Research Handbook on Abuse of Dominance and Monopolization (Research Handbooks in Competition Law series) 1839108711, 9781839108716

Table of contents :
Front Matter
Copyright
Contents
Contributors
Preface
Acknowledgements
Introduction
PART I ESTABLISHING DOMINANCE
1. Market definition
2. Understanding market power: an economics perspective
3. Concentration and rising market power: fears and facts
4. Understanding market power: a legal perspective
PART II ABUSE OF DOMINANCE AND MONOPOLIZATION: CONDUCT AND THEORIES OF HARM
5. What is an abuse of a dominant position? Deconstructing the prohibition and categorizing practices
6. Exploitative abuses: recent trends and comparative perspectives
7. A re-evaluation of the abuse of excessive pricing
8. Predation and discrimination
9. The future of refusals to deal and margin squeezes in the face of sector-specific regulation
10. Tying and bundling
11. Rebates
12. The role of intent in abuse of dominance and monopolization
13. The special responsibility of dominant undertakings
14. Abuse without dominance and monopolization without monopoly
PART III ENFORCEMENT
15. Remedies, sanctions and commitments
16. Burden of proof and judicial review
PART IV CROSS-CUTTING ISSUES
17. Intersection of competition and regulation in abuse of dominance and monopolization
18. Mandated neutrality, platforms, and ecosystems
19. Killing nascent innovation as abuse of dominance and monopolization
20. Monopolization and intellectual property rights
21. Public procurement and abuse of dominance
22. Big data and data-related abuses of market power
Index

Citation preview

RESEARCH HANDBOOK ON ABUSE OF DOMINANCE AND MONOPOLIZATION

RESEARCH HANDBOOKS IN COMPETITION LAW This highly-topical series addresses some of the most important questions and areas of research in competition law and antitrust. Each volume is designed by a leading expert to appraise the current state of thinking and probe the key questions for future research on a particular topic. The series encompasses some of the most pressing issues as well as the foundational pillars of the field, including: merger control, competition damages, abuse of dominance and cartels, amongst others. Each Research Handbook comprises specially-commissioned chapters from leading academics, and practitioners, as well as those with an emerging reputation and is written with a global readership in mind. Equally useful as reference tools or high-level introductions to specific topics, issues and debates, these Research Handbooks will be used by academic researchers, postgraduate students, practising lawyers, competition authority officials and policy makers. Titles in the series include: Research Handbook on Asian Competition Law Edited by Steven Van Uytsel, Shuya Hayashi and John O. Haley Research Handbook on Methods and Models of Competition Law Edited by Deborah Healey, Michael Jacobs and Rhonda L. Smith Research Handbook on the Law and Economics of Competition Enforcement Edited by Ioannis Kokkoris Research Handbook on Abuse of Dominance and Monopolization Edited by Pınar Akman, Or Brook and Konstantinos Stylianou

Research Handbook on Abuse of Dominance and Monopolization Edited by

Pınar Akman Professor of Law, School of Law, University of Leeds, UK

Or Brook Associate Professor of Competition Law and Policy, School of Law, University of Leeds, UK

Konstantinos Stylianou Professor of Competition Law and Regulation, School of Law, University of Glasgow, UK

RESEARCH HANDBOOKS IN COMPETITION LAW

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: 2022950597 This book is available electronically in the Law subject collection http://dx.doi.org/10.4337/9781839108723

ISBN 978 1 83910 871 6 (cased) ISBN 978 1 83910 872 3 (eBook)

EEP BoX

Contents

List of contributorsvii Prefacexi Acknowledgementsxii Introduction1 Pınar Akman, Or Brook and Konstantinos Stylianou PART I

ESTABLISHING DOMINANCE

1

Market definition Sean P. Sullivan

2

Understanding market power: an economics perspective Nicolas Petit

26

3

Concentration and rising market power: fears and facts Gregory J. Werden

42

4

Understanding market power: a legal perspective Nicolas Petit

60

PART II

8

ABUSE OF DOMINANCE AND MONOPOLIZATION: CONDUCT AND THEORIES OF HARM

5

What is an abuse of a dominant position? Deconstructing the prohibition and categorizing practices Pablo Ibáñez Colomo

6

Exploitative abuses: recent trends and comparative perspectives Marco Botta

101

7

A re-evaluation of the abuse of excessive pricing Kathryn McMahon

119

8

Predation and discrimination John B. Kirkwood

140

9

The future of refusals to deal and margin squeezes in the face of sector-specific regulation Inge Graef

10

Tying and bundling A. Douglas Melamed

81

162 181

v

vi  Research handbook on abuse of dominance and monopolization 11 Rebates Viktoria H.S.E. Robertson

200

12

The role of intent in abuse of dominance and monopolization Mariateresa Maggiolino

222

13

The special responsibility of dominant undertakings Giorgio Monti and Ekaterina Rousseva

239

14

Abuse without dominance and monopolization without monopoly Or Brook and Magali Eben

259

PART III ENFORCEMENT 15

Remedies, sanctions and commitments Florian Wagner-von Papp

283

16

Burden of proof and judicial review Andriani Kalintiri

317

PART IV CROSS-CUTTING ISSUES 17

Intersection of competition and regulation in abuse of dominance and monopolization338 Katalin J. Cseres

18

Mandated neutrality, platforms, and ecosystems Barak Orbach

359

19

Killing nascent innovation as abuse of dominance and monopolization Alan J. Devlin

374

20

Monopolization and intellectual property rights Christopher R. Leslie

397

21

Public procurement and abuse of dominance Grith Skovgaard Ølykke

420

22

Big data and data-related abuses of market power Tilman Kuhn, Kristen O’Shaughnessy, Tobias Pesch, Jaclyn Phillips and D. Daniel Sokol

438

Index456

Contributors

Pınar Akman is a Professor of Law at the School of Law, University of Leeds, where she is also a Director at the Jean Monnet Centre of Excellence on Digital Governance. She has published widely in competition law and economics, particularly on abuse of dominance, vertical restraints, and competition in digital markets. Her publications include the sell-out monograph The Concept of Abuse in EU Competition Law: Law and Economics Approaches (Hart 2012) alongside numerous peer-reviewed publications. She is the recipient of a Philip Leverhulme Prize (Leverhulme Trust) for her research into competition law in the digital economy. Marco Botta is Part-time Professor at the European University Institute, where he is the coordinator of the Florence Competition Programme (FCP) and ENTraNCE for Judges. In addition, he is Affiliated Research Fellow at the Max Planck Institute for Innovation and Competition (MPI) in Munich (Germany). Finally, he is Adjunct Professor at the Law Faculty of the University of Vienna (Austria), where he teaches EU, competition and State aid law. Or Brook is an Associate Professor of Competition Law and Policy at the Centre for Business Law and Practice, School of Law at the University of Leeds, where she teaches EU and international competition law, business regulation, and quantitative research methods. Holding an academic background in law and economics, she employs empirical approaches to study questions regarding the goals of competition law, the role of public policy considerations, decentralised enforcement, and the exercise of enforcement discretion. She is co-director of ASCOLA UK. Katalin J. Cseres is an Associate Professor of Law at the Amsterdam Centre for European Law and Governance (ACELG), University of Amsterdam. She is a Senior Fellow of the Amsterdam Center for Law & Economics (ACLE) and the academic coordinator for the Masters Programme on European Competition Law and Regulation. Her research expertise lies in (EU) competition law, consumer law and sector regulation, most notably energy law. Alan J. Devlin is a Partner at Latham & Watkins LLP, and an Adjunct Professor at Georgetown University Law Center. Magali Eben is a Lecturer in Competition Law at the School of Law, University of Glasgow, where she teaches EU and UK competition law, and US antitrust law. Her research focuses on market definition, digital markets, and legal approaches to anticompetitive unilateral conduct. She was the American Bar Association’s Antitrust Law International Scholar in Residence (2018) and a visiting researcher at Georgetown University. She consults on a variety of EU law topics. Dr Eben is co-director of ASCOLA UK. Inge Graef is Associate Professor of Competition Law at Tilburg University, where she is affiliated to the Tilburg Institute for Law, Technology, and Society (TILT) and the Tilburg Law and Economics Center (TILEC). She holds expertise in the areas of competition law, platform regulation and the governance of data-driven innovation. vii

viii  Research handbook on abuse of dominance and monopolization Pablo Ibáñez Colomo is Professor of Law and Jean Monnet Chair in Competition and Regulation at the London School of Economics. He is also a Visiting Professor at the College of Europe (Bruges), Joint General Editor of the Journal of European Competition Law & Practice and co-editor of the Chillin’Competition Blog. He received a PhD from the European University Institute in June 2010 and an LLM from the College of Europe in 2004. Andriani Kalintiri is a Lecturer in Competition Law at King’s College London. Previously she was a Lecturer at City University and a Fellow at LSE. She holds a PhD from Queen Mary University of London and an LLM from Cambridge. Among others, she is a Fellow of the UK Higher Education Academy and a member of the editorial board of the Journal of European Competition Law & Practice. John B. Kirkwood is a Professor at Seattle University School of Law, where he writes extensively about antitrust. He has been quoted by the Supreme Court and three of his articles have won national awards for pathbreaking scholarship. He has testified before Congress and the federal enforcement agencies and is often quoted in the media. Prior to joining Seattle University, he directed the Planning Office, the Evaluation Office, and the Premerger Notification Program at the Federal Trade Commission’s Bureau of Competition in Washington, D.C. and later managed cases and investigations at the Northwest Regional Office. He has received the Outstanding Faculty Award and the Dean’s Medal. Tilman Kuhn is a Partner in the Düsseldorf and Brussels offices of White & Case LLP. Christopher R. Leslie is a Chancellor’s Professor of Law at the University of California Irvine School of Law. Professor Leslie received his JD from UC Berkeley, his MPP from the Kennedy School of Government at Harvard University, and his BA in Economics and Political Science from UCLA. He is the founding director of the Competition, Antitrust Law, and Innovation Forum at UC Irvine. Kathryn McMahon is an Associate Professor at the School of Law, University of Warwick, where she researches and teaches UK, EU and international competition law. She has spent time as a visiting scholar at the Centre for Antitrust and Regulatory Studies (CARS), University of Warsaw, University of Texas, University of California, Berkeley, and as a Fernand Braudel Senior Fellow at the European University Institute in Florence. She is an Associate Editor of the Global Journal of Comparative Law. Mariateresa Maggiolino is the Director of the five-year Masters Program in Law of the Bocconi University, where she is also Associate Professor of Business Law. She has been a visiting professor at Fordham Law School, New York, in 2015. Her main fields of interest are antitrust law, data law, banking law and intellectual property law. She has published three monographs and many articles in both national and international journals. A. Douglas Melamed is Professor of the Practice of Law at Stanford Law School. From 2009 until 2014, he was Senior Vice President and General Counsel of Intel Corporation. Prior to joining Intel, he was a partner in the Washington, DC office of WilmerHale and chair of the firm’s Antitrust and Competition Practice Group. From 1996 to 2001, he served in the US Department of Justice as Acting Assistant Attorney General in charge of the Antitrust Division and, before that, as Principal Deputy Assistant Attorney General. Doug received his BA from Yale and his JD from Harvard.

Contributors  ix Giorgio Monti is a Professor of Competition Law at Tilburg University and TILEC. Barak Orbach is a Professor of Law at the University of Arizona James E. Rogers College of Law, as well as a member of the American Law Institute and the American College of Governance Counsel. His areas of expertise are antitrust, regulation, corporate governance and the digital economy. Kristen O’Shaughnessy is a Partner in the New York office of White & Case LLP. Tobias Pesch is an Associate in the Düsseldorf office of White & Case LLP. Nicolas Petit is Joint Chair in Competition Law at the Department of Law and at the Robert Schuman Centre for Advanced Studies. He is also an invited Professor at the College of Europe in Bruges. He is the author of Big Tech and the Digital Economy: The Moligopoly Scenario (Oxford University Press 2019), co-author of EU Competition Law and Economics (Oxford University Press 2012) and author of Droit européen de la concurrence (Domat Montchrestien 2013 and 2018), a monograph which was awarded the prize for the best law book of the year at the French Constitutional Court. Jaclyn Phillips is an Associate in the Washington, DC office of White & Case LLP. Viktoria H.S.E. Robertson is Professor of Competition Law and Digitalization at the Vienna University of Economics and Business, and Professor of International Antitrust Law at the University of Graz. She has been a visiting academic at Oxford University’s Centre for Competition Law and Policy, Stanford University and the FGV-Rio Law School. She has taught competition law at numerous universities and is a member of the European Law Institute and the Academic Society for Competition Law. Her current research focuses on the application of competition law in digital markets, as well as on competition law and sustainability. Ekaterina Rousseva is a Member of the Legal Service of the European Commission and a Visiting Lecturer at the University of Sofia. D. Daniel Sokol is a Professor of Law at the USC Gould School of Law and an Affiliate Professor of Business at the USC Marshall School of Business.  Additionally, he serves as Senior Advisor at White & Case LLP. Grith Skovgaard Ølykke is a commercial law consultant at the law firm Poul Schmith, where she advises mainly public clients on state aid and general EU law. Prior to that, she was a Professor of Commercial Law and EU Market Law at the Copenhagen Business School (on leave 2017–19). She has published extensively on state aid law and public procurement law. Konstantinos Stylianou is a Professor of Competition Law and Regulation at the University of Glasgow School of Law. His research focuses on online markets, blockchain, and high technology industries. He holds degrees from Penn, Harvard, and Aristotle University. Sean P. Sullivan is an Associate Professor of Law at the University of Iowa College of Law. Prior to joining the University of Iowa, he served as an antitrust attorney at the United States Federal Trade Commission. Florian Wagner-von Papp is Professor of Private and Economic Law at the Helmut Schmidt University for the Armed Forces in Hamburg, Germany. Previously, until 2019, he was

x  Research handbook on abuse of dominance and monopolization Professor of Antitrust and Comparative Law & Economics at UCL in London, UK. He holds a doctorate degree (Dr iur.) from the University of Tuebingen, Germany, and an LLM from Columbia Law School, New York. Gregory J. Werden earned a PhD in economics in 1977 then began a 42-year career in the Antitrust Division of the US Department of Justice. He contributed to numerous guidelines, policies, investigations, litigations and appeals, including more than 80 Supreme Court briefs. In 2019 he received the Mary C. Lawton Lifetime Service Award. While in government service, he actively contributed to scholarship on competition policy. In retirement he published a book: The Foundations of Antitrust: Events, Ideas, and Doctrines (Carolina Academic Press 2020).

Preface

This Research Handbook offers a comprehensive and up-to-date guide to the competition law (antitrust) prohibition of abuse of a dominant position and monopolization. Directed at competition law stakeholders around the globe, it offers a vital resource for academics, practitioners, competition officials, and policymakers alike. Bringing together a team of authoritative legal and economic experts, it provides an in-depth analysis of foundational legal and economic principles which guide the analysis of the competition law prohibition of abuse of dominance and monopolization. The Research Handbook draws from the long and influential traditions of leading jurisdictions such as the European Union and the United States to analyse applicable rules and policy in these jurisdictions, as well as comparatively to identify common threads and differences. Organized in four Parts across 22 chapters, the Handbook adopts the familiar analytical framework of abuse of dominance and monopolization. Each chapter addresses a key aspect of the competition law prohibition of unilateral conduct by consolidating existing theory and practice and breaking new ground. Part I is dedicated to establishing dominance and monopoly power. It includes chapters on the legal and economic tests and notions of market power and market definition, as well as on market concentration and its implications for competition policy. Taken together, these chapters present established and new theories and evidence on the key requirement of market (or monopoly) power in establishing abuse of dominance or monopolization. Part II focuses on the element of ‘abuse’ or ‘bad conduct’ in the prohibition, and it analyses different theories of harm. It covers both the theoretical aspects of the concept of abuse, including the role of intent, the special responsibility of the dominant undertaking, and the relationship between abuse and market power, as well as specific types of abuses, such as exploitative abuses, excessive pricing, predation and discrimination, refusal to supply, tying, and rebates. Part III examines enforcement-related matters pertaining to investigating and litigating cases involving abuse of a dominant position and monopolization. It covers the issues of available and appropriate remedies, sanctions, and commitments, as well as the allocation of the burden of proof in prosecutorial procedures and in judicial review. Part IV focuses on the interplay between abuse of dominance and monopolization and other areas of law, and on special challenges in the theory and application of the prohibition. Combining interdisciplinary expertise, it covers the intersection of competition law and regulation, as well as the intersection of intellectual property and competition law in abuse of dominance and monopolization cases. Further, it offers contemporary analyses of emerging issues, such as the legal status of platform neutrality and ‘self-preferencing’, killing nascent innovation as a potential abuse of dominance and monopolization, data-related abuses, and the application of abuse of dominance in procurement and services of general economic interest. With its comprehensive breadth, interdisciplinary approach and trans-Atlantic coverage, as well as the law and economics approach it takes, this Research Handbook can serve as the most up-to-date and authoritative resource on abuse of dominance and monopolization.

xi

Acknowledgements

This book project kicked off just before the start of the COVID-19 pandemic and at the time of its completion the pandemic is still not over. We are immensely grateful to all of the contributors to this volume for their professionalism and sustained commitment to the project during this tumultuous period of research and writing. Without their dedication during these difficult times, this project could not have been completed. On a personal note, Pınar Akman would like to thank Laura Mann at Edward Elgar Publishing for approaching her to take on the project of compiling this Research Handbook. She would also like to thank Or Brook and Konstantinos Stylianou for agreeing to join her as co-editors in this project; like the contributors, they also did not quite know what they signed up to, but they similarly proved their utmost professionalism and commitment to the successful and timely delivery of the Handbook. Pınar would like to also gratefully acknowledge support by the Leverhulme Trust (Philip Leverhulme Prize) which allowed her to dedicate the necessary time to this project. Finally, she would like to thank Peter Whelan who had to live with her through the ups and downs of the entire process. We are grateful to Laura Mann and the entire editorial and production team at Edward Elgar Publishing for their professionalism and assistance throughout the project. Last but not least, we would like to thank the Centre for Business Law and Practice, the Jean Monnet Centre of Excellence on Digital Governance and the School of Law at the University of Leeds for providing us with a supporting environment for this project. Prof. Pınar Akman, Dr Or Brook and Dr Konstantinos Stylianou Leeds, 9 June 2022

xii

Introduction Pınar Akman, Or Brook and Konstantinos Stylianou

The competition law prohibitions of an abuse of a dominant position and monopolization have traditionally registered less frequently on the radar of competition authorities compared to cartels and merger control, both areas of competition law which have attracted most of their attention and resources.1 However, the growing importance of the digital economy over the past two decades and the concomitant rise of ‘tech giants’ have added fresh impetus to the investigations of unilateral conduct. At the same time, socio-economic transformations have contributed to calls for a re-evaluation of both the legal and the economic underpinnings of abuse of dominance and monopolization,2 creating some uncertainty, but also opportunities for further developing the intellectual foundations of this aspect of competition law. Against this backdrop, the Research Handbook on Abuse of Dominance and Monopolization offers a comprehensive, timely and up-to-date guide for scholars and practitioners who wish to gain an in-depth understanding of the European Union (EU) prohibition of the abuse of a dominant position and the United States (US) prohibition of monopolization. Each of the 22 chapters included in this volume consolidates existing knowledge, but also breaks new ground. The chapters consider both legal and economic principles of the subject matter and provide a solid understanding of the law of two of the most influential antitrust jurisdictions – the EU and the US. Each chapter engages with both EU and US competition law in order to analyse applicable rules and policy, and comparatively to identify common threads and differences. The volume is organized into four parts which mirror the familiar analytical framework of the abuse of dominance and monopolization prohibitions. Part I considers the fundamental requirement of market power and establishing dominance. Part II considers in detail the concept of abuse and monopolization, with a view to setting out the main theories of harm and the types of conduct that can violate the relevant legal provisions. Part III is dedicated to the enforcement of the abuse of dominance and monopolization prohibitions. Finally, Part IV adopts a forward-looking perspective in identifying cross-cutting issues which transcend the traditional borders of antitrust. Each of the six chapters in this final part is dedicated to contemporary challenges faced by competition authorities and regulators across the globe. Part I starts off by providing a detailed discussion of dominance and market power: it is only when firms are dominant or have monopoly power that they can distort competition unilaterally. Traditionally, the inquiry into market power starts with market definition, and so does this Research Handbook.

‘OECD Competition Trends 2022’ (2022), available at http://​www​.oecd​.org/​competition/​oecd​ -competition​-trends​.htm, accessed 25 October 2022. 2 See Barak Orbach, ‘The Present New Antitrust Era’ (2018) 60 William & Mary Law Review 1439; Hal Singer, ‘The Monopoly Harms that Antitrust Keeps Missing’ ProMarket (12 August 2020), available at https://​www​.promarket​.org/​2020/​08/​12/​the​-monopoly​-harms​-that​-antitrust​-keeps​-missing/​, accessed 18 January 2023; Schools Brief, ‘What More Should Antitrust Be Doing?’ The Economist (6 August 2020). 1

1

2  Research handbook on abuse of dominance and monopolization In Chapter 1, Sean P. Sullivan argues that market definition is one of the few areas where the EU and the US competition regimes have converged, but that the common approach to market definition may be problematic. After offering a brief review of the common justifications for defining markets and a restatement of the common standards for defining markets, he explains that markets should not be defined by the rote application of omnibus standards, but by tests tailored to the specific purposes that relevant markets are meant to serve in a given application. Market definition sets the boundaries wherein market power is assessed. In Chapter 2, Nicolas Petit thus analyses the concept of market power. Given that market power is an economic concept and that competition authorities and academic scholarship routinely rely on economic definitions of market power, Chapter 2 focuses on market power from an economics perspective. In that direction, Petit offers different economic conceptions of market power, ranging from the basic concept of power over price to more technical definitions, such as control over output. Recent economic observations of market power have led some economists to conclude that market power is on the rise across industry segments, partly driven by increasing concentration.3 If this is true then it may be a cause for concern, because increasing market power creates the conditions for anticompetitive unilateral conduct. In Chapter 3, Gregory J. Werden tackles this timely and controversial issue by critically reviewing recent claims of increasing concentration. He argues that when concentration is reviewed at the correct market level where competition occurs, the cited evidence does not show increasing concentration in the US, but that some evidence of increasing market-level concentration exists for Europe. He concludes that, instead of suggesting a decline in competition, the available evidence offers competitively neutral explanations for the increasing concentration – notably mergers that extend geographic or product reach without affecting market concentration. Following the chapters on the economics of market power, in Chapter 4, Nicolas Petit returns to the legal understanding of market power. He observes that the legal definition used in US antitrust law is more closely aligned with an economic understanding of market power, whereas the definitions of EU competition law are more ‘autonomous’ and sometimes even conflicting. He then goes on to situate the assessment of market power within a chosen defined market, and subsequently analyses the kind of proof and evidence that EU and US competition laws require to establish market power. However market power is defined and measured, it is well established that market power per se is not objectionable. It is the abuse of market power or the monopolizing conduct that competition law pursues. Thus, the chapters in Part II analyse a wide range of conduct and theories of harm that may implicate the prohibitions against abuse of a dominant position and monopolization. Chapter 5 by Pablo Ibáñez Colomo opens this part by investigating the constituent elements of the notion of abuse and defining the fundamental criteria to distinguish between the main categories of potentially abusive conduct. He shows, however, that an all-encompassing legal test has proven elusive, and argues that it is not possible meaningfully to distinguish between inherently ‘improper’ and valid expressions of competition on the merits, since most practices demonstrate at least some procompetitive aspects and do not necessarily or invariably lead to anticompetitive effects. 3 Gustavo Grullon and others, ‘Are US Industries Becoming More Concentrated?’ (2019) 23 Review of Finance 697.

Introduction  3 The ensuing chapters analyse the different types of abusive or monopolizing conduct. A common distinction with respect to the types of abuse is between exploitative and exclusionary abuses. Exploitative abuses are those where market power is used to extract rents from customers, whereas exclusionary abuses result in the foreclosure of effective competition from the market. Chapter 6 and Chapter 7 address exploitative abuses. Marco Botta shows how the EU has pursued exploitative practices in the form of unfair purchase or selling prices, unfair trading conditions, and discrimination more vigorously than the US. He presents the relevant legal tests and inquires whether and to what extent we are witnessing a revival of the scrutiny of exploitative conduct in Europe. Kathryn McMahon focuses on excessive pricing. She introduces the theory of harm around such activity, namely the wealth transfer from consumers to producers, but also documents the legal and practical difficulties both with identifying excessive pricing, and with remedying the conduct. Similar to Botta, McMahon emphasizes that excessive pricing is pursued predominantly by EU enforcers, especially in the pharmaceutical and intellectual property (IP) sectors. McMahon analyses the applicable legal rules and concludes with suggestions on the rules and limiting principles which need to be established if the prohibition of abuse is to be successfully enforced. In Chapter 8, John B. Kirkwood moves on to exclusionary abuses by examining predatory pricing and platform discrimination. He contrasts the two practices by clarifying that predatory pricing involves a dominant firm which competes more aggressively in the short term, losing money in order to weaken or exclude a rival, whereas in platform discrimination a dominant firm disadvantages a competitor not by competing more aggressively itself, but by forcing up the rival’s costs. Despite their differences, he observes that both forms of exclusion enhance the market power of the dominant firm and, where unjustified, they restrict choice, raise prices and may diminish innovation. These adverse consequences notwithstanding, US law imposes serious obstacles to challenging either practice; by contrast, EU law is less restrictive, but perhaps should be more so. Inge Graef in Chapter 9 turns to another key exclusionary abuse, namely refusal to supply (and its lesser form, margin squeeze). Graef first lays out the applicable legal test for the two common forms of refusal to supply, namely outright and constructive refusal to supply. She argues, however, that the distinction is unworkable, and instead proposes the existence of a regulatory duty under sector-specific regulation as a better indicator for deciding the legality of refusal to supply. She also posits that the key criterion of indispensability should be interpreted flexibly on a per-case basis. The approach should be neither so strict that it can rarely be met in practice nor so lenient that it facilitates access to inputs which are not indispensable, but merely ‘convenient’ for rivals to compete. Graef places particular emphasis on recent case law, such as Google Search (Shopping),4 in order to highlight the most recent application of the refusal to supply doctrine. In Chapter 10, A. Douglas Melamed follows on with a legal and economic analysis of the practices of tying and bundling. Tying exists when a firm permits a customer, by sales conduct or product design, to purchase a tying product only if it also purchases a separate, tied product. Although tying can increase economic welfare in some circumstances, it is generally deemed unlawful in both the US and the EU if the firm has market power in the tying product and the



4

Google Search (Shopping) (Case AT.39740) Commission Decision [2018] OJ C 9/11.

4  Research handbook on abuse of dominance and monopolization tying forecloses from competitors a substantial amount of sales in the tied product market. Mixed bundling exists when a firm sells two or more separate products at lower prices than if the customer buys the products in separate transactions. The lower prices often reflect cost savings or other efficiencies from the bundled sales. Mixed bundling is likely to be unlawful in both the US and the EU if the firm has market power in one or more products in the bundle and the price of any one product in the bundle, after allocating to it the entire discount for the bundle, is below the firm’s average variable cost for introducing and selling that product. Another common practice which has recently come into the spotlight again with the controversial Intel case5 is that of rebates. In Chapter 11, Viktoria H.S.E. Robertson first distinguishes the different types of rebates (loyalty-enhancing, exclusivity, mixed) and then offers a comparative analysis between the EU and the US regimes. She finds that the two jurisdictions come from different places when it comes to assessing loyalty-enhancing rebate schemes, but have gradually moved towards greater convergence in their treatment of exclusivity rebates. Nevertheless, antitrust liability for a dominant company will more readily be established in the EU, where the applicability of economics-based tests is still being navigated. In the US, rebates by a monopolist will usually be found to be anticompetitive where they constitute predatory pricing, although they might also engender antitrust liability where they constitute exclusive dealing arrangements. So far, however, the US Supreme Court has not weighed in on the antitrust assessment of rebates. The remaining three chapters in Part II examine key conceptual issues around the abuse of dominance and monopolization which transcend specific types of conduct. In Chapter 12, Mariateresa Maggiolino tackles the role of intent in unilateral anticompetitive conduct. She notes that, under Section 2 of the Sherman Act and Article 102 of the Treaty on the Functioning of the European Union (TFEU), antitrust decision-makers use firms’ intent for evidentiary purposes, namely, to understand if the conduct under their scrutiny is likely to harm competition. However, she questions whether the analysis of firms’ intent may actually reveal the competitive impact of a given behaviour, as well as whether and why antitrust decision-makers should ever be permitted to use firms’ intent to prove the occurrence of a conduct capable of harming competition under Section 2 and Article 102 TFEU. In Chapter 13, Giorgio Monti and Ekaterina Rousseva address the ‘special responsibility’ of the dominant firm – a concept that is unique to EU competition law. They note that the notion of special responsibility does not create an additional requirement in the identification of an abuse for the purposes of Article 102 TFEU, but its evolution has nevertheless been intertwined with the evolution of the notion of abuse. The chapter traces the relationship between these two notions in four different ways: first, by exploring the evolution of the case law on exclusionary abuse, which affirms the need of showing a qualified effect on the competitive process; second, by exploring the links between dominance and abuse and illuminating how the degree of dominance impacts the assessment of unilateral conduct for the finding of an abuse; third, by explaining why sometimes dominant firms have positive obligations; and, finally, by clarifying the role of regulation in the identification of abuse. In Chapter 14, Or Brook and Magali Eben conclude this part by examining the issue of abuse without dominance and monopolization without monopoly. While the very definitions of abuse of dominance and monopolization require significant market power (ie market dom5 Case T-286/09 RENV Intel Corporation v Commission ECLI:EU:T:2022:19 (Appeal Case before the Court of Justice C-240/22 P).

Introduction  5 inance or monopoly), Brook and Eben present existing laws and planned legislative proposals diverging from this premise. They highlight the different degrees and types of economic power embedded in the EU and US rules, which limit the conduct of firms holding lower thresholds of market power, relative power (eg economic dependence, unfair practices) or gatekeeper powers. The authors expose the gap between the notions of power and the theory of harm, challenging the dividing line between protection against harm to effective competition and harm to competitors. Accordingly, Brook and Eben call upon legislators and enforcers to align the notions of power with the harm and interests covered by the laws. Part III on Enforcement starts off with Chapter 15, where Florian Wagner-von Papp explores the diverse goals pursued by competition law enforcement. These goals include: restoring competition; preventing future infringements; educating market participants about the content of the law and deterring them from breaking it; punishing past infringements; and curing their effects. The chapter also discusses the different types of remedies and sanctions used to achieve these diverse goals. Wagner-von Papp investigates the right mix between different remedies and sanctions, in particular, when accepting commitments, and examines whether granting the competition authorities more remedial discretion might not, overall, result in more effective enforcement as well as better protection of the alleged infringers’ rights. Next, in Chapter 16, Andriani Kalintiri considers the burden of proof and judicial review which, coupled with substantive liability rules, can be dispositive of the outcome of unilateral conduct enforcement. Comparing the EU and the US regimes, this chapter points to remarkable similarities: in both jurisdictions, the burden of proof is shared between the parties, while public enforcers are afforded a degree of deference. At the same time, the chapter reveals significant differences, stemming from the specific features of the EU and the US substantive rules and enforcement contexts. Kalintiri argues that greater clarity and consistency in judicial pronouncements and practice concerning evidentiary matters and in the intensity of courts’ control would be desirable. Part IV on Cross-Cutting Issues starts with Chapter 17 by Katalin J. Cseres, which analyses the intersection of competition law and regulation in the context of abuse of dominance and monopolization. She argues that as innovative technologies, shifts in consumer preferences, or other conditions transform markets, the legal and policy frameworks also need to adjust. Legislators and policymakers must decide whether to intervene, and if so, what respective roles regulation and competition law should play. Such questions are particularly timely following recent financial, climate and health crises, digitalization, and the rising of corporate and political powers. The chapter traces how this complex interplay has evolved in the EU and the US, and how the relationship between competition law and regulation is changing in law and policy-making processes. In Chapter 18, Barak Orbach assesses the conceptual foundations of mandated neutrality standards (MNS) prescriptions, such as ‘platform neutrality’ and bans on ‘self-preferencing’. MNS prescriptions require dominant digital intermediaries to deal with all interested parties on fair and equal terms. They oblige dominant digital ecosystems to treat rivals as they treat their own subsidiaries and units, and treat all trade partners alike, regardless of the attributes of the trade relations. Extreme forms of MNS prescriptions seek to break up digital ecosystems and outlaw business models which integrate platforms and other lines of business. Orbach submits that MNS prescriptions are shallow populist reactions to serious problems, and that inquiries into the intellectual foundations of MNS prescriptions tend to frustrate serious antitrust thinkers. In particular, MNS prescriptions do not seriously address the economics of the

6  Research handbook on abuse of dominance and monopolization digital economy, but rather offer variants of old regulatory schemes that were formulated for the brick-and-mortar economy. Alan J. Devlin in Chapter 19 moves to an elaboration of killing nascent innovation as a form of abuse of dominance and monopolization. He bases the discussion on the premise that an effort by a monopolist to extinguish a potential competitive threat can violate both Section 2 of the Sherman Act and Article 102 TFEU. He further posits that the existing competition laws in the US and the EU can adequately police sources and channels of technological distribution to facilitate sustained competitive pressure. With that pressure will come incentives to innovate and, in time, user migration. Devlin warns that it would be a profound mistake to abandon an effects-based framework that requires agencies to show that a transaction, practice or agreement will harm likely competition. In Chapter 20, Christopher R. Leslie reviews the relationship between monopolization and intellectual property (IP) rights. Those two areas of law may conflict when IP owners use their exclusionary rights to harm competition in ways beyond the scope of their legitimate IP rights. Both US antitrust law and EU competition law restrict how IP owners may exercise their exclusionary rights. Several types of anticompetitive conduct by IP owners can violate Section 2 of the Sherman Act and/or Article 102 TFEU. This conduct includes procuring patents through fraud, pursuing sham infringement litigation to harm competitors, using IP rights to impose tying arrangements, refusing to license and breaching one’s commitments to charge a fair, reasonable and non-discriminatory (FRAND) royalty for standard-essential patents. Grith Skovgaard Ølykke in Chapter 21 explores an EU law-specific issue of how abuse of a dominant position may take place in a public procurement procedure. The chapter begins by considering the concept of a ‘dominant position’ in a public procurement context, noting that the common way of establishing a dominant position can be applied, albeit with consideration to the specifics of the public procurement context. Those specifics include the procedural framework, transparency and the design of the tender. The chapter then examines several types of exclusionary and exploitative abuses by analysing cases where an abuse of dominance has been established or might have been involved. Ølykke concludes that there is ample scope for various kinds of abusive behaviour in the public procurement context and that this context may facilitate special types of abuse. The book’s final chapter, Chapter 22, concludes with a discussion of big data and data-related abuses of market power. Tilman Kuhn, Kristen O’Shaughnessy, Tobias Pesch, Jaclyn Phillips and D. Daniel Sokol provide an overview of the developments in EU and US case law relevant to such abuses, as well as the economic underpinnings of such an analysis. They discuss the procompetitive and anticompetitive concerns surrounding big data and the current legal landscape for evaluating them in the US and the EU. The authors maintain that much of the case law in both jurisdictions remains case specific and that the law is in flux. While there are few cases specific to this interface, related areas of law such as merger control and legislation that regulate certain types of conduct engaged in by big tech firms can provide a template for how competition authorities, courts and academics should frame such issues.

PART I ESTABLISHING DOMINANCE

1. Market definition Sean P. Sullivan1

I. INTRODUCTION Monopolization, in the United States (US), and abuse of dominance, in the European Union (EU), embody different philosophies about how best to police single firm conduct in competition law. Surprisingly, their disagreements end at market definition. Both doctrines define relevant markets by similar processes and use relevant markets for similar purposes. In some contexts, this type of agreement would be a welcome sight. Here, it reflects a pocket of confusion in each area of law. This chapter describes the confusion of current market definition practices and takes some initial steps toward a more coherent approach. It begins, in Section II, with a brief review of the common justifications for defining markets in single firm conduct cases, and a brief restatement of the common standards for defining markets. Section III then turns a critical eye to the assumed connections between current market definition practices and the substantive purposes the resulting markets are meant to serve. Sections IV and V sketch the outline of a more coherent approach to market definition in single firm conduct cases. The idea behind this approach is that markets should not be defined by rote application of omnibus standards, but should be defined by tests tailored to the specific purposes that relevant markets are meant to serve in a given application.

II.

COMMON STANDARDS

Everyone knows the standards for defining markets in single firm conduct cases. Advocates in both the US and the EU have spent decades intoning the same familiar language about interchangeability and the substitutes to which consumers may turn. The stated reasons for defining markets, and the uses to which those markets are put, are also similar across jurisdictions. A.

