also like the WTO’s GATT, DEPA contains general exceptions which are modelled on those of GATT’s Article XX. It is amazing that after many decades of FTA negotiation, as well as ongoing ambiguity in the understanding of WTO general exceptions, the parties of such an innovative agreement did not choose to spell out exceptions with greater clarity. Interestingly, however, DEPA includes the specific addition of ‘environmental’ measures, which is missing from the old GATT16 – a refreshing update of the old 1940s language on exhaustible natural resources. There is also a self-judging essential security exception, now standard in modern FTAs, despite the possibility of abuse associated with the use of essential security to justify economic treaty breaches. WTO practice now suggests that such provisions are indeed justiciable by arbitral tribunals.17 Finally, in terms of its institutional structure for the purposes of administration, as with many modern FTAs, DEPA establishes a Joint Committee to enable further dialogue among the parties, recognizing that it is a living agreement designed to adapt as new technologies and innovations emerge.18 These kinds of permanent bodies are essential to facilitate the renegotiation of the treaty text, which is inevitably incomplete, ensuring the longevity of the instrument.19

3.

A DIGITAL-ONLY APPROACH TO ECONOMIC INTEGRATION

Generally speaking, DEPA offers much promise in creating a framework for the digital economy in the four signatory states, with the potential for other countries to join. It is sufficiently flexible to allow members to adapt the rules to local conditions yet manages to achieve a level of certainty that will foster business opportunities and promote innovation. The long-term success of DEPA will require ongoing participation by firms which engage in digital trade in order to ensure that the regulations remain useful and relevant.20 Its format as a stand-alone agreement, covering digital matters only but omitting commitments in conventional areas such as tariff reductions and market access for services, is an untested approach in digital trade, if not international economic law generally. It remains to be seen whether ‘digital-only’ international agreements catch on as instruments of trade liberalization in this specialized area. The fact that digitization touches on so many economic issues indirectly, including for example trade facilitation in goods, suggests it may be uniquely suited to treatment in isolation. In approaching digital trade in any international agreement, as with domestic legislation, governments must strike the right balance between, on the one hand, realizing the vast economic value of data, which a lighter regulatory touch may better facilitate, and safeguarding national security and the data privacy of citizens. This can be seen for example in DEPA’s Module 4 on data issues which sets out that ‘each party shall adopt or maintain a legal frame-

Art 15. Russia – Measures Concerning Traffic in Transit, WT/DS512/7 (29 April 2019). 18 Art 12. 19 TJ Park, Incomplete International Investment Agreements (Edward Elgar Publishing, 2022). 20 D Elms, ‘Talking Trade: Unpacking the Digital Economic Partnership Agreement’ (28 Jan 2020) Asian Trade Centre. 16 17

96  Research handbook on digital trade work that provides for the protection of the personal information of the users of electronic commerce and digital trade’. This provision is followed by the caveat that each party may take different ‘legal approaches’ to protecting personal information and establishes a set of guiding principles for promoting the interoperability of standards. One of DEPA’s strengths is its potential to expand in scope. While new countries can accede to the Agreement, each ‘module’ of DEPA can easily be inserted into other trade agreements – hence the use of the term ‘module’, which implies ready re-application from one context to another. Countries doing business with DEPA signatories will likely be inclined to align their domestic policies with DEPA. For example, in DEPA’s Art. 2.7 on e-payments, the parties agree to foster ‘the adoption and use of internationally accepted standards, promoting interoperability and the interlinking of payment infrastructures, and encouraging useful innovation and competition in the payments ecosystem’. The UK’s signing of the digital-only Digital Economic Agreement with Singapore, which is substantially similar to DEPA, has compelled it to pursue domestic legislation in the area of electronic trade documents, which it had previously lacked.21 These standard-setting and innovative aspirations of DEPA underscore the objective of the Agreement, also acknowledging the need to achieve balance between regulating for certainty while maintaining sufficient flexibility for innovation. Many of its provisions are ‘soft law’ rather than legally binding rules and flexibility might in fact be an advantage in this fast-moving area22 – hard rules established under the agreement may well become outdated as new technologies emerge, requiring costly renegotiation. Establishing a stand-alone trade agreement on a single subject matter (digital trade) could offer some benefits with respect to the scope and depth of the final outcome of negotiations. FTAs generally involve multiple issues and industries at the same time, inevitably leading to diverging interests across the different policy areas. Depth and scope is often sacrificed as a result. Concentrating on a single topic may help remove any need to reduce the width and breadth of the agreement in order to reach a compromise on other negotiating areas. DEPA negotiators may have found it easier to delve in greater detail into how best to address all the issues relevant to the digital economy and promote digital trade than would have been possible in a conventional FTA covering multiple fields.23 The sensitivity of unrelated fields, such as agriculture, can often act as a stumbling block to progress in areas where agreement might otherwise be reached. Furthermore, a unique agreement such as DEPA enabled negotiators to focus on the specific needs of trade in the digital economy, extending the scope of negotiations beyond the traditional, and arguably outdated, goods and services dichotomy under the WTO framework and enabling the incorporation of WTO-plus issues (such as data localization) that may otherwise have been omitted. It was consequently able to generate new ideas, for example in relation to e-invoicing and AI, offering insight into subjects which are evolving at a rapid pace, typically by encouraging cooperation rather than setting proscriptive rules which could end up being restrictive. In the words of one commentator: ‘DEPA provides an understanding of the scope

Electronic Trade Documents Bill (HL Bill 57). S Honey, ‘Enabling Trust, Trade Flows and Innovation’, The DEPA at Work (21 July 2021) The Hinrich Foundation. 23 Soprana above n 3 at 166. 21 22

The Digital Economy Partnership Agreement (DEPA)  97 and depth that a stand-alone agreement on this topic could reach when all the parties involved are interested in achieving a high-level and meaningful outcome.’24 One of the drawbacks to a subject-specific agreement such as DEPA is the potential for overlap and inconsistency with existing FTAs. This is especially since the distinction between trade in services and digital trade can often be blurred. Many services are delivered digitally, often yielding the (arguably redundant) provision in FTAs that obligations contained in digital trade chapters incorporate obligations found in services chapters.25 The lack of concurrent negotiations on aspects of digital trade typically addressed in the trade in services chapters of FTAs could prove problematic since digital trade and trade in services, especially of the Mode 1 cross-border supply variety, are undoubtedly intertwined. This could be exacerbated in countries which have limited negotiating capacity, or which lack the ability to fully assess the impact that new disciplines on digital trade introduced with a single-topic agreement could have on commitments and obligations arising from other existing trade agreements they have signed. In many respects this is a modern recapitulation of the infamous ‘noodle bowl’ problem identified by commentators in the era of proliferation of regional agreements.26 Moreover, a stand-alone agreement like DEPA raises some questions about its legal interaction within the WTO framework. In particular, it is unclear whether this type of agreement satisfies the ‘substantial sectoral coverage’ requirement of Article V of the GATS as an exception to most favoured nation non-discrimination and therefore should be notified under that provision. DEPA would need to be assessed in terms of number of sectors, volume of trade affected and modes of supply to evaluate whether the agreement meets this requirement, and whether it eliminates substantially all discrimination particularly when DEPA parties do not engage in the negotiation of specific commitments.27 Furthermore, some of DEPA’s provisions may need to be inscribed by the parties in their GATS schedules as additional commitments under Article XVIII. This article covers a wide range of measures affecting trade in services that are not subject to scheduling obligations under market access and national treatment contained in Articles XVI and XVII of the GATS. Several of the disciplines negotiated under DEPA arguably fall under the scope of application of Article XVIII of the GATS.28 It is quite possible that, were a dispute to arise between the parties, some of these issues could surface, particularly if there were corresponding allegations of GATS commitments being breached and a WTO panel were to be appointed for the resolution of the matter.

4. ACCESSION DEPA was envisioned as an agreement to which other countries would be able to accede – expanding its reach well beyond the original four signatories. The authority to consider and approve the terms of accession, including the period for the deposit of an instrument of accession, vests in the Joint Committee, established under Article 12.1. This body consists of government representatives of each party. Aspiring members will be required to demonstrate

26 27 28 24 25

Ibid at 166. E.g. USMCA Art 19.2.4. J Bhagwati, Termites in the Trading System (Oxford University Press, 2008). Soprana above n 3 at 168. Soprana above n 3 at 167.

98  Research handbook on digital trade their efforts to date, and the changes that will need to be made to their domestic laws and regulations – if any – to meet the obligations under the agreement. DEPA disciplines have not yet been embedded into other trade arrangements, although the momentum for doing so might be more compelling as the agreement itself expands.29 4.1 Canada In December 2020, Canada notified DEPA parties of its interest in joining the Agreement. In February of 2022, Canada officially began exploratory discussions with those parties. One month later, Canada began public consultations with individuals and stakeholders on the current DEPA text and how DEPA could potentially be updated. On 22 May 2022, Canada submitted a formal request to launch negotiations for Canada’s accession to DEPA. In August 2022, the Vice Ministers of Commerce of Singapore, New Zealand and Chile decided to create the Accession Working Group (AWG) for the purposes of assessing Canada’s application. This group is in charge of evaluating compliance with the regulatory standards of the applicant country in relation to DEPA. The working group, which comprises government representatives from member countries, will also discuss Canada’s ability to comply with the standards and commitments of the pact, and submit a report to the DEPA Joint Committee on the proposed terms and conditions of Canada’s accession to the partnership. Singapore's Trade and Industry Minister stated that Singapore welcomes Canada’s application to join DEPA.30 Joining DEPA could have significant benefits for Canadian exporters as the digital economy becomes more vital for businesses. In 2020, retail e-commerce sales in Canada increased by 70.5 per cent compared to 2019, in part spurred by the Covid-19 pandemic.31 DEPA fits well with Canada’s international and domestic policy objectives. This includes encouraging e-commerce as a means of facilitating international trade and enabling trade diversification for Canadian businesses, including SMEs, in the near term. Canada has also been focused on inclusivity in its trade agenda, as seen for example in the CPTPP. Given that digital trade is evolving rapidly along with advances in technology as well as changes in consumer habits and business models, past FTAs are not sufficiently advanced to serve the needs of businesses operating in the digital economy. The CPTPP is already beginning to look outdated. There may also be value in Canada joining DEPA early so that it is positioned to contribute to the future course of the agreement. Many of DEPA’s provisions are similar to those undertaken by Canada in the CPTPP. This means Canada would, in DEPA, essentially agree to provisions already in force. There is an advantage to Canada’s joining DEPA at an earlier stage in order to avoid finding itself in the difficult position it faced when applying to enter the CPTPP late in the game, after many key issues had already been settled. Some believe that Canada should push for greater cooperation that goes beyond DEPA, perhaps in relation to matters such as privacy and cybersecurity. For

29 D Elms, ‘China Applies to Join DEPA’ Talking Trade, Asian Trade Centre https://​asiantradecentre​ .org/​talkingtrade/​china​-applies​-to​-join​-depa undated (accessed October 2022). 30 CY Ting, ‘Singapore, Chile, New Zealand to Assess Canada’s Application to Join Digital Trade Pact’, The Straits Times (25 August 2022). 31 Asia Pacific Foundation of Canada www​.asiapacific​.ca/​publication/​depa​-worlds​-first​-digital​-only​ -trade​-agreement (accessed October 2022).

The Digital Economy Partnership Agreement (DEPA)  99 the most part, substantive regulation of the digital economy and its interface with society is being conducted outside negotiations towards digital economy trade agreements.32 In some respects Canada is well suited to undertake the obligations contained in DEPA. DEPA’s requirement that parties ‘ensure that the implementation of measures related to e-invoicing in its jurisdiction is designed to support cross-border interoperability’ (Article 2.5.2) has already been implemented by Canada which introduced e-invoicing in 2017 adopting of Universal Business Language (UBL) as a national standard.33 While e-invoicing is allowed in Canada, it is not mandatory. If a Canadian business decides to issue invoices electronically, it is recommended to keep paper evidence of integrity and authenticity. Invoices must be stored for six years from the end of the fiscal year in which they were issued. It is also possible to store invoices abroad, provided that they can be accessed online.34 On AI, one of DEPA’s other highly innovative aspects, Canada is also already a world leader, having been at the forefront of advances in this area for decades, with many labs, government funding and global investment. It is currently ranked fourth in the world, among 62 countries, for research into AI and machine learning by the Global AI Index.35 Canada is also known for its highly progressive approach to trade agreements,36 incorporating recognition of indigenous and women’s rights in line with DEPA’s agenda for inclusivity. 4.2

Republic of Korea

In July 2020, the Ministry of Science and ICT of Korea announced the ‘Digital New Deal’ as part of the ‘Korean New Deal Comprehensive Plan’.37 Based on this national policy, the Korean government is focusing its efforts on digital innovation throughout the economy, such as accelerating digital transformation, 5G and AI within industries as a new springboard for economic development. Furthermore, recent pandemics have provided an opportunity for Korea to strengthen its digital capabilities and digital business models. With this background in mind, Korea is also actively participating in international discussions to establish new rules on digital trade. Korea’s efforts towards digital trade rule-setting go back to the FTA era: 13 out of its 18 FTAs concluded so far have a separate chapter dedicated to e-commerce.38 It is noteworthy that among them, the Korea–US FTA (KORUS FTA) contains a few advanced provisions such as ‘cross-border information flow’ under the e-commerce chapter and ‘cross-border transfer of information by financial institutions’ under the financial services chapter. Even though 32 R Fay and D Curiak, ‘Digital Economy Partnership Agreement Is a Big Step Forward: Canada Should Join’ CIGI Online (14 April 2022). 33 Adopting this standard also satisfies DEPA Art 2.5.2’s requirement that e-invoicing rules should be based on international standards. 34 ‘How Does Canada Implement E-Invoicing’, INPOSIA, www​.inposia​.com/​en/​canadas​-journey​ -towards​-e​-invoicing/​ (accessed October 2022). 35 Behind the US, China and the UK: The Global AI Index, www​.tortoisemedia​.com/​intelligence/​ global​-ai/​(accessed October 2022). 36 Seen for example in the Preamble of the USMCA. 37 Ministry of Science and ICT, press release, ‘Ministry of Science and ICT Rolls Out Digital New Deal to Leap Forward into a New Economy beyond COVID-19’, 14 June 2021, available at https://​digital​ .go​.kr/​front/​main/​eng​.do​#news. 38 Bo Min Ko, ‘Digital Trade Negotiation in Korea-China-Japan FTA: Minimax Strategy in the Process of Making Consensus’, 18(1) Journal of Global Business and Trade February 2022 at 50.

100  Research handbook on digital trade these are so-called ‘endeavours’ clauses rather than obligatory provisions, they recognize the ‘importance of the cross-border free flow of information’ and to refrain from imposing or maintaining unnecessary barriers to electronic information flows across borders.39 In addition, KORUS FTA underlines the significance of ‘cross-border transfer of data even in the finance sector’ by allowing foreign financial service providers to transfer information electronically.40 Among the agreements that have emerged since then, it is indispensable to pay attention to e-commerce rules of the Regional Comprehensive Economic Partnership (RCEP), which was signed in 2020.41 RCEP, going one step further than KORUS FTA, laid the groundwork for vitalizing digital trade by covering the principle of cross-border free flow of data and prohibition of localization of computing facilities. Even though new governance on cross-border data flows is welcomed, these provisions are flawed and vulnerable in some ways. RCEP leaves room for policy adjustment to impose any restrictions as long as they are applied in a non-discriminatory manner, since it explicitly stipulates that implementing party holds the authority to determine ‘necessity’ with a view to safeguarding its own legitimate interests.42 Moreover, it has no remedy under the dispute settlement system of RCEP provided that the restrictive measures are taken for the essential security interests.43 However, Article 14.14 (Cross-Border Transfer of Information by Electronic Means) under the Korea–Singapore Digital Partnership Agreement of 2021 carves out the aforementioned drawbacks and adopts CPTPP language.44 Recently, Korea has actively participated in WTO joint statement initiatives (JSI) at the multilateral level. Tremendous efforts have been consistently made by Korea to establish rules on digital trade at the bilateral level. First, Korea has tried to set up a new digital partnership agreement by upgrading the e-commerce chapter of the existing FTAs. For instance, the Korea–Singapore Digital Partnership Agreement (KSDPA) paves the way for a stand-alone type of digital treaty closely linked to current FTA regime.45 KSDPA is the first digital trade agreement with full-fledged and top-standard digital trade rules concluded by Korea. Second, Korea announced it would join the Indo-Pacific Economic Framework (IPEF), whose first round of negotiations was to be held in December 2022. Rule-making on digital trade under IPEF must be mainly centred on Pillar I (Trade).46 Finally, it is pursuing the conclusion of a separate and independent digital agreement like DEPA.

KORUS FTA Article 15.8. BR Williams et al., ‘The U.S.–South Korea Free Trade Agreement (KORUS FTA): Provisions and Implementation’, CRS Report, U.S. Congress, 16 Sep 2014 at 35, https://​sgp​.fas​.org/​crs/​row/​RL34330​ .pdf (accessed October 2022). This is stipulated in paragraph 6(b) of 13-A and Section B of Annex 13-B of Chapter 13(Financial Services) of KORUS FTA. 41 RCEP came into effect on 1 January 2022. 42 RCEP Article 12.5 and its footnote 14. 43 RCEP Articles 12.4 and 12.5. 44 CPTPP Article 14.11 (Cross-Border Transfer of Information by Electronic Means). 45 Ministry of Trade, Industry and Energy, press release, ‘Korea and Singapore Announce Conclusion of Korea’s First Digital Trade Agreement, Korea–Singapore Digital Partnership Agreement (DPA)’, 15 December 2021, https://​english​.motie​.go​.kr/​en/​pc/​pressreleases/​bbs/​bbsView​.do​?bbs​_seq​_n​ =​898​&​bbs​_cd​_n​=​2 (accessed October 2022). 46 Ministerial Text for Trade Pillar of the Indo-Pacific Economic Framework for Prosperity, Pillar I (trade), https://​ustr​.gov/​sites/​default/​files/​2022​-09/​IPEF​%20Pillar​%201​%20Ministerial​%20Text​ %20(Trade​%20Pillar)​_FOR​%20PUBLIC​%20RELEASE​%20(1)​.pdf (accessed October 2022). Among the three pillars, Pillar 1 (Trade) focuses on promoting not only eco-friendly and low-carbon trade and 39 40

The Digital Economy Partnership Agreement (DEPA)  101 The Korean government has already completed the domestic procedures to initiate negotiations on Korea’s accession to DEPA, including a feasibility study, public hearing and report to the National Assembly. In September 2021, one month before Chinese notification of accession,47 the Korean government officially notified New Zealand, the Depositary of DEPA, of Korea’s intent to join and the Accession Working Group was established. A first round of negotiations was launched in January 2022. Considering the continued expansion of DEPA and its position as a platform for building a global digital cooperation network, negotiations will be expected to be fruitful. Against this backdrop, a thorny issue in the DEPA text should be mentioned. Even though DEPA contains essential elements of obligation clauses such as non-discriminatory treatment of digital products, information and communication technology products using cryptography, cross-border transfer of information electronically and prohibition of location of computing facilities, its Annex I could severely undermine these provisions.48 It is necessary to see what this could mean and how it should be discussed in the future.

5. CONCLUSION DEPA is an important new economic integration agreement which reflects the growing importance of digital trade to the global economy. Its unique digital-only content sets it apart from many of the modern FTAs from which it draws inspiration, including the CPTPP, but DEPA goes further, delving into innovative fields such as AI and breaking new ground in matters such as electronic trade facilitation. While many of DEPA’s obligations are framed in non-binding, best-efforts language, it places obligations on parties to engage in cooperation in areas vital to the expansion of the digital economy, signalling each signatory’s willingness to adapt their own approaches to new ways of doing business mindful of the need to balance this against social costs, for example in relation to privacy. DEPA’s status as a living agreement open to modification in line with changes in technology is further reflected in its capacity to encompass new signatories. Canada and Korea are among the first countries to pursue accession to stimulate their respective economies in a manner that is responsive to the needs of their citizens.

investment, agricultural technology innovation and food security, digitization of customs procedures, but also ‘digital trade’. IPEF plans to establish high-level rules and discuss cooperation agendas for vitalization of digital trade. 47 Ministry of Commerce of the People’s Republic of China, press release, ‘中方正式提出申请加入 《数字经济伙伴关系协定》（DEPA）’, available at 中方正式提出申请加入《数字经济伙伴关系 协定》（DEPA） (mofcom.gov.cn). 48 Annex I – UNDERSTANDING ON THIS AGREEMENT […] For greater certainty, the Parties record their understanding that the following Articles do not create any rights or obligations between or among the Parties under this Agreement: (a) Article 3.3: Non-Discriminatory Treatment of Digital Products; (b) Article 3.4: Information and Communication Technology Products that Use Cryptography; (c) Article 4.3: Cross-Border Transfer of Information by Electronic Means; and (d) Article 4.4: Location of Computing Facilities.

7. Placing gender equality at the centre of global digital trade policy Michael Geist1

1. INTRODUCTION Digital trade offers both benefits and disadvantages to the development of gender equality. From a benefits perspective, it notably offers opportunities for women, including lower barriers to entry for women-run SMEs and more flexible working arrangements that allow for better balance between work and home obligations. Increased choice and competition may allow for greater access to new products and better pricing. Yet as consumers, business owners and employees, women face challenges when trying to operate and succeed in the digital trade space. While infrastructure, and many logistical, challenges are universal to all, women still face barriers that restrict them from reaching their full potential. These barriers include financing, government procurement, logistics and supply chains, information technology infrastructure, online payments, workplace protections, consumer protection, competition, online harassment and risks associated with artificial intelligence. Notwithstanding the many concerns associated with digital trade and gender, gender-equality concerns in digital trade or e-commerce trade agreements remain a relatively novel concept. While many trade agreements reference gender equity or equality, the trade commitments are typically modest and rarely address the many challenges identified in this chapter. Indeed, the inclusion of digital trade or e-commerce in free trade agreements or at the World Trade Organization (WTO) primarily focuses on trade facilitation without much regard for the broader policy and societal impacts of digital trade and e-commerce. This chapter examines both the benefits and challenges faced by women in digital trade and the role those issues have played in trade agreements and work at the WTO. It identifies several models that could be emulated within agreements and points to potential policy initiatives that could better address the role of women within digital trade.

2.

DIGITAL TRADE AND GENDER: THE OPPORTUNITIES

Digital trade offers considerable promise for the future of global commerce with notable opportunities for women. For example, digital trade allows for a new level of flexibility for women-run SMEs. Throughout the pandemic, many women faced enormous challenges

An earlier version of this paper was commissioned by the International Trade Centre. My thanks to the ITC for its support and commitment to identifying gender as a policy priority within e-commerce law and policy development, to Micheala De Groot and Garry Balaganthan for their research assistance, and to the Social Sciences and Humanities Research Council of Canada and Canada Research Chair program for their support. 1

102

Placing gender equality at the centre of global digital trade policy  103 involving providing primary care in the home and other domestic responsibilities that rendered non-flexible workplaces all but inaccessible.2 Through digital trade, women may be able to run their businesses with flexible time commitments, hours and locations.3 As women shift towards digital trade, they are reportedly more open than men to support offered through training, programmes, financial advice and support, as well as help from online platforms.4 Digital trade has enabled women to enter more lucrative areas such as electronics and sales of appliances and automotive merchandise with higher success rates than in regular commerce.5 Furthermore, the opportunities are not limited to entrepreneurial initiatives. Women-related employment in areas such as software coding or data processing may benefit from digital trade. Similarly, digital trade has key implications and opportunities for women as consumers, including the prospect of greater choice, better pricing and electronic-based services such as banking that may be more convenient for women facing many demands on their time. Digitization may enable women to gain a comparative advantage by relying on the social skills that they are more likely than men to possess.6 Indeed, research has found that women’s social strengths give them advantages for interacting with others online, particularly in consumer-specific product fields where there is greater customization and differentiation.7 For example, research by e-commerce giant Alibaba found that women-led businesses receive higher ratings on their platform for customer service and better reviews on communicating accurately about their goods.8 Digital technologies require communication and collaboration skills, and women tend to excel and take on more digital work rather than rigid physical work, aligning with the demanded skill set of the digital technologies.9 More women are choosing digital work because of its flexibility and alignment with their skills. Further, women may gravitate towards digital trade and e-commerce positions given the potential for greater workplace flexibility that offers the promise of better balance of their own economic situation and their caregiver responsibilities.10 Women tend to face more time constraints in their lives with unpaid domestic labour and caregiver expectations, so digital trade offers some women a flexible work environment where they can still financially support themselves and their families.11

2 International Finance Corporation, (2021) Women and E-commerce in Southeast Asia. See www​ .ifc​.org/​wps/​wcm/​connect/​topics​_ext​_content/​ifc​_external​_corporate​_site/​gender+​at+​ifc/​resources/​ women​-and​-ecommerce​-sea at 22 [IFC Asia]. 3 Ibid 22. 4 Ibid 33.3. 5 International Finance Corporation (2021). Women and E-commerce in Africa. See www​.ifc​ .org/​wps/​wcm/​connect/​47361305​-6ebe​-431a​-8dd9​-db2290919823/​202105​-digital2equal​-women​-and​-e​ -commerce​-africa​.pdf​?MOD​=​AJPERES​&​CVID​=​nCGRGTr 3, 27–9 [IFC Africa]. 6 World Bank Group and WTO (2020). Women and Trade: The Role of Trade in Promoting Gender Equality 142, www​.wto​.org/​english/​res​_e/​booksp​_e/​women​_trade​_pub2807​_e​.pdf [World Bank Group and WTO]. 7 Ibid. 8 Ibid. 9 Ibid. 10 Ibid at 143 and IFC Africa, at 22. 11 World Bank Group and WTO at 141–4.

104  Research handbook on digital trade

3.

DIGITAL TRADE AND GENDER: THE BARRIERS

As consumers, business owners and employees, women face numerous barriers when trying to operate and succeed in the digital trade space. While infrastructure and many logistical challenges are universal to all, certain obstacles make it especially challenging for women to reach their full potential. 3.1 Financing Financing is a challenge for women-owned SMEs. Businesswomen are less likely than men to finance their businesses through a bank loan. Instead, they tend to rely on personal savings or funding from friends and family. For those who turn to financial intermediaries, gender-based barriers remain. Women are not only less likely than men to have a bank account, but they are also less likely to be permitted to have a bank account or financial independence.12 This creates a gender-specific barrier to create and scale a business. Few financial intermediaries, particularly in developing countries, have internal systems to target lending to women-owned SMEs.13 As a result, data is limited on whether loans are being directed to these firms.14 Where data is collected, it points to significant barriers. For example, during the COVID-19 pandemic women-owned/led businesses were twice as likely to have their loan applications rejected than men-led firms.15 Loans to women-owned SMEs also tend to be smaller than those granted to men-led companies.16 Programmes to assist small women-owned firms include the Women Entrepreneurs Finance Initiative (We-Fi) and IFC ScaleX. The former aims to increase access to finance through mixed financial instruments, developing digital financial solutions for women-owned SMEs to access online loans, working with banks to adjust internal processes to better serve women and leading gender-sensitivity training for loan officers to decrease gender discrimination.17 Yet room remains to improve businesswomen’s access to finance, with an ongoing need to emphasize financial literacy and clear opportunities and to create internal infrastructure in banks and with other investors for working with women-owned SMEs.18 3.2

Government Procurement

Women-owned businesses typically enjoy less access to and success in public procurement projects.19 Public procurement can produce up to 40 per cent of the gross domestic product of

IFC Africa at 19 and Yasmin Ismail and Hiral Hirani, Addressing the Gender Dimension of E-Commerce: Towards a Holistic Analytical and Policy Framework, CUTS International, Geneva, 2021 at 12 [Ismail and Hirani]. 13 We-Fi (2021). Pivoting toward a Resilient Future: Annual Report 2021, 33. https://​we​-fi​.org/​wp​ -content/​uploads/​2022/​03/​We​-fi​-2021​-Annual​-Report​.pdf [We-Fi]. 14 Ibid. 15 Ibid at 15. 16 Ibid. 17 Ibid at 30–5. 18 Ibid. 19 Ibid. 12

Placing gender equality at the centre of global digital trade policy  105 some developing countries, yet some data suggest that women-owned SMEs supply just 1 per cent of these projects.20 This points to barriers including awareness of government procurement opportunities and limited skill and experience in navigating the government procurement process.21 In fact, the emergence of mandated electronic procurement for government services is likely to widen the digital divide, adding to the challenges women-owned businesses face. As noted in a World Bank blog, There are opportunities for countries, particularly those in initial stages of digital development, to ‘leapfrog’ through learning from the lessons of other countries’ e-procurement experiences and take advantage of the latest technology that exists. However, basic building blocks have to be in place first, including legal, institutional, capacity building and related arrangements.22

The emphasis on basic building blocks is well placed and speaks to the potential digital trade barriers. In some cases, the issue of public procurement can be addressed through specific targets. For example, the Indian government has established rules that target 25 per cent of procurement awards to micro and small enterprises, including at least 3 per cent owned by women entrepreneurs.23 3.3

IT Infrastructure

Limited internet connectivity and poor information technology infrastructure are major barriers in developing and least developed countries.24 Women-owned SMEs reported that gaps in connectivity resulted in reduced logistical capacity, online payment constraints and difficulty engaging with consumers.25 Women also tend to have less access to internet-connected devices to operate online companies.26 Reliable internet access remains a challenge across areas of Africa and Southeast Asia.27 Without reliable internet access, women-owned/led SMEs struggle to maintain or start digital trade businesses and work in digital trade jobs. 3.4

Workplace Protections

While the entrepreneurial opportunities for women in digital trade are frequently emphasized, the reality is that many gravitate to digital trade for work opportunities. The ability to work from home with flexible hours is extremely important to women in both developing and devel-

Ibid. Ibid. 22 Y. Taylor and N. Langburd Wright (July 2018). The Future of Public Procurement in the Era of Digitalization. World Bank Blogs, https://​blogs​.worldbank​.org/​governance/​future​-public​-procurement​ -era​-digitalization. 23 https://​pib​.gov​.in/​Pressreleaseshare​.aspx​?PRID​=​1580277. 24 International Labour Organization at 140. 25 Ibid 141 and IFC Africa at 14. 26 United Nations (2021). COVID-19 and E-Commerce: A Global Review. https://​unctad​.org/​system/​ files/​official​-document/​dtlstict2020d13​_en​_0​.pdf [United Nations]. 27 IFC Africa, at 48–9 and World Bank Group and WTO (2020), at 14. 20 21

106  Research handbook on digital trade oped countries.28 These jobs may offer the benefits of flexibility, but suffer from drawbacks such as limited workplace protections, no pensions or equal pay assurances, and few labour safeguards.29 Indeed, workers for digital trade firms are often misclassified as self-employed when they are employees.30 This prevents them from qualifying for social security benefits and the rights that come with employment.31 They also do not enjoy union or collective negotiation rights, and competition laws in some regions prevent self-employed workers from collective bargaining.32 Gaining this bargaining and negotiation power could advance rights, wages and benefits for digital trade workers who interact with platforms in a manner where work hours are frequently determined by the workers themselves (known as gig workers).33 In some instances, there may be a risk that these workers will not be paid their full fees. This can also amount to being penalized for not taking on certain tasks or creating flexibility in their work days.34 This can happen to all platform workers, but women are already often paid less than men for the same work.35 As discussed below, women working in digital trade and e-commerce also face a larger risk of discrimination and harassment, including risk of physical assault when working delivery or taxi digital platform jobs.36 3.5

Consumer Protections

Limited consumer protections create risks for women as consumers. In particular, women who are in difficult financial situations are generally more likely to be vulnerable compared to other consumers. Consumer vulnerability is most frequent when consumers are faced with complex marketing and are unable to make the best choices.37 Consumer protections are frequently missing in countries – particularly in developing economies. For example, some estimates find that only 6 per cent of African countries have adequate consumer protection laws.38 For their citizens, their businesses and their economy to benefit from digital trade, countries must enact laws that protect consumers who buy goods or services online. Such laws should confer protection comparable to that enjoyed by consumers in offline

28 International Labour Organization (2021). World Employment and Social Outlook: The Role of Digital Labour Platforms in Transforming the World of Work, 22 [International Labour Organization]. 29 Anita Gurumurthy and Nandini Chami, From Ill-Founded Delusions to Real Possibilities: An E-Commerce Agenda for Women’s Empowerment, 2019 https://​itforchange​.net/​sites/​default/​files/​1786/​ Feminist​%20Digital​%20Justice​%20Issue​%20Paper​%202​_​%20updated​%20name​%20and​%20logo​.pdf. 30 Janine Berg, Marianne Furrer, Ellie Harmon, Uma Rani and M. Six Silberman (2018). Digital Labour Platforms and the Future of Work: Towards Decent Work in the Online World, 105. International Labour Organization, Geneva. www​.ilo​.org/​wcmsp5/​groups/​public/​-​-​-dgreports/​-​-​-dcomm/​-​-​-publ/​ documents/​publication/​wcms​_645337​.pdf [Berg, Furrer]. 31 International Labour Organization at 23. 32 Berg, Furrer, at 105–6. 33 Ibid. 34 Ibid 106. 35 International Labour Organization at 163–5. 36 Ibid at 171. 37 European Commission (2016). Understanding Consumer Vulnerability in the EU. See https://​ec​ .europa​.eu/​info/​sites/​default/​files/​consumer​-vulnerability​-factsheet​_en​.pdf. 38 IFC Africa, at 16.

Placing gender equality at the centre of global digital trade policy  107 transactions, with special consideration given to the unique circumstances of digital trade. Moreover, these laws must be backed by effective enforcement mechanisms. 3.6 Competition The power imbalance between women-owned/led SMEs and large digital trade platforms may also represent a major barrier. The former rarely have input into how digital trade platforms are designed or operate, yet have few alternatives but to work with these platforms. Competition laws are a critical tool to ensure a fair marketplace that does not disadvantage SMEs, including those owned by women. Yet concerns regarding fair competition even extend to how those rules are enforced, notably in relation to gender bias in the enforcement (and on the part of the enforcers) of competition rules.39 The competition issues associated with large digital trade platforms extend beyond these gender concerns, however. In some countries, large platforms are viewed as the internet, with many users rarely venturing outside the platform that offers communication services, e-commerce and social media interaction. This creates power imbalance risks and suggests that competition authorities must more aggressively ensure that antitrust and competition laws are effectively enforced to prevent marketplace abuse, self-dealing and unfair pricing. 3.7

Online Harassment

The COVID-19 pandemic increased women’s vulnerability to gender-based violence both at home and online.40 Online harassment and violence against women has grown in recent years, including through digital trade platforms.41 Women and girls are also more likely to be targets of bullying, cyberstalking and harassment.42 Digital trade platforms have often been slow to implement policies to curb gender-based harassment and violence.43 With digital privacy threats having real-world implications, the platforms’ privacy approach may put privacy at risk for users, while facilitating the ability for bad actors to monitor, harass and pursue others. A 2017 study by Hyperwallet found that 33 per cent of women work online under a pseudonym, or have a username that does not reveal their gender, in an attempt to avoid discrimination. Among them, 72 per cent choose to work under a gender-neutral name to maintain anonymity, 14 per cent do so to increase bids on online selling platforms and the remaining 14 per cent explicitly do so to avoid sexism or hostility.44

Alessandra Tonazzi (24 September 2020). Gender Inclusive Competition Policy. See www​.oecd​ .org/​daf/​competition/​gender​-inclusive​-competition​-policy​-other​-proposals​.pdf. 40 IFC Africa, at 18; United Nations, and IFC Asia at 19. 41 International Labour Organization, at 165, 171; United Nations, and IFC Asia at 19. 42 Ismail and Hirani, at 15. 43 Ibid. 44 Hyperwallet (2017). The Future of Gig Work Is Female: A Study on the Behaviors and Career Aspirations of Women in the Gig Economy. See www​.hyperwallet​.com/​app/​uploads/​HW​_The​_Future​_of​ _Gig​_Work​_is​_Female​.pdf. 39

108  Research handbook on digital trade 3.8

Artificial Intelligence Barriers

As discussed in several chapters in this handbook, artificial intelligence and machine learning have emerged as critical elements of digital trade as platforms use the technology for a wide range of decision-making processes. The implications for gender bias within these systems can create serious barriers for women in digital trade. Data scientists have pointed to at least two ways in which artificial intelligence perpetuates gender bias. First, algorithm and design bias or flaws may lead to unfair decisions made towards people of a certain gender, ethnicity, religion or geographic location.45 Second, reproduction of gender stereotypes in new digital products may occur due to artificial intelligence.46 Despite its outward gender-neutral presence, technology is frequently gendered, with users affecting daily outputs and creating a framework of understanding around women despite being largely underrepresented in the design process. Several tools may help address artificial intelligence/algorithmic gender bias. For example, multi-accuracy auditing improves the transparency of artificial intelligence systems by quantifying how their performance varies across race, age, sex and intersections of these attributes.47 Word-embedding association tests, which measure the impact of removing part of the training data, can be used to test the fairness of artificial intelligence systems.48 White box automatic machine learning models can help address the transparency and accountability of artificial intelligence systems.49

4.

DIGITAL TRADE AND GENDER: TRADE PROVISIONS

Notwithstanding the many concerns associated with digital trade and gender, gender-equality concerns or principles in digital trade or e-commerce, chapters in trade agreements remain a relatively novel concept that continues to challenge the notion that digital trade is a gender-neutral platform. While many trade agreements reference gender equality or gender equity, the trade commitments are typically modest and rarely address the numerous challenges identified in this chapter. Indeed, critics fear that the inclusion of digital trade or e-commerce in the agreements or at the WTO primarily focuses on trade facilitation, without due regard for their broader policy and societal impacts.

45 Alina Kochling and Marius Claus Wehner (2020). Discriminated by an Algorithm: A Systematic Review of Discrimination and Fairness by Algorithmic Decision-Making in the Context of HR Recruitment and HR Development 13 Business Research 795–848. 46 Ibid. 47 United Nations Educational, Scientific and Cultural Organization (2020). Artificial Intelligence and Gender Equality: Key Findings of UNESCO’s Global Dialogue 25. See https://​unesdoc​.unesco​.org/​ ark:/​48223/​pf0000374174. 48 Ibid at 36. 49 Kishore Sugali, Chris Sprunger and Venkata N. Inukollu (2021). Software Testing: Issues and Challenges of Artificial Intelligence & Machine Learning 103. 12:1 IJAIA 101.

Placing gender equality at the centre of global digital trade policy  109 4.1

WTO E-Commerce Agenda

Digital trade is not a new issue at the WTO, which typically considers the issue within the context of e-commerce. In fact, the WTO first officially recognized the growth of global e-commerce and the new opportunities it created for trade in the Declaration on Global Electronic Commerce adopted at the Second Ministerial Conference held in Geneva in May 1998.50 This declaration called for a Work Programme on Electronic Commerce, which was established four months later and covered issues related to trade arising from global e-commerce. It also affirmed that ‘[w]ithout prejudice to the outcome of the work programme or the rights and obligations of Members under the WTO Agreements, we also declare that Members will continue their current practice of not imposing customs duties on electronic transmissions’.51 This moratorium does not include goods, meaning that offering and selling products on the internet is subject to tariffs and other General Agreement on Tariffs and Trade obligations in force. The e-commerce customs moratorium was designed to stop countries from applying tariffs or duties on electronic transmissions across borders. The approach ensured that merely transmitting information across borders would not be subject to additional customs duties that might slow or block electronic transmissions. However, the moratorium does not prevent a country from imposing internal taxes, fees or other charges on electronic transmissions, including content transmitted electronically, provided that those taxes, fees or charges are imposed in a manner consistent with the WTO Agreements. In other words, electronic transmissions, including content transmitted therein, are exempt from customs duties, but internal taxes can be applied such that online sale of physical goods may still be subject to taxes, tariffs or other fees. After years of little change with respect to e-commerce, the chair of the WTO General Council convened an informal open-ended meeting in November 2017 to consider a request by India and South Africa for a fresh examination of the moratorium’s revenue impact. Both countries expressed concern that the growing size of e-commerce transactions might call for a re-evaluation of the long-standing customs moratorium. The two countries suggested that its continuation, if any, should be based on concrete facts and statistics. Several delegations saw merit in further studying the scope of the moratorium and factoring in other issues, including implications beyond revenues, such as the developmental dimension. The review session of the Work Programme on Electronic Commerce held in December 2018 resulted in the continued discussion on the moratorium on the WTO Members’ custom duties for electronic transmissions, with the expectation that a decision on the issue would be made by the end of 2019. That deadline was extended in December 2019.52 While the initial plan was

50 WTO General Council (September 1998). Work Programme on Electronic Commerce, Adoption, WT/L/274. WTO. docs​.wto​.org/​dol2fe/​Pages/​SS/​directdoc​.aspx​?filename​=​q:/​WT/​L/​274​.pdf​&​Open​=​ True. 51 WTO General Council, Declaration on Global Electronic Commerce (May 1998). WTO Doc WT/ MIN(98)/DEC/2. WTO. www​.wto​.org/​english/​thewto​_e/​minist​_e/​min98​_e/​ecom​_e​.htm. 52 WTO, Work Programme on Electronic Commerce, Decision (10 December 2019), WTO Doc WT/L/1079. https://​docs​.wto​.org/​dol2fe/​Pages/​SS/​directdoc​.aspx​?filename​=​q:/​WT/​L/​1079​.pdf​&​Open​ =​True.

110  Research handbook on digital trade to address the issue at the 12th WTO Ministerial Conference in Kazakhstan in June 2020, the meeting was postponed in March of that year due to the COVID-19 pandemic.53 Another important milestone in the WTO e-commerce work occurred in January 2019, when 49 Members alongside the European Union issued the Joint Statement on Electronic Commerce.54 The document noted the intention of Members to negotiate trade-related aspects of e-commerce, while recognizing the challenges faced by developing and least developed Members, as well as SMEs. Members held seven rounds of negotiations in 2019, categorized under six themes: enabling e-commerce; openness and e-commerce; trust and e-commerce; cross-cutting issues; telecommunications; and market access. Each round, referred to as a ‘focus group’, dealt with a particular set of issues, which included cybersecurity, capacity building, paperless trading, open government data, open internet access, business trust, telecommunications, the e-commerce moratorium, open data flows, consumer protection, data and privacy protection and market access. In December 2021, the envoys to the WTO of Australia, Japan and Singapore issued a joint statement stating that significant progress was being made in digital regulation among the 86 Members55 taking part in the E-Commerce Joint Statement Initiative (JSI).56 Reflecting the ambitions and concerns of both developed and developing nations, a global framework would represent a major development in e-commerce regulation and an important moment to include gender-related issues within global e-commerce legal frameworks. While the draft text has not been released publicly, there have been some well-publicized leaks of earlier drafts.57 These leaks suggest that the JSI is broadly following the structure found in the digital trade and e-commerce chapters of bilateral and regional trade agreements, discussed further below and in several chapters in this handbook. Work is ongoing, with plans for continued discussion in 2023. Reports from the talks suggest that convergence has been achieved in several areas, including online consumer protection, electronic signatures and authentication, unsolicited commercial electronic messages,

53 WTO (26 November 2021). ‘General Council Decides to Postpone MC12 Indefinitely’. See www​ .wto​.org/​english/​news​_e/​news21​_e/​mc12​_26nov21​_e​.htm. 54 WTO (25 January 2019). Joint Statement on Electronic Commerce WTO Doc WT/L/1056. See https://​docs​.wto​.org/​dol2fe/​Pages/​SS/​directdoc​.aspx​?filename​=​q:/​WT/​L/​1056​.pdf​&​Open​=​True. 55 The 86 Members participating in the JSI: Albania; Argentina; Australia; Austria; Bahrain; Belgium; Benin; Brazil; Brunei Darussalam; Bulgaria; Burkina Faso; Cameroon; Canada; Chile; China; Chinese Taipei; Colombia; Costa Rica; Côte d’Ivoire; Croatia; Cyprus; Czechia; Denmark; Ecuador; El Salvador; Estonia; Finland; France; Georgia; Germany; Greece; Guatemala; Honduras; Hong Kong, China; Hungary; Iceland; Indonesia; Ireland; Israel; Italy; Japan; Kazakhstan; Kenya; Korea, Republic of; Kuwait; Lao People’s Democratic Republic; Latvia; Liechtenstein; Lithuania; Luxembourg; Malaysia; Malta; Mexico; Moldova; Mongolia; Montenegro; Myanmar; Netherlands; New Zealand; Nicaragua; Nigeria; North Macedonia; Norway; Panama; Paraguay; Peru; Philippines; Poland; Portugal; Qatar; Romania; Russian Federation; Saudi Arabia; Singapore; Slovak Republic; Slovenia; Spain; Sweden; Switzerland; Penghu, Kinmen and Matsu; Thailand; Türkiye; Ukraine; United Arab Emirates; United Kingdom; United States; Uruguay. WTO (28 September 2022). Joint Initiative on E-Commerce. www​.wto​.org/​english/​tratop​_e/​ecom​_e/​joint​_statement​_e​.htm. 56 WTO (December 2021). Joint Statement Initiative on E-commerce Statement by Ministers of Australia, Japan, and Singapore. See www​.wto​.org/​english/​news​_e/​news21​_e/​ji​_ecom​_minister​ _statement​_e​.pdf. 57 See, for example, www​.bilaterals​.org/​?wto​-plurilateral​-ecommerce​-draft.

Placing gender equality at the centre of global digital trade policy  111 open government data, electronic contracts, transparency and paperless trading. Issues that remain subject to discussion include the e-commerce moratorium, data flows and privacy.58 The e-commerce gender issues discussed in this chapter have been largely absent from proposals put forward by WTO Members, however. One notable exception is a 2019 concept paper submitted by the Government of Canada on preventing the use of personal information for the discrimination or persecution of natural persons.59 The paper proposes including the following provision in the data protection and privacy section of the JSI: ‘No Party/Member shall use the personal information of users of digital trade to persecute or discriminate against a natural person on the basis of race, colour, sex, sexual orientation, gender, language, religion, political or other opinion, national or social origin, property, birth or other status, or disability [emphasis added].’60 The absence of substantive gender commitments in the JSI may reflect the fact that advanced e-commerce work at the WTO remains a relatively new phenomenon. Delegations may only be learning about e-commerce more broadly and have yet to fully incorporate gender-based analysis in their policy development. Given the current stage of negotiations, there is still time to enhance the role of gender within the work programme. Alternatively, negotiators may consider that the rules they are discussing will provide notable benefits to microenterprises, SMEs and women. The issue seems most salient for member states that have already grappled with the issues within the context of regional or bilateral trade agreements. 4.2

E-Commerce and Digital Trade within Regional Trade Agreements

With work continuing on e-commerce regulations in the WTO, countries often rely on free trade agreements to address digital trade issues. In developed nations such as Canada, these agreements have sought to establish gender as a priority. For example, the Canada–Chile Free Trade Agreement, which entered into force in February 2019,61 says ‘the Parties acknowledge that international trade and investment are engines of economic growth, and that improving women’s access to opportunities and removing barriers in their countries enhances their participation in national and international economies, and contributes to sustainable economic development’.62 A review of global free trade agreements identifies at least five with explicit chapters referencing gender.63 None requires a change in domestic regulations. Rather, most refer to 58 WTO (13 June 2022). Co-Convenors Welcome Good Progress in E-Commerce Talks, Launch Capacity Building Framework. www​.wto​.org/​english/​news​_e/​news22​_e/​jsec​_13jun22​_e​.htm. 59 Government of Canada (4 September 2019). Concept Paper: Preventing the Use of Personal Information from being used for the Discrimination or Persecution of Natural Persons. See www​ .international​.gc​.ca/​trade​-agreements​-accords​-commerciaux/​topics​-domaines/​other​-autre/​statement​ -concept​-ecom​-declaration​-reflexion​-09​.aspx​?lang​=​eng. 60 Ibid. 61 Government of Canada (5 February 2019). Canada–Chile Free Trade Agreement. See www​ .international​.gc​.ca/​trade​-commerce/​trade​-agreements​-accords​-commerciaux/​agr​-acc/​chile​-chili/​fta​-ale/​ index​.aspx​?lang​=​en​&​_ga​=​2​.232675101​.321642341​.1660269162​-1603950834​.1658169228. 62 Ibid. 63 Bomin Ko (October 2020). Analysis of Gender Chapter in Five Free Trade Agreements and Its Lessons for Korea 89. Journal of Korea Trade. www​.researchgate​.net/​publication/​346757054​_Analysis​ _of​_Gender​_Chapters​_in​_Five​_Free​_Trade​_Agreements​_and​_Its​_Lessons​_for​_Korea [Ko].

112  Research handbook on digital trade cooperation changes, without requirements to enact substantive reforms. For example, the four agreements where Chile is a signatory – Chile–Uruguay, Chile–Argentina, Chile–Canada and Chile–Brazil) contain near-identical provisions focused on skills enhancement, financial inclusion, access to technology and leadership. Yet there are no specific targets, budgets or objectives that can be evaluated. Similar limitations can be found in multilateral accords. While agreements such as the Comprehensive and Progressive Agreement for Trans-Pacific Partnership explicitly identify gender equality as a priority, no targets or standards are in place. The remaining 75 free trade agreement that refer to gender do so in a passive, non-binding manner.64 Generally referring to the importance of greater inclusion and equal treatment of women, the provisions are merely a recognition of gender with limited impact. Despite limited enforceability or specific regulatory commitments, these agreements reaffirm support for the United Nations Sustainable Development Goal for the promotion of gender equality. While recognizing that the promotion of gender equality improves the socioeconomic development of society, whether tangible effects are produced remains to be seen. Instead of creating gender-related standards for trade in the agreements, they simply mention implementing gender-equality commitments. While progress at the WTO and in most bilateral and regional trade agreements has been limited, there are some promising signs. For example, Canada’s agreement with Israel envisions forming a more comprehensive committee to assess changes and issues that arise with gender equality and trade.65 A dedicated chapter on gender encourages corporate responsibility, hiring practices and supporting small and medium-sized enterprises, though again, no measurable results are tracked. This chapter includes provisions such as ‘the Parties also acknowledge that women’s enhanced participation in the labour market and their economic independence and access to, and ownership of, economic resources contribute to sustainable and inclusive economic growth, prosperity, competitiveness and the well-being of society’.66 Canada has adopted a three-pronged strategy on trade and gender in its free trade agreements, including a trade and gender chapter, gender mainstreaming across departments and Gender Based Analysis Plus (GBA+).67 Seen as a pivotal instrument in policymaking, GBA+ seeks to inform policymakers that trade and various policies affect the genders differently. Under this approach, the objective is to evaluate policies regularly with a gendered lens to narrow the divide. The agreement with Israel signals a change towards binding mechanisms for gender commitments between countries. Most related commitments remain binding, with the language updated to include that ‘parties may consent to submit the matter to dispute settlement in accordance with Chapter Nineteen Dispute Settlement’.

José-Antonio Monteiro (July 2020). Gender-Related Provisions in Regional Trade Agreements Database. WTO Economic Research and Statistics Division. 65 Ko, at 89. 66 Government of Canada (25 May 2018). Canada–Israel Free Trade Agreement. www​.international​ .gc​.ca/​trade​-commerce/​trade​-agreements​-accords​-commerciaux/​agr​-acc/​israel/​fta​-ale/​text​-texte/​13​.aspx​ ?lang​=​eng. 67 UNCTAD (n.d.). Gender and Trade: Assessing the Impact of Trade Agreements on Gender Equality: Canada–EU Comprehensive Economic and Trade Agreement 15. See https://​unctad​.org/​ system/​files/​official​-document/​UNWomen​_2020d1​_en​.pdf. 64

Placing gender equality at the centre of global digital trade policy  113 4.3

Digital Economy Partnership Agreement

The Digital Economy Partnership Agreement (DEPA) ratified by New Zealand, Singapore and Chile, and discussed in a chapter in this handbook, provides an example of balancing digital trade and domestic policies.68 A dedicated digital trade agreement – in contrast to other trade accords that have chapters on digital trade – DEPA works as a modular agreement. It consists of 16 different modules designed to increase digital business and trade. These modules were created to work in conjunction with one another to combine issues that are inherently affected by the digital economy. Working independently or together with any combination of the 16 modules, the agreement’s flexibility remains one of its foremost strengths. DEPA mandates collaboration on artificial intelligence, government procurement and competition policy to foster inclusion. Based on building trust with data flows between nations, DEPA is guided by the principle that products and services can increase when the cross-border digital environment is more open. With several exceptions to the sharing of information available, the continued flexibility poses an appealing approach to governance in the digital space. Using a capacity-based approach, DEPA explicitly recognizes that SMEs require greater engagement with policymakers. In addition, underrepresented groups including women are governed through mandated programming aimed at increasing digital skills and the evidence base for policymaking.69 For example, Article 11.3 says: ‘To this end, the Parties shall cooperate on matters relating to digital inclusion, including participation of women, rural populations, low socio-economic groups and Indigenous Peoples in the digital economy. Cooperation may include: (a) sharing of experiences and best practices, including exchange of experts, with respect to digital inclusion.’70 DEPA is a marked departure from the standard trade agreements and its ratification signals a willingness among nations to tackle new challenges. Given the rapid development of digital technologies in modern trade, changing the approach in the formulation of trade agreements from static to flexible will not only increase responsiveness, but also help to address regulatory challenges, improve social development and ultimately increase digital trade.

5.

GENDER IN E-COMMERCE AND DIGITAL TRADE NEGOTIATIONS

While new trade agreements increasingly include measures to narrow the gender divide, these accords represent a small fraction of enacted free trade agreements. Moreover, much of this progress has been limited to countries in the developed world. Developing countries have traditionally been cautious about incorporating indirect trade concerns in trade agreements,

68 Government of Canada (9 June 2022). Background: Canada’s Possible Accession to the Digital Economy Partnership Agreement. See www​.international​.gc​.ca/​trade​-commerce/​consultations/​depa​ -apen/​background​-information​.aspx​?lang​=​eng. 69 DEPA (12 June 2020) Module 11 Digital Inclusion. See www​.sice​.oas​.org/​trade/​DEPA/​DEPA​ _Module11​_e​.pdf. 70 Ibid Article 11.3.

114  Research handbook on digital trade fearing these considerations may lead to trade barriers or that their implementation may constitute an excessive burden in terms of financial and human resources. Trade agreements require both short- and long-term approaches to remedy the current situation. In the short term, gender-related commitments need to be tailored to the economic and political contexts of the countries involved. An explicit reference to gender in the core text of agreements helps increase political commitment and may increase the availability of capital for funding women-owned SMEs. In some trade agreements, such as the North American Agreement on Labour Cooperation (a subsidiary of the North America Free Trade Agreement), gender has been raised in accords that accompany the core text. By targeting labour practices affecting women, the specific measures allow goals to be evaluated and critiqued more effectively. For example, the free trade agreement between the United Kingdom of Great Britain and Northern Ireland and New Zealand includes the following provision: 3.

The Parties recognize the role played by SMEs, including Maori-led and women-led enterprises, in economic growth and job creation, and the need to address the barriers to participation in digital trade for those entities. To this end, the Parties shall: a) foster close cooperation on digital trade between SMEs of the Parties; b) encourage their participation in platforms that help link them with international suppliers, buyers, and other potential business partners; and c) share best practices in improving digital skills and leveraging digital tools and technology to improve access to capital and credit, participation in government procurement opportunities, and other areas that could help SMEs adapt to digital trade.

The free trade agreement between the United Kingdom and Australia has a near-identical provision.71 While considerable room remains for further incorporation of gender issues in global digital trade and e-commerce work, it is notable that many of the lead facilitators of the JSI working groups at the WTO are women. In fact, according to a WTO source, 80 to 90% per cent of small group facilitators today are women, including those leading on cybersecurity, privacy, source code, telecommunications and the customs moratorium. Gender equality is a concern for all, but ensuring that women hold leadership positions with respect to the negotiations is an encouraging move. In the long term, initiatives targeting digital education for both women and girls are needed. Improving the digital literacy standards for women is not only an empowering measure for gender equality, but an expected requirement for future labour participation. Reducing information asymmetry and training gaps is important to level the gender playing field. Educational programming and technological competency effectively reduce deterrents to entry and level the playing field for women. Compounded with financial intermediaries providing initial capital for women-owned SMEs, the likelihood of success is exponentially increased for new and emerging small firms.

71 United Kingdom (16 December 2021). Free Trade Agreement (FTA) between the United Kingdom of Great Britain and Northern Ireland and Australia, Chapter 14. See https://​assets​.publishing​.service​ .gov​.uk/​government/​uploads/​system/​uploads/​attachment​_data/​file/​1040551/​uk​-australia​-free​-trade​ -agreement​-fta​-chapter​-14​-digital​-trade​.pdf.

Placing gender equality at the centre of global digital trade policy  115 As women play an increasingly important role in the JSI working groups, it is notable that delegations identified gender issues in the outcome document from the 12th WTO Ministerial Conference in June 2022. It states: ‘We recognize women’s economic empowerment and the contribution of MSMEs to inclusive and sustainable economic growth, acknowledge their different context, challenges and capabilities in countries at different stages of development, and we take note of the WTO, UNCTAD and ITC’s work on these issues.’72 The statement is a positive indicator for future prioritization of gender issues within digital trade and e-commerce.

6.

ADDRESSING GENDER IN DIGITAL TRADE AND E-COMMERCE AGREEMENTS

Although 75 bilateral and regional trade agreements include explicit provisions on gender and another 170 refer indirectly to gender equality through references to human rights, sustainable development and labour discrimination, the absence of a global framework may add to the existing discrimination towards women in digital trade and e-commerce.73 There are opportunities to address these frameworks in several forums, including the WTO, the International Labour Organization and UNCTAD.74 Formulating new regional or bilateral trade agreements that apply a gender lens is a positive step, but countries that have neglected these issues may need capacity-building assistance. Empowered by the consensus-based framework of the WTO, developing nations are better suited under a global framework rather than regional and bilateral free trade agreements that can create imbalances in their negotiation and implementation stages. Beyond international consensus at the WTO and trade agreements, other measures could include: ● Removing trade barriers by lowering or eliminating tariffs, increasing access to trade facilitation measures, and improving access to trade finance. ● Working with established WTO programmes and initiatives on services, agriculture, e-commerce, and micro, small and medium-sized enterprises to further empower women. ● Adopting policies that complement trade policies, such as improving access to education, financial resources, digital technologies, information and infrastructure. ● Adopting targeted and well-funded labour market adjustment policies to maximize benefits for women of trade openings and mitigating potential risks. The WTO and regional/bilateral trade agreement experience points to the potential to include gender-sensitive digital trade and e-commerce policies in the policy development process, with implications for both domestic and international rules. Capturing that potential requires

72 WTO (22 June 2022). MC12 Outcome Document. https://​docs​.wto​.org/​dol2fe/​Pages/​SS/​directdoc​ .aspx​?filename​=​q:/​WT/​MIN22/​24​.pdf​&​Open​=​True. 73 International Institute for Sustainable Development ISD, (November 2021). Mainstreaming Gender in Trade Policy: Practice, Evidence, and Ways Forward, 2. See www​.iisd​.org/​system/​files/​2021​ -11/​mainstreaming​-gender​-in​-trade​-policy​.pdf. 74 International Labour Organization.

116  Research handbook on digital trade both the capacity building and systemic frameworks designed to ensure that gender is a core policy consideration throughout the development process. While there are no easy mechanisms to achieve this, Canada’s implementation of GBA+ analysis throughout its policy development process serves as an example of how to better ensure that gender issues are placed at the forefront rather than as an afterthought. Applying GBA+ to policy development is important, because diverse groups of people can experience the same issue differently. Even in cases where the issues do not seem gendered, there may be varying impacts on diverse populations that extend beyond gender. GBA+ enhances policy development by guiding the gathering of information and analysis needed to inform the best decisions. Indeed, indirect discrimination against women and detrimental effects on women traders can occur if a gendered lens is not used when trying to address issues involving women in digital trade.75 Without a well-rounded analytical approach that considers different aspects of women’s lives, it is extremely difficult to produce productive and proactive solutions to gender-divide issues in digital trade.

Caroline Dommen (2021). How Can WTO Talks on E-Commerce Benefit Women? If Only We Knew. Human Rights Economics https://​humanrightseconomics​.ch/​2021/​gender​-equality/​gendered​ -impacts​-of​-wto​-e​-commerce​-talks/​. 75

PART II REGIONAL APPROACHES TO DIGITAL TRADE: NORTH AMERICA AND EUROPE

8. The EU as a digital trade actor Elaine Fahey

1.

OVERVIEW: THE EU AS A GLOBAL DATA ACTOR

EU data protection law and EU processing of data is broadly understood to have had global reach and effects, as does its cyber policy and increasingly its digital trade agenda.1 EU law has defied the odds and granted rights to those whom it was thought fell beyond regulatory reach.2 Key CJEU decisions3,4 have generated a global debate on new frontiers of privacy and privacy as a core EU value. The EU now unambiguously advocates the global reach of EU data protection standards in its policies.5 The EU’s new General Data Protection Regulation (GDPR) Regulation EU 2016/69 on the protection of natural persons with regard to the processing of personal data and the free movement of such data is landmark law-making. The GDPR has ignited a global wave of reflection upon global compliance with EU law, with more than 100 jurisdictions copying it, aligning with it or engaging with its substance, and at least 14 jurisdictions wholesale complying therewith.6 The new Regulation was perceived to mark a significant extension of the extra-territorial application of EU law. Its impact upon the EU’s external law-making is of much interest, partly because of its increasingly high-profile salience and partly also because of the challenges the EU faces as to the direction of its digital trade policy. Yet the EU has increasingly begun to advocate digital sovereignty,7,8 which mostly refers to Europe’s ability to act independently in the digital world9 but also links closely to ideas of 1 The piece develops further Elaine Fahey, ‘The European Union as a Digital Trade Actor: The Challenge of Being a Global Leader in Standard-Setting’ (2021) 27(2) International Trade Law and Regulation 155 and Elaine Fahey, The EU as a Global Digital Actor (Hart Publishing 2022) ch 2. 2 Orla Lynskey, ‘Deconstructing Data Protection: The Added Value of a Right to Data Protection in the EU Legal Order’ (2014) 63(3) ICLQ 569. 3 See Case C-131/12 Google Spain SL, Google, Inc. v Agencia Española de Protección de Datos, Mario Costeja González ECLI:EU:C:2014:317; See Case C-362/14 Maximillian Schrems v Data Protection Commissioner ECLI:EU:C:2015:650 See Case C‑293/12 and C‑594/12 Digital Rights Ireland and Others ECLI:EU:C:2014:238. 4 See Case C-230/14 Weltimmo s.r.o. v Nemzeti Adatvédelmi és Információszabadság Hatóság ECLI:EU:C:2015:639. 5 European Commission, ‘Communication from the Commission to the European Parliament and the Council: Exchanging and Protecting Personal Data in a Globalised World’ COM (2017) 07 final. 6 Anu Bradford, Brussels Effect: How the European Union Rules the World (OUP 2020). 7 European Council, ‘Special meeting of the European Council (1 and 2 October 2020) – Conclusions’ EUCO 13/20 (2020). 8 European Parliament, ‘Digital Sovereignty for Europe’ (2020) EPRS Ideas Paper Briefing www​ .europarl​.europa​.eu/​RegData/​etudes/​BRIE/​2020/​651992/​EPRS​_BRI(2020)651992​_EN​.pdf accessed 31 March 2023. 9 Cf European Commission, ‘Europe: The Keys to Sovereignty’ (News) (11 September 2020), https://​ec​.europa​.eu/​commission/​commissioners/​2019​-2024/​breton/​announcements/​europe​-keys​ -sovereignty​_en accessed 31 March 2023; Frances G Burwell and Kenneth Propp, ‘The European Union

118

The EU as a digital trade actor  119 digital protectionism, which is a feature of the high standards in international data transfer law that the CJEU appears to promote.10 How this will affect digital trade going forward remains to be seen. Increasingly, data sovereignty is emphasised by countries such as Russia and Vietnam as much as by the EU.11 Data in trade agreements takes many forms and relates to many regulatory issues increasingly conventionally outside of trade agreements. Cybersecurity incidents and concerns increasingly present challenges, not least the significant barriers they present to trade and investment.12,13 EU law appears to have only recently begun to bring cybersecurity within the ambit of trade agreements with more detail, agency, operations and standards-based reflections. There is still a lot of uncertainty about the interaction of cybersecurity and international trade.14 More than 50 countries have in recent years begun publishing cybersecurity strategies to define the security of the online environment and the EU has spearheaded many significant legislative and regulatory invocations as to cybersecurity.15 Beyond these significant innovations, it is difficult to evaluate the likely place of cybersecurity in EU trade agreements. The ‘backdrop’ of EU data regulation is also difficult to pinpoint as a dizzying morass of extraordinary regulation. A European Strategy for Data was published in 2020 to develop a Single Market in Data by 2025 and a Common European Data Space in nine areas, ranging from industrial manufacturing to health, financial, energy and agricultural data.16 In the 2020 work programme, ‘A Europe fit for the digital age’, the Commission proposed a new Digital Services Act and a new Digital Markets Act in 2020 in order to reinforce the single market for digital services and help provide smaller businesses with the legal clarity and level playing field they need.17 It develops complex means to understand gatekeepers on a company level. The EU has sought through its two key Acts – the Digital Services Act and the Digital Markets Act – and to a degree through a third, the Digital Governance Act, to build on the political successes of the GDPR. For example, it looks likely to adopt the world’s first binding AI

and the Search for Digital Sovereignty: Building “Fortress Europe” or Preparing for a New World?’ (Atlantic Council, June 2020) www​.atlanticcouncil​.org/​wp​-content/​uploads/​2020/​06/​The​-European​ -Union​-and​-the​-Search​-for​-Digital​-Sovereignty​-Building​-Fortress​-Europe​-or​-Preparing​-for​-a​-New​ -World​.pdf accessed 31 March 2023. 10 Susan Aaronson, ‘What Are We Talking about When We Talk about Digital Protectionism?’ (2019) 18(4) World Trade Review 541. 11 Emily Laidlaw, ‘Privacy and Cybersecurity in Digital Trade: The Challenge of Cross Border Data Flows’ (2021) Global Affairs Canada Paper https://​ssrn​.com/​abstract​=​3790936 accessed 31 March 2023. 12 Keman Huang, Stuard Madnick and Simon Johnson, ‘Framework for Understanding Cybersecurity Impacts on International Trade’ (2019) CISL Working Paper No 2019-23, 3 https://​ssrn​.com/​abstract​=​ 3555341 accessed 31 March 2023. 13 ibid. 14 Huang, Madnick and Johnson (n 12). 15 ibid, citing Alexander Klimburg, National Cyber Security Framework Manual. NATO CCD COE Publication (NATO 2013). 16 European Commission, ‘Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions “A European Strategy for Data”’ COM (2020) 66 final. 17 European Commission, ‘Commission Work Programme: A Union that Strives for More’ COM (2020) 37 final.

120  Research handbook on digital trade Act.18 The external reach of all this into digital trade seems, on the one hand, at this remove difficult to predict and, on the other hand, increasingly likely if its ambitions as to data are to be furthered. This is because of the complex position of the EU, advancing world-leading positions on data but also imposing regulations that are complex for third countries and parties to fathom, initially at least. This chapter assesses the EU as a global digital trade actor. It chooses case studies of EU law, of EU external relations law and in particular of ‘third country’ partnerships or relations of the EU of note, examining the US and UK for their complexity, their closeness and their innovations. It considers their features descriptively in the text that follows. The chapter thus considers the place of EU digital trade in international economic law (section 2). The chapter then outlines the evolution of the EU’s horizonal data strategy model clauses (section 3). It considers EU–US innovations as to trade and technology (section 4) and then (section 5) the EU–UK Brexit Trade and Cooperation Agreement (TCA) as a shift in digital trade, followed by offering conclusions.

2.

SITUATING THE PLACE OF THE EU DIGITAL TRADE

EU trade agreements have historically merged trade in services, establishment and electronic commerce rather than giving e-commerce a standalone chapter. In more contemporary post-Lisbon trade agreements with developed global economies such as the EU–Canada Comprehensive Economic and Trade Agreement (CETA), initial EU practice then saw a standalone e-commerce chapter. The EU subsequently negotiated the EU–Japan Economic Partnership Agreement (EPA) that has an e-commerce chapter covering trade in services, investment liberalisation and e-commerce therein. More recent negotiations with Australia, Mexico and the UK have seen the EU adopt US terminology on digital trade, and having a separate chapter on digital trade aligned carefully with services, but apart.19 Digital trade is still conventionally understood to have evolved in two genres – narrow and broad – with the EU traditionally falling between these extremes: the more digital-focused understanding of trade of the US being far away from the Chinese approach focused upon the ‘trade’ perspective.20 The EU is thought to stand out among WTO members for having made the greatest changes in its approach to e-commerce in PTAs over the year, although for very different economic and geopolitical factors than others such as the US and China.21 Yet

See European Commission, ‘Proposal for a Regulation of the European Parliament and of the Council Laying Down Harmonised Rules On Artificial Intelligence (Artificial Intelligence Act) and Amending Certain Union Legislative Acts’ COM (2021) 206 final. 19 See Datahub Global Data Governance Mapping Project: https://​datagovhub​.elliott​.gwu​.edu/​ accessed 31 March 2023; Mira Burri, ‘The Governance of Data and Data Flows in Trade Agreements: The Pitfalls of Legal Adaptation’ 51(56) University of California Davis Law Review 65. 20 Henry Gao, ‘Digital or Trade? The Contrasting Approaches of China and US to Digital Trade’ (2017) 21 Journal of International Economic Law 297. 21 Pierre Sauvé and Marta Soprana, ‘The Evolution of the EU Digital Trade Policy’ in Michael Hahn and Gulliame Van der Loo (eds), Law and Practice of the Common Commercial Policy (Brill Nijhoff 2020) 287; Martina Francesca Ferracane, Hosuk Lee-Makiyama and Erik van der Marel, ‘ECIPE Digital Trade Restrictiveness Survey Index’ (2018) ECIPE https://​ecipe​.org/​wp​-content/​uploads/​2018/​05/​DTRI​ _FINAL​.pdf accessed 31 March 2023. 18

The EU as a digital trade actor  121 it is regarded as a latecomer to many key issues in digital trade.22 As noted above, the EU has post-Lisbon merged e-commerce with trade in services and only recently moved to separate digital trade. It has thus moved closer to the US lexicon but also notably merged digital trade and services in some key recent trade agreements, such as the UK–EU TCA. However, the economic as opposed to rights-based construction of privacy as between APEC and the GDPR constitutes a significant challenge going forward in the search for global standards or the workability of data localisation in an era of the rising salience of privacy. The place of the EU as a first mover internationally on best practice in data protection and data flows on account of the high standards of the General Data Protection Regulation (GDPR) will be unavoidably significant for many countries. The EU’s Digital Strategies published in 2018 and 2020 indicated unambiguously a global approach confirmed in policy.23 The EU has been accused of digital protectionism in many contexts, despite its broader organisational goals to promote free and fair trade and to protect consumer rights. Recent trade agreements increasingly provide for free flows of data and set out provisions prohibiting ‘bad’ data localisation but with caveats, considered here below. Indeed, it is this complex approach to digital trade that characterises the EU. Similar to the approximately 80 or more FTAs globally that include provisions on privacy in varied form,24 the EU also has a variety of forms of trade agreement – such as agreements that are part of a broader framework agreement, encompassing Deep and Comprehensive Free Trade Agreements including Associate Agreements and free trade agreements.25 Some, such as26 EU–Central America, EU–Columbia or EU–Peru, provide that e-commerce shall be consistent with international standards of data protection, while others, such as EU–Ukraine, provide that e-commerce must be fully compatible with the highest international standards of data protection.27 EU agreements always include qualifications to these standards, for example

Sauvé and Soprana (n 21) 288. European Commission, ‘Communication from the Commission: European Commission Digital Strategy; Digitally Transformed, User-Focused and Data-Driven’ C (2018) 7118 final; European Commission, Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions, ‘Shaping Europe’s Digital Future’ COM (2020) 67 final. 24 José-Antonio Monteiro and Robert Teh, ‘Provisions on Electronic Commerce in Regional Trade Agreements’ (2017) WTO Working Paper ERSD-2017-11; Mira Burri and Rodrigo Polanco, ‘Digital Trade Provisions in Preferential Trade Agreements: Introducing a New Dataset’ (2020) 23 Journal of International Economic Law 187. 25 Sauvé and Soprana (n 21) 286; Burri, ‘The Governance of Data and Data Flows in Trade Agreements: The Pitfalls of Legal Adaptation’ (n 19). 26 Free Trade Agreement Between the European Union and the Republic of Singapore [2019] OJ L 294/2 art 8.57.4; Comprehensive and enhanced Partnership Agreement between the European Union and the European Atomic Energy Community and their Member States, of the one part, and the Republic of Armenia, of the other part [2018] OJ L23/4 art 197.2; Economic Partnership Agreement between the CARIFORUM States, of the one part, and the European Community and its Member States, of the other part [2008] OJ L289/3 art 119.2; Agreement establishing an association between the European Community and its Member States, of the one part, and the Republic of Chile, of the other part [2002] OJ 359/3, art 202; See Monteiro and Teh (n 24) 52; Magdalena Slok-Wodkowska and Joanna Mazur, ‘The EU’s Regional Trade Agreements: How the EU Addresses Challenges Related to Digital Transformation’ (2021) 57 International Journal of Management and Economics 105. 27 EU–Ukraine is a notable agreement establishing a particularly novel model of integration into the EU legal order (see for example Roman Petrov, Guillaume Van der Loo and Peter Van Elsuwege, ‘The 22 23

122  Research handbook on digital trade to give the parties the right to define and regulate their own levels of protection of personal data in pursuit of public policy objectives and not to be required to disclose confidential or sensitive information or data.28 Assertions of uncompromisingly high standards on the part of the EU in the post-GDPR context are arguably, then, far from accurate, but are rather more complex. Electronic commerce is the subject of a joint initiative by the WTO, the World Economic Forum and the Electronic World Trade Platform (eWTP), in the form of the Joint Statement Initiative (JSI) on e-commerce.29 The EU has been leading technical discussions in plenary and in small groups and the co-convenors include Australia, Japan and Singapore. Notably, the EU is actively seeking to upload its model horizontal articles. The WTO e-Commerce negotiations have been heavily hyped, understandably, for their potential and reach.30 The European Union has been placed in a position of unprecedented expectation as to its leading position as the privacy forerunner. Yet a closer inspection of the operation of its complex and nuanced model clauses suggests that the path upward is not straightforward. The idea of global provisions emerging appears increasingly more complex.31 In the EU proposal for WTO rules on electronic commerce, it adopts a highly distinctive approach of endorsing and protecting privacy as a fundamental right and seeks to ban data localisation measures and provide for a free data flow but also subject to conditions.32 The cleavages emerging globally appear created or shaped by complex EU positions. As a key supporter of multilateralism,

EU–Ukraine Association Agreement: A New Legal Instrument of Integration Without Membership?’ (2015) 1 Kyiv-Mohyla Law and Politics Journal 1) but notably pre-dating the GDPR. 28 Agreement between the European Union and Japan for an Economic Partnership (EU–Japan EPA) [2018] OJ L 330/3, art 18.1.2.h and art 18.16.7. 29 WTO, ‘Joint Statement on e-Commerce News Archives’ www​.wto​.org/​english/​news​_e/​archive​ _e/​jsec​_arc​_e​.htm accessed 31 March 2023. The negotiations have been structured around eight focus groups on ‘enabling digital trade/e-commerce’, ‘openness and digital trade/e-commerce’, trust, cross-cutting issues and telecommunications. Negotiations on a plurilateral agreement on e-commerce covering a range of rules on digital trade have yet to come to a conclusion; see Gautami Govindrajan and Ayushi Singh, ‘Curb Your Enthusiasm: The WTO E-Commerce Negotiations and the Developing World’ (2021) 13 Trade Law and Development 1; Ines Willemyns, ‘Agreement Forthcoming? A Comparison of EU, US, and Chinese RTAs in Times of Plurilateral E-Commerce Negotiations’ (2020) 23 Journal of International Economic Law 221; Slok-Wodkowska and Mazur (n 26); Mira Burri, ‘A WTO Agreement on Electronic Commerce: An Enquiry into Its Legal Substance and Viability’ (2021) Trade Law 4.0 Working Paper No 01/2021. 30 Digital Europe, ‘WTO eCommerce negotiations – A Unique Opportunity for Digital Trade’ (2020) www​.digitaleurope​.org/​wp/​wp​-content/​uploads/​2020/​10/​DIGITALEUROPE​_​-WTO​-eCommerce​ -negotiations​_A​-unique​-opportunity​-for​-digital​-trade​.pdf accessed 31 March 2023; Gary Clyde Hufbauer and Zhiyao (Lucy) Lu, ‘e-Commerce Talks Stumble on Data Issues, Privacy, and More’ (PIIE, October 2019) www​.piie​.com/​sites/​default/​files/​documents/​pb19​-14​.pdf accessed 31 March 2023; Henry Gao, ‘e-Commerce Governance: Back to Geneva?’ (CIGI, 14 February 2022) www​.cigionline​.org/​articles/​e​ -commerce​-governance​-back​-to​-geneva/​ accessed 31 March 2023. 31 Burri, ‘A WTO Agreement on Electronic Commerce: An Enquiry into Its Legal Substance and Viability’ (n 29); Svetlana Yakovleva, ‘EU’s Policy on Cross-Border Data Flows: Navigating the Thin Line between Liberalizing Digital Trade, Promoting Rules-Based Multilateralism and Safeguarding Fundamental Rights and Values’ in Elaine Fahey and Isabella Mancini (eds), Understanding the EU as a Good Global Actor: Whose Metrics? (Edward Elgar Publishing 2022). 32 WTO, ‘Joint Statement on Electronic Commerce: EU Proposal for WTO Disciplines and Commitments Relating to Electronic Commerce, Communication from the European Union’ (26 April 2019) INF/ECOM/22.

The EU as a digital trade actor  123 the WTO has appeared as a key place where the EU could advance its convergence agenda.33 There are highly divergent global expectations and hopes as to its outcomes. Global cleavages emerging in digital trade often appear rooted in EU-based norms. For example, literature on digital trade arguably frames EU initiatives, developments and response as a highly esoteric form of engagement with multilateralism. The capacity of the EU to upload its values here will be highly critical to the future success of EU global reach – can it make it a multilateral settlement?34 The chapter next considers a key manner in which the EU expresses its ‘values’ as to digital trade, with respect to personal data and its model clauses for data flows.

3.

THE EU HORIZONTAL STRATEGY FOR DATA AND MODEL CLAUSES: QUICKLY FORGOTTEN?

In the wake of the introduction of the EU’s far-reaching GDPR, the European Commission developed highly significant so-called model horizontal clauses on cross-border data flows and personal data protection in EU trade and investment agreements after pressure from the European Parliament.35 Institutional balance or conflict has evolved greatly but also has new dimensions, with more pushback from the Member States.36 The model clauses include so-called Articles A, B and X. Article A provides for a declaratory commitment on cross-border data flows and prohibits restrictions in four data and IT localisation requirements. Article B sets out counterbalancing provisions for national measures as to personal data. Article X on regulatory cooperation with respect to digital trade then provides a carve-out for cross-border data flows and the protection of personal data from the dialogue on regulatory issues. The Commission thereafter submitted these provisions to Australia, Chile, Indonesia,

33 Elaine Fahey (ed), Framing Convergence with the Global Legal Order: The EU and the World (Bloomsbury 2020). 34 Martina Ferracane and Li Mosi, ‘What Kinds of Rules Are Needed to Support Digital Trade?’ in Bernard Hoekman, Tu Xinquan and Dong Wang (eds), Rebooting Multilateral Trade Cooperation: Perspectives from China and Europe (CERP Press 2021); Sangeeta Khorana and W Gregory Voss, ‘The Digital Single Market: Move from Traditional to Digital?’ in Sangeeta Khorana María García, Handbook on the EU and International Law (Edward Elgar Publishing 2018); Ines Willemyns, ‘Agreement Forthcoming? A Comparison of EU, US, and Chinese RTAs in Times of Plurilateral e-Commerce Negotiations’ (2020) 23 Journal of International Economic Law 221; Burri, ‘A WTO Agreement On Electronic Commerce: An Enquiry into Its Legal Substance and Viability’ (n 29); Mira Burri, ‘Towards a New Treaty on Digital Trade’ (2021) 55 Journal of World Trade 77. 35 European Commission, ‘Horizontal Provisions for Cross-Border Data Flows and for Personal Data Protection (in EU Trade and Investment Agreements)’ (2018) Tradoc 156884 https://​trade​.ec​ .europa​.eu/​doclib/​docs/​2018/​may/​tradoc​_156884​.pdf accessed 31 March 2023; European Commission, ‘Communication from the Commission to the European Parliament and the Council: Exchanging and Protecting Personal Data in a Globalised World’ COM(2017) 07 final; European Commission, ‘Communication from the Commission to the European Parliament and the Council: Exchanging and Protecting Personal Data in a Globalised World’ COM(2017) 07 final. 36 See Council of the European Union (as published by Statewatch), ‘Working Paper on EU– UK Trade and Cooperation Agreement – Compilation of Written Questions and Answers’ (2021) WK 793/2021 INIT www​.statewatch​.org/​media/​1759/​eu​-uk​-agreement​-questions​-com​-wk​-793​-21​.pdf accessed 31 March 2023.

124  Research handbook on digital trade Mexico, New Zealand, Tunisia and the UK – mostly successfully – and has sought to replace the EU–Japan EPA rendez vous clause, to a degree also in the EU–Mexico modernised FTA.37 The model clauses are understood to contain a narrower prohibition on restrictions of cross-border data flows than in so-called US models implemented in the CPTPP, USMCA, US–Japan Digital Trade Agreement and China’s model implementation of the RCEP.38 Whether the EU has actually successfully ‘uploaded’ these views to the WTO remains to be seen.39 The horizontal strategy is nonetheless understood to be highly ambitious and of much significance to the deeper trade era – but its implementation in practice has shown itself to be more complex and the EU Member States’ views of the swings in its negotiation position have proven to be complex.40 The process of EU’s ‘repositioning’ on the issue of data flows has arguably begun to mature in a very short space of time.41 In fact, it is said that the clauses are ‘fully’ endorsed in the EU’s currently negotiated deals with Australia, New Zealand and Tunisia, which include in their draft digital trade chapters norms on the free flow of data and data localisation bans.42 Third countries with no comprehensive data law have clearly had a more challenging time or been more complex to deal with. Yet this may be open to much contestation. The place of free flows of data and privacy was left undetermined and postponed as to the EU–Japan EPA, despite having many notable provisions on privacy.43 As Burri states, however, the EU’s repositioning and newer commitments are also linked with higher levels of the regulation of data protection.44 Indeed, one can say that their place, or otherwise, in any agreement more reflects the shifts of other parties, usually significantly ‘towards’ the EU, perhaps philosophically. The EU has also sought to make provision for review of the operation of its data flow provisions with respect to privacy and thus what were presumed to be controversial EU–Japan EPA provisions, discussed further below, are now widespread as more flexible rendez vous clauses.45 The EU gives itself

37 Jan A Micallef, ‘Digital Trade in EU FTAs: Are EU FTAs Allowing Cross Border Digital Trade to Reach Its Full Potential?’ (2019) 53 Journal of World Trade 855, 867; See also Svetlana Yakovleva and Kristina Irion, ‘Pitching Trade against Privacy: Reconciling EU Governance of Personal Data Flows with External Trade’ (2020) 10(3) International Data Privacy Law 201, 214. 38 See Yakovleva (n 31). 39 Communication from the European Union, Joint Statement on Electronic Commerce, ‘EU Proposal For WTO Disciplines and Commitments Relating to Electronic Commerce’ INF/ECOM/22 (2019) https://​trade​.ec​.europa​.eu/​doclib/​docs/​2019/​may/​tradoc​_157880​.pdf accessed 31 March 2023. 40 Yakovleva and Irion (n 37) 219–20. 41 Burri, ‘A WTO Agreement On Electronic Commerce: An Enquiry into Its Legal Substance and Viability’ (n 29). 42 ibid. 43 E.g. EU–Japan EPA: Article 10.4.2 (subjecting data sharing regarding temporary entry of business persons to each party’s privacy and data protection law); Article 20.5 (affirming that intellectual property-related disclosure of information was not required if except under either party’s privacy laws; Article 21.4.e (subjecting provision of proposed regulations to applicable privacy law) or Article 32.1 of the Protocol on rules of origin and origin procedures (affirming furnishing or access to information was not required if contrary to either party’s personal data protection and privacy law). 44 Burri, ‘A WTO Agreement on Electronic Commerce: An Enquiry into Its Legal Substance and Viability’ (n 29). 45 Article 5(2) of draft Trade Agreement between the European Union and Australia (EU–Australia FTA) https://​trade​.ec​.europa​.eu/​doclib/​press/​index​.cfm​?id​=​1865 accessed 31 March 2023; Draft European Union-New Zealand Free Trade Agreement (EU–New Zealand FTA) https://​ec​.europa​.eu/​ trade/​policy/​countries​-and​-regions/​countries/​new​-zealand/​ accessed 31 March 2023 and the European

The EU as a digital trade actor  125 ample regulatory leeway for its current and future data protection measures.46 There are also broad carve-outs providing for the right to regulate.47 The EU provides for data sovereignty of a broad nature within its recently agreed or recently negotiated clauses.48 However, the EU–UK TCA operated as a significant exception or turning point in some respects as to the model clauses. On the one hand, provisions of the TCA provide that there are significant consequences for the UK leaving the Council of Europe or eradicating its Human Rights Act given the joint commitments to human rights and the rule of law.49 The digital trade chapter clauses however span an article providing that data protection is a ‘right’, but not expressed as a ‘fundamental right’.50 The reformulation of privacy other than as a fundamental right in the EU–UK TCA drew disquiet from the EU Member States as to the position of the European Commission in negotiating downwards its model clauses with a departing state.51 The model causes here turn out to be a template of flexibility, not absolutism. Whether the EU’s clauses cause difficulty for future public policy or ultimately undermine the EU’s goals as to liberalising data flows remains to be seen. The next section examines in more detail one of the EU’s most significant ongoing bilateral developments: the evolution of an EU–US Trade and Technology Council (TTC).

4.

THE EU–US TRANSATLANTIC TRADE AND TECHNOLOGY COUNCIL: SHIFTING MULTILATERALISM FORWARD?

Transatlantic data flows amount to some of the most significant for the global economy.52 Even in the midst of a pandemic, the transatlantic economy amounted to one third of global

Union–Tunisia Free Trade Agreements (EU–Tunisia FTAs) https://​ec​.europa​.eu/​trade/​policy/​countries​ -and​-regions/​countries/​tunisia/​ accessed 31 March 2023. 46 Mira Burri, ‘Interfacing Privacy and Trade’ (2021) 53 Case Western Reserve Journal of International Law International 35, 80. 47 ‘The Parties reaffirm the right to regulate within their territories to achieve legitimate policy objectives, such as the protection of public health, social services, public education, safety, the environment including climate change, public morals, social or consumer protection, privacy and data protection, or the promotion and protection of cultural diversity’; see e.g. Article 2 draft EU–Australia FTA; draft EU–New Zealand and the EU–Tunisia FTAs. See Burri, ‘Interfacing Privacy and Trade’ (n 46). 48 See Burri, ‘Interfacing Privacy and Trade’ (n 46); ‘Each Party may adopt and maintain the safeguards it deems appropriate to ensure the protection of personal data and privacy, including through the adoption and application of rules for the cross-border transfer of personal data. Nothing in this agreement shall affect the protection of personal data and privacy afforded by the Parties’ respective safeguards.’ See Article 6(2) draft EU–Australia FTA, draft EU–New Zealand and the EU–Tunisia FTAs (n 45). 49 Trade and Cooperation Agreement between the European Union and the European Atomic Energy Community, of the one part, and the United Kingdom of Great Britain and Northern Ireland, of the other part (EU–UK TCA) [2021] OJ L149/10. 50 See EU–UK TCA, art 202. 51 ‘Brexit: Goodbye and Hello: The New EU–UK Security Architecture, Civil Liberties and Democratic Control’ (Statewatch, 20 January 2022) https://​www​.statewatch​.org/​brexit​-goodbye​-and​ -hello​-the​-new​-eu​-uk​-security​-architecture​-civil​-liberties​-and​-democratic​-control accessed 31 March 2023. 52 See US Chamber of Commerce, ‘Transatlantic Data Flows: Moving Data with Confidence’ (20 September 2021) www​.uschamber​.com/​technology/​data​-privacy/​transatlantic​-dataflows accessed 31 March 2023.

126  Research handbook on digital trade GDP.53 At the recent announcement of a transatlantic data privacy framework, the US government stressed that more data flows between the United States and Europe mattered more than anywhere else in the world, enabling the $7.1 trillion US–EU economic relationship.54 The Transatlantic Trade and Investment Partnership (TTIP), the largest-scale form of transatlantic collaboration in recent history, expressly excluded data flows from its negotiations. Its negotiation of e-commerce could have been pivotal, given the gap between the TPP and EU agreements emerging as to data flows but also the gap emerging as to the formulation of digital trade between the EU and US.55 TTIP texts (initially leaked) exposed the significant divergence between the US and the EU on data protection, particularly given that TTIP was being negotiated in advance of the GDPR. The US shift towards the need for federal privacy laws, agencies and standards could radically shift the understanding of the outcome of the TTIP negotiations, discussed further below.56 The EU proposed an EU–US Joint Agenda for Global Change to the incoming Biden administration very early on in the new relationship, which would include a Transatlantic Trade and Technology Council to engage with key global challenges.57 However, civil society responded adversely to its inception.58 A Transatlantic Trade and Technology Council, as proposed by the EU in late 2020 and already in place by autumn 2021, could provide an important bedrock from which multilat Daniel Hamilton and Joseph Quinlan, US Chamber of Commerce, The Transatlantic Economy 2021 (2021) www​.amchameu​.eu/​sites/​default/​files/​publications/​files/​tran​satlantice​conomy2021​_fullreporthr​ .pdf accessed 31 March 2023; Daniel Hamilton and Joseph Quinlan, US Chamber of Commerce, The Transatlantic Economy 2022 (2022) www​.uschamber​.com/​assets/​documents/​TE2022​_Report​_LR​.pdf accessed 31 March 2023. 54 The White House, ‘FACT SHEET: United States and European Commission Announce Trans-Atlantic Data Privacy Framework’ (25 March 2022) www​.whitehouse​.gov/​briefing​-room/​ statements​-releases/​2022/​03/​25/​fact​-sheet​-united​-states​-and​-european​-commission​-announce​-trans​ -atlantic​-data​-privacy​-framework/​ accessed 31 March 2023. 55 European Commission, ‘TTIP: Initial Proposal on Trade in Services, Investment and e-Commerce’ (31 July 2015) http://​trade​.ec​.europa​.eu/​doclib/​docs/​2015/​july/​tradoc​_153669​.pdf accessed 31 March 2023; European Commission, ‘TTIP: Annexes to the Services, Investment and e-Commerce Initial Proposal’ (31 July 2015) http://​trade​.ec​.europa​.eu/​doclib/​docs/​2015/​july/​tradoc​_153670​.pdf accessed 31 March 2023; European Commission, ‘A Reading Guide to the EU Proposal on Services, Investment and e-Commerce for the Transatlantic Trade and Investment Partnership’ (31 July 2015) http://​trade​.ec​ .europa​.eu/​doclib/​docs/​2015/​july/​tradoc​_153668​.pdf accessed 31 March 2023; European Parliament, ‘TTIP Legislative Train Schedule’ (2020) www​.europarl​.europa​.eu/​legislative​-train/​theme​-international​ -trade​-inta/​file​-ttip​-services​-investment​-and​-e​-commerce accessed 31 March 2023; See Mira Burri, ‘The Regulation of Data Flows through Trade Agreements’ (2017) 48 Georgetown Journal of International Law 407; Walter Berka, ‘CETA, TTIP, TiSA, and Data Protection’ in Stefan Griller, Walter Obwexer and Erich Vranes (eds), Mega-Regional Trade Agreements: CETA, TTIP, and TiSA: New Orientations for EU External Economic Relations (OUP 2017); Andrea Renda and Christopher Yoo, ‘Telecommunications and Internet Services: The Digital Side of the TTIP’ (2015) CEPS, TIPP in Balance Project Paper No 8 www​.ceps​.eu/​download/​publication/​?id​=​9012​&​pdf​=​SR112​%20Renda​%20and​%20Yoo​%20Telecoms​ %20TTIP​.pdf accessed 31 March 2023. 56 Thomas Streinz, ‘Digital Megaregulation Uncontested? TPP’s Model for the Global Digital Economy’ in Benedict Kingsbury and others (eds), Megaregulation Contested (OUP 2019). 57 ‘Joint Communication to the European Parliament, the European Council and the Council: A New EU–US Agenda for Global Change’ JOIN (2021)22 final. See the EU–US Trade and Technology Council Website: https://​futurium​.ec​.europa​.eu/​en/​EU​-US​-TTC accessed 31 March 2023. 58 Trans Atlantic Consumer Dialogue (TACD), ‘Lack of Transparency Could Thwart the Strong Consumer Safeguards that Must Be the Goal of EU–US Cooperation Dialogues’ (Press Release, 28 September 2021) https://​tacd​.org/​eu​-us​-organisations​-transparency​-ttc​-pr/​ accessed 31 March 2023. 53

The EU as a digital trade actor  127 eral e-commerce developments can flourish.59 The EU–US Joint Agenda for Global Change included a Transatlantic Trade and Technology Council, putatively developing a loose institutionalisation of key global challenges currently not well covered by, for example, the WTO. Thus at the outset of the Biden administration, the EU proposed as part of its global change a Transatlantic Trade and Technology Council (TTC), centred on multiple working groups: Technology Standards Cooperation; Climate and Clean Tech; Secure Supply Chains; ICT Security and Competitiveness; Data Governance and Technology Platforms; Misuse of Technology Threatening Security & Human Rights; Export Controls Cooperation; Investment Screening Cooperation; Promoting SME Access to and Use of Digital Technologies; and Global Trade Challenges.60 Notably, seven of the ten working groups address themes that refer to technology, either with a security angle or from a competition perspective.61 What is noticeable is the evolution of security between the early meetings, in 2021 into 2022, as the Ukraine crisis unfolded. Yet also notable is the speed with which the entity evolved. Whether it will generate global law-making in any of these fields remains to be seen. As to Working Group 10 on global trade challenges, industry and academia were united on the need for WTO reform to be prioritised as a transatlantic goal, emphasising the acute necessity of multilateralism within this bilateral engagement.62 The initial TTC meeting was poorly received by NGOs.63 The first post-meeting consensus was, however, that the TTC was off to a ‘promising start’, but observers also noted that the bar for success in the first meeting was low.64 In fact, the external regulatory landscape was understood to be an advantage for the EU in taking the lead – perhaps so much so as to make the rule-making exercise questionable bilaterally. As Demertzis aptly states: The EU has discovered that there is a first-mover advantage when it comes to regulating. Following GDPR, they have proposed a Digital Markets Act, a Digital Services Act and regulations on artificial intelligence […] The United States, in contrast, has thus far failed to adopt national policies in any of those areas […] and the U.S. will once again be reminded of the old axiom of politics that you can’t fight something with nothing.65 59 The initial Joint Statement, the so-called Pittsburgh Statement: see European Commission, ‘EU–US Trade and Technology Council Inaugural Joint Statement’ (Press Release, 29 September 2021) https://​ec​.europa​.eu/​commission/​presscorner/​detail/​en/​STATEMENT​_21​_4951 accessed 31 March 2023; Chad P Bown and Cecilia Malmström, ‘What Is the Transatlantic Trade and Technology Council’ (PIIE, 24 September 2021) www​.piie​.com/​blogs/​trade​-and​-investment​-policy​-watch/​what​-us​-eu​-trade​ -and​-technology​-council​-five​-things​-you​-need accessed 31 March 2023. 60 See the EU–US Trade and Technology Council Website: https://​futurium​.ec​.europa​.eu/​en/​EU​-US​ -TTC accessed 31 March 2023. 61 Maria Demertzis, ‘US–EU Relations in the First Year of President Biden: A View from Brussels’ (Transatlantic, 8 November 2021) www​.transatlantic​.org/​wp​-content/​uploads/​2021/​12/​11​-10​-2021​ -Demertzis​-US​-EU​-trade​-Challenges​-v2​.pdf accessed 31 March 2023. 62 Cecillia Bonefeld-Dahl, ‘Becoming Tech Allies: 24 Targets for the EU–US Trade & Technology Council by 2024’ (Digital Europe, 2 February 2022) www​.digitaleurope​.org/​resources/​becoming​-tech​ -allies​-24​-targets​-for​-the​-eu​-us​-trade​-technology​-council​-by​-2024/​ accessed 31 March 2023. 63 Trans Atlantic Consumer Dialogue (TACD), ‘Lack of Transparency Could Thwart the Strong Consumer Safeguards that Must Be the Goal of EU–US Cooperation Dialogues’ (n 58). The Futurium platform where its work took effect appeared operationally challenging to see as an engagement platform. 64 Demertzis (n 61). 65 The next TTC was scheduled for 15–16 May 2022 in France. See European Commission, ‘CSD Meeting on the EU–US Trade and Technology Council’ (2022) https://​policy​.trade​.ec​.europa​.eu/​events/​ csd​-meeting​-eu​-us​-trade​-and​-technology​-council​-2022​-05​-11​_en accessed 31 March 2023.

128  Research handbook on digital trade Digital trade has increasingly many key features for EU law across a range of regulatory fields with significant implications for trade, technology, consumer rights and fundamental rights. The struggle here to evolve policy is thus salient for EU law internally as much as externally. There was significant parliamentary interest in this cooperation which reflects the heightened salience of digital trade in EU–US relations, although the place of the EP appears curious and excluded de jure.66 The TTC has concrete and broad ambitions to align standards in times of geopolitical shift, with the EU and US coming together to align technical standards and joint mapping for certain key goods supply chain. It can be said that digitisation and greening are global challenges but cannot per se be resolved by standards alignment alone or by like-minded cooperation, however noble-minded – raising questions as to whether reform of the WTO should not in fact be the key focus. Whether the working groups’ outcomes align well with the WTO agenda more broadly also remains to be seen. The challenges for civil society engaging with the breadth of the issues proposed and in this fashion could be arguably more relevant than ever. As Hamilton states: ‘These important goals can position both parties as standard-setters with regard to responsible use of technology and an open Internet that respects human rights […] a democratic internet requires coordination on regulation, not patchwork legislation affording users’ different democratic rights when they cross borders.’67 Separately of note is that while the CJEU struck down the EU–US Privacy Shield in Schrems II, the EU and US finally agreed a new transatlantic data privacy framework principle in March 2022, to include a ‘trans-atlantic court’ and independent oversight.68 In October 2022, President Biden signed an Executive Order ‘enhancing safeguards for United States signals intelligence activities’, which along with Regulations issued purport to address all issues raised by the CJEU, in particular ‘binding safeguards’, a first instance Civil Liberties Protection Officer and a Data Protection Review Court, as a means to procure an adequacy decision.69 Yet the US Biden administration has also shifted its focus in trade negotiations 66 European Parliament: Marcin Szczepanski, ‘EU–US Trade and Technology Council: New Forum For Transatlantic Cooperation’ (2021) European Parliamentary Research Service PE 698.037; European Parliament, ‘Statement on the Inaugural Meeting of the Trade and Technology Council (TTC)’ (Press Release, 30 September 2021) www​.europarl​.europa​.eu/​news/​en/​press​-room/​20210930IPR13906/​ statement​-on​-the​-inaugural​-meeting​-of​-the​-trade​-and​-technology​-council​-ttc accessed 31 March 2023; European Parliament, ‘The Outcome of the EU–US Trade and Technology Council (TTC) (debate)’ (11 November 2021) www​.europarl​.europa​.eu/​doceo/​document/​CRE​-9​-2021​-11​-11​-ITM​-003​_EN​.html accessed 31 March 2023; European Parliament, ‘Resolution of 6 October 2021 on the Future of EU–US Relations (2021/2038(INI))’ (6 October 2021) www​.europarl​.europa​.eu/​doceo/​document/​TA​-9​-2021​ -0410​_EN​.html accessed 31 March 2023; European Parliament, Committee on International Trade, ‘Opinion of the Committee on International Trade for the Committee on Foreign Affairs on the Future of EU–US relations (2021/2038(INI))’ (17 June 2021) www​.europarl​.europa​.eu/​doceo/​document/​INTA​ -AD​-691423​_EN​.pdf accessed 31 March 2023. 67 Daniel Hamilton, ‘Getting to Yes: Making the U.S.–EU Trade and Technology Council Effective (Summary Brief)’ (Transatlantic, 6 March 2022) www​.transatlantic​.org/​wp​-content/​uploads/​2022/​03/​ TTC​-summary​-brief​-final​-March​-6​-2022​.pdf accessed 31 March 2023. 68 European Commission, ‘European Commission and United States Joint Statement on Trans-Atlantic Data Privacy Framework’ (Press Release, 25 March 2022) https://​ec​.europa​.eu/​commission/​presscorner/​ detail/​en/​IP​_22​_2087 accessed 31 March 2023. 69 See US Government: www​.justice​.gov/​opcl/​redress​-data​-protection​-review​-court; www​ .whitehouse​.gov/​briefing​-room/​presidential​-actions/​2022/​10/​07/​executive​-order​-on​-enhancing​ -safeguards​-for​-united​-states​-signals​-intelligence​-activities/​.

The EU as a digital trade actor  129 towards soft law frameworks. For instance, the Indo-Pacific Economic Partnership (IPEF), to which the US is party, among others in the Indo-Pacific region, has many outline objectives on data flows and data localisation and the digital economy across borders based upon soft law. It mimics the allegedly most cutting-edge digital governance trade agreement which is also a soft law non-binding partnership, the 2020 Digital Economy Partnership Agreement (DEPA) between Chile, New Zealand and Singapore.70 It remains to be seen what the outcome of the latest EU–US agreement and US Executive Order is and if it shifts the debates. The EU–US TTC may yet need to further refocus upon the place of institution in proposing solutions. As noted above, several US states have recently adopted the GDPR de facto if not de jure, including California. The capacity for this convergence to evolve and to align bilaterally, particularly through institutions, could become pivotal going forward. Its global effects, however, remain to be seen. The chapter finally considers the question of another significant but less ‘close’ bilateral arrangement, the EU–UK TCA, as a challenge to the EU as a global actor.

5.

EU–UK TRADE AND COOPERATION AGREEMENT: A CHALLENGE TO THE EU AS A GLOBAL DIGITAL ACTOR?

The post-Brexit free trade agreement signed between the EU and the UK in December 2020, the UK–EU Trade and Cooperation Agreement (TCA) is a very modern and contemporary agreement but still a very slim one, falling short of a single market membership and a very ‘hard’ Brexit that had been negotiated.71 There had been a tremendous amount of differences for businesses, importers and so on, aiming at the provisions of this in practical operation. But in terms of digital trade, it is arguably aligned to other more contemporary EU agreements. The word data appears more than a thousand times in the TCA. Data transfer is similarly voluminous: 32 times in the PNR chapter, for example. The chapter contains numbers clauses similar to most EU negotiations and recently agreed provisions at the time, for example EU–Mexico and EU–Australia as to regulatory cooperation and data flows. One of the most interesting chapters in the TCA is arguably the digital trade chapter in Title III, largely not for what is in the agreement, but rather for what is not agreed or what diverges from what might have been there. The UK was at great pains to emphasise in press releases on the negotiation of other rollover agreements with third countries during its exit process that its newer digital trade provisions were consistently more modern, contemporary and supportive of free flows of data than those of the EU.72 The TCA digital trade chapter is a slim and modern chapter with many contemporary positions of the EU expressed therein, in terms of the model clauses, but See www​.mfat​.govt​.nz/​en/​trade/​free​-trade​-agreements/​free​-trade​-agreements​-in​-force/​digital​ -economy​-partnership​-agreement​-depa/​depa​-text​-and​-resources/​ 71 See Federico Fabbrini (ed), The Law & Politics of Brexit (Oxford University Press 2017); Federico Fabbrini (ed), The Law & Politics of Brexit. Volume 2: The Withdrawal Agreement (Oxford University Press 2020); Federico Fabbrini (ed), The Law & Politics of Brexit. Volume 3: The Framework of New EU–UK Relations (Oxford University Press 2021); Federico Fabbrini (ed), The Law & Politics of Brexit. Volume 4: The Protocol on Ireland/Northern Ireland (Oxford University Press 2022, forthcoming). 72 For example, as to the EU–Japan agreement: see UK Government, Department for International Trade, ‘Digital and Data Provisions in the UK–Japan CEPA’ https://​assets​.publishing​.service​.gov​ 70

130  Research handbook on digital trade also betraying some evidence of the haste of the negotiations.73 The EU’s horizontal strategy has only partially been accepted by the UK in the EU–UK TCA, it having clear its ambitions to join CPTPP.74 Data flows and localisation bans feature in the TCA in Article 201,75 alongside the model clauses’ replication as to the protection of data as a right. However, the removal of the term ‘fundamental’ from data as a fundamental right has caused considerable concern as between the Member States.76 Whether this is more representative of the challenges that the EU is likely to face when negotiating with other third countries remains to be seen. The place of the TCA in ‘downgrading’ privacy is probably overstated. Yet its fit with a more innovative and dynamic global Britain is far from certain if the UK is fixated upon developing a ‘rival’ GDPR, as it began to do in 2022.77 The EU has also sought to include regulatory cooperation in the area of digital trade or e-commerce in all of its post-Lisbon trade agreements with all major developed global economies, as shown in EU–Canada chapter 16 or EU–Japan Economic Partnership Agreement (JEEPA) chapter 8, at least until the UK–EU TCA. It is important to state that many such articles of the regulatory cooperation chapters are embedded heavily in multilateralism, but not so as to the TCA.78 In fact its regulatory cooperation provisions are considerably muted in contrast to, for example, Article 16.4 CETA or Article 8.80 EU–Japan Economic Partnership .uk/​government/​uploads/​system/​uploads/​attachment​_data/​file/​933990/​uk​-japan​-cepa​-digital​-and​-data​ -explainer​.pdf accessed 31 March 2023. 73 Mira Burri, ‘Approaches to Digital Trade and Data Flow Regulation Across Jurisdictions: Implications for the Future EU–ASEAN Agreement’ (2022) 49 Legal Issues of Economic Integration 149. 74 UK Government, ‘UK Applies to Join Huge Pacific Free Trade Area CPTPP’ (Press Release, 30 January 2021) www​.gov​.uk/​government/​news/​uk​-applies​-to​-join​-huge​-pacific​-free​-trade​-area​-cptpp accessed 31 March 2023. 75 See EU–UK TCA, art 201: Article 201 Cross-border data flows 1. The Parties are committed to ensuring cross-border data flows to facilitate trade in the digital economy. To that end, cross-border data flows shall not be restricted between the Parties by a Party: (a) requiring the use of computing facilities or network elements in the Party’s territory for processing, including by imposing the use of computing facilities or network elements that are certified or approved in the territory of a Party; (b) requiring the localisation of data in the Party's territory for storage or processing; (c) prohibiting the storage or processing in the territory of the other Party; or (d) making the cross-border transfer of data contingent upon use of computing facilities or network elements in the Parties' territory or upon localisation requirements in the Parties’ territory. The Parties shall keep the implementation of this provision under review and assess its functioning within three years of the date of entry into force of this Agreement. A Party may at any time propose to the other Party to review the list of restrictions listed in paragraph 1. Such a request shall be accorded sympathetic consideration. 76 Council of the European Union (n 36). 77 UK Government, ‘UK Unveils Post-Brexit Global Data Plans to Boost Growth, Increase Trade and Improve Healthcare’ (Press Release, 26 August 2021) www​.gov​.uk/​government/​news/​uk​-unveils​ -post​-brexit​-global​-data​-plans​-to​-boost​-growth​-increase​-trade​-and​-improve​-healthcare accessed 31 March 2023. 78 Fahey, ‘The European Union as a Digital Trade Actor: The Challenge of Being a Global Leader in Standard-Setting’ (n 1).

The EU as a digital trade actor  131 Agreement (JEEPA). These are important discernible differences which evidence the rancour and contestability of the relationship, but also perhaps a lack of respect between global Britain and the EU as a global digital actor.79 There was also a very important UK adequacy decision finally been granted in 2021 to enable free data flows. The TCA made provision for data flows pending the adoption of an adequacy decision.80 The TCA provided for lengthy provisions as to flows of data pending the adoption of an adequacy decision. It proved to be a controversial process, with vocal opposition from the European Parliament. Ultimately, a UK–EU adequacy decision was adopted on 28 June 2021 on the basis of its alignment to the GDPR, but only after highly vocal opposition from the EP. There are notable carve-outs for immigration and a sunset clause which limits its duration.81 A review of all EU adequacy decision clauses was planned by the Commission in 2022, thus providing an increasingly interesting backdrop. Notable proposals from the UK to show Brexit ‘being done’ have included recently a so-called bonfire of retained EU law: that is, to ‘get rid of’ all EU law or for the UK to depart from the bureaucracy of the GDPR despite a unified assertion from businesses, consumer groups and the like as to the need for an infrastructure of this kind and the existing compliance.82 The impact of these proposals remains to be seen. Exit of the UK from the ECHR and the ‘reform’ of UK human rights law have long been on the agenda of the UK Conservative Party.83 Post-Brexit these issues may assume a heightened salience with the UK in the event of complex divergences emerging. The GDPR has ignited a global wave of reflection upon global compliance with EU law, with dozens of jurisdictions copying it, aligning with it or engaging with its substance, and at least 14 jurisdictions complying wholesale therewith.84 This could have significant consequences However, in the UK–EU TCA there are only very small pockets of internationalisation evident, and not in the digital chapter: contrast Cybersecurity in Article 704 ‘Cooperation on cyber issues’, providing that for a distinctive commitment to developing international standards and practices at international level: 2. The Parties shall also endeavour to cooperate in relevant international bodies and forums, and endeavour to strengthen global cyber resilience and enhance the ability of third countries to fight cybercrime effectively. 80 EU–UK TCA, art 782. 81 European Commission, ‘Commission Implementing Decision of 28.6.2021 pursuant to Regulation (EU) 2016/679 of the European Parliament and of the Council on the Adequate Protection of Personal Data by the United Kingdom’ (EU–UK Adequacy Decision) C (2021) 4800 final. 82 UK Government, ‘The Benefits of Brexit: How the UK Is Taking Advantage of Leaving the EU’ (31 January 2022) www​.gov​.uk/​government/​publications/​the​-benefits​-of​-brexit accessed 31 March 2023; Lord Frost, ‘Statement to the House of Lords: 16 September 2021’ (Oral statement to UK Parliament); See Michael Cross, ‘Government Plans Bonfire of Retained EU Law’ (The Law Society Gazette, 17 September 2021) lawgazette​.co​.uk/​law/​government​-plans​-bonfire​-of​-retained​-eu​ -law/​5109831​.article accessed 31 March 2023; Catherine Barnard, ‘Retained EU Law and Brexit Opportunities’ (UK in a Changing World, 14 February 2022) https://​ukandeu​.ac​.uk/​retained​-eu​-law​-and​ -brexit​-opportunities/​accessed 31 March 2023; Catherine Barnard, ‘Retained EU Law in the UK Legal Orders: Continuity between the Old and the New’ in Adam Lazowski and Adam Cygan (eds), Research Handbook on Legal Aspects of Brexit (Edward Elgar Publishing 2022, forthcoming). 83 ‘Data Case Defeat Increases Tory Pressure to Quit ECHR’ Financial Times (13 February 2020) www​.ft​.com/​content/​7daf254e​-4e5c​-11ea​-95a0​-43d18ec715f5 accessed 31 March 2023; ‘UK Must Leave European Convention on Human Rights, Says Theresa May’ The Guardian (25 April 2016) www​ .theguardian​.com/​politics/​2016/​apr/​25/​uk​-must​-leave​-european​-convention​-on​-human​-rights​-theresa​ -may​-eu​-referendum accessed 31 March 2023. 84 See Bradford (n 6) ch 5. 79

132  Research handbook on digital trade for the adequacy decision and the TCA itself. The ‘exit’ from the GDPR has been proposed as part of a bonfire of retained EU law in the UK legal order. Additionally, the UK appeared to engage in a form of half-hearted gesture to develop a British alternative to the GDPR, setting up an expert advisory body comprised predominantly of industry relating to Big Tech and a small number of individuals but largely lacking independent academics. The irony of the place of Big Tech here cannot be lost in view of the decision of Facebook and Google to move all UK account holders under US law in the wake of Brexit.85 At the WTO, the UK has notably been seeking to lead smaller groups but largely has not followed any EU lead or initiative on electronic commerce as part of its intentional and deliberate process of extrication. The UK has repeatedly emphasised how ‘global Britain’, as a motto or slogan, was defined by its new capacity for regulatory divergence in areas of data regulation and governance, for example as to AI and data flows, in ways which being part of the EU had supposedly inhibited.86 The place of digital trade and data flows remains to be seen: as a barometer of first-mover advantage, innovation or simply rancour. Initial appearances are, however, that the UK–EU relationship stands markedly apart from the TTC bilateral arrangement. It signals perhaps muted ambitions in digital trade, and possibly more downward pressure on standards.

6. CONCLUSIONS Digital trade forms a key case study of the EU and its partners’ development of institutional design and the rollout of high binding standards on data flows, previously neglected in international economic law. With mostly developed third countries, the EU advocates regulatory cooperation predicated upon international standards and seeks regulatory cooperation also in the absence of global consensus of standards. This has formed part of the EU’s key push for trade in the post-Lisbon era, striving to set the global rulebook. The global legal order appears increasingly fragmented as between multiple regimes in digital trade with respect to data flows – split between the EU and many other forms of regime. The high standards of the EU represent an important counter-weight to a variety of global actors promoting economic standards over individual rights. The WTO appears unlikely to constitute the forum for significant change here for now. Where international leadership in this domain will come from in an era of soft law dominance remains to be seen. The EU and US TTC in theory represents a significant bilateral development, not least its emphases in parallel and related developments as to binding principles and a court; however, its broader impact remains to be seen. The gap between the TTC and Brexit Britain as a bilateral partnership has been shown to be both significant and distinctive. The case study of the EU is distinctive as an actor shifting away from its traditionally middle-ground position. The EU thus constitutes an important study of the hardening of its rules, values and principles in an era of soft law proliferation.

85 ‘Facebook to Move all UK Users onto US Agreements’ BBC (16 December 2020) www​.bbc​.co​ .uk/​news/​business​-55328376 accessed 31 March 2023. 86 See UK Government, ‘The Benefits of Brexit: How the UK Is Taking Advantage of Leaving the EU’ (n 80); Barnard, ‘Retained EU Law and Brexit Opportunities’ (n 80).

9. Federalism and digital trade Patrick Leblond1

1. INTRODUCTION Impediments to digital trade are no different than traditional trade barriers in that they take two main forms: tariff and non-tariff.2 The World Trade Organization (WTO) defines tariffs as ‘customs duties on merchandise imports’, which ‘give a price advantage to locally-produced goods over similar goods which are imported, and they raise revenues for governments’.3 So, tariffs apply to physical goods, whether they are physically or digitally ordered across borders.4 However, they could also apply to ‘digital goods’ such as songs, videos, movies or software when they are downloaded from the internet.5 In 1998, WTO members agreed to a moratorium on imposing customs duties on electronic transmission.6 This moratorium

The author would like to thank Hesameddin Abbaspour Tazehkand for his assistance with the chapter’s research and the Social Sciences and Humanities Research Council of Canada (Partnership Development Grant #890-2016-0118) for its financial support of the research. 2 The Organisation for Economic Cooperation and Development (OECD), the World Trade Organization (WTO) and the International Monetary Fund (IMF), in their joint framework on measuring digital trade, define digital trade as ‘all trade that is digitally ordered and/or digitally delivered’. Digitally ordered trade corresponds to the ‘sale or purchase of a good or service, conducted over computer networks by methods specifically designed for the purpose of receiving or placing orders’ while digitally delivered trade represents ‘transactions that are delivered remotely in an electronic format, using computer networks specifically designed for the purpose’. Organisation for Economic Cooperation and Development, World Trade Organization and International Monetary Fund, Handbook on Measuring Digital Trade (Version 1, 2020) 11. 3 World Trade Organization, ‘Tariffs’, www​.wto​.org/​english/​tratop​_e/​tariffs​_e/​tariffs​_e​.htm accessed 18 May 2022. 4 In addition to regular tariffs, special trade-defence custom duties can affect digital trade: anti-dumping duties, countervailing duties and safeguard measures. Martina Francesca Ferracane, Hosuk Lee-Makiyama and Erik van der Marel, Digital Trade Restrictiveness Index (European Centre for International Political Economy, April 2018). 5 According to the OECD–WTO–IMF framework for measuring digital trade, goods cannot be delivered in digital form, only services can: ‘by this definition, digital equivalents of goods – such as e-books or digital software – would be considered as the delivery of a licence to use the product and not physical ownership of the product’. OECD et al. (n 2); Swapna Nair, Choosing the Right Ruler: Approaches to Measuring Digital Trade (Impact Paper, The Conference Board of Canada, 17 November 2021) 14. The logic is that ‘digital goods’ such as books, movies, songs or software that are downloaded (after being purchased) onto an electronic device can usually not be transferred (for free or a price) to someone else (unless it is done illegally). The advent of so-called non-fungible tokens (NFTs), where a digital good has a specific identity and associated ownership registered on a blockchain, is likely to change this definitional approach. This is because digital goods registered as NFTs can have their ownership transferred to someone else. 6 Declaration on Global Electronic Commerce (World Trade Organization) [1998] WT/MIN(98)/ DEC/2. 1

133

134  Research handbook on digital trade has been in effect ever since, having been regularly renewed.7 But what if a subnational government decides to impose a tax on digitally delivered services imported from outside its jurisdiction? If the tax also applies in a non-discriminatory way to digitally delivered services provided by residents, then it is not a problem; it is not akin to a tariff. If the tax only applies to external service providers, then it would be equivalent to a tariff, but one that applies only at the borders of the subnational entity (for example, a province or state). With physical goods, such a situation does not arise as the good must ‘clear customs’ at a country’s national borders. Subnational jurisdictions do not usually control the entry of goods for tax purposes at their borders. A discriminatory tax imposed by a subnational government on digitally delivered services provided by a non-resident would contravene trade agreements to which the country is a party (for example, see articles 14.3, para 2 and 14.4, para 1 in the Comprehensive and Progressive Agreement on Trans-Pacific Partnership [CPTPP]). This could lead to a dispute initiated by another party to the trade agreement if the tax is maintained; however, the dispute would only engage the national government as the subnational one is not party to the trade agreement.8 The issue then becomes whether the national government has the constitutional authority to force the subnational government to put an end to the discriminatory tax. If not, the dispute could lead to retaliatory duties imposed by the aggrieved party, which would not resolve the issue and would ultimately end up hurting both parties. More likely, given the relatively small sums involved, no dispute-settlement process would be launched and the subnational government would just continue imposing its discriminatory tax on digitally delivered services provided by non-residents. What about non-tariff barriers (NTBs)? Could a similar scenario involving subnational governments occur? Susan Aaronson lists the following NTBs to international digital trade: localization requirements (data storage, processing or other digital trade activities), other data flow restrictions (for example, privacy protection), infringement of intellectual property rights, filtering/blocking of websites or applications, cybersecurity (too little or too much), other national standards and conformity assessments (for example, obligation to divulge source codes or other trade secrets).9 To this list, one could add barriers to foreign technologies, preferential subsidies, restrictions on foreign investment, local presence requirements and buy-local requirements in public procurement contracts.10 Presumably, subnational governments could adapt several of these NTBs to digital trade. For example, the province of Quebec in Canada has adopted its own personal information protection regime, which is modelled after the EU’s General Data Protection Regulation (GDPR). It also has requirements that cloud services used by the government localize the data in Canada. In the United States (US), an increasing number of states are adopting their own privacy protection regimes in the absence of a national one. These policies and regimes could potentially have a negative impact on digital trade and even contravene provisions found in trade agreements that their country has signed on to.

The moratorium was last extended in June 2022. Ministerial Decision on Work Programme on Electronic Commerce (World Trade Organization) [22 June 2022] WT/L/1143. 8 Australia’s states and territories as well as Canada’s provinces and territories are not signatories to the CPTPP. 9 Susan Ariel Aaronson, ‘What Are We Talking about When We Talk about Digital Protectionism?’ (2019) 18 World Trade Review 541, table 1. 10 Ferracane et al. (n 4). 7

Federalism and digital trade  135 So, the above discussion makes clear that federalism and multilevel governance are something that digital trade scholars must pay attention to. Yet, this is an area of research that has been neglected so far.11 An important question to answer is whether federal systems are more prone to digital trade protectionism than non-federal ones. Presumably, provincial/ state/territorial governments could adopt rules and regulations that run counter to the federal government’s international commitments on digital trade through trade agreements. The adoption of incoherent laws and regulations affecting digital trade by federal and provincial/ state governments should lead businesses to engage less in digital trade than they otherwise would because of higher compliance costs associated with such trade. For instance, a study of Swedish enterprises from a wide range of sectors found that the ability to move data across borders easily was crucial for the functioning of firms’ global value chains.12 Another study found that restrictions on cross-border data flows are associated with lower imports of data-intensive services.13 The extent to which there is incoherence between national and subnational government policies should, however, depend on the degree of the federal architecture’s institutionalization (that is, modes of interaction in intergovernmental relations).14 A related question pertains to the ability of digital trade provisions in international trade agreements to address policies adopted by subnational governments to regulate data and the digital economy. To begin with, the effectiveness of digital trade provisions in international trade agreements in preventing protectionist policies adopted by national governments remains to be established.15 If one adds the challenges that international trade agreements have in addressing policies adopted by subnational governments, then we can expect federal systems to be generally more prone to digital protectionism and, therefore, to engage in less digital trade internationally.16 11 There is, however, an emerging literature on international trade and federalism, multilevel politics and multilevel governance: Christopher J Kukucha, The Provinces and Canadian Foreign Trade Policy (UBC Press 2008); Patrick Fafard and Patrick Leblond, ‘Twenty-First Century Trade Agreements: Challenges for Canadian Federalism’ (The Federal Idea 2012); Christopher J Kukucha, ‘Canadian Sub-Federal Governments and CETA: Overarching Themes and Future Trends’ (2013) 68 International Journal 528; J Anthony VanDuzer and Melanie J Mallet, ‘Compliance with Canada’s Trade and Investment Treaty Obligations: Addressing the Gap between Provincial Action and Federal Responsibility’ (2016) 54 Alberta Law Review 89; Michelle Egan and Maria Helena Guimarães, ‘The Dynamics of Federalism, Subnational Markets and Trade Policy-Making in Canada and the US’ (2019) 29 Regional & Federal Studies 459; Jörg Broschek and Patricia Goff (eds), The Multilevel Politics of Trade (University of Toronto Press 2020); Christian Freudlsperger, Trade Policy in Multilevel Government: Organizing Openness (Oxford University Press 2020); Stéphane Paquin, ‘Trade Paradiplomacy and the Politics of International Economic Law: The Inclusion of Quebec and the Exclusion of Wallonia in the CETA Negotiations’ (2022) 27 New Political Economy 597. 12 Magnus Rentzhog, No Transfer, No Production — A Report on Cross-Border Data Transfers, Global Value Chains, and the Production of Goods (National Board of Trade Sweden [Kommerskollegium] 2015). 13 Erik van der Marel and Martina Francesca Ferracane, ‘Do Data Policy Restrictions Inhibit Trade in Services?’ (2021) 157 Review of World Economics 727. 14 Jörg Broschek and Patricia Goff, ‘Introduction: The Evolution of Multilevel Trade Politics’ in Broschek and Goff (n 11). 15 Patrick Leblond, ‘Uploading CPTPP and USMCA Provisions to the WTO’s Digital Trade Negotiations Poses Challenges for National Data Regulation: Example from Canada’ in Mira Burri (ed) Big Data and Global Trade Law (Cambridge University Press 2021). 16 On international trade agreements’ ability to address subnational impediments to cross-border trade: Fafard and Leblond (n 11); VanDuzer and Mallet (n 11); Christian Hederer and Patrick Leblond,

136  Research handbook on digital trade This chapter is a first attempt to answer these questions and understand how federalism and multilevel governance affect international digital trade. To do so, it begins by examining the relationship between federalism and digital trade in Canada and the United States, focusing on policies adopted by subnational governments that affect digital trade. After that, the chapter analyses how digital trade provisions in the United States–Mexico–Canada Agreement (USMCA) – as it applies to both Canada and the US – impose constraints on subnational governments’ ability to adopt policy measures that could impede cross-border digital trade.

2.

FEDERALISM AND DIGITAL TRADE IN CANADA AND THE UNITED STATES

This section examines some of the laws and regulations recently adopted by subnational governments in Canada and the United States that could have an impact on digital trade by acting as (potential) tariff and non-tariff barriers to such trade. The goal herein is not to provide an exhaustive list of all laws and regulations adopted by provincial or state governments that could affect their country’s digital trade with the rest of world, but to convey the importance that subnational policies have for digital trade and its governance. 2.1 Canada In Canada, there are five main areas where provincial governments’ policies could impede international digital trade: personal information (privacy) protection, which may include local data storage and processing requirements; buy-local requirements for digitally delivered services; subsidies for local providers of digitally delivered services; taxation.17 With respect to the protection of personal information protection, several Canadian provinces have adopted laws to protect personal information. British Columbia, for instance, has adopted the Freedom of Information and Protection of Privacy Act (FOIPPA), which covers public organizations. Concerning digital trade, the FOIPPA requires that ‘[a] public body must protect personal information in its custody or under its control by making reasonable security arrangements against such risks as unauthorized collection, use, disclosure or disposal’.18 Before 25 November 2021, to protect personal information held by public bodies in British Columbia, FOIPPA specifically required that the information could only be stored and accessed in Canada.19 This data localization requirement prevented data-storage and/or processing firms located outside Canada from offering their services to public bodies in British

‘Implementation of Twenty-First-Century Trade Agreements in Canada: CETA and Intergovernmental Cooperation’ in Broschek and Goff (n 11). 17 The focus herein is on provinces because territories have a more limited range of competencies. 18 Freedom of Information and Protection of Privacy Act [RSBC 1996, c 165], s 30. 19 Before it was repealed by the Freedom of Information and Protection of Privacy Amendment Act 2021, section 30.1 of the FOIPPA stated: ‘A public body must ensure that personal information in its custody or under its control is stored only in Canada and accessed only in Canada, unless one of the following applies: (a) if the individual the information is about has identified the information and has consented, in the prescribed manner, to it being stored in or accessed from, as applicable, another jurisdiction; (b) if it is stored in or accessed from another jurisdiction for the purpose of disclosure allowed under this Act; (c) if it was disclosed under section 33.1 (1) (i.1).’

Federalism and digital trade  137 Columbia unless they had servers and/or other computing facilities located in Canada. The FOIPPA’s requirement, therefore, represented an impediment to digital trade between Canada and the rest of the world. The FOIPPA’s data localization provision raised an important concern with respect to cloud services. Although the personal data of British Columbia residents were stored in Canada, it could be routed outside the country (that is, via the United States) when accessed by users (for example, if the data is stored in another province). Michael Power describes the situation as follows: When clients use a cloud-based service, the transmitted packets are usually encrypted but require decryption at servers to determine where they need to be forwarded. Once that is determine, the packets are re-encrypted for onward transmission. […] The decryption, reading of routing information, and re-encryption occurs at each Internet node and is usually automated with no human intervention and no human view of the decrypted routing information. However, one can argue that the decryption at a server for routing constitutes ‘access’ for the purpose of FOIPPA’s s. 30.1 (personal information may only accessed in Canada). If decryption constitutes access, or the ability to provide access, at a node and it occurs outside of Canada then, arguably, there is a violation of s. 30.1.20

This possible violation of FOIPPA’s section 30.1 limited the ability of British Columbia public bodies to use cloud-based services, even if the data were stored on servers in Canada. As a result, services to British Columbia residents and the security of their personal information could be compromised if public bodies were not able to procure digitally delivered services based on the latest technology.21 To address the above concerns, the British Columbian government adopted the Freedom of Information and Protection of Privacy Amendment Act 2021, which entered into force on 25 November 2021. As a result, the FOIPPA’s data localization requirement (section 30.1) was repealed and replaced by new provisions regarding the disclosure of personal information.22 The FOIPPA’s amended section 33 now provides a long list of circumstances in which disclosure is permitted. With respect to cloud-based storage, disclosure is allowed ‘if the disclosure is necessary for the processing of information and the following apply: (i) the processing does not involve the intentional accessing of the information by an individual; (ii) any processing done outside of Canada is temporary’.23 Disclosure is also permitted ‘if the information is metadata and the following apply: (i) the metadata is generated by an electronic system; (ii) the metadata describes an individual’s interaction with the electronic system; (iii) if practicable, information in individually identifiable form has been removed from the metadata or destroyed’.24 Finally, section 33.1 was amended to give the minister responsible for the FOIPPA the authority to allow the disclosure of personal information outside of Canada by a public body through regulations, presumably to provide flexibility in contracting services for data storage and/or processing as technologies evolve over time.

Michael Power, ‘BC Amends Privacy Statute to Facilitate Canadian Cloud Computing’ (LinkedIn, 1 November 2019) www​.linkedin​.com/​pulse/​bc​-amends​-privacy​-statute​-facilitate​-canadian​-cloud​ -computing​-power/​accessed 20 May 2022. 21 ibid. 22 Freedom of Information and Protection of Privacy Act [RSBC 1996, c 165], s 33. 23 ibid, s 33(2)(u). 24 ibid, s 33(2)(v). 20

138  Research handbook on digital trade In September 2021, Quebec’s National Assembly adopted Bill 64 to update its existing law protecting personal information in the private sector.25 The new law, whose main provisions will come into effect in September 2023, draws its inspiration from the EU’s GDPR but doesn’t follow it to the letter. With respect to cross-border data flows, which is most relevant to digital trade, the GDPR allows firms to transfer personal information outside the EU if the individual has given explicit consent to do so or if the European Commission has deemed the jurisdiction to which the information is being transferred to be ‘adequate’ in providing a level of protection similar in nature to the GDPR’s.26 For its part, Bill 64 offers no such options for allowing the transfer of personal information outside Quebec’s borders. According to the new law, such transfers are only possible if (1) a business operating in the province of Quebec has informed individuals that their personal information might be transferred outside the province’s borders, (2) it has conducted a privacy impact assessment (PIA) and (3) it has a written agreement with the third party to which the personal information is transferred, including measures to mitigate risks to privacy protection identified in the PIA.27 According to Karine Joizil et al., ‘unlike with GDPR, the burden to determine if a jurisdiction is adequate now falls solely on the shoulders of organization making the cross-border data transfer’.28 These new requirements regarding the transfer of personal information outside Quebec’s borders will affect the cost of conducting digital trade for businesses operating in the province, especially if they differ from those imposed by other jurisdictions.29 Consequently, some businesses might decide to stop engaging in digital trade to and from the province.30

An Act to modernize legislative provisions as regards the protection of personal information [Bill

25

64].

https://​ec​.europa​.eu/​info/​law/​law​-topic/​data​-protection/​international​-dimension​-data​-protection/​ adequacy​-decisions​_en/​accessed 27 July 2022. 27 Karine Joizil, Michael Scherman, Francis Langlois and Philippe April, ‘Quebec’s Bill 64 Introduces New Operational Requirements for Cross-Border Transfers of Personal Information’ (McCarthy Tetrault, 26 October 2021) www​.mccarthy​.ca/​en/​insights/​blogs/​techlex/​quebecs​-bill​-64​ -introduces​-new​-operational​-requirements​-cross​-border​-transfers​-personal​-information accessed 27 July 2022. 28 ibid. 29 The Canadian federal government introduced Bill C-27 in parliament on 16 June 2022: An Act to enact the Consumer Privacy Protection Act, the Personal Information and Data Protection Tribunal Act and the Artificial Intelligence and Data Act and to make consequential and related amendments to other Acts [Bill C-27] www​.parl​.ca/​legisinfo/​en/​bill/​44​-1/​c​-27 accessed 27 July 2022. At the time of writing, Bill C-27 does not include requirements along the lines of Quebec’s Bill 64 for the transfer of personal information outside Canada’s borders. The only stated requirement is that the level of protection be equivalent, without any indication as to how such equivalence is to be established. Kirsten Thompson, ‘Canada’s New Federal Privacy Bill C-27 – Summary of Significant Impacts and New Proposals’ (Dentons, 20 June 2022) www​.acc​.com/​sites/​default/​files/​resources/​upload/​canadas​-new​ -federal​-privacy​-bill​-c27​-summary​-of​-significa​.pdf accessed 27 July 2022. 30 The provinces of Alberta and British Columbia are considering updating their private-sector privacy laws, which are outdated. For its part, the province of Ontario is looking at adopting its own law to protect personal information held by the private sector, launching a public consultation process in the summer of 2021. If these new laws are aligned with Quebec’s Bill 64, especially with respect to cross-border data requirements, they would help mitigate the risk that some businesses abandon the Quebec market for digital trade. Murad Hemmadi and David Reevely, ‘Provinces Moving to Update Private-Sector Privacy Rules as Federal Legislation Stalls’ (The Logic, 9 August 2021) https://​thelogic​.co/​ news/​provinces​-moving​-to​-update​-private​-sector​-privacy​-rules​-as​-federal​-legislation​-stalls/​  accessed 27 July 2022. 26

Federalism and digital trade  139 Regarding the purchase of digitally delivered services, the Quebec government announced in February 2019 that, to save money, it would contract out 80 per cent of its data hosting and management operations to private firms offering cloud services. The service providers’ origin does not matter as long as the data are hosted and processed on servers located in Canada (preferably in Quebec).31 This data localization requirement is similar to what the British Columbian government had in place to protect personal information held by public bodies before it updated the FOIPPA in 2021. It is also in line with the Government of Canada’s policies towards data storage and processing.32 In January 2021, it was announced that one third of identified cloud-service providers had their headquarters in Quebec.33 However, in terms of value, 85 per cent of cloud contracts had gone to Microsoft, IBM and Amazon in 2020.34 Therefore, the Quebec government’s data localization requirements have not prevented foreign-based cloud-service providers from obtaining the larger share of contracts in value terms. On 1 January 2019, the province of Quebec began applying its provincial sales tax to digitally delivered services provided from outside Canada.35 On 1 September 2019, the tax’s application was extended to non-residents from the rest of Canada. The objective was to create a level playing field between resident and non-resident providers. Before, only digital service providers located within the province were obligated to collect the tax from their customers and remit it to the provincial government, because they were under the obligation to register their business with Revenu Québec (the province’s tax authority). This gave non-resident providers an unfair price advantage vis-à-vis their resident competitors. The province of Saskatchewan did the same on 1 January 2020, followed by British Columbia on 1 April 2021 and Manitoba on 1 December 2021.36 Even if the requirements that non-resident providers of digitally delivered services register, collect and remit provincial sales taxes are meant to create a level playing field for provincially based providers, some non-resident service providers could decide to stop offering their services in those provinces, likely the ones with smaller populations, because the administrative costs make doing business there unprofitable. As such, provincial taxes on digitally delivered services by non-resident providers represent an obstacle to digital trade in those provinces. To conclude on the Canadian case, privacy protection, public procurement and tax policies adopted by provincial governments in Canada represent potential impediments to digital trade, within Canada as well as internationally. The absence of digital trade statistics makes it Martin Croteau, ‘Nos données livrées au renseignement américain ?’ (La Presse+, 5 February 2019) https://​plus​.lapresse​.ca/​screens/​b23f54dd​-b9e3​-4cc7​-9762​-b82dbf88cebf​_​_7C​_​_​_0​.html accessed 19 May 2022. 32 Government of Canada, ‘Directive on Service and Digital’ (1 April 2020), s 4.4.3.14 www​.tbs​-sct​ .canada​.ca/​pol/​doc​-eng​.aspx​?id​=​32601 accessed 28 July 2022. 33 Karim Benessaieh, ‘Une entreprise sur trois est québécoise’ (La Presse, 5 January 2021) www​ .lapresse​.ca/​affaires/​entreprises/​2021​-01​-05/​stockage​-des​-donnees​-gouvernementales/​une​-entreprise​ -sur​-trois​-est​-quebecoise​.php accessed 28 July 2022. 34 Ulysse Bergeron, ‘La transition numérique de Québec profite surtout aux géants américains’ (Le Devoir, 12 May 2021) www​.ledevoir​.com/​politique/​quebec/​600565/​infonuagique​-quebec​-a​-octroye​-85​ -des​-contrats​-sans​-appel​-d​-offres​-a​-trois​-geants​-americains accessed 28 July 2022. 35 The Government of Quebec required non-resident providers of digitally ordered goods and digitally delivered services to register for the Quebec sales tax. 36 Brian Morcombe, ‘Digital Sales Tax in Canada’ (BDO Canada, 18 October 2021) www​.bdo​.ca/​en​ -ca/​insights/​tax/​tax​-articles/​digital​-sales​-tax​-in​-canada/​ accessed 19 May 2022. 31

140  Research handbook on digital trade impossible to determine the impact that these policies have actually had on Canada’s digital trade; however, some anecdotal evidence with respect to cloud services suggests that the larger foreign providers are able to satisfy the requirements imposed by provincial governments. Whether small- and medium-sized enterprises (SMEs) can also do so remains to be established. 2.2

United States

Unlike in Canada, there is no federal-level law to protect personal information held by private-sector organizations in the US.37 It has been left to the states to adopt such laws to protect their residents.38 There has been rapid growth in state-level comprehensive privacy legislation since 2018, although only a few have been adopted to date.39 In 2018, California was the first US state to adopt a comprehensive law to protect individuals’ privacy.40 The California Consumer Privacy Act (CCPA) came into force on 1 January 2020. At the time of writing (July 2022) it is the only such law in force, as the other state-level legislation that has been adopted is to become effective in 2023. As a result, we limit our focus to the CCPA in terms of how it might affect US digital trade. The CCPA applies to enterprises that do business in California and process the personal information of the state’s residents, regardless of their base of operation. This means that firms located outside California but that do business with California residents must comply with the CCPA’s provisions. As other states’ comprehensive privacy and data protection laws come into force, businesses conducting digital trade in the US are likely to see their compliance costs increase, especially if there are significant differences between the various state-level legislations. With respect to cross-border data transfers, the CCPA does not impose any restriction on such transfers. The only requirement is that the business that collects personal information and sells or shares it with a third party, including a service provider, enters into a written agreement with this third party.41 Notably, this agreement must specify ‘that the personal information is sold or disclosed by the business only for limited and specified purposes’ and that the third party has to comply with the applicable obligations under the CCPA in order to provide the same level of privacy protection. Therefore, the CCPA does not represent an impediment to digital trade when it comes to transferring personal information outside the state’s borders.

There are some federal laws and regulatory provisions regarding the protection of personal information in particular circumstances or sectors (e.g., telecommunications). Thorin Klosowski, ‘The State of Consumer Data Privacy Laws in the US (And Why It Matters)’ (The New York Times, 6 September 2021) www​.nytimes​.com/​wirecutter/​blog/​state​-of​-privacy​-laws​-in​-us/​ accessed 28 July 2022. 38 All 50 states have laws requiring private businesses to notify individuals when there has been a security breach of their personal information. ‘Security Breach Notification Laws’ (National Conference of State Legislatures, 17 January 2022) www​.ncsl​.org/​research/​telecommunications​-and​ -information​-technology/​security​-breach​-notification​-laws​.aspx accessed 28 July 2022. 39 ‘The Growth of State Privacy Legislation’ (International Association of Privacy Professionals, November 2021) https://​iapp​.org/​resources/​article/​the​-growth​-of​-state​-privacy​-legislation​-infographic/​ accessed 28 July 2022; Taylor Kay Lively, ‘US State Privacy Legislation Tracker’ (International Association of Privacy Professionals, 7 July 2022) https://​iapp​.org/​resources/​article/​us​-state​-privacy​ -legislation​-tracker/​accessed 28 July 2022. 40 TITLE 1.81.5. California Consumer Privacy Act of 2018 [1798.100–1798.199.100]. 41 ibid, s 1798.100(d). 37

Federalism and digital trade  141 Compliance with multifarious state laws protecting privacy and data could end up being costly for firms conducting digital trade across the US. This is why a number of digital firms – including Big Tech ones such as Apple, Google, Meta and Microsoft – have been supportive of a federal law to protect individuals’ privacy: ‘A number of tech companies are embracing federal legislation, to a point, because would otherwise have to abide by a hodgepodge of state laws.’42 After years of the US Congress unsuccessfully trying to pass a law that would protect personal information collected and used by the private sector, a bipartisan bill created some hope for supporters of a federal privacy law during the summer of 2022.43 The bill, known as the American Data Privacy and Protection Act (ADPPA), was introduced in the House of Representatives on 21 June 2022.44 Like the CCPA, the proposed ADPPA does not impose restrictions on the cross-border flow of personal information. It follows the CCPA’s logic in requiring a written agreement with a service provider to which the data are transferred, making it clear that ultimate responsibility for ensuring that data are protected remains with the ‘covered entity’.45 Therefore, the ADPPA does not represent a potential impediment to US digital trade. On the contrary, it could make digital trade easier by superseding state laws and thus reducing compliance costs for businesses. This is because the proposed bill would pre-empt privacy and data protection laws enacted by US states, with some exceptions: ‘No State or political subdivision of a State may adopt, maintain, enforce, or continue in effect any law, regulation, rule, standard, requirement, or other provision having the force and effect of law of any State, or political subdivision of a State, covered by the provisions of this Act, or a rule, regulation, or requirement promulgated under this Act.’46 Unlike in Canada, the US federal government has the legal authority, even in areas of shared competence, to adopt legislation that supersedes state-level laws (for example, under the commerce clause).47 This pre-emption clause proved controversial. On the one hand, some argue that pre-emption does not go far enough because of the long list of exceptions.48 On the other hand, US states argue that Congress should set a floor for states to build upon and innovate: ‘A federal legal framework for privacy protections must allow flexibility to keep Danielle Abril, ‘This Is What Tech Companies Want in Any Federal Data Privacy Legislation’ (Fortune, 21 February 2019) https://​fortune​.com/​2019/​02/​21/​technology​-companies​-federal​-data​ -privacy​-law/​ accessed 28 July 2022. See also Sam Pfeifle, ‘US Federal Privacy Law? Apple, Google, Facebook, Microsoft All Hope So’ (International Association of Privacy Professionals, 25 October 2018) accessed 28 July 2022. 43 Rebecca Kern, ‘Bipartisan Draft Bill Breaks Stalemate on Federal Data Privacy Negotiations’ (Politico, 3 June 2022) www​.politico​.com/​news/​2022/​06/​03/​bipartisan​-draft​-bill​-breaks​-stalemate​-on​ -federal​-privacy​-bill​-negotiations​-00037092 accessed 28 July 2022. 44 American Data Privacy and Protection Act [H.R.8152] www​.congress​.gov/​bill/​117th​-congress/​ house​-bill/​8152 accessed 28 July 2022. 45 ibid, s 302(b). 46 ibid, s 404(b). 47 ‘Subsequently, the US dormant commerce clause sought to prohibit individual states from passing legislation that discriminates against or overly burdens interstate commerce, while Canadian constitutional protections prohibiting provincial trade barriers have been interpreted more narrowly to include only customs duties.’ Egan and Guimarães (n 11) 462. 48 Daniel Castro, ‘Review of the Proposed “American Data Privacy and Protection Act,” Part 1: State Preemption and Private Right of Action’ (Information Technology & Innovation Foundation, 6 June 2022) https://​itif​.org/​publications/​2022/​06/​06/​american​-data​-privacy​-and​-protection​-act​-review​-part​-1​ -state​-preemption​-and​-private​-right​-of​-action/​ accessed 28 July 2022. 42

142  Research handbook on digital trade pace with technology; this is best accomplished by federal legislation that respects – and does not preempt – more rigorous and protective state laws.’49 The Electronic Frontier Foundation went further by claiming that the ADPPA was ‘rolling back state privacy protections to meet a lower federal standard’.50 In terms of government procurement for cloud services, US state and local governments do not appear to have explicit requirements for using US-based providers or storing personal data locally (that is, in the US). They seem to follow the federal government’s lead when it comes to security and data protection. For instance, they predominantly rely on providers that are FedRamp authorized.51 FedRamp is the Federal Risk and Authorization Management Program that ‘provides a standardized approach to security authorizations for Cloud Service Offerings’.52 In its latest cloud strategy, which refers explicitly to the FedRamp process, the US federal government does not mention using US-based providers or storing personal or sensitive data on US territory.53 The following sums up the recommended approach, which is case by case: To ensure continuity of information security during and after the migration process, it is incumbent upon agencies to thoroughly assess their operational, policy, and business requirements and advocate for themselves when brokering new arrangements with cloud service providers. Regardless of provider type – commercial or federal – agencies should consider having agreements in place with all providers regarding access to, and use of, log data given its importance in effectively conducting information security operations. Moreover, as each agency is the custodian of its information on behalf of the public, each agency is responsible for determining its own governance model for cloud-hosted data that aligns with its identity and credential management systems. To that end, where a cloud solution is deployed by a vendor, a Service Level Agreement (SLA) should be in place that provides the agency with continuous awareness of the confidentiality, integrity, and availability of its information.

The US approach to cloud-computing services for governments is much more integrated and coordinated than in Canada, where provincial governments act independently of the federal government. Like with Canadian provinces, several US states require online sellers of (digitally ordered) products and services (digitally ordered and delivered) to collect applicable sales taxes.54 This was made possible by the US Supreme Court’s 2018 Wayfair decision, which ‘says that states can require that out-of-state online sellers collect and remit sales tax, regardless of whether

49 Rob Bonta et al., Letter to Congressional Leaders, Office of the Attorney General, State of California, 19 July 2022 https://​oag​.ca​.gov/​system/​files/​attachments/​press​-docs/​Letter​%20to​ %20Congress​%20re​%20Federal​%20Privacy​.pdf accessed 28 July 2022. 50 Hayley Tsukayama, Adam Schwartz, India McKinney and Lee Tien, ‘Americans Deserve More than the Current American Data Privacy Protection Act’ (Electronic Frontier Foundation, 24 July 2022) www​.eff​.org/​es/​deeplinks/​2022/​07/​americans​-deserve​-more​-current​-american​-data​-privacy​-protection​ -act accessed 28 July 2022. 51 ‘FedRAMP Survey Results Report’ (Maximus/Genesys and Market Connections, 2020) https://​ maximus​.com/​sites/​default/​files/​documents/​FedRAMP​_Survey​_Results​.pdf accessed 29 July 2022. 52 www​.fedramp​.gov accessed 29 July 2022. 53 CIO Council, ‘From Cloud First to Cloud Smart’ (Federal Cloud Computing Strategy, 2019) https://​cloud​.cio​.gov/​strategy/​#security accessed 29 July 2022. 54 Joshua Sophy, ‘The Ultimate Guide to Internet Sales Tax’ (Small Business Trends, 7 January 2022) https://​smallbiztrends​.com/​2022/​01/​internet​-sales​-tax​.html accessed 28 July 2022.

Federalism and digital trade  143 or not the retailer had a physical presence in that state’.55 Different rules and thresholds (for example, for individual transactions or gross value of transactions over a certain period of time) apply in different states. Therefore, for businesses conducting digital trade with several US states, the costs of complying with state-level sales tax provisions can be extensive and, as a result, constitute an impediment to digital trade. To encourage tax compliance and support interstate commerce, a majority of US states (44) and the District of Columbia have signed on to the Streamlined Sales and Use Tax Agreement (SSUTA), which is ‘an agreement between states that have agreed to simplify their sales and use tax laws to make it easier for out-of-state sellers to comply’.56 To conclude on the US case, digital trade is less affected by subnational policies adopted by US state governments with respect to privacy protection, government procurement of cloud-computing services and taxation. Compared to Canada, there is more coordination between states as well as between state governments and the federal government. The stronger centralization powers of the US federal government compared to the Canadian federal government relative to the provinces would explain this difference in approaches.57 Furthermore, from a political economy perspective, the US federal government has been a proponent of the liberalization of international digital trade, because it benefits large US technology firms. In Canada’s case, the federal government’s approach to digital trade has been more passive and reactive, domestically as well as internationally.58

3.

AMERICAN AND CANADIAN FEDERALISMS AND THE USMCA

The previous section shows that the US federal system is better equipped and more inclined to address potential digital protectionism from subnational governments when compared to the Canadian system. But what about the constraints imposed by international trade agreements? After all, such agreements are generally meant to liberalize cross-border trade and increase market access. It stands to reason that they would impose limits on subnational governments’ ability to adopt policy measures that negatively affect cross-border trade. This section analyses the USMCA’s digital trade provisions to determine the extent to which they apply to both federal and subnational governments.59 The USMCA is the appropriate international trade

ibid. ibid. 57 Martha A Field, ‘The Differing Federalisms of Canada and the United States’ (1992) 55 Law and Contemporary Problems 107; Herman Bakvis and Douglas Brown, ‘Policy Coordination in Federal Systems: Comparing Intergovernmental Processes and Outcomes in Canada and the United States’ (2010) 40 Publius: The Journal of Federalism 484. 58 Rabii Haji and Patrick Leblond, ‘L’ACÉUM et le commerce numérique’, (2022, special issue) Revue québécoise de droit international 197; Patrick Leblond, ‘Is Canada’s digital-trade policy decided in Washington?’ (iPolitics, 2 May) www​.ipolitics​.ca/​opinions/​is​-canadas​-digital​-trade​-policy​-decided​-in​ -washington accessed 29 July 2022. 59 Agreement between the United States of America, the United Mexican States, and Canada [7/1/20 Text] https://​ustr​.gov/​trade​-agreements/​free​-trade​-agreements/​united​-states​-mexico​-canada​-agreement/​ agreement​-between accessed 29 July 2022. 55 56

144  Research handbook on digital trade agreement to examine here because it applies to both Canada and the US, and it has the most extensive provisions to promote cross-border digital trade.60 Regarding the adoption of privacy and data protection laws and regulations by provinces or states, the USMCA’s chapter on digital trade requires its signatories to ‘adopt or maintain a legal framework that provides for the protection of the personal information of the users of digital trade’.61 This legal framework should be based on the following principles: ‘limitation on collection; choice; data quality; purpose specification; use limitation; security safeguards; transparency; individual participation; and accountability.’62 It adds that ‘Each Party shall endeavor to adopt non-discriminatory practices in protecting users of digital trade from personal information protection violations occurring within its jurisdiction’.63 In addition, the USMCA stipulates that the parties ‘recognize the importance of […] ensuring that any restrictions on cross-border flows of personal information are necessary and proportionate to the risks presented’, thereby providing some limit on the extent to which data protection legislation or regulation, whether federal or subnational, can constrain the transfer of (personal) data between the parties.64 Therefore, the adoption of privacy and data protection laws by subnational governments in Canada or the US does not infringe the USMCA’s digital trade provisions; on the contrary, these laws fulfil them. If a particular obligation arising from a subnational privacy and data protection law were deemed by one of parties to not be ‘necessary and proportionate to the risks presented’ (for example, a requirement for private organizations in Canada to obtain explicit consent from individuals before the latter’s data are transferred across the jurisdiction’s borders), then it would be left to a dispute-settlement panel under the USMCA’s chapter 31 to determine if such an obligation were necessary and proportionate to the risks presented. Should a panel rule that it is not, then the problem is how to get the targeted subnational government to comply with the panel’s decision since subnational governments are not signatories to the USMCA (and other international trade agreements).65 In Canada, the federal government lacks effective means to obtain such compliance from provincial and territorial governments.66 The situation is similar in the United States, unless the federal level legislates in a particular policy area in a way that pre-empts state-level laws as in the ADPPA case mentioned above.67 In terms of data localization, the USMCA states that ‘No Party shall require a covered person to use or locate computing facilities in that Party’s territory as a condition for conduct-

Joshua P Meltzer, ‘The United States-Mexico-Canada Agreement: Developing Trade Policy for Digital Trade’ (2019) 11 Trade, Law and Development 239; Haji and Leblond (n 58); Patrick Leblond, ‘The USMCA and Digital Trade in North America’, in Joshua P Meltzer and Brahima S Coulibaly (eds) USMCA Forward 2022: Building a More Competitive, Inclusive, and Sustainable North American Economy (Brookings Institution 2022). 61 Agreement between the United States of America, the United Mexican States, and Canada, art 19.8(2). 62 ibid, art 19.8(3). 63 ibid, art 19.8(4). 64 ibid, art 19.8(3). 65 The federal governments are legally responsible for compliance with international trade agreements by subnational government. 66 Fafard and Leblond (n 11); Hederer and Leblond (n 16). 67 Michelle Egan, ‘Multilevel Trade in the United States: Federalism, Internal Markets, and Intergovernmental Relations’ in Broschek and Goff (n 11). 60

Federalism and digital trade  145 ing business in that territory’.68 However, the USMCA’s digital trade provisions do not apply to government procurement.69 This means that the federal and subnational governments in Canada and the US are allowed under the USMCA to impose data localization requirements in their procurement of cloud-computing services, as the Quebec government now does. Finally, with respect to taxation, the USMCA allows internal taxes on digital trade transactions if they do not discriminate against firms from other USMCA parties. Customs duties on electronic transactions and digital products or services are, however, prohibited.70 Therefore, provincial and state governments are allowed under the USMCA to impose sales taxes on digitally ordered goods and digitally delivered services within their jurisdictions, as long as these taxes apply equally to businesses inside as well as outside their borders. In sum, the USMCA’s digital trade provisions impose few constraints on subnational governments in Canada and the US in terms of policy measures that could negatively affect cross-border digital trade. And in the case that a policy measure could be considered as infringing a USMCA digital trade provision, then the ability of the federal governments in both countries to ensure subnational compliance with a dispute-settlement panel’s decisions against this measure is very limited.

4. CONCLUSION This chapter examines the relationship between federalism and digital trade, a research topic to which scholars have so far paid little attention. Provincial/state/territorial governments adopt laws and regulations that affect digital trade, domestically as well as internationally. Incoherent laws and regulations adopted by federal and provincial/state governments are likely to cause businesses to engage less in digital trade than they otherwise would because of higher compliance costs associated with such trade. The extent to which there is incoherence between national and subnational government policies affecting digital trade should, however, depend on the degree of the federal architecture’s institutionalization (that is, modes of interaction in intergovernmental relations). Similarly, it is also important to examine how digital trade provisions in international trade agreements address policies adopted by subnational governments with respect to the digital economy. To study the relationship between federalism and digital trade, this chapter looks at the cases of Canada and the United States and their membership in the USMCA, which has the most extensive set of digital trade provisions among existing international trade agreements. For each case, the chapter examines three policy areas: privacy and data protection, government procurement of cloud-computing services and taxation of online commercial transactions. Digital trade is less affected by subnational policies adopted by US state governments than those adopted by Canadian provinces. There is also more coordination between states as well as between state governments and the federal government in the US than in Canada, where intergovernmental cooperation on digital economy issues is minimal. The stronger centralization powers of the US federal government compared to the Canadian federal government rel68 Agreement between the United States of America, the United Mexican States, and Canada, art 19.12. 69 ibid., art 19.2(3)(a). 70 ibid., art 19.3.

146  Research handbook on digital trade ative to the provinces would explain this difference in intergovernmental cooperation. Greater support for the liberalization of international digital trade in the US than in Canada could also contribute to explaining the difference in approach between the two North American countries. As for the USMCA’s digital trade provisions, they impose few constraints on subnational governments in Canada and the US in terms of policy measures that could negatively affect cross-border digital trade. And in the case where a policy measure could be considered as infringing a USMCA digital trade provision, then the ability of the federal governments in both countries to ensure subnational compliance with a dispute-settlement panel’s decisions against this measure is very limited. Therefore, coordination between various levels of government in Canada and the US is important to ensure that businesses, most especially SMEs, do not face undue compliance burdens because of a multitude of laws and regulations within federal countries, which would ultimately hurt their ability and willingness to engage in digital trade.

PART III REGIONAL APPROACHES TO DIGITAL TRADE: ASIA

10. China and WTO e-commerce negotiations Henry Gao1

1.

CHINA AND E-COMMERCE

‘Across the Great Wall we can reach every corner of the world.’ Such is the prescient message in the very first email from China, sent on 20 September 1987 by a group of researchers at the Institute for Computer Science of China’s State Commission of Machine Industry to the University of Karlsruhe in Germany.2 However, it was not until 20 April 1994 that the first connection to international network was established by China Education and Research Network, which marked the launch of the internet in China.3 Since then, the Chinese internet has grown by leaps and bounds, despite occasional hiccups such as Google’s exit from China in 2009.4 In 2013, China’s e-commerce volume exceeded 10 trillion RMB and it overtook the US as the largest e-commerce market in the world.5 Nowadays, Chinese e-commerce giants such as Alibaba are among the biggest online retailers in the world and Chinese online shopping festivals such as the Singles Day (11.11) Sale have gained loyal followings all around the world.6 1 This research is supported by the National Research Foundation, Singapore under its Emerging Areas Research Projects (EARP) Funding Initiative. Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not reflect the views of National Research Foundation, Singapore. Parts of the chapter draw from some of the author’s earlier works: Henry Gao, ‘Digital or Trade? The Contrasting Approaches of China and US to Digital Trade’ (2018) 21(2) Journal of International Economic Law 297–321, https://​doi​.org/​10​.1093/​jiel/​jgy015; H Gao, ‘Data Regulation with Chinese Characteristics’ in M. Burri (ed), Big Data and Global Trade Law (Cambridge, Cambridge University Press, 2021), 245–67. doi:10.1017/9781108919234.017; H Gao (2021). ‘Across the Great Wall: E-commerce Joint Statement Initiative Negotiation and China’ in S Peng, C Lin, and T Streinz (eds), Artificial Intelligence and International Economic Law: Disruption, Regulation, and Reconfiguration (Cambridge: Cambridge University Press, 2021), 295–318. 2 W Li, ‘In the beginning...’ (China Daily, 17 March 2008) www​.chinadaily​.com​.cn/​bw/​2008​-03/​ 17/​content​_6540460​.htm (last accessed 10 July 2020). 3 Guowuyuan Xinwen Bangongshi [State Council Information Office], ‘《中国互联网状况》 白皮书 [China’s White Paper on the State of the Internet]’ (SCIO, 8 June 2010) www​.scio​.gov​.cn/​tt/​ Document/​1011194/​1011194​.htm (last accessed 10 July 2020). 4 For a review of the background of the case and the trade law issues it raised, see H Gao, ‘Google’s China Problem: A Case Study on Trade, Technology and Human Rights under the GATS’ (2011) 6 Asian Journal of WTO & International Health Law and Policy 347, at 347–85. 5 ‘我国大宗电子商务交易额已超10万亿元 [China’s Bulk e-Commerce Transaction Value Exceeds 10 Trillion]’ (Zhongguo Caijing Bao [China Financial and Economic News], 7 August 2014) www​.mof​.gov​.cn/​zhengwuxinxi/​caijingshidian/​zgcjb/​201408/​t20140807​_1123621​.html (last accessed 10 July 2020). See also ‘中国电子商务报告（2013） [China e-Commerce Report of 2013]’ (MOFCOM, 23 September 2014) dzsws​.mofcom​.gov​.cn/​article/​ztxx/​ndbg/​201409/​20140900740745​ .shtml (last accessed 10 July 2020). 6 See M Smith, ‘Australian Brands Woo Shoppers at China’s Singles’ Day Sales’ (Financial Review, 12 November 2018) www​.afr​.com/​news/​world/​asia/​australian​-brands​-woo​-shoppers​-at​-chinas​ -singles​-day​-sales​-20181112​-h17sxb (last accessed 10 July 2020); J Lim, ‘Singles’ Day Sales in S[inga]

148

China and WTO e-commerce negotiations  149 Notwithstanding the phenomenal growth in the e-commerce sector, the internet remains under tight regulation in China.7 This started with hardware regulations in the early days of the Chinese internet, which required that all internet connections must go through official gateways sanctioned by the Chinese government. Then the government moved to software regulation and started to require that software used for internet access must be sanctioned by the government. The latest iteration is content and data regulation, which culminated in the introduction of the Cybersecurity Law in 2016, elevating internet regulation to a matter of national security. Internationally, China has engaged with e-commerce regulation at both the multilateral and regional levels. In the WTO, China’s first encounter with data regulation started on the wrong foot as it concerns a sensitive area: China’s regulation of publications and audio-visual products.8 In that case, the US complained that China has failed to grant foreign firms the right to import and distribute publication and audio-visual products. One of the key issues in the case is whether China’s commitments on ‘sound recording distribution services’ cover ‘electronic distribution of sound recordings’ as alleged by the US.9 China disagreed with the US approach and argued instead that such electronic distribution ‘in fact corresponds to network music services’,10 which only emerged in 2001 and were totally different in kind from the ‘sound recording distribution services’. According to China, the most fundamental difference between the two is that, unlike ‘traditional’ sound recording distribution services, network music services ‘do not supply the users with sound recordings in physical form, but supply them with the right to use a musical content’.11 In response, the US cited the Panel’s statement in US–Gambling12 that ‘the GATS does not limit the various technologically possible means of delivery under mode 1’, as well as the principle of ‘technological neutrality’ mentioned in the Work Programme on Electronic Commerce – Progress Report to the General Council,13 and argued that electronic distribution is merely a means of delivery rather than a new type of service.14 Furthermore, the US argued that the term ‘distribution’ encompasses not only the distribution of goods, but also distribution of services.15 After a lengthy discussion canvassing the ordinary meaning, the context, the provisions of the GATS, the object and purpose and pore Doubled from a Year Before: ShopBack’s Data’ (Today Singapore, 12 November 2018) www​ .todayonline​.com/​singapore/​singles​-day​-sales​-spore​-doubled​-year​-shopbacks​-data (last accessed 10 July 2020). 7 For a detailed analysis of the evolution of internet regulation in China, see H Gao, ‘Data Regulation with Chinese Characteristics’ in M. Burri (ed), Big Data and Global Trade Law (Cambridge, Cambridge University Press, 2020), available at https://​ssrn​.com/​abstract​=​3430284 (last accessed 10 July 2020). 8 Panel Report, China – Measures Affecting Trading Rights and Distribution Services for Certain Publications and Audiovisual Entertainment Products, WT/DS363/R and Corr.1, adopted 19 January 2010, as modified by Appellate Body Report WT/DS363/AB/R, DSR 2010:II, 261 [hereinafter ‘Panel Report: China – Publications and Audiovisual Products’]. 9 Ibid, at paras 4.49–4.71. 10 Ibid, at para. 4.147. 11 Ibid, at para. 4.149. 12 Panel Report, United States – Measures Affecting the Cross-Border Supply of Gambling and Betting Services, WT/DS285/R, adopted 20 April 2005, as modified by Appellate Body Report, WT/ DS285/AB/R, DSR 2005:XII, 5797. 13 Work Programme on Electronic Commerce, Progress Report to the General Council, adopted by the 230 Council for the Trade in Services on 19 July 1999, S/L/74, circulated 27 July 1999, at para. 4. 14 Ibid, at para. 4.69. 15 Ibid, at para. 7.1156.

150  Research handbook on digital trade various supplementary means of interpretation, the Panel concluded that the term ‘sound recording distribution services’ does extend to distribution of sound recording through electronic means.16 China appealed the Panel’s findings, but they were upheld by the Appellate Body, which largely adopted the Panel’s reasoning.17 The case was also the first WTO case concerning China’s censorship regime. It is interesting to note, however, that the US did not challenge the censorship regime per se.18 Instead, the US only challenged the alleged discrimination in the operation of the regime, where imported products were subject to more burdensome content review requirements.19 Ironically, the US even proposed, as the solution to the alleged discrimination, that the Chinese government itself shall shoulder the sole responsibility for conducting content review, rather than outsourcing it to importing firms.20 After such an unpleasant experience, China took a cautious approach on the inclusion of internet or data regulation in other fora, such as free trade agreements. While it has signed more than a dozen FTAs so far, most of them have not included provisions on such regulations. The only ones with included stand-alone chapters on e-commerce are the two FTAs China signed with Korea and Australia in 2015,21 as well as the recently upgraded FTAs with Chile22 and Singapore.23 However, unlike the US FTAs, which often include provisions on free flow of data and ban on data localization requirements,24 the four FTA chapters only address e-commerce-related issues such as moratoria on customs duties on electronic transmission; electronic authentication and electronic signature; protection of personal information in e-commerce; and paperless trading.25 Ibid, at paras 7.1168–7.1265. Appellate Body Report, China – Measures Affecting Trading Rights and Distribution Services for Certain Publications and Audiovisual Entertainment Products, WT/DS363/AB/R, adopted 19 January 2010, DSR 2010:I, 3, at paras 338–413 [hereinafter ‘ABR, China – Publications and Audiovisual Products’]. 18 Ibid, at para. 20. 19 Panel Report: China – Publications and Audiovisual Products, supra note 8 paras 4.72–4.85. 20 Ibid, at para. 7.875; ABR, China – Publications and Audiovisual Products, supra note 17, at para. 72. 21 See H Gao, ‘e-Commerce in ChAFTA: New Wine in Old Wineskins?’ in C Piker et al (eds), The China Australia Free Trade Agreement: A 21st-Century Model (Oxford, Hart Publishing, 2018). 22 ‘Protocol to Amend the Free Trade Agreement and the Supplementary Agreement on Trade in Services of the Free Trade Agreement Between the Government of the People's Republic of China and the Government of the Republic of Chile’ (2017) fta​.mofcom​.gov​.cn/​chiletwo/​xieyi/​bcyds​_en​.pdf, at Chapter 4. 23 ‘Protocol to Upgrade the Free Trade Agreement between the Government of the People's Republic of China and the Government of the Republic of Singapore’ (2018) www​.enterprisesg​.gov​.sg/​-/​media/​ ESG/​Files/​Non​-Financial​-Assistance/​For​-Companies/​Free​-Trade​-Agreements/​CSFTA/​Legal​-Text/​ Appendix6​_E​-Commerce​_Chapter​.pdf, at Appendix 6, New Chapter 15. 24 See H Gao, ‘Regulation of Digital Trade in US Free Trade Agreements: From Trade Regulation to Digital Regulation’ (2018) 45 Legal Issues of Economic Integration 47; M Wu, ‘Digital Trade Related Provisions in Regional Trade Agreements: Existing Models and Lessons for the Multilateral Trade System’ (2017), RTA Exchange http://​e15initiative​.org/​wp​-content/​uploads/​2015/​09/​RTA​-Exchange​ -Digital​-Trade​-Mark​-Wu​-Final​-2​.pdf (last accessed 10 July 2020); RF Fefer et al, ‘Digital Trade and U.S. Trade Policy’ (2019), CRS Report for Congress R44565 https://​fas​.org/​sgp/​crs/​misc/​R44565​.pdf (last accessed 10 July 2020). 25 See H Gao, ‘Digital or Trade? The Contrasting Approaches of China and US to Digital Trade’ (2018) 21 Journal of International Economic Law 297. 16 17

China and WTO e-commerce negotiations  151 Over the past five years, capitalizing on the enormous success of the Chinese e-commerce market, China has been pushing for wider adoption of its e-commerce model beyond its own shores. At the regional level, China has been building the electronic silk road, which provides online e-commerce platforms to facilitate both the exports of Chinese products abroad and the imports of foreign products into China.26 At the multilateral level, Alibaba, with the support of the Chinese government, has been aggressively promoting its Electronic World Trade Platform (‘eWTP’) concept, which led to the launch of ‘Enabling e-commerce’ initiative along with the WTO and the World Economic Forum in late 2017.27 As discussed later in the chapter, these initiatives have also found their way into China’s e-commerce proposals in the JSI.

2.

CHINA AND JSI: RESISTANCE AND ACCEPTANCE

Recognizing the growing importance of e-commerce, WTO Members adopted the Declaration on Global Electronic Commerce at the second Ministerial Conference in 1998.28 In addition to establishing a temporary moratorium on customs duties on digital transmission, the Declaration also calls WTO Members to ‘examine all trade-related issues relating to global electronic commerce’. Pursuant to the Declaration, the General Council adopted the Work Programme on Electronic Commerce,29 which divided up the work among several WTO bodies such as the Council for Trade in Services, the Council for Trade in Goods, the Council for Trade-Related Aspects of Intellectual Property Rights and the Committee on Trade and Development. While the division of work among the different bodies provided an opportunity for in-depth discussions on the impact of e-commerce in different areas, such compartmentalized approach was not really conducive to the negotiations due to the inherent complexity of e-commerce, which does not fit neatly into the pigeonholes of goods, service and intellectual property rights. Thus, by July 1999, the bodies had reached an impasse in their respective discussions and the discussions were suspended. As WTO Members started to grasp the cross-cutting nature of e-commerce issues, the General Council decided on 8 May 2001 to have dedicated sessions to discuss cross-cutting issues in e-commerce, with the first held on 15 June 2001.30 After this a total of 12 dedicated session were held, with the last one on 18 October 2016.31 However, other than agreeing to continue the moratorium on customs duties on electronic transmission periodically, these cross-cutting discussions failed to produce substantive results and the Members remain divided

‘跨境电商连接网上丝绸之路 [Cross-Border e-Commerce Connects Cyber Silk Road]’ (People’s Daily, 12 June 2018) world​.people​.com​.cn/​n1/​2018/​0612/​c1002​-30051604​.html (last accessed 10 July 2020). 27 See Gao, supra note 25, at 308–10. 28 WTO, Declaration on Global Electronic Commerce, adopted on 20 May 1998 at the Second WTO Ministerial Conference in Geneva, WT/MIN(98)/DEC/2, 25 May 1998. 29 WTO, Work Programme on Electronic Commerce: Ministerial Decision of 13 December 2017, Ministerial Conference, Eleventh Session, Buenos Aires, 10–13 December 2017, WT/MIN(17)/65, WT/L/1032, 18 December 2017. 30 Dedicated Discussion on Electronic Commerce under the Auspices of the General Council on 15 June 2001, Summary by the Secretariat of the Issues Raised, WT/GC/W/436, 6 July 2001. 31 ‘Electronic Commerce’ (WTO) www​.wto​.org/​english/​tratop​_e/​ecom​_e/​ecom​_e​.htm (last accessed 10 July 2020). 26

152  Research handbook on digital trade on even the most basic issues such as the mode of supply and classification of e-commerce. Indeed, the division among the Members was so wide that no substantive discussion was held at the 12th dedicated session due to the procedural concerns raised by some.32 Such opposition means discussions have only been held in informal open-ended meetings convened by the General Council Chair since then and the process is basically stalled. In view of the lack of progress under the formal Work Programme, proponents of the e-commerce negotiation started to explore alternative ways to advance the negotiation. This was recognized by the Ministerial Declaration at the Nairobi Ministerial Conference in December 2015, which acknowledged that some Members ‘believe new approaches are necessary to achieve meaningful outcomes in multilateral negotiations’.33 The US was even more explicit in its statement, with the then United States Trade Representative (‘USTR’) Michael Froman declaring the Nairobi Ministerial to begin ‘the road to a new era for the WTO’ and stating that ‘[a]s WTO Members start work next year, they will be freed to consider new approaches to pressing unresolved issues and begin evaluating new issues for the organization to consider’.34 After Nairobi, e-commerce gained ‘renewed interests’ among WTO Members.35 On 1 July 2016, the first post-Nairobi submission was made by the US. Likely in anticipation of the strong resistance from developing countries, the US took a rather cautious approach and labelled its submission as a ‘non-paper’ ‘intended solely to contribute to constructive discussion among Members’ rather than to advance ‘specific negotiating proposals’.36 While the non-paper repeatedly emphasizes that the US has ‘no preconceived views on best approaches, or on whether negotiations on specific aspects of e-commerce should be pursued, and if so on what bases’,37 many of the examples raised in the paper reiterated the US proposals in the negotiations of the Trade in Services Agreement (‘TiSA’) and Transpacific Partnership Agreement (‘TPP’) and brought into the WTO new issues such as free flow of data, a ban on data localization and forced transfer of source code for the first time.38 The US submission spurred a new wave of activity from other Members, with major players such as Japan, EU, Brazil, Canada and Singapore all making submissions within the same month.39 The work intensified over the next 16 months, and at the 11th Ministerial Conference held in December 2017 in Buenos Aires, 71 Members led by three co-conveners – Australia, Japan and Singapore – launched a Joint Statement to ‘initiate exploratory work together 32 General Council, 7 December 2016, Item 6 – Work Programme on Electronic Commerce: Progress Report by the Chairman, WT/GC/W/728, 8 December 2016. 33 ‘Nairobi Ministerial Declaration’ (WTO) www​.wto​.org/​english/​thewto​_e/​minist​_e/​mc10​_e/​ mindecision​_e​.htm, at para. 30. 34 ‘Statement by Ambassador Michael Froman at the Conclusion of the 10th World Trade Organization Ministerial Conference’ (Office of the United States Trade Representative (USTR), 19 December 2015) https://​ustr​.gov/​about​-us/​policy​-offices/​press​-office/​press​-releases/​2015/​december/​ statement​-ambassador​-michael (last accessed 10 July 2020). 35 WTO General Council, ‘Item 4 – Work Program on Electronic Commerce – Review of Progress: Report by Ambassador Alfredo Suescum – Friend of the Chair’, WT/GC/W/721, 1 August 2016. 36 WTO, Work Program on Electronic Commerce: Non-paper from the United States, JOB/GC/94, 4 July 2016, at para. 1.3. 37 Ibid, at para. 1.2. 38 Gao, supra note 25, at 307–8. 39 JOB/GC/96 (Japan et al); JOB/GC/97 (EU et al); JOB/GC/98 (Brazil); JOB/GC/99 (MIKTA countries); JOB/GC/100 (Japan); JOB/GC/101/Rev.1 (Singapore et al).

China and WTO e-commerce negotiations  153 toward future WTO negotiations’ on e-commerce.40 Nine meetings were held in 2018, and the negotiations were finally launched by 76 Members at the sidelines of the World Economic Forum Annual Meeting in Davos on 25 January 2019.41 Initially, China was quite reluctant to support the launch of e-commerce negotiations. In its first submission on e-commerce at the WTO, China tried to pre-empt the upcoming e-commerce negotiation in several ways.42 First, reflecting its long-standing position that only goods-related e-commerce issues should be discussed, China proposed, at the outset, that the scope of e-commerce discussions should ‘focus on promotion and facilitation of cross-border trade in goods enabled by internet, together with services directly supporting such trade in goods, such as payment and logistics services’.43 Second, China also indicated that it was not ready to negotiate new rules for e-commerce by stating that e-commerce discussions are ‘to clarify and to improve the application of existing multilateral trading rules’, which are normally understood not to include issues such as free flow of data, data localization, and so on.44 Third, China also tried to prevent e-commerce negotiations from being used as a Trojan horse for ‘new market access commitments including tariff reductions’.45 By taking out new rules and new tariff concessions, the Chinese submission then spelt out the only issues China might be willing to consider, that is, trade facilitation, transparency, digital certificates, electronic signature, electronic authentication and consumer protection and privacy.46 The same elements were reiterated in China’s submission to the General Council and the three subsidiary councils on GATT, GATS and Development in October 2017, which deemed these issues as ‘elements acceptable to Members’.47 Trying to steer the course on e-commerce negotiations at the Ministerial Conference, China even submitted a draft Ministerial Decision on Electronic Commerce, which suggested continuing the work under the Work Programme on Electronic Commerce in the General Council, while raising the possibility of reinvigorating the dedicated discussions on ‘elements acceptable to Members’ such as ‘facilitating cross-border e-commerce; promoting paperless trading; transparency; as well as development and cooperation’.48 Most of the draft made its way into the final Ministerial Decision,49 promoting Chinese Ministry of Commerce (‘MOFCOM’) Vice Minister Wang Shouwen to boast that ‘China has become a participant and even leader

40 WTO, Joint Statement on Electronic Commerce, Ministerial Conference, Eleventh Session, Buenos Aires, 10–13 December 2017, WT/MIN(17)/60, 13 December 2017. 41 WTO, Joint Statement on Electronic Commerce, WT/L/1056, 25 January 2019. 42 WTO General Council, Council for Trade in Goods, Council for Trade in Services, Committee on Trade and Development, Work Programme on Electronic Commerce: Aiming at the 11th Ministerial Conference, Communication from the People’s Republic of China and Pakistan, Revision, JOB/GC/110/ Rev.1, JOB/CTG/2/Rev.1, JOB/SERV/243/Rev.1, JOB/DEV/39/Rev.1, 16 November 2016. 43 Ibid, Introduction. For a detailed explanation of the meaning of ‘trade in goods enabled by internet’, see Gao, supra note 25, at 314. 44 Ibid, Introduction. 45 Ibid. 46 See Gao, supra note 25, at 314–15. 47 E-Commerce Elements for MC11, Communication from China, JOB/GC/142, JOB/CTG/9, JOB/ SERV/271, JOB/DEV/49, 19 October 2017. 48 Work Programme on Electron-Commerce, Communication from China, JOB/GC/150, 10 November 2017. 49 Work Programme on Electronic Commerce, Ministerial Decision of 13 December 2017, WT/ MIN(17)/65, WT/L/1032, 18 December 2017.

154  Research handbook on digital trade in rule-making’.50 However, by abandoning the consensus-based approach and launching the Joint Statement Initiative via the plurilateral route, the US and the other 70 Members have turned China’s success into a Pyrrhic victory. In a way, the e-commerce Joint Statement caught China by surprise. For China, the most important issue at the 11th Ministerial Conference was investment facilitation, which China has been pushing for at the WTO since 2014 as the coordinator of the group on ‘friends of investment facilitation for development’.51 Designed to provide support for its Belt and Road Initiative, China successfully persuaded 70 WTO Members to co-sponsor a Joint Statement on the issue.52 While China was also interested in e-commerce, its main task at the 11th Ministerial Conference was to push the WTO and World Economic Forum to officially endorse the Enabling e-commerce initiative – the brainchild of the Alibaba-backed Electronic World Trade Platform (‘eWTP’) – a mission which was also accomplished.53 In contrast, the e-commerce Joint Statement, as a US-led initiative, made China quite wary. Thus, many observers were surprised by China’s ‘last-minute’ decision to join the second e-commerce Joint Statement on 25 January 2019.54 However, a careful reading of the events in 2018 still reveals many hints explaining China’s shift. After the 11th Ministerial, the sponsors of the e-commerce Joint Statement did not waste any time in getting to business and held nine meetings over the short span of one year. Such frenzy of activities was unheard of in the WTO and prove that they are quite serious. Moreover, the key players in the e-commerce Joint Statement – US, EU and Japan – referred to e-commerce in the three trilateral statements on WTO reform they issued in 2018.55 In the last one, issued in September 2018, they further agreed to ‘intensify and accelerate this process’ to achieve ‘the timely launch of negotiations of a high standard agreement with the participation of as many Members as possible’. Three more trilateral statements were issued after 2018, with the latest one being issued on 14

50 热点问答：世贸组织第11届部长级会议中国怎么看[Hot Questions Q&A: China’s Opinion on the 11th WTO Ministerial Meeting], (Xinhua News, 14 December 2017) xinhua​-rss​.zhongguowangshi​ .com/​233/​-​5166990033​302484302/​2775895​.html (last accessed 10 July 2020). 51 ‘Investment Facilitation for Development’ www​.unescap​.org/​sites/​default/​files/​%28China​%29​ %20Statement​%20of​%20Investment​%20facilitation​%20for​%20development​.pdf (last accessed 10 July 2020). 52 Joint Ministerial Statement on Investment, Facilitation for Development, WT/MIN(17)/59, 13 December 2017. 53 ‘WTO, World Economic Forum and eWTP Launch Joint Public-private Dialogue to Open Up E-commerce for Small Business’ (WTO, 11 December 2017) www​.wto​.org/​english/​news​_e/​news17​_e/​ ecom​_11dec17​_e​.htm (last accessed 10 July 2020). 54 B Baschuk and S Donnan, ‘China to Join Talks on $25 Trillion E-Commerce Market at Last Minute’ (Bloomberg, 25 January 2019) www​.bloomberg​.com/​news/​articles/​2019​-01​-25/​china​-is​-said​-to​ -join​-global​-e​-commerce​-talks​-at​-last​-minute (last accessed 10 July 2020). 55 The three statements are: ‘Joint Readout from Meeting of the United States, European Union and Japan in Brussels’ (USTR, 10 March 2018) https://​ustr​.gov/​about​-us/​policy​-offices/​press​-office/​ press​-releases/​2018/​march/​joint​-readout​-meeting​-united​-states; ‘Joint Statement on Trilateral Meeting of the Trade Ministers of the United States, Japan, and the European Union’ (USTR, 31 May 2018) https://​ustr​.gov/​about​-us/​policy​-offices/​press​-office/​press​-releases/​2018/​may/​joint​-statement​-trilateral​ -meeting; ‘Joint Statement on Trilateral Meeting of the Trade Ministers of the United States, Japan, and the European Union’ (USTR, 25 September 2018) https://​ustr​.gov/​about​-us/​policy​-offices/​press​-office/​ press​-releases/​2018/​september/​joint​-statement​-trilateral.

China and WTO e-commerce negotiations  155 January 2020.56 All these developments reminded China that the e-commerce Joint Statement parties are taking the issue very seriously and China could not just ignore it. Indeed, China learned its lesson the hard way when its attempt to join the TiSA negotiations was blocked by the US, making it impossible for China to shape the rules on services trade, where e-commerce was a major issue.57 The first indication for the policy change can be detected when China released its Position Paper on WTO Reform on 23 November 2018.58 While the position paper took the cautious approach and did not explicitly mention e-commerce, at the press conference held on the same day, Vice Minister Wang made a direct reference to e-commerce in response to a question from a journalist for examples on how to ‘keep the WTO rules relevant’, a key objective for China in WTO reform.59 After China joined the second e-commerce Joint Statement on 25 January 2019, Chinese WTO Ambassador Dr Zhang Xiangchen also gave the official explanation for the shift in position.60 First, referring to the critical juncture the WTO was at, Zhang pointed to the special significance the launch of the e-commerce negotiation could have on reinvigorating the negotiation function of the WTO and boosting people’s confidence in the multilateral trading system and economic globalization. Second, Zhang also regarded China’s participation as a good opportunity for it to play an active role in the negotiations, especially in reflecting the participation of developing countries and designing a flexible framework to reflect the reasonable demands of different parties. For long-time observers of China’s trade policy, such shifts in position are not unprecedented. For example, in early stages of the Doha Round negotiations, China sided with developing countries. Before the Cancun Ministerial Conference in September 2003, China and 16 other developing countries formed the ‘Core Group’, which resisted the push by the ‘Colorado Group’ of developed countries to start negotiations on the Singapore issues including trade facilitation.61 However, when the General Council decided to start negotiations on trade facilitations on 1 August 2004, China became an active participant.62 This makes sense because China, as one of the top exporters in the world, would benefit from more efficient and The three statements are: ‘Joint Statement of the Trilateral Meeting of the Trade Ministers of the European Union, Japan and the United States’ (USTR, 9 January 2019) https://​ustr​.gov/​about​-us/​policy​ -offices/​press​-office/​press​-releases/​2019/​january/​joint​-statement​-trilateral​-meeting; ‘Joint Statement of the Trilateral Meeting of the Trade Ministers of the United States, European Union, and Japan’ (USTR, 23 May 2019) https://​ustr​.gov/​about​-us/​policy​-offices/​press​-office/​press​-releases/​2019/​may/​joint​-statement​ -trilateral​-meeting; ‘Joint Statement of the Trilateral Meeting of the Trade Ministers of Japan, the United States and the European Union’ (USTR, 14 January 2020) https://​ustr​.gov/​about​-us/​policy​-offices/​ press​-office/​press​-releases/​2020/​january/​joint​-statement​-trilateral​-meeting​-trade​-ministers​-japan​-united​ -states​-and​-european​-union. 57 Gao, supra note 25, 301–4. 58 MOFCOM, 中国关于世贸组织改革的立场文件[China’s Position Paper on WTO Reform] http://​sms​.mofcom​.gov​.cn/​article/​cbw/​201812/​20181202817611​.shtml (last accessed 2 August 2020). 59 商务部召开世贸组织改革有关问题新闻吹风会 [China’s Ministry of Commerce Opens News Conference for Response to WTO-related Reforms], (PRC Gov, 23 November 2018) www​.gov​.cn/​ xinwen/​2018​-11/​23/​content​_5342934​.htm (last accessed 10 July 2020). 60 世贸组织成员在达沃斯签署电子商务联合声明 [WTO Members Signs Joint Statement on E-commerce at Davos] (Xinhua News, 25 January 2019) www​.xinhuanet​.com/​fortune/​2019​-01/​25/​c​ _1210047708​.htm (last accessed 10 July 2020). 61 S Zhenyu (ed), WTO多哈回合谈判中期回顾 [Mid-term Review of the WTO Doha Round Negotiations] (Beijing, People’s Publishing House [Renmin Chubanshe] 2005), at 178–81. 62 Ibid, 194–5. 56

156  Research handbook on digital trade cheaper customs processes.63 Like trade facilitation, China’s decision to join the e-commerce negotiations demonstrated once again China’s flexibility when it comes to specific trade issues and its willingness ‘to take up commitments commensurate with its level of development and economic capability’, as stated in its Position Paper on WTO Reform.64

3.

THE CHINESE PROPOSALS

From an initial group of 76 Members in January 2019, the JSI has grown to include 86 Members as of 1 June 2022, with Ecuador the newest participant. Together, they represent more than 90 per cent of global trade and more than half of the WTO’s membership. In addition, the JSI also remains open for participation by non-Members, which include Senegal, the LDC signatory of the Osaka Declaration on e-commerce, which has yet to join the JSI as a formal member.65 Before January 2019, the JSI was framed around four themes: A: Enabling Digital Trade/E-Commerce; B: Openness and Digital Trade/E-Commerce; C: Trust and Digital Trade/E-Commerce; and D: Cross-Cutting Issues, including Development, Transparency and Cooperation.66 During the exploratory discussions held in 2018, each theme was further divided into several sub-themes, resulting in 13 sub-themes in total. Selected issues and topics were further identified under each sub-theme, resulting in 40+ issues in total.67 After January 2019, the group moved on to the plurilateral negotiations phase and the themes were also expanded to include two new ones, E: Telecommunications and F: Market Access.68 The six themes were further divided into 15 sub-themes and 35 selected issues/topics.69 In 2020, intensive work was carried out in ten small groups focusing on the following issues70: (i) consumer protection; (ii) spam; (iii) e-signatures and electronic authentication; (iv) paperless trading; (v) digital trade facilitation; (vi) source code; (vii) open government data; (viii) market access; (ix) customs duties on electronic transmissions, and (x) open internet access. In December 2021, the Ministers of the three co-conveners – Australia, Japan and Singapore – reported that they have achieved ‘good convergence in negotiating groups on eight articles – online consumer protection; electronic signatures and authentication; unsolicited commercial electronic messages; open government data; electronic contracts; transparency; paperless trading; and open internet access’.71 H Gao, ‘China’s Ascent in Global Trade Governance: From Rule Taker to Rule Shaker, and Maybe Rule Maker?’ in C. Deere-Birkbeck (ed), Making Global Trade Governance Work for Development (Cambridge, Cambridge University Press, 2001), at 166. 64 MOFCOM, supra note 57, at 4. 65 IDEAS, ‘WTO Joint Statement on Electronic Commerce: Advancing the Search for Convergence’ (IDEAS) https://​ideascentre​.ch/​newslettre​_2019/​(last accessed 10 July 2020). 66 Ibid. 67 Ibid. 68 Ibid. 69 Ibid. 70 Yasmin Ismail, ‘E-Commerce Joint Statement Initiative Negotiations among World Trade Organization Members: State of Play and the Impacts of COVID-19’ (International Institute for Sustainable Development) www​.iisd​.org/​publications/​e​-commerce​-negotiations​-wto​-members​-covid​-19 accessed 10 April 2022, at 11≠12. 71 WTO Joint Statement Initiative on E-commerce Statement by Ministers of Australia, Japan and Singapore, December 2021, www​.wto​.org/​english/​news​_e/​news21​_e/​ji​_ecom​_minister​_statement​_e​ .pdf. 63

China and WTO e-commerce negotiations  157 In the negotiation process, China has emerged as one of the most active participants, with three submissions out of a total of 48 substantive submissions so far.72 This section examines the submissions in detail. 3.1

The First Submission

The first submission was submitted by China on 23 April 201973 and reiterated its general positions made on prior occasions leading to China’s participation in the JSI. The first part sets out China’s overall approach to the JSI negotiation, which covers four areas: the objective; relationship with the WTO; negotiation process; and its direction and focus. It started by noting that development should be the objective of the JSI and calling participant to help ‘Members, particularly developing Members and LDCs, to integrate into global value chains, bridge the digital divide, seize development opportunities and benefit from inclusive trade, and hence better participating in the economic globalization’. Consistent with the developing country position, China also stated that the JSI negotiation ‘should be complementary to the electronic commerce discussion in relevant subsidiary bodies of the WTO’ and ‘ultimately achieve a multilateral outcome’. This approach is also reflected in China’s proposal for the negotiation process, where it noted that the JSI negotiation ‘should be open, inclusive and transparent’ with ‘well-designed frameworks and flexible approaches on the implementation of negotiation outcomes’. This point probably reflects China’s unhappy experience with the TiSA negotiations, where the US reportedly blocked its request to participate in the closed, exclusive and non-transparent negotiation. The mentioning of ‘flexible approaches on the implementation of negotiation outcomes’, on the other hand, indicates that China might not accept all obligations but prefers a tiered approach on commitments, which again affirms its willingness to ‘take up commitments commensurate with its level of development and economic capability’.74 With regard to the scope of the JSI negotiation, China again emphasized that it should focus on the discussion of cross­border trade in goods enabled by the internet, together with relevant payment and logistics services while paying attention to the digitalization trend of trade in services, and explore the way to develop international rules for electronic commerce centering on a sound transaction environment and a safe and trust­worthy market environment.

This is again unsurprising given China’s strong interest in trade in goods and the relevant trade facilitation and electronic payment issues, as evidenced by the enormous success of its homegrown e-commerce model with Alibaba as the e-commerce platform, Alipay as the payment gateway, and the many courier services companies as the distributor of goods.

A search on WTO’s Documents Online system on 1 July 2020 using the document symbol for the e-commerce JSI ‘INF/ECOM’ generated 60 submissions, but 6 of them are revisions or addendum to original submissions, while the other 6 are just communications from new participants informing their decisions to participate, which include, for example, INF/ECOM/18 by Benin; INF/ECOM/37 by Kenya; INF/ECOM/38 by Côte D’Ivoire; INF/ECOM/48 by Cameroon; INF/ECOM/50 by the Philippines; INF/ ECOM/53 by Burkina Faso. 73 Joint Statement on Electronic Commerce, Communication from China, INF/ECOM/19, 24 April 2019. 74 MOFCOM, supra note 58, at 4. 72

158  Research handbook on digital trade What follows further elaborates the focus of the negotiation by listing China’s priority issues, which are grouped into four action areas: 3.1.1 Definition and clarification China calls for Members to define terms such as trade-related aspects of electronic commerce and electronic transmission, and to clarify the relationship between future electronic commerce rules and existing WTO Agreements. Both tasks appear innocuous, but as the history of the e-commerce Work Programme has shown, even such mundane discussions could become contentious, especially given the open hostility some WTO Members have displayed towards the JSI. Thus, it seems that the more sensible approach is to adopt the ‘constructive ambiguity’ approach and leave these issues undisturbed. 3.1.2 Trade facilitation measures China also calls for Members to ‘establish a sound environment for electronic commerce transaction[s]’, which includes two types of measures. The first are measures to facilitate customs process, such as the improvement of customs procedures, electronic payment of customs fees and electronic customs documentation, establishment of free zones and customs warehouses and a moratorium on customs duties. The second is mainly the establishment of the necessary legal framework to enable the recognition of electronic signatures, electronic authentication and electronic contracts. These measures are mostly uncontroversial as they largely copy from the provisions under the Agreement on Trade Facilitation, of which China is a main proponent. The only exception is the moratorium on customs duties on e-commerce, which became a contentious issue in 2019 due to the opposition of India and South Africa to extend.75 While a decision to extend the moratorium until the 12th Ministerial Conference was finally made by the General Council on 10 December 2019, there is still a possibility for the revocation of the moratorium due to the growing interest among WTO Members in collecting tax on digital services and e-commerce. Thus, instead of a permanent moratorium, China only suggested to maintain the practice ‘until the next Ministerial Conference’. This implies that China has yet to decide where its interests lie on the issue and wants more time to study the issue. 3.1.3 Safety and security This part focuses on measures to ‘create a safe and trustworthy market environment for electronic commerce’, which mainly includes consumer safety regulations, such as measures for online consumer protection, personal information protection and fighting spam or unsolicited electronic commercial messages. Interestingly, the submission also includes a paragraph on ‘cyber security’, which, in addition to language on enhancing e-commerce security and safeguarding cyber security, also calls for members to ‘respect […] internet sovereignty’. As I have discussed elsewhere, ‘internet sovereignty’ has been a favourite slogan of the Chinese government, which elevated internet regulation to the level of national security or even sovereignty to justify its draconian laws.76 As shown by the latter parts of the submission B Reinsch et al, ‘Ongoing Goings On: A News Update on WTO’ (2020) www​.csis​.org/​analysis/​ ongoing​-goings​-news​-update​-wto​-0 (last accessed 10 July 2020). 76 H Gao, supra note 7. 75

China and WTO e-commerce negotiations  159 discussed below, the reference to ‘internet sovereignty’ is more than empty propaganda, but does reflect the seriousness China attaches to certain issues and is indicative of China’s position on these issues. 3.1.4 Development The submission also encourages Members to ‘promote pragmatic and inclusive development cooperation’, including measures to help developing countries to improve the e-commerce infrastructure and bridge the digital divide, to share best practices on e-commerce development and help them build up their capacity, and also to ‘establish an Electronic Commerce for Development Program under the WTO framework’. These initiatives, if successfully implemented, will definitely help developing countries to boost their e-commerce development, which, in turn, could also facilitate the expansion of Chinese e-commerce giants such as Alibaba in these countries, especially in regions covered by the Belt and Road Initiative. In the last part, entitled ‘Other Issues’, China also discussed the main demands of the US in the JSI, that is, data flow, data storage, treatment of digital products, and so on. By addressing them directly and acknowledging them as issues of concern for some Members, China has broken from its traditional approach of simply ignoring them. This itself is a positive sign, as it indicates China’s willingness to engage on these issues. At the same time, China also indicated that it was not ready to discuss these issues, at least not in the early stages of the negotiation. Citing the ‘complexity and sensitivity’ of these issues, as well as ‘the vastly divergent views among the Members’, China stated that ‘more exploratory discussions are needed before bringing such issues to the WTO negotiation, so as to allow Members to fully understand their implications and impacts, as well as related challenges and opportunities’. Such approach is all too familiar to those who follow WTO negotiations closely, as it is basically a polite way of saying ‘we don’t want to discuss these issues now’. In particular, China singled out the issue of cross-border data flow, by stating that ‘[i]t’s undeniable that trade-related aspects of data flow are of great importance to trade development’. It is interesting to note, however, what China did and did not say in this sentence. It did not, for example, use ‘free flow of data’, which is how the US has always referred to the issue in its submissions.77 On the other hand, it qualified ‘data flow’ with ‘trade-related aspects’. This implies that China is not willing to address all kinds of data flows, just those related to trade. In other words, to the extent that some data flows do not have a trade nexus, they could be legitimately excluded. This qualification could have wide implications as it could be employed to justify restrictions on data flows in sectors that China have not made commitments to, or even on those covered by existing commitments but provided free of charge (such as Google’s search engine services) as they are not technically ‘traded’. Moreover, in an effort to turn the tables, China also prefaced the discussion on these ‘other issues’ with the affirmation of ‘the legitimate right’ by Members ‘to adopt regulatory measures in order to achieve reasonable public policy objectives’. This language is reminiscent of the 77 See Work Programme on Electronic Commerce, Non-paper from the United States, JOB/GC/94, which refers to ‘free flow of information’ in para. 2.3; and Joint Statement on Electronic Commerce Initiative, Communication from the United States, INF/ECOM/5, which refers to ‘free flows of information’ in Section 2.

160  Research handbook on digital trade calls for more ‘policy space’, a term often employed in trade negotiations to justify special and differential treatment and resort to exceptions clauses. As the China–Publications and Audiovisual case mentioned above has illustrated, China will, most likely, invoke the public order exception contained in the General Exceptions clauses of both the GATT and GATS to justify its online censorship regime. In particular, on data flow, China emphasized that it ‘should be subject to the precondition of security’78 and should ‘flow orderly in compliance with Members’ respective laws and regulations’.79 This extends China’s domestic narrative of cybersecurity to the international level, which is made complete with the earlier reference for all Members to ‘respect the internet sovereignty’ of other Members. By elevating the issue to one of ‘sovereignty’, China has shown the seriousness it attaches to the issue of regulating data flow. In summary, China made it clear that it was not yet ready to discuss these sensitive issues, at least not in the early stages of the negotiations. There is a possibility that it will consider some of the issues further down the road, but such negotiations will not be easy given China’s guarded position on these issues. 3.2

The Second Submission

In its second submission dated 8 May 2019,80 China spelt out its detailed proposals on its priority issues. As China’s first substantive proposal, the 12 draft articles in the submission largely correspond to three of the four main action areas mentioned in Section 3 of its first submission, that is, Section 3.1 on definition and clarification, Section 3.2 on trade facilitation and Section 3.3 on safety and security. The first draft article is entitled ‘scope’, but actually dealt with the definition issue by proposing that the agreement ‘apply to measures affecting the production, distribution, marketing, sale or delivery of goods and services by electronic means’. This language copies verbatim the language from the 1998 e-commerce Declaration81 and confirms China’s position that the JSI should ‘support the multilateral trading system’ and ‘keep WTO rules relevant’. In the alternative, China suggests that the agreement ‘apply to measures adopted or maintained by Members that affect trade by electronic means’, which mirrors the language in its FTAs.82 The next draft article addresses relationship with existing WTO Agreements by noting first that in the event of conflicts between the new agreement and the WTO Agreements, those in Annex 1 to the Marrakesh Agreement shall prevail. The next paragraph explicitly states that the new agreement ‘shall not be construed to have changed or modified Members’ market access commitments made under the [GATT or GATS]’. This partly reflects China’s sour experience in the China–Publications case discussed earlier, where the US used the technology neutrality principle to persuade the Panel that China’s services schedule also includes commitments on electronic distribution of audio-visual products. Thus, this article was proposed in an attempt

Communication from China, supra note 71, at para. 4.3. Ibid. 80 Joint Statement on Electronic Commerce, Communication from China, INF/ECOM/32, 9 May 2019. 81 Work Programme on Electronic Commerce, WT/L/274, adopted by the General Council on 25 September 1998. 82 China–Singapore Free Trade Agreement (CSFTA), in Chapter 15 at Art. 2.2. 78 79

China and WTO e-commerce negotiations  161 to seal the loophole and ensure that China would not inadvertently modify its commitments by participating in the JSI. The third draft article deals with exceptions, and starts by explicitly noting that Article XX of the GATT 1994 and Article XIV of the GATS ‘shall apply to this Agreement to the extent applicable’ and their provisions ‘shall be incorporated into and made an integral part of this Agreement, mutatis mutandis’. Again, like the previous article, this provision is partly the result of the hard lessons China learned in the China–Raw Materials83 and China–Rare Earth84 cases, where due to the lack of explicit reference to the general exception clause of the GATT, China was denied the right to justify its export restrictions under GATT Article XX. In addition, China also specifically pointed out that the new agreement shall not prevent Members from adopting or maintaining any measures for the purposes of guaranteeing cybersecurity, safeguarding cyberspace sovereignty, protecting the lawful rights and interests of its citizens, juridical persons and other organizations and achieving other legitimate public policy objectives, provided that such measures are not applied in a manner which would constitute a means of arbitrary or unjustifiable discrimination or a disguised restriction on trade, and are no more than necessary to achieve the objectives.

This strong language confirms once again China’s obsession with cybersecurity, which is elevated to the level of sovereignty and thus non-negotiable. The second part of the article focuses on ‘Security Exceptions’, where China proposes that the agreement shall not be construed ‘to require any Member to furnish any information, the disclosure of which it considers contrary to its essential security interests’ or ‘to prevent any Member from taking any action which it considers necessary for the protection of its essential security interest’. Again, it is probably not unreasonable to surmise that cybersecurity would be considered a matter of ‘essential security interest’. The rest of the proposal is mostly unexciting, dealing either with the issue of trade facilitation, with four articles on electronic authentication and electronic signatures, electronic contracts, electronic invoices and maintaining domestic legal frameworks governing electronic transactions; or the issue of e-commerce safety, with three clauses on unsolicited commercial electronic messages, personal information protection and online consumer protection. Then there are two articles on good governance, with one focusing on transparency and calling for publication of ‘all measures of general application which pertain to electronic commerce’ and ‘all measures relating to public telecommunications networks or services’ and the other on domestic regulation. The last draft article is particularly interesting, as it, in addition to incorporating GATS Article VI, also specifically states that ‘[n]othing in this Agreement shall be construed to affect any Member’s right to conduct a content review for the purposes of achieving legitimate public policy objectives’. Again, the inspiration for this clause comes from the China–Publications case. Even though China’s right to conduct content review was 83 Panel Reports, China – Measures Related to the Exportation of Various Raw Materials, WT/ DS394/R, Add.1 and Corr.1/WT/DS395/R, Add.1 and Corr.1/WT/DS398/R, Add.1 and Corr.1, adopted 22 February 2012, as modified by Appellate Body Reports WT/DS394/AB/R/WT/DS395/AB/R/WT/ DS398/AB/R, DSR 2012:VII, at 3501. 84 Panel Reports, China – Measures Related to the Exportation of Rare Earths, Tungsten, and Molybdenum, WT/DS431/R and Add.1/WT/DS432/R and Add.1/WT/DS433/R and Add.1, adopted 29 August 2014, upheld by Appellate Body Reports WT/DS431/AB/R/WT/DS432/AB/R/WT/DS433/ AB/R, DSR 2014:IV, at 1127.

162  Research handbook on digital trade explicitly affirmed by the AB in that case, China’s inclusion of this draft clause indicates that it is not taking any chances and attaches high importance to its censorship regime, which is another non-negotiable item. 3.3

The Third Submission

Compared to the second submission, China’s latest submission,85 made on 20 September 2019, has fewer draft articles – 8 instead of 12 – but is longer – six instead of five pages. This is because the draft articles are more detailed in the third submission, indicating that China has probably put more effort into drafting these articles. With the exception of the last article, the third submission mainly focuses on trade facilitation measures. These include three articles on streamlining customs administration, such as transparency and non-discrimination of trade policy, paperless trading and various measures to enhance trade facilitation, including implementation of trade facilitation agreement, advance electronic data for customs clearance, electronic payment of duties, bounded warehouse and free zones, regional distribution centres and expedited clearance for low-risk cargo and collective clearance. Three other articles call on Members to improve their e-commerce-related services commitments, such as online trade facilitating and supporting services such as those provided by Alibaba, logistics services such as those provided by SF Express, and electronic payment services like those provided by Alipay. Together, they help to solve three common problems faced by developing countries when they try to develop e-commerce: lack of a good e-commerce platform, a slow or non-existent logistics network and the inability to transfer payments between buyers and sellers. Of course, all these are likely to be achieved with the help of Chinese firms, which are now the world leaders in providing e-commerce solutions on platform, logistics and payment. Even though such services are mainly provided online, they might need the physical presence of e-commerce-related personnel to set up, maintain and repair. Thus, another article suggests that Members facilitate the temporary entry and sojourn of such personnel. This is similar to the GATS visa proposal by India, albeit further limiting the beneficiaries to e-commerce-related personnel. The last draft article in the submission is on ‘Electronic Commerce-Related Network Equipment and Products’. Ostensibly, it can be said to be related to trade facilitation in e-commerce, but it is quite obvious that such equipment and products are capable of much wider use in the telecom sector, especially in view of the expansive definition provided in the article, which covers ‘all hardware and related software and services that can be used to support transactions done by electronic means, including telecommunication network equipment, products, resources, and related services such as installation, trial operation, testing, optimization, maintenance and repair services and etc., and other related equipment, products, resources and related services’. The article calls on Members to not discriminate against ‘network equipment and products of any other Member’, which are further elaborated in three successive substantive paragraphs to mean not to exclude such network equipment and products, not to prevent public telecommunications networks or their services suppliers from choosing them, and not to ‘block the supply chains of electronic commerce-related network equipment and products, in particular those based on long-term commercial cooperation, 85 WTO, Joint Statement on Electronic Commerce, Communication from China, INF/ECOM/40, 23 September 2019.

China and WTO e-commerce negotiations  163 including cutting or prohibiting the supply to enterprises of any other Member of necessary raw materials, components, parts, software, technologies and their updates for electronic commerce-related network equipment and products’. As this proposal was submitted after the widely reported exclusion of Huawei in 5G network in Europe and Australia, and the ban on the sales of chips and the licence of the Android system to Huawei by the US, the inclusion of the article on network equipment and products is probably far from mere coincidence. It reflects China’s attempt to fight what it perceives as ‘technology protectionism’ using trade rules, which along with the ‘Made in China 2025’ plan is another key component of China’s quest for technological supremacy. But for two reasons, China might see its initiative thwarted. First, this is more of a telecom issue, which is arguably beyond the scope of e-commerce negotiation. Even though telecommunication has been added as one of the focus groups of discussion, past experiences in GATS negotiations such as the Reference Paper have shown that Members are more concerned with services regulatory issues such as competitive safeguards, licensing and regulatory requirements rather than hardware-related issues.86 Instead, technical issues on hardware and software have traditionally been dealt with at the International Telecommunication Union (ITU). This is also confirmed by the recent discussions of the issue in the JSI, where several Members noted either that ‘the JSI was not the appropriate forum to discuss this topic’ or ‘some topics were more appropriate to be discussed at the ITU’.87 Second, even if JSI participants agree to engage in discussions on the issue, it would not be hard for them to justify any restrictions they might introduce or maintain with the security exception, which, ironically, also features prominently in China’s second submission discussed earlier, where China advocates broad leeway for Members to take ‘any action which it considers necessary for the protection of its essential security interest’. Interestingly, even though China addresses – albeit in a negative manner – the issues of data flow and localization in its first submission, neither the second nor the third submission contain language on these issues. Nor was the moratorium on customs duty mentioned.

4.

CONCLUDING THOUGHTS

Initially reluctant to join the JSI negotiation on e-commerce over concerns it might be a US plot, China finally jumped on the JSI bandwagon at its launch in Davis in January 2019 and has emerged as one of the most active participants. Such policy shift is the result of China’s realization that it is important to enhance its rule-making power in e-commerce and cyberspace, as noted by President Xi in his speech at the 36th Collective Study Session of the Politburo.88

See H Gao, ‘Telecommunications Services: Reference Paper’ in R Wolfrum et al (eds), Max Planck Commentary on World Trade Law, Volume VI: Trade in Services (Leiden, Martinus Nijhoff Publishers, 2008), at 718–47. 87 Joint Statement on Electronic Commerce, 11–14 February 2020, Facilitator’s Reports, Seventh Negotiating Round, INF/ECOM/R/7, 25 February 2020. 88 习近平：加快推进网络信息技术自主创新 朝着建设网络强国目标不懈努力 [Xi Jinping: Accelerate the Promotion of Indigenous Innovation on Internet Information Technology, Strive Unrelently Towards the Objective of Building the Internet Power] (Xinhua News, 9 October 2016) news​ .xinhuanet​.com/​politics/​2016​-10/​09/​c​_1119682204​.htm (last accessed 10 July 2020). 86

164  Research handbook on digital trade Despite being a world leader in e-commerce – or in China’s own words, ‘trade in goods facilitated by the internet’ – China’s draconian approach to cybersecurity has made people doubtful as to whether it would make a positive contribution to global e-commerce governance, with some even calling for ‘disqualifying’ China from participation in the JSI negotiation.89 Indeed, as reviewed earlier in this chapter, while many of China’s detailed proposals, especially those on trade facilitation and consumer protection, seem rather innocuous or even benevolent, as they do offer good lessons for developing countries eager to catch the e-commerce train, its proposals on security exception and content review raise concern regarding whether China would be willing to accept the main demands of the US and other Western countries, that is, free flow of information across borders, a free and open internet and prohibition of localization requirements, forced technology transfer and transfer of source code.90 However, these considerations do not necessarily have to spell the end of China’s participation in the JSI, especially if one takes a closer look at the nuances of the contrasting positions between China and the West. Here I will illustrate the potential for compromise with a few key examples. 4.1

Free Flow of Information

Many commentators, especially those with a technology or internet background, tend to believe that the free flow of data should be absolute, that is, should apply to all data. While this could be a laudable ultimate goal, at present this is far from how the principle is understood in trade agreements. Take for example the relevant provisions in the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) and the United States– Mexico–Canada Agreement (USMCA), both of which are regarded as providing a ‘gold standard for digital trade’ – at least in the eyes of the US, the main drafter of the rules.91 Instead of calling for a blanket free flow of information, both agreements only require the parties to allow the cross-border transfer of information by electronic means ‘when this activity is for the conduct of the business of a covered person’.92 This is entirely understandable because trade agreements, at the end of the day, are not human rights agreements. Instead, they are designed to facilitate cross-border trade, which means data flows are protected only when they contribute to the trade flow. Thus, to the extent that China does not wish to allow data flow for a specific type of service activity, it can simply carve out an exception for that specific sector. Indeed, this is probably why Google, despite the noises it made when it was forced to pull out of China in 2009, never successfully persuaded the USTR to bring a WTO case against China. As I analysed in the case study of the merits of such case in 2011, such a complaint would be

89 N Cory, ‘Why China Should Be Disqualified From Participating in WTO Negotiations on Digital Trade Rules’ (2019) https://​itif​.org/​sites/​default/​files/​2019​-china​-disqualified​-wto​.pdf (last accessed 10 July 2020). 90 WTO, Work Program on Electronic Commerce, Non-paper from the United States, JOB/GC/94, 4 July 2016. Also affirmed in Joint Statement on Electronic Commerce, Communication from the United States, INF/ECOM/23, 26 April 2019. 91 J Garber, ‘USMCA Is “Gold Standard for Digital Trade”: Trade Chief Robert Lighthizer’ (Fox News, 17 December 2019) www​.foxbusiness​.com/​markets/​usmca​-gold​-standard​-digital​-trade​-robert​ -lighthizer (last accessed 10 July 2020). 92 USMCA, Art. 19.11; CPTPP, Art. 14.11.

China and WTO e-commerce negotiations  165 doomed as China has not made any commitments on the search engine services provided by Google.93 Moreover, both agreements also provide, in the same article, an exception clause that allows parties to adopt or maintain inconsistent measures ‘to achieve a legitimate public policy objective’ so long as they do not constitute ‘arbitrary or unjustifiable discrimination or a disguised restriction on trade’ and fulfils the necessity requirement. Thus, if needed, China could also invoke the exception clause to justify its data flow restrictions, as it did in the China–Publications case. Moreover, as shown by the case, the US does not have problem with the exception per se; instead, its main concern is that it is discriminatory and not necessary. 4.2

Data Localization

Another oft-mentioned concern is data localization, where people believed that China requires the localization of all data. Again this is another misconception as the provision in question, Article 37 of China’s Cyber Security Law, only requires local storage for ‘personal information and important data collected and generated by operators of critical information infrastructure from its operations within the people’s Republic of China’.94 Thus, there are important qualifiers on the types of data, that is, only personal information and important data; types of operators, that is, only operators of critical information infrastructure; and geographical scope, that is, only those data generated from its operations in China. Moreover, the final version of the law already improves over previous drafts. For example, the first draft of the law applies the localization requirement to all such data generated by such operators from its operation,95 and the final text greatly reduces the impact by limiting the geographical scope. Of course, the final provision on data localization is far from perfect for several reasons. First, in addition to the commonly used concept of ‘personal information’, the law also includes ‘important data’, a concept that has yet to be defined by Chinese law. Second is what constitutes ‘critical information infrastructure’. Article 31 of the Cyber Security Law defines it as those in ‘important industries and fields such as public communications and information services, energy, transport, water conservancy, finance, public services and e-government affairs’, as well as those ‘that will result in serious damage to state security, the national economy and the people’s livelihood and public interest if it is destroyed, loses functions or encounters data leakage’. Such broad definition could potentially capture everything and is not really helpful, which is why the same Article also directed the State Council to develop the ‘specific scope of critical information infrastructure’. In 2016, the CAC issued the National Network Security Inspection Operation Manual96 and the Guide on the Determination of Critical Information

See Gao, supra note 4. 中华人民共和国网络安全法 [Cyber Security Law of the People’s Republic of China], 7 November 2016, www​.cac​.gov​.cn/​2016​-11/​07/​c​_1119867116​.htm (last accessed 2 August 2020). 95 Article 31 of the first draft dated 6 July 2015, see 网络安全法（草案）全文 [Cyber Security Law (Draft) Full Text], 6 July 2015, www​.npc​.gov​.cn/​npc/​c1481/​201507/​82​ce4cb5549c​4f56be8a67​ 44cf2b3273​.shtml (last accessed 2 August 2020). 96 Central Leading Group on Cyber Security and Informatisation General Office, ‘国家网络安全检 查操作指南[Network Security Coordination Bureau, National Network Security Inspection Operation Manual]’, June 2016 (on file with author). 93 94

166  Research handbook on digital trade Infrastructure,97 which clarified the scope of critical information infrastructure by grouping them into three categories: websites, which includes the websites of government and party organizations, enterprises and public institutions, and news media; platforms, which includes internet service platforms for instant messaging, online shopping, online payment, search engines, emails, online fora, maps and audio video; production operations, which includes office and business systems, industrial control systems, big data centre, cloud-computing and TV broadcasting systems. They also laid down three steps in determining the critical information infrastructure, which starts with the identification of the critical operation, then continues with the determination of the information system or industrial control system supporting such critical operation, and concludes with the final determination based on the level of the critical operations’ reliance on such systems and possible damages resulting from security breaches in these systems. More specifically, they listed 11 sectors: energy, finance, transportation, hydraulic, medical, environmental protection, industrial manufacturing, utilities, telecom and internet, radio and TV and government agencies. The detailed criteria are both quantitative and qualitative. For example, critical information infrastructure includes websites with a daily visitor count of more than one million people and platforms with more than ten million registered users or more than one million daily active users, or daily transaction value of ten million RMB. On the other hand, even those that do not meet the quantitative criterion could be deemed to be critical information infrastructure if there are risks of security breaches that would lead to leakage of lots of sensitive information about firms or enterprises, or leakage of fundamental national data on geology, population and resources, or seriously harming the image of the government or social order, or national security. The potentially wide reach of the criteria was well illustrated by the case of the BGI Group, which was fined by the Ministry of Science and Technology in Oct 2018 for exporting certain human genome info abroad via the internet without authorization.98 Given the nature of their business, the BGI case could fall under the category of ‘leakage of fundamental national data on […] population’ as mentioned earlier. The last problem with China’s data localization policy is that, according to Article 37, only the export of personal information and important data requires security review, while there is no such requirement for domestic use. This could be interpreted as discriminatory and arbitrary, and constitute disguised restrictions in international trade. Of course, this does not mean that all hope is lost regarding a potential deal on data localization. Instead, as I have explained elsewhere, the key to understand China’s data regulation is national security, which translates into the ability to maintain its censorship regime. So long as the Chinese regulators can continue to conduct content view and block foreign websites on security grounds, where the data is stored would be much less important. Actually, given the sophistication of the Great Firewall, data stored in offshore servers would be easier to block and filter. In this regard, it is instructive to study the evolution of the US approach on data localization for financial services. In the TPP negotiation, the US carved out the entire finan-

97 关键信息基础设施确定指南（试行） [Guide on the Determination of Critical Information Infrastructure (Trial)], under 关于开展关键信息基础设施网络安全检查的通知 [Notice on Conducting Network Security Inspections of Key Information Infrastructure], Zhongwangban Fawen [2006] #3, Annex 1, July 2016, available at www​.xxrd​.gov​.cn/​notice/​794​.html (last checked 1 July 2020). 98 S An, ‘数据出境如何’安检’ [How to Conduct ‘Safety Check’ for Exporting Data] (Zhihu) https://​ zhuanlan​.zhihu​.com/​p/​65413452 (last accessed 10 July 2020).

China and WTO e-commerce negotiations  167 cial services sector from the scope of its e-commerce chapter, including prohibition of data localization requirements.99 In the new USMCA, however, the US explicitly brought over the ban to the financial services sector by stating that data localization should not be required ‘so long as the Party’s financial regulatory authorities, for regulatory and supervisory purposes, have immediate, direct, complete, and ongoing access to information processed or stored on computing facilities that the covered person uses or locates outside the Party’s territory’.100 If such language can successfully overcome the grave concerns of the US Federal Reserve, then probably the Chinese regulators would also have less reason to insist on local storage instead of having ‘immediate, direct, complete, and ongoing access to information processed or stored on computing facilities outside the Party’s territory’.101 To summarize, while China’s participation in the JSI would make the negotiations difficult, it also provides an opportunity to better understand the policy rationale of China’s data regulation and explore avenues for convergence and compromise. 4.3

RCEP, CPTPP and DEPA

China’s willingness to accept both commitments is confirmed by its obligations in the RCEP, which includes provisions disallowing its Members to ‘require a covered person to use or locate computing facilities in that Party’s territory as a condition for conducting business in that Party’s territory’,102 or ‘prevent cross-border transfer of information by electronic means where such activity is for the conduct of the business of a covered person’.103 While both articles are subject to exceptions allowing a Party to adopt or maintain measures that such Party considers necessary ‘to achieve a legitimate public policy objective’ or ‘for the protection of its essential security interests’, the very fact that China is willing to accept such obligations is encouraging. Moreover, unlike 4 of the 15 RCEP Members,104 China did not seek any transition period for both obligations to take effect. While the chapter on e-commerce is not subject to the normal dispute settlement procedure under the RCEP, the importance China attaches to the RCEP and the peer pressure under the consultation and Joint Committee procedures could provide some incentive for China to implement the obligations in good faith. If we can learn anything from the RCEP, it is that actively engaging China in e-commerce negotiations is much better than leaving it in its own cyber-enclave. It is in the same spirit that China’s recent applications to join the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) and the Digital Economy Partnership Agreement (DEPA) shall be handled. The internet was built to transcend walls. International negotiations on the internet and e-commerce should also help people reach across walls, no matter how great they might be.

TPP, Art. 14.1. USMCA, Art. 17.18.2. 101 Ibid. 102 RCEP, Art. 12.14. 103 Ibid, Art. 12.15. 104 For the implementation of both obligations, Cambodia, Lao PDR, and Myanmar were granted a transition period of up to eight years, while Viet Nam was granted five years. See footnotes 11 and 13 in Chapter 12 of the RCEP. 99

100

11. Facilitating digital trade in Hong Kong Bryan Mercurio1

1. INTRODUCTION The emergence of the COVID-19 pandemic in January 2020 and the associated ‘covid zero’ policies initiated and maintained by the government through 2022 severely strained the Hong Kong economic climate. While most of the world has moved on from the challenges of the pandemic, the Hong Kong economy continues to be moribund and the city struggles to recover. Competitors such as Singapore are gaining market share from Hong Kong in key sectors, including banking and finance. This chapter argues that Hong Kong should not return to business as usual following the pandemic but instead should seek to expand its economic footprint, widen its geographical range and diversify in order to transition to higher value-added services and industries. For several decades, Hong Kong has relied almost exclusively on its connection to and special relationship with mainland China for its continuing economic prosperity. China’s rapid economic growth and steady movement towards greater liberalisation served as not only a springboard but also a cushion for Hong Kong. While manufacturing virtually ceased to exist in Hong Kong in this century, its services sector more than made up for the decline of the legacy industry. Hong Kong’s economy transformed to become reliant on professional services and trade logistics. By 2020, Hong Kong’s service industry accounted for 93.4 per cent of gross domestic product (GDP), with financial services and logistics contributing 43.2 per cent of GDP, with 23.4 per cent and 19.8 per cent, respectively.2 Such staggering figures largely reflect the opening of China’s capital market, its export-oriented growth strategy and the size of the Chinese market.3 While many in Hong Kong simply assume that ‘what is will continue to be’, this assumption is risky and history proves it to be incorrect. Many small traders in Hong Kong became wealthy in the 1990s and early 2000s by serving as a ‘middleman’ between buyers in the West and sellers in China. Technological advancements, combined with trade and investment liberalisation, made this role redundant, as buyers no longer have the need for a conduit and instead can simply order directly from the supplier. We may be seeing another paradigm shift as the drivers of China’s growth shift from heavy industry and infrastructure investment to technological innovation and domestic consumption. Stated differently, China is transitioning to a mature, services-oriented digital economy. At the same time, China is better integrating into the Asia-Pacific region through the negotiation of free trade agreements (FTAs). The most notable among these is the Regional Comprehensive Economic Partnership (RCEP), an agreement covering 2.3 billion people This chapter draws from Bryan Mercurio and Ronald Yu, Regulating Cross-Border Data Flows: Issues, Challenges and Impact (Anthem Press, 2022). 2 Hong Kong Census Statistics Department, National Accounts, Table 36, www​.censtatd​.gov​.hk/​en/​ web​_table​.html​?id​=​36. 3 Ibid. 1

168

Facilitating digital trade in Hong Kong  169 accounting for more than $26 trillion in output and $12 trillion in trade.4 While the economic benefits of RCEP will be incremental rather than transformative, it is estimated that the mega-regional agreement could add $186–209 billion annually to world income and nearly $400–500 billion to world trade by 2030.5 Of note, 90 per cent of the income gains and 88 per cent of the trade will accrue to China, South Korea and Japan (three large economies that have not negotiated bilateral FTAs or a trilateral FTA between and among themselves).6 China has also applied to join the Comprehensive and Progressive Trans-Pacific Partnership (CPTPP),7 a mega-regional agreement once led by the United States (US), from which President Trump withdrew in his first days in office.8 Far from 2013, when it first expressed interest to join, China now seems serious and while it understands that accession would require a commitment to further liberalisation, openness and reform, it also recognises that increasing standards may better integrate China into the global economy. Importantly, China does not believe that the level of commitments necessary for accession to the CPTPP will be an insurmountable hurdle. Chinese membership in the CPTPP would not only bring with it economic benefits for other partner countries but would also demonstrate that China is seeking to play a leadership role in trade governance at a time when the US is withdrawing from its traditional leadership role. Hong Kong needs to ensure that it is able to leverage and gain from these developments. In order to do this, it must transform once again by enhancing its international competitiveness. The way to do this is to further expand its global network and connectivity. Section 2 of this chapter provides an overview of Hong Kong’s economic activities and regulatory framework on digital trade. Section 3 provides recommendations for Hong Kong in order to maintain its status as a services hub and to fully capitalise on the digital transformation. Section 4 concludes.

2.

RETHINKING THE LEGAL FRAMEWORK

Hong Kong has historically been a flexible, resilient and entrepreneurial city. It has thrived due to its ability to adapt and position itself as a player in the broader global economy. Declared a free port in 1841,9 Hong Kong quickly established itself as a trading entrepôt before transitioning to the manufacture of light goods and textiles following the First World War. This 4 See https://​asean​.org/​rcep​-agreement​-enters​-into​-force/​. The RCEP entered into force on 1 January 2022. 5 For estimates see Peter A. Petri and Michael G. Plummer, ‘East Asia Decouples from the United States: Trade War, COVID-19, and East Asia’s New Trade Blocs’, Peterson Institute for International Economics Working Papers 20-9, https://​www​.piie​.com/​publications/​working​-papers/​east​ -asia​-decouples​-united​-states​-trade​-war​-covid​-19​-and​-east​-asias​-new; New Zealand Ministry of Foreign Affairs and Trade, www​.mfat​.govt​.nz/​en/​trade/​free​-trade​-agreements/​free​-trade​-agreements​-in​-force/​ regional​-comprehensive​-economic​-partnership​-rcep/​rcep​-overview/​. 6 Ibid. 7 See Mireya Solis, ‘China Moves to Join the CPTPP, but Don’t Expect a Fast Pass’, Brookings Blog (23 September 2021) www​.brookings​.edu/​blog/​order​-from​-chaos/​2021/​09/​23/​china​-moves​-to​-join​ -the​-cptpp​-but​-dont​-expect​-a​-fast​-pass/​. 8 Office of the United States Trade Representative, ‘The United States Officially Withdraws from the Trans-Pacific Partnership’, Press Release (30 January 2017) https://​ustr​.gov/​about​-us/​policy​-offices/​ press​-office/​press​-releases/​2017/​january/​US​-Withdraws​-From​-TPP. 9 See www​.mardep​.gov​.hk/​theme/​port​_hk/​en/​p1ch2​_2​.html.

170  Research handbook on digital trade process accelerated following the Communist revolution in China, as factories relocated from the mainland to Hong Kong. Beginning in the 1980s, however, the trend reversed as China began opening up to the world. Hong Kong became a less efficient, uneconomic place of manufacture and attempted to transition to a service hub for the manufacturing industry, before booming in the 1990s as an international finance centre and shipping hub. The city now seeks to become a Fintech and global technology hub. The time has come for Hong Kong to shift; while less radical than previous shifts, the emergence of COVID-19, increasing geopolitical tensions between China and the West and what may be longer-term inflationary pressures necessitates a repositioning in order to ensure that Hong Kong continues to grow and prosper. At the same time, global supply-chain bottlenecks have forced companies to focus on agility and resilience. Hong Kong must capitalise on these realities, as well as on the fact that regional trade between and among Asian countries is increasing, assisted by mega-regional FTAs such as the CPTPP and RCEP.10 To date, however, Hong Kong has focused its trade and logistics efforts on supply chains linking China to the United States and Europe. This must change, and Hong Kong should extend its existing supply-chain network to the region to link South Asia, Southeast Asia and Oceania to each other and to the West. Hong Kong is well placed to do so, with the burgeoning economies of China to the north, India to the west and ASEAN to the south. The shift becomes even more necessary when one considers that technologies, including blockchain, will continue to play an important role in the logistics and payment structures that enable cross-border trade. In this regard, Hong Kong must ensure that its digital infrastructure and legal framework allows the city to play a role in facilitating global trade through finance, logistics, transport and other professional services which enable and contribute to trade and economic development. More specifically, Hong Kong should seek to further extend its existing capabilities and role to become the so-called digital command centre for the region’s supply chains. Positioning itself at the heart of a complex network will not be easy, and will require the government to be more proactive than usual to enable business interests to build, implement and operate the necessary digital infrastructure that links factories, logistics operations, ports and airports into one seamless network. In order to fully take advantage of the digital transformation, jurisdictions must have a clear data strategy. While jurisdictions with larger economies have clear policies in place, many smaller jurisdictions have yet to adopt a strategy or framework and instead operate on a ‘by default’ combination of piecemeal legislation and obligations undertaken in trade agreements. Notable exceptions include Singapore, Chile and New Zealand, which are leading the efforts to codify an open and accessible framework through FTAs and other subject-specific agreements.11 In formulating such a strategy, jurisdictions should consider national competitiveness, business attractiveness and regulatory structure.12 The convergence of multiple interests and issues create challenges for policymakers, as each of these areas impacts on subsets of the

10 Intra-Asian trade is on the rise. See Asian Development Bank, ‘Trade Integration Deepens in Asia and the Pacific amid Pandemic’ (9 February 2022) www​.adb​.org/​news/​trade​-integration​-deepens​-asia​ -and​-pacific​-amid​-pandemic. 11 See eg the Digital Economy Partnership Agreement. 12 See Bryan Mercurio and Ronald Yu, Regulating Cross-Border Data Flows: Issues, Challenges and Impact (Anthem Press, 2022), p.40.

Facilitating digital trade in Hong Kong  171 economy or social structure. For instance, while it is clear that cross-border data impacts on trade (for instance, issues of digital trade, e-commerce and banking services) and finance (including issues of trade finance, compliance, blockchain/cryptocurrencies and regulation), it also impacts on key domestic drivers, such as innovation. This then raises issues such as the impact of data flows on new technologies and new product development, as well as unintended consequences of policy decisions regarding privacy, intellectual property and competition. Hong Kong is essentially wholly reliant on data, and digital trade, for its prosperity. Given its advantages in legal infrastructure, protection of IPRs, buoyant capital markets and strategic location to the mainland cities in the Greater Bay Area (GBA) that are spearheading China’s economic prosperity, Hong Kong views itself as a financial corridor to China and the Belt and Road Initiative (BRI) and is seeking to become a data depository and processing centre for the GBA.13 Data is already connecting and driving Hong Kong’s economy and future prospects, yet Hong Kong lacks any clear strategy. Indeed, there is little domestic legislation in place. The most specific legislation is s33(2) of the Personal Data (Privacy) Ordinance (PDPO) (1996/2012), which would provide the relevant legislative framework in specifying that a data user shall not transfer data to a place outside Hong Kong unless one of the following conditions is met: ● The place is ‘whitelisted’ by the Office of the Privacy Commissioner for Personal Data (PCPD) in that there is in force any law which is substantially similar to, or serves the purpose of the PDPO; ● The data user has reasonable grounds for believing that the destination has in force laws which are substantially similar to, or serve the same purposes as the PDPO; ● The data subject has consented in writing to the transfer; ● The data user has reasonable grounds for believing that the transfer is for the avoidance or mitigation of adverse action against the subject; or ● The data user has taken all reasonable precautions and exercised all due diligence to ensure that the data will not be handled in a manner that would be a contravention of the PDPO. With concerns from the business community that s33(2) could have a significant and substantive negative impact on small and medium-sized enterprises – which account for 98 per cent of the business units and provide 45 per cent of the total employment in Hong Kong – the section has not been put into effect.14 While a consultation process remains ongoing, there is no timetable for its completion or enactment of the section. This is not to say that Hong Kong does not have any framework in place. On the contrary, while there is no legislation the city has undertaken several digital trade-related commitments in FTAs in what has essentially established a ‘de facto’ policy direction. The most comprehensive of these FTAs is the Australia–Hong Kong Free Trade Agreement (AUHKFTA), which

See Nathan Chau, ‘HK Has Opportunities and Obligations as a “Super Connector”’, China Daily (7 September 2022) www​.chinadailyhk​.com/​article/​289265​#HK​-has​-opportunities​-and​-obligations​-as​-a​ -'super​-connector'. 14 Hong Kong Trade and Industry Department, ‘Support for Small and Medium Enterprises’ www​ .tid​.gov​.hk/​english/​smes​_industry/​smes/​smes​_content​.html. 13

172  Research handbook on digital trade entered into force in January 2020. Chapter 11 of the AUHKFTA on e-commerce contains dedicated provisions on: ● Electronic signatures and electronic authentication; ● A legal framework governing electronic transactions consistent with the principles of the UNCITRAL Model Law on Electronic Commerce 1996 or the UN Convention on the Use of Electronic Communications in International Contracts 2005; ● Consumer protection; ● Prohibition of customs duties on electronic transmissions, including content transmitted electronically; ● Freedom of movement of information, including financial services; ● Ban on localization of computing facilities, including financial services; ● Protection of personal information; ● Paperless trading; ● Unsolicited commercial electronic messages; ● Prohibition on requiring transfer or access to source code; and ● Cooperation on development, enforcement and by other means.15 As is common in other trade agreements, the AUHKFTA contains a wide exception clause for both the provisions on the movement of information and data localization which allows a party to adopt or maintain inconsistent measures ‘to achieve a legitimate public policy objective, provided that such measures are not applied in a manner which would constitute a means of arbitrary or unjustifiable discrimination, or a disguised restriction on trade’.16 While Hong Kong should be mindful of and join initiatives that continue to set the global standards (such as the Digital Economic Partnership Agreement (DEPA)), the problem is that relying on such an approach could lead to inconsistent commitments between various agreements. This has already been the case in Hong Kong, as the requirements contained in the FTAs with EFTA and Chile differ. For example, Hong Kong’s FTA with Chile is more permissive of domestic regulation than the agreement with EFTA. As it would be administratively difficult, if not impossible, to regulate on a country-dependent basis, the practical effect is that Hong Kong will be bound to the lowest level of regulation allowed for in its FTAs. Another issue with the ‘by default’ policy is that the obligations undertaken in FTAs is not quite aligned with s33 of the PDPO, as Hong Kong’s FTAs take a similar approach to that of the US – minimal regulation and a commitment to free flows – whereas s33 of the PDPO resembles the European approach of mutual recognition. Smaller economies, like that of Hong Kong, must ensure their policies are consistent with the data transfer rules set by larger economies, most notably the US, EU and China. But this is not enough for an economy that relies on data for economic prosperity and has ambitions to become a major financial, Fintech and technology player while also remaining a leading logistics and innovation hub. Hong Kong must thus balance the aims and objectives of global initiatives with considerations of how regulations could impact business efficiency and innovation – and how to differentiate itself from its competitors.

See www​.dfat​.gov​.au/​trade/​agreements/​in​-force/​a​-hkfta/​a​-hkfta​-text/​Pages/​default. See AUHKFTA arts 11.8 and 11.9. Interestingly, Hong Kong FTAs with ASEAN, Georgia and Macao regarding flows of financial data do not contain similar safeguards. 15 16

Facilitating digital trade in Hong Kong  173 For these reasons, Hong Kong must develop an explicit and strategic policy framework on cross-border data flows in order to provide legal certainty, maintain competitiveness and deliver opportunities to identify new areas of growth, maintain its comparative advantages and foster more diversified economic development.

3.

CONSIDERATIONS IN DESIGNING A CROSS-BORDER DATA POLICY FOR HONG KONG

Hong Kong has a long history of laissez-faire-style governance, and has greatly benefited from letting the markets dictate the direction of economic development. That being the case, in the modern environment, it is not possible to leave the digital environment unregulated. Hong Kong has, indeed, regulated in a de facto manner by virtue of its FTA commitments and could benefit from a greater level of policy direction and coordination in the regulation of cross-border data flows. Such an approach could be done in a manner that eschews a heavily centralised policy undertaking and retains a more libertarian, free-flowing direction. In a book published earlier this year, Ronald Yu and I made five recommendations for Hong Kong to so that it can retain its place in the world and continue to prosper in a rapidly evolving digital environment.17 Recommendation One: Further Commitments and Updated Trade Agreements While Hong Kong has made several important commitments relating to cross-border data flows in its recent FTAs that send a signal to investors and interested stakeholders that it will continue to respect an open internet, it should continue to negotiate rules that encourage and facilitate digital trade and ensure the free flow of data. The A-HKFTA offers the logical starting point as it already compares well with the USMCA, CPTPP and other agreements negotiated by (among others) the US, Japan, Korea and Singapore. That said, advances have been made in the DEPA and other more recent agreements that Hong Kong’s commitments are yet to capture. For example, Hong Kong could seek to join the DEPE or use this modern agreement as a model to advance beyond the A-HKFTA by adding provisions from these more recent agreement on digital identities to its future agreements. In order to suit the needs and developmental objectives of Hong Kong, such provisions should include a best endeavours clause to foster technical interoperability and comparable levels of protection of digital identities (Art. 7.1), cooperation and development of solutions in Fintech (Art. 8.1) – all of which are important for cross-border trade and specifically to ensure that electronic payments systems, digital currencies and software applications, such as e-wallets, can be easily integrated, as these technologies will increasingly play a more important part of the financial system. Moreover, Hong Kong should also seek to negotiate for a specific clause on AI (Art. 8.2) to promote the adoption of ethical and governance frameworks that support trusted, safe and responsible AI

17

These recommendations are sourced from Mercurio and Yu, above n 12, Ch. 5.

174  Research handbook on digital trade usage.18 Like existing agreements, Hong Kong could make its commitments subject to appropriate exceptions for privacy, national security and other legitimate public policy objectives. By so doing, Hong Kong would have a blueprint on which to build a framework on cross-border data flows. The intention is therefore not to entrench a ‘by default’ policy but rather to establish a foundation that would feed into a more comprehensive strategic policy and framework. In this regard, it would be useful, and send a further positive signal, if Hong Kong were to entrench the commitments in domestic legislation. Recommendation Two: Copyright Law – Consider a TDM Exemption In order to ensure it can remain competitive and retain its place as a finance and logistics leader as well as develop into an innovation hub, Hong Kong should add an exemption for text and data mining (TDM) in the copyright law. TDM is the process of copying existing electronic information (such as that in scientific journals) to analyse the data it contains for patterns, trends and other useful information;19 it is increasingly necessary in research and the development and training of AI applications, which require access and analysis to vast amounts of data to allow the learning of patterns and the ability to do so. While countries have allowed temporary ephemeral acts of copying (that is, cache copies) for some time, many technical processes first require the creation of a reference corpus in which content is permanently stored for further processing. It is questionable whether TDM would survive legal scrutiny in Hong Kong,20 and such an outcome would prevent Hong Kong from becoming a centre for AI innovation and R&D. As it stands, Hong Kong’s copyright law is outdated as it has not been updated to respond to the innovative ways content is now created, distributed, and used.21 Singapore offers a useful example for Hong Kong to emulate, with its Copyright Bill (© Bill) offering a user-friendly blueprint in designing legislation that considers future national development as well as feeding into and supporting Singapore’s National AI Strategy to transform Singapore into ‘a leader in developing and deploying scalable, impactful AI solutions, in key sectors of high value and relevance to our citizens and businesses’.22 In this regard, the © Bill introduces three critical exceptions. The first exception allows a user to legally perform ‘computational data analysis (CDA)’, defined as ‘using a computer program to identify, extract and analyse information or data from the work’ or ‘using the work as an example of a type of information or data to improve the functioning of a computer program’ (which would cover training algorithms), on a work protected by copyright.23 The second exception permits Dan Ciuriak, ‘The Challenge of Updating Institutions for Digital Trade’, Centre for International Governance Innovation’ (16 July 2021) www​.cigionline​.org/​articles/​the​-challenge​-of​-updating​ -institutions​-for​-digital​-trade/​. 19 John Kelly, ‘The Text and Data Mining Copyright Exception: Benefits and Implications for UK Higher Education’ JISC (9 February 2016) www​.jisc​.ac​.uk/​guides/​text​-and​-data​-mining​-copyright​ -exception; University of Western Australia, ‘Text Mining and Data Mining: What Is Text Mining and How Does It Work?’, https://​guides​.library​.uwa​.edu​.au/​textmining. 20 Mercurio and Yu, above n 12, 49–51. 21 See contra, European Copyright Directive (Directive (EU) 2019/790); Singapore © Bill. 22 Government of Singapore, Smart Nation Digital Government Office, National AI Strategy, November 2019, p.17, www​.smartnation​.gov​.sg/​docs/​default​-source/​default​-document​-library/​national​ -ai​-strategy​.pdf​?sfvrsn​=​2c3bd8e9​_4. 23 © Bill, p.152. The Bill also includes the following plain English illustration: ‘An example of computational data analysis under paragraph (b) is the use of images to train a computer program to recognise.’ Ibid. 18

Facilitating digital trade in Hong Kong  175 a user to copy and/or communicate a copy of the work or recording of a protected performance for computational data analysis and preparation of the words for such analysis, without having to seek the rights holder’s permission subject to certain safeguards (for example, not to use the reproduction of the work for purposes other than computational data analysis).24 Finally, the Bill updates a general fair dealing exemption introduced in 2004 to make it a much wider US-style fair use exception (Division 2–Fair use). In so doing, Singapore is making it clear that innovation is a priority, and that the narrow rights traditionally exercised by rights holders will no longer be available.25 This is important to the development of AI, as the shift to a fair use exception means that developers are more likely to successfully defend the use of copyrighted content as fair if such use is for the purposes of training AI and does not affect the value of such copyrighted content. Hong Kong failed to pass the Copyright (Amendment) Bill 2014 due, in large part, to resistance from a self-interested collection of copyright owners who even objected to a ‘limited fair use’ concession.26 To proceed with reform, the government would have to convince the business community that a TDM exemption is in the interests of local businesses, innovators, creators and other stakeholders. Without such reform, Hong Kong’s copyright laws will increasingly disadvantage and serve as a drag on the local AI industry. In short, Hong Kong will become a less attractive place for R&D and innovation. Recommendation Three: A More Holistic, Coordinated Approach The third recommendation is for Hong Kong to develop a holistic, coordinated approach to cross-border data. Government departments must work together to formulate a comprehensive and coherent policy on cross-border data flows that creates an environment conducive to innovation and to attracting and retaining funds, business and skills, and that makes it possible to leverage the opportunities presented by emerging regional developments. Inter-departmental government cooperation is imperative: many cross-border data issues are not only complex but also interrelated. For example, developing an AI research hub requires a refocusing of tertiary education and promoting innovation in trade and logistics for Hong Kong would necessitate training people on how to use the latest products or employ the latest technologies. In developing such a policy, close connections and reliance on China will obviously play a large role in the discussions and outcomes. It is therefore important that Hong Kong provides clarity to the increasing integration of Hong Kong–China cross-boundary data flows. Such flows considerably impact interests across cross-industries and while the Chinese legal framework and practices may prove challenging from an implementation standpoint, Hong Kong’s increasing integration into the Chinese framework will allow it to better exploit its strengths and capitalise on its unique position.

© Bill, p.153a. See Kok Keng Lau and Nicholas Lauw, Impending Changes to the Copyright Regime – Copyright Bill Introduced in Parliament, July 2021, https://​eoasis​.rajahtann​.com/​eoasis/​lu/​pdf/​2021​-07​_Impending​ -Changes​-Copyright​-Regime​.pdf. 26 See Stuart Lau, ‘Five Reasons the Hong Kong Copyright Bill Failed’, South China Morning Post, 4 March 2016, https://​www​.scmp​.com/​news/​hong​-kong/​politics/​article/​1920569/​five​-reasons​-hong​ -kong​-copyright​-bill​-failed. 24 25

176  Research handbook on digital trade Hong Kong should be able to leverage its strategic positioning to benefit from China's growth in tech. However, in order to do so, Hong Kong stakeholders may require additional certainty with regard to cross-boundary data flows. This could be accomplished in a number of ways. For instance, Hong Kong could complete the process of updating the cross-border data transfer aspects of the PDPO by either implementing or amending s33. However, such an option may not be entirely satisfactory, as (as noted earlier) while s33 resembles a European-style ‘adequacy decision’ with the whitelisting of jurisdictions with substantially similar regimes, Hong Kong has undertaken commitments in trade agreements that are more in line with the US approach to cross-border data flows. Recommendation Four: Develop a Matrix and Checklist The fourth recommendation is that Hong Kong develops a system regulating when data can be sent cross-border. This proposal follows the established practice of leading jurisdictions and, in the Hong Kong context, could see the establishment of a task force to coordinate data transfer issues and establish a ‘whitelist’ which permits certain categories of data to enter and exit in certain circumstances and a ‘negative list’ to ascertain which data cannot be freely transferred.27 Use of a whitelist and negative list is preferred as it is transparent and clear, and the identification of specific activities, industries and transfer purposes within the scope of the initiative would enhance certainty and efficiency of data flows to China and within the increasingly important Greater Bay Area (GBA). This is not to say that the list approach should be limited to cross-boundary trade; indeed, Hong Kong should develop a directed impact assessment matrix and checklist offering a decision support tool for decision-makers in different disciplines and across jurisdictions. The matrix could map data-related interrelationships while the checklist would establish procedures or items to check as well as the relevant laws and requirements. Recommendation Five: Establish a Think Tank to Assist Policy Formulation and Implementation The final recommendation is that the government should create an independent, interdisciplinary think tank to assist with the formulation and eventually evolution of a comprehensive policy framework on cross-border data flows. The current data-intensive and competitive online operating environment is one of rapid change and incomplete information. Policymakers often work with incomplete information and can struggle to judge which potential courses of action will result in optimal outcomes or potential ruin. In order to aid in the development of laws and implementation of the strategic policy, a think tank could produce and implement (as well as update) the matrix and checklist decision-making tool. In order to successfully fulfil its mandate, the think tank must be staffed and supported by various stakeholders with differing expertise and from different disciplines. There is a great need for stakeholders from various disciplines to mix and better understand each other’s per27 See Lei Chen, Chunyan Ding, Tianxiang He, Pinxin Liu and Rostam J. Neuwirth, Legal Research Project: Proposal for Hong Kong to Be a Data Centre Hub for the Greater Bay Area & China, 16 September 2021, https://​www​.cityu​.edu​.hk/​slw/​lib/​doc/​rccl/​201901​_RCCL​_Report​-HK​_as​_Data​ _Centre​_Hub​-ES​.pdf.

Facilitating digital trade in Hong Kong  177 spectives, especially in the context of policy formulation on a complex topic that inherently crosses and intersects with multiple disciplines. The advantage of a think tank like this is that, unlike individual bodies whose analyses are focused on specific topics and generating recommendations to support senior organizational leaders, an independent think tank surveying the data flow landscape more broadly would not be (or at least would be less) constrained by existing cognitive biases or perceptual limitations stemming from over-immersion in organizational or industry-specific issues. Thus, the think tank would be well placed to present alternative hypotheses or employ alternative analytical methodologies. Given the importance of the issue to Hong Kong and the complex interconnections involved, a data flow policy failure for Hong Kong could represent a catastrophic setback for the city’s place in the digital trade ecosystem. Thus, while the establishment of an independent and interdisciplinary think tank would not be a guarantee of success, it would be useful in assisting policymakers to confront and mitigate the challenges and threats that characterize the competitive environment of today’s data-driven world.

4. CONCLUSIONS Hong Kong’s future lies in building on its existing strengths of connecting East and West and as a gateway to China as well as adapting these strengths to the changing environment. For Hong Kong to remain competitive and realise its ambitions, it must recognise the importance of data and place it at the core of its policy choices. Digital trade is at the heart of Hong Kong’s economy, and its continued growth and social stability depend on continuing to excel in professional services revolving around finance and logistics. In order to do so, Hong Kong must have a stable and predictable policy framework on cross-border data flows in order to provide comfort to stakeholders that issues such as the localization, collection and storage of data, cybersecurity and privacy are adequately addressed and managed. The government must also ensure stakeholders that the city has an important role in the digital transformation, and that it can serve as a technological and R&D hub. This too relies on getting the policies right, and should be included in a holistic, comprehensive strategic policy framework.

12. Digital trade agreements and digital policy space in the Republic of Korea Tae Jung Park1 and Joo Hyun Park2

1. INTRODUCTION The connectivity of our global society is accelerating owing to innovative technologies such as 5G communication, AI services, and blockchain. Therefore, in the international commerce sector, competition among digital commerce platforms is fierce as they implement new technologies to secure partnerships with global high-tech companies. Correspondingly, the scale of digital trade through e-commerce on these platforms is growing rapidly.3 Digitalization of transactions in trade is recognized as an inevitable international trend. Likewise, the advent of the digital trade era is regarded as a new opportunity for many exporting and importing companies. The innovative digitization of transactions in trade not only reduces transaction costs but also helps to pioneer new—and previously unimaginable— markets through new methods such as the establishment of export strategies using big data. The rapid development of digital technology encouraged many governments to conclude new forms of bilateral or multilateral digital trade agreements, known as digital economy agreements, to implement frameworks that foster digitalization of trade. However, these governments have simultaneously made “policy space”4 in digital trade as the priority of governments’ domestic policy agendas.5 Many governments have a mutual interest in ensuring that they clarify and secure the states’ policy space for legitimate objectives while they fulfill their purpose of liberalizing digital trade markets. Yet, searching for an equilibrium between

The views or opinions expressed herein are the authors’ alone and do not reflect the views or opinions of the Ministry of Justice or the Ministry of Trade, Industry and Energy of the Republic of Korea. All remaining errors and misconceptions are entirely the authors’ responsibility. 2 The views or opinions expressed herein are the authors’ alone and do not reflect the views or opinions of Yulchon LLP, the Board of Audit and Inspection, and the Ministry of Justice of the Republic of Korea. 3 United Nations Conference on Trade and Development, Digital Economy Report 2021 (UNCTAD/ DER/2021). 4 “Regulatory power,” “policy space,” “sovereignty power,” “regulatory space,” and “policy autonomy” are interchangeably used in the literature. This article uses the term “policy space” for the purpose of this article. For the definition of “policy space,” see Veijo Heiskanen, The Regulatory Philosophy of International Trade Law, 38 J. World Trade 1, 2–5 (2004); Michael J. Trebilcock and Robert Howse, Trade Liberalization and Regulatory Diversity: Reconciling Competitive Markets with Competitive Politics, 6 Eur. J. Law Econ. 5 (1998). 5 See e.g., Andrew Mitchell and Theodore Samlidis, Cloud Services and Government Digital Sovereignty in Australia and Beyond, 29 International Journal of Law and Information Technology, 364–94 (2022). 1

178

Digital trade agreements and digital policy space  179 liberalization or market openings and states’ policy space is still the most challenging aspect of this trend.6 With respect to that challenge, this chapter examines the Republic of Korea’s (hereinafter ROK) recent movements in digital trade areas, exploring how ROK is balancing out its policy space with liberalization by looking at recently implemented domestic laws. The chapter will show, in particular, how these domestic laws comply with ROK’s recently concluded trade agreements in digital trade. There is a compelling reason for analyzing ROK’s movement in digital trade. ROK has an externally dependent economic system in which foreign trade accounts for more than half of its gross domestic product.7 Countries with such an economic system have no choice but to respond sensitively to changes in the environment of internal and external international trade; thus, their approach to the establishment of new international trade norms is worth noting. In addition, ROK has rich experience with implementing new trade rules while successfully overcoming the East Asian economic crisis of the late 1990s as well as the global financial crisis of 2008. Furthermore, ROK demonstrates how a country can actively promote trade liberalization policies. Moreover, though the multilateral international trade system represented by the World Trade Organization (hereinafter “WTO”) has been insecure in the past, ROK has actively promoted bilateral or regional free trade agreements and has taken the lead in embracing changing trends. As a result, since its free trade agreement (hereinafter “FTA”) with Chile went into effect in 2004, ROK has built 17 FTA networks with a total of 57 countries, including major markets such as the European Union (hereinafter “EU”), the United States (hereinafter “the US”), China, and ASEAN.8 It is particularly noteworthy that ROK has pursued a comprehensive FTA by including norms in various fields such as services, investment, government procurement, intellectual property rights, and technical standards, as well as tariff abolition on goods, in order to maximize the economic impact of the FTA. To further elaborate on ROK’s policies, this chapter first discusses ROK’s recently enforced policies regarding digital trade and introduces the characteristics and status of the trade agreements (multilateral or bilateral) within which ROK is participating. Next, this chapter details ROK’s approach to representative issues of digital trade agreements—specifically, (ⅰ) the prohibition of free cross-border movement of data and localization of data and (ⅱ) the prohibition of an obligation to disclose or transfer source code. In doing so, the article will examine how ROK’s domestic laws interact with its previously concluded digital trade agreements.

6 Breg Natens and Jan Wouters, Regulatory Autonomy Constraints from GATS’ Unconditional Obligations: The Case of the European Union (Working Paper No.132, KU Leuven, Leuven Centre for Global Governance Studies 24, 2014); Panagiotis Delimatsis, The Fragmentation of International Trade Law, 45 J. World Trade 87, 96 (2011). 7 Commission of National Statistics Office, “National Indicator: Import/Export Ratio (to GDP)” www​.index​.go​.kr/​unify/​idx​-info​.do​?idxCd​=​4207​&​clasCd​=​7 accessed April 30, 2022. 8 Ministry of Trade, Industry and Energy, “Status” (February 2022) www​.fta​.go​.kr/​main/​situation/​ kfta/​ov/​accessed April 30, 2022.

180  Research handbook on digital trade

2.

KOREA’S RECENT MOVEMENTS IN DIGITAL TRADE

In 2020, the government of ROK announced the “New Deal Comprehensive Plan” to overcome the pandemic-induced economic crisis and design a brighter future. The Korean version of the New Deal is a declaration for a “great transformation of ROK” to leap forward into an advanced country and consists of four goals: digital, green, human, and regional balance.9 The “Digital New Deal” policy changed the basis of foreign trade policy. This Digital New Deal is a development strategy that utilizes three key concepts: data, network, and artificial intelligence. This strategy is based on the reality, which was triggered by COVID-19, that online consumption and non-face-to-face services such as remote work are spreading. As such, “digital competency” is highlighted as a key element of national competitiveness. Simply put, the strategy is a nationwide digital transformation project that aims to overcome the pandemic-driven economic crisis and presents a new vision of the future by converging information communication technology with all industrial fields of the country.10 This project includes a plan to graft data, network, and AI technologies—key elements of the digital society—into all industries of ROK. Ultimately, the goal is to enhance ROK’s national competitiveness by creating new products or services using these digital technologies, advancing domestic industries, and reducing the time and cost required for land or sea logistics by establishing a smart logistics system. In other words, the Digital New Deal is generating a digital transformation throughout Korean society as a strategic and preparatory investment to lead the global digital economy by developing Korea’s digital capabilities in the post-COVID era. One of the preeminent objectives of these efforts is to create the standards of a new global digital commerce norm that will eventually be applied across various fields such as manufacturing and services. Once this initiative has been implemented, the ROK government expects that the Korean economy will overcome the pandemic crisis and lead innovation in all industries. In addition, the worldwide popularity of Korean content, as represented by the award-winning film “Parasite,” the global fandom of the boy band BTS (Bang Tan Sonyundan), and the Netflix hit “Squid Game” helped the ROK government recognize the limits of international trade norms centered on in-kind products. The rapid development of digital technology has made it easier for the general public to access digital goods. This change has increased the scale of digital goods and services trade explosively. As a result, ROK is preparing for the “digitization of trade.” Specifically, ROK is actively participating in discussions on digital commerce in the existing multilateral system. However, it is difficult for multiple countries with diverse interests to come up with a single agreement. In fact, cases in which new multilateral negotiations have been concluded are rare; aside from the 2013 Trade Facilitation Agreement in the WTO, the only other examples are the objectives outlined in the WTO 12th Ministerial, which have yet to become full agreements.

Ministry of Strategy and Finance, “Korean New Deal” www​.knewdeal​.go​.kr/​#firstPage accessed April 30, 2022. 10 Ministry of Science and Technology Information and Communication, “Korean Digital New Deal” https://​digital​.go​.kr/​resources/​UPLOAD/​2021/​07/​09/​125/​cbc29d4f​-f3a6​-43fb​-834b​-9c55517c310b​.jpeg accessed April 30, 2022. 9

Digital trade agreements and digital policy space  181 Accordingly, ROK recognizes the limitations of this multilateral system and actively promotes digital trade agreements at the bilateral and regional levels. To that end, ROK joined the Regional Comprehensive Economic Partnership (hereinafter “RCEP”) and signed the Korea–Singapore Digital Partnership Agreement (hereinafter “KSDPA”) with Singapore— focusing on digital commerce. In addition, ROK is considering joining the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (hereinafter “CPTPP”), Indo-Pacific Economic Framework (hereinafter “IPEF”), and Digital Economy Partnership Agreement (hereinafter “DEPA”), which include chapters related to digital commerce. By joining these new digital trade agreements, ROK believes that it will be able to ease the digital trade barriers of the other party and support its exporting enterprises in their business activities. Meanwhile, in addition to the above digital trade instruments themselves, ROK is actively participating in the Base Erosion and Profit Shifting (hereinafter “BEPS”) project, which is being considered as one of the tax rules suitable in the era of digital transformation. This project was launched at the G20 Summit in June 2012 to establish fair taxation rights in Organisation for Economic Co-operation and Development (hereinafter “OECD”) member countries. Specifically, BEPS aims to respond to concerns about tax equity between digital companies and traditional manufacturing companies as well as tax erosion and income transfer owing to tax evasion by multinational companies. Thus, BEPS has introduced the so-called digital tax and minimum corporate tax to prevent tax evasion by digital service providers and to project a joint international response to the digital economy. The next section explains in detail ROK’s participation in digital trade agreements.

3.

ACTIVE PARTICIPATION IN THE FORMATION OF NEW INTERNATIONAL NORMS RELATED TO DIGITAL TRADE

3.1

World Trade Organization (WTO) Electronic Commerce Negotiations

During the 2019 World Economic Forum Annual Meeting in Davos, 76 members declared a joint statement11 expressing the members’ strong interest in initiating negotiations at the WTO on “trade-related aspects of electronic commerce.” The number of negotiating countries now stands at 86, including the ROK. Currently, the negotiations cover six main themes: enabling e-commerce; openness and e-commerce; trust and e-commerce; cross-cutting issues; telecommunications; and market access. The negotiation team has successfully concluded eight articles12 and created a consolidated text to conclude articles on customs duties on electronic transmissions, cross-border data

E-commerce discussions in the WTO currently take place in two parallel tracks. The first is known as “the multilateral track,” which was launched in 1998, following the Ministerial Declaration on Global Electronic Commerce. The second involves a subset of WTO members, which are currently part of the WTO JSI on e-commerce. 12 Articles on online consumer protection; electronic signatures and authentication; unsolicited commercial electronic messages; open government data; electronic contracts; transparency; paperless trading; and open internet access. 11

182  Research handbook on digital trade flows, data localization, source code, electronic transactions frameworks, cybersecurity, and electronic invoicing, as well as advanced discussions on market access.13 Countries have different priorities and perspectives regarding the agreement, and there appears to be a wide range of negotiating positions, especially on free flows of cross-border data and privacy. For instance, the US advocates free flows of cross-border data, including personal data, with exceptions for a legitimate public policy objective. The US also wants to include provisions from the US–Mexico–Canada Agreement (hereinafter referred to as “USMCA”) requiring parties to adopt or maintain a legal framework to protect personal information and encouraging the development of interoperability mechanisms.14 On the other hand, China advocates the position that data flows and data storage should be subjects for exploratory discussions rather than solid commitments.15 Regarding privacy, China would like to secure policy space by including the statement that parties “should adopt measures that they consider appropriate and necessary to protect the personal information of electronic commerce users” in the agreement.16 The EU proposed to allow cross-border data flows and prohibit localization requirements while allowing members to “adopt and maintain the safeguards they deem appropriate to ensure the protection of personal data and privacy, including through the adoption and application of rules for the cross-border transfer of personal data.”17 ROK is also currently participating in the negotiations and actively reflecting its position by strengthening ties with like-minded countries. For instance, like the US, ROK also advocates the free flow of cross-border data but wishes to carve out security exceptions to maintain policy space over security issues.18 The major challenge in concluding the negotiations is the digital divide between developing and developed economies. The discussions are very technical in nature, involving much explanation of terminologies and policies applied by developed countries to other developing countries that are not familiar with the technology and do not have a suitable domestic law in place.19 For instance, among the developing countries negotiating group, only 67 per cent of countries have enacted domestic laws related to online consumer protection and data protection and privacy laws, respectively, while the developed countries negotiating group all have the relevant domestic laws in place.20 This reflects a dramatic discrepancy between developed and

13 WTO joint statement initiative on e-commerce statement by Ministers of Australia, Japan, and Singapore. www​.wto​.org/​english/​news​_e/​news21​_e/​ji​_ecom​_minister​_statement​_e​.pdf accessed April 30, 2022. 14 United States, Joint Statement on Electronic Commerce (WTO INF/ECOM/23, April 26, 2019). 15 China, Joint Statement on Electronic Commerce (WTO INF/ECOM/19, April 23, 2019). 16 Ibid. 17 European Union, Joint Statement on Electronic Commerce (WTO INF/ECOM/22, April 26, 2019). 18 Hyo Young Lee, “Recent Updates and Critical Issues in International Regulations regarding Digital Trade,” Institute of Foreign Affairs & National Security, Korea National Diplomatic Academy (2021), p.34. 19 Y. Ismail, “E-commerce in the World Trade Organization: History and Latest Developments in the Negotiations under the Joint Statement,” International Institute for Sustainable Development and CUTS International, Geneva (2020), p.24 www​.iisd​.org/​publications/​e​-commerce​-world​-trade​-organization​ -history​-and​-latest​-developments​-negotiations​-under accessed April 30, 2022. 20 WTO E-Commerce Negotiation Prospects and Korea’s Agenda (World Economy Today Vol 21, No.3), Korea Institute for International Economic Policy, www​.google​.com/​url​?sa​=​t​&​rct​=​j​&​q​=​&​esrc​=​ s​&​source​=​web​&​cd​=​&​ved​=​2ahUKEwiK6​_zp​6Kb1AhVUZd​4KHYScALEQ​FnoECAQQAQ​&​url​=​https​

Digital trade agreements and digital policy space  183 developing countries in terms of readiness to conclude the negotiations. The lack of domestic laws addressing the issue of e-commerce impedes developing countries from examining their markets and coming up with their own appropriate positions and negotiation strategies. Likewise, ROK is also in the process of implementing various domestic laws relating to e-commerce. ROK carefully needs to examine its existing and currently reforming laws to find a suitable position to conclude the ongoing WTO e-commerce negotiations. 3.2

Regional Comprehensive Economic Partnership (RCEP)

Given the digital disparity between developing and developed countries and the resulting varied perspectives on digital commerce, it is unlikely the WTO e-commerce negotiations will be concluded quickly. Therefore, individual countries seek to establish regional economic cooperation systems based on historical ties, geographical proximity, and economic usefulness. Along with this trend, RCEP is one of the new representative economic cooperation systems in Asia in which ROK actively participates. One of the promotional goals of RCEP is regional economic integration centered on Asia-Pacific countries. To that end, a total of 15 countries21 signed RCEP on November 15, 2020. Those countries represent a total population of 2.3 billion—30 percent of the world’s population—and a total gross domestic product (GDP) of around $38,813 billion—30 percent of global GDP. In addition, five economies (Australia, China, Indonesia, Japan, and South Korea) among the RCEP are members of the world’s 20 largest economies (that is, the Group of 20 or G20).22 In relation to digital commerce, the most noteworthy chapter among RCEP’s 20 chapters is the one that discusses e-commerce. Scholars contend that this e-commerce chapter (Chapter 12) has introduced rules to promote e-commerce and create an environment of trust, reflecting the digital economic activity trend that increased rapidly following the COVID-19 pandemic.23 This e-commerce chapter builds on the agreement to maintain the current practice of not levying tariffs on electronic transmissions, as agreed at the 12th WTO Ministerial Meeting (where the moratorium was extended until at least 2024). Specifically, this chapter of the RCEP contains provisions promoting paperless trade, cross-border trade using e-authentication and e-signatures, online consumer protection and privacy, and cross-border e-commerce. Within the RCEP, the Articles on Promotion of Cross-Border E-Commerce contain new rules governing the actions of a party affecting e-commerce. According to these rules, except in circumstances where it is deemed necessary to achieve legitimate public policy or security

%3A​%2F​%2Fwww​.kiep​.go​.kr​%2FgalleryDownload​.es​%3Fbid​%3D0003​%26list​_no​%3D9356​%26seq​ %3D1​&​usg​=​AOvVaw0zV23wSRzwd4fQB8VgAvqm accessed April 30, 2022. 21 These countries include the ten members of the Association of Southeast Asian Nations (ASEAN): Brunei-Darussalam, Cambodia, Indonesia, Laos, Malaysia, Myanmar, Philippines, Singapore, Thailand, and Viet Nam. Also, five regional countries with which ASEAN has free trade agreements: Australia, China, Japan, South Korea and New Zealand. 22 New Zealand Foreign Affairs & Trade, Regional Comprehensive Economic Partnership www​.mfat​.govt​.nz/​en/​trade/​free​-trade​-agreements/​free​-trade​-agreements​-in​-force/​regional​ -comprehensive​-economic​-partnership​-rcep/​rcep​-overview/​ accessed April 30, 2022. 23 For more details in RCEP, see, e.g., Eun Jung Oh, Digital Trade Regulation in the Asia-Pacific: Where Does It Stand? Comparing the RCEP E-Commerce Chapter with the CPTPP and the JSI, 48 Legal Issues of Economic Integration (2021) p.403.

184  Research handbook on digital trade objectives, a party cannot—as a condition for conducting business in that party’s territory—(ⅰ) require that computer equipment be installed in its territory (data localization)24 or (ⅱ) obstruct the free flow of data by electronic means.25 The RCEP, which came into effect in Korea on February 1, 2022, is the first mega FTA signed by the ROK. Compared to the previous bilateral (1:1) FTA, the RCEP is a 15:15 multilateral FTA. Therefore, the government is preparing, both at the institutional and the administrative levels, to help domestic companies more easily understand the contents of the agreement. One such method of preparation is seen where each government department is enacting or revising RCEP implementation laws. For example, on December 22, 2021, the ROK Ministry of Strategy and Finance announced that it would promote the revision of the enforcement ordinance and enforcement rules of the Act on Special Cases of Customs Act for FTA Implementation.26 Accordingly, a table of tariff rates applied to imported goods originating in RCEP member countries was prepared. Likewise, the procedures required (for example, notification procedure to the other country) in the case of taking trade remedies (safeguarding, anti-dumping duties, subsidy countervailing duties, and so on) were included in the laws and regulations. In addition, for the smooth implementation of the agreement, details of the agreement, such as the criteria for determining the country of origin to be applied to imported goods, the issuance method and period of validity of the certificate of origin, and the format of the certificate of origin, have been reflected in the enforcement ordinance and enforcement rules. 3.3

Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP)

On December 13, 2021, ROK’s Deputy Prime Minister for Economic Affairs announced that he would initiate the procedure based on social discussions with various stakeholders, stating at the Ministerial Meeting: “Considering economic and strategic values through trade and investment expansion, the CPTPP will be joined in earnest.”27 Such a declaration meant that Article 12.14: Location of Computing Facilities 1. The Parties recognise that each Party may have its own measures regarding the use or location of computing facilities, including requirements that seek to ensure the security and confidentiality of communications. 2. No Party shall require a covered person to use or locate computing facilities in that Party’s territory as a condition for conducting business in that Party’s territory. 3. Nothing in this Article shall prevent a Party from adopting or maintaining: (a) any measure inconsistent with paragraph 2 that it considers necessary to achieve a legitimate public policy objective, provided that the measure is not applied in a manner which would constitute a means of arbitrary or unjustifiable discrimination or a disguised restriction on trade; or (b) any measure that it considers necessary for the protection of its essential security interests. Such measures shall not be disputed by other Parties. 25 Ibid. 26 Ministry of Strategy and Finance, “Press Release” (December 22, 2021) www​.moef​.go​.kr/​nw/​nes/​ detailNesDtaView​.do​?menuNo​=​4010100​&​searchNttId1​=​MOSF​_000000000057854​&​searchBbsId1​=​ MOSFBBS​_000000000028 accessed April 30, 2022. 27 Ministry of Strategy and Finance, “Press Release” (December 13, 2021) www​.moef​.go​.kr/​ nw/​nes/​detailNesDtaView​.do​?searchBbsId1​=​MOSFBBS​_000000000028​&​searchNttId1​=​MOSF​ _000000000057723​&​menuNo​=​4010100 accessed April 30, 2022. 24

Digital trade agreements and digital policy space  185 ROK officially joined the CPTPP about eight years after reviewing its 2013 Trans-Pacific Partnership (hereinafter “TPP”) membership, the predecessor of the CPTPP. As it relates to digital commerce, the main provisions of the e-commerce chapter of CPTPP (Chapter 14) include non-discriminatory treatment of digital goods (Article 14.4), cross-border Transfer of Information by Electronic Means (Article 14.11), and location of computing facilities (Article 14.13). With respect to the article on location of computing facilities, the text offers policy space for states to carve out their domestic laws if such laws achieve a legitimate public policy objective and do not constitute arbitrary or unjustified discrimination or disguised restrictions on trade. 3.4

Indo-Pacific Economic Framework (IPEF)

On November 11, 2021, at the East Asia Summit, the United States introduced the IPEF. The Biden administration is expected to push the IPEF forward vigorously, with the goal of launching the framework in 2022. The agenda of this new framework includes three main topics: supply chain, the digital economy, and climate change. In an effort to promote the IPEF, the US emphasized its pro-labor nature, alluding to this strong new trade agreement’s ability to create jobs, enhance supply chain resilience, decarbonize trade, establish worker standards and infrastructure, and even enact standards and norms in the digital field, which are not covered by existing trade rules.28 The US expects its Indo-Pacific regional friends will participate actively in the IPEF to counter and defend against Chinese economic influence.29 However, to date there have been no specific discussions of digital commerce-related norms to be included in the IPEF. Nevertheless, many potential members of the IPEF, including the leaders of the US House Foreign Affairs Subcommittee on Asia and Republican members of the US Senate Finance committee, have expressed support for a digital trade agreement. By lowering barriers to digitally enabled trade and establishing rules and standards that allow for nondiscriminatory competition, digital trade agreements may help US companies compete effectively in the international marketplace, while also expanding US firm and consumer access to international products that rely on digital technologies. Scholars opine that a US-led digital agreement could help establish regional and global US leadership on an emerging trade policy issue and serve as a counterweight to China’s recent advances into regional trade agreements.30 To that end, the US House of Representatives held various hearings in relation to the IPEF's digital norms on January 1, 2021, and discussed the strategic importance of the new digital trade agreement.31 28 Inside US Trade, “Eyeing an Indo-Pacific Framework: Waiting for More Flesh on the Bones” (November 24, 2021) https://​insidetrade​.com/​daily​-news/​eyeing​-indo​-pacific​-framework​-waiting​-more​-​ %E2​%80​%98flesh​-bones​%E2​%80​%99 accessed April 30, 2022. 29 Nikkei Asia, “U.S. Pushes for Indo-Pacific Rules on Digital Trade, AI” (November 18, 2021) https://​asia​.nikkei​.com/​Economy/​Trade/​U​.S​.​-pushes​-for​-Indo​-Pacific​-rules​-on​-digital​-trade​-AI accessed April 30, 2022. 30 Congressional Research Service, “Biden Administration Plans for an Indo-Pacific Economic Framework” (February 25, 2022) https://​crsreports​.congress​.gov/​product/​pdf/​IN/​IN11814 accessed April 30, 2022. 31 House of Foreign Affairs Committee, “Strategic Importance of Digital Economic Engagement in the Indo-Pacific” (January 19, 2022) https://​foreignaffairs​.house​.gov/​hearings​?ID​=​AF933B39​-5030​ -4CFB​-934D​-08D476FC7284 accessed April 30, 2022.

186  Research handbook on digital trade Meanwhile, on November 19, 2021, Catherine Tai, head United States Trade Representative (hereinafter “USTR”), visited ROK and inquired about ROK’s interest and participation in the IPEF.32 Accordingly, on December 29, 2021 the government of ROK also decided to conduct an in-depth analysis of the background and structure of the US IPEF initiative, and its impact on the ROK. As a result, an internal review of the ROK’s response to the future cooperation agenda for each sector in the Indo-Pacific region has been initiated.33 3.5

Promotion of Bilateral or Small-Scale Digital Trade Agreements

In addition to the multilateral cooperation system discussed above, the ROK is pushing for accession to an agreement that specializes in digital commerce norms. ROK has already had experience introducing standards related to digital commerce in the Korea–US FTA signed with the United States in 2007. Specifically, the Korea–US FTA made mandatory the non-taxation of digital products, mutual recognition of digital signatures and certifications, non-discriminatory treatment for digital goods, and provisions for cooperation on paperless trade (Article 15.8). In addition, the Korea–US FTA stipulated cooperation between the two parties in relation to the smooth transfer of information across borders and free access to the other country’s internet. Based on the standards at the time of the negotiations in 2007, the content was evaluated as very progressive.34 However, as some commentators note, ROK’s previously concluded e-commerce chapters indicate that “while provisions on consumer protection, paperless trading, and personal information protection show relatively high levels of substantive consistency, others on non-discrimination, defining ‘electronic transmissions’, domestic regulatory framework, and electronic authentication do not.”35 This failure to include highly liberalized non-discrimination clauses before enacting digital trade agreements more recently has been seen not only in Korea but also in other East Asian countries such as Japan and China.36 To overcome these limitations, the government of ROK plans to adapt to the latest norms related to digital trade by joining a bilateral or small-scale digital trade agreement that sets out highly liberalized rights and obligations. 3.5.1 Korea–Singapore Digital Partnership Agreement (KSDPA) On June 22, 2020, ROK initiated the KSDPA negotiations. On December 15, 2021, after ten rounds of negotiations, ROK reached a final agreement and issued a joint statement with 32 USTR, “Readout of Ambassador Katherine Tai’s Meetings with Korean Minister for Trade Yeo Han-Koo” (November 19 2021) https://​ustr​.gov/​about​-us/​policy​-offices/​press​-office/​press​-releases/​ 2021/​november/​readout​-ambassador​-katherine​-tais​-meetings​-korean​-minister​-trade​-yeo​-han​-koo accessed April 30, 2022. 33 Ministry of Trade, Industry and Energy, “Press Release” (December 29, 2021) www​.motie​.go​ .kr/​motie/​ne/​presse/​press2/​bbs/​bbsView​.do​?bbs​_seq​_n​=​165098​&​bbs​_cd​_n​=​81​&​currentPage​=​1​&​search​ _key​_n​=​title​_v​&​cate​_n​=​&​dept​_v​=​&​search​_val​_v​=​%EC​%9D​%B8​%EB​%8F​%84​%ED​%83​%9C​%ED​ %8F​%89​%EC​%96​%91​%20​%EA​%B2​%BD​%EC​%A0​%9C​%20​%ED​%94​%84​%EB​%A0​%88​%EC​ %9E​%84​%EC​%9B​%8C​%ED​%81​%AC accessed April 30, 2022. 34 Evan Y. Kim, E-commerce in South Korean FTAs: Policy Priorities and Provisional Inconsistencies, 18 World Trade Review (2019), 85. 35 Ibid at 97. 36 See generally, David Collins et al, A Soft Landing for Developing Countries and Non-Discrimination in Digital Trade: Possible Lessons from Asian Countries, 55 Journal of World Trade, 649, 649 (2021).

Digital trade agreements and digital policy space  187 Singapore.37 According to the agreement, both ROK and Singapore decided to expedite the domestic procedures for the entry into force of the KSDPA, including legal scrubbing of the agreement, to formally sign the KSDPA as soon as possible. The KSDPA is the first digital-only trade agreement signed by the ROK. Both sides carried out the negotiation process by maximizing the use of digital technology. For example, both sides used a real-time videoconference method based on digital technology to declare the start of negotiations for the agreement, and the joint declaration related to the conclusion of the negotiation was recorded as a digital file. The signature was electronically signed and exchanged through a tablet PC. The two sides have not yet released the text of the KSDPA, but according to press releases from both sides the KSDPA is known to contain a wide range of mandates, such as requiring the parties to: (ⅰ) adopt transparent and facilitative rules to promote secure cross-border e-payments (E-Payment); (ⅱ) encourage innovation by making government information open and accessible (Open government information); (ⅲ) accept electronic versions of trade administration documents for efficient cargo clearance and facilitate cross-border supply chain digitalization (Paperless trading); (ⅳ) remove access and transfer of private keys and related technologies, as a condition of market access (Cryptography); (ⅴ) guard against fraudulent or misleading conduct online through regulations (Online consumer protection); and (ⅵ) promote jobs for small and medium sized enterprises (hereinafter “SMEs”) and facilitate their connection with suppliers, buyers, and business partners (SME cooperation).38 Likewise, the KSDPA carved out policy space to prohibit data localization. Specifically, Art. 14.15 (Location of Computing Facilities) allows parties to take measures to achieve a legitimate public policy objective as long as the measures would not constitute arbitrary or unjustifiable discrimination or a disguised restriction on trade.39 3.5.2 Digital Economy Partnership Agreement (DEPA) On May 11, 2021, ROK held a public hearing to collect public opinions on joining the DEPA, which was signed previously by New Zealand, Singapore, and Chile.40 The following September, ROK initiated the process for joining the DEPA.41 The DEPA contains a variety of articles, including the Non-Discriminatory Treatment of Digital Products (Article 3.3), Information and Communication Technology Products that Use Cryptography (Article 3.4), Personal Information Protection (Article 4.2), Cross-Border Transfer of Information by Electronic Means (Article 4.3), and Location of Computing Facilities (Article 4.4). Notably,

Ministry of Trade, Industry and Energy, “Press Release” (December 15, 2021) https://​english​ .motie​.go​.kr/​en/​pc/​pressreleases/​bbs/​bbsView​.do​?bbs​_cd​_n​=​2​&​bbs​_seq​_n​=​898 accessed April 30, 2022. 38 Korea Ministry of Trade, Industry and Energy, “Korea–Singapore Digital Partnership Agreement (KSDPA)” www​.fta​.go​.kr/​webmodule/​_PSD​_FTA/​ksdpa/​1/​DPA​_eng​.pdf accessed April 30, 2022. 39 Ibid. 40 According to the domestic law of the Republic of Korea, opinions of stakeholders and related experts must be collected prior to the conclusion of a trade treaty. Act on the Conclusion Procedure and Implementation of Commercial Treaties, Article 7(1): “The Minister of Trade, Industry and Energy shall hold a public hearing to hear opinions from interested persons and relevant experts before formulating a plan for concluding a commerce treaty.” 41 Ministry of Trade, Industry and Energy, “Press Release” (May 15, 2021) http://​english​.motie​.go​ .kr/​en/​pc/​pressreleases/​bbs/​bbsView​.do​?bbs​_cd​_n​=​2​&​bbs​_seq​_n​=​870 accessed April 30, 2022. 37

188  Research handbook on digital trade the DEPA incorporates two articles that are not addressed by the CPTPP—e-invoicing and e-payment. Furthermore, the DEPA contains several articles that have the potential to impact future regulation of the digital economy. For instance, Module 8 of the DEPA addresses emerging trends and technologies, while Module 9 talks about innovation and the digital economy. Although these modules do not have legally binding obligations, their inclusion in the DEPA generates opportunities for scholars to discuss their provisions. For example, academic discussions on FinTech (Financial Technology) and/or AI (Artificial Intelligence) topics may increase because of Module 8’s inclusion of articles on FinTech cooperation— encouraging parties to promote development of FinTech solutions as well as cooperation between firms—and AI technologies—encouraging governments to adopt AI governance frameworks.42 3.5.3 Base Erosion Profit Shifting (BEPS) Traditionally, tax corporations’ authority has been exercised by the state having jurisdiction over the physical location of the corporation’s permanent establishment. However, this traditional rule has allowed multinational corporations that do not have a separate permanent establishment in a specific country to avoid taxation in the countries where their markets are located by transferring profits from one country to another with a lower tax rate. This pattern is particularly popular among multinational companies whose main business activity is the provision of digital services. In fact, initial discussions surrounding the new international tax system revolved around multinational “digital” service providers. According to the OECD, the magnitude of tax avoidance by such multinational corporations varies between US$100 and 240 billion annually for each OECD member, which is roughly equivalent to the amount of their respective total corporate tax revenues. It is estimated that these annual tax revenue losses account for about 4–10 percent of each state’s global corporate income tax revenue.43 To address this issue, in June 2021 the Group of 7 (hereinafter “G7”) and the OECD designed and announced preliminary details regarding a new international tax system to guarantee that taxation authorities of countries where multinational corporations operate prevent profit taking. At the inaugural G7 Finance Ministers and Central Bank Governors Meeting, held on June 4–5, 2021 in London, the G7 agreed on the conditions under which the jurisdiction of the territory where direct sales are derived can tax multinational corporations, regardless of the location of their permanent establishment (hereinafter “G7 Agreement”). Specifically, under the G7 Agreement, the world’s largest and most profitable multinational corporations are now taxable by countries that are hosts to markets where those corporations derive profits of at least 20 percent in excess of 10 percent of their global margins.44

42 Rachelle Taheri et al, “DEPA: The World’s First Digital-Only Trade Agreement,” Asia Pacific Foundation of Canada, available at www​.asiapacific​.ca/​publication/​depa​-worlds​-first​-digital​-only​-trade​ -agreement accessed August 2, 2022. 43 For more information, see OECD/G20 Inclusive Framework on BEPS, Base Erosion and Profit Shifting, www​.oecd​.org/​tax/​beps/​flyer​-inclusive​-framework​-on​-beps​.pdf accessed April 30, 2022. 44 For more information, see G7 Finance Ministers and Central Bank Governors Communiqué, https://​assets​.publishing​.service​.gov​.uk/​government/​uploads/​system/​uploads/​attachment​_data/​file/​ 991640/​FMCBGs​_communique​_​-​_5​_June​.pdf accessed April 30, 2022.

Digital trade agreements and digital policy space  189 The G7 Agreement was embodied at the 12th General Assembly of the OECD/G20 Inclusive Framework on Base Erosion and Profit Shifting (hereinafter “IF BEPS”)45 held in July 2021 (hereinafter “OECD Agreement”). The OECD Agreement presented two approaches (the so-called Two-Pillar Solution) to prevent tax challenges by multinational corporations.46 In the end, as a result of the 13th IF BEPS General Assembly held on October 8, 2021, the final agreement on Pillar 147 and Pillar 248 of the OECD Agreement was drawn up.49 ROK is a member of the IF BEPS Council, which consists of a total of 24 countries, and participates in all IF BEPS steering committees and working-level meetings. ROK’s active participation in resolving BEPS issues is not limited to the Council; in fact, in February 2021, the ROK’s Ministry of Strategy and Finance’s Customs Office established the New International Tax Standards Division to design the domestic taxation system in line with the establishment of new international tax rules. As demonstrated by its New International Tax Standards Division, the government of ROK has become heavily involved with the international tax sector that arose with the growth of global giant digital companies. ROK’s involvement with the international tax sector aims to keep pace with the master plan of the OECD Agreement, which is scheduled to be implemented in 2023 (once discussions finalize, the agreement is ratified and signed, and domestic legislation is enacted). Through the OECD Agreement, the ROK government fully expects and intends to increase tax revenues as the agreement grants the government the right to validly tax multinational corporations conducting business in its territory. Nevertheless, the Korean government acknowledges that it will have to share the authority to tax multinational corporations headquartered in Korea with other countries in which those corporations also operate.

For more information, see OECD BEPS, International collaboration to end tax avoidance, www​ .oecd​.org/​tax/​beps/​ accessed April 30, 2022. 46 For more information, see OECD/G20 Base Erosion and Profit Shifting Project, Statement on a Two-Pillar Solution to Address the Tax Challenges Arising From the Digitalisation of the Economy, www​.oecd​.org/​tax/​beps/​statement​-on​-a​-two​-pillar​-solution​-to​-address​-the​-tax​-challenges​-arising​-from​ -the​-digitalisation​-of​-the​-economy​-july​-2021​.pdf accessed April 30, 2022. 47 The first approach (“Pillar 1”) is called the “unified approach.” This approach prescribes that when a multinational corporation generates sales and profits beyond a certain threshold (25 percent of residual profit defined as profit in excess of 10 percent of revenue) in a given market, tax authority is granted to the country in which that market is located, regardless of whether the company has a permanent establishment or physical presence there. 48 The second approach (“Pillar 2”) establishes the lowest corporate tax rate at a global level so that the income of multinational corporations is not taxed at an effective rate below the threshold (the minimum tax rate will be 15 percent). This is to prevent tax avoidance not addressed by Pillar 1. The specific minimum corporate tax rate will be finalized at the time of agreement in October 2021. For now, the OECD Agreement tentatively set a minimum corporate tax rate at least 15 percent for all OECD members. The ultimate goal of Pillar 2 is to prevent countries from competing against one another in excessively cutting corporate tax rates to provide tax havens, and ultimately to deter multinational corporations from engaging in tax shopping. 49 For more information, see Statement on a Two-Pillar Solution to Address the Tax Challenges Arising from the Digitalisation of the Economy, www​.oecd​.org/​tax/​beps/​statement​-on​-a​-two​-pillar​ -solution​-to​-address​-the​-tax​-challenges​-arising​-from​-the​-digitalisation​-of​-the​-economy​-october​-2021​ .pdf accessed April 30, 2022. 45

190  Research handbook on digital trade

4.

IMPLEMENTATION OF ROK’S DIGITAL TRADE AGREEMENT

This section examines ROK’s articles on freedom of cross-border transfer of data, prohibition of data localization, and prohibition of obligations on disclosure or transfer of core technologies. The freedom of cross-border data transfer refers to the notion that individual countries are prohibited from hindering cross-border movement of information by electronic means in accordance with business activities. The prohibition of data localization refers to an individual country’s inability to require other countries to use or have a computing facility in that country as a condition of conducting business in that country. The prohibition on data localization is increasingly important because it encourages states with cloud technology industries to continue their development.50 The successful use of cloud technology requires the industries to fragment data and store it in an optimal location. Therefore, to further enhance the use of cloud technology, states should be prohibited from forcing a business to locate data in a certain country or area.51 With respect to the prohibition on the disclosure or transfer of source code (that is, algorithms), parties cannot require transfer of or access to the source code of the software owned by either party as a condition of the import, distribution, sale, or use of the software. While ROK accepts trade norms on disclosure and transfer of core technologies, exceptions are generally permitted for legitimate public policy objectives. This following section discusses ROK’s trade agreements relating to the two issues explained above. In addition, the section examines ROK’s domestic laws and explores their compliance with those trade agreements. 4.1

Freedom of Cross-Border Transfer of Data and Prohibition of Data Localization

4.1.1 Analysis of regulations in trade agreements concluded by ROK ROK has prior experience including the free movement of data and the prohibition of data localization in both bilateral and multilateral trade agreements. For example, Article 15.8 of the Korea–US FTA stipulates that a party shall “endeavor to refrain” from taking or maintaining unnecessary trade-restricting measures on the flow of electronic information across borders.52 In addition, Articles 12.14 and 12.15 of the RCEP specify obligations to prohibit localization of computer equipment and restrictions on cross-border information transfer.53

Nam Jong Kim, “Global Trend and Its Response in Data Localization in Digital Trade” 30(2) Financial Research Brief, Korea Institute of Finance (2021), the brief is available at www​.google​.com/​ url​?sa​=​i​&​rct​=​j​&​q​=​&​esrc​=​s​&​source​=​web​&​cd​=​&​cad​=​rja​&​uact​=​8​&​ved​=​0CAMQw7AJahcKEwig6​-q​ -2a​f5AhUAAAAA​HQAAAAAQAg​&​url​=​https​%3A​%2F​%2Fwww​.kif​.re​.kr​%2Fkif2​%2Fpublication​ %2Fviewer​.aspx​%3Fismail​%3D1​%26controlno​%3D282583​%26email​%3D​%5B​%24email​%24​%5D​&​ psig​=​AOvVaw2yTNTUtD8x9U2MlFb​-PbyH​&​ust​=​1659513881949612 accessed August 2, 2022. 51 Ibid. 52 Article 15.8 of the Korea–US FTA is available at https://​ustr​.gov/​sites/​default/​files/​uploads/​ agreements/​fta/​korus/​asset​_upload​_file816​_12714​.pdf accessed April 30, 2022. 53 The RCEP text is available at https://​fta​.go​.kr/​webmodule/​_PSD​_FTA/​rcep/​doc/​eng/​1/​23​ %20Chapter​%2012​.pdf accessed April 30, 2022. 50

Digital trade agreements and digital policy space  191 Given that the Korea–US FTA was concluded in June 2007, it seems that ROK acknowledged the significance of digital trade early in the digital trade era. However, as the peculiar wording of Article 15.8 of the Korea–US FTA—“endeavor to refrain”54—indicates, this article does not go so far as to impose binding legal obligations. Meanwhile, the RCEP, which concluded in 2019, contains more compulsory language than the corresponding provisions of the Korea–US FTA. For example, it states, “[n]o Party shall require a covered person to use or locate computing facilities in that Party’s territory as a condition for conducting business in that Party’s territory” and “[a] Party shall not prevent cross-border transfer of information by electronic means where such activity is for the conduct of the business of a covered person.”55 Such language mandates stronger obligations compared to the relevant provisions of the Korea–US FTA.56 As noted previously, prohibiting data localization encourages states to develop and grow the cloud technology industry. Thus, these binding provisions will help RCEP members eliminate domestic measures that force foreign firms to locate data in their countries. Nevertheless, it is worth noting that the RCEP member states explicitly allow policy space that may weaken the existing obligations. For example, the RCEP states that “[t]he Parties recognize that each Party may have its own measures regarding the use or location of computing facilities, including requirements that seek to ensure the security and confidentiality of communications” and “[t]he Parties recognize that each Party may have its own regulatory requirements concerning the transfer of information by electronic means.”57 Moreover, the RCEP also allows a party to adopt or maintain any measures it deems necessary to (i) achieve legitimate public policy objectives or (ii) protect its essential security interests, and specify the permitted exceptions.58 Most notably, it should be emphasized that the RCEP provides that the “necessity behind the implementation of such legitimate public policy” shall be determined by the state party.59 Thus, without third party oversight of these public policy exceptions, the mandates may be seriously weakened. In essence, these police space provisions may serve as the basis for introducing measures that are inconsistent with the stipulated obligation to “prohibit free border movement of data and localization of data.” For instance, states may opportunistically utilize the policy space provisions to postpone liberalization or market openings in digital trade for the purpose of protectionism. In the negotiating practice, however, these policy space provisions are frequently the consequence of accepting liberalization articles and are therefore—to a certain extent—unavoidable.

54 Article 15.8 CROSS-BORDER INFORMATION FLOWS […] “Recognizing the importance of the free flow of information in facilitating trade, and acknowledging the importance of protecting personal information, the Parties shall endeavor to refrain from imposing or maintaining unnecessary barriers to electronic information flows across borders.” 55 RCEP, supra note 53. 56 The Korea–US FTA text is available at https://​fta​.go​.kr/​webmodule/​_PSD​_FTA/​rcep/​doc/​eng/​1/​ 23​%20Chapter​%2012​.pdf accessed April 30, 2022. 57 Article 12.14 and 12.15 in the RCEP text, supra note 53. 58 Ibid. 59 Ibid at footnote 12 and 14.

192  Research handbook on digital trade 4.2

ROK’s Approach to Freedom of Cross-Border Transfer of Data and Prohibition of Data Localization

4.2.1

Enactment of the Framework Act on Data Industry Promotion and Utilization Promotion On October 19, 2021, ROK enacted the Framework Act on Data Industry Promotion and Utilization Promotion (hereinafter “Data Industry Act”) to lay the foundation for the development of the data industry (the effective date was April 20, 2022).60 The Data Industry Act stipulates that it is the government’s responsibility to ensure that data can be moved in and out of the country to the extent necessary for facilitating data production, transaction, and utilization.61 The Act imposed an obligation to establish an institutional basis for smooth transfer to the person or a third party in a form that the information-processing device can process.62 It also requires that international trends in the data industry be identified and international cooperation with foreign governments, international organizations, or foreign companies and organizations be promoted.63 As the Data Industry Act enacts regulations for the “facilitation of data movement” everywhere, it helps gauge the ROK government’s approach to liberalization of data movement. This Act will significantly impact the interpretation and future enactment and revision of provisions related to data movement under other laws such as the Personal Information Protection Act and the Credit Information Act. In addition, the enforcement of the Data Industry Act on its own is unlikely to constitute a violation of the ROK’s concluded digital agreement. There is no specific content that might violate the digital agreement itself. 4.2.2

Addition of conditions for international movement of personal information; forced localization of cloud service-related facilities in the financial sector or public institutions The ROK’s domestic law may conflict with the obligations of trade agreement provisions related to the free movement of data regarding personal information protection, administrative and public cloud security, financial service stability, and supervisory authority. As stated above, the provisions related to “free border movement of data and the prohibition of data localization” in the RCEP concluded by ROK offer more flexibility and broader policy space exceptions. This allows ROK to judge the justification of the public policy measures on its own. In this regard, it is unlikely that the special measures to secure the ROK’s policy space will constitute a violation of the agreement unless there are special circumstances. For example, according to the Personal Information Protection Act, information and communications service providers are required to obtain the user’s prior consent to provide (including inquired) or transfer users’ personal information overseas.64 In addition, it stipulates that sufficient protection measures must be taken when personal information is transferred

62 63 64 60 61

Data Industry Act. Para. 2 of Art. 3, Data Industry Act. Art. 15 of the Data Industry Act. Art. 29 of the Data Industry Act. Para. 2 of Art. 39-12 in the Personal Information Protection Act.