Market Definition in Monopolization Cases

In the US, Section 2 of the Sherman Act makes it illegal for any person to ‘monopolize, or attempt to monopolize, or combine or conspire … to monopolize’ any trade within the scope of interstate commerce.2 Interpreting this sparse language, the US Supreme Court has held that the offence of monopolization consists of two elements: ‘(1) the possession of monopoly power in the relevant market and (2) the willful acquisition or maintenance of that power as distinguished from growth or development as a consequence of a superior product, business I am extremely grateful to Alexander Asawa for research assistance on this project. Amy Koopmann and the University of Iowa Law Library also provided invaluable support and assistance. 2 15 USC s 2. 1

8

Market definition  9 acumen, or historic accident’.3 The related offence of attempted monopolization edges away from the threshold requirement of monopoly power, replacing it instead with the firm’s probable acquisition of monopoly power as a result of its anticompetitive conduct.4 Monopoly power, or its imminent acquisition, is thus a necessary element in proving any violation of Section 2 of the Sherman Act. So, what is monopoly power? The Supreme Court has described it as ‘the power to control prices or exclude competition’.5 But this is a very broad net. Nearly every competitor has some power to control prices, as well as some capacity to exclude at least certain types of competition.6 Alternatively, monopoly power is sometimes defined as something like substantial market power.7 This is a little more descriptive, but only a little. If market power is defined as the ability to price above a competitive level, then identifying monopoly power by this definition requires that two judgements be made, neither guided by clear standards. First, a determination must be made about what a ‘competitive’ price would be.8 Second, another determination must be made about whether the current or predicted price is ‘substantially’ greater than this competitive price. The practical difficulty of applying either of the above definitions of monopoly power drives most litigants to follow a different path when seeking to establish a violation of Section 2. Ever since the decision of United States v Aluminum Company of America in 1945, courts have endorsed an assumption that a large share of a relevant market – something in the order of 80 or 90 per cent – is evidence of a firm’s monopoly power in that market.9 This is not to say that a large market share is always indicative of monopoly power. But, according to conventional thinking, a large share is at least a necessary condition for possessing monopoly power.10 This, of course, requires the definition of relevant markets, and the Supreme Court has supplied several standards for the exercise. By one standard, the scope of a relevant market is to be ‘drawn narrowly to exclude any other product to which, within reasonable variations in price, only a limited number of buyers will turn; in technical terms, products whose “cross-elasticities of demand” are small’.11 By another standard, relevant markets are ‘com United States v Grinnell Corp 384 US 563, 570–71 (1966). See Spectrum Sports Inc v McQuillan 506 US 447, 459 (1993) (‘[P]etitioners may not be liable for attempted monopolization under [s] 2 of the Sherman Act absent proof of a dangerous probability that they would monopolize a particular market and specific intent to monopolize’); Image Technical Services Inc v Eastman Kodak Co 125 F 3d 1195, 1202 (9th Cir 1997) (observing that monopolization and attempted monopolization offences ‘are similar, differing primarily in the requisite intent and the necessary level of monopoly power’). 5 United States v E. I. du Pont de Nemours & Co 351 US 377, 391 (1956) (Cellophane case). 6 See Northern Securities Co v United States 193 US 197, 406 (1904) (Holmes J, dissenting) (‘According to popular speech, every concern monopolizes whatever business it does, and if that business is trade between two states it monopolizes a part of the trade among the states. Of course, the statute does not forbid that. It does not mean that all business must cease’). 7 See Eastman Kodak Co v Image Technical Services Inc 504 US 451, 481 (1992) (‘Monopoly power under [s] 2 requires, of course, something greater than market power under [s] 1’). 8 See Daniel A Crane, ‘Market Power without Market Definition’ (2014) 90 Notre Dame Law Review 31, 38–9. 9 See United States v Aluminum Co of America 148 F 2d 416, 424 (2d Cir 1945) (Alcoa); Grinnell (n 3) 571 (‘The existence of [monopoly] power ordinarily may be inferred from the predominant share of the market’). 10 See, eg, Duncan Cameron and others, ‘Good Riddance to Market Definition?’ (2012) 57 Antitrust Bulletin 719, 720–21. 11 Times-Picayune Publishing Co v United States 345 US 594, 612 n 31 (1953). 3 4

10  Research handbook on abuse of dominance and monopolization posed of products that have reasonable interchangeability for the purposes for which they are produced – price, use and qualities considered’.12 By another standard, relevant markets can be identified by the recognition of a product’s ‘peculiar characteristics and uses’.13 By yet another standard, markets – or, at least, ‘submarkets’ – may be recognized by ‘practical indicia’ such as industry or public recognition of a market, the presence of unique production facilities, or the observation of distinct customers, distinct prices or other distinct features.14 Other tests supplement these standards. Initially developed for horizontal merger analysis,15 the Hypothetical Monopolist Test (HMT) defines markets around groups of competitors who, if they were to band together as monopolists, would elect to raise prices by at least a small but significant amount for some period of time. In the merger context, the HMT is an iterative algorithm that starts from a candidate market as small as the merging parties and progressively expands the candidate market until the group of competitors it encompasses would have the joint market power and profit motive to exercise the hypothesized price increase.16 The HMT can be used to define relevant markets in monopolization cases,17 but sometimes requires modification to function usefully in this context – a point to which we will return shortly. B.

Market Definition in Abuse of Dominance Cases

In the EU, Article 102 of the Treaty on the Functioning of the European Union (TFEU) prohibits the ‘abuse by one or more undertakings of a dominant position’ within at least a substantial part of the internal market.18 In contrast to the Sherman Act, Article 102 TFEU spells out several of the specific acts it aims to prohibit. It explains that abuse of a dominant position includes ‘imposing unfair purchase or selling prices’, ‘limiting production … to the prejudice of consumers’, and excluding rivals or otherwise distorting the competitive process, among other things.19 By its terms, Article 102 TFEU only applies to the conduct of an undertaking with a ‘dominant position’.20 So, what is a dominant position? In United Brands, the Court of Justice described dominance as ‘a position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market by giving it the power to behave to an appreciable extent independently of its competitors, customers and ultimately of its consumers’.21 In guidance, the Commission has also contributed its view

Cellophane (n 5) 404. United States v E. I. du Pont de Nemours & Co 353 US 586, 593–4 (1957) (du Pont-General Motors case). 14 Brown Shoe Co v United States 370 US 294, 325 (1962); see also Grinnell (n 3) 572 (endorsing, in dicta, the use of this test in monopolization cases). 15 See Gregory J Werden, ‘The 1982 Merger Guidelines and the Ascent of the Hypothetical Monopolist Paradigm’ (2003) 71 Antitrust Law Journal 253. 16 United States Department of Justice and Federal Trade Commission, Horizontal Merger Guidelines (19 August 2010) s 4.1.1. 17 See, eg, United States v Microsoft Corp 253 F 3d 34, 81 (DC Cir 2001). 18 Consolidated Version of the Treaty on the Functioning of the European Union [2012] OJ C326/47, art 102. 19 ibid. 20 ibid. 21 Case 27/76 United Brands Co v Commission [1978] ECR 207, para 65. 12 13

Market definition  11 that dominance occurs when ‘competitive constraints are not sufficiently effective’, such that an undertaking ‘enjoys substantial market power over a period of time’.22 As in the case of US monopolization, these efforts to define dominance do little more than gesture at the concept. Again, the definitions fail to clearly separate dominant undertakings from the large mass of undertakings with at least some market power. Again, the effort to equate dominance with ‘substantial market power’ stumbles at the difficulty of identifying substantial market power. And, again, the cases have addressed these difficulties by interpreting large shares of relevant markets as the primary indicator of dominance.23 Market shares of 70 to 80 per cent have been described as ‘a clear indication of the existence of a dominant position’,24 with shares greater than this extending beyond dominance into something closer to monopoly.25 This places market definition in roughly the same position in abuse of dominance cases that it occupies in monopolization cases. The processes of market definition are also similar. For example, implementing regulation defines a relevant market by language similar to the US standard of reasonable interchangeability: ‘A relevant product market comprises all those products and/or services which are regarded as interchangeable or substitutable by the consumer, by reason of the products’ characteristics, their prices and their intended use’.26 The cross-elasticity standard is also used: ‘Factors relevant to the assessment of the relevant product market include the analysis of … cross-price elasticity of demand’.27 As explained by the Commission in the Notice on Market Definition, the goal of market definition is to identify the competitive constraints that act upon an undertaking: ‘Basically, the exercise of market definition consists in identifying the effective alter­native sources of supply for the customers of the undertakings involved’.28 This recalls the HMT, which is also expressly adopted as a potential standard when defining relevant markets in abuse of dominance cases.29

Guidance on the Commission’s enforcement priorities in applying Article 82 of the EC Treaty to abusive exclusionary conduct by dominant undertakings [2009] OJ C45/7, para 10. 23 See Case C-62/86 AKZO Chemie BV v Commission [1991] ECR 3359, para 60 (stating that ‘the Court has held that very large shares are in themselves, and save in exceptional circumstances, evidence of the existence of a dominant position’, citing Case 85/76 Hoffman-La Roche & Co AG v Commission [1979] ECR 461, and holding that the inference of dominance applies ‘where there is a market share of 50% such as that found to exist in this case’); Commission Notice on the Definition of Relevant Market for Purposes of Community Competition Law [1997] OJ C372/5, para 10 (‘[A dominant position] would usually arise when a firm or group of firms accounted for a large share of the supply in any given market, provided that other factors analysed in the assessment (such as entry barriers, customers’ capacity to react, etc.) point in the same direction’ (internal footnotes omitted)). 24 Case T-30/89 Hilti AG v Commission [1991] ECR II-1439, para 92. 25 See Hoffman-La Roche (n 23) para 39 (distinguishing a dominant position from ‘a monopoly or a quasi-monopoly’). 26 Commission Implementing Regulation (EU) No 1269/2013 of 5 December 2013 amending Regulation (EC) No 802/2004 implementing Council Regulation (EC) No 139/2004 on the control of concentrations between undertakings [2013] OJ L336/1, Annex 1, sec 6.1; see also Commission Notice on the Definition of Relevant Market (n 23) para 9 (adopting this definition in other contexts). 27 Commission Implementing Regulation (n 26). 28 Commission Notice on the Definition of Relevant Market (n 23) para 13. 29 ibid paras 15–17. 22

12  Research handbook on abuse of dominance and monopolization

III.

COMMON PROBLEMS

Much could be said about the various standards for defining relevant markets,30 but the more intriguing puzzle is the assumption – common to both the US and the EU – that current market definition practices connect relevant markets to the underlying concerns of the substantive law. What makes a large share of a relevant market evidence of monopoly power or occupancy of a dominant position? What makes any of the market definition standards helpful in assessing allegedly exclusionary or abusive conduct? The conventional answers to these questions are not inspiring. They do, however, offer glimpses into the confusion that currently envelops market definition practice. A.

Markets as Threshold Tests

One conventional argument for defining markets in single firm conduct cases is the notion that only the conduct of a firm with a large enough share of a relevant market warrants special scrutiny. Consistent with this idea, market shares below about 50 per cent are treated as inadequate to establish monopoly power in US courts,31 while something like 40 per cent or a little less gates enforcement of Article 102 TFEU in the EU.32 This positions market definition, and market share analysis, at the threshold of the competitive effects inquiry,33 with market shares below minimum thresholds ending inquiries before they get to competitive effects analysis.34 That raises a question: why should low shares of a relevant market have this abortive effect? One possible justification for the approach is the frequent suspicion that a firm without a large share of a relevant market cannot have the market power necessary to raise monopolization or abuse of dominance concerns. This might make sense if market definition controlled the price elasticity of market demand. But no current market definition standard except the HMT has any serious connection to market demand elasticity. The problem with this justi See Sean P Sullivan, ‘Modular Market Definition’ (2021) 55 UC Davis Law Review 1091. See, eg, Bailey v Allgas Inc 284 F 3d 1237, 1250 (11th Cir 2002) (‘A market share at or less than 50% is inadequate as a matter of law to constitute monopoly power’); Rebel Oil Co v Atlantic Richfield Co 51 F 3d 1421, 1438 (9th Cir 1995) (collecting ‘numerous cases’ that hold ‘a market share of less than 50 percent is presumptively insufficient’); see also Phillip E Areeda and Herbert Hovenkamp, Antitrust Law, vol 2B (4th edn, Wolters Kluwer 2014) para 532c (collecting other holdings). 32 Guidance on the Commission’s enforcement priorities (n 22) para 14 (‘The Commission considers that low market shares are generally a good proxy for the absence of substantial market power. The Commission’s experience suggests that dominance is not likely if the undertaking’s market share is below 40% in the relevant market’); Commission, ‘DG Competition discussion paper on the application of Article 82 of the Treaty to exclusionary abuses’ (2005) para 31, https://​ec​.europa​.eu/​ competition/​antitrust/​art82/​discpaper2005​.pdf, accessed 10 June 2021 (DG competition discussion paper) (‘[A]lthough also undertakings with market shares below 40% could be considered to be in a dominant position. … [U]ndertakings with market shares of no more than 25% are not likely to enjoy a (single) dominant position on the market concerned’). 33 See William M Landes and Richard A Posner, ‘Market Power in Antitrust Cases’ (1981) 94 Harvard Law Review 937, 972 (describing the usual treatment of market power analysis in rule of reason cases as ‘a threshold condition’); Guidance on the Commission’s enforcement priorities (n 22) para 9 (‘The assessment of whether an undertaking is in a dominant position and of the degree of market power it holds is a first step in the application of Article 82’). 34 See Frank H Easterbrook, ‘The Limits of Antitrust’ (1984) 63 Texas Law Review 1, 17–18 (advocating for the use of a market power ‘filter’ in antitrust analysis). 30 31

Market definition  13 fication is thus that the relationship it posits is not a reliable product of current practice. As William Landes and Richard Posner observed – over 40 years ago – without controlling for demand elasticity, ‘a given market share is neither necessary nor sufficient for a firm to be able to raise prices above the competitive level’.35 While tests like the HMT offer one context where market share may have a consistent relationship with market power, the traditional practice of lumping together all different tests of market definition leaves Landes and Posner’s critique as true today as it was back then.36 A large share of a relevant market does not always indicate market power, and a small share does not preclude it.37 Another possible justification for the threshold-test interpretation is that a firm without a large share of a relevant market is not economically significant enough to warrant scrutiny under monopolization or abuse of dominance law. A focus on economic significance motivated the US Supreme Court in its articulation of the practical indicia standard,38 and similar justifications have been offered for the size of a hypothesized price increase under the HMT.39 This focus on economic significance might make sense if market definition did indeed reliably identify markets of significant economic importance. But current practice does no such thing. Markets defined by the HMT are often narrow40 and bear no guarantee of economic significance.41 Markets defined by substitutability inquiries similarly focus on the closeness of products, not the economic significance of those products as a class. The practical indicia test might capture markets of reliable economic significance42 but, again, the typical practice of defining markets by lumping together different standards dilutes this relationship. In short, under current practice, a firm’s large share of a relevant market says little to nothing about its size or importance in the local, national or global economy. B.

Markets as Sets of Substitutes

Another common argument for defining markets in single firm conduct cases is the idea that doing so helps to identify sets of reasonable substitutes for the products of the firm in question. The intuitive justification for this project appears to be a belief that markets are identifiable scopes of trade in the world, bounded by the similarity of products in serving particular

Landes and Posner (n 33) 952–3. See Commission Notice on the Definition of Relevant Market (n 23) para 52 (contemplating the definition of relevant markets by a combination of different factors); Sullivan (n 30) Section I (observing the apparently simultaneous application of different market definition standards in antitrust opinion). 37 Franklin M Fisher, ‘Diagnosing Monopoly’ (1979) 19 Quarterly Review of Economics & Business 7, 18 (commenting on false belief that small market share shows the absence of monopoly power while large market share shows its presence and noting that what matters is what happens to market share when monopoly profits are sought). 38 See, eg, Brown Shoe Co v United States 370 US 294, 325 (1962). 39 See Areeda and Hovenkamp (n 31) para 537a. 40 See Horizontal Merger Guidelines (n 16) s 4, para 9 (‘Relevant antitrust markets defined according to the hypothetical monopolist test are not always intuitive and may not align with how industry members use the term “market”’); Sullivan (n 30) Section II.A (discussing the tendency of most modern market definition standards to identify narrow markets focused on particular concerns). 41 See David Glasner and Sean P Sullivan, ‘The Logic of Market Definition’ (2020) 83 Antitrust Law Journal 293, 339–40. 42 See Sullivan (n 30) Section I.B. 35 36

14  Research handbook on abuse of dominance and monopolization purposes or satisfying particular needs.43 The current author and others have argued that it is a category error to think of relevant markets as constructs with any presence in the world or permanence outside of a given inquiry.44 But, even setting aside this error, what relevance would identification of a given set of substitutes have for assessing monopolization and abuse of dominance concerns? That question can be sharpened by the well-known Cellophane fallacy. As generations of students have learned, the US Supreme Court made a serious mistake in its market definition analysis in United States v E.I. du Pont de Nemours & Co.45 Translating the error into HMT terms, the Court defined the relevant market by asking whether DuPont would have found it profitable to increase the price of its cellophane product. Finding this would not be profitable – since DuPont was presumably already pricing as high as its market power would allow – the Court expanded the relevant market to include products that were only realistic substitutes at DuPont’s already elevated prices. This expansion of the relevant market reduced DuPont’s share of the market – perversely allowing DuPont’s exercise of market power to masquerade as evidence of DuPont’s lack of market power. The error of the Cellophane fallacy has been well aired in the scholarly literature,46 and it now motivates cautionary qualifications in prescriptions of the HMT for use in monopolization and abuse of dominance cases.47 For example, it is frequently said that a competitive price must be used as the baseline price when attempting to apply the HMT to single firm conduct analysis.48 This is not exactly right. We will shortly return to this point with examples where the Cellophane fallacy does and does not apply in the single firm conduct context. First, however, a few additional points about current practice need to be addressed. The underlying problem at issue in the Cellophane fallacy extends beyond the HMT. All efforts to define markets as sets of substitute products suffer from the same fundamental problem that beguiled the Court in Cellophane. Reasonable interchangeability, cross-price

See ibid 303–7 (discussing attempts to identify ‘natural’ market boundaries by reference to substitutability). 44 ibid 307–12; Magali Eben, ‘The Antitrust Market Does Not Exist: Pursuit of Objectivity in a Purposive Process’ (2021) 17 Journal of Competition Law and Economics 586; Steven C Salop, ‘The First Principles Approach to Antitrust, Kodak, and Antitrust at the Millennium’ (2000) 68 Antitrust Law Journal 187, 188–9. 45 Cellophane (n 5) 380–81. 46 The earliest clear articulation of the error seems to be George W Stocking, ‘Economic Tests of Monopoly and the Concept of the Relevant Market’ (1957) 2 Antitrust Bulletin 479. For modern recitations, see Salop (n 44) 197–8; Richard A Posner, Antitrust Law (2nd edn, University of Chicago Press 2001) 150–51. 47 Eg, Commission Notice on the Definition of Relevant Market (n 23) para 19 (noting the Cellophane problem); OECD, ‘Abuse of Dominance in Digital Markets’ (2020) 14, www​.oecd​.org/​daf/​competition/​ abuse​-of​-dominance​-in​-digital​-markets​-2020​.pdf, accessed 10 June 2021 (‘The core concepts of market definition in abuse of dominance cases are the same as those applied in merger cases. However, care must be taken when applying analytical techniques such as the hypothetical monopolist test in markets whose conditions may already have been shaped by market power’); DG Competition discussion paper (n 32) para 15 (‘It is essential to take account of the fact that the SSNIP-test normally is based on the assumption that prevailing prices constitute the appropriate benchmark for the analysis. This assumption often does not hold in Article 82 cases’). 48 Eg, Patrick Massey, ‘Market Definition and Market Power in Competition Analysis: Some Practical Issues’ (2000) 31 Economic & Social Review 309, 323 (‘The cellophane trap means that a different approach is required in abuse of dominance cases’). 43

Market definition  15 elasticity, the HMT – every substitution-based test of market definition is exposed to an interpretive problem known to economists since well before any of these tests were developed. As Fritz Machlup explained in 1952: [O]ne must have certain prices, price ranges, or price relations in mind when one speaks of particular demand elasticities. The cross-elasticity of demand may be of very different magnitudes at different price relations and, hence, when making an estimate of the elasticity, one obviously thinks of the currently existing price relations. If we now say that the cross-elasticity of demand between two products is zero, it may refer only to the given price relation, while at others the cross-elasticities may be positive. Where this is the case, the monopoly position exists only within certain price ranges and is therefore an imperfect one.49

Put another way, the very project of trying to identify a single set of substitute products is economically unsound. There is no one set of substitute products, only different sets of substitutes at different prices.50 To focus on a useful set of substitutes, one must start from a clear understanding of what price range is relevant to an inquiry. That is possible – as addressed in more detail below – but it is not how relevant markets are defined or interpreted today. C.

Markets as Collections of Competitive Constraints

One last argument for defining markets in single firm conduct cases is that doing so helps to identify the competitive constraints acting upon the firm in question. Franklin Fisher was an early advocate of this understanding of market definition in antitrust cases. In 1979, he explained that, if market definition is to facilitate analysis, it needs ‘to place in the relevant market those products and services and firms whose presence and actions can serve as a constraint on the policies of the alleged monopolist’.51 Fisher contended that the breadth of a relevant market should be whatever was needed to embrace the ‘significant constraints’ on market power: [A] ‘market’ is something that can be monopolized. If you have left out significant constraints on power, the ‘market’ is too small. If you have kept in firms and products or services that are not significant constraints, the ‘market’ is too large.52

Something analogous to Fisher’s constraints interpretation of market definition motivates enforcement practices in the EU. A 2005 DG Competition discussion paper states: ‘The main purpose of market definition is to identify in a systematic way the immediate competitive constraints faced by an undertaking’.53 This language parallels that of the Commission’s 1997 Notice on the Market Definition,54 which devotes an entire section to commenting on Fritz Machlup, The Economics of Sellers’ Competition (John Hopkins Press 1952) 547. See Franklin M Fisher, ‘Economic Analysis and “Bright-Line” Tests’ (2008) 4 Journal of Competition Law and Economics 129, 132 (‘The relevant facts of Cellophane are undisputed. At a high enough price for cellophane, there was substitution of other flexible wrapping papers. At lower, still profitable prices, there was not. Once one has stated that (and specified the prices), one has said all there is to say’). 51 Fisher (n 37) 13. 52 Fisher (n 50) 133. 53 DG Competition discussion paper (n 32) para 12. 54 Commission Notice on the Definition of Relevant Market (n 23) para 2. 49 50

16  Research handbook on abuse of dominance and monopolization the various competitive constraints that may be identified in the process of defining relevant markets.55 For abuse of dominance analysis, this focus on constraints is tethered to the substantive understanding of dominance itself. The Commission explains this point in the same 1997 notice: The concept of relevant market is closely related to the objectives pursued under Community competition policy. … Under the Community’s competition rules, a dominant position is such that a firm or group of firms would be in a position to behave to an appreciable extent independently of its competitors, customers and ultimately of its consumers.56

Other Commission guidance explains dominance as a situation in which ‘the undertaking’s decisions are largely insensitive to the actions and reactions of competitors, customers and, ultimately, consumers’.57 Insensitivity to customers and competitors is thus seen as a necessary condition for dominance and market power.58 There is undoubtedly something helpful in the constraints approach, but the usual articulation fails to capture what it is. To start, the simple equation of market power with the absence of competitive constraints is inconsistent with basic economics. All firms face competitive constraints. Even those with significant market power. Examples include workhorse models in industrial organization economics. A single-price monopolist facing a competitive fringe exercises durable market power despite the presence of competitive constraints.59 In many oligopoly models, firms likewise exercise durable market power despite the presence of direct competition from other firms.60 These examples illustrate a general point. As courts have intuited since the dawn of modern competition law, all market power is exercised within the bounds of some set of competitive constraints.61 A similar error underlies the notion that dominance over a market traces to the independence or insensitivity of a firm to the actions of others in the market. Consider the textbook single-price monopolist. No firm better captures the idea of dominance, yet the single-price monopolist is emphatically sensitive to, and dependent upon, the actions of its consumers. The monopolist raises its price until it is in the elastic portion of the demand curve, and the limit of its price is the threatened exit of the marginal consumer still willing to buy at the monopolistic ibid paras 13–24. ibid para 10. This definition of dominance was articulated in Hoffman-La Roche (n 23) [38]. 57 Guidance on the Commission’s enforcement priorities (n 22) para 10. 58 ibid para 11 (‘[A]n undertaking which is capable of profitably increasing prices above the competitive level for a significant period of time does not face sufficiently effective competitive constraints and can thus generally be regarded as dominant’); DG Competition discussion paper (n 32) para 23 (‘For dominance to exist the undertaking(s) concerned must not be subject to effective competitive constraints. In other words, it thus must have substantial market power’). 59 See, eg, Louis Kaplow and Carl Shapiro, ‘Antitrust’ in A Mitchell Polinsky and Steven Shavell (eds), Handbook of Law and Economics, vol 2 (Elsevier 2007) 1073, 1180–83. 60 See, eg, ibid 1083–6. 61 See, eg, United States v Addyston Pipe & Steel Co 85 F 271, 292 (6th Cir 1898) (observing that firms possessed a window of market power despite the presence of competitive constraints: ‘Within the margin of the freight per ton which Eastern manufacturers would have to pay to deliver pipe in pay territory, the defendants, by controlling two-thirds of the output in pay territory, were practically able to fix prices’); Alcoa (n 9) 426 (similarly finding that ‘within the limits afforded by the tariff and the cost of transportation, “Alcoa” was free to raise its prices as it chose’). 55 56

Market definition  17 price. The same point applies for competitors. Even a monopolist is typically sensitive to the prices of marginal substitute products and to changes like the expansion of firms in its competitive fringe.62 Again, all market power is subject to constraints. There is, of course, an important role for competitive constraints in single firm conduct cases. In any given equilibrium, the addition of a new constraint might depress a firm’s market power while the elimination of an existing constraint might increase it. The identification of these competitive constraints can be helpful in evaluating potential changes in a firm’s market power as a result of its conduct.63 What the identification of these constraints does not do is shed much light on the thing it is commonly said to illuminate: whether a given firm is a monopolist, occupies a dominant position or is presently able to exercise substantial market power.

IV.

A MORE COHERENT APPROACH

The single most reliable way to get market definition wrong is to try to define markets before deciding what questions need to be answered. Contrary to what seems to be standard practice today, market definition cannot usefully come before a determination of what exactly the competitive concern entails.64 A single market also cannot generally serve every analytical purpose that might be needed. Helpful and reliable market definition starts from the questions that need to be answered and reasons backwards to the identification of markets responsive to those needs.65 This may compel the definition of different relevant markets for different parts of the evaluative process.66 All depends on what questions need to be answered. So, what questions do need to be answered in single firm conduct cases? Let us focus, for now, on traditional market power concerns. Three categories of concerns could arise from the challenged conduct: (1) it could allow the firm to gain new market power, (2) it could allow the firm to maintain existing market power, or (3) it could reflect the firm’s exercise of its existing market power. Within each of these broad categories of concern, subcategories could arise. Single firm conduct is not conventionally broken apart to this level of granularity – and we must save for another day a full exploration of all the possible permutations of concern – but this deconstruction helps to uncover some useful rules and guides to follow when defining relevant markets.

62 Lawrence J White, ‘Market Power and Market Definition in Monopolization Cases: A Paradigm Is Missing’ in W Dale Collins (ed), Issues in Competition Law and Policy, vol 2 (American Bar Association Section of Antitrust Law 2008) 913, 920 (‘[D]emand for a monopolist’s product should be expected to be sensitive (at the margin) to the prices of sellers of some substitutes … and thus the monopolist’s price should be expected to vary as well’). 63 See Eben (n 44) 30 (suggesting something similar through a focus on the ‘competitive constraints … most appropriate to the alleged conduct and theory of harm’). 64 See Glasner and Sullivan (n 41) 312–15 (describing the theory-dependence of market definition). 65 See Sullivan (n 30) 33 (proposing a purpose-based approach to selecting between different methods of market definition). 66 For specific discussion of defining multiple relevant markets, see Glasner and Sullivan (n 41) 330–33; Sullivan (n 30) 26–7.

18  Research handbook on abuse of dominance and monopolization A.

Conduct that Increases Market Power (Results in Higher Prices)

To start, consider a firm not presently in possession of monopoly power or a dominant position, but which threatens to acquire that status through its challenged conduct. The concern is a substantial increase in the firm’s market power: something that would allow it to raise prices relative to where they are now. Example fact patterns could include a merger from duopoly to monopoly or the acquisition of an essential patent by one of several current competitors. In the US, this type of conduct could be challenged as attempted monopolization.67 In the EU, Article 102 arguably does not fit the facts – the conduct in question is creating a dominant position, not exploiting it – but a flexible enough interpretation of dominance could capture many of the interesting cases.68 Helpful market definition, in this context, can borrow from the standard practices of market definition in horizontal merger analysis. Where exclusionary conduct is used to obtain new market power, a helpful relevant market collects those competitors that currently constrain the firm, and thus whose exclusion could result in an increase in the firm’s market power. Something analogous to the HMT, starting from the current price, would be an appropriate test for identifying this type of market. The problem of the Cellophane fallacy does not apply, since these markets are being defined to evaluate potential increases in market power relative to whatever market power the firm may already have. The HMT may, however, require a larger-than-average hypothesized price increase to approximate the concept of substantial market power, at least where the firm in question appears to have little market power at present.69 Next, consider a firm already in possession of monopoly power or a dominant position, but which threatens to obtain even more market power via the challenged conduct. In the US, this type of conduct could be challenged as simple monopolization.70 In the EU, the dominant position of the firm would bring its conduct within the reach of Article 102. In either case, the inquiry involves two steps: (1) assessing whether the firm has monopoly power or a dominant position, and (2) assessing whether the firm’s conduct is anticompetitive or constitutes an abuse of a dominant position. Each of these inquiries may be aided by a distinct relevant market. As discussed above, large shares in a relevant market are often interpreted as proof of substantial market power. Without necessarily endorsing this practice, we can observe three properties of market definition necessary for a large share of a relevant market to support this interpretation. First, the market would need to be defined by a process that controls in some way for demand elasticity.71 A process like the HMT does this indirectly through its hypothesized price increase: by construction, a monopolist in an HMT market has at least enough See n 2 (prohibiting ‘attempt to monopolize’). See n 18 (prohibiting ‘abuse … of a dominant position’ (emphasis added)); Pınar Akman, The Concept of Abuse in EU Competition Law: Law and Economic Approaches (Hart Publishing 2012) 94 (tracing the original intent of Article 102 to the prevention of ‘the dominant undertaking receiving advantages that would not be possible but for its dominance’). But cf n 32 (providing some flexibility for dominance to be found even at moderate share thresholds). 69 See Glasner and Sullivan (n 41) 316–17 (explaining why the size of the hypothesized price increase in the HMT must match the magnitude of the anticipated exercise of market power). 70 See Grinnell (n 3) 570–71. 71 See generally Landes and Posner (n 33). 67 68

Market definition  19 market power to implement a price increase of the hypothesized magnitude. Markets defined by abstract reference to substitutability or interchangeability offer no such guarantee. Second, the hypothesized price increase in the HMT would need to be large enough to equate a large market share with the possession of substantial market power.72 Exactly what constitutes substantial market power may be a fact-bound question and has not been adequately explored to date. Third, the problem of the Cellophane fallacy emphatically does apply in this situation, so care must be taken not to assess patterns of substitution at already elevated prices. If something like the HMT is used, then it must be implemented as a hypothesized price increase above an estimate of a competitive base price.73 The above considerations apply only to the first inquiry in the challenge. Now consider the second inquiry. Even if the firm possesses significant market power, is its conduct reasonably capable of increasing that market power?74 If market-based inferences are to be used in this second inquiry,75 it will typically be helpful to define a different relevant market when doing so. The reason for this is that the purpose of the relevant market is now different. The point is no longer to establish the firm’s market power – the purpose of the first relevant market – but to help evaluate the firm’s ability to obtain additional market power through its challenged conduct. Intuitively, this second market definition exercise is analogous to that involved in the attempted monopolization challenge. Helpful market definition can again follow the pattern of horizontal merger analysis and the Cellophane fallacy again ceases to be a problem.76 As discussed above, the equation of monopoly power or a dominant position with a firm’s possession of ‘substantial market power’ requires the hypothesis of an appropriately substantial price increase when using shares of an HMT market to assess the firm’s possession of monopoly power or a dominant position. Does the same apply to HMT markets used to assess the firm’s challenged conduct? How large must the hypothesized price increase be in this second market? These are difficult questions. While insignificant changes in market power may not warrant intervention, one would expect both monopolization and abuse of dominance to prohibit conduct which, though insignificant in isolation, may in aggregate have significant effects.77 A reasonable minimum requirement is that the conduct ‘relates’ to the firm’s current market power78 or that it is reasonably capable of contributing to that market power.79 Beyond this, some form of sliding scale may be appropriate, such that larger exclusionary effects

See n 69. ibid 320–21. 74 See Salop (n 44) 195 (observing that a firm’s ability to price above marginal cost ‘does not mean that the firm can maintain or enhance its power by engaging in specific conduct alleged to be anticompetitive’). 75 See Herbert Hovenkamp, Federal Antitrust Policy: The Law of Competition and its Practice (5th edn, Wolters Kluwer 2016) 110–11 (citing predatory pricing, foreclosure, and tying as examples of conduct offences that only makes sense if the firm in question has a large market share). 76 Here, the focus of concern is an increase in market power, and that may be appropriately modelled by a price increase over the current-price baseline. 77 See Phillip E Areeda and Herbert Hovenkamp, Antitrust Law, vol 3 (4th edn, Wolters Kluwer 2015) para 651g (‘Any single exclusionary act may seem trivial. … [Yet] it may be fitting to presume the exclusionary act ‘significant’ or ‘causally related’ to the monopoly power being challenged’). 78 ibid (‘[I]t must at least appear plausible to an informed observer that the exclusionary act could have had, or would probably have, a significant relationship to the defendant’s monopoly’). 79 ibid (rephrasing as restraints that ‘reasonably appear capable of making a significant contribution to creating or maintaining monopoly power’). 72 73

20  Research handbook on abuse of dominance and monopolization are demanded of firms with lower initial market power. This would smooth the distinction between monopolization and attempted monopolization in US law.80 B. Conduct that Maintains Market Power (Preserves Current Prices) Now, consider single firm conduct that does not risk the creation of new market power, but instead risks entrenching, preserving or otherwise operating to maintain the firm’s existing market power. Example fact patterns may include a dominant incumbent’s acquisition of a nascent but potentially disruptive rival or a monopolist’s adoption of contracting practices that limit the ability or incentive of customers to trade with its rivals. In the US, conduct which has the effect of maintaining existing market power can be challenged as monopolization under Section 2.81 Recent cases have, however, narrowed the situations in which a monopolist may be obligated to allow its competitors to grow.82 In the EU, conduct that would limit production or development in order to maintain a dominant position is proscribed by Article 102.83 Indeed, undertakings which hold a dominant position are frequently said to have ‘a special responsibility not to allow [their] conduct to impair genuine undistorted competition in the internal market’.84 As in the previous discussion of monopolization challenges and their EU analogues, multiple relevant markets may be helpful in evaluating maintenance-of-market-power concerns. The first relevant market to consider is a market defined and used to establish the firm’s current possession of substantial market power. This is the same exercise discussed in relation to the first inquiry in an acquisition-of-market-power challenge and does not require repetition here.85 It should, however, be noted that the effectiveness of some tools of exclusion may depend upon a firm’s size and importance relative to its rivals. Examples include exclusion through volume discounts or exclusivity contracts, where the exclusionary potential of the conduct arises primarily because of the large size or importance of the firm. In some cases, the same relevant market used to establish the firm’s present market power may double as a way of

See n 69 and accompanying text. See Grinnell (n 3) 570–71 (‘willful acquisition or maintenance of [monopoly] power’ (emphasis added)); Aspen Skiing v Aspen Highlands Skiing 472 US 585, 602 (1985) (‘purpose to create or maintain a monopoly’ (emphasis added)). See also Herbert Hovenkamp, ‘The Monopolization Offense’ (2000) 61 Ohio State Law Journal 1035, 1041 (‘[A] great deal of strategic behavior is concerned with the “maintenance” rather than the acquisition of monopoly power’). 82 See Verizon Communications Inc v Law Offices of Curtis V. Trinko LLP 540 US 398, 415–16 (2004) (‘The Sherman Act … does not give judges carte blanche to insist that a monopolist alter its way of doing business whenever some other approach might yield greater competition’). 83 See Hoffman-La Roche (n 23) para 91 (including in the concept of abuse, behaviour which ‘has the effect of hindering the maintenance of the degree of competition still existing in the market or the growth of that competition’ (emphasis added)). 84 Case C-52/09 Konkurrensverket v TeliaSonera Sverige AB [2011] ECR 527, para 24; see also Eleanor Fox, ‘Monopolization and Abuse of Dominance: Why Europe is Different’ (2014) 59 Antitrust Bulletin 129, 135 (‘[D]ominant firms have a special responsibility not to erect or maintain barriers that frustrate the access of nondominant firms to markets’). 85 See nn 71–73 and accompanying text. 80 81

Market definition  21 establishing the firm’s relative size and importance when evaluating this type of exclusionary conduct.86 Since a firm’s current possession of substantial market power is no guarantee that its challenged conduct operates to maintain that power,87 a separate inquiry is needed to identify which competitors may be limited or excluded by the challenged conduct and to assess whether these competitors would have more tightly constrained the dominant firm – eroding the firm’s market power – but for the effect of the challenged conduct. As Lawrence White has explained, this aspect of maintenance-of-market-power challenges can itself be usefully deconstructed into two questions: (1) what would the relevant rivals’ performance be (or have been) but for the exclusionary conduct of the dominant firm, and (2) what would the effect of this performance be (or have been) on the behaviour of the dominant firm and the options available to consumers?88 If market-based inferences are to be used in addressing either part of this inquiry, one market concept responsive to the maintenance-of-market-power concern would be something like a reverse HMT: a relevant market defined as a group of competitors who, if unrestrained by the firm’s conduct, would likely expand or take other actions that would lead to a small but substantial depression of prices for some product sold by the monopolist or dominant firm. Note that this definition of a relevant market is easier to state than it is to operationalize. The hypothesis involves predicting both the future actions of firms and the effects of those actions on prices. Also, markets defined by this process may not be helpful in every case. Where only a single competitor is included in the relevant market, for example, the market adds nothing to a direct evaluation of White’s two inquiries. But where multiple competitors are simultaneously constrained, or where future competitive constraints could arise from different sources, the above process of defining a relevant market may provide helpful context for the requisite analysis.89 Two additional observations about market definition are noteworthy. First, relevant markets defined for purposes of evaluating the firm’s conduct may be quite different from relevant markets defined for purposes of evaluating the firm’s market power. In United States v Microsoft, for example, Microsoft was found to have market power in the worldwide market for ‘Intel-compatible PC operating systems’.90 This would be the relevant market for the first inquiry above. But Microsoft’s exclusionary conduct occurred in a broader scope of trade including ‘middleware’ products like Java and Netscape.91 That broader scope of trade could be seen as the relevant market for the second inquiry. In Microsoft, the DC Circuit rightly

See Hovenkamp (n 81) 1041 (‘Much monopolistic conduct is rational behavior only on the premise that the firm is already a monopolist, and frequently the conduct is designed not so much to create monopoly in a secondary market as to maintain the dominant firm’s position in the primary market’). 87 See n 74; Thomas G Krattenmaker and others, ‘Monopoly Power and Market Power in Antitrust Law’ (1987) 76 Georgetown Law Journal 241, 255 (offering similar comments). 88 White (n 62) 923. 89 See Thomas G Krattenmaker and Steven C Salop, ‘Anticompetitive Exclusion: Raising Rivals’ Costs to Achieve Power over Price’ (1986) 96 Yale Law Journal 209, 255 (‘[E]nough of the purchaser’s actual and potential rivals must suffer the price increase so that remaining unexcluded rivals cannot or will not prevent the purchaser from exercising power over price’). 90 Microsoft (n 17) 52. 91 See ibid 53–4 (discussing the future threat that middleware products posed for Microsoft). 86

22  Research handbook on abuse of dominance and monopolization rejected Microsoft’s argument that the targets of alleged exclusion needed to be included in the relevant market – that is, the first relevant market.92 The court rested this conclusion on a temporal distinction between nascent and present substitutes,93 but it would have been clearer to observe that different relevant markets may be defined for different purposes. Here, the relevant market used to evaluate Microsoft’s current market power simply involved a different set of competitors than the relevant market used to evaluate Microsoft’s ability to maintain its market power against future erosion by middleware products.94 Second, a remaining consideration, applicable to most single firm conduct cases but especially important for maintenance concerns, is the role that the dominant firm’s size can play in explaining its incentive to exclude rivals or raise their costs.95 In many cases, the larger the dominant firm’s base of existing sales, the more benefit it will extract from obstructing the competitive inroads of rivals. Intuitively, a 5 per cent market-wide price reduction may equate to a far greater profit loss for the dominant firm with an 85 per cent market share than it would for smaller competitors in the same market. The dominant firm may thus find it profitable to invest in forestalling future price decreases where smaller competitors would not.96 When shares of a relevant market are used to assess the comparative incentives of firms to engage in exclusionary or cost-raising conduct, it seems intuitive that the relevant market for the second inquiry will often be the appropriate basis for share computations. This is only intuition, however, and exceptions may arise. C.