Digital trade agreements and digital policy space  193 abroad with consent.65 Although this may violate Article 12.15(2) of RCEP, which states that measures restricting cross-border transfer of information are prohibited, it is a measure that is particularly sensitive to arbitrary or undue discrimination or trade for legitimate public policy objectives of the protection of personal data. Unless it is used as a disguised restriction, it is deemed justifiable by the exception clause. In addition, according to the Electronic Financial Supervisory Regulations of the ROK Financial Services Commission, financial companies or electronic financial companies are permitted to process and use personally identifiable information and credit information (collectively, personal financial information) through the cloud. However, when personal financial information is included in the contents of such business processing, the information-processing system must be installed in the ROK.66 In addition, computer rooms and disaster recovery centers of financial companies headquartered in ROK must be installed in ROK.67 Likewise, according to the standards for data protection of cloud computing services of the ROK Ministry of Science and ICT, localization of cloud facilities in the public sector is required. In other words, in order for private operators to provide cloud services for public institutions, cloud systems and data must be located in Korea.68 Based upon these, ROK is forcing the localization of certain parts of facilities in relation to cloud services in the financial sector or public institutions. However, this is done such that the measures may also fall under those necessary to achieve legitimate public policy goals, such as the safe management of personal financial information or public institution information and the need to access information in case of emergency. Simply put, the introduction of such measures does not seem to be a violation of the ROK’s concluded trade agreement. 4.3

Prohibition of the Obligation to Disclose or Transfer Source Code

4.3.1 Analysis of regulations in trade agreements concluded by ROK Source code is the set of instructions that generates digital software. The provision on the prohibition of the obligation to disclose or transfer source code refers not only to the source code itself but also the “algorithm for the source code.” These provisions reassure foreign companies that they do not have to disclose or transfer the source code underlying their software.69 The USMCA, for example, states that

Para. 4 of Art. 39-12. Para. 8 of the Art. 14-2. 67 Art. 1 Item #11. 68 Attached Table #4 , Item #14.2.1 in the standards for data protection of cloud computing services of the ROK Ministry of Science and ICT. The text is available at www​.law​.go​.kr/​%ED​%96​%89​%EC​ %A0​%95​%EA​%B7​%9C​%EC​%B9​%99/​%ED​%81​%B4​%EB​%9D​%BC​%EC​%9A​%B0​%EB​%93​%9C​ %EC​%BB​%B4​%ED​%93​%A8​%ED​%8C​%85​%EC​%84​%9C​%EB​%B9​%84​%EC​%8A​%A4​%EC​%A0​ %95​%EB​%B3​%B4​%EB​%B3​%B4​%ED​%98​%B8​%EC​%97​%90​%EA​%B4​%80​%ED​%95​%9C​%EA​ %B8​%B0​%EC​%A4​%80 accessed 30 April 2022. 69 For the first comprehensive paper on source code, see generally, C. Dorobantu, F. Ostmann, and C. Hitrova (2021). Source code disclosure: A primer for trade negotiators. In I. Borchert and L.A. Winters (eds), Addressing Impediments to Digital Trade (pp.105–40). London: CEPR Press. 65 66

194  Research handbook on digital trade [n]o Party shall require the transfer of, or access to, a source code of software owned by a person of another Party, or to an algorithm expressed in that source code, as a condition for the import, distribution, sale or use of that software, or of products containing that software, in its territory.70

Notably, the ROK’s previously concluded trade agreements do not contain norms that prohibit the compulsory disclosure or transfer of software source codes (algorithms). Likewise, the RCEP also lacks such provisions relating to source code or algorithm handling. However, the KSDPA does contain a Source Code Protection clause, requiring a party to remove access and transfer of source codes, including algorithms expressed within, as a condition of market access.71 In addition, the CPTPP, which the ROK is attempting to join, also has a regulation relating to the prohibition of obligation to disclose or transfer source code.72 The KSDPA is currently in the process of ratification under the domestic laws of both parties, and the original text has not been made public. Therefore, we will briefly review the applicable provisions of the CPTPP. According to Art 14.17 of the CPTPP e-commerce chapter, “[n]o Party shall require the transfer of, or access to, source code of software owned by a person of another Party, as a condition for the import, distribution, sale or use of such software, or of products containing such software, in its territory.”73 These principles, however, are essentially limited to mass-market software or products containing such software, and do not apply to software used for critical infrastructure.74 For reference, the USMCA and the Agreement Between the USA and Japan Concerning Digital Trade (hereinafter “USJDTA”) include algorithms that make up the source code in addition to source code as protection targets, while the CPTPP has a feature that limits the protection target to source code. Furthermore, CPTPP has a policy space exception clause. Specifically, in the following cases, even if the obligation to disclose the source code is imposed, the agreement is not to be violated: (ⅰ) the inclusion or implementation of terms and conditions related to the provision of source code in commercially negotiated contracts or (ⅱ) a party requiring the modification of source code of software necessary for that software to comply with laws or regulations that are not inconsistent with the agreement.75 In other words, CPTPP limits the scope of the provision and has broader policy space provisions than USMCA or other agreements. Although the text strictly prohibits disclosure or transfer of source code, the policy space provisions allow states to require the inclusion or implementation of terms and conditions of source code or modification of source code.

USMCA Art. 19.16 and USJDTA Art. 17. Singapore Ministry of Trade, Industry and Energy, “Korea–Singapore Digital Partnership Agreement (KSDPA)” www​.mti​.gov​.sg/​Improving​-Trade/​Digital​-Economy​-Agreements/​KSDPA accessed April 30, 2022. 72 Art. 14.17 (Source Code) in the CPTPP. The text is available at www​.mfat​.govt​.nz/​assets/​Trade​ -agreements/​TPP/​Text​-ENGLISH/​14​.​-Electronic​-Commerce​-Chapter​.pdf accessed April 30, 2022. 73 Ibid. 74 Para. 2 of Art.14.17 in the CPTPP. 75 Part 3 of Art. 14.17 in the CPTPP. 70 71

Digital trade agreements and digital policy space  195 4.4

ROK’s Approach on the Prohibition of the Obligation to Disclose or Transfer Source Code

4.4.1 Cloud Security Assurance Program If the KSDPA goes into effect or ROK joins the CPTPP, the ROK’s Cloud Security Assurance Program (hereinafter “CSAP”) will likely be the most controversial one. The purpose of CSAP is to supply private cloud services with proven safety and reliability to the public sector through the verification of a publicly trusted certification body. When a private business provides cloud services to public institutions, the CSAP system requires a certification body to evaluate compliance with information protection standards and obtain cloud security certification in accordance with the “Act on the Development of Cloud Computing and Protection of its Users.”76 In ROK, the public sector includes about 13,000 institutions, including administrative institutions, public institutions, local corporations and industrial complexes, schools, and government-funded research institutions. ROK has been implementing CSAP since 2016 to resolve the security concerns of administrative or public institutions following the introduction of cloud services. According to the Act on the Development of Cloud Computing and Protection of its Users, a company that intends to provide cloud services to such public fields must meet the CSAP requirements for information protection standards.77 According to the CSAP, security requirements for cloud computing services for public institutions are very diverse. For example, based on the standards for data protection of cloud computing services of the ROK Ministry of Science and ICT, business operators must use encryption technology that has passed the deliberation of the Cryptographic Verification Committee to be recognized as fulfilling the conditions for technical protection measures. In other words, those who wish to conduct a cloud service provision business in the public sector in ROK must provide the source code (or algorithm constituting the source code) of their cloud service to the Cryptographic Verification Committee. In addition, they are required to receive confirmation from the committee that the technology they use does not pose a safety risk. In sum, given these requirements, if the KSDPA or CPTPP enters into force in the ROK, the CSAP is highly likely to be evaluated as a system inconsistent with the ROK’s digital agreement. The current CSAP, which requires that the source code of cloud service technology be disclosed for uniform “safety” reviews, may violate the “obligation to prohibit source code disclosure” of the digital agreement. In addition, the fact that the current law does not allow an exception to CSAP’s application may constitute a discriminatory trade barrier for foreign companies. In fact, among major foreign cloud service companies, such as Microsoft, Google, Amazon Web Services, Oracle, IBM, and SAP, there were no CSAP certified companies as of 2021.78 The Foreign Trade Barriers Report, which the US Trade Representative submits to the US Congress every year, points out that certain security regulations in ROK for cloud computing are an obstacle to the

Act on the Development of Cloud Computing and Protection of Its Users. Ibid at para. 2 of Art. 23 (specifically, it is based on Attached Table 4 of the standards for data protection of cloud computing services announced by the Ministry of Science and ICT). 78 Eon-ju Ryu, “Public Cloud Essential Gateway CSAP Acquired Number of Companies Increase to 33” (October 28, 2021) http://​it​.chosun​.com/​site/​data/​html​_dir/​2021/​10/​27/​2021102702212​.html accessed April 30, 2022. 76 77

196  Research handbook on digital trade revitalization of digital trade. In particular, “[i]t also appears unprecedented among developed countries, which, apart from national security applications, have generally embraced a ‘multi-tenant’ architecture, allowing both commercial and public sector customers to share the same computing resources, subject to robust access controls.”79 On January 11, 2022, the ROK government partially revised the Korean Act on the Development of Cloud Computing and Protection of its Users, which is the basis for CSAP (to be implemented on January 12, 2023). However, in the revised bill, the CSAP remained unchanged. Although, as previously introduced, matters related to security certification for cloud computing services in the past were stipulated in the “notice” of the Ministry of Science and ICT, the same content was transferred to the amended law. The “law” enacted by the National Assembly grants a more secure regulatory power than the “notice” enacted by the administrative department in Korea. In other words, ROK attempted to strengthen information security matters by shifting from the notice to the law. 4.4.2 Exceptions related to regulatory or judicial authorities Once ROK joins the CPTPP, ROK’s domestic law relating to regulatory or judicial authority exceptions may also be criticized for being inconsistent with the CPTPP due to the CPTPP’s lack of exceptions for regulatory or judicial authorities. The obligation not to enforce the transfer of source code generally does not prevent a national regulatory or judicial body from requiring other parties to preserve and provide the source code of the software or the algorithms expressed in that source code for specific investigations or inspections. In other words, the principle of prohibition of disclosure or transfer of source code (algorithms) allows that a regulatory or judicial authority of a party may provide the source code of software to a regulatory or judicial authority in a particular investigation, examination, enforcement action, or judicial proceeding. The USMCA, for instance, states that the [o]bligation not to request disclosure of source code does not preclude a regulatory body or judicial authority of a Party from requiring a person of the other Party to preserve and make available the source code of software, or an algorithm expressed in that source code, for a specific investigation, inspection, examination, enforcement action, or judicial proceeding, subject to safeguards against unauthorized disclosure.80

In this context, the ROK stipulates in domestic law that the regulatory body can request the transfer of a specific source code or algorithm in the process of exercising its authority in, for example, an investigation pursued by the regulatory body. For example, the Korea Communications Commission may, if it is recognized that there is an act in violation of the Telecommunications Business Act, conduct a necessary investigation—led by a public official under its jurisdiction—to determine whether there was a violation. Accordingly, the assigned public official may order the telecommunications business operator to submit necessary data or materials.81 Similarly, according to the Act on Promotion of Information and Communications Network Utilization and Information Protection, the Minister of Science, ICT and Future United States Trade Representative, 2021 National Trade Estimate Report on Foreign Trade Barriers, pp.328–9 https://​ustr​.gov/​sites/​default/​files/​files/​reports/​2021/​2021NTE​.pdf accessed April 30, 2022. 80 USMCA Art. 19.16 Paragpaph.2 and USJDTA Article 17 Paragraph 2. 81 Para. 1 and 5 of Art. 51 in the Telecommunications Business Act. 79

Digital trade agreements and digital policy space  197 Planning or the Korea Communications Commission allows information and communications service providers to submit related data or documents under certain circumstances.82 Although many trade agreements have regulatory or judicial authority exceptions, the CPTPP does not contain this exception clause. Therefore, if ROK decides to join the CPTPP, the two domestic laws described above may be regarded as inconsistent with the CPTPP’s source code provision. To avoid this potential violation, the Korean government could carve out those domestic laws in the accession process of the CPTPP. Given that the CPTPP text is already concluded by previous member states, using the side letters seems to be the most reasonable way to carve out the laws and maintain the policy space. Unlike many other carve-out devices that allow states to carve out during the negotiation phase, the side letter allows the states to carve out even after the main text is concluded. The flexible nature of the letter offers a comparative advantage, as parties can freely add any content as long as the other parties agree to it.83 4.4.3 Exceptions to the Internet mass media industry In certain business areas (internet portals, internet news service providers, and so on), there is increasing deliberation on whether operators should be obligated to disclose source codes or source code algorithms. Accordingly, it is noteworthy that a number of new laws imposing the obligation to disclose source codes to operators in the above business areas are being proposed by the ROK National Assembly. For example, the Amendment to the Newspaper Act (Proposition No: 2109919) imposes an obligation on internet news service providers to disclose the basic policy of and specific criteria for arranging articles (Article 10(2) of the Act) as well as the arrangement of articles to the User Committee Authority. The Newspaper Act also permits requests for disclosure or verification of key components of the algorithm (Article 10-5, Paragraph 1, Item 3 of the Act).84 In addition, in the amendment to the Act on Promotion of Information and Communications Network Utilization and Information Protection (Agenda No: 2108913), the Minister of Science and ICT and the Chairman of the Korea Communications Commission imposed an obligation to submit algorithm-related content (Article 64-6 of the Act).85 The amendment was promulgated to form a fair market order and secure the rights of information and communication users. However, this obligation is limited to cases that directly affect the safety and reliability of user information within the scope that does not infringe the company's trade secrets. Para. 1 of Art. 64 in the Act on Promotion of Information and Communications Network Utilisation and Information Protection (when there is or is suspected of violating the Act on Promotion of Information and Communications Network Utilization and Information Protection, etc, when the safety and reliability of user information is likely to be compromised, or when it is necessary to protect other users). 83 For more reference on the side letter, see generally, Tae Jung Park, Uses and Advantages of side letters in the Investment Chapter in Preferential Trade Agreements, 12 J. Int’l Dispute Settlement 84–103 (2021). 84 National Assembly Bill Information, the Amendment to the Newspaper Act (Proposition No: 2109919) (April 6, 2021) https://​likms​.assembly​.go​.kr/​bill/​billDetail​.do​?billId​=​PRC​_​D2Y1R0W3S1​ A9J1M8K3I1O3O9N6K8V5 accessed April 30, 2022. 85 National Assembly Bill Information, The Amendment to the Act on Promotion of Information and Communications Network Utilization and Information Protection (Agenda No: 2108913) (March 19, 2021) https://​likms​.assembly​.go​.kr/​bill/​billDetail​.do​?billId​=​PRC​_​L2U1Z0X3M1​G7N1Q7G4S4Z1F4V6J9G8 accessed April 30, 2022. 82

198  Research handbook on digital trade At their core, the two bills discussed above enhance the objectivity and fairness of the press. However, both bills require internet portals or internet news service providers, which have a huge influence on public opinion formation, to disclose source codes or algorithms related to the distribution of major news articles to regulatory agencies for verification. Thus, if the bills are finalized and implemented in ROK, it may generate inconsistency in relation to the prohibitions on disclosure or transfer of source codes (algorithms) under the CPTPP—to which ROK is expected to join—or other future digital trade agreements. It is important to note that in addition to developed nations, developing nations are also implementing domestic laws that may be inconsistent with the concluded treaty. It is difficult for the legislative body in any state to fully account for the rights and obligations in the concluded treaty and implement domestic laws that reflect the needs of the rapidly changing economy. It is even more challenging for the treaty negotiation team to predict the potential domestic measures that may be enacted by legislative bodies and preemptively carve out such laws in the treaty to avoid any disputes. In most cases, states simultaneously implement domestic law while concluding a treaty to act in accordance with the changing society and economy; thus, the laws may be inconsistent with each other. Although it may be difficult in practice, once a legislative body implements a certain domestic law, the negotiation team and the relevant ministry should consider such law and propose any relevant or necessary changes in subsequent treaty negotiations.

5. CONCLUSION Following the onset of the COVID-19 pandemic, the global economy entered a turbulent period of transition to become a more fully digitized society. As non-face-to-face use of new digital technologies is gradually increasing, commerce centered on real products is also changing. The global economy centered on trade in goods and services is expanding into cross-border transactions of digitized products, services, and data that can be supplied through the internet. Accordingly, the international community is beginning to take a more contemplative approach to digital trade norms. Specifically, taking advantage of the sluggish performance of WTO e-commerce negotiations, countries with matching interests are pursuing bilateral or multilateral digital agreements. In line with this global trend, ROK is also actively participating in international discussions to establish new digital trade norms to expand trade in digital goods and services. In addition, it is working on reorganizing domestic laws and regulations in line with the newly established digital commerce standard. At this time, it appears that some domestic laws in ROK may be in violation of obligations under the provisions of general digital agreements. ROK’s domestic laws that force the localization of certain facilities in relation to cloud services in the financial sector or public institutions are necessary to achieve legitimate public policy goals. ROK may, therefore, assert that they fall under the exceptions stipulated in various agreements. However, with respect to the CSAP, the obligation to disclose source code to regulatory or business organizations and to the internet mass media industry may be a violation of the obligation to prohibit the disclosure or transfer of source code (algorithm) stipulated in the KSDPA or the CPTPP. Accordingly, ROK may have to carve out such domestic laws in the CPTPP negotiation or subsequent negotiations in the future.

13. Towards digital special economic zones: new technology, digitalization and transformation Julien Chaisse1

1. INTRODUCTION Special Economic Zones (SEZs) are present in almost three-quarters of all countries and have been rapidly expanding in recent decades.2 SEZs can be precisely defined as the geographical regions that are included inside the national limits of a country but have their own set of regulations for doing business, which may or may not be consistent with those of the national territory. Furthermore, unlike the rest of the country, SEZs have preferential treatment from the government in terms of laws and regulations, especially in the fields of foreign commerce and customs, taxes and regulation.3 The global battle to identify a complete regulatory framework that should attract and promote investment from international investors has been brought to light by growing World Trade Organization (WTO) membership and offshore production encouraged by globalization.4 Foreign investment inflows are essential, especially for developing countries, as they create jobs, increase revenue, ameliorate export performance and create the conditions for positive spillovers to national governments.5 Developing countries are thoughtfully promoting and establishing such industrial enclaves to improve their economic growth. SEZs establish ‘jurisdiction within a jurisdiction’, where firms enjoy benefits such as better and favourable business-friendly laws as compared to the rest of the country.6 On the other hand, in developed countries, SEZs are not restricted to any specific location or export performance. Governmental territories do not always run SEZs. In addition, often private developers invest effort and time in such ‘charter cities’ to take benefit of business-friendly laws in

This chapter benefited greatly from the insights provided by my colleagues at the 2021 World Free Zone Convention (WFZC) conference ‘Special Economic Zones: Multiple Models, Emerging Challenges, and Future Directions’. The author thanks Huw Davies, Jacopo Dettoni, Xueliang Ji, Susan L. Karamanian, Graham Mather,’ and Sue Wake for their indispensable feedback and comments. The opinions expressed herewith are the author’s own. 2 UNCTAD, World Investment Report, Special Economic Zones (2019). 3 Tom W. Bell, ‘Special Economic Zones in the United States: From Colonial Charters, to Foreign-Trade Zones, toward US SEZs’ (2016) 64 Buff L Rev 959. 4 Julien Chaisse and Georgios Dimitropoulos, ‘Special Economic Zones in International Economic Law: Towards Unilateral Economic Law’ (2021) 24 J Int Econ Law 237. See also UNCTAD, World Investment Report, Special Economic Zones (2019). 5 Sarbajit Chaudhuri and Ujjaini Mukhopadhyay, Foreign Direct Investment in Developing Countries: A Theoretical Evaluation (Springer, 2014). 6 Susanne Frick, Andrés Rodríguez-Pose and Michael D. Wong, ‘Toward Economically Dynamic Special Economic Zones in Emerging Countries’ (2019) 95 Econ Geogr 1. See also Julien Chaisse and Xueliang Ji, The Pervasive Problem of Special Economic Zones for International Economic Law: Tax, Investment, and Trade Issues (2020) 19(4) World Trade Review 567–88. 1

199

200  Research handbook on digital trade respect of the land acquisition, labour, business operations and trade. Moreover, enterprises face increasingly liberalized circumstances in terms of sector-specific market access and post-establishment concerns. Governments aim to concentrate crucial economic activity at the national level within SEZ. However, sometimes selected domestic industries can also benefit under the protection of SEZ.7 The constellation, with a focal point on preferential tariffs for selected intermediate goods in an SEZ, can offer protection to selected domestic industries. This protection could open the way for legal claims by other trading partners. One of the essential characteristics of SEZs is their self-contained regimes. They often form part of broader strategic governmental planning to eradicate poverty, promote regional development and advance technology because of their association with investment facilitation guidelines and national industrial policy. They can also perform as a testing ground for the government to experiment with related legislation and policy reforms.8 Moreover, SEZs host various important laboratories to be used as pilots to assess the advantages and disadvantages of liberalization and market openness before introducing new liberalization legislation to the whole country.9 Globalization escalates competition among countries keen to attract foreign capital. To compete, countries create distortions within economies and permit the preceding valuable tax revenue.10 Several countries in Latin America have fermented substantial tax exemptions – from short to long tax holidays.11 Politicians aiming to boost growth rates at the regional level are attracted by SEZs. For instance, China established Shenzhen as an SEZ to compete with Hong Kong and Singapore.12 Another example would be the establishment of tax havens, as in the case of the Cayman Islands.13 Simultaneously, SEZs would also facilitate much-needed technology and knowledge transfer to domestic firms, empowering their presence in the region. It allows domestic firms to gradually integrate with global economy through operationalization of policies.14 Another advantage to domestic firms is that SEZs are generally set up in the industrially developed belt of the state, so the already established domestic firms in that industrial belt will get all trade

7 Douglas Zhihua Zeng, Building Engines for Growth and Competitiveness in China: Experience with Special Economic Zones and Industrial Clusters (World Bank Group, 2010). 8 Yehua Dennis, ‘Decentralization, Marketization and Globalization: The Triple Processes Underlying Regional Development in China’ (2001) 1 Asian Geogr 2. 9 Henry Gao, ‘Digital or Trade: The Contrasting Approaches of China and US to Digital Trade’ (2018) 21 J Int’l Econ L 297. 10 Thomas Farole and Gokhan Akinci, Fostering Innovation through Special Economic Zones Progress, Emerging Challenges, and Future Directions (World Bank Group, 2011). 11 Frederik Heitmuller and Irma Mosquera, ‘Special Economic Zones Facing the Challenges of International Taxation: BEPS Action 5, EU Code of Conduct, and the Future’ [2021] J Int Econ Law 5. See also Julien Chaisse, ‘Toward a Big Bang for the Taxation of the Digitalized Economy: a Business Retrospective, Perspective, and Prospective’ (2022) 41(3) Virginia Tax Review 66. 12 Zhaodong Jiang, ‘China’s Tax Preferences to Foreign Investment: Policy, Culture and Modern Concepts’, (1998) 18 Nw J Int’l L & Bus 569. See also Julien Chaisse, ‘Dangerous Liaisons: The Story of Special Economic Zones, International Investment Agreements, and Investor-State Dispute Settlement’ (2021) 24 J Int Econ Law 444. 13 Chris Jones and Yama Temouri, ‘The Determinants of Tax Haven FDI’ (2016) 51 JWorld Bus 241. 14 Carol Newman et al, ‘Linked-in by FDI: The Role of Firm-Level Relationships for Knowledge Transfers in Africa and Asia’ (2020) 56 J Dev Stud 3.

Towards digital special economic zones  201 benefits without moving to any other place.15 Although SEZs could turn into a disadvantage for the domestic firms because they also attract multinational enterprises (MNE) which in turn provides massive competition to the domestic firms and can become a reason for their failure.16 However, promising results from SEZs on GDP growth is premature at this stage. Furthermore, the positive correlation between the establishment of SEZs and improved export performance is difficult to comprehend, especially with respect to export diversification, and evidence for positive spillovers on the economy is indecisive. SEZs evoked mixed reactions worldwide, which shows that getting SEZs right is not easy and will take a decade more to show results, only if the right approach is undertaken.17 A clear objective and long-term commitment should be taken into consideration by the policymakers for better results. For SEZs to flourish, the essential elements are strategic location; incorporation of zonal strategy with overall developmental strategy; knowledge of the market and grasping the comparative advantage; and ensuring business-friendly regulations even if there is an external shock, such as the COVID-19 pandemic.18 These elements should be fulfilled otherwise SEZs will turn into a colossal failure, and the possible reasons for those failures could be poor governance, political influence and SEZ location. Various countries, such as India, China and Russia, have amended their legislation to fulfil the purpose of SEZs in their region.19 Location and excessive bureaucracy were considered as one of the main reasons for the failure of the Costa Rica (1993) and Dakar EPZ (2001).20 Indeed, location is the main factor in the success of SEZs; apparently, if the location is provided in the already established industrial belt then it will also ensure availability of workforce, power supply and other utilities.21 The chapter focuses on the digitalization and evolution of SEZs as a type of unilateral economic development instrument. Since the introduction of the first modern zone in Ireland, SEZs have considerably evolved, from initial ‘enclaves’ to today’s ‘Economic Zone 5.0’, founded in the fast expanding trade digitalization and its integration with urban development. The chapter provides a detailed overview of the progressive role of special economic zones, from the time when the export-processing zones were established to the latest digital trading

15 This practice can be seen in India. See Aradhna Aggrawal, ‘Promoting Agglomeration Economies And Industrial Clustering Through SEZs: Evidence From India’ (2011) 2 J Int Commer Econ Policy 201–27. 16 Rajneesh Narula and James X. Zhan, ‘Using Special Economic Zones to Facilitate Development: Policy Implementations’ (2019) 26 Transnatl Corp. 17 Aradhna Aggarwal, ‘Special Economic Zones: Revisiting the Policy Debate’ (2006) 41 EPW 44. 18 Kalpana Gopalan, ‘Special Economic Zones – Economic Panacea or Legal Nightmare? A Discussion on Elements of Public Law in SEZs’ (2007) https://​papers​.ssrn​.com/​sol3/​papers​.cfm​ ?abstract​_id​=​2393391 accessed 5 January 2021. 19 Karim Khan, Krim Khan and Saba Anwar, Special Economic Zones (SEZs) and CPEC: Background, Challenges and Strategies (The Pakistan Development Review, 2016, 32nd Conference). 20 Michael Engman, Osamu Onodera and Enrico Pinali, ‘Export Processing Zones: Past and Future Role in Trade and Development’, OECD Trade Policy Working Paper No. 53 (2007) 40, www​ .oecd​.org/​officialdocuments/​publi​cdisplaydo​cumentpdf/​?cote​=​TD/​TC/​WP​%282006​%2939/​FINAL​&​ docLanguage​=​En accessed 9 January 2021. 21 Aradhna Aggarwal, ‘Performance of Export Processing Zones: A Comparative Analysis of India, Sri Lanka and Bangladesh’, Working Paper No. 155 (Indian Council for Research on International Economic Relations (ICRIER), New Delhi, 2006). See also Julien Chaisse and Georgios Dimitropoulos, ‘Special Economic Zones in International Economic Law: Towards Unilateral Economic Law’ (2021) 24 J Int Econ Law 237.

202  Research handbook on digital trade zones.22 Further, this chapter describes the new generation of digitized SEZs and analyses the viability of such models over the evolution of these zones referencing the recent groundbreaking developments made by China in cross-border e-commerce and Malaysia in digital free trade zones.23

2.

SPECIAL ECONOMIC ZONES AND INTERNATIONAL TRADE

The connection between globalization and development due to trade is complex. There is a vocal minority of academics that dispute the idea that trade boosts economic growth. To support their view, they referred to environmental harm, inequality between and within countries, and examples of those developing countries trapped in the specialization of low-value products.24 If trade is observed as a technology towards an end, then free trade can facilitate environmental protection and sustainable development; and if we look at development from the perspective of inequality and economic growth, then development is served by trade. In the past two centuries, globalization and trade contributed the most to counterbalance inequality.25 The SEZ has benefited the world by boosting the economic growth of various places such as Shenzhen, where the average annual rate of gross domestic product (GDP) has risen from 32 per cent to 49 per cent.26 Despite the Asian economic crisis, Shenzhen exports have continued to grow by 9 per cent annually.27 However, on the other side, if we consider the examples of Russia and Kazakhstan, the SEZ has shown a negative impact: in SEZ Titanium Valley (Russia), for example, it has caused immense air pollution and chemical contamination of soil. SEZ Chemical Park in Taraz (Kazakhstan) has increased the salt concentration in drinking water and has also affected the greenery of nearby areas in a drastic manner.28 The correct approach to rising inequality would be more equitable trade and globalization, and not less trade. The state should engage with a more international perspective and disconnect with restrictions put in place by the lens of the nation-state, clarifying a comprehensive bottom line on the link between globalization, trade and development.29 This topic has been divided into two subsections: ‘SEZ, Trade and Development’ and ‘Digital Trade’. The first describes how SEZs and multilateral institutions have facilitated the 22 W. Gregory Guedel and Philip H. Viles Jr, ‘Digital Economic Zones: A Program for Comprehensive Tribal Economic Sovereignty’ (2022) 57 Tulsa L Rev 591. 23 Bin Xue Sang, ‘Pudong: Another Special Economic Zone in China – An Analysis of the Special Regulations and Policy for Shanghai's Pudong New Area’ (1993) 14 Nw J Int’l L & Bus 130. 24 Lurong Chen and Philippe De Lombaerde, ‘Testing the Relationships between Globalization, Regionalization and The Regional Hubness of the BRICs’ (2014) 36 J Policy Model 1. 25 François Bourguignon, The Globalization of Inequality (Princeton University Press, 2015). 26 Xu Xiujun & Lin Kaiwen, ‘Global Economic Governance Reform and China's Strategy in the Digital Age’ (2022) 93 China Int’l Stud 48. 27 Xiaozi Liu and others, ‘Interactions between Economic Growth and Environmental Quality in Shenzhen, China’s First Special Economic Zone’ (International Institute for Apllied Systems Analysis Interim Report, 2006), https://​core​.ac​.uk/​download/​pdf/​33899534​.pdf accessed 5 January 2021. 28 Irina Turgel and others, ‘Impact of Zones with Special Status on the Environment (Experience of Russia and Kazakhstan)’ (2019) 23 Environ Clim Technol 102–13. 29 Pol Antràsa, Alonsode Gortaria and Oleg Itskhokib, ‘Globalization, Inequality and Welfare’ (2017) 108 Joint Econ.

Towards digital special economic zones  203 growth of the developing economies through SEZ-friendly policies.30 It lays out the increasing regional trade agreements, their advantages and how rising unilateral acts have coexisted with multilateralism. The second discusses the recently developed trade platform that has been embraced by business owners and the growth of an online business platform as a key method of digital trade. Its advantages include increasing the market area of business, and its disadvantages the cross-border IP violations associated with it.31 2.1

SEZ, Trade and Development

Trade is only a single factor in the broader strategy of development. Domestic and multilateral institutions also play a significant role. Multilateral institutions connect various national institutions of different levels of development and especially help SEZs of developing countries.32 Moreover, governing bodies of SEZs are crucial for the success of SEZs. Regionalism is another essential aspect in discussion of the concept of trade, development and unilateralism. Regional and preferential trade agreements have increased in numbers in the past decade.33 The regional trade agreement (RTA) has the ability to attract FDI because of its advantages such as tax-free inputs, a tariff-free export market and much more.34 This will definitely help when a small country enters into an RTA with a large country, and the same positive effect was seen when Mexico signed the North American Free Trade Agreement (NAFTA).35 These agreements are supplementary to multilateralism and do not diminish it. It is a given that the law permits unilaterally constituted SEZs to be changed within their own borders.36 However, unilateral strictures should not necessarily be antagonistic to multilateralism.37 Through past experiences, SEZs of developing countries can be moulded into unilateral institutions of multilateralism.38 According to the World Bank simulations regarding trade and developing countries, there could be a possible gain of 1.7 per cent in unweighted average in real income in developing

30 Francois Bost, ‘Special Economic Zones: Methodological Issues and Definition’ (2019) 26 Transnatl Corp. 31 Kristi L. Bergemann, ‘A Digital Free Trade Zone and Necessarily-Regulated Self-Governance for Electronic Commerce: The World Trade Organization, International Law, and Classical Liberalism in Cyberspace’ (2002) 20 J Marshall J Computer & Info L 595. 32 Arpita Mukherjee and Tanu Goyal, ‘SEZs in the Multilateral Trading System’ in Arpita Mukherjee et al., Special Economic Zones in India: Status, Issues and Potential (Springer, 2016, ebook). 33 Manjio Chi, ‘Regulation of Special Economic Zones through Regional Trade Agreements: Confronting the Synergy Issue’ (2021) J Int Econ Law 1. 34 Dilip K Das,‘Regional Trading Agreements: The Contemporary Scenario' (2001) 2 J World Invest 347. 35 Michael Engman, Osamu Onodera and Enrico Pinali, ‘Export Processing Zones: Past and Future Role in Trade and Development’, OECD Trade Policy Papers No. 53 (OECD Publishing, 2007). 36 Julien Chaisse, ‘Dangerous Liaisons: The Story of Special Economic Zones, International Investment Agreements, and Investor-State Dispute Settlement’ (2021) 24 J Int Econ Law 444. 37 Jiaxiang Hu, ‘From SEZ to FTZ: An Evolutionary Change toward FDI in China’, in Julien Chaisse, Leila Choukroune and Sufian Jusoh (eds), Handbook of International Investment Law and Policy (Springer, 2019). 38 Naoko Koyama, SEZs in the Context of Regional Integration: Creating Synergies for Trade and Investment (World Bank Group, 2011).

204  Research handbook on digital trade countries from stronger trading agreements,39 whereas with individual agreements developing countries could suffer losses of around 0.4 per cent – a loss which could expand to 1.0 per cent for low-income countries. Therefore, trade multilateralism and globalization are favourable to development. The progressive globalization school suggested that neither trade protectionism nor complete global deregulation is sufficient for inclusive international growth and development.40 It assumed that certain viewpoints on world order and development encouraged legal reforms to fuel globalization in the 1980s and 1990s.41 The WTO,42 preferential trade agreements (PTA)43 and bilateral investment treaties (BIT)44 operated as a pedestal for the wave of globalization. Progressive globalization proposed the drafting of new trade agreements comprising transaction taxes which would impart social compensation to developing countries without producing problems regarding social dumping for rich countries.45 The school viewed trade law as an interface between different policies of the countries. An analysis of SEZs would place them as a national tool for progressive globalization.46 Developing countries that adopted a progressive globalization agenda could benefit from the flexibility and tendency in controlling the reforms related to trade and investment. Nevertheless, due to the flexibility and practicability of SEZs, they could also become vital for states that pursue different strategies towards development through their positioning as an additional or alternative means of integrating into the world market the mechanisms of international economic law. 2.2

Digital Trade

Digital trade refers to the buying, selling and exchanging of products and services between individuals, businesses and governments, and may entail either the digital or the physical transfer of goods and services. While digital trade can thrive upon both digital deliverable and physical deliverable of the required goods or services through online marketplaces, the underlying premise is the movement of consumer and seller data and contactless interactions. To do

39 ‘Making Regionalism Complementary to Multilateralism’, Global Economic Prospects (World Bank Group, 2005). 40 Anthony Bende-Nabende, Globalisation, FDI, Regional Integration and Sustainable Development: Theory, Evidence and Policy (Routledge, 2017). 41 Julien Chaisse and Georgios Dimitropoulos, ‘Special Economic Zones in International Economic Law: Towards Unilateral Economic Law’ (2021) 24 J Int Econ Law 237. 42 An international organization that attempts to standardize trade practices worldwide. Its primary goal is to facilitate free and orderly exchanges between trading partners. 43 World Trade Organization uses these agreements to describe unilaterally offered trade advantages, such as reduced or eliminated tariffs. The North American Free Trade Agreement and the ASEAN Free Trade Area are examples of PTAs. 44 These treaties provides that the restrictions should be set on the expropriation of investments, and when that does happen, compensation should be paid out quickly and in a way that is fair and effective. 45 V.N. Balasubramaniam, FDI in Developing Countries: Impact and Determinants (OECD Global Forum on International Investment New Horizons for Foreign Direct Investment, OECD Publishing, 2002). 46 Henry Veltmeyer, Globalization and Anti-globalization: Dynamics of Change in the New World Order, New Regionalisms Series (Taylor & Francis, 2017).

Towards digital special economic zones  205 business online, most people use e-commerce services used by companies and entrepreneurs because of the low cost and little paperwork involved in implementing the system.47 The digital transformation has significantly reduced the costs of engaging in international trade on account of global value chains.48 Digitization of trade helps in increasing viability and scalability, thereby reducing physical barriers of establishing the registered office, maintaining warehouse facilities and customer interactions. It enables even small businesses to compete with established businesses with equal access to digital payments, enabling fruitful collaborations and avoiding fixed investments. With increased and continuous access to affordable internet, duty and tariff-free zones and marketplaces globally that facilitate such digital trade provide options to resort to state-of-the-art technological developments through blockchain, artificial intelligence or the internet of things. While there are inherent advantages associated with digital trade of goods and services, certain pitfalls arising out of it include restricted access for brick-and-mortar stores, increasing data localization requirements, cross-border IP violations, forced technology transfer, web filtering, economic espionage, absence of full duty drawback, taxation of digital transactions without establishing permanent establishments, and so on.49 Cross-border data flow, compatibility and related topics dominate the regulatory and legal landscape of digital commerce. The digital trade has also caused huge threats to consumers’ personal information. A 2016 consumer survey response showed that even then, almost 64 per cent of Americans had experienced data breach issues, and various institutions have been unsuccessful in protecting their consumers’ data.50 Data localization is one of the problems plaguing electronic commerce that needs fixing. It means storing data which is physically present within the boundaries of that country.51 The main aim of data localization is to prevent data from leaving the country.52 It generally costs millions of dollars to set up a data centre: in Brazil around 60.9 million dollars; in the US around 40 million dollars.53 Another drawback is restrictions on the flow of data across various economies; while less extreme than data localization, lots of rules and procedures are necessary to perform the function of information exchange. Restrictions can vary from place to place: China’s Firewall, for instance, restricts certain websites such as Twitter, Facebook

47 International Chamber of Commerce ‘E-Commerce’, https://​iccwbo​.org/​global​-issues​-trends/​ trade​-investment/​digital​-trade/​ accessed 9 January 2021. 48 OECD, ‘The impact of digitalisation on trade’, www​.oecd​.org/​trade/​topics/​digital​-trade/​accessed 9 January 2021. 49 Usman Ahmed and Grant Aldonas, Addressing Barriers to Digital Trade: Strengthening The Global Trade And Investment System For Sustainable Development (E15 Expert Group on the Digital Economy International Centre for Trade and Sustainable Development and World Economic Forum, 2015). See also Julien Chaisse and Xueliang Ji, ‘The Pervasive Problem of Special Economic Zones for International Economic Law: Tax, Investment, and Trade Issues’ (2020) 19(4) World Trade Review 567–88 and Julien Chaisse (2022) Toward a Big Bang for the Taxation of the Digitalized Economy: a Business Retrospective, Perspective, and Prospective 41(3) Virginia Tax Review 66. 50 Stacy Ann Elvy, ‘Paying For Privacy and The Personal Data Economy’ (2017) 117 Colum L Rev 1369–1459. 51 Sam Mulopulos, ‘Digital Trade Zones: Answering Impediments To International Trade In Information’ (2018) 21 Chapman L Rev. 52 Anupam Chander and Uyen P. Le, ‘Data Nationalism’ (2015) 64 Emory LJ. 53 Anupam Chander and Uyen P. Le, ‘Breaking The Web: Data Localization vs The Global Internet’, University of California, Working Paper No. 378 (2014).

206  Research handbook on digital trade and a few others.54 IP infringement is another serious issue that needs to be solved, as there are instances where websites in different countries sell stolen content or pirated content.55 The OECD study states that the trade of such pirated or fake goods was worth 461 billion dollars, equivalent to 2.5 per cent of global trade.56 As far as the global trade law on digital trade is concerned, the WTO’s General Agreements on Trade in Services (GATS), being the first multilateral instrument on liberalization of international trade in services, aim at eliminating trade barriers between different countries.57 The non-discrimination principles are adhered to based upon the level of commitment agreed upon by the countries, and data localization requirements can create situations where foreign suppliers are treated less favourably than domestic service suppliers.58 Whether the foreign product is handled ‘less generously’ is at issue under Article III of the General Agreement on Tariffs and Trade (GATT). Article III of the GATT applies to both legal and informal forms of bias. When a measure expressly conditions the award of a benefit on the use of local content, as is the situation in local content instances, the discrimination is always de jure because of the law’s bias against imported products based on their place of origin. This need will therefore be met almost mechanically in situations involving local content. The Appellate Body has expanded the focus of this factor beyond a purely formalistic evaluation to include a consideration of whether or not the product at question will suffer economically as a result of the varied treatment, therefore altering the competitive landscape. When it comes to situations involving local content, the legal distinction in treatment will undoubtedly alter competitive circumstances to the prejudice of imported items.59 The global trade and privacy law on cross-border transfers suffers from global inconsistency, with developing countries and least developing countries having no restrictions on the movement of data and developed countries having an extremely regulated data localization environment. This inconsistency between national legal systems is vividly shown by the recent Schrems II ruling by the Court of Justice of European Union (CJEU). The CJEU invalidated the traditionally utilized EU–US Privacy Shield for legitimizing transfers of personal data to the US and consequently now recourse can be had only to the European Commission’s (EC) Standard Contractual Clauses, subject to numerous caveats based upon restrictive clauses under the EU General Data Protection Regulation.60 This ruling is definitely going to have

54 Office of the US Trade Representative, ‘National Trade Estimate Report on Foreign Trade Barriers’ (2017) https://​ustr​.gov/​sites/​default/​files/​files/​reports/​2017/​NTE/​2017​%20NTE​.pdf accessed 5 January 2021. 55 Rachel G. Fefer, Shayerah Ilias Akhtar and Wayne M. Morrison, Digital Trade And U.S. Trade Policy (Cong. Res. Serv., 2017) 13. 56 Office of the Intell. Prop. Enforcement Coordinator, U.S. Joint Strategic Plan On Intellectual Property Enforcement Fy 2017–2019 (2016) 20, www​.whitehouse​.gov/​sites/​whitehouse​.gov/​files/​omb/​ IPEC/​20​16jointstr​ategicplan​.pdf accessed 12 October 2022. 57 WTO, ‘What Role for International Cooperation on Services Trade Policy?’ World Trade Report 2019: The Future of Services Trade (2019) 158, www​.wto​.org/​english/​res​_e/​booksp​_e/​06​_wtr19​_5​_e​ .pdf accessed 9 January 2021. 58 Holger Hestermeyer and Laura Nielsen, ‘The Legality of Local Content Measures under WTO Law’ (2014) 48 J World Trade 553. 59 Lothar Ehring, ‘De Facto Discrimination in World Trade Law: National and Most-Favoured-Nation Treatment – or Equal Treatment?’ (2002) 36 J. World Trade 921–77. 60 Anupam Chander, ‘Is Data Localization a Solution for SchremsII?’ (2020) 23 J Int’l Econ L 3.

Towards digital special economic zones  207 implications for all the American companies, and also for any other third country which is currently planning to have adequacy with the EU under the GDPR.61

3.

THE FOUNDATIONS: THE FIRST FOUR GENERATIONS OF ZONES

SEZs from the past have evolved in various forms, such as Free Trade Zones (FTZ),62 Export Processing Zones (EPZ), Digital Trade Zones (DTZ) and a few more. The initial FTZs were generally located near seaports, and the trade was also limited to few sectors of business.63 Gibraltar (1704) and Singapore (1819) are famous examples of Free Trade Zones near seaports, both becoming successful because of their strategic geographical location and capacity to be influential manufacturing hubs.64 The main aims of these zones have been employment generation, attracting foreign direct investment (FDI) and increasing foreign exchange earnings; to achieve all these goals, special zone policies are framed which discuss tax avoidance, infrastructure provisions, export and import restrictions and so on.65 Industrial parks are also considered as a significant step to attract FDI and strengthen the economy of the country.66 These parks provide all essential services to industries in a single place ranging from technology transfer to technical infrastructure.67 Availability of buyers, suppliers and producers at the same place and exceptional modern administrative services are the prominent reasons which make them different from other industrial locations. Industrial parks have the ability to develop themselves as an important asset to national economic development.68 Industrial parks saw their inception in the United States in 1900 in Chicago, and they gained success in their early period due to a well-established rail service,69 whereas in China,

61 Ian Gauci and Terence Cassar, ‘Schrems 2 Ruling: A Brief Commentary on Its Effects’ (Mondaq, 2 Oct 2020), www​.mondaq​.com/​data​-protection/​990356/​schrems​-2​-ruling​-a​-brief​-commentary​-on​ -its​-effects​#:​~:​text​=​The​%20Court​%20of​%20Justice​%20of​,Facebook​%20and​%20similar​%20US​ %20companies accessed 12 October 2022. 62 Sam Mulopulos, ‘Digital Trade Zones: Answering Impediments to International Trade in Information’ (2018) 21 Chap L Rev 443. 63 Xianming Chen, ‘Change and Continuity in Special Economic Zones: A Reassessment and Lessons From China’ (2019) 26 Transn’l Corp. 64 Julien Chaisse and Xueliang Ji, ‘The Pervasive Problem of Special Economic Zones for International Economic Law: Tax, Investment, and Trade Issues’ (2020) 19(4) World Trade Review, 567–88. See also FIAS, Special Economic Zones Performance, Lessons Learned and Implications for Zone Development (World Bank 2008). 65 Andrew Cheesman, ‘Special Economic Zones and Development: Geography and Linkages in The Indian EOU Scheme’, DPU Working Paper No. 145 (UCL 2012). 66  Ratnakar Adhikari, ‘How Can Least Developed Countries Attract Foreign Direct Investment in the Context of COVID-19?’, Trade for Development News (2020) https://​trade4devnews​.enhancedif​ .org/​en/​op​-ed/​how​-can​-least​-developed​-countries​-attract​-foreign​-direct​-investment​-context​-covid​-19 accessed 9 January 2021. 67 Jarmila Vidova, ‘Industrial Parks – History, Their Present and Influence on Employment’ (2010) 10(1) Rev Econ Perspect 412–58. 68 Europe and Central Asia Regional Conference on Industrial Parks, 17–18 April, 2012. 69 Michael T. Peddle, ‘Planned Industrial and Commercial Developments in the United States: A Review of The History, Literature, and Empirical Evidence Regarding Industrial Parks and Research Parks’ (1993) 7 Econ Dev Q.

208  Research handbook on digital trade Special Economic Zones were launched between 1970 and 1980 after the Cultural Revolution. There were four Economic Zones in China, and after a few years, those four zones had 59.8 per cent of China’s entire FDI share.70 Target objectives vary from place to place: developing countries generally aim at narrow economic goals; some other countries focus on technology transfer and broader regional development.71 SEZs have existed for more than five decades, and therefore various lessons can be taken from them. First, it takes almost a decade to generate employment on a large scale through SEZs, and this has been visible in China and Malaysia.72 Second, FDI will only be attracted if the zone guarantees sufficient cost advantages to the investors.73 A number of other important lessons have been taught by SEZs in the past decades,74 and maybe because of this condition-specific success, many economists consider them as only the second or third best solution to competitiveness.75 The following subsections address the regulatory evolution of SEZs. First, they explain how the concept of SEZ was brought into existence by the countries of East Asia and Ireland. Second, they set out how countries moved towards Economic Zone 2.0 to explore new trade options and the benefits enjoyed by China by introducing the whole city as a Special Economic Zone and taking the first-mover advantage. Third, they explain how Eco-Industrial Parks were created as part of Economic Zones 3.0 with the goal of protecting the world’s flora and fauna and following the low-carbon emission trend. Fourth, they describe how China’s initiatives, together with those of the Shanghai FTZ and the novel policies that contributed to its success, have been widely hailed as positive. 3.1

First Generation: ‘Economic Zones 1.0’

The modern FTZ was first established in Shannon, Ireland in 1959.76 From the 1970s, SEZs were established by countries in East Asia and Latin America to increase exports in labour-intensive manufacturing sectors by alluring FDIs.77 Initially, export-processing zones (EPZs) were established by these countries. This approach was different from the traditional import-substitution policies. EPZs were located in fenced estates, having strict customs controls, and many products manufactured in these zones were exported. EPZs focused on

Douglas Zhihua Zeng, China’s Special Economic Zones and Industrial Clusters (Lincoln Institute of Land Policy, 2012) www​.jstor​.com/​stable/​resrep18404​.5 accessed 5 January 2021. 71 Xiangming Chen, ‘The Changing Roles of Free Economic Zones In Development: A Comparative Analysis of Capitalist and Socialist Cases in East Asia’ (1994) 29 Studies in Comparative International Development 3–25. 72 Charles C. Valauskas, ‘China Special Economic Zones in Perspective: A Contextual Discussion with Emphasis on the Shekou Industrial Zone’ (1986) 9 Hastings Int'l & Comp L Rev 149. 73 Jai S. Mah, ‘Foreign Direct Investment Inflows and Economic Development: The Case of Shenzhen Special Economic Zone in China’ (2008) 9 J World Investment & Trade 319. 74 Peter Warr and Jayant Menon, ‘Cambodia’s Special Economic Zones’ (2016) 33 J Southeast Asian Econ. 75 Hamada K, ‘An Economic Analysis of The Duty Free Zone’ (1974) 4 J Int Econ 225–41. 76 Jose Daniel Amado, ‘Free Industrial Zones: Law and Industrial Development in the New International Division of Labor’ (1989) 11 U Pa J Int’l Bus L 85. 77 Thomas Farole, ‘Special Economic Zones: What Have We Learned?’, Economic Premise (The World Bank, 2011) 1 http://​documents1​.worldbank​.org/​curated/​en/​275691468204537118/​pdf/​64430​ 0NEWS0Eco0​00PUBLIC00​BOX361537B​.pdf accessed 9 January 2021. 70

Towards digital special economic zones  209 implementing a unilateral economic development strategy of implementing national laws instead of international law for encouraging trade and foreign investment.78 The SEZ works as a testing ground for states to experiment and develop new policies.79 It is not wrong to regard it as a ‘pilot’ version of IEL, respecting the sovereignty of a state, with flexibility. The flexible nature of the model was destined to evolve as multiple progenies of the order grew to be multilateral in nature. For instance, labour rights became more intersectional between national and international forms of regulation.80 This model has been demonstrated as successful after a decade of development in countries such as the Republic of Korea, Taiwan, China, Vietnam, Bangladesh, Mauritius, the Dominican Republic, El Salvador, and others. The EPZs could be referred to as ‘Economic Zone 1.0’.81 3.2

The Second Generation: ‘Economic Zones 2.0’

Economic Zones 1.0 were appealing but had limitations. This first-generation version depended heavily upon fiscal incentives as there was a tendency for zones to become enclaves without a prior connection with the national economy. Therefore, countries moved forward to adapt Economic Zones 2.0, which focused on greater size, more connection with the national economy, multifunctionality and less reliance on incentives.82 Furthermore, China took it a step further by announcing a whole city or province as an SEZ for testing the market-oriented economic reforms.83 The rapid development of China and other Asian tigers was helped along by this strategy. The rapid proliferation of SEZs, part of a new wave of industrial policies, led to an increase in competition for foreign capital.84 3.3

The Third Generation: ‘Economic Zones 3.0’

Economic Zones 3.0 focused on concerns raised on global climate change and environmental sustainability. They took into account both the previous approaches to evolve integrated solutions that address both low-carbon trends in the world and trade and investment policies with domestic institutional frameworks. These zones are also called ‘Eco-Industrial Parks’ (EIPs). An EIP can broadly be defined as ‘a dedicated area for industrial use at a suitable site that ensures sustainability through the integration of social, economic, and environmental quality

78 Malini Tantri, ‘The Dynamics of SEZs over EPZs: A Case Study of Santacruz SEZ’ (2011) 2 J Int Econ 1. 79 Douglas Zeng, ‘The Past, Present, and Future of Special Economic Zones and Their Impact’ (2021) Journal of Int Econ Law 5. 80 Lorenzo Cotula and Liliane Mouan, ‘Labour Rights in Special Economic Zones: Between Unilateralism and Transational Law Diffusion’ (2021) J Int Econ Law 3. 81 Julien Chaisse and Jiaxiang Hu, International Economic Law and the Challenges of the Free Zones (Kluwer, 2019). 82 Jieming Zhu, ‘Changing Land Policy and Its Impact on Local Growth: The Experience of the Shenzhen Special Economic Zone’ (1994) 31 Urban Stud 1611–23. 83 Gianmatteo Sabatino, ‘Chinese Special Economic Zones and International Economic Law: Diversification, Expansion, Containment and Circulation of a Cryptic Model’ (2020) TDM 2. 84 Jin Wang, ‘The Economic Impact of Special Economic Zones: Evidence from Chinese Municipalities’ (2013) 101 J Dev Econ.

210  Research handbook on digital trade aspects into its siting, planning, management and operations’.85 It overcame the challenges by adopting an inclusive and sustainable industrial development approach, within the scope of the Sustainable Development Goals (SDGs).86 Moreover, it responded to the worldwide demand for a green supply chain and reduction in resource constraints by enhancing the resource management and conservation strategy, and at the same time securing both national and international climate change commitments. EIPs are also used as a mode to promote sustainable development in developing countries, as countries such as India, China, Singapore, and the like, which are going through rapid industrialization, can adopt this formula to support environmental objectives.87 Some states are gaining success in EIP models, but some are struggling to achieve the Sustainable Development Goals: for example, construction of highways and infrastructure projects in Rio De Janeiro has led to huge environmental issues.88 In principle, EIP testifies that states can adopt an approach that balances SEZ incentives with the sustainability behaviour of companies. This is predominantly based on two concomitant factors: ‘targeting incentives towards sustainability sectors and investors and making incentives condition on investor behaviour.’89 3.4

The Fourth Generation: ‘Economic Zones 4.0’

In Economic Zone 4.0, SEZs are modifying to offer a higher value-added service sector and increased business-friendly regulations. One of the examples is the Shanghai Pilot Free Trade Zone (FTZ), established in 2013 as a platform to experiment with new policies to create more open trade and service sectors by alleviating restrictions on national and international companies.90 More than 50,000 companies had been attracted to Shanghai Pilot FTZ by 2018 and China had approved 18 FTZs by 2019.91 Furthermore, China introduced one of the significant reforms by announcing a ‘negative list’ of the investment sector, which expanded over time.92 Fundamentally, FTZs were used as grounds to upgrade the business investment environment by simplifying filing and registration procedures, encouraging the inception of services industries and institutional reform for management of foreign investment, and testing capital

Sergey Sosnovskikh, ‘Industrial Clusters in Russia: The Development of Special Economic Zones and Industrial Parks’ (2017) 3 RuJE 2. 86 Jacqueline Li, ‘Building Green Supply Chains in Eco-Industrial Parks towards A Green Economy: Barriers and Strategies’ (2015) 162 J Environ Manage 1. 87 Anthony Chiu, ‘On the Industrial Ecology Potential in Asian Developing Countries’ (2004) 12 J Clean Prod. 88 Lilian Bechara Elabras Veiga and Alessandra Magrini, ‘Eco Industrial Park Development in Rio De Janeiro, Brazil: A Tool For Sustainable Development’ (2009) 17 Journal of Cleaner Production. 89 Mathhew Stephenson et al., ‘More and Better Investment Now! How Unlocking Sustainable and Digital Investment Flows Can Help Achieve SDGs’ (2021) 4 J Int Bus Policy 156. 90 Guangwen Meng and Douglas Zhihua Zeng, ‘Structural Transformation through Free Trade Zones: The Case of Shanghai’ (2019) 26 Transnl Corp 95. 91 Daqing Yao and John Whalley, ‘The China (Shanghai) Pilot Free Trade Zone: Background, Developments and Preliminary Assessment of Initial Impacts’ (2016) 39 World Econ 2–15. 92 China Briefing, ‘Shanghai Releases “Negative List” for Foreign Investment in Shanghai Free Trade Zone China Briefing (2013) www​.china​-briefing​.com/​news/​shanghai​-releases​-negative​-list​-for​ -foreign​-investment​-in​-shanghai​-free​-trade​-zone accessed 9 January 2021. 85

Towards digital special economic zones  211 account convertibility and financial sector liberalization.93 The Shanghai pilot FTZ was adapted on the basis of international experience and introduced more than 60 innovative measures covering different grounds, including maritime affairs, quarantine and better custom clearance services.94 At the same time, FTZ developed three ways to facilitate trade: first, a regulatory classification for bonded, non-bonded and offshore goods: second, standardized and normalized procedures; third, regulating various departments and setting up a single-window system to provide a systematic service for enterprises.95

4.

ECONOMIC ZONES 5.0: DIGITAL SEZ AND THE ADVENT OF DIGITAL TECHNOLOGIES

The unfolding of digital technologies and the fourth industrial revolution laid the foundation of ‘Economic Zone 5.0’ as countries built more futuristic zones. An important change involves the use of digitized services such as registration, licensing and one-stop-shop services.96 To make customs clearance faster and more accurate, the automated, digitized services allowed parties to submit standardized information at a single entry point that manages all regulatory requirements related to imports, exports and transit.97 The emergence of new technologies will help zones become more efficient, greener and less labour-intensive. For instance, in February 2020 South Korea’s Ministry of Trade, Industry and Energy introduced the first batch of locations for the establishment of smart industrial zones, to capitalize on opportunities brought by digital technologies for utilizing high-end information and communication technologies to upgrade the productivity of companies.98 The first two smart industrial zones are located in Changwon and Banwol-Sihwa.99 Highlighting China’s Cross-Border E-Commerce Pilot Zones and Malaysia’s Digital Free Zones, the next section describes the current models of digital SEZs. Both models are investigated to determine whether or not digital free trade is feasible, and further models are recommended for adoption to grow domestic and regional business.

93 Gladie Lui, ‘Shanghai Pilot Free Trade Zone: Shaping of China’s Future Foreign Investment Environment’ (2014) 40 Int Tax J 31. See also Julien Chaisse, ‘Dangerous Liaisons: The Story of Special Economic Zones, International Investment Agreements, and Investor-State Dispute Settlement’ (2021) 24 J Int Econ Law 444. 94 Gonzalo Villalta Puig and Sabrina Leung Tsam Tai, ‘China (Shanghai) Pilot Free Trade Zone Investor-State Dispute Settlement: An Uncertain Experiment' (2017) 18 J World Investment & Trade 673. 95 Guangwen Meng and Douglas Zhihua Zeng, ‘Structural Transformation through Free Trade Zones: The Case of Shanghai’ (2019) 26 Transn’l Corp 2. 96 EV Kryukova and DS Smrinova, ‘Smart Technologies and Logistics as Drivers of Development of a Special Economic Zone’, in E.G. Popkova and B.S. Sergi (eds), Smart Technologies for Society, State and Economy (Springer, 2020). 97 Ekaterina Kryukova and others, ‘Functioning of Special Economic Zones in the Digital Economy Era’, Proceedings of International Conference on Economics, Management and Technologies (Atlantis Press, 2020). 98 Yonhap, ‘Govt. picks two locations for smart industrial zones’, (The Korea Herald, 2019), www​ .koreaherald​.com/​view​.php​?ud​=​20190220000221 accessed 9 January 2021. 99 Ibid.

212  Research handbook on digital trade The introduction of digital special economic zones has presented both an opportunity and a dilemma. Among the obstacles that lie ahead are the need for new skill sets as well as new technological service providers. The opportunity, however, is that now countries can target new digital firms and this will definitely sort out the problem of location because a foreign company can sell their product in another country without transferring their business physically to that country.100 Digital Free Trade Zones will also connect small and medium enterprises with a wide range of consumers, and now a small business will also be able to manage cross-border trade,101 and can be trained via Alibaba’s Netpreneur Training Course.102 This course will help them expand their business from the local to the international market.103 4.1

Cross-border E-commerce (CBEC) Pilot Zones (Cities)

Cross-border E-commerce (CBEC) Pilot Zones (Cities) were introduced by China in 2015.104 By May 2020, China had established 105 CBEC pilot zones, covering the whole country except Tibet. Most of these zones are entire cities and are located in coastal regions such as Beijing (1), Shanghai (1), and the provinces of Guangdong (13), Zhejiang (10), Jiangsu (10), Shandong (7) and Fujian (6). These zones have boosted China’s imports and exports and assisted in promoting new industrial chains. Furthermore, they stimulate entrepreneurship and aid domestic SMEs to access the international market.105 In six years, CBEC contributed to a total foreign trade increase from 2.2 per cent to 11.3 per cent.106 Cross-border E-commerce Retail Import (CERI) can be conducted in two ways. The first possible way is Cross-border Direct Purchase and Imports, whereby a consumer from China can directly place an online order with a foreign supplier who delivers the product from abroad directly to the residence of a consumer in China (‘Direct Import’). Direct Shipping Mode comes under customs supervision code 9610.107 The second possible way to conduct CERI is Online Purchase of Bonded Products. In this, a foreign distributor, supplying goods into China, stores his product in an approved CERI special customs supervision zone or the B-type bonded logistics hub (collectively ‘special customs supervision zones’), then sells the product to Chinese consumers (‘Bonded Import’). The Bonded Import lower the cost of transportation and reduce the possibility of damage to products during transportation. These products are kept under the supervision of customs authority and cannot be removed or transferred without their permission. Once a consumer UNCTAD, World Investment Report, Special Economic Zones (2019). Sam Mulopulos, ‘Digital Trade Zones: Answering Impediments to International Trade in Information’ (2018) 21 Chap L Rev 443. 102 Alibaba Global Initiatives, https://​activity​.alibaba​.com/​supplier/​alibaba​-netpreneur​-training​.html accessed 9 January 2021. 103 Eugene Mark, ‘Malaysia–China Economic Relations Post-GE 14: Rheotric vs Reality’ RSIS Commentaries (Nanyang Technological University, Singapore 2019). 104 Francisco Jose Leandro and Yichao Li, ‘Tranformative SEZ: China’s Changing the Paradigm Along the “Belt and Road” and in the “Greater Bay Area”’ (2020) TDM 21. 105 Qiuyan Fan, ‘An Exploratory Study of Cross Border E-commerce (CBEC) in China: Opportunities and Challenges for Small to Medium Size Enterprises (SMEs)’ (2019) 9 IJEEI 1. 106 Yanbin Tu and Joe Shangguan, ‘Cross-Border E-Commerce: A New Driver of Global Trade’, in J. Agarwal and T. Wu (eds), Emerging Issues in Global Marketing (Springer, 2018). 107 Jeanne Huang, ‘The Latest Generation of SEZs: Consumer Oriented Unilateralism in China’s E-Commerce Trade’ (2021) 24 J Int Econ Law 300. 100 101

Towards digital special economic zones  213 places an order, special supervision zones promptly conduct the custom clearance before the delivery. As a result, consumers can expect to receive the product within 2–3 days of placing an order. Consumers who order their product using Direct Import usually wait for a much longer period as the product is directly delivered from abroad. There are two custom supervision codes, 1210 or 1239, to conduct Bonded Import, depending upon the location of special supervision zones.108 1210 Bonded Import should be conducted in the Cross-border E-commerce Pilot Cities (‘Pilot Cities’), and 1239 Bonded Import in other special supervision zones not located in the Pilot Cities. The reason for this is that 1239 Bonded Import needs to be reviewed and verified by customs for customs clearance forms; this is not required under 1210 Bonded Import. Furthermore, products falling under 1239 Bonded Import must acquire a first-time import licence, registration or record-filing, but these requirements do not apply to 1210 Bonded Import. Technology plays a significant role in achieving the purpose of CERI to allow consumers to shop from foreign markets virtually and directly buy products. Moreover, consumers should authenticate their identity information with customs transparently, as the tax reduction is connected to the identity of each and every individual consumer.109 The New Deal of 2016 was created to offer an online system for cross-border e-commerce customs service, and it digitized the clearing service, in order to meet the problems posed by digital technology. The portal requires the submission to customs of data on consumers’ identities, products, transactions, payments and logistics, by foreign suppliers, e-commerce platform companies, payment companies and logistics companies.110 CERI import and export inspection and quarantine information management system was established in 2017, making it essential for foreign suppliers or e-commerce platforms to convey transaction data to the General Administration of Quality Supervision, Inspection and Quarantine. The transfer of information has been increased since 1 January 2019; foreign suppliers or e-commerce platforms must transfer the details of payment-related data to customs, including the order number, product names, transaction amount, currency, information of the seller, product display website address, payment transaction serial number, verification institution and transaction complete time. If, in any case, foreign suppliers or e-commerce platforms are not able to submit data electronically to customs, then logistics companies should be responsible for providing all the mandatory data and accept the legal liability for importation.111 China has cleverly based CERI on its unilateral trade liberalization policy, which means that there is no treaty obligation for China to carry on this liberalization.112 So, the right question is whether China will continue to further liberalize CERI. The answer is affirmative, based on

108 Zoey Zhang, ‘An Introduction to China’s Cross-Border E-Commerce Pilot Zones and Pilot Cities’, China Briefing (2020), www​.china​-briefing​.com/​news/​cross​-border​-e​-commerce​-china​-introduction​ -cbec​-pilot​-zones​-pilot​-cities/​ accessed 9 January 2021. 109 Zhiyong Liu and Zipei Li, ‘A Blockchain-Based Framework of Cross-Border E-Commerce Supply Chain’ (2020) 52 Int J Inf Manag. 110 Miriam Ruiz, ‘Chinese Customs Regulations on Cross-Border E-Commerce: A Growth Opportunity for Foreign Enterprises and Chinese Commercial Platforms’ (2018) 6 China Studies Review 133–56. 111 Shuzhong Ma, Yuxi Chai and Hongsheng Zhang, ‘Rise of Cross‐border E‐commerce Exports in China’ (2018) 26 China World Econ 63–87. 112 Li Quan, E-commerce, Free Trade Zone and the Linkage Effect to Trade Upgrading (Peking University, 2020).

214  Research handbook on digital trade four reasons. First, China is determined to encourage CERI, as is evident from the fact that China doubled the number of pilot cities in 2020. Second, Chinese e-commerce platforms have strongly supported CERI because it helps them to compete with other sellers. Third, CERI is import-focused and will flourish in the China–US trade war.113 Finally, CERI is focused on promoting balanced development in international trade by expanding imports. CERI is a successful model for doing business because it satisfies the demand of Chinese customers to buy things made in other countries that are of excellent quality. CERI will not be replaceable by import substitution until it is clear to consumers that Chinese standards for product quality and food safety have increased, and ‘made-in-China’ products are of similar quality to the products of developed countries.114 4.2

Malaysia and the World’s First Digital Free Trade Zone (DFTZ)

Malaysia became the first country to establish a DFTZ in 2017. The world’s first DFTZ is designed to provide a friendly business environment to ease the entry of SMEs into the e-commerce market. Nearly 2,072 SMEs had joined the DFTZ by 2018. The DFTZ requires improved digital infrastructure to assist the private digital platforms. Some have referred to it as a ‘borderless zone’, where AI, 3D and the IoT are implanted in products and services and such products and services are merged with cities and eco-developmental concepts.115 This could also be referred to as a ‘future city’. China has also seen a rapid proliferation of such zones. Framing of policies is not only sufficient to develop a successful DFTZ; other requirements, such as using Kuala Lumpur International Airport (KLIA) for logistics and air cargo, have also supported Malaysia’s vision of a Digital Free Trade Zone, which will definitely reduce export and import time.116 In the IMD World Digital Competitiveness 2017 rankings, Malaysia was ranked 4 out of 63 countries for high-tech exports; this has helped, as it has attracted a good amount of FDI from the EU (RM 106 billion) and China (RM 63 billion).117 4.3

Other Digital SEZs in the Making

India’s efforts show that they are willing to create the next digital SEZ; the two main changes brought by the current NDA government in foreign trade policy show that the main focus of the Indian government is to attract foreign trade to work with Indian enterprises,118 by which

113 Caixia Liu and Jinhwa Hong, ‘The Effects of Consumer Cosmopolitanism and Consumer Ethnocentrism on Cross-Border E-Commerce in China’ (2020) 16 J Int Trade & Commerce 3. 114 Marta Majcherczyk and Bai Shuqiang, ‘Digital Silk Road –­ The Role of Cross-Border E-Commerce in Facilitating Trade’ (2019) 9 Journal of WTO & China 106. 115 Ismail Sissani and Midoun Bengana, ‘Digital and Free Trade Zones Impact on Malaysia’s Economy and Its Prospects’ (2018) 8 IJEFI 4. 116 ‘Economic Management and Prospects’, Economic Outlook (2019), www​.treasury​.gov​.my/​pdf/​ economy/​er/​1718/​chapter1​.pdf accessed 12 October 2022. 117 Zamri Mohamed and others, ‘Malaysia Industrial Master Plans (IMPs) and the Focus on the Nation Technology and Innovation Development’ (2018) 4 Perdana Center UTM. 118 Arpita Mukherjee and Angana Parashar Sarma, ‘Special Economic Zones (SEZs) and the WTO: The Case of India’ (2020) TDM 21.

Towards digital special economic zones  215 they are serving the CERI motive. This move can be termed as a foundation stone of the Indian government’s efforts to create the next digital SEZ.119 The US–Mexico–Canada Agreement (USMCA) contains important provisions for digital trade ranging from IP protection to anti-spam laws, and its main aim is free flow of digital trade across borders. USMCA can be considered as a promising agreement for countries who are willing to set up digital free trade zones.120 In the same vein, in Honduras Próspera and its subsidiaries have made major investments in Honduras based on the ZEDE legal framework and Honduran legislation, international treaties and legal stability agreements signed by the country. According to a business statement, Honduras has been a stable legal environment for 50 years, and the company wants to spend millions further.121 Próspera ZEDE investors have been implementing a business model of offering high-integrity, high-efficiency rule-of-law governance that meets or surpasses US anticorruption criteria in order to attract new foreign investment to Honduras at a scale never previously witnessed. These recent developments in North and South America and in Asia can only generate more experiments in the rest of the world as countries move away from old-generation SEZs.

5. CONCLUSION The relation between SEZs and international trade shows that SEZs have that capability to boost the economy of a particular country by growing their trade and business. Businesses are now leading the march into the digital sphere, with e-business rapidly becoming the dominant means of doing business and indeed posing many challenges, such as cross-border IP violations and data breach issues. The solution could be, as seen above, the model of USMCA, which mentions everything ranging from IP violations issues to anti-spam issues; therefore, for cross-border trade such types of agreements should be made mandatory between countries who are willing to do so, as it will help to reduce such crimes. The SEZ evolved in the twentieth century and has since seen further major changes – from the very first zones, limited to only a few trade options, to Malaysia, where small and medium enterprises are made eligible for cross-border trade. Economic Zone 1.0 saw the initiative of the modern industrial zone established in Ireland, its expansion into Asian and Latin American countries and its transformation into export-processing zones. Economic Zone 2.0 was focused on widening trade possibilities and multifunctionality and the benefits enjoyed by China by declaring whole cities SEZs. Economic Zone 3.0 focused on the new challenge, which was environmental damage, and the launch of EIPs, focusing on low-carbon emission trends. The Sustainable Development Goals (SDGs) were formed, and an effort was made to serve the

Biswajit Dhar, ‘India’s New Foreign Trade Policy: Pluses and Minuses’ (2018) 50 EPW. Robert K. Knake, Weaponizing Digital Trade: Creating a Digital Trade Zone to Promote Online Freedom and Cybersecurity (Council on Foreign Relations, 2020). 121 Peter Mihalick, ‘Honduras’ Fight for Prosperity’, Washington Times, 9 August 2022 www​ .washingtontimes​.com/​news/​2022/​aug/​9/​honduras​-fight​-for​-prosperity/​. See also Wayne Winegarden, ‘Incentivizing Honduran Prosperity and Reducing Migration to the U.S. through ZEDEs’, Forbes, 17 August 2022 www​.forbes​.com/​sites/​waynewinegarden/​2022/​08/​17/​incentivizing​-honduran​-prosperity​ -and​-reducing​-migration​-to​-the​-us​-through​-zedes/​?sh​=​75a587935279. 119 120

216  Research handbook on digital trade motive of the green supply chain. Economic Zone 4.0 highlights the FTZ concept, which aims to lighten the restrictions to facilitate open trade and services. The digital SEZ is emerging as the new normal for businesses and its benefits include greater efficiency, no compulsion for a physical presence, and the like. It will give small and medium enterprises the chance to engage in cross-border trade. The first DFTZ was established in Malaysia and thousands of SMEs have joined the digital platform, making it a grand success. In fact, there are other digital SEZs in the making (India, United States, Honduras, and so on). A challenge for the community is data breaches – a serious issue which needs to be resolved because a single company holds data from millions of consumers, and a leak can harm those consumers in the worst possible manner. Also, given the amount of trade in fake goods and pirated contents, there is a need to criminalize this by agreements and provide efficient dispute resolution methods so that justice can be served in the best possible manner. In sum, SEZs are ‘high-risk, high-reward’ programmes, and they have seen different phases and faced changes raised by global industrial and market landscapes as well as technologies. They were launched with an export-focused approach but have now evolved to accept the green and digital transformations. SEZs have been subject to different modalities, yet the core of success is still the same. Despite their new digital evolution, SEZs face the same old challenges. It is not guaranteed that SEZs would lead to economic development and structural transformation. The purpose of SEZs is to overcome certain market failures and government coordination failures and they should not be used in pursuit of political agendas. There is a need for sound strategic planning, strong business demand and a friendly business environment to be successful. SEZ should encourage business-friendly operations and to do so, a reform-oriented mindset is necessary. Moreover, a strategic location with good connectivity, active private sector participation and strong implementation capacity is also required for an SEZ to be successful. Furthermore, SEZs should secure sustainability and competitiveness by adhering to international standards. Adaptation is the main key to success as SEZs are complex operations, and there is no ‘one-size-fits-all’ approach, even for digital SEZ. SEZ should consider adaptations from any successful experiences, and a wise approach is necessary in any context.

PART IV REGIONAL APPROACHES TO DIGITAL TRADE: AFRICA, THE MIDDLE EAST AND THE AMERICAS

14. Streamlining the law on digital trade: prospects for the African Continental Free Trade Area (AfCFTA) negotiations Cheryl Dine

1. INTRODUCTION The regulation of digital trade is a relatively new concept, especially on the African continent. Much like the Big Bang, it is not exactly easy to direct oneself back to the precise moment that the digital age commenced. This challenging feat is perhaps best left to historians. Nevertheless, it is known that the digital age has indeed started and the world is certainly in it, whether it is prepared for it or not. From the creation of the ARPAnet, to Cerf’s Transmission Control Protocol, to the invention of the World Wide Web, it is unquestionable that information and communication technology, through which much of digital trade occurs, has developed at a rapid pace.1 As the internet and other digital technologies continue to grow, they provide more opportunities for private parties to trade among themselves while also calling for continuous improvements to mercantile law. In the legal sphere, the key considerations that often arise are typically questions of clarity, certainty, predictability, recognition and enforcement. Therefore, the establishment of rules that ensure the recognition of imperative documents and the enforceability of rights is generally considered paramount in digital trade. Moreover, it is equally important that the world also makes sense of the safety and security risks that exist in the context of digital trade and how recourse can be sought in the event of a breach of contract or other matters of dispute. In special consideration of trade law, the primary technical objective would be to minimise as many obstacles to trade as possible in keeping with general governance to ensure a level playing field. This is an objective that is shared between the World Trade Organization (WTO) and the United Nations Commission on International Trade Law (UNCITRAL). Without going into the details of the work of the WTO and the UNCITRAL, it is however commonly known that international trade law is struggling to keep up with the swift development of emerging technologies and digital trade.2 More alarmingly, African countries run the risk of being seriously left behind if they do not prioritise the development of suitable laws on digital trade and find ways to build the necessary infrastructure so as to ensure adequate engagement within the intra-African trading bloc. As outlined in the WTO Chairs Programme (WCP) (2021), most African countries do not have strong infrastructure, and as such, require robust regulatory systems in order to

1 The Invention of the Internet, 28 October 2019, History, www​.history​.com/​topics/​inventions/​ invention​-of​-the​-internet​#section​_5 accessed 8 November 2022. 2 Wentong Zheng, ‘The Digital Challenge to International Trade Law’ (2020), 52 NYU J. Int’l L. & Pol. 539, 542.

218

Prospects for the African Continental Free Trade Area (AfCFTA) negotiations  219 advance digital trade in Africa, for which the African Continental Free Trade Area (AfCFTA) plays a vital role.3 The AfCFTA presently has at least 40 state parties, among which are 14 English-speaking countries, 13 French-speaking countries, 7 Arabic-speaking countries, 3 Portuguese-speaking countries, 2 Swahili-speaking countries and 1 Spanish-speaking country.4 These are the African Union (AU) recognised languages and do not truly reflect the 2,000 different languages spoken across the African continent, making a single or manageable group of common languages a significant challenge, given that not all communities within the continent may necessarily speak and understand the AU languages with sufficient proficiency. In May 2021, the Council of Ministers established the Committee on Digital Trade to coordinate and facilitate the negotiations of the protocol on digital trade under the AfCFTA.5 However, there remain various digital and non-digital barriers to trade that could hinder the benefits of implementing the various strategies stemming from the AfCFTA.6 It is widely known that there are at least three principles that are integral to the regulation of e-commerce, namely non-discrimination, functional equivalence and technological neutrality.7 However, for a protocol on digital trade, many more factors need to be taken into account. This article will thus aim to uncover certain manifestations of legal principles that emerge under certain international texts and free trade agreements (FTAs). It will subsequently discuss certain trends in FTAs that may be indicating some commonality in approaches as regards to the law on digital trade, before considering certain obstacles that must be taken into account in the AfCFTA negotiations, and finally positing certain considerations for the anticipated AfCFTA Protocol on Digital Trade.

2.

PRINCIPLES OF DIGITAL TRADE

Prior to delving deeper into the subject of digital trade and what it means for the AfCFTA, it is crucial to understand the difference between digital trade and electronic commerce (or ‘e-commerce’). While there is no single definition of ‘digital trade’, there is a common understanding of what it encompasses. The Organisation for Economic Co-operation and Development (OECD) defines ‘digital trade’ as ‘all trade that is digitally ordered and/or digitally delivered’,8 and similarly the United Nations Conference on Trade and Development (UNCTAD) refers to ‘digital trade’ as ‘trade in goods and services that are digitally ordered and/or digitally delivered’.9 Neither, thus, distinguish between trade in physical goods and M. Smeets (ed.), Adapting to the Digital Trade Era: Challenges and Opportunities (World Trade Organization 2021), 21. 4 Status of Ratification, AfCFTA, au​-afcfta​.org/​state​-parties/​accessed 5 November 2022. 5 AU Decision on the African Continental Free Trade Area (AfCFTA) dated 10 February 2020 (Assembly/AU/Dec.751(XXXIII)). 6 Alberto Lemma, Max Mendez-Parra and Laura Naliaka, ‘The AfCFTA: Unlocking the Potential of the Digital Economy in Africa’ (2022) ODI Report, 45. 7 Lillyana Daza Jaller, Simon Gaillard and Martín Molinuevo, The Regulation of Digital Trade: Key Policies and International Trends (2020) World Bank, 5. 8 Nadim Ahmad, Handbook on Measuring Digital Trade, Organisation for Economic Co-operation and Development (OECD), World Trade Organization (WTO) and International Monetary Fund (IMF), 2020, www​.oecd​.org/​sdd/​its/​Handbook​-on​-Measuring​-Digital​-Trade​.htm accessed 4 November 2022. 9 UNCTAD, Digital Trade: Opportunities and Actions for Developing Countries, UNCTAD Policy Brief No. 92 (UNCTAD/PRESS/PB/2021/10), dated 7 January 2022, 1. unctad​.org/​system/​files/​official​ -document/​presspb2021d10​_en​.pdf. 3

220  Research handbook on digital trade services versus trade in digital goods and/or digital services provided through artificial intelligence (AI), for instance. However, they do specify that trade must be occurring through digital means. UNCITRAL, despite having a mandate on e-commerce, does not provide a definition for it.10 The WTO, however, defines ‘e-commerce’ as the production, distribution, marketing, sale or delivery of goods and services by electronic means.11 It thus appears that in some instances, e-commerce and digital trade can be used interchangeably. In 1998 the WTO Ministerial Conference issued a declaration on e-commerce with a view to establish a comprehensive work programme to examine all trade-related issues relating to global e-commerce.12 The same declaration also stated that WTO Members ought to continue their practice of not imposing customs duties on electronic transmissions.13 It was decided by the 12th Ministerial Conference (MC12) in June 2022 that this practice would continue at least until the next Ministerial Conference (MC13) in late 2023 or early 2024.14 As regards to customs duties on electronic transmissions, it is thus clear where the law on digital trade should stand at this point in time, which is an application of a zero-tax principle for electronic transmissions. The subject of e-commerce or digital trade (as the world has come to understand it thus far) has therefore been around for at least two decades and remains largely unresolved, particularly in developing countries. While those who are members of the WTO can seek some guidance from the law on trade in services,15 which WTO jurisprudence has confirmed applies to digital technologies like electronic payment systems16 and the internet,17 those African countries which are not members of the WTO are left to decide whether they will become WTO-compliant without necessarily being part of it or whether they will press on with their own interests, perhaps providing more innovative contributions. A joint work by the Overseas Development Institute (ODI) and the United Nations Economic Commission for Africa (UNECA) revealed that only 32 per cent of South–South preferential trading areas include provisions on digital trade, and many countries do not have adequate laws on digital trade still.18 Given that all 54 African states are part of the 10 UNCITRAL does not define ‘e-commerce’, although a definition is provided for ‘electronic data exchange’ (or ‘EDI’) under Art. 2 of the UNCITRAL Model Law on Electronic Commerce (1996). See also the UNCITRAL Guide to Enactment of the UNCITRAL Model Law on Electronic Commerce (1996) and the UNCITRAL ‘Report of the Working Group on International Payments on the work of its Twenty-Fourth Session’ (Twenty-Fifth Commission Session, New York, 4–22 May 1992) A/CN.9/360, paras 28–29. 11 E-commerce, WTO, www​.wto​.org/​english/​thewto​_e/​whatis​_e/​tif​_e/​bey4​_e​.htm accessed 5 November 2022. 12 WTO, Ministerial Declaration on Global Electronic Commerce, adopted 20 May 1998, WT/ MIN(98)/DEC/2. 13 Ibid. 14 WTO, 12th Ministerial Conference (MC12) Decision on the Work Programme on Electronic Commerce, adopted 17 June 2022, WT/MIN(22)/32. 15 General Agreement on Trade in Services (GATS), Art. VI. 16 Panel Report, China – Certain Measures Affecting Electronic Payment Services (2012), WT/ DS413/R. 17 Appellate Body Report, China – Measures Affecting Trading Rights and Distribution Services for Certain Publications and Audiovisual Entertainment Products (2010), WT/DS363/AB/R. 18 Alberto Lemma, Max Mendez-Parra and Laura Naliaka, ‘The AfCFTA: Unlocking the Potential of the Digital Economy in Africa’ (2022) ODI Report, 34.

Prospects for the African Continental Free Trade Area (AfCFTA) negotiations  221 South–South Cooperation (or ‘G-77’), this ultimately reveals that African countries are not as engaged in dialogues on digital trade and could signal a lack of enthusiasm for cooperation and legal development in that area. For this reason, it may be worth exploring what principles of digital trade are presently known and considering certain others that may be energing under trade or trade-related agreements, all of which may be yet to be streamlined and understood coherently. As the WTO does not currently have a legal instrument on digital trade specifically, the closest one could get to the idea of international law on digital trade would be to consider some of the harmonised rules developed by UNCITRAL thus far. However, it is important to note that most of UNCITRAL’s texts are not necessarily binding and they are limited to commercial transactions insofar as they are understood in contract law. For example, it only covers areas such as legal recognition of electronic communications, the treatment of e-signatures, and the formation and validity of contracts made through electronic means, among a few others that are more closely linked to trade facilitation or ‘paperless trade’. It does not cover issues of treatment of tariffs under digital trade, technical and non-technical barriers to trade and other trade-related matters as they are understood under WTO jurisprudence. The UNCITRAL Model Law on Electronic Commerce (1996) (‘MLEC’) was the first legislative text to provide for the fundamental principles that are recognised under e-commerce in present times.19 However, it falls short in addressing novel conditions in digital trade. While it can provide a good basis on which Africa can begin its conversation on laws on digital trade in terms of electronic communications, validity and enforceability of contracts through digital means, there are at least ten other technical areas that the MLEC does not provide for and other principles that are finding themselves in FTA chapters on digital trade. Subsequent to the 1998 MLEC, UNCITRAL later developed a specific Model Law on Electronic Signatures, 2001 (‘MLES’) in the context of commercial activities – the aim of which was to enable and facilitate the use of electronic signatures by establishing criteria of technical reliability for the equivalence between electronic and hand-written signatures.20 Although these two UNCITRAL texts provide ideas as regards the recognised principles in e-commerce, it is important to note that they are model laws and do not have ‘real’ legal effect unless incorporated in respective countries’ national laws, whereby they would be enforceable insofar as they are applied within those respective jurisdictions. Nevertheless, under the auspices of the UN, there exists a treaty for international contracts and electronic communications used to that effect. The United Nations Convention on the Use of Electronic Communications in International Contracts, 2005 applies to electronic communications in relation to contracts between parties whose places of business are set in different states.21 This UN Convention aims to facilitate international trade by ensuring validity and enforceability of electronically exchanged communications concerning transnational contrac-

UNCITRAL Model Law on Electronic Commerce (1996) with additional article 5 bis as adopted in 1998 uncitral​.un​.org/​en/​texts/​ecommerce/​modellaw/​electronic​_commerce accessed 5 November 2022. 20 UNCITRAL Model Law on Electronic Signatures (2001) uncitral​.un​.org/​en/​texts/​ecommerce/​ modellaw/​electronic​_signatures accessed 5 November 2022. 21 United Nations Convention on the Use of Electronic Communications in International Contracts, Chapter 1, Art. 1. 19

222  Research handbook on digital trade tual relations. However, as noted prior, it limits the dialogue on digital trade law to contractual communications through digital means. 2.1

UNCITRAL Principles

On this note, it is worth noting the fundamental legal developments that have been made by these UNCITRAL texts in terms of establishing certain principles to which the world has become more accustomed, namely, the principle of non-discrimination, the principle of functional equivalence and the principle of technological neutrality. Considering these principles briefly in turn, the principle of non-discrimination is one that aims to ensure legal validity and enforceability of documents in commercial transactions made through electronic means. It establishes, essentially, that contracts made in digital formats are just as binding as the more traditional paper-based transactions. In support of this, the principle of functional equivalence aims to ensure clarity and certainty with respect to electronic communications. It entails specific requirements that electronic communications need to meet in order to fulfil the same purposes and functions as the traditional paper-based system, for example in terms of signatures and authentication methods. Finally, the principle of technological neutrality urges the adoption of legal provisions that do not differentiate between the types of technology used for electronic transmissions. In this way, many emerging technologies can still fall within the meaning of ‘electronic means’. These principles are becoming, or perhaps arguably have become, integral to the law on digital trade because of their appearance in UNCITRAL texts concerning e-commerce and their growing application the world over. However, as will be discussed further below, an emergence of other principles can also be observed through FTAs with digital trade provisions or digital economy agreements (or ‘DEA’), including but not limited to principles of consent, autonomy, transparency, integrity and new ways of looking at rights in trade law. 2.2

The Principle of Consent

Among the already established principles of non-discrimination, functional equivalence and technological neutrality, a principle of consent also seems to emerge in digital trade, particularly with regard to unsolicited commercial electronic messages. For example, under the recent FTA between the United Kingdom (UK), Norway, Iceland and Liechtenstein, state parties are bound to adopt or maintain measures that require suppliers of such messages to allow recipients to choose whether to consent to receiving these messages or decline the receipt of such messages.22 It suggests that digital trade law should afford traders and consumers the right to consent to the receipt of unsolicited electronic messages of a commercial nature. While they do not offer an in-depth description of how such messages are determined to be unsolicited commercial electronic messages, it is evident that there is a certain legal development in that regard that is recognising a principle of consent, especially for unsolicited electronic messages.

UK–Norway, Iceland and Liechtenstein FTA, Art. 4.9; UK–Singapore DEA, Art. 8.61-N.

22

Prospects for the African Continental Free Trade Area (AfCFTA) negotiations  223 2.3

The Principle of Autonomy

The principle of autonomy or principle of choice appears mostly in relation to parties’ choice in their authentication methods, disclosure of source codes and cross-border flow of data. The provisions on source codes, for example, tend to aim particularly at ensuring that software owners are protected from being required to disclose their source code for trading in another market, while also being permitted to transfer their source code insofar as this transfer or access is on a voluntary basis. The UK–Norway, Iceland and Liechtenstein FTA, for instance, provides a similar approach.23 What can be drawn from this is an understanding that business owners ought to be primarily safeguarded from forced disclosure of source code, while also being conferred the right to disclose it if they so wish, implying a principle of autonomy for private parties in deciding when and how to disclose their source codes. 2.4

The Principle of Transparency

Transparency is a principle that emerges in most areas of law in different respects. Under the WTO, transparency is very much closely linked to notifications of measures that may or may not present an obstacle to trade. It is therefore not surprising that it would also appear in digital trade provisions. It becomes especially present in terms of provisions on cross-border data flows,24 open internet access25 and open government data.26 The principle of transparency in these contexts is related to accessibility in that it essentially advocates for the free flow of data and open access to technological devices, the internet and information that ought to be made available to the public. 2.5

The Principle of Integrity

Equally and to some degree interconnectedly to the principle of transparency, a principle of integrity can also be identified in terms of the treatment of data and assurance against safety and security risks. This notion particularly comes through for provisions on online consumer protection,27 protection of personal data and privacy28 and cybersecurity.29 The Australia–Singapore DEA, for example, intercorrelates the principle of transparency and integrity in its provision on online consumer protection. It indicates that both parties recognise ‘the importance of adopting and maintaining transparent and effective measures to protect consumers from misleading and deceptive commercial activities, unfair contract terms and unconscionable conduct when they engage in electronic commerce’,30 before agreeing more strongly to ‘adopt or maintain consumer protection laws to proscribe misleading and deceptive commercial activities that cause harm or potential harm to consumers engaged in online com-

25 26 27 28 29 30 23 24

UK–Norway, Iceland and Liechtenstein FTA, Art. 4.10. UK–Norway, Iceland and Liechtenstein FTA, Art. 4.11. Ibid, Art. 4.13. Ibid, Art. 4.14. UK–Norway, Iceland and Liechtenstein FTA, Art. 4.8. Ibid, Art. 4.12. Ibid, Art. 4.15. Australia–Singapore DEA, Art. 15.1.

224  Research handbook on digital trade mercial activities’.31 This suggests that in some FTAs, countries are implying that their laws on digital trade must have integrity, especially when it comes to consumer rights and private parties’ personal data.

3.

CURRENT TRENDS IN FREE TRADE AGREEMENTS (FTAS)

According to the UNCTAD Global Cyberlaw Tracker, 79 per cent of 194 countries have adopted laws on electronic transactions and 72 per cent have implemented cybercrime laws.32 Information is however lacking as regards to non-discrimination in digital trade, technology and internet access, online licences, cross-border data flows, standardisation and removing technical barriers to digital trade, artificial intelligence (AI) regulation, and access to justice through digital means, all of which are increasingly appearing in digital trade chapters under the most recent FTAs. In a World Bank paper, Daza Jaller and colleagues noted that their study of e-commerce legislation across the world suggests that having an ‘e-commerce law’ does not necessarily equate to a sound regime on digital trade.33 They continue that many of the laws pertaining to the protection of privacy date back to the 1980s and thus feature inadequate provisions for the challenges of data privacy in present times.34 The same is especially true in Africa. Seychelles, for instance, has a Commercial Code dating back to 1976, which is hardly capable of adequately governing present digital trade conditions, including those concerning emerging technologies. Seychelles did enact an Electronic Transactions Act in 2001; however, the content of this piece of legislation is very much limited to the bare minimum standards provided in the UNCITRAL MLEC and thus does not, as Daza Jaller and colleagues discuss, indicate an efficient legal framework for digital trade in its entirety. The provisions that are included in recent FTAs seem to indicate four general themes. For one, it suggests that there ought to be laws that protect traders and consumers’ rights in order to inspire confidence in digital trade. Another apparent theme is the harmonisation of electronic transaction laws so as to provide legal recognition and enforceability to digitally based forms of exchange. A theme that is more explored in developed countries’ regulation are data protection laws in order to ensure that the data of buyers and sellers are protected and ethically managed. Last but not least, there is also a theme around cybercrime or cybersecurity laws so as to protect e-commerce and digital trade from corrupt or criminal activities.35 Several of the FTAs that have provisions on digital trade now tend to include elements of UNCITRAL texts with some modifications to build on the standards appearing in UNCITRAL texts, while also taking into consideration the respective policy interests of the parties to those FTAs. One such FTA is the UK–Norway, Iceland and Liechtenstein FTA, which applies

Ibid, Art. 15.3. Lillyana Daza Jaller, Simon Gaillard and Martín Molinuevo, ‘The Regulation of Digital Trade: Key Policies and International Trends’ (2020) World Bank, 4. 33 Lillyana Daza Jaller, Simon Gaillard and Martín Molinuevo, ‘The Regulation of Digital Trade: Key Policies and International Trends’ (2020) World Bank, 4. 34 Ibid. 35 Alberto Lemma, Max Mendez-Parra and Laura Naliaka, ‘The AfCFTA: Unlocking the Potential of the Digital Economy in Africa’ (2022) ODI Report, 20. 31 32

Prospects for the African Continental Free Trade Area (AfCFTA) negotiations  225 to measures affecting trade enabled by electronic means.36 It does not, however, apply to audio-visual services, gambling services and some aspects of digital government procurement.37 The regulation of unsolicited commercial electronic messages and the protection of source codes are areas that have been added beyond the standards apparent in UNCITRAL texts. Among the more recent agreements relating to digital trade, the UK–Singapore DEA is perhaps one of the most comprehensive to date.38 The UK–Singapore DEA, which was presented to Parliament in March 2022,39 forms a vital part of the UK–Singapore FTA40 and provides for a wide range of new areas relevant to digital trade rules under Annex A on Digital Trade and the Digital Economy.41 The DEA includes six bilateral agreements between the UK and Singapore, on various other areas such as fintech (UK–Singapore Fintech Bridge), digital customs (DEA Digital Customs Agreement), cybersecurity (DEA Cybersecurity Agreement), digital identities (DEA Digital Identities Agreement), electronic trade documents and electronic invoicing (DEA Electronic Trade Documents and Electronic Invoicing Agreement). Under the chapter for digital trading systems, the UK–Singapore DEA includes: modern electronic contracts, digital authentication, paperless trading, digital customs and modern logistics. In America, and as researched by Lopez and colleagues in 2021, the US–Mexico–Canada FTA and the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) are among the more advanced FTAs when it comes to the inclusion of provisions on digital trade in that region.42 However, the Singapore–Chile and New Zealand DEA appears to be more expansive globally, especially as regards to the treatment of digital products. A similar trend can be found in the Australia–Singapore DEA. All of these would be interesting examples of what works and what does not work, should AfCFTA negotiators look to find the more relevant trends to them.

4.

OBSTACLES FOR DIGITAL TRADE REGULATION IN THE AFRICAN REGION

In 2017, e-commerce accounted for 12 per cent of global trade in goods.43 In the 2020s, the innovative and digital economies, as well as engagement with emerging technologies, play significant roles in determining a country’s comparative advantage in global trade.44 With UK–Norway, Iceland, Liechtenstein FTA, Chapter 14. Ibid, Art. 4.3. 38 www​.gov​.uk/​government/​publications/​uk​-singapore​-digital​-economy​-agreement​-explainer/​uk​ -singapore​-digital​-economy​-agreement​-final​-agreement​-explainer. 39 UK–Singapore DEA [CS Singapore No.1/2022] www​.gov​.uk/​government/​publications/​ uksingapore​-digital​-economy​-agreement​-cs​-singapore​-no12022. 40 UK–Singapore DEA, Art. 5. 41 Ibid. 42 Dorotea Lopez, Bradly Condon and Felipe Munoz, ‘The New Rules on Digital Trade in Latin America: Regional Trade Agreement’ in M. Smeets (ed.) Adapting to the Digital Trade Era: Challenges and Opportunities (World Trade Organization 2021), 214. 43 African Union (AU), ‘The Digital Transformation Strategy for Africa (2020–2030)’, 18 May 2020. 44 M. Smeets (ed.), Adapting to the Digital Trade Era: Challenges and Opportunities (World Trade Organization 2021), 339. 36 37

226  Research handbook on digital trade barriers to digital trade on the rise due to certain data localisation laws, lack of cooperation with global partners could seriously damage a country’s ability to optimise digital trade opportunities and the economic growth offered therefrom.45 According to UNCTAD, digital trade presents an array of opportunities for developing countries. In UNCTAD’s Policy Brief No. 92, it was particularly emphasised that the new opportunities created by digital trade have the potential to inspire more resilience.46 However, the African region, which is home to 33 least developed countries (LDCs), stands to lose substantial benefits if it does not ambitiously develop laws on digital trade and seek to set up adequate infrastructure. The AfCFTA negotiations on digital trade thus present opportune conditions to cooperate on addressing existing barriers and key digital trade areas such as digital consumer protection, the protection of intellectual property rights in the digital realm, authentication, data localisation, cross-border data flows, cybersecurity and data protection.47 At WTO level, the Africa Group has consistently expressed its concerns as regards implementation, and continues to echo this position.48 In support of this, some studies made concerning the agility of African countries to actively engage in the digital economy reveal more distressing realities. A 2022 ODI Report by Lemma and colleagues states that the percentage of individuals who use the internet in Africa is the lowest among all regions, amounting to only 30 per cent, significantly lower than the global average of 60 per cent and the Asian average of 57 per cent.49 This ultimately indicates that the pool of participants in digital trade are rather small.50 What can be drawn from this is that AfCFTA negotiators need to be especially considerate of the cooperation needed to address the lack of engagement and lack of access to digital trade. The subject of digital trade itself was not even considered a priority for the African region until February 2020, when the AU Heads of State and Government Assembly decided that e-commerce would be included in the AfCFTA negotiations under Phase III, immediately after the conclusion of Phase II of the AfCFTA negotiations, which is still ongoing.51 This position does not seem to align with the position of the Africa Group at the WTO. However, this may also signal a preference by African countries to address digital trade regionally prior to engaging more actively on global fora. Limited engagement by African countries can also be observed in consideration of the Information Technology Agreement (ITA), which makes it a requirement for countries to eliminate and bind custom duties at zero for both hardware and

The Digital Trade Revolution: How U.S. Workers and Companies Can Benefit from a Digital Trade Agreement, US Chamber of Commerce, 9 February 2022. 46 UNCTAD, Digital Trade: Opportunities and Actions for Developing Countries, UNCTAD Policy Brief No. 92 (UNCTAD/PRESS/PB/2021/10), dated 7 January 2022, 1. 47 Alberto Lemma, Max Mendez-Parra and Laura Naliaka, ‘The AfCFTA: Unlocking the Potential of the Digital Economy in Africa’ (2022) ODI Report, 6, cdn​.odi​.org/​media/​documents/​Digital​_trade​ _report​_AfCFTA​_​_​_A7wsHot​.pdf. 48 C. Igue, A. Alinsato and T. Agadjihouédé, ‘E-commerce in Africa: Issues and Challenges’ in M. Smeets (ed.), Adapting to the Digital Trade Era: Challenges and Opportunities (World Trade Organization 2021). 49 Alberto Lemma, Max Mendez-Parra and Laura Naliaka, ‘The AfCFTA: Unlocking the Potential of the Digital Economy in Africa’ (2022) ODI Report, 17. 50 Ibid. 51 AU Decision on the African Continental Free Trade Area (AfCFTA), dated 10 February 2020 (Assembly/AU/Dec.751(XXXIII)). 45

Prospects for the African Continental Free Trade Area (AfCFTA) negotiations  227 software high-technology products. Only four African countries have signed the ITA, namely Egypt, Mauritius, Morocco and the Seychelles.52 In the AU Digital Transformation Strategy for Africa (2020–30), much more emphasis was placed on digital financial services than on other digital trade topics found in FTAs the world over.53 This suggests that Africa’s strategy is perhaps to address the digital economy by sector, starting with financial services, which possibly has the more potential to grow the African economy. Nevertheless, Africa’s concerns as regard to readiness are particularly founded when it comes to legal taxonomy given that certain methodologies still remain to be clarified even at a global level.54 Furthermore, while this distinction does not seem to exist under UNCITRAL, the World Bank does distinguish between electronic signatures and digital signatures.55 It is thus worth considering whether the AfCFTA will also distinguish between e-signatures and digital signatures and whether there would be a certification authority for the African continent should AfCFTA negotiators prefer to undertake an approach that includes the enforceability of digital signatures across Africa. More importantly, the digital divide and what might be termed the ‘digital brain fog’ that persists across the African continent are among the most pressing challenges in maximising digital trade in Africa, which should be at the centre of negotiators’ objectives for the AfCFTA Protocol on Digital Trade. 4.1

The Digital Divide

The digital divide is a commonly known concept. According to a 2022 economic study by Welde, the digital divide remains a pertinent issue in the African region.56 The statistics explored in this paper presented trends in the use of digital technologies and suggests that mere access to technology remains a major problem in Sub-Saharan Africa.57 Therefore, it is also expected that there is a lack of access to technologies for businesses and consumers to utilise e-commerce platforms for trade. 4.2

The Digital Brain Fog

Aside from the digital divide, there is also a condition, as mentioned above, that may be termed the ‘digital brain fog’, which poses another great challenge in both the development of a legal framework in African countries and the actual utilisation of digital technologies for trading purposes. Digital brain fog can be understood as the lack of understanding in regard to digital

Ibid, 29. African Union (AU), ‘The Digital Transformation Strategy for Africa (2020–2030)’, 18 May 2020. 54 UNCITRAL, ‘Legal issues related to digital economy – advancing work on automated contracting and other progress’, Note by the Secretariat, A/CN.9/1116. 55 Lillyana Daza Jaller, Simon Gaillard and Martín Molinuevo, ‘The Regulation of Digital Trade: Key Policies and International Trends (2020), World Bank, 9. 56 A.A. Welde, ‘Digital Divide in Sub-Saharan Africa: A Panel Analysis of Internet Diffusion’ (2022), Conference Paper, in Proceedings Of The Eighteenth International Conference on the Ethiopian Economy, Ethiopian Economics Association (EEA), 314. 57 Ibid. 52 53

228  Research handbook on digital trade aspects. In this context, the main challenge would be policy makers’ ability to understand the technical aspects of the area that they are negotiating. It is widely known that African countries tend to argue that there is a lack of understanding of what digital trade is and what it should comprise. However, there is not much suggestion on how to proactively reach the desired level of understanding. There is thus a seemingly perpetual digital brain fog that needs to be effectively addressed. Solutions can be found by looking to theories under the branch of cognitive economy. By understanding that learned helplessness is a phenomena that plagues several policy makers, AfCFTA negotiators may then help Africa reach a level where its people can successfully partake in digital trade with adequate laws to both govern and protect them.

5.

CONSIDERATIONS FOR THE AFCFTA PROTOCOL ON DIGITAL TRADE

In a recent study, the US Chamber of Commerce found that digital trade is boosting US small business exports and creating more job opportunities – an increase that is especially prominent in digital advertising, digital payment systems and shipment procedures.58 It is thus indicative of the potential that awaits African countries if both the governments and the private sector view the prospects of the digital economy more enthusiastically. Lemma and colleagues also found, in relation to the African region specifically, that although there is not much actual engagement, the rate at which the use of online purchasing systems, for example, is growing in Africa is higher than in other regions, with such use growing by 240.4 per cent in three years.59 More notably, the leading countries in regard to digital trade seem to be coastal, namely, Nigeria, South Africa, Kenya and Ghana, suggesting access to underwater optical fibre cables is a probable key advantage. South Africa, being one of the leading countries on the African continent in this respect, has developed an advanced legal framework for consumer protection and data protection, possibly instilling more trust among its constituents than other African countries.60 Ghana, as another example, has already introduced model legislation for e-commerce, including dispute resolution mechanisms for cross-border digital trade.61 While this may not be enough for a multilateral agreement on digital trade at the WTO, expert contributions from these leading countries could be key to the development of an AfCFTA Protocol on Digital Trade that works for African countries. Additionally, it may also be worth adopting Zheng’s back-to-basics approach.62 Zheng argues that instead of undertaking far-reaching negotiations on controversial issues, the global community should target its focus on negotiating a basic framework for rules and applying widely accepted and existing WTO legal principles to digital trade. While this approach can John G. Murphy, The Case for a Digital Trade Agreement, U.S. Chamber of Commerce (Washington, 17 August 2021). 59 Alberto Lemma, Max Mendez-Parra and Laura Naliaka, ‘The AfCFTA: Unlocking the Potential of the Digital Economy in Africa’ (2022) ODI Report, 13. 60 Alberto Lemma, Max Mendez-Parra and Laura Naliaka, ‘The AfCFTA: Unlocking the Potential of the Digital Economy in Africa’ (2022) ODI Report, 36–7. 61 Ibid, 40. 62 Wentong Zheng, ‘The Digital Challenge to International Trade Law’, 52 NYU J. Int’l L. & Pol. 539 (2020). 58

Prospects for the African Continental Free Trade Area (AfCFTA) negotiations  229 certainly work in the beginning, it may not resolve issues around legal gaps indefinitely. Therefore, as regards the AfCFTA negotiations, it is perhaps more favourable to approach the development of a legal framework on digital trade progressively. The Digital Transformation Strategy for Africa aims to build on the existing initiatives and frameworks such as the AfCFTA, Policy and Regulatory Initiative for Digital Africa (PRIDA), the Programme for Infrastructure Development in Africa (PIDA), the AU Financial Institutions (AUFIs), the Smart Africa Initiative and others, in order to support the development of a Digital Single Market (DSM) for Africa.63 All of these initiatives can together support the AfCFTA negotiations on digital trade. Furthermore, and in light of the discussions in the previous sections, a protocol on digital trade for the AfCFTA can benefit from exploring possibilities for: rules on digitalisation of commercial transactions; rules on technical aspects of digitally enabled trade; rules on data governance and transparency mechanisms; and rules on the protection of rights in the digital sphere. Rules on digitalisation of commercial transactions would be a desirable starting point for AfCFTA negotiations on digital trade. It should include, inter alia, rules governing the validity, enforceability and automation of procedures in trade, including paperless systems, such as registration, verification, authentication and certification, beyond that which is already provided for under the WTO Trade Facilitation Agreement, albeit drawing largely on UNCITRAL’s current texts and ongoing work in regard to the particulars of e-commerce.64 Additionally, it is expected that the AfCFTA Protocol on Digital Trade should comprise rules on technical aspects of digitally enabled trade in goods and services. Explicitly, it should include rules on the treatment of trade in digital goods and digital services. For example, traditional trade rules as are found in the General Agreement on Tariffs and Trade (GATT) 1994 include rules on the application of customs duties on goods. Following the introduction of the ITA and the WTO’s various ministerial conferences since 1998, it is clear that the trend is to avoid imposing tariffs on electronic transmissions and digital technologies. However, discussions still need to be made on how to coherently treat trade in digital goods, especially given the challenge of tracing such transactions which can occur in the space of milliseconds on informal platforms that are not necessarily intended for digital trade, for example social media platforms such as Instagram, Facebook, LinkedIn and others. Consideration must also be given to digital advisory services, such as chatbots, which bring about questions of ethics and recourse, notwithstanding technical challenges that are posed by data localisation, such as those applied by the European Union (EU).65 Aside from these two categories, rules on data governance and transparency mechanisms are also imperative considerations for AfCFTA negotiations on a protocol for digital trade. These would include rules on access to information, access to technological devices, access to the internet, access to and choice of storage, among other areas relevant to the principle of transparency and data governance. It may also be worth considering in critical depth how or to

63 African Union (AU), ‘The Digital Transformation Strategy for Africa (2020–2030)’, 18 May 2020. 64 See UNCITRAL, ‘Legal issues related to digital economy – advancing work on automated contracting and other progress’, Note by the Secretariat, A/CN.9/1116. 65 Regulation (EU) 2016/679 on the protection of natural persons with regard to the processing of personal data and the free movement of such data (General Data Protection Regulation (GDPR)) (2016).

230  Research handbook on digital trade what extent a reasonable balance can be achieved between free flow of data and the protection of data for personal or security reasons. Finally, another consideration for the AfCFTA negotiations on a protocol for digital trade, based on FTA trends, would be to consider strongly the rules that ought to govern the protection of rights in the digital sphere. In this respect, the protection of rights should include all relevant rights to a trader or consumer, including the obvious data and privacy protection rules, consumer protection online or through other digital means and the protection of intellectual property rights (not limited to copyright). The AfCFTA negotiators may need to consider in particular how they would address e-signatures and authentication so as to ensure that the e-signatures used in an online transaction are not denied their value or legal effect because they are submitted in electronic format. This subject matter would also present an opportunity to address the treatment of unsolicited commercial messages, source code disclosures and open government data.66 Another key area of focus may be cross-border data flows.67 The world has become so much more connected, and with that connection lawmakers are called upon to set boundaries on such interconnectedness innovatively and effectively.68 In Mexico–Telecom (2004), the Panel affirmed that ‘interconnection’ of public telecommunications networks with foreign networks is carried out through agreements entered into by the interested parties.69 The AfCFTA Protocol on Digital Trade is therefore a fundamental determinant of how cross-border data flows will be treated on the African continent, bearing in mind that not all African states are members of the WTO. This view also echoes Banga and colleagues’ 2021 study which indicated that ‘AfCFTA negotiations could provide a common framework for data flow governance, building on existing approaches in SADC and ECOWAS to create a large enough marketplace where investment in data infrastructure (e.g. data centres) becomes commercially sustainable’.70 With regard to the protection of rights in the digital sphere, data and privacy protection have the potential to create barriers to digital trade. It is therefore crucial that AfCFTA negotiators are meticulous in maintaining fair protection of personal rights while also not restricting accessibility to digital trade within Africa. In support of this, Kugler (2022) makes a valid point with regard to the liberalisation of data flows, stating that any data localisation requirement could result in a violation of a number of provisions under the Services Protocol. For example, the most-favoured nation principle under Article 4 of the AfCFTA

Alberto Lemma, Max Mendez-Parra and Laura Naliaka, ‘The AfCFTA: Unlocking the Potential of the Digital Economy in Africa’ (2022) ODI Report, 29. 67 See Alexander Beyleveld and Franziska Sucker, ‘Cross-Border Data Flows in Africa: Policy Considerations for The AfCFTA Protocol on Digital Trade’, Mandela Institute Report, published October 2022; Appellate Body Report, United States – Measures Affecting the Cross-Border Supply of Gambling and Betting Services (2005), WT/DS285/AB/R. 68 See Gregory Shaffer, ‘Trade Law in a Data-Driven Economy: The Need for Modesty and Resilience’ [2021] World Trade Review. 69 Panel Report, Mexico – Measures Affecting Telecommunications Services (Mexico–Telecom), WT/DS204/R, adopted 2 April 2004, 140. 70 K. Banga, J. Macleod and M. Mendez-Parra, ‘Digital Trade Provisions in the AfCFTA: what can we learn from South–South Trade Agreements’ (2021), Supporting Economic Transformation (SET) Working Paper Series, ODI, London. 66

Prospects for the African Continental Free Trade Area (AfCFTA) negotiations  231 might be violated if a state party permits the transfer of data to another state party that it considers has adequate data protection laws, while it refuses the transfer to a country with similar data protections.71

This is therefore another valuable consideration for negotiations on digital trade. Lastly, another key consideration around the subject of digital trade is the subject of access to justice beyond the UNCITRAL Technical Notes on Online Dispute Resolution, 2017. While establishing laws favourable to the engagement of buying and selling is principal to trade effectively via digital means, it is equally important to provide for laws and mechanisms that would facilitate access to justice through digital means as well. Therefore, the formalisation of the use of digital technology for access to justice, whether it be for advisory services or dispute resolution, is just as fundamental to the preservation and enforcement of the rights and obligations of traders and consumers in relation to the digital economy.

6. CONCLUSION Streamlining the law on digital trade is evidently an extremely intricate task that would require substantial study, expertise and rigour by all nations, and not only the African region. The WTO’s zero-tax principle and the UNCITRAL texts on e-commerce provide a basic foundation for rules on electronic transactions, whereas several FTAs provide trending areas of consideration from which AfCFTA negotiators can seek inspiration for a protocol on digital trade; for example, those FTAs discussed in previous sections, as well as the US–Japan Digital Trade Agreement. This would, however, only work as a starting point. AfCFTA negotiators would also need to give considerable thought to addressing the obstacles and challenges also explored in previous sections, namely, the digital divide and digital brain fog. The AfCFTA may not immediately address all disparities in law, in infrastructure (such as broadband and fibre coverage) and in technical capacity that are essential to engaging in digital trade. However, it provides great potential towards achieving this. If the AfCFTA does indeed move forward with a protocol on digital trade, it must be able to provide clarity, certainty and predictability of law as regards to (as explored under Section 5 above): rules on digitalisation; rules on digitally enabled trade in goods and services; rules on trade in digital goods; rules on trade in digital services; rules on data governance and transparency (in particular, determining how much transparency is freedom of information and technology transfer and how much transparency infringes on privacy rights); and rules on the protection of traders’ and/or consumers’ rights in the digital sphere, including access to justice mechanisms. Following or even among these considerations, it is also worth taking into account how emerging technologies (such as AI, distributed ledger technologies, quantum technologies, immersive technologies and the Internet of Things) ought to be treated, especially as facilitators of digital trade.72 Despite the complexities of digital trade law and the multitude of concerns around it, it is clear that there are at least basic foundations on which negotiations can commence and that

71 Kholofelo Kugler, ‘The Impact of Data Localisation Laws on Trade in Africa’ (2022), Policy Brief 08, Mandela Institute, School of Law, University of The Witwatersrand, 7. 72 M. Smeets (ed.), Adapting to the Digital Trade Era: Challenges and Opportunities (World Trade Organization 2021), 339.

232  Research handbook on digital trade even with a small amount of interest, there is great potential to streamline digital trade law in a way that is sensible and hopefully effective.

15. Legal approaches to the regulation of digital trade by Middle Eastern countries Bashar Malkawi

1. INTRODUCTION Although there is no single definition of digital trade, the phrase is most commonly used to refer to “digitally-enabled transactions of trade in goods and services […] and that involve consumers, firms, and governments.”1 On the other hand, electronic commerce, or e-commerce, can be defined as “the transfer of data, products, or services by electronic means, usually the Internet.”2 Digital trade and e-commerce are often treated interchangeably, although there are key differences. Furthermore, digital trade can play a role in e-commerce by “facilitating the buying, selling, and servicing of physical goods and services.”3 The unique characteristics of digital trade and electronic commerce—the fact that trade can be facilitated through digital means and that goods and services may be digital—are extraordinary things that were not possible before the introduction of the internet, modern technology, and the digitization of society. As digital trade, e-commerce, and the digitalization of society and populations continue, innovation and new possibilities in the market continue to grow. Digital trade and electronic commerce have become increasingly important as technology continues to advance and as the digitalization of trade progresses. As a result of technological advancements, goods and services can more easily be traded through digital means, due to increased scope, speed, and scale of trade. This is to say, digital goods and services have increased in prevalence and value in modern society, and digital means have facilitated access to them, even across international boundaries.4 Furthermore, technology is changing how goods and services are provided, even blurring the lines between the two, as well as the modes of delivery. For example, an article produced by 3D printing is a design service, but also a good when it is consumed.5

See Digital Trade, OECD, www​.oecd​.org/​trade/​topics/​digital​-trade/​ (last visited Jan 28, 2022). See Emily Benson, E-commerce vs. Digital Trade: Digital World E-Commerce vs. Digital Trade | Digital World (2019), www​.bfna​.org/​digital​-world/​e​-commerce​-vs​-digital​-trade​-3tsxyi3owh/​ (last visited Jan 28, 2022). 3 Ibid. 4 See Nikolas Guggenberger, Reallocating Network Effects: An Industrial Policy for Digital Pluralism. Paper presented at Works in Progress for Intellectual Property Scholars Colloquium (WIPIP). Saint Louis University School of Law, St. Louis, Missouri, United States (2022), pp.10–12. 5 See Emily Benson, E-commerce vs. Digital Trade: Digital World E-Commerce vs. Digital Trade | Digital World (2019), www​.bfna​.org/​digital​-world/​e​-commerce​-vs​-digital​-trade​-3tsxyi3owh/​ (last visited Jan 28, 2022). 1 2

233

234  Research handbook on digital trade The digital economy has emerged as a critical driver of growth, prosperity, and dynamism for national economies.6 Exports of digitally tradable services support millions of jobs. Also, the digital trade revolution offers impressive new opportunities for small businesses. Issues that have always plagued regular trade have also arisen in digital trade, albeit with vastly different consequences. For example, some countries might have cumbersome border procedures or restrictions on newly tradable services. Furthermore, the digitalization of trade has produced new problems, such as dissonance between the regulations of different countries on digital trade. Still, the prevalence of digital trade has proven to be a powerful tool for growth in many countries. This is especially true for many Middle Eastern countries, depending on how they have chosen to embrace digital trade and e-commerce. Thus, studying digital trade and e-commerce regulations and the legal approaches of these regulations by each country can help explain the influence of digital trade in these countries and the role of digital trade moving forward.

2.

DOMESTIC ELECTRONIC COMMERCE LAWS

Middle Eastern countries have, in general, been attempting to broaden their engagement in trade with other countries and different systems, including through digital trade. Even as these countries have tried to engage in trade and digital trade on a larger scale, many of them find themselves unable to establish a modern governance framework for the digital economy.7 Overall, it is a general trend among countries in the Middle East that the regulation of their digital markets is still in its infancy, being mostly governed by general laws that were not originally intended for the digital era.8 And yet, the Middle East is one of the most promising regions for electronic commerce.9 Indeed, some of these countries—such as the United Arab Emirates—are leading the digital consumer charge, which means that they have serious potential to transform into leading digital economies.10 One of the reasons that the Middle Eastern region is so promising for digital trade and electronic commerce advancements is that it is larger than the United States in terms of population, although it is smaller than economic powerhouses such as China and India. One key difference, however, is that Arab countries hold a GDP greater than that of India and a higher per capita GDP than both Indian and China. Another reason why this region is so auspicious is that digitization of the country and internet usage are rapidly increasing. Companies in the region would be apt to take advantage of the opportunities that Arab countries have to

6 See US Chamber Staff, The Digital Trade Revolution (2022), pp.2–10, www​.uschamber​ .com/​international/​trade​-agreements/​the​-digital​-trade​-revolution​-how​-u​-s​-workers​-and​-companies​-can​ -benefit​-from​-a​-digital​-trade​-agreement (last visited Mar 3, 2022). 7 See Lillyana Daza Jaller and Martin Molinuevo, Digital Trade in MENA: Regulatory Readiness Assessment Open Knowledge Repository (2020), pp.31–2. 8 See Lillyana Daza Jaller and Martin Molinuevo, Digital Trade in MENA: Regulatory Readiness Assessment Open Knowledge Repository (2020), pp.31–2. 9 Expert Analysis, How to Make Inroads in the MENA E-Commerce Market, Law360 (2014), www​ .law360​.com/​articles/​565225/​how​-to​-make​-inroads​-in​-the​-mena​-e​-commerce​-market (last visited Jan 29, 2022). 10 See Tarek Elmasry et al., Digital Middle East: Transforming the Region into a Leading Digital Economy (2016), p.41.

Legal approaches to the regulation of digital trade  235 offer, especially since both financial and cultural reasons make e-commerce suitable for their populations. An example is Saudi Arabian women, who, in reflection of their country, may be wealthy but are traditionally not legally allowed to drive, making it difficult for them to shop. However, e-commerce may permit them to spend their money online instead, essentially eliminating that barrier.11 Electronic commerce, however, has not been able to become completely widespread over the region, mostly due to high levels of skepticism among the population, because not many people have credit cards, and because many people prefer to use cash on delivery (COD).12 A study of domestic e-commerce laws in key Middle Eastern countries, such as the United Arab Emirates, Egypt, Jordan, and Saudi Arabia, allows for a deeper understanding of the modern-day treatment of e-commerce in each of these countries more specifically. 2.1

United Arab Emirates

The United Arab Emirates leads e-commerce among states in the Gulf Cooperation Council (GCC). In 2020 alone, the market jumped by 53 percent, with a record of $3.9 billion in e-commerce sales.13 In the United Arab Emirates, the Telecommunications Regulatory Authority (TRA) is primarily responsible for the regulation of electronic commerce and transactions. Additionally, intellectual property rights and ownership play a significant role in e-commerce in this country. Apart from leading the region in the digitization of its population, the UAE has also begun to implement core digitization initiatives.14 Many retailers in the UAE are improving their traditionally physical shopping experience by offering online experiences, especially in consumer electronics and appliances, media products, computers, fashion accessories, women’s apparel, cosmetics, and perfumes, as these are the categories that produced the most revenue in 2017. In fact, Carrefour and Lulu Group, large regional retailers, introduced online shopping portals for their customers in the UAE. Another notable milestone for e-commerce in 2017 was the acquisition of Souq.com by Amazon.com, which was significant since Souq.com is the largest online merchant in the UAE. Furthermore, in 2020, the UAE released the “E-Commerce Guide (VATGEC1) providing guidance on the VAT treatment of the supply of goods and services through electronic means.”15 In addition to rapid growth in previous years, there has been additional growth in e-commerce in the United Arab Emirates in recent years, mostly due to the COVID-19 pan-

Expert Analysis, How to Make Inroads in the MENA E-Commerce Market Law360 (2014), www​ .law360​.com/​articles/​565225/​how​-to​-make​-inroads​-in​-the​-mena​-e​-commerce​-market (last visited Jan 29, 2022). 12 Expert Analysis, How to Make Inroads in the MENA E-Commerce Market Law360 (2014), www​ .law360​.com/​articles/​565225/​how​-to​-make​-inroads​-in​-the​-mena​-e​-commerce​-market (last visited Jan 29, 2022). 13 See International Trade Organization, United Arab Emirates: E-Commerce, www​.trade​.gov/​ country​-commercial​-guides/​united​-arab​-emirates​-ecommerce (last visited Jan 28, 2022). 14 See Tarek Elmasry et al., Digital Middle East: Transforming the Region into a Leading Digital Economy (2016), p.22. 15 See PricewaterhouseCoopers, United Arab Emirates: FTA Publishes an E-Commerce VAT Guide PwC, www​.pwc​.com/​m1/​en/​services/​tax/​me​-tax​-legal​-news/​2020/​united​-arab​-emirates​-fta​-publishes​-e​ -commerce​-vat​-guide​.html (last visited Jan 28, 2022). 11

236  Research handbook on digital trade demic. The dramatic increase in e-commerce in the UAE was paired with an increase in the trust that the population had for online products, as well as increased smartphone usage in this region. Another factor that has made the public more comfortable with online transactions is the introduction of e-government services, whereby public services such as utility billing, traffic services, licenses, visa issuance, and more can be accessed through online platforms. Furthermore, the Dubai Free Zone Council has announced e-commerce regulations for free zones that will allow increased foreign direct investment, encouraging e-commerce. Thus, e-commerce in the United Arab Emirates is especially promising, not only because of the young and technologically educated population, but because of initiatives being taken to encourage e-commerce. This is not to say, however, that there are no barriers to e-commerce in the UAE. The same issues—preference for cash on delivery (COD), consumer wariness, and security concerns—prevent e-commerce from becoming widespread, and must be addressed. 2.2 Egypt Out of the entire Middle Eastern and North African region, the Egyptian market has the largest population of internet users.16 In 2019, internet penetration reached 53.5 million people, and yet only 8 percent of those made regular online purchases. Most of these online purchases were in categories such as electronics, entertainment, airline tickets, and fashion. Thus, e-commerce has many barriers in Egypt, one of the most prominent being the low number of credit and debit card holders. Only around 10 million people in Egypt have credit or debit cards, meaning that about 80 percent of e-commerce relies upon cash on delivery. Egypt enacted its Electronic Signature Law (ESL) in 2004. The statute is remarkable because it is one of the few in the world that does not contain exclusions.17 The ESL created the Information Technology Industry Development Authority (ITIDA), a public corporation affiliated with the Ministry of Communications and Information. In recent years, however, credit card issue has grown to about 40 percent, which is promising. Furthermore, e-commerce has increased in Egypt in response to the COVID-19 pandemic and the need for social distancing. As a result, many businesses began operating remotely and measures were taken by the Egyptian government to ensure a smooth digital transformation. For example: Upper Egypt, 8 technology parks will be completed and operational in 2020 to help the small SMEs and entrepreneurs work on finding solutions for the digitalization to be effective. It will also include training institutions, integrated systems, and startups incubators, hardware design labs, training halls for Artificial Intelligence (AI), data science, and cybersecurity.18

Furthermore, since 2021, the Ministry of Communication and Information Technology has been providing electronic services for private businesses.

16 See International Trade Administration, Egypt: E-Commerce, www​.trade​.gov/​country​-commercial​ -guides/​egypt​-ecommerce (last visited Jan 28, 2022). 17 See Stephen E. Blythe, E-Commerce Security in the Land of the Pharaohs: Refining Egypt’s Electronic Signature Law, 21:2 Ind. Inte’l & Comp. L. Rev. (2011), pp.195–8. 18 See Jordanian Market: E-Commerce (2021), www​.objectif​-import​-export​.fr/​en/​international​ -marketplaces/​country/​jordan/​market​-ecommerce (last visited Jan 28, 2022).

Legal approaches to the regulation of digital trade  237 Overall, Egypt has great potential when it comes to e-commerce, as a great portion of the population is online already. Still, the country has numerous obstacles to overcome when it comes to making e-commerce appealing to its population. Although, as noted, this has improved due to social distancing protocols in the COVID-19 pandemic and aid offered by the government, additional measures would be helpful. 2.3 Jordan Jordan is more advanced than other countries in this region when it comes to e-commerce, for several reasons. First, as of 2017, around 88 percent of the country’s population used the internet, with high-speed services being widely available from various service providers.19 In previous years, Jordan has begun to introduce legislation relating to e-commerce, although it has not been entirely comprehensive. Jordan is also undergoing a digital transformation in terms of the technological skills of its population, with many calling for the incorporation of these skills in schools.20 In 2017 alone, e-commerce sales are estimated to have amounted to approximately $662 million. In more recent years, e-commerce has become more popular due to the ongoing COVID-19 pandemic. Due to social distancing and lockdown protocols, e-commerce was the only way that some people were able to purchase products.21 Despite this great success, Jordan faces several challenges that other countries in this region must also address. For example, e-commerce has not progressed as rapidly due to the lack of e-commerce regulations that protect consumers, changes to licenses and taxes that burden e-commerce companies, and the lack of electronic payment systems.22 Thus, although Jordan is quite well developed in e-commerce compared to other countries, it is not at its full potential due to a lack of coordination and clear strategy as well as a lack of awareness about e-commerce more broadly among the public.23 2.4

Saudi Arabia

E-commerce in Saudi Arabia is projected to increase significantly in upcoming years and is estimated to exceed $13.2 billion (SAR 50 billion) by 2025.24 This sudden explosion in 19 See International Trade Administration, Jordan: E-Commerce, www​.trade​.gov/​country​ -commercial​-guides/​jordan​-ecommerce (last visited Jan 28, 2022). 20 See Batool Ghaith, Jordan Sees “Enormous Growth” in E-Commerce, Yet More Needed to Improve Sector—Experts, The Jordan Times (2021), jordantimes​.com/​news/​local/​jordan​-sees​-​%E2​ %80​%98enormous​-growth​%E2​%80​%99​-e​-commerce​-yet​-more​-needed​-improve​-sector​-​%E2​%80​%94​ -experts (last visited Jan 28, 2022). 21 Ibid. 22 See Jordanian Market: E-Commerce (2021), www​.objectif​-import​-export​.fr/​en/​international​ -marketplaces/​country/​jordan/​market​-ecommerce (last visited Jan 28, 2022). 23 See Batool Ghaith, Jordan Sees “Enormous Growth” in E-Commerce, Yet More Needed to Improve Sector—Experts, The Jordan Times (2021), jordantimes​.com/​news/​local/​jordan​-sees​-​%E2​ %80​%98enormous​-growth​%E2​%80​%99​-e​-commerce​-yet​-more​-needed​-improve​-sector​-​%E2​%80​%94​ -experts (last visited Jan 28, 2022). 24 See Pablo Martinez et al., The Sar 50 billion E-Commerce Opportunity in Saudi Arabia (2021), www​.bcg​.com/​en​-mideast/​publications/​2021/​e​-commerce​-market​-50​-billion​-sar​-opportunity​-in​-saudi​ -arabia (last visited Jan 28, 2022).

238  Research handbook on digital trade e-commerce is, in part, due to the global COVID-19 pandemic and related lockdown and store closures, as customers shifted to online shopping for a safer overall experience. This is evidence by the category of products that experienced the fastest growth—food and drink, and home-related products. Although Saudi Arabia lags behind more mature e-commerce markets, it certainly has potential to become a major player in e-commerce, especially if e-commerce continues to thrive after the COVID-19 pandemic. Online sales in Saudi Arabia have increased by, on average, 60 percent in all categories. Major categories include media products and apparel and footwear. E-commerce has surpassed traditional retail at this point in time. There are many favorable factors in Saudi Arabia that have facilitated e-commerce growth, such as telecommunications, internet, legislation, and financial services.25 The population in Saudi Arabia has high smartphone usage rates— among some of the highest smartphone penetration rates in the world—further contributing to this influx. Additionally, the government supports e-commerce laws to establish payment protocols and customer protections. Nonetheless, e-commerce in Saudi Arabia has not reached full maturity and is still in its infancy, as evidenced by its total retail sales of 6 percent, compared to leading e-commerce markets at 18 percent. One of the challenges that Saudi Arabia faces in expanding e-commerce is the limited size of product assortment and inventory. Product delivery speed is also rather limited. Overall, the framework necessary for e-commerce to grow to its full potential has not been developed, but there are promising elements at play within the country that could help e-commerce transform even further in the future. The E-Commerce Law of 2019 encourages and increases e-commerce activities in Saudi Arabia. The new law enhances the reliability of e-commerce and protects merchants and consumers from fraudulent, deceptive, and misleading practices in e-commerce transactions.26 Article 8 of E-Commerce Law 2019 requires that service providers give consumers an invoice after the conclusion of a contract. The invoice must list the cost of each product or service; the total price, including fees, taxes or additional amounts related to delivery, if applicable; the date; and the place of delivery, as specified in the regulations.

3.

DATA PROTECTION REGULATIONS

Data protection regulations are, as their name implies, legal frameworks that set guidelines for the collection of data for people who live in certain countries or regions.27 Data plays an important role in the contemporary digital economy as informational capitalism.28 Indeed, data is a key feature of what makes the digital economy profitable. One of the major factors influencing the growth of digital trade and e-commerce is how well customers are protected. Many countries have introduced digital sovereignty laws of varying scope on account of con Ibid. See Smith Russell and Urbas Gregor, Controlling Fraud on the Internet: A CAPA Perspective, A Report for the Confederation of Asian and Pacific Accountants, Confederation of Asian and Pacific Accountants (Malaysia: Australian Institute of Criminology, Research and Public Policy Series, No. 39 (2001), p.63. 27 See Jake Frankenfield, General Data Protection Regulation (GDPR) Investopedia (2021), www​ .investopedia​.com/​terms/​g/​general​-data​-protection​-regulation​-gdpr​.asp (last visited Jan 28, 2022). 28 See Salomé Viljoen, A Relational Theory of Data Governance (2021), pp.875–6. 25 26

Legal approaches to the regulation of digital trade  239 cerns about cybersecurity, data privacy and sensitivity, and cyber capabilities, often imposing broad restrictions on cross-border data transfer or introducing local content requirements for digital-related services.29 In many Middle Eastern countries, like all countries around the world, consumers are wary of online transactions and maintaining an online presence because they are not certain of how their personal information will be used. Thus, many of these countries have taken the initiative to create regulations to dictate how businesses and companies can use consumer data. All GCC countries, with the exception of Kuwait, have enacted personal data protection laws.30 These laws introduces rights for individuals to access, rectify, correct, delete, restrict processing, request cessation of processing or transfer of data, and object to automated processing. These laws impose strict obligations in relation to how, when, and why personal data is collected, stored, and used. 3.1

United Arab Emirates

The United Arab Emirates has just recently issued its first federal data protection regulation, Federal Decree Law No. 45/2021 on the Protection of Personal Data (the Data Protection Law).31 In addition to this regulation, the UAE government also issued a law establishing the new UAE Data Office—Federal Decree Law No. 44/2021 on Establishing the UAE Data Office. This law was issued in late 2021 and came into force on January 2, 2022. The new UAE data protection regulation follows a trend of similar regulations that are being passed and proposed throughout the Middle Eastern region. The new data protection regulation passed by the UAE government applies to both controllers and processors that are located in the UAE and those located outside of the countries that process the personal data of individuals within the UAE. There are, however, several exclusions, such as government data and authorities, the processing of health, credit, and backing data, and companies and institutions located in free zones, such as the Dubai International Finance Centre (DIFC) and the Abu Dhabi Global Market (ADGM), which have their own data protection laws. The data protection law passed in the UAE closely resembles the General Data Protection Regulation (GDPR), which is an international data protection law, although there are some key differences.32 One of those differences is that the UAE law has a more limited legal basis, with consent as the primary one. Furthermore, there is no legitimate interest ground, unlike the GDPR. Another difference between the Data Protection Law and the GDPR is that there are

29 See Andrew D. Mitchell and Theodore Samlidis, Cloud Services and Government Digital Sovereignty in Australia and Beyond, International Journal of Law and Information Technology (2022), p.4. 30 See Nadim Al Jisr, Data Privacy: The GCC Landscape, Thomson Reuters (2022), support​ .thomsonreuters​.ae/​product/​westlaw​-middle​-east​-0/​updates​-alerts/​data​-privacy​-gcc​-landscape (last visited Mar 6, 2022). 31 See Brian A. Meenagh, Alexander Hendry and Lucy Tucker, UAE Publishes First Federal Data Protection Law Global Privacy & Security Compliance Law Blog (2021), www​.globalprivacyblog​.com/​ legislative​-regulatory​-developments/​uae​-publishes​-first​-federal​-data​-protection​-law (last visited Jan 28, 2022). 32 See Manu J. Sebastian, The European Union's General Data Protection Regulation: How Will It Affect Non-EU Enterprises? 31 Syracuse Journal of Science and Technology (2015), pp.216–17.

240  Research handbook on digital trade less burdensome transparency requirements and no specific privacy notice requirement. The final major difference between the two laws is that the UAE provides a more detailed record of processing requirements. In additional to domestic data affairs, the new law accounts for international data transfers. International data transfers are allowed to countries that have been approved by the UAE Data Office, as per the Federal Decree Law No. 44/2021 on Establishing the UAE Data Office; those that have a data protection agreement with the UAE; or “or where certain exceptions apply, such as where data transfer clauses are in place, the data subject has provided consent, or the transfer is necessary for a contract with a data subject.”33 3.2 Egypt Much like the United Arab Emirates and other Middle Eastern countries, Egypt has promulgated a data protection laws fairly recently. In mid-July of 2020, Egypt passed its Data Protection Law (Law No. 151 of 2020) with the aim of protecting the rights of individuals in Egypt as they relate to their personal data, as well as to ensure that business process personal data correctly. In general, the Data Protection Law concerns “the handling and management of all personal data, except that is handled by the Central Bank of Egypt and all the entities under its supervision.”34 The law went into effect in October of 2020, with the Government of Egypt allowing for a grace period for companies to make adjustments in order to comply with the law. The Data Protection Law resembles the General Data Protection Regulations from the European Union. Similar to the GDPR, the Data Protection Law requires that the processing of personal data be conducted in accordance with specific principles, such as data minimization, accuracy and security, lawfulness and purpose limitation, and storage limitation.35 There are some key distinctions, however. For example, according to Article 2 of the Data Protection Law, “personal data may not be collected, processed, or disclosed by any means except with the explicit consent of the data subject.” Furthermore, Article 6 sets consent by the data subject as one of the conditions for the processing of personal data.36 In addition to general data protection, the Data Protection Law also created licensing requirements for data processing, data control, the handling of sensitive data, electronic marketing, and cross-border transfers of data. The law provides for fines and also imposes sanctions in the case of violations, some acts even resulting in imprisonment. The law also created a new regulatory authority for personal data protection, the Centre.37 Unlike other data

33 See Brian A. Meenagh, Alexander Hendry and Lucy Tucker, UAE Publishes First Federal Data Protection Law Global Privacy & Security Compliance Law Blog (2021), www​.globalprivacyblog​.com/​ legislative​-regulatory​-developments/​uae​-publishes​-first​-federal​-data​-protection​-law (last visited Jan 28, 2022). 34 See International Trade Administration, Egypt Data Protection (2020), www​.trade​.gov/​market​ -intelligence/​egypt​-data​-protection (last visited Jan 28, 2022). 35 See Masha Ooijevaar and Dino Wilkinson, Egypt's Data Protection Law Enters into Force Clyde & Co. (2020), www​.clydeco​.com/​en/​insights/​2020/​10/​egypt​-s​-data​-protection​-law​-enters​-into​-force (last visited Jan 28, 2022). 36 See Mohamed Hashish and Farida Rezk, Egypt: Data Protection Overview Data Guidance (2021), www​.dataguidance​.com/​notes/​egypt​-data​-protection​-overview (last visited Jan 28, 2022). 37 Ibid.

Legal approaches to the regulation of digital trade  241 protection regulations in the region, however, the Data Protection Law does not allow for and generally prohibits the transfer of personal data to foreign countries—notwithstanding several exceptions—unless a license has been obtained and the level of protection is not less than that offered by the law.38 The Egyptian Data Protection Law prohibits, in Article 14, the transfer of personal data to a third country unless such country guarantees a level of protection of personal data that does not fall below that which is stipulated under the law, and subject to obtaining a relevant license or permit. Article 15 of the law gives an exception in which it is possible to transfer personal data to a country that does not offer the required level of protection, if direct consent of the data subject or his representative is obtained. Also, Article 2 of the law also provides that personal data may not be collected, processed, disclosed, or revealed by any means except with the explicit consent of the data subject or in cases permitted by law. This obligation of information and consent is one of the criteria of transparency that the European Union and many other countries require before accepting any cross-border transfer of data. Before this law was passed in Egypt, there were still some protections for personal data, albeit not nearly as extensive. For instance, the Egyptian Constitution of 2014 provided for the protections of privacy and secrecy in relation to communications; the Penal Code No. 58/1994 imposed criminal sanctions for unlawful collection of images or recordings of individuals in private places; and the Cyber Security Law No. 175 of 2018 placed a duty on service providers to maintain the privacy of the data that they store. The enactment of the Law brought a more stable and comprehensive data protection and privacy regime to Egypt.39 3.3 Jordan Although Jordan does not currently have any data protection laws, as of early 2022 a data protection law had been proposed. A draft of the Personal Data Protection Law of 2021 was proposed in order to protect the personal data of citizens. The law was deemed necessary “in light of the ease of collection, retention and processing” and in order to protect citizens’ right to privacy.40 The draft of the Personal Data Law of 2021 provides a regulatory framework for maintaining and processing personal data, as well as a legal framework balancing individuals’ rights to personal data protection with technological advances. The law would also establish a Personal Data Protection Board tasked with protecting personal data.41 Jordan does not have a specific law or centralized framework that regulates data protection, nor is there a regulatory body that oversees data protection, but there are several laws that have been made to satisfy different necessities that provide some kind of safeguard for data. One of the laws that relates to data protection is the Telecommunications Law No. 13 of 1995 and its Amendments, which regulates the telecommunications sector in Jordan and imposes confidentiality requirements against telephone calls and private communications but

Ibid. See Masha Ooijevaar and Dino Wilkinson, Egypt’s Data Protection Law Enters into Force, Clyde & Co. (2020), www​.clydeco​.com/​en/​insights/​2020/​10/​egypt​-s​-data​-protection​-law​-enters​-into​-force (last visited Jan 28, 2022). 40 See Jordan Drafts Law to Protect Citizens’ Personal Data, IAPP (2022), iapp​.org/​news/​a/​jordan​ -drafts​-law​-to​-protect​-citizens​-personal​-data/​ (last visited Jan 28, 2022). 41 Ibid. 38 39

242  Research handbook on digital trade does not regulate online platform providers. Another law that involves data protection is the Cybercrime Law No. 27 of 2015, which criminalizes unlawful access to websites or information systems. Thus, the provisions of the Cybercrime Law regulate unlawful access to data, but do not dictate how data can lawfully be collected and processed. Finally, multiple laws regulating the Jordanian banking system exist to preserve the confidentiality of client data and ensure that there are sufficient security checks to achieve this. These include the Banking Law No. 28 of 2000 and the Electronic Payment and Transfer of Money Regulation No. 111 of 2017, which make disclosures on client transactions and information subject to rigorous requirements. The Cloud Computing Guidelines of 2018 also apply to regulate the use and integration of Cloud Consumers in the financial sector, defined as parties requesting and using resources and services offered via cloud technology. Although comprehensive when taken altogether, these laws are limited to the financial sector and do not comprise a cohesive data protection framework.42 3.4

Saudi Arabia

Like many other Middle Eastern countries, Saudi Arabia has issued its first national data protection law. This law, the Personal Data Protection Law, entered into force in early 2022 with the purpose of regulating the collection, processing, and use of personal data within the country. Like many other data protection laws, this law provides businesses, organizations, and other data-processing entities with a grace period, at the end of which they should be in compliance with the new law. The supervisory body for the law, the Saudi Data & Artificial Authority, has also issued several regulations that supplement the law.43 The Personal Data Protection Law does not deal with the transfer of personal data comprehensively. Data controllers cannot transfer personal data outside Saudi Arabia unless required to comply with an international agreement to which Saudi Arabia is party, to serve Saudi interests, or for other purposes that will be set out in the Executive Regulation. There are several requirements outlined in the new law. Some of these requirements include residency, extraterritoriality, restrictions on cross-border transfers, consent, and sensitive data.44 Similar to other data protection laws in the region, the Personal Data Protection Law relies on consent, which is to say that consent must be obtained in writing in order for personal data to be legally processed, notwithstanding certain exceptions. Under the Personal Data Protection Law, the SDAIA and data subjects must be notified immediately in the case of breaches, leakages, or unauthorized uses of personal data. The law also includes criminal penalties, such as up to two years’ imprisonment and fines of up to SAR 3 million (approximately USD $800,000).45

See Khaled Saqqaf, Dana Abduljaleel and Hakam Al Shawwa, Data Protection in Jordan: An Overview of the Current and Future Framework, Al Tamimi & Company (2021), www​.tamimi​.com/​ law​-update​-articles/​data​-protection​-in​-jordan​-an​-overview​-of​-the​-current​-and​-future​-framework/​ (last visited Jan 28, 2022). 43 See Tarek Khanachet, Julie Teperow and Antonio Michaelides, Saudi Arabia Issues New Personal Data Protection Law, Inside Privacy (2021), www​.insideprivacy​.com/​privacy​-and​-data​-security (last visited Jan 28, 2022). 44 Ibid. 45 Ibid. 42

Legal approaches to the regulation of digital trade  243 Something important to note about Saudi Arabia is that the paramount body of law is the Sharīʿah, which is composed of “a collection of fundamental principles derived from a number of different sources, which include the Holy Qu’ran and the Sunnah, which are the witnessed sayings and actions of the Prophet Mohammed.”46 This means that prior to the Data Protection Law that went into effect in early 2022, Sharīʿah principles dictated data protection. In general, Sharīʿah principles protect the individual right to privacy and prohibit invasions. Furthermore, the disclosure of secrets is prohibited except when the public interest requires such disclosure. The Holy Qu’ran and the Sunnah do not specify penalties in the case that secrets are disclosed; it may be punishable by a penalty that a judge, in his discretion, deems appropriate and equitable. This penalty can be anything, such as a fine, imprisonment, or suspension of certain rights.47 On a different front, there is a lack of general privacy protections in Saudi Arabia. The Credit Information Law is the only source of privacy protections.48 The law is limited in its application to government entities, companies, and other entities maintaining credit information. Still, the law is a step in right direction. The E-Commerce Law of 2019, in Article 8, handles the privacy of receipt data, with comprehensive information to be provided by service providers. However, Article 8 omits a precise time for the provider’s obligation of contractual knowledge. Service providers’ obligation of contractual knowledge, according to this Article, should come after the conclusion of the contract. Article 8, concerning consumers’ ability to download invoice data, is an important legal and technical guarantee for consumers, especially in that the invoice data must be secure and easy to read. Under Article 18 of E-Commerce Law 2019, the penalties for service providers that breach this law or its implementing regulations include a warning, a fine not exceeding 1 million SAR, stoppage of e-commerce temporarily or permanently, and withholding the electronic shop partially or completely, temporarily or permanently. The Saudi legislator placed importance on consumers’ personal data and electronic communications. Service providers have a responsibility to take the necessary methods to protect these types of data and communications. Article 5 of the E-Commerce Law defines consumers’ personal data to include their name, ID number, address, contact details, license numbers, records, personal property, bank account numbers, and credit cards. Service providers are obligated to protect consumers’ personal data from access, disclosure, expositure, exchange, or use for illegal reasons. Service providers must not save consumers’ personal data except to execute the obligations. If such data are hacked, then the service provider must notify the Ministry of Commerce and the affected consumer(s) within three days of the date of its knowledge. In notifying the affected consumer(s) of the hacking, service providers must submit an analytical report stating the key reasons for the hacking, its effects, and the measures taken to resolve it. The notification does not exempt service providers from responsibility to the affected consumer(s). Service providers abide by what is issued by the competent authorities. According to Article 2 of the Saudi Cloud Computing Regulatory Framework, cloud computing means the use of a scalable and elastic pool of shareable physical or virtual resources See Data Protection in the Kingdom of Saudi Arabia: A Primer, pp.1–2. Ibid. 48 See Adam Eichelberger, Global Employee Privacy: A Case Study on the Minefield of Employee Privacy Rights in the EU, USA, and KSA 31: 117 Indiana Int’l Comp. Law Review (2021), p.201. 46 47

244  Research handbook on digital trade such as servers, operating systems, networks, software, applications, and storage equipment with self-service provisioning and administration on demand.49 The Saudi Cloud Computing Regulation Framework (CCRF) does not use the concept of personal data, but the concept of subscriber data. According to the framework, subscriber data refers to any content (software, text, files, audio and video, and so on), related to at least one of the following categories: – Any data related to a natural person, who is specifically identified, directly or indirectly, as a cloud computing user. – Any data related to any Cloud Subscriber’s business activity or business information or financial affairs. – Any data generated by or for a CSP concerning any subscriber activity log, billing, and so on.50 Regarding the cross-border transfer of data, the CCRF makes a clear distinction between Saudi government data and non-government data. Government data includes four different levels (top secret, secret, confidential, and public). Non-government data includes data received from government entities and other data. Article 3.3.8 of the CCRF requires registered cloud computing service providers to ensure that no Saudi government data content is transferred outside the Kingdom for whatever purpose and in whatever format, whether permanently or temporarily, unless it is expressly stated that is permitted according to the laws or Regulations of the Kingdom, other than the CCRF. The Saudi CCRF provides that registered CSPs must inform and get the approval of the CITC and cloud subscribers if cloud subscriber content is to be transferred, stored, or processed outside the Kingdom, permanently or temporarily.51 According to the Saudi Regulation, the CSPs may not provide or authorize another party to provide to any third party, including, but not limited to, any individuals, legal entities, domestic or foreign government, or public authorities, subscriber content or subscriber data,52 or process or use subscriber content or subscriber data for purposes other than those allowed under the Cloud Computing Agreement with the cloud subscriber concerned.53 Nevertheless, the Cloud Computing Regulatory Framework allows the transfer of personal data in response to a foreign jurisdiction requirement, or if it is required by Saudi law and after getting customer consent for any non-governmental data (type ‘Other Data’) for which the CSP has to acquire only cloud customer consent.54 The customer has the possibility to give prior consent via an opt-in or opt-out form, from which the customer can withdraw at any time.55 Any transfer of

49 Cloud Computing Regulation, Saudi Communication and Information Technology Commission, (Feb. 6, 2018) as updated on 3 December 2020 (Version 3). 50 Ibid Art. 2.2.15. 51 CCRF, Art. 3.3.10. 52 According to Art. 3.4.2.1 of the CCRF, CSPs may not ‘provide or authorize another party to provide any third-party (including, but not limited to, any individuals, legal entities, domestic or foreign government or public authorities) Customer content or customer Data. 53 Art. 3.4.2.2 of the CCRF. 54 CCRF, Art. 3.4.3.2. 55 CCRF, Art. 3.4.3.

Legal approaches to the regulation of digital trade  245 data based on a foreign jurisdiction request may not happen unless it is allowed by the Saudi laws.56 Regarding the right to access to one’s personal data, the CCRF provides that CSP shall guarantee to cloud customers the technical capability to access, verify, correct, or delete their subscriber data.57 The provision does not explain how this right can be guaranteed but gives the CSP the possibility to choose the appropriate way to provide these rights, especially the right to delete the customer data, known as the ‘Right to be forgotten’.

4.

DIGITAL TRADE IN THE US AND EU: FREE TRADE AGREEMENTS WITH ARAB COUNTRIES

A free trade agreement (FTA) between two or more countries where they collaborate to create certain obligations that affect trade in goods and services, and protections for investors and intellectual property rights, among other topics.58 The United States and European Union—as two major economies that regularly conduct trade and other business with numerous other countries—have countless FTAs with different countries, which can help companies to enter and compete more easily in the global marketplace, as FTAs address foreign government activities that affect businesses. As such, both the US and EU have a few FTAs with Arab countries concerning digital trade and e-commerce. It is important to note, however, that many Arab countries lack FTAs with the EU and US on the subject of digital trade and e-commerce. As a result, digital trade and e-commerce do not have favorable environments in which they can thrive and grow. Jordan is an example of a country that has a FTA with the US concerning e-commerce; however, it is actually more of an exception than an example, since FTAs dealing with digital trade with the US and the EU are scarce in the Middle Eastern region. 4.1

United Arab Emirates

Although the United Arab Emirates does not have a free trade agreement on digital trade with the European Union, nor with the United States, it does have several adjacent agreements with the US that may influence digital trade and e-commerce.59 First, it is important to note that in 1994 the UAE became a party to the General Agreement on Tariffs, and then in 1996 to the World Trade Organization. The UAE is also party to the Information Technology Agreement, a treaty within the WTO, which sought to eliminate tariffs on IT products. In 2004 the United States and the United Arab Emirates signed a Trade and Investment Framework Agreement (TIFA), which provided a formal framework for the discussion of economic reform and trade liberalization. In general, TIFAs promote the establishment of legal protection for investors, improvements in intellectual property rights protection, more transparent and efficient customs procedures, and greater transparency in government and

Ibid. Art. 3.4.4 of the CCRF. 58 See Negotiations and Agreements, European Commission, ec​.europa​.eu/​trade/​policy/​countries​ -and​-regions/​negotiations​-and​-agreements/​ (last visited Jan 28, 2022). 59 See International Trade Administration, United Arab Emirates: Trade Agreements, www​.trade​ .gov/​country​-commercial​-guides/​united​-arab​-emirates​-trade​-agreements (last visited Jan 28, 2022). 56 57

246  Research handbook on digital trade commercial regulations.60 Then, in 2005, the US and the UAE began negotiations for a FTA, although negotiations were not completed, and no further process has been made since. More recently, in 2012, the UAE became a party to the US–GCC Framework Agreement for Trade, Economic Investment, and Technical Cooperation. The UAE subsequently ratified the US–GCC agreement in 2014 through Federal Decree No. 86. Furthermore, since 2012 the US and the UAE have held “several iterations of the U.S.–UAE Economic Policy Dialogue, which provides a platform to collaborate on economic issues and address irritants to the bilateral commercial relationship.”61 Even though many of these agreements between the United States and the United Arab Emirates have not been formally related to digital trade or e-commerce, likely due to the novelty of these advancements within the country, existing trade agreements nonetheless influence e-commerce. More often than not, however, ordinary trade agreements are not compatible with the specific issues concerning e-commerce and digital trade, limiting their growth. Therefore, free trade agreements that are specifically tailored to address e-commerce and digital trade are favorable and, depending on the terms, can result in a favorable environment that can further development. 4.2 Egypt Like the United Arab Emirates, Egypt does not have any free trade agreements with the United States or European Union on the subject of digital trade or electronic commerce. Egypt does have a free trade agreement with the European Union, however.62 The European Union–Egypt Free Trade Agreement (or EU–Egypt Association Agreement) has been in force since 2004 and was designed to create a free trade area between the two countries by “removing tariffs on industrial products and making agricultural products easier to trade.”63 Originally drafted in 2001, this FTA provided for immediate duty-free access of Egyptian products into EU markets, while duty-free access for EU products was phased in over a 12-year period. In 2010, the EU and Egypt created an agricultural amendment, liberalizing trade in more than 90 percent of agricultural goods. In addition to the EU–Egypt FTA, Egypt is also a party to multiple multilateral trade agreements. For example, Egypt is part of the General Agreement on Tariffs and Trade (GATT), which deals with international trade, and the General Agreement on Trade in Services (GATS), which deals with trade in services. Egypt also has many bilateral agreements with many countries on other subjects, albeit none involving digital trade nor electronic commerce. Like other countries without free trade agreements that directly target digital trade and e-commerce, however, there is a sort of dissonance between regular FTAs involving trade and actual digital trade and e-commerce. The reason for this dissonance is that digital trade comes with much different challenges than regular trade, and thus, FTAs might not be sufficient to address these challenges. It is for this reason that actual FTAs on e-commerce and digital trade are needed, as well as to ensure that they have a proper environment to thrive in.

Ibid. Ibid. 62 Ibid. 63 See European Commission, Egypt: Countries and Regions, ec​.europa​.eu/​trade/​policy/​countries​ -and​-regions/​countries/​egypt (last visited Jan 28, 2022). 60 61

Legal approaches to the regulation of digital trade  247 4.3 Jordan Unlike other countries in the Middle Eastern region, Jordan and the United States have a free trade agreement on e-commerce. In fact, the US–JO FTA is the first FTA to include a provision concerning e-commerce.64 The e-commerce provision of the US–JO FTA applies to goods and services traded over the medium of the internet. The FTA covers e-commerce but does not define the term.65 The FTA also uses the term “digitized products” without providing a definition. The FTA could have listed illustrative examples of digitized products, such as electronically traded software, books, and music. The FTA requires that the parties do not deviate from their “existing practice” of not imposing tariffs on electronic transmissions.66 This language is based on the US Internet Tax Freedom Act of 1998.67 The FTA also requires that the parties do not establish “unnecessary” barriers on electronic transmissions, which include digitized products. The FTA requires the US and Jordan to make publicly available all laws, regulations, and requirements affecting e-commerce. The e-commerce provision of the US–JO FTA does not extend to all e-commerce matters, such as domain name, internet security, and intellectual property.68 The FTA covers border See Thomas Cottier, The Impact of New Technologies on Multilateral Trade Regulation and Governance, 72 Chi.-Kent L. Rev. 415, 426, 435 (1996) (concluding that the new technologies are the most important driving force in the process of globalization of the economy). 65 There is no universally agreed definition of e-commerce. However, in the WTO Work Program on Electronic Commerce, e-commerce is understood to mean the production, distribution, marketing, sale or delivery of goods and services by electronic means. A commercial transaction can be divided into three main stages: the advertising and searching stage, the ordering and payment stage, and the delivery stage. Any or all of these stages may be carried out electronically and may therefore be covered by the concept of e-commerce. In other words, a buyer may purchase a book via the internet and be delivered physically later on or he can purchase and download the book via the internet. 66 See United States (US)–Jordan: Agreement Between the United States of America and the Hashemite Kingdom of Jordan on the Establishment of a Free Trade Area, supra n. 695, art. 7.1.a. 67 The act, which has the purpose of promoting universal access and less burdensome internet tax policy, imposes a moratorium on all taxation of internet access and on “multiple” or “discriminatory” taxes on e-commerce. The act also includes a declaration that the internet should be free of tariffs, trade barriers, and other restrictions. Moreover, the act asks the US President to pursue “international agreements” to ban such tariffs and other trade barriers. See Internet Tax Freedom Act of 1998, 47 U.S.C. §151 (2000). The moratorium begins on Oct. 1, 1998, and ends on Nov. 1, 2003. On Nov. 19, 2004, Congress passed legislation S.150 that reinstated a four-year moratorium on internet access taxes and multiple and discriminatory taxes on e-commerce. The new legislation applies retroactively to Nov. 2003. Thus, the new legislation will expire in 2007. 68 Legal questions brought about by e-commerce but not discussed here include, but are not limited to, domain names in cyberspace and trademarks. See Suzanna Sherry, Haste Makes Waste: Congress and the Common Law in Cyberspace, 55 Vand. L. Rev. 309 (2002). See also Serge G. Avakian, Global Unfair Competition in the Online Commerce Era, 46 UCLA L. Rev. 905 (1999). For the effects of new methods, such as price comparison, cookies, exclusive provision by Internet Service Provider (ISP), and business alliances, used by e-retailers to sell their products online that may raise issues of anticompetitive practices, see Alan E. Wiseman, The Internet Economy: Access, Taxes, and Market Structure, chapter 4 (Brookings Instn. Press 2000). For discussion of privacy issues such as cookies on users’ personal computers and unsolicited commercial communications without the consent of the addressee (opt-in system) in the age of e-commerce and the U.S. reliance on market oriented industry self-regulation approach of handling personal data see Joel R. Reidenberg, Restoring Americans’ Privacy in Electronic Commerce, 14 Berkeley Tech. L.J. 771 (1999). For discussion on consumers concern regarding identity, privacy, 64

248  Research handbook on digital trade trade measures only, such as tariffs, unnecessary trade barriers, and services that facilitate e-commerce. The FTA does not determine if digital products should be treated as goods or as services.69 Determining whether an e-product is a good or service is a crucial assessment. If an e-product is a good, then it will be subject to the national treatment rules of the FTA. In contrast, if an e-product is a service, then each party may impose restrictions on market access and national treatment.70 The vagueness of the e-commerce provision of the US–JO FTA suggests that negotiators faced a dilemma while drafting it. On the one hand, the FTA parties had to recognize the importance of e-commerce to world trade. On the other, there is no universal regulatory system for e-commerce within the WTO that would have guided the FTA negotiators.71 The U.S. and Jordan adopted an approach whereby the current rules of trade that apply in the physical world would also apply in the virtual world. For example, the e-commerce provision of the US–JO FTA holds to the existing rules regarding market access, services, and transparency. Thus, the drafters of the US–JO FTA did not create entirely new rules for the new virtual environment. The next question that must be addressed is how to identify the party benefits from the e-commerce provision of the US–JO FTA. The US, as of 2004, had the largest number of internet users, estimated at 200 million, and about 70 percent of all websites are based in the US.72 Even more, 85 percent of the world’s internet revenue in 1996–7 was generated in the US.73 In 2003, online retail sales in the US were $55 billion.74 In contrast, in Arab countries,

data collectors, and security of financial information see Nicole Ladouceur, Calibrating the Electronic Scale: Tipping the Balance in Favor of Vigorous and Competitive Electronic Market for Consumers, 25 Can.–US L. Rev. 295 (1999). For discussion of cyber-security issues and how not only legal but architectural responses are needed see Neal Kumar Katyal, Digital Architecture as Crime Control, 112 Yale L.J. 2261 (2003). 69 There is disagreement among WTO members on this point. In other words, it is unclear whether digital products delivered over the Internet should be classified as goods or as services for the purpose of WTO regimes and rules. E-products include books, music, videos, software, and other newly emerging media, which in a rapidly growing digital market may arguably be developed in strictly computerized forms simply to avoid tariffs and strict international legal standards on physical goods and services. See Kristi L. Bergemann, A Digital Free Trade Zone and Necessarily-Regulated Self-Governance For Electronic Commerce: The World Trade Organization, International Law, and Classical Liberalism in Cyberspace, 20 J. Marshall J. Computer & Info. L. 595, 600–1 (2002) (discussing the option for self-governance of the Internet based on existing quasi-legal norms within the cyber-community, rather than regulatory framework within national governments or supranational organization). 70 The US has been the primary advocate of the position that digitized products should be classified as goods. See Stewart A. Baker et al., E-Products and the WTO, 35 Int’l Law 5,7 (2001). 71 The first time the WTO addressed internet trade, although under GATS, was its ruling on US restrictions on cross-border internet gambling services. The ruling found that online gambling restrictions imposed by the US at the federal and state levels violated its market access commitments under sub-sector 10.D (other recreational services) of its GATS schedule which corresponds to CPC heading 96492. The restrictions denied Antiguan-based gaming operators access to the US market. See Report of the Panel on United States-Measures Affecting the Cross-border Supply of Gambling and Betting Services, Nov. 10, 2004, WTO Doc. No. WT/DS285/R. 72 See E-Commerce Takes Off, Economist 3, 20 (May 15, 2004) (in the US, women outnumber men online). 73 See WTO Secretariat, supra note 870, at para.21. See also Catherine L. Mann et al., Global Electronic Commerce: A Policy Primer, 16 (Inst. Intl. Econ. 2000). 74 See The Economist, supra n. 72, at 3 (the figure, according to the US Department of Commerce, excludes online travel services. For example, the owner of Seattle-based expedia.com and hotels.

Legal approaches to the regulation of digital trade  249 including Jordan, online commerce is negligible.75 In 2001, Jordan had only 120,000 internet users.76 Most commercial websites in Jordan are aimed at cataloging and advertising products rather than selling online. Online commerce requires use of modern financial and credit transactions and payments via consumer credit cards. One of the obstacles for the development of e-commerce in Jordan is the fact that the number of credit card holders is small.77 One can describe Jordan as a cash-transaction society.78 In contrast, the US is a cashless society. The differences between the US and Jordan with respect to internet accessibility, efficient telecommunication infrastructure, and volume of e-commerce should have been an indication for the FTA negotiators that a more balanced provision would be needed to accommodate the e-commerce environments in each country. The FTA lacks legal obligations or recommendations that the two parties share information and experience regarding policies intended to encourage the development of e-commerce. 4.4

Saudi Arabia

Saudi Arabia is yet another example of an Arab country that does not have an FTA with the United States or the European Union on digital trade and e-commerce. Saudi Arabia does, however, have many multilateral and bilateral trade agreements, and even a framework agreement with the United States, which no doubt has continued to influence digital trade and e-commerce since the agreement was signed in 2003. The Trade and Investment Framework Agreement (TIFA) provided both countries with the opportunity for ongoing structured dialogue on economic reform and trade liberalization. This agreement helped promote “the establishment of legal protections for investors, improvements in intellectual property protection, more transparent and efficient customs procedures, and greater transparency in government and commercial regulations.”79 Furthermore, the United States and Saudi Arabia hold TFA negotiations on various trade and policy-related issues every one to two years. Apart from the Trade and Investment Framework Agreement that Saudi Arabia has with the United States, Saudi Arabia does not have any notable FTAs with the US or the EU. As of 2021, however, Saudi Arabia has announced that it is looking to engage in free trade agreements with 11 countries in the near future, including China, India, Pakistan, Australia, New

com sold $10 billion worth of travel in 2003 (some 20 percent of travel in the US is bought online). It also excludes pornography which made $2 billion in the US in 2003. The figure also excludes sales of financial services, ticket-sales agencies, online dating, tracing ancestors, and gambling. The gambling business is worth $6 billion. About $24 billion worth of trade was done in 2003 on the California-based eBay, the biggest online auctioneer. Used-car sales are now one the biggest online growth areas). See also Barrett J. Willingham, Electronic Commerce and the Free Trade of the Americas, 6 NAFTA: L. & Bus. Rev. Am. 483, 491 (2000) (by 2003, business-to-business trade within the US would soar to $1.3 trillion, an amount that would represent more than 9 percent of total US business). 75 See Trade Facilitation and E-commerce in the ESCWA Region, UN Economic and Social Commission for Western Asia, at 5, U.N Doc. E/ESCWA/ED/2001/2 (2001). 76 Ibid at 36. 77 Ibid. 78 Buying through using credit card is not the norm. Cash is the norm. Generally, credit cards are accepted at major hotels, restaurants, and other establishments. There might be a lack of system of credit history check. 79 See International Trade Administration, Saudi Arabia Trade Agreements, Export.gov (2020), www​.export​.gov/​apex/​article2​?id​=​Saudi​-Arabia​-Trade​-Agreements (last visited Jan 28, 2022).

250  Research handbook on digital trade Zealand, Britain, Indonesia, the Philippines, Bangladesh, Sri Lanka, and the United States of America.80 As an effort to increase GDP through non-oil related exports, Saudi Arabia has announced that it would like to export services including “transport, distribution, professional and financial services, communication services, postal services as well as express mail, media, hotel, construction and contracting, education and training, travel and tourism, environmental, and entertainment.”81 Undoubtedly, digital trade and e-commerce will play a large role in these FTAs. Without formal FTAs on the subject of digital trade and e-commerce, however, development in these areas may become stagnant, as a result of an unfavorable environment. As such, FTAs and other agreements that support the growth of digital trade and e-commerce and address their unique challenges and needs will likely be the key to further growth in these areas, especially as the public in Saudi Arabia continues to become more comfortable with digital trade and e-commerce.

5. CONCLUSION In recent years, trade has been heavily influenced by the emergence of new technology, the internet, and the digitalization of society. Thus, digital trade and electronic commerce have picked up traction throughout the world, including the Middle East, bringing about change and new issues to this once well-established area. Middle Eastern countries, in particular, are very recent developers in terms of digital trade and e-commerce, as key changes in society take place. The COVID-19 pandemic and increased smartphone and internet usage among the population in countries such as the United Arab Emirates, Egypt, Jordan, and Saudi Arabia are especially salient in the development of digital trade and e-commerce. Since many Middle Eastern countries are barely in their infancy in terms of developing digital trade and e-commerce frameworks, however, these areas may be burdened by ill-fitting existing laws and regulations. That is to say, digital trade and e-commerce need to be addressed individually, apart from regular trade, so as to target issues that are unique to these areas. Still, changes in society have triggered an uptick in digital trade and e-commerce, which have proven to be a powerful tool for growth in many Middle Eastern countries, making this region promising for digital trade and electronic commerce advancements. Furthermore, these countries have been in the process of developing data protection regulations to further encourage advancements in digital trade and e-commerce. The reason for this is that one of the major factors influencing the growth of digital trade and e-commerce is how well customers are protected. In many Middle Eastern countries, like all countries around the world, consumers are wary of online transactions and maintaining an online presence because they are not certain of how their personal information will be used. Thus, many of these countries have taken the initiative to create regulations to dictate how businesses and companies can use consumer data.

80 See Jana Salloum, Saudi Arabia Looks for Free Trade Agreement with 11 Countries in Exports Push Arab News (2021), www​.arabnews​.com/​node/​1914176/​business​-economy (last visited Jan 28, 2022). 81 Ibid.

Legal approaches to the regulation of digital trade  251 Certain countries, such as Jordan, have also been in the process of developing free trade agreements with countries such as the United States in order to better accommodate digital trade and e-commerce. It is important to note, however, that many Arab countries lack FTAs with the EU and US on the subject of digital trade and e-commerce. This is the case with countries such as the United Arab Emirates, Egypt, and Saudi Arabia. This means that these countries do not have favorable environments for digital trade and e-commerce to further develop. It is relevant that several of these Middle Eastern countries have announced their intent to develop FTAs with the United States and the European Union, which could influence the advancement of digital trade and e-commerce in the near future. Many Middle Eastern countries are thus on the cusp of great developments in this area. In conclusion, an in-depth study of digital trade and e-commerce in key countries such as the aforementioned reveals a general trend—that many countries in this region are at the beginning stage of large advancements in these areas. Furthermore, an analysis of the situations in each country reveals that although each country faces its own challenges when it comes to the growth of digital trade and e-commerce, many of the conditions are favorable or can be improved upon to foster further growth. Thus, the Middle Eastern region, and countries such as the United Arab Emirates, Jordan, Egypt, and Saudi Arabia, have proven to be very promising in terms of digital trade and e-commerce, although they still have quite a way to go before they can compare to major powerhouse countries.

16. Latin America: the leap from the single window to the Single Submission Portal as a way of internalizing SMEs Manuel Quindimil

1. INTRODUCTION The digitization process has had a huge impact on international trade. However, the great economic benefits generated by the application to trade of information and communications technologies (ICT) are not sufficiently reaching developing countries. The irruption of ICT still presents many challenges for international trade players and, especially, for governments that must adopt the appropriate public policies to take advantage of technological advances. The continuous appearance of concepts such as digital transformation, digital trade and the study of the application of technologies such as blockchain to international trade, makes institutional and regulatory systematization more complex to maximize the economic benefits of the digital economy. This problem is accentuated in the case of developing countries that lack the necessary physical and human resources for the digital transformation of their societies. One of the greatest advances in digital trade has been the automation of processes linked to international trade. The fundamental tool to achieve this objective has been the design and implementation of international trade single windows. Although the obligation to put into operation the international trade single windows has been foreseen in the WTO Trade Facilitation Agreement,1 there are different models and stages of development of this trade facility. The range of platforms fluctuates from those that still have to present physical papers and where there is little interoperability between State agencies, to those that constitute a single point of delivery of electronic documents and operations are carried out with regulatory agencies (business-to-government, B2G), to the relationship with carriers, banks, cargo agents, and all international trade operators (business-to-business, B2B). These latest generation platforms are called Single Submission Portals. The countries most advanced in these projects, such as Singapore, South Korea, Japan, and the ASEAN region, are already beginning to interoperate. The enormous leap in competitiveness generated by the

1 The Trade Facilitation Agreement of the World Trade Organization establishes the obligation of the parties to implement a single window for foreign trade: Trade Facilitation Agreement. 4.1 Members shall endeavour to establish or maintain a single window, enabling traders to submit documentation and/or data requirements for importation, exportation, or transit of goods through a single-entry point to the participating authorities or agencies. After the examination by the participating authorities or agencies of the documentation and/or data, the results shall be notified to the applicants through the single window in a timely manner.

252

Latin America  253 complete digitization of international trade processes has direct effects on the export performance of small and medium enterprises (SMEs) that can interact in this digital scenario. This article will focus on analyzing whether Latin American countries (LAC), which are experiencing delays in their process of digitalization and automation of international trade processes, could advance by taking a technological leap with the adoption of a single point of presentation of e-documents that performs operations on a B2G and B2B basis. This enormous effort will have the reward of internationalizing Latin American SMEs, economic units that have little participation in international trade. Aware of the great difficulties of implementing single submission point schemes, the most plausible solutions may be taken at the regional or subregional level. The lack of sufficient resources to achieve these goals will also require international cooperation. Section 2 of this chapter reviews the difficulty of coping with digital transformations in Latin America. Reference will be made to the difficulties that LAC face in inserting themselves into the digital economy. In section 3, analyzing how digital transformation impacts international trade, the recommended public policies are reviewed to obtain the benefits of what has been called digital trade. In this area, LAC also has many challenges in relation to public matters—a problem that is analyzed here. The state of the art regarding single windows is examined in section 4. This information is contrasted with the state of evolution of said tool in Latin America. The advanced pilot projects in the region are described to analyze the possibility of promoting them to reach 100 percent of the digitalization of processes. Special emphasis is given to the different Latin American examples to evaluate good practices so that they are scalable throughout the region. The few measures to facilitate digital trade in the region are studied. Finally, some suggestions are presented to advance the automation of international trade processes and how these policies can improve the internationalization of SMEs.

2.

THE SLOW DIGITAL TRANSFORMATION IN LATIN AMERICA

Prior to considering whether it is possible to reach a single portal for international trade procedures in Latin America, the first thing that is required is to review how digital transformation is being carried out in Latin America. Digitization has profoundly changed the global production of goods and services. Technological evolution has also had a strong effect on international trade. As always happens, technological progress is not simultaneously reflected, either in national regulations or in international trade standards. The United Nations Conference on Trade and Development (UNCTAD)2 indicates that as the world is only in the early stages of digitalization, the evolving digital economy and several other related economic terms lack widely accepted definitions. This situation generates the difficulty of deeply understanding the categories and definitions that allow for developing the conceptual frameworks that come from the economy based on data. This phenomenon was

UNCTAD “Digital Economy Report 2019 Value Creation and Capture: Implications for Developing Countries Value Creation and Capture: Implications for Developing Countries,” 2019, UNCTAD/DER/2019 https://​unctad​.org/​webflyer/​digital​-economy​-report​-2019 accessed May 29, 2022. 2

254  Research handbook on digital trade highlighted in the Organisation for Economic Co-operation and Development (OECD) Going Digital Summit in March 2019:3 Digital transformation is complex and evolving rapidly. Policy decisions must increasingly be made under uncertainty about future digital and other developments […] While progress has been made in answering some of the most pressing and difficult questions that governments face today, more work is needed to understand some complex issues and to design resilient policy frameworks in response.

Facing these complex scenarios posed by digitization presents greater challenges to developing countries to seeking regulate in favor of their own economies. Most developing countries face many constraints in trying to benefit from the digital economy. Governments and other stakeholders need a basic understanding of the dynamics of the digital economy before they can formulate and implement relevant policies. Digitization requires a broad set of complex interrelated public policies. Among the fundamental ones are expanding ICT connectivity, cross-border data flows, skills and education, trade, and international support.4 Consequently, e-trade facilitation is framed within a range of regulations in multiple areas that must function simultaneously and in sync. Moreover, actions need to be taken simultaneously in several areas, as isolated actions in only one area at a time will have limited impact.5 As already indicated, analysis of how to improve the digitization of international trade processes requires a prior review of the state of the digitization development in Latin America. As there is no exhaustive definition of what is meant by digital transformation, different reports will be taken that indicate a modest performance by the different LAC. From a general point of view, Juan Carlos Navarro6 maintains that access to new ICT by LAC has been late and partial, as illustrated by all available indicators, such as the number of personal computers, internet access, and access to broadband. These deficiencies should be considered when analyzing the effects of innovation on productivity, since ICT is a general-purpose technology that has a cross-sectional impact on all economic sectors. This means digitization has not been adopted by regional SMEs. Regarding the preparedness of Latin America for the digital economy, the disadvantages observed are the underperforming education systems, limited broadband coverage, and most firms not being engaged in innovation activities, much less in digital transformation.7 Juan Carlos Navarro,8 citing the conclusions of the Ministerial Conference on the Information Society in Latin America and the Caribbean, enumerates the region agreed on a regional digital agenda running up to 2022. To adapt and reap the benefits of the data-driven economy, Latin American governments should undertake reforms in the following areas: 3 OECD “The OECD Going Digital Summit. A Summary,” March 11–12, 2019, www​.oecd​.org/​ digital/​going​-digital​-project/​ accessed May 29, 2022. 4 UNCTAD “Information Economy Report 2017: Digitalization, Trade and Development” 2017 UNCTAD/IER/2017/Corr.1 https://​unctad​.org/​webflyer/​information​-economy​-report​-2017 accessed May 29, 2022. 5 UNCTAD (n 2) 2019. 6 Juan Carlos Navarro “The Digital Transformation Imperative. An IDB Science and Business Innovation Agenda for the New Industrial Revolution” Aug 2018 https://​publications​.iadb​.org/​en/​digital​ -transformation​-imperative​-idb​-science​-and​-business​-innovation​-agenda​-new​-industrial accessed May 29, 2022. 7 ibid. 8 ibid.

Latin America  255 – – – – – –

Improve the level of digital education of the population. Government support for the digital transformation of businesses, particularly SMEs. Constitute a digital ecosystem, promoting the development of start-ups. Strengthen the supply of human capital for the ICT sector. Undertake regulatory reform to adapt to the digital economy. Invest in broadband infrastructure.

The OECD9 observes that policy areas related to access to and use of digital technologies, communication infrastructure and the future of jobs are currently incorporated more in National Development Plans (NDP) and the Digital Agenda in LAC than any other area. The digital transformation takes in a series of public policies that would have to be consistent and coordinated between the National Development Policies and the Digital Agendas. Due to the transversal nature and the large number of public entities involved, not all public policies are always included, nor do all stakeholders participate. It should be noted that there are different guide schemes in the region to achieve digital transformation in LAC. Each of the countries vary in the way they adopt their guidelines and their institutional structure. On the other hand, the transversal nature of digitization and the number of state entities makes it very difficult for countries to take a holistic and effective approach to this mission. In general terms, in most LAC, either the Digital Agenda or National Development Plans instruments have not been planned in coordination with development policies and international trade policies. This situation is already an obstacle that would have to be considered, mainly by each of the countries. This objective should mainly be fulfilled at the national level. The implementation of public policies for incorporation into the digital economy must respond to the specific characteristics of each of the LAC.

3.

DIGITAL TRADE POLICIES ENFORCEMENT IN LAC

As stated by the OECD,10 digital transformation has reduced the costs of engaging in international trade, facilitated the coordination of global value chains (GVCs), helped diffuse ideas and technologies and connected a greater number of businesses and consumers globally. Kati Suominen,11 in relation to the changes introduced by the digital transformation in commerce, points out the following benefits: greater variety of goods at lower cost; services incorporated into higher value-added manufactures; more agile and traceable processes, and the internationalization of SMEs. As technological progress has deepened, the concept of digital trade has started to gain popularity. In this sense, it is necessary to define what is meant by the concept of digital trade.

9 OECD et al. “Latin American Economic Outlook 2020: Digital Transformation for Building Back Better” OECD Publishing, Paris, https://​doi​.org/​10​.1787/​e6e864fb​-en. 10 OECD “Digital Trade. The Impact of Digitalisation on Trade” www​.oecd​.org/​trade/​topics/​digital​ -trade/​accessed May 29, 2022. 11 Suominen with ConnectAmericas and Asociación Latinoamericana de Internet “Accelerating Digital Trade in Latin America and the Caribbean” 2017.

256  Research handbook on digital trade Although there is no unequivocal definition of digital trade, López-González and Jouanjean12 indicate that it involves digitally enabled or digitally ordered cross-border transactions in goods and services which can be digitally or physically delivered. Despite the benefits that digital trade brings, as Javier López-González13 maintains, the benefits of digitalization for trade are not automatic and can be compromised by restrictive trade policies. The World Economic Forum14 points out the following digital trade priorities: – Establish the conditions for the development of e-commerce. – Generate interoperability of cross-border data flow through framework trade agreements and co-regulation. – Promote the rapid expansion of digital services. – Support framework agreements for the proper functioning of digital services, from remote work to digital taxation. – Identify new technologies—including cloud services, DLT, and 3D printing—and evaluate the pros and cons of the public policies to be adopted. Regarding the discipline of digital trade, different public policies are pointed out that governments must adopt to obtain the greatest benefits that this new way of trading can bring. Joshua P. Meltzer15 suggests that to maximize the potential of the internet and data to expand international trade, the LAC region needs a digital trade policy and complementary domestic reforms. Among the reforms that must be adopted in national and regional trade policies, he lists the following: allow cross-border data flows; ensure non-discrimination; eliminate any data localization requirements; expand market access for services; recognize digital signatures; ensure effective intellectual property protection; implement a third-party intermediary liability regime; agree a common de minimis customs duty; do not impose customs duties on electronic transmissions; ensure consumer protection; enable digital payments; have an effective competition policy; protect privacy; increase regulatory certainty; expand cooperation on cybersecurity, among others. The possibility of achieving an automated digitalization process in Latin American international trade requires a sophisticated combination of public policies. To do this, it is necessary to review and articulate the digital agendas and trade policies that may lead to the facilitation of digital trade. Because of the COVID-19 pandemic, many have taken various measures LAC to promote digital trade. However, despite the efforts made, there are many challenges indicated by the Inter-American Development Bank (IADB)16 in the matter: – Promote internet infrastructure and regulations, encouraging public–private participation. – Foster trade facilitation and logistics to stimulate e-commerce. 12 Javier López González and Marie-Agnes Jouanjean “Digital Trade: Developing a Framework for Analysis” OECD Trade Policy Papers 205, OECD Publishing 2017. 13 Javier López-González “The Changing Nature of Digital Trade, Current and Future Barriers and Ideas to Overcome Them” Wilson Center, April 21, 2021. 14 World Economic Forum “Digital Trade” www​.weforum​.org/​projects/​digital​-trade accessed May 29, 2022. 15 Joshua P. Meltzer “A Digital Trade Policy for Latin America and the Caribbean” (IDB Technical Note; 1483) 2018. 16 “Post Pandemic COVID-19 Economy Recovery Enabling Latin America and the Caribbean to better harness e-commerce and digital trade” Discussion Paper Nº IDB -DP – 00881.

Latin America  257 – Modernize regulations that allow electronic and digital payments. – Take measures to reduce entry barriers for SMEs and avoid unfair practices and against free competition in e-commerce. – Training SMEs for the new business models proposed by e-commerce. – Promote cooperation in digital policies to enhance cooperation at the regional level, especially within the context of the Digital Agenda for Latin America and the Caribbean (eLAC2022).17 To carry out an analysis of the impact of the automation of foreign trade processes on the internationalization of SMEs, it is necessary to determine the definition of digital trade facilitation. Digital trade facilitation refers to the application of modern information and communication technologies (ICT) to simplify and automate international trade procedures.18 E-trade facilitation is encapsulated within the concepts of digital transformation and digital trade. From the discipline of digital trade, emphasis is placed on what Jonathan Koh19 points to as making processes, documents, and information digital and transmissible in a legally recognized manner with the aim of reducing the costs of companies linked to international trade. In this sense, this author emphasizes that trade digitization clearly has massive benefits. It enables dealing with cross-border trades to be easier, with less paperwork; faster; and less costly. Koh insists that the digitization of documents must aim to reduce the costs that SMEs have when they operate in international trade. Regarding the efficiency of the processes involved in international trade LAC performs better than lower-income regions such as Sub-Saharan Africa and South Asia in the trading across borders component of the World Bank’s Doing Business ranking; however, it scores considerably worse than the developed countries of the OECD.20 In terms of trade facilitation, considering the WTO Trade Facilitation Agreement (TFA), in most disciplines (particularly in terms of transparency and formalities) LAC has had a high content of compliance in its implementation. Where the greatest challenges arise is in what can be considered some TFA+ disciplines, specifically cross-border paperless, international cooperation arrangements, trade facilitation for women, and trade facilitation for SMEs. To be able to specify the degree of progress in the adoption of information technologies in international trade processes, the results obtained by a study carried out by the Economic

17 The Digital Agenda for Latin America and the Caribbean (eLAC) is a strategy aimed at 2022, which promotes the use of digital technologies as instruments for sustainable development. Its mission is to encourage the development of the digital ecosystem in Latin America and the Caribbean through a process of integration and regional cooperation, strengthening digital policies that drive knowledge, inclusion, and equality, innovation, and environmental sustainability: eCLAC Digital Agenda for Latin America and the Caribbean (eLAC2022), www​.cepal​.org/​en/​projects/​digital​-agenda​-latin​-america​-and​ -caribbean​-elac2022 accessed May 29, 2022. 18 Y. Duval and K. Mengjing “Digital Trade Facilitation: Paperless Trade in Regional Trade Agreements” ADBI Working Paper 747. Tokyo: Asian Development Bank Institute 2017, www​.adb​.org/​ publications/​digital​-trade​-facilitation​-paperless​-trade​-regional​-trade​-agreements. 19 Jonathan Koh, “Digital Trade Deals Should Focus More on Small Businesses” (DIGITAL TRADE, September 14, 2021) www​.hinrichfoundation​.com/​research/​article/​digital/​digital​-trade​-deals​ -should​-focus​-more​-on​-small​-businesses/​ accessed May 29, 2022. 20 S. Herreros, “Digital and Sustainable Trade Facilitation in Latin America and the Caribbean” Regional Report 2021 (LC/TS.2021/126), Santiago, Economic Commission for Latin America and the Caribbean (ECLAC), 2021.

258  Research handbook on digital trade Commission for Latin America and the Caribbean (ECLAC) are considered.21 According to the definition of the survey carried out by ECLAC Digital Trade Facilitation Measures includes two categories: Paperless Trade22 and Cross-Border Paperless Trade. ECLAC includes ten measures in the paperless category, these being: electronic-automated customs established; internet connection available to customs and other trade control agencies at border crossings; electronic single window system; electronic submission of customs declarations; electronic application and issuance of import and export permits; electronic submission of sea and air cargo manifests; electronic applications and issuance of preferential certificates of origin; electronic payment of customs duties and fees; and electronic application for customs reforms. In this field of paperless measures (the use of information and technology to fulfill trade-related formalities) very good levels of compliance were recorded in the region as an 87 percent rate. Within this group of measures, with 8 of its 10 measures showing implementation rates above 80 per cent. Nevertheless, ECLAC indicates that there is there is considerable variation in implementation levels across individual measures and subregions.23 One of the topics to be analyzed in this article is the operation of single windows for foreign trade, a measure that is included in the category of paperless. Progress in the implementation of single windows has not been a simple process for the countries of the region, involving various complex issues such as financial, technological, and interagency coordination requirements. In line with the survey, this is reflected in the fact that LAC obtained an 81 per cent implementation rate, with five countries (Argentina, Cuba, Ecuador, Guatemala, and Panama) reporting partial implementation. All of them indicated that there were still relevant stakeholders not connected to the single electronic window. Regarding the measures aimed at cross-border paperless,24 ECLAC considers six measures within the category of cross-border paperless trade, among which: laws and regulations for electronic transactions (for example, e-commerce law, e-transactions law); recognized certifications authorities issue digital certificates to traders to conduct electronic transactions; documents electronically exchanged between countries such as customs declarations, certificates of origin, and sanitary and phytosanitary certificates. From private sector agents, banks and insurers retrieve letters of credit electronically without lodging paper-based documents. The evaluation indicates that this category is the one with the lowest degree of compliance by LAC (57 percent). The assessment specifies the measures that have presented inconveniences in their implementation, such as the electronic cross-border exchange of customs declarations (43 percent), certificates of origin (57 percent), and sanitary and phytosanitary certificates (45 percent). ECLAC seeks the explanation for this low performance in the fact that the implemen-

21 The survey was considering 14 participating LAC countries: Argentina, Colombia, Chile, Costa Rica, Cuba, Dominican Republic, Ecuador, Guatemala, Mexico, Panama, Paraguay, Peru, Saint Kitts and Nevis, and Uruguay. 22 Article 3(c) of the Framework Agreement defines paperless trade as “trade in goods, including their import, export, transit and related services, taking place on the basis of electronic communications, including exchange of trade related data and documents in electronic form.” The Framework Agreement on Facilitation of Cross-border Paperless Trade in Asia and the Pacific, adopted by ESCAP in May 2016 and which entered into force on February 20, 2021. 23 S. Herreros (n 20) 2021. 24 ibid.

Latin America  259 tation of these measures requires the support of a sophisticated ICT infrastructure and close cooperation between the relevant agencies of the countries exchanging information.25

4.

THE TRANSITION FROM THE SINGLE WINDOW FOR FOREIGN TRADE TO THE SINGLE SUBMISSION PORTALS

To propose that SMEs can gain competitiveness through the automation and digitization of processes, it is necessary to study the evolution of the implementation of international trade single windows. One of the most important items in the automation of processes related to foreign trade is the development of electronic single window schemes. One of the keys to improving the competitiveness of SMEs through a single trading platform is to understand how Latin America can speed up the process of modernizing its foreign trade single windows to reach a Single Submission Portal. To start, it is necessary to define what is meant by single window in this article. To this end, the provisions of the United Nations Centre for Trade Facilitation and Electronic Business (UN/CEFACT) Recommendation No 33 defines Single Window as “a facility providing trade facilitation that allows parties involved in trade and transport to lodge standardized information and documents with a single-entry point to fulfill all import, export, and transit-related regulatory requirements. Individual data elements should only be submitted once electronically.”26 The United Nations Economic and Social Commission for Asia and the Pacific and the United Nations Economic Commission for Europe UNESCAP/UNECE27 has determined, based on the development of single windows, the following evolutionary stages: Level 1: Paperless Customs (Development of paperless customs declaration system). Level 2: Regulatory Single Window (Integration of paperless customs with other regulatory bodies issuing trade/import/export/transit-related permits and certificates, and other related documents). Level 3: Port Single Window or B2B Port Community System (Extension of the single window to serve entire trade and logistics communities within the airports, seaports, and/or dry ports). Level 4: Fully Integrated Single Window (Creation of an integrated domestic logistics platform interlinking the administrations, companies, and the service sectors to better manage the entire chain of import–export operations). As some countries achieved high levels of interoperability, trade windows reached the Regional Information Exchange, managing to incorporate cross-border operations. At this stage, there was already talk about the Single Submission Portals projects, developed at a national level

S. Herreros (n 20) 2021. UN/CEFACT No.33 Recommendation and Guidelines on Establishing a Single Window to Enhance the Efficient Exchange of International Trade Information Between Trade and Government, ECE/TRADE/352/Rev.1 eISBN 978-92-1-005201-6 2020 Edition p.3. 27 UNESCAP/UNECE Single Window Planning and Implementation 2012 Guide ECE/TRADE/404, www​.unescap​.org/​resources/​single​-window​-planning​-and​-implementation​-guide accessed 29 May 2022. 25 26

260  Research handbook on digital trade and that are beginning to interoperate at a regional level.28 The proposal to convert a digital platform for the entry of data aimed at the government and the exchange of information of international trade actors requires the analysis of the concept and implementation process of the Single Submission Portals. As technological progress led to countries such as Singapore, South Korea, and Japan beginning to integrate their foreign trade single windows with platforms that provide services to private operators, a new level was added. The concept of Single Submission Portals is established in the UN/CEFACT Recommendation No 37.29 A Single Submission Portal is ‘an access point that allows traders to exchange information, in a standard format and related to a specific activity, with relevant parties including government agencies’. The Single Submission Portal allows B2B operations to be carried out such as contracting for transport, logistics, and financial services, and makes regulatory processes easier through B2G information exchange. The recommendation indicates that these portals have been implemented at the national level. Likewise, it indicates the following platforms that would fulfill the functions of this new modernization scheme: Port Community Systems, Cargo Community Systems, Data Pipelines, Customs Clearance Systems, and Integrated Services for micro, small, and medium-sized companies (MSME) for International Trade.30 The evolution that has been registered in the digitization of processes is that it has passed to a single platform by which governments regulated B2G operations, to the constitution of a new platform or the complementation of the previous one that includes B2B operations. In this way, all international trade players are integrated into a digital environment. These new platforms would be providing the following services: data reuse and data accuracy, clearance by border authorities, trade finance, and logistics. The NTT Data Institute of Management Consulting, Inc informs that, in addition, a side-by-side overview of Japan, ASEAN, China, and South Korea on the status of single window construction shows that Japan is aiming for private-sector-led cooperation with ASEAN centered on a trade platform, while China is promoting cooperation with the ASEAN Single Window (ASW), focusing on the computerization of procedures for certification of origin as part of its One Belt, One Road strategy. China is promoting cooperation with the ASEAN Single Window (ASW) as part of its One Belt, One Road strategy, focusing on the computerization of origin verification procedures, while South Korea is focusing on demonstration experiments to expand its own infrastructure.31 As has already been stated, LAC lacks national policies that target the process of digital transformation—trade policies that are adjusted to the disciplines of digital trade and do not fully automate international trade processes. For the LAC, advancing the process of digitizing documents related to international trade has been very complex. Being essential for the automation of international trade processes, a complete and unique implementation of the foreign trade single windows has not yet been achieved in Latin America. As mentioned above, LAC have not been able to consolidate the implementation of its single electronic windows. LAC 28 Committee on Trade and Investment Asia-Pacific Economic Cooperation APEC Policy Support Unit Study on Single Window Systems' International Interoperability: Key Issues for Its Implementation August 2018, www​.apec​.org/​Publications/​2018/​08/​Study​-on​-Single​-Window​-Systems​-International​ -Interoperability accessed May 27, 2022. 29 UN/CEFACT Recommendation no 37 2019 ECE/TRADE/447, https://​unece​.org/​trade/​uncefact/​ tf​_recommendations accessed May 27, 2022. 30 ibid. 31 ibid.

Latin America  261 has not managed to consolidate the implementation of its single electronic windows, with only 5 countries managing to develop these projects out of the 14 countries surveyed. The ECLAC Regional Report 2021 reports all the 14 participants indicated that there are still relevant stakeholders not connected to the single window.32 The restrictions to advancement in the process of modernizing single windows have been due to different barriers. In this sense, the IADB–WEF report33 lists the pain points in single windows related to interoperability among the stakeholders, paperless trade, traceability of goods, documents and payment processing, and trustworthiness of data. As said by Volpe Martincus,34 these technological platforms include a regionally coordinated national logistics vision, as a way of ensuring interoperability and promoting technological integration through adherence to regional standards that allow for interoperability of the different national logistics systems, just as South Korea, Japan, and Singapore have already done. The interoperability of single windows was one of the first problems to emerge. The concept of interoperability was specified by the UN/CEFACT Recommendation No. 36 Single Window Interoperability,35 which said that interoperability is “the ability of two or more systems or components to exchange and use information across borders without additional effort on the part of the user.” An IADB report36 highlights that the single window concept is based on the efficient exchange of data and information between operators and government agencies (B2G, business-to-government) as well as between the regulatory agencies themselves (G2G, government-to-government). The IADB compendium states that to meet these objectives, it is necessary that the different systems used, both in the regulatory agencies and by the operators and the rest of the agents involved, be interoperable with each other. It goes on to state that interoperability enables communication between different computer applications running on different technology platforms, using different communication protocols. When the systems are more interoperable with each other, their integration to exchange information requires less effort and costs.37 Years after the launch of the single windows, the Port Community System platforms were slowly designed in LAC in recent years – unlike what was seen in several countries in Europe, which began adopting them back in the 1970s. Krista Lucenti and Jonas Mendes Constante38 define the Port Community System platform as an electronic platform that enables intelligent and secure exchange of information between public and private stakeholders. These authors S. Herreros (n 20) 2021. Alejandra Radl, Sandra Corcuera-Santamaría, Lorena Cano, Ziyang Fan, Jesse Lin and Kati Suominen “Windows of Opportunity: Facilitating Trade with Blockchain Technology” Jan 2020, https://​ publications​.iadb​.org/​en/​windows​-opportunity​-facilitating​-trade​-blockchain​-technology accessed May 27, 2022. 34 C.Volpe Martincus “How Does Trade Respond When Borders Are Simplified via Single-Window Systems?” January 31, 2018 https://​blogs​.iadb​.org/​integration​-trade/​en/​how​-does​-trade​-respond​-when​ -borders​-are​-simplified​-via​-one​-stop​-systems/​ accessed May 20, 2022. 35 UN/CEFACT Recommendation no 36 Single Window Interoperability 2017 ECE/TRADE/431, https://​unece​.org/​trade/​uncefact/​tf​_recommendations accessed May 27, 2022. 36 BID Academy, Module 4 Components and Strategies to Consider for the Implementation of a Single Window. The Single Windows: Tools of Trade Facilitation Course. 37 ibid. 38 Krista Lucenti and Jonas Mendes Constante “What Global Ports Did to Become More Efficient” June 11, 2019 https://​blogs​.iadb​.org/​integration​-trade/​en/​what​-global​-ports​-did​-to​-become​-more​ -efficient​-without​-spending​-too​-much​-money/​ accessed May 22, 2022. 32 33

262  Research handbook on digital trade indicate that the objective of these platforms is to optimize, manage, and automate logistics processes through a single submission of data that connects transport and logistics chains. The delay in the setting-up of the Port Community System platforms is another indicator of the lag in the region in the automation of foreign trade processes. Only Chile, Jamaica, Panama, and Brazil have, or are implementing, advanced, digitally connected port systems. The IADB is cooperating with Argentina, Peru, Suriname, and Trinidad and Tobago to support the design and deployment of Port Community System solutions. Among these few examples, there are also no high degrees of interoperability between the foreign trade single windows and the platforms that digitize port procedures. Another barrier to the advancement of digital trade in Latin America is the low level of digitization in the logistics processes of Latin American companies. The lack of incorporation of technology in their business processes does not stimulate governments to deepen the necessary reforms. This situation leads to the lack of digital platforms managed by private sector companies. In turn, there was also no development of platforms managed by the private sector that would allow digital logistics services to be provided to importers and exporters. The difficulties in developing technological projects to automate processes at the national level have led to various efforts by regional international organizations to design platforms with a sub-regional scope. At the national level, a major obstacle is that border agencies operate with isolated data. The difficulty of achieving full interoperability of all state agencies still represents an enormous challenge in the countries of the region. From the regional perspective, the national single windows are disconnected from each other. In the case of LAC countries, given the difficulty of consolidating a single window for the national sphere, the objective has changed to establishing them at the subregional level. It should be clarified that, although there are many free trade agreements that include matters such as “Paperless Trading” or “Paperless Trade Administration,” these provisions are not included in a single chapter of the free trade agreement.39 The provisions of digital trade are not only found in the sections of trade facilitation or e-commerce, but also in those of covering rules of origin, technical barriers to trade, sanitary and phytosanitary measures, financial services, as well as electronic commerce.40 This dispersion of obligations related to the digitalization of processes in free trade agreements hinders the regional implementation of said mechanisms. A good practice that LACs should consider is that of the Framework Agreement on the Facilitation of Cross Border Paperless Trade in Asia and the Pacific (CPTA).41 The objective of this agreement is to facilitate cross-border paperless trade (data exchange) among willing UN Economic and Social Commission for Asia and the Pacific (ESCAP) member states by providing a dedicated intergovernmental framework to develop legal and technical solutions. It is an agreement that complements the TFA and consists of an implementation plan that outlines a roadmap and establishes different checklists for compliance with technical requirements.

Y. Duval and K. Mengjing (n 18) 2017. ibid. 41 ESCAP “Framework Agreement on the Facilitation of Cross Border Paperless Trade in Asia and the Pacific (CPTA)” December 25, 2020 www​.unescap​.org/​kp/​cpta​#. 39 40

Latin America  263 ECLAC42 describes this process by stating that in LAC, there are several examples of work already underway on a foreign trade single window 2.0 aimed at better interconnection, services, and security, both internally, with all the entities linked to foreign trade processes, and with windows in other countries and trade blocs of interest. This same institution highlights the efforts aimed at providing interoperability to different countries of the trading blocs. They are cited as examples to promote greater interoperability of the foreign trade single windows for the countries of the Pacific Alliance (Chile, Colombia, Mexico, and Peru) and of the Southern Common Market (MERCOSUR), made up of Argentina, Brazil, Paraguay, and Uruguay, along with the Bolivarian Republic of Venezuela and the Plurinational State of Bolivia (in the process of joining). A cornerstone to achieve regional interoperability has been the creation of the Inter-American Network of International Trade single windows (RedVUCE), a regional forum for dialogue and collaboration promoted by the IADB that includes the governmental, public, and private agencies in charge of the design, development, and administration of electronic trade windows in the LAC.43 Recently, regional cooperation organizations have aided with the implementation of pilots, using blockchain technology. As technological progress develops, in terms of facilitating international trade processes, different authors analyze this phenomenon to the extent that blockchain technology began to be applied in different processes of international trade. In this sense, the WEF and WTO44 coined the term TechTrade to refer to the set of technologies and innovations that enable global trade to become more efficient, inclusive, and sustainable. As different digital technologies were developed, various projects began to be implemented to automate processes in the field of foreign trade through blockchain. Concerning the adoption of blockchain technology in Latin America, it is being implemented by the customs of the Plurinational State of Bolivia, Chile, Colombia, Costa Rica, Ecuador, Guatemala, Mexico, and Peru. This process has been possible thanks to the CADENA Project, promoted jointly by the IADB and Microsoft. CADENA basically consists of enhancing the security and efficiency of their customs-related mutual recognition agreements (MRA) and Authorized Economic Operator (AEO) programs. This project has allowed and accelerated the processing of goods in destination countries, and the established information flow supports.45 A recent report by the WCO and the WTO46 states that the CADENA that the project has contributed to strengthening the security of supply chains by ensuring that AEO information on certifications, suspensions, and cancellations executed by a customs authority is shared in real time with customs authori L. Valdés Figueroa and G. Pérez “Digitization of Trade Logistics in Landlocked Countries in South America” 2021 FAL Bulletin Nº 385 number 1 https://​repositorio​.cepal​.org/​handle/​11362/​46177 accessed May 22, 2022. 43 ibid. 44 This new concept focuses on studying the potential of artificial intelligence (AI), blockchain and distributed ledger technology (DLT), and the internet of things (IoT) to shape the global trade ecosystem: WEF and WTO “The Promise of TradeTech Policy Approaches to Harness Trade Digitalization 2022” www​.wto​.org/​english/​res​_e/​publications​_e/​tradtec​hpolicyhar​ddigit0422​_e​.htm​#:​~:​text​=​promise​%20of​ %20TradeTech​-​,Policy​%20approaches​%20to​%20harness​%20trade​%20digitalization​,and​%20creating​ %20new​%20trade​%20opportunities accessed May 22, 2022. 45 Radl (n 33) 2020. 46 WCO and WTO, “The Role of Advanced Technologies in Cross-border Trade: A Customs Perspective” 2022 www​.wto​.org/​english/​res​_e/​publications​_e/​wcotech22​_e​.htm​#:​~:​text​=​publications​ -​,The​%20role​%20of​%20advanced​%20technologies​%20in​%20cross​%2Dborder​%20trade​%3A​%20A​ ,flow​%20of​%20goods​%20across​%20borders accessed May 22, 2022. 42

264  Research handbook on digital trade ties of MRA partners, which allows them to take appropriate action. Finally, the report47 points to a set of benefits that CADENA has brought to LAC: – Point-to-point automation to exchange data through the digitization of AEO certificate status. – A secure, reliable, and traceable mechanism to exchange data on AEO certificates and to maintain an historical record of all shared information relating to each certificate. – Data integrity and access control though authentication of customs officers assigned to specific roles and authorizations. – Transparency of exchanged data. – Potential reduction of time and costs for AEOs by guaranteeing that the application of the benefits both at origins and destinations will be automatic from the moment they receive the AEO certification from customs authorities of MRA partners. Another project developed in the region is Mercosur’s bConnect. This mechanism is used by Mercosur members’ customs authorities and AEOs to exchange information while ensuring the integrity of any information shared and verifying the identities of those inputting information onto the platform. The platform makes sharing information by customs authorities faster, more efficient, and safer, and includes Argentina, the Plurinational State of Bolivia, Brazil, Paraguay, and Uruguay. Most of these pilot projects were driven by the Global Knowledge Alliance for the Development of Blockchain Ecosystem in Latin America and the Caribbean (LACChain). LACChain is a global public–private alliance supported by the IDB Lab (the innovation laboratory of the Inter-American Development Bank Group) to promote integration and economic and social development among Chile, Colombia, Mexico, and Peru by providing the infrastructure to develop interoperable blockchain applications in Latin American and the Caribbean. This alliance works in four specific areas: partnerships between public–private stakeholders, technological infrastructure, marketplace of applications, and data analytics to measure social impact.48 Beyond these advances, the projects are pilots and testing are required to obtain the greatest possible benefits from this technology. Likewise, it is not yet possible to determine that blockchain is the best technology for the automatic clearance of goods.49

5.

THE AUTOMATION OF PROCESSES AS A CATALYST FOR THE INTERNALIZATION OF SMES

Added to the problem of poor automation of foreign trade processes, the different countries of the region have a low level of internationalization of their SMEs. Both problems are intertwined, since one of the main reasons for the lack of competitiveness of SMEs is the difficulty in complying with all the documentation to carry out an export.

ibid. Radl (n 33) 2020; WEF and WTO (n 41) 2022. 49 Radl (n 33) 2020. 47 48

Latin America  265 López-González and Sorescu50 highlight the importance of small and medium-sized enterprises for the economic growth of different countries. This is also applicable to Latin America since they represent more than 99.5 percent of the companies in the region, they generate 60 percent of formal productive employment. For the OECD/CAF,51 SMEs are critical engines for growth and social inclusion at the regional level, as well as a key priority for policymakers in the region. Nevertheless, Latin American SMEs face competition from a large informal sector and a significant productivity gap with large companies.52 The lack of productivity has generated that only about 10 percent of Latin American SMEs export part of their production.53 This problem also drags on the lack of participation of SMEs in the digitization process. This deficiency means that both governments and regional organizations seek the international insertion of this type of production unit. Within a wide range of public policies, trade policy can help SMEs start operating in international trade. Given their limited human and financial resources, these firms tend to be disproportionately affected by complex documentary and procedural requirements, to the extent that these may become insurmountable obstacles to their participation in international trade.54 The UN Global Survey on Digital and Sustainable Trade Facilitation55 indicates that trade agenda that explicitly recognizes and acts on the facilitation of SMEs can be a strong force for inclusive trade. On the other hand, López-González and Sorescu56 point out that trade facilitation measures that contribute to inserting SMEs in international markets include streamlining of procedures, automation of the border process, simplification of fees, or consultations with traders. Although the adoption of trade facilitation measures can have a positive impact for SMEs to access international markets, these policies are applied by a small number of countries. Single windows can solve some of the challenges faced by SMEs in their process of internationalization. This scheme is a trade facilitation tool that promotes the rationalization and standardization of foreign trade procedures, which significantly lowers trade costs for businesses, especially SMEs, and reduces the complexity of customs formalities. The most successful experiences to provide competitiveness to SMEs are the implementation of Single Submission Portals. The Single Submission Portals generate immediate and

50 J. López González and S. Sorescu “Helping SMEs Internationalise through Trade Facilitation” OECD Trade Policy Papers, No. 229, OECD Publishing 2019 https://​doi​.org/​10​.1787/​2050e6b0​-en accessed May 31, 2022. 51 OECD/CAF “Latin America and the Caribbean 2019: Policies for Competitive SMEs in the Pacific Alliance and Participating South American Countries, SME Policy Index” OECD Publishing 2019 https://​doi​.org/​10​.1787/​d9e1e5f0​-en accessed May 22, 2022. 52 ibid. 53 ECLAC “Acerca de Microempresas y Pymes” www​.cepal​.org/​es/​temas/​pymes/​acerca​ -microempresas​-pymes accessed May 31, 2022. 54 World Trade Organization “World Trade Report 2016. Levelling the Trading Field for SMEs” Geneva www​.wto​.org/​english/​res​_e/​publications​_e/​wtr16​_e​.htm accessed May 31, 2022. 55 United Nations Regional Commissions (UNRCs)—ECA, ECE, ECLAC, ESCAP and ESCWA “UN Global Survey on Digital and Sustainable Trade Facilitation. Based on the United Nations Global Survey on Digital and Sustainable Trade Facilitation 2021” 2022 www​.unescap​.org/​kp/​2022/​untf​-survey​ -2021​-global accessed May 22, 2022. 56 López González and Sorescu (n 50) 2019.

266  Research handbook on digital trade direct benefits to SMEs. The UN/CEFACT Recommendation No 3757 lists the benefits that this automation mechanism brings for SMEs: – Single submission: SMEs just need to submit all the required information (for example, customs, tax, inspection) once and do not need to submit information to different places. – Easier clearance: SMEs would continue to send their data electronically to carry out the customs clearance of merchandise. – Better financial support: SMEs can get better financial support from banks due to these platforms—this type of platform speeds up trade finance collection security and provides better business risk control. – More efficient logistics: SMEs can get more efficient and cheaper logistics and transport services because SSPs can offer a wide range of services connecting transport and logistics chains. – Reduced business transaction costs: SMEs can interact with the standard import and export service ecosystem with lower costs and higher efficiency. This development, whose priority was to improve the internationalization of SMEs, has taken place particularly in Asian countries. LAC and regional cooperation organizations should look for good practices, mainly in countries such as Singapore, South Korea, and Japan. Asian trade platforms such as Japan, South Korea, and Singapore are characterized by their linkage with national public systems (single-window and customs systems), and in some cases, the platforms themselves are operated under national initiative.58 LAC have tried to advance in modernizing and deepening their foreign trade single windows, but, in general terms, the results have not yet been achieved. J.D. Thorrens59 states that the cost reduction has been evident using the single foreign trade window for the different foreign trade agents in the region. However, much work remains to be done to take advantage offered by this tool to support the internationalization process of SMEs, although recently some countries have already taken the first steps in this direction. Most single windows in the region offer services mainly related to paperwork and foreign trade formalities, which is consistent with the conceptual definition of this instrument. Although great returns have been achieved with the B2G approach (red tape), a profound change in the conception of the single window is required to make it a fundamental piece for the internationalization of SMEs. In addition to the automation of the processes carried out by the international trade window, there should be another one that allows the management of business services for the SMEs.60 Herreros61 evaluates the different measures that have facilitated trade aimed at SMEs. The region still has much room for improvement in terms of easing documentary and procedural UN/CEFACT (n 29) 2019. NTT Data Institute of Management Consulting, Inc. “International Economic Research Project for the Establishment of an Integrated Domestic and International Economic Growth Strategy (International Economic Research Project on Japan’s Measures to Enhance Trade Facilitation (including the Use of FTAs))” Surbey Report, March 19, 2021. 59 J.D. Thorrens “La ventanilla única de comercio exterior: ¿cuánto aporta a la internacionalización de las pequeñas y medianas empresas?” Documentos de Proyectos (LC/TS.2019/110), Santiago, Comisión Económica para América Latina y el Caribe (CEPAL), 2020 https://​repositorio​.cepal​.org/​ handle/​11362/​45395 accessed May 22, 2022. 60 ibid. 61 S. Herreros (n 20) 2021. 57 58

Latin America  267 barriers to SME internationalization. The 14 countries studied in Latin America have shown a low level of implementation of trade facilitation measures for SMEs. Only 50 percent have established regulations, but there is a lot of disparity between the promotion of SMEs among the countries surveyed.62 If it is the case that single trading platforms contribute to the economic development of SMEs, the ECLAC survey provides information related to SME access to trade-related information. The region performs best in facilitating SME access to trade-related information—arguably the least demanding measure in this category—with an 86 percent average implementation rate. All participating countries reported at least partial implementation of this measure.63 The second area where most progress has been made relates to easing SME access to electronic single windows (60 percent). Next comes ensuring adequate SME participation in National Trade Facilitation Committees (43 percent) and other measures to reduce the cost for SMEs of complying with trade procedures (40 percent). The least implemented measure concerns facilitating SME participation in AEO schemes (19 percent).64 The reality indicates that this low level of adoption of this program is due to the criteria to obtain AEO certification are the same regardless of a firm’s size. Across all five measures, South America scores above the regional average (58 percent), while the three Caribbean participants achieve a 49 percent implementation rate, almost identical to the regional average, and Central America and Mexico comes last at 35 percent.65

6.

CONCLUDING REMARKS

LAC has shown delays in the digitization processes of its economies. In turn, the countries of the region have not yet been able to articulate their national development policies, digital agendas, and trade policies to take advantage of the data-driven economy. Considering the concept of digital trade, it has not been easy to advance to Latin American regulators either. A fundamental milestone in digital trade policies is the automation of international trade processes. What is considered facilitation of digital trade has as its basic structure the implementation of a single window for international trade. One of the successes of this mechanism is to achieve the interoperability of government agencies and foreign trade agents. Asian countries (especially South Korea, Singapore, and Japan), based on the development of legislation that promoted the digitization of their processes and the internationalization of SMEs, launched Single Submission Portals. These portals allow B2B operations to be carried out such as contracting for transport, logistics, and financial services, and make regulatory processes easier through B2G information exchange. SMEs are directly benefited since they can obtain information through unique government or private portals that interconnect information related to international trade. In Latin America, progress in the implementation of international trade single windows has presented some drawbacks in different countries of the region. The modernization of single windows is seeking to strengthen them through deepened interoperability with their internal regulators and those of the different countries. The countries with the greatest progress have ibid. ibid. 64 ibid. 65 ibid. 62 63

268  Research handbook on digital trade managed to electronically exchange documents such as certificates of origin and sanitary and phytosanitary certificates, and to allow the cross-border operation of AEOs. The most advanced countries in the region are conducting some pilots to implement blockchain technology to digitize international trade procedures. Likewise, they are also advancing in integrating the Port Community Systems to their foreign trade single windows. Although there are several regional free trade agreements that include provisions for “Paperless Trading” or “Paperless Trade Administration,” the dispersion of the obligations assumed in the matter prevents the countries from coordinating among themselves to advance in these matters. In this sense, it would be advisable for the LACs to sign a supplementary agreement such as the Framework Agreement on the Facilitation of Cross Border Paperless Trade in Asia and the Pacific (CPTA). This type of agreement facilitates the implementation of the different technical aspects between the countries and the cooperating entities. The great effort that a Single Submission Portal will require should be an incentive to internationalize their SMEs. Countries like South Korea that have designed projects so that their SMEs can interconnect electronically with their official agencies and the various private actors in international trade should be imitated. In Latin America, in addition to a modest advance in e-trade facilitation, a technological leap is required that points to the realization of a single foreign trade portal. This task must be interconnected with the promotion of the internalization of SMEs. This great effort in the acquisition of digital, human, and financial infrastructure must aim at improving the competitiveness of Latin American SMEs. LAC should adopt some policies to achieve the mentioned results. In the first place, at the national level, an institutional engineering must be carried out that allows the articulation of economic development, digital transformation, and digital trade. The promotion of public– private articulation is essential to achieve the stated objectives. Relative progress has also been made in the region in the digitization of logistics processes in private companies, so these collaborative policies will lead to the incorporation of technology into the different business models. At the regional level, the signing of agreements to implement the facilitation of cross-border paperless measures would be important. Following the Asian experience, it would be important to consolidate and form a regional team of experts in e-trade facilitation issues. The application of international standards in the digitization of processes will make it easier for trade platforms in the region to interoperate with those outside the region. Due to the need for financing and technical advice, projects sponsored by regional cooperating entities such as the IADB and ECLAC should be fine-tuned to become more efficient in digitizing international trade processes.

PART V DIGITAL TRADE LEGAL AND POLICY CHALLENGES: FINANCE, INVESTMENT AND TAXES

17. The compatibility of digital services taxes with World Trade Organization law David Collins

1. INTRODUCTION The advancement of technology has led to widespread digitalization in the global economy. This trend poses significant challenges to regulatory regimes, especially those which have an international element, one of the most important of which is corporate tax. The determination of the boundaries of a state’s authority to impose tax on foreign enterprises has become an increasingly contentious battlefield between governments and multinationals.1 Efforts to establish more fair taxation regimes on the providers of digital services through digital services taxes (DSTs) raise a host of issues, one of which is their compatibility with international trade law – meaning the imposing country’s obligations as a member of the World Trade Organization (WTO). This chapter will consider the WTO implications of DSTs, assessing whether the imposition of a DST could constitute an illegitimate trade barrier by arbitrarily discriminating against a service supplier on the basis of its nationality, violating the Most Favoured Nation (MFN) or national treatment provisions of WTO law.2 The WTO instrument which is most relevant to a consideration of the DST is the General Agreement on Trade in Services (GATS), although there may also be a claim under the WTO’s moratorium on customs duties on electronic transmissions. Other potentially unlawful elements of DSTs, including their compatibility with international tax principles, will not be discussed, nor will claims of discrimination which might be brought under preferential trade agreements. The potential for DSTs to impact foreign investment will also not be examined. The chapter will begin by illustrating some of the common features of DSTs, highlighting their controversy in terms of their trade impacts.

2.

DSTS: CONTROVERSY AND MAIN FEATURES

Normally a corporation pays tax in jurisdictions in which it has a physical presence. But for the large tech companies, a physical presence is not necessary to do business in a given jurisdiction. The lack of physical presence of companies which provide digital services within countries where they do business makes it nearly impossible for tax authorities to collect taxes on profits made by multinational technology firms serving customers in their jurisdiction. The growth of the digital economy has accordingly raised fundamental questions about how residence is determined as well as the jurisdiction in which value its creation occurs for tax



1 2

K Freida, Cybertaxation: The Taxation of E-Commerce (Arthur Andersen, 2000) at 263. As found in GATS II and GATS XVII in the case of services.

270

The compatibility of digital services taxes with WTO law  271 purposes.3 Added to this is the common perception that certain large companies, especially in the tech sector, do not pay their fair share of tax – a concern compounded by the global financial crisis of 2008–9 and the ensuing fiscal austerity policies imposed by many countries which acted as a catalyst for coordinated action against large multinationals seen as not paying their share.4 The Organization for Economic Cooperation and Development (OECD) has engaged in work on a framework for addressing base erosion and profit shifting (BEPS), which refers to tax-planning strategies used by multinational enterprises that exploit gaps and mismatches in tax rules to avoid paying tax, paying specific regard to issues arising from digitalization.5 Yet a global approach to tax losses arising in the digital economy has yet to be actualized. Concerned about erosion of their tax base and pessimistic about the possibility of reaching a global consensus on the issue via a multilateral treaty, approximately 30 countries have proposed or implemented DSTs on some of the world’s largest technology companies. These taxes are designed to redeem the value of ‘user-created’ data allegedly exploited by those digital service providers. Among the countries pursuing DSTs are France,6 the UK7 and, more recently, Canada.8 Unilateral DSTs have been advocated for developing countries which rely heavily on corporate income tax and consequently suffer from tax evasion.9 The US does not appear to be considering a DST, likely because it is home to many of the world’s largest tech companies. The main targets of the DSTs are the technology behemoths such as Google, Amazon and Facebook. It may come as no surprise that the US strongly objects to DSTs, as it is the home jurisdiction of these companies. In response to French, British and other DSTs, the Trump administration proposed retaliatory tariffs under Section 301 of the US Trade Act of 1974. Under WTO law, retaliation can be pursued against other products and sectors from the one on which the alleged illegal measure is imposed.10 The US imposed tariffs of up to 100 per cent on French luxury goods exports such as handbags and soaps,11 notwithstanding the fact that the DSTs such as that of France are imposed only on digital services. In June 2020, the US Trade Representative initiated an investigation of the UK DST under the US Trade Act. More recently, the threat of a trade war over DSTs has been held in abeyance because of the negotiation of a global tax treaty, signed in October 2021 under the auspices of the OECD. The US agreed to conditionally withdraw the threatened Section 301 tariffs in exchange for

A de Jonge, ‘The Evolving Nature of the Transnational Corporation in the 21st Century’ in A de Jonge and R Tomasic eds, Research Handbook on Transnational Corporations (Edward Elgar Publishing, 2017) at 33. 4 M Geist, ‘The Law Bytes Podcast, Episode 115: Reuven Avi-Yonah on the Past, Present and Future of Digital Services Taxes’ (31 Jan 2022) www​.michaelgeist​.ca/​2022/​01/​law​-bytes​-podcast​ -episode​-115/​. 5 Programme of Work to Develop a Consensus Solution to the Tax Challenges Arising from the Digitalisation of the Economy, OECD (31 May 2019). 6 Bill no 2019-759 of 24 July 2019 on the creation of a tax on digital services, NOR: ECOE1902865L. 7 Finance (Digital Services Tax) Bill 2020, HC Bill [114]. 8 Notice of Ways and Means Motion to introduce an Act to Implement a Digital Services Tax (December 2021). 9 M Magwape, ‘Unilateral Digital Services Tax in Africa: Legislative Challenges and Opportunities’ Intertax 50:5 (May 2022). 10 Dispute Settlement Understanding, Art 22.3. 11 A Williams, ‘US Threatens $1.3bn Worth of French Goods with 25% Tariff’ The Financial Times (10 July 2020). 3

272  Research handbook on digital trade a two-year moratorium, expiring in December 2023, on new DSTs. The OECD agreement calls for tax imposing countries to attribute, for tax purposes, a portion of earnings of their large corporations to foreign countries where goods or services are consumed rather than the traditional practice of attributing all earnings to the home country. With regard to DSTs themselves, the OECD agreement provides ‘[n]o newly enacted Digital Services Taxes or other relevant similar measures will be imposed on any company from 8 October 2021 and until the earlier of 31 December 2023 or the coming into force of the [Multilateral Convention which embodies the agreement]’.12 DSTs, including those proposed by France, the UK and Canada, have similar features. DSTs only target companies that provide digital services whose revenues largely derive from user data generated within the territory of imposing countries.13 The taxes depart from the permanent establishment rule which limits the source countries’ tax jurisdiction, only allowing them to collect taxes on profits attributable to the physical presence/permanent establishment of companies within their own territory.14 Instead, DSTs are imposed on digital companies at the group level and collected on their worldwide revenues as long as they are ‘generated’ within the imposing countries’ territory.15 As such, they are ‘destination-based taxes’ – levied where services are consumed. DSTs are applied at a flat rate on the gross revenue of affected companies. The amount of the tax tends to be small: the UK, French and Canadian proposals are each at 3 per cent.16 DSTs apply to any kind of digital service such as advertising, streaming audio and movies. Controversially, DSTs apply only to the largest companies, typically those with over US $750 million revenue. For example, the UK DST applies to companies with £500 million in annual revenue. Approximately 30 companies are expected to be caught by the DSTs,17 most of which, because of the size threshold, are US-based. Many of the large technology companies have stated that they will simply transfer the tax on to consumers,18 precluding the element of fairness which tends to be associated with these, and other, taxes. This aspect of DSTs tends not to be disclosed by the politicians who have advocated for them.19 A DST could have an impact on a wide range of services delivered over the internet. Those that could be affected, either directly or indirectly, include online advertising, computer and related services, travel agencies, news and press agency services, betting exchanges, suppliers of video and audio streaming (where funded through advertising), non-regulated financial intermediaries, peer-to-peer marketplaces including crowdfunding, and distribution. Educational and professional services which are delivered over the internet could also be

12 GC Hufbauer and M Hogan, ‘Canada’s Digital Services Tax Threatens Global Effort to Curb Tax Havens’ Pearson Institute for International Economics (15 December 2021). 13 See e.g. French DST, above n 6, art. 299-II.1; the UK DST, above n 7, cl. 42. 14 See e.g. OECD Model Tax Convention on Income and on Capital 2017, art. 7 (21 November 2017). 15 French DST, above n 6, art. 299 III; UK DST, above n 7, cl. 45. 16 French DST, above n 6, art. 299 I.A; UK DST, above n 7 cl. 45. 17 United States Trade Representative, ‘Section 301 Investigation Report on the United Kingdom’s Digital Services Tax’ (13 January 2021) at 12. 18 A Barker, ‘Google to Pass Cost of Digital Services Taxes on to Advertisers’ Financial Times (1 September 2020). 19 Geist, above n 4.

The compatibility of digital services taxes with WTO law  273 impacted due to the aspect of some DSTs that apply to intermediaries.20 It is difficult to determine whether a business falls under the DSTs’ scope as business models are rapidly evolving because of their use of dynamic technology. The lack of predictability with regards to tax exposure is itself a burden for digital services companies.21

3.

TAX AS A TRADE BARRIER UNDER WTO RULES

WTO law does not prohibit a tax, customs duty or tariff. The Appellate Body has specifically stated that tariffs are legitimate instruments to accomplish certain trade policy or other objectives such as to generate fiscal revenue and that they are not ‘inherently discriminatory’.22 Of the various taxes that have been examined by WTO tribunals, most have been primarily concerned with those applied to goods rather than services, although some of the principles may be relevant for an assessment of the legality of DSTs. The question often arises before WTO panels and the Appellate Body whether foreign goods were subject to less favourable competitive conditions than domestic products, for example by facing a heavier tax burden or a more burdensome administrative method of tax collection. The tax disputed measures tend to be ‘origin neutral’ but their application can violate WTO rules prohibiting de facto discrimination, causing adverse effects such as lost sales.23 In assessing the compatibility of a DST with WTO law it must be understood that the purpose of the WTO in relation to tax is to establish the limits of tax as it is applied to goods and services which are traded internationally. WTO law marks a sharp distinction between goods and services, arguably artificially so, particularly in an era of services bundled into goods and vice versa. WTO provisions also distinguish between customs duties, discriminatory taxes on imports, export subsidies (including tax relief on exports), and taxes on the supply of cross-border services.24 Commentators urge that comparisons with existing kinds of tax are helpful in assessing the legality of new taxes such as DSTs, but only under very specific circumstances. The use of analogy can help policymakers anticipate the incompatibility of a new tax with WTO law when the specific characteristics of the new tax resemble features of taxes that have been found compatible (or not) with WTO law by panels or the Appellate Body.25 WTO members are less likely to challenge tax measures that are similar to taxes that are part of their domestic tax system. Therefore, new taxes that follow the design features of existing taxes benefit from what has been described as ‘political protection’ under WTO law. If new destination-based taxes, such as DSTs, simply replicated existing taxes, such as value added taxes, there would probably be no reason to consider them at risk of violating WTO

PWC, ‘A White Paper Analysing the EU’s 2018 Proposed Digital Services Tax (Interim Measure) under WTO Law’ (undated) at 7 https://​thesuite​.pwc​.com/​media/​10060/​dst​-under​-wto​-law​.pdf (April 2022). 21 USTR, above n 17 at 11. 22 India–Additional Duties on Imports from the United States, Appellate Body Report, WT/DS360/ AB/R (17 November 2008) at [159]. 23 E.g. Japan–Taxes on Alcoholic Beverages, Appellate Body Report, WT/DS8/AB/R (1 November 1996). 24 See section 2.2.2. 25 A Pirlot, ‘Don’t Blame It on WTO Law: An Analysis of the Alleged WTO Law Incompatibility of Destination-Based Taxes’ 23:1 Florida Tax Review 432 (2019) at 442. 20

274  Research handbook on digital trade law.26 However, the fact that an existing tax has not been challenged in the past does not necessarily mean that it is fully WTO-compliant. There are two WTO legal instruments relevant to an analysis of DSTs. First is the WTO moratorium on customs duties on electronic transmissions; second, and more importantly, is the GATS. Each will be explored in turn. 3.1

DST Compliance with the WTO Moratorium on Customs Duties on Electronic Transmissions

In 1998, as the internet began to mature into a driving force behind global commerce, WTO members agreed not to impose customs duties on electronic transmissions as part of the Global Declaration on Electronic Commerce.27 This moratorium has been renewed several times and remains in effect, having been renewed at the WTO’s 12th Ministerial Conference in 2022. Customs duties do not generally apply to services because they do not cross borders in the way that goods do. Still, it is not unreasonable that a DST could be characterized as a customs duty on electronic commerce.28 But the WTO moratorium is limited specifically to formal ‘customs duties’ whereas DSTs are internal taxes – a crucial difference. Customs duties are indirect taxes levied at the border and paid by the importer while DSTs are probably best viewed as direct taxes paid by the supplier of the service at the place of consumption, likely passed on to consumers in the form of higher prices (although not as a formal tax burden). Moreover, unlike most taxes, customs duties generally do not have a purpose of raising money, but are designed to regulate the flow of trade,29 suggesting that the moratorium is most likely inapplicable to DSTs which are aimed at revenue generation. Commentators point out that the moratorium on customs duties on electronic transmissions therefore does not apply to direct taxes like the DST.30 Others contend that a DST would constitute a tariff which would conflict with the moratorium on customs duties.31 The characterization as a tariff seems somewhat strained since the DST will not be paid at the border, as is the case of a tariff. The applicability of the moratorium to the DST therefore appears to rest on how the tax is categorized. Regardless of how the DST is characterized, the customs duty on electronic transmission moratorium is not subject to the WTO’s dispute settlement system, so it cannot be used as the basis for a legal claim under WTO law. This suggests that the DST’s alleged transgression of the moratorium could not be used to compel imposing countries to rescind its DST.32 Until it I Grinberg, ‘A Destination-Based Cash Flow Tax Can Be Structured to Comply with World Trade Organization Rules’ 70 National Tax Journal 803 (2017) at 811. 27 Declaration on Global Electronic Commerce, WTO Doc WT/MIN(98)/DEC/2 (25 May 1998, adopted on 20 May 1998). 28 A Mitchell, T Voon and J Hepburn, ‘Taxing Tech: Risks of An Australian Digital Services Tax under International Economic Law’ 20:1 Melbourne Journal of International Law (2019) at 95. 29 Frieda, above n 1 at 389. Tariffs play a more important role in revenue generation for developing countries. 30 I Willemyns, Digital Services in International Trade Law (Cambridge University Press, 2021) at 280. 31 GC Hufbauer and Z Lu, ‘UK Money Grab: Proposed Digital Tax’ Pearson Institute for International Economics (1 November 2018). 32 C Forsgren, S Song and D Horváth, ‘Digital Services Taxes: Do They Comply with International Tax, Trade, and EU Law?’ Tax Foundation (29 May 2020) https://​files​.taxfoundation​.org/​20200522152239/​ Digital​-Services​-Taxes​-Do​-They​-Comply​-with​-International​-Tax​-Trade​-and​-EU​-Law​.pdf (April 2022). 26

The compatibility of digital services taxes with WTO law  275 solidifies into a binding treaty, the moratorium is best described as a political agreement which does not impose legal obligations on WTO members and as such should not affect the DST.33 Some have suggested that while the moratorium itself might not ground a claim against a DST, as a ministerial statement, the WTO’s Declaration on Global Commerce might be relevant for interpreting GATS with regard to establishing the ‘object and purpose’ of the latter agreement under art 31(1) of the Vienna Convention on the Law of Treaties (VCLT) or even as ‘subsequent agreement’ or ‘subsequent practice’ under art 31(3).34 This could render an interpretation of the GATS more supportive of a challenge to a DST. 3.2

DST Compatibility with the General Agreement on Trade in Services (GATS)

3.2.1 Taxation and non-discrimination based on national origin The most important agreement for the purposes of assessing the WTO law compatibility of a DST is GATS. WTO caselaw has established that GATS is relevant to tax issues.35 The GATT Secretariat noted the applicability of GATS to tax measures that affect services suppliers, particularly the National Treatment obligation (art XVII) and the MFN obligation (art II).36 GATS does not prohibit any type of differential treatment between service suppliers but only those differentiations that affect the supply of services. A DST might have a discriminatory effect on cross-border services relative to those produced at home. Indeed, the US government criticized DSTs, including the UK DST in particular, for being discriminatory.37 GATS requires WTO members to adhere to the two aforementioned non-discrimination principles, National Treatment and MFN. In that sense GATS limits the adoption of tax measures that impact trade in services with other WTO members. The National Treatment obligation of GATS explicitly requires members to ‘accord to services and service suppliers of any other member, in respect of all measures affecting the supply of services, treatment no less favourable than that it accords to its own like services and service suppliers’.38 In contrast to the GATT, GATS National Treatment applies only to services that have been included by WTO members in their Schedule of Specific Commitments under the conditions that these commitments are not subject to additional limitations. WTO members can limit the scope of their commitments by means of horizontal limitations (meaning limitations that apply to all sectors listed in the schedule) or sectoral limitations (those specific to certain sectors). 3.2.2 Assessing the scope of a member’s National Treatment commitments For the DST to violate another member’s GATS National Treatment commitments, it would first have to be shown that imposing country accepted a National Treatment commitment in respect of the services impacted by the DST and with regard to the mode of supply. A WTO member could adopt a horizontal limitation for National Treatment in order to exclude taxes Willemyns, above n 30 at 279–80. Mitchell, Voon and Hepburn above n 28 at 95. 35 See generally Argentina–Measures Relating to Trade in Goods and Services, Appellate Body Report, WT/DS/453/AB/R (9 May 2015). 36 GATT Multilateral Trade Negotiations Uruguay Round, Group of Negotiations on Services, ‘The Applicability of GATS to Tax Measures’ Note by the Secretariat, MTN. GNS/W/210 (1 December 1993). 37 USTR, above n 17. 38 Art XVII. 33 34

276  Research handbook on digital trade from its GATS commitments. All tax measures, including DSTs, would thereafter fall outside of the scope of GATS obligations. This would remove the risk that any tax measures would be captured by GATS.39 WTO members often resist claims for breach of GATS National Treatment on the grounds that they never intended to bind itself in that sector or that mode.40 Commitments made 20+ years ago may no longer correspond to services which exist today. Members may have listed a National Treatment obligation in their schedule of GATS commitments for a service or mode of supply mode that technology has transformed, resulting in the commitment being broader than that member might have been anticipated during the original GATS negotiations. For example, China listed a National Treatment commitment for the distribution of sound recordings through electronic means. China’s GATS schedule entry referred to ‘sound recording distribution services’. However, China claimed that it only covered physical distribution. The Appellate Body disagreed, finding that this covered both physical and electronic distribution.41 This ruling is also consistent with the generally agreed view that the GATS is a technologically neutral agreement which does not distinguish between the various means through which a service can be supplied.42 The result is, however, that a member may be held to a commitment which it did not intend to make. It is not only WTO law that has lagged behind technology – crucially, tax law has also struggled to keep up to date with the transformation of the global economy. The telecommunications industry is perhaps the best example of one where domestic tax legislation has failed to adapt to technological advances.43 Clearly, when assessing whether a member has made a GATS commitment which may be triggered by a DST, it must be ascertained precisely what services the DST is intended to cover. This will not be an easy task, again because digital trade is a rapidly evolving field where technological changes often outpace legal terminology.44 A WTO panel would, following the VCLT, have to construe this concept by reference to the ordinary understanding of the term.45 Taking the UK’s DST as an example, the tax only applies to ‘online services’, although this phrase is not defined in the legislation. The revised guidance of the UK tax authority, Her Majesty’s Revenue and Customs (HMRC), contains some commentary on whether activities are sufficiently separate to be considered a ‘service’ in their own right, or if they are ancillary to a wider business function. There are three sub-categories of online services specified in the UK DST, the first of which is social media services. The definition of ‘social media service’ requires a main purpose of promoting user interactions and sharing of user-generated content, which must be ‘a significant feature of the service’. This condition means that businesses will not be covered by the DST simply because users share content on their platform even if, for example, that content is not significant because it merely relates to a comment on content

39 JE Farrell, The Interface of International Trade Law and Taxation (IBRD Doctoral Series, 2013) at 189. 40 E.g. United States–Measures Affecting the Cross-Border Supply of Gambling and Betting Services, WT/DS285/AB (5 April 2005). 41 China–Publications and Audiovisual Products, Appellate Body Report, WT/DS363/AB/R (19 January 2010). 42 D Curiak, ‘The Challenge of Updating Institutions for Digital Trade’, CIGI (16 July 2021). 43 Frieda, above n 1 at 402. 44 Willemyns, above n 30 at 95. 45 Article 31.1.

The compatibility of digital services taxes with WTO law  277 provided by the business.46 For the second category, online games, the HMRC’s guidance notes that a case-by-case approach will be needed to determine whether user-generated content is a significant part of the game. Any in-game messaging and user-generated items would be considered user-generated content, although playing the game itself will not.47 The UK DST also covers ‘internet search engines’, which is similarly undefined. The legislation excludes search engines which search a single (or closely related) website. The search engines that are covered by the DST are those whose core business involves the operation of a search engine. HMRC guidance notes that these engines will in principle search the whole of the internet. Further, where a website has a ‘search box’ that uses third-party technology to display results from external websites, the third-party technology provider typically would be considered performing an in-scope activity, rather than the website owner. The other kind of online service which falls under the UK’s DST is ‘online marketplaces’ but only where the marketplace has a main purpose to facilitate the sale or hire of goods or services offered by users, and the marketplace enables users to sell or advertise. The use of the ‘main purpose’ test here may be controversial, given the difficulty that there has been historically in the application of this test in WTO law.48 HMRC’s guidance on the definition of ‘online marketplace’ sets out a list of factors that it considers to be relevant in determining whether there is an online marketplace. These include competition between sellers, the existence of a recognizable place or portal and features that allow customers to search for products or services.49 As noted earlier, sectors that could be impacted (either directly or indirectly) by the DST therefore include advertising services, computer and services, travel agencies, news services, betting exchanges, suppliers of video and audio streaming services (where funded through advertising), non-regulated financial intermediaries, peer-to-peer marketplaces including crowdfunding and distribution services. Educational and professional services could also be affected because of the feature of the DST that applies to intermediaries. Returning to the scope of National Treatment commitments in these sectors, the appropriate mode of supply (cross-border, consumption abroad, commercial presence and movement of natural persons) must also be identified to ascertain whether commitments have been made by members imposing the DST with regards to the relevant digital service. Evaluating the applicability of different modes is further relevant where a member has scheduled ‘unbound’50 or a limitation in respect of Mode 1 supply of a service but accepted full commitments respect of Mode 2 of that same service (‘none’ in the schedule, meaning full commitment undertaken). A DST could conceivably have an impact on all four modes of supply, although the effect on certain modes may be more direct than others.51 DST may impinge many different types of trade in services spread across modes as enabled through connections established by multiple interfaces and subsequently supplied through a variety of modes. To appreciate the application of a DST to services commitments it is necessary to understand the intangible supply chain or the ‘footprint’ of the supply,52 which could be much broader than is appreciated. It would seem 46 HMRC, ‘Internal Services Tax Manual’ (19 March 2020, updated 14 June 2021) www​.gov​.uk/​ hmrc​-internal​-manuals/​digital​-services​-tax (April 2022). 47 Ibid. 48 Ibid. 49 Ibid. 50 No commitments, therefore free to impose restrictions. 51 PWC, above n 20 at 9. 52 PWC, ibid at 8–9.

278  Research handbook on digital trade that basic commitments in relation to various traditional kinds of services could be construed to engage GATS disciplines in those areas. 3.2.3 National Treatment and ‘likeness’ National Treatment requires that the imposing country not discriminate against foreign services or service suppliers in favour of ‘like’ domestic services or services suppliers.53 The WTO Appellate Body has indicated that the concept of likeness under GATS with respect to non-discrimination is primarily ‘concerned with the competitive relationship of services and service suppliers’.54 This involves a ‘holistic analysis’, considering both services and service suppliers together.55 Consequently, a key consideration in evaluating the likeness of digital services such as advertising is the extent to which the advertising services and service suppliers actually compete for customers. The more competition to be found to exist between digital and non-digital advertising services, the more probable it is that these types of services would be regarded as ‘like’.56 The likeness of digital advertising and conventional advertising is reasonably clear, as both tend to compete for the same customers. The likeness of digital interfaces is less straightforward. For example, while a small retail business that sells its products on Amazon’s platform would be covered by a DST, Amazon selling its own products through the platform would not be. But these services are obviously in a competitive relationship and could therefore be considered ‘like’. Similarly, while Uber would fall under the coverage of a DST, a taxi company that operates an app would not because the latter company is selling its own service, which excludes it from the scope of a DST.57 A close consideration of each situation will therefore be required to assess the application of the non-discrimination rule. Factors such as the technological equipment, size and scale of a business such as number of employees and types of assets could be relevant in determining supplier likeness. GATS does provide for supplier likeness to be taken into account; however, this is not for the purpose of distinguishing suppliers on the basis of size but rather a recognition that the quality of a service and nature of a supplier are inextricably linked. Whether such characteristics of services suppliers could inform an assessment of likeness has not been clarified by the Appellate Body. In particular, it is not clear how criteria such as size would be relevant if the services compete in the same given market. Distinctions based on size are problematic because this could make it practically impossible to find ‘like’ suppliers, undermining the effectiveness of National Treatment in preventing discrimination. There is a danger that size or other such factors become a proxy that could target digital companies in order to distinguish them from non-digital ones.58 Assessing likeness based on business model appears to jeopardize the principle of technological neutrality – meaning that regulations should not favour one technology for delivering a service over another one. The concept of technological neutrality has reasonably strong support across WTO membership,

General Agreement on Trade in Services, art. XVII (15 April 1994), Marrakesh Agreement Establishing the World Trade Organization, Annex 1B, 1869 U.N.T.S. 183, (1994). 54 Argentina–Financial Services, above n 35 at [6.25]. 55 Ibid. 56 Mitchell, Voon and Hepburn, above n 28 at 100. 57 Willemyns, above n 30 at 278. 58 PWC, above n 20. 53

The compatibility of digital services taxes with WTO law  279 although it remains under active discussion.59 A determination by the WTO Panel or Appellate Body that suppliers of services supplied digitally are not ‘like’ suppliers of like services that are supplied through more ‘traditional’ means, such as in-person, would have significant implications, even more so because almost all types of commercial activity are becoming digitized. Digital delivery of services can complement traditional ones, with education being a good example. Indeed, the same supplier could offer services digitally that compete with ‘like’ services that it supplies on an in-person basis.60 Monetary thresholds below which service suppliers are exempt from DSTs enhance the risk of a breach of an imposing countries’ National Treatment obligation. This is especially so because the ‘likeness’ comparison would therefore be between digital service suppliers below the threshold and digital service suppliers above the threshold (rather than between digital and non-digital services). The WTO member imposing the DST could conceivably argue that the service suppliers below the chosen threshold are not like the service suppliers above the threshold,61 which effectively would be a distinction based on size. Under National Treatment, a complainant country must show that the only meaningful difference between its digital service suppliers, who must pay the DST, and suppliers of the country imposing the DST, such as the UK, of the same covered digital services, who are not subject to the DST is national origin. The complainant country must then demonstrate that the DST’s facially neutral revenue thresholds which capture that country’s suppliers but not UK suppliers are merely a proxy for national origin. The complainant’s argument would be that the DST is designed to capture companies with a business model typical of US digital service suppliers but not UK service suppliers – de facto discrimination. Taking the US as the most likely example of a complainant given that it is home to most of the big tech companies, it must also prove that US digital service suppliers subject to the DST and their excluded UK competitors cannot be distinguished from one another on the grounds that covered companies benefit from so-called user value creation. The US could achieve this by showing that either US digital service providers do not, in fact, derive meaningful value from UK users or it could show that UK service providers not captured by the DST also benefit from user value creation. The USTR has argued that the theory of ‘user created value’ is unfounded, in part because it is unclear why firms of a certain size benefit from this value when smaller firms do not.62 Theories of network effects suggest otherwise – networks of users interacting with each other, such as Twitter, or dating sites such as Tinder only possess value when there is a critical mass of users. ‘The more users a communication network has, the more value it offers to every user.’63 On the other hand, some hold that the value of ‘peer production networks’ are often over-estimated and that such services are unlikely to compete with commercial, professionally produced sources.64 That the intangible value of data may not easily be reflected in profits nor recorded on a balance sheet, rendering its applicability to

Curiak, above n 21. PWC, above n 20. 61 Mitchell, Voon and Hepburn, above n 28 at 101. 62 USTR, above n 17 at 18. 63 J Montero and M Finger, The Rise of the New Network Industries Regulating Digital Platforms (Routledge, 2021) at 65. 64 A Guadamuz, Networks, Complexity and Internet Regulation (Edward Elgar Publishing, 2011) at 143. 59 60

280  Research handbook on digital trade various kinds of legislation, such as taxation, highly complicated.65 The greater the relevance of user-created value, the more likely that ‘likeness’ will not be found, precluding a successful claim of National Treatment. 3.2.4 DST affecting trade in services In order to demonstrate that DST is discriminatory, the complainant must also demonstrate that the tax ‘affects’ trade in services. The concept of ‘affect’ is broadly understood in WTO law. A measure ‘affects’ trade in services where it changes the conditions of competition in the supply of a service – a concept wider in scope than merely regulating the service. Having ‘an effect on’ trade in services is sufficient. Indirect effect is enough – even a measure designed to regulate goods can be found to have an indirect effect on services.66 The DST must be found to modify the conditions of competition of the foreign company on which the tax is imposed with regards to a ‘like’ domestic service or service supplier of the imposing country. Regarding the adverse impact of DST, firms to which the tax applies will not only pay higher taxes and thereby suffer a weakening of their competitive position, but will also need to fulfil additional compliance requirements for tax collection. These firms will face administrative difficulties in obtaining information on revenue generated from taxable activities. The fact that the burden of the DST will be passed on to consumers, as some companies such as Google have conceded, will not offset these costs. The USTR argued that DSTs have the effect of shifting advertising spending away from larger US companies with revenues that exceed the thresholds, to domestic companies with digital advertising revenues that do not meet the thresholds. In that way, DSTs unfairly advantage smaller domestic companies against leading US companies.67 WTO jurisprudence has found less favourable treatment when the design leads disproportionately capturing foreign products under higher tax rates.68 Some DST proposals, such as that of France, may have been designed in this way.69 Unlike the GATT National Treatment provision which mentions the use of measures ‘so as to afford protection’, the GATS National Treatment provision does not make any such reference. Regulatory intent to afford protection to domestic suppliers is not required (nor is it relevant for the purpose of determining ‘likeness’).70 The intent behind the measure could, however, be relevant for fitting the DST under one of the GATS’ general exceptions. 3.2.5 The DST as a breach of Most Favoured Nation (MFN) There may be potential to bring an MFN claim against a DST. Recall that MFN requires that the country imposing the tax not to discriminate between the services and service suppliers of one foreign country in favour of ‘like’ services and service suppliers of another foreign country.71 For example, the US could claim that a Canadian digital services company receives more favourable treatment under the UK DST than ‘like’ US digital services companies. The

67 68 69 70 71 65 66

Magwape, above n 9 at 452. China–Publications and Audiovisual Products, above n 41. USTR, above n 17 at 17. See e.g. Japan–Alcoholic Beverages, above n 23 (which concerned goods rather than services). PWC, above n 20 (referring to Article 12 of the French DST). Argentina–Financial Services, above n 35. GATS, art. II.

The compatibility of digital services taxes with WTO law  281 strength of such a claim will depend on which other foreign service supplier it uses for its comparison. As with National Treatment, the main difficulty here will be proving that the US service supplier and the services it provides are ‘like’ the other foreign service supplier and its services.72 Since, based on their construction, the burden of DSTs appears to fall particularly on services or service suppliers from the US, this constitutes less favourable treatment of US services and service suppliers in comparison with the treatment of services and service suppliers of other WTO members, evidently breaching MFN. As discussed above, the inclusion of a threshold exempting smaller suppliers from the DST would increase this risk if suppliers of WTO members other than the US are more likely to be exempt.73 It is noteworthy that in France the DST is referred to as the ‘taxe GAFA’, referring to Google, Apple, Facebook and Amazon.74 Therefore, as the taxes appear to disproportionately target US companies it is quite likely that an MFN violation could be found. 3.3

The DST and GATS Exceptions

3.3.1 Securing compliance with domestic laws and direct taxes Commentators have argued that it is unlikely that the discriminatory nature of DSTs would be saved by GATS general exceptions.75 In particular, it is improbable that a DST would be considered necessary to protect public morals or to maintain public order, despite the broad scope of these concepts. There is arguably a moral dimension to compliance with an obligation to pay taxes that are owed,76 but it is hard to see how the crafting of tax policy could possibly be considered an element of ‘public morals’ or vital to the preservation of order in the sense of preventing civic unrest. A footnote to art XIV(a) on public morals states that this exception may only be invoked where there is a genuine and sufficiently serious threat posed to one of the fundamental interests of society, which does not appear to be engaged with regard to taxation of digital services. GATS art XIV(c), entitled ‘Securing Compliance with WTO-Consistent Regulations’, could conceivably apply to a DST. In Argentina­–Measures Relating to Trade in Goods and Services, the panel found that, by taxing the profits earned from certain financial services supplied from service suppliers of non-cooperative countries at a higher rate than like services from service suppliers of cooperative countries, the measure contributed to protecting the tax base. This was because it discouraged the undeclared outflow of capital and the false payment of interest.77 For a digital services tax to be justified under GATS art XIV(c), the country imposing the DST would have to show that the tax was ‘necessary’ to secure compliance with other domestic laws or regulations that were themselves consistent with WTO law – a high threshold. However, since DSTs appear to address tax avoidance only in a general sense, Pirlot, above n 25. Mitchell, Voon and Hepburn, above n 28 at 102–3. 74 Willemyns, above n 30 at 278. 75 Mitchell, Voon and Hepburn, above n 28 at 88. 76 See e.g. D Gonzalez, ‘Morality, Consciousness and Tax Discipline: Their Role in Tax Compliance’ Inter-American Centre of Tax Administrators (13 July 2020) www​.ciat​.org/​morality​-conscience​-and​-tax​ -discipline​-their​-role​-in​-tax​-compliance/​?lang​=​en (April 2022). 77 Argentina–Financial Services, Report of the Panel, above n 35. This decision was ultimately rejected by the Appellate Body. 72 73

282  Research handbook on digital trade rather than targeting conduct that is unlawful in the imposing countries, a defence under art XIV(c) is unlikely to succeed.78 Most consequently agree that it would be difficult to argue that a DST is necessary to secure compliance with certain laws or regulations given that these taxes were adopted to address tax avoidance in general.79 Unlike the GATT, GATS contains specific tax exceptions: GATS art XIV(d) refers to direct taxes as a the carve-out from the National Treatment principle ‘provided that the difference in treatment is aimed at ensuring the equitable or effective imposition or collection of direct taxes in respect of services or service suppliers of other Members’. This exception could conceivably justify the features of destination-based taxes such as DSTs that are potentially problematic under the National Treatment principle. It is not available for breaches of MFN. GATS’ tax carve-out enables members to differentiate between resident and non-resident service suppliers as well as between ‘service suppliers subject to tax on worldwide taxable items’ and ‘other service suppliers’. When considering the scope of the direct tax exception, the Appellate Body recognized the importance of a WTO member’s ability to protect its tax system, combat harmful tax practices and have effective tax collection.80 It could be argued that the DST seeks to secure compliance with domestic tax law, although the relevant tax law would be the DST itself (since no equivalent exists under most domestic tax regimes), rendering the measure problematically self-justifying. The exemption of certain resident service suppliers from DSTs could be saved by GATS art XIV(d) under the condition that they are subject to a broader personal income tax than non-residents (so as to meet the condition of the chapeau that they are not arbitrary, of which more below). For example, some large UK technology companies would be subject to the UK DST while smaller foreign ones are not. It is noteworthy that GATS art IV(d) only justifies inconsistencies with the National Treatment obligation – it would not operate as a defence to a breach of MFN. In assessing the validity of the ‘direct tax’ exception GATS art XXVII(o) usefully defines ‘direct taxes’ to comprise all taxes on total income, on total capital or on elements of income or of capital, including taxes on gains from the alienation of property, taxes on estates, inheritances and gifts and taxes on the total amounts of wages or salaries paid by enterprises, as well as taxes on capital appreciation. In one sense this definition captures DSTs because these are taxes on income generated by the provision of digital services. On the other hand, the practical consequence of DSTs is that the service supplier will pass the tax on to consumers, effectively acting as an intermediary, although only in a de facto sense as there is no requirement for consumers to pay this tax. A DST applicable to each digital service transaction, similar to an excise tax, would likely be characterized as an indirect rather than direct tax. In contrast, a digital services tax imposed on a proportion of a service supplier’s income (that portion attributable to the supply of digital services in the imposing country) would be more likely to qualify as a tax on ‘elements of income’ and consequently be viewed as ‘aimed at ensuring the equitable or effective imposition or collection of direct taxes’ within the meaning of the exception in GATS art XIV(d), as elaborated in GATS footnote (6).81 Of course, the fact that a tax (or a tax regime) is described as such is not determinative. In other words, both indirect

80 81 78 79

Michell, Voon and Hepburn, above n 28 at 105. Willemyns, above n 30 at 279. Argentina–Financial Services, Appellate Body Report, above n 35 at [6.191]. Mitchell, Voon and Hepburn, above n 28 at 106.

The compatibility of digital services taxes with WTO law  283 and direct taxes can violate the GATT if they are implemented in a discriminatory manner with respect to imported products.82 3.3.2 The chapeau While DSTs could conceivably fit under at least one of the enumerated exceptions to the GATS, it would be difficult for the DST-imposing country to satisfy GATS art XIV’s gatekeeping provision, known as ‘the chapeau’, which is necessary to successfully invoke any of the exceptions: ‘measures [must] not [be] applied in a manner which would constitute a means of arbitrary or unjustifiable discrimination between countries where the same conditions prevail, or a disguised restriction on international trade.’ Many believe that there is a good chance that a claimant could succeed in a WTO case challenging a DST on the basis of the chapeau.83 In assessing the applicability of the chapeau, it must be recognized that the tax base of a digital services company will be some measure of global revenue attributed to website users in the imposing country. But since users do not pay anything to tech firms, the attribution of advertising and other revenues, perhaps paid by firms located in the US or France, to the UK tax authority in the case of the UK DST will be arbitrary.84 Thus, DSTs arguably constitute ‘arbitrary or unjustifiable discrimination between countries where the same conditions prevail’. Indeed, the United States Trade Representative (USTR) claimed that the UK DST was designed as ‘a narrowly-targeted tax’ on revenues from specific digital platform business models to ensure it is established tech giants – rather than UK tech start-ups – that shoulder the burden. The USTR argued that references to ‘established tech giants’ by the UK government clearly refer to US companies and disclose the intention to hit US companies while excluding similarly situated UK digital service companies. The USTR further asserted that the UK DST only pertains to three specific categories in which US firms are marketplace leaders – namely, certain search engines, social media platforms and online marketplace. It therefore unfairly targets US companies.85 DSTs have been viewed by others as discriminatory precisely because they apply almost exclusively to US companies.86 In assessing the discriminatory effect of a measure for the purposes of satisfying the chapeau of art XIV, the DST’s design, structure and intended operation must be examined along with its implications for the marketplace – not necessarily its actual effects. The USTR stated that the design of the UK’s DST was such that leading digital services firms in the UK and Europe would not be affected. It cited Spotify and Monzo as two companies which would not be caught by the DST because they do not operate search engines, social media services and online marketplaces.87 Neither Spotify nor Monzo are British; however, the fact that the services provided by these firms are excluded calls into question why the UK DST was crafted in the way that it was, suggesting arbitrariness. The USTR further argued, in the case of the UK DST, that it appears as though the UK government intends to interpret its DST purposefully in a manner that would exclude UK companies. Notably, ‘marketplace delivery fees’

84 85 86 87 82 83

Pirlot, above n 25. Forsgren, Song and Horváth, above n 32. Hufbauer and Lu, above n 31. USTR, above 14 at 16. Geist, above n 4. USTR, above 17 at 15.

284  Research handbook on digital trade may be excluded, precluding the application of the DST to two British companies that might have otherwise been subject to the DST: Just Eat and Deliveroo.88 The USTR also criticized the UK DST on the basis that it was crafted so that the thresholds will apply on a group-wide basis, not on a per business activity or per company basis. This means that revenues will be counted towards DST thresholds even if they are recognized in entities which do not have a UK taxable presence for corporation tax purposes. The USTR claimed that the purpose of this application of the tax is precisely to ensure that leading US firms’ revenues are captured, which it views as indicative of unfair discrimination against US companies.89 A measure that applied particularly to foreign service suppliers, such as those from the US, as the consequence of a lower threshold, may be seen to constitute trade discrimination or restriction, contrary to the chapeau. The country imposing the DST would need to demonstrate that this distinction had a legitimate rationale linked to the overarching purpose of the measure,90 namely ensuring the fair collection of tax. But it seems that the DSTs’ exemption for smaller suppliers undermines the purpose of ensuring tax coverage of digital services.91 A measure’s absence of consistency is thought to indicate arbitrary discrimination or a disguised restriction on international trade.92 Establishing the policy purpose behind the DST would assist in determining whether it satisfies the art XIV chapeau. Ideally there should be some proportionality between the objective and the measure chosen. The greater the contribution to the stated objective, the easier it would be to establish that the DST is ‘necessary’.93 It is useful to consider the stated objective of the UK DST, as explained by the HMRC: The application of the current corporate tax rules to businesses operating in the digital economy has led to a misalignment between the place where profits are taxed and the place where value is created. Many of these digital businesses derive value from their interaction and engagement with a user base. Under the current international tax framework, the value businesses derive from user participation is not taken into account when allocating the profits of business between different countries. This measure will ensure the large multinational businesses in-scope make a fair contribution to supporting vital public services.

The objective of the DST therefore seems to be, rather vaguely, one of ‘fairness’. But commentators have noticed that the DST does not seem to support the objective of establishing a tax system which is ‘fair’. The scope of the DST seems counterproductive to a fair tax regime, particularly because of thresholds and the services covered by the tax.94 In the context of DSTs imposed by EU members, notably France, the EU position appears to be that profit margins of the big tech companies are so high that the 3 per cent DST imposed by France, for

Ibid at 16. Ibid at 18. 90 EC–Seal Products, Appellate Body Report, WT/DS400/AB/R and WT/DS401/AB/R at [5.306] (adopted 18 June 2014). 91 Mitchell, Voon and Hepburn, above n 28 at 106. 92 US–Gambling, Report of the Panel, WT/DS285/R (20 April 2005). 93 A Mitchell, ‘Proportionality and Remedies in WTO Disputes’ 17:5 European Journal of International Law 985 (2006). 94 PWC, above n 20. 88 89

The compatibility of digital services taxes with WTO law  285 example, will have no practical bearing on the firm,95 detracting from the claim that the tax is somehow fair in terms of a company’s ability to pay. Furthermore, the rationale underlying taxing user value creation, identified by HMRC as one of the justifications for the UK DST, does not seem credible. This is because the use of digital interfaces creates value for all companies that use them including excluded businesses, for example those that supply certain investment services. The application of the measure indicates that, rather than taxing user value creation, which would capture a broader range of services/suppliers, the DST seeks to tax companies where the degree of value created by users is at a particular level, which happens primarily to be US companies. With regards to preventing the distortion of competition, it is not clear that the tax profile of suppliers targeted by the DST will beneficially impact competition in that sector. Competition may be more the result of technological advances and consumer habits which have led consumers to use digitally supplied services.96 Arbitrariness is identified in WTO law by reference to reasonably available less trade-restrictive alternative measures. The trade-restrictiveness of a measure is typically a question of degree. The Appellate Body ruled recently that the mere modification of the conditions of competition in a market does not suffice to indicate the level of restrictiveness – evidence regarding the actual trade effects may also be needed.97 Less trade-restrictive measures, instead of a DST, are conceivable, suggesting that the DST is not ‘necessary’ within the meaning of art XIV. Commentators have suggested that ‘a less blunt and discriminatory alternative DST might cover an expanded scope of services supplied digitally and capture more services that are in a competitive relationship by using different revenue thresholds’.98 The revenue thresholds, which lack a clear rationale, might also pose challenges under the chapeau. If the objective of the UK’s DST is truly to provide a regulatory environment in which smaller start-ups can grow and thrive, to the ultimate betterment of consumers, then surely competition law is a more suitable mechanism to address the dominance of the large tech companies that have crowded out the market. The EU’s new Digital Markets Act, designed to control abuse of market dominance of the large tech companies, is a good example of this approach.99 Getting rid of cumbersome data privacy requirements such as the EU’s General Data Protection Regulation (GDPR), which disadvantages smaller firms, would be another sensible move.100 Moreover, evidence indicates that suppliers caught by the DST currently pay about the same percentage of direct taxes as multinational enterprises which fall outside its scope. This in turn suggests that the DST may constitute arbitrary or unjustifiable discrimination against foreign/ US services and suppliers and a disguised restriction on trade in digital services.101 In addition to GATS direct tax exception, there is also room to challenge the application of a DST on the basis that the DST is covered by an existing double taxation treaty between Ibid at 10–11. Ibid at 15. 97 Australia–Tobacco Plain Packaging, WT/DS435/AB/R (9 June 2020). 98 Ibid at 13–14. 99 Proposal for a Regulation of the European Parliament and of the Council on Contestable and Fair Markets in the Digital Sector (Digital Markets Act) COM(2020) 842, 2020/0374(COD). 100 CB Frey and G Presidente, ‘The GDPR Effect: How Data Privacy Regulation Shaped Firm Performance Globally’ Vox EU CEPR (10 March 2022) https://​voxeu​.org/​article/​how​-data​-privacy​ -regulation​-shaped​-firm​-performance​-globally (April 2022). 101 PWC, above n 20 at 15. 95 96

286  Research handbook on digital trade the countries under art XIV(e). This defence reflects the WTO dispute settlement body’s cognizance of international treaties between the parties.102 In order to defend its DST against a National Treatment claim, the imposing country would have to respond successfully to the claimant’s demonstration that the foreign country’s DST is covered by a double taxation treaty between the two countries. If such a treaty exists, a National Treatment claim against a foreign country’s DST will be barred by GATS art XXII. Commentators have suggested that, in the case of a US claim against the French DST for example, the US would be able to establish that the French DST is not covered by the non-discrimination provision of the US–France Bilateral Income Tax Treaty.103

4. CONCLUSION The DSTs proposed by a number of countries poses significant challenges in terms of compliance with WTO law. The clear target of DSTs – US big tech companies such as Google, Meta and Amazon – may prompt the US to bring WTO discrimination claims under the GATS against countries imposing DSTs. While the success of these claims will ultimately depend on how DSTs are applied in practice, there is good reason to believe that they will be viewed as discriminatory and potentially arbitrary, primarily because of their apparently purposeful design to hit US companies as well as the debatable value of user-generated content. The US’s threat of trade retaliation against every country imposing a DST is not credible. It is conceivable that the US could impose retaliatory tariffs against France and a few other large economies, but not likely against 30-plus countries. Ultimately retaliatory tariffs are also harmful to consumers, undermining many of the gains from free trade. This is particularly so since the tax burden of DSTs is fairly small and, again, it will for the most part be passed on to consumers.104 The fair contribution of large tech companies has exposed the inadequacy of traditional tax laws in capturing profit-generating activity which occurs in a digital environment. Some argue that WTO law should not be used as an excuse to disregard a move towards destination-based taxation as a relevant policy option to reform outdated tax systems.105 In contrast, others contend that national DSTs undermine global efforts through the OECD to establish a multilateral destination-based tax,106 although the proposed OECD convention will probably never be implemented because it will require changes to every tax treaty in existence.107 This may explain why countries such as Canada are preparing to go ahead with DSTs on their own. Others hold that multilateral conventions regarding taxation of digital presence are inferior to unilateral DSTs which can be structured to address the particular needs of each country.108 The possibility of the US retaliating against DSTs by imposing tariffs on the countries which have imposed it highlights the larger issue in the use of DSTs. Until a global treaty on P van den Bossche and W Zdouc, The Law and Policy of the World Trade Organization (Cambridge University Press, 2022) at 68–70. 103 Forsgren, Song and Horváth, above n 32. 104 Geist, above n 4. 105 Pirlot, above n 25. 106 USTR, above n 17 at 2. 107 Geist, above n 4. 108 W Cui, ‘The Superiority of the Digital Services Tax over Significant Digital Presence Proposals’ 72:4 National Tax Journal 839 (Dec 2019). 102

The compatibility of digital services taxes with WTO law  287 digital tax is established, DSTs constitute a unilateral response to some of the tax problems posed by the digital economy. Unilateral measures of this kind can have unintended and unnecessary consequences, furthering the growing trend towards economic nationalism, which runs counter to the universal understanding that globalization, in particular freer trade, has contributed to economic and social prosperity.109 Apart from imposing taxes on a unilateral or even multilateral basis, a better way to ensure that the big digital services providers like Google and Amazon ‘pay their way’ and are making a fair contribution to society, for which payment of tax tends to be a proxy, is to mandate that these companies genuinely make such contributions. But as one commentator puts it, digital services providers ‘do not have the economic incentive nor the legal obligation to ensure public values such as reliability, diversity or pluralism in the public debate that they increasingly manage and shape’.110 If consumer-led demands for more socially minded digital services providers are unheeded, additional regulation may be needed to ensure that digital services providers fulfil these ‘duties’. Regulation of this kind will raise even more controversies than the debates surrounding DSTs. On the other hand, the mere enabling of online interactions among communities of people, detached from particular social objectives, has itself facilitated participatory democracy and the cultivation of civic identity. So-called ‘voice’ is often viewed as an essential component of citizenship and digital media has unquestionably facilitated this.111 If this can be believed, then it is at least plausible that the large tech companies should not bear the brunt of additional taxes that appear to run counter to established principles of international trade.

Mitchell, Voon and Hepburn, above n 28 at 124. Montero and Finger, above n 63 at 118. 111 B Bimber, ‘The Impact of Digital Media on Citizenship from a Global Perspective’ in E Andvizen, M Jensen and L Jorba eds, Digital Media and Political Engagement Worldwide (Cambridge University Press, 2012) at 30. 109 110

18. The feasibility and desirability of applying international investment agreements to digital assets Nicolette Butler and Jasem Tarawneh

1. INTRODUCTION The digital economy is rapidly expanding, and states are increasingly recognising its potential to contribute to economic growth by increasing gross domestic product (GDP). This will, in turn, potentially translate to improvements in the standard of living and poverty reduction.1 At the heart of the expansion of the digital economy is digital trade, which is defined as ‘digitally-enabled transactions of trade in goods and services that can either be digitally or physically delivered’.2 Growth of the digital economy is achieved through the increasing the volume of digital trade. A closely related concept is that of technological advancement: technological innovation and advancement drive digital trade, which in turn leads to growth of the digital economy. However, as digital technologies evolve and expand, governments are faced with ever more complex regulatory challenges in order to ensure that other (potentially competing) legitimate aims and policy goals are satisfied (or at least not harmed) as technology progresses and as the digital economy flourishes through digital trade. There is acknowledgement that regulation of the digital economy and digital trade ‘is an area where little progress has been made at the international level in building a framework for an open, rules-based global digital economy’.3 Two of the most often cited regulatory challenges pertaining to digital trade and the digital The extent to which increased economic growth (measured through GDP) leads to raised standards of living and poverty reduction is debated in the literature. See for example M Roemer and M Gugerty, ‘Does economic growth reduce poverty?’ (Technical Paper, Harvard Institute for International Development, April 1997) available at https://​pdf​.usaid​.gov/​pdf​_docs/​PNACA656​.pdf accessed 4 August 2022. However, it is generally accepted that ‘sustained and inclusive economic growth can drive progress, create decent jobs for all and improve living standards’ (United Nations, Sustainable Development Goals website, available at www​.un​.org/​sus​tainablede​velopment/​economic​ -growth/​accessed 4 August 2022). It is important to note that levels of development can also affect the extent to which economic growth can raise living standards and reduce poverty: ‘The extent to which growth reduces poverty depends on the degree to which the poor participate in the growth process and share in its proceeds. Thus, both the pace and pattern of growth matter for reducing poverty’ (Department for International Development, ‘Growth: building jobs and prosperity in developing countries’, available at www​.oecd​.org/​derec/​unitedkingdom/​40700982​.pdf accessed 4 August 2022). 2 New Zealand Foreign Affairs & Trade, ‘What is “the digital economy” and “digital trade”?’ website, available at www​.mfat​.govt​.nz/​en/​trade/​free​-trade​-agreements/​free​-trade​-agreements​-in​-force/​ digital​-economy​-partnership​-agreement​-depa/​what​-is​-the​-digital​-economy​-and​-digital​-trade/​ accessed 4 August 2022. 3 R Atkinson and N Cory, ‘Cross-border data policy: opportunities and challenges’ in H Wang and A Michie (eds), Consensus or Conflict? China and Globalization (2021, Springer). 1

288

Applying international investment agreements to digital assets  289 economy are data privacy and cyber security. For example, law and policy makers are confronted with a whole range of issues relating to privacy and cybersecurity concerns when it comes to managing cross-border data flows. On the one hand, states wish to encourage digital trade and data flows, but on the other, they need to balance the economic benefits of this with the rights of citizens to data privacy and national security concerns.4 This highlights one of the major challenges for governments in terms of regulating digital trade; states strive to expand their digital economies through digital trade, but states must also protect state interests and their citizens from any potential adverse effects. The central objective of this chapter is to examine the regulation of digital assets through an international investment law lens. Accordingly, the aim of the chapter is to determine whether the current regime for the regulation of international investment is applicable to digital assets, and further to examine whether the current regime provides an appropriate regulatory framework for such assets. International investment law has not yet been confronted with issues pertaining to the digital sector, and specifically the status of digital assets.5 This is due to the fact that there are no known investment disputes dealing with digital assets to date. The lack of disputes, the relative novelty of the topic and the vagueness of the concepts underlining it mean that the digital sector and its interaction with the international investment law regime remain largely unexplored in academic literature; herein lies the originality of this chapter. In order to achieve its aim, this chapter is structured into five sections. Section 2 will provide background information on the digital economy, digital assets and the international investment law regime in order to contextualise the discussion. Section 3 will consider the applicability of the current international investment law regime to digital assets, examining whether they can be characterised as protected investments. Section 4 will question the suitability of the potential application of the current international investment regime to digital assets. The fifth and final section will offer some concluding remarks.

2.

BACKGROUND AND CONTEXT

2.1

Digital Economy

In recent years, digital technologies have provided exponential opportunity for economic growth. Such growth has been further fuelled by the Covid-19 pandemic, which demonstrated the importance of technology and digital platforms in almost every aspect of our lives. The development of digital technology and its contribution to economic growth is expected to continue to accelerate in the future. These developments have led to increased emphasis on the importance of the so-called digital economy by many states, including the UK.6 Despite

Ibid. E Horváth and S Klinkmüller, ‘The concept of “investment” in the digital economy: the case of social media companies’ (2019) 20 The Journal of World Investment & Trade 577, p.580. 6 UK Government Department for Digital, Culture, Media and Sport Policy paper, ‘Digital regulation: driving growth and unlocking innovation’ (9 March 2022) available at www​.gov​.uk/​ government/​publications/​digital​-regulation​-driving​-growth​-and​-unlocking​-innovation/​digital​-regulation​ -driving​-growth​-and​-unlocking​-innovation​#:​~:​text​=​Digital​%20technologies​%20are​%20the​%20engine​ ,1​.6​%20million​%20jobs​%20in​%202019 accessed 2 June 2022. 4 5

290  Research handbook on digital trade a wealth of literature,7 there is no universally accepted definition of the term digital economy.8 Notwithstanding this, the OECD suggests: The Digital Economy incorporates all economic activity reliant on, or significantly enhanced by the use of digital inputs, including digital technologies, digital infrastructure, digital services and data. It refers to all producers and consumers, including government, that are utilising these digital inputs in their economic activities.9

It is important to note that this definition is not utilised by all nations, and without a common benchmark it is near impossible to assess the true size or significance of digital economies of nations in comparison to one another. Nonetheless, attempts have been made to calculate digital competitiveness; the IMD’s World Digital Competitiveness Ranking 2021 places the USA at the top, followed by Hong Kong and Sweden. Meanwhile, the UK is ranked 14th.10 Despite this comparatively low ranking, at present the digital economy is credited as the ‘engine driving the UK’s economic growth’,11 contributing £151 billion in output, and accounting for 1.6 million UK jobs in 2019. The importance of the digital economy is not a UK-specific phenomenon; the digital economy accounted for 9.6 per cent of US gross domestic product (GDP) in 2019 and supported 7.7 million jobs in the US (5.0 per cent of total US employment) in the same year. One study found that the digital economy added 1.4 million US jobs between September 2017 and September 2021, and was the main job producer in 40 states.12 In actual fact, the digital economy currently represents 15.5 per cent of global GDP,13 and it is expected that this will continue to increase in the future. The statistics demonstrate that the digital economy is indeed extremely lucrative. It therefore attracts considerable attention from state governments wishing to capitalise on its potential to accelerate national economic growth and improve standards of living. However, in endeavouring to grow the digital economy, law and policy makers also have to grapple with the regulation of new technologies and technological developments. Indeed, ‘in this rapidly evolving environment, governments are facing growing regulatory challenges in ensuring that

See for example R Bukht and R Heeks, ‘Defining, conceptualising and measuring the digital economy’ (2017) Development Informatics Working Paper Series (No. 68) available at http://​hummedia​ .manchester​.ac​.uk/​institutes/​gdi/​publications/​workingpapers/​di/​di​_wp68​.pdf accessed 3 August 2022. 8 OECD, ‘A roadmap towards a common framework for measuring the digital economy’ (2020) available at www​.oecd​.org/​sti/​roadmap​-toward​-a​-common​-framework​-for​-measuring​-the​-digital​ -economy​.pdf accessed 4 June 2022. 9 Ibid, p.34. 10 IMD World Digital Competitiveness Ranking (2021) available at www​.imd​.org/​centers/​world​ -competitiveness​-center/​rankings/​world​-digital​-competitiveness/​ accessed 4 June 2022. 11 UK Government Department for Digital, Culture, Media and Sport Policy paper, ‘Digital regulation: driving growth and unlocking innovation’ (9 March 2022) available at www​.gov​.uk/​ government/​publications/​digital​-regulation​-driving​-growth​-and​-unlocking​-innovation/​digital​-regulation​ -driving​-growth​-and​-unlocking​-innovation​#:​~:​text​=​Digital​%20technologies​%20are​%20the​%20engine​ ,1​.6​%20million​%20jobs​%20in​%202019 accessed 2 June 2022. 12 R Fefer, F Akhtar and D Sutherland, ‘Digital trade and US trade policy’ (2021, Congressional Research Service) available at https://​crsreports​.congress​.gov/​product/​pdf/​R/​R44565 accessed 2 June 2022. 13 The World Bank, IBRD, ‘Understanding poverty: digital development’ available at www​ .worldbank​.org/​en/​topic/​digitaldevelopment/​overview​#1 accessed 2 June 2022. 7

Applying international investment agreements to digital assets  291 the opportunities and benefits from digital trade can be realised and shared more inclusively’.14 The legal and regulatory challenges emerging from the digital economy and rapid digitalisation are many and varied. 2.2

Digital Assets

Digital assets are one very important aspect of the digital economy. Despite the fact that digital assets have the potential to contribute to exponential economic expansion, there is no universally agreed definition of a digital asset, let alone a comprehensive legal regulatory regime for such assets.15 Part of this chapter’s originality is that it will contribute to clarification of such regulatory issues, particularly pertaining to investment in digital assets. In order to achieve this aim, this chapter proceeds on the understanding that a digital asset is one that exists only in digital form, and to which is attached an ownership or usage right. These can be distinguished from physical assets which have a tangible form as well as an ownership or usage right.16 Digital assets can take many different forms.17 Nowadays, much of the discussion on digital assets tends to focus on cryptocurrencies, such as bitcoin, but the term technically refers to much more than such currencies.18 There is no widely agreed system of classification of digital assets: ‘at present, a common system of categorisation does not exist for digital assets. This is a barrier to the regulation and management of digital assets which often exist in an international and multi-jurisdictional environment.’19 That said, Wilshire’s Digital Asset Taxonomy seems to provide a useful trilateral classification of digital assets. Wilshire finds three categories of digital assets: digital currencies, computation platforms and financial instruments.20 Digital currencies (such as bitcoin) are assets ‘whose main objective is to replicate the fundamental functions of money: store of value, medium of exchange, and unit of account’.21 Computation platforms are ‘assets that exist within networks that support highly expressive, Turing-complete smart contracts’.22 Gaming services, non-fungible tokens, digital art and social networks are all examples of computation platforms.23 Finally, financial instrument digital assets are those ‘that apply the decentralized properties of digital assets to financial contracts and corporate structures that exist in traditional finance’.24 Examples of

14 J López González and J Ferencz, ‘Digital trade and market openness’ (2018) OECD Trade Policy Papers, No. 217, OECD Publishing, available at http://​dx​.doi​.org/​10​.1787/​1bd89c9a​-en accessed 3 June 2022. 15 J Bick, ‘All digital assets are not legally equal’ (Law Journal Newsletters November 2017), accessed 7 June 2022. 16 J Allen, M Rauchs, A Blandin and K Bear, ‘Legal and regulatory considerations for digital assets’ (2020) Cambridge Centre for Alternative Finance available at www​.jbs​.cam​.ac​.uk/​wp​-content/​uploads/​ 2020/​10/​2020​-ccaf​-legal​-regulatory​-considerations​-report​.pdf accessed 1 June 2022. 17 Ibid. 18 Ibid. 19 Ibid. 20 Wilshire, ‘Digital Asset Taxonomy System’ (December 2021) available at https://​assets​ -global​.website​-files​.com/​60f80​38183eb84c​40e8c14e9/​61e9​588916c9f6​7577825da2​_Digital​-Asset​ -Taxonomy​-System​-DATS​.pdf accessed 10 June 2022. 21 Ibid. 22 Ibid. 23 Ibid, noting that there are many more computation platform digital assets. 24 Ibid.

292  Research handbook on digital trade financial instrument digital assets include staking instruments and security tokens (such as tokenised hedge funds and debt).25 Although it is helpful to understand the range of digital assets available, especially given the general widespread limited understanding of the term, it should be noted that this chapter proceeds on the assumption that all digital assets (as described in the preceding paragraph) can be regulated in the same manner from an international investment perspective, at least. This is based on the view that their central common characteristics (that they exist purely in the digital sphere and are attached with some right of usage or ownership) mean that they warrant the same treatment vis-à-vis investment law. 2.3

International Investment Law

International investment (also known as foreign direct investment or FDI) takes place when a national of one state (or a company registered in one state) starts a business from scratch or takes over a business in a different state. International investment law refers to the legal norms that govern the relationship between said foreign investors and the host states in which they invest.26 Although the domestic law of the host state clearly has the potential to affect the investment relationship, international investment law is concerned only with the relevant sources of international law, such as customary international law and international investment agreements (bilateral investment treaties and treaties with investment provisions, that is, BITs and TIPs). Customary international law provides the most basic protections to foreign investors, many of which are expanded upon greatly in BITs and TIPs.27 The so-called spaghetti bowl28 network of the 222929 BITs and TIPs currently in force include similar but differently worded provisions providing what have become fairly extensive protections to foreign investors. BITs were originally intended to provide a ‘transparent framework of investment protections and state obligations’30 as well as to offer an adaptable form of regulation. This adaptability stems from the fact that the provisions of BITs can be applied by arbitrators in a flexible manner, in order to respond to future unforeseen developments. That said, ‘in order to maintain their legitimacy, BITs must not extend too far beyond what the states envisioned at the time of signing’.31 It should be noted that BITs/TIPs can also relatively easily be terminated or renegotiated, giving an added layer of flexibility to states in terms of the regulation of international investment.

Ibid. S Wittich, ‘International investment law’ in K Miles (ed), The Origins of International Investment Law Empire, Environment and the Safeguarding of Capital (Cambridge University Press, 2013) 822. 27 S Subedi, International Investment Law: Reconciling Policy and Principle (3rd edition, Hart, 2020). 28 UNCTAD, ‘Investment provisions in economic agreements’ (2006), available at https://​unctad​ .org/​system/​files/​official​-document/​iteiit200510ch1​_en​.pdf accessed 3 June 2022. 29 UNCTAD International Investment Agreements Navigator, available at https://​investmentpolicy​ .unctad​.org/​international​-investment​-agreements accessed 4 June 2022. 30 See M Sornarajah, The International Law on Foreign Investment pp.205­–6 (Cambridge University Press, 2017) as cited in J Chaisse and C Bauer, ‘Cybersecurity and the protection of digital assets: assessing the role of international investment law and arbitration’ (2019) 21 Vanderbilt Journal of Entertainment and Technology Law 549. 31 Ibid Sornarajah pages 225 – 226. 25 26

Applying international investment agreements to digital assets  293 The first BIT was signed in 1959, and such agreements proliferated in the 1980s and 1990s. Thus, many of the 2229 agreements have been in operation for 30–40 years. The negotiators of such agreements likely did not envision many of the issues pertaining to investment in digital assets that we are confronted with today. Answers to important questions on the regulation of investment in digital assets are therefore unlikely to be found wholesale in the current international investment agreements, most of which were negotiated prior to the development of much of the newest technology and the advancement of the digital economy that have recently been witnessed. It is for this reason that the evaluation of the applicability of the current regime of international investment law to digital assets is crucial. In most cases (that is, most BITs), the extension of investment protection to today’s digital assets was likely not contemplated by the signatory states. Indeed, ‘the operations [of entities operating within the digital economy] do not fit neatly into existing frameworks for investment protection, which were created primarily for the brick-and-mortar investments that remain the primary subject of disputes even today’.32 Accordingly, it is crucial to determine first the applicability of the current regime to digital assets, and further, the desirability of protecting investment in digital assets through the current regime of investment protection.

3.

APPLICATION OF INTERNATIONAL INVESTMENT AGREEMENTS

The importance of determining the applicability of an investment treaty cannot be overstated as it is ‘key to the scope of application of rights and obligations of investment agreements and to the establishment of the jurisdiction of investment treaty-based arbitral tribunals’.33 Further, ‘given the potential for investment claims in the digital era, the jurisdiction and admissibility of such claims is likely to be a divisive issue’.34 Thus, in order to determine the applicability of a BIT/TIP, generally two criteria must apply: investors must have a covered investment in the territory of the investment host state, and the investment be made ‘in the territory’ of the host state. Each of these requirements will be dealt with in turn. 3.1

Covered Investment

3.1.1 International investment agreement provisions: BITs and TIPs From the outset, it is important to note that to date, there is no known investment treaty that includes reference to digital assets explicitly as covered investments.35 Given the fact that digital assets are a relatively new phenomenon this is not necessarily surprising. Perhaps more surprisingly, there is also no universally accepted definition of the term ‘investment’.36 Each Horváth and Klinkmüller supra note 5. OECD, ‘International investment law: understanding concepts and tracking innovations’ (2008) available at www​.oecd​.org/​investment/​inte​rnationali​nvestmenta​greements/​40471468​.pdf accessed 1 June 2022. 34 Chaisse and Bauer, supra note 30. 35 An extensive search of the UNCTAD International Investment Agreements Navigator by the authors confirms this: supra note 29. 36 OECD, ‘International investment law: understanding concepts and tracking innovations’, supra note 33. 32 33

294  Research handbook on digital trade individually negotiated investment treaty (BIT/TIP) defines ‘investment’ for the purposes of the application of that particular treaty. That said, there is considerable convergence towards a broad approach to defining the term. BITs tend to take a broad view of the definition of investment, with many referring to ‘every kind of asset followed by an illustrative but usually non-exhaustive list of assets, recognising that investment forms are constantly evolving’.37 Assets that are often listed include: An enterprise; shares, stock, and other forms of equity participation in an enterprise; bonds, debentures, other debt instruments, and loans; futures, options, and other derivatives; turnkey, construction, management, production, concession, revenue-sharing, and other similar contracts; intellectual property rights; licenses, authorizations, permits, and similar rights conferred pursuant to domestic law; and other tangible or intangible, movable or immovable property, and related property rights, such as leases, mortgages, liens, and pledges.38

Clearly, then, under such a broad conception of the term, intangible assets can be categorised as an ‘investment’ under such a broad conception of the term. Indeed, this is evident from the very first bilateral investment treaty, which was signed in 1959 by Germany and Pakistan. Article 8 states: (1)(a) The term ―investment shall comprise capital brought into the territory of the other Party for investment in various forms in the shape of assets such as foreign exchange, goods, property rights, patents and technical knowledge. The term ―investment shall also include the returns derived from and ploughed back into such ―investment.39

The inclusion of intangible assets as a form of investment even in the earliest days of the BIT regime is significant, as it shows the intention from the outset that investment should not be limited only to physical assets. Despite this broad approach witnessed in many BITs, other treaties do offer a more restrictive approach to the definition of investment. Article I of the Slovakia–Iran BIT for example contains a lengthy definition of investment which includes an exhaustive list of covered investments, with additional requirements including the Salini criteria as well as a fifth element requiring ‘the expectation of regularity of profit’.40 The treaty even goes on to set out activities which will not be counted as investments: Notwithstanding the above, for the avoidance of any doubt, ‘investment’ shall not include: a) goodwill or market share; b) portfolio investment, which is 10% or less shareholding; c) claims to money deriving solely from commercial contracts for the sale of goods or services to or from the territory of a Contracting Party to the territory of another country, or to a State enterprise; d) futures, swaps, forwards, options, and other derivatives;

Ibid. US Model BIT 2012, available at https://​ustr​.gov/​sites/​default/​files/​BIT​%20text​%20for​ %20ACIEP​%20Meeting​.pdf accessed 1 June 2022. 39 Germany–Pakistan BIT 1959, available at https://​investmentpolicy​.unctad​.org/​international​ -investment​-agreements/​treaty​-files/​1387/​download accessed 1 June 2022. 40 Slovakia–Iran BIT 2016, available at https://​investmentpolicy​.unctad​.org/​international​-investment​ -agreements/​treaty​-files/​3601/​download accessed 1 July 2022. 37 38

Applying international investment agreements to digital assets  295 e) assets used for non-business purposes, other than assets of research and development non-profit organizations; f) funds; g) the following loans and debt securities: i. debt securities and loans with the original maturity of less than three years; ii. a loan to or debt security issued by a financial institution, which is not treated as regulatory capital by the Contracting Party in whose territory the financial institution is located; iii. the extension of credit in connection with a commercial transaction, such as trade financing.41

Additionally, the treaty requires a ‘significant physical presence of the investment in the territory of the Host State’42 if the purported investor is claiming that the investment is in a research and development non-profit organisation (which is included in the exhaustive list of possible covered investments). In this situation, the treaty defines such physical presence as ‘not includ[ing], for example, sales offices without other operational facilities, post office box-based businesses, internet-based business or other types of business with no or limited physical presence in the Host State’.43 3.1.2 ICSID decided cases As evidenced, there is no generally accepted definition of the term investment in international investment treaties. Accordingly, an examination of decided cases may aid in understanding the concept more concretely. The International Centre for the Settlement of Investment Disputes (ICSID), which is the primary institution which administers investment disputes, does not define investment. Article 25(1) of the ICSID Convention states that the Centre’s jurisdiction extends to a legal dispute arising from an ‘investment’ between a contracting state and a national of another contracting state.44 However, nowhere does the Convention define what ‘investment’ actually means. Indeed, the travaux préparatoires of the Convention demonstrate that several definitions were written and scrapped.45 Ultimately it was decided that such a definition was not necessary due to the ‘essential requirement of the consent of the parties’.46 Notwithstanding this, there have been a number of ‘generally expansive statements by [ICSID] tribunals’47 on what constitutes investment. Although there is no doctrine of precedent applicable to investment arbitration, examining individual cases does hold value as evidence of how key terms may be interpreted by arbitrators. Thus, an examination of leading cases can provide an illuminating discussion of arbitra Ibid. Ibid. 43 Ibid. 44 Article 25(1) ICSID Convention 1965 available at https://​icsid​.worldbank​.org/​en/​Documents/​ icsiddocs/​ICSID​%20Convention​%20English​.pdf accessed 1 June 2022. 45 History of the ICSID Convention Vol I-1, p.116, Vol II-1, pp.285–6 and 492–3 and Vol II-2 pp.843–4 and 972, online versions available at https://​icsid​.worldbank​.org/​resources/​publications/​the​ -history​-of​-the​-icsid​-convention accessed 15 July 2022. 46 Report of the Executive Directors of the International Bank for Reconstruction and Development on the Convention on the Settlement of Investment Disputes between States and Nationals of Other States, 1 ICSID Reports 23 (1993), section 22. 47 A Mitchell and J Hepburn, ‘Don’t fence me in: reforming trade and investment law to better facilitate cross-border data transfer’ (2017) 19 Yale Journal of Law and Technology 182. 41 42

296  Research handbook on digital trade tors’ approaches to defining investment. Various cases have been dealt with under the auspices of ICSID which have attempted to define the meaning of the term under the Convention itself. The most important case to date on the issue is undoubtedly Salini.48 According to the case, in order to be considered an investment, four criteria must be satisfied: a contribution of money or assets; an element of risk; for a certain duration; and a contribution to the host state’s economy.49 Many subsequent cases have accepted and endorsed the Salini criteria without hesitation.50 Other tribunals have not accepted Salini and have chosen to advance alternative views of the definition of the term investment.51 In terms of the four Salini requirements specifically, the first three criteria are not particularly problematic. However, the fourth element that requires a contribution to the host state’s economy has proven decidedly more controversial. In declaring this requirement for an investment, the Salini tribunal relied on the preamble to the ICSID Convention as the foundation.52 In contrast, in the Quiborax53 case, the tribunal suggests that ‘such contribution may well be the consequence of a successful investment; it does not appear as a requirement’.54 The tribunal in Quiborax asserted that this dropping of the fourth Salini criterion is an evolution executed and endorsed by several ICSID tribunals over time. In LESI v Algeria,55 the first three Salini criteria were met but the fourth was not. They considered the fourth criteria, ‘a requirement that is any event difficult to establish and implicitly covered by the other elements reviewed’.56 Similarly, in Victor Pey Casado v Chile57 a contribution to the development of the host state was deemed a potentially expected consequence of an investment, but not a requirement for its existence. In Phoenix Action v Czech Republic58 the tribunal were of the opinion that what matters is a contribution to the economy of the host state, and that requirement was subsumed by the other three Salini criteria. Finally, in Saba Fakes59 the requirement of contributing to the development of the host state was unequivocally rejected. Commentators have sought to explain the trend towards rejection of the fourth Salini criteria by the fact that, ‘while the ICSID Convention attempts to foster economic development via international investment, such development is not a necessary element of investment’.60 Whereas the Salini tribunal itself reasoned that requiring a contribution to the host state was Salini Costruttori S.p.A. and Italstrade S.p.A. v Kingdom of Morocco (ICSID Case No. ARB/00/4). Salini Costruttori, ibid, Decision on jurisdiction, 23 July 2001, available at www​.italaw​.com/​sites/​ default/​files/​case​-documents/​ita0738​.pdf accessed 1 June 2022. 50 See for example Joy Mining Mach. Ltd. v Egypt, ICSID Case No. ARB/03/11. 51 Quiborax v Bolivia, ICSID Case No. ARB/06/2, Saba Fakes v Turkey, ICSID Case No. ARB/07/20, Victor Pey Casado and President Allende Found v Chile, ICSID Case No. ARB/98/2, LESI S.p.A. et Astaldi S.p.A. v Algeria, ICSID Case No. ARB/05/3. 52 ICSID Convention, supra note 44, as referenced by the Salini tribunal in its Decision on Jurisdiction, supra note 49, para 52. 53 See for example Quiborax, supra note 51. 54 Ibid, Decision on Jurisdiction, 27 September 2012, available at www​.italaw​.com/​sites/​default/​ files/​case​-documents/​italaw1098​.pdf accessed 10 July 2022, para 220. 55 LESI, supra note 51. 56 Ibid LESI, Decision on Jurisdiction, 12 July 2006, para 72. 57 Victor Pey Casado, supra note 51. 58 Phoenix Action Ltd. v Czech Republic, ICSID Case No ARB/06/5, Decision on Jurisdiction, 15 April 2009, para 85. 59 Saba Fakes v Turkey, supra note 51, paragraphs 110–111. 60 A Grabowski, ‘The definition of investment under the ICSID Convention: a defense of Salini’ (2014) 15 Chicago Journal of International Law 287. 48 49

Applying international investment agreements to digital assets  297 justified in Salini by reference to the ICSID Convention preamble, rather than specific provisions of the Convention itself.61 In international law, the value of preambular language and the extent to which it is binding is not completely clear.62 The issue is not clarified by the Vienna Convention,63 which appears to take an ambivalent approach to preambles.64 A move away from requiring a contribution to the host state development may have also been motivated by the inherent difficulties in defining and also quantifying and attributing such a contribution.65 For example, would only an increase in GDP amount to a contribution to economic development of the host state? If GDP does qualify, it would be nearly impossible to measure and attribute such an increase to a specific investment, given the many other factors which can influence economic growth. Additionally, scholars questioned whether the contribution to the host state should be only economic, and whether social, political or cultural contributions should count.66 Whatever the exact reason(s) for any given tribunal to move away from requiring a contribution to the development of the host state, the fact that it happens is most relevant to this discussion. So too is the fact that some tribunals appear to have moved away from contribution to development and towards an examination of whether the investor has made a ‘contribution […] apt to create the value that is protected under the BIT’.67 This may not be a positive development, as it could be argued that such a contribution is even more difficult to define and measure than a contribution to the development of the host state developed by the Salini tribunal. To summarise, then, decided cases may not be very helpful in the quest to try to understand the definition of investment. This is because different cases have put forward different definitions of the term, and also because there is a lack of de jure precedent when it comes to international investment disputes. Nonetheless, cases may aid us in understanding the types of issues that tribunals will take into account when making a decision about what will or will not constitute an investment in a given case. Applying this to digital assets, then, it may (or may not) be necessary for a purported investor to demonstrate that there has been some contribution to the economic development of the host state as a result of the investment in the digital asset. Given the online and often delocalised nature of such assets, this may prove tricky in many instances.