Conduct that Exercises Market Power (Sets Current Prices)

Finally, consider single firm conduct risking neither the creation nor the maintenance of market power but instead merely reflecting the exercise of market power. Examples of this type of conduct include a firm charging monopolistic prices or insisting upon terms of trade which could not be demanded if it did not already possess substantial market power. Under US law, this conduct falls squarely outside the reach of Section 2. Cases have long held that a monopolist commits no wrong if it obtains monopoly power without engaging in prohibited types of exclusionary conduct,97 and recent cases have settled what little doubt may have

92 ibid. See also Case T-83/91 Tetra Pak International SA v Commission [1994] ECR II-00755, para 116 (rejecting the argument that abuse of dominance necessarily requires abuse to occur in the same market that is dominated). 93 ibid 54 (‘Because middleware’s threat is only nascent, however, no contradiction exists. Nothing in [s] 2 of the Sherman Act limits its prohibition to actions taken against threats that are already well-developed enough to serve as present substitutes’). 94 See n 75. 95 Franklin M Fisher, ‘The IBM and Microsoft Cases: What’s the Difference?’ (2000) 90 American Economic Review 180, 180 (‘An anticompetitive act by a single firm is one that is not profit-maximizing without the monopoly rents that it creates or maintains but is profit-maximizing with those rents included’). 96 See Krattenmaker and others (n 87) 259 (‘The greater the disparity in market shares between the firm seeking to raise its rivals’ costs and the rivals, the greater the firm’s anticipated reward for achieving a higher price for its output. Hence, such a firm would be willing to spend more in attempting to exclude rivals to gain power over price’). 97 See Grinnell (n 3) 571 (distinguishing illegal monopolization from ‘growth or development as a consequence of a superior product, business acumen, or historic accident’); Alcoa (n 9) 429 (inquiring whether a firm had ‘monopolized’ a market or whether ‘monopoly [had] been thrust upon it’).

Market definition  23 remained about the legality of exercising monopoly power, lawfully obtained.98 In Pacific Bell v linkLine Communications, the Supreme Court declared that ‘antitrust law does not prohibit lawfully obtained monopolies from charging monopoly prices’.99 For the same reasons, Section 5 of the Federal Trade Commission Act is unlikely to prohibit the mere exercise of market power today.100 Under EU law, dominant undertakings are less free to exercise their market power. Article 102 includes in its definition of abuse the act of ‘imposing unfair purchase or selling prices’. Case law supports the obvious interpretation of this language as reaching something like monopolistic pricing. As the Court of Justice held in United Brands, ‘charging a price which is excessive because it has no reasonable relation to the economic value of the product supplied would be [within the scope of the abuse of imposing an unfair selling price]’.101 The Commission has also issued guidance identifying as illegal conduct which is ‘directly exploitative of consumers, for example charging excessively high prices’.102 At least in principle, this means that excessive pricing, or equivalent contracting practices, could be challenged as abuse of dominance under EU law.103 In evaluating exercise-of-market-power concerns, it is hard to find any analytical need for relevant markets or market definition.104 The traditional inference of substantial market power from a large share of a relevant market could still be drawn. (And, if this inference is to be drawn, the relevant market should be defined subject to the practices discussed above.) But, unless something like excessive pricing is to be inferred from the possession of market power alone, evidence of the exercise of market power must still be produced, and whatever evidence is produced in that regard would seem better proof of the firm’s possession of market power than the market definition exercise would support.105 Perhaps this limits the role of market definition to the – presumably unusual – situation in which the challenged exercise of market power is not itself a maximal exercise of the firm’s available market power.

Cf Areeda and Hovenkamp (n 77) para 650a (‘[A]s the cases have clearly held, [s]2 is concerned not only with remedying monopoly power to which undesirable conduct has made a substantial contribution, but also with enjoining the conduct itself’); Trinko (n 82) 407 (‘The mere possession of monopoly power, and the concomitant charging of monopoly prices, is not only not unlawful; it is an important element of the free-market system’ … ‘To safeguard the incentive to innovate, the possession of monopoly power will not be found unlawful unless it is accompanied by an element of anticompetitive conduct’). 99 555 US 438, 454 (2009). 100 See, eg, Official Airline Guides, Inc. v FTC, 630 F 2d 920, 927–8 (2d Cir 1980) (acknowledging, in a Section 5 case, ‘the long recognized right of trader or manufacturer … freely to exercise his own independent discretion as to parties with whom he will deal’ and concluding that ‘even a monopolist, as long as he has no purpose to restrain competition or to enhance or expand his monopoly, and does not act coercively, retains this right’). 101 United Brands (n 21) paras 248–54. 102 Guidance on the Commission’s enforcement priorities (n 22) para 7. 103 For a discussion of apparent legal standards for the offence of excessive pricing, see Akman (n 68) 194–5. 104 See Glasner and Sullivan (n 41) 336–9 (critiquing the possibility that market definition should be legally required even where not analytically helpful). 105 See generally Louis Kaplow, ‘Why (Ever) Define Markets?’ (2010) 124 Harvard Law Review 437, 466 (providing a related but broader critique of market definition). 98

24  Research handbook on abuse of dominance and monopolization

V.

QUESTIONS OF SIZE

While the previous discussion focused on traditional market power concerns, monopolization and abuse of dominance are not necessarily so limited. Other concerns, and other aspects of what it might mean to dominate or monopolize trade, can be usefully collected under the heading of a firm’s size. These, too, may influence how relevant markets should be interpreted and defined. A.

Does Size Matter?

As a thought experiment, suppose that fictional firm ZedCo has a 90 per cent share of a relevant market containing only it and one other competitor. This market was defined by the HMT with a current-price baseline and a hypothesized price increase of 5 per cent, so if ZedCo’s rival were eliminated, ZedCo would be able to raise its prices by at least 5 per cent. Now suppose ZedCo pays a critical input supplier to refuse future deliveries to its rival, eventually driving the rival out of business. Does it matter, for monopolization or abuse of dominance purposes, that ZedCo is a single-employee business with market power limited to the isolated rural village in which it operates? Many would say yes: neither monopolization nor abuse of dominance feel like appropriate labels for the conduct of a firm this small,106 even if the firm has substantial market power in its limited scope of operation. Now change the facts. Suppose the same conduct presents only a 3 per cent chance of driving the rival from the market and suppose further that successful exclusion of the rival would only buy ZedCo the power to raise its prices by 0.001 per cent. Does it matter, for monopolization or abuse of dominance purposes, that ZedCo is a tech giant with yearly revenue greater than the GDP of many countries? Again, many would say yes. Indeed, there is an economic basis for this distinction. As Landes and Posner observed long ago, ‘actual economic injury’ is a function not only of the change in price ‘but also the amount of economic activity over which the [change in price] occurs’.107 Put another way, a modest market power gain could have disastrous welfare consequences when multiplied across a large enough base of transactions. Our intuition in this thought experiment contrasts with the limited role that size plays in modern market definition practice. Relevant markets are not regularly defined by processes designed to capture economically significant scopes of trade.108 Nor are they interpreted in these terms. True, consistent attention to market shares keeps the relative size of firms in focus. But an overemphasis on relative size may actually obscure the role that absolute size – amount of economic activity affected – also plays in explaining competitive harm. Particularly with the growing attention being paid to large tech companies, matters of size may soon become a more important part of single firm conduct cases. This is no justification

106 White (n 62) 923 (‘[T]here must be some de jure or de facto lower limit on a defendant’s size and importance … Antitrust enforcement should not care if Joe (the owner of Joe’s Unique Coffee Shop) insists that his manager Nora signs a “non-compete” clause when she decides to leave and start her own eatery …’). 107 Landes and Posner (n 33) 953; see also Louis Kaplow, ‘On the Relevance of Market Power’ (2017) 130 Harvard Law Review 1303, 1361 n 130 (noting the potentially countervailing concern that the costs of enforcement errors may also scale with firm size). 108 Sullivan (n 30) Section II.A.

Market definition  25 for changing market definition standards – making economic significance a necessary requirement of market definition, for example. But it is excellent justification for making economic significance a regular part of the interpretation of relevant markets and how the challenged conduct would affect them. B.

Does Size Alone Matter?

The previous discussion might undersell the role that size plays in monopolization and abuse of dominance concepts. One could take the position that size alone is a proper concern in single firm conduct cases. Under this view, firms too great in size are a danger themselves, separate from any market power concerns.109 Markets dominated by large firms may similarly be seen as deficient themselves, separate from any specific connection between concentration and power in pricing or negotiation. Competition policy could simply prefer more and smaller competitors in important areas of trade. Here, again, growing agitation over large tech companies is surfacing many of these policy positions in debates about enforcement practices and priorities. It is not obvious that market definition is needed to apply ‘size alone’ considerations. But, if relevant markets are to be used in evaluating single firm conduct under these policy goals, then new processes of market definition may be required. Markets defined by a process like the HMT, though potentially helpful in assessing the market power implications of different types of conduct, have no necessary relationship to these concerns. A firm’s control of 99 per cent of a relevant market defined by the HMT is simply not reliable evidence that the firm is economically or politically significant, or that any other non-market-power objective would be furthered by reduced concentration in this market. Helpful relevant markets must be defined by processes matched to the concern to be addressed. For policy objectives not recently at the forefront of competition policy, there is little reason to suppose that previous market definition processes match the concern.

VI. CONCLUSION The aspiration of this chapter is to take some initial steps toward a more coherent approach to market definition in monopolization and abuse of dominance cases. In brief, the proposed approach is to match market definition with the concerns to be addressed. If we want markets to be useful in evaluating potential wrongs, then we must define markets by processes that are sensitive to the wrongs we seek to avoid. For single firm conduct, this places the onus on the substantive law to clearly identify what specific wrongs it seeks to avoid. Greater clarity in substantive law is the most promising path to coherence in this area of market definition.

109 For a variety of arguments in this vein, see Tim Wu, The Curse of Bigness: Antitrust in the New Gilded Age (Columbia Global Reports 2018).

2. Understanding market power: an economics perspective Nicolas Petit

I. INTRODUCTION Antitrust laws are concerned with undue market power. In an economic conception of the law, antitrust rules of liability strike down anticompetitive business conduct or mergers that represent illegitimate market power strategies. Throughout history, the antitrust literature has been an ebb and flow of commentary on antitrust laws’ costly and ineffective checks against undue market power.1 Part of the controversies owes to misconceptions about market power. Antitrust lawyers lack a common understanding of market power.2 The result is a great deal of confusion about what antitrust laws achieve. At a time of concern towards rising levels of market power in the past decades, empirical evidence suggests that high prices have been localized in the United States (US), and that a stronger approach to antitrust law and economic regulation in the European Union (EU) has supplied welfare gains to consumers on the old continent.3 However, when European antitrust lawyers think about market power, they do not direct their attention to consumer prices. They think about corporate size and industrial concentration, see giant American firms, and deduce that they have a domestic market power problem. Different concepts of market power additionally shape conflicting views about what antitrust laws can achieve. According to one school, excessive market power produces economic inefficiency in the form of missing output at positive rates of return as well as distributional harms through transfers from consumers to managers and owners of firms which are arguably wealthier.4 Consequently, antitrust law can be an uncostly instrument of choice to address economic inequality. Stronger antitrust rules and enforcement policies against weaker forms of market power are desirable. On the other side, a conception acknowledges that market power produces inefficiency as well as diverse distributional harms and benefits. But antitrust law should treat distributional effects as indifferent because they involve hard-to-resolve trade-offs for antitrust courts and agencies.5 Instead, the selection of a value choice criterion and the col-

Eleanor M Fox, ‘The Battle for the Soul of Antitrust’ 75 California Law Review 917. And antitrust lawyers do not realize this. 3 Thomas Philippon, The Great Reversal: How America Gave Up on Free Markets (Belknap Press 2019). On concentration and rising market power, see also Chapter 3 in this volume. 4 Lina M Khan and Sandeep Vaheesan, ‘Market Power and Inequality: The Antitrust Counterrevolution and Its Discontents’ (2017) 11 Harv L & Pol’y Rev 235. 5 Oliver E Williamson, ‘Economies as an Antitrust Defense: The Welfare Tradeoffs’ (1968) 58 The American Economic Review 18. 1 2

26

Understanding market power: an economics perspective  27 lection of data on relative income distributions among consumers, workers, and firm owners fall more clearly within the province of taxation, expenditure, and transfer payment activities.6 The state of affairs never ceases to surprise economists. For them, market power is a term of art. However, judges and agencies are not economists. And statutory law, judicial interpretation, the constraints of antitrust practice, and the influence of ideology combine to produce diverse market power conceptions. Besides, the market power story is not always uniformly told by economists. Subtle definitional differences that matter get overlooked in favour of ‘simplistic notions’,7 and all the more if lawyers are in the audience. A review of the economics of market power can therefore help to shed preliminary light on how to evaluate what US and EU antitrust laws do to control undue market power, and what can be done better. Put differently, an economic perspective can be a first step towards the development of a unified and coherent theory of market power control in antitrust. The issue is approached in three steps. The chapter starts with the basic definition of market power as power over price (II).8 Because, however, nearly every economist accepts that a concept of market power as power over price is too simplistic, the chapter surveys more technical definitions of market power in use in the literature (III). Last, the chapter describes economists’ common understanding of classes of pricing phenomena that do not constitute market power problems (IV).

II.

POWER OVER PRICE

When there is market power, one (or more) firm(s) can raise prices above the cost of production. Customers that would benefit from buying units at the cost of production – including at an ordinary rate of return to the firm – are not served. And units that would be profitable to sell are not produced. Sales are missing, even though benefits to buyers exceed the cost to sellers. That outcome, called a ‘deadweight loss’, is inefficient from a social perspective. Some wealth, value, and prosperity are lost.9 Buyers direct their expenditure to ‘less satisfactory purchases’ without this being compensated by gains to the seller with market power.10 Sellers of less valuable commodities increase their output. But the aggregate size of the economic pie remains smaller than without market power, because the worth of these trades is inferior.11 6 Joseph Farrell and Michael L Katz, ‘The Economics of Welfare Standards in Antitrust’ (2006) UC Berkeley: Competition Policy Center, available at https://​escholarship​.org/​uc/​item/​1tw2d426, 26 October 2022. 7 Franklin M Fisher, ‘Diagnosing Monopoly’ (1927) 27 J Reprints Antitrust L & Econ 66 (noting ‘economists often use words which are in common use and whose everyday meanings are not in fact the same as their technical definition’ and adding: ‘the legal profession and the economist (not to mention competitors of the alleged monopolist) all have something different in mind’). 8 Chad Syverson, ‘Macroeconomics and Market Power: Context, Implications, and Open Questions’ (2019) 33 Journal of Economic Perspectives 23, 25. 9 It cannot be observed in the GDP statistics. 10 Abba P Lerner, ‘The Concept of Monopoly and the Measurement of Monopoly Power’ (1934) 1 Review of Economic Studies 157. 11 What should happen is that the resources (ie the capital and labour) used in lower profit industries where purchasers now buy would be more optimally reallocated to the high-profit industry, to supply the output not furnished by the monopoly firm. For an exposition of this idea, see Arnold C Harberger, ‘Monopoly and Resource Allocation’ (1954) 44 The American Economic Review 77.

28  Research handbook on abuse of dominance and monopolization The cost, efficiency, or innovation constraints bearing on firms with market power are also laxer. Market power, simply put, is inefficient. Harberger lumped all this under a concept of ‘malallocative’ or ‘misallocation’ effects of monopoly.12 Now, why do firms with market power leave profits from additional transactions on the table? Economics explains that when firms have no rivals,13 profit maximization leads monopoly firms to choose to reduce output as the privately efficient equilibrium.14 The logical implication is that deviations from competitive conditions engender market power exploitation. Private and social benefits are misaligned. The policy implication is that lost output constitutes a pecuniary externality from rational profit maximization that stands on equal footing with other market failures addressed by public intervention (or private ordering), like information asymmetries, free riding, missing markets, regulatory distortions, and so on.15

III.

TECHNICAL DEFINITIONS OF MARKET POWER

In the economics literature, several technical definitions of market power exist. They arose in response to the classical economists’ initial understanding of monopoly and competition as government privilege. These are reviewed in turn.16 A.

Classical Economics

Classical economists envisioned monopoly as a problem caused by the government. Interference with markets through royal privileges, licences, patents, and tariffs was the main source of monopoly power.17 The problems besetting free markets were different. The

ibid. Economics adds another condition, which is that firms cannot price discriminate, meaning that they cannot charge each user with a distinct reservation value an individual price. 14 Note though, that another privately and possibly even socially efficient equilibrium would arise if the firm was able to perfectly price discriminate, and charge each consumer the maximum price he or she is willing to pay. Yet, the conditions for this to be possible are difficult to fulfil, and seldom encountered in practice. 15 In some economic works, the externalities that are mediated by markets are called pecuniary externalities. Some consider that pecuniary externalities are not market failures, or at least not as problematic from a welfare standpoint. For the concept of pecuniary externalities (an externality manifest in a change in prices), see Martin Shubik, ‘Pecuniary Externalities: A Game Theoretic Analysis’ (1971) 61 American Economic Review 713. 16 Estimation techniques are not discussed. Before estimating, one needs a definition, to know what is measured. Market power estimation techniques are a field of economics of their own, and involve statistical, accounting, and data techniques beyond the scope of this chapter. 17 During the 1890s, despite the passing of the Sherman Act, US economists showed overall little interest in pursuing the analysis of monopoly. Cournot’s pivotal study on duopoly, although being published in 1838, would be recognized only posthumously, having had before only limited influence. The reasons are probably to be found in a mathematical approach which was uncommon at the time, together with the focus on output as a key variable, disregarded as unrealistic. See Irving Fisher, ‘Cournot and Mathematical Economics’ (1898) 12 The Quarterly Journal of Economics 119; JM de Bornier, ‘The “Cournot-Bertrand Debate”: A Historical Perspective’ (1992) 24 History of Political Economy 623; TAB Corley, ‘Emergence of the Theory of Industrial Organization, 1890–1990’ (1990) 19 Business and Economic History 83. 12 13

Understanding market power: an economics perspective  29 opposite of competition was not a monopoly, but ‘cooperation’.18 Concerted action amongst independent firms came first from agreements, congers, guilds, pools, trade associations, and trusts (sometimes supported by governments). More generally, classical economists understood competition as an ‘active process of jockeying for advantage’, that is, the race to get the best deal on the market. Prices above costs were signs of lively competition, not its elimination. When high prices were observed, potential competition was deemed a disciplining force. There was no discussion of monopoly in terms of a falling demand curve, price above costs, or reduced output. B.

Falling Demand Curve

Augustin Cournot, a French (late classical) economist of the nineteenth century, was the first to offer a technical treatment of the issue. Cournot considered that market power stemmed from the falling demand curve. With help of the now familiar ‘demand curve’,19 Cournot established that the units of a good that are demanded (D) are a continuous function of its price (p).20 The functional relation F(p) between p to D is, ‘in general’ that a price increase will be followed by a corresponding decrease in demand.21

Figure 2.1

Demand curve

Mark Blaug, ‘Is Competition Such a Good Thing? Static Efficiency versus Dynamic Efficiency’ (2001) 19 Review of Industrial Organization 37. FA Walker, later the first president of the American Economic Association, wrote in the main textbook of the time that ‘Competition signifies the operation of individual self-interest among the buyers and sellers of any article in any market. It implies that each man is acting for himself solely, by himself solely, in exchange, to get the most he can from others, and to give the least he must himself.’ See Francis Amasa Walker, Political Economy (3rd edn, Macmillan 1982) 91–2. 19 In Chapter IV of Antoine-Augustin Cournot, Recherches Sur Les Principes Mathématiques de La Théorie Des Richesses (Hachette 1838). See also Fisher (n 17). 20 Actually, Cournot talked more of a law of falling demand or a law of falling sales (débit), than of the falling demand curve; Cournot (n 19). 21 Cournot was careful to stress that the relationship needed not to be proportional, and constant, at all levels of prices; ibid, para 23. 18

30  Research handbook on abuse of dominance and monopolization Table 2.1

Example of supply point with falling demand

Price (p)

Units F(p)

Total Revenue pF(p)

Direction

0

10

0

↑

1

9

9

↑

2

8

16

↑

3

7

21

↑

4

6

24

↑

5

5

25

=

6

4

24

↓

7

3

21

↓

8

2

16

↓

9

1

9

↓

10

0

0

↓

So the law of demand means that a firm can sell more by lowering the price. But how can this ever create market power or, in the popular saying, ‘power over price’? The seminal demonstration of Cournot is counterintuitive, but compelling. Each firm, Cournot hypothesized, wants to maximize the ‘total value’, revenue or profit from its ‘things’ or ‘labour’.22 Now, when the demand curve is negatively sloped, it is wrong to assume that maximizing sales maximizes total value.23 Of course, a firm can sell more units by lowering p. However, the theoretical price that maximizes sales is 0, and at p = 0, the revenue to the firm is null. In addition, another property of the falling demand curve must be that an ‘infinite’ price also gives a demand of 0.24 With this background, total value, revenue, or profit maximization requires a firm to pick a price point on the demand curve between zero and a high price. Cournot established another continuous function pF(p) that can be derived from the falling demand curve and that shows that revenue (not demand) tends to increase with p moving away from 0, until a point where it decreases.25 This particular pricing point is the supply point that any firm, including a monopolist, will select on the demand curve (see Table 2.1, for an example). Now, in what sense is the falling demand curve a source of market power? There are at least two ways to see this. The first requires focusing on the differences between the demand curves faced by a monopolist and by a firm operating under perfect competition. In perfect competition, which Cournot also studied, the demand curve faced by a firm is horizontal, because there are multiple alternative sellers of goods. A high price is impossible because it means an absence of sales. By contrast, when the demand curve is falling, a high price means

Cournot indistinctly uses terms like value or profits; ibid. Lawyers usually struggle to capture the depth of this idea, because they implicitly assume that, to sell more, you need to lower the price, and make the hidden assumption that profit maximization means selling more. 24 Cournot said that demand ‘vanishes’ when p becomes infinite. He remarked that a mineral water monopoly can indeed impose an unrealistic price of 100 francs a litre. But soon enough, the mineral water monopolist will face a scarcity, if not a disappearance, of buyers due to the negative slope of the demand curve; Cournot (n 19), paras 24–6. Robinson, in stricter terms, once said that infinite elasticity was an ‘absurdity’; Joan Robinson, The Economics of Imperfect Competition (Springer 1969). 25 The total value function expresses the total value of quantities demanded and sold, which can be noted as pF(p). 22 23

Understanding market power: an economics perspective  31 serving fewer customers, but more revenue than selling to all customers. Relative to perfect competition, the monopoly firm has more power over demand. The second approach consists in reading the falling demand curve as an aggregate population of ‘varied’ consumers who enjoy distinct utility from consumption, itself dependent, said Cournot, upon ‘customs’, ‘norms’, and the ‘scale’ and ‘distribution’ of ‘wealth’ in society.26 Simply put, the falling demand curve means that some customers are ready to pay higher prices than others.27 The firm faces an environment different from one with a flat demand curve where not one customer will accept a higher price. The firm with a falling demand curve can extract a ‘rent’ by selling to some customers, but not to all, says Cournot. The magnitude of that rent will only depend on the ‘nature’ – the slope – of the falling demand curve. A conventional economics presentation is to discuss the falling demand curve as a constraint on monopolists, not as a profit maximization opportunity.28 Lawyers often struggle with this idea. But it has a kernel of truth. The proposition that a monopoly is constrained by demand makes sense if understood as the idea that a monopoly firm is ‘only’ constrained by demand, nothing else. The monopolist maximizes profit given the falling demand curve. This explains that Cournot literally discussed the problem in terms of how the monopoly firm ‘fixes’ a price (this relates to the contemporary concept of ‘price setter’). This is unlike consumers, who ‘bear’ the price set by the monopoly firm.29 C.

Price Above (Marginal) Cost

Another definition characterizes market power as prices exceeding production costs. This definition is the work of neoclassical economists.30 Enlightened, but half-fed by Cournot’s positive theory of monopoly, neoclassical economists had just figured out that monopoly was endogenous.31 Something could be done to address the welfare effects of monopoly. Unfortunately, Cournot’s work was frugal in insights about how to rearrange markets in ways more beneficial to society. In addition, Cournot had described abstract functional relations, not algebraic ‘determinations’ that would have been more directly useful to practitioners. Neoclassical economists, who also understood that market power was a question of degree, needed a more

Cournot (n 19) paras 21–2. Compared to a situation in which the firm must sell all the output that is demanded by all consumers at the lowest price point. The monopoly firm that faces a falling demand curve no longer needs to sell all the output demanded. 28 It is truly a constraint, because the firm, even a monopoly, cannot both select price and output, just one of them. 29 A mainstream account of market power is that the monopoly firm ‘sets’ prices, while consumers ‘take’ prices. To avoid confusion here, it is worth recalling the contrast with Cournot’s oligopoly model where firms set quantities. 30 This idea was already in Ricardo, a cornerstone of classical economics. See David Ricardo, On the Principles of Political Economy and Taxation (John Murray 1817). 31 In this regard, Giocoli writes: ‘Shockingly, American economists discovered that market power could well be the natural product of competitive markets and that, under certain conditions, the inevitable outcome of free competition was monopoly, i.e. the demise of competition itself’; Nicola Giocoli, ‘Free from What? Classical Competition and the Early Decades of American Antitrust’ (2021) 26 New Political Economy 86, 91. 26 27

32  Research handbook on abuse of dominance and monopolization operational procedure to estimate both its magnitude and welfare effects.32 They attacked the problem of the ‘determination’ of the output of the individual firm head-on. Building on Cournot’s insight that the monopolist would proceed by ‘trial and error’, neoclassical economics formalized the process of setting prices as a marginalist thought experiment.33 Assuming profit maximization, the monopolist grows output and lowers prices up to the level where marginal revenue (‘MR’) equals marginal cost (‘MC’) (see Table 2.2 and Figure 2.2).34 Put differently, the monopolist decides to produce an extra quantity of output if (and only if) this yields a revenue increment greater than the costs incurred to produce a marginal unit.35 The neoclassical redux of market power emphasizes the role of two constraints, not just one, on the monopoly firm. The first constraint is the falling demand curve, with a tweak.36 The falling demand curve must be read as meaning that as served buyers experience less satisfaction through consumption, they derive marginally less benefit from extra units, and are thus willing to pay less for them. Importantly, MR is lower than the price at each level of output, because all previous units sell at a lower price too. The second constraint is the existence of increasing (or constant) costs that, neoclassical economists believe, is a good empirical generalization of a firm’s decisional context.37 Costs increase with quantity because the labour force becomes less productive as the scale of production expands,38 and the opportunity cost of leisure increases. There is a declining marginal product of labour. Most supply curves are drawn with an upward slope to show this. The implication is that to produce one more unit, a monopolist must spend more.39 Now, the joint consideration of MR and MC allows a prediction about the level of output of the monopoly firm. The level of output lies at the intersection of MR and MC. It is an equilibrium point, because, under this condition, the level of output will not be altered.40

32 Piero Sraffa, ‘The Laws of Returns under Competitive Conditions’ (1926) 36 The Economic Journal 535. 33 Cournot used the French word ‘tâtonnement’; Cournot (n 19). 34 This is not true for a firm in a competitive market, which sets price equal to marginal cost without considering marginal revenue. The firm in a competitive market is said to be a ‘price taker’, whereas the firm in a monopoly market is said to be a ‘price setter’. 35 Here is an untechnical example to help the non-economist understand the monopolist’s thinking: do marginal returns on producing ten additional pages in a long working paper compensate for the marginal costs of writing them? In this example, marginal returns are the reader’s interest, downloads, or citations to the paper. In both the metaphor and the model, marginal returns tend to decrease when more pages are added to the paper, at least when the reader is a journalist or policymaker. 36 Lerner (n 10). 37 Robinson writes that ‘in general, it may be supposed that in the short period marginal costs begin to rise at a fairly low level of output’; see Robinson (n 24) 50. 38 The earnings of a worker tend to be equal to the net product due to the additional labour of the relevant worker (assuming the marginal product of capital is constant). 39 Sraffa discussed the idea that ‘each competing producer necessarily prices normally in circumstances of individual increasing costs’; see Sraffa (n 32) 543. 40 In Robinson’s terms, there is equilibrium when output is not altered, when there is no tendency to contract or expand its output; Robinson (n 24) 57.

Understanding market power: an economics perspective  33 Table 2.2 Price (p)

Monopoly equilibrium Units F(p)

Total Revenue pF(p)

Marginal Revenue

Marginal Cost (MC)

(MR) 0

10

0

0

4

1

9

9

9

3

2

8

16

7

2

3

7

21

5

1

4

6

24

3

2

5

5

25

1

3

6

4

24

(1)

4

7

3

21

(3)

5

8

2

16

(5)

6

9

1

9

(7)

7

10

0

0

(9)

8

Figure 2.2

Monopoly equilibrium

The neoclassical perspective on market power is ‘artificial’.41 In real life, monopolists do not practice the marginalist thought experiment.42 The process requires unattainable precision in demand estimation.43 Neoclassical economists have overcome this objection by advancing, quite reasonably, that firms behave ‘as if’, balancing ballpark estimations of MCs and receipts. More fatal to the neoclassical perspective is its arbitrary baseline assumption of decreasing returns. Pietro Sraffa, in a 1926 paper, established a functional connection between cost and quantity, and showed that in a large dimension firm, a greater division of labour is possible,

ibid 107. What is meant here is that firms do not undertake the marginal computation that is described here, but may iteratively discover the profit-maximizing price. 43 If they had this information, the monopoly firm would not set a single price, but instead, price discriminate perfectly by charging each customer its own reservation price. Robinson herself said that ‘even the most up-to-date businesses have only the vaguest notion of what kind of demand curves they have to deal with’; Robinson (n 24) 55. 41 42

34  Research handbook on abuse of dominance and monopolization leading to increases in output. Sraffa appeared to lament that a consideration of decreasing costs ‘was entirely abandoned, as it was seen incompatible with competitive conditions’.44 The problem is that disregard of supply curves showing decreased costs risks overpredicting monopoly power.45 Aware of this problem, neoclassical economics developed a ‘more reasonable procedure’ that consists in considering the ‘actual conditions of the monopoly equilibrium’ to study the ‘excess of price over marginal cost’.46 Lerner stated that ‘the loss involved in monopoly can be seen in the divergence between price and this marginal cost’. His famous index,47 which is a ratio of profit (P – MC) to price (p), gives algebraic expression to the ‘degree of monopoly power’.48 The definition of market power as price above MC is equally questionable. Market power has nothing distinctive under the definition. Firms in the real world ordinarily set price above marginal costs.49 Pricing above cost is necessary for every economic activity, to cover the cost of capital, insurance against loss, and managerial earnings.50 Acknowledging this constraint, Marshall conceded that only the revenues above earnings necessary for economic activity would be the ‘exceptional gains’ that are ‘the nature of monopoly’.51 But one may find very legitimate gains in the higher strata of profits reaped by monopoly firms. For accounting or fiscal reasons, firms may decide to ‘price in’ substantial expenses, rather than capitalizing them (like research and development and advertisement). A rebuttal to this issue has consisted in focusing analysis on market power that is ‘substantial’, not trivial (an idea that has been embedded in antitrust law frameworks). Much literature states that market power exists only when deviations from competitive levels are ‘important’ or when a ‘degree of monopoly power’ is reached.52 These qualifications are unhelpful. First, there is the question of whether substantial market power is what is at play in differentiated

44 Sraffa, who had noted that external economies in a growing industry also create increasing returns, remarked that the importance of ‘external economies’ was more and more emphasized – that is, of the advantage derived by individual producers from the growth not of their individual undertakings, but of the industry in its aggregate; Sraffa (n 32) 537–8. 45 Cournot, who had considered different cost patterns (increasing, decreasing, or constant) in his Chapter V, maintained that the falling demand curve was the main explanatory factor of monopoly power because costs can go in any direction; see Cournot (n 19). By contrast, Robinson, and many other neoclassicals based on the assumption that marginal costs would in general increase, considered that costs were at least equally important in determining output levels as the falling demand curve; see Robinson (n 24). 46 Marshall, for example, said that artificial monopoly price is a price ‘determined with little direct reference to cost of production’. He added ‘but chiefly by a consideration of what the market will bear’; Alfred Marshall, Principles of Economics (8th edn, Macmillan and Co 1920) 274. 47 Invented by Italian economist Luigi Amoroso; see Nicola Giocoli, ‘Who Invented the Lerner Index? Luigi Amoroso, the Dominant Firm Model, and the Measurement of Market Power’ (2012) 41 Review of Industrial Organization 181. 48 Lerner (n 10) 168. 49 Benjamin Klein, ‘Market Power in Antitrust: Economic Analysis after Kodak’ (1993) 3 Supreme Court Economic Review 43, 72. 50 Marshall (n 46) 396. Marshall added that the amount produced under a monopoly is not necessarily less than the amount produced under competition or no monopoly, because often the monopoly will have economies of scale; see ibid 400–401. 51 Marshall (n 46) 396. 52 Kurt W Rothschild, ‘The Degree of Monopoly’ (1942) 9 Economica 24, 26.

Understanding market power: an economics perspective  35 goods markets where suppliers compete for demand at prices substantially higher than marginal costs. Even more fatal is the methodological objection. The economic literature seldom formulates an explicit threshold of substantiality.53 Subjectivity, not the ‘scientific’ exactitude called for by Lerner, feeds back into the definition (and later estimation) of market power. A major contribution of the neoclassical economics perspective lies in an understanding of monopoly power in terms of outcome (compared to the process view of the classical economists). But short of a counterfactual, deviations from MC pricing are uninformative. And because it defies human ingenuity ‘to compare the monopoly position with the competitive position’,54 neoclassical works often end up falling back on the idea that a firm’s market power depends on the elasticity of the demand, which itself is chiefly influenced by ‘the number of other firms selling the same commodity and the degree to which substitution is possible’.55 In sum, the neoclassical characterization of market power as price above marginal cost helped raise the legitimacy of welfarist government interventions against monopoly. At the same time, however, it increased the costs of decision-making by creating insuperable measurement issues. D.

Control of Output

The growth in applied economics research in the twentieth century has steered the economic literature towards a more actionable understanding of market power. The definition avoids the Charybdis of over-inclusiveness characterizing the ‘P above MC’ definition. It also escapes the Scylla of abstraction found in the classical definition.56 Economics has no academy to standardize terminology, thus, there is no official version of the definition. However, regularities encountered in the market power literature allow the identification of some building blocks, and the construction of a mainstream definition that corresponds to the one that economists use when they speak about market power. Broadly, the market power definition can be stated as follows: market power consists of one (or more) firm(s)’s freedom purposefully to influence the price by the control of market output and by benefitting from constraints on industry supply. Every element of the formulation matters. The definition starts with the firm, because the market power that matters (to the observer that accepts its endogeneity) lies in the firm. This eliminates the concept deviations from competitive conditions caused by market-, industry-, or economy-wide events. This happens when the demand or supply curve shifts upwards due to a temporary shock. Trade wars, inflation, or environmental hazards are common examples.57 The definition also covers collective action by several firms.58

53 Richard Schmalensee, ‘Another Look at Market Power in Antitrust’ (1982) 95 Harvard Law Review 1789. 54 Lerner (n 10) 476. See also, Jonathan B Baker and Timothy E Bresnahan, ‘Empirical Methods of Identifying and Measuring Market Power’ (1997) 27 J Reprints Antitrust L & Econ 743, 454: ‘it is rarely if ever possible to know what the competitive equilibrium would look like’. 55 Robinson (n 24) 50. 56 Marshall (n 46). 57 All firms, not just monopolies, see their supply curve shift upwards when there is inflation. 58 Marshall talks of monopoly through agreement: ‘monopoly values, that can be traced with more or less distinctness in every case in which a single person or association of persons has the power of fixing

36  Research handbook on abuse of dominance and monopolization The concept of freedom is important, because it insists on the idea that the market power firm’s choices are independent of other firms, unlike in oligopoly or perfect competition. The emphasis on ‘purposefully’ should not be overlooked. As Hicks noted, a monopolist incurs ‘subjective costs of securing a close adaptation to the monopoly output’.59 It is not inconceivable that the private costs of constructing a demand curve sufficiently close to reality outweigh the benefits of supra-competitive prices. Market power may not be exercised even under monopoly conditions. When market power is unused, it can be ignored.60 Moreover, in the case of a tax increase, a firm with monopoly power will only decrease output if it adopts a deliberate decision to pass on the surcharge to users, instead of booking it on its profits. A deliberate course of action by the firm is required. Focus on purpose additionally allows filtering out market power that results from external events, accidents, luck, or legacy investments over which the firm has no choice. If the firm operates under exogenous constraints, no government policy can change firm behaviour in ways which are more beneficial to society without great expense and friction. Economists discuss market power in terms of ‘influence over price’ to avoid using rigid concepts like ‘setting’ or ‘raising’ prices. The idea of price-setting wrongly suggests that the monopoly fixes a price. This is not the case in all industries where firms set quantities first, like airlines, hospitals, or steel. Transaction prices remain indeterminate until demand clears. Plus, ‘influence’ over price captures the economic case of the market power of buyers, which consists of price going down, not up. It also covers all cases in which a firm may find it strategically beneficial to lower its short-term price, while pursuing longer-term monopoly profits. The requirement of ‘control of market output’ is decisive. A firm can only confidently raise (or decrease) prices if it enjoys some control over other market participants’ output.61 Otherwise, any output lost to a price increase will be replaced by an expansion of the output of other firms. Often, people define market power as pricing power by reduction of output, and assume that it is enough that the monopoly firm decreases its own output to raise prices above costs. The idea is intuitive. Reduced output increases the competition between buyers against scarcity. While this is a necessary condition, it is not a sufficient one. There is no scarcity if rival output can flood the market. Some ‘control’ of other firms’ output is also required. A case in point is when a monopoly firm controls market output by securing preferential access to most or all the input supply in the market. Or, when a monopolist invests in excess capacity to deter competitors from expanding. Last, the idea of the benefit of constraints on industry supply means that suppliers of other goods or services that are at best remote substitutes must be prevented from repositioning their resources in response to missing output. ‘Benefit’ denotes that the constraints are not imposed by the firm. It can be conjectured that the costs involved in active entry deterrence towards the vast population of suppliers of other products by far outweigh the gains from reduced either the amount of a commodity that is offered for sale or the price at which it is offered’; Marshall (n 46) 275. 59 JR Hicks, ‘Annual Survey of Economic Theory: The Theory of Monopoly’ (1935) 3 (1) Econometrica 1, 8. This was already in Robinson too, who talked of ‘a maldistribution of resources as between different use’; Robinson (n 24). 60 Lerner (n 10) 170. 61 Even more clearly, John Hicks wrote that if entry into the industry is free, it is ‘impossible to earn more than normal profits’; see Hicks (n 59) 1, 9. Before him, Lerner had talked of ‘protection of competition from rest of supply’; Lerner (n 10) 482.

Understanding market power: an economics perspective  37 output. Exogenous constraints must be present to give confidence to the firm that it is protected against potential competition from industry participants.

IV.

NON-MARKET POWER

Observable increases in prices or above cost prices do not necessarily implicate market power. Economics allows drawing a line between what constitutes usage of market power, and what does not. A.

Demand and Supply Shifts

Prices can increase as a result of upward shifts in the demand or supply curves. Such shifts in the demand curve occur when benefits from consumption increase. For example, heavy rainfall raises the utility of umbrellas. Shifts in the supply curve arise when the marginal costs of production increase. For example, shortages of chipsets raise the manufacturing cost of personal computers. Upward shifts in the demand or supply curve that raise prices do not necessarily involve the exercise of market power. Economics captures this idea by distinguishing movements of the demand/supply curve from movements along the demand/supply curve. A monopolist will typically raise prices by selecting a higher price point on the demand curve. By contrast, a monopolist (and its competitors without market power) will both benefit from an upward shift in the demand curve. A firm with market power can internalize a shift of the supply curve on its own profits, and limit price increases to buyers. This will not happen with firms operating under perfect competition. Firms in perfect competition make zero profits. Increases in costs must be passed on through a price increase. In this case, evidence of increased prices to consumers denotes more a competition case than a market power one. The cause of the change in demand and supply can be exogenous or endogenous to the firm. Endogenous changes are particularly confusing because it is tempting to equate them with monopolistic practices. The issue of an endogenous shift in the supply curve can be left aside as an anomaly. By contrast, demand shifts can also come from a firm deciding to add a new line of complements to existing products (including those of rivals), thereby raising utility levels for the whole market, or from a decision to increase marketing expenditure to raise users’ willingness to buy. Even more problematic is the case in which a firm decision will cause an uncertain risk of increased market power concomitant with a shift in the demand curve. Figure 2.3 shows the effect of a horizontal merger in a concentrated industry that leads to (i) a probability of increased market power and (ii) a probabilistic innovation (new product) with high marginal benefits for users. Both effects can materialize cumulatively or alternatively. The post-merger shift in the demand curve produces at least three times the value of the deadweight loss arising from a post-merger usage of monopoly power. If the examiner gets the market power prediction wrong, and post-merger prices are competitive (Pc), a substantial value will be destroyed. Now, the picture below shows a movement of the demand curve that might be extreme in the real world. But it gives a sense of the general orders of magnitude. Even if we halved the shift in the demand curve, the gains in value would still remain above

38  Research handbook on abuse of dominance and monopolization the monopoly losses. And it is only reasonable to use a shift of such magnitude, which corresponds to the small base of the deadweight loss triangle. In addition, if there are plausible cost efficiencies from mergers (see Figure 2.4), the gains are even higher.

Figure 2.3

Horizontal merger in a concentrated industry

Figure 2.4

Horizontal merger in a concentrated industry with cost efficiencies

B.

Profits

Profits entail prices above costs. But economics does not treat profits as market power. Economists ‘like’ to ‘count’ the return that a firm gets from its services as a ‘cost’ rather than

Understanding market power: an economics perspective  39 as a profit.62 The reason for this convention is that returns earned on capital invested (including capital goods, physical labour, and human capital), as well as on other expenditure like research and development or marketing, constitute the ‘normal profits’ that reward competitive activity.63 If profits are less than normal, ‘firms will tend to leave the industry’.64 Economists interested in market power are concerned about profits ‘beyond’ these.65 Excess profits that add up to the normal rate of return cannot persist in the long term. They disappear in the short term with the entry or expansion of new firms. The short term is the time before which the firm earning profits can reinvest in new capacity (or the time during which its productive equipment, and costs, are fixed).66 The long term is when the technique of production can be altered.67 If long-term profits are observed, an inference can be drawn that ‘something blocks entry’.68 A restriction of entry becomes a condition to consider profits ‘abnormal’. In some cases, ‘abnormal’ profits are legal. For example, laws granting firms a legal monopoly lay down entry restrictions towards third parties. The point is to give the protected firm the opportunities to earn rents beyond the normal rate of return and incentivize investments. In other cases, firms will earn differential rates of return, and it will be wrong to consider that the firm earning the highest profits holds market power. Ricardian rents arise from productivity differences amongst the resources under the control of firms. They will not be dissipated by entry. A firm with better land, well-positioned estate, or a superior method of organizing production, providing services, or establishing buyer confidence will produce at a lower cost ‘than can be obtained by newer or smaller firms’.69 Productivity differentials between energy generation between nuclear, coal, and renewables cause producers to display large discrepancies in profitability.70 Sometimes, new entrants themselves provide services that command higher Ricardian rents, because old firms cannot expand as cost effectively to satisfy new demand as new entrants. For example, entrants in digital advertisement services might command higher profits than incumbent sellers of display, TV, or classified advertising, because of higher productivity differentials, that is a higher sales conversion rate.71

Hicks (n 61). In his famous 1954 paper, Harberger did just this, ‘build[ing] a 10 per cent return into the cost curve’; Harberger (n 11) 78. Robinson also includes in the costs of production what she calls the ‘minimum reward of the entrepreneur’; Robinson (n 24) 47. 63 Hicks (n 61). 64 Robinson (n 24). 65 Also known as ‘exceptional’ profits, supernormal profits or superprofits. 66 Robinson (n 24) 47. Robinson draws a difference between the short period, the quasi-long period, and the long period. 67 ibid 48 68 Fisher (n 17). Also remember that Hicks writes: ‘if entry in industry is free, it is impossible for firms to earn more than “normal profits”’; Hicks (n 61) 9. 69 Harold Demsetz, ‘Two Systems of Belief about Monopoly’ in Harvey J Goldschmid and others (eds), Industrial Concentration: The New Learning (Little Brown 1974). 70 See Roger G Noll, ‘“Buyer Power” and Economic Policy’ (2005) 72 Antitrust Law Journal 589. In particular, he takes the example of productivity differentials in natural gas extraction among wells and fields, where ‘competitive pricing causes most producers to earn very large Ricardian rents, while the market price equals the average cost of the most costly well in production’. 71 Ricardian rents need not attract, or discourage entry. A firm might enter or expand, on the view that some customers are ready to pay lower rents for a product of lower quality. 62

40  Research handbook on abuse of dominance and monopolization Economists’ approach to profits has two practical implications. Accounting profits are poor proxies of market power. This is because accounting profits comprise the normal rates of returns required to induce firms to do business. In recent years, these ‘imperfections’ have not prevented a strand in the macroeconomic literature from using accounting data to produce evidence of increased markups across the US economy.72 The second is that inferring market power from cost benchmarks might be attempted, but is fallacious. Joan Robinson laid out a very important, but subtle distinction.73 Average costs determine whether a firm enters an industry and continues in business. Average costs do not determine price or output. Marginal costs do. This conception has led to the belief that market power can be observed by tracking deviations from marginal costs. The problem is not that it is impossible (as Robinson conceded) or wrong. It is that it is fallacious, because this convention sweeps away the necessary return that was embedded in the cost function, and that is the predicate for business activity without which competition would not exist. This return would be seen with an average cost definition. C.

Bargaining Power

It is inappropriate to discuss all economic transactions at prices above costs in terms of market power. Buyers (and sellers) on markets happen to make bad deals. Paying too much for a good or service is not a market power problem. It is an income redistribution one. As long as the price a buyer pays remains below her marginal benefit, there is no inefficiency. No gains from trade are lost. Economists look at the ‘excessive’ price paid to a seller as a transfer of resources. Economic theory supplies no consequential or moral reason to deem increases in sellers’ utility less socially worthy compared to increases in buyers’ utility. Unequal prices or terms between parties to economic exchange occur when both sides do not have similar bargaining power. Differences in opportunity costs, and knowledge of these differences, create asymmetries in bargaining power. Bargaining power means that negotiating parties do not stand to lose equally if one walks away from the transaction.74 In general, the party that least needs the agreement gets the majority of the proceeds.75 This might have to do with information asymmetries, incomplete contracts, risk aversion, transaction costs, or asset-specific investments.76 These market imperfections, which can exist independently of market power, are determinants in outcomes often confused with market power, like holdup, consumer exploitation, lock-in, and economic dependence. Of course, the existence of competition decreases the bargaining power of the seller. However, the existence of competition does not imply the absence of bargaining power. For example, sellers in a competitive industry might lose less from a negotiation breakdown than 72 Syverson (n 8). Note that in the past, Harberger had done this himself, saying ‘it is only reasonable to identify monopoly power with high rates of profit’; Harberger (n 11) 84. 73 Robinson (n 24) 48. 74 ‘The power to withhold from making a transaction is probably the most general content of the term “bargaining power”’; John T Dunlop and Benjamin Higgins, ‘“Bargaining Power” and Market Structures’ (1942) 50 Journal of Political Economy 1, 2. 75 Preston McAfee, Competitive Solutions: The Strategist’s Toolkit (Princeton University Press 2005). 76 Benjamin Klein and others, ‘Vertical Integration, Appropriable Rents, and the Competitive Contracting Process’ (1978) 21 The Journal of Law and Economics 297.

Understanding market power: an economics perspective  41 their buyers who operate in a concentrated market. In the gasoline industry, competitive suppliers meet highly inelastic demand from airlines.77 Absence of competition on the buying side can also bring prices closer to marginal costs. The practical implication is that charging what the market can bear is not a market power problem. Conversely, even under conditions of monopoly, the existence of supra-competitive prices charged on the units supplied is not a market power problem. It is just bargaining power exerted on the share of demand that is served. Those ‘inframarginal’ customers that pay units at a monopoly price do not divert expenditure to less satisfactory channels.78 The same is true of perfect price discrimination. When a monopoly seller can charge each consumer her maximum reservation price, all surplus is absorbed, but the output is similar to the level that obtains under perfect competition.

V. CONCLUSIONS With all their imperfections, the above definitions are small steps towards a better understanding of what makes markets competitive or not. The robustness of the above definitions has, however, come under heavy stress in the contemporary economic context. The falling demand curve or rising supply curve are both weak descriptors of the situation of firms in network industries where individual users register growing marginal benefits as the demand served by the network monopolist increases, as well as substantial economies of scale and scope due to decreasing or zero marginal costs. Network industries display, at least to some extent, an upward-sloping demand curve and a downward-sloping cost curve. More generally, the digital age of today features costs and demand conditions different from the agrarian or industrial age described above. Faced with this issue, some antitrust scholars have advanced simpler definitions of monopoly power. One popular idea is to define it as the absence of alternatives, a bit like the classical economists’ view of government privileges. Another popular definition of monopoly is in terms of bigness. The definition has several variations. It alternatively treats large size, scale, share, breadth, and/or growth of the business corporation as signs of monopoly. An influential example is Lina Khan’s discussion of monopoly in terms of structural dominance, used to ‘connote that the company controls a significant share of market activity in a sector’.79 It might be questionable to use such a broad and amorphous concept of market power because it says nothing of its determinants. Bigness is a symptom, not a cause. How can effective policy against market power ever be designed absent a good understanding of its determinants?80

Kenneth Hendricks and R Preston Mcafee, ‘A Theory of Bilateral Oligopoly’ (2010) 48 Economic Inquiry 391, 406. 78 Lerner (n 10) 158. Even though it might be objected that they enjoy fewer resources to purchase in other markets, where production is efficient. For instance, a customer of milk that pays a monopoly price enjoys fewer resources for pizza night. 79 Lina M Khan, ‘Amazon’s Antitrust Paradox’ (2017) 126 The Yale Law Journal 710. 80 Unless, of course, direct regulation of prices is the preferred option. 77

3. Concentration and rising market power: fears and facts Gregory J. Werden

I. INTRODUCTION In the Second Industrial Revolution, entrepreneurs invested in new machinery that lowered cost and increased output. Fierce competition among efficient factories then pushed down prices and put substantial investments at risk. To keep prices up, entrepreneurs resorted to output-restricting cooperation and consolidation.1 Concern about the exercise of market power and fear of large corporations motivated the first antitrust laws in Canada (1889), the United States (US) (1890) and many individual US states (beginning in 1888).2 First-generation antitrust laws did not directly address mergers, and by prohibiting cooperation, they encouraged consolidation.3 As a great merger wave peaked in 1899,4 increasing concentration again became a concern,5 and a grassroots antitrust movement sprang up in the US. Congress made the first appropriation earmarked for antitrust enforcement in 1903; cases filed in 1906 and 1907 against Standard Oil and American Tobacco resulted in their dissolution in 1911; antitrust was a central issue in the three-way 1912 presidential election campaign; and major antitrust legislation was enacted in 1914.6 The Roaring Twenties produced a second merger wave, which ended with a financial crash and the onset of the Great Depression. This merger wave spawned the first systematic studies of economic concentration, beginning with a landmark work of Adolf A. Berle and Gardiner

See, eg HA Marquand, The Dynamics of Industrial Combination (Longmans, Green and Co. 1931) 3 (‘the offspring of intensified competition was combination’). 2 The ‘trust’ from which ‘antitrust’ derived, originally took a particular organizational form with features of cooperation and consolidation. The 1882 Standard Oil Trust was the most prominent example. But the word ‘trust’ came to embrace all large companies. On the rise of the trusts and the political response, see Gregory J Werden, The Foundations of Antitrust: Events, Ideas, and Doctrines (Carolina Academic Press 2020) chs 2–4. The trust movement was not confined to North America. See, eg Herman Levy, Monopolies, Cartels and Trusts in British Industry (Macmillan 1927); Henry W Macrosty, The Trust Movement in British Industry: A Study of Business Organization (Longmans, Green and Co. 1907). 3 Economists observed this at the time. See, eg Richard T Ely, Monopolies and Trusts (Macmillan 1900) 244. 4 See Ralph L Nelson, Merger Movements in American Industry 1895–1956 (Princeton University Press 1959) ch 3. Great Britain experienced a less pronounced merger movement at the same time; ibid 128–9. 5 Speakers at an 1899 conference noted the ongoing wave of consolidations. See Franklin H Head (ed), Chicago Conference on Trusts (Civic Federation of Chicago 1900). See also Luther Conant, Jr, ‘Industrial Consolidations in the United States’ (1901) 7 Publications of the American Statistical Association 1. 6 See Werden (n 2) chs 8–12. 1

42

Concentration and rising market power: fears and facts  43 C. Means. They observed that the 200 largest non-financial corporations in the US accounted for 49 per cent of the assets owned by such companies.7 President Franklin D. Roosevelt’s January 1938 State of the Union Address to a joint session of Congress drew attention to economic and social problems arising ‘out of the concentration of economic control’.8 On the day of the address, Congress received a report by Robert H. Jackson, Assistant Attorney General in charge of the Antitrust Division, providing evidence of ‘increasing concentration of business control’ ‘which eliminates or hampers competition’ and promotes inequality ‘in the distribution of national income’.9 Roosevelt relied on Jackson’s report in a 29 April 1938 message to Congress declaring that: Among us today a concentration of private power without equal in history is growing. This concentration is seriously impairing the economic effectiveness of private enterprise as a way of providing employment for labor and capital and as a way of assuring a more equitable distribution of income and earnings among the people of the nation as a whole.10

Congress finally amended US antitrust law to prohibit anticompetitive mergers in 1950, and judicial interpretation of the law in the 1960s rendered horizontal mergers nearly per se illegal. This helped precipitate a wave of conglomerate mergers giving the 200 largest non-financial corporations control of 58 per cent of the assets owned by such companies.11 On 6 June 1969, Attorney General John N. Mitchell declared, ‘The danger that this super-concentration poses to our economic, political and social structure cannot be overestimated’.12 Richard W. McLaren, Assistant Attorney General in charge of the Antitrust Division, then led a crusade against conglomerate mergers. It failed because he could not convince the courts that conglomerate mergers threatened competition. Horizontal merger enforcement in the United States was relaxed significantly in the mid-1980s. The courts also became ever more demanding of proof of likely anticompetitive effects. Many twenty-first century mergers would not have been permitted without this evolution in policy and law. Concerns were raised about several large mergers, and the intensity of

Adolf A Berle and Gardiner C Means, The Modern Corporation and Private Property (Macmillan 1932, rev edn, Harcourt, Brace & World 1967) ch 3. Berle and Means projected that the asset share of the top 200 non-financial corporations would increase to 70 per cent by 1950, but their share did not increase at all. See Joe S Bain, Industrial Organization (2nd edn, John Wiley 1968) ch 4. 8 ‘Annual Message to Congress’ (3 January 1938), www​.presidency​.ucsb​.edu/​documents/​annual​ -message​-congress​-0, accessed 7 May 2021. 9 ‘Report of Assistant Attorney General Robert H. Jackson’ in Annual Report of the Attorney General of the United States for Fiscal Year 1937 (USGPO 1937). Jackson made similar comments in a 17 September 1937 speech: Robert H Jackson, ‘Should the Antitrust Laws Be Revised?’ (1937) 71 United States Law Review 575. Jackson worked closely with the president at this time. See William Kolasky, ‘Robert H. Jackson: How a “Country Lawyer” Converted Franklin Roosevelt into a Trustbuster’ (2013) 27 Antitrust 85, 89–90. Jackson became Solicitor General in 1938, Attorney General in 1940 and Justice of the Supreme Court in 1941. 10 ‘Message to Congress on Curbing Monopolies’ (29 April 1938), www​.presidency​.ucsb​.edu/​ documents/​message​-congress​-curbing​-monopolies, accessed 7 May 2021. 11 See, eg John M Blair, Economic Concentration: Structure, Behavior and Public Policy (Harcourt Brace Jovanovich 1972) ch 12. 12 John N Mitchell, ‘Address to the Georgia Bar Association’ (6 June 1969), www​.justice​.gov/​sites/​ default/​files/​ag/​legacy/​2011/​08/​23/​06​-06​-1969b​.pdf, accessed 5 April 2021. 7

44  Research handbook on abuse of dominance and monopolization the concerns grew in recent years. For the fourth time in US history, increasing concentration became a political issue,13 and this time, it also became an issue in Europe. A watershed was the 2016 ‘issue brief’ released by the US Council of Economic Advisers (CEA) identifying ‘trends that are broadly suggestive of a decline in competition: increasing industry concentration, increasing rents accruing to a few firms, and lower levels of firm entry and labor market mobility’.14 The evidence of increasing concentration was mainly a comparison of 1997 and 2012 data for broad sectors of the economy. The CEA pointed to significant increases in concentration in transportation and warehousing, retail trade, finance and insurance, and wholesale trade. The CEA encouraged alarm despite noting that ‘antitrust authorities direct their attention to concentration at the relevant market level’ and conceding that the evidence presented was ‘neither a necessary nor sufficient condition to indicate an increase in market power’.15 To understand the evidence on concentration trends, it is necessary to revisit the history of industrial organization economics. Section II provides essential background information, mainly by reviewing what students were once expected to know on concentration measurement and on how markets naturally concentrate. Section III reviews recent claims of increasing concentration and explains that the cited evidence does not relate to markets, although that is the level at which competition occurs. The available market-level data for the US does not show increasing concentration, but some evidence of increasing market-level concentration exists for Europe. Section IV explains why other evidence cited by the CEA does not suggest a decline in competition. Section V offers final thoughts, including that the evidence cited in support of increasing concentration reflects mergers that extend geographic or product reach without affecting market concentration.

II.

OLD LEARNING ON MARKET CONCENTRATION

A.

Concentration Data from the US Bureau of the Census

Industrial organization economists characterize a size distribution of firms with a ‘concentration index’ reflecting the number of firms and their size inequality.16 Economists have proposed numerous concentration indexes, beginning in the 1930s when Gardiner C. Means worked with the US Bureau of the Census to construct concentration indexes for 275 manufacturing industries.17 Census disclosure rules allowed the publication of data only if four

13 As this chapter was finalized in the Summer of 2021, antitrust was a more prominent public issue in the US than it had been during the previous century. 14 Council of Economic Advisers, ‘Benefits of Competition and Indicators of Market Power’ (Issue Brief, May 2016) 4, https://​obamawhitehouse​.archives​.gov/​sites/​default/​files/​page/​files/​20160502​ _competition​_issue​_brief­​_updated​_cea​.pdf, accessed 1 April 2021. 15 ibid. 16 See MA Adelman, ‘The Measurement of Industrial Concentration’ (1951) 33 Review of Economics and Statistics 269; Gideon Rosenbluth, ‘Measures of Concentration’ in Business Concentration and Price Policy: A Conference of the Universities–National Bureau Committee for Economic Research (Princeton University Press 1955). 17 National Resources Committee, The Structure of the American Economy, Pt I: Basic Characteristics (USGPO 1939) app 7. Grace W Knott compiled the data. A comparable UK study followed: H Leak and

Concentration and rising market power: fears and facts  45 or more companies were aggregated, so Means used four-firm and eight-firm concentration ratios to characterize industries.18 The n-firm concentration ratio, CRn, is the percentage of any measure of firm size, for example turnover, accounted for by the n largest firms when ranked by that measure. With shares of 35, 20, 15, 10, 8, 7 and 5, CR1 = 35, CR2 = 55, CR4 = 80 and CR8 = 100. President Roosevelt’s 29 April 1938 message to Congress proposed a ‘study of the concentration of economic power in American industry’, and Congress responded by creating the Temporary National Economic Committee (TNEC). It produced 37 volumes of hearings and 43 monographs.19 One monograph painted a statistical portrait of US industry,20 including Walter F. Crowder’s data on product-level concentration. Crowder stressed that the ‘product concept approaches more closely that of the economic commodity as employed in the usual type of economic analysis’ and that many non-competing products are produced by the firms in most industries.21 Crowder sampled 117 industries from Means’s study and found that they contained 1,807 products. He also found that 76 per cent of the products had a CR4 in excess of 50 (which was a benchmark often used to denote high concentration), and those products accounted for 57 per cent of the total turnover for the industries.22 In comparison, only 34 per cent of industries in Means’s study had a CR4 in excess of 50.23 Higher concentration at the product level than at the industry level arises from specialization, and the CR4 for a properly defined market can be much higher than that for the corresponding industry. While the foregoing studies were underway, the Census began developing the Standard Industrial Classification (SIC), which divided economic activity into industries represented by four-digit codes. The first SIC manual in book form was published in 1941, and the first SIC data was that for the 1947 Census of Manufacturers.24 From the outset, the Census published concentration ratios for four-digit manufacturing industries. Concentration ratios for five-digit product classes were added beginning with the 1954 economic census. The SIC system was supplanted by the North American Free Trade Agreement. It required consistent data for the three signatory countries, although they were permitted to use more refined categories internally. The North American Industry Classification System (NAICS) is much like the SIC system but uses broader groupings.25 NAICS currently divides the economy

A Maizels, ‘The Structure of British Industry’ (1945) 108 Journal of the Royal Statistical Society 142. UK industries were more concentrated than those in the US. See Rosenbluth (n 16) 71–7. 18 Different disclosure rules in other countries led to the use of three-firm and five-firm concentration ratios. 19 On the TNEC and its work, see David Lynch, The Concentration of Economic Power (Columbia University Press 1946). 20 TNEC, Investigation of Concentration of Economic Power, Monograph No 27, The Structure of Industry (prepared under the direction of Willard L. Thorp and Walter F. Crowder) (USGPO 1941). 21 Walter F Crowder, ‘The Concentration of Production in Manufacturing’ pt V of TNEC Monograph No 27 (n 20) 275. Genevieve Beckwith Wimsatt compiled the data. 22 ibid 292. 23 National Resources Committee (n 17) 240–48. Although this study did not weight by turnover, it did sort industries by economic importance. Only 29 per cent of the large industries had a CR4 in excess of 50. 24 In 1948 the UK introduced a Standard Industrial Classification, which remains in use. 25 The EU adopted the Statistical Classification of Economic Activities in the European Community (NACE, based on the French name) in 1990.

46  Research handbook on abuse of dominance and monopolization into 20 sectors, 99 subsectors, 311 four-digit groups, and 709 five-digit industries. As it is used in the United States, the NAICS also has 1,057 six-digit industries, whereas the SIC system had 1,530 four-digit industries. Since 1997 the Census has published CR4, CR8, CR20 and CR50 at each level of the US NAICS and for nearly every sector of the economy. For the manufacturing sector, the Census also has published the Herfindahl-Hirschman Index (HHI), which is the most popular concentration index among industrial organization economists. It is computed by summing the squared shares (expressed as percentages) of all firms in an industry. With shares of 35, 20, 15, 10, 8, 7 and 5, the HHI is 2088. B.

Census Data Do Not Reflect Market Concentration

In the 1970s, SIC data was a mainstay of empirical research in industrial organization economics, and economists knew it had significant limitations.26 In his 1970 textbook, F.M. Scherer wrote: Unfortunately, Census Bureau industry and product class definitions do not always conform consistently to the criteria economists would like to apply. To get its difficult job done at all, the Bureau must use definitions facilitating accurate reporting by business firms, which usually means that they must follow the way firms have grouped or segregated their production operations. Emphasis often is on similarity of production processes, which may not reflect competitive interrelationships.27

Much as the decennial census counted people, the quinquennial economic census counted ‘establishments’ – facilities engaged in economic activity, such as factories, stores and offices.28 The Census classified establishments based on their primary activity. Many four-digit SIC industries were defined by processes,29 for example paper mills were in SIC 2621 and steel mills were in SIC 3312. Establishments in these industries produced numerous non-competing products.30 On the other hand, classification sometimes was done on basis of inputs, so two different industries could produce the same product. Cane sugar refineries were in SIC 2062, while beet sugar refineries were in SIC 2063, and both mainly produced crystalline sucrose. Problems with SIC data were compounded when firm-based data were forced into the molds of SIC industries by assigning firms to industries based on the plurality of their activity. A tabulation of 1947 data for 23 industries found wide disparity between CR4 calculated by the Federal Trade Commission using firm-based data and CR4 calculated by the Census using

See Bain (n 7) 126–33; FM Scherer, Industrial Market Structure and Economic Performance (Rand McNally 1970) 52–7. 27 Scherer (n 26) 53. 28 The past tense is used here to describe what was known decades ago. The basic methods used by the Census have not changed. 29 See Maxwell R Conklin and Harold T Goldstein, ‘Census Principles of Industry and Product Classification, Manufacturing Industries’ in Business Concentration and Price Policy: A Conference of the Universities–National Bureau Committee for Economic Research (Princeton University Press 1955) 17–22. 30 In addition, less-important industries were combined, eg SIC 2869 (industrial organic chemicals, not elsewhere classified) and SIC 3429 (hardware, not elsewhere classified) included thousands of distinct products made with hundreds of distinct processes. Empirical research generally dropped these catchall industries. 26

Concentration and rising market power: fears and facts  47 establishment-based data. The firm-based CR4 was at least 25 per cent greater for eight industries.31 The largest firms in an industry tended to be the most diverse, so assigning all of their turnover to their primary industry exaggerated industry concentration. The leading old-school industrial organization textbook considered any study using firm-based data to be ‘virtually worthless’.32 Just as Census industries could be much broader than economic markets in the product dimension, they could be much broader in the geographic dimension. Census concentration data were available only at the level of the entire United States, but some manufacturing industries had regional markets, and some markets were local, as in asphalt, bread33 and ready-mixed concrete. Local markets were common in non-manufacturing industries, such as construction, personal services and retail trade. Since the 1970s, a serious problem for manufacturing industries has been the failure of Census data to reflect imports. Long after the focus of empirical research in industrial organization shifted away from Census data, merger filings still provided the US antitrust enforcement agencies with data for four-digit SIC industries. In a 1988 article, the present author examined the utility of SIC data by comparing the scope of cartels prosecuted by the US Department of Justice with the scope of the corresponding four-digit SIC industries.34 The comparison divided (1) the annual ‘volume of commerce’ (turnover) affected by a cartel by (2) the annual ‘value of shipments’ for the corresponding SIC industry. The resulting ‘Commerce Quotient’ built on the idea that the relevant market in a competition case had the same scope as the optimal cartel,35 and the insight that actual cartels should not be much narrower than optimal cartels. Because economists had used SIC data only for manufacturing industries, the study was limited to cases within that sector. For 52 of the 80 manufacturing cases filed during 1970–80, the Commerce Quotient was less than 0.01. Special attention was paid to cases in which the corresponding SIC industry had been thought to comport tolerably well with an economic market. For 16 of those 19 cases, the Commerce Quotient was less than 0.1. The US Department of Justice promulgated merger guidelines in 1982 incorporating the hypothetical monopolist paradigm for market delineation.36 A critical insight motivating the paradigm was that relevant markets used to assess market power issues must be delineated on the demand side of the market. Beginning in 1982, the Department of Justice alleged relevant markets for which it was prepared to prove that a hypothetical monopolist would possess significant market power. Application of the hypothetical monopolist paradigm in merger cases provided a second test for how well SIC industries comported with economic markets.

Conklin and Goldstein (n 29) 32. FM Scherer and David Ross, Industrial Market Structure and Economic Performance (Houghton Mifflin 1990) 418. 33 A special tabulation for the bread industry found far higher concentration in local markets than indicated by the national Census data. Committee on the Judiciary, United States Senate, ‘Administered Prices: Bread’ (Senate Report No 1923, 86th Congress) (USGPO 1960) 119–25. 34 Gregory J Werden, ‘The Divergence of SIC Industries from Antitrust Markets: Some Evidence from Price Fixing Cases’ (1988) 28 Economics Letters 193. 35 See, eg Kenneth D Boyer, ‘Is There a Principle for Defining Industries?’ (1984) 51 Southern Economic Journal 761; Gregory J Werden, ‘Market Delineation and the Justice Department’s Merger Guidelines’ [1983] Duke Law Journal 514. 36 See Werden (n 35); Gregory J Werden, ‘Market Delineation under the Merger Guidelines: A Tenth Anniversary Retrospective’ (1993) 38 Antitrust Bulletin 517. 31 32

48  Research handbook on abuse of dominance and monopolization In a 1990 article, Russell W. Pittman and the present author computed Commerce Quotients for Justice Department merger cases filed during the first seven years after release of the 1982 merger guidelines.37 Those cases alleged violations in 47 relevant markets. The Commerce Quotient was greater than 0.25 for 12 of the 47, and it was less than 0.01 for 26. For 12 of the markets, the corresponding four-digit SIC industry had been thought to comport well with an economic market. For five of those 12, the Commerce Quotient was less than 0.1, and it was greater than 0.25 for just two. In 2018, when NAICS concentration data were used in support of claims of increasing concentration, Luke M. Froeb and the present author calculated Commerce Quotients for the relevant markets alleged in merger complaints filed by the US Justice Department during fiscal years 2013–15.38 This study used Census data for six-digit NAICS industries. For 26 of the 44 relevant markets, the Commerce Quotient was less than 0.005, meaning that the finest level of data in the NAICS could aggregate more than 200 distinct markets. The Commerce Quotient was greater than 0.1 for just three of the 44 markets. The Commerce Quotients were far lower in the 2010s than in the 1980s mainly because the 1980s cases were all in manufacturing, whereas many of the 2010s cases were in non-manufacturing industries with local markets, such as crushed stone, film exhibition, radio advertising and waste management. In addition, six-digit industries under the NAICS are, on average, broader than four-digit SIC industries. Data for six-digit NAICS industries are as close as the Census now gets to the market level, but the relevant markets in competition cases tend to be far narrower. In most US Justice Department merger cases, the relevant market makes up less than 0.5 per cent of the corresponding six-digit NAICS industry. Higher levels of aggregation are further removed from the market level. The 20 NAICS sectors contain 1,057 six-digit industries, so a single sector could aggregate 10,000 distinct markets. C.

Natural Causes of Increasing Market Concentration

In 1931 French engineer Robert Gibrat published the first formal model with firm and market dynamics and observed Gibrat’s Law of Proportionate Growth.39 It holds that the increment to a firm’s size over any period of time is proportionate to its initial size. Gibrat’s Law implies a skew distribution of firm sizes – in particular, the log-normal distribution. Gibrat’s data indicated that firm size was log-normally distributed in French industries, with a few large firms and a long tail of smaller firms. In the 1950s economists propounded stochastic growth models based on Gibrat’s Law and used the models to explain the skew distribution of firm sizes observed in the United Kingdom (UK) and US.40 The models featured several different entry processes, producing different Russell W Pittman and Gregory J Werden, ‘The Divergence of SIC Industries from Antitrust Markets: Indications from Justice Department Merger Cases’ (1990) 33 Economic Letters 283. 38 Gregory J Werden and Luke M Froeb, ‘Don’t Panic: A Guide to Claims of Increasing Concentration’ (2018) 33 Antitrust 74. 39 Robert Gibrat, Les Inégalités Économiques; Applications: Aux Inégalités des Richesses, à la Concentration des Entreprises, aux Populations des Villes, aux Statistiques des Familles, etc., d’une Loi Nouvelle, la Loi de l’Effet Proportionnel (Librairie du Recueil Sirey 1931). 40 PE Hart and SJ Prais, ‘The Analysis of Business Concentration: A Statistical Approach’ (1956) 119 Journal of the Royal Statistical Society 150; Herbert A Simon and Charles P Bonini, ‘The Size 37

Concentration and rising market power: fears and facts  49 long-run distributions of firms, but all were skew distributions implying high concentration. For his 1970 industrial organization textbook, F.M. Scherer illustrated these models by simulating industry evolution in a stochastic growth model calibrated to data for large US firms. He posited an industry initially having 50 identical firms, yielding a CR4 of just 8. In a series of simulations, the average CR4 was 27 after 40 years, 42 after 80 years and 53 after 120 years.41 The foregoing models were not meant to be realistic.42 They suppressed things usually of interest to economists – investment, pricing, mergers and the like. But they did usefully demonstrate that increasing market concentration is the natural consequence of a competitive process in which firms have differing degrees of success. In these models, all firms were created equal, and the market evolved through random chance. In real-world industries, firms can be created unequal, and economic forces can speed the process of market concentration. A firm can grow by offering a better product than its rivals, and it can keep growing due to size-related advantages such as economies of scale or network effects. To the extent that market concentration reflects the growth of firms that innovate and develop products to which consumers flock, it should not be a concern for competition policy.43

III.