Salini supra note 48. See for example M Hulme, ‘Preambles in treaty interpretation’ (2016) 164 University of Pennsylvania Law Review 1281. 63 Vienna Convention on the Law of Treaties (1969) available at https://​legal​.un​.org/​ilc/​texts/​ instruments/​english/​conventions/​1​_1​_1969​.pdf accessed 15 July 2022. 64 J Klabbers, ‘Treaties and their preambles’ in M Bowman and D Kritsiotis, Conceptual and Contextual Perspectives on the Modern Law of Treaties (Cambridge University Press, 2018) pp.172–200. 65 M Hwang and J Fong, ‘Definition of “investment”: a voice from the eye of the storm (2011) 1 Asian Journal of International Law 1. 66 C Schreuer, The ICSID Convention: A Commentary on the Convention on the Settlement of Investment Disputes Between States and Nationals of Other States (2nd edition, Cambridge University Press, 2009). 67 Abaclat and Others v Argentine Republic, ICSID Case No ARB/07/5, para 365. 61 62

298  Research handbook on digital trade 3.2

Territorial Requirement

Aside from the necessity of establishing the existence of a covered investment, treaties also typically require that the investment be made ‘in the territory’ of the host state.68 In relation to whether digital assets can indeed be protected or covered investments, this requirement is the most problematic, given the non-tangible and internet-based nature of digital assets. The territoriality requirement primarily exists because an investment is meant to benefit the host economy in a manner that trading goods and/or services would not, for example by increasing GDP, creation of employment for host state citizens or technology transfer.69 In practice, this requirement may be difficult to separate from the fundamental notion of an investment. Mitchell and Hepburn note: The territorial requirement is connected to the basic (though sometimes elusive) distinction between trade and investment. Broadly speaking, while cross-border traders operate from their home state even if selling goods or services into another state, cross-border investment by its nature involves more integration of business operations within the host state.70

This is a key distinction, because without an investment in the host state’s territory, the would-be claimant – for example, an individual or company offering products or services for sale via the internet to consumers in another country – risks their status being viewed as that of simply a trader (that is, not a true investor), and consequently they may find themselves ineligible to access investment treaty protections. The territoriality requirement when it comes to investment in digital assets is potentially much more complex than when it comes to physical assets. It is usually quite clear, for example, that an entrepreneur is indeed a foreign investor when a foreign national purchases land in order to build and operate a warehouse or factory in a state other than their own. It is equally clear that a company incorporated in a different state to the one in which the company itself is operating is also a foreign investor. The non-tangible, online nature of digital assets can make the determination of territoriality much more complex. The ‘[technology] sector […] is perhaps least likely to have a physical presence in the countries in which it is able to operate, precisely because many of its products and services can be delivered electronically via the internet’.71 Bick proposed the categorisation of digital assets based on their location; such categorisation may be helpful when it comes to the territoriality requirement. According to Bick, The first class of digital assets is contained on a device that is in the owner’s control. Usually, this device is a computer or storage device. Class one digital assets include emails, software, and content and data stored in tangible property, typically a decedent’s home computer… A second class of digital assets are access rights and use rights to Internet assets located in a computer or other storage device owned by a person other than the digital asset owner. Class two digital

C McLachlan, L Shore and M Weiniger, International Investment Arbitration: Substantive Principles (Oxford University Press, 2017) as cited in Mitchell and Hepburn, supra note 47. 69 J Salacuse, The Law of Investment Treaties (Oxford University Press, 2010) p.210. 70 Mitchell and Hepburn, supra note 47. 71 Ibid. 68

Applying international investment agreements to digital assets  299 assets are emails, software, content and data stored in tangible property on a third-party’s computer or other tangible property… Class-three digital assets are access and use rights related to internet assets, but unlike class-two digital assets, class-three digital assets do not have any physical point of presences (i.e., their existence is not dependent upon storage), hence they need not be stored anywhere. A domain name is an example of a class three digital asset72

This categorisation of digital assets then appears to be essentially founded on three potentially relevant factors: location, possession and control. Applying the categorisation in order to resolve the territoriality conundrum then, class one assets could be established as investments relatively easily where an investor has some kind of physical presence or physical location in the investment host state. Such a physical presence might be established where the investor has local servers. Of the three classes, class one most overtly establishes a territorial link to the host state, as the assets are ‘physically’ located within the host state and they are also under the control of the investor. Class two assets are similar to class one, in that they have some kind of physical form. However, class two can be distinguished by control; the storage or device is controlled by a third party (as opposed to it being controlled by the asset owner him/ herself, as per class one). This class would include investors who grant the use of or access to third parties through licensing agreements. Unlike class one or two assets, class three assets have no physical presence anywhere. Would-be class three investors might include businesses involved data processing or social media companies. Class two (digital assets being outside the control of the investor) and class three (where the digital assets have no physical connection) are more difficult for establishing a territorial connection to the host state. As such, it may be more difficult to satisfy the territoriality requirement.73 Bick’s categorisation might be theoretically helpful to understanding the threshold of the territoriality requirement of an investment. However, examining how arbitrators have tackled the issue in practice is even more useful. There are no known cases dealing with digital assets per se.74 However, there are a number of cases where the would-be investment involved an intangible asset. For example, the Abaclat case concerned sovereign bond investments (that is, an intangible asset). The Abaclat panel helpfully set out a territoriality test for such assets. The panel explained that ‘determination of the place of the investment firstly depends on the nature of such investment’.75 According to Abaclat, the criteria applied to the territoriality requirement are different in the case of an investment in intangible assets. In this case, ‘the

72 UNCTAD International Investment Agreements Navigator, supra note 29; see J Bick, ‘All digital assets are not legally equal’, Law Journal Newsletters (November 2017), https://​www​.l​awjournaln​ ewsletters​.com/​sitesila​wjournalne​wsletters/​2017/​11101/​all​-digitalassets​-are​-not​-legally​-equal/​ as cited in Chaisse and Bauer, supra note 30. 73 Ibid; see also A Chakravarty, ‘Challenges to the assessment of damages claims involving crypto-assets in investment arbitration’ Global Jurist, 20, 2, 2020, available at https://​doi​.org/​10​.1515/​gj​ -2019​-0044 accessed 23 August 2022. 74 Horváth and Klinkmüller, supra note 5. 75 Abaclat and Others v Argentine Republic, ICSID Case No ARB/07/5, Decision on Jurisdiction (4 August 2011) para 374, available at www​.italaw​.com/​sites/​default/​files/​case​-documents/​ita0236​.pdf accessed 4 August 2022. For other examples see L Cuatrecasas, ‘International Investment Policy and the Coming Wave of Data-Flow Disputes’ (4 September 2021) 11 Michigan Business & Entrepreneurial Law Review (forthcoming), available at SSRN: https://​ssrn​.com/​abstract​=​3917552 or http://​dx​.doi​.org/​ 10​.2139/​ssrn​.3917552 accessed 20 August 2022.

300  Research handbook on digital trade relevant criteria should be where and/or for the benefit of whom the funds [were] ultimately used, and not the place where the funds were paid out or transferred’.76 Applying this test to digital assets might be helpful as it would certainly eliminate some of the uncertainties related to the location of such assets, shifting the emphasis towards the creation of benefits in the host state. That said, this reliability of the Abaclat test was weakened by the dissenting opinion of Professor Abi-Saab who argued that the alleged investments ‘have been sold in international financial markets, outside Argentina, with a choice of law and forum selection clauses subjecting them to laws and fora foreign to Argentina […] they were intentionally situated outside Argentina and out of reach of its laws and tribunals […] there is no legal basis for saying that they are located in Argentina’.77 In other cases, for example SGS,78 the tribunal took a more literal approach to the territoriality requirement, stating that it would assist the investor’s claim of territoriality if evidence of expenditures to establish the investment within the host state are demonstrated. In this way, an investor could show evidence of payment to a local (host state-based) web hosting company to host its website, or perhaps that it owns or rents property located in the host state which houses the server. Thus, it would seem that an investor may need some kind of physical presence within the territory of the host state in order to satisfy the territoriality requirement, even with digital assets. Tribunals have upheld claims where investors constitute minority shareholders in a subsidiary business incorporated solely to conduct the investors’ business within the host state as well.79 Some types of business operations may find this requirement much more challenging to satisfy – social media companies, for example, whose ‘digital operations do not take place in the territory of any given State, but in cyberspace, and involve no obvious flow of capital or other resources into a host State’.80 In reality, social media companies participate in the host state’s economy ‘only to the extent that their commercial clients and the users of any given platform, especially those targeted by the aforementioned commercial clients, are present in the territory of that State’.81 Other purely online companies may have similar issues concerning the territoriality requirement; for example, online ‘cloud’ data storage companies that can offer storage facilities to internet users anywhere in the world. Some tribunals have interpreted the territoriality requirement more creatively, finding that ‘the location of an investment project is where its “center of gravity” or “focal point” is found’.82 Thus, services (including digital ones) might therefore be deemed to be situated in the territory of the state where they are deemed to have the most impact.83 In short, there is Ibid. Abaclat and Others v Argentine Republic, ICSID Case No ARB/07/5, Dissenting Opinion of Professor Abi-Saab, available at www​.italaw​.com/​sites/​default/​files/​case​-documents/​italaw4085​.pdf accessed 4 August 2022. 78 SGS Société Génerale de Surveillance SA v Islamic Republic of Pakistan, ICSID Case No. ARB/01/13 (6 August 2003). 79 Kristian Almås and Geir Almås v Poland, UNCITRAL Award at 201 (27 June 2016). 80 Horváth and Klinkmüller, supra note 5. 81 Ibid. 82 Alpha Projektholding GmbH v Ukraine, ICSID Case No ARB/07/16, Award (8 November 2010) para 279, citing SGS v Philippines paras 101–112 and Inmaris Perestroika Sailing Maritime Services GmbH and others v Ukraine, ICSID Case No ARB/08/8, Decision on Jurisdiction (8 March 2010) paras 124–125. 83 C Schreuer and others, supra note 66, p.140. 76 77

Applying international investment agreements to digital assets  301 no single criteria or method applied to the territoriality requirement of intangible assets. This lack of clarity will be problematic if and when disputes arise involving digital assets, as the outcome will be highly unpredictable. 3.3

Summary: Can Digital Assets Be Protected Investments?

Broadly speaking, under most BITs/TIPs, there are two related requirements that must be fulfilled in order for an investment to be recognised as such. The consequence of such recognition is that the investment protections contained in the treaty would be applicable to the investor’s investment. First, it is important to note that the term investment is generally defined in the particular BIT/TIP that is applicable. Many treaties take a wide definition of the term, utilising the every asset approach. Ultimately, though, whether something is an investment will be decided according to the individual treaty text. Some ICSID tribunals have pronounced on the matter, which has resulted in the application of the widely accepted Salini test. Taking the four-prong Salini test renders matters a bit more complicated when it comes to digital assets, as it may be more difficult to satisfy some elements of the criteria, particularly the requirement to contribute to investment host state development. However, there is no doctrine of precedent in investment law, therefore the requirements laid down in Salini can be freely ignored. Therefore, there is likely no real impediment to digital assets being recognised as investments in this regard. However, digital assets would need to overcome a second obstacle, namely territoriality. When it comes to territoriality, it seems that an investor in digital assets may be able to satisfy the location requirement in a number of different ways. Although meeting this requirement may be more complex when it comes to digital assets than traditional assets/investments, it is theoretically possible to do so on the part of an investor. That said, some business entities operating in the digital economy may struggle to meet this requirement. Thus, in theory it seems that digital assets might be covered investments in most existing BITs and TIPs. The most difficult obstacle that aspiring investors will have to overcome is the territoriality requirement, which might be problematic for certain types of entities and operations. It is therefore likely that any evaluation of whether the threshold for a covered investment has been met will need to be undertaken on a case-by-case basis. This is a natural consequence of attempting to extend the application of existing BIT/TIP provisions (which were largely negotiated with traditional physical assets and bricks and mortar investments in mind) to the digital economy and digital assets. Indeed, this case-by-case approach has been adopted by many tribunals who have opted to view the operation of a business entity in a holistic manner when it comes to assessing whether or not a covered investment exists. For example, in CSOB v Slovakia: An investment is frequently a rather complex operation, composed of various interrelated transactions, each element of which, standing alone, might not in all cases qualify as an investment. Hence, a dispute that is brought before the Centre must be deemed to arise directly out of an investment even when it is based on a transaction which, standing alone, would not qualify as an investment under the

302  Research handbook on digital trade Convention, provided that the particular transaction forms an integral part of an overall operation that qualifies as an investment.84

This approach was also adopted in Joy Mining v Egypt,85 where the tribunal stated that ‘a given element of a complex operation should not be examined in isolation because what matters is to assess the operation globally or as a whole’. Many other tribunals also followed suit.86 It seems then that arbitrators have not taken a singular approach to the issue of territoriality; rather, they appear to have done a case-by-case evaluation, which often involves examining the issue alongside the wider considerations of what constitutes a protected investment. The central problem with such a holistic approach is it is likely to turn on the views of the individual tribunal members. This undermines the principle of legal certainty to a great extent, and, further, serves to elevate the parties’ choice of arbitrators in any given dispute as the pivotal decision which may completely alter the outcome of the dispute. However, unless BIT/TIP provisions are reviewed and reformed to clarify these issues, in terms of practicalities, it may be the only viable approach to determining whether a digital asset is a protected investment within the current framework of international investment law.

4.

THE DESIRABILITY OF DESIGNATING DIGITAL ASSETS AS PROTECTED INVESTMENTS

The preceding discussion has highlighted the difficulties at play in evaluating with any degree of certainty whether a digital asset may be categorised as an investment that attracts the protections enshrined in an international investment agreement. In this section we move beyond analysis of the legal status quo, in order to examine the normative and policy considerations within the international investment regime specifically related to its potential application to the regulation of digital assets. In essence, we seek to demonstrate that the current international investment regime does not provide an appropriate regulatory framework for digital assets. 4.1

Digital Assets as Protected Investments: Perpetuating Problems

In recent years the international investment regime has been at the centre of much criticism. The controversy has largely been focused on the use of Investor-State Dispute Settlement (ISDS) to resolve disputes between aggrieved investors and investment host states. Although Ceskoslovenska Obchodni Banka, AS v The Slovak Republic, ICSID Case No ARB/97/4, Decision of the Tribunal on Objections to Jurisdiction (24 May 1999) para 78. 85 Joy Mining Machinery Limited, supra note 50, para 54. 86 ADC Affiliate Limited and others v The Republic of Hungary, ICSID Case No ARB/03/16, Award (2 October 2006) para 331, Inmaris Perestroika Sailing Maritime Services GmbH and others v Ukraine, ICSID Case No ARB/08/8, Decision on Jurisdiction (8 March 2010) paras 124–125, Saipem SpA v The People’s Republic of Bangladesh, ICSID Case No ARB/05/07, Decision on Jurisdiction and Recommendation on Provisional Measures (21 March 2007) paras 110–114, Mamidoil Jetoil Greek Petroleum Products Societe Anonyme SA v Republic of Albania, ICSID Case No ARB/11/24, Award (30 March 2015) para 288, Vestey Group Ltd v Bolivarian Republic of Venezuela, ICSID Case No ARB/06/4, Award (15 April 2016) para 196, and Koch Minerals Sarl and Koch Nitrogen International Sarl v Bolivarian Republic of Venezuela, ICSID Case No ARB/11/19, Award (30 October 2017) paras 6.57–6.59. 84

Applying international investment agreements to digital assets  303 no international investment disputes have specifically involved the status of digital assets specifically, criticisms of the mechanism would be amplified significantly if such a digital asset dispute were to arise. The backlash against ISDS is essentially premised on rule of law concerns (such as consistency, predictability, correctness of outcome, independence and impartiality of arbitrators, duration and costs), as well as a purported lack of democratic legitimacy (including transparency and selection of arbitrators, and so on).87 While many of these are valid concerns, a number of commentators seem to be increasingly unconvinced that reform of the dispute resolution procedure will cure the ills associated with international investment law.88 Such commentators are actually taking aim at the foundations of the law of foreign investment itself (that is, investment agreements): ‘The central issue is whether investment treaties should exist in the form and substance they do at all.’89 The present authors tend to agree with this argument. Therefore, promoting the application (and possible expansion) of the international investment regime and its composite treaties to digital assets cannot be advocated in good conscience. The only purpose thus served would be to perpetuate a bad regime by entrenching its reach through its application to digital assets. In order to avoid a regulatory race to the bottom and competition among states through the provision of ever-increasing protections for foreign investors (including investors in digital assets), the present authors suggest that law and policy makers refrain from revising investment agreements to explicitly include digital assets. Digital assets are fundamentally different to more traditional assets; their non-tangible existence is the cause of their often flimsy link to the host state territory. This, in turn, means that the purported benefits of investment for the host state are not guaranteed (or even as likely to be realised). Thus, expanding extensive investment protections to digital assets does not make much practical sense. Although digital assets are likely to become progressively important for increasing digital trade and expanding the digital economy, expansive investment protections are not warranted. The international investment regime was not created, and international investment agreements were not negotiated, with digital assets and digital investments in mind; the regime is therefore ill-equipped to deal with the nuances and specificities associated with them. 4.2

Would International Investment Agreements Promote Investments in Digital Assets?

The ills of the international investment regime are well documented in recent literature.90 If the investment protections contained in international investment agreements are so problematic, one may legitimately ask why states choose to sign them. Traditionally, it was thought that international investment treaties serve to promote and attract foreign investments into states

87 S Schill and G Vidigal, ‘Cutting the Gordian knot: investment dispute settlement à la carte’ (2018) available at https://​uncitral​.un​.org/​sites/​uncitral​.un​.org/​files/​rta​_exchange​_​-​_investment​_dispute​ _settlement​_​-​_schill​_and​_vidigal​.pdf accessed 4 August 2022. 88 M Sornarajah, ‘Disintegration and change in the international law on foreign investment’ (2020) 23 Journal of International Economic Law 413 and J Linarelli, M Salomon and M Sornarajah, ‘Foreign investment: property, contract and protecting private power’ in J Linarelli, M Salomon and M Sornarajah, The Misery of International Law: Confrontations with Injustice in the Global Economy (Oxford University Press, 2018). 89 Ibid, Linarellli, Salomon and Sornarajah, p.173. 90 Ibid.

304  Research handbook on digital trade that sign them.91 However, empirical evidence does not bear this out. Studies have shown that investment treaties (BITs and TIPs) do not substantially affect investment flows.92 If this is the case, there is no reason to expect that the explicit application of such treaties to digital assets/investments why would attract investment in such assets, and therefore boost the digital economy of a particular state. We should be cautious in assuming that protecting digital assets as investments would promote or increase investment activity in digital assets. There is no good reason to believe that would be the case.

5. CONCLUSION This chapter has highlighted that there appears to be a common issue with anything that is prefaced by the word ‘digital’ (for example, digital economy or digital assets): that there are no universally accepted definitions of such terms. Digital versions of traditional concepts or items are relatively novel, and the novelty is evident through a lack of certainty and universality of understanding. This is problematic, particularly from a regulatory perspective, as it may lead to generalisation and over-inclusion. This is certainly the case with investment in digital assets, where we are essentially attempting to understand investments in digital assets through a regulatory regime that was, for all intents and purposes, developed to deal primarily with traditional assets. What follows then is attempting to fit a square peg into a round hole, as the analogy goes. This lack of understanding of digital assets leads to confusion about the level of protection needed and the legal and regulatory tools that should be applied to them, particularly within the context of international investment. Applying the existing provisions of international investment agreements (BITs and TIPs) to digital assets simply does not work. An overhaul of the international investment agreement regime would be required in order to ensure their smooth application to digital assets. However, such an overhaul may not be entirely appropriate. Simply revising investment agreements (specifically their definition of the term ‘investment’ to accommodate digital assets), and in doing so perpetuating the current international investment regime, is not a desirable outcome. States, stakeholders and commentators alike are increasingly critical of the international investment framework, from overprotective provisions to generous dispute settlement options. Calls for reform have centred around investment dispute resolution, with reforms to and/or replacement of investment treaty arbitration. However, many commentators are sceptical that this will address the many ills of the investment regime. Thus, reforming international investment agreements (that is, BITs and TIPs) specifically to include protection of digital assets (investments) would simply serve to enshrine problematic levels of investment protection and perpetuate the problems of the system. Given that the evidence is mixed (at best) as to whether international investment agreements actually serve

See for example Subedi, supra note 27, p.115. See also BIICL, ‘Risk and return: foreign direct investment and the rule of law’ (2015) available at https://​binghamcentre​.biicl​.org/​documents/​49​_risk​_and​_return​_fdi​_and​_the​_rol​_compressed​.pdf accessed 4 August 2022 and J Yackee, ‘Do BITs really work? Revisiting the empirical link between investment treaties and foreign direct investment’ (2007) University of Wisconsin Legal Studies Research Paper No. 1054 available at https://​papers​.ssrn​.com/​sol3/​papers​.cfm​?abstract​_id​=​1015083 accessed 4 August 2022. 91 92

Applying international investment agreements to digital assets  305 to promote or increase international investment, introducing reformed IIAs (including digital assets as investment) may not be a worthwhile endeavour. A broader re-think of the regulation of international investment may be required. However, a significantly less radical and less complex solution might be for interested states to negotiate a multilateral treaty dealing with the status of digital assets as investments, in order to clarify their position. In the same vein as the Mauritius Convention,93 states could choose to apply their network of investment agreements to digital assets or not by choosing whether to ratify the Convention. It should be noted though that this suggestion could be problematic, leading to further fragmentation of an already disjointed investment regulatory framework. What is clear is that, in the context of investment in digital assets, regulation and the scholarly debate are lagging behind real world developments. This is problematic given both the increasing importance of the digital economy and digital assets, and also the propensity for states’ vulnerability to potentially costly investment claims at the hands of would-be investors in digital assets. Indeed, Chaisse and Bauer recognise that ‘the gravity of the interpretation of the investment definition with regard to digital assets cannot be emphasized enough, as the definition is a threshold criterion and is inextricably linked to the power and force of the substantive BIT protections’.94 Accordingly, law and policy makers should mobilise in this regard; the regulatory framework governing international investments in digital assets should therefore be clarified as a matter of urgency. Whether the decision is to include or exclude digital assets in the international investment regime, there is no room for a grey area in this regard. The present authors would favour the exclusion of such assets, for the reasons elucidated above.

93 United Nations Convention on Transparency in Treaty-based Investor-State Arbitration (2014), full text available at https://​uncitral​.un​.org/​en/​texts/​arbitration/​conventions/​transparency accessed 4 August 2022. 94 Chaisse and Bauer, supra note 30.

19. Fintech: technology-enabled financial innovation for digital trade Lerong Lu

1. INTRODUCTION1 Financial technology (fintech) refers to the latest technological innovations in the financial services industry, giving rise to various new-type financial services, financial products and financial institutions that have been playing an increasingly important and indispensable role in facilitating digital trade.2 Over the past decade, the world has witnessed the advancement and mass application of key technologies in relation to artificial intelligence (AI), blockchain, cloud computing and data analytics, underpinning the digital transformation and further growth of most business sectors, including banking, finance and insurance.3 The rising trend of fintech has been reinforced by the growing BigTech companies, such as Google (Alphabet), Amazon, Apple and Facebook (Meta) in the US as well as Alibaba and Tencent in China, which constantly increase their presence in financial services by leveraging on their existing strengths in user base, data and information technologies.4 Most recently, the emergence of Metaverse, built on augmented reality (AR) and virtual reality (VR) technologies, is said to seamlessly connect our physical and virtual worlds, for it is creating a unified global community where people are able to work, play, relax, transact and socialise together.5 The number of fintech companies nearly tripled from 2018 to 2021, with a total of $254 billion of global venture capital being invested into 18,000 fintech start-ups.6 Financial institutions and financial markets, due to their data-intensive and technology-driven nature, have been the best test ground for novel technologies which could be applied promptly and extensively after their initial invention. Therefore, fintech and other forms of digital

The author would like to thank colleagues and graduate students from the Dickson Poon School of Law, King’s College London, who have participated in the LLM module – The Law and Policy of Financial Technologies (FinTech) – of which he is the co-module leader. He is grateful for the support from research assistants Shunqi Yang, Ci Ren and Hang Chen when preparing the early draft of this chapter. 2 Lerong Lu, ‘Promoting SME Finance in the Context of Fintech Revolution: A Case Study of the UK’s Practice and Regulation’ (2018) 33 Banking and Finance Law Review 317, 319. 3 Shahriar Akter et al., ‘Transforming Business Using Digital Innovations: The Application of AI, Blockchain, Cloud and Data Analytics’ (2022) 308 Annals of Operations Research 7–39. 4 Parma Bains, Nobuyasu Sugimoto, and Christopher Wilson, ‘BigTech in Financial Services (FinTech Notes No. 2022/002)’, International Monetary Fund (IMF), available at www​.imf​.org/​en/​ Publications/​fintech​-notes/​Issues/​2022/​01/​22/​BigTech​-in​-Financial​-Services​-498089, accessed 1 May 2022. 5 JPMorgan, ‘Opportunities in the Metaverse’ (2022), available at www​.jpmorgan​.com/​content/​ dam/​jpm/​treasury​-services/​documents/​opportunities​-in​-the​-metaverse​.pdf, accessed 1 May 2022. 6 Statista, ‘Number of Fintech Startups Worldwide by Region’, available at www​.statista​.com/​ statistics/​893954/​number​-fintech​-startups​-by​-region/​, accessed 1 May 2022. 1

306

Fintech  307 finance could be widely found in today’s financial world, including but not limited to banking, lending, payment, virtual currency, wealth management and insurance.7 Clearly, fintech has made the provision of financial services more efficient, customer-focused and profitable, which is beneficial for both financial institutions and their customers. Fintech has impressively changed the lifestyle of many consumers, as it offers the convenience of mobile payments such as Apple Pay, Google Pay and Alipay; the speediness of affordable peer-to-peer loans from online lending platforms such as Zopa and Funding Circle; and the anonymous, borderless, and swift money transactions involving Bitcoin, Ethereum, and other crypto currencies.8 The fintech revolution has undoubtedly transformed the outlook of the financial sector by lowering costs, improving service quality, assessing financial risks intelligently and inventing a diverse and stable credit sector.9 After one decade of rapid growth, fintech itself has become an enormous and complex ecosystem comprising various market participants, including fintech corporations, fintech consumers and fintech regulators interacting with each other in shaping the appearance and evolution of the system. According to Ernst & Young, a well-functioning fintech ecosystem involves the sound interconnection of four elements: talent, capital, policy-making and client demand.10 Obviously, fintech has produced lots of business opportunities for both existing financial institutions and tech start-ups who could exploit big data and other cutting-edge technologies to rewrite the rules in the financial industry. For instance, Ant Group, launched by Alibaba’s founder Jack Ma, has become the world’s most valuable fintech company by serving more than 1 billion consumers with an annual transaction volume of US$17 trillion, and its Alipay payment service is supporting millions of financial transactions on Alibaba’s e-commerce platforms, including Tmall and Taobao.11 Despite their enormous commercial value, innovative financial services and products could also lead to extra financial risks, posing new challenges for policy-makers and financial regulators, in particular, how to balance financial innovation with financial stability and consumer protection.12 Supporting the growth of fintech has become a key national strategy for an increasing number of governments, as fintech is said to create job opportunities, increase business efficiency and promote the competitiveness of financial industry.13

Lerong Lu, ‘Financial Technology and Challenger Banks in the UK: Gap Fillers or Real Challengers?’ (2017) 32 Journal of International Banking Law and Regulation 273, 273. 8 Ibid. 9 The Economist, ‘The Fintech Revolution’ (9 May 2015), p.13. 10 Ernst & Young, ‘FinTech and Ecosystems’, available at www​.ey​.com/​en​_gl/​banking​-capital​ -markets/​fintech​-ecosystems, accessed 1 May 2022. 11 Lerong Lu and Alice LS Zhang, ‘Regulating Fintech Corporations amidst Covid-19 Pandemic: An Analysis of Ant Group (Alipay)’s Suspension of IPO and Business Restructuring’ (2021) 42 The Company Lawyer 341, 341–2. 12 A number of jurisdictions have introduced the Regulatory Sandbox regimes to allow fintech firms to test their innovative propositions in the market with real consumers under a controlled environment to limit financial risks. See, for example, Financial Conduct Authority (UK), ‘Regulatory Sandbox’, available at www​.fca​.org​.uk/​firms/​innovation/​regulatory​-sandbox, accessed 1 May 2022. 13 See, for instance, the UK Government, ‘Kalifa Review of UK Fintech’ (February 2021); The People’s Bank of China, ‘Fintech Development Plan (2019–2021)’ (August 2019); and The Monetary Authority of Singapore, ‘FinTech and Innovation’, available at www​.mas​.gov​.sg/​development/​fintech, accessed 1 May 2022. 7

308  Research handbook on digital trade Against this background, the chapter aims to introduce and analyse the rise of fintech and explains how it supports digital economy and trade. After the introduction, Section 2 explains the terminology of fintech and how various innovative technologies have been applied in the financial industry. Section 3 examines the business organisations providing fintech services and products, especially fintech start-ups, BigTech and Metaverse firms. Section 4 assesses a number of global cities, most of which are top international trade and financial centres, as they compete to become global fintech hubs, like London, New York, Shanghai, Dubai and Mumbai. Section 5 continues to present case studies of fintech in banking and non-bank lending sectors that offer alternative financing channels for consumers and businesses. Section 6 discusses fintech in the field of monetary and payment systems, leading to the rise of crypto assets and central bank digital currencies (CBDCs). Finally, section 7 draws a conclusion.

2.

WHAT’S FINTECH? ARTIFICIAL INTELLIGENCE, BLOCKCHAIN, CLOUD AND DATA TECHNOLOGIES IN FINANCIAL INDUSTRY

Despite the global popularity of fintech, there has not yet been consensus on the definition and exact scope of fintech. According to the Cambridge English Dictionary, fintech refers to ‘the business of using technology to offer financial services in new and better ways’.14 The Financial Stability Board (FSB) regards fintech as ‘technologically enabled innovation in financial services that could result in new business models, applications, processes or products with an associated material effect on financial markets and institutions and the provision of financial services’.15 Moreover, the United Nations (UN) set up a Digital Financing Task Force with an objective to harness digitalisation in creating a citizen-centric financial system and to accelerate the financing of the UN’s Sustainable Development Goals.16 The study of fintech is of great importance for researchers, policy-makers and regulators, so it is necessary to demystify the term ‘fintech’ at first to enable readers to better understand the nature of fintech and what kind of financial activities, products, services and institutions could be caught by the definition. Therefore, this section will look at both the ‘fin’ and ‘tech’ components of the term respectively. In terms of the ‘fin’ or ‘finance’ part of fintech, we shall bear in mind that no matter how technologies evolve and innovate, the nature and basic functions of financial markets and institutions remain mostly similar. Financial activities match money supply and money demand in any economies, as they facilitate the risk management transcending time and space.17 Nonetheless, the new technologies have greatly improved the efficiency of financial activities, which increased profit margins for financial institutions and enriched consumers’

14 Cambridge English Dictionary, ‘Financial Technology (Fintech)’, available at https://​dictionary​ .cambridge​.org/​dictionary/​english/​financial​-technology, accessed 1 May 2022. 15 Financial Stability Board (FSB), ‘Fintech’, available at www​.fsb​.org/​work​-of​-the​-fsb/​financial​ -innovation​-and​-structural​-change/​fintech/​, accessed 1 May 2022. 16 United Nations (UN), ‘Digital Financing Task Force’, available at www​.un​.org/​en/​digital​ -financing​-taskforce, accessed 1 May 2022. 17 Peter Christoffersen, Elements of Financial Risk Management (2nd edition, 2012, Academic Press), p.3.

Fintech  309 choice.18 From the functional perspective, fintech is playing exactly the same role as traditional finance in our economy, despite protean forms and more efficient delivery methods. Consequently, the key to understanding fintech lies in the ‘tech’ or ‘technology’ part. In fact, the development of the financial industry has always been driven by technological advances throughout human history, such as the telegram, telephone, computer and internet. The interaction between financial services and technologies has a long history, dating back to the nineteenth century.19 A lot of common financial services that we enjoy today used to be a great invention just few decades ago. For example, in June 1967 the world’s first automated teller machine (ATM) was unveiled at a branch of Barclays Bank in London, allowing customers to withdraw cash outside of banking hours at a time when banks were only open until 3:30pm.20 The section continues to assess four important technologies that have greatly transformed the financial sector over the past decade: artificial intelligence, blockchain, cloud computing and data technology (big data). They are often collectively referred to as ‘ABCD’ as we abbreviating their initial capitals. 2.1

Artificial Intelligence

Artificial intelligence (AI) aims to simulate human intelligence in machines or robotics to enable them to think and act either humanly or rationally.21 AI represents a multi-disciplinary area of study involving computer science, mathematics, behavioural science, psychology and linguistics. Accordingly, AI can be applied in extensive business scenarios, including natural language processing, intelligent retrieval from databases, automatic consulting systems, theorem proving, robotics, automatic programming, and so on.22 The term AI was first invented in 1956 by Professor John McCarthy, who was a computer scientist at Stanford University, when he held the first academic conference on this subject.23 AI is further divided into strong AI and weak AI; the two have different goals.24 Weak AI refers to machine intelligence which simulates basic human cognition and has limited application in certain pre-defined areas. Weak AI do not have their own mind; examples are Apple’s Siri, Amazon’s Alexa and Baidu’s Xiaodu. In contrast, strong AI are created to have their own mind, as machine intelligence which could mostly match human intelligence and possess strong human cognition abilities

In most cases, fintech has lowered the threshold of accessing financial services and reduced the service costs for financial consumers. For example, small businesses and consumers could now borrow fast and affordable loans via P2P lending platforms as they might not be accepted banks due to the strict and onerous lending criteria. See Lerong Lu, ‘Solving the SME Financing Puzzle in the UK: Has Online P2P Lending Got the Midas Touch?’ (2018) 33 Journal of International Banking Law and Regulation 449. 19 Douglas W. Arner, Janos Barberis and Ross P. Buckley, ‘The Evolution of FinTech: A New Post-Crisis Paradigm?’ (2016) 47 Georgetown Journal of International Law 1271, 1273. 20 Barclays Bank, ‘From the Archives: The ATM Is 50’ (27 June 2017), available at https://​home​ .barclays/​news/​2017/​06/​from​-the​-archives​-the​-atm​-is​-50/​, accessed 1 May 2022. 21 Stuart Russell and Peter Norvig, Artificial Intelligence: A Modern Approach (4th edition, 2021, Pearson), pp.19–22. 22 Nils J. Nilsson, Principles of Artificial Intelligence (Morgan Kaufmann 2014), p.2. 23 University of Washington, ‘The History of Artificial Intelligence (December 2016)’, available at http://​courses​.cs​.washington​.edu/​courses/​csep590/​06au/​projects/​history​-ai​.pdf, accessed 1 May 2022. 24 Stanford University, ‘Artificial Intelligence (July 2018)’, available at http://​plato​.stanford​.edu/​ entries/​artificial​-intelligence/​, accessed 1 May 2022. 18

310  Research handbook on digital trade like reasoning and logical thinking. Strong AI has been a hypothetical concept so far, as we have not yet seen mature commercial products in this field. AI has been widely deployed in the financial industry, producing substantial changes to financial products and services in terms of delivery methods, risk management, credit rating, and investment decision-making. JPMorgan set up an AI Research team headquartered in New York to explore cutting-edge technologies in the fields of Machine Learning and Cryptography to develop practical solutions for the bank’s clients and its own businesses.25 AI can be found in many areas, including smart payment, insurance claims, robo-advice, customer service, smart investment research and intelligent risk control. For example, intelligent customer service has been used by an increasing number of financial institutions to provide better service for customers. The Bank of Communications has introduced the smart AI robots ‘Jiao Jiao’ as lobby managers in its branches in more than 30 Chinese cities; they are able to welcome guests, provide service guidance and check account information for customers using speech recognition, touch interaction and body language.26 Ping’an Bank has utilised AI robots to build a 24/7 customer service system, replacing 80 per cent of its customer service staff.27 According to a survey, 56 per cent of hedge funds in the US have employed AI to assist in the making of their investment decisions.28 2.2 Blockchain Blockchain refers to a decentralised public ledger system holding the database of transactional records which will be shared by network nodes, updated by miners and monitored by all participants.29 ‘Block’ stands for the digital information regarding any transactions, such as the transactional parties, date, time and price amount, while ‘chain’ means the way in which information within blocks is being stored in the database as well as how one block is linked to another. Blockchain is the underlying technology for Bitcoin, Ethereum, Ripple, Litecoin and other cryptocurrencies whose prices are highly volatile.30 As the technology enabling public record-keeping, blockchain has gained wide popularity thanks to its distinct characteristics such as decentralisation, transparency and immutability.31 The most unique feature of blockchain is that there is no one possessing central authority or ownership over the entire system, as opposed to the centralised data services that most businesses and financial institutions have 25 JPMorgan, ‘Artificial Intelligence Research’, available at www​.jpmorgan​.com/​global/​technology/​ artificial​-intelligence, accessed 1 May 2022. 26 Bank of Communications, ‘BOC News’, available at www​.bocomgroup​.com/​BankCommSite/​ shtml/​jyjr/​cn/​7158/​7162/​39814​.shtml, accessed 1 May 2022. 27 Xinhua, ‘Pingan Bank’s Financial Technology Investments Witnessed Strategic Outputs’ (9 October 2019), available at www​.xinhuanet​.com/​money/​2019​-10/​09/​c​_1125082104​.htm, accessed 1 May 2022. 28 Peter Salvage, ‘Artificial Intelligence Sweeps Hedge Funds’ (BNY Mellon, March 2019), available at www​.bnymellon​.com/​us/​en/​what​-we​-do/​business​-insights/​artificial​-intelligence​-sweeps​-hedge​ -funds​.jsp, accessed 1 May 2022. 29 Melanie Swan, Blockchain: Blueprint for a New Economy (O’Reilly, 2015), p.1. 30 Bank of England, ‘What Are Cryptoassets (Cryptocurrencies)?’, available at www​.bankofengland​ .co​.uk/​knowledgebank/​what​-are​-cryptocurrencies, accessed 1 May 2022. 31 United Nations (UN), ‘Let’s Learn Blockchain’ (11 April 2018), available at https://​unite​.un​.org/​ sites/​unite​.un​.org/​files/​technovation/​1​_blockchain​_101​_ariana​_fowler​_consensys​.pdf, accessed 1 May 2022.

Fintech  311 been using for a long time.32 Owing to blockchain’s peer-to-peer method of data storage and sharing, the information system is considered as secure, transparent and easily accessible. Moreover, the immutability of blockchain suggests that no one is able to tamper with the information once it has been entered into the system, as it uses the hashing algorithm which takes an input of information at any length and then produces an output of information at a fixed length.33 What is more, blockchain has an obvious economic advantage as relevant transactions do not incur transactional costs, but they do produce certain infrastructure costs to be borne by all users. Besides, the identities of users will be hidden by complicated cryptography so people can only see their public addresses instead of real account names, which protects the privacy of users but also makes Bitcoin a hotbed for money laundering activities.34 Blockchain can be applied across different industries and business scenarios, including banking, insurance, ticket selling, intellectual property protection, sharing economy, supply chain management, property record, smart contract and the internet of things. As blockchain enables the tamper-proof documentation of sensitive data, it is perfectly suited for making international payments and money transfers. In April 2018, Santander launched the world’s first blockchain-based international money transfer services, known as the ‘Santander One Pay FX’.35 It has made international payments more efficient and cheaper for retail customers as the number of intermediaries within the payment process has been reduced, and there is no need for banks to settle payment transactions manually. In November 2019, the US Securities and Exchange Commission (SEC) approved a pilot project led by Paxos which relies on blockchain to settle stock market transactions in a faster and cheaper way, as the current settlement standard takes two business days.36 The secure feature of blockchain renders it a useful tool for accounting and auditing firms where exists high regulatory standards regarding validity and integrity, for it allows companies to record their transactions directly into a joint register, creating an interlocking system of enduring accounting records and reducing the possibility of human errors and deliberate falsifying of accounting data.37

32 By using centralised data services, businesses keep the information of all users through one set of data servers. Each user has to interact with the sole centralised body to submit and retrieve information whenever and wherever they need it. For instance, banking institutions represents the centralised money system where their customers need to deposit and withdraw cash through banks’ official branches. 33 University of Missouri – St. Louis, ‘Hashing Functions and Their Uses In Cryptography’, available at www​.umsl​.edu/​~siegelj/​information​_theory/​projects/​HashingFun​ctionsInCr​yptography​.html, accessed 1 May 2022. 34 Lerong Lu, ‘Bitcoin: Speculative Bubble, Financial Risk and Regulatory Response’ (2018) 33 Butterworths Journal of International Banking and Financial Law 178, 180. 35 Santander, ‘Santander Launches the First Blockchain-Based International Money Transfer Service across Four Countries’ (April 2018), available at www​.santander​.com/​content/​dam/​santander​-com/​en/​ documentos/​historico​-notas​-de​-prensa/​2018/​04/​NP​-2018​-04​-12​-Santander​%20launches​%20the​%20first​ %20blockchain​-based​%20international​%20money​%20transfer​%20service​%20across​%20​-en​.pdf, accessed 1 May 2022. 36 Alexander Osipovich, ‘Blockchain Makes Inroads into the Stock Market’s $1 Trillion Plumbing System’, Wall Street Journal (7 November 2019), available at www​.wsj​.com/​articles/​blockchain​-makes​ -inroads​-into​-the​-stock​-markets​-1​-trillion​-plumbing​-system​-11573131600, accessed 1 May 2022. 37 Deloitte, ‘Blockchain Technology: A Game-Changer in Accounting?’ (2016), available at www2​ .deloitte​.com/​content/​dam/​Deloitte/​de/​Documents/​Innovation/​Blockchain​_A​%20game​-changer​%20in​ %20accounting​.pdf, accessed 1 May 2022.

312  Research handbook on digital trade 2.3

Cloud Computing

Cloud computing refers to the supply of computing services, including servers, storage, databases, networking, software, analytics and intelligence over the internet (‘the cloud’), to offer faster innovation, flexible resources and economies of scale.38 It represents a major transformation of how modern businesses organise their IT resources, as in the past, businesses had to set up their on-site data centres, requiring heavy investments in hardware, software and daily infrastructure management. In contrast, cloud computing allows enterprises to directly use professional cloud services provided by IT companies like Microsoft and Amazon, which is likely to save costs and increase the capacity and speed of data storage. A study suggested that the use of cloud computing could save 50–67 per cent of the lifecycle cost for a 1,000-server deployment.39 Cloud computing is often provided as on-demand services, so businesses are allowed to request any amount of computing power within a few seconds, meeting their elastic needs of computing capacity. Cloud computing services are broadly divided into three categories: Software as a Service (SaaS), Platform as a Service (PaaS) and Infrastructure as a Service (IaaS).40 They can be categorised into public cloud, private cloud and hybrid cloud, depending on the place where business users store their data.41 Some well-known public cloud providers include Microsoft Azure, Amazon Web Services (AWS), Google Cloud Platform, AliCloud, Tencent Cloud and Baidu Cloud. The application of cloud computing in financial services is likely to raise productivity for financial institutions and to provide better and cheaper products and services for consumers. Migrating to the cloud is recommended for most financial institutions as cloud computing resources bring benefits such as agile innovation, risk mitigation, cost reduction and computing solutions for multinational companies.42 The cloud transformation also allows financial institutions to synchronise the whole enterprise by actively integrating different business units through data sharing, so as to have strong analytical power and to make fast and uniform decisions.43 So far, the financial industry has already adopted a variety of cloud services and models to suit their specific needs. For example, there are banking-focused cloud services such as the core banking cloud service, lending and leasing cloud service and billing cloud service, as well as insurance-focused cloud services such as the annuity cloud service, health insurance

38 Microsoft, ‘What Is Cloud Computing? A Beginner’s Guide’, available at http://​azure​.microsoft​ .com/​en​-gb/​overview/​what​-is​-cloud​-computing/​, accessed 1 May 2022. 39 Kevin L. Jackson, ‘The Economic Benefit of Cloud Computing’, Forbes (17 September 2011), available at www​.forbes​.com/​sites/​kevinjackson/​2011/​09/​17/​the​-economic​-benefit​-of​-cloud​-computing/​ , accessed 1 May 2022. 40 IBM, ‘IaaS, PaaS and SaaS Cloud Service Models’, available at www​.ibm​.com/​cloud/​learn/​iaas​ -paas​-saas, accessed 1 May 2022. 41 Alibaba Cloud, ‘What Is the Difference among Public Cloud, Private Cloud and Hybrid Cloud?’, available at www​.alibabacloud​.com/​knowledge/​public​-cloud​-private​-cloud​-hybrid​-cloud, accessed 1 May 2022. 42 British Bankers Association (BBA), ‘Banking on Cloud: A Discussion Paper by the BBA and Pinsent Masons (December 2016)’, p.3; and Institute of International Finance, ‘Cloud Computing in the Financial Sector (August 2018)’, pp.3–4. 43 Deloitte, ‘Cloud Computing: More than Just a CIO Conversation (2019)’, available at www2​ .deloitte​.com/​content/​dam/​Deloitte/​global/​Documents/​Financial​-Services/​gx​-cloud​-banking​-2030​-fsi​ .pdf, accessed 1 May 2022.

Fintech  313 cloud service, and rating and underwriting cloud service.44 Cloud computing also allows fintech platforms such as P2P lenders to better assess the credit profile of certain borrowers who might be refused by mainstream banks, which improves their chance of obtaining a loan at a fair price, reflecting the real credit risks.45 2.4

Data (Big Data)

Big data describes the large volume of data being collected, processed and utilised by modern businesses in their daily operation.46 It requires the production and accumulation of large amounts of data from various sources such as personal computers, smartphones, tablets and other electronic devices. The data will then be processed by complicated algorithms and analytical tools to extract valuable information for any business purpose. Nowadays, internet companies and tech giants are able to gather massive amounts of information from users. Facebook had 2.9 billion monthly active users as of the first quarter of 2022.47 The e-commerce giant Alibaba and its affiliated Taobao and Tmall online shopping platforms serve 700 million consumers.48 Obviously, big data is beneficial for most modern corporations as it reduces the time and monetary costs of business operation and helps businesses to make smarter decisions based on precise data analysis. Big data is also advantageous to financial consumers and investors, since more tailored products and services can be created according to their real needs or risk appetites. With the aid of big data, financial corporations are able to provide better and bespoke services for consumers. In total 62 per cent of businesses believe that big data has significant potential to create competitive advantages.49 Big data assists financial institutions in having more precise market segmentation that divides customers into sub-groups based on their information, including gender, age, marital status, occupation, income and location. Depending on the classification, banks and fintech companies are able to predict customers’ demands and offer loan and investment products that suit their real needs. Barclays Bank, for example, has around 17 million customers in the UK, and it can rely on customers’ data from their current accounts and credit card transactions to identify market trends, growth opportunities and

Oracle, ‘Cloud Computing In Financial Services: A Banker’s Guide’ (November 2015), p.4. Thomas Hale, ‘How Big Data Really Fits into Lending’, Financial Times (13 March 2019), available at https://​ftalphaville​.ft​.com/​2019/​03/​13/​1552488421000/​How​-big​-data​-really​-fits​-into​-lending/​, accessed 1 May 2022. 46 The concept of big data encompasses structured, semi-structured and unstructured information including consumers’ demographic and psychographic information, product reviews and commentaries, blogs, social media and data streamed 24/7 from mobile devices, sensors and technical devices. See PwC, ‘Where Have You Been All My Life? How the Financial Services Industry Can Unlock the Value in Big Data’ (October 2013), available at www​.pwc​.com/​us/​en/​financial​-services/​publications/​viewpoints/​ assets/​pwc​-unlocking​-big​-data​-value​.pdf, accessed 1 May 2022. 47 Statista, ‘Number of Monthly Active Facebook Users Worldwide’, available at www​.statista​.com/​ statistics/​264810/​number​-of​-monthly​-active​-facebook​-users​-worldwide/​, accessed 1 May 2022. 48 Alibaba, ‘Alibaba Group Announces December Quarter 2018 Results’, available at www​ .alibabagroup​.com/​en/​news/​press​_pdf/​p190130​.pdf, accessed 1 May 2022. 49 PwC, ‘Where Have You Been All My Life? How The Financial Services Industry Can Unlock The Value In Big Data’ (October 2013), available at www​.pwc​.com/​us/​en/​financial​-services/​publications/​ viewpoints/​assets/​pwc​-unlocking​-big​-data​-value​.pdf, accessed 1 May 2022. 44 45

314  Research handbook on digital trade customer behaviours based on actual spending.50 As more personalised services are offered to customers, financial institutions are expected to gain higher customer satisfaction and increase their profits level. In all, 84 per cent of US executives expressed that their companies had discovered a trend towards customers looking for a more individualised and tailored experience, and the same survey suggested that an additional 18 per cent of annual revenue would be created if a personalised experience was offered to customers. 51 Furthermore, big data can enhance the fraud detection mechanism for financial companies. JPMorgan Chase has relied on big data to spot fraudsters trying to hack into client accounts or cash machines, and the bank also used the technology to prevent terrorist activities and spot fraud risks among its own 250,000 employees.52 Finally, big data lies at the foundation of regulatory technology (Regtech), which offers technologically advanced solutions to help financial institutions cope with onerous regulatory demands and improve the efficiency of financial compliance.53 In a word, big data and other fintech have been transforming most aspects of the financial services industry, for the benefit of consumers in digital trade, by delivering faster and better quality results with lower expenditure.

3.

FINTECH, BIGTECH AND METAVERSE: HOW TECH CORPORATIONS CONTRIBUTE TO DIGITAL ECONOMY

Fintech has seen explosive growth since the late 1990s, when the internet and e-commerce businesses started to prevail. At present, it is widely used in the backend systems of financial institutions, resulting in the digitalisation of financial transactions that optimises information processing and data analysis. It in turn facilitates digital trade and its finance, which ultimately promotes the development of the real economy. For example, fintech could reduce the volume of documents and streamline their flow in trade transactions, thus reducing transaction costs, speeding up transaction timeframes and contributing to greater transparency and security for contracting parties.54 Fintech has been impacting the development of digital economy on several fronts. On the one hand, it promotes the operational efficiency of traditional financial institutions that provides direct financial support for digital economy; on the other hand, the technological progress has expanded application scenarios of finance, that act as an external driving force for various industries to achieve innovative development indirectly.55 The rapid growth of

Barclays, ‘Insights by Barclays’, available at http://​insights​.uk​.barclays/​, accessed 1 May 2022. Oracle, ‘The Era I Enterprise: Ready for Anything’ (April 2016), available at www​.oracle​.com/​us/​ industries/​oracle​-era​-ready​-anything​-2969053​.pdf, accessed 1 May 2022. 52 Richard Waters, ‘Counter-Terrorism Tools Used to Spot Fraud’ Financial Times (13 December 2012), available at www​.ft​.com/​content/​796b412a​-4513​-11e2​-838f​-00144feabdc0, accessed 1 May 2022. 53 Douglas W. Arner, Janos Barberis and Ross P. Buckley, ‘FinTech, RegTech and the Reconceptualization of Financial Regulation’ (2017) 37 Northwestern Journal of International Law and Business 371, 377. 54 Tarsem Bhogal and Arun Trivedi, International Trade Finance A Pragmatic Approach (2019 Springer), p. 303. 55 Chris Brummer and Yesha Yadav, ‘Fintech and the Innovation Trilemma’ (2019) 107 Georgetown Law Journal 235. 50 51

Fintech  315 fintech has increased competition from non-financial institutions such as third party payment institutions and internet corporations for incumbent institutions which also face peer pressure among each other to upgrade their financial infrastructure.56 Clearly, fintech companies, like neo-banks and P2P lending platforms, are offering new solutions for customers who are unable to access services or get loans from banks with no or limited credit scores.57 Take Ant Group, one of the largest fintech companies in the world, as an example. In 2017, Ant Group handled more payments than Mastercard by completing more than $8 trillion worth of transactions through its digital payment platforms, equivalent to more than twice the gross domestic product (GDP) of Germany.58 By providing full-scale financial services, including money market funds, insurance, a credit rating system and personal loans, Ant Group provides convenient and affordable financing services for more than one billion customers.59 Such external environment has forced traditional financial institutions to either collaborate with new fintech firms or develop their own fintech solutions that integrate cutting-edge internet technologies and business ideas into their existing product portfolios. In total 48 per cent of financial institutions have embedded fintech into their strategic operating model, and 37 per cent have incorporated emerging technologies into their products and services.60 In 2019, JP Morgan invested $25 million in fintech start-ups, Capital One created technology-focused banking cafes, and Citi launched the Citi Developer Hub to invite third party programmers to test on their application programming interfaces (APIs).61 The involvement of BigTech companies in fintech activities further boosts economic efficiency and digital trade. BigTech refers to giant tech companies, such as Amazon, Apple, Facebook (Meta) and Google (Alphabet), the concept of which not only implies the size of such companies but also emphasises their enormous influence on people’s lives and work.62 For instance, Apple launched a credit card with Goldman Sachs in 2019 that has eliminated processing fees and comes with an extra layer of protection of privacy and security.63 Google stepped up its fintech ambition by launching a checking account product with Citigroup in

Yi-An Lai, ‘Alipay and the Impact of E-Payment Systems Resulting in New Regulations in China and Other Jurisdictions’, International Bar Association, available at www​.ibanet​.org/​article/​5D63B47D​ -B8C6​-47E9​-92ED​-D32EEDEB00F8, accessed 1 May 2022. 57 For detailed analysis on how alternative finance facilitates digital trade by serving more financial consumers, see this chapter’s section 5. 58 Stella Xie, ‘Jack Ma’s Giant Financial Startup Is Shaking the Chinese Banking System’ The Wall Street Journal (2018), available at www​.wsj​.com/​articles/​jack​-mas​-giant​-financial​-startup​-is​-shaking​ -the​-chinese​-banking​-system​-1532885367, accessed 1 May 2022. 59 Lerong Lu, ‘How a Little Ant Challenges Giant Banks? The Rise of Alipay (Ant Financial)’s FinTech Empire and Regulatory Options’ (2018) 29 International Company and Commercial Law Review 12, 18–23. 60 PwC, ‘Crossing the Lines: How Fintech Is Propelling FS and TMT Firms Out of Their Lanes (Global Fintech Report 2019)’, available at www​.pwc​.co​.uk/​financial​-services/​fintech/​assets/​global​ -fintech​-report​-2019​.pdf, accessed 1 May 2022. 61 Stephanie Walden, ‘What Is Fintech and How Does It Affect How I Bank?’ Forbes (2020), available at www​.forbes​.com/​advisor/​banking/​what​-is​-fintech/​, accessed 1 May 2022. 62 Miguel de la Mano and Jorge Padilla, ‘Big Tech Banking’ (2018) 14 Journal of Competition Law and Economics 494–526. 63 Goldman Sachs, ‘Goldman Sachs Partners with Apple on a Game-Changing Credit Card’ (2019), available at www​.goldmansachs​.com/​our​-firm/​history/​moments/​2019​-apple​-card​.html, accessed 1 May 2022. 56

316  Research handbook on digital trade 2020, which is available on the Google Pay app.64 Amazon has been offering payment and lending services to its more than 100 million Prime customers.65 Based on their large global user base and leading technology reserves, BigTech corporations are able to achieve economies of scale quickly and are well positioned to provide financial services. Most recently, the concept of the Metaverse and its application in several commercial scenarios has gained popularity around the world, which has profound impacts on digital trade.66 In October 2021 Facebook announced its plan to change the company’s name to Meta, intending to bring the Metaverse to life and ‘help people connect, find communities, and grow business’.67 In fact, Facebook’s Metaverse is not a new concept: the concept of the Metaverse firstly appeared in the science fiction novel Snow Crash in 1992, and its prototypes have existed in some online games and chat communities for a long time.68 The Metaverse is mostly about virtual experiences aided by hardware such as digital glasses, headsets and contact lenses, supported by technologies including virtual reality (VR) and augmented reality (AR). It allows the convergence of our life in both digital and physical worlds regarding wealth, socialisation, productivity, shopping and entertainment.69 The Metaverse is a fusion of multiple technologies regarding the internet, blockchain, 5G, digital currency and AI, and its influence is likely to far exceed that of the internet. If the internet is something that people can browse, then the Metaverse is where people can live, play and work in virtual communities such as Decentraland, Somnium Space and Sandbox. The cornerstone of the Metaverse economy is non-fungible tokens (NFTs), a record of digital ownership stored in the blockchain which authenticates individuals’ possession of property and identity.70 Metaverse has been transforming business models and lifestyles in the digital age. Some real estate investment trusts (REITs) are finding opportunities in the Metaverse. In June 2021, Republic Realm, a digital property investment fund, bought a parcel of land in Decentraland for more than $900,000 and then turned the plot into a virtual mall called Metajuku, modelled after Harajuku in

Paresh Dave and Anna Irrera, ‘Google Redesigns Pay App, Opens Waitlist for Bank Accounts with Citi’ Reuters (2020), available at www​.reuters​.com/​article/​us​-alphabet​-google​-payments​ -idINKBN27Y2LH, accessed 1 May 2022. 65 Graham Ruddick, ‘Amazon Starts Offering Loans to Customers with Pay Monthly Option’ The Guardian (2015), available at www​.theguardian​.com/​business/​2015/​dec/​31/​amazon​-loans​-customers​ -pay​-monthly​-option, accessed 1 May 2022. 66 Metaverse has become an strategic business segment of several BigTech companies including Facebook, Microsoft, Alibaba and Tencent, as they try to reinforce their industry status and influence in digital economy. 67 Meta (Facebook), ‘Introducing Meta: A Social Technology Company’ (2021), available at https://​ about​.fb​.com/​news/​2021/​10/​facebook​-company​-is​-now​-meta/​, accessed 1 May 2022. 68 Neal Stephenson, Snow Crash (Re-issue edition, 2011 Penguin). 69 Peter Allen Clark, ‘The Metaverse Has Already Arrived. Here’s What that Actually Means’ TIME (2021), available at https://​time​.com/​6116826/​what​-is​-the​-metaverse/​, accessed 1 May 2022. 70 Oleg Fonarov, ‘What Is the Role of NFTs in the Metaverse?’ Forbes (2022), available at www​ .forbes​.com/​sites/​forbestechcouncil/​2022/​03/​11/​what​-is​-the​-role​-of​-nfts​-in​-the​-metaverse/​, accessed 1 May 2022. 64

Fintech  317 Tokyo.71 Even some governments have shown their interest in increasing their presence in the Metaverse: Barbados planned to open a diplomatic embassy in the online world Decentraland.72

4.

GLOBAL FINTECH HUBS AND NEW FINANCIAL INFRASTRUCTURE FOR DIGITAL TRADE

So far, fintech and Metaverse firms, along with more established BigTech corporations, have been concentrated in a small number of global cities, given their unparalleled strengths in location, talent, capital, education and technology. This section assesses the competitive advantages of those global cities that are likely to become international and regional fintech hubs in the twenty-first century. 4.1

New York, USA

New York enjoys a leading position in the fintech industry, being home to 690 Fintech companies.73 New York has been a top international financial centre since the twentieth century. Access to finance has been identified as the most important factor to generate a competitive environment for fintech providers, which is an obvious strength of the city.74 With two of the largest stock exchanges in the world (that is, the New York Stock Exchange and Nasdaq) as well as powerful financial institutions on Wall Street, New York is considered the prime financial centre of the world.75 Education is another vital factor for New York to cultivate potential fintech talents. There are more than 120 higher education institutions in New York, including Columbia University, Cornell University, New York University and others. New York has displayed its strength in business-to-business (B2B) fintech, cryptocurrencies, lending and wealth management. Some famous fintech companies in New York include Policygenius (consumer web and financial services), Anchorage Digital (blockchain and financial services), Riskified (big data), Enova (artificial intelligence), Simon (cloud) and Paddle (information technology). 4.2

San Francisco, USA

San Francisco is also a top-tier player in the fintech industry. The ability to provide information and communication technology infrastructure and systems has been considered as the

Saniya More, ‘Decentraland Estate that Sold for $913,000 Will Become a Virtual Shopping District’ The Block (2021), available at www​.theblockcrypto​.com/​post/​109960/​decentraland​-estate​-that​ -sold​-for​-913000​-will​-become​-a​-virtual​-shopping​-district, accessed 1 May 2022. 72 Jim Wyss, ‘Barbados Is Opening a Diplomatic Embassy in the Metaverse’ Bloomberg (2021), available at www​.bloomberg​.com/​news/​articles/​2021​-12​-14/​barbados​-tries​-digital​-diplomacy​-with​ -planned​-metaverse​-embassy, accessed 1 May 2022. 73 Built in NYC, ‘690 Fintech Companies in NYC’, available at www​.builtinnyc​.com/​companies/​ type/​fintech​-companies​-nyc, accessed 1 May 2022. 74 Long Finance, ‘Global Financial Centres Index 31’ (March 2022), available at www​.longfinance​ .net/​media/​documents/​GFCI​_31​_Report​_2022​.03​.24​_v1​.0​.pdf, accessed 1 May 2022. 75 Statista, ‘Largest Stock Exchange Operators Worldwide’, available at www​.statista​.com/​statistics/​ 270126/​largest​-stock​-exchange​-operators​-by​-market​-capitalization​-of​-listed​-companies/​, accessed 1 May 2022. 71

318  Research handbook on digital trade most important element for propelling the development of fintech.76 This is the exact strength of San Francisco, where Silicon Valley is located: it is home to thousands of US and global technology companies, leading to a robust fintech ecosystem in the city. It houses some of the most valuable fintech start-ups, such as Stripe (online payment platform), Coinbase (blockchain and cryptocurrency exchange) and SpotOn (E-commerce), as well as the headquarters of multinational technology companies such as Apple, Google (Alphabet), Facebook (Meta), Wells Fargo and Visa. In terms of educational resources, San Francisco houses a number of globally prestigious universities, such as Stanford University; University of California, Berkeley; and the University of San Francisco. 4.3

London, UK

London is a top financial centre in Europe and globally, often ranked first among global cities in attracting capital, enterprise and people.77 There are more than 1,600 fintech companies in the UK, raising £20.8bn in equity funding in total.78 London has established itself as Europe’s fintech capital with 17 unicorn companies, including Monzo and Revolut as great examples. London’s fintech industry is known for digital banking, blockchain and digital currencies, alternative lending and insurtech. The UK government’s fintech review measured the UK’s success as a centre for finance, innovation and technology in five dimensions: skills; investment; policy and regulation; national connectivity; and international competitiveness.79 With world-class educational establishments including universities such as Imperial College London, London School of Economics and Political Science, University College London and King’s College London, the city has a high concentration of fintech talents and experts. London also has the well-recognised London Stock Exchange and so the city’s fintech sector can easily access global capital markets. According to a survey, the UK’s financial regulatory regime, including its Regulatory Sandbox, is considered the most fintech-friendly in the world, which is said to encourage innovation and competition while minimising financial risks.80 4.4

Paris, France

Paris performs well in its fintech industry and possesses a sound ecosystem for fintech start-ups. It is home to Euronext, the EU’s largest stock exchange in terms of market capitalisation and value of equity trading, creating an ideal financing environment for its fintech 76 Long Finance, ‘Global Financial Centres Index 31’ (March 2022), available at www​.longfinance​ .net/​media/​documents/​GFCI​_31​_Report​_2022​.03​.24​_v1​.0​.pdf, accessed 1 May 2022. 77 See, for example, Globalization and World Cities Research Network (GaWC), ‘Classification of Cities 2020’, available at www​.lboro​.ac​.uk/​microsites/​geography/​gawc/​world2020​.html; and Mori Memorial Foundation, ‘Global Power City Index’ (November 2021), available at https://​mori​-m​ -foundation​.or​.jp/​pdf/​GPCI2021​_summary​.pdf, accessed 1 May 2022. 78 Department for International Trade (UK), ‘UK FinTech: State of the Nation’, available at https://​ assets​.publishing​.service​.gov​.uk/​government/​uploads/​system/​uploads/​attachment​_data/​file/​801277/​UK​ -fintech​-state​-of​-the​-nation​.pdf, accessed 1 May 2022. 79 UK Government, ‘Kalifa Fintech Review Final Report’ (April 2021), available at https://​www​.gov​ .uk/​government/​publications/​the​-kalifa​-review​-of​-uk​-fintech, accessed 1 May 2022. 80 Caroline Binham, ‘UK Regulators Are the Most Fintech Friendly’, Financial Times (12 September 2016), available at www​.ft​.com/​content/​ff5b0be4​-7381​-11e6​-bf48​-b372cdb1043a, accessed 1 May 2022.

Fintech  319 businesses.81 Paris has established extensive collaboration with international fintech firms and regulatory authorities by hosting the ‘Paris Fintech Forum’, an annual forum for digital finance and fintech.82 Paris has some top-tier financial engineering programmes in prestigious universities such as HEC Paris and ESSEC Business School, which provide the city with finance and tech talents. Some well-known fintech companies in Paris include Le Swave (innovation fintech platform), Alan (digital health insurance platform), Kayrros (data analytics), Payfit (payroll) and Ledger (cryptocurrencies platform). 4.5

Shanghai, China

Shanghai, as an international financial centre in China and East Asia, is taking the lead in fintech innovation and business growth. Shanghai has built its unique fintech advantages through a combination of elements, such as its good geographical location in the region, a robust expansion of e-commerce and a supportive regulatory environment.83 As an existing financial and commercial centre, Shanghai presents fintech providers with sufficient capital, a vibrant financial ecosystem and skilled expertise. Shanghai features strongly in digital banking, blockchain, artificial intelligence and cloud services. There are 288 fintech start-ups in Shanghai, such as Lufax (P2P lending platform), Wind Information (online financial data and information portal), XTransfer (cross-border payment platform), IceKredit (credit assessment) and so on. 4.6

Hong Kong SAR, China

Already a prominent financial and commercial hub in Asia, Hong Kong is also emerging as a global fintech hub. It is home to more than 600 fintech companies and 12 unicorns, like Airwallex (online payments), TNG (e-wallet) and WeLab (virtual banking).84 Consumers in Hong Kong are very willing to try fintech services, with an application rate of 67 per cent – higher than that in the US.85 The local fintech community has been nurtured by two government-led innovation and technology flagship programmes in Cyberport and Science Park for new enterprise. Hong Kong has recognised the significance of obtaining sufficient funding for fintech business and its monetary authority, the HKMA, has launched the Fintech Supervisory Sandbox 3.0 and provides up to HK$1 million funding to local tech companies.86 Hong Kong also emphasises the importance of attracting talent in building the fintech industry, as the government has introduced a series of talent schemes to welcome immigration.

81 Euronext, ‘Euronext Paris’, available at www​.euronext​.com/​en/​markets/​paris, accessed 1 May 2022. 82 Paris Fintech Forum, ‘Paris Fintech Forum Communities’, available at https://​members​ .parisfintechforum​.com/​, accessed 1 May 2022. 83 Shanghai Municipal Government, ‘Shanghai Finance’, available at http://​en​.jrj​.sh​.gov​.cn/​, accessed 1 May 2022. 84 Hong Kong Government, ‘New Initiatives Set to Boost Fintech’ (4 November 2021), available at www​.news​.gov​.hk/​eng/​2021/​11/​20211104/​20211104​_161538​_408​.html, accessed 1 May 2022. 85 Ibid. 86 Hong Kong Monetary Authority, ‘Fintech Supervisory Sandbox (FSS)’, available at www​ .hkma​.gov​.hk/​eng/​key​-functions/​international​-financial​-centre/​fintech/​fintech​-supervisory​-sandbox​-fss/​ , accessed 1 May 2022.

320  Research handbook on digital trade 4.7

Dubai, UAE

Dubai, an international financial centre, has developed a dynamic fintech sector with an annual growth rate of 30 per cent in the region since 2017 and is noted for its world-class regulatory environment in fintech.87 The Dubai Financial Services Authority (DFSA) aims to ensure that the regulatory system is sound, stable, secure and growth-oriented for technological and innovative businesses.88 The fintech sectors most common in the city include payments and remittances, insurance, digital banking, crowdfunding and cryptocurrency. There are 515 fintech start-ups in Dubai, including some well-known companies such as TSLC (digital banking), Postpay (online financing), Bayzat (insurtech) and BitOasis (cryptocurrency). 4.8

Mumbai, India

Mumbai, as the commercial and financial centre in India, has been making its way to a top fintech hub in recent years, thanks to the country’s fast-growing economy and consumer market. In 2021, India is said to have the third largest fintech ecosystem worldwide with a market size of $31bn.89 There are 1,040 fintech companies in Mumbai, including Acko (insurance and banking infrastructure), SBI Payment Services (online payment), InCred (lending) and Fino Paytech (banking services). Mumbai’s fintech sector is particularly strong in fields such as payment and remittance, personal finance, lending and equity funding. The Mumbai Fintech Hub is an official investment platform, initiated by the Government of Maharashtra to further promote the development of fintech ecosystem.90

5.

ALTERNATIVE FINANCE FOR DIGITAL TRADE: DIGITAL CHALLENGER BANKS AND ONLINE P2P LENDING PLATFORMS

Prior to the global financial crisis of 2007–8, consumers and businesses had been mostly reliant on traditional financial institutions such as banks to raise funds if there was a financing need.91 However, in the post-crisis era, the world has seen the rise of various alternative financing methods and distributive channels emerging outside the regulated bank sector and capital

87 Jackson Mueller and Michael S. Piwowar, ‘The Rise of FinTech in the Middle East’ (Milken Institute, September 2019), available at https://​milkeninstitute​.org/​sites/​default/​files/​reports​-pdf/​FinTech​ %20in​%20the​%20Middle​%20East​-FINAL​-121119​.pdf, accessed 1 May 2022. 88 Dubai Financial Services Authority (DFSA), ‘Innovation’, available at www​.dfsa​.ae/​innovation, accessed 1 May 2022. 89 Ishwari Chavan, ‘India’s Fintech Market Size at $31 Billion in 2021, Third Largest in World’ India Times (10 January 2022), available at https://​bfsi​.economictimes​.indiatimes​.com/​news/​fintech/​ indias​-fintech​-market​-size​-at​-31​-billion​-in​-2021​-third​-largest​-in​-world​-report/​88794336, accessed 1 May 2022. 90 Government of Maharashtra (India), ‘Mumbai Fintech Hub’, available at https://​fintech​ .maharashtra​.gov​.in/​, accessed 1 May 2022. 91 Some large corporations, especially those publicly listed in New York, London, Hong Kong and other major stock exchanges, could have access to global capital markets to obtain funding as they are able to issue shares and bonds to public investors.

Fintech  321 markets.92 The growth of alternative finance has been largely attributed to the proliferation of fintech enabling entrepreneurs to offer low-cost and convenient financial services that could easily reach millions of consumers via internet or mobile networks (such as 3G, 4G and, more recently, 5G). This section mainly introduces two alternative financing methods for digital trade: digital-based challenger banks and online peer-to-peer (P2P) lending platforms. An increasing number of digital-based banks, equipped with fintech, have been established in recent years with an objective to challenge dominant high street lenders for a larger proportion of market share.93 Digital banks operate without physical branches as customers will access financial services via online websites or smartphone apps. Some famous examples of digital banks include Monzo, Revolut and Starling in the UK, Axos Bank and Varo Bank in the US, N26 in Germany and WeBank in China. The term ‘challenger banks’ has been widely used in the UK as such digital lenders are grabbing market share from incumbent banks such as the Big Five (HSBC, Barclays, Lloyds Banking Group, Royal Bank of Scotland and Santander), which hold 85 per cent of personal current accounts in the country.94 The digital banks, aided by cutting-edge information technology, have got rid of conventional branch networks and built a simple, straightforward and low-cost business model. They have been offering financial products and services for customers underserved by established banks, including small and medium-sized businesses (SMEs). Digital banks have grown their businesses rapidly: their total lending volume in the UK increased by 31.5 per cent in 2015, compared with a decline of 4.9 per cent for the Big Five.95 Digital challenger banks were an immediate success after the global financial crisis as they are said to better serve customers in the digital age, for the following three reasons. First, during the crisis we saw the failure of some well-established banking institutions, such as Lehman Brothers in the US and the Royal Bank of Scotland in the UK. The crisis and other financial scandals, such as Libor manipulation and the mis-selling of payment protection insurance and interest rate swaps, greatly undermined people’s trust in the traditional banking industry.96 A survey suggested that only 16 per cent of people tend to agree that ‘banks generally provide good quality products and services which are sold responsibly’, 83 per cent believe that ‘bankers are greedy and get paid too much’ and 80 per cent say ‘banks aren’t doing enough to support the economy’.97 Traditional banks suffered severe reputational damage after the crisis, while digital challenger banks were able to create new brands to distinguish themselves from 92 Lerong Lu, ‘Promoting SME Finance in the Context of Fintech Revolution: A Case Study of the UK’s Practice and Regulation’ (2018) 33 Banking and Finance Law Review 317, 321. 93 Such digital-based challenger banks are often referred to as neo-banks, virtual banks, digital banks or online-only banks, and the concept sometimes includes other digital platforms that have not obtained a banking licence in a given jurisdiction but play a similar role as banks in the economy like taking deposits, making loans, and forming the payment network. See Lerong Lu, ‘Financial Technology and Challenger Banks in the UK: Gap Fillers or Real Challengers?’ (2017) 32 Journal of International Banking Law and Regulation 273, 276. 94 Elaine Moore, ‘Challengers Line Up to Take on the Big Banks’, Financial Times (14 July 2012), p.3. 95 Emma Dunkley, ‘Challengers Prise Open Grip of Larger Rivals’, Financial Times (4 May 2016), p.23. 96 Adam Jones and Jennifer Thompson, ‘RBS Prepares for Libor Settlement Talks’, Financial Times (3 November 2012), p.10. 97 YouGov, ‘Public Trust in Banking’ (April 2013), p.16, available at http://​cdn​.yougov​.com/​ cumulus​_uploads/​document/​ylf7gpof19/​Public​_Trust​_in​_Banking​_Final​.pdf, accessed 1 May 2022.

322  Research handbook on digital trade existing lenders, making them more appealing to consumers. Second, digital banks are said to accommodate the financing needs of certain customers who have limited access to banks, such as SMEs and start-ups that could not pass the strict lending criteria of mainstream lenders or provide valid collateral. According to the Bank of England, SMEs received only 37 per cent of total bank credits in the UK, with 63 per cent of bank loans being given to large corporations.98 Therefore, some challenger banks focus their businesses on SME financing, filling a major market gap. Other digital banks target the buy-to-let property market by providing mortgage loans for property investors. In 2016, Aldermore originated £1 billion buy-to-let loans, seeing 25 per cent growth from £0.8 billion in 2015.99 Most digital banks try to offer differentiated products and services to win market share. Third, digital banks have used new methods of distribution that is more cost-saving and effective. As virtual banks, they have been using brokers, online platforms or smartphone apps to reach customers, instead of building an expensive network of brick-and-mortar branches as traditional banks do. For instance, the publicly listed Shawbrook Bank does not have a single branch and only employs 569 member staff.100 Digital banks, therefore, do not have geographical limitations for doing businesses, and can expand their customer base quickly without hefty infrastructure investments. Online peer-to-peer (P2P) lending platforms have been another popular alternative financing method for modern consumers and businesses in the digital economy. P2P lending refers to the practice of lending to unrelated individuals or businesses through online portals, the transactions which do not involve financial intermediaries such as banks.101 It is often dubbed ‘banking without banks’, which offers a better deal on both sides of the lending agreement.102 P2P lending leads to the reduction of transactional costs for borrowers and provides more investment opportunities and higher returns for lenders. In 2005, Zopa, the world’s first online lending platform, was launched in the UK.103 Through the internet, P2P lending platforms will be able to match individual and business borrowers with investors who have spare funds and would like to earn interest above bank savings. Prospective borrowers, when making P2P loan applications, have to submit relevant documents online, after which P2P lending platforms will conduct identity checks, credit risk evaluations and other due diligence. If borrowers meet relevant lending criteria, their loans will be listed on the e-marketplace to be funded by registered investors quickly. The UK has the largest alternative finance market in

98 Bank of England, ‘BankStats (Monetary & Financial Statistics)’, available at www​.bankofengland​ .co​.uk/​statistics/​Pages/​bankstats/​current/​default​.aspx, accessed 1 May 2022. 99 Aldermore, ‘Annual Report and Accounts 2016’, available at www​.shawbrook​.co​.uk/​media/​ 228610/​Shawbrook​-Report​-and​-Accounts​-2016​.PDF, accessed 1 May 2022. 100 Shawbrook PLC, ‘Annual Report and Accounts 2016’, p.130. 101 According to the P2P Finance Association (UK), P2P lending means ‘a debt-based funding arrangement facilitated by an electronic platform that comprises, to a significant extent, direct one-to-one contracts between a single recipient and multiple providers of funds, where a significant proportion of lenders are generally retail consumers and where borrowers are generally retail consumers or small businesses’: see P2PFA, ‘Rules of the Peer-To-Peer Finance Association’ (May 2015), para. 2.1; and P2P lending is also known as ‘loan-based crowdfunding’ in the UK, as the concept was invented by the Financial Conduct Authority (FCA): see FCA, ‘Crowdfunding’, www​.fca​.org​.uk/​consumers/​financial​ -services​-products/​investments/​types​-of​-investment/​crowdfunding, accessed 1 May 2022. 102 The Economist, ‘Peer-To-Peer Lending: Banking Without Banks’ (1 March 2014), p.70. 103 Lerong Lu, ‘Solving the SME Financing Puzzle in the UK: Has Online P2P Lending Got the Midas Touch?’ (2018) 33 Journal of International Banking Law and Regulation 449, 452.

Fintech  323 Europe, representing 74.3 per cent of total market share.104 The UK has more than 70 active online lending websites which have lent £4.4 billion in loans to consumers and SMEs.105 In the US, the online lending industry has grown rapidly with two main players: Lending Club and Prosper Marketplace.106 China has one of the largest P2P lending markets, with its total lending amount reaching CNY975 billion ($141.9 billion) in 2015.107 In today’s digital economy, an increasing number of individual and business borrowers tend to favour P2P loans over traditional bank loans, as a result of multiple benefits brought by financial disintermediation.108 The advent of online P2P lending has made financial intermediaries less attractive. Banks serve as credit intermediaries which form a creditor–debtor legal relationship with both depositors and borrowers.109 P2P lending platforms appear to serve a very similar function to banks, but in fact they are mere information intermediaries rather than credit intermediaries.110 P2P lending has five distinctive advantages for supporting digital trade. First of all, the funding source of P2P lending is diverse and has no geographical limitation. Banks rely heavily on deposits to fund their lending businesses, but P2P lending platforms could utilise various funding sources, including retail investors, institutional investors and government agencies, to arrange loans for borrowers. In the US, around 80 per cent of funds in P2P platforms come from institutional investors.111 In the UK the government plays an active role in P2P business lending, as in 2017 the British Business Bank lent £40 million to small businesses on Funding Circle.112 Second, the cost of borrowing is affordable on P2P lending platforms as there exists no financial intermediary extracting value from lending transactions. Small businesses can borrow up to £500,000 at a 4.50 per cent annual rate.113 Many platforms do not charge an early repayment fee for P2P borrowers, so small businesses only need to pay interest on the actual borrowing period. Third, P2P lending provides better and safer investment opportunities for ordinary savers with decent returns. Its crowdfunding nature has greatly lowered the investment threshold, as a loan will be funded by hundreds of investors to diversify risks. The threshold for purchasing Zopa’s classic P2P lending products is only

University of Cambridge and EY, ‘Moving Mainstream: The European Alternative Finance Benchmarking Report’ (February 2015), p.13. 105 P2P Finance Association, ‘Data’, available at http://​p2pfa​.org​.uk, accessed 1 May 2022. 106 Peter Rudegeair, ‘Layoffs Mount at Online Lenders’, Wall Street Journal (8 July 2016), C1. 107 Lingyi Finance, ‘P2P Lending Industry Annual Report 2015’, available at http://​01caijing​.baijia​ .baidu​.com/​article/​286273, accessed 1 May 2022. 108 Financial disintermediation means the process of eradicating any forms of financial intermediaries from lending and investment transactions, as the ubiquity of intermediaries in financial markets has been discussed extensively in corporate finance literature. See Lily Fang, Victoria Ivashina and Josh Lerner, ‘The Disintermediation of Financial Markets: Direct Investing in Private Equity’ (2015) 116 Journal of Financial Economics 160, 160–78. 109 Ross Cranston et al., Principles of Banking Law (3rd edition, Oxford University Press, 2018), p.190. 110 The primary function of online P2P lending portals is to facilitate the exchange of information between investors and borrowers and to provide relevant services supporting lending agreements like money collection. 111 Amy Cortese, ‘Loans that Avoid Banks? Maybe Not’, New York Times (4 May 2014), available at www​.nytimes​.com/​2014/​05/​04/​business/​loans​-that​-avoid​-banks​-maybe​-not​.html, accessed 1 May 2022. 112 Emma Dunkley, ‘Funding Circle: Small Business Backing’, Financial Times (6 Jan 2017), p.17. 113 Funding Circle, ‘Business Loans & Funding in the UK’, available at www​.fundingcircle​.com/​uk/​ businesses/​, accessed 1 May 2022. 104

324  Research handbook on digital trade £10, and more complicated Zopa products require a minimum capital of £1,000.114 The safety of P2P transactions is enhanced by technical methods whereby any investor’s money will be divided into small portions that are allocated to different loans, limiting the risk exposure and impact of default. P2P lending platforms such as Lending Works and Landbay have set up special reserve funds to recover investors’ losses when there is a default.115 The average annual return for P2P lending investors at Funding Circle was around 5.70–7.20 per cent.116 Fourth, P2P lending is very time-efficient compared with the lengthy, opaque and complex procedure of bank lending. For example, Funding Circle operates a simple and transparent loan application process for business borrowers, who only need to file applications online, with the loan to be advertised and funded within no more than seven days.117 Fifth, P2P lending platforms could potentially solve the information asymmetry problem and promote market transparency. By using big data and artificial intelligence, P2P platforms can accurately measure the size and location of credit risks of borrowers within seconds, as some of them have designed advanced risk models to assess multiple variables of borrowers, including their gender, age, marital status, educational level, working years, company size, monthly payment, loan amount, debt to income ratio and delinquency history.118

6.

DIGITAL MONEY FOR DIGITAL TRADE: CRYPTO ASSETS, STABLECOINS AND CENTRAL BANK DIGITAL CURRENCIES

This section considers fintech in the area of monetary and payment systems, leading to the rise of crypto assets, stablecoins and central bank digital currencies (CBDCs). It focuses on the traditional functions of money and how it has been digitalised to assist digital economy and trade. The term money traditionally refers to physical coins and banknotes. Economists define money as any item that is generally being treated as payment for goods, services and repayment of debts by a community.119 It is a social convention facilitating global trade when there is a lack of a double coincidence of wants, which solves the problem of lack of trust in commercial transactions. From a state’s point of view, money is regarded as legal tender or fiat currency, prescribed by authorities to circulate in the market, that will be widely accepted by the public with the backing of state credit.120 Throughout history, money has developed 114 Zopa, ‘What Is the Minimum Amount that I Can Lend in Each Product?’, available at http://​help​ .zopa​.com/​customer/​portal/​articles/​2329337, accessed 1 May 2022. 115 Tara Evans, ‘Peer-to-Peer Lending: Everything You Need to Know about the Leading Websites’, The Telegraph (18 July 2016), available at www​.telegraph​.co​.uk/​personal​-banking/​savings/​peer​-to​-peer​ -lending​-everything​-you​-need​-to​-know​-about​-the​-leadi/​, accessed 1 May 2022. 116 Funding Circle, ‘Lend to UK Businesses’, available at www​.fundingcircle​.com/​uk/​investors/​, accessed 1 May 2022. 117 Funding Circle, ‘Business Loans & Funding In The UK’, available at www​.fundingcircle​.com/​uk/​ businesses/​, accessed 1 May 2022. 118 Xuchen Lin, Xiaolong Li and Zhong Zheng, ‘Evaluating Borrower’s Default Risk in Peer-to-Peer Lending: Evidence from a Lending Platform in China’ (2017) 49 Applied Economics 3538, 3538–54. 119 Ross Cranston et al., Principles of Banking Law (3rd edition, 2018, Oxford University Press), p.363. 120 However, fiat money does not have the intrinsic value or use value. For example, the British pound, US dollar, euro and Chinese yuan have been perceived as money owing to the governmental or monetary authorities’ official recognition and their maintenance of purchasing power.

Fintech  325 from commodities such as cowrie shells to precious metals such as gold and silver, and then to government-issued paper notes. The emergence of fiat money and related payment tools such as cheques and bankcards have boosted commercial efficiency in an economy and stimulated the overall trading activities. Thanks to the role played by money in any economic and trading activities, it has solved the trust issue between transacting parties, both of whom would recognise the value of money being used for the underlying exchange of goods or services. Money has the basic function of a medium of exchange, a unit of account and a store of value.121 The past decade has seen the accelerated growth of various forms of digital currency such as Bitcoin, Ethereum, Facebook’s Diem, USDT, the digital yuan and the digital euro. Nearly 4,000 digital currencies have been issued by the private sector since the launch of Bitcoin as the very first cryptocurrency in 2009, partly because of the global financial crisis of 2007–8, which led some people to lose confidence in the existing financial system and rethink the future of money.122 The Covid-19 pandemic also contributed to the popularity of digital currencies as more people tend to work from home and shop online, resulting in more non-physical commercial transactions taking place that call for the frequent use of digital forms of money.123 What is more, building a cashless society has become a priority in some countries’ political agendas, as governments try to limit the use of cash and promote the digital economy. A research paper suggested that the number of cash and cash-like transactions has been falling significantly in most countries, including Australia, Denmark, China, Germany, India, Japan, Netherlands, Norway, Singapore, the UK and the US, declining at an annual rate of 6 per cent between 2006 and 2016.124 Sweden is a front-runner in achieving the cashless society, with one report predicting that the country would have fully embraced digital money by March 2023.125 The first popular type of digital money is blockchain-based cryptocurrencies, with some well-known examples such as Bitcoin and Ethereum.126 In 2008, the concept of Bitcoin was invented by Satoshi Nakamoto, who published an online paper advocating a new form of electronic cash that people can send to each other without going through disgraced banks.127 Bitcoin is a digital, decentralised and partially anonymous currency that is not backed by any government or legal entity and not redeemable for gold or other commodities.128

Paul Davidson, ‘Money and the Real World’ (1972) 82 The Economic Journal 101–15. The digitisation has raised the issue of currency competition, as both private sectors and public authorities compete to create new forms of money and to make their own monetary standards for the future economy. 123 Lerong Lu and Alice LS Zhang, ‘Regulating Fintech Corporations amidst Covid-19 Pandemic: An Analysis of Ant Group (Alipay)’s Suspension of IPO and Business Restructuring’ (2021) 42 The Company Lawyer 341, 341. 124 Tanai Khiaonarong and David Humphrey, ‘Cash Use across Countries and the Demand for Central Bank Digital Currency’ (2019) 13 Journal of Payments Strategy and Systems 32–46. 125 The Wharton School of the University of Pennsylvania, ‘Going Cashless: What Can We Learn from Sweden’s Experience?’ (31 August 2018), available at https://​knowledge​.wharton​.upenn​.edu/​ article/​going​-cashless​-can​-learn​-swedens​-experience/​, accessed 1 May 2022. 126 For a general discussion of blockchain technology, see this chapter’s section 2. 127 Satoshi Nakamoto, ‘Bitcoin: A Peer-To-Peer Electronic Cash System’ (2008), available at www​.ussc​.gov/​sites/​default/​files/​pdf/​training/​annual​-national​-training​-seminar/​2018/​Emerging​_Tech​ _Bitcoin​_Crypto​.pdf, accessed 1 May 2022. 128 Lerong Lu, ‘Bitcoin: Speculative Bubble, Financial Risk and Regulatory Response’ (2018) 33 Butterworths Journal of International Banking and Financial Law 178, 178–9. 121 122

326  Research handbook on digital trade Decentralisation and anonymity are two special characteristics that distinguish Bitcoin from other forms of digital money and payment systems.129 Crypto money relies on a decentralised system of invested individuals to verify, record and confirm each transaction. The peer-to-peer exchange system uses anonymous IP addresses instead of real names to protect the identities of its users. As of May 2022, Bitcoin had a market capitalisation of $723.35 billion and a daily transaction volume of $23.96 billion, having been widely used globally for both commercial transactions and investments.130 However, Bitcoin and other cryptocurrencies have seen huge volatility in their daily prices due to speculative activities. For example, Bitcoin’s price reached a peak of $68,000 in November 2021 before falling sharply to below $40,000 in early 2022.131 Clearly, the volatile pricing has made cryptocurrencies less popular for large-scale commercial transactions, which need money with stable value when it serves the medium of exchange and unit of account, which partly restricts the use of Bitcoin in global trade. Also, cryptocurrencies have been easily used for illegal or even criminal activities relating to drugs, weapons and money laundering.132 It has also raised regulatory concerns about fund safety and data protection for consumers.133 The second type of digital money is called the stablecoin and includes, for example, Tether (USDT) and Facebook’s Diem, which have been created by modern corporations and other private entities to alleviate the high volatility of crypto currencies. The value of stablecoins are pegged to one or more global currencies or to value-stable assets such as gold and silver to mitigate their price volatility.134 However, it is controversial whether stablecoins could really achieve price stability over a long period of time and become a good substitute for traditional currencies in global trade, given that their underlying assets could still undergo substantial price change within a short period of time – as we have experienced in relation to the Covid-19 pandemic and other geo-political incidents that have caused severe supply-chain crises in recent years. Clearly, cryptocurrencies and the more refined stablecoins have been increasingly used in global commercial and financial transactions. However, considering their speculative nature and other regulatory concerns such as money laundering and data protection, it is less likely that they would become mainstream currencies for global trade. Thus, we have seen governmental responses to digital money as central banks have been experimenting with official versions of sovereign digital currencies. The third type of digital money is state-issued central bank digital currencies (CBDCs), which aim to address certain limitations of privately issued digital currencies. CBDCs are the digital representation of sovereign currency made by a jurisdiction’s monetary authority that will appear on the liability side of the monetary authority’s balance sheet.135 Contrary Ibid. Coinbase, ‘Bitcoin Price’, available at www​.coinbase​.com/​price/​bitcoin, accessed 1 May 2022. 131 Ibid. 132 Sean Foley, Jonathan R Karlsen and Talis Putnins, ‘Sex, Drugs, and Bitcoin: How Much Illegal Activity Is Financed through Cryptocurrencies?’ (2019) 32 The Review of Financial Studies 1798–1853. 133 Lerong Lu, ‘Bitcoin: Speculative Bubble, Financial Risk and Regulatory Response’ (2018) 33 Butterworths Journal of International Banking and Financial Law 178, 180. 134 Douglas Arner, Raphael Auer and Jon Frost, ‘Stablecoins: Risks, Potential and Regulation’ BIS Working Papers No. 905 (November 2020), available at www​.bis​.org/​publ/​work905​.pdf, accessed 1 May 2022. 135 John Kiff et al., ‘A Survey of Research on Retail Central Bank Digital Currency’ IMF Working Paper No. 20/104 (June 2020), available at www​.imf​.org/​en/​Publications/​WP/​Issues/​2020/​06/​26/​A​ -Survey​-of​-Research​-on​-Retail​-Central​-Bank​-Digital​-Currency​-49517, accessed 1 May 2022. 129 130

Fintech  327 to decentralised money such as bitcoins or stablecoins, CBDCs are in a form of centralised system similar to conventional banknotes. All transaction records and stored data of CBDCs will be accessible to authorities, echoing the government push for digital payments and a cashless society.136 CBDCs also contribute to effective monetary policy solutions, as they are not only acting as official money and payment methods but also working as an interest leverage tool. CBDCs could even be charged with extra regulatory functions such as combating money laundering and terrorist financing. According to the Bank for International Settlement (BIS), 86 per cent of 65 surveyed central banks have been engaging in some form of CBDC work, with 60 per cent of them having progressed from conceptual research to experiences or proof of concept and 14 per cent moving forward to development and pilot arrangements.137 So far, China has been playing a leading role in the trial of CBDCs, when the People’s Bank of China (PBOC) started to test the feasibility of digital fiat currency in 2014 and rolled out the use of digital yuan on a large scale in Shenzhen and other cities.138 In 2018, the Bank of England has further explored the design mode of freely traded bank deposits against CBDCs by elaborating several core principles of CBDC issuance.139 In January 2022, the US Federal Reserve expressed potential interest in developing its digital currencies as digital dollars and issued a discussion paper examining the pros and cons of US CBDCs.140 At present, CBDCs seem to be the most viable non-physical currency and payment options for digital trade due to their advanced regulatory function, stable value and, most importantly, official endorsement and wide acceptability.

7. CONCLUSION This chapter has provided a holistic and in-depth analysis of the global fintech sector. Artificial intelligence, blockchain, cloud computing and data technologies have greatly transformed every aspect of the financial industry, as they enable financial institutions, both old and new, to deliver convenient and high-quality services with reduced expenditure. Fintech, BigTech and Metaverse corporations have further boosted the economic efficiency of digital trade. Global cities such as New York, London, Paris, Shanghai, Dubai and Mumbai have been reliant on their unparalleled advantages in location, capital, talent and education to build global fintech 136 CBDCs offer other advantages that are typically found in central bank-issued money, such as settlement finality, liquidity and integrity, and such features make them more widely acceptable money for digital trade. 137 Codruta Boar, Henry Holden and Amber Wadsworth, ‘Impending Arrival – A Sequel to the Survey on Central Bank Digital Currency’ BIS Papers No. 107 (January 2020), available at www​.bis​.org/​ publ/​bppdf/​bispap107​.pdf, accessed 1 May 2022. 138 Lerong Lu and Hang Chen, ‘Digital Yuan: The Practice and Regulation of China’s Central Bank Digital Currency’ (2021) 36 Butterworths Journal of International Banking and Financial Law 601, 601–3. 139 Michael Kumhof and Clare Noone, ‘Central Bank Digital Currencies – Design Principles And Balance Sheet Implications’ Bank of England Staff Working Paper No. 725 (May 2018), available at www​.bankofengland​.co​.uk/​-/​media/​boe/​files/​working​-paper/​2018/​central​-bank​-digital​-currencies​ -design​-principles​-and​-balance​-sheet​-implications, accessed 1 May 2022. 140 The Federal Reserve (US), ‘Federal Reserve Board Releases Discussion Paper that Examines Pros and Cons of a Potential U.S. Central Bank Digital Currency (CBDC)’ (January 2022), available at www​ .federalreserve​.gov/​newsevents/​pressreleases/​other20220120a​.htm, accessed 1 May 2022.

328  Research handbook on digital trade hubs offering top-notch financial infrastructure for tomorrow’s digital economy. Fintech hubs, technology corporations, financial institutions, consumers and regulators have interacted with each other to form a complex and dynamic fintech ecosystem. The key to success in the fintech sector lies in balancing the goal of promoting financial innovation with other traditional regulatory objectives such as ensuring financial stability and protecting financial consumers. This has presented mounting challenges for global policy-makers and financial authorities. The chapter has also examined some detailed fintech services and products as case studies. For instance, it looked at fintech-driven alternative finance methods such as digital-based challenger banks and online P2P lending platforms, which have been serving millions of consumers and small businesses that have limited access to formal financial institutions. This has contributed to financial inclusion and equality. Finally, the discussion of digital money, which includes crypto assets, stablecoins and CBDCs, predicts the future of money, which has always been a vital element in digital economy and trade. It makes a tentative suggestion that, since privately issued bitcoins and stablecoins are still mainly used for financial investment, official CBDCs are likely to be the most feasible currency solution for tomorrow’s digital trade thanks to their regulatory mandate, stable value and wider acceptability.

20. The digital nomad and the emergence of global labour mobility David Collins and Ben Robinson

1. INTRODUCTION The widespread and rapid advancement of information technology has digitalised the global economy to such an extent that, where ‘digital trade’ was once a sufficiently fine distinction as to have real utility, one can now simply say ‘trade’ with a degree of certainty that it will possess a digital element. That is to say that the majority of the globe now spends a significant portion of time, discretionary income and attention in an increasingly hybrid digital/physical existence. Although the most narrowly defined measure of digitally enabled trade or trade in digital products remains comparatively small, at 3.5–6 per cent of total global trade,1 this narrow definition greatly understates the impact and presence of digitalisation on trade, particularly in the form of labour that is mediated through digital technologies. Even more, there is good reason to imagine that digitalisation will eventually impact the overwhelming majority of all global trade activity – 70 per cent or more by some estimates.2 This process has suddenly made practically accessible the mass ‘relativization of geographical distances made possible by working virtually from any location’3 implied in the transformation to a digital nomad lifestyle. In theory, a digital nomad is a knowledge worker who requires nothing more than an internet connection and a personal computing device to earn a living without regard to geographical location.4 In popular culture, this individual has long been characterized as young, typically working on a self-employed or freelance basis, often involved in software development, graphic design or a social media-adjacent area of work – all of which have some basis in reality, particularly among full-time digital nomads living in countries where they have neither citizenship nor other territorial links.5 While there is currently no more than a modest probability of a substantial segment of the global labour force abruptly relocating to Bali, few of the substantive barriers that once ruled out this course of action still exist. This chapter will examine the implications of the increasingly practical accessibility of the digital nomad phenomenon and the information technology changes that have caused this. It will then consider the extent to which international rules on digital trade have enabled this kind of commercial lifestyle through the provisions aimed A Bhattacharya and C Ketels, Global Trade Goes Digital (BCG Global, 12 August 2019). Ibid. 3 M Orel, ‘Coworking Environments and Digital Nomadism: Balancing Work and Leisure whilst on the Move’ (2019) 61(3) World Leisure Journal 2 www​.tandfonline​.com/​doi/​abs/​10​.1080/​16078055​.2019​ .1639275 (accessed October 2022). 4 O Hannonen. ‘In Search of a Digital Nomad: Defining the Phenomenon’ (2020) 22(3) Inf Technol Tourism 2 https://​link​.springer​.com/​article/​10​.1007/​s40558​-020​-00177​-z (accessed October 2022). 5 I Wayan Sukma Winarya Prabawa and PR Pertiwi, ‘The Digital Nomad Tourist Motivation in Bali: Exploratory Research Based on Push and Pull Theory’ (2020) 7(3) Athens Journal of Tourism 169. 1 2

329

330  Research handbook on digital trade at assisting small and medium-sized enterprises (SMEs) contained in free trade agreements (FTAs), and to a lesser extent the World Trade Organization (WTO).

2.

THE RISE OF DIGITAL NOMADS

The digitalisation process alluded to above has progressed to such an extent that an increasing proportion of the average citizen’s daily life is mediated through digital technologies. In particular, this refers to the internet and the digitalisation of traditional notions of ownership toward a state where ‘half the world’6 now mediates consumption in the more ‘liquid ownership state’7 of licensed or accessed digital goods. This trend has long been present in education, entertainment and trade in goods and services, primarily in e-commerce ordering of physical and digital goods such as Software as a Service (SaaS) subscriptions. However, for a sizeable proportion of the population, this digitalisation has now had sudden and potentially lasting impacts in the area of work and employment. No period has demonstrated the changing nature of work or digitalisation’s impact on it more clearly than the response to the COVID-19 pandemic. In an effort to reduce the spread of this novel respiratory virus, governments enacted regional and national lockdowns with ‘work from home’ (WFH) orders of varying severity. Major transnational corporations enacted global voluntary WFH policies that, in some cases, pre-empted and outlasted the orders of regulators. This process caused the most rapid and prolonged period of dislocation and digitalisation in modern history, radically accelerating pre-existing trends and reshaping entire industries in a space of weeks. The superficial impacts of these trends have been well established and have dominated the news cycles since early 2020 as nearly all knowledge workers found themselves barred from their offices and required to work from home under unprecedented lockdown measures. While this transition was not without challenge, the pre-existing trends of digitalisation made what was previously imagined to be a complete global shutdown into a period of rapid, but ultimately successful adaptation. Zoom and other videoconference meetings replaced in-person meetings and then eclipsed them, going on to take up more and more of the average knowledge worker’s day. Young urban knowledge workers abandoned cramped apartments in major urban centres and fled to the comfort and perceived safety of parental homes in the countryside – in some cases, in other countries. This last example holds the most relevance for the concept of the digital nomad and its likely impact on digital trade. Yet, although much has been made of this concept, it has failed to significantly impact the global labour market or the operations of major transnational companies because, despite the globalisation of capital and goods, the global market for labour has remained relatively constrained, largely because of immigration rules and the functional impracticability of working remotely over a long period. With the rise of asynchronous communications tools, geographically independent working and digital citizen or e-residency

6 CK Morewedge and others, ‘Evolution of Consumption: A Psychological Ownership Framework’ (2021) 85(1) Journal of Marketing 196 https://​journals​.sagepub​.com/​doi/​full/​10​.1177/​0022242920957007 (accessed October 2022). 7 Ibid at 197.

The digital nomad and the emergence of global labour mobility  331 schemes such as the well-established programme in Estonia,8 as well as the newer schemes in Lithuania, Portugal, South Africa and Ukraine, the barriers to self-employed work on a digitally nomadic basis have significantly fallen away. Moreover, the remote working experience of the COVID-19 pandemic has further demonstrated that, even in an employment context, the need to be in an office arises primarily from the intersection of habit/convention, immigration and taxation rules. Further trends and changes in the global workforce suggest that these changes will have broader and farther-reaching impacts over time, as fewer people are traditionally employed, and many large organisations make greater use of temporary and outsourced work. The greater the degree to which large organisations become accustomed to remote and geographically distributed working, the more these trends may become self-reinforcing. Coupled with the expansion of automation and artificial intelligence in a wide variety of tasks, these trends are unlikely to reverse. Rather, it is probable that they will accelerate exponentially, as has consistently been the case for information technology and the companies that leverage these technologies effectively.9 However, there is some confusion around the concept of the digital nomad. Specifically, its ultimate and most common form is not the wholly realised nomadic ‘influencer’ originating from the Global North or West, sitting on a beach in a tropical country in the Global South and hopping to other exotic destinations every three months while monetising social media, dodging immigration controls and avoiding – or evading – taxation. Few people are likely to live this exotic lifestyle in full, and endless global travel is not the likely final form of the digital nomad. Rather, this phenomenon is likely to be diffused and distributed, spreading across the sharing economy or ‘semi-automated gig economy’,10 and spanning Global North and South. The practical accessibility of the digital nomad lifestyle – or, rather, of remote and asynchronous working in a knowledge economy context – offers the potential for offshoring, outsourcing and relatively free movement of labour – even without freedom of movement of people – in a way and to a degree never before possible.

3.

DIGITAL ISOLATION AND LABOUR MARKETS

The ordinary conception of the labour market is necessarily bounded by national borders. However, the escalating digitalisation of global trade (and, increasingly, work) has posed successive challenges to this status quo. This accelerated rapidly during the COVID-19 pandemic, during which many service workers adapted to WFH in response to government-mandated ‘lockdowns’. WFH in some cases became working from abroad – employees were able to return to their home countries while continuing to work for their employer overseas. However, even in the aftermath of COVID-19, it is commonplace for workers to come into the office only one or two days per week. Globally, the de facto practice of work has pulled away from the conventional conception of what it is to be a worker or employee. While it is clear that these changes are here to stay 8 ‘Estonian e-Residency’ https://​e​-estonia​.com/​solutions/​e​-identity/​e​-residency/​ (accessed October 2022). 9 N Kander, ‘The Exponential Age Will Transform Economics Forever’ (2021) Wired UK. 10 Ibid.

332  Research handbook on digital trade and only likely to accelerate, it is not yet clear precisely how they will change in the future. Nevertheless, governments and legislators have begun to recognise the significance of these changes and make sporadic and half-hearted responses to them. Regulators do not want to stifle innovation – and yet, inaction has led to a situation in which the previously stable labour markets have become spatially dynamic, notwithstanding the presence of rigid or semi-rigid rules on immigration. GATS Mode Four of supply of services (Movement of Natural Persons) no longer requires positive commitments on the part of states to open their borders and allow temporary workers. Liberalisation of GATS Mode One supply of services (Cross-Border Supply) is sufficient. Indeed, as economic activity becomes intensively digitalised, there will a corresponding decline in relevance of Mode Four supply. Mode Two (Consumption Abroad) may also decline in relevance – applicable only to services such as tourism and other which are necessarily place-dependent. This is crucial because the negotiation of commitments under Mode Four has often been fraught with difficulty (for example the UK’s negotiations for an FTA with India, still ongoing at the time of writing). With Mode 4 marginalised as an objective for treaty partners, efforts on furthering the liberalisation of Mode 1 can be pursued with greater vigour. Progress on services domestic regulation, for example through the WTO’s recent 64-member initiative, could be vital in this regard since professional qualifications will be essential for some kinds of services, such as legal advice. GATS disciplines on Mode Two are far from complete, as many WTO members retain barriers in this regard for a wide variety of professional services.11 Greater liberalisation may be needed in preferential FTAs for digital nomads to have sufficiently wide access to foreign markets. Mutual recognition agreements, pursued at a bilateral or regional level, which acknowledge professional qualifications of digital nomads would also be helpful here. If the modern digital economy has evolved into one in which the skill of the individual supplier can be monetised and traded across borders, then it is crucial to consider what kind of regulatory barriers could act as a drag on this kind of economic activity. Tax is the obvious example. World Trade Organization (WTO) law, which governs trade in goods and services, does not prohibit tax, whether the relevant measure is described as a tax, as a customs duty or as a tariff. As discussed in Chapter 17 of this book by Collins, digital services taxes (DSTs) have been proposed by a number of countries, raising potential issues of WTO compliance in terms of non-discrimination obligations (National Treatment and Most Favoured Nation). As generally envisioned, DSTs would apply only to the largest companies, such as the Silicon Valley Tech Giants (such as Amazon, Facebook and Netflix) with large turnovers. This means that small and micro-sized enterprises would likely be unaffected. Neither would digital nomads – the lone individuals offering digital services across international borders. In the event that some DSTs were to apply to smaller suppliers by virtue of lower revenue thresholds, the business model of the digital nomad could suffer or become unsustainable because tariffs on services delivery could render their businesses prohibitively costly. The WTO compliance (or non-compliance) of DSTs is therefore a question of paramount concern, as discussed in the chapter by Collins. In 1998, as the internet began to mature into a driving force behind global commerce, WTO members agreed not to impose customs duties on electronic transmissions as part of their work 11 See further, D Collins, The Public International Law of Trade in Legal Services (Cambridge University Press, 2019).

The digital nomad and the emergence of global labour mobility  333 package on electronic commerce. Such duties could also undermine the business models of digital nomads. This moratorium has been renewed several times, most recently at the 12th Ministerial Conference in June 2022. Prohibitions on customs duties on electronic transmissions can also be found in numerous FTAs, including the Digital Economic Partnership (DEPA) between New Zealand, Chile and Singapore as well as the Comprehensive and Progressive Trans-Pacific Partnership (CPTPP), which will be discussed in further detail below. Perhaps more than any other legal innovation, the moratorium has enabled the delivery of digital services across international borders, facilitating the mobility of those who deliver such services. Without customs duties, the suppliers of digital services are free to live wherever they wish without regard to ‘tariff walls’ or other such taxes on their services. It is important to recognise also that digital nomads do not only supply services across the internet. Many will also be exporters to international markets. Customs procedures and paperwork associated with this process can be daunting for smaller firms lacking stable office or warehouse space, intensifying the need for trade facilitation. Paperless trading, now a common feature of digital trade agreements, could help incorporate firms of this nature into the global economy. Electronic invoicing, contemplated by agreements such as DEPA (Article 2.5) and contained in domestic legislation, such as the highly anticipated UK Electronic Trade Documents Bill (Bill 57, currently before the House of Lords) could lower some of the transaction costs associated with goods-oriented digital nomads.

4.

DIGITAL NOMADS AS SMES

4.1

Digital Challenges to Smaller Firms

While digital nomads are typically individuals or single-member companies rather than companies with employees, they still fit within the conventional understanding of small and medium-sized enterprises (SMEs). As such, the opportunities and challenges faced by digital nomads are similar to those which are experienced by SMEs operating in a digital environment. The smaller size of a business often operates as a barrier to adoption of a digital business strategy. Indeed, this may be the reason SMEs continue to lag in the digital economy relative to larger firms. SMEs may not possess the internal resources, awareness or technological skills, or may have problems raising capital. These shortcomings can restrict productivity and contribute to the firm’s underperformance.12 Barriers faced by SMEs which supply services internationally include nationality requirements (where physical movement is required – typically not an issue for digital nomads), minimum capital requirements to enter foreign markets, restrictions on multi-disciplinary activities (such as legal services bundled with accounting or financial services), lack of mutual recognition of licensing, credentials and standards, and intellectual property and piracy issues.13

12 S Hensellek and N Puchala, ‘The Emergence of the Digital Nomad: A Review and Analysis of the Opportunities and Risks of Digital Nomadism’ in M Orel, O Dvouletý and V Ratten, eds, The Flexible Workspace (Springer 2021). 13 R Adlung and M Soprana, ‘Trade Policy for SMEs from a GATS Perspective’ in T Rensmann, ed, Small and Medium Sized Enterprises in International Economic Law (Oxford University Press, 2017) at 14.

334  Research handbook on digital trade Regulatory barriers impeding services related to communications infrastructure and connectivity are especially problematic for digital nomads because they affect all trade supplied through digital networks. In many respects, the landscape for digitally enabled services has become more restrictive in the past ten years. Encouragingly, as a consequence of remote working and changing consumer habits during the COVID-19 pandemic in favour of online shopping, fewer obstacles to cross-border trade in digitally enabled services were introduced in 2020 than in 2019. In fact, the implementation of new policy measures that facilitated digital trade, notably regarding e-signatures and e-payments, increased during this period.14 These are the lifeblood of digital nomads for whom concluding contracts in person, or even through international post, would be unthinkable. Digital nomads may often work in the creative industries, including music, literature and art. In these cases intermediaries tend to be closer to customers and therefore removed from the riskier activity of developing original products or content. The asymmetry in levels of risk is captured by the relative scale of organisations in the creative industries, with small independent content creators, rooted in localised collaborative networks, orbiting larger media conglomerates.15 The success of settled intermediaries with well-developed networks across a multiple of formats (see for example, books and films) appears to have established a situation in which cultural producers are relegated to a peripheral role, unable to establish a direct connection with consumers.16 This asymmetry is exacerbated as markets mature, with SMEs or digital nomads tending to be crowded out by the market leaders.17 Enabling digital nomads to capitalise on the advantages of the digital economy requires support from policy makers and on the national and, increasingly, international level.18 The enforcement of strict privacy rules, such as the EU’s General Data Protection Regulation (GDPR), entails significant resources, and may not be consistent with the available regulatory capacity in many developing countries where digital nomads choose to locate because of lower costs of living. Contractual and certification mechanisms for cross-border data transfer are unaffordable for SMEs and require long processing times, hampering business opportunities for smaller data-driven work undertaken by digital nomads in developing countries.19 Although some impediments to digital trade have been lowered recently to assist in the adaptation to remote purchasing, for example as discussed in Hulicki’s chapter on smart contracts, the global regulatory environment for digital trade remains complicated and highly variable across jurisdictions. International laws governing trade in digital services, such as those contained in multilateral and international treaties, play a key role in addressing some of these

14 OECD, OECD Services Trade Restrictiveness Index: Policy Trends up to 2021 (OECD, February 2021) https://​issuu​.com/​oecd​.publishing/​docs/​oecd​-stri​-policy​-trends​-2021​?fr​=​sMGVlMjI5ODk2NDE (accessed October 2022). 15 C Bilton, The Disappearing Product: Marketing and Markets in the Creative Industries (Edward Elgar Publishing, 2017) at 41. 16 Ibid, at 42. 17 Ibid, at 42; noting the example of major studios investing heavily in a small number of blockbuster films. This process can also lead to opportunities for SMEs as the market leaders end up focusing on a smaller set of large projects with high returns, leaving SMEs space for smaller niche projects. 18 OECD, The Digital Transformation of SMEs, OECD Studies on SMEs and Entrepreneurship (OECD Publishing, 2021) doi​.org/​10​.1787/​bdb9256a​-en accessed 29 November 2021. 19 Ibid, at 131.

The digital nomad and the emergence of global labour mobility  335 barriers. The main regulatory challenges include reducing impediments to digital infrastructure and connectivity, such as laws that limit access to high-quality communications services, as well as measures that inhibit the transfer of data across borders. Reforms are also required to eliminate data localisation requirements, seen in some FTAs, and limitations on online content providers. Compliance with strict localisation policies and complex data regulations that may have been designed to limit the power of large technology companies may actually be more affordable for these larger firms than for SMEs or digital nomads. This is seen for example in smaller firms’ difficulty in operating in the EU due to the elaborate regulatory requirements of the GDPR.20 Data localisation rules could undermine digital nomads’ ability to access affordable technologies, such as cloud-based software, which may not be present in their own jurisdiction.21 This is why states should avoid data regulations that can harm the growth of smaller businesses or affect consumer confidence in their domestic economies.22 A legally fragmented internet in which many of these barriers are enabled lowers economic opportunities for SMEs, stifling their efforts to serve the most lucrative worldwide markets, restricting them to local ones where growth has already been exhausted.23 4.2

Digital SMEs and FTAs

More countries are beginning to include e-commerce or digital trade provisions in their FTAs, helping to address some of the legal obstacles faced by the smaller firm seeking to engage in digital trade. Improvements in digital connectivity have been shown to lead to even greater increases in trade when they are supported by the liberalising provisions found in preferential FTAs.24 As discussed throughout this edited volume, a growing number of preferential FTAs contain legally binding commitments on digital trade, such as Chapter 19 of the United States– Mexico–Canada Agreement (USMCA), Module 10 of the Digital Economy Partnership Agreement (DEPA) and, perhaps most notably given the number of countries which have joined, the Comprehensive Progressive Trans-Pacific Partnership (CPTPP), an 11-nation FTA comprising roughly 13 per cent of global GDP.25 The importance of addressing the needs of SMEs operating in international digital markets is conveyed in the statement in DEPA: parties ‘recognise the fundamental role of SMEs in maintaining dynamism and enhancing competitiveness in the digital economy’.26

20 N Martin, C Matt, C Niebel and K Blind, ‘How Data Protection Regulation Affects Startup Innovation’ (2019) 21 Information Systems Frontiers 1307–24. 21 L Guglya and M Maciel, ‘Addressing the Digital Divide in the Joint Statement Initiative on E-Commerce’ International Institute for Sustainable Development, (30 December 2020) at 43. www​.iisd​ .org/​publications/​addressing​-digital​-divide​-e​-commerce accessed October 2022. 22 See Digital Economy Report, UNCTAD (October 2021), https://​unctad​.org/​system/​files/​official​ -document/​der2021​_overview​_en​_0​.pdf at 134 (accessed October 2022). 23 Ibid, at 116. 24 See ‘Making the Case for G7 Digital Trade Principles’, OECD, 25 Oct 2021 at 2. Examples of software include platforms like Upwork, Freelancer, and Shopify: at 1. 25 The parties are: Canada, Japan, Australia, New Zealand, Vietnam, Mexico, Peru, Chile, Brunei, Singapore and Malaysia. The UK acceded to the CPTPP in 2023, although its access has not yet been ratified by the other members at the time of writing. 26 Art 10.1.1.

336  Research handbook on digital trade The CPTPP could become a future hub of global trade governance, establishing benchmarks in various areas, such as digital trade.27 Should more countries accede to it as expected,28 the CPTPP could become another global regime for trade law, alongside or perhaps in competition with the WTO. The CPTPP’s e-commerce chapter (Chapter 14) contains several elements that could help expand growth in digital trade among the signatories with beneficial impacts on digital nomads and SMEs. Moreover, the agreement’s rules on services (Chapter 10) encompass both internet service providers and internet users as well as both buyers and sellers. It is unclear whether the CPTPP’s provisions on digital trade contained in its e-commerce chapter apply to all cross-border information flows by all internet actors, meaning if they cover both suppliers and consumers of digital transmissions.29 Chapter 14 applies to ‘measures adopted or maintained by a Party that affect trade by electronic means’.30 This would seem to encompass more than measures that govern or ‘apply to’ trade. Measures affecting the supply of a service delivered or performed electronically are subject to the obligations contained in the relevant provisions of Chapter 9 (Investment), Chapter 10 (Cross-Border Trade in Services) and Chapter 11 (Financial Services).31 Article 14.4 on Non-Discriminatory Treatment of Digital Products prohibits signatory parties from favouring domestic products and their creators and owners or from discriminating between products or producers from home versus those from abroad. This encompasses a National Treatment rule for data which is key to the success of digital nomads serving international clients. ‘Products’ here includes computer programs, videos and recordings produced for sale and transmitted electronically. Again, this provision enables a level playing field in which firms can compete based on the quality and price of their products rather than their nationality – a critical requirement for small foreign providers aiming to compete with well-established local firms. Digital nomads handling client data will have to satisfy the privacy requirements of the markets in which they operate, including those outside the country in which they are physically located at any given time. Chapter 14 of the CPTPP requires that ‘Each Party shall adopt or maintain a legal framework that provides for the protection of the personal information of the users of electronic commerce’.32 Parties further undertake to publish information on personal privacy protection and ‘endeavour to adopt non-discriminatory practices’.33 There is also an agreement among the parties to develop mechanisms to promote compatibility among different privacy regimes. Wide variation in approach to privacy at the national level could make this obligation difficult to fulfil, or it could prevent the accession of countries with

See for example JA Huerta-Goldman and D Gantz (eds) The Comprehensive and Progressive Trans-Pacific Partnership Analysis and Commentary (Cambridge University Press, 2021). 28 China, Taiwan and Ukraine have expressed an interest in joining in the future. 29 Article 14.11 on information flows states that ‘each party shall allow the cross-border transfer of information by electronic means […] when this activity is for the conduct of the business of a covered person’. A ‘covered person’ is defined in article 14.1 as an ‘investment, investor or service supplier’. The agreement only mentions ‘users’ in article 14.8, where it recognises the benefits of protecting users’ personal information. Some information flows are not for the conduct of the business of a covered person. These flows do not involve the exchange of money. 30 Art 14.2.2. 31 Art 14.2.4. 32 Art 14.8. 33 Art 14.8. 27

The digital nomad and the emergence of global labour mobility  337 the strictest privacy regimes; for example, the EU’s GDPR. A digital nomad resident in one CPTPP country might therefore find it easier to serve clients across other CPTPP countries than those located elsewhere. Article 14.10 of the CPTPP builds on recognised principles for internet governance aimed at empowering consumers. These kinds of provisions may be especially important for consumers receiving services (or goods) from digital nomads because these smaller, itinerant suppliers will almost certainly be less well vetted than larger ones, for which there will likely be online rating services and customer satisfaction reviews. Signatories recognise the importance of consumers being able to make their own choices; to connect their own devices to networks and to access information on the practices of their internet service suppliers. These requirements promote competition, which can only be beneficial to SMEs operating in a market dominated by large incumbents, as well as, obviously, the customers that they serve. The CPTPP further states that parties must adopt or maintain consumer protection laws for an online environment.34 Again, compliance with these kinds of laws is a significant but legitimate cost for SMEs and digital nomads because consumer trust is key to the expansion of electronic commerce. Governments of CPTPP parties cannot force suppliers to disclose their source code to foreign governments, even for national security reasons.35 While this rule supports the development of commercially valuable intellectual property, it could work against cybersecurity initiatives. On the other hand, this rule should provide some assurances to innovative, technology-focused digital nomads who have expended significant resources on software development. The CPTPP prohibits data localization requirements through Article 14.13 which states: ‘No Party shall require a covered person to use or locate computing facilities in that Party’s territory as a condition for conducting business in that territory.’ This feature is vital to digital nomads seeking to serve clients online in multiple jurisdictions. While forced localisation of technologies and servers may be advantageous in terms of security and data protection, it creates the need for costly and often redundant data storage, likely operating as a market entry barrier for SMEs and sole traders such as digital nomads. It should be noted that the prohibition on data localisation in the CPTPP is subject to three exceptions: government data, financial services and a general four-step test exception. The governmental data exception enables governments to retain the rights for data localisation measures for government data that it holds or that is held by third parties under contract. Establishing categories with different legal treatment, in some ways an inescapable feature of the regulation of digital trade, might be prohibitively expensive for digital nomads. The application of strict measures to certain types of data, such as the personal data of customers, effectively amounts to measures affecting all types of data.36 The four-step test to permit data localisation found in the CPTPP uses General Agreement on Tariffs and Trade (GATT) Article XX language: Nothing in this Article shall prevent a Party from adopting or maintaining measures […] to achieve a legitimate public policy objective, provided that the measure: (a) is not applied in a manner which would constitute a means of arbitrary or unjustifiable discrimination or a disguised restriction on trade; and (b) does not impose restrictions on the use or location of computing facilities greater than are required to achieve the objective.

Art 14.7. Art 14.7. 36 See Guglya and Maciel, supra n 21, at 29. 34 35

338  Research handbook on digital trade Some feel that this highly discretionary test could lead to regulatory chill among implementing governments, potentially preventing them from applying stronger data export limitations.37 This could work to the advantage of digital nomads eager to operate in a lighter-touch regulatory environment. On the other hand, jurisdictionally mobile service providers could escape tighter regulation of this nature by moving – one of the advantages of the nomadic commercial lifestyle. The scope of the exceptions could be important for digital nomads creating digital content concerned about the trade implications of censorship and filtering.38 A CPTPP signatory which censored or filtered the internet could cause rerouting of information flows, hampering market access for digital nomads. Given the difficulty of satisfying General Exception tests under GATT case law,39 CPTPP parties would likely struggle to raise issues such as privacy and data security as justified obstacles to cross-border digital trade. Domestic regulation of privacy will clearly inform the content of FTAs even though they can only be meaningfully enforced only within national jurisdictions. Banning future data localisation requirements as well as data transfer restrictions consistent with privacy, security and innovation policy needs may be an unwise stance that unnecessarily inhibits the capacity of policy makers to enable data governance within their own borders.40 The business models of digital nomads could be vulnerable to these strategies. FTAs with digital trade chapters must maintain a balance between data localisation and data transfer restrictions and public policy needs. For digital nomads to be able to survive, this approach must ensure that reasonable privacy, security and public policy measures will not be disallowed, provided that they are reasonably administered and not unduly burdensome for firms that do not have large compliance departments. Careful application of the four-step test by dispute settlement panels such as the CPTPP’s state-to-state arbitration system found in Chapter 2841 should help to ensure that the general four-step exception can be reconciled with an open internet which is compatible with domestic privacy and security but which also fosters innovation and market participation by smaller entities.42 CPTPP signatories agree that each party ‘shall allow the cross-border transfer of information by electronic means, including personal information, when this activity is for the conduct of the business of a covered person’.43 ‘Covered person’ includes a ‘covered investment’, which is defined in the CPTPP’s investment chapter (Chapter Nine); an investor of a CPTPP party other than a financial institution; or a service supplier of a party as defined by the CPTPP’s

37 M Geist, ‘Data Rules in Modern Trade Agreements: Toward Reconciling an Open Internet with Privacy and Security Safeguards’ Centre for International Governance and Regulation (4 April 2018) 38 See S Aronson, ‘The Digital Trade Imbalance and Its Implications for Internet Governance’ (February 2016) Global Commission on Internet Governance, Paper Series No 25, at 10. 39 For example, EC–Seal Products, WT/DS400/AB/R (report adopted 18 June 2014, in which the respondent failed to satisfy the chapeau because of uneven application of the indigenous hunting exception) and China–Periodicals, WT/DS363/AB/R (report adopted 19 January 2010, in which the respondent failed to demonstrate that the measure was necessary – there was an alternative, less trade restrictive method reasonably available). 40 See Geist, supra n 37. 41 Art 14.18 states that the CPTPP’s dispute settlement applies to Chapter 14, with some reservations for Malaysia and Viet Nam. 42 See Geist, supra n 37. 43 Art 14.11.2.

The digital nomad and the emergence of global labour mobility  339 services chapter.44 The wide scope of this definition should help facilitate the business activities of digital nomads. The CPTPP goes some way further than other FTAs, some of which simply seek to promote regulatory cooperation in this area.45 The CPTPP’s data transfer rule is subject to the same general four-step test exception noted above in relation to data localisation. Limits on data transfers may be seen, in principle, as a positive development for SMEs, including digital nomads, who seek open access to the internet and therefore global customers. Of course, data transfer restrictions may be legitimately used to protect the privacy and security of consumers. Finally, Chapter 14 of the CPTPP permits governments of signatory parties to provide subsidies or grants to their own producers or creators. SMEs may benefit from various kinds of assistance programmes adopted for cultural purposes or to spur innovation. Many digital nomads, however, will be virtually unknown to the states in which they operate, meaning that their capacity to avail themselves of grants will be diminished. It may be worth states considering how they could proactively provide more assistance to these types of businesses as the number of digital nomads rises. Greater availability of (free) access to the internet in designated areas could be one such way.46 In addition to its e-commerce chapter, the CPTPP has a dedicated chapter on SMEs (Chapter 24). These are beginning to appear in various modern FTAs, such as the new ones signed by the United Kingdom as well as the stand-alone digital agreement, DEPA. This element of the agreement was intended to enable SMEs to benefit from international trade to a greater extent than under other FTAs or WTO rules. The SME chapter achieves this objective essentially through transparency. It requires signatories to make available online information about the obligations and rights contained in the CPTPP as well as information specifically designed to assist SMEs in pursuing trade opportunities provided under other chapters of the CPTPP, such as the chapter on e-commerce. Each signatory is required to establish its own publicly accessible website containing information regarding the CPTPP which is of relevance to SMEs.47 The chapter further establishes a committee on SMEs aimed at helping these businesses take advantage of the treaty.48 This important feature of the CPTPP acknowledges the information asymmetry under which SMEs often operate. They lack the resources to acquire sufficient information about the regulatory conditions of foreign markets relative to larger firms. It would make sense to include material on digital nomads in the SME committee, since they may be even more vulnerable to information asymmetries then smaller firms geographically established with permanence. Some believe that the digital trade provisions of the CPTPP have been constructed to protect the first-mover advantage and oligopolistic power of large technology multinationals. In that sense, the rules are of limited use to smaller firms49 such as digital nomads. Given the rise in prominence of digital nomads, it is unfortunate that the highly innovative DEPA did not make Art 14.1. For example, Canada–Colombia FTA, Art 1507. 46 Only about 35 per cent of the population in developing countries has access to the Internet (versus 80% in developed countries): ‘Connecting for Inclusion: Broadband Access for All’, The World Bank, www​.worldbank​.org/​en/​topic/​digitaldevelopment/​brief/​connecting​-for​-inclusion​-broadband​-access​-for​ -all (undated, accessed October 2022). 47 Art 24.1. 48 Art 24.2. 49 J Kelsey, ‘How a TPP-Style E-commerce Outcome in the WTO Would Endanger the Development Dimension of the GATS Acquis (and Potentially the WTO)’ (2018) 21(2) J. of Int’l. Econ. L. 273. 44 45

340  Research handbook on digital trade specific reference to them in its module on Digital Inclusion, as a category of business which has not been prioritised in typical FTAs. One might expect this addition in a future iteration of the DEPA, given that it is conceived as a ‘living agreement’. Furthermore, the CPTPP’s obligations, as with most FTAs, are enforceable through only through state-to-state and in some cases investor-state arbitration, both of which can be prohibitively costly for small firms, whether acting as a complainant (in investor-state arbitration) or as a lobbyist (in state-to-state arbitration). These mechanisms are often seen as precisely those which multinational enterprises would seek to put in place to maintain their dominance.50 4.3

Digital SMEs and the WTO

As for multilateral rules under the WTO, there is not much in the way of recognition of the needs of digital nomads. SMEs do not feature at all in the text of the GATT or the GATS, for example.51 Furthermore, while large enterprises and industry associations are able to initiate complaints through the WTO Dispute Settlement System as a kind of ‘indirect access’,52 the political power of individual SMEs is limited, and that of digital nomads even less so. Some believe that SMEs from developing countries in particular have suffered under the WTO because it has exposed these more fragile firms to the rigours of global trade competitiveness53 without commensurate access to justice. Who speaks for the digital nomad who has had their market access curtailed, for example, contrary to GATS commitments? Encouragingly, the WTO has begun to acknowledge the need to empower SMEs as participants in the global economy. This could afford some protection for digital nomads. In 2021, WTO Deputy Director-General Xiangchen Zhang drew attention to the importance of targeted policies and a coordinated global response to mitigate the impact of the COVID-19 pandemic on SMEs. As suggested earlier, the COVID-19 pandemic is in large part responsible for the need for individuals to pursue nomadic commercial endeavours as opposed to stable, geographically rooted ones. Observing that SMEs ‘are the backbone of many economies’, Zhang called on WTO members to ‘foster a transparent, inclusive, non-discriminatory, and predictable global trade environment that supports and enhances MSMEs’ [Micro, Small and Medium-Sized Enterprises] involvement in international trade’.54 A number of initiatives designed to assist SMEs have been put forth by the WTO in recent year. The Informal Working Group on MSMEs, consisting of 91 WTO members, was launched at the 11th Ministerial Conference in December 2017. It aims to identify and address obstacles 50 G Greenleaf, ‘Looming Free Trade Agreements Pose Threats to Privacy’ (2018) 152 Privacy L. & Bus. Int’l. Rep. 23; [2018] UNSWLRS 38, at 5. 51 Art 6.13 of the Anti-Dumping Agreement requires national authorities to take the special circumstances of SMEs into account and provide them with any practical assistance when conducting dumping investigations. 52 P van den Bossche and W Zdouc, The Law and Policy of the World Trade Organization (4th ed, Cambridge University Press, 2017) at 184. 53 J Sonwalkar and N Soni, ‘Challenges Faced by Small and Medium Sized Enterprises (SMEs) Including Indian Small Scale Industries and WTO’ (2017) 9(3) Int’l. J. of Info., Bus. and Management 62 who write at 64 that SMEs in India ‘lost the shield of protective policy of the government and have been showing to the onslaught of multinational organization and international competition’. 54 WTO, ‘Coordinated Global Response Key to MSMEs’ Post-Pandemic Economic Recovery – DDG Zhang’ (WTO, 22 October 2021) www​.wto​.org/​english/​news​_e/​news21​_e/​msmes​_22oct21​_e​.htm accessed October 2022.

The digital nomad and the emergence of global labour mobility  341 to SME participation in international trade.55 There is also the Global Trade Helpdesk, an online platform that simplifies market research for SMEs by integrating trade and business information into a single online portal. In effort to address the issue of information asymmetry, this portal provides data on applicable tariffs and taxes; relevant health and safety standards and compliance procedures; details about export and import procedures such as pre-shipment formalities; and current trade patterns and trade agreements.56 Portals of this nature could be very useful to digital nomads, who may lack access to legal advice. There is also the WTO Working Group on Trade, Debt and Finance. This body investigates ways to improve the availability of trade finance, including for SMEs. Access to finance is one of the chief barriers to smaller firms participating in global markets, although digitalisation of financial services may have diminished this problem to a degree.57 It is conceivable that digital nomads could take advantage of these services. The Council for Trade-related Aspects of Intellectual Property Rights (TRIPS) allows WTO members to exchange information about their policies designed to support SMEs’ creativity, inventiveness and investments in research and technology. This could be vitally important for digital nomads working in innovative sectors where new ideas and approaches are key to securing customers. Intellectual property could be the primary asset for many digital nomads. The SME-related policies discussed in this Council include financial assistance schemes, streamlining application procedures and enhanced transparency of intellectual property rules.58 The WTO’s Trade Dialogues series, where stakeholders in the private sector can highlight issues of relevance to their engagement in international trade, was established in part to address the concerns of SMEs.59 Most importantly for the needs of digital nomads, Digital Champions for Small Business is a WTO joint initiative launched by the Informal Working Group on Micro, Small and Medium-sized Enterprises in partnership with the International Chamber of Commerce and the International Trade Centre to help small businesses go digital and increase their participation in international trade.60

5. CONCLUSION Although much has been made of post-COVID trade and workplace trends such as the ‘Great Resignation’61 or the ‘Great Disengagement’, the evidence does not suggest that a plurality of 55 WTO, ‘New Initiatives on Electronic Commerce, Investment Facilitation and MSMEs’ (WTO, 13 December 2017) www​.wto​.org/​english/​news​_e/​news17​_e/​minis​_13dec17​_e​.htm accessed 29 November 2021. 56 WTO, ‘ITC, UNCTAD, WTO Launch Global Trade Helpdesk’ (WTO, 11 December 2017) www​ .wto​.org/​english/​news​_e/​news17​_e/​bus​_11dec17​_e​.htm accessed October 2022. 57 WTO, ‘Trade Finance’ www​.wto​.org/​english/​thewto​_e/​coher​_e/​tr​_finance​_e​.htm accessed October 2022. 58 WTO, ‘TRIPS – Trade-Related Aspects of Intellectual Property Rights’ www​.wto​.org/​english/​ tratop​_e/​trips​_e/​trips​_e​.htm accessed 29 November 2021. 59 WTO, ‘Trade Dialogues’ www​.wto​.org/​english/​res​_e/​reser​_e/​tradedialogues​_e​.htm accessed 29 November 2021. 60 ICC, ITC, and WTO MSME Group, ‘Digital Champions for Small Business: Call For Proposals’ www​.wto​.org/​english/​tratop​_e/​msmesandtra​_e/​dcsm​_callforproposals​.pdf accessed 29 November 2021. 61 A Serenko, ‘The Great Resignation: The Great Knowledge Exodus or the Onset of the Great Knowledge Revolution?’ (2022) Journal of Knowledge Management www​.emerald​.com/​insight/​content/​ doi/​10​.1108/​JKM​-12​-2021​-0920/​full/​html (accessed October 2022).

342  Research handbook on digital trade workers will soon be sipping beverages on a tropical beach while taking glamorous images of their new WFH reality. However, the location flexibility, labour mobility, workplace atomisation presented by the digital nomad lifestyle pose major challenges to the status quo in digital trade regarding the openness of markets to the provision of services across borders. While capital, goods and – to a lesser extent – services have long enjoyed significant freedom of movement, the same has never been entirely true of workers. Even with the EU’s ‘Four Freedoms’, only the movement of people/workers saw serious ongoing constraints. Until now, the de facto position was, therefore, that labour mobility was seriously constrained as a knock-on effect of the limited mobility of human workers. Labour challenges, such as the rights of digital nomads to unionise in order to address declines in working conditions, have not been discussed in this chapter. These concerns will only intensify as digitalisation enables white-collar work to be performed anywhere – spurring a shift in offshoring to jurisdictions where labour costs are lowest.62 While the notion of a digital nomad is burdened by some of the same assumptions and linguistic connotations as such terminology as ‘influencers’ versus ‘creators’63 and ‘migrants’ versus ‘expats’,64 it is clear that some elements of the technologically mobile digital nomad lifestyle has become a practical (if not operational) reality for more than 40 per cent of the modern workforce.65 It is mainly cultural assumptions and expectations that currently hold back the more widespread adoption of geographically distributed and asynchronous patterns of work. Current trends in platform-based employment and crowd-sourced work suggest that – if or when these assumptions and expectations change – a significant proportion of the global population will at least partially embrace the digital nomad lifestyle. This will have unpredictable impacts on population density in major urban areas as well as employment rates throughout the Western world and the Global North. Coming alongside the increasing use of automation and artificial intelligence for commercial purposes, it is clear that information technology has fundamentally changed the nature of work, and that the resulting global mobility of labour is poised to totally upend the traditional concept of employment and accelerate the third wave (we may later term it the start of the fourth wave) of globalisation,66 with comparatively reduced ecological impacts and dramatically increased socioeconomic and cultural impacts. This is because it is no longer necessary to physically transport human beings across the planet in order to engage in meaningful conversation and communication. Since digital nomads are ultimately a variant of SMEs, international rules on digital trade aimed at addressing the needs of SMEs could be relevant in ensuring the ongoing market access of these modern forms of business. Provisions eliminating customs duties on electronic

62 R Foroohar, ‘Digital Trade Must Not Become a Zero Sum Game’ The Financial Times (28 November 2022). 63 EG Ellis, ‘Why Women Are Called “Influencers” and Men “Creators”’ (2019) Wired Magazine www​.wired​.com/​story/​influencers​-creators​-gender​-divide/​ accessed October 2022. 64 S Kunz. ‘Expatriate, Migrant? The Social Life of Migration Categories and the Polyvalent Mobility of Race’ (2020) 46(11) Journal of Ethnic and Migration Studies 2145 www​.tandfonline​.com/​ doi/​abs/​10​.1080/​1369183X​.2019​.1584525 (accessed October 2022). 65 M Orel, ‘Coworking Environments and Digital Nomadism: Balancing Work and Leisure whilst on the Move’ (2019) 61(3) World Leisure Journal 2 www​.tandfonline​.com/​doi/​abs/​10​.1080/​16078055​ .2019​.1639275 (accessed October 2022). 66 F J. Lechner and J Boli, The Globalization Reader (6th edn, Wiley, 2019) 574 https://​elib​.maruzen​ .co​.jp/​elib/​html/​BookDetail/​Id/​3000098227.

The digital nomad and the emergence of global labour mobility  343 transmissions and prohibiting data localisation requirements found in FTAs such as the CPTPP can work to level the playing field for the nomadic commercial model. For its part, the WTO has shown an awareness of the importance of addressing the need of SMEs, since they are becoming key drivers of the digital economy. With these developments in mind, digital nomads, which may have adopted their business practices out of exigency, may ultimately choose this lifestyle because it is as economically profitable as conventional trade thanks to the enabling force of digital technology, once properly liberalised from unnecessary regulatory encumbrances.

PART VI DIGITAL TRADE LEGAL AND POLICY CHALLENGES: ARTIFICIAL INTELLIGENCE

21. The principles of algorithmic justice in the digital market Maciej Hulicki

1. INTRODUCTION1 The notion ‘algorithm’ lacks a legal definition; therefore, to determine its scope, it will often be necessary to refer to computer sciences. There, an algorithm can be defined as a ‘well-ordered collection of unambiguous and effectively computable operations that when executed produces a result and halts in a finite amount of time’.2 Other definitions emphasise other aspects, indicating that an algorithm is a solution to a given problem,3 a pre-set decision mechanism4 or a ‘recipe’, consisting of an input, a set of instructions and an output.5 In other words, it is a technical instruction as to how various systems, applications and devices operate. Moreover, advanced algorithms comprise the essence of artificial intelligence (hereinafter AI) systems.6 This paper is part of the research project ‘Algorithmic contract as a challenge for commercial law’ (no. 2019/35/D/HS5/04377), financed by the National Science Center, Poland. 2 Schneider M., Gersting J., An Invitation to Computer Science, New York 1995, p.9. 3 Britannica, T. Editors of Encyclopaedia, ‘Algorithm’, in: Encyclopedia Britannica, www​.britannica​ .com/​science/​algorithm, accessed 15 April 2022. 4 Gal M., Algorithmic-facilitated Coordination, OECD’s Roundtable on Algorithms and Collusion, 22 June 2017, p.7. 5 The notion of an algorithm is similar to the concept of a computer program: algorithm has a more general meaning, whereas program denotes rather an implementation of an algorithm, which has been specified in a programming language, in a sense resembling the Turing machine: ‘The idea behind digital computers may be explained by saying that these machines are intended to carry out any operations which could be done by a human computer.’ Turing A.M., Computing Machinery and Intelligence, Mind, Vol. LIX, Issue 236, October 1950, pp.433–60, available at: https://​academic​.oup​.com/​mind/​article/​LIX/​ 236/​433/​986238. 6 The term ‘artificial intelligence’ is complex, but one of the most prominent definitions, expressed by J. McCarthy, explains that ‘it is the science and engineering of making intelligent machines, especially intelligent computer programs’ (see: McCarthy J. (2007). What Is Artificial Intelligence? Technical report, Stanford University, http://​jmc​.stanford​.edu/​artificial​-intelligence/​what​-is​-ai/​index​ .html, accessed 31 October 2022). Algorithms constitute the core of any computer program and can be regarded as ‘building blocks of computer science’ (see: Hill R.K., What an Algorithm Is, Philosophy and Technology, No. 29/2016, pp.35–59). In fact, one can note that definitions of ‘algorithm’ and ‘computer program’ to a great extent overlap (for example, see ‘algorithm’ and ‘computer program’ in Britannica Encyclopedia [britannica.com], accessed 31 October 2022). Algorithms implement the methods of solving complex information processing problems, what can be seen as the essence of AI (Marr D., Artificial Intelligence – A Personal View, Artificial Intelligence, Vol. 9/1, pp.37–48). AI systems are based on algorithms employing advanced technologies, such as machine learning, neural networks, and deep learning (see: Kavlakoglu E., AI vs. Machine Learning vs. Deep Learning vs. Neural Networks: What’s the Difference? IBM www​.ibm​.com/​cloud/​blog/​ai​-vs​-machine​-learning​-vs​-deep​-learning​-vs​ -neural​-networks, accessed 31 October 2022). However, what differs between an AI system and other computer programs, is that it has capacity to take autonomous actions (see definition of ‘artificial intelligence’ in the European Parliament’s resolution of 20 January 2021 on artificial intelligence: questions 1

345

346  Research handbook on digital trade Therefore, algorithms play a fundamental role in the digital economy. In fact, an assumption can be made that the digitalized economy is based on three pillars: sets of data (personal and non-personal); algorithms, which process that data to operate different products and services; and platforms, which use them to facilitate the functioning of various digital solutions and accomplishing remote/intelligent services.7 The COVID-19 pandemic was a catalyst for the processes of digitalization, where remote work, and the use of numerous digital services, have developed to an unparalleled level, and therefore the post-pandemic world will see a rapid application of AI in new fields and solutions. Generally, the legal situation of algorithms’ use in the digital economy remains unregulated. Whereas there might be some exceptions, it may even seem paradoxical that modern societies regulate most aspects of economic activity, but algorithms, which constitute the very fundament of the new economy, remain to a great extent unregulated. Taking into account that in cyberspace algorithms can regulate which actions are allowed/disallowed or which content is available, or enforce contracts, the scope of the digital market will be determined by the scope of the algorithm implementation.8 The crucial role of algorithms and related technologies (for example, machine learning, big data, automated decision-making) in modern society should not lead us to disregard the risks associated with their implementation.9 Algorithms have been studied from a legal perspective in numerous contexts, in particular concerning their impact on law,10 the need for regulation,11 their transparency and accounta-

of interpretation and application of international law in so far as the EU is affected in the areas of civil and military uses and of state authority outside the scope of criminal justice (2020/2013(INI)), Official Journal of the European Union 10.11.2021, C 456/34). 7 At present every other organization uses AI systems in conducting at least one of their core business functions. See: Balakrishnan T. et al., The State of AI in 2020, McKinsey 2020, www​.mckinsey​.com/​~/​ media/​McKinsey/​Business​%20Functions/​McKinsey​%20Analytics/​Our​%20Insights/​Global​%20survey​ %20The​%20state​%20of​%20AI​%20in​%202020/​Global​-survey​-The​-state​-of​-AI​-in​-2020​.pdf, accessed 15 April 2022. It is estimated that up to 2030 the contribution of such solutions to the global economy will equal 15.7 trillion USD. See: Rao A., Verweij G., Sizing the Prize: What’s the Real Value of AI for Your Business and How Can You Capitalise?, PwC, 2017, www​.pwc​.com/​gx/​en/​issues/​analytics/​ assets/​pwc​-ai​-analysis​-sizing​-the​-prize​-report​.pdf, accessed 15 April 2022. The World Economic Forum estimates that in 2022 AI will be responsible for the creation of 58 million jobs. Chowdhry, A., Artificial Intelligence to Create 58 Million New Jobs By 2022, Says Report, Forbes, www​.forbes​.com/​sites/​ amitchowdhry/​2018/​09/​18/​artificial​-intelligence​-to​-create​-58​-million​-new​-jobs​-by​-2022​-says​-report/​ #74dd80624d4b, accessed 15 April 2022. 8 See: Kenney M., Zysman J., The Rise of the Platform Economy, Issues in Science and Technology, No. 32/3, 2016, pp.61–9, https://​issues​.org/​rise​-platform​-economy​-big​-data​-work. 9 Blacklaws C., Algorithms: Transparency and Accountability, Philosophical Transactions of the Royal Society A, No. 376/2128, 2018, p.1. 10 See e.g.: Lessig L., Code Is Law: On Liberty in Cyberspace, Harvard Magazine, January 2000, www​.harvardmagazine​.com/​2000/​01/​code​-is​-law​-html, accessed 15 April 2022; Noto La Diega G., Against the Dehumanisation of Decision-Making – Algorithmic Decisions at the Crossroads of Intellectual Property, Data Protection, and Freedom of Information, Journal of Intellectual Property, Information Technology and E-Commerce Law, Vol. 9/2018, pp.3–34; Hildebrandt M. Algorithmic Regulation and the Rule of Law, Philosophical Transactions of the Royal Society A, No. 376/2128, 2018, pp.1–11. 11 See e.g.: Buiten M., Towards Intelligent Regulation of Artificial Intelligence, European Journal of Risk Regulation, Vol. 10/1, pp.41–59.