EVALUATING RECENT CLAIMS OF INCREASING CONCENTRATION

A.

Claims of Increasing Concentration

A March 2016 story in The Economist was not the first twenty-first century claim of increasing concentration in the US economy, but it was the first to get much attention. Staff researchers compared the 1997 CR4 for six-digit NAICS industries to the most recent data – 2007 data for 2 sectors and 2012 data for 13 sectors.44 The Economist defined an industry to be ‘concentrated’ if the CR4 was between 33.3 and 66.7, and reported that the share of revenue in ‘concentrated’ industries increased from 24 to 33 per cent. But six-digit NAICS industries are

Distribution of Business Firms’ (1958) 48 American Economic Review 607; Irma G Adelman, ‘A Stochastic Analysis of the Size Distribution of Firms’ (1958) 53 Journal of the American Statistical Association 893. 41 Scherer (n 26) 125–7. 42 For an introduction to more complex models, see John Sutton, ‘Gibrat’s Legacy’ (1997) 35 Journal of Economic Literature 40, 47–8. More recent literature includes Gabriel Y Weintraub, C Lenier Benkard and Benjamin Van Roy, ‘Markov Perfect Industry Dynamics with Many Firms’ (2008) 76 Econometrica 1375; Richard Ericson and Ariel Pakes, ‘Markov-Perfect Industry Dynamics: A Framework for Empirical Work’ (1995) 62 Review of Economic Studies 53. 43 A few scholars argued in the 1970s that the correlation between concentration and profits was generated by efficiency: The most efficient firms were the most profitable and the fastest growing, so profits and concentration increased together. See Harold Demsetz, ‘Industry Structure, Market Rivalry, and Public Policy’ (1973) 16 Journal of Law and Economics 1; Sam Peltzman, ‘The Gains and Losses from Industrial Concentration’ (1977) 20 Journal of Law and Economics 229. 44 ‘Too Much of a Good Thing’ The Economist (24 March 2016). The Economist reported observations for 893 industries, which implies data for six-digit industries. The online version of the article originally had interactive graphics, but the interactions no longer function; see www​.economist​.com/​ news/​briefing/​21695385​-profits​-are​-too​-high​-america​-needs​-giant​-dose​-competition​-too​-much​-good​ -thing, accessed 8 April 2021.

50  Research handbook on abuse of dominance and monopolization apt to be hundreds of times broader than properly delineated markets, and an industry with a CR4 of, say, 40 actually is considered unconcentrated.45 Subsequent studies using six-digit Census data have found modest increases in concentration. Sharat Ganapati found that the average CR4 increased about five points between 1997 and 2012.46 Robert D. Atkinson and Filioe Lage de Sousa found that the average CR4 increased by just one point between 2002 and 2017.47 Finally, Mary Amiti and Sebastian Heise addressed the failure to include firms that sell in the US but manufacture elsewhere. They merged Census shipments data with import-export data, then examined trends in the CR4, CR20 and HHI for five-digit NAICS manufacturing industries. Between 1992 and 2012, they found that the average level of concentration was stable.48 The Economist identified three major industries in what it called the ‘oligopolistic corner of the economy’, with CR4 greater than 66.7 – ‘telecoms, pharmacies, and credit cards’. Telecoms was an industry with better data because the Federal Communications Commission tracked mobile wireless concentration in local markets, and those data exhibited just a slight upward trend in average concentration.49 Census concentration data for pharmacies were highly problematic because they omitted more than 10,000 pharmacies operating within larger stores (typically supermarkets) and because the data were national although the relevant markets were local. Credit card processing has been highly concentrated for the entire history of the industry. Shortly after publication of The Economist story, the CEA published the issue brief mentioned in the introduction. The CEA performed much the same analysis as The Economist but at the level of NAICS sectors,50 and the CEA examined CR50 rather than CR4.51 The CEA found the largest concentration increases in the transportation and warehousing sector and in the retail trade sector, but those increases are meaningless. To presume that the individual markets in a sector exhibit the same trends as the sector as a whole is Aristotle’s fallacy of division. These two sectors contain 123 six-digit NAICS industries, which generally are much broader than relevant markets. Important retail trade The HHI almost certainly would be below 1000, whereas any market with an HHI below 1500 is considered unconcentrated under the US Horizontal Merger Guidelines. 46 Sharat Ganapati, ‘Growing Oligopolies, Prices, Output, and Productivity’ (2021) 13 American Economic Journal: Microeconomics 309. 47 Robert D Atkinson and Filioe Lage de Sousa, ‘No, Monopoly Has Not Grown’ (Information Technology & Innovation Foundation, June 2021), https://​itif​.org/​sites/​default/​files/​2021​-no​-monopoly​ -has​-not​-grown​.pdf, accessed 1 July 2021. 48 Mary Amiti and Sebastian Heise, ‘U.S. Market Concentration and Import Competition’ (Federal Reserve Bank of New York Staff Report, May 2021), https://​www​.newyorkfed​.org/​medialibrary/​media/​ research/​staff​_reports/​sr968​.pdf, accessed 29 July 2021. 49 Federal Communications Commission, ‘Annual Report and Analysis of Competitive Market Conditions with Respect to Mobile Wireless, Including Commercial Mobile Services’ (2016) 18, https://​ docs​.fcc​.gov/​public/​attachments/​DA​-16​-0161A1​.pdf, accessed 9 April 2021. 50 CEA chairman Jason Furman previously had done a similar analysis, but it was published subsequently: Jason Furman and Peter Orszag, ‘A Firm-Level Perspective on the Role of Rents in the Rise in Inequality’ in Martin Guzman (ed), Toward a Just Society: Joseph Stiglitz and Twenty-First Century Economics (Columbia University Press 2018). 51 The usual CR4 was wrong for the sector level: Suppose a sector consisted of 50 markets of comparable size and no firm operated in more than one market. The sector CR50 then would approximate the weighted average CR1 across the markets, while the sector CR4 would approximate the maximum CR1 across the markets. 45

Concentration and rising market power: fears and facts  51 industries – such as automobile dealers, gasoline stations and grocery stores – have local markets. In long-distance transportation industries, like airlines, origin-destination pairs are distinct markets, and in short-distance transportation industries, like taxis, the markets are local. Concentration trends are meaningless when the data are aggregated over hundreds of products and hundreds of locations, as the CEA later acknowledged.52 The most widely noted claim of increasing concentration in an academic journal is that of Gustavo Grullon, Yelena Larkin and Roni Michaely, who claimed that ‘industry concentration over the last two decades has markedly increased’.53 Their work is part of the finance literature, although they presented it to competition policy audiences, notably at the Stigler Center conference ‘Is There a Concentration Problem in America?’54 As is typical in finance,55 Grullon, Larkin and Michaely used firm-based data for traded companies, in particular the Compustat data. They sorted the companies into NAICS subsectors, then computed HHIs. Thus measured, concentration increased in 80 per cent of the subsectors, with a median increase of 41 per cent and a mean increase of 90 per cent. The claim of increasing concentration was for the period 1997–2014, even though the study went back to 1972. Grullon, Larkin and Michaely acknowledged problems with their firm-based data and found reassurance in the fact that the distribution of HHI increases in their data is much the same in Census manufacturing subsectors. But research discussed above found that concentration increased far less in the Census data at the six-digit level. Grullon, Larkin and Michaely referred to ‘market power’ throughout their article and purported to find evidence that increasing concentration enhanced or created market power, but they did not look for increases in concentration that plausibly enhanced or created market power, that is, increases in concentrated industries or that caused industries to become concentrated. Moreover, an NAICS subsector is apt to aggregate more than 1,000 distinct markets, so subsector concentration trends can shed no light on market concentration trends or on market power. Trends in concentration outside the US have gotten less attention. Two studies compared sector-level concentration trends in Europe to those in the US. Researchers at the OECD included more and different European countries but examined a shorter time period than

Due to excessive aggregation at the product and geographic levels, the CEA commented that the data presented in the issue brief ‘shed little light on the state of competition’. Council of Economic Advisers, Economic Report of the President (USGPO 2020) 210. Jason Furman argued in 2018 that aggregated data were useful despite all the flaws. Jason Furman, ‘Antitrust in a Changing Economy and Changing Economics’ (ProMarket, 30 November 2018), https://​promarket​.org/​antitrust​-in​-a​-changing​ -economy​-and​-changing​-economics/​, accessed 12 April 2021. 53 Gustavo Grullon, Yelena Larkin and Roni Michaely, ‘Are US Industries Becoming More Concentrated?’ (2019) 23 Review of Finance 697, 700. Other academic articles presented data indicating increasing concentration. See, eg David Autor and others, ‘Concentrating on the Fall of the Labor Share’ (2017) 107 American Economic Review (Papers and Proceedings) 180; Simcha Barkai, ‘Declining Labor and Capital Shares’ (2020) 75 Journal of Finance 2421. 54 Stigler Center for the Study of the Economy and the State, University of Chicago, Booth School of Business, ‘Is There a Concentration Problem in America?’ (record of conference, 27−29 March 2017), https://​promarket​.org/​wp​-content/​uploads/​2018/​04/​Is​-There​-a​-Concentration​-Problem​-in​-America​.pdf, accessed 8 April 2021. 55 See Jan Keil, ‘The Trouble with Approximating Industry Concentration from Compustat’ (2017) 45 Journal of Corporate Finance 467. 52

52  Research handbook on abuse of dominance and monopolization researchers at the European Central Bank.56 The OECD researchers found no concentration trend in Europe, but the European Central Bank researchers found a significant upward trend. Of course, sector-level data cannot reveal market-level concentration trends, and both studies relied on firm-based data, which can bias the trends. The UK data look very much like the US data. The Competition and Markets Authority (CMA) found sector-level increases in concentration (measured by CR10) like those reported by the CEA issue brief (using CR50).57 The CMA also examined concentration at the four-digit level of the UK SIC system (comparable to the NAICS five-digit level), while acknowledging that the product dimensions of relevant markets in CMA cases are much narrower.58 The CMA found that average concentration (CR10) increased somewhat from 1998 to 2012, just as Sharat Ganapati found for the US. The CMA also found that concentration declined after 2012, but not to its prior level.59 B.

How Industry Dynamics Affect Measured Concentration Trends

Measurement of concentration levels and trends can be highly misleading when the data are not at the market level. Imagine a world in which national champions compete in global markets, but concentration is measured at the national level. Further imagine that competing national champions merge from time to time. In this world, the use of national data when markets are global overstates the level of market concentration and understates the trend in market concentration. The foregoing illustrates one possible error in measuring levels and trends in economic concentration. A far more common error, however, is excessive aggregation. Excessive aggregation tends to impart a downward bias in the level of market concentration and an upward bias in the trend of market concentration. The former bias was explained above. To appreciate how the latter bias arises, consider the following scenarios in which concentration is measured 56 The OECD examined 2006–15 and included companies from Austria, Belgium, Denmark, Germany, Finland, France, Hungary, Norway, Portugal and Sweden. Matej Bajgar and others, ‘Industry Concentration in Europe and North American’ (OECD Productivity Working Papers, January 2019), www​.oecd​-ilibrary​.org/​economics/​industry​-concentration​-in​-europe​-and​-north​-america​_2ff98246​-en, accessed 17 April 2021. The European Central Bank examined 2000–14 and included companies from Germany, France, Italy and Spain. Maria Chiara Cavalleri and others, ‘Concentration, Market Power and Dynamism in the Euro Area’ (European Central Bank Discussion Paper, March 2019), www​.ecb​.europa​ .eu/​pub/​pdf/​scpwps/​ecb​.wp2253​~cf7b9d7539​.en​.pdf, accessed 17 April 2021. A less aggregated study of just Germany found no trend. Monopolkommission, ‘Trends in Indicators of Marker Power in Germany and Europe’ (2018), www​.monopolkommission​.de/​images/​HG22/​Main​_Report​_XXII​_Market​_Power​ .pdf, accessed 29 July 2021. 57 Competition and Markets Authority, ‘The State of UK Competition’ (30 November 2020) 28–30, https://​assets​.publishing​.service​.gov​.uk/​government/​uploads/​system/​uploads/​attachment​_data/​ file/​939636/​State​_of​_Competition​_Report​_Nov​_2020​_Final​.pdf, accessed 29 July 2021. 58 Competition and Markets Authority, ‘The State of UK Competition Annexes’ (November 2020) 2, 4, https://​assets​.publishing​.service​.gov​.uk/​government/​uploads/​system/​uploads/​attachment​_data/​file/​ 939639/​State​_of​_Competition​_Report​_Nov​_2020​_​-​_Annexes​.pdf, accessed 29 July 2021. 59 CMA (n 57) 27. Similar findings were reported by Torsten Bell and Dan Tomlinson, ‘Is Everybody Concentrating? Recent Trends in Product and Labour Market Concentration in the UK’ (Resolution Foundation, July 2018), www​.​resolution​foundation​.org/​app/​uploads/​2018/​07/​Is​-everybody​ -concentrating​_Recent​-trends​-in​-product​-and​-labour​-market​-concentration​-in​-the​-UK​-​.pdf, accessed 29 July 2021.

Concentration and rising market power: fears and facts  53 at the industry level but an industry consists of five distinct markets. In each scenario, concentration is measured in the years 2000 and 2020. First, posit an industry that experienced a huge merger wave in the intervening 20 years: The 2000 data indicate that the industry had 25 equal-sized firms, and the 2020 data indicate that it had just five equal-sized firms. The industry HHI increased from 400 to 2000, but this tells us nothing about the trend in market concentration. The data make all of the mergers appear to have been horizontal. But if they all were horizontal, the measured concentration trend conceals a competition catastrophe, as each market became a monopoly. And if none of the mergers were horizontal, market concentration did not increase at all. In this example, industry-level data simply are uninformative about the trends in market-level concentration. Second, posit an industry that experienced no mergers, no entry and no exit. Assume five, equal-sized markets in 2000 – four markets with five equal-sized firms and one market with a single firm. Suppose that the structure of all markets was the same in 2020, but the single-firm market quadrupled in size between 2000 and 2020, while the other markets experienced no growth. The industry HHI increased from 720 to 2625, and the industry share of the largest firm increased from 20 to 50 per cent, but the true market shares did not change. In this example, industry-level data indicate a sharp upward trend in concentration, but market-level concentration is stable. Third, suppose the composition of the industry in 2000 was exactly as in the preceding scenario and again assume that the size of the single-firm market quadrupled by 2020. Further suppose that each of the five-firm markets were entered by five new firms that grew to the size of the incumbents by 2020. And suppose that the single-firm market was entered by one new firm that also achieved the size of the incumbent by the year 2020. In every market, the true HHI in 2020 was half what it was in 2000, but the industry HHI increased from 720 to 1313. In this example, industry-level data indicate a sharp upward trend in concentration, but market-level concentration sharply decreased in every market. The fact that entry eliminated a monopoly went unnoticed because no one was looking. C.

How Firm Dynamics Affect Measured Concentration Trends

Measuring concentration at the industry or sector level can cause data to exhibit concentration trends that do not exist at the market level, and using firm-based data can heighten the illusion. The use of firm-based data does not have this effect if firms are no broader in scope than the industries or sectors used to compile the data, but the use of firm-level data can exaggerate bias from excessive aggregation when firms operate in multiple industries or sectors. Consider the impact of the merger of two multi-industry firms assigned to different industries on the basis of the plurality of their operations, and assume that the successor firm is assigned to the same industry as the larger merging firm. Following the merger, the smaller merging firm disappears from the data, causing an increase in measured concentration for its assigned industry.60 The turnover of the smaller merging firm is credited to the larger merging firm, causing an increase in measured concentration for its assigned industry.61 In both affected 60 Observed concentration could fall if the smaller merging firm was the largest firm in its assigned market. 61 Observed concentration could fall if the larger merging firm was a small firm in its assigned market.

54  Research handbook on abuse of dominance and monopolization industries, however, measured concentration increases only because firm-level data is used to measure it.62 To illustrate the potential magnitude of distortion of concentration trends, consider an economy with ten industries corresponding to markets. In the year 2000, suppose that all firms were the same size, that each firm operated in a single industry/market, and that each industry had five firms. Now suppose that the only change over time was the birth of a conglomerate formed by merging two firms from one industry and one firm from each of the other nine industries. The actual HHI in the first industry increased from 2000 to 2800, but the measured HHI increased to 6327 because the conglomerate’s entire operations were assigned to the industry. In the other nine industries, acquired firms disappeared from the data, so the measured HHI increased to 2500, although the actual HHI remained at 2000. The weighted average measured HHI for the economy increased a whopping 79 per cent, although the weighted average actual market HHI increased just 4 per cent. The finance literature has recognized that Census concentration measures are not highly correlated with Compustat-based concentration measures but has not identified the all the reasons.63 The finance literature has identified the problem that Compustat data omits some firms: Compustat includes only companies traded on US exchanges, but some large US companies are not traded, and some companies doing business in the US are based elsewhere and not traded in the US.64 D.

Available Evidence on Market Concentration

The available evidence on market-level concentration trends is hardly comprehensive, but it does undermine the foundations of claims of increasing concentration in the US. Researchers can access market-level data for a few US industries that are subject to special government oversight, notably airlines and banking. Despite numerous mergers that increased concentration at the national level, studies consistently find no increases in average market concentration. Economists at the Federal Reserve Board found that more than 10,000 mergers between 1980 and 2010 did not increase average concentration in local banking markets.65 Over roughly the same time period, academic and government economists found slight decreases in average concentration for airline city-pair markets.66 62 The successor firm could be assigned to an industry other than that to which merging firms had been assigned, in which case the merger would increase observed concentration in three industries. 63 See Keil (n 55); Ashiq Ali, Sandy Klasa and Eric Yeung, ‘The Limitations of Industry Concentration Measures Constructed with Compustat Data: Implications for Finance Research’ (2009) 22 Review of Financial Studies 3839. 64 ibid. The largest US non-traded companies are Cargill, Chrysler, Koch Industries and Mars. Many companies based outside the US cross-list their shares on a US exchange, but some foreign companies with large US sales do not, including BMW, Hyundai and Samsung. 65 Stephen A Rhoades, ‘Bank Mergers and Banking Structure in the United States, 1980–98’ (Board of Governors of the Federal Reserve System, August 2000), www​.federalreserve​.gov/​pubs/​staffstudies/​ 2000​-present/​ss174​.pdf, accessed 4 May 2021; Robert M Adams, ‘Consolidation and Merger Activity in the United States Banking Industry from 2000 through 2010’ (FEDS Working Paper No 2012-51, 8 August 2012), https://​ssrn​.com/​abstract​=​2193886, accessed 4 May 2021. 66 Severin Borenstein, ‘The Evolution of U.S. Airline Competition’ (1992) 6 Journal of Economic Perspectives 45; Kai Hüschelrath and Kathrin Müller, ‘Low Cost Carriers and the Evolution of the Domestic U.S. Airline Industry’ (2012) 13 Competition and Regulation in Network Industries 133; US

Concentration and rising market power: fears and facts  55 An analysis of the private National Establishment Time Series for the US found declining concentration at an extremely disaggregated level. Esteban Rossi-Hansberg, Pierre-Daniel Sarte and Nicholas Trachter measured concentration at the postal code and eight-digit SIC level.67 They found pronounced concentration decreases for sectors in which the markets tend to be local, even if not as narrow as postal codes – finance, insurance and real estate; retail trade; and services. C. Lanier Benkard, Ali Yurukoglu and Anthony Lee Zhang used consumer surveys to construct concentration measures at levels approximating markets. They found slightly declining average concentration from 1994 to 2019, with the sharpest decline in the most concentrated markets.68 A study of market-level concentration in Europe undermines the foundations of denials of increasing concentration in Europe. Pauline Affeldt and others relied on the markets defined in European Commission merger decisions and constructed lower bounds for the post-merger HHI when the decisions provided sufficient information. Although Commission merger decisions provide a non-random sample, concentration in the sampled markets increased significantly from 1995 to 2014, especially in the service sector.69

IV.

EVALUATING RECENT CLAIMS ON DYNAMISM AND MARGINS

A.

The Dynamism of the Economy

Increasing concentration was not the only indicator of declining competition cited by the 2016 CEA issue brief. It also cited decreased dynamism of the economy and increased corporate margins. On dynamism the issue brief presented a graph depicting the entry and exit rates in the Business Dynamics Statistics (BDS) compiled by the US Census Bureau. The data indicated that the entry and exit rates both steadily declined from the inception of the program in 1978 to 2013, the most recent year available to the CEA.70 Researchers who drew attention to these declines had cited them as evidence of diminishing vigor in the process of creative

Government Accountability Office, ‘Airline Competition: The Average Number of Competitors Serving the Majority of Passengers Has Changed Little in Recent Years, but Stakeholders Voice Concerns about Competition’ (June 2014). 67 Esteban Rossi-Hansberg, Pierre-Daniel Sarte and Nicholas Trachter, ‘Diverging Trends in National and Local Concentration’ (2001) 35 NBER Macroeconomics Annual 115. The seventh and eighth digits are not part of the SIC system as developed by the Census. 68 C Lanier Benkard, Ali Yurukoglu and Anthony Lee Zhang, ‘Concentration in Product Markets’ (NBER Working Paper, April 2021), www​.nber​.org/​system/​files/​working​_papers/​w28745/​w28745​.pdf, accessed 4 May 2021. 69 Pauline Affeldt and others, ‘Market Concentration in Europe: Evidence from Antitrust Markets’ (DIW Berlin Working Paper, January 2021), https://​papers​.ssrn​.com/​ abstract_id=3775524, accessed 4 May 2021. 70 Issue Brief (n 14) 5. A decline in the entry rate also was observed in Orbis data (provided by Bureau Van Dijk), which includes 28 countries. See Ufuk Akcigit and others, ‘Rising Corporate Market Power: Emerging Policy Issues’ (IMF Staff Note, March 2021), www​.imf​.org/​en/​Publications/​Staff​ -Discussion​-Notes/​Issues/​2021/​03/​10/​Rising​-Corporate​-Market​-Power​-Emerging​-Policy​-Issues​-48619, accessed 29 July 2021.

56  Research handbook on abuse of dominance and monopolization destruction that drives economic growth.71 The declines did not continue after 2013, and they appear to reflect a shift in the nature of startups. The Census BDS program tracks businesses with at least one employee, and as of 2018 the US had 5.3 million such businesses. The Nonemployer Statistics compiled by the Census72 indicate that, as of 2018, the US had 26.5 million nonemployer businesses with turnover of $1,293 billion. During 1997–2018, the number of nonemployers increased 71.5 per cent, while the number of employers increased just 10.5 per cent. Both the entry and exit rates for nonemployers are far higher than those for employers.73 During 1997–2010 nonemployers accounted for over 90 per cent of the entry.74 Combining the BDS data with the Nonemployer Statistics, Pedro Bento and Diego Restuccia found that business dynamism was responsible for more than half of US productivity growth between 1982 and 2014.75 B.

Corporate Margins

Since Abba P. Lerner’s 1934 article,76 economists have looked to price–cost margins as an indicator of exercised market power. The price–cost margin for a product is its price minus its marginal cost, all divided by its price. The price–cost margin is bounded by one from above, and it is zero when price equals marginal cost under perfect competition. In some oligopoly models, price–cost margins are closely related to market concentration.77 Direct measurement of price–cost margins can be difficult because marginal costs can be poorly reflected in accounting data, but nonpublic, product-level data are routinely used to measure price–cost margins in competition cases. See, eg Ryan A Decker and others, ‘Declining Business Dynamism: What We Know and the Way Forward’ (2016) 106 American Economic Review (Papers and Proceedings) 203. Concern about declining dynamism is widespread. See Flavio Calvino, Chiara Criscuolo and Rudy Verlhac, ‘Declining Business Dynamism: Structural and Policy Determinants’ (OECD Science, Technology and Innovation Papers, Nov 2020), www​.oecd​-ilibrary​.org/​science​-and​-technology/​declining​-business​ -dynamism​_77b92072​-en, accessed 25 June 2021. The entry rate declines are much the same for the US and Europe; ibid 61. 72 See US Census Bureau, ‘Nonemployer Statistics’, www​.census​.gov/​programs​-surveys/​ nonemployer​-statistics​.html, accessed 15 April 2021. The Census provides limited data combining employers and nonemployers. See US Census Bureau, 2018 CBP and NES Combined Report, www​ .census​.gov/​data/​tables/​2018/​econ/​cbp/​2018​-combined​-report​.html, accessed 15 April 2021. 73 See Steven J Davis and others, ‘Measuring the Dynamics of Young and Small Businesses: Integrating the Employer and Nonemployer Universes’ in Timothy Dunne, J Bradford Jensen and Mark J Roberts (eds), Producer Dynamics: New Evidence from Micro Data (University of Chicago Press 2009) 329. 74 See Robert Fairlie, Javier Miranda and Nikolas Zolas, ‘The Integrated Longitudinal Business Database: Measuring Job Creation among the Universe of Business Entities’ (slide presentation, 14 October 2016), www​.nationalacademies​.org/​event/​10​-14​-2019/​docs/​D2AC​137B733A75​7F2D5940AE​ 5AFB555ED8​5979DD272E, accessed 15 April 2021. 75 See Pedro Bento and Diego Restuccia, ‘The Role of Nonemployers in Business Dynamism and Aggregate Productivity’ (University of Toronto Working Paper 19 January 2021), www​.economics​. utoronto​.ca/​public/​workingPapers/​tecipa​-686​.pdf, accessed 7 May 2021. 76 AP Lerner, ‘The Concept of Monopoly and the Measurement of Monopoly Power’ (1934) 1 Review of Economic Studies 157. 77 Firms in the Cournot model noncooperatively choose the quantities they sell into the market. In equilibrium the weighted average price–cost margin for all firms in the market equals the market HHI divided by the market elasticity of demand. See, eg Scherer and Ross (n 32) 200. 71

Concentration and rising market power: fears and facts  57 Markups are closely related to margins. A product’s markup is simply the ratio of its price to its marginal cost. The markup is one under perfect competition and unbounded from above. With simple production technology (a single output and homogeneous inputs), the economics of cost minimization allows the inference of markups from firm-level accounting data,78 and recent studies have taken this approach to track margins over time. Jan De Loecker, Jan Eeckout and Gabriel Unger found that the weighted average markup of traded US companies increased from 1.21 in 1980 to 1.61 in 2016. Their detailed findings are more revealing, however: They found that the median markup was constant, and the increase in the weighted average markup was due to the growth (and increasing weight) of high-margin firms. When De Loecker, Eeckout and Unger took out the effects of entry, exit, and different rates of grown, the average markup rose from 1.21 in 1980 to about 1.3 in 2000 and then fell. Thus, their data showed that markups were quite stable, except for those of the highest-markup firms. The average markup of the top 10 per cent of firms, ranked by markup, rose from 1.5 to 2.5.79 De Loecker, Eeckout and Unger indicated that the markup trends are much the same in other parts of the world, and studies using the same methods confirm that for Europe.80 David Autor and others used methods similar and made similar findings in trying to explain why the OECD countries have experienced a decline in the share of national income going to labor.81 They proposed and tested a ‘superstar’ model, in which network effects and other comparable forces made a small number of firms much more successful than their rivals. The authors tested their model with data for 1982–2012, assigning firms to industries at roughly the SIC four-digit level of aggregation.82 Autor and others found that the weighted average markup increased from 1.2 in 1982 to 1.8 in 2012, and they found that ‘Rising aggregate (weighted-average) markups are driven by the changing market shares and markups of the largest firms …’. A few superstar tech giants can skew the aggregate data but cannot demonstrate a general decline in competition. In any event, the foregoing markups are not high. Markups of 1.61 and

78 See Susanto Basu, ‘Are Price-Cost Markups Rising in the United States? A Discussion of the Evidence’ (2019) 33 Journal of Economic Perspectives 3. The reliability of these methods has been questioned. See Steve Bond and others, ‘Some Unpleasant Markup Arithmetic: Production Function Elasticities and their Estimation from Production Data’ (2021) 121 Journal of Monetary Economics 1; Devesh Raval, ‘Testing the Production Approach to Markup Estimation’ (21 October 2020), https://​ papers​.ssrn​.com/​abstract​_id​=​3324849, accessed 27 June 2021. 79 Jan De Loecker, Jan Eeckout and Gabriel Unger, ‘The Rise of Market Power and the Macroeconomic Implications’ (2020) 135 Quarterly Journal of Economics 561. 80 See Ufuk Akcigit and others, ‘Rising Corporate Market Power: Emerging Policy Issues’ (IMF Staff Note, March 2021), www​.imf​.org/​en/​Publications/​Staff​-Discussion​-Notes/​Issues/​2021/​03/​10/​ Rising​-Corporate​-Market​-Power​-Emerging​-Policy​-Issues​-48619, accessed 29 July 2021; Tommaso Aquilante and others, ‘Market Power and Monetary Policy’ (Bank of England Staff Paper, May 2019), www​.bankofengland​.co​.uk/​-/​media/​boe/​files/​working​-paper/​2019/​market​-power​-and​-monetary​-policy​ .pdf, accessed 29 July 2021. 81 David Autor and others, ‘The Fall of the Labor Share and the Rise of Superstar Firms’ (2020) 135 Quarterly Journal of Economics 645. 82 The authors observed that measured concentration increased and found that rising concentration correlated with the decline in labor’s share of national income. But they measured concentration at too high a level of aggregation and with firm-based data, so they had no reliable evidence of a decline in competition.

58  Research handbook on abuse of dominance and monopolization 1.8 are equivalent to price–cost margins of 37.9 and 44.4 per cent,83 which are lower than what typically are observed in competition cases. The markup of 2.5 is equivalent to a price–cost margin of 60 per cent, which is in the range of margins observed in many old economy industries, while new economy industries tend to have higher margins. While the foregoing studies were circulated as working papers, industrial organization economists Steven Berry, Martin Gaynor and Fiona Scott Morton reviewed them and other evidence, concluding: ‘In our opinion, both industry studies and accounting data studies point to the broad category of endogenously increasing fixed and sunk costs as an important, perhaps the most important, source of the apparent pattern of rising global markups’.84 Rapidly growing new economy industries have higher fixed and sunk costs than old economy industries, so higher margins are required to yield competitive returns. Claims of increasing margins elicited a call for stricter enforcement in unilateral effects merger cases.85 But that reaction overlooked two key points: First, unilateral effects analysis already accounted for margins.86 Second, higher margins have no consistent effect on unilateral anticompetitive effects.87 Working through a standard model of competition used to predict unilateral effects also revealed that margins can increase for reasons other than increasing concentration. In particular, stronger brand preference increases margins, all other things being equal.

V. CONCLUSIONS Recent claims of increasing concentration sounded a false alarm. The claims were not based on measurements of concentration at the market level, although that is the level at which competition takes place, and the available evidence for the US does not indicate increasing market-level concentration. Increasing concentration has been less of a concern in Europe, but there is some evidence of increasing market-level concentration in Europe. Recent claims of increasing concentration of the US economy rely on excessively aggregated data. Excessive aggregation imparts a substantial downward bias in market concentration levels and can impart a substantial upward bias in market concentration trends. The finance literature’s reliance on firm-based data exaggerates the upward bias in concentration trends seen in excessively aggregated data.

It is not possible to translate average markups into average margins because markups and margins have a non-linear relationship. Consider a two-firm economy with an average markup of 1.6. If the individual firms’ markups were 1.5 and 1.7, the average margin would be 37.3 per cent, but if the markups were 1.1 and 2.1, the average margin would be 30.7 per cent. 84 Steven Berry, Martin Gaynor and Fiona Scott Morton, ‘Do Increasing Markups Matter? Lessons from Empirical Industrial Organization’ (2019) 33 Journal of Economic Perspectives 44, 56. 85 Tommaso Valletti and Hans Zinger, ‘Should Profit Margins Play a More Decisive Role in Horizontal Merger Control? A Rejoinder to Jorge Padilla’ (2018) 9 Journal of European Competition Law and Practice 336. 86 See Gregory J Werden and Luke M Froeb, ‘Unilateral Competitive Effects of Horizontal Mergers’ in Paolo Buccirossi (ed), Handbook of Antitrust Economics (MIT Press 2008). 87 See Gregory J Werden and Luke M Froeb, ‘Increased Margins and Merger Assessment: No Need to Fret’ (2018) 9 Journal of European Competition Law and Practice 519. 83

Concentration and rising market power: fears and facts  59 Census data indicate that the US economy became a bit more concentrated at the industry and sector levels,88 which can arise from what older literature called market-extension and product-extension mergers. Such mergers combine firms in the same industry or sector but not in the same markets. These mergers do not affect market concentration but can generate synergies by creating regional or national chains or by expanding product lines. For all the concern about the decline of the US economy, it did not decline. As gauged by total factor productivity (TFP), the US economy performed better than other large industrialized countries between 1989 and 2019. TFP increased 24.7 per cent in the US, as compared with 22.2 per cent in Germany, 18.0 per cent in the UK, 17.7 per cent in Sweden, 11.3 per cent in Canada, 10.0 per cent in France, 5.1 per cent in Japan and −15.4 per cent in Italy.89 Historian Richard Hofstadter lamented in 1964 that: ‘The antitrust movement is one of the faded passions of American reform’.90 He regretted that antitrust became more technocratic than democratic when it became important. He preferred the old populism, and a half-century after his death, the old populism is threatening to make a comeback. The facts about increasing concentration might not matter as much as the fears. In July 2021 President Biden issued an executive order declaring that: ‘We must act now to reverse these dangerous trends, which constrain the growth and dynamism of our economy, impair the creation of high-quality jobs, and threaten America’s economic standing in the world’. Among those trends was that ‘over the last several decades, as industries have consolidated, competition has weakened in too many markets’.91 If market concentration were increasing, that still would not be a reason to revamp competition enforcement policy. Market concentration naturally trends up, and the trend can be accelerated by innovation, which drives economic growth. When firms grow because they provide great value to consumers, a sound competition policy does not begrudge their well-earned success, no matter how much market concentration increases. Stopping large horizontal mergers makes sense, but that is a pillar of existing competition policy. Finally, improving competition enforcement is more difficult than is supposed. Beyond increasing funding for the competition agencies, meaningful reform is not easy to implement, and most reforms could do more harm than good. If there is a serious systemic problem in US antitrust enforcement, it is that the excessive skepticism of judges prevents meritorious cases from succeeding in court, and that problem defies easy solutions.

Nevertheless, aggregate economic concentration of the US economy did not increase. See Lawrence J White, ‘Trends in Aggregate Concentration in the United States’ (2002) 16 Journal of Economic Perspectives 137; Lawrence J White and Jasper Yang, ‘What Has Been Happening to Aggregate Concentration in the US Economy in the Twenty-First Century?’ (2020) 38 Contemporary Economic Policy 483. 89 The raw data are from Penn World Table 10.0, maintained by the Groningen Growth and Development Centre, www​.rug​.nl/​ggdc/​productivity/​pwt/​?lang​=​en, accessed 12 May 2021. 90 Richard Hofstadter, ‘What Happened to the Antitrust Movement?’ in The Paranoid Style in American Politics and Other Essays (Alfred A. Knopf 1965). 91 ‘Promoting Competition in the American Economy’ (Executive Order 14036 of 9 July 2021) (2021) 86 Federal Register 36987, 36987. 88

4. Understanding market power: a legal perspective Nicolas Petit

I. INTRODUCTION Contemporary antitrust laws involve the containment of market power. In the United States (US) and the European Union (EU), the growth of antitrust as a system of market power control is the result of a slow developmental process of law in action, shaped by judicial interpretation, administrative intervention, and private adjudication.1 Market power has received different understandings in US and EU antitrust law. The chapter studies, and compares, the definition (II), proof (III), and evidence (IV) of market power in the US and the EU. The discussion relies on the economics of market power discussed in Chapter 2. The four main conceptions of market power found in the literature are used to guide the legal analysis. A study of how antitrust systems consider market power is useful. In the course of its history, antitrust law has been repeatedly subject to contestations suggesting it does either too little or too much. A comparative understanding of where US and EU laws set boundaries to the market power concept can help clarify how much each antitrust law does.2

II.

DEFINITIONS OF MARKET POWER

A.

Economic Interpretation of Market Power under US Law

The definition of market power used in US antitrust law fits with an economic understanding of the concept. Two conceptions dominate the cases. The first is a view of market power centred on the ability to raise prices above the competitive level.3 In Fortner, a Sections 1 and 2 case about US Steel’s tying of loans to the purchase of prefabricated houses, the Court For a comprehensive discussion of the history of market power in US and EU law, see N Petit, ‘Understanding Market Power’, EUI RSC, 2022/14, http://​hdl​.handle​.net/​1814/​74347. 2 While the focus of the chapter is on single firm conduct, occasional reference is made to coordinated conduct law. When courts consider market power, they do not necessarily change their mindset depending on the particular area of antitrust law that is mobilized. Moreover, often, antitrust cases have formally or informally been tried as both coordinated and single firm conduct violations. 3 In NCAA v Board of Regents, the Court held that ‘[m]arket power is the ability to raise prices above those that would be charged in a competitive market’. The Court referred to its prior case law in Jefferson Parish, but this reference is inconclusive, because in this case, the Court decided to condemn another form of economic power; NCAA v Board of Regents 468 US 85 (1984); Jefferson Parish Hospital Dist No 2 v Hyde 466 US 2 (1984). In Copperweld, the Court, at footnote 9, talked of ‘the ability to raise prices above those that would be charged in a competitive market’; Copperweld Corp v Independence Tube Corp 467 US 752 (1984). 1

60

Understanding market power: a legal perspective  61 described its understanding of market power in terms of the falling demand curve introduced by Cournot as follows: Market power is usually stated to be the ability of a single seller to raise price and restrict output, for reduced output is the almost inevitable result of higher prices. Even a complete monopolist can seldom raise his price without losing some sales; many buyers will cease to buy the product, or buy less, as the price rises.4

But ‘Cournotism’ was already present in older opinions whose wording showed interest towards the price.5 Many cases concerned themselves with problems arising from situations of monopolies that ‘fix’ price6 or ‘control’ price.7 In International Harvester, the Court wrote that influence over price defines power. The combination, said the Court, ‘controlled or dominated the harvesting machinery industry by the compulsory regulation of prices’.8 Later opinions added that the power to raise prices was a problem compared to the situation that could be expected in a competitive market.9 The second conception gives more weight to power over output. Power over price still matters. But it is looked at as an implication of control over output, which becomes the decisive element. Standard Oil gives an illustration. The case saw the Court dismiss charges of unlawful monopoly against a patent pooling, cross-licensing, and royalty-sharing scheme over oil cracking technology. The Court considered that short of ‘control of supply’ of cracked gasoline – the members of the pool enjoyed 55 per cent of oil cracking capacity – no power to fix price could ever result from a licensing agreement, regardless of its sophistication. The Court suggested that for such power to arise, the parties should have controlled ‘through some means, the total gasoline production from all sources’. Learned Hand repeated the same idea a few years later in Alcoa: ‘It is only when a restriction of production inevitably affects prices, or is intended to do so, that it violates S1 of the Act.’10

4 See Fortner Enterprises, Inc v United States Steel Corp 394 US 495 (1969). The cited definition of marked power can be retrieved at ibid 503. 5 We borrow this expression from JR Hicks, ‘Annual Survey of Economic Theory: The Theory of Monopoly’ (1935) 3 Econometrica 1, 9. 6 Appalachian Coals, Inc v US 288 US 344 (1933). 7 United States v Trenton Potteries 273 US 392 (1927). 8 United States v Int Harvester Co 274 US 693 (1927). The Court wrote: ‘in fact controlled or dominated the harvesting machinery industry by the compulsory regulation of prices’; ibid 708. 9 See Eastman Kodak Co v Image Technical Services, Inc 504 US 451 (1992). The Court defines market power as the power ‘to force a purchaser to do something that he would not do in a competitive market’. See also in the lower courts, for example, in LePage’s v 3M, where the Court wrote: ‘Once a monopolist achieves its goal by excluding potential competitors, it can then increase the price of its product to the point at which it will maximize its profit. This price is invariably higher than the price determined in a competitive market. That is one of the principal reasons why monopolization violates the antitrust laws.’ See LePage’s Inc v 3M 324 F3d 141 (3d Cir 2003) 169. 10 United States v Aluminum Co of America 148 F2d 416 (2d Cir 1945).

62  Research handbook on abuse of dominance and monopolization In contrast, few antitrust cases endorse the neoclassical idea that deviations from marginal cost pricing constitute the definitional touchstone of market power.11 Similarly, the case law at best occasionally refers to a non-technical idea of market power as one of ‘freedom’ of trade.12 Two additional details of considerable significance allow a claim that US jurisprudence endorses a modern conception of market power. First, the Court considers that power over price is the dependent variable of power over output. To see this, it is necessary to realize that many old cases used to define market power in terms of power over price or control over output. In Appalachian Coals, for example, the Court talked of establishing ‘monopoly control of a market or power to fix monopoly prices’.13 Hay has underlined the economic problem of this approach as follows: The definition suggests that there are two different possible tests for monopoly power and that a firm will have monopoly power if either is satisfied. For the economist, however, the power to control prices in any meaningful way depends on the absence of competition […].14

Some opinions from the Supreme Court had tackled this ambiguity by making price plus control of output (or competition) a key condition for a showing of market power. In 1927, in Cline v Frink Dairy, the Court had talked of the ‘power to control the market and to fix arbitrary and unreasonable prices’.15 Closer to present times, the Jefferson Parish Court defined market power as ‘the ability of a single seller to raise price and restrict output’.16 But the replacement of ‘or’ with ‘and’ added further confusion. In du Pont, the Supreme Court discussed market power in terms of a reciprocal influence, saying that ‘it is inconceivable that price could be controlled without power over competition or vice versa’.17 The

11 See, for example, FTC v Indiana Fed’n of Dentists 476 US 447 (1986). Here, the Court suggested that a competitive market is one with ‘ability’ to ‘advance social welfare by ensuring the provision of desired goods and services to consumers at a price approximating the marginal cost of providing them’. 12 United States v Colgate & Co 250 US 300 (1919): ‘The purpose of the Sherman Act is to prohibit monopolies, contracts and combinations which probably would unduly interfere with the free exercise of their rights by those engaged, or who wish to engage, in trade and commerce’; ibid 307; Eastman Kodak Co v Southern Photo Materials Co 273 US 359 (1927): ‘the plaintiff being thus deprived, by reason of the monopoly, of the ability to obtain the defendant’s goods and supply them to its trade, its business had been greatly injured’; ibid 369; Kiefer-Stewart Co v Seagram & Sons, Inc 340 US 211 (1951): ‘the Act forbids through its contract with each distributor. Interstate was able to acquire the control and impose its will by force of its monopoly of first-run theatres in the principal cities of Texas and the threat to use its monopoly position against copyright owners who did not yield to its demands’; ibid 228; NYNEX Corp v Discon, Inc 525 US 128 (1998): ‘The freedom to switch suppliers lies close to the heart of the competitive process that the antitrust laws seek to encourage’; ibid 137; agreements ‘which cripple the freedom of traders’, and ‘ability to sell according to their own judgment’; ibid 213; in Interstate Circuit, Inc v United States 306 US 208 (1939) the Court looks at Interstate imposing on film distributors to raise price over subsequent run movies, and this coercion defines monopoly power: ‘The restrictions imposed upon Interstate’s competitors did not have their origin in the voluntary act of the distributors or any of them. They gave effect to the will and were subject to the control of Interstate, not by virtue of any copyright of Interstate, for it had none, but through its contract with each distributor.’ 13 See Appalachian Coals (n 6). See also Geddes v Anaconda Mining Co 254 US 590 (1921), where the Court talked of the control ‘of production or prices’. 14 George A Hay, ‘Market Power in Antitrust’ (1992) 60 Antitrust Law Journal 807. 15 Cline v Frink Dairy Co 274 US 445 (1927). 16 Jefferson Parish (n 3). 17 United States v E L du Pont de Nemours & Co 351 US 377, 391 (1956).

Understanding market power: a legal perspective  63 beginning of the dictum is right. The end is not. No economist would consider that power over competition comes from power over price.18 What is the conventional understanding? Power over price might be indicative of power over competition because it requires control over output.19 But no further inference can be drawn. Besides, inferences from high prices are confounding. Any price increase will tend to weaken power over competition by raising incentives to enter or expand. Today, the case law has dissipated most doubts by replacing ‘or’ and ‘and’ with ‘by’. In Amex, the Court defined market power as ‘the ability to raise price profitably by restricting output’.20 The case shows that emphasis on output is not just merely a lawyer’s terminological quarrel. With eyes on output, the Court observed historical growth of credit card transactions. This fact was relevant to counter allegations that American Express had breached Section 1 by preventing merchants to steer shoppers towards other payment cards.21 The Amex definition of market power appears to be good law.22 In NCAA, the Supreme Court found that the NCAA power over price was uncontroversial given the fact that ‘student-athletes had nowhere else to sell their labor’.23 The Court observed a situation of complete control over output. It inferred power over price. The lower courts are aligned. In Ball Memorial Hospital, the Court of Appeals (7th Cir) wrote that ‘[m]arket power comes from the ability to cut back the market’s total output and so raise price’.24 The second detail is that, in US law, market power does not cover any price increase situation. Recall that since very early, the Court has tried to avoid constructing the Sherman Act in ways that would reach ‘normal and usual contracts incidents’.25 The same concern of reasonable construction has governed the jurisprudential definition of market power. According to the Supreme Court, prices over costs caused by product differentiation do not

Unless one engages in a rhetorical understanding of power over price as the ability to set prices below costs. 19 In Olympia, Judge Posner described this inference as follows: a ‘steep increase in the price’ that does not cause switching can be seen as ‘implying low elasticity of demand […] and therefore monopoly power’; Olympia Equipment Leasing Co v Western Union Telegraph Co 797 F2d 370 (1986). 20 Amex cites scholarship and sections in previous opinions in Kodak and Business Electronics. But in Kodak, the Court did not define market power by reference to price control by restriction of output (but by comparison to what a firm would do in a competitive market. See Ohio v American Express Co 585 US ___ (2018); Kodak (n 9). And in Business Electronics, there is no straight reference to this issue. See Business Electronics v Sharp Electronics Corp 780 F2d 1212 (5th Cir 1986). In Fortner, cited in Kodak, the Court talked of raising price ‘and’ output. See Fortner Enterprises, Inc v United States Steel Corp (n 4). 21 Amex (n 20). 22 ibid. See also, Ball Memorial Hospital, Inc v Mutual Hospital Insurance, Inc 784 F2d 1325, 1335 (1986): ‘Market power comes from the ability to cut back the market’s total output and so raise price.’ 23 NCAA v Alston 141 S Ct 2141 (2021). 24 Ball Memorial Hospital (n 22). 25 Eastern States Lumber Ass’n v United States 234 US 600 (1914): ‘its proper construction the act was not intended to reach normal and usual contracts incident: to lawful purposes and intended to further legitimate trade’. See also, and more generally, United States v Hopkins 427 US 123 (1976): ‘The act of Congress must have a reasonable construction or else there would scarcely be an agreement or contract among business men that could not be said to have, indirectly or remotely some bearing upon interstate commerce, and possibly to restrain it. We have no idea that the act covers or was intended to cover such kinds of agreements.’ From the very beginning, the Court refused effects which were too ‘remote’, ‘not intended’, and ‘too small’ under the Act. 18

64  Research handbook on abuse of dominance and monopolization constitute antitrust market power. In du Pont, the Court said that ‘t[he] power that let us say, automobile or soft-drink manufacturers have over their trademarked products is not the power that makes an illegal monopoly’.26 The same exclusion applies to price increases arising from demand shocks. In Brooke Group, the Court held that ‘[i]f prices rise in response to an excess of demand over supply […] the market is functioning in a competitive manner. Consumers are not injured from the perspective of the antitrust laws by the price increases; they are in fact causing them’.27 In conclusion, it can be said that US antitrust law adopts an economic definition of market power. The definition is in essence close to Cournot’s, but in application tends to focus on control of output. By contrast, the very limited reference to ‘price above marginal costs’ as a definitional element of market power will strike an economically-minded reader of the case law as bizarre. B.

Autonomous Notions of Market Power in EU Law

EU law is characterized by conflicting definitions of market power. On the one hand, the EU courts have followed an original interpretation of the term ‘dominant position’ that has a casual connection with the economic definitions of market power. On the other hand, the Commission and EU lawmakers have adopted a conception of Articles 101 and 102 Treaty on the Functioning of the European Union (TFEU) that draws inspiration from economic definitions of market power. Judicial interpretation is discussed first. As if market power, and monopoly power, were taboo, the terms rarely appear in the case law. So, if the Court does not expressly mention market power, what does it talk about when it talks about dominance? In United Brands, the Court defined a ‘dominant position’ as follows: a position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market by giving it the power to behave to an appreciable extent independently of its competitors, customers and ultimately of its consumers.28

The definition is long. Its wording is not economic. There is no reference to the terms of art used in the market power literature like ‘price’, ‘cost’, ‘profits’, or ‘output’. When words like ‘effective’, ‘independently’, and ‘appreciable’ suggest measurement, no benchmark is provided. The main connection with market power concepts is a reference to ‘economic strength’ or ‘economic power’. In subsequent cases, the Court elaborated on its doctrine of dominance. A constant in the case law has been that the economics of market power is never expressly acknowledged.29

du Pont (n 17). Brooke Group Ltd v Brown & Williamson Tobacco Corp 509 US 209, 235 (1993). 28 See Case 27/76 United Brands Company and United Brands Continentaal BV v Commission of the European Communities [1978] ECR 207, para 81. In Case 322/81 NV Nederlandsche Banden Industrie Michelin v Commission of the European Communities [1983] ECR 3461, para 30, the Court affirms that a dominant position ‘enables [the firm holding] it to hinder the maintenance of effective competition on the relevant market by allowing it to behave to an appreciable extent independently of its competitors and customers and ultimately of consumers’. 29 Some like to see this as a sign of autonomy of the law. 26 27

Understanding market power: a legal perspective  65 Instead, the Court has appeared willing to develop its own conception of ‘dominant position’. The result is a definition that focuses on independence from competitors, on absence of alternatives for customers, or on advantages of the dominant firm. Sometimes all are relevant. The three alternatives will be considered in turn. The most often encountered definition of dominant position in the cases concerns itself with ‘independence’. Some clarity was added in Hoffmann-La Roche,30 where the Court said that an absence of pricing ‘pressure’ characterizes dominance.31 In Michelin I, the Court repeated that a decisive element of dominance consists in conducting itself to a large extent without having to take into account competition.32 Concepts of independence, freedom of action, and limited pressure are not completely detached from economics. As Franklin Fisher once observed, monopoly power is the ‘ability to act in an unconstrained way’.33 Concretely, the judicial concept of dominance appears to cover a range of situations spanning monopoly and oligopoly structures with fringe competitors. Another definition in the case law focuses on the idea that the dominant firm is an ‘unavoidable trading partner’. In multiple judgments, the Court has emphasized that customers (or suppliers) of a dominant firm are forced into economic transactions.34 They have no ability to switch entirely to alternative products, to delay or withdraw purchases, or to self-supply. That conception of dominant power in terms of coercion of customers resembles the monopoly concerns expressed in classical economics works. A last definition centred on ‘economic strength’ rests on firm-level advantages. In United Brands, the Court noted in conclusion that the ‘cumulative effect of advantages makes a dominant position’.35 In AKZO, the Court singled out the defendant’s ‘most highly developed marketing organization both commercially and technically, and [its] wider knowledge than that of their competitors with regard to safety and toxicology’.36

30 In Hoffmann-La Roche, the Court also said that ‘the special feature of a dominant position’ is ‘freedom of action’; Case 85/76 Hoffmann-La Roche & Co AG v Commission of the European Communities [1979] ECR 461, para 41. 31 ibid. The Court stated that ‘the fact that an undertaking is compelled by the pressure of its competitors’ price reductions to lower its own prices is in general incompatible with that independent conduct which is the hallmark of a dominant position’. The Court stated additionally that no ‘competitive pressure’ was ‘likely to jeopardize the market degree of independence’. Note, however, in Michelin I, the statement whereby, in response to the argument that ‘[s]ince 1979 Michelin NV has made a loss’, the Court said that ‘temporary unprofitability or even losses are not inconsistent with the existence of a dominant position. By the same token, the fact that the prices charged by Michelin NV do not constitute an abuse and are not even particularly high does not justify the conclusion that a dominant position does not exist’; Michelin I (n 28) para 59. 32 See ibid para 48. The case concerned discounts granted by a leading supplier on the Dutch market for replacement tyres for heavy vehicles. On account from competition from retread tyres, the Court added that a ‘complete absence of competition’ needed not be present to establish dominance, just ‘partial competition’. 33 Franklin M Fisher, ‘Diagnosing Monopoly’ (1997) 27 J Reprints Antitrust L & Econ 669. 34 The cases are too numerous to mention here. Among them, see Hoffmann-La Roche talking of ‘unavoidable trading partner’; Hoffmann-La Roche (n 30) para 41. See also, Case T-219/99 British Airways plc v Commission of the European Communities [2003] ECR II-5917, para 75, in which the Court talks of ‘unavoidable business partner’. 35 See United Brands (n 28) para 129. 36 Case C-62/86 AKZO Chemie BV v Commission of the European Communities [1991] ECR I-3359/3362, para 61. See also Michelin I (n 28) para 58, where the Court insisted on the defendant’s lead

66  Research handbook on abuse of dominance and monopolization Only in rare cases has an idea of control over output been endorsed. Continental Can is one of them. The Court suggested that once a firm had strengthened its dominant position through abuse, it was able to make life-or-death decisions about competitors.37 Occasionally, the Court has used non-economic concepts to designate high levels of dominance. The expression ‘economic dependence’ often connotes dominant market power.38 More recently, the General Court discussed Google’s position in general search engine services in terms of ‘super’ dominance.39 The EU courts’ conception of dominance in terms independent from economics is mysterious.40 From a historical standpoint, the market power literature was well developed when the treaties that contain the basic EU antitrust rules were drafted. As seen previously, compared to US law, the language of the EU provisions is more economic. Moreover, the judicial institution in charge of interpreting the Treaty of Paris that preceded the EU treaties had come close to an economic definition of market power. A 1960 judgment about a joint agency agreement for the selling of coal amongst West German mines stated that: [t]he power to determine prices, however, resides in a power, given to the undertaking in a position to exercise it, to establish prices at a level appreciably different from that which would be established by the effects of competition alone.41

over its competitors in relation to ‘investment and research’, its ‘range of products’, and the ‘efficiency and quality’ of its dealer network. 37 See Case 6/72 Europemballage Corporation and Continental Can Company Inc v Commission of the European Communities [1973] ECR 215, para 26: ‘Abuse may therefore occur if an undertaking in a dominant position strengthens such position in such a way that the degree of dominance reached substantially fetters competition, i.e. that only undertakings remain in the market whose behaviour depends on the dominant one.’ If this is the case, then the dominant position that pre-exists the abuse implies some lesser control, that is not necessarily existential, over competitors. 38 Case C-226/84 British Leyland Public Limited Company v Commission of the European Communities [1986] ECR 3263, para 9; Case C-583/13 P Deutsche Bahn AG and Others v European Commission ECLI:EU:C:2015:404, para 57. 39 Case T-612/17 Google LLC, formerly Google Inc and Alphabet, Inc v European Commission [2021] ECR II-3601. 40 In 2007, the General Court accepted a description of Microsoft’s grip on the personal computer market as a ‘quasi-monopoly’; Case T-201/04 Microsoft Corp v Commission of the European Communities [2007] ECR II-3601, para 775. See also, for a previous case, Case C-333/94 Tetra Pak International SA v Commission [1996] ECR I-5951, para 31. Also, the jurisprudence on collective dominance appears to draw inspiration from game theoretic insights, but the cases are extremely rare. See Nicolas Petit, ‘The Oligopoly Problem in EU Competition Law’ in Ioannis Liannos and Damien Geradin, Research Handbook in European Competition Law (Edward Elgar Publishing 2013). 41 Case no 13-60 ‘Geitling’, Ruhrkohlen-Verkaufsgesellschaft mbH and others v High Authority of the European Coal and Steel Community ECLI:EU:C:1962:15, 83. The Court wrote: ‘Thus, to show the existence of a power to determine prices, it is necessary to establish that the actual prices are, or could be, different from what they would have been in the absence of any power to fix prices. Such a proposition involves a subtle comparison between the actual and the potential, of a kind which must rest to a considerable extent on informed speculation.’

Understanding market power: a legal perspective  67 The judgment also held that the power to determine prices was commensurate with the ‘volume of production’ under internal control of the combination, and was inversely related to the weight of external competition.42 As hinted above, the Court’s autonomous definition of dominance is inconsistent with legislative and policy developments. In sectoral telecommunications regulation, EU lawmakers started in the 2000s to explicitly discuss dominance in terms of ‘significant market power’.43 In December 2005, the Directorate General for Competition of the European Commission published a ‘Discussion Paper’ on the application of Article 82 EC, now 102 TFEU, to exclusionary abuses. Attempting to breathe economics into the legal concept of independence, the paper stated that ‘[f]or dominance to exist the undertaking(s) concerned must not be subject to effective competitive constraints. In other words, it thus must have substantial market power’.44 In a following 2009 Guidance Paper on exclusionary abuse, the Commission framed dominance as a market power issue, and defined it as the capacity to ‘profitably increas[e] prices above the competitive level for a significant period of time’.45 Today, the lack of a single definition of dominance based on economics can be contrasted with the more economically minded opinions of the Court in Post Danmark I and Intel.46 The law on single firm conduct moves towards economics when the courts interpret the concept of abuse. But it remains quite ‘esoteric’ when the focus is on dominance.47 The result is a mixed approach.

III.

PROOF OF MARKET POWER

Substantive antitrust law infers anticompetitive purpose or effect from proof of market power facts (A) and sub-facts (B).

A similar economic understanding of monopoly power was pushed forward in a 1966 Memorandum of Concentration in the Common Market commissioned by the Commission. The authors wrote that a monopoly position ‘leads to a limitation of production, so that profits are maximised through prices above the level that would prevail in an oligopolistic market, where output levels would be higher’ (free translation of: une situation de monopole ‘conduit souvent à limiter la production, de telle sorte qu’il y a maximalisation des profits, obtenue par des prix qui se situent à un niveau plus élevé que ce ne serait le cas dans un marché oligopolistique qui, lui, conduirait à un niveau de production plus élevé’). See Etudes ‘Le Problème de la Concentration dans le Marché Commun’ (1966) 3 Series Concurrence 26. 43 See Directive 2002/21/EC of the European Parliament and of the Council of 7 March 2002 on a common regulatory framework for electronic communications networks and services, (2002) OJ L108, Art 14(2), which provided: ‘An undertaking shall be deemed to have significant market power if, either individually or jointly with others, it enjoys a position equivalent to dominance, that is to say a position of economic strength affording it the power to behave to an appreciable extent independently of competitors, customers and ultimately consumers.’ 44 DG Competition, ‘Discussion Paper on the Application of Article 82’ (2005), accessed at https://​ eur​-lex​.europa​.eu/​legal​-content/​EN/​ALL/​?uri​=​CELEX​%3A52009XC0224​%2801​%29. 45 Communication from the Commission – Guidance on the Commission’s enforcement priorities in applying Article 82 of the EC Treaty to abusive exclusionary conduct by dominant undertakings (2009) OJ C 45/07, para 11. 46 See Case C-23/14 Post Danmark A/S v Konkurrencerådet [2015] ECLI:EU:C:2015:651; Case T-286/09 RENV Intel Corporation Inc v European Commission ECLI:EU:T:2022:19. 47 See Michel Waelbroeck and Aldo Frignani, European Competition Law (Transnational Publishers 1999). 42

68  Research handbook on abuse of dominance and monopolization A.

Three Main Market Power Facts

Three types of market power facts can be relevant to a violation of antitrust law. The first is a market power base. An antitrust agency or court can use a market power base as a screen for anticompetitive purposes or effects. A market power base throws contextual light on anticompetitive motives.48 Further, lasting injury to consumers can be inferred from a market power base.49 The second is market power utilization. An antitrust agency or court will determine whether control over output has been deployed to exploit trading parties or exclude rivals.50 Market power utilization happens, for example, if a multi-product firm forces inelastic demand in product A to purchase its output in product B. The third is a market power injury. To establish harm under the consumer welfare standard that guides the practical application of US and EU law, an antitrust agency or court will focus on market power harms.51 Considerable differences exist between US and EU law concerning market power facts. In EU law, a market power base will allow the inference that pricing below cost is intentionally and consequentially predatory. This is not the case in US law, which does not cover market power utilization ‘by merely enhancing the price’ of a product.52 EU law, however, covers acts of monopolistic exploitation, though enforcement has historically been limited.53 Last, both US and EU law assume market power injury in aftermarkets, upon proof of dominance in primary markets.54 48 In AKZO, the Court held that ‘a dominant undertaking has no interest in applying such [below cost] prices except that of eliminating competitors so as to enable it subsequently to raise its prices by taking advantage of its monopolistic position’. See AKZO (n 36) para 71. 49 Though rendered in a merger context, the General Dynamics opinion explains this well: ‘In most situations, of course, the unstated assumption is that a company that has maintained a certain share of a market in the recent past will be in a position to do so in the immediate future.’ The Court added that firms’ ‘past performances imply an ability to continue to dominate with at least equal vigor’; United States v General Dynamics Corp 415 US 486, 501 (1974). 50 For example, in Interstate Circuit, the Court noted that a powerful theatrical exhibitor of movies in the main cities of Texas ‘was able to acquire the control and impose its will by force of its monopoly of first-run theatres […] and the threat to use its monopoly position against copyright owners who did not yield to its demands’; Interstate Circuit (n 12) 228. Similarly, EU law is concerned that strong distributors with market power may ‘be able to force/convince one or more suppliers to fix their resale price above the competitive level’; European Commission, Guidelines on Vertical Restraints [2010] OJ C 130/1 63. 51 US law tends to use the concept of market power as a workable surrogate for anticompetitive effects. Though in Actavis, the Court hinted that market power covers effects narrower than anticompetitive effects, by enumerating as traditional factors of antitrust analysis, ‘anticompetitive effects, redeeming virtues, market power, and potentially offsetting legal considerations’. See Fed Trade Comm’n v Actavis, Inc 570 US 136 (2013). 52 In Jefferson Parish, the Court drew a line between the ‘exploitation of market power by merely enhancing the price of the tying product, on the one hand, and by attempting to impose restraints on competition in the market for a tied product, on the other’. Utilization of market power is only unlawful when exclusionary; Jefferson Parish (n 3). 53 Case C-177/16 Autortiesību un komunicēšanās konsultāciju aģentūra / Latvijas Autoru apvienība v Konkurences padome, EU:C:2017:689. 54 As far as US law is concerned, see Kodak (n 9). As far as Europe is concerned, see Case C-56/12 P European Federation of Ink and Ink Cartridge Manufacturers (EFIM) v European Commission

Understanding market power: a legal perspective  69 Antitrust regimes use proxies of power and/or structure for market power facts. In the case of power, the problem of proof consists in establishing a degree of influence on price.55 In the case of structure, the problem of proof consists in establishing market ‘control’, or ‘dominance’56 or monopoly.57 US antitrust law allows a fact finder to establish a market power fact by the scrutiny of power or structure. In Appalachian Coals, the Court talked of the ‘control of a market or power to fix monopoly prices’.58 In Fortner, a similar idea was expressed with indiscriminate reference to standards of ‘sufficient economic power’, ‘dominant position’, or ‘monopoly’.59 In Amex, the Court blurred the clarity of the principle. In a dictum, the Court implied that the rule of reason implied a facts-specific assessment of ‘market power and market structure’.60 EU antitrust law appears to place a stronger emphasis on structural proxies. As far as a market power base is concerned, the preference for structural proxies can be seen in the mention of the dominant position in Article 102 TFEU. Structural proxies also dominate the inquiry of market power injury. In Intel, a case celebrated for its embrace of the effects-based approach, the Court mostly pointed out six structural factors for the analysis of exclusivity enhancing rebates: extent of the dominant undertaking market position, share of market covered by the challenged practice, conditions governing the grant of the rebates, duration, amount, and possible exclusionary aim.61 B.

Subsidiary Fact: The Relevant Market

US and EU antitrust law require market power facts to be assessed by reference to a ‘relevant market’. The relevant market is a subsidiary fact specific to antitrust law.62 The history, meaning, and practice of the ‘relevant market’ in both legal systems are reviewed below.

ECLI:EU:C:2013:575. The Court adopted a presumption of dominance, which can be ruled out if competition in primary markets exist under four conditions identified in the case. See also Case T-427/08 Confédération européenne des associations d’horlogers-réparateurs (CEAHR) v European Commission [2010] ECR II-5865. 55 In Alcoa, Learned Hand acknowledged ‘certain limits’ to the power of a monopolist, consisting in the ‘expansion of small producers’ or in the costs of ‘keeping idle any part of a plant or of personnel’; Alcoa (n 10). 56 US v Standard Oil talks of ‘dominancy’. See US v Standard Oil 32 US 301, 67 S Ct 1604 (1947). 57 The Court is not always clear on the meaning to give to monopoly. In Grinnell, it held that ‘domination or control of it makes out a monopoly’; United States v Grinnell Corp 384 US 563 (1966). But in US v Knight, the Court appeared to define a ‘monopoly’ as a single supplier; United States v E C Knight Co 156 US 1 (1895). 58 See Appalachian Coals (n 6) 375. 59 See Fortner Enterprises (n 4) 499. 60 See Amex (n 20). In Kodak, for example, the Court redundantly found that a printer manufacturer with a monopoly position on the relevant parts market had actually engaged in conduct with market power effects on consumers; Kodak (n 9). 61 See Intel (n 46). 62 In Astra Zeneca, the EU General Court stressed that the meaning of the concept differs from its conventional use in economics, law, and social sciences. See Case C457/10 P AstraZeneca AB and AstraZeneca plc v European Commission ECLI:EU:C:2012:770, where the Court affirmed that ‘the concept of a relevant market is different from other definitions of markets often used in other contexts, such as the area in which companies sell their products or, more broadly, the industry or sector to which the companies belong’. For a detailed discussion of market definition, see Chapter 1 in this volume.

70  Research handbook on abuse of dominance and monopolization a. US law The Sherman Act invites fact finders to search for market power in ‘any part’ of commerce. The Court added a requirement in International Shoe Company, that the ‘existence of competition is a fact disclosed by observation, rather than by the processes of logic’.63 So where in the economy should fact finders turn their eyes when they look for competition or market power?64 The answer is in the question. Antitrust laws’ unit of analysis is the market. But what exactly is the market over which the power or position of an antitrust defendant has to be evaluated?65 Plausibly conscious of the subjectivity of the issue, the Court in Appalachian Coals warned that the application of the statute should ‘not to be determined by arbitrary assumptions’.66 In Alcoa, Learned Hand echoed the problem, writing that there were ‘various ways of computing Alcoa’s control of the aluminium market – as distinct from its production’, ‘depending upon what one regards as competing in that market’.67 A first attempt to deal with the subjectivity issue appeared in Columbia Steel. The US Government was trying to block US Steel’s acquisition of the largest steel supplier of the West coast. The Court supplied a first holding that one ought to ‘delimit the market in which the concerns compete’, and then to ‘determine the extent to which the concerns are in competition in that market’.68 The elaboration of a complete method of market definition came, however, in du Pont. The Court had to decide whether du Pont’s 75 per cent share of US cellophane trade was an unlawful monopoly under Section 2. A previous ruling had found that ‘competition from […] other materials prevented du Pont from possessing monopoly powers’.69 The Court said that the true test of competition lies in ‘considering what is the relevant market’, and that ‘no more definite rule can be declared than that commodities reasonably interchangeable by consumers for the same purposes make up that “part of the trade or commerce”’. In practice, the resolution of this issue, said the Court, ‘called for an appraisal of the “cross-elasticity” of demand in the trade’.70 The legacy of du Pont is controversial. The issue is not so much with the adoption of a rule of interchangeability for purposes of market definition. There is strong support for this idea, even with opponents to market definition devices in antitrust application.71 Rather, the issue is with the application of a test of cross-elasticity to an existing monopoly firm. In du Pont, the Court held on the facts that a market for all wrapping materials existed, stressing the ‘[g]reat sensitivity of customers in the flexible packaging markets to price or quality changes’ prevented du Pont from possessing monopoly control. But perhaps this was only true because International Shoe Co v FTC 280 US 291 (1930). In Standard Oil, the Court acknowledged the geographic and distributive significance of this inquiry. But the Court abstractly instructed a search for market power in relation to ‘classes of things’; Standard Oil Co of NJ v United States 221 US 1 (1910). 65 In the formative era, the Supreme Court considered industry level market shares to establish monopoly. Swift & Co v United States 286 US 106 (1932) provides an illustration. The Court lumped meat, fish, vegetables, either fresh or canned, fruits, cereals, milk, poultry, butter, eggs, cheese, ‘and other substitute foods ordinarily handled by wholesale grocers or produce dealers’ in the same market. 66 See Appalachian Coals (n 6) 474. 67 Alcoa (n 10) 424. 68 See United States v Steel Corp 251 US 417 (1920). 69 See du Pont (n 17) 380. 70 ibid. 71 Like, for instance, Louis Kaplow, ‘Antitrust, Law & Economics, and the Courts’ (1987) 50 Law and Contemporary Problems 181. 63 64

Understanding market power: a legal perspective  71 du Pont was already charging the monopoly price, and demand was more elastic at this level. Economists discuss this problem in terms of the ‘cellophane fallacy’. Besides, important ambiguities remain in the interpretation of the market definition doctrine. Is market definition a necessary step that conditions the validity of a market power evaluation? Is this true in relation to both Sections 1 and 2?72 In Section 2 cases, the Court has required an inquiry into the relevant product and geographic market.73 But in Section 1 cases, the doctrine is confusing. While application of the rule of reason appears to require a market definition,74 the doctrine in relation to other categories of analysis like the quick look rule or the per se prohibition rule is chaotic.75 A dictum in Broadcast Music v CBS held that the analysis of the ‘portion of the market’ affected was required to apply the per se prohibition rule against price fixing.76 Subsequently, per se analysis cases like NCAA v Board of Regents muddied the waters.77 Reviewing the facts, the Court confirmed the finding of a ‘separate market for telecasts of college football which “rest[s] on generic qualities differentiating” viewers’ over which the NCAA enjoyed a monopoly. The passage was, however, preceded by a statement suggesting that the NCAA restrictions could attract liability ‘without proof of market power’, and ‘even in the absence of a detailed market analysis’. Later, in Ind Fed’n of Dentists, the Court held more clearly that since the function of ‘market power is to determine whether an arrangement has the potential for genuine adverse effects on competition, “proof of actual detrimental effects, such as a reduction of output,” can obviate the need for an inquiry into market power’.78 The Supreme Court might have laid this discussion to rest in recent opinions. In Amex, the Court required a preliminary market definition in a case involving a vertical restraint, even though the evidence on the record pointed out clearly to adverse price effects.79 And in NCAA v

Note that in merger cases, the Court has held that market definition was a necessary predicate. See Brown Shoe Co v US 370 US 294 (1962). In that case, the Court held: ‘Thus, again, the proper definition of the market is a “necessary predicate” to an examination of the competition that may be affected by the horizontal aspects of the merger’; ibid 335. See also US v Pabst Brewing 384 US 546 (1966). 73 See Spectrum Sports: In Section 2 the Court ‘requires inquiry into the relevant product and geographic market and the defendant’s economic power in that market’; Spectrum Sports, Inc v McQuillan 506 US 447 (1993). 74 On the rule of reason, see Chicago Board of Trade v United States 246 US 231 (1918): ‘The effects of the rule: As it applies to only a small part of the grain shipped to Chicago and to that only during a part of the business day and does not apply at all to grain shipped to other markets, the rule had no appreciable effect on general market prices; nor did it materially affect the total volume of grain coming to Chicago’; ibid 240. 75 As the Court noted in California Dental Assn v FTC 526 US 756 (1999), ‘categories of analysis of anticompetitive effect are less fixed than terms like “per se,” “quick look,” and “rule of reason” tend to make them appear’; at (779). In intermediate analytical categories, once can find a range of practices that will require some analysis, but not a full-blown market investigation. This is the case when ‘likelihood’ of ‘anticompetitive effects is obvious’ (Areeda talked of the ‘twinkling of an eye’; see Phillip Areeda, ‘The “Rule of Reason” in Antitrust Analysis: General Issues’ (1981) 81 Federal Judicial Center 1) or where a ‘confident conclusion about the principal tendency’ of the anticompetitive effects of the restraint can be reached. 76 Broadcast Music, Inc v CBS, Inc 441 US 1 (1979), 1552. We are well aware that BMI is a rule of reason case, but the dictum we discuss concerns the per se rule. 77 NCAA v Board of Regents (n 3). 78 Ind Fed’n of Dentists (n 11) 460. 79 See Amex (n 20). 72

72  Research handbook on abuse of dominance and monopolization Alston, the Court did not challenge the identification, by the lower courts, of a relevant market in a case where the defendants had admittedly engaged in horizontal price fixing.80 Bottom line? A market definition screen appears warranted in all or most cases. The amount of market definition work required, however, will depend on the type of restraint. If the challenged conduct falls within a category of behaviour which has been ruled to be one that knowingly threatens to raise prices and decrease output like price fixing, the market definition work will be limited.81 When this is not the case, the amount of market definition work will be higher. The rule that requires a market definition is sensible in most cases. True, a straight focus on price or output might save a lot of time. But what price, or what output, is it relevant to focus on? Do the prices or output observed cover a significant proportion of buyers? And if price and output are relevant, how fixed in time are they? A handle on the relevant market provides answers. Direct observations of price or output effects do not. b. EU law EU law benefited from cross-fertilization. Compared to US law, EU law developed more quickly a market definition custom. The first cases adopted under Articles 101 and 102 TFEU were built upon an explicit consideration of relevant markets.82 A more important difference, however, is that EU law assigns a critical role to market definition in single firm conduct cases, and a lesser one in coordinated conduct cases. The courts require a market definition in Article 102 TFEU cases. In United Brands, the Court said that market definition is ‘necessary’ to determine a dominant position.83 Other abuse of dominance cases kept stressing that the definition of the market is of ‘essential significance’84 or ‘fundamental significance’.85 The Court’s insistence on the market definition is logical. The wording of Article 102 TFEU requires proof of a base of market power for its application. And the term ‘position’ hints at a structural proxy. The market definition helps to the extent that it gives a market share number. As far as Article 101 TFEU is concerned, the production of a market definition as grounding for a finding of liability appears unnecessary. In Volkswagen, the Court held that the definition of the relevant market differs ‘according to whether Article [101] or Article [102] of the Treaty is to be applied’, and said it is only a ‘necessary precondition’ for the application of Article 102 TFEU.86 Now, as is often the case in EU competition law, the clear rule in the case law is belied by judicial and administrative practice. The reality is this: market definition is key in Article 101 TFEU cases. Antitrust legislation as well as soft law documents allow parties to agree-

See NCAA v Alston (n 23). Of course, if a possibility is reserved to salvage the conduct from per se illegality by application of a de minimis rule, then the market definition work increases. 82 The first opportunity to assess the relevance of market definition was identified by the Court in Continental Can (n 37) paras 32−3. On this, see Miguel S Ferro, Brief History of Market Definition (Edward Elgar Publishing 2019). 83 United Brands (n 28). 84 Continental Can (n 37) para 32. 85 Case 31/80 Loreal v De Nieuwe AMCK [1980] Rec 1980/3775, para 25. 86 See Case T-62/98 Volkswagen AG v Commission of the European Communities [2000] ECR II-2713, para 230. 80 81

Understanding market power: a legal perspective  73 ments the benefit of individual or categorical exceptions conditional on the respect of market share limits.87 In Delimitis, a case involving a dispute between a brewery and a tenant over a comprehensive beer supply agreement, the Court held that in any competitive evaluation of effects ‘the relevant market must first be determined’.88 Later, in Langnese Iglo, a case over similar arrangements in the ice cream industry, the Court held that the delimitation of the relevant market is ‘essential in order to analyse the effects of the exclusive agreements on competition’.89 The market definition exemption of Volkswagen might therefore apply only to a subset of naked, or ‘by object’, restrictions. Yet even in such cases, a definition of the relevant market will be useful for agencies and courts in the estimation of antitrust fines or damages. One important nuance matters still. The EU courts ascribe greater consequences to errors in market definition under Article 102 TFEU, compared to Article 101 TFEU. In Langnese Iglo, the Court faulted the Commission for not including scooping ice cream sold in the street in its definition of a market for impulse ice cream. But the Court reviewed other facts, and found on that additional examination that the market definition error would have produced marginal consequences on the Commission’s evaluation of anticompetitive effects. By contrast, in Article 102 TFEU, market definition errors will be presumed to invalidate findings related to dominance, without further need to analyse the facts. In the first CEAHR case, the General Court refused to follow the Commission’s holding that luxury and prestige watches, spare parts, and repairs and maintenance services formed a single market. The case was discarded. The Commission adopted a new decision, this time distinguishing between three separate markets.90 Servier shows even more neatly how market definition errors can collapse a liability finding like a house of cards. Here, the Commission had found that a pharmaceutical manufacturing firm had unlawfully breached Article 102 TFEU by entering into pay for delay practices to protect its dominant position over Perindropril.91 The Commission had excluded from the relevant market a class of heterogeneous medical products known as ACE inhibitors. The General Court faulted the Commission’s conclusion on the ground of survey evidence and internal documents that showed clear patterns of substitution with prescribers.92 It annulled the finding of abuse. The general takeaway is this. Under Article 102 TFEU, market definition is the analytical foundation on which everything rests.