The principles of algorithmic justice in the digital market  347 bility,12 their explainability13 and their relevance to competition law14 and contract law15. With the ever-rising importance of algorithms for the economy, the amount of legal scholarship is increasing in parallel. In particular, algorithmic justice has been the subject of various research papers concerning automated decision-making in the legal environment, especially in the criminal law and civil rights fields.16 This chapter analyses what the actual meaning of algorithmic justice is, and its goal is to illustrate how this notion should be understood in the circumstances of the digital market and determine what features algorithm decision-making should have in order to ensure that the entire process is following the rule of law. The chapter will analyse documents of international organizations concerning ethical AI to find out what could be the common standard for algorithms utilized in the digital economy. Contrary to the frequently expressed idea that algorithms pose a threat to modern society and the rule of law, this paper follows the assumption that they are not necessarily hazardous, but if applied properly could enhance the due functioning of the market.

2.

THE CONCEPT OF ALGORITHMIC JUSTICE

Simultaneously with the development of novel technologies and their application in the economy, almost every aspect of life is subject to an irreversible and comprehensive transfor-

See e.g.: Hulicki M., Algorithm Transparency as a Sine Qua Non Prerequisite for a Sustainable Competition in a Digital Market? EU and Comparative Law Issues and Challenges Series (ECLIC), No. 5/2021, pp.238–61. Blacklaws, Algorithms, Felzmann H., Fosch-Villaronga E., Lutz C. et al., Towards Transparency by Design for Artificial Intelligence, Science Engineering Ethics, Vol. 26/2020, pp.3333–61; Olhade S., Rodrigues R., Fairness and Transparency in the Age of the Algorithm, Significance Magazine, April 2017, pp.8–9; Monterossi M.W., Algorithmic Decisions and Transparency: Designing Remedies in View of the Principle of Accountability, The Italian Law Journal, Vol. 5/2, 2019, pp.711–30. Desai, D.R., Kroll, J.A., Trust But Verify: A Guide to Algorithms and the Law, Harvard Journal of Law and Technology, Vol. 31/1, 2017, pp.1–64. 13 See e.g.: Deeks A., The Judicial Demand for Explainable Artificial Intelligence, Columbia Law Review, Vol. 119, 2019, pp.1829–50; Malgieri G., Automated Decision-Making in the EU Member States: The Right to Explanation and Other ‘Suitable Safeguards’ in the National Legislations, Computer Law and Security Review, Vol. 35, 2019, pp.1–26. 14 See e.g.: Ezrachi A., Stucke M.A., Virtual Competition: The Promise and Perils of the Algorithm-Driven Economy, Harvard University Press, 2016; Mehra S., Algorithmic Competition, Collusion, and Price Discrimination, in: Barfield W. (ed.), The Cambridge Handbook of the Law of Algorithms, Cambridge University Press, 2020, pp.199–208; Mehra S., US v. Topkins: Can Price Fixing Be Based on Algorithms? Journal of European Competition Law & Practice, Vol. 7/7, July 2016, pp. 470–474. 15 See e.g.: Scholz L., Algorithmic Contracts, Stanford Technology Law Review Vol. 20/2, 2017, pp.128–69. 16 See e.g.: Završnik A., Algorithmic Justice: Algorithms and Big Data in Criminal Justice Settings, European Journal of Criminology, 2019, pp.623–42; Freeman K., Algorithmic Injustice: How the Wisconsin Supreme Court Failed to Protect Due Process Rights in State v. Loomis, North Carolina Journal of Law & Technology, 2016, Vol. 18/5, pp.75–106. Williams J.D., Lopez D., Shafto P. et al., Technological Workforce and Its Impact on Algorithmic Justice in Politics. Customer Needs and Solutions, Vol. 6, 2019, pp.84–91. Toohey L., Moore M., Dart K., Toohey D., Meeting the Access to Civil Justice Challenge: Digital Inclusion, Algorithmic Justice, and Human-Centered Design, Macquarie Law Journal, vol. 19, 2019, pp.133–56. 12

348  Research handbook on digital trade mation. The information society is no longer only an interesting concept but an all-embracing reality, conjoining the tangible world and cyberspace, which are no longer alternative realities but a common dimension of life. Nonetheless, the legal domain has not yet changed in a complex way. Technological solutions are increasingly used in law, but decision-making is still largely reserved for humans. The main argument against automated decision-making is based on the belief that only people can comprehend the complexity of legal problems and make decisions which would reflect legal norms fairly. However, the development of intelligent systems based on machine learning and neural networks can change such attitudes, and in fact, exploitation of the AI systems in the legal domain seems inevitable. Most often the term ‘algorithmic justice’ (or ‘automatic justice’17) is used to refer to automation in the decision-making process within the criminal justice system,18 or more broadly enhancing the non-discriminatory nature of how algorithms work (and are being trained). Such an approach to algorithms is based on a premise that they often involve a pre-coded bias.19 The popularity of the abovementioned notion intensified due to the prominent case State v Loomis (2016), where the IT system used in precognition of committing a crime20 in the future helped in the conviction of Mr Loomis for taking part in a shootout.21 It is not unusual for the conviction courts in the United States to utilize recidivism risk assessments,22 and evidence-based decision-making systems are exploited not only in the law (in fact, in 17 Marks A., Bowling B., Keenan C., Automatic Justice? Technology, Crime and Social Control, in: Brownsword R., Scotford E., Yeung K. (eds), The Oxford Handbook of the Law and Regulation of Technology, Oxford University Press, 2017. 18 Završnik, Algorithmic Justice, pp.3–5. 19 Based on such an argument, an organization called Algorithmic Justice League was founded. It aims to raise public awareness and fight against algorithms which are discriminatory and threaten democratic rules and constitutional rights. See Algorithmic Justice League’s website: www​.ajl​.org/​about, accessed 15 April 2022. 20 Such precognition systems resemble stories from sci-fi literature (e.g. P.K. Dick’s Minority Report, where a dystopic world was described in which police use a tool to predict crime before it is even committed, and where police can arrest potential offenders before the crime has been committed), but similar mechanisms are in fact utilized in contemporary judicial system. 21 State v Loomis, 2016 WI 68, 371 Wis. 2d 235, 881 N.W.2d 749. The COMPAS software was used for retrospective analysis of risk factors for deciding on the penitentiary/custody/resocialization measures and prospective risk analysis and potential for more severe punishment. The judgment referred to the decision of analysis conducted with the use of COMPAS software and on such grounds, a higher sentence was imposed. The software analysis showed that there was a high risk of recidivism and further crime, thus also the risk for the society was assessed as high. Mr Loomis claimed that by using COMPAS the court infringed his right to due process and the right to individualized punishment. Moreover, he claimed that as the software used such data as race or gender it was discriminatory and violated his right to be convicted based on accurate information. What is important is the fact that the methodology of the algorithm operation was protected through trade secrets, therefore the court had access only to the conclusions of the analysis. However, the court noted that the software was not a key factor in the case; facts such as committing crime, fleeing from the scene or posssession of a criminal record were substantial aspects considered for the conviction. Despite using the software, the court maintained judicial discretion and it could still acknowledge or discard the results of the analysis. The computer program served only as a useful tool; the defendant could question the facts, which were inputted into the system, and the inclusion of the social factors in the analysis made it more precise. However, the court has stated that such software should not be the sole basis for a decision on the conviction. 22 Criminal Law – Sentencing Guidelines – Wisconsin Supreme Court Requires Warning Before Use of Algorithmic Risk Assessments in Sentencing – State v Loomis, 881 N.W.2d 749 (Wis. 2016), Harvard Law Review, vol. 130/2017, p.1530.

The principles of algorithmic justice in the digital market  349 some states they are part of the criminal justice system) but also, for example, in the medical domain.23 As the Electronic Privacy Information Center’s report has highlighted, such systems are used in courts to set bail, make judgments and sometimes even to decide on guilt.24 In 2017 the American Law Institute acknowledged the value of evidence-based sentencing through tools which can assess risks that offenders pose to public safety.25 In the academic literature, the Wisconsin Supreme Court’s decision in State v Loomis generally was a subject of critique.26 Criticism of involving risk assessment software in the courts concerned a lack of judicial scepticism, the impossibility of assessing the software itself and informal pressure on the courts to use such systems in convicting decisions.27 Moreover, ProPublica’s report claimed that the court’s assessment involved inappropriate risk analysis based on racial factors and therefore comprised machine bias.28 Parties affected by automated decisions wish to know how the algorithm works and on what basis the results of the automated analysis were made.29 Incomprehensive information, which disallows challenging the accuracy of information may violate the right to due process.30 One of the main objections to the use of automated systems in the legal domain is focused on the opacity of such systems, which are secretive and protected through intellectual property rights. Since parties do not have access to them, they will find it exceptionally difficult to challenge them. Hence, reform of trade secret law has been proposed in the literature31 and some authors suggest that in the criminal justice system only open-source algorithms should be used, in particular where they will be audited in terms of the scope and methodology of data being processed.32 Even though ‘algorithmic justice’ is a term which refers generally to the use of advanced algorithms in the domain of criminology, criminal procedure and civil rights law, it can be

Forward J., The Loomis Case: The Use of Proprietary Algorithms at Sentencing, Inside Track. Bi-Weekly Newsletter of the State Bar of Wisconsin, Vol. 9/14, 19 July 2017, www​.wisbar​.org/​ NewsPublications/​InsideTrack/​Pages/​Article​.aspx​?Volume​=​9​&​Issue​=​14​&​ArticleID​=​25730, accessed 15 April 2022. 24 Liberty at Risk: Pre-Trial Risk Assessment Tools in the U.S., Electronic Privacy Information Center, September 2020, p.1. 25 Model Penal Code: Sentencing § 6B.09(3), American Law Institute, 2017. 26 See: Sidhu D., Moneyball Sentencing, Boston College of Law Review, Vol. 56, 2015, p.671; Freeman, Algorithmic Injustice; Israni E., Algorithmic Due Process: Mistaken Accountability and Attribution in State v. Loomis, Jolt Digest, Harvard University, 31 August 2017, https://​jolt​.law​ .harvard​.edu/​digest/​algorithmic​-due​-process​-mistaken​-accountability​-and​-attribution​-in​-state​-v​-loomis​ -1, accessed 15 April 2022. See also an earlier work: Starr S.B., Evidence-Based Sentencing and the Scientific Rationalization of Discrimination, Stanford Law Review, Vol. 66/4, 2014, pp.803–72. 27 Criminal Law – Sentencing Guidelines, p.1531. 28 Angwin J., Larson J., Mattu S. and Kirchner L., Machine Bias. There’s Software Used across the Country to Predict Future Criminals. And It’s Biased against Blacks, ProPublica, 23 May 2016, www​ .propublica​.org/​article/​machine​-bias​-risk​-assessments​-in​-criminal​-sentencing, accessed 15 April 2022. 29 Villasenor J. and Foggo V., Algorithms and Sentencing: What Does Due Process Require? Brookings Institution, 21 March 2019 www​.brookings​.edu/​blog/​techtank/​2019/​03/​21/​algorithms​-and​ -sentencing​-what​-does​-due​-process​-require/​, accessed 15 April 2022. 30 Ibid. 31 Moore T., Trade Secrets and Algorithms as Barriers to Social Justice, Center for Democracy & Technology, 2017, https://​cdt​.org/​files/​2017/​08/​2017​-07​-31​-Trade​-Secret​-Algorithms​-as​-Barriers​-to​ -Social​-Justice​.pdf, accessed 15 April 2022. 32 Freeman, Algorithmic Injustice. 23

350  Research handbook on digital trade also used in a broader sense in the context of algorithmization of law, that is, the application of computer programs, especially AI systems, in decision-making processes, having legal effects.33 Importantly, the problems associated with the notion of algorithmic justice are common across different legal domains, although the implications in specific areas can be very different. Hence, the evaluation of this term should not be limited purely to the field of criminal law. The essence of the problem may have been formulated by A. Wolfe, who understands algorithmic justice as a finite system, where there are ‘no truths or standards outside the operation of a system’ and ‘the rules that structure the system lie within the system’, where ‘each system is governed by its own laws, and such laws have as their goal the reproduction of whatever system in which they are found’.34 In such a meaning ‘algorithms are rules designed to be followed with as little interpretive variation as possible’.35

3.

A MODEL OF ALGORITHMIC JUSTICE

Since algorithms play a pivotal role in the digital economy, their utilization may also pose many fundamental challenges for the legal system, and at times they can have detrimental effects on the digital market. It is therefore key to search for legal solutions which would provide for the proper functioning of algorithms in the market and therefore ensure algorithmic justice, a notion that could very well apply not only in the field of criminal law but also in the context of the digital market. Although the theory of justice is a subject matter which extends beyond the scope of this article, one can refer to the definition of justice, which envisages that it is ‘the ethical, philosophical idea that people are to be treated impartially, fairly, properly, and reasonably by the law and by arbiters of the law, that laws are to ensure that no harm befalls another, and that, where harm is alleged, both the accuser and the accused receive a morally right consequence merited by their actions’.36 Hence, it is apparent that the concept of justice implies such elements as non-discrimination, fairness and non-arbitrary treatment. In the academic literature, the model of algorithmic decision-making has been a subject of intensive critique, and the notion of ‘algorithmic pollution’ was coined. This notion refers to ‘the presence of unjustified, unfair, discriminatory, or other harmful consequences of algorithmic decision-making for individuals, groups, organizations, sections of the population, the economy, or society at large’.37 If the algorithmic justice is an antitype of

33 In particular with respect to the public institutions. Keddell E., Algorithmic Justice in Child Protection: Statistical Fairness, Social Justice and the Implications for Practice. Social Sciences, Vol. 8/10, 2019, p.281; Sun M., Gerchick M., The Scales of (Algorithmic) Justice: Tradeoffs and Remedies, AI Matters, Vol. 5, 2019, pp.30–40. Bavitz C., Hessekiel K., Algorithms and Justice. Examining the Role of the State in the Development and Deployment of Algorithmic Technologies, The Berkman Klein Center for Internet & Society at Harvard University, 11 July 2018, https://​cyber​.harvard​.edu/​story/​2018​ -07/​algorithms​-and​-justice, accessed 15 April 2022. 34 Wolfe A., Algorithmic Justice, Cardozo Law Review, Vol. 11/1990, p.1414. 35 Ibid., pp.1414–15. 36 ‘Justice’, Wex, Legal information Institute, Cornell University, www​.law​.cornell​.edu/​wex/​justice, accessed 15 April 2022. 37 Marjanovic O., Cecez-Kecmanovic D., Vidgen R., Algorithmic Pollution: Understanding and Responding to Negative Consequences of Algorithmic Decision-Making. Working Conference on Information Systems and Organizations (IS&O), December 2018, San Francisco. pp.31–47. See also:

The principles of algorithmic justice in the digital market  351 algorithmic pollution, then it denotes such functioning of the algorithms, which is justified, fair, non-discriminatory, and unharmful. A common standard of ‘algorithmic justice’ could be extracted from the official documents of international organizations (and a few other scientific institutions) which have taken initiatives to determine the principles for ethical AI. In the European Union (hereinafter EU), in 2019 the High-Level Expert Group on Artificial Intelligence produced ethical guidelines for trustworthy AI, which included seven fundamental requirements for ensuring the practical realization of the concept of trustworthy AI. They include human agency and oversight, technical robustness and safety, privacy and data governance, transparency, diversity, non-discrimination and fairness, societal and environmental well-being and accountability. Moreover, these criteria are based on four main ethical principles – respect for human autonomy, prevention of harm, fairness, and explicability – and should be evaluated throughout the entire lifecycle of an AI system.38 On the other hand, the Organization for Economic Cooperation and Development’s (hereinafter OECD) Council recommended only five principles for trustworthy AI, which are focused on inclusive growth, sustainable development and well-being, human-centred values and fairness, transparency and explainability, robustness, security and safety, and accountability. All of them are complementary to each other.39 A different approach was taken in the Montreal Declaration for responsible AI development, which was announced in 2017 at the Forum on the Socially Responsible Development of AI in Montreal. The document aims to stimulate the debate on the development of AI and promote inclusivity in its development. The Montreal criteria comprise the well-being of all sentient beings, privacy and intimacy, respect for autonomy, responsibility, democratic participation, equity, solidarity, diversity inclusion, prudence and sustainable development.40 A more comprehensive approach, reconciling requirements from other documents can be observed in ‘A Framework for Ethical AI at the United Nations’ produced by the United Nations (hereinafter UN) Office for Information and Communications Technology includes just a few crucial aspects of the ethical AI, which are however further elaborated in the text of the document. They include beneficence, non-maleficence, accountability, justice, competence, and governance.41 A slightly different approach is visible in the Statement on Algorithmic Transparency and Accountability of Association for Computer Machinery’s (hereinafter ACM) US Public Policy Council and of ACM Europe Policy Committee of 2017, which indicates seven principles for algorithmic transparency and accountability: awareness,

Marjanovic O., Cecez-Kecmanovic D., Vidgen R., Theorising Algorithmic Justice, European Journal of Information Systems, Vol. 31/3, 2022, pp.269–87. 38 High-Level Expert Group on Artificial Intelligence, Ethics Guidelines for Trustworthy AI, Brussels 2019, p.18. 39 Recommendation of the Council on Artificial Intelligence, OECD/LEGAL/0449, OECD 2021, p.8. 40 Montreal Declaration for a Responsible Development of Artificial Intelligence, Montreal 2018 www​.montrealdeclaration​-responsibleai​.com/​the​-declaration, accessed 15 April 2022. 41 A Framework for Ethical AI at the United Nations, UN Office for Information and Communications Technology, 15 March 2021, Unite Paper 2021(1), https://​unite​.un​.org/​sites/​unite​.un​.org/​files/​ unite​_paper​_​-​_ethical​_ai​_at​_the​_un​.pdf, accessed 15 April 2022.

352  Research handbook on digital trade Table 21.1 Organization

Requirements

European Union (Ethics guidelines for trustworthy AI,

● Human agency and oversight

High-Level Expert Group on Artificial Intelligence)

● Technical robustness and safety ● Privacy and data governance ● Transparency ● Diversity, non-discrimination and fairness ● Societal and environmental well-being ● Accountability

United Nations (A Framework for Ethical AI at

● Beneficence

the United Nations, UN Office for Information and

● Non-maleficence

Communications Technology)

● Accountability ● Justice (includes various aspects) ● Competence ● Governance

Organization for Economic Cooperation and

● Inclusive growth, sustainable development, and well-being

Development (Recommendation of the Council on

● Human-centered values and fairness

Artificial Intelligence)

● Transparency and explainability ● Robustness, security, and safety ● Accountability

Association for Computing Machinery (Statement on

● Awareness

Algorithmic Transparency and Accountability)

● Access and redress ● Accountability ● Explanation ● Data provenance ● Auditability ● Validation ● Testing

Montreal Declaration for responsible AI development

● Well-being ● Privacy and intimacy ● Respect for autonomy ● Responsibility ● Democratic participation ● Equity ● Solidarity ● Diversity inclusion ● Prudence ● Sustainable development

access, and redress; accountability; explanation; data provenance; auditability; validation; and testing.42 Whereas the approach taken in each of these documents is somewhat different, including the diverse scope and goals of specific requirements, overall comparison of them indicates several common features. Many of the abovementioned aspects are covered in all these documents,

Statement on Algorithmic Transparency and Accountability, Association for Computing Machinery, www​.acm​.org/​binaries/​content/​assets/​public​-policy/​2017​_joint​_statement​_algorithms​.pdf, accessed 15 April 2022. 42

The principles of algorithmic justice in the digital market  353 Table 21.2

Comparison of requirements of ethical AI, where the EU guidelines for trustworthy AI are a point of reference

European Union’s

United Nations’

OECD’s

Montreal Declaration for

Association for

ethics guidelines for

Framework for

Recommendation on

responsible AI development

Computer Machinery

trustworthy AI

Ethical AI

Artificial Intelligence

Human agency and

governance

respect for autonomy

awareness

+

prudence

testing, validation

+

+

oversight Technical robustness

non-maleficence,

and safety

competence

Privacy and data

non-maleficence

human-centered values

Transparency

accountability

+

Diversity,

justice

human-centered values

equity, diversity inclusion

inclusive growth,

well-being, solidarity,

governance auditability, explanation

non-discrimination and fairness Societal and

beneficence

sustainable development, democratic participation,

environmental well-being Accountability

+

and well-being

sustainable development

+

responsibility

accountability, access, redress

yet sometimes they are grouped in one category, are somehow more elaborate or simply have a different naming. However, one can notice that the ACM’s approach is more technical and focused on the issue of algorithmic transparency and accountability, hence aspects such as societal well-being or fairness are sidelined, whereas the Montreal Declaration is motivated by societal aspects in the development of AI systems. Nevertheless, there are noticeable similarities between the documents, and with some minor exceptions, they overlap. Since the EU approach seems most comprehensive, it can be used as a point of reference when analysing the notion of algorithmic justice concerning its application in the digital market. It should be highlighted that one of the UN’s requirements for ethical AI involves ‘justice’, which includes many aspects, such as fairness, diversity or redress. However, in terms of determining the scope of algorithmic justice it is important not to focus solely on such aspects, as other issues, such as accountability or transparency, are central to ensuring the proper functioning of algorithms.

4.

PILLARS OF ALGORITHMIC JUSTICE

Since the EU document on trustworthy AI is the most comprehensive one, it is used as a point of reference concerning the analysis of the notion of algorithmic justice. Hence, seven requirements listed in the EU guidelines are analysed: they include human agency and oversight; technical robustness and safety; privacy and data governance; transparency; diversity, non-discrimination and fairness; societal and environmental well-being; and accountability.

354  Research handbook on digital trade 4.1

Human Agency and Oversight

EU guidelines understand this requirement as supporting human actions and decision-making in line with the general principle of respect for human autonomy. This means that this requirement encompasses two main elements: AI systems should enable and support fundamental rights (for example, applications enhancing education or tracking exploitation of personal data), and if the software utilization poses risks to fundamental rights and can hamper them an impact assessment should be conducted before the introduction of the product to the market, and should preferably involve a certain degree of external feedback.43 Human agency is an important aspect of this requirement. Individuals should be able to comprehend the functioning of such a system, interact with it, reasonably assess its functioning and be given tools which will assist them in this process. Additionally, the user’s autonomy should not be hindered by the functioning of the system, which translates to a ‘right not to be subject to decision based solely on automated processing when this produces legal effects on users or similarly significantly affects them’.44 The key aspect is the human oversight of the AI system and ensuring that it will not negatively impact individuals/society. Such oversight can take place in various timeframes, either during the design of the system or even throughout the whole lifecycle of the software, and should imply the involvement of the public authorities as well. Importantly, the less oversight is provided, the more testing and governance procedures are needed.45 It is also important that algorithms should not promote only one point of view/one lifestyle, and the AI should not only respect human autonomy but actually increase it.46 Moreover, all stakeholders should be aware of the potential biases in and the impact of the algorithms on society.47 It is complicated to regulate and have oversight of algorithms as such, since in the realities of a globalized digital economy one cannot impose responsibilities on them, and their creation may often be a subject of a distributed process. Hence, cooperation between oversight bodies and developers is needed to efficiently regulate them.48 4.2

Technical Robustness and Safety

Algorithms should be designed in such a way as to prevent inexcusable harm and minimize risks of unintentional wrongdoing.49 It is difficult to avoid faults and defects in algorithms, as they are developed by fallible human beings, and mistakes are calculated into the essence of algorithm making.50 That is why the OECD’s document refers to ‘unreasonable safety risk’.51 It is however crucial that the safety of AI systems should be ensured by adopting preventive measures. Such systems should be built to be safe and resilient to cyberattacks, in particular High-Level Expert Group on Artificial Intelligence, pp.15–16. Ibid., p.16. 45 Ibid. 46 Montreal Declaration for a Responsible Development of Artificial Intelligence, p.9. 47 Statement on Algorithmic Transparency and Accountability, p.2. 48 A Framework for Ethical AI at the United Nations, pp.19, 22. 49 High-Level Expert Group on Artificial Intelligence, p.16. 50 Although accuracy, reliability and reproducability comprise important elements of the technical robustness requirement. 51 Recommendation of the Council on Artificial Intelligence. 43 44

The principles of algorithmic justice in the digital market  355 concerning exposure to unexpected situations or malicious attacks, and fallback plans should be implemented for problematic situations. In some instances this may require human intervention. The level of safety measures is closely associated with the level of risks posed by the AI system, and with respect to the riskiest systems a proactive approach may be needed.52 The ACM highlights that organizations should use rigorous methods for validating algorithms through routine tests, should document this process and should make its results publicly available.53 Moreover, system developers should ensure traceability of processes, whereas autonomous decisions should apply systematic risk management across the entire lifecycle of the software product.54 The errors and flaws that are found within the systems should be publicly (and globally) distributed. If the system is found to pose a significant risk to the public, it is recommended to restrict access to it.55 4.3

Privacy and Data Governance

The AI should be deployed on the market in a manner which protects privacy. Data governance mechanisms employed should protect the quality and integrity of the data. Therefore, algorithms need to utilize data that is unbiased, accurate and correct. Additionally, it is important to inspect whether the data within the system has not been manipulated, for example, to provide more desirable results of data processing for manipulating entities. This requirement aims at protecting privacy as one of the fundamental rights, and the overarching goal is to prevent harm to individuals concerned.56 It should cover any aspect of data and situations which occur during the entire lifecycle of the product/service. Users should trust the system in the sense that it will not use their data unlawfully, contrary to their interests, to inflict harm on them or to discriminate against them. Access to personal data should be limited only to qualified and designated personnel within the organization. Moreover, the systems solicited from third parties should fulfil the same general requirements concerning data protection.57 What is more, individuals should be able to have control over their data and their preferences, in particular in terms of how it is used, collected and disseminated.58 In a wider context, including also non-personal data, the data which was used to train the system should be maintained for later scrutiny for possible biases in the data-gathering processes. Data management mechanisms should include state-of-the-art technology and methods.59 An important consideration, which must be resolved, is the tradeoff between privacy, trade secret protection and public scrutiny of the data.60

High-Level Expert Group on Artificial Intelligence, 16–17. Statement on Algorithmic Transparency and Accountability. 54 Recommendation of the Council on Artificial Intelligence. 55 Montreal Declaration for a Responsible Development of Artificial Intelligence, p.15. 56 High-Level Expert Group on Artificial Intelligence, p.17. 57 Ibid. 58 Montreal Declaration for a Responsible Development of Artificial Intelligence, p.10. 59 Recommendation of the Council on Artificial Intelligence. 60 Statement on Algorithmic Transparency and Accountability. 52 53

356  Research handbook on digital trade 4.4 Transparency Transparency is a focal point of algorithms regulation, and for ensuring the proper functioning of the algorithms in the market. Transparency is a broad concept, yet in essence it denotes providing a certain degree of access to algorithms, although different levels of access, based on various regulatory models, may be envisioned.61 Transparency is also interconnected with explainability, that is, a process of tracking the steps taken to achieve an autonomous decision,62 and auditability, that is, the possibility to record functioning of algorithms, their models and data utilized by them in order to screen them in situations when they cause adverse effects.63 As the OECD points out, transparency serves four primary functions: fostering a general understanding of AI systems, raising stakeholders’ awareness about their interactions with such systems, enabling affected individuals to understand the outcome of automated decision-making, and empowering individuals affected by such decisions to challenge the outcome of the AI system.64 One of the most challenging aspects in this context is a tradeoff between increasing the algorithm’s explainability (which can reduce its accuracy) and enhancing its accuracy (which can reduce explainability).65 The EU guidelines indicate the relevance not only of algorithmic but also of business model transparency, concerning any decision which can affect the design and rationale of algorithms. Moreover, entities interacting with an AI system should be informed about its capacities and limitations.66 4.5

Diversity, Non-discrimination, and Fairness

OECD notes that ‘AI actors should respect the rule of law, human rights, and democratic values, throughout the AI system lifecycle’. In practice, this means freedom, dignity and autonomy, privacy and data protection, non-discrimination and equality, diversity, fairness, social justice and internationally recognized labour rights should be respected by the autonomous systems. Safeguards, which AI developers should introduce, include in particular human intervention.67 The goal of such a requirement is to ensure that AI works for the creation of a just and equitable society. Therefore, access to digital tools should not be restricted, and ideas of open data and open algorithms should be supported.68 Algorithms should promote diversity in the society, inter alia through diversified tools to prevent monopolies, promote free expression and not lock individuals into user profiles.69 UN documentation places several interconnected aspects into the category ‘justice’, such as fairness, democracy, diversity,

See: Hulicki M., Algorithm Transparency as a Sine Qua Non Prerequisite for a Sustainable Competition in a Digital Market? EU and Comparative Law Issues and Challenges Series (ECLIC), No. 5/2021, pp.238–61. 62 Besides transparency and auditability, explainability also assures validity of decisions, and maintainability – providing data scientists with insight into the algorithmic models to maintain and improve them. See: A Framework for Ethical AI at the United Nations, p.15. 63 Statement on Algorithmic Transparency and Accountability. 64 Recommendation of the Council on Artificial Intelligence. 65 High-Level Expert Group on Artificial Intelligence, p.18. 66 Ibid. 67 Recommendation of the Council on Artificial Intelligence. 68 Montreal Declaration for a Responsible Development of Artificial Intelligence, p.13. 69 Ibid., p.14. 61

The principles of algorithmic justice in the digital market  357 non-discrimination, genderless, unbiased AI, equity, respect for human rights, access and redress and data agency. The risks which can be associated with not fulfilling this requirement are bias, discrimination and deception.70 The EU guidelines indicate biases of commercial nature: unfair competition and intentional harm to the consumers (for example, untransparent markets, homogenization of prices through collusion).71 In business-to-consumer applications, systems should be designed to be user-centric and should allow all people to engage with the AI systems. Therefore, such systems should be adjustable to the specificities of a wide range of users (universal design). The guidelines also indicate the need for inclusion of the oversight process from the initial phases of the system design. Moreover, this process should during the entire lifecycle of the product involve stakeholder participation (persons directly or indirectly affected by the product).72 4.6

Societal and Environmental Well-being

Societal and environmental well-being is an aspect on which most of the analysed documents place special emphasis. AI solutions through their lifecycle are expected to support objectives which are beneficial to society and the environment, such as goals of sustainable development. Hence, it is anticipated that developers of AI systems will take into consideration issues such as energy consumption, responsible use of resources or climate protection. EU guidelines expect these systems to be eco-friendly.73 Moreover, it is expected that advanced algorithms will benefit society, enhance social interaction and foster democratic processes.74 It is feared that the lack of acknowledging such elements in the AI design and operation will result in social erosion and exclusion.75 This context is very important for the Montreal Declaration, which highlights that AI should work for the well-being of people in line with principles of environmental sustainability, enhance cooperation within society, foster social bonds and have a positive impact on society. What is more, AI systems should be subjected to public scrutiny and debate and engage various stakeholders, and the role of public authorities in this process should be substantial.76 4.7 Accountability OECD notes that AI actors are to be accountable for the correct functioning of AI systems and for respecting the principles of ethical AI.77 Accountability is also closely related to transparency, lawfulness, use of ‘ethical black boxes’78 and auditability.79 While AI systems imply at least a certain degree of autonomy in making a decision, the Montreal Declaration

A Framework for Ethical AI at the United Nations, p.18. High-Level Expert Group on Artificial Intelligence, p.18. 72 Ibid. 73 Ibid. 74 Ibid. 75 A Framework for Ethical AI at the United Nations, p.18. 76 Montreal Declaration for a Responsible Development of Artificial Intelligence, p.8, 11–12, 17. 77 Recommendation of the Council on Artificial Intelligence. 78 ‘Ethical black box’ denotes recording the decision-making process and its main elements, so it is later possible to scrutinize it and discover the steps taken by an automated system to reach the decision. 79 A Framework for Ethical AI at the United Nations, p.17. 70 71

358  Research handbook on digital trade constitutes that ‘only human beings can be held responsible for decisions stemming from recommendations made by AIS, and the actions that proceed therefrom’.80 Responsibility should not be rejected when it is unclear how the decision was made, and affected entities should have a right to redress.81 In cases when intellectual property rights and secretive business models would make it difficult to disclose the algorithm, the trustworthiness of an AI system could be achieved through internal and external audits, which can verify compliance with the abovementioned principles. The EU guidelines include several considerations which should be taken into account to achieve accountability of AI systems: reporting mechanisms, effective responses to outcomes of automated decisions, protection of whistleblowers and other entities engaged in the accountability process, impact assessment processes, and acknowledging and evaluating trade-offs concerning compliance with ethical principles.82

5.

IDENTIFICATION OF THE MAIN PROBLEMS POSED BY ALGORITHMS IN THE DIGITAL MARKET

In the digital market, algorithms are commonly utilized to conduct various operations, from simple interactions between the parties via chatbots, through contract formation/enforcement, to autonomous decision-making.83 In a sense, nowadays algorithms regulate the markets, and their decisions evoke legal effects. In cyberspace it is not a question of whether the programming code regulates online actions, but one of how it regulates. In other words, algorithms comprise an architecture of the digital world, but it is the decisions of human beings (programmers, government, regulators) that put in place how this architecture will be structured.84 Hence, algorithms can constitute a step towards more rationalized decision-making and can limit prejudice which is somehow inherent to human choices. As the above analysis proved, the requirements for ethical AI have common features and comprise a new standard of how intelligent systems should be deployed on the market. Furthermore, it is symptomatic that analogous aspects are already being applied in the private sector by the main actors in AI research and use. Facebook’s five pillars of Responsible AI include privacy and security, fairness and inclusion, robustness and safety, transparency and control, accountability and governance.85 Google has set out seven objectives for smart applications: to be socially beneficial; avoid creating or reinforcing unfair bias; be built and tested for safety; be accountable to people; incorporate privacy design principles; uphold high standards of scientific excellence; and be made available for uses that accord with these prin-

Montreal Declaration for a Responsible Development of Artificial Intelligence, p.16. Statement on Algorithmic Transparency and Accountability, p 2. 82 High-Level Expert Group on Artificial Intelligence, p.20. 83 Some of examples may include remote content verification by social media, data profiling by advertising companies, digital rights management, granting a loan on a basis of automated credit score evaluation, algorithmic grading systems, or even automatic termination of employment communicated through algorithms. 84 Lessig, Code Is Law. 85 Facebook’s Five Pillars of Responsible AI, Facebook.com, https://​ai​.facebook​.com/​blog/​facebooks​ -five​-pillars​-of​-responsible​-ai/​, accessed 15 April 2022. 80 81

The principles of algorithmic justice in the digital market  359 ciples.86 Microsoft’s principles include fairness, reliability and safety, privacy and security, inclusiveness, transparency and accountability.87 Other leading companies in the IT sector have taken a similar approach.88 Therefore, the standard of algorithmic justice, understood as the ethical utilization of algorithms should be based on universally accepted criteria, such as impartiality, fairness, accurateness, reasonability, harmlessness, accountability, and so on. In the digital market, entities which are most impacted by algorithmic decision-making will be competitors, consumers, undertakings using digital platforms to conduct business and, in the broader sense, the society and the economy as a whole. Although algorithms play a fundamental role in the functioning of digital goods and services and one cannot imagine a modern economy without them, it should be kept in mind that they entail a wide range of problems for the participants of the digital market. Some of the main problems include:89 (a) generalization (b) bias (c) faults and defects (d) opacity (e) abuses and manipulation Digital services are dependent on data. Algorithms collect and process big datasets to find similar patterns. On such a basis it is possible to predict market trends, determine customer preferences and dynamically adjust services to the expectations of consumers. Whereas it may seem that it leads to the personalization of services (to lesser extent products), in practice through automated profiling individuals are assigned to specific profiles, as such algorithms conduct generalization of big data, based on probability calculations and finding objects with similar characteristics. However, such a deindividualized approach makes the market participants prisoners of advanced statistics. The already invoked notion of ‘algorithmic pollution’ refers to such a situation, where individuals are ‘datafied’ in terms of being represented by a limited set of attributes which are relevant for a certain process. The process of individual datafication is amplified by acquiring data from third parties, who have already transformed individual characteristics into data.90 When the historical data is used a person may not be actually or accurately described (‘datafied’).91 Use of single data factors, such as age, may be inaccurate, irrelevant or even harmful, in particular when they constitute a significant element of the methodology.92 Research has illustrated that the accuracy of consumer profiling is very

Artificial Intelligence at Google: Our Principles, Google, https://​ai​.google/​principles/​, accessed 15 April 2022. 87 Our approach, Microsoft.com, www​.microsoft​.com/​en​-us/​ai/​our​-approach​?activetab​=​pivot1​ %3aprimaryr5, accessed 15 April 2022. 88 See e.g. Cisco Principles for Responsible Artificial Intelligence, Cisco.com, www​.cisco​.com/​ c/​dam/​en​_us/​about/​doing​_business/​trust​-center/​docs/​cisco​-responsible​-artificial​-intelligence​-principles​ .pdf​?CCID​=​cc000742​&​DTID​=​odicdc000016, accessed 15 April 2022. 89 This list is of exemplary character and should not be considered as a comprehensive list of issues which may relate to algorithmic activities in the digital market. 90 Marjanovic, Cecez-Kecmanovic, Vidgen, Algorithmic Pollution. 91 Ibid. 92 Stevenson M., Slobogin C., Algorithmic Risk Assessments and the Double-Edged Sword of Youth, Washington University Law Review, Vol. 96, 2018, pp.681–706; Stevenson M., Doleac J.L., 86

360  Research handbook on digital trade differentiated and often unreliable,93 and users of digital platforms are very often surprised about how they were profiled.94 Additionally, profiling impacts human autonomy in making reasonable socio-economic choices, ultimately triggering a less efficient market.95 Another aspect of algorithms’ performance on the market that is often a subject of criticism concerns frequent bias implied in many algorithms, either directly coded in the structure of the algorithm as such or be indirectly impacting it through the input of incorrect, incomplete or outdated data.96 Bias might concern such data components as demographics, social profiles or other aspects, and may impact the output received by the end-user of an algorithm-based product/service. Sometimes, it will not be relevant if the results of the algorithm are accurate, but the use of certain data in the algorithms may amount to discrimination and be deemed unconstitutional or against basic rules of law.97 Biased algorithms may create a disadvantageous cycle for affected individuals, especially when they will refer to the historical performance of such a person. The output of such systems can further impact one’s statistical situation and again worsen his perspectives.98 Resolving such a situation may prove to be exceptionally onerous since it would require finding the core of the problem and cutting all dependencies between data. An algorithm is precise, as it has been precisely designed, and accurate where the relevant data have been provided to the system. In consequence, many algorithms will be faulty and defective, and it is almost impossible to completely avoid the possibilities of faults and defects. The main responsibility for the safe, accurate and correct operation of the software rests with the programmer and data-inputter. Many of the faults will rely on incorrect or insufficient data. However, it is also important to appropriately interpret the data, to make accurate conclusions and predictions. Some research results show that AI decisions are 90 per cent effective and do not deviate from human-made decisions.99 However, some authors suggest that to protect affected individuals, opt-in/opt-out mechanisms should be introduced for automated decision-making.100 Individuals using services based on algorithms will rarely know their technicalities. The situation is additionally complicated by the fact that algorithms are usually protected by trade secrets, and as such, only the right holder can grant access to them. Moreover, enabling the

Algorithmic Risk Assessment in the Hands of Humans, 21 April 2021, Social Sciences Research Network, https://​ssrn​.com/​abstract​=​3489440, accessed 15 April 2022. 93 Neumann N., Tucker C. and Whitfield T., Is Third-Party Consumer Profiling Effective? Evidence from Field Studies, Marketing Science, Vol. 38/6, 2019, pp.918–26. 94 Büchi M., Fosch-Villaronga E., Lutz C., Tamò-Larrieux A., Velidi S., Making Sense of Algorithmic Profiling: User Perceptions on Facebook, Information, Communication & Society, 2021, pp.1–17. 95 Büchi M., Fosch-Villaronga E., Lutz C., Tamò-Larrieux A., Velidi S., Viljoen S., The Chilling Effects of Algorithmic Profiling: Mapping the Issues, Computer Law & Security Review, Vol. 36, 2020, p.7. 96 See: Marjanovic, Cecez-Kecmanovic, Vidgen, Algorithmic Pollution. 97 Starr, Evidence-Based Sentencing; Sidhu, Moneyball Sentencing. 98 Marjanovic, Cecez-Kecmanovic, Vidgen, Algorithmic Pollution. 99 Liu H., Lin C., Chen Y., Beyond State v. Loomis: Artificial Intelligence, Government Algorithmization, and Accountability, International Journal of Law and Information Technology, Vol. 27/2, 2019, p.141; Karamouzis S., Harper D.W., An Artificial Intelligence System Suggests Arbitrariness of Death Penalty, International Journal of Law and Information Technology, Vol. 16/1, 2008, p.5. 100 Ibid.

The principles of algorithmic justice in the digital market  361 parties with mere access to algorithms without unfolding the entire machine-learning process might not provide any meaningful information. Algorithms do not operate in a vacuum, but they always work in a specific context, for a specific purpose, and are trained with specific data. Even for a specialist, understanding how the algorithm works may be very burdensome. Some experts in the field call for algorithm ‘black box tinkering’, which would demonstrate the decision-making process of the algorithm by providing it with a different set of values and examining numerous scenarios.101 Solutions to the problem of opacity even in cases of full algorithm transparency have downsides as well, and both static and dynamic methods of software analysis are unsatisfactory.102 Interconnection of different sets of algorithms can make it more difficult to find out how they interact and operate, and it may be burdensome even for their programmers.103 Algorithmic intransparency, therefore, causes a situation of significant information asymmetry in the market where entities deploying AI systems to the market will have a substantial advantage over individuals using them. Algorithms’ utilization in the market can also be abusive for the market participants. For example, companies using algorithms for the operation of their services in the digital market may promote and favour their own products and services, or dominant entities may impose unfavourable terms on digital contracts. In fact, most anti-competitive actions can occur in cyberspace, and usually through exploitation algorithms. Consumers may be affected as well, as digital companies (especially those having a dominant position in the market) may control the content that users are viewing and create filtering bubbles. Consumers may be affected also through manipulation conducted by algorithms, for example through price differentiation mechanisms or dynamic pricing. Moreover, algorithms can be manipulated by other market participants, which is particularly a problem with those algorithms which are disclosed transparently. One of the side effects of ensuring algorithmic transparency may be the phenomenon of manipulation of algorithms.

6.

TOWARDS THE REGULATORY FRAMEWORK FOR ‘ALGORITHM JUSTICE’

The system of algorithm justice described in this chapter has a few common features. The AI tools should be safe and aim at positive effects for society. Moreover, they should be based on accurate and fair data and should function transparently. Such tools should also be subject to internal and external control procedures from the time of their design to final market implementation. One could ask if those requirements fit well into the market realities? To a great extent, the answer will depend on how one sets objectives of undertakings deploying such systems in the market. Are they more focused on promoting societal well-being or do they aim at taking competitive advantage and maximizing profits? Hence, sometimes referring to the guidelines of ethical AI in the market situation may seem impractical, as from some perspectives it may appear maladjusted to market realities. Also, applying these criteria to

Filmar M., Elkin-Koren N., Black Box Tinkering: Beyond Transparency in Algorithmic Enforcement. SSRN Electronic Journal, 2016, 10.2139/ssrn.2741513. See also: Liu, Lin, Chen, Beyond State v. Loomis. 102 Desai, Kroll, Trust But Verify, pp.37–9. 103 Marjanovic, Cecez-Kecmanovic, Vidgen, Algorithmic Pollution. 101

362  Research handbook on digital trade Table 21.3 Requirement of

Desired actions and risk in case of incompliance with the requirements of ethical AI Desired actions in the digital market

ethical AI

Main risks in case of non-compliance (a market perspective)

Human agency and

Algorithms which protect fundamental rights

1. Limitation of individual fundamental rights (e.g. loss

oversight

(e.g. privacy, freedom of speech, freedom to

of privacy)

conduct business); impact assessments including 2. Creation of monopolies – market control, less external feedback; right to opt-out from an

competition, less choice for the consumers

automated decision; providing information on the AI system and supporting the user; implementing oversight of the AI system. Technical robustness

Systems designed not to inflict harm; taking

1. Digital products, which are unsafe for the consumers

and safety

preventive measures against cyberthreats;

and other market participants

routinely testing the AI, and making results

2. Untransparent decision-making mechanisms

of such tests public; traceability of the decision-making processes; systematic risk assessments; access to potential harmful applications should be restricted. Privacy and data

Protection of privacy by design/default;

1. Loss of privacy

governance

implementing state-of-the-art data governance

2. Manipulation of data, e.g. for anticompetitive

mechanisms; choosing accurate data; monitoring purposes the data used to prevent manipulations; access

3. Biased outcomes of autonomous decision-making

to personal data should be restricted; solicited systems should meet similar thresholds. Transparency

Access to algorithms and explainabilty of their

1. Untransparent decision-making mechanisms –

decisions, should be provided. Decision-making inability to ensure compliance with norms, standards processes should be recorded for future audits. Diversity,

Entities impacted by automated decision should

non-discrimination and have a right to human intervention; digital fairness

and ethics. 1. Biased outcomes of autonomous decision-making 2. Creation of monopolies – market control, less

tools should not be limited for certain groups;

competition, less choice for the consumers

implementing open data policy; prevention

3. Consumer harm

of monopolies and promotion of diversity in

4. Unfair competition

the market; actions taken to tackle the bias; adjustable systems to different individuals; stakeholder participation in the oversight processes. Societal and

Supporting democratic processes, social bonds,

1. Social erosion

environmental

and solidarity within the society; eco-friendly,

2. Unsustainable products/services

well-being

sustainable solutions; public scrutiny.

Accountability

Documented decision-making process; regular

1. Untransparent decision-making mechanisms –

audits checking for compliance with laws,

inability to ensure compliance with norms, standards,

policies, and ethics, availability of redress;

and ethics

engagement of AI actors in the accountability

2. Lack of incentives to enhance AI systems and employ

process.

preventive measures for making them safe and fair 3. Injured persons are unable or have burdensome task to pursuing claims

businesses may seem extensively complicated, and may discourage innovation. But, the role

The principles of algorithmic justice in the digital market  363 of algorithms in the society/economy is becoming so essential that regulating their functioning should not be underestimated. Furthermore, if it is agreed that algorithm developers and companies using them in their products and services should follow these guidelines, one could ask the question: how to enforce them? Would it be best if they become a market standard, or should they just remain non-binding guidelines, or maybe they should be given legally binding effect? Some past examples indicate that a transformation from non-binding guidelines to legal obligations for corporations is feasible (for example, it can be seen in the evolution of the principles of corporate social responsibility, which resemble guidelines on ethical AI). Moreover, some legislative advances in this respect have already been made in the EU. In particular, the proposal for an Artificial Intelligence Act (hereinafter AIA)104 should be mentioned. Some aspects are apparent in some other legal acts (in force or prospective) in the EU law. AIA is the EU’s legal approach to the topic of AI, echoing many of the issues described in this chapter. The extensive scope of the document may even seem to be relatively bureaucratic. The main sense of AIA is the division of AI systems into four categories (unacceptable risk, high risk, limited risk and minimal risk) and adjusting the level of measures to each of them. The definition of the AI systems to which the regulation would apply covers a very broad range of technologies, and includes machine learning approaches, logic, knowledge-based approaches, statistical approaches, Bayesian estimation and search and optimization methods.105 Many of the pillars of algorithmic justice described before are found in the proposal. Protection of fundamental rights is a point of reference for the classification of risk posed by the AI system, in particular in terms of designation and oversight of the high-risk systems. Such systems (and their modifications) are also subject to ex ante conformity assessment before they will be deployed on the market, a process which includes the participation of external bodies.106 One of the assessment criteria used in verifying if the system poses risks of harm to the health and safety of, or has adverse impacts on, fundamental rights is a situation in which it is not reasonably possible to opt out from that outcome produced by an AI system.107 Additionally, there are informative requirements for the high-risk AI systems, since they need to be accompanied by instructions that include concise, complete, correct, and clear information that is relevant, accessible and comprehensible to users. These instructions should inform the user about the high-risk system’s intended purpose and its proper use, including the specific geographical, behavioural or functional settings in which such system is intended to be used.108 Also, according to Article 14(1) AIA, a high-risk AI systems needs to be designed and developed in such a way that they can be effectively overseen by natural persons during the period in which the AI system is in use. This includes appropriate human–machine interface tools.109 The aim of this measure is the prevention or minimization of the risks to health, safety or fundamental rights linked with the utilization of high-risk systems, not only within their

104 Proposal for a Regulation of the European Parliament and of the Council Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act) and Amending Certain Union Legislative Acts, COM/2021/206 final. 105 See: Annex I to AIA. 106 See: Article 16(e), and Article 43 AIA. 107 See: Article 7(1), and (2) AIA. 108 See: Article 3(15), and Article 13(2) AIA. 109 See: Article 14(1) AIA.

364  Research handbook on digital trade intended purposes but also in situation misuses, which could be reasonably foreseen.110 If it is technically possible, measures ensuring human oversight should be identified and developed by the provider (or implemented by the user) of the system before placing it on the market or putting it into service.111 Additionally, high-risk systems are evaluated in terms of harm to health and safety, and the most harmful have been classified as prohibited artificial intelligence practices.112 With reference to high-risk systems also a risk management mechanism needs to be established, implemented, documented and maintained,113 and they should be tested for the purposes of identifying the most appropriate risk management measures.114 AIA implements a requirement for the high-risk systems to be able to conduct the automatic recording of events while operating, which should ensure traceability of the AI system’s functioning throughout its lifecycle within its intended purpose.115 AIA relatively sparsely focuses on personal data, as already-in-force rules of Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data and repealing Directive 95/46/EC (General Data Protection Regulation)116 (hereinafter GDPR) provide for a complex set of regulations on the processing of personal data. Nevertheless, some prohibited practices by the AIA relate to the processing of biometric data.117 Moreover, it should also be kept in mind that privacy is a fundamental right, which is protected by AIA.118 And the high-risk systems should contain appropriate data governance and management practices.119 AIA imposes an obligation on developers of high-risk AI systems to design and develop the in a manner that they achieve an appropriate level of accuracy, robustness and cybersecurity, and their performance should be consistent in those respects throughout their lifecycle. Additional measures are provided to ensure resilience to unauthorized alteration of their use or performance through exploitation of the system vulnerabilities. In particular, this regards a process called ‘data poisoning’, that is, manipulating datasets used in training the algorithms.120 Special provisions were introduced to AIA for purposes of ensuring transparency of AI systems. Providers of high-risk systems can be requested by a national competent authority to provide all the information and documentation necessary to demonstrate the conformity of the high-risk AI system.121 Moreover, Article 13 requires designing and developing it to an extent

See: Article 14(2) AIA. See: Article 14(3) AIA. 112 See: Articles 5–6 AIA. 113 See: Article 9(1) AIA. 114 See: Article 9(5) AIA. Moreover, the market surveillance authorities are to be granted full access to the testing datasets used by the provider. See: Article 64 AIA. 115 See: Article 12(1) and (2) AIA. 116 Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation), OJ L 119, 4.5.2016, pp.1–88. 117 See: Article 5 AIA. 118 See: Articles 7 and 8 Charter of Fundamental Rights of the European Union, OJ C 326, 26.10.2012, pp.391–407. 119 See: Article 10(2) AIA. 120 See: Article 15(1) and (4) AIA. 121 Article 23 AIA. 110 111

The principles of algorithmic justice in the digital market  365 to ensure that its operation is sufficiently transparent to enable users to interpret the system’s output and use it appropriately.122 Also for the purposes of transparency, some AI systems need to consider the specific risks of manipulation associated with them and provide individuals with the information that they are dealing with an AI system. This refers to the emotion recognition system, biometric categorization system and deep-fake producing systems.123 AIA provides measures against feedback loops. Namely, AI systems that pose high risks and which are learning continuously after being placed on the market or put into service have to be developed in a manner to ensure that possibly biased outputs due to outputs used as an input for future operations are duly addressed with appropriate mitigation measures.124 While in the context of diversity, non-discrimination, fairness, and societal and environmental well-being the proposal is fairly underdeveloped, the realization of many of such elements can be achieved through general rules of AIA, such as prohibited practices and classification of high-risk systems. Still, AIA encourages providers to voluntarily apply any additional requirements, in particular, to guarantee environmental sustainability, accessibility to persons with disability, stakeholders’ participation in the design and development of AI systems, and diversity of the development teams.125 Some of the principles of ethical algorithms have already been enacted into EU law, whereas other measures have been proposed. Some examples may include principles of privacy by default/design, the right to opt out from automated decision-making, and the right to human intervention, which can be found in the GDPR.126 Recently adopted consumer protection rules offer enhanced protection to consumers with respect to information, which needs to be provided to them on the main parameters determining the ranking of products presented to them as a result of the search query and the relative importance of those parameters, when providing them with the possibility to search for products offered by different traders or by consumers on the basis of a query in the form of a keyword, phrase or other input, irrespective of where transactions are ultimately concluded.127 Then, the Proposal for a Directive of the European Parliament and of the Council on improving working conditions in platform work, in Article 19 provides protection from dismissal for platform workers, who may request the digital labour platform to provide duly substantiated grounds (in writing) for the dismissal. Such measure protects them against automated dismissal, which is sometimes practiced by digital platforms.128 Moreover, Article 12 provides access to relevant information on platform work to labour, social protection and other relevant authorities, who exercise their functions in ensuring compliance with legal obligations applicable to the employment status of persons performing platform work, and who could also ask digital labour platforms for additional

Article 13(1) AIA. Article 52 AIA. 124 Article 15(3) AIA. 125 Recital 81 AIA. 126 Article 22(1), and (3), and Article 25 GDPR. 127 Article 3.4(b) Directive (EU) 2019/2161 of the European Parliament and of the Council of 27 November 2019 amending Council Directive 93/13/EEC and Directives 98/6/EC, 2005/29/EC and 2011/83/EU of the European Parliament and of the Council as regards the better enforcement and modernisation of Union consumer protection rules, OJ L 328, 18.12.2019, pp.7–28. 128 See Article 19 Proposal for a Directive of the European Parliament and of the Council on improving working conditions in platform work. Brussels, 9.12.2021 COM(2021) 762 final. 122 123

366  Research handbook on digital trade clarifications and details regarding any of the data provided.129 Additionally, Article 7 provides for human monitoring of automated systems, whereas Article 8 ensures that platform workers have the right to obtain an explanation from the digital labour platform for any decision taken or supported by an automated decision-making system that significantly affects the platform worker’s working conditions.130 Furthermore, Regulation (EU) 2019/1150 of the European Parliament and of the Council of 20 June 2019 on promoting fairness and transparency for business users of online intermediation services increases the level of transparency to ensure that business users of online intermediation services and corporate website users in relation to online search engines are granted appropriate transparency, fairness and effective redress possibilities,131 generally through informative measures. Finally, with reference to algorithm transparency the EU’s Digital Services Package, which comprise the Digital Services Act (DSA), and the Digital Markets Act (DMA) should be mentioned.132 While DSA aims at standardizing rules on liability, due diligence and regulation/monitoring of the functioning of providers of intermediary services in the internal market, in the context of algorithmic transparency, DSA’s main focus is on prioritization and targeting of information, content moderation and recommendation systems.133 On the basis of DSA, the European Commission (hereinafter EC) and independent auditors are granted special powers to access algorithms of the very large platforms.134 DMA aims to ensure contestable and fair markets in the EU digital sector. In particular, Article 3(1) introduces the notion of the gatekeeper, who can be defined

Ibid., Article 12. Ibid., Articles 7–8. 131 Regulation (EU) 2022/2065 of the European Parliament and of the Council of 19 October 2022 on a Single Market For Digital Services and amending Directive 2000/31/EC (Digital Services Act) (Text with EEA relevance), PE/30/2022/REV/1, OJ L 277, 27.10.2022. 132 Regulation (EU) 2022/1925 of the European Parliament and of the Council of 14 September 2022 on contestable and fair markets in the digital sector and amending Directives (EU) 2019/1937 and (EU) 2020/1828 (Digital Markets Act), OJ L, 265/1, 12.10.2022. 133 See e.g. DSA, recitals 45, 70 and 96. 134 See e.g. DSA art. 72(1) and art. 69(2)(d). Additionally, recital 141 provides that EC ‘should have access to any relevant documents, data and information necessary to open and conduct investigations and to monitor the compliance with the relevant obligations […] should be able to directly require by means of a duly substantiated request for information that the provider of the very large online platform or of the very large online search engine concerned as well as any other natural or legal persons acting for purposes related to their trade, business, craft or profession that may be reasonably aware of information relating to the suspected infringement or the infringement, as applicable, provide any relevant evidence, data and information. […] should be able to require access to, and explanations by means of exercise of investigatory powers, such as requests for information or interviews, relating to documents, data, information, data-bases and algorithms of relevant persons, and to interview, with their consent, any natural or legal persons who may be in possession of useful information and to record the statements made by any technical means. […], whereas recital 92 provides that given the need to ensure verification by independent experts, very large online platforms should be accountable, through independent auditing, for their compliance with the obligations […] Auditors should guarantee the confidentiality, security and integrity of the information, such as trade secrets, that they obtain when performing their tasks and have the necessary expertise in the area of risk management and technical competence to audit algorithms. Moreover, transparency is ensured by art. 14(1) DSA: Providers of intermediary services shall include information on any restrictions that they impose in relation to the use of their service in respect of information provided by the recipients of the service, in their terms and conditions. That information shall include information on […] algorithmic decision-making and human review. There are further transparency requirements for very large online platforms, e.g. in art. 27, art. 39, art. 40. 129 130

The principles of algorithmic justice in the digital market  367 as a provider of core platform services, who has a significant impact on the internal market, and operates a core platform service which serves as an important gateway for business users to reach end users and enjoys (or is foreseen to enjoy) an entrenched and durable position in its operations. New obligations were imposed on gatekeepers135 and DMA provisions safeguard the market against some practices of gatekeepers by prohibiting some actions. Also, the DMA granted the EC access to the algorithms.136

7. CONCLUSIONS Algorithmic decision-making may be introduced in three different layers: first, as a computer program assisting humans in decision-making processes; second, as a fully automated system working under the supervision of a human; third, as a completely autonomous system of decision-making that is independent of humans. Nevertheless, the introduction of any automated decision-making in the digital market would require social acceptance. To achieve it, some aspects of algorithm functioning need to be regulated and provide necessary safeguards for impacted entities. Therefore, rules such as human oversight, safety, privacy protection, transparency, accountability, fairness and general beneficence to society should be ensured. The analysis conducted in the chapter proved that these rules are already de facto standards in the digital realm. Moreover in the EU, they are becoming de iure standards as well, as legal developments focus on the implementation of guidelines on ethical AI in practice. And the field is just beginning to provide legal solutions to the problem of the ethical functioning of AI in the digital environment, even though the solutions presented in the chapter are far-reaching and may seem complex. Algorithmization of law, in parallel to developments of ICT technology, is inevitable, and one should not be concerned about it. Algorithmic justice, in fact, may prove to enhance justice, by eliminating judicial prejudice and subjectivity, and may bring enhanced predictability of legal outcomes and greater transparency to the decision-making process. However, new problems may emerge, such as manipulation of algorithms, and they should be addressed by the legislator. It seems widely accepted that AI systems should be developed with the aim to benefit humanity. Similarly, the debate on the role of AI in society demonstrates that the main concerns are based on the prospect of AI systems getting out of human control and inflicting negative effects on society. Now, when the impact of advanced algorithms on the economy is 135 Obligations/prohibitions for gatekeepers involve e.g. combination personal data; automatic signing in of end users to other services; free setting of market pricing; concluding contracts with end users outside the platform; access to platform services acquired outside the platform; interoperability of software; preferential treatment of services and products; access to the performance measuring tools for advertisement; portability and access to data, in particular, access to search data; conditions of access for business users to the platform’s software application store. 136 See: DMA art. 21(1) and art. 23(2)(d), and art. 23(4). Moreover, art. 21(3) DMA provides that where the EC requires undertakings to provide access to its data-bases and algorithms, it shall state the legal basis and the purpose of the request, and fix the time-limit within which it is to be provided. Recital 81 envisages EC should be ‘empowered to request information […] In particular […] should have access to any relevant documents, data, database, algorithm and information necessary to open and conduct investigations and to monitor the compliance with the obligations […], irrespective of who possesses such information, and regardless of their form or format, their storage medium, or the place where they are stored’. Failure to provide access to algorithms was sanctioned with fines. See: DMA art. 30(3)(e).

368  Research handbook on digital trade profound, the vital choice of how to shape the algorithmic reality still needs to be made. If no action is being taken a situation may emerge where the prominence of algorithms will be even greater and they will continue to impact the digital market and the society at large, often with detrimental effects, including algorithmic manipulation, abuses of market power, biased outcomes of automated decisions, ‘data traps’ and the commercial use of untransparent, defective and harmful systems. However, the key aspect of technology regulation is ensuring the effectiveness of its solutions, and it can be assumed that the principles of algorithmic justice will be difficult to implement in practice. Enacting unenforceable laws relating to the regulation of technology should be avoided, as they may constitute a burden for businesses, stifling market competition and innovation while having little positive effect on society. Hence, regulation of the AI area should be very cautious and focus on the most important aspects to provide for the adaptability of laws to changing environments, which is one of the most important aspects of regulating the digital realm. The EU, through the introduction of AIA and other related legal acts, wants to be a trendsetter in the regulation of emerging technologies, and its proposals in the context of AI regulation seem advanced. While it is very important to protect society against insecure technologies, simultaneously, developers of AI software await clarity about their responsibilities. Therefore, it can be expected that adopting AIA will lead to enhanced legal certainty. However, being a pioneer in the field entails a certain number of risks. The proposed system may seem formalistic and burdensome for the IT sector and potentially may lead to overregulation, especially taking into consideration such a new, dynamic and crucial area of the modern economy.

22. Artificial intelligence and law: emerging divergent national regulatory approaches in a changing landscape of fast-evolving AI technologies Lijun Zhao

1. INTRODUCTION Back in the mid-1950s, some pioneers of artificial intelligence (AI) set themselves an impossibly lofty but well-defined mission: to recreate human intelligence in machines. In 1955, in Dartmouth, USA, John McCarthy and his talented collaborators coined the term ‘artificial intelligence’ for a branch of computer science with this mission.1 In the early days of AI research, Alan Turing, usually considered the father of modern intelligent machines, proposed that the true indication of computer intelligence is when a question-asker cannot distinguish between answers from humans and from computers.2 Today, AI is no longer mere science fiction. It is a reality. From using a virtual personal assistant to schedule our workdays, to language translation, to our mobile phones recommending video clips, news, music or restaurants that we might like, to travelling in self-driving vehicles, AI has become a part of our lives. While AI science has made steady but slow progress, only recently has it accelerated rapidly, which enabled academic achievements to be transformed into real-world use cases. So many AI experts point out that AI is having a renaissance, and we have embarked on the era of AI.3 Sixty years after its origin, AI has transformed into an enabling technology with the power to reshape every aspect of the real world in which we live.4 Nowadays, AI is characterised by a number of applications, including computers that play games with/against humans, understanding human languages, virtual personal assistants, computer vision, and robotics that involve computer vision, hearing and responding to sensory stimuli.5 Much of the recent unprecedented progress in AI has been largely driven by a combination of three factors – the dramatic technological advances in computing power and capacity, avail-

1 John McCarthy and others, ‘A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence’ (1955) p.11. 2 Alan Turing, ‘Computing Machinery and Intelligence’ (1950) 59 Mind 433. 3 Kai-Fu Lee and Yonggang Wang, Artificial Intelligence (1st edn, Cultural Development Press 2017). 4 See details in Artificial Intelligence (AI) worldwide – Statistics & Facts, www​.statista​.com/​topics/​ 3104/​artificial​-intelligence​-ai​-worldwide/​#topicHeader​_​_wrapper, Statista, 2021. All internet resources in this chapter last accessed on 1 January 2023. 5 Rahul Pareek, ‘Web Intelligence – An Emerging Vertical of Artificial Intelligence’, International Journal of Engineering and Computer Science [2012], 9430–6.

369

370  Research handbook on digital trade ability of massive amounts of data and progress in algorithms.6 With the help of all these three factors collectively, AI applications, such as deep learning, have marked the most significant leap forward in the past 60 years; a leap on this scale rarely comes more than once every few decades. This contribution argues that, as AI encompasses a broad range of technologies and is evolving rapidly, regulators and policy-makers at both national and global levels are facing the ‘pacing problem’ – technology is developing faster than the policy-makers’ ability to keep up.7 Moreover, while AI largely is unregulated across WTO members on a global scale, a handful of AI powers – the US, EU, UK and China – have started to regulate AI to secure the first-mover advantage. This contribution argues that several AI powers have developed divergent regulatory responses to handle the pacing problem at the national level. Subsequently, this contribution argues that the pacing problem also exists at the international level, by examining the deficiency of global trade law governing the areas of trade in goods, services and trade-related intellectual property rights.

2.

BENEFITS AND COSTS ASSOCIATED WITH DEVELOPMENT AND DEPLOYMENT OF AI

2.1 Opportunities AI is critical to global trade law and governance in the fast-changing landscape. Just as one coin has two sides, AI is changing the world, bringing opportunities and challenges to international organisations, governments and civil society.8 AI expert Kai-Fu Lee predicts the impact of AI will be ‘more than anything in the history of mankind’.9 Other sources, such as McKinsey, also foresee that AI will trigger advances that transform lives, business and the global economy.10 Many sources forecast the enormous benefits that AI will bring about. According to McKinsey, AI has enormous potential to contribute to global economic activity, and it could generate additional global economic activities worth approximately $13 trillion by 2030.11 Another source suggests that, by 2025, the automation of knowledge work, robotics and 6 International Telecommunication Union (ITU), Assessing the Economic Capacity of Artificial Intelligence (2018), pp.1–2. Lee and Wang (n 3). stating Deep learning + Large Data = the Renaissance of AI. 7 Braden R. Allenby in The Growing Gap Between Emerging Technologies and Legal-Ethical Oversight: The Pacing Problem (Gary E Marchant, Braden R Allenby and Joseph R Herkert eds, 1st edn, Springer 2010) (coining the term ‘pacing problem’). Lyria Bennett Moses and Monika Zalnieriute, Law and Technology in the Dimension of Time, in Sofia Ranchordas and Yaniv Roznai (eds), Time, Law and Change: An Interdisciplinary Study (Hart, 2020), 10.5040/9781509930968.ch-014, 303–26. 8 Deloitte, Areas of Preparedness and Concern among AI Adopters Worldwide as of 2020, Statista, July 2020, www​.statista​.com/​statistics/​1136694/​artificial​-intelligence​-concern​-preparedness​-2020/​. 9 Lee and Wang (n 3). 10 McKinsey Global Institute, Disruptive Technologies: Advances that Will Transform Life, Business, and the Global Economy, 2013. 11 Jacques Bughin and others, ‘Notes from the AI Frontier: Modeling the Impact of AI on the World Economy’ (McKinsey Global Institute Discussion Paper, 2018) https://​perma​.cc/​6ZFR​-2LRT accessed 31 December 2021.

Artificial intelligence and law  371 self-driving vehicles will generate 6.5–12 trillion euros per year (including improved productivity and higher quality of life in ageing populations).12 Still, the full potential of AI has yet to be fully realised, waiting to be unleashed in the coming decades and even centuries. Technologists offer a wide range of predictions about upcoming AI developments, from its use as a tool to aid relatively simple processes (through weak AI technologies) to robots with human-like mental capabilities (through general AI technologies).13 As explained later in Section 3, AI can be divided into two kinds: weak and general. In a specific field, weak AI systems have already done better jobs than humans, as outlined at the outset. An example of weak AI beating human performance is the 2017 AlphaGo and Ke Le match. 2.2 Challenges On the other hand, due to the scale and complexity of AI technologies, AI has addressed, and will address, unprecedented challenges to global trade governance and policy-makers. First, AI encompasses a wide range of different technologies, as illustrated above. Also, AI has become a commonplace feature in many products and services, and the division line between products and services is blurring. Given AI’s ability to act autonomously and its wide range, AI systems cannot easily be classified as goods or services. This contritions finds that, at an international level, the ‘goods vs services’ dichotomy inherited under the global trade law for decades is no longer suitable. This issue will be explained in detail later in Section 4. Similar to the above issue, another key challenge is the pacing problem due to a mismatch between legal and technological developments. One example is the dichotomy on which global trade law draws between goods and services. Another example is that it is unclear to what extent AI-driven products are subject to product liability law.14 Much literature has discussed the interplay between AI and product liability law, and whether AI products should be granted personhood or not;15 so, this contribution links its analysis with the existing literature but omits the details to avoid repetition.16 Furthermore, one of the most critical unprecedented challenges is that, globally, AI will widen gaps among countries, companies and workers.17 For instance, while AI will generate economic interests and boost international trade as mentioned above, beneficiaries will primarily be a handful of developed countries, where AI will significantly enhance their annual

Accenture, Why AI Is the Future of Growth, 2016. Peter Diamandis, ‘The World in 2025: 8 Predictions for the Next 10 Years’. Singularity Hub, 11 May 2015. Weak and general AI will be explained later in Section 3. 14 See more in Carpenter Wellington PLLC, ‘The Legal Landscape of Artificial Intelligence (AI) Law’ [2022] Lexology www​.lexology​.com/​library/​detail​.aspx​?g​=​d95bfa51​-db03​-47e0​-8c63​ -d7c7311d84f1. 15 ibid. 16 Carpenter Wellington PLLC (n 14). So far, the EU has revised its regulation on product liability directive 1985 to cover AI applciations. Tiago Sérgio Cabral, ‘Liability and Artificial Intelligence in the EU: Assessing the Adequacy of the Current Product Liability Directive’ (2020) 27 Maastricht Journal of European and Comparative Law 615. In the USA, the application of product liability law in the AI context has been addressed by case law, e.g. Cruz v Raymond Talmadge d/b/a Calvary Coach (2017) https://​casetext​.com/​case/​cruz​-v​-talmadge. 17 Bughin and others (n 11). 12 13

372  Research handbook on digital trade economic growth rate by 2035.18 Even though the adoption of AI technologies has been gradually diffused around the globe through – online and offline – transnational business activities, the AI technologies are primarily dominated by a handful of leading tech giants, mainly in developed countries such as the US, the UK and some European countries (also in two emerging economies – China and South Korea).19 As illustrated in Section 3, this contribution finds that those powerful countries have begun to regulate AI technologies to secure their first-mover advantages and ‘export’ their national/regional rules as global standards for other countries in the world.20 At a macro level, from a country’s perspective, AI is reshaping the landscape of superpowers in the AI arena and has created emerging superpowers that are clusters of non-Western and non-traditional developed countries. As per a worldwide survey on AI performance as of 2018, while the US shows the best performance in regards to AI, followed by Germany and the UK, there are two emerging economies (namely, China and South Korea) rank among the top ten countries in terms of their overall AI performance benchmarks.21 AI specialist Lee points out that the AI revolution consists of four arenas: ‘Business AI’ remains the only arena in which the US maintains clear leadership globally; China is in a strong position to (co)lead in ‘Internet AI’ and ‘Perception AI’, and will likely catch up with the US in ‘Autonomous AI’.22 At a micro level, from a company’s perspective, those start-ups are now scrapping for a slice of an AI landscape increasingly dominated by a handful of major players. Besides IBM (the globally largest AI patent owner),23 there are the so-called Seven Giants of the AI Age, including Google, Facebook, Amazon, Microsoft, Baidu, Alibaba and Tencent.24 These corporate juggernauts are ‘almost evenly split between the United States and China’, and they are making bold plays to dominate the AI economy.25 Given the factors above, Sections 3 and 4 of this contribution find that policy-makers from a handful of countries have used different national regulatory responses to address AI-associated opportunities and challenges in the AI age. The remainder of this contribution, in Section 5, highlights the rise of a phenomenon and the emerging challenges facing global trade law and governance. 2.3

Emerging Divergent Regulatory Approaches in a Few AI Leading Countries

AI will (re)shape countries and market-players’ global competitiveness and productivity in the coming decades/centuries, empowering early adopters of significant societal, economic and strategic advantages. As the pace of AI innovation and development accelerates – under-

18 Artificial Intelligence (AI) worldwide – Statistics & Facts, www​.statista​.com/​topics/​3104/​artificial​ -intelligence​-ai​-worldwide/​#topicHeader​_​_wrapper, Statista, 2021. 19 Artificial intelligence (AI) performance benchmark by country as of 2018, www​.statista​.com/​ statistics/​942041/​ai​-performance​-benchmark​-by​-country/​, Statista, July 2018. 20 See more in section 4. 21 ibid. 22 Kai-Fu Lee, AI Superpowers: China, Silicon Valley, and the New World Order (1st edn, Houghton Mifflin Harcourt 2018) pp.110–11. 23 Artificial Intelligence (AI) worldwide – Statistics & Facts, www​.statista​.com/​topics/​3104/​artificial​ -intelligence​-ai​-worldwide/​#topicHeader​_​_wrapper, Statista, 2021. 24 Lee (n 22) pp.110–11. 25 ibid p.89.

Artificial intelligence and law  373 pinned by advances in big data and high-performance computing – both the US and China are dominating, and both countries have adopted hard and soft laws to regulate AI.26 Apart from the AI superpowers, other AI leading countries, notably the EU, have joined the race and added further regulatory complexity to the global trade governance. In recent years, European policy-makers have recognised the importance of not falling behind on AI and have sought to raise their ambitions by launching EU’s AI regulations. The remainder of this contribution discusses the regulatory responses of the US, China, EU and UK, respectively, to the fast-evolving AI technologies.

3.

DEFINITION OF ARTIFICIAL INTELLIGENCE: PARTICULARITY OF AI

Unlike in the previous six decades since the origin of AI, the most remarkable feature of the current AI renaissance is that the applications of AI – in speech recognition, machine vision, data mining and other fields – have entered the real application scenarios of the industry and are closely linked to business models. Together, they are starting to show real value in the industry. For instance, in addition to playing Go, the availability of deep learning-based programs and massive amounts of data has enabled AI to do better than human counterparts in many areas, such as identifying faces, recognising speech and issuing loans; these are just what weak AI can do already, let alone general AI.27 Due to the significant scale and complexity of AI technology as outlined above, the overarching question for policy-makers and researchers is what AI is, that is, the definition of AI. The answer to this question is vital for the law-making process at both national and international levels, as it usually affects whether a particular AI technology falls within the scope of application of a specific law, regardless of soft or hard law. AI technologies today can be divided into five broad categories in terms of the areas of applications, according to McKinsey and many other sources: computer vision, data mining, machine learning, natural language and robotic process automation.28 Thus, there is a wide range of AI technologies, and more importantly, these technologies are fast-evolving, with great potential to be unleashed, as pointed out above. Thus, this attribute and particularity of the current AI renaissance address a question for all policy-makers at national and international levels: any static definitions will lead to the pacing problem,29 immediately or in the short run. Namely, it is likely that AI-related legislation or regulations will quickly fall behind technological development. In order to resolve the pacing problem, this contribution finds that in the US, EU and China, AI is widely defined, as discussed below.

See section 4. Lee and Wang (n 3). Lee (n 22). 28 Bughin and others (n 11). See also EqualOcean, Share of AI Technology Applications by Technology, Statista, 2020, www​.statista​.com/​study/​102790/​artificial​-intelligence​-in​-china/​. 29 Braden R. Allenby (n 7). 26 27

374  Research handbook on digital trade 3.1

Dictionary Definition

What is AI? That is a question. As WTO adjudicators have pointed out, in order to determine the ordinary meaning of a term, ‘a Panel [or tribunal] may start with the dictionary definitions of the terms to be interpreted’.30 So, it is meaningful to refer to the dictionary definition of AI. According to the Oxford Dictionary, ‘artificial’ means something ‘made or produced by human beings rather than occurring naturally’, and ‘intelligence’ means ‘the ability to acquire and apply knowledge and skills’.31 Put the two words together and ‘artificial intelligence’ refers to the ‘theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages’.32 As seen in these three dictionary definitions, it is clear that AI must be a man-made system which has the ability to acquire and apply knowledge and skills; however, the words ‘normally’ and ‘such as’ in the dictionary definitions fail to clarify to what extent or level the ability is. Moreover, technology usually develops over time, so the openness in the above dictionary definitions may mean a computer system that outperforms its human counterparts is as an AI for some time, but not as an AI later, after technology has advanced. For instance, in a 1997 match dubbed ‘The Brain’s Last Stand’, IBM’s Deep Blue defeated the world chess champion, Garry Kasparov.33 Deep Blue’s victory was considered a milestone in the history of AI, since it was the very first machine to beat a reigning world chess champion. Many people at that time thought Deep Blue was AI; however, by the time of the 2017 AlphaGo and Ke Jie match most people were already used to playing computer games programmed like Deep Blue, so few people would consider Deep Blue-like systems as AI. Therefore, due to the openness and elasticity in the dictionary definitions of AI, this contribution seeks to address the overarching question by defining the concept of AI to lay down a common ground for the discussion for academics and policy-makers across countries. 3.2

Definitions Featured by the AI Pioneers

Among those pioneers who laid down the cornerstone of today’s AI were the aforementioned British mathematician Alan Turing and American computer scientist John McCarthy. Turing poses the famous ‘Imitation Game’ (also known as the ‘Turing Test’), which is still being used today by computer scientists and many companies. In 1950 Turing published a ground-breaking work entitled ‘Computing Machinery and Intelligence’, in which he addressed the question ‘Can machines think?’ To answer this question, he further posed the Turing Test, which assesses whether a machine can generate human-like responses so that its behaviour cannot be distinguished from that of a human.34 The Turing Test fills in the gap See, e.g., Appellate Body Report, United States—Measures Affecting the Cross-Border Supply of Gambling and Betting Services [hereinafter AB Report, US—Gambling], WT/DS285/AB/R (20 April 2005), Paragraph 164. 31 Artificial, Oxford Living Dictionary, www​.lexico​.com/​definition/​artificial. Intelligence, Oxford Living Dictionary, www​.lexico​.com/​definition/​intelligence. 32 Artificial Intelligence, Oxford Living Dictionary, www​.lexico​.com/​definition/​artificial​ _intelligence. The italics is emphasis added by the current author. 33 It is a chess-playing expert system run on a particular purpose-built IBM supercomputer. 34 Turing (n 2) p.442. 30

Artificial intelligence and law  375 in the dictionary definition and elucidates the extent to which the ‘ability’ in a man-made machine can be ‘intelligence’; nevertheless, it should be noted that the Turing Test tactfully avoids the conundrum of what ‘intelligence’ is. Furthermore, even though some existing AI systems have outperformed their human counterparts, it is still arguably difficult to pass the Turing Test.35 John McCarthy and his collaborators – a group of leading researchers including John Nash, a Nobel Prize winner, with combined expertise from a broad spectrum of mathematics, computer science, neurology and so on – coined the term ‘artificial intelligence’, which first appeared in 1955 in their paper entitled ‘A Proposal for the Dartmouth Summer Research Project on Artificial Intelligence’.36 Unfortunately, McCarthy and his collaborators did not offer a solid definition in that paper. Nevertheless, these pioneers laid down the cornerstone of AI technologies, as mentioned at the outset, by setting themselves an impossibly lofty but well-defined mission to an impossibly lofty but well-defined mission: creating intelligent machines that work and react like humans.37 In 2007, McCarthy loosely defined AI as ‘the science and engineering of making intelligent machines, especially intelligent computer programs’.38 However, when it comes to the fundamental question of ‘what intelligence is?’, he explained that intelligence is the ‘computational part of the ability to achieve goals in the world. [There exist] varying kinds and degrees of intelligence […] in […] machines’.39 Even so, McCarthy admitted that there had not yet been a solid definition of ‘intelligence’ that does not depend on relating it to ‘human intelligence’, since ‘we cannot yet characterise in general what kinds of computational procedures we want to call intelligent’.40 Simply put, the point McCarthy made can be illustrated by the fact that, as AI technology develops over time, many people changed their answers to the same question mentioned above – that is, whether Deep Blue, AlphaGo or other man-made machines are AI or not. 3.3

Modern Approach to Defining AI

Even though there is no single definition of AI universally accepted by practitioners, in recent years, as McCarthy points out, a modern approach has emerged to define the term AI by referring to its multi-faceted characteristics. Russel, one of the leading AI researchers, has teamed up with Norvig, who led research work at Google (one of the seven AI tech giants), to publish a leading AI textbook.41 This book defines AI through the modern approach that US policy-makers have incorporated into US legislation,42 so here this contribution looks at the definition of AI in this leading textbook.

Aleksandar Todoroviæ, Has The Turing Test Been Passed? http://​isturingtestpassed​.github​.io/​. John McCarthy and others (n 1). 37 ibid. 38 John McCarthy, ‘What Is Artificial Intelligence?’ (2007) 73 American Scientist 258 http://​jmc​ .stanford​.edu/​articles/​whatisai​.html pp.1–2. 39 ibid pp.1–2. 40 ibid pp.1–2. 41 Stuart Russell and Russel Norvig, Artificial Intelligence: A Modern Approach (4th edn, Pearsons 2020). pp.19-23. 42 See the next section. 35 36

376  Research handbook on digital trade Russel and Norvig summarise the existing efforts in defining AI and then define the term through the use of four taxonomies to accommodate the multi-faceted characteristics of AI as follows: Acting humanly (that is, the Turing Test approach). Thinking humanly (that is, the cognitive modelling approach). Thinking rationally (that is, the ‘laws of thoughts’ approach). Acting rationally (that is, the ‘rational agent’ approach).43

● ● ● ●

Before proceeding further in this contribution, two interim findings need to be shared with readers as the stepping stones for further discussion. First, AI is a rapidly evolving family of technologies and a ‘truly universal’ field.44 This can be seen from the existence of a wide range of AI applications in real-world scenarios, as discussed above. Due to the scale and complexity of AI, a solid, static definition of AI in soft and/or hard law will quickly become obsolete; if such a definition were included in a piece of legislation, the legislation would become obsolete due to the ‘pacing problem’ in law-making and policy-making in the field of AI.45 Second, this modern approach to defining AI has generated an impact beyond academia and has influenced the US – an AI superpower – in drafting its new bills, as discussed below.

4.

STATUTORY DEFINITIONS OF AI: EMERGING DIVERGENT DOMESTIC REGULATORY RESPONSES

Since 2017, at least 60 countries have adopted some forms of AI policy, with a torrent of activities seeking to promote and match the pace of AI technology development with the pace at which policy-makers update rules.46 However, the expansion of AI governance has raised concerns about looming challenges to international cooperation. The growing ubiquity of AI in the online services and physical devices used in our daily lives means that any new regulations will significantly impact global trade governance. Thus, this section identifies and reviews the statutory definitions of AI, with a tendency for divergent national regulatory approaches, in a handful of AI leading countries – the US and China (the two AI superpowers), the EU and the UK. 4.1

From a US Perspective: A Proactive and Future-Proofing Approach

The US is not only an AI superpower but also a pioneer in regulating AI globally. Through following Russel and Norvig’s modern approach in its statutory definitions, the US has adopted and incorporated the modern approach and its well-defined taxonomies with a forward-facing nature to accommodate the rapidly evolving nature and a broad spectrum of AI applications. By adopting the modern approach to defining AI, the US law defines AI broadly: first in Russell and Norvig (n 41). ibid. 45 Braden R. Allenby (auth.) (n 7). See more details in sections 4 and 5. 46 OECD.AI, ‘National AI Policies & Strategies’ https://​oecd​.ai/​en/​dashboards accessed 1 May 2022. (This online database provides a live repository of more than 700 AI policy initiatives from 60 countries, territories and the EU.) 43 44

Artificial intelligence and law  377 the 2016 White House report on AI,47 and later in the bill for ‘the FUTURE of Artificial Intelligence Act of 2017’.48 The FUTURE bill governs not only any ‘artificial systems that perform tasks under varying and unpredictable circumstances, without significant human oversight’ but also a set of ‘techniques, including machine learning to approximate some cognitive task’.49 Moreover, in order to widen the scope of application, the bill defines AI in a highly broad manner. Section 3(a) of the FUTURE bill stipulates: In General. — Except as provided in subsection (b), in this Act: (1) ARTIFICIAL INTELLIGENCE. —The term ‘artificial intelligence’ includes the following: (A) Any artificial systems that perform tasks under varying and unpredictable circumstances, without significant human oversight, or that can learn from their experience and improve their performance. Such systems may be developed in computer software, physical hardware, or other contexts not yet contemplated.50 They may solve tasks requiring human-like perception, cognition, planning, learning, Communication, or physical action. In general, the more human-like the system within the context of its tasks, the more it can be said to use artificial intelligence. (B) Systems that think like humans, such as cognitive architectures and neural networks. (C) Systems that act like humans, such as systems that can pass the Turing Test or other comparable test51 via natural language processing, knowledge representation, automated reasoning, and learning. (D) A set of techniques, including machine learning, that seek to approximate some cognitive task. (E) Systems that act rationally, such as intelligent software agents and embodied robots that achieve goals via perception, planning, reasoning, learning, communicating, decision making, and acting. (2) ARTIFICIAL GENERAL INTELLIGENCE. — The term ‘artificial general intelligence’ means a notional future artificial intelligence system that exhibits apparently intelligent behavior at least as advanced as a person across the range of52 cognitive, emotional, and social behaviors. (3) NARROW ARTIFICIAL INTELLIGENCE. — The term ‘narrow artificial intelligence’ means an artificial intelligence system that addresses specific53 application areas such as playing strategic games, language translation, self-driving vehicles, and image recognition.

As seen from the breakthrough definitions of AI cited above, the US approach in the FUTURE bill has shown great flexibility in dealing with the pacing problem, through incorporating not only the academic definitions (such as ‘think like humans’ and ‘act like humans’)54 but also

47 White House, Preparing for the Future of Artificial Intelligence 6–7 (2016), https://​obamawhitehouse​ .archives​.gov/​sites/​default/​files/​whitehouse​_files/​microsites/​ostp/​NSTC/​preparing​_for​_the​_future​_of​ _ai​.pdf. 48 FUTURE of Artificial Intelligence Act of 2017, H.R. 4625, 115th Congress, (2017) www​.congress​ .gov/​bill/​115th​-congress/​house​-bill/​4625/​text​#:​~:​text​=​Introduced​%20in​%20House​%20(12​%2F12​ %2F2017)​&​text​=​To​%20require​%20the​%20Secretary​%20of​,Intelligence​%2C​%20and​%20for​%20other​ %20purposes. 49 US bill for the FUTURE of Artificial Intelligence Act of 2017, H.R. 4625, 115th Congress, Section 3(a)(1)(A), (D)(2017). 50 Emphasis added by the current author. 51 Emphasis added by the current author. 52 Emphasis added by the current author. 53 Emphasis added by the current author. 54 Turing (n 2). Russell and Norvig (n 41).

378  Research handbook on digital trade definitions advanced by AI experts and practitioners in the industry (such as the Turing test, ‘general AI’ and ‘narrow AI’).55 Moreover, the statutory definitions in the FUTURE bill have become part of effective legislation. The John S. McCain National Defense Authorization Act of 2018, which became law on 13 August 2018, followed the approach of the FUTURE bill to define AI. In the 2018 Act, AI is widely defined again, as follows:56 (1) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets.57 (2) An artificial system developed in computer software, physical hardware, or another context that solves tasks requiring human-like perception, cognition, planning, learning, Communication, or physical action. (3) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (4) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (5) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decision-making, and acting.58

It is also observed that the US is developing and expanding its statutory definition of AI to some extent. More recently, the bill for the Algorithmic Accountability Act of 2019 has not defined the term AI at all. Instead, this 2019 bill resorts to a more generic term – ‘automated decision system’ under its Section 2(1) – which broadly embraces ‘a computational process, including one derived from machine learning, statistics, or other data processing or artificial intelligence techniques, [which] facilitates human decision making, that impacts consumers’.59 As seen above, therefore, the US legislative approach nicely avoids the pacing problem and makes the US national legislation forward-facing and future-proof to counteract AI-associated challenges due to the fast-evolving nature of AI technology. Aiming at ‘maintaining American Leadership in Artificial Intelligence’,60 in 2020 the White House issued a memorandum, ‘Guidance for Regulation of Artificial Intelligence Applications’, which guides federal agencies when they develop regulatory approaches to AI.61 While the guidance is not hard law, it sets out a series of policy considerations for US policy-makers that should guide, to the extent permitted by law, regulatory and non-regulatory approaches to governing AI applications:

Lee and Wang (n 3). The John S. McCain National Defense Authorization Act for Fiscal Year 2019, Pub. L. No. 115-232, 132 Stat. 1636, 1695 (13 August 2018) (codified at 10 U.S.C. § 2358, note). See full text of the Act www​.congress​.gov/​bill/​115th​-congress/​house​-bill/​5515/​text. 57 Emphasis added by the current author. 58 Section 238(g) of the John S. McCain National Defense Authorization Act. 59 H.R.2231 – Algorithmic Accountability Act of 2019, 116th Congress (2019–2020), www​ .congress​.gov/​bill/​116th​-congress/​house​-bill/​2231. 60 White House, Exec. Order No. 13,859 of Feb. 11, 2019, Maintaining American Leadership in Artificial Intelligence, 84 Fed. Reg.3967 (14 February 2019), www​.whitehouse​.gov/​presidential​-actions/​ executive​-order​-maintainingamerican​-leadership​-artificial​-intelligence/​. 61 The White House, Guidance for Regulation of Artificial Intelligence Applications (11 November 2020) www​.whitehouse​.gov/​wp​-content/​uploads/​2020/​11/​M​-21​-06​.pdf. 55 56

Artificial intelligence and law  379 (a) Consider new regulations only after carefully assessing whether existing or evolving regulatory frameworks are sufficient; (b) avoid hampering AI innovation and growth; (c) contribute to the public trust in AI by promoting reliable robots, and trustworthy AI applications that do not pose risks to fundamental human rights and freedoms; (d) embed public participation, disclosure, and transparency in rule-making processes; (e) apply risk assessment and risk management frameworks and be transparent about the evaluation of risks; (f) carefully consider the full societal benefits and costs associated with the development and deployment of AI; (g) take flexible approaches that are technology-neutral and do not harm innovation;62 (h) consider issues of fairness and non-discrimination in regard to outcomes and decisions produced by AI applications; (i) promote the development of AI systems that are safe, secure, and operate as intended; (j) consider any national security implications raised by AI applications; and (k) coordinate with other agencies to ensure the consistency and predictability of AI policies.63

Moreover, the White House Guidance introduces recommendations for non-regulatory approaches to AI, for instance sector-specific policy guidance and voluntary consensus standards.64 The 2020 Guidance also exemplifies a list of actions that federal agencies can take to reduce barriers to the deployment and use of AI (from facilitating access to federal data, to engaging with the private sector in the development of standards).65 Most recently, since Biden’s administration began in 2020, a range of further regulatory changes have signalled a more proactive stance by the US federal government towards AI regulation. The US bill for the National Artificial Intelligence Initiative Act of 2020 acknowledges the importance of artificial intelligence as ‘a tool that has the potential to change and possibly transform every sector of the United States economy and society’.66 Furthermore, in this Act, AI is loosely defined, as below. Section 3(3) of this Act stipulates: ARTIFICIAL INTELLIGENCE.—The term ‘artificial intelligence’ means a machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments. Artificial intelligence systems use machine and human-based inputs to — (A) perceive real and virtual environments; (B) abstract such perceptions into models through analysis in an automated manner; and (C) use model inference to formulate options for information or action.67

It should be noted that this Act embraces both existing AI technologies (such as weak AI) and those that have not yet arrived (or are at least far from being mature, such as general AI). In order to understand the difference between weak AI and general AI (also called strong AI), it is meaningful to refer to AI expert Kai-Fu Lee, who divides the ongoing AI revolution into

Emphasis added by the current author. White House (n 60). 64 ibid. 65 ibid. 66 US National Artificial Intelligence Initiative Act of 2020, H.R.6216, 16th Congress (2019–2020) www​.congress​.gov/​bill/​116th​-congress/​house​-bill/​6216/​text. 67 ibid. 62 63

380  Research handbook on digital trade four waves, briefly mentioned above.68 In detail, Lee claims that the complete AI revolution will take some time and will ultimately wash over us in a series of ‘four waves – Internet AI, Business AI, Perception AI, and Autonomous AI’.69 The first two waves are already all around us, reshaping our digital and financial worlds. Although acting in ways we can barely recognise, they are tightening internet companies’ grip on our attention; replacing paralegals, journalists and personal assistants with algorithms; trading stocks; and diagnosing illnesses. Furthermore, Lee points out that Perception AI is still ongoing, with more AI potential to be unleashed and developed in the future, which will digitise our physical world. Consider, for instance, the Metaverse. This wave of Perception AI is ongoing, with the potential to blur the lines between the digital and physical worlds (this has been referred to in several places among the aforementioned US legislation and soft law), revolutionising how human beings experience and interact with our world. Ultimately, Autonomous AI will arrive, but it will have the most profound impact on our lives. For example, fully autonomous vehicles and intelligent robots might take over factories, transforming everything from organic farming to highway driving and fast food. As discussed in the above US legislation, US policy-makers also take Autonomous AI into consideration. Thus, the US approach to regulating AI is proactive and future-proofing. Last but not least, the US government under the Biden administration has signalled a more proactive stance towards AI regulation, which brings the US closer to the EU.70 For instance, the new EU–US Trade and Technology Council (TTC), an online consultation platform between the US and the EU, was established in 2021, with AI issues included in the TTC.71 The EU’s regulatory approach to AI will be discussed later. In summary, the US is the AI superpower, with the aim to maintain American leadership in this field; as such, US policy-makers have employed a loose, open-ended and future-proofing approach in soft and hard laws regarding AI. In its regulatory and policy initiatives, the US has considered both AI technologies that already exist (weak AI, such as ‘internet AI’ and ‘business AI’) and future AI technologies (artificial general intelligence, such as ‘perception AI’ and ‘autonomous AI’). 4.2

From a European Perspective: A Flexible, Coordinated Approach

Considering the vital role of massive data in AI technology, the EU’s ground-breaking data regime – the General Data Protection Regulation (GDPR) of 2016 – has impacted AI applications.72 Nevertheless, it should be noted that GDPR is technology-neutral in its application rather than focusing on AI. While the GDPR applies to AI-related products and services in today’s data-driven, changing economy, there are currently no specific EU laws to regulate AI.

Lee (n 22) p.110. ibid p.110. 70 Alex Engler, ‘The EU and U.S. Are Starting to Align on AI Regulation’ (Techtank, 2022) www​.brookings​.edu/​blog/​techtank/​2022/​02/​01/​the​-eu​-and​-u​-s​-are​-starting​-to​-align​-on​-ai​-regulation/​ accessed 31 December 2021. 71 European Commission, EU–US Trade and Technology Council: Commission launches consultation platform for stakeholder’s involvement to shape transatlantic cooperation, 10 October 2021, https://​ ec​.europa​.eu/​commission/​presscorner/​detail/​en/​IP​_21​_5308. 72 Regulation (EU) 2016/679 – General Data Protection Regulation (GDPR). 68 69

Artificial intelligence and law  381 In order to fill this gap, since 2018 the EU has been working to establish a clear and predictable legal framework to govern AI. Notably, the EU’s latest effort – the 2021 AI package – is eye-catching since the EU policy-makers are working on the very first detailed set of comprehensive AI regulations on a global scale; even AI superpowers like the US and China have not yet done this. For these reasons, this section evaluates the EU’s AI legal framework. 4.2.1 The Communication on AI for Europe of 2018 In the Communication on AI for Europe of 2018,73 the European Commission (the Commission) introduced the EU’s first definition of AI. Furthermore, in this Communication, the Commission developed an AI strategy across EU members at the EU level74 and proposed measures to streamline research and policy options for AI regulation, which paved the way for a more comprehensive AI package for the EU in 2021.75 Thus, it is meaningful to examine some details of the Communication. At the outset of the Communication, AI is defined by the Commission as follows: Artificial intelligence (AI) refers to systems that display intelligent behaviour by analysing their environment and taking actions – with some degree of autonomy – to achieve specific goals. AI-based systems can be purely software-based, acting in the virtual world (e.g. voice assistants, image analysis software, search engines, speech and face recognition systems) or AI can be embedded in hardware devices (e.g. advanced robots, autonomous cars, drones or Internet of Things applications).76

Besides this first definition of AI from the EU policy-makers, it is worth mentioning that in various places the Commission refers to AI with nuances. In some places of the Communication, the Commission considered AI to be ‘a generic term that refers to any machine or algorithm that is capable of observing its environment, learning, and based on the knowledge and experience gained, taking intelligent action or proposing decisions’.77 This broad definition can cover a wide range of different AI technologies. So, when facing the pacing problem and the competition from other AI leading countries, the EU policy-makers’ approach is to adopt AI as a ‘generic term’ which covers a wide range of technologies, and the EU policy-makers consider that AI should be defined flexibly to promote innovation. It is also observed that, after the EU launched its first AI definition, the Commission’s High-Level Expert Group on AI maintained but further refined the first definition. The Expert Group released ‘A Definition of AI’ on 18 December 2018 (updated on 8 April 2019).78 In this document, the Expert Group describes AI as ‘systems that display intelligent behaviour by

European Commission, ‘Communication from the Commission to the European Parliament, the European Council, the Council, the European Economic and Social Committee and the Committee of the Regions on Artificial Intelligence for Europe’ (2018) https://​eur​-lex​.europa​.eu/​legal​-content/​EN/​TXT/​ ?uri​=​COM​%3A2018​%3A237​%3AFIN. 74 ibid. 75 ibid p.1. 76 ibid p.1. 77 European Commission Joint Research Centre, Artificial Intelligence: A European Perspective 18 (2018). 78 High-Level Expert Group on Artificial Intelligence—Set up by the European Commission, A Definition of AI: Main Capabilities and Disciplines (first released on 18 December 2018, updated on 8 April 2018). https://​ec​.europa​.eu/​futurium/​en/​system/​files/​ged/​ai​_hleg​_definition​_of​_ai​_18​_december​ _1​.pdf. 73

382  Research handbook on digital trade analysing their environment and taking actions – with some degree of autonomy – to achieve specific goals’ [for example, narrow AI mentioned above], and ‘AI-based systems’ can be ‘embedded in hardware devices’ or ‘purely software-based, acting in the virtual world’.79 So, the Expert Group’s definitions are nearly identical to the first AI definition of the EU that was introduced in 2018. In addition, the Expert Group defines ‘AI’ and ‘AI systems’, respectively, as follows: Artificial intelligence (AI) systems are software80 (and possibly also hardware) systems designed by humans that, given a complex goal, act in the physical or digital dimension by perceiving their environment through data acquisition, interpreting the collected structured or unstructured data, reasoning on the knowledge, or processing the information, derived from this data and deciding the best action(s) to take to achieve the given goal. 81 AI systems can either use symbolic rules or learn a numeric model, and they can also adapt their behaviour by analysing how the environment is affected by their previous actions.82

Subsequently, in 2020 the EU released a number of publications on AI regulation, arising from the Commission’s intention to accelerate digital development in the EU through its adoption of a pan-European Digital Strategy. In February 2020 the Commission published a Report on the safety and liability implications of AI (hereinafter ‘the AI Report’) and a White Paper on AI (hereinafter the ‘AI White Paper’)83 in order to address the aforementioned gap that exists concerning the lack of specific AI regulation in the EU. Following the Report and White Paper, the European Parliament (hereinafter ‘the Parliament’) adopted the texts of legislative proposals in October 2020, to introduce the EU’s AI comprehensive regulatory frameworks, including ‘a legal and ethical framework for AI’84 and ‘a civil liability framework in respect of AI products’.85 4.2.2 The EU’s AI Package of 2021 The EU’s latest effort is the Commission’s long-awaited comprehensive AI package, which was unveiled on 21 April 2021. Aiming to establish a ‘global gold standard’ for AI and posi-

ibid. Emphasis added by the current author. For instance, internet AI as mentioned above is not a physical form of hardware. 81 High Level Expert Group, A Definition of AI, p.8. 82 ibid. 83 European Commission, AI White Paper, 19 February 2020, https://​ec​.europa​.eu/​info/​sites/​default/​ files/​commission​-white​-paper​-artificial​-intelligence​-feb2020​_en​.pdf. See more www​.arthurcox​.com/​ knowledge/​eu​-developments​-in​-ai​-regulation/​. 84 European Parliament, https://​www​.europarl​.europa​.eu/​doceo/​document/​TA​-9​-2020​-0275​_EN​ .html. 85 European Parliament, https://​www​.europarl​.europa​.eu/​doceo/​document/​TA​-9​-2020​-0276​_EN​ .html. 79 80

Artificial intelligence and law  383 tion Europe to play a leading role globally, the Commission’s AI package consists of three parts, including: ● A proposal for an AI Regulation (also called ‘the AI Act’ of the EU) which is the first-ever legal framework on AI in the world and lays down harmonised rules on AI across the EU member states,86 ● a Coordinated Plan with Member States of the EU,87 and ● a Communication on Fostering a European Approach to Artificial Intelligence.88 4.2.3 Aim to make the EU a global hub Although the EU is not an AI superpower like the US and China, the two primary aims of the EU’s proposed AI Act are ‘excellence AI’ and to ‘turn Europe into the global hub for trustworthy AI’.89 Following the model of the EU GDPR 2016 that has had an impact on the EU member states and beyond, through the long-awaited AI package, the Commission aims to align the new AI rulebook with the fundamental EU values ​​that AI systems must adhere to in design, development and use.90 So, as the first-mover in launching comprehensive AI rules on the global scale, the EU’s AI-specific legislation and regulatory approach will likely set standards beyond Europe, even globally, or at least work as a model for other developing countries to follow in managing AI-associated risk within those countries. 4.2.4 EU’s novel risk-based approach The proposed AI Act is the result of the publication of the aforementioned AI White Paper91 and several years of work by the Commission. In the AI Act, the Commission introduces a legal framework for AI that employs a broad AI definition, as discussed below, with a novel ‘risk-based approach’ (see Appendix) to regulate a wide range of AI systems. According to the Commission, ‘these rules will also provide Europe with a leading role in setting the global gold standard’,92 and its objective is to strike a balance between building citizens’ trust in AI systems to mitigate associated risks and boosting investment and innovation in the further development of AI systems which are built upon high-quality data sets.93 As such, the Commission has adopted a novel so-called risk-based approach to regulating

European Commission, Proposal for a Regulation laying down harmonised rules on artificial intelligence, COM(2021) 206 final, 2021/0106(COD), (21 April 2021), https://​digital​-strategy​.ec​.europa​ .eu/​en/​library/​proposal​-regulation​-laying​-down​-harmonised​-rules​-artificial​-intelligence. 87 European Commission, Coordinated Plan on Artificial Intelligence 2021 Review, 21 April 2021, https://​digital​-strategy​.ec​.europa​.eu/​en/​library/​coordinated​-plan​-artificial​-intelligence​-2021​-review. 88 European Commission, Communication on Fostering a European approach to Artificial Intelligence, 21 April 2021, https://​digital​-strategy​.ec​.europa​.eu/​en/​library/​communication​-fostering​ -european​-approach​-artificial​-intelligence. 89 Commission, Europe Fit for the Digital Age, 21 April 2021, https://​ec​.europa​.eu/​commission/​ presscorner/​detail/​en/​IP​_21​_1682. 90 Francine Cunningham, First-ever Legal Framework for AI Proposes Obligations for Developers, Users and Importers, 4 May 2021, www​.twobirds​.com/​en/​insights/​2021/​global/​first​-ever​-legal​ -framework​-for​-ai​-proposes​-obligations​-for​-developers​-users​-and​-importers. 91 European Commission, AI White Paper (n 83). 92 European Commission, ‘A European Approach to Artificial Intelligence’ (2018) https://​digital​ -strategy​.ec​.europa​.eu/​en/​policies/​european​-approach​-artificial​-intelligence. 93 ibid. 86

384  Research handbook on digital trade AI, which abides by an overall rule that ‘the higher the risk is, the stricter the rule is’ (see the table in Appendix). Accordingly, the Commission’s Regulatory Framework divides associated different AI systems into four levels of risk (see Appendix). The EU policy-makers have used a holistic yet flexible approach to regulating AI systems, ranging from weak AI to strong AI, from software-based AI to hardware-based AI, to some unknown means that are yet to come into existence. Before the publication of the proposed AI Act, an online public consultation was launched on 19 February 2020 along with the publication of the AI White Paper, and ran until 14 June 2020.94 Stakeholders have mostly requested a narrow, clear and precise definition for AI.95 Besides the need to clarify the term ‘AI’, stakeholders also highlighted that it is important to define ‘risk’, ‘high-risk’, ‘low-risk’, ‘remote biometric identification’ and ‘harm’.96 Most stakeholders explicitly supported a risk-based framework that the EU proposed to regulate AI: using the risk-based framework was considered a better option than blanket regulation of all AI systems.97 The types of risks and threats should be based on a sector-by-sector and case-by-case approach. The risks of different AI systems should also be calculated, considering the impact on rights and safety.98 Bearing the above feedback from stakeholders to the public consultation in mind, when looking at the final version of the AI Act, it can be seen that the Act has taken into account the feedback from the public consultation but it is not always adopted. The AI Act has not defined ‘AI’ but rather an ‘AI system’, stating that ‘the definition of AI system in the legal framework aims to be as technology neutral and future proof as possible, taking into account the fast technological and market developments related to AI’.99 Furthermore, the Commission noted that ‘the notion of AI system should be clearly defined to ensure legal certainty, while providing the flexibility to accommodate future technological developments’.100 On the one hand, Article 3 of the AI Act stipulates that

94 European Commission, Proposal for a Regulation laying down harmonised rules on artificial intelligence (AI Act), COM(2021) 206 final, 2021/0106(COD), (21 April 2021), https://​digital​-strategy​.ec​ .europa​.eu/​en/​library/​proposal​-regulation​-laying​-down​-harmonised​-rules​-artificial​-intelligence. During the consultation, in total, 1,215 contributions were received, of which 352 were from companies or business organisations/associations, 406 from individuals (92 per cent of individuals from the EU), 152 on behalf of academic/research institutions and 73 from public authorities. Civil society’s voices were represented by 160 respondents (among which 9 consumers’ organisations, 129 non-governmental organisations and 22 trade unions) and 72 respondents contributed as ‘others’. Of the 352 business and industry representatives, 222 were companies and business representatives, 41.5 per cent of which were micro, small and medium-sized enterprises. The rest were business associations. Overall, 84 per cent of business and industry replies came from the EU-27. Depending on the question, between 81 and 598 of the respondents used the free text option to insert comments. More than 450 position papers were submitted through the EU Survey website, either in addition to questionnaire answers (more than 400) or as stand-alone contributions (more than 50). 95 ibid. 96 ibid. 97 ibid. 98 ibid. 99 ibid. 100 ibid. EU AI Act of 2021, Preamble, Paragraph (6).

Artificial intelligence and law  385 ‘artificial intelligence system’ (AI system) means software that is developed with one or more of the techniques and approaches listed in Annex I101 and can, for a given set of human-defined objectives, generate outputs such as content, predictions, recommendations, or decisions influencing the environments they interact with.102

On the other hand, the AI Act states that the techniques and approaches listed in Annex I can be updated.103 This arrangement means the Commission recognises that AI is fast-evolving and comprises an array of technologies, so the law should be flexible and future-proof. Addtionally, the EU’s regulatory approach regarding AI demonstrates two characteristics. On the one hand, the EU employs a proactive approach to regulating AI, similar to that of the US. On the other, the EU’s approach is in a way that also showcases the EU’s attributes and particularity. Like the US, European countries have taken a proactive stance towards AI regulations and policy, such as the AI-specific regulations, since 2018. Meanwhile, despite having certain competitive advantages, such as a strong scientific and industrial base to build on, leading AI research and talent and recognised leadership in robotics and innovative start-ups, the EU has recognised it has the potential to become, but is not yet, a world leader like the US and China in the AI arena.104 Furthermore, the EU’s AI strategy emphasises that European countries must jointly commit to making Europe an AI superpower at the EU level, and it is working towards a clear and predictable legal framework to ensure consumer protection and legal certainty.105 In summary, while the EU shares some similarities with the US and China in adopting flexible, broad AI definitions to promote innovation and ‘excellence AI’ in Europe, the way the EU regulates also shows its particularity, such as focusing on ‘trustworthy AI’ and the EU’s fundamental values.106 Furthermore, it is worth noting that the EU has recognised that its competitive advantages in AI have not made it an AI superpower, so European countries need to join forces at the EU level.107 The EU’s AI agenda states that ‘the main ingredients are there for the EU to become a leader in the AI revolution, in its own way and based on its values’, such as safeguarding people’s safety and fundamental rights and protecting consumers.108 Thus,

101 ibid. EU AI Act of 2021, ANNEX I (ARTIFICIAL INTELLIGENCE TECHNIQUES AND APPROACHES) referred to in Article 3, point 1: (a) Machine learning approaches, including supervised, unsupervised and reinforcement learning, using a wide variety of methods including deep learning; (b) Logic- and knowledge-based approaches, including knowledge representation, inductive (logic) programming, knowledge bases, inference and deductive engines, (symbolic) reasoning and expert systems; (c) Statistical approaches, Bayesian estimation, search and optimization methods. 102 EU AI Act Title 1 (General Provisions) Article 3 (1) Definitions. 103 EU AI Act Article 4 (Amendments to Annex I). The Commission is empowered to adopt delegated acts in accordance with Article 73 to amend the list of techniques and approaches listed in Annex I, in order to update that list to market and technological developments on the basis of characteristics that are similar to the techniques and approaches listed therein. 104 European Commission (n 73). COM(2018) 237 final. Communication Artificial Intelligence for Europe (25 April 2018), https://​digital​-strategy​.ec​.europa​.eu/​en/​library/​communication​-artificial​ -intelligence​-europe (hereinafter Communication on AI for Europe). 105 ibid, Communication on AI for Europe. 106 European Commission (n 92). 107 European Commission (n 73). https://​digital​-strategy​.ec​.europa​.eu/​en/​library/​communication​ -artificial​-intelligence​-europe, https://​eur​-lex​.europa​.eu/​legal​-content/​EN/​TXT/​?uri​=​COM​%3A2018​ %3A237​%3AFIN. 108 European Commission (n 92). Emphasis by the current author.

386  Research handbook on digital trade divergent AI regulations have been emerging on the global scale among the first movers – the US, the EU and China. 4.3

From a Chinese Perspective

While some WTO members appear to be following the US approach by capturing AI along the same lines,109 it can be seen that an operative definition of AI still eludes many other countries when contemplating their recent soft and hard law on AI. Back in 2017, China was not an AI superpower. Surprisingly, just five years after Ke Jie’s loss to AlphaGo, which attracted the Chinese central government’s attention and saw it begin to take action, China has become an AI superpower. So, it is interesting to examine how China upgraded its AI capacity and the interplay between law and technology in advancing AI in China. Notably, China, another global superpower in shaping AI development, has yet to offer a clear-cut definition of AI.110 This contribution argues that China prefers a cross-sectoral, piecemeal but flexible approach, which seems to suit China’s AI technological development, since this approach has successfully boosted AI development in China and made it an AI superpower like the US in the short run (from 2017 to now). In 2017, with AlphaGo – a product of the British AI start-up DeepMind, which Google had acquired in 2014 – the West seemed poised to continue its dominance in the AI era.111 However, shortly after Ke Jie’s loss to AlphaGo, the Chinese central government released a comprehensive blueprint for Chinese leadership in AI. It is noted that China’s AI plan originated at the highest level of the central government,112 such as the ‘Mass Innovation and Mass Entrepreneurship’ campaign.113 Moreover, China’s AI plan is turbocharging growth through a flood of new funding, including subsidies for AI start-ups, generous government contracts and soft laws to accelerate the adoption of AI.114 For example, in 2017, the State Council released the country’s strategy for developing AI, entitled ‘New Generation Artificial Intelligence Development Plan’ (AIDP).115

109 For instance, the EU seems increasingly to align its AI strategy with the US. Taiwan, for instance, has attempted to define AI as broadly covering, among others, ‘systems acting as humans’ and ‘systems acting rationally’. See, e.g., Rengong Zhihui Fazhan Jibenfa Caoan (人工智慧發展基本法(草案) [Draft Bill of Basic Law Governing Development of Artificial Intelligence (Taiwan)], article 2 (hereinafter Taiwan AI Development Basic Law (Bill)). 110 See, e.g., Guowuyuan guanyu yinfa xin yidai rengong zhineng fazhan guahua de tongzhi [Notice of the State Council on Issuing the Development Plan on the New Generation of Artificial Intelligence] (promulgated by the State Council, 8 July 2017). 111 Lee (n 22) p.13. 112 E.g. the Communist Party of China (CPC) Central Committee and the State Council of PR China. 113 State Council of PR China, More actions to promote mass entrepreneurship and innovation, 25 June 2021, http://​english​.www​.gov​.cn/​premier/​news/​202106/​25/​content​_WS60d5​89a4c6d0df​ 57f98dbddd​.html. 114 Lee (n 22) pp.89–90. 115 State Council of China, A New Generation of Artificial Intelligence Development Plan (AIDP) [新 一代人工智能发展规划], State Council Document [2017] No. 35; there is no official translation of the strategy document at the time of writing. An unofficial translation might be found at https://​flia​.org/​wp​ -content/​uploads/​2017/​07/​A​-New​-Generation​-of​-Artificial​-Intelligence​-Development​-Plan​-1​.pdf. For an exposition of the AIDP, see H., Roberts, J. Cowls, J. Morley et al., The Chinese Approach to Artificial Intelligence: An Analysis of Policy, Ethics, and Regulation. AI and Society (2021) 36, 59–77.

Artificial intelligence and law  387 This strategy paper of 2017 sets out China’s aims to become the world leader in AI by 2030.116 In particular, China sees an opportunity to monetise AI into a trillion-yuan industry. The strategy also sets out in general terms the commitment to establish ethical norms and standards for AI, at least within the jurisdiction of China. Driven by the Chinese central government’s and regional initiatives, China has become an AI superpower like the US in some areas of AI applications.117 Against this backdrop, China, as an AI superpower nowadays, has recently passed three AI-related statutes: ● Cybersecurity Law of the People’s Republic of China, 2016;118 ● Data Security Law of the People’s Republic of China, 2021;119 ● Personal Information Protection Law of the People’s Republic of China, 2021.120 In summary, China prefers a cross-sectoral, piecemeal and flexible approach. It is worth noting that the above three pieces of legislation are AI-related but not AI-specific rules, unlike the EU’s AI package. Similar to the EU GDPR, these three pieces of legislation govern AI applications, due to the importance of massive amounts of data, as pointed out at the outset. However, none of them is AI-specific legislation, which currently does not exist in China so far. Through checking the texts of the three legislation, the author finds that neither the Chinese Cybersecurity Law nor the Data Security Law uses the term AI even once; the Chinese Personal Information Protection law mentions the term AI only once, without any elaboration or definition. On the other hand, the central and local soft laws in China that have been adopted since 2017 still play an vital role in the field of AI in China, as explained above. 4.4

From a UK Perspective

Besides China, some other countries also seem reluctant to crystallise the statutory definition of AI further. For instance, the UK House of Lords simply refers to ‘AI’ as technologies with the ability to ‘perform tasks that would otherwise require human intelligence’, bearing in mind that contemporary AI systems generally have the capacity to ‘learn or adapt to new experiences’.121 Similar to China, the UK has also chosen to adopt a more technically focused but piecemeal approach to regulating AI. On the one hand, the EU GDPR 2016, which was passed before

ibid. Lee (n 17). 118 Cybersecurity Law of the People’s Republic of China [中华人民共和国网络安全法] (Adopted at the 24th meeting of the Standing Committee of the Twelfth National People's Congress on 7 November 2016), www​.cac​.gov​.cn/​2016​-11/​07/​c​_1119867116​.htm. 119 Data Security Law of the People’s Republic of China [中华人民共和国数据安全法] (Adopted at the 29th meeting of the Standing Committee of the 13th National People's Congress on 10 June 2021), www​.npc​.gov​.cn/​npc/​c30834/​202106/​7c​9af12f5133​4a73b56d79​38f99a788a​.shtml. 120 China, Personal Information Protection Law of the People's Republic of China [中华人民共和 国个人信息保护法] (Adopted at the 30th meeting of the Standing Committee of the 13th National People’s Congress on 20 August 2021) www​.npc​.gov​.cn/​npc/​c30834/​202108/​a8​c4e3672c74​491a80b53a​ 172bb753fe​.shtml. 121 UK House of Lords Select Committee on Artificial Intelligence, ‘AI in the UK: Ready, Willing and Able?’ (2019) https://​publications​.parliament​.uk/​pa/​ld201719/​ldselect/​ldai/​100/​10002​.htm at 14. Likewise, other common law countries seem to loosely define the term AI. 116 117

388  Research handbook on digital trade Brexit, still applies to the UK and regulates AI due to AI’s reliance on data.122 On the other, instead of adopting single all-encompassing legislation on AI, like the EU AI package, the UK chose to publish a National AI Strategy on 22 September 2021.123 Subsequently, the UK government has launched an AI Standards Hub for the UK to shape and improve global technical standards for artificial intelligence. For instance, the UK is currently developing and agreeing on international technical standards, working with the International Organization for Standardisation and International Electrotechnical Commission (ISO/IEC) and the Industry Specification Group on Securing AI at the European Telecommunications Standards Institute (ETSI).124

5.

AN OVERVIEW OF CHALLENGES TO GLOBAL TRADE LAW AND GOVERNANCE IN THE AGE OF AI

Rapidly developing AI technologies have led to the pacing problem, not only at a national level for policy-makers from WTO members with the emerging divergent national regulatory responses, as discussed above, but also at the international level, challenging global trade governance. The aims of the remainder of this contribution are twofold: first, to illustrate AI-associated legal challenges to the global trade governance (see Figure 22.1 below); second, to shed light on the WTO Plurilateral Negotiations on Trade-Related Aspects of Electronic Commerce.125 The global trading system under the WTO framework mainly governs trade in goods, services and trade-related intellectual property rights (see the figure below), all of which are facing the pacing problem. Two of the most important WTO-based multilateral agreements are the General Agreement on Tariffs and Trade (hereinafter ‘GATT’)126 and the General Agreement on Trade in Services (hereinafter ‘GATS’).127 The third major multilateral trade agreement of the WTO is the Agreement on Trade-Related Aspects of Intellectual Property Rights (hereinafter ‘TRIPS’).128 This section explores the legal issues facing the global trade governance under the WTO legal framework, particularly the GATT–GATS legal framework and TRIPS, brought about by AI technologies.

122 The Data Protection Act 2018 is the UK's implementation of the General Data Protection Regulation (GDPR). See more www​.gov​.uk/​data​-protection​#:​~:​text​=​The​%20Data​%20Protection​ %20Act​%202018​%20is​%20the​%20UK's​%20implementation​%20of​,used​%20fairly​%2C​%20lawfully​ %20and​%20transparently. 123 UK Government, National AI Strategy: Our Ten-Year Plan to Make Britain a Global AI Superpower, www​.gov​.uk/​government/​publications/​national​-ai​-strategy/​national​-ai​-strategy​-html​ -version. 124 ibid. 125 WTO, E-Commerce, www​.wto​.org/​english/​tratop​_e/​ecom​_e/​ecom​_e​.htm. 126 General Agreement on Tariffs and Trade, 30 October 1947, 61 Stat. A-11, 55 U.N.T.S. 194 (hereinafter GATT). 127 General Agreement on Trade in Services Article XIV, 15 April 1994, Marrakesh Agreement Establishing the World Trade Organization, Annex 1B, 1869 U.N.T.S. 183 (hereinafter GATS). 128 Agreement on Trade-Related Aspects of Intellectual Property Rights, Marrakesh Agreement Establishing the World Trade Organization, Annex 1C, 15 April 1993, 1869 U.N.T.S. 299; 33 I.L.M. 1197 (1994) (hereinafter TRIPS).

Artificial intelligence and law  389

Figure 22.1

WTO agreements

Note: The figure above illustrates the overall structure and content of global trade law. Source: Author. See the full texts of the Uruguay Round Agreements on the WTO official website, www.wto.org/ english/docs_e/legal_e/legal_e.htm

5.1

The Pacing Problem Arising from the Goods/Services Dichotomy

In today’s global trading system, GATT is the multilateral agreement that applies to international trade in goods, whereas GATS is the multilateral agreement that governs international trade in services (see Figure 22.1). In addition, the application of GATS varies according to how a particular service is supplied; GATS defines four modes of services as to how the service is so supplied.129 These are Mode 1 (Cross-border supply),130 Mode 2 (Consumption abroad),131 Mode 3 (Commercial presence)132 and Mode 4 (Cross-border movement of natural persons supplying services).133 The renaissance of AI with real-life applications has posed increasing challenges to the GATT–GATS legal framework, particularly the goods/services dichotomy. This is because the GATT–GATS legal framework was negotiated from 1986 to 1994, and came into force in 1995.134 It is therefore worthwhile to note that the global trading system, including its GATT–

GATS Article I.2. Services supplied from the territory of one WTO Member into the territory of any other. 131 Services supplied in the territory of one WTO Member to the service consumer of any other Member. 132 Services supplied by a service provider of one WTO Member, through commercial presence (e.g., companies), in the territory of any other Member. 133 Services supplied by a service provider of one WTO Member, through the presence of natural persons of a Member in the territory of any other Member. 134 WTO, Legal Texts, www​.wto​.org/​english/​docs​_e/​legal​_e/​legal​_e​.htm. See also Lijun Zhao, Transportation, Cooperation and Harmonization: GATS as a Gateway to Integrating the UN Seaborne Cargo Regimes into the WTO (2015) 27 Pace International Law Review 60, available at SSRN: https://​ ssrn​.com/​abstract​=​3226314. 129 130

390  Research handbook on digital trade GATS legal framework, was negotiated and concluded at a time when much of the online activities with which we are familiar today did not exist. So the negotiators had not considered the post-1994 blurring of the boundary between tangible goods and intangible services, which is inconsistent with the WTO’s goods/services dichotomy.135 Furthermore, the proliferation of e-commerce over the past two decades has revealed the limitations of the existing global trade law. With the widespread use of the internet, various sorts of ‘digital products’ – electronic goods or intangible goods that exist in digital form – which possess conventional characteristics of both ‘goods’ and ‘services’ as defined under GATT and GATS, respectively, have caused difficulties in applying the GATT–GATS legal framework.136 5.1.1 Whether ‘digital products’ fall within the scope of GATT, GATS, or both There have been debates among academics, practitioners and policy-makers on the issue of whether businesses that supply digital products, such as books or music in digital form, over the internet should fall within the scope of GATT, GATS or both.137 This question is fundamental, and it matters to global trade law. For example, regarding market access of ‘digital products’ in a WTO member state, whether such products are subject to the application of GATT and/or GATS matters, because they adopt different approaches in terms of their applications: GATT adopts the negative list, but GATS utilises the positive list and has no guarantee, regarding market access.138 Likewise, this aforementioned fundamental question remains unresolved and addresses the same legal problems and challenges to AI-enabled products/services. With the rise of the internet and various sorts of ‘digital products’, the WTO has been trying to update the GATT– GATS legal framework and the corresponding goods/services dichotomy since 1998 through its Electronic Commerce Work Programme.139 Nevertheless, after many years of hard work, little progress has been made in updating the WTO agreements. Some WTO members, notably from the EU, argue that the ‘products’ in question lack physical attributes; thus, they should be a service(s) and governed by GATS.140 In contrast, other WTO members, especially the US, insist that digitally delivered content-based products should be considered ‘goods’ instead of ‘services’, because of their ‘durability’ and ‘inseparability’ from the physical medium.141 5.1.2 Attempted solutions within and outside of the global trade system There have been many attempts, both within and outside of the global trading system, to resolve the pacing problem. Within the global trading system, the pacing problem facing

ibid. Rolf H. Weber and Mira Burri, Classification of Services in the Digital Economy (Springer 2013). See details of the negotiating history in Sacha Wunsch-Vincent, The WTO, The Internet and Trade in Digital Products: EU–US Perspectives (Hart 2006) 35, 55–62. 137 Mark Wu, Digital Trade-Related Provisions in Regional Trade Agreements: Existing Models and Lessons for the Multilateral Trade System (E15 Initiative 2017). 138 ibid. See also Lijun Zhao, Transportation, Cooperation and Harmonization: GATS As a Gateway to Integrating the UN Seaborne Cargo Regimes into the WTO, 27 Pace International Law Review 60 (2015). 139 WTO, E-Commerce, www​.wto​.org/​english/​tratop​_e/​ecom​_e/​ecom​_e​.htm. 140 Wunsch-Vincent (n 136). 141 ibid. 135 136

Artificial intelligence and law  391 global trade law, particularly the GATT–GATS legal framework, has been resolved to some extent by the WTO adjudicators. In the landmark decisions of US–Gambling142 and China– Publications and Audiovisual Products,143 GATS was interpreted by the WTO Appellate Body to govern the cross-border trade in digital content.144 Beyond the global trading system, the pacing problem associated with the goods/services dichotomy has been resolved to some extent. In the context of preferential/regional trade agreements (PTAs/RTAs), such as the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP), electronic commerce provisions have been introduced under Chapter 14 of CPTPP.145 In accordance with Article 14.1 of CPTPP, ‘digital product’ has been defined as ‘a computer programme, text, video, image, sound recording or other product that is digitally encoded, produced for commercial sale or distribution, and that can be transmitted electronically’.146 Nevertheless, this article is accompanied by remarks in the footnotes of CPTPP147 stating that this definition ‘should not be understood to reflect a Party’s view on whether trade in digital products through electronic transmission should be categorised as trade in services or trade in goods’.148 5.1.3 Legal gaps and uncertainty of the GATT–GATS legal framework in the AI era There are two observations regarding the legal gaps in the GATT–GATS legal framework. First, as discussed above, the two streams of aforementioned efforts, that are within and beyond the global trading system, have addressed but not fully resolved the legal gap and uncertainty regarding digital products. Second, in the field of AI, the unresolved legal gap, as part of the pacing problem of global trade law under the goods/services dichotomy, will be more problematic and perplexing when the law meets the AI-enabled products/services. As noted above, different national approaches to regulating AI have emerged in a handful of AI leaders; there is an emerging tendency that WTO members will likely adopt different national approaches to regulating AI-enabled products/services. For instance, while the EU launched the first-ever comprehensive AI legal/regulatory framework (that is, the EU’s AI pack, mentioned previously) with an ambition to make its proposed AI Act to be a model for the rest of the world to follow when regulating AI technology,149 this contribution has identified that other AI powers, particularly the US, China and the UK, have been working on their own

United States–Measures Affecting the Cross-Border Supply of Gambling and Betting Services (AB Report, US–Gambling) (2005) 20 April 2. 143 China–Measures Affecting Trading Rights and Distribution Services for Certain Publications and Audiovisual Entertainment Products (AB Report, China–Publications and Audiovisual Products) (2010) WT/DS363/A. 144 See details in Peter Van den Bossche and Zdouc Werner, The Law and Policy of the World Trade Organization Text, Cases, and Materials (5th edn, Cambridge University Press 2021). 145 CPTPP, Full Text, www​.mfat​.govt​.nz/​vn/​trade/​free​-trade​-agreements/​free​-trade​-agreements​-in​ -force/​comprehensive​-and​-progressive​-agreement​-for​-trans​-pacific​-partnership​-cptpp/​comprehensive​ -and​-progressive​-agreement​-for​-trans​-pacific​-partnership​-text​-and​-resources/​. 146 ibid, CPTPP, Article 14.1 (Definitions – ‘Digital Products’). 147 ibid, CPTPP, Article 14.1 (Definitions – ‘Digital Products’), and accompanying footnotes 2 and 3. 148 ibid. 149 Euronews, EU’s Artificial Intelligence Law Should Serve as ‘Model across the Globe’, 25 October 2021. 142

392  Research handbook on digital trade national approaches to regulating AI. As such, the rest of the WTO members are facing at least a handful of emerging divergent models for regulating AI technology. In order to explain the pacing problem under the GATT–GATS legal framework, this contribution uses AI-enabled legal services (that is, legal AI, AI lawyers or robot lawyers150) to illustrate the mismatch between fast-developing AI technology and the existing global trade law. Assuming that AI technology will continue to develop and mature as predicted by AI experts such as Kai-Fu Lee, legal AI can be divided into four categories, depending on the level of intelligence and automation (from low to high), as follows: ● ● ● ●

Level 1 Legal AI: Specialised standalone technologies (for example, legal chatbots); Level 2 Legal AI: Enablers of legal advice (for example, automated document review); Level 3 Legal AI: Further enablers of legal advice (for example, legal data analytics); and Level 4 Legal AI: Human-free smart contracts (for example, a kind of general AI; not necessarily an AGI, but an AI with a high degree of automation).151

While we are not yet at Level 4 (that is, general AI, as explained earlier), AI lawyers, with ability at the other three levels, already exist in our lives. The existing AI lawyers are usually run through website platforms or mobile apps, so foreign users may access such an AI lawyer easily via the internet. For instance, there exist some website-based AI lawyers in the US and China, including: ● ‘ROSS Intelligence’ in the US and Europe, a leading AI-enabled computer program that uses natural language processing to help conduct legal research and document review on US laws;152 ● ‘Xiao Xia’,153 a leading legal AI on Chinese law which is based on an online platform called China Legal Service Network,154 by the Chinese Ministry of Justice; ● ‘Bao Xiaohei’,155 which is owned and run by a company in Hangzhou, China; and ● ‘AI Legal Counsellor’ of Zhongshan Municipal Bureau of Justice of China.156 Readers should remember that all existing AI lawyers are weak AI, not general AI, at least at the moment. However, it is foreseen by many AI experts that AI will replace human labour,

AI in Law and Legal Practice – A Comprehensive View of 35 Current Applications (July 2021) https://​emerj​.com/​ai​-sector​-overviews/​ai​-in​-law​-legal​-practice​-current​-applications/​. 151 ibid. See also Han-Wei Liu and Ching-Fu Lin, ‘Artificial Intelligence and Global Trade Governance: A Pluralist Agenda’ (2020) 61 Harvard International Law Journal 61 at 421. 152 ROSS Intelligence (‘ROSS’) builds AI-driven products to augment lawyers’ cognitive abilities: https://​rossintelligence​.com/​about​-us. As of 31 January 2021 the ROSS platform became unavailable and was shut down because Thomson Reuters and Westlaw brought a spurious lawsuit against ROSS on the ground of copyrighted data on judgments being used by ROSS without authorisation. 153 Lv Pin Ltd, Xiao Xia Legal AI, www​.lvpin100​.com/​#/​pro​_xiaoxia. 154 Through an online platform of the ministry of justice https://​ai​.12348​.gov​.cn/​api/​speed​-front/​ freeReading​?url​=/​pc/​, this legal AI issue advances legal opinions for clients free of charge. Up to 1 June 2022, 2,876,071 professional legal opinions had been issued. 155 Bao Xiaohei, Free AI lawyer, https://​pc​-bxh​.ai​-indeed​.com/​. 156 AI Legal Counsel of Zhongshan Municipal Bureau of Justice (Guangdong province, China), launched from 2 April 2020, https://​m​.thepaper​.cn/​baijiahao​_6796505. 150

Artificial intelligence and law  393 including human lawyers – especially those in entry-level positions, such as paralegals and first-year associates.157 Therefore, advanced legal AI of Level 3 or above is technologically feasible, and it is merely a matter of time before they emerge to present new challenges to the GATT–GATS legal framework. Take a legal AI that is run in China as an example in this section, since literature in English rarely discusses AI lawyers in China, but there are some about legal AI in the US.158 In China, several AI lawyers projects have been undertaken by the central government, local governments, and enterprises.159 Thus, the availability of massive data (such as massive information on judgments for AI lawyers) is essential to legal AI systems: this is a problem for some legal AI systems in the US, such as the ROSS as mentioned above, since data on judgments are not available in the public domain but are privately owned by databases (such as Westlaw) of big corporations (such as Thomson Reuters); however, the availability of massive data for legal AI is not a problem in China, because Chinese courts publish judgments to the public for free on a database launched by the Supreme Court of China that includes all judgments from all Chinese courts since 2013.160 As such, similar to the statutory definitions of AI discussed above, this contribution identifies and discusses legal challenges associated with legal AI over global trade law, when AI-enabled legal services in question are supplied in a cross-border context. Based on the discussion above, current WTO law has not clarified whether digital products/ services, such as legal AI, are subject to GATT, GATS or both. There is not much difference between AI and the aforementioned digital products/services. Next, if the AI lawyer were treated as ‘services’, a follow-up question is whether the services provided by a foreign AI lawyer are through Mode 1 or 2 ‘modes of provision’161 of services in the eyes of GATS. However, as noted above, this legal issue is also unresolved both within and beyond the WTO, which is built mainly upon the GATT–GATS legal framework and the corresponding goods/ services dichotomy. Let us now use AI lawyers as an example of AI-enabled digital products/services to illustrate the legal challenges that AI poses to the GATT–GATS legal framework. First, it is unclear whether a WTO member can ban website-based ‘foreign AI lawyers’, such as ROSS or Xiao Xia, from providing legal advice in cross-border situations. Technically speaking, legal AI is not a ‘lawyer’ under current law, unless most WTO member countries were to amend their national laws on AI’s personhood (legal personality or the like) to recognise AI’s independent personhood from its inventors (such as programmers). For a detailed analysis, China and its GATS commitments are used here162 to illustrate the legal uncertainty of legal AI products/services in cross-border trade settings. Suppose

Lee and Wang (n 3). See legal AI in the US in Liu and Lin (n 151). 159 AI legal services embedded in Gov.cn, https://​ai​.12348​.gov​.cn/​pc/​(having issued 2,698,767 pieces of legal opinions, each of which covers legal analysis and advice in detail. Zhongshan Municipal Bureau of Justice, AI legal advisor, www​.zs​.gov​.cn/​zssfj/​gkmlpt/​content/​1/​1722/​post​_1722215​.html​#1128. Bao Xiaohei project, a personal legal adviser, in Hubei Provincial Government Portal of China, The AI​​ Entrepreneurship Competition of the Central Venture Capital Conference (2019-06-28). 160 China Judgment Online [database], https://​wenshu​.court​.gov​.cn/​ (judgments from 2013 are all published in this Chinese Supreme Court-led database). 161 See GATS (nn.130, 131, 132, 133, and 134). 162 The selection is because of China’s prominent role in global trade and AI techonogy. 157 158

394  Research handbook on digital trade a website-based ‘foreign’ AI lawyer provides its products/services in a WTO member state, such as China; in this hypothetical scenario, China’s GATS Schedule on the Chinese legal services sector is governing law. After searching the WTO’s GATS Schedule database, it is found that China has submitted its GATS-specific schedule on the legal services sector (CPC 861, excluding Chinese law practice).163 When applying the GATS-specific schedule to website-based ROSS-like ‘foreign’ AI lawyers in China, some legal uncertainties exist, as follows. First, regarding Modes 1 and 2 and corresponding concessions that China made to the legal service sector, the specific schedule shows that China inscribes ‘None’ in relation to this sector.164 Thus, there are no limitations specific to this sector.165 Also, it is controversial whether such a foreign AI lawyer, such as a ROSS-like legal AI, has received the legal education and training to be eligible for sitting in an accredited bar exam in the country of the legal AI, for example, the US bar exam for ROSS.166 That is why the previous analysis mentioned that it is unclear in reality whether a WTO member would ban website-based ‘foreign’ AI lawyers. Second, regarding Mode 3 and corresponding concessions that China made to the legal service sector, ‘foreign law firms’ and ‘foreign representative offices’ can consult the laws of a jurisdiction in which ‘the lawyers of the law firm are permitted to engage in the lawyer’s professional work’.167 When it comes to a foreign AI lawyer, it is problematic whether the foreign AI lawyer with the presence of physical hardware or equivalent in China constitutes and fulfils the concepts of ‘foreign law firms’ and ‘foreign representative offices’ stated in China’s concession mentioned under Mode 3. So, this is another legal uncertainty. In addition, assuming that AI will continue to mature and legal AI will reach Level 4, regardless of such legal AI system being embodied in physical robotics, hardware, software only, or any future forms yet unknown, the Level 4 legal AI will also propose new challenges to the WTO legal framework. One of the key challenges is that if a Level 4 foreign AI lawyer (not necessarily an AGI, but an AI with a high degree of automation) is embodied in a physical form (not necessarily a humanoid) and it is traded in a cross-border context, then it is legally uncertain under current GATT–GATS legal framework whether GATT or GATS regulates such a transaction; if GATS regulated the hypothetical scenario, the probably most significant legal uncertainty would exist when the legal AI is providing a service in Mode 4 of GATS, compared with that involving Modes 1, 2 and 3. The legal uncertainty is caused by several reasons. First, the current WTO case law has identified ambiguity in the goods/services dichotomy. Notably, in China–Publications and Audiovisual Products, the WTO Appellate Body held that

WTO, China’s GATS Schedules: Specific Commitment (legal services sector included), Doc. GATS/SC/135 (14 February 2002). See full text from https://​docs​.wto​.org/​dol2fe/​Pages/​ FE​_Search/​FE​_S​_S009​-DP​.aspx​?language​=​E​&​CatalogueIdList​=​71613​,11725​,21775​,25776​,34016​&​ Cur​rentCatalo​gueIdIndex​=​2​&​FullTextHash​=​&​HasEnglishRecord​=​True​&​HasFrenchRecord​=​True​&​ HasSpanishRecord​=​True. 164 ibid. 165 WTO, Guide to Reading the GATS Schedules of Specific Commitments and the List of Article II (MFN) Exemptions, www​.wto​.org/​english/​tratop​_e/​serv​_e/​guide1​_e​.htm. 166 Liu and Lin (n 151). 167 See China–Schedule of Specific Commitments, WTO Doc. GATS/SC/135 (14 Feburary 2002) (China–GATS Commitments). 163

Artificial intelligence and law  395 ‘where the content of a film is carried by a physical delivery [of] materials’, China’s restrictions will ‘inevitably regulate who may import goods for the plain reason that the content of a film is expressed through, and embedded in, a physical good’.168 So, GATT may potentially apply, provided that the legal AI is embodied in the physical medium, regardless of being human-shaped or other-shaped.169 Second, WTO members have developed divergent national regimes on the above legal issue of whether the presence or absence of physical forms affect the application of law and adopted different national laws. Some countries, such as the US and the EU countries, have recognised software as ‘goods’.170 By contrast, in English law, notably in the case of St Albans City and District Council v International Computers, ‘as discs were certainly goods, software that was delivered on a disc would be goods too, but a software program, in itself, does not amount to goods’.171 Therefore, it is likely that divergent national regimes to this legal question will continue to exist for some time, affecting whether GATT or GATS is applicable. Another key challenge is that the above legal problems will be intensified because some WTO members are considering recognising the legal personhood of robot lawyers – such as the EU172 – but some other countries have no such plan yet. If this sort of law confirming AI’s legal personhood becomes effective in some WTO members, it is uncertain whether robot lawyers (for example, Legal 4 legal AI) should be subject to Mode 4 of GATS.173 Similar to the emerging divergent national regimes on AI definitions, WTO members have revealed divergent answers to the question of AI’s legal personhood. 5.2

Emerging Divergent National AI Regulations and Impact on TRIPS Agreement

TRIPS, which came into effect on 1 January 1995, is the most comprehensive existing multilateral agreement on intellectual property rights.174 This contribution argues that, in addition to challenges facing the GATT–GATS legal framework as discussed above, TRIPS also faces the pacing problem and has been shown to fall short in keeping pace with the fast-evolving AI technologies.175 AI has become a tool for painting, poetry writing and journalism;176 for instance, the Washington Post, Forbes, Bloomberg, Reuters and the Guardian have used

168 China–Measures Affecting Trading Rights and Distribution Services for Certain Publications and Audiovisual Entertainment Products (AB Report, China–Publications and Audiovisual Products) (n 144) para 188. 169 See also UK case law on the Sale of Goods Act 1979 as amended in 1994 and 1995 on software, and relevant case law. 170 see e.g. CJEU Case referred by UK – Software Incubator v Computer Associates (2021). 171 [1996] 4 All ER 481 (CA). 172 For instance, during the public consultation the EU proposed legal personhood of AI, but objections were addressed by AI experts in a joint letter, and the EU parliament abandoned the proposal. 173 Liu and Lin (n 151). 174 WTO, Overview: The TRIPS Agreement, www​.wto​.org/​english/​tratop​_e/​trips​_e/​intel2​_e​.htm. 175 For a general overview of challenges, see WTO, World Trade Report 2018: The Future of World Trade: How Digital Technologies are Transforming Global Commerce (2018). 176 Lee and Wang (n 3).

396  Research handbook on digital trade robot-writing AI systems.177 However, it is legally unclear who is the ‘author’ or ‘inventor’ of the AI-enabled algorithmically authored works. Therefore, while TRIPS has facilitated the harmonisation of intellectual property law globally through implementing minimum standards among WTO members, TRIPS has shown a tendency to fail this task in the age of AI. On the one hand, TRIPS is a minimum standards agreement and enhances the harmonisation of the law across the world. In respect of each of the main areas of intellectual property covered by TRIPS, TRIPS sets out the minimum standards of protection to be provided by each WTO member. On the other hand, it is difficult to maintain the minimum standard approach of TRIPS in the AI era. In regard to the emergence of divergent national regulatory approaches on a global scale, at least three different styles of national regulatory approaches have emerged to govern algorithmically authored works. The first national approach to regulating algorithmically authored works, exemplified by the UK legislation under its Copyright, Designs and Patents Act 1988 (CDPA),178 features more flexibility and could readily fit into the new setting of AI.179 For example, under Section 9(3) of CDPA, for a literary, dramatic, musical or artistic work that is computer-generated, the author shall be ‘the person by whom the arrangements necessary for the creation of the work are undertaken’. Additionally, the term ‘computer-generated’ is defined under Section 178 of CDPA as the work ‘generated by computers in circumstances such that there is no human author of the work’. Therefore, the UK approach protects algorithmic creations, though it leaves no room for AI itself to be an ‘author’. Instead, authorship of algorithmically authored works will be attributed to ‘the person by whom the arrangements necessary for the creation of the work are undertaken’, for example, programmers of computer software.180 Some other countries have adopted the second national approach to regulating algorithmically authored works, notably the US. The US Copyright Law only protects ‘the fruits of intellectual labor’ that ‘are founded in the creative powers of the mind’.181 Therefore, under US-style national legislation, such a nation ‘will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author’.182 So, under this type of national legislation, it seems neither the human creator nor the AI itself is entitled to claim authorship. More recently, a third regulatory approach has emerged in Europe alongside increased computer power and the EU’s ambition to become a global hub in the AI age. In the EU, according to Article 2(1) of the Computer Directive and Article 4(1) of the Database Directive, ‘author’ is defined as a ‘natural person’ or group of natural persons who create something (such as AI); however, there is a proviso here that the EU permits national laws of EU member countries

Jaclyn Peiser, The Rise of the Robot Reporter, New York Times, 5 February 2019, www​.nytimes​ .com/​2019/​02/​05/​business/​media/​artificial​-intelligence​-journalism​-robots​.html. 178 Legislation.gov.uk, UK Copyright, Designs and Patents Act 1988, www​.legislation​.gov​.uk/​ukpga/​ 1988/​48/​contents. 179 Colin R. Davies, An Evolutionary Step in Intellectual Property Rights – Artificial Intelligence and Intellectual Property, 27 Computer L. & Sec. Rev. 601 (2011) 155. 180 UK Copyright, Designs and Patents Act 1988, Section 9(3). 181 US Copyright Act of 1976, 17 U.S.C. §§ 101–102 (1976). 162. See also US Copyright Office, Compendium of U.S. Copyright Office Practices § 306 (3rd ed., 2017). 182 US Copyright Act of 1976, 17 U.S.C. §§ 101–102 (1976). 162. See also US Copyright Office, Compendium of US Copyright Office Practices § 306 (3rd ed., 2017). 177

Artificial intelligence and law  397 to designate the legal person as the right holder.183 This leads the discussion to the aforementioned question on the personhood of AI applications and emerging divergent national answers to this question.184 Therefore, considering the existing three national approaches of AI powers in the field of trade-related intellectual property rights in mind and the leading roles of the US, EU and UK in the AI domain, this contribution argues that with the emergence and rise of AI, the big picture of global trade law in the area of TRIPS will become more fragmented, perplexing and diversified.