These are the de minimis rules. See Communication from the Commission, Notice on agreements of minor importance which do not appreciably restrict competition under Article 101(1) of the Treaty on the Functioning of the European Union (De Minimis Notice), OJ [2014] C291/1; Commission Regulation (EU) No 330/2010 of 20 April 2010 on the application of Article 101(3) of the Treaty on the Functioning of the European Union to categories of vertical agreements and concerted practices, [2010] OJ L102/1; Communication from the Commission, Notice – Guidelines on the application of Article 81(3) of the Treaty [2004] OJ C101/97. 88 See Case C-234/89 Stergios Delimitis v Henninger Bräu AG [1991] ECR I-935, para 16. 89 Case T-7/93 Langnese Iglo GmbH v Commission of the European Communities [1995] ECR II-1533, para 60. 90 See CEAHR (n 54). 91 Perindopril (Servier) (Case COMP/AT.39612) Commission Decision of 9 July 2014. 92 Case T-691/14 Servier SAS and Others v European Commission ECLI:EU:T:2018:922, para 1480. 87

74  Research handbook on abuse of dominance and monopolization c. Conclusions When antitrust experts disagree, it is often about market definition. A usual controversy concerns the right degree of aggregation of products, services, or user groups into a relevant market. To see the issue clearly, consider the following example. Assume three products A, B, and C exist. The products are not substitutable. A firm supplies A, B, and C. Is it in a ‘single’ market for a cluster of products A+B+C? Or is it in ‘separate’ markets A, and B, and C? Now assume that A, B, and C are respectively right shoe, left shoe, and laces. Clearly, the single market appears to be the correct answer. But consider now that A, B, and C are respectively printers, ink cartridges, and repair services. The single market conclusion is less obvious. What rule of method should allow the selection of a single or a separate market? Both US and EU law have struggled to bring a coherent response to this issue. Often, agencies and courts look at average and historical diversification trends in the industry.93 If most firms are multi-product ones, then the analysis will lean towards a single market.94 If not, that is, some firms are diversified and others are not, then the analysis will follow separate markets.95 But this logic does not hold in all cases. In Hoffmann-La Roche, the Court followed a separate market analysis, despite the industry being composed of multi-product firms.96 In AKZO, the industry configuration looked similar, but the Court took another tack.97 Sometimes, the logic followed has nothing to do with facts.98 In Hilti, the Court supported an idea of separate markets for nail guns, cartridge strips (and cartridges), and nails even if most suppliers sold all three components as ‘systems’. The Court said that going for a single market would prevent applying the law on tying,99 and suggested this would be a problem. Market definition also receives severe criticism on the ground that it downplays competition.100 With an analytical focus on short-term substitutability, market definition would by design neglect the competitive pressure arising from supply-side elasticity, or from potential competitors. Moreover, market concern would be gameable. Because market definition is

For instance, in Michelin I, the Court looked at a single market for replacement tyres for lorries, buses, and similar vehicles, confirming the industry-level tendency; Michelin I (n 28). 94 See, for example, Grinnell (n 57). Here, a group of related companies involved in protection and security solutions had acquired 87 per cent of the national central station service business. The Court held that 87 per cent looked clearly like a monopoly but added with a doubt, ‘if that business is the relevant market’; ibid 571. It proceeded to ask, ‘[t]he only remaining question therefore is, what, is the relevant market?’; ibid. In turn, the Court said the answer to the question requires analysis of ‘commodities reasonably interchangeable’; ibid. With this background, the Court controversially refused to consider each protection and security service one by one: burglar protection, fire protection, and so on. Instead, it found that the right approach consisted in considering that firms competed with a ‘cluster’ of security services, and that the defendant was dominant; ibid 573. 95 See the first CEAHR judgment where the General Court noted that the existence of independent repairers that were not manufacturing luxury and prestige watches was key to annulling the EC’s decision to treat the markets for luxury watches and luxury watches’ aftersales services as a single one; see CEAHR (n 54). 96 Hoffmann-La Roche (n 30). 97 AKZO (n 36). 98 Note here that the Court has said in Servier that ‘the definition of the boundaries of the market must be made by examining empirical evidence, aimed at making an effective use of all information which is relevant in an individual case’; see Servier (n 92) para 1383. 99 Case T-30/89 Hilti AG v Commission of the European Communities [1991] ECR II-1439. 100 In the past, market definition was said to exonerate too much monopoly power. Critics liked to point out this to the Supreme Court decision in du Pont; see du Pont (n 17). 93

Understanding market power: a legal perspective  75 determinant in liability outcomes, Mr Justice Fortas warned in his Grinnell dissent of risks that agencies and plaintiff promote ‘procrustean’ definitions that ‘gerrymander’ the boundaries of a relevant market around a firm’s product.101 In contrast, defendants pretend that they compete against a large population of firms which often will comprise either suppliers of complements, differentiated, or bundled products. Last, the market definition has also stressed the costs of applying the market definition to environments where firms intermediate the interdependent demand of distinct user groups and/ or where suppliers charge no monetary price for their goods. These issues are especially salient in digital markets. It is a pity that market definition fuels division in antitrust law. As Judge Posner rightly observed, monopoly power is a ‘question of fact’.102 Facts being imperfect, people will inevitably disagree over them. In principle, a focus on method, if reasonable, should enable consensus.103 The preliminary delimitation of the relevant market arguably provides just this.104 It sets fact finding on neutral grounds, promotes cooperation with defendants, and maintains the costs of market power evaluation within reasonable bounds. It is likely that antitrust disagreements about market definitions owe more to priors about the solution of particular cases, than over the analytical method itself.

IV. EVIDENCE In US and EU law, market power facts must be proved on the evidence. But what sort of evidence can be proved to establish a market power fact? In contemporary practice, two types of evidence can be used to establish market power facts. Courts accept direct or indirect evidence. Direct evidence focuses on the exercise of power to infer its existence. The relevant data points are higher prices, lower output, or reduced quality. Indirect evidence focuses on market power circumstances (also called circumstantial evidence) to infer its existence. The relevant data points are dominant market shares (or a foreclosure share leading to dominance), entry barriers, the regulatory environment, and other economic characteristics associated with market power. Both ideas of exercise and circumstances of market power match conceptually well with the proxies of power and structure described above. But they are not identical. Proxies of power and structure are intellectual constructs. They are not legal principles like the rules of evidence. Besides, direct evidence of market power focuses on a class of issues narrower than the abovementioned concept of power. For example, coercive acts like threats to withhold supply to buyers will rarely be deemed direct evidence of market power. By contrast, circumstantial evidence embodies more than structural indicators. For example, exclusionary practices of

See Justice Fortas’ dissent in Grinnell (n 57). Olympia (n 19). 103 In du Pont, the Court held that ‘Section 2 requires the application of a reasonable approach in determining the existence of monopoly power just as surely as did § 1’; du Pont (n 17). 104 Justice Fortas described it in his dissent in Grinnell as ‘an economic task put to the uses of the law’. Mr Justice Fortas, however, also pointed out to risks of distortion, unless it is ‘well done’; Grinnell (n 57). 101 102

76  Research handbook on abuse of dominance and monopolization a firm with a low market share might be deemed unlawful upon diagnosis of an incipient trend toward concentration. This sub-section addresses the following question: do US and EU antitrust laws prefer one type of evidence over the other to establish market power facts? We focus on facts of market power base and injury, and leave aside market power utilization whose proof can often be observed directly.105 A.

US Law

In US law, direct and circumstantial evidence can be used to establish market power facts. The choice of one evidential route over the other depends on what market power fact the agency or plaintiff wants to prove. That said, a general rule of practice is the following. Indirect evidence often supports proof of a base of market power fact. Direct evidence is often relied upon to show a fact of market power injury. In other words, US antitrust practice tends to treat direct evidence as inferentially stronger in a dynamic context (has price increased?), rather than in a static context (are prices high?).106 One qualifying addition is in order. Direct evidence only dominates indirect evidence to prove a fact of actual market power injury. In Ind Fed of Dentists, the Court said that adverse output effects ‘obviate the need for an inquiry into market power’ – meaning structure – ‘which is a surrogate for detrimental effects’.107 But indirect evidence will remain easier to adduce in cases where a fact of likely market power injury constitutes the allegation. Moreover, direct evidence is not a smoking gun.108 The case law of the Supreme Court shows that cases brought on the basis of direct evidence are fragile. In du Pont, the Court refused to consider high profits as evidence of market power absent a benchmark for comparison.109 In Brooke Group and Amex, the Court noted that confounding factors such as output expansion discarded the inferential value of evidence of rising prices.110 This is not to mean 105 Note that sometimes, direct evidence, or absence thereof, is used to establish market power utilization. For example, in Matsushita, the Court rejected a notion that there had been a conspiracy on grounds of direct market power evidence. Put differently, direct evidence of absence of monopoly power disproves the act of conspiracy. The Court stated: ‘The alleged conspiracy’s failure to achieve its ends in the two decades of its asserted operation is strong evidence that the conspiracy does not in fact exist’; Matsushita v Zenith Radio Corp 475 US 574, 592 (1986). 106 A long line of cases looks at whether defendants possess market power by considering market share evidence. See, for instance, Lorain Journal Co v United States 342 US 143 (1951). Here, the Court held: ‘The court below describes the position of the Journal, since 1933, as “a commanding and an overpowering one. It has a daily circulation in Lorain of over 13,000 copies and it reaches ninety-nine per cent of the families in the city”’; at 149. 107 Ind Fed’n of Dentists (n 11) 461. 108 Contrary to a widespread presentation (that is not based on Supreme Court doctrine) that holds that direct evidence cannot be rebutted simply by the production of indirect evidence. See Andrew I Gavil, ‘Burden of Proof in U.S. Antitrust Law’ (2008) 1 Issues in Competition Law and Policy 125. 109 See du Pont (n 17), where the Court affirmed: ‘Nor can we say that du Pont’s profits, while liberal (according to the Government 15.9% net after taxes on the 1937−1947 average), demonstrate the existence of a monopoly without proof of lack of comparable profits.’ And: ‘There is no showing that du Pont’s rate of return was greater or less than that of other producers of flexible packaging materials’; ibid 404. 110 See Brooke Group (n 27). The Court held that: ‘Even in a concentrated market, the occurrence of a price increase does not in itself permit a rational inference of conscious parallelism or supracompetitive

Understanding market power: a legal perspective  77 that direct evidence never proves antitrust injury. In Conwood, evidence that industry output had been higher before the challenged conduct allowed the Court to establish a violation of the Sherman Act.111 Against this background, one can question whether direct evidence constitutes the preferential route for an antitrust plaintiff or agency. Cases advanced on grounds of likely effects might not be harder. Recall that, in Microsoft, the Court did not require a very large foreclosure share to establish Section 2 liability.112 It also refused to impose a strict counterfactual demonstration that Microsoft’s conduct was the main cause of the foreclosure of competitors. Last, a possible exception to the preference for direct evidence concerns market definition. The Supreme Court has accepted both direct (in du Pont) and indirect (in Grinnell) evidence of product interchangeability.113 The reason for the lower reliance on direct evidence owes to the cellophane fallacy. When a 5 to 10 per cent price increase is applied on top of a monopoly price – a test known as the Small but Significant Non-transitory Increase in Prices (‘SSNIP’) test – unreal patterns of substitution emerge. For example, a consumer of monopolized TV entertainment subject to a price increase above its reservation value might choose to redirect purchases towards books, online language courses, or knitting material.114 Yet, it is a stretch to consider that all these products compete in the same market as TV entertainment channels. B.

EU Law

EU law prefers indirect evidence to prove many market power facts. This is true for facts of market power base. Since Hoffmann-La Roche, ‘very large market shares’ must be presumed to establish a dominant position.115 In AKZO, the Court held that a market share of 50 per cent qualified as ‘very large’.116 AKZO marked an evolution. In Hoffmann-La Roche, the Court had refused application of the presumption to a share approaching 51 per cent,117 and reserved it

pricing. Where, as here, output is expanding at the same time prices are increasing, rising prices are equally consistent with growing product demand. Under these conditions, a jury may not infer competitive injury from price and output data absent some evidence that tends to prove that output was restricted or prices were above a competitive level’; ibid 237. See also Amex (n 20). 111 Thus controlling for what Harold Demsetz called the Nirvana Fallacy, see Harold Demsetz, ‘Information and Efficiency: Another Viewpoint’ (1969) 12 The Journal of Law & Economics 1. In Conwood, the defendant was saying that there was no injury to competition because the number of moist snuff products had increased, and the market had expanded, compared to other tobacco products that had decreased. The court, however, found that Conwood’s growth post defendant conduct had been lower (2.5 per cent), than pre conduct (11 per cent) in the 10 years prior to 1990; Conwood Company v United States Tobacco Company 290 F3d 768 (2002). 112 See United States v Microsoft Corp 584 US, 138 S Ct 1186 (2018). Note there another inconsistency in a discrepancy between Sections 1 and 2. The case law does require different levels of market share foreclosure under Sections 1 and 2 when the indirect evidence route is followed. But it does not discriminate equally between both provisions under the direct evidence route. 113 In Grinnell, the Court noted that watchman services or local services did not compete with central station ones, on the ground that they were less reliable, and gave rise to less premium discount from insurers; Grinnell (n 57). 114 Lawrence White and Philip B Nelson, ‘Market Definition and the Identification of Market Power in Monopolization Cases’ (2003) NYU Working Paper No EC-03-26. 115 Hoffmann-La Roche (n 30) para 41. 116 AKZO (n 36) para 60. 117 See Hoffmann-La Roche (n 30) para 58.

78  Research handbook on abuse of dominance and monopolization to markets in which the defendant held shares of between 75 and 88 per cent.118 The result of these generous rules has been to create a bias in practice towards mobilizing indirect evidence in establishing a market power base. When high market shares are not identifiable, EU law again defaults to indirect, not direct, evidence. A ‘derivation’ method must be followed.119 Dominance can be evaluated on the basis of a ‘combination of several factors which, taken separately, are not necessarily determinative’.120 No statement of the Court appears to permit a showing of a market power base by adducing direct evidence. Occasionally, direct evidence has been used to corroborate a showing of market power based on indirect evidence.121 In Hoffmann-La Roche, the Court opened a narrow possibility for a defendant to invalidate indirect evidence of a market power position by presenting direct evidence. The Court held that ‘the fact that an undertaking is compelled by the pressure of its competitors’ price reductions to lower its own prices is in general incompatible with that independent conduct which is the hallmark of a dominant position’ – however, the Court denied the necessity of engaging in a full review of the facts to determine whether price reductions were a sign of competitive pressure, and held that this would be in any case ‘inconsistent with the existence of a dominant position’.122 As far as facts of market power injury are concerned, the situation is mixed. The case law has long been rife with statements whereby ‘concrete’ effects were not required. A showing of a ‘likely’ market power injury sufficed for Article 101 or 102 TFEU liability. The case law concept of likely effects also covered injuries that might have happened in the past. With this understanding, indirect evidence became the preferred proof of likely effects, including in cases where actual effects could have been established. More recently, however, the Court has straightened its case law on market power injury. The judgments in Post Danmark I and, more importantly, Intel 1 and 2, show that direct and indirect evidence can both establish proof of market power injury. In Post Danmark I, the Court has recommended the usage of direct price-costs tests.123 Intel 1 and 2 suggest focusing on indirect evidence of foreclosure, but recognize some potential application to the price cost test of the Guidance Paper on Article 102 TFEU.124 In market definition, indirect evidence is preferred to direct evidence. In Topps, the General Court held that a SSNIP test is ‘not the only method’.125 On appeal, the Court further noted ‘the absence of an obligation to carry out a SSNIP test which consists in a mental exercise presupposing a small (5 to 10%) but permanent variation’.126 With this, multiple factors bearing on product substitutability are to be considered. Antitrust lawyers like to illustrate the method that is applied with reference to United Brands. The case featured a dispute between United Brands and the Commission over whether the product market comprised all fresh fruits or whether

ibid para 56. ibid para 66. 120 ibid. 121 See the General Court in British Airways (n 34) para 218: with eyes on a monopsony power configuration, the Court observed that British Airways was in a position to ‘impose a reduction as from 1 January 1998 of its rates of commission in force’ (emphasis added). 122 Hoffmann-La Roche (n 30) para 71. 123 See Post Danmark I (n 46). 124 See Intel (n 46) para 129. 125 Case T-699/14 Topps Europe Ltd v European Commission ECLI:EU:T:2017:2, para 82. 126 ibid para 78. 118 119

Understanding market power: a legal perspective  79 there was a ‘specific demand for bananas’. In a famous passage, the Court held that ‘[t]he banana has certain characteristics, appearance, taste, softness, seedlessness, easy handling, a constant level of production which enable it to satisfy the constant needs of an important section of the population consisting of the very young, the old and the sick’.127 Today, it is clear that a quantitative test is not required. In many recent cases involving digital industries, the Court and agencies have found the good old qualitative method helpful. For indeed, a qualitative assessment appears less costly than the application of a SSNIP test in zero price markets and/or multisided context.

V. CONCLUSION Despite a common focus, important differences exist between the US and the EU antitrust laws’ treatment of market power. US antitrust law uses an economic conception of market power. EU antitrust law has developed a more idiosyncratic understanding of market power. Besides, US antitrust law grew into an explicit system of market power control, while EU competition law can be better characterized as a system of market share control. The legal implications of the different transatlantic focus on market power proper versus market power share are ambiguous. Economics suggests that market shares are uninformative of market power, and confounding in the presence of productive efficiency, technological innovation, or temporary shocks in the economy. That said, a focus on market shares offers one advantage. It reduces enforcement costs. The loss in diagnosis accuracy from a focus on market share might be compensated by gains of observability, and in turn, lower enforcement costs that are particularly suited in regimes with higher risk aversion and uncertainty (like perhaps the EU).128

United Brands (n 28) para 31. Nicolas Petit, ‘A Theory of Antitrust Limits’ (2021) 28 Geo Mason L Rev 1399.

127 128

PART II ABUSE OF DOMINANCE AND MONOPOLIZATION: CONDUCT AND THEORIES OF HARM

5. What is an abuse of a dominant position? Deconstructing the prohibition and categorizing practices Pablo Ibáñez Colomo

I. INTRODUCTION Drawing the line between the lawful and the unlawful exercise of substantial market power is a complex endeavour. Provisions dealing with abusive conduct come into play in circumstances where – due to the very presence of a dominant firm – effective competition is significantly weakened. As a result, practices that would otherwise be unproblematic may be subject to scrutiny. In this sense, there is by definition an element of ‘exceptionality’ about the scenarios to which these provisions apply. This point is effectively captured in, for instance, the idea of the ‘special responsibility’ of dominant firms, which is central to the European Union (EU) competition law system.1 On the other hand (and this adds to the complexity of the exercise), the vast majority of practices caught by these provisions are known to be capable of leading to pro-competitive gains. In addition, they do not necessarily or invariably have anticompetitive effects. Against this background, it is difficult to define legal tests that are not over- or underinclusive. Such tests risk being underinclusive, if they do not account for the fragile conditions of competition in which dominant players operate; and overinclusive, if they fail to acknowledge the nature and (pro- and anticompetitive) effects of the practice under examination. The task of defining the boundaries of the notion of abuse has been a source of challenges virtually since the inception of the discipline. Efforts in this sense can be broken down into three main dimensions. First, substantial resources have been devoted to the development of an overarching analytical framework for the evaluation of potentially abusive conduct, and this, on both sides of the Atlantic. The policy overhaul that would lead to the adoption of the European Commission’s Guidance Paper is an example.2 Second, it is commonplace to see discussions about the legal test that should apply to a specific practice – or set of practices. Just to mention a prominent example, much ink has been spilled, both in the EU and the United States (US), on the appropriate legal treatment of exclusivity obligations and rebate schemes applied by dominant firms.3 1 Case 322/81 NV Nederlandsche Banden Industrie Michelin v Commission (‘Michelin I’), EU:C:1983:313, para 57. See also Chapter 13 in this Handbook, which is devoted to the concept of ‘special responsibility’. 2 Guidance on the Commission’s enforcement priorities in applying Article 82 of the EC Treaty to abusive exclusionary conduct by dominant undertakings [2009] OJ C45/7. 3 See, among the countless contributions, Paul Nihoul, ‘The Ruling of the General Court in Intel: Towards the End of an Effect-based Approach in European Competition Law?’ (2014) 5 Journal of European Competition Law & Practice 521; Luc Peeperkorn, ‘Conditional Pricing: Why the General

81

82  Research handbook on abuse of dominance and monopolization The rise of the digital economy and the reorientation of enforcement priorities towards online platforms adds a third dimension to ongoing discussions. Some of the strategies implemented in the digital arena defy an easy categorization, in the sense that they do not fit neatly within the legacy frameworks and tests crafted over the years. The emerging questions relate to whether – and if so to what extent – the practices are comparable to pre-existing categories and to the definition of an appropriate legal framework. So-called self-preferencing conduct, whereby a vertically integrated platform (such as a search engine or a marketplace) treats more favourably its own activities than those of rivals on a neighbouring market, is arguably the single most salient example.4 This chapter addresses these three dimensions. First, it examines the various criteria that have been explored, in the case law and the literature, to tell abusive conduct apart from lawful behaviour. One conclusion from the analysis is that there is no such thing as an all-encompassing approach or test to identify abuses. Not all practices are inherently against the ideal of competition on the merits. Thus, assessing their lawfulness demands a case-by-case evaluation that considers a number of factors, including the relevant economic and legal context. In the absence of an all-encompassing approach, the legal conditions to establish a prima facie prohibition will depend on, inter alia, the nature of the practice and its potential to harm competition and yield pro-competitive gains. Second, the chapter considers the criteria to classify potentially abusive practices. The systematic analysis of the case law is typically structured, in the literature, around some categories (the most common of which include refusal to deal, exclusive dealing and rebates, tying or margin squeeze). In this regard, the chapter seeks to tease out the constituent elements of these various categories with a view to identifying the similarities and differences between them. The analysis focuses on several criteria, including whether the concern is with exploitation or exclusion, whether the practice is price-based or not, whether exclusionary concerns relate to the strengthening or leveraging of a dominant position as well as the remedies required to bring the infringement to an end. Structuring the debate around these constituent elements assists the analysis in a number of ways. In the first place, it is useful to tell one category apart from another. In the second place, it makes it possible to explain why the legal status of some practices is or has been contentious. Finally, it provides a template to navigate emerging strategies in digital markets and sheds light on the most appropriate criteria to evaluate their lawfulness. The analysis intends to be relevant across legal regimes, and thus addresses, irrespective of the jurisdiction, issues of principle as well as the fundamental choices to be made when interpreting provisions dealing with potentially prohibited conduct. This said, it builds on the case law and administrative practice developed over the years in the EU (in particular) and the US systems. Any references to abusive conduct in the text are used as a shorthand for breaches of both Article 102 Treaty on the Functioning of the European Union (TFEU) and Section 2 of the Sherman Act.

Court Is Wrong in Intel and what the Court of Justice Can Do to Rebalance the Assessment of Rebates’ (2015) 12 Concurrences 43; and Wouter Wils, ‘The Judgment of the EU General Court in Intel and the So-called “More Economic Approach” to Abuse of Dominance’ (2014) 37 World Competition 405. 4 Pablo Ibáñez Colomo, ‘Self-preferencing: Yet Another Epithet in Need of Limiting Principles’ (2020) 43 World Competition 417.

What is an abuse of a dominant position?  83

II.

CRITERIA TO DRAW THE LINE BETWEEN LAWFUL AND ABUSIVE CONDUCT

A.

What Is ‘Competition on the Merits’?

When addressing the notion of abuse, it is not unusual to find, both in the case law and the literature, references to the notion of ‘competition on the merits’.5 This expression is sometimes relied upon as a guide to distinguish between lawful and unlawful conduct. The legal assessment, pursuant to this approach, would involve ascertaining whether the practice under examination departs from legitimate efforts to gain an edge in the marketplace. If it does, it would be prohibited as abusive. In the EU, the Court of Justice (hereinafter, the ‘Court’ or the ‘CoJ’) has hinted at the use of this technique. In its landmark Hoffmann-La Roche ruling, it referred to the notion of abuse as concerning ‘methods different from those which condition normal competition’.6 Thus, it distinguished between loyalty rebates (which were deemed abusive) and quantity rebates – which were found to be a valid expression of competition on the merits.7 It is possible to discern similar attempts in the case law of the US Supreme Court. In Grinnell, for instance, Justice Douglas drew a line between the ‘wilful acquisition and maintenance of [monopoly] power’, on the one hand; and ‘growth or development as a consequence of a superior product, business acumen, or historic accident’, on the other.8 An approach that revolves around the notion of ‘competition on the merits’ is premised on the assumption that some practices are inherently ‘improper’ or anticompetitive, whereas others are pro-competitive (or, if one prefers, legitimate manifestations of rivalry). One can think of two (alternative or cumulative) reasons why practices could be deemed to be inherently at odds with competition on the merits. First, conduct could be considered abusive by its very nature where its objective purpose is anticompetitive. In Hoffmann-La Roche, for instance, the Court categorized exclusive dealing and loyalty rebates as abusive insofar as they were deemed to be ‘designed to deprive the purchaser of or restrict his possible choices of sources of supply and to deny other producers access to the market’.9 In other words, they were found to serve an exclusionary aim. Second, conduct may be said to be inherently abusive where it can be safely presumed to have anticompetitive effects irrespective of the context of which it is a part. A clean divide between ‘improper’ conduct and valid expressions of (normal) competition on the merits would fail to capture the majority of potentially abusive practices (and most behaviour scrutinized by competition authorities). What is more, the approach lacks explanatory power, in the sense that it is useful to make sense of just a fraction of the EU and US case law. A majority of practices routinely evaluated under Article 102 TFEU and Section 2 of the Sherman Act can be rationalized as pro-competitive strategies. Thus, they cannot be presumed to serve an ‘improper’, abnormal or exclusionary aim. Economic analysis shows that poten-

For an extensive discussion, see Konstantinos Stylianou, ‘Can Common Business Practices Ever Be Anticompetitive? Redefining Monopolization’ (2020) 57 American Business Law Journal 169. 6 Case 85/76 Hoffmann-La Roche & Co AG v Commission EU:C:1979:36, para 91. 7 ibid para 90. 8 United States v Grinnell Corp 384 US 563 (1966). For a similar attempt, see Aspen Skiing Co v Aspen Highlands Skiing Corp 472 US 585 (1985). 9 ibid. 5

84  Research handbook on abuse of dominance and monopolization tially abusive behaviour is, more often than not, a plausible means to attain legitimate objectives. Exclusive dealing and loyalty rebates, for instance, give certainty to the manufacturer and allow it to recover fixed costs.10 In addition, they may allow for more intense competition and address externalities.11 Some of these pro-competitive gains have been expressly acknowledged in the case law.12 The same can be said of other traditional categories, including price discrimination, tying and bundling13 as well as refusals to deal.14 In addition, all these practices are commonplace in a market economy. There is therefore nothing abnormal or exceptional in them. Experience shows, for instance, that exclusive dealing and tying are widespread and as such not a manifestation of substantial market power.15 The same can be said, in general, of unilateral conduct, including refusals to deal.16 Similarly, potentially abusive practices do not invariably have anticompetitive effects, in the sense that they are not necessarily restrictive in all scenarios. At most, one can claim that they have, in the abstract, an exclusionary and/or exploitative potential. It is not possible to assume that they will have, or are very likely to have, a restrictive impact in a specific context. Decades of economic analysis suggest that anticompetitive effects can be expected only where certain conditions are fulfilled. Accordingly, it is necessary to formulate a theory of harm (that is, the mechanism through which the impact would be manifested) and examine the extent to which the available facts support the theory in the specific circumstances of the case.17 The conclusion that follows, from a legal standpoint, is that a case-by-case assessment is necessary to consider the probable restrictive impact of the practice in a given economic and regulatory scenario. Just to mention an example, the anticompetitive effects of exclusive dealing are likely where the dominant supplier deals with a multitude of relatively small buyers.18 As can be seen from the above, most practices – if not the vast majority of them – are not inherently abusive and cannot be presumed to be part of an exclusionary or exploitative strategy or to lead, inevitably, to anticompetitive outcomes. Accordingly, whether or not they are prohibited necessitates a context-specific assessment of their object and/or effect. In other words, conduct may or may not go against the ideal of competition on the merits depending on its nature and the circumstances in which it is implemented. The very same behaviour (say,

See, for an overview, EAGCP, ‘An Economic Approach to Article 82’ (July 2005), at http://​ec​ .europa​.eu/​dgs/​competition/​economist/​eagcp​_july​_21​_05​.pdf, accessed 1 November 2022. 11 Guidelines on vertical restraints [2010] OJ C130/1, para 107. 12 Case C-234/89 Stergios Delimitis v Henninger Bräu AG EU:C:1991:91, paras 10−12. 13 Michael D Whinston, ‘Tying, Foreclosure, and Exclusion’ (1990) 80 American Economic Review 837. 14 Einer Elhauge, ‘Defining Better Monopolization Standards’ (2003) 56 Stanford Law Review 253, 295−8. 15 See in this sense Delimitis (n 12), which involved exclusive dealing obligations implemented by a firm with a 6.4 per cent market share. As far as tying is concerned, the US Supreme Court introduced a market power requirement to filter the instances in which the practice is subject to scrutiny under US antitrust law. See in particular Jefferson Parish Hospital District No 2 v Hyde 466 US 2 (1984). 16 Phillip Areeda, ‘Essential Facilities: An Epithet in Need of Limiting Principles’ (1989/90) 58 Antitrust Law Journal 841, 844, where the author notes that ‘unilateral action’ (including refusal to deal) is ‘omnipresent’. 17 On theories of harm, generally, see Hans Zenger and Mike Walker, ‘Theories of Harm in European Competition Law: A Progress Report’ in Jacques Bourgeois and Denis Waelbroeck, Ten Years of Effects-based Approach in EU Competition Law: State of Play and Perspectives (Bruylant 2012). 18 Eric B Rasmusen and others, ‘Naked Exclusion’ (1991) 81 American Economic Review 1137. 10

What is an abuse of a dominant position?  85 a standardized rebate scheme) may amount to an infringement in a given economic and legal scenario and escape the prohibition in a different one. In many instances, whether or not there is a breach of the provision depends on factors such as the features of the relevant market as well as the specific nature of the conduct and of the product. In fact, what has marked the evolution of Article 102 TFEU and Section 2 of the Sherman Act is the realization that only a small fraction of practices can be safely presumed to be ‘improper’ by their very nature. The prime example of conduct that is inherently at odds with competition on the merits is pricing below average variable costs. As explained by Areeda and Turner in a seminal piece, it is in principle irrational for a dominant firm (or indeed any firm) to price at that level.19 Accordingly, where such behaviour is observed, a court or authority can ground their analysis on the premise that it serves an exclusionary purpose.20 Other examples include the strategic use by a dominant firm of administrative and judicial processes with a view to foreclosing rivals or exploiting its substantial market power. For instance, a dominant firm may provide misleading information to a patent authority with the sole aim of extending its patent (and thus delaying any prospect of potential competition).21 One can identify, in the EU legal order (but not the US one), a divide between conduct that is deemed to be inherently at odds with competition on the merits (and as such prima facie prohibited) and conduct that demands a case-by-case assessment of its impact in the relevant economic context. It is indeed possible to distinguish, under Article 102 TFEU, between practices that are presumptively abusive irrespective of their impact and those that are only subject to the prohibition where they have, or are likely to have, anticompetitive effects. Underlying the stricter legal treatment of the former there is a presumption that they have an anticompetitive object. There is, thirdly, conduct that is prima facie lawful, irrespective of its effects, under Article 102 TFEU. The latter is presumed to serve a pro-competitive purpose and as such is deemed to be a valid expression of competition on the merits. Thus, and in line with what has been suggested above, pricing below average variable costs is prima facie prohibited, irrespective of its effects, in the EU system.22 The same is true of prices above average total costs where there is specific evidence that below-cost pricing is part of a strategy to drive a rival out of the market.23 More controversially, practices such as exclusive dealing and loyalty rebates (already mentioned above),24 as well as tying, are also presumptively prohibited.25 In any event (and just like in the US), the majority of potentially abusive conduct is only caught by Article 102 TFEU following a case-by-case assessment of its likely impact on competition. Accordingly, whether or not such conduct amounts to competition on the merits depends on an evaluation that takes into consideration not only the economic and legal context of which the practice is a part but also whether there is evidence

19 Phillip Areeda and Donald F Turner, ‘Predatory Pricing and Related Practices under Section 2 of the Sherman Act’ (1975) 88 Harvard Law Review 697. 20 ibid 712. 21 Case C-457/10 P AstraZeneca AB and AstraZeneca plc v Commission EU:C:2012:77. 22 Case C-62/86 AKZO Chemie BV v Commission EU:C:1991:286, para 71. 23 ibid para 72. 24 Hoffmann-La Roche (n 6) para 90 and Case C-413/14 P Intel Corp v Commission EU:C:2017:632, para 137. 25 See in particular Case T-30/89 Hilti AG v Commission EU:T:1991:70 and Case T-83/91 Tetra Pak International SA v Commission (‘Tetra Pak II’) EU:T:1994:246.

86  Research handbook on abuse of dominance and monopolization of an anticompetitive strategy.26 As explained in greater detail below, there are some cases in which the threshold of intervention is even stricter. B.