6. CONCLUSION Believe it or not, that age of AI has already arrived. According to AI experts, what we already have in AI is a whole spectrum of abilities, from programs that are smart but not as smart as humans, to super-clever programs in specific areas that outperform humans, even human champions.185 Take playing Go, one of the specific areas, as an example. The remarkable duel between AlphaGo and Ke Jie in 2017 attracted worldwide attention and became a showcase of the capability of existing AI systems. Google’s AI – AlphaGo – runs on deep learning, a ground-breaking approach to AI that has turbocharged the cognitive capabilities of machines. This contribution finds that AI lacks a universally agreed definition. The EU and the US have been the first movers, passing some soft and/or hard laws. At the same time, some other AI leading countries, such as China and the UK, are reluctant to crystallise the definition of AI and have only passed AI-related legislation instead of AI-specific legislation. Nevertheless, one can extract some commonalities from these emerging regulatory approaches adopted by the major AI countries. First of all, most policy- and rule-makers share a view that the concept of AI is a generic one that encompasses a wide range of technologies – algorithms, big data, expert systems, machine learning, deep learning, robotics, and so forth.186 Many regulators/legislators also agree that AI is not limited to a specific form: it can be run through software (such as a virtual assistant) or embodied in hardware devices (such as a robotic vacuum), or a mixture of both.187 Second, while the term ‘intelligence’ remains a context-specific one, it has become increasingly common for many rule-makers to join industries and academics by classifying AI as ‘narrow/weak AI’ and ‘general/strong AI’.188 Narrow AI is capable of performing specific tasks.189 By contrast, general AI refers to that which ‘exhibits apparently intelligent behaviour at least as advanced as a person across the full range of cognitive tasks’;190 it is essentially

Directive 2009/24/EC, of the European Parliament and of the Council of 23 April 2009 on the Legal Protection of Computer Programs, 2009 O.J. (L 111/16). 184 See more in Liu and Lin (n 151) pp.433–5. 185 Lee and Wang (n 3). 186 See, e.g., UK House of Lords Select Committee on Artificial Intelligence (n 121) at 14–15; Communication on AI for Europe, at 8–10. 187 Communication on AI for Europe, at 1. 188 See, e.g., UK House of Lords Select Committee on Artificial Intelligence (n 121) at 15; US Preparing for the Future of AI (n 47), at 7. 189 ibid at 15. 190 US Preparing for the Future of AI (n 47), at 7. 183

398  Research handbook on digital trade ‘intellectually indistinguishable from a human being’.191 According to the two definitions above, both AI experts and policy-makers agree that we are not yet at the stage of general AI so far; all existing AI – including those that outperform their human counterparts, even human champions, in many areas, such as playing Go, disease diagnosis, car driving, and drone aircraft – still fall within the scope of narrow AI.192 Still, AI technology is fast developing and will continue to mature and become general AI. This is why much of the aforementioned legislation/regulations nicely avoid static definitions of AI in legislation and regulations to make the law future-proof. Policy-makers, at both international and national levels, can utilise these commonalities to create a common ground for initial talks on managing the challenges in the field of AI. When designing a new legal framework, one should also bear in mind the fast-evolving landscape of technology and the inequality between countries, to capture this trendy term by allowing flexibility to accommodate different approaches to defining AI. Based on examining definitions of AI, this contribution concludes that the term AI is broadly understood to cover both narrow AI and general AI, which follows the approach of the AI experts and the US and EU legislators/regulators. The different national regulatory approaches further reinforce the divergent landscape for global trade governance. A handful of AI powers like the EU and US have passed specific law governing AI. Some countries have chosen to pass AI-related law by regulating data which play a critical role in AI technology. Due to the significant role of data in AI systems, these laws also affect AI. For example, the EU (as well as the UK) introduced its data regulation, the GDPR, in 2016; China passed the Personal Information Protection Law of the People’s Republic of China in 2021.193 Based on comparative analysis of emerging divergent regulatory responses to the rise of AI technologies, this contribution has examined the gaps and legal uncertainty in existing global trade law. It argues that, in addition to challenges facing policy-makers at national levels, policy-makers also face the pacing problem at the international level. This contribution evaluates the insufficiency of global trade law in the AI age through examining the GATT–GATS legal framework and TRIPS, and it is shown that all of the three multilateral agreements of the WTO have problems in keeping pace with the fast-evolving AI technologies.

UK House of Lords Select Committee on Artificial Intelligence (n 121) at 15. ibid at 15; Lee and Wang (n 3). 193 See the comparision between the EU and Chinese Data Protection Laws in Osborne Clark, The PRC Personal Information Protection Law, China’s GDPR – in a nutshell (6 September 2021) www​ .osborneclarke​.com/​insights/​prc​-personal​-information​-protection​-law​-chinas​-gdpr​-nutshell. 191 192

Artificial intelligence and law  399

APPENDIX Table 22A.1

The European Commission’s regulatory framework proposal on artificial intelligence

Risk levels associated with AI systems

Effects

Unacceptable risk

Such AI systems are prohibited, since they Examples of the prohibited AI systems

Examples

are deemed to be against EU fundamental include: ● Exploitative or manipulative practices, such

rights and values.

as ‘practices that have a significant potential to manipulate persons through subliminal techniques’. ● AI-based social scoring carried out by High risk

public authorities. Such AI systems are allowed only ● The identification of high-risk AI will be if they comply with certain mandatory

closely linked to their intended purpose

requirements

data

and includes systems used in the following

governance, documentation and record

sectors: critical infrastructure, educational

keeping, transparency and provision of

training, hiring services, migration and

information to users, human oversight,

border control tools, justice administration

comprising:

robustness, accuracy and security, and ex-ante conformity assessments.

and law enforcement. ● Within this high-risk AI, the Commission has included real-time biometric systems (e.g. facial recognition), which will be prohibited unless considered strictly necessary to search for a missing child, to prevent a specific and imminent terrorist threat or to locate the suspect of a serious criminal offence.

Limited risks

AI systems to which only specific Chatbots. transparent obligations will apply, such as making citizens aware that on the other side there is a machine (and not a human) interacting with them.

Minimal or no risks

This last group comprises AI systems that are considered not to constitute a risk or pose a threat to citizens’ fundamental rights and to which no specific obligation will be applied.

Note: Table compiled by the current author. Source: Commission, Regulatory framework proposal on artificial intelligence. See more: https://​eur​-lex​.europa​.eu/​ legal​-content/​EN/​TXT/​HTML/​?uri​=​CELEX:​52021PC0206​&​from​=​EN; see also https://​digital​-strategy​.ec​.europa​.eu/​ en/​policies/​regulatory​-framework​-ai; https://​digital​-strategy​.ec​.europa​.eu/​en/​policies/​european​-approach​-artificial​ -intelligence; www​.lawfareblog​.com/​artificial​-intelligence​-act​-what​-european​-approach​-ai.

23. The role of artificial intelligence in international commercial arbitration in the post-Covid era Muhammad Hassan Idrees and Annabelle Onyefulu

1. INTRODUCTION There is no universally accepted definition of the term AI; this term has been defined by several authors, scholars, scientists and mathematicians, among other thinkers. John McCarthy defined the term AI as ‘the science of creating, developing and use of intelligent machines, software, data processing machines amongst others to analyse macro data using algorithms’.1 Ryan Calo defined it as ‘a process which creates a machine that consists of a well-designed and analytic approximation of human and animal cognition’.2 The use of AI-based technologies in the twenty-first century, especially in our everyday lives, cannot be overemphasized. With the emergence of the global pandemic in 2019, technological advancements have sped up as business transactions, legal cases, arbitral proceedings, academic classes and even yoga lessons can all be done virtually. Developments in science and technology during the pandemic significantly affected international dispute resolution resulting in foreseeable transformations in international arbitration. The possibility of technological inclusion in international arbitration has puzzled scholars for quite some time. It has been argued that while some scholars have embraced the idea that parties, counsels and arbitrators are in dire need of this inclusion, others are vehemently opposed to this.3 However, the fact remains that humans are constantly in dire need of ways to make their lives somewhat easier; hence, the constant need to develop and advance in technology. The foreseeable transformations proposed by artificial intelligence include but are not limited to, the application of algorithms in the arbitral process and the replacement of human arbitrators with AI-based arbitrators. International arbitration is an essential component of international trade, enabling the relatively quick resolution of disputes between companies, consumers and states without recourse to domestic courts, which may be more expensive and time-consuming. The replacement of human adjudicators by AI adjudicators is already in existence in some countries.4 Some jurists have contended that this is indeed a welcome innovation as AI arbi-

1 John McCarthy, ‘What Is Artificial Intelligence’ (2007), see http://​jmc​.stanford​.edu/​articles/​ whatisai/​whatisai.pdf accessed 14 February 2022. 2 Ryan Calo, ‘Artificial Intelligence Policy: A Primer and Roadmap’ (2017), see https://​ssrn​.com/​ abstract​=​3015350 accessed 14 February 2022. 3 H. Eidenmüller, F. Varesis, ‘What Is an Arbitration? Artificial Intelligence and the Vanishing Human Arbitrator’ (2020) see https://​ssrn​.com/​abstract​=3​ 629145 accessed 14 February 2022. 4 For example, Estonia has developed an AI judge which is capable of adjudicating small claims cases with a monetary jurisdiction of about 7,000. Also, China has digitalized courts and AI judges as

400

The role of AI in international commercial arbitration in the post-Covid era  401 trators can produce expert decisions in a human-readable form that will be transparent and devoid of bias.5 Some other jurists are against this as they contend that AI judges will lack the most important element in their decision, which is judicial persuability.6 This will however be extensively discussed in this chapter.

2.

PRACTICABILITY OR OTHERWISE OF REPLACING HUMAN ARBITRATORS WITH AI ARBITRATORS

AI has led to proposals for software or programs set to be adjudicatory in nature, therefore threatening the jobs of arbitrators.7 The possibility of AI arbitrators replacing human arbitrators has been a topic of contention by several scholars over the past few years.8 Professor Eugene Vodoh, a renowned professor of law, posed the rhetorical question: ‘if Al advances to the point where it can adequately mimic judicial opinion writing, should we accept an Al judge?’9 This has raised contention not just in the legal community but among those involved with other forms of dispute resolution mechanisms, such as arbitration. Some people welcome this change with open arms, while others are vehemently opposed. This debate will be explored in this chapter.10 Professor Vodoh expressed interest in welcoming this change in the legal community. He expressly stated that ‘If a system proposes change and it is seen by members of that system as a form of development to that system, It should be ACCEPTED rather than resisted’.11 As previously stated, some scholars have however criticized this opinion as they continue to contend that a judgment is said to be reasoned if there is an epitome of judicial persuadability which is derived from factual legal arguments from parties.12 They contend that an AI judge or arbitrator lacks human feelings and cannot be persuaded, and even if the software/program is designed otherwise, there would be a huge problem in deciphering the factors that would persuade an AI judge/arbitrator.13 This, therefore, means that one of the elements in determining head adjudicators. See Eric Niiler, ‘Can AI Be a Fair Judge in Court? Estonia Thinks So’, Wired Mag. (25 Mar. 2019). See www​.wired​.com/​story/​can​-ai​-be​-fair​-judge​-court​-estonia​-thinks​-so/​ accessed on 26 April 2022. See also Beijing Internet Court Launches AI judge, China Daily (2019), see www​.chinadaily​ .com​.cn/​a/​201906/​28/​WS5d15​6cada3103d​bf1432ac74​.html accessed on 26th April 2022. 5 Richard E. Susskind, ‘Expert Systems in Law: A Jurisprudential Approach to Artificial Intelligence and Legal Reasoning’ (1986) 49 Modern L. Rev. 168, 173. 6 Andrew C. Michaels, ‘Artificial Intelligence, Legal Change, and Separation of Powers’ (2020) 88 U Cin L Rev 1083. 7 Christine Sim, ‘Will Arbitration take over Arbitration?’ (2018) 13(1) Asian Journal of Intl Arbitration 1–14. 8 Young-Yik Rhim and Kyung Bae Park, ‘The Applicability of Artificial Intelligence in International Law’ (2019) 12 J. East Asia & Int’l L. 1, see http://​dx​.doi​.org/​10​.14330/​jeail​.2019​.12​.1​.01 accessed 14 February 2021. 9 Eugene Volokh, ‘Chief Justice Robots’ (2019) 68 Duke L.J. 1135, 1138. 10 Ibid. 11 Ibid. 12 Andrew C. Michaels, ‘Artificial Intelligence, Legal Change, and Separation of Powers’ (2020) 88 U Cin L Rev 1083. 13 Emily Berman, ‘Individualized Suspicion in the Age of Big Data’ (2020) 105 Iowa L. Rev. 463, 502.

402  Research handbook on digital trade who can be an arbitrator is the ability of the person or robot to be judicially persuaded. Thus, this chapter proceeds by asking the pressing question: who can be an arbitrator?

3.

WHO CAN BE AN ARBITRATOR?

Upon the commencement of a dispute in international arbitration, parties or their appointing authority are at liberty to choose an arbitrator or arbitrators subject to the dictates of their arbitral contract.14 The court also has the inherent power to make a such appointment in certain circumstances, for example where a party to a dispute is a state entity15 or in the event that parties or their appointing authority fail to appoint an arbitrator.16 Parties can choose to appoint a natural person or an arbitral institution to precede the dispute.17 Several procedural laws, conventions and scholars have expressly or by implication provided that arbitrators must be natural persons. Article 1450 of the Code of Civil Procedure of France expressly defines an arbitrator as a natural person having the full legal capacity to render adjudicatory powers on an arbitral dispute who is duly appointed by parties, their appointing authority or the court.18 By virtue of Article 3 Scottish Rule of Arbitration, an arbitrator is defined as a sole arbitrator who must be an individual or person.19 The parties to the dispute or their appointing authority are at liberty to appoint an arbitrator if expressly stated in their agreement.20 However, it is important to note that whomever they select as an arbitrator must pass the eligibility test, which includes necessary qualifications, impartiality and independence.21 Article 11 of the UNCITRAL Model Law impliedly provides for arbitrators to be natural persons. It provides that a person shall not be stopped from representing parties as an arbitrator by his/her nationality.22 It is in doubt whether an intelligent machine can possibly have a nationality. However, according to law, there are only two types of legal entities in law with nationalities. They are natural persons and juridical persons.23 Nonetheless, parties or their appointing authority have the authority to appoint arbitrators subject to the contract. The parties can state the procedure for appointing arbitrators and the number of arbitrators to be appointed. It is important to note that if this procedure is not adhered to, a party may request that the court steps in.24 It is worthy of note that the requirement for appointment of arbitrators under the UNCITRAL Model Law remains: qualifications, impartiality and independence.25 Article 11(2) UNCITRAL Model Law. Article 13(4) ICC Rules. 16 Section 18 Arbitration Act 1996. 17 Article 11 UNCITRAL Model Law. 18 Article 1450 Code of Civil Procedure of France. 19 Article 3.1 Scottish Short form Arbitration Rules 2012, Section 8 Arbitration (Scotland) Act 2010. 20 Ibid. 21 IBA Guidelines on Conflicts of Interest in International Arbitration, Part I (2014) n 71, Article 3.8 Scottish Short form Arbitration Rules 2012. 22 Article 11 UNCITRAL Model Law. 23 Elvia Arcelia Quintana Adriano, ‘The Natural Person, Legal Entity or Juridical Person and Juridical Personality’ (2015) 4 PSJL 5. See https://​elibrary​.law​.psu​.edu/​cgi/​viewcontent​.cgi​?article​=​ 1117​&​context​=​jlia accessed on 21 March 2022. 24 Article 11(4) UNCITRAL Model Law. 25 Article 12 UNCITRAL Model Law. 14 15

The role of AI in international commercial arbitration in the post-Covid era  403 Furthermore, Article 13 of the ICC rules also impliedly provides for natural persons as proposed arbitrators, because the nationality and residence of a proposed arbitrator act as criteria for the appointment of an arbitrator.26 The court under these rules, in appointing a sole arbitrator, must appoint an arbitrator of a different nationality and/or residence from the parties unless the parties expressly agree otherwise.27 The ICC rules also state the basic requirements to be qualifications, impartiality and independence.28 Article 6 of the LCIA rules also refers to the nationality of arbitrators as criteria for appointment as the presiding arbitrator29 or emergency arbitrator.30 This inherently means that an arbitrator is a person in law capable of having a nationality and who is eligible, that is, qualified according to the dictates of parties in the agreement, impartial, and independent. An intelligent machine does not appear to possess all of these qualities. Nevertheless, it is worth mentioning that in all the arbitration laws mentioned above, not even one expressly disqualifies an AI-based tool acting as an arbitrator.31 There is also no provision for the unenforceability of an award rendered by an AI arbitrator under the New York Convention.32 The point has, however, been argued by several jurists that the convention was ratified in 1958 and as of then, the draftsmen could not possibly envisage the possibility of the use of AI arbitrators.33 This is true; however, the position remains that by virtue of Article II(2) of the New York Convention, a court can enforce an award rendered by a corporate or legal entity.34 If AI tools are ratified as a legal entity capable of rendering an award in the future, then their awards will fall under this subsection for enforcement.35

4.

PRACTICABLE GROUNDS FOR CHALLENGING THE ENFORCEABILITY OF AN ARBITRAL AWARD RENDERED BY AI ARBITRATOR

4.1 Attestation Most jurists have contended that a possible ground for the challenge of the enforcement of an arbitral award rendered by an AI arbitrator is the attestation of an award.36 They contend Article 13 ICC Rules. Article 13(5) and (6) ICC Rules. 28 Article 14 ICC Rules. 29 Article 6 LCIA Rules. 30 Article 9B LCIA Rules. 31 Christine Sim, ‘Will Artificial Intelligence take over Arbitration?’ (2018) n 8, p.3. 32 Ibid. 33 Ibid; Gülüm Bayraktaroğlu-Özçelik and Ş. Barış Özçelik, ‘Use of AI-Based Technologies in International Commercial Arbitration’ (2021) p.9 see https://​ejlt​.org/​index​.php/​ejlt/​article/​view/​814/​ 1033 accessed on 14 February 2021. 34 Article II(2) New York Convention provides: ‘The term “arbitral awards” shall include not only awards made by arbitrators appointed for each case but also those made by permanent arbitral bodies to which the parties have submitted.’ 35 Christine Sim, ‘Will Artificial Intelligence take over Arbitration?’ (2018) n 8, p.3. 36 Gülüm Bayraktaroğlu-Özçelik and Ş. Barış Özçelik, ‘Use of AI-Based Technologies in International Commercial Arbitration’ (2021) n 34, p.9. 26 27

404  Research handbook on digital trade that much national and international legislation has stated that one of the elements as to the validity of an arbitral award is the attestation or signature of the arbitrator(s).37 Article IV(1) (a) of the New York Convention expressly states that an arbitral award shall be recognized and enforced by the court if a party seeking its enforcement submits a duly authenticated award to the court or a certified copy thereof.38 Article 31 of the UNCITRAL Model Law provides that an arbitral award shall be in writing and signed by the arbitrators.39 Thus, since an AI tool cannot produce a signature, it can be deciphered that an AI cannot sufficiently render a final and binding arbitral award.40 Some other jurists are of the opinion that having human arbitrators alongside an AI arbitrator may solve the problem of admissibility of the arbitral award and issues resulting therefrom, such as a signature.41 Adopting this system will, however, pose a lot of unanswered questions, such as: who becomes the umpire if there are two human arbitrators and an AI arbitrator?42 Where there are conflicts in the decisions/opinions of arbitrators, how will the opinion of the AI arbitrator be explained?43 If two AI arbitrators and one human arbitrator are appointed, will the authentication of only one arbitrator suffice?44 And so on. The creation of a distinctive mark or stamp for AI arbitrators may solve the issue of signatures.45 Consequently, in the event that a human arbitrator is paired up with an AI arbitrator, a signature might not necessarily be an issue. 4.2

Selection of Arbitrators in Accordance with the Arbitral Agreement

Awards rendered by an AI arbitrator can also be challenged on grounds stated in Article 34(2) (a)(IV) of the UNCITRAL Model Law and Article V(1)(d) of the New York Convention which provides that an award can be challenged if the arbitral tribunal is not composed of arbitrators agreed to by parties or is in huge contravention of the law of the seat of arbitration. Parties involved in an arbitral proceeding have autonomy to select their arbitrators46 as well as determining how the proceedings will be conducted.47 This autonomy is, however, subject to their arbitral agreement. When the arbitral agreement states that natural persons should be appointed as arbitrators and there is a subsequent appointment of an AI Arbitrator, an unsatisfied party can raise that as a ground for challenging the award under Article 34 of the UNCITRAL Model Law and Article V of the New York Convention. However, in the case where parties to an arbitral proceeding expressly agree to the appointment of AI arbitrators and the law of the seat of arbitration (lexarbitri) only permits the appointment of a natural person, Ibid. Article IV(1)(a) New York Convention 1958. 39 Article 31 UNCITRAL Model Law. 40 Gülüm Bayraktaroğlu-Özçelik and Ş. Barış Özçelik, ‘Use of AI-Based Technologies in International Commercial Arbitration’ (2021) n 34, p.9. 41 Epaminontas E. Triantafilou, ‘Non-Lawyer Arbitrators and the Deliberative Balance of International Arbitral Tribunals’ (2016) TDM 7. 42 Christine Sim, ‘Will Artificial Intelligence take over Arbitration?’ (2018) n 8, p.4. 43 Ibid. 44 Ibid. 45 Ng (Huang Ying) and Benedetti del Rio, ‘When the Tribunal Is an Algorithm: Complexities of Enforcing Orders Determined by a Software under the New York Convention’ (2019) 121 Kluwer 34. 46 Article 11 UNCITRAL Model Law. 47 Article 19 UNCITRAL Model Law. 37 38

The role of AI in international commercial arbitration in the post-Covid era  405 the law of the seat shall prevail by virtue of Article 34(2)(a)(IV) UNCITRAL Model Law.48 That is, an award rendered by an AI arbitrator duly appointed by parties can be challenged if such an appointment is contrary to the lexarbitri. It is, however, very important to note that for a party to successfully raise that as a ground for challenging the arbitral award, the party must have successfully raised that issue at the early stages of the proceedings.49 4.3

Contravention of the Law of the Seat of Arbitration

An award rendered by an AI arbitrator can be challenged under Article 34 of UNCITRAL Model Law and Article V of the New York Convention on the ground that the composition of the Arbitral Tribunal is in huge contravention of the law of the seat of arbitration, as discussed previously. Most national legislation does not explicitly bar the use of AI arbitrators; nor does it permit the use of AI-based arbitrators.50 Some jurists have contended that since this lacuna exists, it may be resolved in favour of AI-based arbitrators.51 This lacuna is found in most national and international legislation, as mentioned earlier. There is nothing in the UNCITRAL Model Law barring the use of AI arbitrators; neither is there anything under the New York Convention that bars the enforcement and recognition of awards rendered by such AI arbitrators. Nevertheless, it is important to note that some legislation is now being enacted that expressly bars the use of AI-based arbitrators. For example, under Article 1450 of the Civil Code of Conduct of France only natural persons can be appointed as arbitrators; Article 7 of the Turkish International Arbitration Law also stipulates that only natural persons can be appointed as arbitrators. Some other legislation has stated that ‘individuals or persons who have rights and capacity should be appointed as arbitrators’.52 The interpretation of this section in the future might become a herculean task as the European Parliament has passed a resolution that addresses robots as electronic people, so they may fall into this category of persons.53

5.

THE PUBLIC POLICY REQUIREMENT

Public policy has no specific definition. It has been described by a scholar as principles of morality and justice that citizens of a nation consider to be of national interest and it varies from state to state.54 Public policy is a concept that may change from time to time in a soci-

Article 34(2)(a)(IV) UNCITRAL Model Law. Gülüm Bayraktaroğlu-Özçelik and Ş. Barış Özçelik, ‘Use of AI-Based Technologies in International Commercial Arbitration’ (2021) n 34, p.11. 50 Nairobi Centre for International Arbitration, ‘Artificial Intelligence (AI) in International Arbitration: Machine Learning’ (2021). See https://​ncia​.or​.ke/​wp​-content/​uploads/​2021/​08/​ARTIFICIAL​ -INTELLIGENCE​-AI​-IN​-INTERNATIONAL​-ARBITRATION​.pdf accessed on 22 March 2022. 51 Ibid. 52 Article 3.1 Scottish Short form Arbitration Rules 2012, Section 8 Arbitration (Scotland) Act 2010. 53 European Parliament Resolution of 16th February 2017 with recommendation to the commission on civil law rules on robotics (2015/2103(INL)). 54 Andrew I. Okekeifere, ‘Public Policy and Arbitrability under the UNCITRAL Model Law’ (1999) 2 Int. A.L.R 70–7. 48 49

406  Research handbook on digital trade ety.55 Article 34(2)(b)(ii) of the UNCITRAL Model Law56 and Article V(2)(b) of the New York Convention57 provide that a court may set aside an award if it is contrary to public policy. While what can violate public policy is not expressly spelled out in the UNCITRAL Model Law or the New York Convention, unlike other public international regulations,58 the construction or interpretation of manifest contravention of public policy can be deciphered through case-by-case analysis.59 It has been contended that an arbitral award rendered by an AI arbitrator may contravene several public policy grounds.60 For example, where the national legislation of the seat of arbitration expressly provides that an arbitrator must be a natural person, any award rendered by an AI arbitrator is in huge contravention of the public policy of the national legislation.61 In addition, the recognition and enforcement of an award rendered by an AI arbitrator which is contrary to the lexarbitri will also amount to a contravention of public policy under Article V(2)(d) New York Convention.62 Furthermore, an award can be set aside and recognition and enforcement denied if a party satisfies the court that the arbitrators were not independent or impartial in their dealings.63 This can serve as a public policy ground.64 It has been contended by several jurists that the use of AI arbitrators in rendering decisions in arbitral proceedings will be flawed with data bias.65 AI arbitrators can only function when the algorithms have been fed enough data by humans.66 This data can be discriminatory as it may contain elements likely to affect race and gender.67 Another potential ground for the challenge of an award or denial of recognition and enforcement of an award on the ground of public policy may be linked to the absence of a reasoned decision in an arbitral award.68 Under Article 32(2) of the UNCITRAL Model Law, an award shall be valid if it contains a reasoned decision of an arbitral tribunal, except if the parties Ibid. Article 34(2)(b)(ii) UNCITRAL Model Law. 57 Article V (2)(b) New York Convention. 58 Regulation (EC) No 593/2008 of the European Parliament and of the Council of 17 June 2008 (‘Rome I’), Art. 21; Regulation (EU) No 1215/2012 of the European Parliament and of the Council of 12 December 2012 on jurisdiction and the recognition and enforcement of judgments in civil and commercial matters (recast), Art. 45(1)(a), Art. 46; Swiss Federal Act on Private International Law, Art. 27(1); Turkish Private International Law Act among others. 59 Gülüm Bayraktaroğlu-Özçelik and Ş. Barış Özçelik, ‘Use of AI-Based Technologies in International Commercial Arbitration’ (2021) n 34, p.12. 60 Ibid. 61 Ibid. 62 H. Eidenmüller, F. Varesis, ‘What Is an Arbitration? Artificial Intelligence and the Vanishing Human Arbitrator’ n 4. 63 IBA Guidelines on Conflicts of Interest in International Arbitration, Part I (2014) n 71, Article 3.8 Scottish Short form Arbitration Rules 2012. 64 M.V. Benedetelli, ‘Human Rights as a Litigation Tool in International Arbitration: Reflecting on the ECHR Experience’ (2015) 4 Arbitration International 59. 65 J.J. Angwin et al., ‘Machine Bias’ (2016) available at www​.propublica​.org/​article/​machinebias​Risk-assessments-in-criminal-sentencing accessed on 22 March 2022. See also A. Caliskan et al., ‘Semantics Derived Automatically from Language Corpora Contain Human-Like Biases’ (2017) 356 Sci. 183–6. See also Gülüm Bayraktaroğlu-Özçelik and Ş. Barış Özçelik, ‘Use of AI-Based Technologies in International Commercial Arbitration’, n 34, p.13. 66 Ibid. 67 Ibid. 68 M. Scherer, ‘International Arbitration – How Artificial Intelligence Will Change Dispute Resolution’ [2019] Austrian Yearbook of International Arbitration 514. 55 56

The role of AI in international commercial arbitration in the post-Covid era  407 expressly agree that the reasoning should not be stated or the award is based on the agreement entered into by parties.69 This, therefore, means that in countries where the national legislation provides for the reasoning of an award as a strict requirement for the validity of an arbitral award, the absence of such will amount to a violation of public policy.70 Algorithms lack feelings and empathy, and certainly cannot make use of their discretion,71 all of which are components of a reasoned award. Justice is much more than processed macro data and a set of precedent decisions. It is not a well-constructed algorithm;72 it is a human virtue that is characterized by a range of factors such as evaluating and analysing key facts and legal arguments, application of the law of the state and the use of discretion to arrive at what the tribunal in any case feels as right and just to all parties involved.73 Nevertheless, awards rendered by AI arbitrators might be welcomed by parties and not challenged on grounds of public policy, as certain advantages might be associated with an AI arbitrator acting as an adjudicatory authority. Such advantages may include the duration of proceedings ceasing to be worrisome, as proceedings will be determined in a matter of months as opposed to years in the current human arbitrator regime.74 Some scholars have widely embraced this concept and stated that the dissatisfaction and grievances occasioned by the resort in Arbitration as a medium of dispute settlement by parties which is associated with time, cost and the nonchalant behavior of members of the arbitral community will be greatly reduced by the introduction of technology to this system and making such technology affordable to parties.75

Algorithms may also be developed which are capable of learning and applying precedent decisions in a matter of minutes, thereby rendering an even better decision than human arbitrators.76 However, it is worth noting that several jurists have contended that algorithms are generally inaccurate and the software is secret, so there is no means of reviewing decisions.77 This will deter the strict adherence to decisions rendered by AI arbitrators.78

Article 32(2) UNCITRAL model law. M. Scherer, ‘International Arbitration – How Artificial Intelligence Will Change Dispute Resolution’ [2019], n 69. 71 Christine Sim, ‘Will Artificial Intelligence take over Arbitration?’ (2018) n 8. 72 Nairobi Centre for International Arbitration, ‘Artificial Intelligence (AI) in International Arbitration: Machine Learning’ (2021) n 51. 73 Ibid. 74 Nairobi Centre for International Arbitration, ‘Artificial Intelligence (AI) in International Arbitration: Machine Learning’ (2021) n 51. 75 Paul Cohen and Sophie Nappert, ‘The March of the Robots’ (2015) see http://​glo​balarbitra​ tionreview​.com/​article/​1080951/​the​-march​-of​-the​-robots accessed on 22 March 2022. See also Jack Wright Nelson, ‘Machine Arbitration and Machine Arbitrators’ (2016). See www​.youngicca​-blog​.com/​ machinearbitration​-and​-machine​-arbitrators/​accessed on 22 March 2022. 76 Nairobi Centre for International Arbitration, ‘Artificial Intelligence (AI) in International Arbitration: Machine Learning’ (2021) n 51. 77 Christine Sim, ‘Will Artificial Intelligence take over Arbitration?’ (2018) n 8. 78 Ibid. 69 70

408  Research handbook on digital trade

6.

IMPLICATION OF THE REPLACEMENT OF HUMAN ARBITRATORS WITH AI ARBITRATORS IN THE FUTURE OF ARBITRATION

The implications of AI-based technologies for the future of international arbitration cannot be gainsaid, as they promise to provide solutions to inherent issues in international arbitration such as duration of proceedings, cost, lack of legal consistency and speed, among others, thereby favouring parties, their counsels and arbitrators.79 AI-based tools are also being designed to render predictive decisions, as parties would prefer to know the likely outcome before embarking on an arbitral proceeding. This may play a major role in deciding whether parties might decide to negotiate or not. The replacement of human arbitrators with AI arbitrators, however, is a game changer in the arbitral community80 as intelligent machines will take over the adjudicatory process of this mechanism. Although no such algorithm has been developed yet, this certainly will change the system of dispute resolution in the coming years.81 Nevertheless, while this replacement is a welcome innovation, the result of this new system will be unemployment as AI is targeted to take over the jobs of arbitrators.82 AI-based arbitrators will be designed in such a way that little or no human intervention is needed, thereby requiring fewer humans and causing unemployment.83 Moreover, there is also the possibility of rendering an erroneous award tainted by gender and/or racial bias. It has already been established that AI arbitrators can only function when fed with data. There have been incidents in the past where existing algorithms have been reported to make judgements in a discriminatory manner due to the information that was fed to them. Some of those algorithms include the Amazon recruitment database, which was proven to be gender-biased and thereby to employ more men than women.84 Furthermore, in cases where an established pattern exists which involves a particular set of people dominating a branch of arbitration, an AI arbitrator is bound to rule in favour of those set of people in any future dispute without looking into the merits of the case due to the data that was fed to it.85 For example, in international investment arbitration where investors have in most cases won cases over host states, an AI arbitrator may, in any future dispute, award the case to the investor without looking at the merit of the case.86 This is because an algorithm’s

Ibid. Gülüm Bayraktaroğlu-Özçelik and Ş. Barış Özçelik, ‘Use of AI-Based Technologies in International Commercial Arbitration’ n 34. 81 Ibid. 82 Arnav Joshi, ‘Replacing Arbitrators with Artificial Intelligence in International Arbitration’ (2020). See www​.th​earbitrati​onworkshop​.com/​post/​replacing​-arbitrators​-with​-artificial​-intelligence​-in​ -international​-arbitration accessed on 22 March 2022. 83 Anandaday Misshra, ‘Artificial Intelligence (AI) and Its effects on Arbitration’ (2020). See https://​ taxguru​.in/​corporate​-law/​artificial​-intelligence​-ai​-effects​-arbitration​.html accessed on 23 March 2022. 84 Jeffrey Dastin, ‘Amazon Scraps Secret AI Recruiting Tool That Showed Bias against Women’ (2018). See www​.reuters​.com/​article/​us​-amazon​-com​-jobs​-automation​-insight/​amazon​-scraps​-secret​-ai​ -recruiting​-tool​-that​-showed​-bias​-against​-women​-idUSKCN1MK08G accessed on 23 March 2022. 85 Jean R. Sternlight, Creeping Mandatory Arbitration: Is It Just? (2005) 57 Stan. L. Rev. 1631, 1650. 86 Young-Yik Rhim and KyungBae Park, ‘The Applicability of Artificial Intelligence in International Law’ n 9, p.14. 79 80

The role of AI in international commercial arbitration in the post-Covid era  409 output largely depends on its input; this is derived from the concept of garbage in, garbage out.87 Another major legal impact that an AI arbitrator could have on arbitration is the issue of absolute immunity granted to the arbitrator by the statute.88 Under Article 41 of the ICC rules, an arbitrator is granted immunity to render awards and is only limited to decisions made in bad faith.89 This consequently means that if it is established that a human arbitrator has rendered one or several awards in bad faith then he will be liable and may face disciplinary action, which may result in him leaving office. However, whether the arbitrator is a machine arbitrator or an AI arbitrator, the ultimate question will be who would be held responsible: the developer of the algorithm? The person who fed data to the algorithm? Or the machine itself?90 Nonetheless, there seem to be positive impacts of an AI arbitrator replacing a human arbitrator. Scholars have opined that the unpredictability associated with arbitral awards will be less than none as AI tools are being developed to predict cases. Moreover, AI arbitrators will solve the problem of prolonged duration of arbitral proceedings, cost and legal inconsistencies associated with most awards.

7.

IMPLICATIONS OF AI ARBITRATORS IN AFRICA

Arbitration in Africa is not a particularly new concept: more than 33 countries out of 54 countries in Africa are signatories to the New York Convention.91 There has been wide adoption and ratification of the UNCITRAL Model Law among countries that are signatories to the New York Convention and actively practice arbitration.92 Most national legislation that provides for arbitration in Africa neither expressly prohibits nor permits the appointment of AI arbitrators.93 For example, the CAJAC (China–Africa Joint Arbitration Centre), an arbitration centre in Johannesburg, South Africa established by Chinese and African stakeholders, adopts in its rules the definition of an arbitrator according to the UNCITRAL Model Law, thereby neither expressly prohibiting the use of AI arbitrators nor permitting their use.94 However, it is pertinent to note that in 2017, a Uniform Arbitration Act was enacted by the Organisation for Harmonisation of Business Law in Africa (OHADA).95 OHADA was formed by about 17 countries in Africa to enhance the investment climate in Africa. The Uniform

Oscar Tang, ‘Think Arbi: Will Artificial Intelligence Help or Harm Arbitration?’ (2021). See http://​arbitrationblog​.kluwerarbitration​.com/​2021/​12/​22/​think​-arbi​-will​-artificial​-intelligence​-help​-or​ -harm​-arbitration/​accessed on 23 March 2022. 88 Mahnoor Waqar, ‘The Use of AI in International Arbitration’ (2021). See https://​papers​.ssrn​.com/​ sol3/​papers​.cfm​?abstract id=3931233 accessed on 23 March 2022. 89 Article 41 ICC Rules. 90 Mahnoor Waqar, ‘ The Use of AI in International Arbitration’ (2021), n 89. 91 Sadaff Habib, ‘The Use of Artificial Intelligence in Arbitration in Africa – Inevitable or Unachievable?’ (2020) IBA. See www​.ibanet​.org/​article/​E62B06F6​-7772​-458A​-A6E7​-1474DB7136B5 accessed on 23 March 2022. 92 Ibid. 93 Lagos State Arbitration Act 2004. 94 See China–Africa Joint Arbitration Centre at www​.shiac​.org/​CAJAC/​index​_E​.aspx accessed on 23 of March 2022. 95 Ibid. 87

410  Research handbook on digital trade Arbitration Act (UAA) expressly provides that an arbitrator must be a natural person.96 This, therefore, means that if parties in arbitral proceedings appoint AI arbitrators, it will be in huge contravention of the lexarbitri and hence it will not stand.97 Nonetheless, in African countries that are not signatories to UAA but have adopted the UNCITRAL Model Law as their national legislation, the appointment of AI arbitrators may be possible as there is no prohibition of such appointment. It is however pertinent to note that the appointment of AI arbitrators in Africa may encounter certain challenges, such as: a. Lack of information: It has been established that the success of an algorithm depends on the information ‘fed’ to it.98 In Africa, arbitration is a relatively new concept, and this mode of dispute resolution is not widely embraced.99 This is because parties prefer to refer their dispute to traditional court or litigation. The resultant effect, therefore, is that very few awards have been rendered by arbitral tribunals.100 This is also associated with the fact that arbitral proceedings are usually prolonged and expensive, and parties cannot afford them. Moreover, the publication of an award rendered by an arbitral tribunal is usually resisted, following the universal doctrine of confidentiality.101 Efforts are, however, being made in Africa to publish certain aspects of the proceedings. For example, under Article 21(6) of the Arbitration Foundation of South Africa rules, the publication of protocols in remote arbitration hearings is permissible.102 b. Lack of trained AI personnel: Knowledge of AI in Africa is quite thin due to very few AI experts or knowledgeable teachers, poorly funded or nonexistent AI research projects, and poorly equipped institutions, among others.103 Governments have also shown little or no interest in funding such projects as they believe they will not yield revenue.104 Nonetheless, the introduction of this type of technology in Africa will however be of great significance, on the one hand, as it will enhance speed and lead to a shorter duration of arbitral decisions, cost, award prediction and much more.105 On the other hand, it may equally lead to several foreseeable difficulties, such as lack of maintenance of the robots as there is a massive problem with power supply in Africa and a lack of experts in the field to update the software when required.106 The problem of unemployment also persists: it was reported in 2018 by scholars at Oxford University that AI is likely to take over jobs in the future as developers are creating algorithms that have the skills of an arbitrator.107

Article 5 UAA 2017. Article 11 UNCITRAL Model Law. 98 Mahnoor Waqar, ‘The Use of AI in International Arbitration’ (2021), n 89. 99 Ibid. 100 Ibid. 101 Dolling-Baker v Merrett (1990) 2 All ER 890. 102 Article 21(6) Arbitration Foundation of South Africa. 103 N. Butcher, M. Wilson-Strydom and Baijnath, ‘Artificial Intelligence Capacity in Sub-Saharan Africa: Compendium Report’ (2021) AI4D. See https://​africa​.ai4d​.ai/​wp​-content/​uploads/​2021/​03/​AI4D​ -Report​%E2​%80​%94AI​-in​-SSA​.pdf accessed on 23 March 2022. 104 Ibid. 105 Mahnoor Waqar, ‘The Use of AI in International Arbitration’ (2021), n 89. 106 Ibid. 107 Ibid. 96 97

The role of AI in international commercial arbitration in the post-Covid era  411 Furthermore, the issue of accessibility to this innovation may be a herculean task in Africa. Some scholars have contended that the cost of acquiring a high-tech AI arbitrator (robot) might be very expensive and perhaps unaffordable in impoverished countries in Africa.108

8.

ETHICAL CONSIDERATIONS

There are several possible ethical concerns associated with the use of AI in international arbitration as well as the replacement of human arbitrators with AI arbitrators. Some of these concerns include, but are not limited to, the following: 8.1

Fairness of Proceedings

There have been major concerns as to the fairness of proceedings that will be conducted by an AI arbitrator.109 This is mainly associated with the quality of output of an AI tool, which has been established to be dependent on the input. That is, an AI tool can only render an award based on the data that has been fed to it. Some scholars have contended that there is a possibility of data bias, and this may likely affect gender and race.110 This inherently means that the fundamental requirement for the appointment of arbitrators, which is impartiality and independence,111 will be breached. Moreover, parties in international arbitration cannot generally appeal the decision of an arbitrator as such decision is final and binding except in limited circumstances, and only on questions of law and not fact.112 This is a major ethical issue because if an award is not fair to a party based on racial or gender bias, the party cannot appeal the decision. However, the law provides that parties can only challenge the recognition of the award.113 This is still not sufficient as it will lead to several unanswered questions, such as who can be held responsible: the developer of the algorithm, the person who implemented the data input, or the AI arbitrator?114 8.2

Validity of the Award

Another ethical concern to be deciphered from the replacement of human arbitrators with AI arbitrators is the fairness or validity of an award. A primary element of an award is the decision of the arbitral tribunal as well as the reason for the decision.115 An award rendered by an AI arbitrator has been argued to lack a reason for its decision.116 This is associated with the fact

Kathleen Paisley and Edna Sussman, ‘Artificial Intelligence Challenges and Opportunities for International Arbitration’ (2018) New York Dispute Resolution Lawyer 11, no. 1: 35–40. 109 Ibid. 110 J.J. Angwin et al., ‘Machine Bias’ (2016) available at www​.propublica​.org/​article/​machinebias​Risk-assessments-in-criminal-sentencing accessed on 22 March 2022. 111 IBA Guidelines on Conflicts of Interest in International Arbitration, Part I (2014) n 71. 112 Article 35 UNCITRAL Model Law, Section 69 English Arbitration Act 1996. 113 Article 35 UNCITRAL Model Law. 114 Mahnoor Waqar, ‘The Use of AI in International Arbitration’ (2021), n 89. 115 Article 32(2) UNCITRAL Model Law. 116 Christine Sim, ‘Will Artificial Intelligence take over Arbitration?’ (2018) n 8. 108

412  Research handbook on digital trade that AI lacks emotions and therefore cannot provide reasoning for its decisions. In the case that a party is not satisfied with the decision of the AI arbitrator and wants to challenge the award, this will lead to a major issue as the court will not be able to decipher the reasoning behind any decision made by the AI tool. 8.3 Confidentiality One of the major attractions of arbitration as a mode of alternative dispute settlement is the privacy and confidentiality associated with the system.117 In the event of the development and subsequent use of an AI arbitrator, there might be demands for the publication of arbitral awards, thereby defeating the parties’ interest. 8.4 Security With the rise of cybercrime, ethical concerns may arise regarding the security measures put in place to protect AI-based tools from atrocities such as hacking, spam and malicious viruses.118 If there is a failure to set up security measures, AI arbitrators can be manipulated into making biased decisions. It is, however, worth noting that the European Commission set up a group in 2018, referred to as the European Commission’s High-Level Expert Group on Artificial Intelligence,119 whose primary responsibility is to evaluate prevalent ethical issues of the use of AI and set guidelines to regulate these issues. These guidelines stipulate policies to ensure that AI tools are human-centric and in compliance with the Charter of Fundamental Rights of the European Union.120 This guideline focuses on the principle of justice as its main point, thereby ensuring that judgments are free from bias or any form of stigmatization.121

9. RECOMMENDATIONS 9.1

Amendments to Legal Instruments

The amendment of the New York Convention and the re-enactment of national and international instruments expressly enabling the use of AI-based technologies and appointment of AI arbitrators and recognizing awards rendered therein will be one of the first steps in accepting this new system. It should be noted that discussions are currently going on as to the amendment

Dolling-Baker v Merrett (1990) 2 All ER 890. James Kwan, James Ng and Brigitte Kiu, ‘The Use of Artificial Intelligence in International Arbitration: Where Are We Right Now?’(2019) 22 Int. A.L.R 19–26. 119 The European Commission’s High Level Expert Group on Artificial Intelligence ‘Draft Ethics Guidelines for Trustworthy AI: Working Document for Stakeholders’ Consultation’ (2018) See https://​ ec​.europa​.eu/​digital​-single​-market/​en/​news/​have​-your​-say​-european​-expert​-group​-seeks​-feedback​-draft​ -ethics​-guidelines​-trustworthy accessed 25 March 2022. 120 Paragraph 4 of the Draft Ethics Guidelines for Trustworthy AI: Working Document for Stakeholders’ Consultation. 121 Ibid. 117 118

The role of AI in international commercial arbitration in the post-Covid era  413 of the New York Convention.122 This amendment will ensure that most public policy grounds for challenging an award or enforcing an award will be silenced as the appointment of an AI arbitrator will not contravene the dictates of any legal instrument. Moreover, the major attraction of arbitration is the autonomy and flexibility of parties to be able to control the system. Mandatory aspects of the legislation which may hinder this flexibility should be abolished. 9.2

Provision of Sufficient Data

Arbitral institutions should start publishing awards or making them accessible when demanded in an anonymized form.123 This will assist in curbing erroneous decisions when this new system is in place. This is because the AI arbitrator will be furnished with a lot of precedent decisions to apply to cases. 9.3

Funding of AI Research Projects in Africa

Adequate funding and deployment of trained personnel to assist in the implementation of this new system in developing countries, especially in Africa, will yield positive results as people will be trained on how to make use of this system. 9.4

Easy Accessibility of These Technological Tools by Less Developed Countries

The less developed countries or countries that cannot easily afford to make use of these technological tools should be assisted in one way or the other. There should be uniform access to these tools by every country that has adopted and is practising arbitration, as failure to do so will lead to the ground for challenging the transparency and equality of this process. 9.5

Feature Selection

There should be a kind of manual or regulation guiding the kind of data that will be ‘fed’ to this software ensuring that it is devoid of discrimination or bias and it can always be updated with the latest precedents and laws.

10. CONCLUSION The entire discourse has demonstrated and critically analysed the consequence of the use of AI-based technologies in international arbitration and the possibility of replacing human arbitrators with AI arbitrators. This chapter analysed the practicability or otherwise of the replacement of human arbitrators with AI arbitrators. Although an algorithm for this has not yet been developed, the chapter

Tirado, Acevedo and Cosio, ‘Draft Convention on the International Enforcement of Arbitration Agreements and Awards’ (2019) 310. 123 ICC, ‘Note to Parties and Arbitral Tribunals on the Conduct of the Arbitration under the ICC Rules of Arbitration’ (2019). See https://​iccwbo​.org/​content/​uploads/​sites/​3/​2017/​03/​icc​-note​-to​-parties​-and​ -arbitral​-tribunals​-on​-the​-conduct​-of​-arbitration​.pdf accessed on 25 March 2022. 122

414  Research handbook on digital trade analysed several national laws on arbitration as well as the UNCITRAL Model Law as to the possibility of this novel innovation to the arbitral community. From this analysis, it can be deciphered that no national legislation expressly bars the appointment of AI arbitrators. Such national legislation, as well as the UNCITRAL Model Law, expressly or by implication states that natural persons be appointed as arbitrators. Although it can be contested on the general principle of law contract, which states that silence does not constitute acceptance, by the authority of McGlove v Lacey,124 it is a known principle of law that what is not expressly admitted is deemed not to be denied and facts not denied are deemed to have been admitted. This has been a subject of contention, especially in relation to the New York Convention wherein Articles II and V do not expressly deny the recognition and enforcement of awards rendered by AI arbitrators. Some scholars have ruled against this fact, stating that the appointment of AI arbitrators was never in contemplation in 1958 when the convention was ratified. However, with the emergence of the Covid-19 pandemic, artificial intelligence gained more popularity and the use of AI technologies in arbitration now seems more plausible. It is a welcome innovation as it has several benefits, such as enhancing speed, shortening the duration of proceedings, reducing costs and generally promoting the efficiency of the system. Some other major concerns arise with the adoption of this system. Such concerns have been discussed throughout this chapter. They include but are not limited to the issue of lack of information, data bias, no reasoned decision and due process of law not followed. Public policy concerns have also been raised in this chapter, as, under Article V(2)(b) of the New York Convention, this could serve as a ground for challenging the recognition and enforcement of an arbitral award rendered by an AI arbitrator. Public policy grounds can stem from several reasons, such as the composition of an arbitral tribunal. If the national legislation expressly provides for the appointment of only natural persons as arbitrators and parties go ahead to appoint AI arbitrators, this is a huge contravention of the national legislation and therefore a valid public policy ground. A potential public policy ground could be the absence of a reason for the decision. A strict element of a valid award is the inclusion of the reason for arriving at such a decision. This is a human element, as a human arbitrator can be persuaded to render an award after a careful examination of evidence, legal arguments and discretion. A scholar has referred to judges as flesh-and-blood persons, as opposed to a robot who merely recites the data which has been fed to ‘it’.125 It is, however, uncertain if an algorithm can be developed to have emotions and function with discretion. Therefore, it can be agreed that public policy requirements serve as a major roadblock to the smooth entering into force of a machine arbitrator’s regime. Thus, the recognition and enforcement of awards rendered as such may still encounter major difficulties under the New York Convention. This chapter concluded by examining the ethical considerations occasioned by the adoption of this new system. A party could contend that the use of AI arbitrators led to severe unfairness and partiality in the case, perhaps as a result of data bias, and that subsequently not being able to appeal has deprived the party of their fundamental right to a fair hearing.

McGlove v Lacey (1968) 288 F. Supp. Ryan Calo, ‘Robots as Legal Metaphors’ (2016) 30 Harv. J.L. & Tech. 209, 218–19.

124 125

PART VII DIGITAL TRADE LEGAL AND POLICY CHALLENGES: DATA, SECURITY AND IP

24. Digital economy agreements: where do we stand and where are we going?1 Dan Ciuriak

1. INTRODUCTION The digital transformation has profound implications for the rules-based trading system. The shift of economic, social and political interaction into online modes has, in the first instance, meant a steep increase in the share of international trade conducted through e-commerce or facilitated by digital technologies. This has led to a commensurate increase in the significance of trade frictions affecting digital commerce and thus amplified the benefits from digital trade facilitation. It has also made the associated data flows an important and coveted new asset class. Data flows have always accompanied trade in goods and services; however, where formerly they were largely ‘data exhaust’, technological breakthroughs in the nexus of big data, machine learning and artificial intelligence have made data the essential capital of the data-driven economy (DDE). This has unleashed a high-stakes ‘winner-take-most’ contest to dominate these powerful new general-purpose technologies. For trade rules, this has meant new chapters in trade agreements addressing data flows and data localization requirements as well as raising issues related to competition policy and industrial policy space. Further, it has generated pervasive structural and qualitative changes that have new consequences for trade commitments made in a pre-digital age and that generate entirely new issues for trade policy. These include the blurring of distinctions between goods and services (for example, through ‘servicification’), raising questions about whether a digital product is a good or a service – a key distinction given that trade rules are organized under mutually exclusive arrangements for goods under the General Agreement on Tariffs and Trade (GATT) and for services under the General Agreement on Trade in Services (GATS). Moreover, it has impacted on the balance of benefits from trade between the leading digital economies and the rest of the world, in particular developing economies, by shifting trade from material goods subject to tariffs to digital products that are protected from tariffs by the World Trade Organization (WTO) Moratorium on the application of tariffs to electronic transmissions, and by enabling firms to operate in jurisdictions without having a taxable, material presence in that jurisdiction. Finally, the digital transformation has created fundamental new social and political conditions that unavoidably elicit social and political responses with potentially profound impli-

This chapter builds on remarks at the Trade Law Forum 2021, Graduate School of International Studies, Seoul National University, 16 September 2021, and the Digital Trade Conference organised by the UK Department for Digital, Culture, Media and Sport, 30 March 2022, with an update to take into account the implications of Russia’s invasion of Ukraine and the ensuing open conflict in the digital domain. 1

416

Digital economy agreements  417 cations for the tenability of commitments to the free flow of data and for the framing of the general and national security carve-outs from commitments under the WTO Agreement and existing bilateral or regional trade agreements (RTAs). While many of the emerging issues are not yet ‘treaty-ready’,2 governments have moved to develop new digital economy agreements (DEAs) to complement the existing rules-based framework. DEAs thus loom large as critical institutional arrangements to support the ongoing shift of national and international commerce into the digital domain and to address the plethora of issues which that shift entails. To date, the reforms to address the digital transformation have mostly focussed on technically accommodating digital and digitally enabled trade and to add a new fifth freedom – the free flow of data across borders – to the four freedoms the system has promoted historically: the free cross-border flow of goods, services, technology and (to a lesser extent) people. The recent Digital Economic Partnership Agreement (DEPA) goes further and provides a more comprehensive template for a trade agreement tailored for the digitally transformed economy. This chapter is organized as follows. As context-setting background, Section 2 sets out the implications for trade and trade rules of the technological changes that have led to the emergence of the data-driven economy, which features two major challenges to achieving cooperative behaviour in the modern technological and economic system: the rise in the share of economic rent in the system and the problematic nature of data as a valuable asset and the essential capital of the DDE. Section 3 reviews the state of DEAs in terms of their issue coverage and the extent to which the texts provide substantive solutions to the pressing governance issues raised by the technological and economic conditions of the DDE. Section 4 then delves into several issues that remain problematic and must be addressed to make the system dynamically stable: (a) the status of digital products; (b) the distributional issues raised by the digital transformation of trade, including the sharing in the value of data; (c) the implications for policy space of the externalities behind the revitalization of industrial policy; and (d) cybersecurity, sovereignty and national security issues in the digital realm that are profoundly different than in the physical realm and that raise questions about the tenability of trade commitments made in a pre-digital context. Section 5 discusses the prospects for a revitalized WTO 2.0 that is fit for purpose for the digital age and thus a DEA in and of itself, taking into account the prospects for some convergence of national interests in regulating the data-driven economy as well as the implications of continued rapid technological change that will drive the formation of a major new capital stock: machine knowledge capital, which will compete with human capital in a way analogous to the competition that machines posed for manual labour.

Dan Ciuriak, ‘Digital Trade: Is Data Treaty-Ready?’ Centre for International Governance Innovation (21 February 2018) www​.cigionline​.org/​publications/​digital​-trade​-data​-treaty​-ready/​ accessed 23 February 2023. 2

418  Research handbook on digital trade

2.

THE TECHNOLOGICAL AND ECONOMIC CONTEXT FOR DEAS

As trade rule-making addresses the issues of the digital age, it is important to recognize that trade policy and indeed trade theory have historically been shaped by the technological and economic conditions of their age. A short 15 years after the establishment of the WTO in 1995, the world for which that agreement was forged was reshaped by the consequences of several connected technological innovations: ● the development in 2006 of deep learning techniques based on stacked neural nets by Geoffrey Hinton at the University of Toronto;3 ● the introduction by Apple of the iPhone in 2007 which launched the age of mobile and massively increased the amount of data collected and streamed into the now rapidly expanding cloud;4 and ● the application by Andrew Ng and his team at Stanford in 2009 of graphics processing units (GPUs) – computer chips designed for the massively parallel processing requirements of videogames – to run stacked neural nets.5 These developments enabled the emergence of a dynamic new economy predicated on leveraging the vast amounts of data generated by mobile devices and flowing into the cloud for processing and exploitation. The sense of profound change ushered in by the digital transformation and the powerful new technological capabilities that enable it is captured in the coining of terms such as the ‘fourth industrial revolution’ and ‘Industry 4.0’.6 2.1

Data is Different

The essential capital asset of this economy – data – has an elusive nature as an economic resource, which complicates greatly the development of a workable governance regime. The first issue is ownership. The defining capital assets of previous ages – land, machinery for mass production, and patented technology – came with property rights. In other words, there was ownership. But ‘ownership’ becomes problematic with data. Data is captured rather than being acquired in commercial transactions with a paper trail of invoices and receipts. In modern digital commercial contexts, numerous intermediaries participate in effecting transactions, from the e-commerce companies and search engines that connect buyers and sellers, to the financial entities that record the financial particulars and execute the payments, to the telecommunications providers that transmit the digital information. There is no unique ‘ownership’ of the information flow.

Kevin Kelly, ‘The Three Breakthroughs that Have Finally Unleashed AI on the World’ Wired (2014) www​.wired​.com/​2014/​10/​future​-of​-artificial​-intelligence/​ accessed 23 February 2023. 4 Rani Molla, ‘How Apple’s iPhone Changed the World: 10 Years in 10 Charts’ Vox (2017) www​ .vox​.com/​2017/​6/​26/​15821652/​iphone​-apple​-10​-year​-anniversary​-launch​-mobile​-stats​-smart​-phone​ -steve​-jobs accessed 23 February 2023. 5 Kelly (n 3). 6 Klaus Schwab, ‘The Fourth Industrial Revolution: What It Means, How to Respond’ World Economic Forum (14 January 2016) www​.weforum​.org/​agenda/​2016/​01/​the​-fourth​-industrial​-revolution​ -what​-it​-means​-and​-how​-to​-respond/​ accessed 23 February 2023. 3

Digital economy agreements  419 The second issue is valuation. The value of data is asymmetric and protean, depending ultimately on its use. To an individual, their own data has no measurable commercial value; to the digital platform firms, the accumulated mass of data translates into trillions of dollars of market capitalization. Accordingly, data does not fit into a neat framework around which markets can easily be structured, with vested ownership rights, and transparent asset values and transactions prices. Nor, seemingly paradoxically, does it fit the description of the alternative to a private good – namely, a freely accessible commons – because access and exploitation depends on investment at scales beyond the reach of most. By the same token, there is no established analogue developed for previous economic eras to apply mutatis mutandis to the data-driven economy. New forms of non-price market competition emerge based on asymmetric access to such data as captured in the term ‘surveillance capitalism’.7 These business models did not exist when the WTO Agreement was negotiated. Further, the pervasive, protean and permanent characteristics of digital data create fundamental new social and political conditions that will necessarily elicit social and political responses. The WTO Agreement was framed for a unipolar world, in which inert goods and services were traded across physical borders that could be rigorously policed. Today, that original, global rules-based framework applies largely unchanged to a multipolar, online world of smart products – and one that is facing further transformation with the rollout of the Internet of Things (IoT), which makes the backbone infrastructure of economies (transportation, telecommunications, energy and finance) as well as their public administration systems an interactive – and hackably vulnerable – central nervous system.8 Data is different, as is the economy which has been transformed by its collection in astronomical amounts and utilization as a new productive asset. 2.2

Economic Rents and Trade Conflict

The incumbent rules-based system was developed in the context of a mature industrial economy that was characterized by constant returns to scale and stability of the shares of national income flowing to capital and labour (the so-called Kaldor facts).9 Competitive markets tend to eradicate economic rents and tend towards a stable equilibrium in which there are no excess profits.10 Such markets also promise a ‘win–win’ sharing of production and consumption across trading economies through the principle of comparative advantage.11 7 Shoshana Zuboff, The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power (PublicAffairs 2019). 8 Jim Balsillie, ‘Why We Need a Second Bretton Woods Gathering’ (2018) Remarks for keynote presentation at the IMF Statistical Forum 6, Measuring Economic Welfare in the Digital Age: What and How? Washington, DC, 20 November. 9 Nicholas Kaldor, ‘Capital Accumulation and Economic Growth’ in F.A. Lutz and D.C. Hague (eds), The Theory of Capital (St. Martins Press 1961) 177–222. 10 The theoretical foundations for this claim are set out in K.J. Arrow and G. Debreu, ‘Existence of an Equilibrium for a Competitive Economy’ (1954) Econometrica 22(3) 265–90. 11 The prediction that economic welfare for the trading community as a whole as well as for its individual participants – including those that are less efficient in the production of all goods – could only increase through the mutual specialization induced by the dismantling of trade barriers gives the theory of comparative advantage persuasive power that rose to the level of ideology. This has been much disputed both as regards the distribution of benefits across countries by level of development,

420  Research handbook on digital trade At the international level, economies of scale were largely exhausted even in scale-intensive sectors such as automobile production, resulting in extensive intra-industry trade and competition based on product differentiation. Notably, these conditions inspired the development of ‘new trade theory’ circa 1980.12 At the same time, there remained incentives for strategic trade policy aimed at capturing global market share (such as through export subsidies).13 Such rivalry did erupt in some sectors – for example, in computer memory chips and civilian aircraft, where protracted trade wars were fought with subsidies, trade remedies and exercise of geopolitical power (for example, the Structural Impediments Initiative forced on Japan by the United States). However, the incentives for cooperation dominated, allowing the completion of the Tokyo and Uruguay rounds, which featured both a deepening of trade liberalization, expansion of the membership in the rules-based system, and the development of disciplines to curb strategic behaviour. These conditions started to change with advent of the KBE, which is based on intangible capital assets (patents, trade marks, copyrights and trade secrets) that require protection to be monetized. Where competition eradicates economic rents, protection creates economic rents. From a policy perspective, this protection is justified on the premise that the short-term costs of protection are outweighed by the long-run acceleration of welfare-enhancing innovation. The question of net benefit is, however, always an empirical one. Since market outcomes depend now on policy choices, calibration of the amount of protection becomes critical. Meanwhile, rent-seeking behaviour is inevitably induced to expand protection while strategic behaviour emerges to capture rents. Market outcomes become inherently uncertain. As the KBE spread, the dynamic in the rules-based system thus started to shift from pro-competitive market liberalization to the spread and intensification of protection for these intangible productive assets.

and within countries across different societal groups. That there are substantial gains from trade in this traditional industrial/agricultural goods setting, which is the point of departure for the present essay, is however almost universally accepted, notwithstanding the many critiques (see, e.g., Dani Rodrik, 'The Trouble with Globalization' (2017) The Milken Institute Review, Fourth Quarter www​ .milkenreview​.org/​articles/​the​-trouble​-with​-globalization​?IssueID​=​26 accessed 23 February 2023). On the technology-based Ricardian formulation of the theory of comparative advantage, see Rudiger Dornbusch, Stanley Fischer and Paul A. Samuelson, ‘Comparative Advantage, Trade, and Payments in a Ricardian Model with a Continuum of Goods’ (1977) American Economic Review 67(5): 823–39. On the factor endowments-based Heckscher-Ohlin formulation, see Bertil Ohlin, Interregional and International Trade (Harvard University Press 1933). On the restatement of the role of comparative advantage in modern heterogenous firms trade theory, see Andrew B. Bernard, Stephen J. Redding and Peter K. Schott, ‘Comparative Advantage and Heterogeneous Firms’ (2007) Review of Economic Studies 74(1) 31–66. On the limitations of the principle of specialization in explaining development and the role of diversification and technological development, see Dan Ciuriak, ‘Diversification vs. Specialization in Economic Development: A Comment’ Ciuriak Consulting Discussion Paper (6 October 2015) https://​ papers​.ssrn​.com/​abstract​=​2632806 accessed 23 February 2023. 12 On the development of new trade theory, which was so called to distinguish it from traditional trade theory based on comparative advantage, see Paul Krugman, ‘The Increasing Returns Revolution in Trade and Geography’ Prize Lecture Princeton University (8 December 2008). 13 James A. Brander and Barbara J. Spencer, ‘Export Subsidies and Market Share Rivalry’ (1985) Journal of International Economics 18 83–100.

Digital economy agreements  421 The DDE greatly intensified this dynamic. In the first instance, the DDE creates a high-stakes ‘winner-take-most’ economic context because of the combination of:14 ● steep economies of scale (which reflect high research and development costs to create intangible capital assets but near-zero marginal costs of replication of digital products); ● powerful economies of scope (as data become more powerful when cross-referenced against other data); ● network externalities in many sectors (especially the platform markets); and ● pervasive information asymmetry. These conditions conspire to give rise to ‘superstar firms’15 that dominate industries on a global scale and in some cases can boast of having more customers than the populations of China, the European Union and the United States combined. Second, just as the KBE accelerated the pace of innovation by industrializing research and development, the DDE further accelerated the pace by industrializing learning and creation through the major breakthroughs in the late 2000s that launched the data-driven economy. The share of protected intangible assets in the economy rose steeply as did the profit share of GDP and thus the scale of economic rents in the system. Reflecting the role of data and algorithms in generating these rents, a new emphasis was placed on protection of trade secrets, the more important mode of IP protection for these intangible assets, through modernized trade secrets legislation in the mid-2010s in all the major DDE jurisdictions. Third, the powerful new general-purpose technologies built on big data, machine learning and artificial intelligence have dual-use applications, adding geopolitical rivalry to the competition to capture economic rents. In the context of the accelerated pace of innovation, strategic behaviour was precipitated. The DDE was replete with possibilities for trade conflict, and those possibilities have been actualized with the trade and technology war between the United States and China and the proliferation of strategic industrial policies worldwide.16

3.

DIGITAL ECONOMY AGREEMENTS

3.1

Electronic Commerce Foundations

While e-commerce emerged only following the conclusion of the Uruguay Round and was thus not directly addressed in the 1995 WTO Agreement, major elements of the agreement – in particular the GATS (including the Telecommunications Services Annex) and the 1996 Telecommunications Reference Paper – helped establish some of the foundations for

Dan Ciuriak, ‘The Economics of Data: Implications for the Data-Driven Economy’ in Data Governance in the Digital Age, Centre for International Governance Innovation (5 March 2018) www​ .cigionline​.org/​articles/​economics​-data​-implications​-data​-driven​-economy/​ accessed 23 February 2023. 15 David Autor, David Dorn, Lawrence F. Katz, Christina Patterson and John Van Reenen, ‘The Fall of the Labor Share and the Rise of Superstar Firms’ (2020) The Quarterly Journal of Economics 135(2) 645–709 https://​doi​.org/​10​.1093/​qje/​qjaa004. 16 Dan Ciuriak and Maria Ptashkina, ‘Technology Rents and the New Great Game’ in Rahul Nath Choudhury (ed.), The China–US Trade War and South Asian Economies (Taylor & Francis/Routledge 2021). 14

422  Research handbook on digital trade e-commerce. The basis for internationally consistent development of national e-commerce laws was established in 1996 with the UNCITRAL Model Law on Electronic Commerce. The UNCITRAL Model Law on Electronic Signatures was adopted in 2001, and the United Nations Convention on the Use of Electronic Communications in International Contracts was adopted in 2005. The development of trade agreements started early as well, with the launch of the WTO work programme on cross-border e-commerce in 1998. RTAs quickly came into play in addressing the conditions of competition as these changed with the digital transformation. For example, in 2000, as the ‘bricks and clicks’ business model was taking hold, United Parcel Service of America (UPS) filed claims under the Canada–US Free Trade Agreement (CUSFTA), alleging anti-competitive practices by Canada Post and discriminatory treatment by the Canada Border Services Agency (CBSA) compared to the services the CBSA provided to Canada Post’s courier services.17 By 2004, express delivery services were included in an FTA (US–Chile). The first e-commerce provision on paperless trading was incorporated in the 2001 FTA between New Zealand and Singapore; in 2003, the first FTA with a chapter dedicated to e-commerce was concluded between Australia and Singapore.18 Supported by the rapid development of the institutional framework, e-commerce has flourished at the national and international levels. Total e-commerce transactions were estimated to be in the order of US$26.7 trillion in 2019, including US$4.9 trillion of business-to-consumer (B2C) transactions and $21.8 trillion of business-to-business (B2B) transactions. Cross-border B2C e-commerce amounted to US$ 440 billion or about 9 per cent of total B2C transactions worldwide.19 In light of well-established benchmarking of country-level e-commerce readiness by UNCTAD and others,20 and the scale of e-commerce activity and its rapid growth, this can be considered a fully mature area for regulation at the national level. This is also the case at the international level, where many trade agreements now include measures relating to digital trade, including chapters dedicated to e-commerce, and standalone

David A. Gantz, ‘The Evolution of FTA Investment Provisions: From NAFTA to the United States–Chile Free Trade Agreement’, American University International Law Review 19(4): 679–767. 18 José-Antonio Monteiro and Robert Teh, ‘Provisions on Electronic Commerce in Regional Trade Agreements’ (2017) WTO Staff Working Paper, No. ERSD-2017-11 https://​doi​.org/​10​.30875/​82592628​ -en accessed 23 February 2023. 19 UNCTAD, ‘Estimates of Global E-Commerce 2019 and Preliminary Assessment of Covid-19 Impact on Online Retail 2020’, UNCTAD Technical Notes on ICT for Development No. 18, United Nations Committee on Trade and Development (2021) https://​unctad​.org/​system/​files/​official​-document/​ tn​_unctad​_ict4d18​_en​.pdf accessed 23 February 2023. 20 See, e.g., UNCTAD, ‘eTrade for All’, UNCTAD Country Profiles, United Nations Committee on Trade and Development https://​etradeforall​.org/​country​-profiles/​accessed 23 February 2023. 17

Digital economy agreements  423 agreements.21 Numerous international organizations have developed instruments or frameworks for digital trade.22 Although issue coverage and organization of the provisions vary across agreements, with differing models adopted by the main e-commerce jurisdictions, the core elements of an e-commerce framework as developed through RTAs are well established and have been picked up in the new-era DEAs, and most issues are now also under discussion in the context of the WTO negotiations on e-commerce pursuant to the 2019 Joint Initiative on E-Commerce.23 Given the rapid development of this area, this section compares the issue coverage of core e-commerce provisions as well as the strength of commitment in agreements negotiated by the United States (Trans-Pacific Partnership24 or TPP and as modified in the 11-member Comprehensive and Progressive Agreement on Trans-Pacific Partnership25 or CPTPP), China (Regional Comprehensive Economic Partnership or RCEP)26 and the EU (EU–New Zealand FTA),27 as well as in the DEPA28 and the Australia–Singapore Digital Economy Agreement (ASDEA),29 which are representative of the DEAs. The issue coverage in the

21 For surveys, see, e.g., Javier López González and Taku Nemoto, ‘Mapping Commonalities in Regulatory Approaches to Cross-Border Data Transfers’ (2021) OECD Trade Policy Papers 248 http://​ dx​.doi​.org/​10​.1787/​0329a4e8​-en accessed 23 February 2023; Mira Burri and Rodrigo Polanco, ‘Digital Trade Provisions in Preferential Trade Agreements: Introducing a New Dataset’ (2020) Journal of International Economic Law 23(1) 187–220 http://​dx​.doi​.org/​10​.1093/​jiel/​jgz044; Monteiro and Teh (n 18); and Mark Wu, ‘Digital Trade-Related Provisions in Regional Trade Agreements: Existing Models and Lessons for the Multilateral Trade System’, RTA Exchange. Geneva: International Centre for Trade and Sustainable Development (ICTSD) and the Inter-American Development Bank (IDB) (2017) www​ .rtaexchange​.org/​accessed 23 February 2023. 22 López González and Nemoto (n 21) develop an inventory that lists 52 instruments directly relevant to digital trade in 24 different fora, including the 2013 WTO Trade Facilitation Agreement (TFA), which has been ratified by 156 WTO Members; see WTO, ‘WTO Trade Facilitation Agreement Database’ https://​tfadatabase​.org/​en/​ratifications accessed 23 February 2023. 23 WTO, ‘E-Commerce – MC12 Briefing Note’, Geneva: World Trade Organization (2022) www​.wto​.org/​english/​thewto​_e/​minist​_e/​mc12​_e/​briefing​_notes​_e/​bfecom​_e​.htm accessed 23 February 2023. 24 Office of the United States Trade Representative, ‘TPP Full Text’ https://​ustr​.gov/​trade​-agreements/​ free​-trade​-agreements/​trans​-pacific​-partnership/​tpp​-full​-text accessed 23 February 2023. 25 Global Affairs Canada, ‘Annex II – List of Suspended Provisions’ www​.international​.gc​.ca/​trade​ -commerce/​trade​-agreements​-accords​-commerciaux/​agr​-acc/​cptpp​-ptpgp/​annex2​-annexe2​.aspx​?lang​=​ eng accessed 23 February 2023. 26 RCEP Secretariat, ‘RCEP Legal Text’ https://​rcepsec​.org/​legal​-text/​accessed 23 February 2023. 27 New Zealand Ministry of Foreign Affairs and Trade, ‘NZ–EU FTA Text and Resources’ www​ .mfat​.govt​.nz/​en/​trade/​free​-trade​-agreements/​free​-trade​-agreements​-concluded​-but​-not​-in​-force/​new​ -zealand​-european​-union​-free​-trade​-agreement/​nz​-eu​-free​-trade​-agreement​-by​-chapter/​ accessed 23 February 2023. 28 New Zealand Ministry of Foreign Affairs and Trade, ‘DEPA Text and Resources’ www​.mfat​ .govt​.nz/​en/​trade/​free​-trade​-agreements/​free​-trade​-agreements​-in​-force/​digital​-economy​-partnership​ -agreement​-depa/​depa​-text​-and​-resources/​ accessed 23 February 2023. 29 Australian Government Ministry of Foreign Affairs and Trade, Australia–Singapore Digital Economy Agreement www​.dfat​.gov​.au/​sites/​default/​files/​australia​-singapore​-digital​-economy​ -agreement​.docx accessed 23 February 2023.

424  Research handbook on digital trade Table 24.1

Market access measures in selected DEAs

Definition of digital

DEPA

ASDEA

TPP

3.1***

1***

14.1***

3.2***

6***

14.4***

3.3***

5***

14.3***

RCEP

EUNZFTA

WTO

product as distinct from electronic transmissions Non-discrimination of

 

In play

digital products Customs duties – non-application to

12.11 (not

12.6.1***

In play

bound)

electronic transmissions or content (products) transmitted digitally

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22-30.

WTO e-commerce negotiations30 and relevant commitments in the WTO Trade Facilitation Agreement (TFA)31 and WTO Telecommunications Agreement32 are also identified. 3.1.1 Market access This includes the now standard provision for non-discriminatory treatment of digital products and commitments regarding application of customs duties on electronic transmissions. Commitments are generally strong; however, the RCEP agreement is weaker. It provides only for complying with the current WTO moratorium on applying tariffs to electronic transmissions, but with non-binding language, thus allowing for future application of such tariffs, were the moratorium to be discontinued. 3.1.2 Digital frameworks for electronic commerce Agreements conventionally establish a requirement to implement a domestic e-commerce framework, drawing on model laws developed by UNCITRAL and on UN conventions, which establish principles of non-discrimination for use of electronic formats, functional equivalence and technological neutrality. Trade agreements also address the main elements of the framework, including the recognition of contracts concluded by electronic means, e-signatures and authentication, e-invoices and e-payments. However, there is some variation across agreements as regards which specific elements to single out for individual treatment. The WTO negotiations address most of these elements, with the exception of e-payments. 3.1.3 Digital trade facilitation A set of provisions have been included in e-commerce chapters of RTAs and in DEAs to facilitate cross-border e-commerce. These focus on automation of customs procedures through paperless trading (including acceptance of electronic copies of trade documents and advance 30 WTO, E-commerce: MC12 Briefing Note: E-commerce www​.wto​.org/​english/​thewto​_e/​minist​ _e/​mc12​_e/​briefing​_notes​_e/​bfecom​_e​.htm accessed 23 February 2023. 31 WTO, Agreement on Trade Facilitation www​.wto​.org/​english/​docs​_e/​legal​_e/​tfa​-nov14​_e​.htm accessed 23 February 2023. 32 WTO, ‘Annex on telecommunications’ www​.wto​.org/​english/​tratop​_e/​serv​_e/​12​-tel​_e​.htm accessed 23 February 2023.

Digital economy agreements  425 Table 24.2

Digital frameworks for e-commerce in selected DEAs DEPA

ASDEA

TPP

RCEP

8***

14.5***

12.10***

Electronic contracts

33*

14.15**

Electronic Signatures

9***

14.6***

Electronic Transactions 2.3***

EUNZFTA

WTO In play

Framework 12.6***

12.8***

In play

12.9***

In play

12.10***

In play

and Authentication Electronic invoicing

2.5***

10***

Electronic payments

2.7**

11***

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22-30.

Table 24.3

Digital trade facilitation measures in selected DEAs ASDEA

TPP

RCEP

EUNZFTA

WTO

Paperless Trading 2.2***

DEPA

12***

14.9**

12.5**

12.15**

In play TFA**

Single Window

2.2.4***

12**

International

2.2.8**

4.3***

TFA*

In play

TFA**

standards to promote interoperability of systems Transparency

13.2***

SME participation 10.2**

14***

26.2***

12.12***

4.11***

36*

14.15

12.4*

12.14d*

facilitation

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22-30.

processing of trade documents in electronic format), provision of a single window and adoption of international standards to promote the interoperability of border systems. Notably, the WTO TFA already includes aspirational/hortatory commitments for acceptance of electronic copies of documents (Article 2.1), implementation of a single window (Article 4) and adoption of international standards (Article 3). Accordingly, there is room for RTAs/DEAs to improve upon WTO commitments. Notably, where RTAs or DEAs decline to specifically address these areas, the default is the TFA measure. We can include transparency requirements to publish information about trade regimes and the cooperative measures to support participation of small and medium-sized enterprises (SMEs) in this group of measures. 3.1.4 Consumer access, protection and privacy E-commerce agreements have developed a set of principles concerning access to the internet for e-commerce, requirements to maintain a regime to protect personally identifiable information, measures to guard against fraudulent and deceptive activities online and measures to address unsolicited marketing communications. These are routinely included in digital economy chapters and are under discussion in the WTO e-commerce negotiations.

426  Research handbook on digital trade Table 24.4

Consumer access, protection and privacy measures in selected DEAs DEPA

ASDEA

TPP

Internet Access

6.4

20

14.10

Personal Information

4.2***

17***

14.8***

RCEP 12.8***

EUNZFTA

WTO

12.16

In play

12.5

In play

Protection Online Consumer Protection

6.3***

15***

14.7***

12.7***

12.12***

In play

Unsolicited Commercial

6.2***

19 ***

14.14***

12.9***

12.13***

In play

5.2

18**

14.15

 

12.14c*

In play

5.1

34

14.16

12.13

12.14.4

In play

Electronic Messages Online safety and security – cooperation Cybersecurity – acknowledgement of importance of cooperation

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22-30.

Table 24.5

Business trust measures in selected DEAs DEPA

Source code – ban on

ASDEA

TPP

RCEP

EUNZFTA

WTO

28***

14.17***

12.16

11***

In play

7***

18.79*** (Suspended

forced access to source code with exception for critical infrastructure Encrypted products –

3.4***

protection subject to law

in CPTPP)

enforcement access Safe harbours – liability

18.82*** (suspended

of intermediary services

in CPTPP)

 

providers Trade secrets protection  

18.78***

11.2 (TRIPS

 

In play

only)

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22–30.

3.1.5 Business trust environment A number of business sector interests are touched on in e-commerce and digital economy agreements. While they are often grouped under a ‘business trust’ rubric, practice has yet to coalesce on a template. These measures include protection of software source code and products that use cryptography, subject to carve-outs for legitimate policy purposes, including as regards software for critical infrastructure and law enforcement. Some agreements also include measures addressing the liability of online services providers, especially the platform firms, for the use made by users of those platforms, including intellectual property (IP) violations. Finally, the digital transformation has made trade secrets protection a much more important IP area; trade agreements have been used to promote stronger IP protection internationally in this area.

Digital economy agreements  427 Table 24.6

Cross-border data flows, data localization and open data measures in selected DEAs DEPA

ASDEA

TPP

RCEP

EUNZFTA

WTO

4.3***

23***

14.11***

12.15***

12.4.2***

In play

Location of computing 4.4***

24***

14.13***

12.14***

12.4.2***

In play

Cross-border transfer of information by electronic means – allowance for business of a covered person with carve-out for legitimate policy exceptions facilities – ban on data localization with carve-out for legitimate policy exceptions Location of computing

25***

12.16

27

 

In play

facilities for financial services Open government data 9.5

 

In play

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22–30.

3.1.6 Commitments on data Measures addressing cross-border data flows and localization of data storage or processing are central to e-commerce trade and are part of the WTO negotiations. Some trade agreements include a separate measure for location of computing facilities for financial services due to the need for access to the data by financial supervisory authorities. Because of rising concerns around privacy and national security in the digital domain, the carve-outs for legitimate policy purposes become all-important in actually determining the extent of free flow of data. 3.1.7 Supporting commercial and physical infrastructure for e-commerce E-commerce includes both trade in intangibles that flow through telecommunications networks and digitally facilitated trade that flows through logistics chains for material products. In RTAs, the e-commerce provisions are complemented by measures in various other chapters (namely, customs and trade facilitation, telecommunications and the general services chapters) that address access to the enabling infrastructure for both digital transactions and digitally facilitated transactions. Digital connectivity is addressed through measure addressing internet interconnection, including submarine cable connections (submarine cables installation and service have also been introduced). Digitally facilitated business-to-consumer cross-border trade is addressed through commitments on courier services, express delivery services and low-value shipments. Notably, the DEAs are incorporating these measures, directly reflecting their centrality to a digital economy operating environment.

428  Research handbook on digital trade Table 24.7

Digital infrastructure measures in selected DEAs DEPA

Internet interconnection

ASDEA

TPP

RCEP

EUNZFTA

21

14.12***

Telecoms 9***

 

13.11***

Telecoms 18***

 

 

 

 

 

– negotiation of interconnection on a commercial basis Submarine cables connections Submarine cables installation

22***

and service Courier services – parcel

Annex 10-B***

delivery Express shipments

2.6***

13***

5.7***

Customs 4.15***

 

Low-value shipments

2.6.4***

13***

5.7.1f***

Customs 4.15**

 

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22–30.

3.1.8 Future areas for development signalled by the cooperation agenda A wide range of issues have been introduced into trade agreements in e-commerce and related chapters with commitments for cooperation and dialogue. These issues point to the future of DEA development. The most extensive coverage of these issues is in the DEPA, which highlights its ambition to provide a template for the future regulation of the digital economy. These issues can be loosely grouped in several categories for the present discussion: consumer-oriented issues, trade facilitation, regulatory issues in the digital economy, public sector management, and IP and data. 3.1.9 Consumer-oriented issues Three areas of relevance for consumers in recent agreements are digital inclusion, improved cooperation on mobile roaming charges and digital identities. The inclusion agenda singled out in the DEPA seeks to broaden the participation in e-commerce of women, rural populations, low socio-economic groups and Indigenous Peoples. This is in line with broader sustainability agendas and will likely be of growing significance for trade agreements given the income-skewing that is characteristic of the digital economy. RCEP features a well-developed cooperation article on roaming that addresses transparency of rates for international mobile roaming services, minimizing impediments to consumers and access to regulated rates for suppliers. The digital identity cooperation programmes included in the DEPA and the ASDEA focus on common standards and technical interoperability between the parties’ digital identity regimes, achieving comparable levels of protection of digital identities, developing use cases for mutual recognition of digital identities and encouraging broader cooperation in this area at the international level. 3.1.10 AI-related trade facilitation The DEPA introduces an important area for the coming era, where rapid development of artificial intelligence capabilities and 5G-enabled automation will impact significantly on digitally enabled trade by facilitating the use of electric, remote-controlled and autonomous vehicles; on-demand and dynamic routing solutions; and use of federated or ‘smart’ lockers for pickup of online retail purchases. The ASDEA meanwhile introduces a cooperation programme on

Digital economy agreements  429 Table 24.8

Consumer-oriented measures in selected DEAs

 

DEPA

ASDEA

TPP

RCEP

EUNZFTA

Digital inclusion

11.1*

 

 

 

 

 

 

Telecoms 22

 

29*

 

 

 

International mobile   roaming Digital identities

7.1**

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22-30.

Table 24.9

AI-related trade facilitation measures in selected DEAs

 

DEPA

ASDEA

TPP

RCEP

EUNZFTA

Logistics

2.4

 

 

 

 

Standards and Conformity Assessment for digital trade

 

30*

 

 

 

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22-30.

conformity assessment for digital trade. While trade in smart devices has grown rapidly, the growing role of increasingly sophisticated AI and the corresponding rise in attention to ethical AI systems portends future standards-related trade issues that trade agreements will need to address.33 3.1.11 Regulatory issues in the digital economy Competition policy, which was deferred during the Uruguay Round negotiations for further work as one of the so-called Singapore Issues, has become increasingly important in the context of the digital economy given the global scale of platform firms and is thus receiving renewed attention in trade agreements. In particular, data portability and interoperability has been identified as an area for cooperation in DEAs given the significant role of network externalities in digital platforms; such externalities raise the cost for consumers of switching between competing suppliers and in turn entrench the position of leading suppliers. In theory, data portability and interoperability mechanisms could increase competition by reducing switching costs for consumers, with consequential benefits to innovation, the development of comparison services in markets with complex pricing structures, and others.34 FinTech has also been an area where regulatory sandboxes have been used to trial regulatory reforms in a controlled experimental environment to identify approaches conducive for financial innovation. A related area is RegTech where the growing complexity of compliance (such as with anti-money laundering laws) and reporting requirements on currency transactions and suspicious activities, and so on, has led financial institutions to seek technological solutions and contributed to the emergence of so-called RegTech start-ups. Cooperation fostered by

33 See, e.g., Dan Ciuriak and Vlada Rodionova, ‘Trading Artificial Intelligence: Economic Interests, Societal Choices and Multilateral Rules’ in Shin-Yi Peng, Ching-Fu Lin and Thomas Streinz (eds), Artificial Intelligence and International Economic Law: Disruption, Regulation, and Reconfiguration (Cambridge University Press 2021). 34 OECD, ‘Data Portability, Interoperability and Digital Platform Competition’ (2021) OECD Competition Committee Discussion Paper http://​oe​.cd/​dpic accessed 23 February 2023.

430  Research handbook on digital trade Table 24.10

Digital economy regulatory measures in selected DEAs

 

DEPA

ASDEA

TPP

RCEP

EUNZFTA

Competition policy

8.4*

16*

 

12.16

 

Data regulation

 

 

 

 

 

8.1**

32

 

 

 

(portability, interoperability etc.) Financial Technologies (FinTech/RegTech)

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22–30.

Table 24.11

Public sector management measures in selected DEAs

 

DEPA

ASDEA

TPP

RCEP

EUNZFTA

Digitization of government procurement

8.3*

 

 

 

 

IP falling into the public domain

9.3

 

 

 

 

E-government

 

 

 

 

12.14e*

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22–30.

Table 24.12

Innovation-related measures in selected DEAs

 

DEPA

ASDEA

TPP

RCEP

EUNZFTA

Innovation

 

 

 

 

 

Data innovation – cooperation on data-sharing

9.4**

26**

 

 

 

AI governance frameworks

8.2**

31**

 

 

 

Notes: Cell entries refer to the agreement article; the strength of the provision is denoted by: *** = mandatory; ** = aspirational; * = cooperation; no asterisk = acknowledgement/recognition. Sources: see notes 22-30.

trade agreements in sharing experience and developing compatible standards is a promising area for development under DEAs. 3.1.12 Public sector management The DEPA identifies a couple of areas where improved government management practices could boost the digital economy, including digitization of government procurement and promoting the construction of publicly accessible databases to identify IP that has fallen into the public domain. These initiatives would promote the participation of SMEs in the digital economy by reducing search costs, including in establishing ‘freedom to operate’ with new technology projects. As well, cooperation on e-government has been flagged as an area for cooperation. 3.1.13 Innovation Reflecting the acceleration in the pace of innovation, DEAs have included commitments to promote innovation through cooperation, with particular attention to data-related innovation and the development of governance frameworks for ethical AI applications and other emerging technologies.

Digital economy agreements  431

4.

MAJOR OUTSTANDING ISSUES

4.1

The Classification of Digital Products

The DEPA differentiates between digital products and electronic transmissions by giving the former a narrower definition. Nonetheless, the DEPA does not pronounce on whether trade in digital products through electronic transmission should be categorized as trade in services or trade in goods and clarifies that the DEPA measures do not commit the parties to any particular view on this issue (DEPA note 10). At the same time, the DEPA clarifies that it binds the parties to not imposing customs duties on content transmitted electronically. This is problematic in numerous ways. First, there is no international consensus on making the WTO moratorium permanent. This is due to the lack of consensus that the moratorium on electronic transmissions should extend to digital products that are transmitted electronically (see, for example, communications from India, South Africa and Indonesia).35 This leads directly to a second unresolved problem, namely which regime – GATT or GATS – would apply to a digital product. As Janow and Mavroidis observe, ‘WTO rules are predicated on an absolute dichotomy between those dealing with goods and those focusing on services trade’.36 This is hardly inconsequential given the rapid development of new digital products that might face market access restrictions, if classified as services that fall under a classification that is not bound under the GATS by particular parties. Further, this puts the spotlight on the distinction between the GATS and GATT as regards technological neutrality. WTO dispute panels in US–Gambling (DS285) and China– Publications and Audiovisual Products (DS363) have held that the GATS is in principle technologically neutral.37 By contrast, WTO adjudicatory bodies have held that the GATT is not technologically neutral.38 Such a distinction between GATS and GATT has no economic grounding and could become problematic with the ongoing ‘servicification’ of goods and technological innovation in digital space. Conconi and Pauwelyn discuss the treatment of this distinction in WTO jurisprudence at length, noting the argument made by the United States in China–Publications and Audiovisual Products that a ruling that the services generated by a good define the product’s status could turn all goods into services.39 A similar argument has been made in the economics literature

35 See, e.g., WTO, ‘The Moratorium on Customs Duties on Electronic Transmissions: Need for Clarity on its Scope and Impact’, Communication from India and South Africa, 8 November (2021) WT/ GC/W/833; and WTO, ‘Statement by Indonesia: Facilitator’s Consultation on Electronic Commerce, MC11 declaration, and other relevant plenary sessions’, 20 December (2017) WT/MIN(17)/68. 36 Merit E. Janow and Petros C. Mavroidis, ‘Digital Trade, E-Commerce, the WTO and Regional Frameworks’ World Trade Review 18 (S1): s1–s7 (2019); at s 2. 37 See: WTO, United States–Measures Affecting the Cross Border Supply of Gambling and Betting Services – Report of the Appellate Body, WT/DS285/AB/R, adopted 20 April 2005; and WTO, China: Measures Affecting Trading Rights and Distribution Services for Certain Publications and Audiovisual Entertainment Products – Report of the Appellate Body, WT/DS363/AB/R, adopted 19 January 2010. 38 Paola Conconi and Joost Pauwelyn, ‘Trading Cultures: Appellate Body Report on China Audiovisuals (WT/DS363/AB/R, adopted 19 January 2010)’, World Trade Review 10(1): 95–118 (2011). 39 Ibid.

432  Research handbook on digital trade that the cross-border trade of, say, a table is merely the cross-border trade of the services of the carpenter who built it, and would be functionally equivalent to the carpenter going to the destination country and building the table there, which would be a mode 4 services transaction.40 Some observers argue that trying to distinguish between goods and services is outdated and that products should be viewed more as ‘bundles’.41 From an economics perspective, the more useful distinction to make in an age of intangibles is not whether a product is tangible or intangible but whether it is ‘durable’. For example, a music file that is shipped electronically across a border to be stored on the customer’s device is not extinguished when listened to; it continues to exist as a product on the device. By contrast, if the music file is streamed, it is in fact extinguished upon ‘consumption’. Durability is a characteristic of goods; extinguishment upon consumption is a characteristic of services. This distinction is reflected in commercial pricing of products and is thus more relevant for trade law purposes than an attempt to treat products as bundles or composites of goods and services. A similar classification issue is raised by cross-border flows of disembodied technology. In the innovation-intensive KBE/DDE, technology is traded both in embedded form (for example, in the value of commercial products such as a jet airliner or the value of a digital application) and in disembodied form (as reflected in payments for licensed IP). Here it is useful to consider the much-discussed question concerning the status of 3D printing instructions transmitted on a cross-border basis.42 Conceptually, these could be argued to fall into the category of ‘electronic transmissions’ that are not digital products in and of themselves, but rather technology flows that serve as intangible capital production inputs for which the payment (whether arm’s length or intra-firm) is then a payment for IP akin to a licence or royalty fee, rather than the payment for a good or a service. These distinctions are likely to become more important for trade negotiations (in terms of understanding national commercial interests) and trade rules (in terms of addressing frictions and adjudicating disputes) as the technology-driven changes in the economy cumulate, driving changes in the structure of trade flows and the nature of impediments to trade. 4.2

The Value of Data

Notwithstanding the enormous value placed on data in the commercial world, it does not show up for the most part in national economic accounts or in the trade statistics. At best, this value can currently only be inferred indirectly from market valuations placed on companies that succeed in exploiting data. For example, much of the market capitalization of the major platform firms is attributed to their data assets. Some sense of the value of data can also be

40 Brian R. Copeland, ‘Benefits and Costs of Trade and Investment Liberalization in Services: Implications from Trade Theory’ in Dan Ciuriak and John M. Curtis (eds) Trade Policy Research 2002 (Department of Foreign Affairs and International Trade 2003). 41 See, e.g., Janja Hojnik, ‘Technology Neutral EU Law: Digital Goods within the Traditional Goods/Services Distinction’ (2017) International Journal of Law and Information Technology 25 63–84 doi: 10.1093/ijlit/eaw009. 42 See, e.g., Robert W. Staiger, ‘On the implications of digital technologies for the multilateral trading system’ in WTO, World Trade Report 2018 (2018); at p.150; and Janow and Mavroidis (n 26).

Digital economy agreements  433 obtained from mergers and acquisitions transactions. For example, PayPal’s acquisition of the firm Honey Science for US$4 billion was described as a ‘pure data play’.43 This gap in the statistical treatment of data makes it difficult, if not impossible, to establish the value of commercial concessions in trade agreements that are under negotiation – or that were made in an era when data flows did not have such value and indeed when the collection and curation of data for administrative reasons represented a cost centre for firms, not a profit centre. At the same time, there is no lack of awareness of the value of data as is reflected in the efforts by the leading DDE states to entrench their first mover advantages by locking in the free flow of data, including by restricting the scope for data localization. While data localization issues had emerged prior to the DDE due to issues such as privacy and local jobs related to data custodial services,44 the economic transformation wrought by the advent of the DDE raised the stakes enormously. To give a sense of the value that have been placed on these measures, the United States International Trade Commission (USITC), in its evaluation of the economic impact of the US–Mexico–Canada Agreement (USMCA), concluded that the data provisions transformed a loss of -0.12 per cent of GDP for the United States from the main body of the agreement into a gain of 0.35 per cent in its ‘moderate’ scenario or as high as 1.21 per cent in its ‘high’ scenario. In dollar terms, these meant the difference between a loss of US$22.6 billion and gains of $68.2 or $235 billion in the moderate and high scenarios respectively.45 For countries with less developed digital economy capabilities, there is painful awareness of the asymmetry across countries in the ability to share in the benefits of the data-driven economy, including through commercial exploitation of their own data, which they see captured by global platform firms that may operate in their jurisdictions largely or entirely on a virtual basis with no local taxable presence. Nonetheless, they have agreed to the routine extension of the WTO moratorium on electronic transmissions since it was first introduced in 1998, including at the WTO’s 12th Ministerial Conference held in June 2022. This means that to date the benefits of the moratorium in terms of enabling the digital economy to take root and flourish have been judged to outweigh the costs in terms of forgone tariff revenue (an important consideration for developing countries), or any distortions to competition based on the above considerations. It is also likely that questions concerning technological feasibility and cost of collection of tariffs on digital product played a role. The effort to capture value in the data-driven economy thus shifted to application of digital services taxes, which drove the development of the OECD/G20 Inclusive Framework, which was designed to share the benefits of the digital economy more fairly as an alternative to digital services taxes. The Inclusive Framework, however, only takes into account current revenues 43 Tom Taulli, ‘Why PayPal Paid $4 Billion for Honey Science’ Forbes (23 November 2019) www​.forbes​.com/​sites/​tomtaulli/​2019/​11/​23/​why​-paypal​-paid​-4​-billion​-for​-honey​-science/​?sh​=​ 1d5cf5f07948 accessed 23 February 2023. 44 Michael Geist, ‘Data Rules in Modern Trade Agreements: Toward Reconciling an Open Internet with Privacy and Security Safeguards’, Centre for International Governance Innovation (4 April 2018) www​.cigionline​.org/​articles/​data​-rules​-modern​-trade​-agreements​-toward​-reconciling​-open​-internet​ -privacy​-and​-security/​accessed 23 February 2023. 45 USITC, ‘U.S.–Mexico–Canada Trade Agreement: Likely Impact on the U.S. Economy and on Specific Industry Sectors’ Washington: United States International Trade Commission, Publication Number: 4889, Investigation Number: TPA 105-003 (April 2019); Tables 2.6 and 2.7 at pp.56–7.

434  Research handbook on digital trade of platform firms, not the value of data as a capital asset nor the rents that accrue to data capital.46 Until a suitable ‘rent-sharing’ agreement is developed for data rents in digital trade agreements, this imbalance in benefits from the DDE threatens to drive a continuing widening in income disparities and consequential distributional tensions. Leaving these issues hanging may therefore create a stumbling block for multilateralization of the DEPA. 4.3

Intellectual Property

The nature of innovation has evolved through industrialization of the process, first through application of computer-assisted design to research and development, then through application of neural nets to discovery or learning, then to the industrialization of the entire process, a step marked by the issuance of a patent to an AI system in 2021. New institutional frameworks have also emerged, including open source and co-creation. Along the way, the pace of innovation has accelerated and the volume of innovation outputs as represented by the amount and value of protected IP has soared. Meanwhile, the basic instruments incentivizing innovation – the various forms of protection for IP – were essentially unchanged and indeed formal protection was expanded as trade secrets protection was upgraded to address the increased emphasis on this IP instrument to cover data and algorithms in the data-driven economy.47 Moreover, even as the accelerated pace of innovation compressed the product life cycle, and income distribution became increasingly skewed due to the capture of economic rents by protected IP, IP reforms mostly worked to lengthen and strengthen protection. The incorporation of IP protection in trade agreements – most importantly the World Trade Organization’s Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) – served to spread such protection globally – and to make it more difficult to implement reforms to address the changing nature of innovation and its role in the economy. While protection is essential for the monetization of intangible assets such as IP, there are rising concerns about unintended consequences. Critics have voiced concerns about the emergence of what has been termed a new form of capitalism – ‘intellectual monopoly capitalism’.48 Moreover, public goods issues have become far more significant, with the major issues of the day – the pandemic, climate change and the regulation of data – all having major public goods characteristics. While the highest profile topic of a TRIPS waiver for IP over COVID-19 vaccines is not a digital economy issue per se, similar issues will be confronted in reconciling public imperatives and private rights in IP in areas that are central to the digital economy. DEAs have flagged cooperation on IP issues in the digital realm as an issue. This will likely be an area where significant development will be required.

Dan Ciuriak and Akinyi J. Eurallyah, ‘Taxing Capital in the Age of Intangibles’, Centre for International Governance Innovation (5 November 2021) www​.cigionline​.org/​articles/​taxing​-capital​-in​ -the​-age​-of​-intangibles/​ accessed 23 February 2023. 47 Dan Ciuriak and Maria Ptashkina, ‘Quantifying Trade Secret Theft: Policy Implications’, Centre for International Governance Innovation (10 May 2021) www​.cigionline​.org/​publications/​quantifying​ -trade​-secret​-theft​-policy​-implications/​ accessed 23 February 2023. 48 See, e.g., Ugo Pagano, ‘The Crisis of Intellectual Monopoly Capitalism’ (2014) Cambridge Journal of Economics 38(6) 1409–29 https://​doi​.org/​10​.1093/​cje/​beu025; and Cecilia Rikap, Capitalism, Power and Innovation: Intellectual Monopoly Capitalism Uncovered (Routledge 2021). 46

Digital economy agreements  435 4.4

Artificial Intelligence and Machine Knowledge Capital

Trade in ‘smart’ devices has grown exponentially with the digital transformation, to all appearance with seemingly little interference from non-tariff barriers, notwithstanding the plethora of potential issues ranging from standards to trust, dual use, and societal impacts in areas such as jobs.49 The DEPA has recognized the importance of AI in the digital economy, and incorporated commitments for the development of ethical and governance frameworks to support trusted, safe and responsible use of AI technologies. This is an area that will likely require very substantial development given truly fundamental technological breakthroughs in scaling AI technology that have been made in recent years:50 ● Specialized AI computer chips have broken through the trillion transistor mark; by 2022, the largest featured 2.6 trillion transistors. ● AI systems grew from the hundred billion parameter scale to the trillion parameter scale, to 10 trillion and to 174 trillion in a matter of a few years. ● Power requirements to run AI chips were reduced to 1/100th through improved protocols. ● The power of AI systems was increased by orders of magnitude to allow systems with 100 billion parameters to achieve the efficiency of trillion parameter systems. Reflecting the increased power generated by the scaling and efficiency gains, specialized AI systems are now routinely breaking through human benchmarks. AI systems have now been awarded patents. AI-piloted fighter jets have out-duelled human-piloted jets in dogfights. The Beijing Winter Olympics showcased a suite of service robots deployed to address pandemic-related concerns. Generative AI systems are generating text, images and music of a quality that has generated both a sensation and profound concerns.51 And this is only the beginning. Just as the industrial revolution transformed the production of goods, our human-capital-intensive services industries are about to be transformed by the widespread deployment of machine knowledge capital (MKC), which will compete with human knowledge capital the way that the machinery of mass production competed with labour, and which will intensify the competition for human labour by making production machinery more flexible and much, much smarter. Importantly, MKC will make services sectors scalable, lifting the constraints imposed by the so-called Baumol effect, which states that as societies shift from industrial production to

Ciuriak and Rodionova (n 33). For discussion and links, see Dan Ciuriak, ‘The Data-Driven Economy Raises New Challenges for Global Governance’ Centre for International Governance Innovation (3 October 2022) www​.cigionline​ .org/​articles/​the​-data​-driven​-economy​-raises​-new​-challenges​-for​-global​-governance/​ accessed 23 February 2023. 51 For a sampling of the controversies stirred by the release of OpenAI’s ChatGPT, see e.g., Kathy Hirsh-Pasek and Elias Blinkoff, ‘ChatGPT: Educational Friend or Foe?’ Brookings (9 January 2023) www​.brookings​.edu/​blog/​education​-plus​-development/​2023/​01/​09/​chatgpt​-educational​-friend​-or​-foe/​ accessed 23 February 2023; and Nestor Maslej, ‘It’s Time to Start Thinking about Politically Biased AI’ Centre for International Governance Innovation (25 January 2023) www​.cigionline​.org/​articles/​its​ -time​-to​-start​-thinking​-about​-politically​-biased​-ai/​ accessed 23 February 2023. On the rapidly evolving generative AI scene, see, e.g., Eric Mack, ‘Generative AI Tools Like ChatGPT and Dall-E Aren't Going Away. Here’s Why’ (CNET 22 February 2023) www​.cnet​.com/​science/​generative​-ai​-tools​-like​-chatgpt​ -and​-dall​-e​-arent​-going​-away​-heres​-why/​ accessed 23 February 2023. 49 50

436  Research handbook on digital trade services, growth slows because the human capital that underpins the traditional services industry cannot be easily scaled. Relieving this constraint promises great benefits in many areas where human resources are strained – for example in nursing and elder care. At the same time, it raises profound new issues for the future of work. The long-standing discussion about AI and jobs has led to the general conclusion that AI will both complement and substitute for human capital, that tasks will be reassigned but that new types of jobs will emerge, and that labour markets will clear. However, this conclusion masks the incalculable impacts on economies and societies organized professionally and socially around the returns to human capital. From a trade system perspective, the growth in the stock of MKC and the incremental share of national income that will flow to the owners of that stock, while disrupting income distribution by in effect cannibalizing the income share flowing to white-collar workers, will likely become a new bone of contention. 4.5

General and National Security Exceptions in the Digital Environment

The digital transformation has generated new vulnerabilities for nation states in terms of protecting democratic processes and national security. These concerns have become acute with Russia’s invasion of Ukraine, which has set in sharp relief the many risks in this area: it is the first full-blown cyber war, featuring both state and non-state actors (including ‘hacktivists’ such as Distributed Denial of Secrets and Anonymous) attacking the information infrastructures of the combatants;52 the first ‘social media war’;53 and the first ‘splinternet war’, in which narratives in different parts of the internet exist in essentially distinct ‘alternative fact’ domains.54 It has also focussed attention to the groundwork for the invasion laid down before the fact through Russia’s influence campaign to sew division in the West.55

Kate Conger and Adam Satariano, ‘Volunteer Hackers Converge on Ukraine Conflict with No One in Charge’ The New York Times (4 March 2022) www​.nytimes​.com/​2022/​03/​04/​technology/​ukraine​ -russia​-hackers​.html accessed 23 February 2023. 53 Peter Suciu, ‘Is Russia's Invasion of Ukraine the First Social Media War?’ Forbes (1 March 2022) www​.forbes​.com/​sites/​petersuciu/​2022/​03/​01/​is​-russias​-invasion​-of​-ukraine​-the​-first​-social​-media​-war/​ accessed 23 February 2023. 54 Dan Ciuriak, ‘Social Media Warfare Is Being Invented in Ukraine’ Centre for International Governance Innovation (15 June 2022) www​.cigionline​.org/​articles/​social​-media​-warfare​-is​-being​ -invented​-in​-ukraine/​accessed 23 February 2023. 55 For example, this includes buying ‘Londongrad’ (Marina Hyde, ‘It’s Putin’s Tale of Two Cities – London for His Oligarchs, Kyiv for His Bombs’, The Guardian (25 February 2022), www​.theguardian​ .com/​commentisfree/​2022/​feb/​25/​tories​-oligarchs​-london​-government​-putin​-donations accessed 23 February 2023.); the ‘kompromat’ (the compromise) of Donald Trump (Julia Ioffe, ‘Now We All Know What Putin Has on Trump’, GQ (2018) www​.gq​.com/​story/​what​-putin​-has​-on​-trump accessed 23 February 2023); and influence over the Republican Party (Tina Nguyen, ‘How the G.O.P. Gave Maria Butina Her Ultimate Cover’, Vanity Fair (2018) www​.vanityfair​.com/​news/​2018/​07/​how​-the​-gop​-gave​ -maria​-butina​-her​-ultimate​-cover accessed 23 February 2023), including through Fox News (Brian Klaas, ‘He Worked in Russian Media. He Recognizes the Same Tactics at Fox News’, The Washington Post (2021) https://​www​.washingtonpost​.com/​opinions/​2021/​03/​23/​he​-worked​-russian​-media​-he​-recognizes​ -same​-tactics​-fox​-news/​ accessed 23 February 2023). Notably, there were links between the backers of Brexit and the Trump campaign, as uncovered by Carole Cadwalladr: ‘The Great British Brexit Robbery: How Our Democracy was Hijacked’, The Guardian (7 May 2017) www​.theguardian​.com/​technology/​ 2017/​may/​07/​the​-great​-british​-brexit​-robbery​-hijacked​-democracy accessed 23 February 2023. 52

Digital economy agreements  437 The vulnerabilities will only expand as the backbone infrastructure sectors (telecommunications, transportation, energy and finance) increasingly come to resemble an interactive central nervous system for the economy. DEAs provide for exceptions to the commitments regarding free flow of data and data localization requirements, as well as making available general and national security exceptions. In the latter regard, DEPA Article 15.1 incorporates the GATT and GATS general exceptions mutatis mutandis with no further elaboration of how those articles might need to be read to reflect the concerns of the digital age. Article 15.2 provides an exception for ‘essential security interests’ but with no elaboration on what might trigger this exception and/or how it might be circumscribed. Relying on flexibilities through open-ended exceptions could, however, prove to be an inefficient and fractious way forward in scoping the tenable extent of data commitments given that invocation of inherently opaque national security grounds invites retaliation and creates a slippery slope for further derogations from trade disciplines (it may be recalled that this was a major concern when the national security exception was first incorporated in the GATT).56 National security is now being advanced in very expansive terms to include everything from access to personal data, to industrial supply chains, to university research partnerships, to all forms of advanced technology. Technological change is non-neutral with regard to the substantive effect of commitments; accordingly, the prudent way forward might be renegotiation in areas that are materially affected by the digital transformation and where the international flow of data poses an unacceptable risk.57 56 Pinchis-Paulsen provides a detailed discussion of the creation of the national security exception in the GATT; the context featured tension between preserving the integrity of the trade system and providing for leeway to address national security concerns; see Mona Pinchis-Paulsen, ‘Trade Multilateralism and U.S. National Security: The Making of the GATT Security Exceptions’ (2020) Michigan Journal of International Law 41 109 https://​repository​.law​.umich​.edu/​mjil/​vol41/​iss1/​4 accessed 23 February 2023. Heath, looking at the contemporary scene, argues that conflating emerging issues with national security creates risk for trade rules in that good faith but novel national security claims may be advanced in areas ranging from economic security to climate change to (presciently) pandemic disease, terrorism, and threats to media. Heath surmises that such good-faith but novel claims may pose a significant challenge to the rules-based system by letting the protectionist genie out of its bottle, as feared by the framers of the GATT – indeed, this risk was actualized by the Trump Administration with its unleashing of Section 232 tariff measures, leading to cascading retaliatory measures. See J. Benton Heath, ‘The New National Security Challenge to the Economic Order’ (2020) The Yale Law Journal 129 1021–92. 57 Peng focuses on the scope of the WTO agreements to cover innovative products based on ‘generic’ characteristics rather than technologically specific instantiations, which is consistent with the concept of ‘technological neutrality’; this reading supports the operation of the WTO system without frequent renegotiation whenever ordinary course innovation changes product delivery – especially in the environment of convergence of technologies that characterized the first decade of the 2000s. See Shin-yi Peng, ‘Regulating New Services through Litigation? Electronic Commerce as a Case Study on the Evaluation of “Judicial Activism” in the WTO’ (2014) Journal of World Trade 48(6) 1189–1222. Streinz, however, questions whether there is a need to ‘recalibrate the temporal mismatch between long lasting obligations under international economic law and the rapid pace of technological development? International economic law’s traditional commitment towards providing “certainty” for transnational business activity seems at odds with the rapid pace of innovation in the digital economy. The principle of technology neutrality may need to be cabined when new technologies transform the economy fundamentally.’ Thomas Streinz, ‘International Economic Law’s Regulation of Data as a Resource for the Artificial Intelligence Economy’ in Shin-yi Peng, Ching-Fu Lin and Thomas Streinz (eds), Artificial Intelligence and International Economic Law: Disruption, Regulation, and Reconfiguration (Cambridge

438  Research handbook on digital trade It may, accordingly, be necessary to incorporate provisions in DEAs modelled on GATT Article XXVIII and GATS Article XXI but expressly tailored for the digital economy to revise bargains made under the WTO Agreement. While these may not be immediately needed between like-minded countries negotiating bilateral or small regional agreements, they might ultimately be necessary if DEAs are to be concluded between countries on opposite sides of the security divide.

5. CONCLUSIONS In retrospect, the rules-based system that was developed in the mature industrial economy of the post-Second World War era was crafted at a uniquely ideal time for such a system: when strategic behaviour was the exception rather than the norm; when the competitive drive to capture additional scale economies drove the formation of global value chains, which spread economic development; and prior to the rise of the intangibles economy with its many problematic features (which include rising societal tensions over skewing of income distribution, increased market concentration, disappointing development outcomes in the Global South, waning business dynamism in the Global North and counter-productive developments in dealing with major societal challenges, including the frictions over patent waivers in the pandemic and trade frictions over measures to address climate change). Moreover, this system evolved within a geopolitically stable order (pre-WTO, the US-led GATT system that excluded the Soviet Union and China; post-WTO, the global system under the undisputed hegemony of the United States). But we are in a profoundly different technological and political environment, one which features intense technological rivalry to capture economic rents and the weaponization of interdependence across increasingly sharp geopolitical divides. These considerations underscore the challenges for rule-making in DEAs and the still greater challenge of establishing a multilateral system for the digital age.

University Press 2021). In this regard, datafication seems sufficiently transformative both in terms of the benefits (and thus the value being contested) and in terms of the security risks to which it gives rise (see Dan Ciuriak, ‘The Challenge of Updating Institutions for Digital Trade’, Centre for Governance Innovation (16 July 2021) www​.cigionline​.org/​articles/​the​-challenge​-of​-updating​-institutions​-for​-digital​ -trade/​accessed 23 February 2023).

25. Digital trade and economic security: considerations on the Japan–US Digital Trade Agreement Rikako Watai1

1. INTRODUCTION Widespread use of the internet began in regular households in the mid-1990s. Amazon.com launched as an online bookstore in July 1995,2 and has since expanded dramatically to become a vast online retailer offering a wide range of merchandise. More recently, the recent global COVID-19 pandemic has also been a contributing factor to wider use of the internet than ever before. There is not yet a single globally accepted definition of digital trade. For that reason, many international organizations have established their own understanding of what digital trade means to them. Among others, one definition, from the Organisation for Economic Co-operation and Development (OECD), is gaining acceptance: ‘that it encompasses digitally enabled transactions in trade in goods and services which can be either digitally or physically delivered involve consumers, firms and governments.’3 Digital trade includes both markets for electronic commerce aimed at consumers (B-to-C EC) and electronic commerce between businesses (B-to-B EC). Recently, transactions in which items are sold by one consumer and purchased by another individual have also become a common form of e-commerce (C-to-C EC). According to the Japan External Trade Organisation (JETRO), global digital trade topped more than 3 trillion US dollars in 20204 and continues to play an increasingly important role in supply chains. The global spread of digital trade will inevitably require unified and universal rules. As the size of the market in digital trade expands, there is likely to be a need for standardisation, particularly on the treatment of data. The World Trade Organization (WTO), since being established in 1995 as a new organization to replace the 1947 General Agreement on Tariffs This chapter was written in the Spring of 2022 and reports part of the results of a research project supported by the Keio Gijuku Academic Development Funds. 2 Richard L. Brandt, ‘Birth of a Salesman: Behind the Rise of Jeff Bezos and Amazon’, Wall Street Journal (Online, New York, 15 October 2011). www​.wsj​.com/​articles/​SB1​0001424052​9702039143​ 0457662710​2996831200 accessed 1 May 2022. 3 J. López González and  M. Jouanjean, ‘Digital Trade:  Developing a Framework for Analysis’,  OECD Trade Policy Papers, No. 205 OECD Publishing 6 (Online, Paris, 2017) https://​doi​.org/​10​.1787/​ 524c8c83​-en accessed 1 May 2022. 4 Japan External Trade Organization, ed., JETRO Global Trade and Investment Report 2021 (Japanese on Demand Version, August 2021) 103. An English summary of the content of the report can be found in JETRO Global Trade and Investment Report 2021 – The World Changed by COVID-19, The Outlook for Sustainable International Business –­ Overview 47　www​.jetro​.go​.jp/​ext​_images/​en/​reports/​ white​_paper/​trade​_invest​_2021​.pdf accessed 1 May 2022. 1

439

440  Research handbook on digital trade and Trade (GATT), has set comprehensive international trade rules for goods and services. Although the original WTO Agreement has been revised to make international trade faster and smoother, this agreement does not include rules on digital trade.5 It may be claimed that, despite the fact that digital trade is subject to the basic rules of international trade law, this was a business model that was not anticipated at the time the WTO’s establishment. In the Work Programme on Electronic Commerce adopted at the General Council in 1998, it was agreed to extend the moratorium on applying tariffs to electronic transmissions. While this has become an established practice and continues as a norm to the present day, it is merely one single rule on digital trade.6 Voluntary negotiations have been held by individual nations on more substantial rules for digital trade. In December 2021, a ministerial statement by the co-conveners of a WTO Joint Statement Initiative on E-Commerce agreed to move forward with negotiations that would aim to ‘secure convergence on the majority of issues by the end of 2022’ and to ‘identify opportunities throughout 2022 for Ministers to provide guidance on key issues in the negotiations’.7 Japan is one of the co-convenors, together with Australia and Singapore, and it is hoped that Japan will play a leading role in the negotiations, expected to develop quickly from this point. Major countries’ ideas on rules for digital trade are not necessarily the same. The influential US internet-based companies Google, Apple, Facebook and Amazon.com (GAFA) are together known as the Big Four.8 This means that private-sector companies exert a powerful influence on rulemaking in the US. In contrast, the EU is taking an approach to rulemaking that focuses on individuals rather than businesses; for example, 2016’s General Data Protection Regulation (GDPR) has followed a policy based on protecting personal information. In China, the government takes the lead on data management through a legal system that limits the cross-border transfer of data and requires all data to be stored within China. Japan has also developed its own approach. At the World Economic Forum annual meeting in January 2019, then Prime Minister Shinzo Abe announced Japan’s Data Free Flow with Trust (DFFT) initiative; he included it in the Leaders’ Declaration at the G20 Osaka Summit in June of that same year. Since taking office in November 2021, the new cabinet under Prime Minister Fumio Kishida has announced that economic security will be its biggest priority.9 The relationship between economic security and digital trade is evolving, but data are clearly of major importance in a digitalized society, and data leaks have the potential to cause national security problems.

The WTO uses the term e-commerce as a synonym for digital trade. World Trade Organization, Work Programme on E-Commerce, www​.wto​.org/​english/​tratop​_e/​ ecom​_e/​ecom​_work​_programme​_e​.htm accessed 1 May 2022. 7 WTO Joint Statement Initiative on E-commerce Statement by Ministers of Australia, Japan and Singapore www​.meti​.go​.jp/​press/​2021/​12/​20211214001/​20211214001​-1​.pdf accessed 1 May 2022. On recent developments in rulemaking on digital trade, see World Trade Organization, ‘Joint Initiative on E-commerce News Archives’, www​.wto​.org/​english/​news​_e/​archive​_e/​jsec​_arc​_e​.htm accessed 1 May 2022. 8 In October 2021, Facebook announced that it would change its company name to Meta. 9 Brad Glosserman, ‘Kishida Doubles Down in Economic Security’ Japan Times (12 October 2021) www​.japantimes​.co​.jp/​opinion/​2021/​10/​12/​commentary/​japan​-commentary/​kishida​-economic​-security/​ accessed 1 May 2022. 5 6

Digital trade and economic security  441 Creating a common set of global rules will not be easy while the basic positions of countries on digital trade do not match. This chapter considers the current state of the international rules on digital trade, focusing on the balance between promoting digital trade and national security.

2.

DIGITAL TRADE REGULATIONS AND THE JAPAN–US DIGITAL TRADE AGREEMENT

It is of course possible to apply the existing rules of the WTO to digital trade. Under the WTO Agreement, which discipline to apply depends on what category digital content falls under. If digital content is interpreted as a transaction of goods, the GATT applies; if it is interpreted as a transaction of services, the General Agreement on Trade in Services (GATS) applies.10 A music CD would thus be subject to GATT if it is traded as a CD, but GATS would apply if the same content is traded through a streaming service.11 Although there is no difference between the basic concepts of GATT and GATS, the different rules applied to various forms of transactions with the same content can become a source of confusion, as seen in the CD example. If it proves impossible to establish a common set of global rules for digital trade, and each country establishes rules of its own, this will lead to fragmentation in cyberspace. Digital trade is based on the idea that data can circulate and move freely across borders, and fragmentation would undermine the meaning of digital markets. To establish rules for digital trade as part of the WTO rules, it will be necessary to follow the revision procedures given force in Article 10 of the WTO Agreement.12 This means that even if negotiations proceed according to the schedule laid out in the December 2021 Ministerial Statement issued by the co-convenors on the Joint Statement Initiative on E-Commerce, it is far from certain that a consensus will be reached that is acceptable to all member countries. To encourage liberalization and to promote smoother trade and investment even outside the WTO framework, countries around the world are working to conclude economic partnership agreements and free-trade agreements.13 These agreements are between states that have basic national legislation in place to facilitate digital trade, such as privacy and data protection, electronic payments and cybercrime legislation. Several economic partnership agreements that Japan has signed cover digital trade. One example is the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP, also known as TPP11), which came into effect on 30 December 2018. The CPTPP Agreement was followed by the Agreement between Japan and the United States of America concerning Digital Trade (Japan–US Digital Trade

Suguru Kamitanida, ‘Nichibei boeki kyotei oyobi nichibei dejitaru boeki kyotei no gaiyo (A Summary of the Japan–US Trade Agreement and the Japan–US Digital Trade Agreement)’ (2019) 417 Rippou to Chousa 106, 112. 11 Ibid. 12 Article 10 of the WTO Agreement decrees that ‘Unless the Ministerial Conference decides on a longer period, for a period of 90 days after the proposal has been tabled formally at the Ministerial Conference any decision by the Ministerial Conference to submit the proposed amendment to the Members for acceptance shall be taken by consensus’. 13 Since Article 5 of the General Agreement on Trade in Services makes it a requirement for regional trade agreements to have ‘substantial sectoral coverage’ of services, an agreement restricted to the field of digital trade alone would be regarded as rulemaking outside the WTO framework. 10

442  Research handbook on digital trade Agreement) in January 2020. The Japan–US Digital Trade Agreement is particularly significant; it is a concrete expression of the DFFT concept. Both Japan and the United States are taking the lead in international rulemaking by presenting high-level rules for digital trade.14 2.1

The Development of DFFT

Japan’s Data Free Flow with Trust (DFFT) outlines the aim of the Japanese government: to achieve a balance in drawing up digital trade rules between the free and open flow of data on the one hand and data security and user confidence on the other, or ‘To aim to promote the free international flow of data, so that data can move freely without regard for being conscious of national borders, bringing advantage beneficial to business and the solution of social issues, while maintaining faith in privacy, security, and intellectual property’.15 The Japan–US Digital Trade Agreement aims to facilitate digital trade between Japan and the United States by establishing a legal foundation to promote smooth, reliable and free digital trade. The Agreement consists of 22 articles in total, and with regard to the DFFT concept, the articles addressing ‘Data Free Flow’ and ‘with Trust’ are addressed below.16 2.1.1

Provisions relating to ‘Data Free Flow’

2.1.1.1 Article 7 Customs Duties The idea that customs duties should not be imposed on electronic transmissions has become established as a WTO norm. As far as electronic transmissions between the United States and Japan are concerned, it is likely that an interpretation will be made that electronic transmissions do not constitute a transaction in goods nor exchange of property. 2.1.1.2 Article 8 Non-Discriminatory Treatment of Digital Products This provision establishes non-discriminatory treatment of digital products. Article 1(g) of the Agreement defines digital product as ‘a computer program, text, video, image, sound recording, or other product that is digitally encoded, produced for commercial sale or distribution, and that can be transmitted electronically’. A digital product created, produced, published, contracted for, commissioned or first made available on commercial terms in the territory of the other party, or a digital product of which the author, performer, producer, developer or owner is a person of the other party, shall not receive less favourable treatment from either party than it does from other similar digital products. The WTO Agreement is based on the basic principles of most-favored-nation treatment and the national treatment principle, and Article 8 shows that the Japan–US Digital Trade Agreement can also be understood to include both these principles.

Cabinet Secretariat TPP Taskforce, Briefing on the Japan–US Trade Agreement and Basic Policy on Comprehensive Revisions to Policy on TPP and Related Matters, Materials 2: Japan–US Digital Trade Agreement (Summary) (27 October 2019). 15 Strategic Conference for the Advancement of Utilizing Public and Private Sector Data, Strategic Headquarters for the Advanced Information and Telecommunications Network Society, New IT Policy Outline for the Digital Age (2019) 19. 16 This draws on Kamitanida, supra note 10 at 116. 14

Digital trade and economic security  443 2.1.1.3 Article 11 Cross-Border Transfer of Information by Electronic Means This article is the central provision with regard to data free flow. It decrees that no restrictions may be imposed on the cross-border transfer of information except in the case of measures necessary to achieve a legitimate public policy objective. It provides, however, that such measures shall not be applied in such a manner as to constitute a means of arbitrary or unjustifiable discrimination or a disguised restriction on trade. Further, these measures shall not impose restrictions on the transfer of information greater than those necessary to achieve the objective; thereby, consideration is given to the risk of abuse. The CPTPP contains a similar provision in Article 14.11, which starts by recognizing that each party may have its own regulatory requirements concerning the transfer of information by electronic means, before laying out provisions that prohibit the imposition of restrictions on the cross-border transfer of information by electronic means.17 The CPTPP and the Japan–US Digital Trade Agreement both insist that cross-border transfers of data should be free and unimpeded; the Japan–US Digital Trade Agreement perhaps goes slightly further in terms of the structure of the wording of the provision. 2.1.1.4 Article 12 Location of Computing Facilities Together with Article 11, Article 12 is one of two provisions that make up the core or the foundation of the provisions on data free flow. This is the provision that prohibits data localization. The purpose of prohibiting requirements for the use or location of computing facilities domestically within a party’s territory is to prohibit the requirement to store data within that country. Freeing companies from the need to establish computing facilities in both countries also helps to lighten the burden on businesses, and is therefore important in terms of promoting digital trade. 2.1.2

Provisions relating to ‘with Trust’

2.1.2.1 Article 17 Source Code This provision prohibits requirements for the transfer of, or access to, the source code of software or an algorithm expressed in that source code as a condition for the import or sale of software. This is important from the perspective of corporate confidentiality and protection of intellectual property. 2.1.2.2 Article 18 Interactive Computer Services Article 18 states that neither party shall adopt or maintain measures that treat a supplier or user of an interactive computer service as an information content provider in determining liability for harms related to information stored, processed, transmitted, distributed or made available by the service, except to the extent the supplier or user has, in whole or in part, created or developed the information. The article contains a side letter, clarifying that the two countries agreed that there are differences in the law governing interactive computer service suppliers’ liability between the two countries and that Japan does not require changes to its legislation. The concept of US law is basically the same as Article 18, but Japanese law differs in some aspects. Namely, Article 3 of Japan’s Act on the Limitation of Liability for Damages Chapter 5 of the GDPR establishes conditions for the transfer of data to a third country outside the EU from the perspective of protecting personal information. This means that, in principle, the consent of the individual is required for the transfer of personal data. 17

444  Research handbook on digital trade Table 25.1

Japan–US Digital Trade Agreement and CPTTP

Japan–US Digital Trade Agreement

Comprehensive and Progressive Agreement for Trans-Pacific Partnership

‘Data

Art. 7 Customs Duties

Art. 14.3 Customs Duties

Free

Art. 8 Non-Discriminatory Treatment

Art. 14.4 Non-Discriminatory Treatment

Flow’

Art. 11 Cross-Border Transfer of

Art. 14.11 Cross-Border Transfer of

Information

Information

Art. 12 Location of Computing Facilities

Art. 14.13 Location of Computing Facilities

‘with

Art. 17 Source Code

Art. 14.17 Source Code

Trust’

Art. 18 Interactive Computer Service

-

Art. 21 Cryptography

-

of Specified Telecommunications Service Providers and the Right to Demand Disclosure of Identification Information of the Sender states that, under certain conditions, interactive computer service suppliers are not liable for compensation for damages caused by the distribution of information that results in an infringement of rights, or for damages caused by measures taken to prevent such infringement. In the case of non-deletion, interactive computer service suppliers are exempt from liability unless (i) the provider knew that the right was being infringed or (ii) there are reasonable grounds to believe that the provider had knowledge of the infringement. Furthermore, in the case of deletion, interactive computer service suppliers are exempt from liability unless (i) when there are reasonable grounds to believe that the right has been unreasonably infringed, or (ii) when the sender has been asked whether he or she agrees to the deletion and no response has been received within seven days. 2.1.2.3 Article 21 Information Communication Technology Goods that Use Cryptography This provision prohibits requirements for the transfer of proprietary information relating to cryptography as a condition of the manufacture of an information and communication technology good that uses cryptography. As well as guarding corporate confidentiality, this provision ensures that cryptographic protection is not compromised. 2.2

Data Localization and Exceptions

In a narrow definition, data localization covers requirements for domestic data storage and domestic facilities.18 Data localization in this sense requires a business to locate computer facilities and related data within a country’s borders as a condition for carrying out business in that country. Regulations based on this requirement may include bans on the transfer of data or requirements affecting the retention and storage of data acquired and collected for the purpose of corporate activities. Articles 11, 12, 17 and 18 of the Japan–US Digital Trade Agreement address these areas and are all therefore relevant to both ‘Data Free Flow’ and ‘with Trust’. Data localization is 18 Data localization in the broad sense includes measures for protecting the privacy and personal information of the public. Yoshinori Abe, ‘Data Localization Measures and International Economic Law: How Do WTO and TPP/CPTPP Disciplines Apply to These Measures?’ (2021) Vol. 16 No. 5 Public Policy Review 1, 2.

Digital trade and economic security  445 essentially a straightforward restriction on the transfer of data, and addressing this issue is therefore key to making DFFT a reality and for achieving the goal of expanding digital trade. If data localization is comprehensive and data transfers are restricted, then the free distribution of data will not be possible. Further, if compulsory access by public institutions (government access) to information held by private-sector companies is used to secure an advantage for that country’s industries, these actions would not only violate WTO agreements but also defy the promotion of digital trade. Balancing data transparency and trade secrecy is an important and difficult challenge. While data transparency is a key principle, allowing a defence on trade secret grounds is seemingly necessary. For this reason, some level of government access, in criminal investigations for example, can be regarded as legitimate, provided proper and defined legal procedures are followed. However, the issue of what other circumstances should be regarded as just reasons for limiting the free flow of data is still open. Articles 3 and 4 of the Japan–US Digital Trade Agreement outline provisions for these exceptions.19 Article 3 covers general exceptions and incorporates into the agreement the provision of Article 14 of the GATS, allowing exceptions under specific circumstances: (a) necessary to protect public morals or to maintain public order; (b) necessary to protect human, animal or plant life or health; and (c) necessary to secure compliance with laws or regulations that are not inconsistent with the provisions of this Agreement. Article 4 lists security exceptions. Demands for data localization based on these perspectives will probably be seen as just and legitimate. However, at the same time, these demands are likely to have the effect of restricting digital trade. As far back as 1947, GATT contained provisions regarding security exceptions,20 and thus the possibility of passing provisions that limit free trade because of national security concerns has been recognized for more than half a century. GATS also has a similar provision in Article XIV bis. The provisions in Article XXI(b) of GATT make clear that measures taken for national security reasons do not contravene GATT, but it is fair to say that the reasons envisaged for these exceptions refer to an extremely limited set of circumstances. By contrast, the security exemptions in Article 4 of the Japan–US Digital Trade Agreement are as follows: Japan–US Digital Trade Agreement: Article 4 Security Exceptions Security Exceptions Nothing in this Agreement shall be construed to: (a) require a Party to furnish or allow access to any information the disclosure of which it determines to be contrary to its essential security interests; or (b) preclude a Party from applying measures that it considers necessary for the fulfilment of its obligations with respect to the maintenance or restoration of international peace or security, or the protection of its own essential security interests.

The scope for applying these exceptions is clearly broader than that of GATT Article XXI.21 The concept of national security that prevailed when Article XXI came into effect was based These exceptions are thought to basically operate in accord with the GATT–WTO framework. In matters unique to digital trade, they can, nevertheless, acquire more significance. It is assumed that they will deal with the data management practices that are already in place in various states. 20 See generally Rikako Watai, ‘Digital Trade and National Security Exceptions’ (2021) Vol. 27 Issue 2 International Trade Law & Regulation 119. 21 Since an equivalent provision is also included in the CPTPP, it seems likely that this kind of provision will become established as a security exception to economic partnership and free trade agreements. 19

446  Research handbook on digital trade exclusively on the assumption of a military confrontation, and the situations imagined in which Article XXI might apply were accordingly quite limited. It is not always appropriate to think of economic security in accordance with Article XXI because this traditional definition of national security does not apply to economic security. Further, Article 4(b) establishes that the final decision on national security measures is left to the country taking said actions. As a result, both parties can claim security exemptions with a considerable measure of political and policy latitude, which may prove a challenge to manage. If these exceptions were frequently applied, they would have the potential to damage the meaning of DFFT and restrict digital trade. On the other hand, since the era of GATT, it has been not uncommon for countries to draft rules among states to reserve policy space for national security issues; it is further likely that such provisions will also be necessary from the perspective of ensuring that the rules operate ‘with Trust’.

3.

DATA AND DIGITAL TRADE

The dividing line between national security and protectionism is especially uncertain, requiring careful and cautious decision-making in any situation where a party might consider applying the security exceptions in Article 4 of the US–Japan Digital Trade Agreement. In the rulemaking process, it is necessary to clarify what is meant by national security and economic security. That said, the concepts of public order and morals have not been clearly defined, and it is undeniable that the concept of obscenity, as an example, has changed over time. The criticism that the concept of national security is not clear is therefore not always appropriate. National security is a value that lies at the foundation of free trade, and what should be considered is how to ensure predictability and due process. Although national security is a broad concept, it is only relatively recently that data has been explicitly understood to be included in national security. A specific example of this is the 2018 amendment to the US law on the regulation of foreign direct investment. In 1988, a regulation on foreign direct investment was introduced, granting the US president authority to stop investments by foreign investors from the viewpoint of national security.22 At the time that the regulation came into effect, the transactions that were subject to screening were investments that might lead to ‘foreign control’ of a US business.23 Initially, transactions were voluntarily submitted for review via a notice.24 The regulations have been revised several times, and the current legislation is the Foreign Investment Risk Review Modernization Act of 2018 (FIRRMA).25 Issues involving data protection were part of the reason that the legislation was revised. 22 Exon-Florio Amendment to the Omnibus Trade and Competitiveness Act of 1988, Pub. L. No. 100–418, 102 Stat. 1107, made permanent law by Section 8 of Pub. L. No. 102-99, 105 Stat. 487 (50 U.S.C. app. 2170) amended by Section 837 of the National Defense Authorization Act for the Fiscal Year 1993, Pub. L. No. 102–484, 106 Stat. 2315, 2463. See generally Rikako Watai, ‘Regulation of Foreign Direct Investment in United States’ in Dennis Campbell (ed), The Comparative Yearbook of International Business Vol. 38 (Kluwer Law International 2016). 23 50 U.S.C. 4565(a)(3), (b)(1)(B). 24 Ibid. 25 John S. McCain National Defense Authorization Act for Fiscal Year 2019, Pub. L. No. 115-232, 132 Stat. 1636, 2174, 50 U.S.C.4565. See generally Rikako Watai, ‘National Security Review of Foreign

Digital trade and economic security  447 In 2016, the German digital mapping company HERE Technologies announced that it had received an offer for an investment from Tencent, the Chinese internet giant, but this investment was abandoned in September 2017 when the proposal failed to get approval from the US regulatory authority.26 HERE has a large share of the market for car navigation systems and rivals Google in the digital mapping field, and US officials were likely concerned that if a Chinese entity became involved in running the company, it would easily gain access to the company’s mapping data. In January 2018, a planned 1.2 billion US dollar acquisition of MoneyGram, an American leader in global payment services, by Ant Financial, an affiliate of China’s Alibaba Group, was abandoned for similar reasons when the acquisition failed to get approval from US regulatory authorities.27 The reasons for not getting approved were unclear, but it is likely that there were concerns that MoneyGram’s accumulated payment data could be accessed by Chinese companies. MoneyGram’s services can be used easily at pharmacy chains and other places with operations nationwide, and the company is established as a familiar money remittance service for global payments. Many members of the US Armed Forces (USAF) use MoneyGram’s services, and concerns were raised in Congress that the deal might pose a national security risk, possibly revealing the position of USAF personnel or information on individuals’ economic circumstances.28 Following these events, when the law was revised in 2018, companies handling sensitive personal data were considered to be significant to US national security and were made subject to review, even in cases where the amount of foreign investment was small and did not rise to the level of foreign control of the company.29 The revised law also introduced a requirement for a mandatory declaration of any plans that would involve a foreign entity acquiring a substantial interest in a US business handling sensitive personal data or that might result in foreign control of the company. In comparison to the prior voluntary notification policy, this mandatory declaration represents a significant change.30 Since then, the United States has continued to increase the stringency of its screenings for foreign investments in entities dealing with data. Presidential orders issued after FIRRMA came into effect in August 2018 covered both the planned acquisition of US hotel guest-centric app developer StayNTouch by a Chinese government-owned company in March 202031 and the acquisition of Musical.ly (TikTok),

Direct Investment: Recent Developments in the United States and Japan’ (2020) International Trade Law & Regulation Vol. 26 No. 2 126, 127–8. 26 Yuan Yang, ‘Chinese Bid for Mapping Company Falls at US Hurdle’ Financial Times (27 September 2017) www​.ft​.com/​content/​6f0e519c​-a33f​-11e7​-9e4f​-7f5e6a7c98a2 accessed 1 May 2022. 27 Greg Roumeliotis, ‘U.S. Blocks MoneyGram Sale to Chinas Ant Financial on National Security Concerns’ Reuters (3 January 2018) www​.reuters​.com/​article/​us​-moneygram​-intl​-m​-a​-ant​-financial/​ u​-s​-blocks​-moneygram​-sale​-to​-chinas​-ant​-financial​-on​-national​-security​-concerns​-idUSKBN1ER1R7 accessed 1 May 2022. 28 Josh Rogin, ‘China’s Jack Ma Has Penetrated the Trump Administration – And He Knows What He Wants’, Washington Post 1 (19 July 2017), www​.washingtonpost​.com/​news/​josh​-rogin/​wp/​2017/​07/​ 19/​chinas​-jack​-ma​-has​-penetrated​-the​-trump​-administration​-and​-he​-knows​-what​-he​-wants/​?utm​_term​=​ .596d2ae70c4b accessed 1 May 2022. 29 50 U.S.C. 4565 (a)(4)(B)(ii), 31 C.F.R. 800.248. 30 31 C.F.R. 800.401(b)(c). 31 Order Regarding the Acquisition of StayNTouch, Inc. by Beijing Shiji Information Technology Co., Ltd. 85 Fed. Reg. 13719 (6 March 2020).

448  Research handbook on digital trade a social media company, by a Chinese company, also in 2020.32 Both of these planned acquisitions were not approved due to concerns over the misuse of personal data. Meanwhile in Japan, a case emerged that challenged the relationship between data and security. LINE is a messaging app that allows people to exchange messages and make free voice and video calls with other LINE subscribers.33 In February 2021, it was discovered that from August 2018 to February 2021, employees of a Chinese affiliate of LINE had access to personal data on servers in China.34 The accessible personal information included names and telephone numbers as well as messages and images saved by the user.35 In response to this issue, a third-party committee set up by LINE’s parent company published a report stating that although there was no actual data leakage, the data management was extremely inappropriate.36 Chinese law establishes an obligation that ‘where a public security organ or national security organ needs to obtain data for the sake of national security’, organizations and individuals are required to cooperate.37 The LINE committee report emphasized the need for a data governance system that takes economic security into consideration and that LINE’s management instruments in the area of economic security, including the risk of government access, were insufficient.38 Further, the report recommended that instead of developing LINE’s own system, the parent company should establish a system to use its knowledge throughout the entire group, as it is crucial for group companies to deal with economic security in an integrated way.39 The LINE incident provided proof that data handling, particularly from the perspective of economic security, is a national security issue. Therefore, it is necessary to check Japan’s legal framework for economic security issues that could impact digital trade.

32 Order Regarding the Acquisition of Musical.ly by ByteDance Ltd. 85 Fed. Reg. 51297 (14 August 2020). 33 LINE https://​line​.me/​en/​accessed 1 May 2022. 34 Osamu Tsukimori, ‘Line data scandal highlights the perils of storing information across borders’ Japan Times (24 March 2021) www​.japantimes​.co​.jp/​news/​2021/​03/​24/​business/​corporate​-business/​line​ -cross​-border​-data/​accessed 1 May 2022. 35 Kenji Minemura and Toshiya Obu, ‘Personal Data of Millions of Line Users Accessed by Affiliates in China’ Asahi Shimbun (17 March 2021) www​.asahi​.com/​ajw/​articles/​14276271 accessed 1 May 2022. 36 Z Holdings Corporation, Final Report Received from the Special Advisory Committee on Global Data Governance, and Future Efforts to Strengthen Group Governance (Japanese Version, October 2021). An English summary of the content of the report can be found in Final Report Received from the Special Advisory Committee on Global Data Governance, and Future Efforts to Strengthen Group Governance www​.z​-holdings​.co​.jp/​en/​news/​press​-releases/​2021/​1018/​ accessed 1 May 2022. 37 Data Security Law of the People’s Republic of China, Article 35. An official English translation is available from the National People's Congress (NPC) of the People's Republic of China website: www​.npc​.gov​.cn/​englishnpc/​c23934/​202112/​1a​bd88297889​46ecab270e​469b13c39c​.shtml accessed 1 May 2022. 38 Z Holdings Corporation, supra note 36 at 80. 39 Ibid.

Digital trade and economic security  449

4.

ECONOMIC SECURITY AND DIGITAL TRADE

4.1

Promotion of Economic Security in Japan

No clear definition of economic security has been established in Japan, but discussions began based on a proposed meaning by the Industrial Structure Council’s Special Sub-Committee for Economic Security Issues in 1982 that stated: ‘Economic security means defending the country’s economy from major threats arising internationally, chiefly by employing economic means’.40 The challenges in economic security are dealing with vulnerabilities in the economy and reducing dependence on other countries. When there is a high level of dependence on other countries, it is necessary to consider stockpiling or manufacturing alternative products. In fact, when China, which controlled 90 per cent of the world’s production of rare earth metals by volume, stopped exports to Japan in 2010, the development of alternatives via collaboration between the government and the private sector enabled Japan to reduce its rare earth imports by half.41 From the perspective of economic security, every country must consider dealing with partners it can trust and must also aim for an ideal balance between economic rationality with national security. Furthermore, some aspects of economic security have the potential to disrupt the rules that have been formed under GATT and the WTO. This is because although attempting to achieve political aims by economic means goes against the WTO Agreement, it can be argued that such steps are necessary when considering national security. The GATT–WTO system formed an international economic order premised on the Western free-trade system, but the value of that system is no longer a premise shared by all member countries today. The difficulty of resolving such differences in values is one possible reason why dispute resolution at the WTO has failed to function properly; the shift to free-trade agreements among regions can also be explained from the same perspective. The importance of international agreements and laws of each country has expanded more than ever before as the conventional structure of international trade law has become less effective. 4.2

Economic Security Promotion Act

As DFFT advocates the free distribution of data, a legal framework has also been put in place in Japan for national security. To understand the impact that the latest trends in Japanese economic security have on digital trade, it is first necessary to outline the legislation that is at the centre of this issue. Prime Minister Fumio Kishida established a new ministerial post for economic security after taking office in the autumn of 2021. As with the concept of national security itself, no fixed definitions exist in Japan for economic security.42 In December 2020, the governing

Industrial Structure Council, ‘Keizai anzenhosho no kakuritsu wo mezashite’ (Toward the establishment of economic security, 28 April 1982) 14, 27. 41 Colonel Charles J. Butler, ‘Rare Earth Elements: China’s Monopoly and Implications for the U.S. National Security’ (2014) Vol. 38 Fletcher F. World Aff. 23, 34. 42 In the sense that a state aims to achieve strategic goals by using economic means to exert an influence over other countries, it might be said to overlap with the US concept of ‘economic statecraft’ in some respects. 40

450  Research handbook on digital trade Liberal Democratic Party defined economic security as ‘ensuring Japan’s independence, survival, and prosperity from an economic perspective’ and pointed to the need to strengthen ‘strategic core industries’ such as energy, telecommunications, food, medical care, finance and distribution as well as the need for Japan to become a strategically indispensable presence in the global community.43 In June 2021, the Japanese government’s Basic Policy on Economic and Fiscal Management and Reform 2021 outlined the strategic direction for economic security as follows: ‘The Government will expand and deepen cooperation with like-minded countries under the international order based on fundamental values and rules, and will seek to ensure Japan’s self-determination and acquire advantages for our country’, noting that ‘from this perspective, the Government will implement concrete measures and policies to strengthen its efforts to identify, protect, and develop critical technologies and to enhance the resilience of essential industries’.44 The Kishida cabinet presented draft legislation to promote economic security in the Ordinary Session of the Diet in 2022. Economic security is a measure to defend the economic and social order of a nation and refers to creating the ability to withstand economic pressure from another nation. In other words, economic security aims to achieve strategic autonomy and strategic indispensability. For this reason, the Bill on the Promotion of Ensuring Security through the Integrated Implementation of Economic Measures (Economic Security Promotion Act) was enacted in May 2022. Economic rationality is an important value of the principles of free trade, dictating that when there is a need for imports, the rational choice is to procure all one’s needs from the vendor selling at the cheapest price. Dealing only with a single company or a single country means however that the supply will be interrupted if it becomes impossible to import from that source. To avoid this risk, it is preferable to secure several sources of imports. While strategic autonomy means defending the economic order, it is necessary to consider the costs that countries are prepared to bear to achieve this aim. In concluding the Regional Comprehensive Economic Partnership (RCEP) Agreement, for example, it would be possible to include a wide range of security exceptions, but if taken too far, this will result in undermining free trade. If a country can increase other countries’ dependence on certain technologies or products that it produces, it can make itself indispensable to those other countries. In comparison with strategic autonomy, strategic indispensability is less about defending the economic order and represents more of an aggressive stance. This approach might be straightforward for state-owned enterprises to adopt, but in countries with an industrial structure like Japan’s, where the national government and private-sector companies are not integrated as a single whole, there is concern that companies may be forced to adopt the policies of the government

43 Strategic Headquarters on the Creation of a New International Order (Policy Research Council, Liberal Democratic Party of Japan), Teigen: Keizai anzen hosho senryaku no sakutei ni mukete (Recommendations: Toward Developing Japan’s ‘Economic Security Strategy’, December 2020) 3, 8–10. Provisional translation available from https://​jimin​.jp​-east​-2​.storage​.api​.nifcloud​.com/​pdf/​news/​ policy/​201021​_5​.pdf accessed 1 May 2022. 44 Cabinet decision, Basic Policy on Economic and Fiscal Management and Reform 2021 (English version) Four Driving Forces that Open the Way to the Future of Japan: Green, Digital, Creation of Vibrant Local Regions, Measures against Declining Birthrate (18 June 2021) 28. The provisional translation available from www5​.cao​.go​.jp/​keizai​-shimon/​kaigi/​cabinet/​2021/​2021​_basicpolicies​_en​.pdf accessed 1 May 2022.

Digital trade and economic security  451

Figure 25.1

Structure of Japan’s Economic Security Promotion Act

of the time. For this reason, it is necessary to carefully consider the strategic direction to make sure that there is no interference with corporate activities. As background for enacting the law, Article 1 of the Economic Security Promotion Act first points out that the complexity of the international climate and changes in the socioeconomic structure have increased the importance of preventing acts that may harm the security of the nation and its citizens conducted with regard to economic activities that ensure national security. Therefore, the purpose of the Economic Security Promotion Act was thus to promote economic measures related to ensuring security in a comprehensive and effective manner. The Act specifies the overarching details to the order and serves as the basis for future rules. The Economic Security Promotion Act incorporates four pillars as mentioned in Article 1: strengthening supply chains, ensuring the safety and reliability of critical infrastructure, promoting public–private cooperation to develop key technologies and providing for nondisclosure of sensitive patents. Of these four pillars, supply chains and critical infrastructure can be regarded as passive or ‘defensive’ aspects of economic security. Conversely, public–private cooperation on technology and nondisclosure of patents represents attempts to secure a position of superiority for one’s own country and may therefore be seen as part of an active or ‘offensive’ position. 4.2.1 Supply chains A supply chain refers to all the transactions and business involved in manufacturing a product, from procuring raw materials and purchasing components to making sales. Reducing costs and making business management more efficient are important priorities for companies, and the role of digital trade in supply chains is likely to increase in the future. Once supply chains develop more on a global level, it is reasonable to assume that foreign-owned companies will be included in these supply chains, and it is possible that, as a result, countries with markedly

452  Research handbook on digital trade different political and economic systems from Japan’s will become part of these supply chains.45 As dependence on foreign countries for resources that are important for citizens’ lifestyles increases, the risk of an interruption to the supply also goes up. From this perspective, it is important to overcome supply chain vulnerabilities and make supply chains more robust and resilient. Since digital trade is closely related to data flows, it is possible that confidential information or information related to technology, as well as other types of data, including personal information, could become vulnerable to access by foreign companies or governments within the supply chain. Deciding whether or not to limit access of this kind by imposing restrictions is likely to become an important issue. 4.2.2 Critical infrastructure Fourteen sectors of Japan’s economy are designated as critical infrastructure: electricity, gas, oil, water, railways, trucking, international marine freight, aviation, airports, telecommunications, broadcasting, postal services, financial businesses and credit cards. The Cybersecurity Policy for CIP (4th ed.), published by the government’s National Center of Incident Readiness and Strategy for Cybersecurity in advance of the Economic Security Promotion Act, also designates 14 sectors as critical infrastructure; 11 of these 14 are the same as those included in the legislation.46 This demonstrates that the challenges for critical infrastructure within economic security are closely related to data management. 4.2.3 Key technologies The third pillar of the Economic Security Promotion Act is aimed at strengthening the technology base. Specifically, the public and private sectors are to cooperate and conduct research and development on advanced and important technologies in the fields of space, oceans, quantum physics and artificial intelligence. This effort includes not only developing high-tech technologies but also preventing their leakage, thereby protecting dual-use technologies. Specifically, a council for research and development will be established, and participants will be subject to the same confidentiality obligations as national public officials in order to share information and not disclose any secrets. Violations of this rule will result in imprisonment of up to one year or a fine of up to 500,000 yen; this severity indicates that data will be carefully managed. 4.2.4 Secret patents The following cases are considered for secret patents: single-use technologies that lead to the development of nuclear weapons and are extremely sensitive to Japan’s national security; dual-use technologies that are the result of government-funded projects and technologies developed for defence; and dual-use technologies that the applicant has agreed not to disclose.47

45 The COVID-19 pandemic exposed problems with supply chains, especially supply chains for medical supplies. 46 The sectors not included under the Economic Security Promotion Act are government and administration services, medical and chemical industries. 47 Advisory panel on Economic Security Legislation, Keizai Anzenhosho ni Kansuru Teigen (Proposal on Economic Security Legislation) (2022) 47–8.

Digital trade and economic security  453 Because secret patents are information-preservation measures, digital trade involving such technologies will naturally be affected. The Economic Security Promotion Act has an element of data management in each of its four pillars. While the details must await the passage of regulations that will eventually be established, it is a natural assumption that the law will have to operate in a way that does not conflict with the GATT–WTO regime in terms of the impact it will have on the relationship with digital trade. 4.3

Economic Security and Digital Trade

Because economic security involves taking measures necessary for a country’s economic autonomy for strategic reasons, it will inevitably have an impact on digital trade. The relationship between DFFT and cybersecurity is important in facilitating digital trade. Although the Economic Security Promotion Act does not mention cybersecurity, it will certainly be an issue considered in the implementation of the law. In fact, though it is not yet clear what will be designated as critical goods in relation to the supply chain, there is little doubt that these items will include semiconductors. In addition, for critical infrastructure, it is a natural assumption that cybersecurity must be addressed. The Economic Security Promotion Act may be seen as playing a role in laying the groundwork for the active development of the DFFT initiative.

5. CONCLUSION National security is an essential value at the foundation of digital trade. To use a computer metaphor, national security can be compared with an operating system, and in an ideal situation, policies that lead to the smooth development of digital trade are applications that run on that operating system. See Figure 25.2.

Figure 25.2

National security and digital trade

While national security is foundational to digital trade, its definition may never be entirely clear. National security is expected to continue to evolve, just as it has grown to include military and homeland security aspects as well as those related to data protection. For this reason, the issue should not be about revealing the content of national security, but rather whether or not regulations can assure transparency and foreseeability. The WTO’s rulemaking on digital

454  Research handbook on digital trade trade, the Japan–US Digital Trade Agreement, and Japan’s Economic Security Promotion Act all share the common challenge of safeguarding transparency and foreseeability for reasons of national security. At the G7 Digital and Technology Ministers’ Meeting in July 2021, agreement was reached on a roadmap for cooperation on DFFT.48 Because the 2023 G7 summit is due to take place in Japan, initiatives to secure the status of DFFT as an international system, along with the efforts taken by the WTO, are likely to increase. In drawing up rules for digital trade, the positions of each country differ, but efforts to bridge these gaps and avoid fragmentation must continue. The very values of preventing fragmentation in the cyber world and developing digital trade are ideally already shared by countries around the world. For this reason, even if the WTO’s rulemaking is not detailed, it will still be possible to develop digital trade based on the DFFT concept among countries with similar values. As an advocate for adopting DFFT, Japan has been examining the framework for the cross-border transfer of data, taking into account the specific circumstances of each country, and has continued to study the policy direction in order to drive this proposal in the future. A February 2022 report from the Expert Group on Data Free Flow with Trust identified five core areas for DFFT to adopt: transparency; technology and standardisation; interoperability; complementarity; and implementation.49 To ensure transparency, it has recently been pointed out that information sharing should be further promoted through international cooperation, and that standardisation, collaboration and involvement from industry for technology should be sought to protect privacy and security.50 Next, the Expert Group’s report pointed out the need to investigate and study policy options for interoperability, based on the differences among the systems of each country.51 Furthermore, regarding complementarity with related systems, the report recommended harmonisation of existing rules and discussions on data handling. Finally, with respect to the implementation of the DFFT, the report noted the need to consider ways to cooperate with endorsed countries in promoting DFFT-friendly policies. Specifically, in order to ensure transparency, one proposal is that each country will be notified of legal reforms and that each country will be reviewed for related efforts.52 It is not yet clear which concrete steps will be taken to achieve DFFT. In the meantime, it is to be desired that in the future, the approach to the free use and application of data put forward in the Japan–US Digital Trade Agreement will continue to find wider acceptance and will eventually be adopted as part of WTO rules on digital trade. The discussion will be watched closely as it evolves, looking specifically at measures taken for economic security. The hope is that these measures will develop within the framework of international cooperation and in a form that prevents them from becoming an obstruction to digital trade.

48 The roadmap included four main points: an assessment of evidence on the impact of data localization, a comparative analysis of each country’s policies on cross-border data transfers, formulation of guidelines for trusted government access and steps to accelerate the development of mutually acceptable sharing of data. 49 Ministry of Economy, Trade and Industry, Expert Group on Data Free Flow with Trust, Interim Report of the Expert Group on Date Free Flow with Trust (28 February 2022). www​.meti​.go​.jp/​ shingikai/​mono​_info​_service/​data​_ekkyo​_iten/​pdf/​20220228​_2e​.pdf accessed 1 May 2022. 50 Ibid at 45–6. 51 Ibid at 46–57. 52 Ibid at 47.

26. Digital trade and intellectual property Marc D. Mimler

1. INTRODUCTION The Covid-19 pandemic has had profound effects on how we interact as human beings. Due to contact and travel restrictions, the ways in which we work together and collaborate with one another have undergone fundamental changes. During lockdowns, work was conducted from home, where possible, by connecting co-workers through digital platforms. Schools and universities also had to rethink their pedagogy and have embraced digital technologies in their curricula. While countries more and more appear to be reverting to pre-Covid habits, the Covid-19 pandemic highlighted the power that the connectivity of the internet can have and showcased the abilities of digital trade, distribution and digital business models. Digitisation has, for instance, changed the way goods are conceived, produced, transferred and consumed1 and has been embraced by many successful companies. Music and movies are less and less consumed by purchasing CDs or DVDs but are increasingly being delivered to end-consumers by the internet. The same applies to video games.2 The physical or virtual goods or items traded digitally are often subject to IP protection. Modern intellectual property law was, however, conceived in the nineteenth century in an analogue world. Digitisation, in conjunction with the ever-increasing and ever-faster connectivity provided by the internet, meant that traditional business models built on analogue IP, as well as the rights themselves, needed to adapt. The internet is the ‘biggest copying machine’, and copied content can be shared in an instant around the globe with a push of a button. This, of course, has challenged IP enforcement as infringers could be based in various jurisdictions, often hiding under the veil of anonymity which the internet might provide. Some IP business models accordingly struggled and perished, while others were able to adapt as fully able to embrace the potential given by the internet. The response by legislators and courts to the ever-increasing digitisation and dissemination was often to widen IP protection and to provide new means to enforce it. This, however, might negatively impact on the flow of digital trade. This chapter will analyse this regulatory dilemma by outlining the current state of affairs of the IP framework for digital trade with a view on the situation in the European Union (EU) and the United States. First, it will trace the development from ‘unshackling’ IP content from its tangible carriers to its current digital manifestations and the problems this entailed. Second,

1 Javier López González and Janos Ferencz, ‘Digital Trade and Market Openness’ (OECD Trade Policy Papers No.217, OECD Publishing, 2018) 9 dx​.doi​.org/​10​.1787/​1bd89c9a​-en accessed 19 July 2022. 2 These were some of the most popular activities for internet users in the EU in 2019: Nadia Iacob and Felice Simonelli, ‘How to Fully Reap the Benefits of the Internal Market for E-Commerce – New Economic Opportunities and Challenges for Digital Services 20 Years after the Adoption of the E-Commerce Directive’ (Study for the IMCO Committee, 2020), p.23. www​.europarl​.europa​.eu/​ RegData/​etudes/​STUD/​2020/​648801/​IPOL​_STU(2020)648801​_EN​.pdf accessed 19 July 2022.

455

456  Research handbook on digital trade as digital trade is global, the ability of IP rights as tradeable commodities and the international legal frameworks enabling this will be discussed. The chapter will then outline IP’s ambiguous operation within digital trade as on the one side, enabling trade in digital IP content, while also setting stumbling blocks. The chapter will finally discuss some current and emerging issues surrounding IP and digital trade, such as digital exhaustion and intermediary liability and also looks at the potential impact of new technologies, such as blockchain technology and non-fungible tokens (NFTs), as well as 3D printing technology.

2.

SETTING THE SCENE: INTELLECTUAL PROPERTY IN THE CONTEXT OF TRADE AND DIGITISATION

The World Intellectual Property Organization (WIPO), a United Nations (UN) specialised agency, defines intellectual property as ‘creations of the mind, such as inventions; literary and artistic works; designs; and symbols, names and images used in commerce’.3 These creations of the mind are protected by different types of IP rights. Copyright, also referred to as author’s rights in many civil law jurisdictions, protects literary and artistic works but can also extend to subject matter, such as computer programs and databases. Patents are granted for new and non-obvious inventions, while trade marks protect signs used in commerce and designs4 protect the appearance of a particular product. Other areas of IP include the protection of geographical indications and plant varieties. While not providing protection through providing exclusive rights, IP protection is complemented by trade secret protection and the various forms of unfair competition law. The different variations of IP rights all share the common theme that the protected object is something intangible. IP rights are formed as exclusive rights. They provide their owners with exclusivity in relation to a particular item with the use of legal mechanisms to stop third parties from using the particular subject matter covered by IP rights without their authorisation. IP’s nature as exclusive rights is usually explained by their purpose in stimulating creativity and inventive behaviour,5 but also to reward inventors and authors for the contributions resulting from such activities.6 Economic theory suggests that the exclusivity provided to the right holder serves to eliminate the so-called free riding problem. This situation occurs in relation to public goods, such as information, which are deemed to be non-excludable and non-rivalrous. Non-excludable means that a good cannot be exhausted by someone’s use but is able to be used simultaneously by others.7 Being a non-rivalrous good entails that it is very difficult, if not impossible, to exclude a third party from using that particular good. IP rights, thus, provide right holders with a means to counteract unauthorised uses of their creations or inventions by free riders. They disincentivise others to infringe the protected items due to the legal sanctions World Intellectual Property Organization, ‘What Is Intellectual Property?’ www​.wipo​.int/​about​ -ip/​en/​accessed 19 July 2022. 4 Design patents under US law. 5 ‘Patents are especially susceptible to the economic argument that industrial innovation requires incentivisation’: Graham Dutfield and Uma Suthersanen, Global Intellectual Property Law (2nd edn, Edward Elgar Publishing 2020) 36. 6 Mikhalien du Bois, ‘Justificatory Theories for Intellectual Property Viewed through the Constitutional Prism’ [2018] P.E.R./P.E.L.J. 19. 7 Robert B. Cooter and Thomas Ulen, Law and Economics (5th edn, Pearson 2004) 120. 3

Digital trade and intellectual property  457 available, such as damages and injunctive relief. IP protection, therefore, is aimed at overcoming a market failure due to creators not feeling compelled to create in the absence of any form of protection or legal intervention8 by addressing the free riding problem. It should be noted here that this economic explanation relates to some IP rights more than others. Copyright, for instance, is usually also explained by deontological rationales in protecting the works as an extension of the individual,9 while the existence of trade mark protection is usually also explained by law and economics but for different reasons.10 2.1

Intellectual Property and Digitisation

Many digital items exchanged in digital trade consist of information contained in computer-readable bits and bytes. Of these digital items, many can be the subject matter of IP.11 The concept and doctrinal underpinnings of contemporary IP protection, however, predate the digital age. They evolved in relation to the protected subject matter of IP being manifested in some form of physical carrier, such as a book or machine. Technological change, however, meant that these physical carriers could be reproduced faster and disseminated more widely, which posed challenges to right holders. Photocopiers and video cassette recorders, for instance, made the reproductions of books, articles and films easier. Technology enabled new ways of distribution of content which led to the emergence of new media, such as film and radio. These developments have often been disruptive in relation to business models that were built on IP protection and have often led to the creation of new rights or the amendment of existing ones.12 The emergence of digitisation and the inception of the internet, however, had the most profound impact. While digitisation decoupled the protected content from its tangible carriers by creating perfect copies, the global reach of the internet meant that it could be disseminated worldwide, without limits, and difficult to trace.13 This has been the cardinal issue for IP enforcement.

Sacha Wunsch-Vincent, ‘The Economics of Copyright and the Internet’ in Johannes M. Bauer and Michael Latzer (eds), Handbook on the Economics of the Internet (Edward Elgar Publishing 2016) 229–46, 230. 9 Graham Dutfield and Uma Suthersanen, Global Intellectual Property Law (2nd edn, Edward Elgar Publishing 2020) 44–9. 10 ‘The primary reasons for the existence and protection of trademarks are that (1) they facilitate and enhance consumer decisions and (2) they create incentives for firms to produce products of desirable qualities even when these are not observable before purchase.’ Nicholas S. Economides, ‘The Economics of Trademarks’ [1988] Trademark Reporter 523–39, 526. 11 Arturo Ancona, ‘Intellectual Property and E-Commerce’ (WIPO-WASME Special Program on Practical IP Issues, Geneva, 6–9 October 2003) 2 www​.wipo​.int/​edocs/​mdocs/​sme/​en/​wipo​_wasme​_ipr​ _ge​_03/​wipo​_wasme​_ipr​_ge​_03​_13​-main1​.pdf accessed 19 July 2022. 12 New business models emerged based on how content could be disseminated and consumed, while others have perished. This phenomenon has been described by the Austrian economist Joseph Schumpeter as ‘creative destruction’: Joseph A. Schumpeter, Capitalism, Socialism and Democracy (Routledge, 1994) 681. 13 Arturo Ancona, ‘Intellectual Property and E-Commerce’ (WIPO-WASME Special Program on Practical IP Issues, Geneva, 6–9 October 2003) 2 www​.wipo​.int/​edocs/​mdocs/​sme/​en/​wipo​_wasme​_ipr​ _ge​_03/​wipo​_wasme​_ipr​_ge​_03​_13​-main1​.pdf accessed 19 July 2022. 8

458  Research handbook on digital trade The so-called WIPO Internet Treaties14 can be seen as a regulatory response in addressing digitisation as an emerging issue for copyright law and its exploitation in the digital age. These treaties have been implemented, for instance, in the United States with the Digital Millennium Copyright Act (DMCA),15 and within the European Union by the Information Society Directive (InfoSoc Directive).16 They clarified that the digitisation of copyright-protected works would constitute an infringement of the reproduction right where it took place without authorisation of the right holder.17 They also entailed an expansion of the scope of existing rights while also adding new ones. The introduction of the making available right,18 which allowed right holders to extend their exclusive rights to uses over the internet, showcases this. It allowed copyright to be exploited in different ways, which has had a profound effect on business models by untying protected content, such as films, computer programs or video games, from its tangible carriers (such as video cassettes, floppy discs and CDs and CD-ROMs or DVDs). The retail models which were built on the sale or renting of physical carriers of IP content became obsolete with increasing digital distribution. This has meant that physical locations, such as warehouses or video rental shops,19 as well as the logistics in relation to transporting physical carriers, are becoming less and less important.20 Another measure provided by the WIPO Internet Treaties surrounded the protection of Digital Rights Management (DRM) and Technological Protection Measures (TPMs) and sought to protect against the circumvention of these ‘digital locks’.21 While the WIPO Treaties have provided templates for signatory states for substantive provisions of how copyright could deal with digitisation and the internet, they were rather vague in relation to one important aspect of intellectual property protection: enforcement.22 Equally, the most comprehensive multilateral agreement worldwide on intellectual property, the Agreement of Trade-Related Aspects of Intellectual Property Rights (TRIPS) – an annex to the WTO Agreement, which will be further explained below – does not provide precise measures on enforcement. There have, of course, been various national or regional approaches in addressing online copyright infringement. Some have attempted to deter individuals from copyright infringement, while others focused on intermediaries, such as internet service providers or ‘platforms’, as having a role in minimising copyright infringement occurring over their

14 The WIPO Internet Treaties consist of the WIPO Copyright Treaty (WCT) which covers the protection for authors of literary and artistic works and the WIPO Performances and Phonogram Treaty (WPPT) which deals with the protection for authors rights of performers and phonogram producers. 15 Digital Millennium Copyright Act, Title I. 16 Recital 15, Directive 2001/29 on the harmonisation of certain aspects of copyright and related rights in the information society [2001] OJ 2001 L 167/10. 17 Agreed statements concerning Article 1(4) WCT; Agreed statements concerning Articles 7, 11 and 16 WPPT. 18 Article 8 WCT (Right of Communication to the Public); Article 10 WPPT (Right of Making Available of Fixed Performances), Article 14 WPPT (Right of Making Available of Phonograms). 19 See the video of the rise and fall of Blockbuster Video rental shops in the United States over the years, with a peak in 2005 and now only one remaining in Bend Oregon, United States - Eric Diaz, ‘This Visualization shows the Rise and Fall of Blockbuster Video’ (The Nerdist, 30 June 2020) nerdist​.com/​ article/​blockbuster​-video​-rise​-and​-fall​-visualization/​ accessed 19 July 2022. 20 Andrew Murray, Information Technology Law: The Law & Society (4th edn, Oxford University Press 2019) 274–5. 21 Article 11 WCT, Article 18 WPPT. 22 See Article 14 WCT; Article 23 WPPT.

Digital trade and intellectual property  459 networks. An effort to provide for a comprehensive international legal framework addressing online copyright infringement can be seen in the Anti-Counterfeiting Trade Agreement (ACTA), a plurilateral treaty23 which sought to improve ‘global standards for the enforcement of IPRs [in order] to more effectively combat trade in counterfeit and pirated goods’.24 While the final treaty text was adopted in December 2010, it faced criticisms due to the secrecy surrounding its negotiations and its far reaching substantial rules25 and has led to public protests in several countries.26 The nail in its coffin was its rejection by the European Parliament by 478 votes to 39.27 The failure of ACTA thus offers a cautionary tale for legislators regarding not overstretching IP protection. 2.2

Intellectual Property and Trade

IP rights are territorial. They are created and protected in the relevant jurisdiction and do not extend beyond the territory of protection.28 This is, on the one hand, easily explained due to state sovereignty extending to the boundaries of that particular jurisdiction. Territoriality of IP protection, on the other hand, also highlights that IP rights and their inception, historical development and scope can be explained by the policy goals to be achieved within that jurisdiction.29 Thus, the territoriality principle can to a certain degree explain the historically divergent approaches and different levels and forms of protection provided by IP. Their rationale to incentive or reward creative enterprise which was explained above suggests that legislators sought to devise laws suiting the level of industrialisation and development of their national industry.30 This also means that from a policy perspective, sovereign nations are able to devise IP legislation as it suits their current economic development. From a trade perspective, diverging rules with regards to IP are not conducive to trade. This is easily explained: whether a particular object traded beyond borders is protected in one jurisdiction but not within the other fundamentally impacts on the outcome of a commercial deal and can lead to uncertainties between the parties. Trade law regards such diverging laws

23 The treaty was negotiated between the USA, Japan and the EU as well as its Member States in their own national capacity, Canada, Australia, Mexico, Morocco, New Zealand, the Republic of Korea, Singapore and Switzerland. 24 Rita Matulionyte, ‘ACTA’s Digital Chapter: Remaining Concerns and What Can Be Done’ [2011] Queen Mary Journal of Intellectual Property 248–71. 25 Michael Geist, ‘ACTA’s State of Play: Looking Beyond Transparency’ [2011] American University International Law Review 543–58. 26 Dave Lee, ‘Acta Protests: Thousands Take to Streets across Europe’ (BBC, 8 March 2012) www​ .bbc​.co​.uk/​news/​technology​-16999497 accessed 19 July 2022. 27 Benjamin Farrand, ‘Lobbying and Lawmaking in the European Union: The Development of Copyright Law and the Rejection of the Anti–Counterfeiting Trade Agreement’ [2015] Oxford Journal of Legal Studies 487–514, 511. 28 ‘IP rights tend to be territorial they only give protection in the countries where they are granted or registered.’ UK Intellectual Property Office, ‘IP Basics’ www​.gov​.uk/​government/​publications/​ip​ -basics/​ip​-basics accessed 19 July 2022. 29 Lydia Lundstedt, Territoriality in Intellectual Property Law (Stockholm University, 2016) 79. 30 The innovation clause within the US Constitution which provides Congress the power ‘to promote the Progress of Science and Useful arts, by securing, for limited Times, to Authors and Inventors, the exclusive Right to their respective Writings and Discoveries’ is a good example for the programmatic role that IP protection takes.

460  Research handbook on digital trade as non-tariff barriers to trade in goods or services.31 A push towards eliminating barriers to trade has been on the international agenda since the end of the Second World War, when many nations put aside their stance of economic protectionism commonly seen during the interwar period. A pinnacle moment was the establishment of the General Agreement on Trade and Tariffs (GATT) in 1947 in the aftermath of the Bretton Woods conference and the failure to establish the International Trade Organization (ITO). In various multilateral rounds, GATT sought to eliminate barriers to trade by creating an international framework which would be conducive for international trade. Initially, these rounds revolved around tariffs, but since the Uruguay Round of GATT the focus shifted to intellectual property as a key element to free trade and was promoted and pushed by developed countries.32 The Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS), an annex to the WTO Agreement, was devised and marked an important shift in international IP law.33 While other international intellectual property agreements exist which predate the TRIPS Agreement, such as the Paris34 and Berne Conventions,35 this has arguably been the most comprehensive and far-reaching. It covers all fields of intellectual property law, such as substantive as well as procedural rules. The agreement wishes to alleviate the aspects of IP which may be hampering international trade. In its first provision, for instance, it sets minimum standards which IP rights in the laws of WTO Member States must have.36 This had the effect that countries, particularly developing ones, among WTO Member States lost some legislative flexibility in tailoring their national IP frameworks. The inception of TRIPS also meant that new subject matter, such as computer programs, databases or geographical indications, was introduced in some WTO Member States. The agreement’s arguably most important feature is that it tethers compliance with its rules to the WTO’s dispute settlement understanding. An example of eliminating trade barriers posed by IP rights can be seen in the thrust by which the European Union (EU) has sought and still seeks to harmonise IP laws within its territories. By means of Regulations and Directives37 and more recently through an Article of the Treaty on the Functioning of the European Union (TFEU), the EU legislator has sought to eliminate trade barriers between its Member States, serving the completion of one of the EU’s main goals: the creation of the Internal Market. The EU legislator has been very active in the area of copyright law and is providing an increasingly harmonised EU copyright framework.

31 Graham Dutfield and Uma Suthersanen, Global Intellectual Property Law (2nd edn, Edward Elgar Publishing 2020) 12. 32 Surendra J. Patel, ‘Intellectual Property Rights in the Uruguay Round: A Disaster for the South?’ [1989] Economic and Political Weekly 978–93, 978. 33 The inclusion of intellectual property within a trade context was controversially regarded as enabling the conclusion of such treaty between developed and developing countries. Peter Drahos, stated that the history of TRIPS is ‘remarkable because one country, the US, was able to persuade more than 100 other countries that, they, as net importers of technological and cultural information, should pay more for the importation of that information’ – Peter Drahos, ‘Global Property Rights in Information: The Story of TRIPS at the GATT’ [1995] Prometheus 6–19, 7. 34 Paris Convention for the Protection of Industrial Property. 35 Berne Convention for the Protection of Literary and Artistic Works. 36 Article 1(1) TRIPS. 37 Article 288 Consolidated Version of the Treaty on European Union [2008] OJ C115/13.

Digital trade and intellectual property  461 In trade mark38 and design law,39 the EU has even established unitary EU IP rights which are granted by a specialised EU agency, the European Union Intellectual Property Office (EUIPO). These rights cover the territory of the European Union and can only be acquired and assigned uniformly.40 While no unitary EU copyright exists, the EU legislator has been very active in harmonising copyright law within the Union by the means of Directives which require Member States to adopt the goals mandated therein. Currently, the Digital Single Market Directive41 is the latest important measure of the EU in the field of copyright law. Another trend that is leading to more harmonised global frameworks of IP can be seen in IP chapters within bi- and multilateral investment treaties. These often go beyond, to those mandated within the TRIPS Agreement.42

3.

INTELLECTUAL PROPERTY AND DIGITAL TRADE

3.1

IP as Enabler of Trade in Digital Assets

Intellectual property protection constitutes a regulatory intervention to overcome the free riding problem, as we have established. For the purposes of digital trade, IP rights serve another important purpose. They also enclose the protected information or content by the means of exclusive rights. Thus, they allocate a particular subject matter to a particular individual or entity by the means of property law.43 This commodification of the protected subject matter enables the commercialisation of intangible goods.44 This can be achieved by assigning/ selling or licensing the content and can therefore be used to generate income through royalties45 or other means.46 The way in which the law enabled digital trade in IP content can be seen in Regulation (EU) 2017/1001 of the European Parliament and of the Council of 14 June 2017 on the European Union trade mark (Text with EEA relevance) OJ L 154 (EU Trade Mark Regulation – EU TMR). 39 Council Regulation (EC) 6/2002 of 12 December 2001 on Community designs OJ EC No L 3 (Community Design Regulation – CDR). 40 Article 1(2) EU TMR; Article 1(3) CDR. 41 Directive (EU) 2019/790 of the European Parliament and of the Council of 17 April 2019 on copyright and related rights in the Digital Single Market and amending Directives 96/9/EC and 2001/29/ EC (Text with EEA relevance). OJ L 130 (Digital Single Market Directive). 42 Henning Grosse Ruse-Khan, ‘Protecting Intellectual Property under BITs, FTAs, and TRIPS: Conflicting Regimes or Mutual Coherence?’ in C. Brown and K. Miles (eds), Evolution in Investment Treaty Law and Arbitration (Cambridge University Press 2011) 485–515, 490. 43 David Lametti, ‘The Concept of Property: Relations through Objects of Social Wealth’ [2003] University of Toronto Law Journal 325, 334. 44 Kenneth J. Arrow, ‘Economic Welfare and the Allocation of Resources of Invention’ in National Bureau of Economic Research, The Rate and Direction of Inventive Activity (Princeton University Press 1962) 615; Sacha Wunsch-Vincent, ‘The Economics of Copyright and the Internet’ in Johannes M. Bauer and Michael Latzer (eds), Handbook on the Economics of the Internet (Edward Elgar Publications 2016) 229–46, 231. 45 Keith Maskus, ‘Fostering Innovation in Digital Trade’ in Xavier Seuba, Christophe Geiger and Julien Pénin (eds), Intellectual Property and Digital Trade in the Age of Artificial Intelligence and Big Data – Global Perspectives for the Intellectual Property System (2018, 5 CEIPI-ICTSD) 19, 25. 46 Rosemary J. Coombe, Steven Schnoor and Mohsen Ahmed, ‘Bearing Cultural Distinction: Informational Capitalism and New Expectations for Intellectual Property’ [2007] UC Davis Law Review 891, 893. 38

462  Research handbook on digital trade the above-mentioned WIPO Internet Treaties. The addition of the right of communication to the public grants right holders the possibility to authorise any communication to the public, by wire or wireless means, including ‘the making available to the public of works in a way that the members of the public may access the work from a place and at a time individually chosen by them’.47 The clarification that the making available of works would fall within this right sought to cover on-demand, interactive communication of works over the internet, such as services provided by YouTube and others. In addition, IP also serves to secure venture capital,48 which firms may need to develop their particular business models. Aside from this, trade in digital IP goods also entails the positive effect of lowering transaction costs.49 As already mentioned, digital distribution makes physical storage space largely redundant due to virtual inventories and does not require the same means of logistics since physical trade routes are shifted to digital ones using the internet. The unbundling from physical carriers also means that consumer preferences can be better served. There is no need to buy a recording artist’s entire album or purchase an entire newspaper. Nowadays one can choose which songs one wishes to listen to or which article to read.50 This has led to fundamental changes in how music and audio-visual content is currently being consumed but has also impacted, for instance, the video game industry. 3.2

IP as Stumbling Block for Digital Trade

While IP has enabled the commodification and distribution of digital goods, it has not only had positive effects on digital trade. One of the deficiencies relates to the territoriality of IP51 in conjunction with the exclusivity it provides to its right holders. This combination allows for the segregation of markets52 according to territorial lines and jurisdictions through selective licensing by right holders which choose to license the content for some territories while refus-

Art. 8 WCT. Mary Juetten, ‘Do Venture Capitalists Care about Intellectual Property?’ (Forbes, 11 August 2015) www​.forbes​.com/​sites/​maryjuetten/​2015/​08/​11/​do​-venture​-capitalists​-care​-about​-intellectual​-property/​ ?sh​=​1204e8d35b87 accessed 19 July 2022. 49 Sacha Wunsch-Vincent, ‘The Economics of Copyright and the Internet’ in Johannes M. Bauer and Michael Latzer (eds), Handbook on the Economics of the Internet (Edward Elgar Publications 2016) 229–46, 233; Graham Dutfield and Uma Suthersanen, Global Intellectual Property Law (2nd edn, Edward Elgar Publications 2020) 483. 50 Sacha Wunsch-Vincent, ‘The Economics of Copyright and the Internet’ in Johannes M. Bauer and Michael Latzer (eds), Handbook on the Economics of the Internet (Edward Elgar Publications 2016) 229–46, 236. 51 Hugenholtz refers to this as the ‘Achilles heel’ of copyright harmonisation within the EU: Bernt Hugenholtz, ‘Is Harmonization a Good Thing?’ in Justine Pila and Ansgar Ohly (eds), The Europeanization of Intellectual Property Law (Oxford University Press, 2013) 57–73, 68. 52 Sacha Wunsch-Vincent, ‘The Economics of Copyright and the Internet’ in Johannes M. Bauer and Michael Latzer (eds), Handbook on the Economics of the Internet (Edward Elgar Publishing 2016) 229–46, 232. This issue is addressed with unified rights, as established in the EU. Another mechanism to overcome this has been conducted with the Portability Regulation (Regulation (EU) 2017/1128 of the European Parliament and of the Council of 14 June 2017 on cross-border portability of online content services in the internal market [2017] OJ L168/1.), which allows subscribers to access online content outside the Member State of their residence while temporarily present in another Member State. 47 48

Digital trade and intellectual property  463 ing to do so for others.53 In addition, the threat that digitisation and the internet would pose to commercialisation of IP was often answered by an expansion of IP rights and their scope. This growing control of content by right holders, however, may impair the overall seamless flow of data and information over the internet, the backbone of digital trade.54 This issue will be discussed in more detail later in this chapter. Transposing traditional IP right doctrines from an analogue world to the digital can generally lead to unwanted consequences for digital trade. An example which relates to the extension to copyright infringement occurring in the case of digital reproductions illustrates this well. While extending the reproduction right to include copies ‘in any material form’,55 which would ensure that unauthorised use of digital copies could be restrained by the right holders, it could also entail overreaching consequences for displaying goods on internet browsers which create reproductions from the cache of the computer.56 The potential chilling effects due to the uncertainty of being sued by right holders could lead to the breakdown of internet communication.57 While this issue was not addressed by the WIPO Internet Treaties, the InfoSoc Directive provided for an exception for temporary acts of reproduction,58 but this example illustrates the regulatory challenges in devising an IP framework conducive to digital trade.

4.

SELECTED ISSUES

4.1

Digital Second-hand Markets and Exhaustion of Copyright

An important issue for digital trade in goods covered by copyright law is whether and when the exclusive rights are exhausted in the digital environment. Similar to the situation offline where a book can be resold without the copyright owner interfering with this transaction, a resale market for second-hand digital goods could be a desirable goal. Copyright law provides its right holders with a set of rights, one being the right of first distribution.59 This gives them the exclusive right to commercialise the work on the market in return for a remuneration. This right, however, is usually exhausted (or would fall within the First-Sale doctrine under US copyright law) and can therefore not be enforced once the work has been placed onto the market with the consent of the right holder.60 Similar considerations could apply to digital 53 Nadia Iacob and Felice Simonelli, ‘How to Fully Reap the Benefits of the Internal Market for E-Commerce – New Economic Opportunities and Challenges for Digital Services 20 Years after the Adoption of the e-Commerce Directive’ (Study for the IMCO Committee 2020) 24 www​.europarl​ .europa​.eu/​RegData/​etudes/​STUD/​2020/​648801/​IPOL​_STU(2020)648801​_EN​.pdf accessed 19 July 2022. 54 Janos Ferencz and Frédéric Gonzales, ‘Barriers to Trade in Digitally Enabled Services in the G20’ (OECD Trade Policy Papers No.232, 2019) 5 read​.oecd​-ilibrary​.org/​trade/​barriers​-to​-trade​-in​-digitally​ -enabled​-services​-in​-the​-g20​_264c4c02​-en​#page6 accessed 19 July 2022. 55 Article 2 InfoSoc Directive. 56 ‘The restricted act of copying is therefore usually implicated in uses of works recorded in digital form.’ Nicholas Caddick et al. (eds), Copinger & Skone James on Copyright (17th edn, Sweet & Maxwell 2016) [7.31]. 57 Manfred Rehbinder and Alexander Peukert, Urheberrecht (17th edn, C.H. Beck 2015) [640]. 58 Article 5(1) InfoSoc Directive. 59 E.g. Article 4(1) InfoSoc Directive, Sec. CDPA 1988, 17 U.S. Code § 106 (3). 60 E.g. Article 4(2) InfoSoc Directive, Section 18(2) CDPA 1988, 17 U.S. Code § 109.

464  Research handbook on digital trade works, such as computer programs, e-books or music files which could be traded on online second-hand markets. Transposing rules for IP content in tangible goods to digital ones, however, is not unproblematic, as already discussed. The exhaustion doctrine was developed in relation to copyright law61 in the analogue world and transposing it to digital goods entails two issues: first, digital copies are not subject to the wear and tear of tangible carriers but stay identical to the original; second, they can easily be disseminated and shared worldwide.62 This would entail a risk of piracy and might impact the market for sales of the originals.63 In the EU context, the exhaustion of rights doctrine is seen as overcoming barriers to trade within the single market64 and applied in relation to the distribution right regarding tangible carriers. For digital trade, the communication to the public right introduced by the WIPO Internet Treaties needs to be considered as it encompasses all forms of transmission in intangible formats, thus being the base for new services on the internet.65 This particular right, however, cannot be exhausted in the EU for works falling within the InfoSoc Directive,66 and has the effect of banning the resale of lawfully acquired digital products67 and a secondary market. The situation is exacerbated where digital works are treated differently within the context of digital trade. The seminal UsedSoft decision by the CJEU68 held with regard to the resale of computer software69 that ‘[t]he on-line transmission method is the functional equivalent of the supply of a material medium’.70 This statement promotes the equal treatment of digital and physical goods in this context. The Court elaborated that, where a permanent copy is received by the end-user in exchange for a fixed purchase price which is retained on a permanent basis, then this act would be covered by the right of distribution under the Software Directive,71 ultimately meaning that it could be subject to being exhausted72 and not a right of communication to the public.73 Stringent criteria were, however, mandated by the Court for the

Péter Mezei, Copyright Exhaustion: Law and Policy in the United States and the European Union (2nd edn, Cambridge University Press 2022) 106. 62 Ariel Katz, ‘Digital Exhaustion: North American Observations’ in John A. Rothchild (ed), Research Handbook on Electronic Commerce Law (Edward Elgar Publishing 2016) 164. 63 Caterina Sganga, ‘A Plea for Digital Exhaustion in EU Copyright Law’ [2018] JIPITEC 212, 213. 64 C-15/74 Centrafarm BV et Adriaan de Peijper v Sterling Drug Inc. [11], [15]. 65 Article 3(1) InfoSoc Directive. 66 Article 3(3) InfoSoc Directive. 67 Stavroula Karapapa, ‘Exhaustion of Rights on Digital Content under EU Copyright: Positive and Normative Perspectives’ in Tanya Aplin (ed), Research Handbook on Intellectual Property and Digital Technologies (Edward Elgar Publishing 2020) 483–505, 493. 68 UsedSoft GmbH v Oracle International Corp (C-128/11) EU:C:2012:407. 69 The CJEU noted that the particular object in question would qualify as falling within the ambit of the Computer Software Directive which does not differentiate between tangible and intangible copies of computer programs in relation to exhaustion. Additionally, these rules were lex specialis to Art. 3 of the InfoSoc Directive. UsedSoft GmbH v Oracle International Corp (C-128/11) EU:C:2012:407 [73], [74]. 70 UsedSoft GmbH v Oracle International Corp (C-128/11) EU:C:2012:407 at [61]. 71 Article 4(1)(2) Directive 2009/24 on the legal protection of computer programs [2009] OJ L111/16 (Software Directive). 72 Article 4(2) Software Directive. 73 This then would suggest that ‘uses that do not lead to the permanent reproduction or sale of any copy of a protected subject matter are governed by the communication or making available to the public right.’ Péter Mezei, ‘The Doctrine of Exhaustion in Limbo – Critical Remarks on the CJEU’s Tom Kabinet Ruling’ [2020] 2 Zeszty Naukowe Uniwersytetu Jagiellonskiego – Prace z Prawa Wlasnosci 61

Digital trade and intellectual property  465 distribution right to be exhausted, to address the issue surrounding the digital copy.74 While this shed some light in relation to works falling within the Computer Software Directive as a lex specialis, the works falling within the ambit of the InfoSoc Directive would still not be subject to digital exhaustion. This issue was discussed by the CJEU in relation to e-books in the Tom Kabinet decision.75 There, the Court took a restrictive approach to digital exhaustion which can be explained from a historical and doctrinal point of view.76 Mézei argued that such approach would lead to the ‘castration’ of the exhaustion doctrine in the digital environment.77 The First-Sale doctrine under US law is based on judicial interpretation with the seminal US Supreme Court decision in Bobbs-Merrill v Straus78 of 1908 setting a precedent. It found its way into the US Copyright Act in 1909 and saw several revisions in order to adapt to technological developments and the impact of international copyright law.79 The first sale doctrine states that ‘the owner of a particular copy or phonorecord lawfully made under this title […] is entitled […] to sell or otherwise dispose of the possession of that copy or phonorecord’.80 In addition, the owner of a copy may also freely rent or lend her copy, unless these are phonograms and software. Reproduction and adaption of a copy is generally not permitted, except a software copy where this is necessary ‘as an essential step in the utilisation of the computer program’,81 subject to certain limitations. The law in the United States in relation to digital exhaustion is equally complex and different rules apply for different types of digital objects. The resale of digital music files was discussed by in the Capitol Records, LLC v ReDigi Inc. decision82 by the US District Court for the Southern District of New York. ReDigi was an online marketplace for used digital music files which would permit the resale of songs bought on iTunes. To be considered to be permissible as a transfer of possession rather than an impermissible reproduction of copies, the process involved migrating the user’s file to a cloud computer so that the particular data would not exists in two places at one time. The Court, however, held these activities, inter alia, not to be covered by the First-Sale doctrine since they would also include the reproduction of files.83 The First-Sale doctrine, however, would only cover the distribution right. In relation to Intelektualnej (Jagiellonian University Intellectual Property Law Review) 139, ssrn​.com/​abstract​=​ 3560138 accessed 19 July 2022. 74 That is, the permanent character of the ‘licence’, the appropriate remuneration of the right holder, and the reseller making their copies unusable for after the resale – Case C-128/11 UsedSoft GmbH v Oracle International Corp EU:C:2012:407 at [87], [88]. 75 Case C-263/18 Nederlands Uitgeversverbond and Groep Algemene Uitgevers v Tom Kabinet Internet BV EU:C:2019:1111. 76 For example, the submission of AG Szpunar in Tom Kabinet EU:C:2019:1111 – Opinion of AG Szpunar, 12 September 2019, at [56]–[63]. 77 Péter Mezei, ‘The Doctrine of Exhaustion in Limbo – Critical Remarks on the CJEU’s Tom Kabinet Ruling’ [2020] 2 Zeszty Naukowe Uniwersytetu Jagiellonskiego – Prace z Prawa Wlasnosci Intelektualnej (Jagiellonian University Intellectual Property Law Review) 139, ssrn​.com/​abstract​=​ 3560138 accessed 19 July 2022. 78 Bobbs-Merrill Co. v Straus, 210 U.S. 339, 350–1 (1908). 79 Péter Mezei, Copyright Exhaustion: Law and Policy in the United States and the European Union (2nd edn, Cambridge University Press 2022) 78. 80 17 U.S.C. §109(a). 81 17 U.S.C. § 117(a)(1). 82 934 F. Supp. 2d 640 (S.D.N.Y. 2013). 83 ‘[T]he plain text of the Copyright Act makes clear that reproduction occurs when a copyrighted work is fixed in a new material object.’ Ibid at 655.

466  Research handbook on digital trade computer software, the First-Sale doctrine would seldom apply. Both source and object code are considered as literary works since the inception of the Computer Software Copyright Act of 1980,84 and would therefore fall within the ambit of 17 US Code § 109. Software producers, however, have sought to label the transaction between them and the purchasers as a licence, which would not trigger the application of the First-Sale doctrine.85 The above-mentioned case law from both sides of the pond showcases the difficulties in placing concepts borne out of a brick-and-mortar world into the digital context. It appears that courts felt constrained by the current legal provisions. Thus, while some acknowledged the new realities of digital trade, others stuck with a very literal interpretation of the positive law. In addition, digitisation has generated issues surrounding legal classification, such as the question whether a digital transaction would constitute a sale or licensing agreement; the relationship between the distribution right and the communication right, which is a stumbling block in the EU context;86 and finally whether the particular transaction has goods or services as their object.87 These issues require attention, though might attract less relevance when the use of licensing overtakes sales contracts in digital trade, which usually occurs in subscription models, such as Netflix, Amazon Kindle or Spotify, where the exhaustion doctrine does not apply. 4.2

Making Gatekeepers Responsible: Online Intermediary Liability for IP Infringements

Digitisation in conjunction with an ever-faster internet created the fastest and most efficient copying machine.88 This has had a severe impact on copyright industries as unauthorised reproductions of copyright works were difficult for copyright holders to address.89 As attempts to act against individual infringers as a potential customer base90 did not appear to be the right move, in addition to being a burdensome and often futile enterprise, the focus was shifted early to other parties which enable the infringing action, by, for instance, providing technologies which enable the reproduction and dissemination of works. Such secondary liability occurs

Act of Dec. 12, 1980, Pub. L. No. 96-517, § 10, 94 Stat. 3028. Lothar Determann, ‘Digital Exhaustion: New Law from the Old World’ [2018] Berkeley Technology Law Journal 177, 192. 86 Other WCT signatories avoided these issues by opting to implement the making available right as a form of the distribution right: Wolf R. Meier-Ewert and Jorge Gutierrez, ‘Intellectual Property and Digital Trade Mapping International Regulatory Responses to Emerging Issues’ (World Trade Organization 2020) 31–2. www​.wto​.org/​english/​res​_e/​reser​_e/​ersd202104​_e​.pdf accessed 19 July 2022. 87 Caterina Sganga, ‘A Plea for Digital Exhaustion in EU Copyright Law’ [2018] JIPITEC 212, 217–22. 88 Former United States Register of Copyrights Marybeth Peters quoted in: Vic Sussman, ‘Policing Cyberspace’ (US News & World Report 23 January 1995) all​.net/​cybercop/​usnews1​.htm accessed 19 July 2022. 89 Alain Strowel, ‘Introduction: Peer-to-Peer File Sharing and Secondary Liability in Copyright Law’ in Alain Strowel (ed), Peer-to-Peer File sharing and Secondary Liability in Copyright Law (Edward Elgar Publishing 2009) 1. 90 Andrew Murray, Information Technology Law: The Law & Society (4th edn, Oxford University Press 2019) 328. 84 85

Digital trade and intellectual property  467 where a party contributes, in some relevant way, to other people’s infringing actions.91 Many copyright cases in the past have dealt with photocopiers and to what extent the entity providing the photocopier to their customers could be held liable for copyright infringement.92 The focus was also on producers of hardware enabling the reproductions. The Betamax case93 in the US and the Amstrad case in the UK94 are examples of such litigation. With regard to infringement occurring over the internet, a secondary liability claim against an intermediary may prove to be more efficient than going against individual infringers in a ‘whack-a-mole attempt’ as it can reach across borders and may shift some costs onto intermediaries.95 Thus, a trend towards ‘more’ secondary liability for online intermediaries can be perceived in the recent policy debates.96 Early case law in relation to secondary liability targeted filesharing platforms. As such, the Napster decision in the United States marked an important turning point in shifting the focus away from individual users sharing the material, to the platform facilitating this and other intermediaries.97 Napster provided a central index server by which users were able to access an index of connected users and the files they would offer to share on their computer, which facilitated the search and download of mp3 music files. Various record companies sued Napster. The Ninth Circuit98 ultimately affirmed the ruling of the District Court for the Northern District of California, which found Napster to be liable for contributory and vicarious infringement of the plaintiffs’ copyright. Another important decision in relation to intermediary liability was handed down by the US Supreme Court in its Grokster decision.99 Grokster’s peer-to-peer filesharing system differed from Napster’s by allowing users to trade files directly, thus skipping a centralised server, which lessened the control Grokster had over what occurred on its platform. The Supreme Court, nevertheless, unanimously found Grokster liable for copyright infringement due to inducing infringement by its users. The examples above highlights, however, that there must be some form of misconduct to trigger secondary liability. In recent years, the creation of a third pillar of liability could be witnessed against innocent third parties that did not engage in any wrongdoing themselves, while wrongdoing of others was still required.100 While they are usually not considered to be

Martin Husovec, ‘Remedies First, Liability Second: Or Why We Fail to Agree on Optimal Design of Intermediary Liability’ in Giancarlo Frosio (ed), Oxford Handbook of Online Intermediary Liability (OUP 2020) 91. 92 Moorhouse v University of New South Wales [1976] RPC 151 (High Court of Australia); CCH Canadian Ltd. v Law Society of Upper Canada [2004] 1 S.C.R. 339. 93 Sony Corp. of America v Universal City Studios, Inc., 464 U.S. 417 (1984). 94 CBS Songs Ltd v Amstrad Consumer Electronics Plc [1988] UKHL 15. 95 Graeme Dinwoodie, ‘A Comparative Analysis of the Secondary Liability of Online Service Providers’ in Graeme Dinwoodie (ed), Secondary Liability of Internet Service Providers (Springer 2017) 1–72, 16–17. 96 Giancarlo F. Frosio, ‘From Horizontal to Vertical: An Intermediary Liability Earthquake in Europe’ [2017] Journal of Intellectual Property Law & Practice 565–75, 565. 97 Annemarie Bridy, ‘Why Pirates (Still) Won’t Behave: Regulating P2P in the Decade after Napster’ [2009] Rutgers Law Journal 565, 583. 98 A&M Records Inc v Napster Inc., 239 F. 3d 1004 (9th Cir. 2001). 99 MGM Studios Inc v Grokster Ltd, 545 U.S. 913 (2005). 100 Martin Husovec, ‘Remedies First, Liability Second: Or Why We Fail to Agree on Optimal Design of Intermediary Liability’ in Giancarlo Frosio (ed), Oxford Handbook of Online Intermediary Liability (OUP 2020) 93. 91

468  Research handbook on digital trade liable, certain obligations would be imposed in holding them accountable for assistance due to efficiency or fairness considerations.101 This leads to a regulatory conflict for legislators in aiming to safeguard intermediaries from liability for actions of the customers while providing an effective mechanism against online copyright infringement.102 On both sides of the pond, this issue has been settled by so-called safe harbour provisions. Such provisions can be found within the US DMCA and the EU E-Commerce Directive. Where they apply, internet service providers would be exempted from any liability caused by the actions of their users. In comparison to the US, which maintained a relatively broad scope of its ‘safe harbour’ provisions and a high threshold for triggering ISP liability,103 the current EU framework appears to be more favourable for IP right holders due to a narrowing scope of the safe harbour in conjunction with an expanding use of injunctive relief against intermediaries.104 The E-Commerce Directive provides for safe harbours for different actions.105 However, the revisions of the framework with the recently enacted Directive on Copyright in the Digital Single Market, along with its infamous Article 17 on the liability of ‘online content-sharing service providers storing and giving access to large amounts of works’, indicate that the responsibility of intermediaries has risen. While this new framework would not change the EU’s safe harbour regime, it sets an enhanced liability regime which introduces new obligations and duties of care for intermediaries.106 Maintaining compliance will inevitably incur cost and may disincentivise smaller companies,107 thus increasing barriers to innovation.108

5.

EMERGING ISSUES

The previous section highlighted some of the current matters that arise with IP and its enforcement in relation to digital trade. This section looks at some emerging issues which may impact

Martin Husovec, ‘Remedies First, Liability Second: Or Why We Fail to Agree on Optimal Design of Intermediary Liability’ in Giancarlo Frosio (ed), Oxford Handbook of Online Intermediary Liability (OUP 2020) 93. 102 Similar considerations apply to the liability of intermediaries for trade mark infringement: see Irene Calboli, ‘Contributory Trademark Infringement on the Internet: Shouldn’t Intermediaries Finally Know What They Need to “Know” and “Control”?’ in John A. Rothchild (ed), Research Handbook on Electronic Commerce Law (Edward Elgar Publishing 2016) 211–31. 103 Wolf R. Meier-Ewert and Jorge Gutierrez, ‘Intellectual Property and Digital Trade Mapping International Regulatory Responses to Emerging Issues’ (World Trade Organization 2020) 24 www​.wto​ .org/​english/​res​_e/​reser​_e/​ersd202104​_e​.pdf accessed 19 July 2022. 104 Valentina Moscon, ‘Free Circulation of Information and Online Intermediaries – Replacing One “Value Gap” with Another’ [2020] I.I.C. 977, 978. 105 For example, arts 13 (caching), 14 (hosting) and 15(1) (no general obligation to monitor) of Directive 2000/31 on certain legal aspects of information society services, in particular electronic commerce in the Internal Market [2000] OJ L178/1 (E-Commerce Directive). 106 Maria Lillà Montagnani, ‘A New Liability Regime for Illegal Content in the Digital Single Market Strategy’ in Giancarlo Frosio (ed), Oxford Handbook of Online Intermediary Liability (OUP 2020) 309. 107 Javier López González and Janos Ferencz, ‘Digital Trade and Market Openness’, OECD Trade Policy Papers No.217 (OECD Publishing 2018) 34 dx​.doi​.org/​10​.1787/​1bd89c9a​-en accessed 19 July 2022. 108 Christophe Geiger, Giancarlo Frosio and Elena Izyumenko, ‘Intermediary Liability and Fundamental Rights’ in Giancarlo Frosio (ed), Oxford Handbook of Online Intermediary Liability (OUP 2020) 149. 101

Digital trade and intellectual property  469 and shape IP rights yet again. While blockchain technology, also referred to as distributed ledger technology (DLT), was devised in the early 1990s, it became most widely known when the person behind the pseudonym Satoshi Nakamoto promoted its use for the bitcoin cryptocurrency in 2009.109 The advantage of DLT is that it does not need a centralised authority to validate a particular piece of information, such as a transaction, as this is done in a decentralised110 way by consensus of nodes within a computer network.111 It provides a data base where information is stored within a block, then chained onto a preceding block, forming a chronological chain. Blockchain technology could have vast applications in relation to IP rights and enforcement. For instance, the creatorship of unregistered IP rights, such as copyright or unregistered design rights could be stored on a blockchain and may serve as evidence.112 They may also be used in customs enforcement where genuine goods are provided with scannable tags or engravings on products.113 In addition, the technology may be used in conjunction with smart contracts for the use of copyright works which would make receiving remuneration in real time by an accompanied smart contract possible.114 A crucial stumbling block for digital exhaustion and a resale market for digital IP assets elaborated above could be alleviated by DLT by guaranteeing some form of control of the files as to their origin. This would provide transparency in the transaction and foreclose the unlawful duplication of files.115 Finally, blockchain technology and smart contracts could be used to overcome the fragmentation of the various national copyrights and the attached rights a particular work entails, which would truly be a milestone in trade in digital IP assets. Thus, while there remain some doubts, for example as to whether the technology is genuinely safe by being ‘unhackable’,116 the technology may have a transformative impact on copyright in the digital environment.117 Very much connected to DLT are so-called non-fungible tokens (NFTs), which have become a buzzword as of late. NFTs are a form of certificate on a suitable blockchain, such

109 Satoshi Nakamoto, ‘Bitcoin: A Peer-to-Peer Electronic Cash System’ (2008) www​.ussc​.gov/​sites/​ default/​files/​pdf/​training/​annual​-national​-training​-seminar/​2018/​Emerging​_Tech​_Bitcoin​_Crypto​.pdf accessed 19 July 2022. 110 Primavera de Filippi and Aaron Wright, Blockchain and the Law: The Rule of Code (Harvard University Press 2018) 21. 111 Sebastian Pech, ‘Copyright Unchained: How Blockchain Technology Can Change the Administration and Distribution of Copyright Protected Works’ [2020] Northwestern Journal of Technology and Intellectual Property 1–50, 10. 112 In June 2018, the Hangzhou Internet Court permitted the use of blockchain-authenticated evidence in a copyright case in June 2018: Kim Lu and Dong Ning, ‘China Patent: Courts Respond Positively to Blockchain Evidence’ (Managing IP, 18 September 2019) www​.managingip​.com/​article/​2a​5bssmgulfp​ v353htybk/​china​-patent​-courts​-respond​-positively​-to​-blockchain​-evidence accessed 19 July 2022. 113 Birgit Clark and Ruth Burstall, ‘Crypto-Pie in the Sky? How Blockchain Technology Is Impacting Intellectual Property Law’ [2019] Stanford Journal of Blockchain Law & Policy 252–62, 255. 114 Balázs Bodó, Daniel Gervais and João Pedro Quintais, ‘Blockchain and Smart Contracts: The Missing Link in Copyright Licensing?’ [2018] International Journal of Law and Information Technology 311–36, 324. 115 Péter Mezei, Copyright Exhaustion: Law and Policy in the United States and the European Union (2nd edn, Cambridge University Press 2022) 193. 116 Birgit Clark and Ruth Burstall, ‘Crypto-Pie in the Sky? How Blockchain Technology Is Impacting Intellectual property Law’ [2019] Stanford Journal of Blockchain Law & Policy 252-262, 261. 117 Balázs Bodó, Daniel Gervais and João Pedro Quintais, ‘Blockchain and Smart Contracts: The Missing Link in Copyright Licensing?’ [2018] International Journal of Law and Information Technology 311–36, 336.

470  Research handbook on digital trade as the Ethereum blockchain, which indicate ownership and are supported by smart contracts.118 NFTs are currently used for digital collectibles but anything that can be represented digitally, including physical goods, can be turned (that is, ‘minted’) into NFTs.119 In comparison to other forms of tokens, NFTs are unique, that is, non-fungible, which attributes value to them.120 Some noticeable examples include an NFT of former Twitter CEO Jack Dorsey’s first tweet from 2006, which sold for an equivalent of 2.5 million US dollars in 2021.121 Earlier that year the auction house Christie’s sold an NFT of the digital artwork ‘Everydays: The First 5000 Days’ by the artist Beeple for 69.3 million US dollars.122 Several IP issues relate to NFTs. With regard to copyright law, there is the question whether minting an NFT could attract copyright protection by regarding it as an artistic performance.123 There is also the elephant in the room: whether minting an NFT of a copyright-protected work would constitute an infringement of the owner’s exclusive rights. Here, it needs to be said that the NFT does not represent the work itself, so when one acquires a token over a work, then this does not transfer ownership of copyright in the underlying work.124 It is rather a cryptographically signed form of receipt that one owns in relation to a particular work.125 This means that only the communication right might potentially be infringed by linking the work;126 this may not arise with regard to other rights, such as the reproduction or distribution right.127 But case law relating to IP and NFTs is currently making its way through the courts. The film director Quentin Tarantino is being sued by the film producers Miramax for creating NFTs of the scripts and scenes of the film Pulp Fiction128 and a Chinese court has found a NFT platform

118 Sebastian Pech, ‘Copyright Unchained: How Blockchain Technology Can Change the Administration and Distribution of Copyright Protected Works’ [2020] Northwestern Journal of Technology and Intellectual Property 1–50, 12. 119 Andres Guadamuz, ‘Non-Fungible Tokens (NFTs) and Copyright’ (WIPO Magazine December 2021) www​.wipo​.int/​wipo​_magazine/​en/​2021/​04/​article​_0007​.html accessed 19 July 2022. 120 Pınar Çağlayan Aksoy and Zehra Özkan Üner, ‘NFTs and Copyright: Challenges and Opportunities’ [2021] JIPLP 115–26, 117–18. 121 Alex Hern and Dan Milm, ‘NFTs: The Great Rush May Be Over – But Are They in Actual Decline?’ (The Guardian 6 May 2022) www​.theguardian​.com/​technology/​2022/​may/​06/​nfts​-rush​ -decline​-jack​-dorsey​-tweet accessed 19 July 2022. 122 ‘Beeple’s opus’ (Christie’s) www​.christies​.com/​features/​Monumental​-collage​-by​-Beeple​-is​-first​ -purely​-digital​-artwork​-NFT​-tocome​-to​-auction​-11510​-7​.aspx accessed 19 July 2022. 123 Pınar Çağlayan Aksoy and Zehra Özkan Üner, ‘NFTs and Copyright: Challenges and Opportunities’ (2021) JIPLP 115–26, 121–22. 124 The transfer of copyright may, however, still be encompassed but often needs to comply with certain formalities to have effect: Andres Guadamuz, ‘Non-Fungible Tokens (NFTs) and Copyright’ (WIPO Magazine December 2021) www​.wipo​.int/​wipo​_magazine/​en/​2021/​04/​article​_0007​.html accessed 19 July 2022. 125 William Holmes, ‘What the NFT Craze Means for IP Law’ (Legal Cheek, 12 March 2021) www​ .legalcheek​.com/​lc​-journal​-posts/​what​-the​-non​-fungible​-token​-craze​-means​-for​-ip​-law/​ accessed 17 July 2022. 126 Pınar Çağlayan Aksoy and Zehra Özkan Üner, ‘NFTs and Copyright: Challenges and Opportunities’ [2021] JIPLP 115–26, 123. 127 Andres Guadamuz, ‘Non-Fungible Tokens (NFTs) and Copyright’ (WIPO Magazine December 2021) www​.wipo​.int/​wipo​_magazine/​en/​2021/​04/​article​_0007​.html accessed 19 July 2022. 128 Winston Cho, ‘Quentin Tarantino Tries for an Early Court Win in “Pulp Fiction” NFT Legal Battle’ (The Hollywood Reporter, 23 June 2022) www​.hollywoodreporter​.com/​business/​digital/​quentin​ -tarantino​-nft​-legal​-battle​-1235171174/​ accessed 19 July 2022.

Digital trade and intellectual property  471 liable for contributary copyright infringement as one of its users created and sold a NFT digital work which was identical to the copyright-protected work in the cartoon series Fat Tiger.129 The impact of digitisation has arguably been most profoundly felt in the area of copyright law. This is also the area where legislators and courts have had most chance to address this issue. The emergence and proliferation of 3D printing technology, also referred to as additive manufacturing (AM), may shift other IP rights, such as patents, designs and trade marks, to a focal point with regard to issues of digitisation.130 With an ever-increasing range of applications, 3D printing can be a crucial element in a market for 3D-printable designs which are delivered digitally via the internet and finally manufactured close to the end-customer. The technology allows for customisation of the end product and also has benefits for the environment through digital storage and digital distribution in lieu of shipping tangible goods.131 Currently, the technology can be used for producing apparel, jewellery and home items but also for other, less ‘trivial’ items, such a spare parts or medical devices. 3D printing technology has been around for some decades now and has initially been used for rapid prototyping.132 The technology uses a digital file that can either be created from scratch using CAD technology or by scanning a real object and converting this to a digital file which can instruct the printer to replicate the object encompassed within the digital file.133 Here, the clash between 3D printing and IP rights becomes clear: the technology can be used to reproduce objects protected by copyright, industrial designs, trade marks or even patents. The files containing the scanned objects which are needed to instruct the 3D printer may be disseminated over the internet and shared globally, and therefore reproduced globally. These points do, to a certain degree, revisit the issues copyright law went through two decades ago with filesharing.134 In relation to other IP rights, questions then arise as to whether the file containing the scanned and digitised object could constitute a patent, design or trade mark infringement. Would intermediaries be regarded as gatekeepers, similar to the situation in copyright law? And can such obligation be constructed for IP rights that did not feel the effects of digitisation in the same way copyright has? EU registered design law, for instance, does not provide for indirect infringement provisions.135 Another question relates to private uses

129 Xiao Baiyang, ‘Shenzhen QiCeDieChu Cultural and Creativity Co. v Hangzhou Bigverse Technology Co. (2022)’ [2022] Journal of Intellectual Property Law & Practice, jpac064, doi​.org/​10​ .1093/​jiplp/​jpac064 accessed 19 July 2022. 130 See the relevant chapters in Dinusha Mendis, Mark Lemley and Matthew Rimmer (eds), 3D Printing and Beyond: The Intellectual Property and Legal and Implications Surrounding 3D Printing and Emerging Technologies (Edward Elgar Publishing 2018). 131 Marc Mimler, ‘3D Printing, the Internet and Patent Law – A History Repeating?’ [2013] La Rivista di Diritto Industriale, 352–70, 353. 132 Matt Ratto and Robert Ree, ‘Materializing Information: 3D Printing and Social Change’ (First Monday, Volume 17, Number 7 – 2 July 2012) journals​.uic​.edu/​ojs/​index​.php/​fm/​article/​download/​ 3968/​3273 accessed 19 July 2022. 133 Marc Mimler, ‘3D Printing, the Internet and Patent Law – A History Repeating?’ [2013] La Rivista di Diritto Industriale, 352–70, 356. 134 Marc Mimler, ‘3D Printing, the Internet and Patent Law – A History Repeating?’ [2013] La Rivista di Diritto Industriale, 352–70, 369. 135 European Commission, Directorate-General for Internal Market, Industry, Entrepreneurship and SMEs, ‘Legal Review on Industrial Design Protection in Europe: Final Report’ (Publications Office, 2016) 132 data​.europa​.eu/​doi/​10​.2873/​056970 accessed 19 July 2022.

472  Research handbook on digital trade which are currently privileged and exempted from infringement.136 This has traditionally been explained as not affecting IP right holders when someone tinkers in their private home and does so for non-commercial purposes. But would this doctrine become obsolete once mass private 3D printing becomes a reality? These points only scratch the surface; many unanswered questions in this regard remain which need to be addressed by legislators and courts in the future.

6.

SOME CONCLUDING THOUGHTS ON INTERNATIONAL GOVERNANCE

Having discussed the operation of IP in digital trade, it becomes clear that its regulation requires international cooperation due to the global reach of the internet. Many of the discussed issues are still subject to national regulation and divergences inevitably lead to barriers to trade. This also applies to rules and regulations on internet governance where diverging national or regional approaches both generally and in relation to IP protection affect the operations of businesses operating in multiple jurisdictions.137 Regulatory challenges posed by digital technologies were initially tackled nationally, but the global nature of the internet has shifted the focus on finding solutions to the international level.138 Some of these international frameworks, such as the TRIPS Agreement, may be able to provide a multilateral framework for discussions, preferable to a set of bilateral or regional responses, which would lead to further fragmentation. This approach should also be prioritised over ‘made to measure’ acts such as the ACTA which seek to safeguard specific regulatory issues which tend to oversee the wider societal issues at stake. In relation to digital trade and internet governance in the context of IP, these are issues such as freedom of expression and operations, but also wider cultural and political measures. This prohibits a trade-specific context but mandates a wider perspective of all these interrelated issues.

7. CONCLUSION This chapter has outlined the importance of intellectual property as a vital component of digital trade. It provides a means to commodify digital assets and has been the foundation of many new business models that create and exploit these. The chapter has also outlined the shortcomings of adapting IP from a world of atoms to one of bits and bytes. Often, the solutions provided have been unsatisfactory, by either not seamlessly fitting the digital era or

136 Rosa Maria Ballardini and Nari Lee, ‘The Private and Non-Commercial Use Defence Revisited: The Case of 3D Printing Technologies’ in Rosa Maria Ballardini, Marcus Norrgård and Jouni Partanen (eds), 3D Printing, Intellectual Property and Innovation Insights from Law and Technology (Kluwer Law International 2017) 169–88. 137 Javier López González and Janos Ferencz, ‘Digital Trade and Market Openness’, (OECD Trade Policy Papers No.217, OECD Publishing 2018) 34–5 dx​.doi​.org/​10​.1787/​1bd89c9a​-en accessed 19 July 2022. 138 Mira Burri and Thomas Cottier, ‘Digital Technologies and International Trade Regulation’ in Mira Burri and Thomas Cottier (eds), Trade Governance in the Digital Age: World Trade Forum (Cambridge University Press, 2012) 1–15, 2.

Digital trade and intellectual property  473 by ‘overprotecting’ IP rights, which may negatively affect growth and innovation in this area. Regulators are tasked with seeking workable global solutions that not only overcome doctrinal reservations and provide holistic answers that must encompass wider societal issues and values, but can also accommodate future technologies.

Index

3D printing technology, intellectual property, copyright 471–2 Aaronson, S 57, 58, 59, 60, 81, 84, 119, 134 accountability algorithmic justice in the digital market and artificial intelligence (AI) 357–8, 364 WTO law and cross-border data flows 84–5, 86, 89 advertising 41, 56, 228, 249, 272, 277, 278, 280, 283 Africa AI arbitrators, accessibility concerns 410 China–Africa Joint Arbitration Centre (CAJAC) 409 Organisation for Harmonisation of Business Law in Africa (OHADA) 409–10 African Continental Free Trade Area (AfCFTA) 218–32 and access to justice 231 AU Digital Transformation Strategy for Africa 227 autonomy principle 223 consent principle 222 consumer protection 226 cross-border data flows 226, 230 current trends in free trade agreements (FTAs) 224–5 cybersecurity laws 224, 226 data governance and transparency mechanisms 229–30 data localisation 226, 230–31 data protection laws 224, 226 digital brain fog 227–8 digital divide 227 digital trade principles 219–22 Digital Transformation Strategy for Africa 229 digitalisation of commercial transactions 229 e-commerce inclusion 226–7 e-signatures and digital signatures, distinction between 227, 230 Information Technology Agreement (ITA) 226–7 integrity principle 223–4 intellectual property rights 226, 230 internet use 226 and least developed countries (LDCs) 226 non-discrimination principle 222

obstacles in Africa 225–8 online purchasing systems 228 open government data 223, 230 paperless systems 229 privacy protection 230–31 Protocol on Digital Trade 228–31 South–South preferential trading areas 220–21 transparency principle 223 UN Convention on the Use of Electronic Communications in International Contracts 221–2 UNCITRAL Model Law on Electronic Commerce (MLEC) 221, 224–5 UNCITRAL Principles 222–4 WTO Chairs Programme (WCP) 218–19 WTO Electronic Commerce Programme 220, 229 AI see artificial intelligence algorithmic justice in digital market and artificial intelligence (AI) 345–68 accountability 357–8, 364 algorithm definition 345–6 algorithmic pollution 350–51, 359 automation in the decision-making process within the criminal justice system 348–9 consumers and automated profiling 359–61, 365 COVID-19 pandemic effects 346 and cyberattacks 354–5 ‘data poisoning’ use 364 diversity, non-discrimination and fairness 356–7, 362, 365 ethics 351, 353, 358–9, 365 EU Artificial Intelligence Act proposal 363–5, 391 EU Digital Services Package 366–7 expert knowledge and ‘black box tinkering’ call 361 feedback loops, measures against 365 fundamental rights 354, 363 gatekeepers 366–7 human agency and oversight 354, 356, 357, 362, 364, 365, 366–7 information society concept 348 legal perspective 346–8, 350–53 market abuse issues 361 model 350–53

474

Index  475 Montreal Declaration 351, 352, 353, 357–8 OECD trustworthy principles 351, 356, 357–8 pillars 353–8, 363 platform work protection 365–6 pre-coded bias premise 348–9, 354, 355, 360, 365 privacy and data governance 355, 362, 364 problem identification 358–61 regulatory framework goal 361–7 risk factors 354–5, 357, 362, 363–5 societal and environmental well-being 357, 362, 365 technical robustness and safety 354–5, 362 trade secrets, protection of 349, 360–61 transparency 351–2, 356, 361, 362, 364–5, 366 UN Framework for Ethical AI 351 Allenby, B 370, 373, 376 Alschner, W 2, 54–72 Angwin, J 349, 406, 411 Anti-Counterfeiting Trade Agreement (ACTA) 459, 472 arbitrator appointment, international commercial arbitration 402–3, 404–5 Argentina–Chile FTA 46 Arner, D 309, 314, 326 Arrow, K 419, 461 artificial intelligence, digital economy agreements 435–6 artificial intelligence (AI) and algorithmic justice see algorithmic justice in the digital market and artificial intelligence (AI) Digital Economy Partnership Agreement (DEPA) 93–4, 96, 99 EU as digital trade actor 119–20 fintech 309–10, 316 gender equality 108 preferential trade agreements (PTAs) 51–2 WTO law and cross-border data flows 80 artificial intelligence (AI), national regulatory approaches 369–99 AI definitions 373–6 AI pioneers 374–5 Autonomous AI 380 Business AI 380 challenges 371–2, 388–97 China 386–7 and data mining 373 developed countries, dominance of 371–2 development opportunities 370–71 Internet AI 380 legal and technological developments, mismatch between 371

Perception AI 380 product liability law 371 superpowers, influence of 372 UK statutory definition 387–8 UK statutory definition, National AI Strategy 388 WTO TRIPS Agreement 395–6 artificial intelligence (AI), national regulatory approaches, EU statutory definition and flexible, coordinated approach 380–86 AI Act proposal 363–5, 383, 384–5, 391 AI package as global gold standard 382–3 Communication on AI for Europe 381–2 competitive advantage 385–6 global hub aim 383, 385 novel risk-based approach 383–6, 399 proactive approach 385 safety and liability implications of AI 382 artificial intelligence (AI), national regulatory approaches, pacing problem and goods/ services dichotomy 371, 389–95 digital products, place of 390 legal gaps and uncertainty 391–5 legal personhood of robot lawyers 395 preferential/regional trade agreements (PTAs/RTAs) 391 presence or absence of physical forms 395 solutions 390–91 website-based AI lawyers 392–5 artificial intelligence (AI), national regulatory approaches, US statutory definition and proactive and future-proofing approach 376–80 Algorithmic Accountability Act 378 and EU–US Trade and Technology Council (TTC) 380 federal agencies, guidance for 378–9 FUTURE of Artificial Intelligence Act 377–8 John S. McCain National Defense Authorization Act 378 National Artificial Intelligence Initiative Act 379–80 sector-specific policy guidance 379 voluntary consensus standards 379 weak and general technologies 379–80 ASEAN Agreement on Electronic Commerce 19 ASEAN Single Window (ASW) 260 Asia-Pacific Economic Cooperation (APEC) 20, 38, 48, 90, 121 Australia Australia–Hong Kong Free Trade Agreement (AUHKFTA) 171–2, 173 Australia–Indonesia FTA 45

476  Research handbook on digital trade Australia–Singapore DEA 20, 45–6, 49, 223, 225, 428–9 Australia–Singapore FTA 29, 422 Australia–Singapore PTA 61, 62, 65 Australia–Thailand PTA 70 Australia–US FTA 36 Moorhouse v University of New South Wales 467 automated profiling, algorithmic justice in the digital market 359–61, 365 Bacchus, J 20, 73, 76, 77, 78, 81, 83, 85, 87, 88 bad faith awards, international commercial arbitration in post-Covid era 409 banks central bank digital currencies (CBDCs) 326–7 challenger banks and fintech 321–2 base erosion and profit shifting (BEPS) digital services taxes (DST) 271 OECD Base Erosion and Profit Shifting (BEPS) project 181, 188–9 Bauer, C 292, 293, 299, 305 Bayraktaroğlu-Özçelik, G 403, 404, 405, 406, 408 Bergemann, K 203, 248 bias algorithmic justice in the digital market 348–9, 354, 355, 360, 365 gender equality see gender equality international commercial arbitration in post-Covid era 408, 411, 413 Bick, J 291, 298–9 Big Data, fintech 313–14 bilateral investment treaties (BITs), international investment agreements and digital assets 292–5, 297, 301, 304 bilateral or small-scale digital trade agreements, promotion of, Korea 186–9 Bitcoin, fintech, digital money 307, 310, 311, 325–6 black box tinkering and expert knowledge, algorithmic justice in the digital market 361 blockchain blockchain-based cryptocurrencies 325–6 copyright 469–70 and DLT infrastructure, WTO law and cross-border data flows 79–80 fintech 310–11, 316, 325–6 Latin America and SMEs 263–4 WTO law and cross-border data flows 75, 87–8 see also cryptocurrencies Bradford, A 18, 118 Broschek, J 135, 136

Bughin, J 370, 371, 373 Burri, M 2, 9–27, 29, 43, 50, 53, 54, 56, 57, 61, 64, 69, 71, 73, 74, 75, 77, 78, 79, 82, 87, 120, 121, 122, 123, 124, 125, 126, 130, 390, 423, 472 business trust digital economy agreements, e-commerce foundations 426 Digital Economy Partnership Agreement (DEPA) 93 economic security, Japan–US DTA 443–4 WTO law and cross-border data flows 85–6 Butler, N 5, 288–305 Calo, R 400, 414 Canada accession interest, Canada, Digital Economy Partnership Agreement (DEPA) 98–9 Canada–Chile FTA 111 Canada–Costa Rica FTA 39, 46 Canada–Israel FTA 112 e-invoicing and Universal Business Language (UBL) 99 EU–Canada Comprehensive Economic and Trade Agreement (CETA) 22, 120, 130 federalism see federalism, Canada Gender Based Analysis Plus (GBA+) 112, 116 Quebec and personal information protection 134 USMCA see US, United States–Mexico– Canada Agreement (USMCA) Caribbean Forum (CARIFORUM)–European Community EPA 41 censorship, China and WTO e-commerce negotiations 150 Chaisse, J 4, 199–216, 292, 293, 299, 305 challenger banks, digital-based, fintech 321–2 Chander, A 9, 17, 80, 205, 206 Chile Argentina–Chile FTA 46 Canada–Chile FTA 111 Chile–Hong Kong FTA 172 Digital Economy Partnership Agreement (DEPA) see Digital Economy Partnership Agreement (DEPA) US–Chile FTA 422 China AI blueprint 386–7 AI lawyers 393–5 AI statutory definition and flexible approach 386–7 as AI superpower 386, 387

Index  477 artificial intelligence (AI), national regulatory approaches 386–7 and ASEAN Single Window (ASW) 260 China–Africa Joint Arbitration Centre (CAJAC) 409 Cross-border E-commerce (CBEC) Pilot Zones (Cities) 212–14 Data Security Law 387 fintech 319 foreign direct investment (FDI) concerns 447–8 New Generation Artificial Intelligence Development Plan (AIDP) 386–7 Personal Information Protection Law 387 Shanghai Pilot Free Trade Zone (FTZ) 210 Special Economic Zones 208, 209, 212–14 state-issued central bank digital currencies (CBDCs) 327 see also Hong Kong China and WTO e-commerce negotiations 148–67 Alibaba and Electronic World Trade Platform (eWTP) 151, 154, 157 Belt and Road Initiative 159, 260 censorship 150 Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) application 16, 26, 167 critical information infrastructure 165–6 cross-cutting issues 151–2 customs duties moratorium 151–2, 158, 163 Cybersecurity Law 149, 160, 161, 165, 387 data flow and storage 159, 164–5, 182 data localisation 165–7 development cooperation measures 159–60 Digital Economy Partnership Agreement (DEPA) application 167 e-commerce volume growth 148–9, 151 e-signatures 150, 153, 156, 158, 161 electronic silk road 151 free trade agreements 150 Great Firewall 166, 205–6 internet regulations and sovereignty 148, 149, 158–9, 160 investment facilitation 154 market access 153, 156, 160 Nairobi Ministerial Conference 152 network equipment and products 162–3, 165–6 new proposals, dealing with 152–4 paperless trading 162 Position Paper on WTO Reform 155–63 public policy objectives 159–60, 161–2, 165, 167

and Regional Comprehensive Economic Partnership (RCEP) 167, 428 relationship with existing WTO Agreements 160–61 safety and security measures 158–9, 160, 161, 166 telecommunication issues 163 trade facilitation measures 158, 162 and Trade in Services Agreement (TiSA) 152, 155, 157 trade-related aspects of e-commerce, clarification request 158, 160 transparency and non-discrimination of trade policy 162 and World Economic Forum 153, 154 WTO E-Commerce Joint Statement Initiative (JSI) 151–6 WTO E-Commerce Joint Statement Initiative (JSI), Chinese proposals 156–63 WTO GATT and general exception clause 161, 165 Ciuriak, D 6, 55, 58, 60, 67, 71, 99, 174, 276, 279, 416–38 cloud facilities Cloud Computing Regulatory Framework (CCRF), Saudi Arabia 243–5 Cloud Security Assurance Program (CSAP), Korea 195–6 fintech 312–13 localisation, Korea 193 Collins, D 1–7, 90–101, 186, 270–87, 329–43 Colombia–Ecuador–EU–Peru FTA 49 commercial arbitration see international commercial arbitration in post-Covid era commercial trust see business trust communication to the public right, intellectual property 462, 464 competition policy and competitiveness artificial intelligence (AI), national regulatory approaches 385–6 digital economy agreements, e-commerce foundations 429–30 gender equality 107 Hong Kong, digital trade facilitation 170–71 international investment agreements and digital assets 290 Korea, digital trade agreements and policy 180 Latin America, Single Submission Portals and SMEs 252–3, 258, 259–64, 265–6 and preferential trade agreements (PTAs) 50 Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) 12–16, 21, 23, 24, 44, 57, 124, 130, 164, 391

478  Research handbook on digital trade China and WTO e-commerce negotiations 16, 26, 167 consumer protection 16, 337 customs duties moratorium 13, 16, 333 cybersecurity 16, 337 data localisation prohibition 337–8 and Digital Economy Partnership Agreement (DEPA) 90–91, 92, 93, 98 digital nomads and global labour mobility, SMEs 335–7 and economic security, Japan–US DTA 441–2, 443 electronic signatures 13 federalism 134 free data flows 14, 336, 338–9 gender equality 112 and Hong Kong, digital trade facilitation 169, 170, 173 Korea, digital trade agreements and policy 181, 184–5, 194, 195, 196, 197, 198 legal framework benchmarks, lack of 15 net neutrality issues 16 non-discriminatory treatment of digital products and National Treatment rule 336 paperless trading 13 privacy rights 15, 336–7 public policy objectives 14 SMEs 333, 335–40 source code 14–15, 16, 337 spam control 16 WTO law and cross-border data flows 78, 82 consent principle, African Continental Free Trade Area (AfCFTA) 222 consumer protection African Continental Free Trade Area (AfCFTA) 226 automated profiling 359–61, 365 Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 16, 337 digital economy agreements, e-commerce foundations 425–6, 428, 429 Digital Economy Partnership Agreement (DEPA) 93 digital nomads and global labour mobility, SMEs 337 e-commerce and digital trade, differences between 65, 66 gender equality 103, 106–7 preferential trade agreements (PTAs) 39–41 Saudi Arabia 243 cooperation China and WTO e-commerce negotiations 159–60

digital economy agreements, e-commerce foundations 428 Digital Economy Partnership Agreement (DEPA) 94, 96, 99 preferential trade agreements (PTAs) 39–40 preferential trade agreements (PTAs), privacy and data protection 46–7, 48 WTO law and cross-border data flows 88, 89 copyright law Hong Kong 174–5 and intellectual property see intellectual property, copyright Cory, N 164, 288 Costa Rica, Canada–Costa Rica FTA 39, 46 Cottier, T 10, 29, 247, 472 COVID-19 effects algorithmic justice in the digital market 346 digital nomads and global labour mobility 330, 331, 334, 340 fintech 325, 326 Hong Kong, digital trade facilitation 168, 170, 173 Latin America and SMEs 256–7 Middle Eastern countries 235–6, 237, 238 post-Covid era see international commercial arbitration in post-Covid era CPTPP see Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) cross-border data flows African Continental Free Trade Area (AfCFTA) 226, 230 China and WTO e-commerce negotiations 159, 164–5, 182 digital economy agreements 422, 432 digital economy agreements, e-commerce foundations 427 digital nomads and global labour mobility, SMEs 338–9 Hong Kong, digital trade facilitation 172 Korea, digital trade agreements and policy 182, 190–93 preferential trade agreements (PTAs) 43–5 Saudi Arabia 244 Special Economic Zones (SEZs) 205 transfer of information by electronic means, economic security, Japan–US DTA 443 and WTO see WTO law and cross-border data flows see also data protection; information exchange; personal data protection; privacy; security cross-border E-commerce Retail Import (CERI), Special Economic Zones (SEZs) 212–15

Index  479 cryptocurrencies blockchain-based 325–6 international investment agreements and digital assets 291 see also blockchain; fintech cryptography use 101, 310, 311, 426, 444 customs duties moratorium China and WTO e-commerce negotiations 151–2, 158, 163 Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 13, 16, 333 Digital Economy Partnership Agreement (DEPA) 92, 333 digital services taxes (DST) 274–5 economic security, Japan–US DTA 442 federalism 133–4, 145 gender equality 109–10 preferential trade agreements (PTAs) 34, 35–6 see also e-commerce cybersecurity African Continental Free Trade Area (AfCFTA) 224, 226 algorithmic justice in the digital market and artificial intelligence (AI) 354–5 China and WTO e-commerce negotiations 149, 160, 161, 165, 387 Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 16, 337 Digital Economy Partnership Agreement (DEPA) 93, 98–9 digital nomads and global labour mobility, SMEs 337 Egypt 241 EU as digital trade actor 119 Japan, Economic Security Promotion Act 452, 453 Jordan 242 preferential trade agreements (PTAs) 66 WTO law and cross-border data flows 82–3, 85 see also source codes data localisation African Continental Free Trade Area (AfCFTA) 226, 230–31 China and WTO e-commerce negotiations 165–7 Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 337–8 digital economy agreements, e-commerce foundations 427

Digital Economy Partnership Agreement (DEPA) 96 digital nomads and global labour mobility, SMEs 337–8 economic security, Japan–US DTA 444–6 federalism 136–7, 139, 144–5 Hong Kong, digital trade facilitation 172 Korea 190–93 preferential trade agreements (PTAs) 45–6 Special Economic Zones (SEZs) 205–6 WTO law and cross-border data flows 85 data mining 60, 65–6, 174–5, 373 ‘data poisoning’ use, algorithmic justice in the digital market 364 data protection African Continental Free Trade Area (AfCFTA) 224, 226 Digital Economy Partnership Agreement (DEPA) 92–3, 95–6 federalism 144 gender equality 107, 111 Middle Eastern countries 238–45 preferential trade agreements (PTAs) see preferential trade agreements (PTAs), privacy and data protection see also cross-border data flows data-driven economy (DDE), digital economy agreements 416, 417, 421 Daza Jaller, D 219, 224, 227, 234 DEPA see Digital Economy Partnership Agreement (DEPA) Desai, D 347, 361 developing countries African Continental Free Trade Area (AfCFTA) 226 digital services taxes (DST) 271 e-commerce and digital trade, differences between 70 Korea, digital trade agreements and policy 182–3, 198 Latin America and SMEs 254 see also indigenous people; individual countries development opportunities, artificial intelligence (AI), national regulatory approaches 370–71 digital brain fog, African Continental Free Trade Area (AfCFTA) 227–8 digital divide 51, 81, 105, 157, 159, 182 digital economy agreements 416–38 artificial intelligence and machine knowledge capital 435–6 commercial exploitation concerns 433 cross-border data flows 422, 427, 432 cryptography use 426

480  Research handbook on digital trade data-driven economy (DDE) 416, 417, 421 Digital Economic Partnership Agreement (DEPA) 417, 431, 434, 435, 437 digital product classification 431–2 digital services taxes (DST) 433–4 goods and services, blurring of distinction between 416, 431–2 intellectual property and WTO TRIPS 434 international investment agreements and digital assets 289–91 labour markets 436 patents 435 product durability classification 432 security issues and exceptions 436–8 social and political responses 416–17 value assessment 416, 419, 432–4 ‘winner-take-most’ contest 416, 421 WTO moratorium on electronic transmissions 431, 433 digital economy agreements, e-commerce foundations 421–30 business trust measures 426 competition policy 429–30 consumer-oriented issues 425–6, 428, 429 cooperation and future development 428 cross-border data flows and localisation of data storage 427 and Digital Economy Partnership Agreement (DEPA) 428, 430 FinTech 429 infrastructure support 427–8 and innovation 430 market access 424, 431 mobile roaming charges and digital identities 428 paperless trading 422 public sector management 430 regional trade agreements (RTAs) 423, 424–5, 427 RegTech start-ups 429–30 trade facilitation 424–5, 428–9 UNCITRAL Model Law on Electronic Signatures 422 WTO Joint Initiative on E-Commerce 423 digital economy agreements, technological and economic context 418–21 data as economic resource 418–19 data ownership 418 economic rents and trade conflicts 419–21, 434 and geopolitical rivalry 421 and Internet of Things (IoT) 419 and knowledge-based engineering (KBE) 420–21 and ‘superstar firms’ 421

Digital Economy Partnership Agreement (DEPA) 20–21, 32, 49, 55, 72, 90–101, 129, 333, 335, 339–40 accession interest 97–101 accession interest, Canada 98–9 accession interest, Korea 99–101 artificial intelligence (AI) governance 93–4, 96, 99 business trust 93 China and WTO e-commerce negotiations 167 and Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 90–91, 92, 93, 98 consumer protection 93 cryptography use 101 customs duties on digital transmissions 92, 333 cybersecurity and online safety 93, 98–9 data localisation 96 data protection 92–3, 95–6 and digital economy agreements, e-commerce foundations 428, 430 digital identities, cooperation on 93 dispute settlement rules 94 e-invoicing 91, 96, 99 economic integration, digital-only approach 95–7 electronic trade facilitation 92 environmental measures 95 gender equality 94, 99, 113 innovation, creativity and cooperation 94, 96, 99 Joint Committee 95, 97, 98 Korea, digital trade agreements and policy 181, 187–8 market access 90–91, 95, 97 national treatment provision 92 non-discrimination of digital products 92 open government data 94 overlap and inconsistency concerns 97 paperless trade 92 and Regional Comprehensive Economic Partnership (RCEP) 100 smart borders 92 SMEs, recognition of importance of 94, 335, 339–40 trade in goods 92 women and indigenous people inclusion 94, 99, 113 and WTO Joint Statement Initiative 90, 100 WTO law and cross-border data flows 78 WTO and substantial sectoral coverage consideration 97

Index  481 Digital Free Trade Zones (DFTZs), Special Economic Zones (SEZs) 214 digital money see fintech, digital money digital nomads and global labour mobility 329–43 COVID-19 pandemic effect 330, 331, 334, 340 digital isolation and labour markets 331–3 digital services taxes (DSTs) 332 e-residency schemes 330–31 immigration rules 330, 330–31, 332 market access 338, 340, 342–3 paperless trading 333, 334 and preferential FTAs 332, 335 professional qualifications, relevance of 332 public policy objectives 337, 338 regulatory responses 332, 334, 335 self-employment 329, 331 temporary and outsourced work 331 videoconferencing 330 WTO and customs duties on e-transmissions moratorium 332–3 WTO and supply of services 332 digital nomads and global labour mobility, SMEs 333–41 consumer protection 337 and CPTPP 335–7 cross-border data transfer 338–9 cybersecurity and source codes 337 and data localisation prohibition 337–8 and preferential FTAs 335–40 risk factors 334 and strict privacy rules 334, 335, 336–7, 338–9 and WTO 340–41 digital services taxes (DST) 270–87 base erosion and profit shifting (BEPS) 271 as destination-based taxes 272 developing countries and unilateral DSTs 271 digital economy agreements 433–4 digital nomads and global labour mobility 332 global tax treaty negotiation 271–2 physical presence, lack of 270, 272 services, impact on 272–3 US objections and tariffs 271–2 ‘user-created’ data 271, 280 value assessment 271, 273–4, 279–80, 284, 285 digital services taxes (DST), as trade barrier under WTO rules 273–86 goods and services, distinction between 273 internet search engines 277 mode of supply identification 277–8

moratorium on customs duties on electronic transmissions 274–5 most-favoured-nation (MFN) obligation 275, 280–81, 282 National Treatment commitments 275–80, 282, 286 non-discrimination principle 275–8, 283–4 online games 277 political protection 273–4 third-party technology provider 277 trade in services affect 280 UK example 276–7, 282, 283–4 digital services taxes (DST), as trade barrier under WTO rules, GATS compatibility 275–80 likeness concept 278–80 National Treatment commitments 278–80, 285 supplier likeness 278–9 US as complainant example 279–80 WTO Declaration on Global Commerce and GATS 275 digital services taxes (DST), as trade barrier under WTO rules, GATS exceptions 281–6 compliance with domestic laws and direct taxes, securing 281–3 double taxation and international treaties 285–6 gate-keeping provision 283–6 gate-keeping provision, discriminatory effect 283–4 gate-keeping provision, thresholds 284 policy purpose and proportionality 284–5 trade-restrictiveness of a measure 284, 285 Dimitropoulos, G 199, 201, 204 Dine, C 4, 218–32 discrimination see non-discrimination distributed ledger technology (DLT) see blockchain double taxation 285–6 see also digital services taxes (DST) Dunkley, E 321, 323 Dutfield, G 456, 457, 460, 462 Duval, Y 257, 262 e-commerce African Continental Free Trade Area (AfCFTA) 226–7 China and WTO e-commerce negotiations 148–9, 151 Cross-border E-commerce (CBEC) Pilot Zones (Cities) 212–14 Cross-border E-commerce Retail Import (CERI) 212–15

482  Research handbook on digital trade customs duties moratorium see customs duties moratorium digital economy agreements see digital economy agreements, e-commerce foundations United Arab Emirates, E-Commerce Guide 235 e-commerce and digital trade, differences between 54–72 analog trade via the internet 58 categorisation and rulemaking issues 56–7 dataset 61–2 digitally-enabled trade versus trade in data 55–61 digitally-enabled trade versus trade in data, trade in data as trade, issues over 59–60 information exchange 58, 59–60, 65–6 non-trade concerns 60 paperless trading 63, 64–5, 70 policy concerns and norm-equivalence 58–60 privacy concerns 57, 60, 69 public policy objectives 64, 65 WTO e-commerce definition 57 e-commerce and digital trade, differences between, future direction and analytical separation 69–72 developing states 70 multilateral negotiations 71 norm colonisation prevention 71–2 normative development in PTAs 70–71 source code rules 71 e-commerce and digital trade, differences between, preferential trade agreements (PTAs) and shifting norms 61–6 consumer protection equivalence 65, 66 conversion challenges 64–5 cybersecurity 66 e-signatures 63, 64–5 general WTO rules and e-commerce 63–4, 65 information transfer and data mining 60, 65–6 technological neutrality of trade norms 63 trade facilitation 63 transparency rules 63–4 and WTO GATT rules 64 e-commerce and digital trade, differences between, trade in data rules 66–9 asset creation and value protection 66, 67–8, 69 data privacy concerns 69 economic and political economy structures 68–9 market access rights 56–7, 63, 64, 66, 68–9

source code protection 67–8 and user control 67 e-invoicing, Digital Economy Partnership Agreement (DEPA) 91, 96, 99 e-residency schemes 330–31 see also digital nomads and global labour mobility e-signatures 17, 22, 25, 186, 187, 334 African Continental Free Trade Area (AfCFTA) 221, 227, 230 China and WTO e-commerce negotiations 150, 153, 156, 158, 161 Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 13 e-commerce and digital trade, differences between 63, 64–5 Egypt 236 preferential trade agreements (PTAs) 36–7 UNCITRAL Model Law on Electronic Signatures 422, 424 see also paperless trading Eco-Industrial Parks (EIPs) 209–10 economic growth 58, 94 and gender equality 111, 112, 115 and international investment agreements and digital assets 288, 289, 290–91, 297 and Special Economic Zones (SEZs) 199, 202 economic rents and trade conflicts, digital economy agreements 419–21, 434 economic security 439–54 data and national security 446–8 digital trade definition 439 and foreign direct investment (FDI) 446 LINE and data handling 448 private sector Big Four influence 440 US Foreign Investment Risk Review Modernization Act (FIRRMA) 446–8 WTO Joint Statement Initiative on E-Commerce 440, 441 see also security economic security, Japan, Economic Security Promotion Act 449–53 cybersecurity 452, 453 digital trade impact 453 secret patents 452–3 and strategic indispensability 451 supply chains 451–2, 453 technology base, strengthening 452 economic security, Japan–US DTA 19, 20, 49, 124, 194, 441–6 and Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) 441–2, 443

Index  483 cross-border transfer of information by electronic means 443 cryptography use 444 customs duties 442 Data Free Flow with Trust (DFFT) initiative 440, 442–6, 449, 454 data localisation 444–6 location of computing facilities 443 non-discriminatory treatment of digital products 442 source code 443 trust provisions 443–4 WTO goods/services transaction 441, 445–6 economy agreements see digital economy agreements Economy Partnership Agreement see Digital Economy Partnership Agreement (DEPA) Ecuador, Colombia–Ecuador–EU–Peru FTA 49 Egan, M 135, 141, 144 Egypt see Middle Eastern countries, Egypt Eidenmüller, H 400, 406 Elms, D 95, 98 Elsig, M 10, 11, 65, 78 Engman, M 201, 203 environmental issues algorithmic justice in the digital market and artificial intelligence (AI) 357, 362, 365 Digital Economy Partnership Agreement (DEPA) 95 Special Economic Zones (SEZs) 202, 209–10 ethics algorithmic justice in the digital market 351, 353, 355, 358–9 international commercial arbitration in post-Covid era 411–12 preferential trade agreements (PTAs) 51–2 EU AI flexible approach see artificial intelligence (AI), national regulatory approaches, EU statutory definition and flexible, coordinated approach Artificial Intelligence Act proposal 363–5, 383, 384–5, 391 Caribbean Forum (CARIFORUM)–European Community EPA 41 Colombia–Ecuador–EU–Peru FTA 49 Communication on AI for Europe 381–2 Computer Directive 396–7 Copyright in the Digital Single Market Directive 468 Database Directive 396–7 Digital Markets Act 119–20, 285 Digital Services Package 366–7

Digital Single Market Directive 461 Digital Strategies 121 E-Commerce Directive 29, 468 ethical guidelines on AI 351, 353, 355 European Strategy for Data 119 European Union Intellectual Property Office (EUIPO) 461 EU–Canada Comprehensive Economic and Trade Agreement (CETA) 22, 120, 130 EU–Egypt FTA 246 EU–Japan Economic Partnership Agreement (EPA) 22, 120, 124–5, 130–31 EU–Mexico Global Agreement 22 EU–South Korea FTA 22 EU–UK Trade and Cooperation Agreement (TCA) 120, 125, 129–32 EU–US Joint Agenda for Global Change proposal 126, 127 EU–US Safe Harbour Agreement 15 EU–US Trade and Technology Council (TTC) 125–9, 380 General Data Protection Regulation (GDPR) 23–4, 81, 93, 118, 121, 123, 126, 131–2, 134, 138, 182, 206, 239–40, 285, 334–5, 337, 364–5, 380–81, 387–8, 440 High-Level Expert Group on Artificial Intelligence 412 Information Society Directive 458, 463, 464 privacy protection and Brussels effect 17–18 Software Directive 464 UK and post-Brexit Trade and Cooperation Agreement (TCA) 24 WTO Joint Statement on Electronic Commerce 110–11, 114, 115 EU, cases Digital Rights Ireland 118 Google Spain SL, Google v Agencia Española de Protección de Datos 118 Schrems 15, 118, 128, 206–7 Software Incubator v Computer Associates 395 Tom Kabinet 465 UsedSoft 464–5 Weltimmo s.r.o. v Nemzeti Adatvédelmi és Információszabadság Hatóság 118 EU as digital trade actor 118–32 artificial intelligence (AI) 119–20 cybersecurity 119 digital sovereignty 118–19, 125 horizontal strategy for cross-border data and model clauses 123–5, 130 and international standards of data protection 121–2

484  Research handbook on digital trade privacy protection 121, 130 protectionism concerns 121 public policy objectives 122, 125 US and Transatlantic Trade and Investment Partnership (TTIP) 126 and WTO Joint Statement Initiative (JSI) on e-commerce 122–3 European Convention on Human Rights (ECHR) 24 European Free Trade Area (EFTA) 11 exhaustion of rights doctrine, and copyright 464, 465–6 export-processing zones (EPZs), Special Economic Zones (SEZs) 208–9 Fafard, P 135, 144 Fahey, E 3, 118–32 Farole, T 200, 208 federalism 133–46 Comprehensive and Progressive Agreement on Trans-Pacific Partnership (CPTPP) 134 customs duties moratorium 133–4, 145 data localisation 144–5 and digital trade protectionism 135 digital trade provisions in international trade agreements 135 dispute settlement 144 government procurement issues 145 non-tariff barriers (NTBs) 134 privacy concerns 139–40, 141–2, 144 and US-Mexico-Canada Agreement (USMCA) 143–5 value assessment 139, 143 federalism, Canada 136–40 British Columbia, Freedom of Information and Protection of Privacy Act (FOIPPA) 136–7, 139 data localisation 136–7, 139 digitally delivered services, purchase of 139 personal information protection 136–7 privacy protection 139–40 public procurement 139–40 Quebec 138, 139, 145 federalism, US 140–43 artificial intelligence (AI) 378–9 and Big Tech firms 141 California Consumer Privacy Act (CCPA) 140, 141 Federal Risk and Authorization Management Program (FedRamp) 142 government procurement for cloud services 142 online sellers (digitally ordered) products and services 142–3

state-level privacy legislation 140, 141–2 Streamlined Sales and Use Tax Agreement (SSUTA) 143 feedback loops, measures against, and artificial intelligence (AI) 365 Fefer, R 150, 206, 290 Ferencz, J 291, 455, 463, 468, 472 Ferracane, M 120, 123, 133, 134, 135 fiat money (legal tender), fintech 324–5 Finger, M 279, 287 fintech 306–28 as alternative finance for digital trade 320–24 Ant Group example 315 artificial intelligence (AI) 309–10, 316 Big Data 313–14 Big Data, fraud detection 314 BigTech companies and economic efficiency and digital trade 315–16 blockchain 310–11, 316, 325–6 blockchain, peer-to-peer method of data storage and sharing 311 cloud computing 312–13 cloud computing, categories 312 Covid-19 pandemic effects 325, 326 cryptocurrencies see cryptocurrencies cryptography use 310, 311 definition 308 and digital economy development 314–15, 429 digital-based challenger banks 321–2 digital-based challenger banks, and SMEs 322 global financial crisis effects 321–2, 325 global hubs 317–20 Metaverse and non-fungible tokens (NFTs) 316–17 online peer-to-peer (P2P) lending platforms 322–3, 326 online peer-to-peer (P2P) lending platforms, purchasing thresholds 323–4 online peer-to-peer (P2P) lending platforms, safety and security 324 UN Digital Financing Task Force 308 value assessment 323, 325, 326, 327 fintech, digital money 324–7 Bitcoin 307, 310, 311, 325–6 Bitcoin, volatile pricing concerns 326 blockchain-based cryptocurrencies 325–6 and decentralisation 326 fiat money (legal tender) 324–5 stablecoin 326 state-issued central bank digital currencies (CBDCs) 326–7 firewall, Great Firewall, China 166, 205–6 first sale doctrine, copyright 463, 465–6

Index  485 foreign direct investment (FDI) 207, 236, 292, 446, 447–8 Forsgren, C 274, 283, 286 France AI-based arbitrators, barring of 405 fintech 318–19 US–France Bilateral Income Tax Treaty 286 fraud, Big Data, fraud detection 314 free trade agreements 9–27 CPTPP see Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) digital economy agreements (DEAs) 19–21 Digital Economy Partnership Agreement (DEPA) 20–21 EU approach 22–4 Latin America and SMEs 262–3 Middle Eastern countries 245–50 open government data 19, 20, 24 RCEP see Regional Comprehensive Economic Partnership (RCEP) United States–Mexico–Canada Agreement see United States–Mexico–Canada Agreement see also individual countries; preferential trade agreements (PTAs) Free Trade Zones (FTZs), Special Economic Zones (SEZs) 207, 208–9, 210–11 Freeman, K 347, 349 Frieden, K 270, 274, 276 Gantz, D 336, 422 Gao, H 4, 11, 120, 122, 148–67, 200 gatekeepers algorithmic justice in the digital market and artificial intelligence (AI) 366–7 digital services taxes (DST) 283–6 Geist, M 1–7, 102–16, 271, 272, 283, 286, 338, 433, 459 gender equality 102–16 artificial intelligence (AI) barriers 108 competition laws 107 Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) 112 consumer protection 103, 106–7 data protection and privacy 107, 111 developing countries 104, 105–7, 113–14 Digital Economy Partnership Agreement (DEPA) 94, 99, 113 discrimination and harassment 106, 107, 111 e-commerce customs moratorium 109–10 and economic growth 111, 112, 115 financial challenges 104 future direction 115–16

government procurement 104–5 international commercial arbitration in post-Covid era 408, 411, 413 International Finance Corporation (IFC) Scale X 104 IT infrastructure issues 105 North American Agreement on Labour Cooperation 114 regional trade agreements 111–12 self-employment issues 106 and SMEs 102–3, 104, 105, 107, 113, 114 trade provisions 108–15 UN Sustainable Development Goal 112 Women Entrepreneurs Finance Initiative (We-Fi) 104 and workplace flexibility 103, 105–6 WTO Work Programme on Electronic Commerce 109–11 Germany–Pakistan BIT 294 Global AI Index 99 global financial crisis effects, fintech 321–2, 325 Global Partnership on Artificial Intelligence 51 global value chains 135, 157, 205, 255, 438 Goff, P 135, 136 goods/services dichotomy 11, 273, 416, 431–2 pacing problem see artificial intelligence (AI), national regulatory approaches, pacing problem and goods/services dichotomy see also WTO headings government procurement 12, 14, 21, 25, 44, 104–5, 142, 145 Guadamuz, A 279, 470 Guglya, L 335, 337 Guimarães, M 135, 141 Gutierrez, J 466, 468 Hamilton, D 126, 128 Hederer, C 135–6, 144 Hepburn, J 295, 298 Herreros, S 257, 258, 259, 261, 266 Hirani, H 1, 104, 107 Honduras, Special Economic Zones (SEZs) 215 Hong Kong, digital trade facilitation 168–77 Australia–Hong Kong Free Trade Agreement (AUHKFTA) 171–2, 173 Belt and Road Initiative (BRI) 171 Chile–Hong Kong FTA 172 and Comprehensive and Progressive Trans-Pacific Partnership (CPTPP) 169, 170, 173 COVID-19 pandemic effects 168, 170, 173 data flows and new technologies and new product development 171, 175–6 data localisation 172

486  Research handbook on digital trade data transfer rules 172 fintech 319 national competitiveness considerations 170–71 Personal Data (Privacy) Ordinance (PDPO) 171, 176 public policy objectives 172, 174 and Regional Comprehensive Economic Partnership (RCEP) 168–9, 170 security and ethical and governance frameworks 173–4 and SMEs 171 supply-chain networks 170 see also China Hong Kong, digital trade facilitation, cross-border data policy recommendations 171, 173–7 copyright law and text and data mining (TDM) exemption 174–5 further commitments and updated agreements 173 holistic, coordinated approach 175–6 impact assessment matrix and checklist development 176 policy formulation and implementation think tank 176–7 Horváth, E 289, 293, 299, 300 Hu, J 203, 209 Hufbauer, G 122, 272, 274, 283 Hulicki, M 6, 345–68 human adjudicators, replacement of, international commercial arbitration in post-Covid era 400–402, 408–9 human agency and oversight, algorithmic justice in the digital market 354, 356, 357, 362, 364, 365, 366–7 Iacob, N 455, 463 ICSID Abaclat and Others v Argentine Republic 297, 299–300 ADC Affiliate Limited v Hungary 302 Alpha Projektholding v Ukraine 300 CSOB v Slovakia 301–2 Inmaris Perestroika Sailing Maritime Services v Ukraine 300, 302 Joy Mining Mach. Ltd. v Egypt 296, 302 Koch Minerals v Venezuela 302 LESI v Algeria 296 Mamidoil Jetoil Greek Petroleum Products v Albania 302 Phoenix Action v Czech Republic 296 Quiborax 296 Saba Fakes 296 Saipem v Bangladesh 302 Salini 294, 296–7, 301

SGS v Philippines 300 Vestey Group Ltd v Venezuela 302 Victor Pey Casado v Chile 296 Idrees, M 6, 400–414 immigration rules, digital nomads and global labour mobility 330–31, 332 India digital Special Economic Zone (SEZ) 214–15 India–Saudi Arabia CEPA 44–5 indigenous people Digital Economy Partnership Agreement (DEPA) 94, 99, 113 see also developing countries Indo-Pacific Economic Framework (IPEF) 100, 129, 181, 185–6 Indonesia, Australia–Indonesia FTA 45 industrial parks, Special Economic Zones (SEZs) 207–8 information exchange e-commerce and digital trade, differences between 58, 59–60, 65–6 international commercial arbitration in post-Covid era 413 preferential trade agreements (PTAs), privacy and data protection 48 see also cross-border data flows information society concept, algorithmic justice in the digital market 348 Information Technology Agreement (ITA), African Continental Free Trade Area (AfCFTA) 226–7 infrastructure support, digital economy agreements, e-commerce foundations 427–8 innovation and digital economy agreements, e-commerce foundations 430 Digital Economy Partnership Agreement (DEPA) 94, 96, 99 Institute of Electrical and Electronics Engineers (IEEE) 85 integrity principle, African Continental Free Trade Area (AfCFTA) 223–4 intellectual property 455–73 African Continental Free Trade Area (AfCFTA) 226, 230 Anti-Counterfeiting Trade Agreement (ACTA) 459, 472 communication to the public right 462, 464 digital distribution 458, 462, 465, 466, 471 digital economy agreements 434 Digital Rights Management (DRM) protection 458 and digital trade 461–3

Index  487 and digitisation 457–9 emerging issues 468–72 enforcement issues 457 as exclusive right 456–7, 461, 462–3 and free riding problems 456–7 gatekeeper responsibility for infringements 466–8, 472–3 gatekeeper responsibility for infringements, internet infringements 467 making available right 458 patents 435, 452–3, 456 royalties 461 safe harbour provisions 468 segregation of markets and selective licensing 462–3 source code 466 Technological Protection Measures (TPMs) protection 458 territoriality principle 459 trade barriers, elimination of 460–61 trade law effects 459–61 trade mark protection 457 venture capital, securing 462 WIPO Internet Treaties 458, 462, 463, 464 WTO TRIPS Agreement 341, 395–7, 434, 458–9, 460, 472 intellectual property, copyright 456, 463 3D printing technology 471–2 blockchain (distributed ledger technology (DLT)) 469–70 digital second-hand markets and exhaustion of copyright 463–6 and exhaustion of rights doctrine 464, 465–6 first sale doctrine 463, 465–6 non-fungible tokens (NFTs) 469–71 and smart contracts 469 Inter-American Develoment Bank (IADB) 256–7, 261, 262, 263 Inter-American Network of International Trade single windows (RedVUCE) 263 international commercial arbitration in post-Covid era 400–414 African AI arbitrators, implications of, accessibility concerns 410 amendments to legal instruments 412–13 appeals 411 arbitrator appointment 402–3, 404–5 arbitrator appointment, as natural person 402 bad faith awards and responsibility 409 confidentiality concerns 412 data publication recommendations 413 enforcement of arbitral award rendered by AI arbitrator 403–5

enforcement of arbitral award rendered by AI arbitrator, contravention of law of seat of arbitration 405 enforcement of arbitral award rendered by AI arbitrator, selection of arbitrators in accordance with arbitral agreement 404–5 ethical considerations 411–12 fairness of proceedings 411 future of arbitration and replacement of human arbitrators 408–9 future of arbitration and replacement of human arbitrators, predictive decisions 408 gender or racial bias concerns 408, 411, 413 human adjudicators, replacement of 400–402, 408–9 public policy requirement 405–7 public policy requirement, and absence of reasoned decision 406–7 public policy requirement, and AI arbitrator award 406, 407 recommendations 412–13 security concerns 412 unemployment concern 408, 410 validity of award 411–12 International Criminal Court (ICC), arbitrator appointment 403 International Finance Corporation (IFC) Scale X 104 international investment agreements and digital assets 288–305 bilateral investment treaties (BITs) 292–5, 297, 301, 304 competitiveness 290 covered investment 293–5 covered investment, intangible assets inclusion 294 cryptocurrencies 291 data privacy and cyber security 289 digital assets definition and categories 291–2, 298–9 digital economy 289–91 and economic growth 288, 289, 290–91, 297 ICSID decided cases 295–7, 301, 302–3 international investment (FDI) law 292–3 investment as contribution to host state’s economy requirement 296–7 protected investments, issues with 302–3 spaghetti bowl network 292 territorial requirement 298–302 territorial requirement, categorisation 298–9 territorial requirement, focal point location 300–301

488  Research handbook on digital trade territorial requirement, investors as minority shareholders 300 treaties with investment provisions (TIPs) 292, 293–5, 301, 304 International Telecommunications Union (ITU) 85 internet, source codes see source codes Internet Engineering Task Force (IETF) 85 internet regulations and sovereignty, China 148, 149, 158–9, 160 Internet of Things (IoT), and digital economy agreements 419 internet use African Continental Free Trade Area (AfCFTA) 226 Middle Eastern countries 234–8, 249 investment agreements see international investment agreements and digital assets Iran, Slovakia–Iran BIT 294–5 Irion, K 24, 124 Ismail, Y 1, 104, 107, 156, 182 Israel, Canada–Israel FTA 112 Janow, M 56, 431, 432 Japan Data Free Flow with Trust (DFFT) initiative 440, 442–6, 449, 454 Economic Security Promotion Act see economic security, Japan, Economic Security Promotion Act EU–Japan Economic Partnership Agreement (EPA) 22, 120, 124–5, 130–31 Japan–Mongolia FTA 45 Japan–UK FTA 44 Japan–US DTA see economic security, Japan–US DTA Limitation of Liability for Damages Act 443–4 Ji, X 199, 205, 207 Jordan see Middle Eastern countries, Jordan Jouanjean, M-A 9, 56, 58, 59, 256, 439 Klinkmüller, S 289, 293, 299, 300 Klotz, S 10, 11, 65, 78 knowledge transfer, Special Economic Zones (SEZs) 200–201 knowledge-based engineering (KBE), and digital economy agreements 420–21 Ko, B 99, 111, 112 Korea Data Industry Act 192 Digital Economy Partnership Agreement (DEPA) 99–101 Digital New Deal 99

Electronic Financial Supervisory Regulations 193 EU–South Korea FTA 22 Korea–Singapore DEA 20, 100 Korea–Singapore Digital Partnership Agreement (KSDPA) 100, 181, 186–7, 194, 195 Korea–US FTA 42, 44, 99–100, 190, 191 Personal Information Protection Act 192–3 smart industrial zones 211 Korea, digital trade agreements and policy 178–98 bilateral or small-scale digital trade agreements, promotion of 186–9 Cloud Security Assurance Program (CSAP) 195–6 competitiveness and digital competency 180 Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) 181, 184–5, 194, 195, 196, 197, 198 cross-border data flow and security 182 cross-Border data transfer and data localisation prohibition 190–93 developing and developed economies, concerns over digital divide 182–3, 198 Digital Economy Partnership Agreement (DEPA) consideration 181, 187–8 Digital New Deal 180 digital trade agreement implementation 190–98 Indo-Pacific Economic Framework (IPEF) consideration 181, 185–6 international norms creation, involvement in 181–9 international tax sector involvement 189 internet mass media industry and source codes 197–8 localisation of cloud facilities in the public sector 193 location of computing facilities 184, 185, 187, 191 market access 181, 182, 187, 194 multilateral negotiation limitations 180–81 OECD Base Erosion and Profit Shifting (BEPS) project 181, 188–9 public policy objectives 182, 183–4, 185, 187, 190, 191, 192, 193 recent movements in digital trade 180–81 Regional Comprehensive Economic Partnership (RCEP) 181, 183–4, 190, 191, 192, 194 Regional Comprehensive Economic Partnership (RCEP), e-commerce promotion 183–4

Index  489 regulatory or judicial authorities, exceptions related to 196–7 source code disclosure 193–8 trade liberalisation policies 179 WTO Electronic Commerce negotiations 181–3 Kroll, J 347, 361 labour mobility see digital nomads and global labour mobility Latin America and SMEs 252–68 automation of processes leading to internationalisation 264–7 blockchain technology use 263–4 CADENA Project 263–4 COVID-19 pandemic effects 256–7 developing countries, constraints on 254 Digital Agenda 255 digital trade definition 256 digital trade facilitation 257, 264–7 digital transformation 253–5 Economic Commission for Latin America and the Caribbean (ECLAC) 257–9, 261, 263, 267 free trade agreements 262–3 Inter-American Network of International Trade single windows (RedVUCE) 263 Mercosur bConnect project 264 Ministerial Conference on the Information Society in Latin America and the Caribbean 254–5 National Development Plans (NDP) 255 paperless trade 258–9, 262–3 Port Community System platforms 261–2 Single Submission Portals and competitiveness 252–3, 258, 259–64, 265–6 Single Submission Portals and competitiveness, interoperability barriers 261, 262, 263 technology incorporation problems 262 trade policies enforcement and domestic reforms 255–9 lawyers, website-based AI lawyers 392–5 Leblond, P 3, 57, 58, 59, 60, 81, 133–46 Lee, J 3, 90–101 Lee, K-F 369, 370, 372, 373, 378, 379–80, 386, 392, 393, 395, 397 Lemma, A 219, 220, 224, 226, 228, 230 Lessig, L 346, 358 likeness concept, digital services taxes (DST) 278–80 Lin, C-F 392, 393, 394, 395, 397 LINE app economic security 448

Liu, H-W 392, 393, 394, 395, 397 location of computing facilities 25, 45, 46, 62, 184, 427 Japan–US DTA 443 Korea, digital trade agreements 184, 185, 187, 191 López González, J 9, 28, 56, 58, 59, 256, 265, 291, 423, 439, 455, 468, 472 Lu, L 5, 122, 306–28 Lu, Z 274, 283 McCarthy, J 309, 345, 369, 374, 375, 400 Maciel, M 335, 337 Magwape, M 271, 280 Malaysia, Digital Free Trade Zone (DFTZ) 214 Malkawi, B 4–5, 233–51 Marjanovic, O 350, 351, 359, 360, 361 market abuse issues 107, 285, 361 market access China and WTO e-commerce negotiations 153, 156, 160 digital economy agreements, e-commerce foundations 424, 431 Digital Economy Partnership Agreement (DEPA) 90–91, 95, 97 digital nomads and global labour mobility 338, 340, 342–3 e-commerce and digital trade, differences between 56–7, 63, 64, 66, 68–9 Korea, digital trade agreements 181, 182, 187, 194 Middle Eastern countries 248 Mavroidis, P 56, 431, 432 Mazur, J 67, 121, 122 Meier-Ewert, W 466, 468 Meltzer, J 82, 144, 256 Mengjing, K 257, 262 Mercosur bConnect project, Latin America 264 Mercurio, B 4, 168–77 Metaverse and non-fungible tokens (NFTs), fintech 316–17 Mexico EU–Mexico Global Agreement 22 Mexico–Panama FTA 44 USMCA see US, United States–Mexico– Canada Agreement (USMCA) Mezei, P 464, 465, 469 Middle Eastern countries 233–51 data protection regulations 238–45 free trade agreements 245–50 internet usage levels 234–8 market access 248 Middle Eastern countries, Egypt 236–7, 240–41, 246 COVID-19 effects 236, 237

490  Research handbook on digital trade credit card use 236 Cyber Security Law 241 data licensing requirements 240 data protection 240–41 Electronic Signature Law (ESL) 236 EU–Egypt FTA 246 free trade agreements 246 internet use 236–7 personal data protection 240–41 Middle Eastern countries, Jordan 237, 241–2, 247–9 coordination and strategy improvements needed 237 COVID-19 effects 237 Cybercrime Law 242 data protection law, lack of 241–2 free trade agreements 247–9 internet use 237, 249 Jordan–US FTA 29, 35, 36, 43, 46, 247–9 Personal Data Protection Law proposal 241 technological skills 237 Telecommunications Law 241–2 Middle Eastern countries, Saudi Arabia 237–8, 242–5, 249–50 challenges to e-commerce use 238 Cloud Computing Regulatory Framework (CCRF) 243–5 consumer protection 243 COVID-19 effects 238 Credit Information Law 243 cross-border transfer of data and exclusions 244 data protection 242–5 E-Commerce Law 238, 243 free trade agreements 249–50 India–Saudi Arabia CEPA 44–5 Personal Data Protection Law 242 and Sharīʿah law 243 smartphone usage rates 238 US Trade and Investment Framework Agreement (TIFA) 249 Middle Eastern countries, United Arab Emirates 235–6, 239–40, 245–6 barriers to e-commerce 236 core digitisation initiatives 235 COVID-19 effects 235–6 data protection 239–40 data protection, exclusions 239 E-Commerce Guide 235 e-government services 236 fintech 320 free trade agreements 245–6 free zones 236 international data transfers 240 internet use 235

US Trade and Investment Framework Agreement (TIFA) 245–6 US–GCC Framework Agreement for Trade, Economic Investment, and Technical Cooperation 246 and WTO GATT 245 Mimler, M 7, 455–83 Mishra, N 58, 59, 60, 72, 81, 82, 83, 84, 85, 86, 89 Mitchell, A 56, 57, 81, 82, 83, 84, 85, 86, 89, 178, 239, 274, 275, 278, 279, 281, 282, 284, 287, 295, 298 mobile roaming charges, digital economy agreements, e-commerce foundations 428 money see fintech, digital money Mongolia, Japan–Mongolia FTA 45 Monteiro, J-A 112, 121, 422 Montero, J 279, 287 Montreal Declaration, algorithmic justice in the digital market 351, 352, 353, 357–8 most-favoured-nation (MFN) obligation, digital services taxes (DST) 275, 280–81, 282 Mukherjee, A 203, 214 Mulopulos, S 205, 207, 212 Murray, A 458, 466 Nakamoto, S 325, 469 national treatment provision Digital Economy Partnership Agreement (DEPA) 92 digital services taxes (DST) 278–80, 285 Neeraj, R 55, 69 Nemoto, T 28, 423 network equipment and products, China and WTO e-commerce negotiations 162–3, 165–6 network neutrality, WTO law and cross-border data flows 85 New York Convention 403, 404–5, 406, 409, 412–13, 414 New Zealand Digital Economy Partnership Agreement (DEPA) see Digital Economy Partnership Agreement (DEPA) New Zealand–Singapore Closer Economic Partnership Agreement (CEPA) 38, 43 New Zealand–Singapore FTA 422 UK–New Zealand FTA 114 Nicaragua, Taiwan–Nicaragua FTA 43 nomads, digital see digital nomads and global labour mobility non-discrimination principle African Continental Free Trade Area (AfCFTA) 222

Index  491 algorithmic justice in the digital market and artificial intelligence (AI) 356–7, 362, 365 China and WTO e-commerce negotiations 162 Digital Economy Partnership Agreement (DEPA) 92 digital services taxes (DST), as trade barrier under WTO rules 275–8, 283–4 economic security, Japan–US DTA 442 gender equality 106, 107, 111 non-fungible tokens (NFTs) copyright 469–71 and Metaverse 316–17 non-tariff barriers (NTBs) 134, 435, 460 North American Agreement on Labour Cooperation 114 North American Free Trade Agreement (NAFTA) 203 Norvig, P 309, 375–6 OECD AI trustworthy principles 351 Base Erosion and Profit Shifting (BEPS) project 181, 188–9 G20 Inclusive Framework 433–4 Going Digital Summit 254 Protection of Privacy and Transborder Flows of Personal Data 48 trustworthy principles 351, 356, 357–8 online games 277, 316 Onyefulu, A 6, 400–414 open government data African Continental Free Trade Area (AfCFTA) 223, 230 Digital Economy Partnership Agreement (DEPA) 94 free trade agreements 19, 20, 24 preferential trade agreements (PTAs) 52 Orel, M 329, 342 oversight and human agency, algorithmic justice in the digital market and artificial intelligence (AI) 354, 356, 357, 362, 364, 365, 366–7 Özçelik, Ş 403, 404, 405, 406, 408 Pacific Alliance Additional Protocol (PAAP) 40, 44, 45 pacing problem, AI see artificial intelligence (AI), national regulatory approaches, pacing problem and goods/services dichotomy Pakistan, Germany–Pakistan BIT 294 Panama, Mexico–Panama FTA 44 paperless trading

African Continental Free Trade Area (AfCFTA) 229 China and WTO e-commerce negotiations 162 digital economy agreements, e-commerce foundations 422 Digital Economy Partnership Agreement (DEPA) 92 digital nomads and global labour mobility 333, 334 e-commerce and digital trade, differences between 63, 64–5, 70 Framework Agreement on the Facilitation of Cross Border Paperless Trade in Asia and the Pacific (CPTA) 262 free trade agreements, Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 13 Latin America and SMEs 258–9, 262–3 preferential trade agreements (PTAs) 38–9 see also e-signatures Park, J 4, 95, 178–98 Park, K 401, 408 Park, T 4, 178–98 patents 435, 452–3, 456 see also intellectual property Pauwelyn, J 57, 431 peer-to-peer (P2P) lending platforms, fintech 322–3, 326 Peng, S 77, 437 personal data protection e-commerce and digital trade, differences between 60 federalism, Canada 136–7 Middle Eastern countries 240–41, 242 preferential trade agreements (PTAs) 44, 48–9 see also cross-border data flows Peru, Colombia–Ecuador–EU–Peru FTA 49 Pirlot, A 273, 281, 283, 286 platform work protection, algorithmic justice in the digital market and artificial intelligence (AI) 365–6 Polanco, R 2, 10, 12, 28–53, 54, 61, 62, 121, 423 policy concerns and norm-equivalence, e-commerce and digital trade, differences between 58–60 political economy structures, e-commerce and digital trade, differences between 68–9 political protection, digital services taxes (DST) 273–4 political responses, digital economy agreements 416–17 Port Community System platforms, Latin America and SMEs 261–2

492  Research handbook on digital trade positive-list approach 75–6 predictive decisions, international commercial arbitration in post-Covid era 408 ‘predominant purpose’ test 76–7 preferential trade agreements (PTAs) 28–53 artificial intelligence (AI) and ethical and governance frameworks 51–2 artificial intelligence (AI), national regulatory approaches 391 and competition law 50 consumer protection 39–41 consumer protection, adoption of domestic standards 40–41 consumer protection, and international cooperation 39–40 customs duties moratorium 34, 35–6 data flows, rules on 43–5 data localisation 45–6 digital divide 51 digital identities 50 and digital nomads and global labour mobility 332, 335–40 e-signatures 36–7 electronic financial transactions and security issues 37 evolution of 29–33 first generation and e-commerce 34–42 open government data 52 open government data, Application Programming Interfaces (APIs) 52 paperless trading 38–9 personal data protection 44, 48–9 and postal costs of delivery of parcels 34–5 public policy objectives 44, 45, 47 second generation and data flows rules 42–9 security concerns 37, 44 and shifting norms see e-commerce and digital trade, differences between, preferential trade agreements (PTAs) and shifting norms spam (unsolicited commercial electronic messages) 41–2 Special Economic Zones (SEZs) 203 Text of Trade Agreement (ToTA) 61 third generation and new data-driven economy 49–52 WTO law and cross-border data flows 78 WTO rules applicability 33–4, 35 see also free trade agreements; individual countries preferential trade agreements (PTAs), privacy and data protection 46–9 cooperation activities 46–7, 48 and free movement of data 49 information exchange 48

and international standards 47–8 personal data protection and domestic standards 47 personal information protection 48–9 treaty bodies’ involvement 49 privacy concerns algorithmic justice in the digital market and artificial intelligence (AI) 355, 362, 364 Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 15, 336–7 digital nomads and global labour mobility, SMEs 334, 335, 336–7, 338–9 e-commerce and digital trade, differences between 57, 60, 69 EU as digital trade actor 121, 130 federalism 139–40, 141–2, 144 gender equality 107, 111 preferential trade agreements (PTAs) see preferential trade agreements (PTAs), privacy and data protection Special Economic Zones (SEZs) 206–7 WTO law and cross-border data flows 81–2 see also cross-border data flows; security product durability classification, digital economy agreements 432 product liability law, artificial intelligence (AI), national regulatory approaches 371 professional qualifications, relevance of, digital nomads 332 proportionality, digital services taxes (DST) 284–5 protectionism EU as digital trade actor 121 federalism 135 international investment agreements and digital assets 302–3 Ptashkina, M 58, 60, 421, 434 public communication right, intellectual property 462, 464 public policy objectives China and WTO e-commerce negotiations 159–60, 161–2, 165, 167 Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 14, 17 digital nomads and global labour mobility 337, 338 e-commerce and digital trade, differences between 64, 65 EU as digital trade actor 122, 125 Hong Kong, digital trade facilitation 172, 174

Index  493 international commercial arbitration in post-Covid era 405–7 Korea, digital trade agreements and policy 182, 183–4, 185, 187, 190, 191, 192, 193 preferential trade agreements (PTAs) 44, 45, 47 Regional Comprehensive Economic Partnership (RCEP) 25–6 public procurement 104–5, 134, 139–40 Quindimil, M 5, 252–68 Radl, A 261, 263, 264 Regional Comprehensive Economic Partnership (RCEP) 24–6, 46, 124 and China and WTO e-commerce negotiations 167, 428 cross-border data flows 25–6 and Digital Economy Partnership Agreement (DEPA) 100 dispute settlement 26 and Hong Kong, digital trade facilitation 168–9, 170 Korea, digital trade agreements and policy 181, 183–4, 190, 191, 192, 194 paperless trading 25 personal information protection 25 public policy objectives 25–6 WTO law and cross-border data flows 78–9 regional trade agreements (RTAs) artificial intelligence (AI), national regulatory approaches 391 digital economy agreements, e-commerce foundations 423, 424–5, 427 gender equality 111–12 Special Economic Zones (SEZs) 203 RegTech start-ups 429–30 regulatory framework goal, algorithmic justice in the digital market 361–7 Rhim, Y-Y 401, 408 risk factors algorithmic justice in the digital market and artificial intelligence (AI) 354–5, 357, 362, 363–5 artificial intelligence (AI), national regulatory approaches 383–6, 399 digital nomads and global labour mobility, SMEs 334 preferential trade agreements (PTAs) 52 Robinson, B 5, 329–43 Rodionova, V 429, 435 royalties, intellectual property 461 Russell, S 309, 375–6

safe harbour provisions, intellectual property 468 safety measures algorithmic justice in the digital market and artificial intelligence (AI) 354–5, 362 artificial intelligence (AI), national regulatory approaches 382 China and WTO e-commerce negotiations 158–9, 160, 161, 166 online peer-to-peer (P2P) lending platforms 324 see also security concerns Samlidis, T 178, 239 Saudi Arabia see Middle Eastern countries, Saudi Arabia Schreuer, C 297, 300 Schwartz, P 9, 15, 17, 46 security concerns China and WTO e-commerce negotiations 158–9, 160, 161, 166 digital economy agreements 436–8 economic see economic security Hong Kong, digital trade facilitation 173–4 international commercial arbitration in post-Covid era 412 Korea, digital trade agreements and policy 182 online peer-to-peer (P2P) lending platforms 324 preferential trade agreements (PTAs) 37, 44 spam control 16, 41–2, 93 trade secrets, protection of 349, 360–61 see also cross-border data flows; privacy concerns; safety measures segregation of markets, intellectual property 462–3 self-employment 106, 329, 331 self-regulation consideration, WTO law and cross-border data flows 86 Sen, N 54, 56 services/goods dichotomy see goods/services dichotomy Seychelles, Electronic Transactions Act 224 Sganga, C 464, 466 Shaffer, G 9, 54, 60, 72, 73, 230 Sidhu, D 349, 360 Sim, C 401, 403, 404, 407, 411 Simonelli, F 455, 463 Singapore Australia–Singapore DEA 49, 223, 225 Australia–Singapore FTA 29, 422 Australia–Singapore PTA 61, 62, 65 Copyright Bill 174–5 Digital Economy Partnership Agreement (DEPA) see Digital Economy Partnership Agreement (DEPA)

494  Research handbook on digital trade Korea–Singapore DEA 20, 100 Korea–Singapore Digital Partnership Agreement (KSDPA) 100, 181, 186–7, 194, 195 National AI Strategy 174 New Zealand–Singapore Closer Economic Partnership Agreement (CEPA) 38, 43 New Zealand–Singapore FTA 422 Singapore–Australia Digital Economy Agreement (ASDEA) 20, 45–6, 428–9 Singapore–Chile and New Zealand DEA 225 UK–Singapore DEA 20, 49, 51, 91, 96 Single Submission Portals, Latin America and SMEs 252–3, 258, 259–64, 265–6 Słok-Wódkowska, M 67, 121, 122 Slovakia–Iran BIT 294–5 smart borders, Digital Economy Partnership Agreement (DEPA) 92 Smeets, M 219, 225, 231 SMEs Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 333, 335–40 Digital Economy Partnership Agreement (DEPA) 94, 335, 339–40 and digital nomads see digital nomads and global labour mobility, SMEs digital-based challenger banks 322 and gender equality 102–3, 104, 105, 107, 113, 114 Hong Kong, digital trade facilitation 171 Latin America see Latin America and SMEs Solove, D 15, 46 Soprana, M 20, 90, 93, 94, 96, 97, 120, 121, 333 Sornarajah, M 292, 303 source codes Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) 14–15, 16, 337 digital nomads and global labour mobility, SMEs 337 e-commerce and digital trade, differences between 67–8, 71 economic security, Japan–US DTA 443 intellectual property 466 Korea, digital trade agreements and policy 193–8 see also cybersecurity; trade secrets South Korea see Korea South–South preferential trading areas 220–21 sovereignty, EU as digital trade actor 118–19, 125 spaghetti bowl network, international investment agreements 292 spam control 16, 41–2, 93 see also security concerns

Special Economic Zones (SEZs) 199–216 China 208, 209, 212–14 cross-border data flow 205 Cross-border E-commerce (CBEC) Pilot Zones (Cities) 212–14 Cross-border E-commerce Retail Import (CERI) 212–15 Cross-border E-commerce Retail Import (CERI), and Online Purchase of Bonded Products 212–13 data localisation issues 205–6 Digital Free Trade Zones (DFTZs) 214 digital Special Economic Zones (SEZs) 211–15 digital trade 204–7 Eco-Industrial Parks (EIPs) 209–10 and economic growth 199, 202 environmental issues 202, 209–10 essential elements and location 201 export-processing zones (EPZs) 208–9 Free Trade Zones (FTZs) 207, 208–9, 210–11 global trade and privacy law on cross-border transfers 206–7 global value chains 205 industrial parks 207–8 and international trade 202–7 origins and evolution 207–11 regional and preferential trade agreements 203 as self-contained regimes 200 Sustainable Development Goals (SDGs) 210 technology and knowledge transfer to domestic firms 200–201 trade and development 203–4 stablecoin, fintech 326 Starr, S 349, 360 Streinz, T 126, 437–8 Subedi, S 292, 304 supply chains 84, 128, 162, 170, 185, 187, 216, 263, 326, 451–2, 453 Sustainable Development Goals (SDGs), Special Economic Zones (SEZs) 210 Suthersanen, U 456, 457, 460, 462 Taiwan and CPTPP 16 Taiwan–Nicaragua FTA 43 Tarawneh, J 5, 288–305 taxation digital services taxes see digital services taxes (DST) Korea, digital trade agreements and policy 189

Index  495 technological challenges, WTO law and cross-border data flows 79–80 technological context, digital economy agreements see digital economy agreements, technological and economic context technological neutrality, e-commerce and digital trade, differences between 63 Technological Protection Measures (TPMs), intellectual property 458 technological skills, Jordan 237 technology incorporation problems, Latin America and SMEs 262 Teh, R 121, 422 telecommunication issues, China and WTO e-commerce negotiations 163 Telecommunications Law, Jordan 241–2 territorial requirement, international investment agreements 298–302 territoriality principle, intellectual property 459 Text of Trade Agreement (ToTA), preferential trade agreements (PTAs) 61 Thailand, Australia–Thailand PTA 70 trade agreements see digital economy agreements; free trade agreements; preferential treade agreements; regional trade agreements; WTO headings trade facilitation China and WTO e-commerce negotiations 158, 162 digital economy agreements, e-commerce foundations 424–5, 428–9 Digital Economy Partnership Agreement (DEPA) 92 e-commerce and digital trade, differences between 63 Hong Kong see Hong Kong, digital trade facilitation Latin America and SMEs 257, 264–7 WTO law and cross-border data flows 79, 87–8 trade law effects, intellectual property 459–61 trade mark protection, intellectual property 457 trade provisions, gender equality 108–15 trade secrets, protection of 349, 360–61 see also security concerns; source codes Trade in Services Agreement (TiSA), China and WTO e-commerce negotiations 152, 155, 157 Trans-Pacific Partnership (TPP) 44, 45, 65, 70 transparency African Continental Free Trade Area (AfCFTA) 223, 229–30

algorithmic justice in the digital market and artificial intelligence (AI) 351–2, 356, 361, 362, 364–5, 366 China and WTO e-commerce negotiations 162 e-commerce and digital trade, differences between, preferential trade agreements (PTAs) 63–4 WTO law and cross-border data flows 84–5, 86, 89 treaties with investment provisions (TIPs), international investment agreements 292, 293–5, 301, 304 trust see business trust Turing, A 345, 369, 374–5, 376, 378 UK

artificial intelligence (AI), national regulatory approaches 387–8 central bank digital currencies (CBDCs) 327 Copyright, Designs and Patents Act (CDPA) 396 and CPTPP 16 digital economy and economic growth 290 digital services taxes (DST) 276–7, 282, 283–4 EU and post-Brexit Trade and Cooperation Agreement (TCA) 24 EU–UK Trade and Cooperation Agreement (TCA) 120, 125, 129–32 fintech 318 Human Rights Act 24, 125 Japan–UK FTA 44 online lending industry 322–3 UK–New Zealand FTA 114 UK–Norway, Iceland and Liechtenstein FTA 222, 223, 224–5 UK–Singapore DEA 20, 49, 51, 91, 96, 225 UK, cases CBS Songs Ltd v Amstrad Consumer Electronics 467 Dolling-Baker v Merrett 410, 412 St Albans City and District Council v International Computers 395 UN Centre for Trade Facilitation and Electronic Business (UN/CEFACT) 259, 260, 261, 266 UN Conference on Trade and Development (UNCTAD) 253–4 UN Convention on the Use of Electronic Communications in International Contracts (UNECC) 13, 33, 221–2 UN Digital Financing Task Force 308 UN Framework for Ethical AI 351

496  Research handbook on digital trade UN Global Survey on Digital and Sustainable Trade Facilitation 265 UN Sustainable Development Goal, gender equality 112 UNCITRAL, Kristian Almås and Geir Almås v Poland 300 UNCITRAL Model Law arbitrator appointment 402, 404, 405, 409, 410 on Electronic Commerce 13, 33, 59 on Electronic Signatures 422, 424 and public policy 406–7 UNCITRAL Principles, African Continental Free Trade Area (AfCFTA) 222–4 United Arab Emirates see Middle Eastern countries, United Arab Emirates Universal Postal Union (UPU) 34–5 US AI lawyers 393 AI proactive and future-proofing approach see artificial intelligence (AI), national regulatory approaches, US statutory definition and proactive and future-proofing approach Algorithmic Accountability Act 378 American Data Privacy and Protection Act (ADPPA) 141–2 Australia–US FTA 36 Bipartisan Trade Promotion Authority Act 29 California Consumer Privacy Act (CCPA) 140, 141 Communications Decency Act 18 Computer Software Copyright Act 466 Copyright Law 396 Copyright Law and First-Sale doctrine 463, 465–6 Digital Agenda 11, 12, 29 digital economy and economic growth 290 Digital Millennium Copyright Act (DMCA) 458, 468 digital services taxes (DST) 271–2, 279–80 EU–US Joint Agenda for Global Change proposal 126, 127 EU–US Safe Harbour Agreement 15 EU–US Trade and Technology Council (TTC) 125–9, 380 federalism see federalism, US fintech 317–18 Foreign Investment Risk Review Modernization Act (FIRRMA) 446–8 FUTURE of Artificial Intelligence Act 377–8 Internet Tax Freedom Act 247 Japan–US DTA see economic security, Japan–US DTA

John S. McCain National Defense Authorization Act 378 Jordan–US FTA 29, 35, 36, 43, 46, 247–9 Korea–US FTA 42, 44, 99–100, 190, 191 National Artificial Intelligence Initiative Act 379–80 online lending industry 323 state-issued central bank digital currencies (CBDCs) 327 Streamlined Sales and Use Tax Agreement (SSUTA) 143 Trade Act and tariffs 271–2 Trade and Investment Framework Agreement (TIFA) 245–6, 249 Transatlantic Trade and Investment Partnership (TTIP) 126 Transpacific Partnership Agreement (TPP) 12, 15, 16, 166 US–Chile FTA 422 US–France Bilateral Income Tax Treaty 286 US–GCC Framework Agreement for Trade 246 US–South Korea FTA 14 US, cases Betamax 467 Bobbs-Merrill v Straus 465 Capitol Records v ReDigi 465–6 Cruz v Raymond Talmadge d/b/a Calvary Coach 371 McGlove v Lacey 414 MGM Studios Inc v Grokster 467 Napster 467 Sony Corp. of America v Universal City Studios 467 State v Loomis 348–9 Wayfair 142–3 US, United States–Mexico–Canada Agreement (USMCA) 16–19, 20, 23, 24, 124, 215, 335, 433 and African Continental Free Trade Area (AfCFTA) 225 algorithms inclusion 18 China and WTO e-commerce negotiations 164, 167 CPTPP model similarities 16–17, 19 customs duties 16 data protection 17–18 e-commerce and digital trade, differences between 54, 57, 66 and federalism 143–5 and Hong Kong, digital trade facilitation 173 and interactive computer services 18 and Korea, digital trade agreements 182, 193, 194

Index  497 open government data and public access 18–19 and preferential trade agreements 44, 45, 48–9, 52 source codes 18 WTO law and cross-border data flows 79, 82 ‘user-created’ data, redeeming value of, digital services taxes (DST) 271, 280 value assessment digital economy agreements 416, 419, 432–4 digital services taxes (DST) 271, 273–4, 279–80, 284, 285 e-commerce and digital trade, differences between 66, 67–8, 69 federalism 139, 143 fintech 323, 325, 326, 327 value chains, global 135, 157, 205, 255, 438 Van den Bossche, P 286, 340, 391 Varesis, F 400, 406 venture capital, securing, intellectual property 462 videoconferencing, digital nomads and global labour mobility 330 Vienna Convention on the Law of Treaties (VCLT) 275, 297 Wang, Y 369, 370, 373, 378, 393, 395, 397 Waqar, M 409, 410, 411 Watai, R 6–7, 439–54 Weber, R 2–3, 73–89, 390 website-based AI lawyers 392–5 Willemyns, I 30, 71, 122, 123, 274, 275, 276, 278, 281, 282 ‘winner-take-most’ contest, digital economy agreements 416, 421 WIPO Internet Treaties 458, 462, 463, 464 see also intellectual property Wolfe, R 18, 22, 56, 68 Women Entrepreneurs Finance Initiative (We-Fi) 104 women inclusion, Digital Economy Partnership Agreement (DEPA) 94, 99, 113 see also gender equality World Customs Organization (WCO) 38 World Economic Forum 77, 153, 154 World Wide Web Consortium (W3C) 85 WTO Chairs Programme (WCP) 218–19 China and e-commerce negotiations see China and WTO e-commerce negotiations and CPTPP see Comprehensive and Progressive Agreement for Transpacific Partnership (CPTPP) Declaration on Global Commerce 275

Digital Champions for Small Business 341 Digital Economy Partnership Agreement (DEPA) 97 digital nomads and global labour mobility 332, 332–3, 340–41 digital services taxes (DST) see digital services taxes (DST), as trade barrier under WTO rules e-commerce and digital trade, differences between 64 E-Commerce Joint Statement Initiative (JSI) 151–6, 156–63 economic security, Japan–US DTA 441, 445–6 Electronic Commerce Programme 9, 13, 20, 26, 28–9, 35, 57, 71, 181–3, 220, 229 and FTAs see free trade agreements goods/services dichotomy see goods/services dichotomy Joint Statement Initiative (JSI) 77, 90, 100, 122–3, 423, 440, 441 Middle Eastern countries and GATT 245, 246 and PTAs see preferential trade agreements and RCEP see Regional Comprehensive Economic Partnership (RCEP) Text of Trade Agreement (ToTA) 61 Trade Facilitation Agreement 180 Trade Facilitation Agreement (TFA) 257 trade finance 341 Work Programme on Electronic Commerce 109–11 WTO Appellate Body Argentina – Trade in Goods and Services 275, 278, 280, 281–2 Australia – Tobacco Plain Packaging 285 China – Electronic Payment Services 220 China – Publications and Audiovisual Services 76, 80, 149–50, 160, 161, 165, 220, 276, 280, 338, 391, 394–5, 431 China – Rare Earth 161 China – Raw Materials 161 EC – Seal Products 284, 338 India – US Imports 273 Japan – Alcoholic Beverages 273, 280 Mexico – Telecom 230 Russia – Traffic in Transit 83, 95 US – Gambling 56–7, 76, 149, 248, 276, 284, 374, 391, 431 WTO law and cross-border data flows 73–89 artificial intelligence (AI) and digital trade 80 blockchain 79–80, 87–8

498  Research handbook on digital trade Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) 78, 82 cybersecurity 82–3, 85 data-driven trade law architecture 86–8 data-localisation requirements 85 Digital Economy Partnership Agreements (DEPAs) 78 digital platform implementation regulation 88 digital trade agreement development 87 digital trust and promotion of business 85–6 digitised assets, treatment of 79–80 distributed ledger technology (blockchain) infrastructure 75 E-Commerce Work Programme 75 future direction 88–9 international regulatory cooperation 88, 89 network neutrality 85 pre-digital legal order 74–7 preferential trade agreements (PTAs) 78 privacy concerns 81–2 Regional Comprehensive Economic Partnership (RCEP) 78–9 regulatory issues 81–4 self-regulation consideration 86 technological challenges 79–80 trade facilitation 79, 87–8 trade governance framework 84–6 trade law 77–9 transparency, accountability and legal interoperability 84–5, 86, 89 United States–Mexico–Canada Agreement (USMCA) 79, 82

WTO Sanitary and Phytosanitary (SPS) agreement 79 WTO Technical Barriers to Trade (TBT) agreement 79 WTO law and cross-border data flows, classification issues 74–7 ongoing revision efforts 76–7 outdated goods and services 75–7 ‘predominant purpose’ test 76–7 and World Economic Forum 77 WTO Joint Statement on Electronic Commerce (JSEC) 77 WTO law and cross-border data flows, trade in services 74–6, 80, 82–4, 89 categorisation 76 digital assets 76 GATT Harmonized System (HS Code) 76 positive-list approach 75–6 privacy protection 82 WTO TRIPS Agreement 341, 395–7, 434, 458–9, 460, 472 see also intellectual property Wu, M 150, 390, 423 Wunsch-Vincent, S 11, 29, 390, 457, 461, 462 Yakovleva, S 23, 59, 69, 72, 124 Yu, R 168, 170, 173–7 Zdouc, W 286, 340 Zeng, D 200, 208, 210, 211 Zhang, A 307, 325 Zhao, L 6, 369–99 Zheng, W 218, 228–9 Zuboff, S 60, 66, 68, 419