The Vain Quest for an All-Encompassing Test

Given how challenging it is to define criteria to draw the line between lawful and unlawful conduct, it is not surprising that attempts to devise an all-encompassing test to identify abuses have proved unfruitful. These efforts are unlikely to ever capture the diversity and richness of potentially anticompetitive practices. Two main factors explain the futility of the exercise. First, some of the tests proposed only seem suitable for certain conduct, and more precisely behaviour that is inherently anticompetitive. If they were to be applied across the board, they would lead to underenforcement in a manner that would not be obvious to justify. From a purely positive standpoint, moreover, such tests lack explanatory power. Second, other approaches are not administrable and as such fail to assist context-specific inquiries in individual cases. More than a test, the latter provide an abstract benchmark that gives a sense of the aims of enforcement at a high level. Two approaches that have been consistently discussed in the literature and in authorities’ reports are the ‘profit sacrifice’ and the ‘no economic sense’ tests. Both share a common philosophy and are sometimes presented as two manifestations of the same idea.27 In essence, they seek to evaluate, by proxy, whether a given practice has an exclusionary aim. In accordance with the profit sacrifice test, a practice would be abusive where it entails, for the dominant firm, a (short-term) profit sacrifice relative to another practice that would not have the same anticompetitive effects.28 In other words, where the conduct is less profitable than another conduct that would not lead to the same outcome, it would be caught by the prohibition. Under the no economic sense test, the question is whether the practice can be explained on grounds other than the restriction of competition.29 Thus, conduct would amount to an infringement where its only plausible purpose is anticompetitive. The test would be administered by asking whether the practice brings any benefit to the firm other than the anticompetitive effects resulting from it. Courts and authorities have occasionally relied on these tests to draw the line between abusive and lawful conduct. It is apparent, from the discussion above, that the AKZO ruling applied a version of the no economic sense test to identify predatory pricing within the meaning of Article 102 TFEU.30 Lithuanian Railways (in which the dominant firm sacrificed

See in particular Case C-209/10 Post Danmark A/S v Konkurrencerådet (‘Post Danmark I’) EU:C:2012:172 and Case C-23/14 Post Danmark A/S v Konkurrencerådet (‘Post Danmark II’) EU:C:2015:651. 27 See, for instance, Renato Nazzini, The Foundations of European Union Competition Law: The Objective and Principles of Article 102 (Oxford University Press 2012), 66. 28 US Department of Justice, ‘Competition and Monopoly: Single-Firm Conduct under Section 2 of the Sherman Act’ (2008) 39. 29 ibid. 30 AKZO (n 22) para 71: ‘[a] dominant undertaking has no interest in applying such prices except that of eliminating competitors so as to enable it subsequently to raise its prices by taking advantage of its monopolistic position, since each sale generates a loss, namely the total amount of the fixed costs (that is to say, those which remain constant regardless of the quantities produced) and, at least, part of the variable costs relating to the unit produced’. 26

What is an abuse of a dominant position?  87 profits to dismantle 19 kilometres of railway with a view to excluding a rival) provides another example.31 There are traces of these tests in US case law, too. A ruling in point is Aspen Skiing, which reached the US Supreme Court. A crucial factor in Justice Stevens’s Opinion is that the dominant firm was willing to sacrifice profits by refusing to deal with its rival (which was willing to pay the retail price of the relevant services).32 For the reasons explained above, these tests capture the logic underpinning intervention in only in a small fraction of cases. More precisely, they are only capable of identifying instances in which a practice is inherently at odds with competition on the merits. As a result, the administration of these tests across the board would lead to the systematic underenforcement of the relevant provisions (at least when compared to the range of conduct that is known from experience to be potentially problematic). One should note, in this regard, that exclusionary behaviour does not necessarily entail a profit sacrifice for the dominant firm.33 More generally (and in line with what has already been mentioned), most conduct caught by provisions on abuse of dominance is known to be plausibly pro-competitive, in the sense that it has an economic rationale beyond its restrictive impact. The consumer welfare test has been occasionally proposed as an all-encompassing approach to the identification of abuses. Pursuant to this test, a practice would be prohibited where its net impact on consumer welfare is negative. It is an approach that demands a balancing of the pro- and anticompetitive dimensions of behaviour to determine whether the latter weighs more than the former, or vice versa. The advantage of the test is that lawfulness would be evaluated against a benchmark that has a clear meaning and has been formalized in the economic literature.34 Its fundamental disadvantage has to do with the fact that it is not easily administrable. The case-by-case balancing of the positive and negative dimensions of a practice is known to be a complex exercise that demands considerable resources and is exceedingly unlikely to assist courts and authorities in individual instances. In reality, reliance on consumer welfare is useful, if at all, as a benchmark rather than as a test. In other words, more than a set of criteria to evaluate, in a specific case, whether a practice amounts to an abuse, the impact on consumers can be valuable as a high-level aspiration, which courts and authorities may choose as the overarching goal of enforcement. The same can be said of the ‘as efficient competitor’ test. According to the latter, the legality of conduct would be assessed by reference to a competitor that is at least as efficient as the dominant firm. This test lacks the requisite precision to be operational in specific cases. It needs to be fleshed out to account for the nature of the practice and the underlying concerns. In other words, rather than a test, it operates as a principle that informs the criteria against which specific issues are assessed and that warns against intervention that might have the unintended effect of subsidizing some firms at the expense of consumers and society at large. Understood in this sense, the ‘as efficient competitor’ principle plays a useful role in the evaluation of the lawfulness of Case T‑814/17 Lietuvos geležinkeliai AB v Commission EU:T:2020:545. Aspen Skiing Co v Aspen Highlands Skiing Corp 472 US 585 (1985). 33 Steven C Salop and David T Scheffman, ‘Raising Rivals’ Costs’ (1983) 73 American Economic Review 267. 34 See in this sense Carl Shapiro, ‘The Consumer Welfare Standard in Antitrust: Outdated, or a Harbor in a Sea of Doubt?’ (US Senate Judiciary Committee – Subcommittee on Antitrust, Consumer Protection and Consumer Rights, 13 December 2017) and Steven C Salop, ‘Question: What Is the Real and Proper Antitrust Welfare Standard? Answer: The True Consumer Welfare Standard’ (2010) 22 Loyola Consumer Law Review 336. 31

32

88  Research handbook on abuse of dominance and monopolization some practices. For instance, above-cost pricing is typically deemed lawful insofar as it is not capable of excluding equally efficient rivals.35 C.

Defining the Scope of the Notion of Abuse: Key Choices

Table 5.1

The scope of the notion of abuse in the EU and US systems EU

US

Exploitative and/or exclusionary?

Both

Exclusionary

Dominance as a prerequisite?

Yes

No

No

No

Yes

Yes

Equally efficient rivals

Consumer welfare

Both

Both

Yes

Yes

Causal link between market power and practice? Effects (established or presumed) required? Benchmark of effects Bright-line rules or case-by-case assessment? Defences available?

The boundaries of the notion of abuse depend, to a significant extent, on the choices made when shaping it. These are summarized in Table 5.1. Substantive choices delineating the outer limits of the prohibition inform how the lawfulness of specific practices is evaluated and which conduct falls within its scope of application. The first and arguably most obvious one relates to whether the relevant provisions capture exploitative behaviour (that is, the exercise of market power vis-à-vis customers and/or suppliers, in the form of, inter alia, excessive pricing and, more generally, the setting of unfair terms and conditions) or exclusionary practices (that is, the foreclosure of actual or potential competitors) alone. As seen in Table 5.1, the gulf between the US and the EU systems on this point is arguably the most significant difference between the two regimes.36 A second choice – and another key difference between the EU and US systems – is whether a finding of dominance is a precondition to trigger the prohibition. The question, in other words, is whether the relevant provisions are applicable only to dominant firms or whether they also come into play prior to firms’ acquiring such a position. Whereas Section 2 of the Sherman Act comprises both the monopolization and the attempt to monopolize a market,37 Article 102 TFEU only applies to firms that already have a dominant position. This divergence came to the fore in Rambus.38 As a matter of principle, the US system was capable of addressing the Areeda and Turner (n 19) 705−6. See also Elhauge (n 14). That Article 102 TFEU covers exploitative practices has been clear from the outset. See in this sense René Joliet, Monopolization and Abuse of Dominant Position: A Comparative Study of the American and European Approaches to the Control of Economic Power (Martinus Nijhoff 1970). That the US system does not encompass such conduct is also clear. See in this sense Verizon Communications v Law Offices of Curtis V Trinko, LLP 540 US 398 (2004), 7: ‘[t]he mere possession of monopoly power, and the concomitant charging of monopoly prices, is not only not unlawful; it is an important element of the free-market system’. 37 US Department of Justice (n 28) 5−7. See also Chapter 14 in this Handbook. 38 See Rambus Inc v FTC 522 F3d 456, 469 (DC Cir 2008) and Rambus (Case COMP/38.636) Commission Decision of 9 December 2009. 35 36

What is an abuse of a dominant position?  89 practice at stake in the case, which took place prior to the firm acquiring a monopoly.39 In the EU, only the behaviour that follows the acquisition of a dominant position can be scrutinized.40 A third choice concerns the causal link between the dominant position and the behaviour. The question, in this regard, is whether the potentially abusive practice must be a manifestation of the substantial market power enjoyed by the firm (for instance, its ability to impose exclusivity obligations to its customers, or to condition the sale of one product to the acquisition of another one) or whether it should be independent from it. If the latter approach is followed, and thus no causal link is required, conduct such as the acquisition of an actual or potential competitor, or bringing proceedings against another firm, can amount to an abuse. This issue is settled on both sides of the Atlantic. Under both regimes, a potential abuse need not flow from the dominant position for it to be caught by the prohibition.41 The fourth choice that shapes the notion concerns the issue of anticompetitive effects, and more precisely the role they play in the system and how they are defined. The issue can be broken down into three separate sub-questions. To begin with, it is necessary to determine whether the prohibition applies only to conduct that is at least a plausible source of restrictive effects in the relevant context or whether, instead, a finding of abuse is independent of its actual or potential impact. The fact that effects are a precondition to trigger the prohibition does not mean that they must necessarily be established by an authority or claimant on a case-by-case basis. The restrictive impact of a practice, even if required, can be presumed. In such cases, it would be for the dominant firm to show why, in the relevant economic and legal context, anticompetitive effects are implausible or unlikely to occur. In addition (this is the second sub-question), it would be necessary to identify the applicable threshold of effects. There is a difference, in theory and practice, between requiring evidence that effects are plausible and that they are likely, or certain, to occur.42 The final sub-question relates to whether a causal link between the behaviour and the effects is required. On each of these points, there seems to be significant alignment between the EU and the US systems. In particular, conduct must – at the very least – be capable of restricting competition for the prohibition to be triggered in both legal orders.43 Thus, where effects are presumed (as is true of prima facie unlawful conduct in the EU system), a dominant firm can provide evidence revealing their absence (or implausibility). Another similarity is that any restriction of competition

For a discussion, see Joel M Wallace, ‘“Rambus v. F.T.C.” in the Context of Standard-setting Organizations, Antitrust, and the Patent Hold-up Problem’ (2009) 24 Berkeley Technology Law Journal 661. 40 For a critical analysis of the decision (and whether it exceeded the boundaries of Article 102 TFEU), see Ryan Stones, ‘Commitment Decisions in EU Competition Enforcement: Policy Effectiveness v. the Formal Rule of Law’ (2019) 38 Yearbook of European Law 361. 41 The Court confirmed that Article 102 TFEU captures acquisitions (which do not flow from the market power enjoyed by the firm) in Case 6/72 Europemballage Corporation and Continental Can Company Inc v Commission, EU:C:1973:22. For a discussion of the US system on this point, see C Scott Hemphill and Tim Wu, ‘Nascent Competitors’ (2020) 168 University of Pennsylvania Law Review 1879. 42 For a discussion, see Pablo Ibáñez Colomo, ‘Anticompetitive Effects in EU Competition Law’ (2021) 17 Journal of Competition Law & Economics 309. 43 See in this sense Case C-307/18 Generics (UK) Ltd and Others v Competition and Markets Authority EU:C:2020:52, para 154 (‘if such conduct is to be characterised as abusive, that presupposes that that conduct was capable of restricting competition and, in particular, producing the alleged exclusionary effects […]’; and US v Microsoft Corporation 253 F3d 34 (DC Cir 2001). 39

90  Research handbook on abuse of dominance and monopolization must be attributable to the conduct.44 The threshold of effects, in turn, may vary depending on the nature of the practice and/or its potential to cause harm.45 A fifth choice relates to the benchmark against which anticompetitive effects are assessed. In line with what has already been pointed out, the impact of a practice can be assessed by reference to, inter alia, its effects on consumers, on the market structure or on equally efficient rivals. According to a consistent line of case law, it would seem that the EU competition law system is concerned – at least as a matter of principle – with the impact of potentially abusive conduct on equally efficient competitors.46 This principle has two implications. In the first place, it is not necessary for a court or an authority to establish harm to consumers for Article 102 TFEU to come into play.47 In the second place, the exclusion of rivals that are less efficient than the dominant firm is considered to be the logical and expected outcome of the competitive process and as such unproblematic.48 The US system, on the other hand, has long been described by the Supreme Court as a ‘consumer welfare prescription’, and Section 2 of the Sherman Act is not an exception.49 A sixth choice, already mentioned, relates to whether practices are deemed abusive (or lawful) by their very nature or following an assessment of their anticompetitive effects. As explained above, the EU system combines both approaches. Thus, some practices are treated as abusive by their very nature. The rest are either prima facie lawful or prohibited following a case-by-case evaluation of their likely restrictive impact. Conversely, a finding of effects appears to be a precondition for intervention in the US in all cases (with the sole exception, more theoretical than real, of tying, discussed below). Finally, the question arises of whether it is possible for the dominant firm to advance a defence for its conduct. The defence may relate to the objective rationale for the behaviour or to the fact that the pro-competitive gains to which it gives rise are sufficient to outweigh any actual or potential anticompetitive effects. There seems to be little doubt that the regimes on both sides of the Atlantic allow for a defence along these lines, even for conduct that is deemed abusive prima facie.50

III.

CLASSIFYING POTENTIALLY ABUSIVE PRACTICES

Typically, the analysis of potentially abusive conduct revolves around some well-established categories, several of which have already been mentioned. They include exclusive dealing, tying and refusal to deal. This section considers, systematically, the criteria on the basis of which the distinctive features of each practice can be classified. The exercise is useful for three

Post Danmark II (n 26) para 47; and US Microsoft (n 43). For a discussion, see Ibáñez Colomo (n 42). 46 Post Danmark I (n 26) para 22; and Intel (n 24) para 134. 47 Case C-95/04 P British Airways plc v Commission EU:C:2007:166, paras 103−8. 48 Post Danmark I (n 26) para 22; and Intel (n 24) para 134 (‘[c]ompetition on the merits may, by definition, lead to the departure from the market or the marginalisation of competitors that are less efficient and so less attractive to consumers from the point of view of, among other things, price, choice, quality or innovation’). 49 See Daniel A Crane, ‘The Tempting of Antitrust: Robert Bork and the Goals of Antitrust Policy’ (2014) 79 Antitrust Law Journal 835. 50 See in this sense Intel (n 24) para 140. On the availability of defences in the US system, see US Microsoft (n 43). 44 45

What is an abuse of a dominant position?  91 main reasons. First, and most obviously, it makes it easier to explain and rationalize the similarities and differences between them. Second, it allows for the identification of tensions and potential contradictions in the case law: like practices are not necessarily treated alike (or not always) by courts and authorities. Finally, the exercise helps makes sense of the best approach to evaluate the lawfulness of the sort of conduct that defies an easy categorization. A.

The Criteria to Differentiate between the Various Categories

There are virtually infinite ways in which potentially abusive practices can be distinguished from one another. A handful of criteria are, however, particularly helpful when telling apart the key similarities and differences between them. Some of these are applicable across all practices. The first and most obvious criterion is the summa divisio that exists between exploitative and exclusionary conduct, which has already been mentioned above. A second across-the-board criterion along which conduct can be categorized revolves around whether the practice is price-based or not, that is, whether the instrument through which the actual or potential effects would be manifested involves the use of prices. From this perspective, a scheme of loyalty rebates would be a price-based practice, whereas an exclusive dealing obligation, whereby a dominant supplier demands its customers not to sell products competing with its own, would not.51 Third, one can differentiate practices based on whether they are a plausible means to attain pro-competitive gains. Put differently, one can draw a line between ‘naked’ abuses – which serve no plausible purpose other than the restriction of competition – and conduct for which there is a potential pro-competitive justification. In line with what has already been explained, pricing below average variable cost is the archetypal example of a ‘naked’ abuse, in the sense that it can only be explained as a means to exclude a rival. Providing misleading information to patent authorities to extend the length of a patent is another one. As already mentioned, there are, on the other hand, practices (in fact, a majority thereof) that are a credible means to achieve pro-competitive gains, such as exclusive dealing, tying or refusal to deal. A final overarching criterion has to do with the nature of the remedy required to bring the infringement to an end. More precisely, one may differentiate between remedies of a reactive and a proactive nature.52 Reactive remedies amount to an obligation of a negative nature that is administered on a one-off basis. Competition law is typically associated with intervention along these lines. There are, on the other hand, remedies that are proactive. Where proactive action relies on behavioural intervention, it amounts to the imposition of a positive obligation, the monitoring of which is typically resource-consuming. For instance, a dominant firm may be required to deal with rivals, and this, on fair, reasonable and non-discriminatory terms and conditions.53 Proactive remedies may also involve structural intervention. In such instances, the dominant firm may be required to agree the divestiture of some of its assets. For instance, 51 It is not unusual for leading commentators to rely on this divide. See for instance Richard Whish and David Bailey, Competition Law (9th edn, Oxford University Press 2018), chs 17 and 18. For a critical perspective, see Joshua D Wright, ‘Simple but Wrong or Complex but More Accurate? The Case for an Exclusive Dealing-based Approach to Evaluating Loyalty Discounts’ (Bates White 10th Annual Antitrust Conference, Washington, DC, 3 June 2013). 52 For a taxonomy of remedies, see Pablo Ibáñez Colomo and Andriani Kalintiri, ‘The Evolution of EU Antitrust Policy: 1966–2017’ (2020) 83 Modern Law Review 321. 53 Article of 5 Microsoft (Case COMP/C-3/37.792) Commission Decision of 24 March 2004.

92  Research handbook on abuse of dominance and monopolization a vertically integrated electricity company may be required to sell its transmission network to a third party.54 Other criteria apply specifically to potentially exclusionary conduct. One may identify, in this regard, a summa divisio between, first, conduct that has the object and/or effect of strengthening a dominant position and, second, behaviour that aims and/or leads to the extension (or leveraging) of such a position to a neighbouring market. In the first scenario (strengthening of dominance), only one market (say, market A, in which the firm enjoys substantial market power) would be affected by the practice. For instance, a dominant supplier on a given market may reinforce its status on the relevant market by means of predatory pricing or exclusive dealing. In the second scenario (leveraging of dominance), the practice would encompass (at least) two markets: one (market A) in which the abuse takes place and a second one (market B) in which the effects are manifested. For instance, a dominant supplier of cola-flavoured carbonated drinks may condition the sale of this product to the acquisition of another product (say, tonic water). It has long been established in the academic literature that leveraging practices may have an ‘offensive’ or a ‘defensive’ object or effect. In other words, the aim or impact of the conduct may be the extension of a dominant position to a neighbouring market (market B) – in which case the practice would qualify as ‘offensive’ – or may instead be the protection of the dominant position on the primary market (market A) – in which case it would qualify as ‘defensive’. Leveraging may be defensive where, for instance, firms operating in market B are a potential threat to the dominant firm’s position on market A, in the sense that they have the potential to undermine its market power. Such was the issue at the heart of the US Microsoft case.55 Irrespective of whether the leveraging is ‘offensive’ or ‘defensive’, it will be implemented in the same way, that is, by means of the exclusion of a rival on the secondary market.56 Finally, and as far as leveraging conduct is concerned, one may distinguish cases based on the relationship between the relevant markets. In some cases, the relationship between markets is vertical, with an upstream and a downstream level. For instance, the upstream market (market A, in which the practice is implemented) may concern an input that is incorporated into a finished product, which is part of market B (where the effects of the practice are manifested). In other cases, the relationship between markets is horizontal in nature, in the sense that the two markets are at the same level of the value chain. Such would be the case where two products are complements. In this sense, one can think, for instance, of two applications running on the same operating system. One can also think of separate products sharing a common pool of customers, such as a cola-flavoured carbonated drink and tonic water. The difference in the relationship between markets – horizontal or vertical – is not inconsequential and provides valuable insights about the likely concerns at stake in a given case. Typically, where the relationship between two products is horizontal, competitors are not in a position in which they are, in addition, customers and/or suppliers of the dominant firm. Where, conversely, the relationship between markets is vertical, rivals are likely to have such dual status. They may receive, as customers, an input from a vertically integrated supplier with which they compete downstream. Alternatively, they may compete upstream with the same 54 German Electricity Balancing Market (Case COMP/39.389) Commission Decision of 26 November 2008. 55 US Microsoft (n 43). 56 Guidance (n 2) para 52.

What is an abuse of a dominant position?  93 firm that operates a platform on which they rely. As a result of the dual status of firms as rivals and customers (or suppliers), intervention will often involve the regulation of the terms and conditions under which vertically integrated firms deal with competitors. In some cases, remedial action may demand that the dominant operator start dealing with would-be competitors. B.

The Main Categories of Potentially Abusive Conduct

It is possible to reconstruct, on the basis of the criteria identified in the preceding subsection, the most frequent categories of potentially abusive conduct. These categories are summarized in Table 5.2. If one focuses on exclusionary conduct, it appears, first, that a practice like predatory pricing, as generally understood (including in the case law), is characterized by the fact that (i) it is a price-based strategy that (ii) is not plausibly pro-competitive, (iii) seeks to strengthen a dominant position and (iv) can be remedied by means of a reactive remedy (namely a cease-and-desist order, which would be enough to bring the infringement to an end). This behaviour stands in contrast with other conduct aimed at strengthening a dominant position, such as rebate schemes and exclusive dealing. A fundamental difference between the latter and predatory pricing – and the key one for the purposes of this analysis – relates to the fact that, as already mentioned, they are known to be a source of pro-competitive benefits. As far as leveraging conduct is concerned, practices may be distinguished on the basis of two dimensions. The fundamental difference between tying and (mixed) bundling, on the one hand, and refusal to deal and margin squeeze abuses, on the other, has to do with the relationship between the relevant markets. The former (at least as traditionally understood) involve horizontally related activities. Thus, the products involved in traditional tying cases may share a common pool of customers. Suffice it to mention, coming back to the example above, a supplier conditioning the sale of cola-flavoured carbonated drink to the acquisition of tonic water by the customer.57 Some traditional tying cases involve complements – think of nails and nail guns, as in Hilti;58 or of salt and a salt processor, as in International Salt.59 It is possible to differentiate between leveraging cases along a different dimension, which is the nature of the remedy. The legal status of refusal to deal has long been controversial because of the draconian consequences of intervention: a duty to deal with third parties60 (that is, the quintessential example of a proactive remedy). Because of its implications from a substantive and an institutional standpoint, forcing a firm to conclude a contract with a would-be rival cannot be likened with the remedies that ban the joint sale of two products (as in traditional instances of tying) or that relate to the conditions under which a firm deals with third parties with which it is already dealing (as is true in the context of a margin squeeze). This difference has been emphasized by courts on both sides of the Atlantic.61

See for instance Coca-Cola (Case COMP/A.39.116/B2) Commission Decision of 22 July 2005. Hilti (n 25). 59 International Salt Co v United States, 332 US 392 (1947). 60 Opinion of AG Jacobs in Case C-7/97 Oscar Bronner GmbH & Co KG v Mediaprint Zeitungs- und Zeitschriftenverlag GmbH & Co KG and others EU:C:1998:264 and Trinko (n 36). 61 Case C-165/19 P Slovak Telekom, as v Commission EU:C:2021:239 and Trinko (n 36). 57 58

94  Research handbook on abuse of dominance and monopolization Table 5.2 Practice

Main categories of potentially abusive conduct Exploitation or

Price-based

exclusion Excessive

Plausibly

Strengthening or Remedy

pro-competitive

leveraging

Relationship between markets

Exploitation

Yes

 

 

 

 

Exclusion

Yes

No

Strengthening

Reactive

 

Exclusion

No

Yes

Strengthening

Reactive

 

pricing Predatory pricing Exclusive dealing Rebates

Exclusion

Yes

Yes

Strengthening

Reactive

 

Tying and

Exclusion

No

Yes

Leveraging

Reactive

Horizontal

bundling Mixed Bundling

Exclusion

Yes

Yes

Leveraging

Reactive

Horizontal

Refusal to deal

Exclusion

No

Yes

Leveraging

Proactive

Vertical

‘Margin

Exclusion

Yes

Yes

Leveraging

Reactive

Vertical

squeeze’

C.

Legal Tests and the Path of the Law

The systematic presentation of the main categories of potentially abusive conduct, summarized in Table 5.2, provides a valuable starting point to make sense of the controversies around the notion of abuse and of the frictions and tensions that exist in the case law. As a preliminary point, and in line with the ideas that have already been introduced, one can identify four main tests against which the lawfulness of conduct can be assessed. These tests are intended to adapt the legal status of the practice to its nature (and, more precisely, the plausibility of a pro-competitive rationale) and potential for harm (or likelihood of anticompetitive effects). Insofar as they are, they seek to minimize errors. The diversity of tests makes it possible to place practices along a spectrum that ranges from the prima facie lawful to the prima facie unlawful. First, conduct that is not capable of having anticompetitive effects, such as above-cost pricing and volume-based rebates, is prima facie lawful in the US and EU systems.62 In the same vein, these practices are presumed to be a valid manifestation of competition on the merits. Aggressive above-cost pricing, for instance, is the sort of behaviour that is expected from firms subject to competitive pressure. Similarly, quantity rebates that are transaction-specific and that reflect the cost savings made by the supplier are beyond reproach, as they are an expression of the dominant firm’s ability to exploit its scale advantage.63 Second, some practices – typically those that serve no plausible purpose other than the restriction of competition – are deemed prima facie unlawful in the EU. As already explained above, pricing below average variable costs falls within this category.64 Third, other practices – typically those that are capable of yielding pro-competitive or at least ambivalent effects on competition – are subject to a case-by-case analysis of their restric See Brooke Group Ltd v Brown & Williamson Tobacco Corp 509 US 209 (1993); and Einer Elhauge, ‘Why Above-cost Price Cuts to Drive out Entrants Are Not Predatory − and the Implications for Defining Costs and Market Power’ (2003) 112 Yale Law Journal 826. 63 Hoffmann-La Roche (n 6) para 90. 64 But not in the US, see in this sense Brooke Group (n 62). 62

What is an abuse of a dominant position?  95 tive impact. In the EU, conduct that demands such an assessment includes margin squeeze behaviour65 and standardized rebate schemes.66 In the US, it is the default approach to the evaluation of potentially anticompetitive conduct. The likelihood of anticompetitive effects depends on a number of considerations, including, in particular, the nature of the behaviour, its coverage and potential for harm, the extent of the dominant position and the features of the relevant market.67 For instance, a restriction of competition is more probable where a margin squeeze involves an indispensable input for downstream rivals.68 Similarly, the effects of standardized rebates depends on factors such as their duration and amount, on the one hand, and the share of the market covered by them, on the other.69 Finally, there are practices that are only deemed abusive in ‘exceptional circumstances’, to be established on a case-by-case basis. Accordingly, a finding of a prima facie infringement would need to satisfy conditions that are more demanding than simply showing the likelihood of anticompetitive effects. This range of conduct can be easily discerned in the EU system. Thus, a refusal to license an intellectual property right is only prohibited where the right in question would be indispensable for competition on a neighbouring market and the practice would prevent the emergence of a new product for which there is potential consumer demand and would lead to the elimination of all competition.70 In relation to tangible property, a refusal to deal would be caught by Article 102 TFEU where, at the very least, it relates to an indispensable input or platform and leads to the elimination of all competition.71 Tensions around the applicable legal test in specific scenarios arise for a number of reasons. The three main reasons are discussed hereinafter. Tensions may emerge, to begin with, where there is a mismatch between the nature and anticompetitive potential of a practice and its legal status. For instance, conduct such as exclusive dealing, loyalty rebates and tying is deemed prima facie abusive, irrespective of its effects, under Article 102 TFEU.72 In the same vein, tying is (at least so nominally) a per se breach of Section 2 of the Sherman Act.73 To the extent that these practices are capable of generating pro-competitive gains and do not invariably cause restrictive effects, they do not seem to be inherently at odds with competition on the merits. For the same reason, it would be natural to require a case-by-case assessment of their restrictive impact before intervening in a given case. This mismatch, which is also manifested in the case law,74 has been consistently contested by commentators.75

Case C-280/08 P Deutsche Telekom AG v Commission EU:C:2010:603, paras 250−51. Post Danmark II (n 26) paras 39−46. 67 ibid. See also Intel (n 24) para 139. 68 Case C-52/09 Konkurrensverket v TeliaSonera Sverige AB EU:C:2011:83, paras 70−71. 69 Intel (n 24) para 139. 70 Joined Cases C-241/91 P and C-242/91 P Radió Telefís Éireann (RTÉ) and Independent Television Publications Ltd (ITP) v Commission (‘Magill’) EU:C:1995:98; and Case C-418/01 IMS Health GmbH & Co OHG v NDC Health GmbH & Co KG EU:C:2004:257. 71 Case C-7/97 Oscar Bronner GmbH & Co KG v Mediaprint Zeitungs- und Zeitschriftenverlag GmbH & Co KG and others EU:C:1998:569. 72 See above nn 24 and 25. 73 Jefferson Parish Hospital District No 2 v Hyde 466 US 2 (1984); and Illinois Tool Works Inc v Independent Ink, Inc 547 US 28 (2006). 74 Suffice it to mention the tension between the assumptions underpinning the CoJ ruling in Hoffmann-La Roche (n 6) paras 89−90 and Delimitis (n 12) paras 10−12. 75 See in particular Denis Waelbroeck, ‘Michelin II: A Per se Rule against Rebates by Dominant Companies?’ (2005) 1 Journal of Competition Law & Economics 149 and references above at n 3. 65

66

96  Research handbook on abuse of dominance and monopolization The conflict between the nature of some practices and their legal status has been gradually addressed in the case law. There has been an incremental shift towards a context-specific evaluation of some practices to adjust, de iure and/or de facto, the applicable test to the reality of the behaviour. As a result, and while not necessarily being fully subject to a case-by-case analysis of their impact, the status of some categories as prima facie infringements has been progressively undermined over time. Suffice it to mention two examples to illustrate the point. The successive refinements introduced over the years show that, in effect, the legal status of tying in the US depends on a context-specific assessment which takes into consideration, in particular, the degree of market power enjoyed by the firm and of any efficiency gains resulting from it.76 The direction of travel is even more marked in technology-related markets.77 A similar trend can be witnessed in the EU system. The European Commission has acknowledged that, at least in certain markets, a case-by-case evaluation of the likely effects of tying should be a precondition for intervention.78 Exclusive dealing and rebate schemes have undergone the same evolution. This is an area of the law that proved to be particularly controversial in the EU system for a number of years.79 The European Commission’s administrative practice from the 1980s to the early 1990s – following, in particular Michelin II and British Airways – led to a situation in which exclusivity obligations and conditional rebates were, for all intents and purposes, treated as presumptively abusive.80 Since then, the case law has moved in the opposite direction. In Post Danmark II, the Court ruled that the legal status of standardized rebates depends on a case-by-case evaluation of, inter alia, the nature and operation of the scheme, its coverage and the relevant economic and regulatory context.81 In this sense, it is no longer sufficient to claim that a particular scheme is ‘loyalty-inducing’ to conclude that it is caught by the prohibition. Formally speaking, however, exclusive dealing and loyalty rebates continue to be (at least so formally) prima facie abusive, irrespective of their effects, in the EU system. However, the CoJ clarified, in Intel, that it is possible for a dominant firm to provide evidence showing that a rebate scheme conditional upon exclusivity is incapable of restricting competition and thus falls outside the scope of Article 102 TFEU.82 By making explicit a principle that was implicit in the case law, the Court provides an opening to evaluate the lawfulness of exclusive dealing and loyalty rebates in a way that is consistent with the nature and potential for harm of these practices. In this sense, Intel judgment comes close to making them subject to a case-by-case analysis of their effects, even if only de facto.83

See Einer Elhauge, ‘Tying, Bundled Discounts and the Death of the Single Monopoly Profit Theory’ (2009) 123 Harvard Law Review 397. 77 ibid 446−7; and US Microsoft (n 43). 78 EU Microsoft (n 53) para 841. See also Google Android (Case AT.40099) Commission Decision of 18 July 2018. 79 See above n 3. 80 Waelbroeck (n 75). 81 Post Danmark II (n 26) paras 39−46. 82 Intel (n 24) paras 138−9. 83 See in this sense Nicolas Petit, ‘The Judgment of the EU Court of Justice in Intel and the Rule of Reason in Abuse of Dominance Cases’ (2018) 43 European Law Review 728; and Pablo Ibáñez Colomo, ‘The Future of Article 102 TFEU after Intel’ (2018) 9 Journal of European Competition Law & Practice 293. 76

What is an abuse of a dominant position?  97 There is a second reason why tensions and controversies may arise when categorizing practices. Because there is not a unique set of conditions, applicable across the board, to evaluate the lawfulness of potentially abusive conduct, how behaviour is labelled can have significant consequences. Consider, for instance, the implications of qualifying, in the EU, a practice as a loyalty rebate scheme (which is prima facie prohibited irrespective of its effects) as opposed to a standardized one (which is subject to a case-by-case assessment). Consider, similarly, the consequences of concluding that a practice amounts to tying (which would be presumptively unlawful) instead of a refusal to deal (which demands evidence, at the very least, of the indispensability of the input or platform and the elimination of all competition). Three main factors explain the observed frictions around the appropriate categorization of practices. In the first place, telling one practice apart from another is not always an obvious exercise. Any refusal to deal, for instance, can be reasonably seen as a form of tying. After all, a dominant firm that denies access to an input or platform conditions the sale of one product to the acquisition of another one. The blurred boundaries between these two categories became apparent in Microsoft II. In that case, the Commission questioned the joint licensing of the dominant firm’s operating system (that is, a platform) and its web browser (that is, an input running on top of the platform).84 Frictions arise, in the second place, because there are functionally equivalent practices that are not always subject to the same legal test. For instance, there are no fundamental differences between rebates that are conditional on exclusivity (or an outright exclusivity obligation) and schemes based on the volume supplied. The difference, to the extent that it exists, is one of degree.85 One can think of a third factor behind these tensions. It relates to the ‘grey areas’ between one practice and the other. A look at Table 5.2 shows that the most common categories of potentially abusive conduct do not cover all possible combinations of the relevant criteria. Slovak Telekom provides an eloquent example of such ‘grey areas’ in between labels. In that case, the European Commission concluded that a vertically integrated firm had abused its dominant position by degrading the quality of the access provided to its downstream rivals. The operator had, inter alia, withheld the information supplied, restricted the scope of the regulatory obligations to which it was subject and, more generally, set unfair terms and conditions.86 This conduct is halfway between a refusal to deal and a margin squeeze. Unlike the latter, it is not based on price; unlike the former, it does not demand the imposition of a duty to deal on the firm (in the specific circumstances of the case there was regulation imposing such a duty). Questions about the applicable legal test arose as a result in the case.

IV.

EMERGING STRATEGIES AND THE BLURRING BOUNDARIES BETWEEN CATEGORIES

Controversies around the appropriate legal treatment of certain practices have intensified over the past decade. This is to a significant extent due to the fact that widespread strategies

Microsoft (tying) (Case COMP/C-3/39.530) Commission Decision of 16 December 2009. See Nicolas Petit and Norman Neyrinck, ‘Tying Law in Microsoft I and II: The Secret Art of Magic?’ (2011) 2 Journal of European Competition Law & Practice 117. 85 Opinion of AG Wahl in Case C-413/14 P Intel Corp v Commission EU:C:2016:788, para 100. 86 Slovak Telekom (Case AT.39523) Commission Decision of 15 October 2014. 84

98  Research handbook on abuse of dominance and monopolization in digital markets do not fit neatly into the common categories described above. It is not surprising, given the fundamental implications, that there are ongoing disputes about the legal status of these practices. The relative paucity of precedents is explained, in turn, by the intensity and consequences of the sort of intervention that competition authorities contemplate in digital markets. In some respects, administrative action in relation to these activities is more ambitious and far-reaching than the cases discussed above, in the sense that it interferes more frequently and decisively with the design of products and with firms’ business models. It is sufficient to mention some examples to illustrate the growing intrusiveness of enforcement. In Google Shopping, for instance, the European Commission challenged the way the dominant firm’s search engine displayed its results. It argued that the product, as designed by Google, gave more prominence to the firm’s own comparison-shopping service, which was displayed in rich format and at the top of the page.87 To the extent that it did, the investigation challenged a key design choice, whereby some information is made more readily available to end-users. It is unusual for competition law to question the integration of, and the interaction between, various features in a complex product. Generally speaking – and the main categories of potentially abusive practices confirm this view – competition law challenges how products are sold, rather than how they are made.88 Interfering with the latter is inevitably more intrusive. The difference between the two is manifested when defining the remedy to bring the infringement to an end. If a design is found to be anticompetitive, remedial intervention is likely to demand positive obligation prescribing the features of a compliant product. Given the scarcity of precedents in this area, it is not surprising that, when Google Shopping was investigated, there was no clarity about the legal test against which the lawfulness of product design would be assessed.89 While the concern was clear – by favouring its affiliated service, Google would be leveraging its dominant position in searches to a neighbouring market – the conditions to establish an infringement were far less so. After all, the status of leveraging practices ranges from the prima facie unlawful to the ‘exceptional circumstances’ test described above.90 A scenario like the one at stake in Google Shopping can be conceptualized both as a form of tying and as a refusal to deal.91 Insofar as a firm’s decision to integrate two features in a product (at the expense of rivals), does not allow end-users to make use of the core product without its complement, it can be seen as a form of tying (or bundling). On the other hand, and to the extent that the concern relates to access by rivals to the core product, it can be seen as a refusal to deal. In 2013, the Federal Trade Commission (FTC) closed an investigation into similar issues once it concluded that there was a plausible pro-competitive rationale for the search engine’s design. Insofar as it did, it concluded that choices about the combination of features in the

Google Search (Shopping) (Case AT.39740) Commission Decision of 27 June 2017. For an in-depth discussion, see Herbert J Hovenkamp, ‘Antitrust and the Design of Production’ (2018) 103 Cornell Law Review 1155. 89 For an overview, see Pablo Ibáñez Colomo, ‘Indispensability and Abuse of Dominance: From Commercial Solvents to Slovak Telekom and Google Shopping’ (2019) 10 Journal of European Competition Law & Practice 532. 90 For an overview of the uncertainties, see Pablo Ibáñez Colomo, ‘Legal Tests in EU Competition Law: Taxonomy and Operation’ (2019) 10 Journal of European Competition Law & Practice 424. 91 Ibáñez Colomo (n 4). 87 88

What is an abuse of a dominant position?  99 product are presumptively lawful.92 In Google Shopping, on the other hand, the European Commission evaluated the legality of self-preferencing against its potential anticompetitive effects, thus suggesting a case-by-case approach.93 The CoJ ruling in Slovak Telekom, mentioned above, gave some indications about the appropriate way to navigate product design cases. The Court held that a case is to be treated as a refusal to deal where intervention would amount to forcing a dominant firm to conclude an agreement with a third party with which it has chosen not to deal.94 The first instance ruling in Google Shopping, which would directly address the status of product design, is pending at the time of writing.95 Second, competition law authorities seem to be willing to intervene, more frequently and more decisively, against dominant firm’s business models. A business model is understood as the set of monetization strategies developed by firms. Competition law is, generally speaking, agnostic about business models.96 For instance, it does not express a preference for vertical integration over selling via third parties. Similarly, it does not favour some vertical restraints (say, franchising, as opposed to selective distribution) over others. As the law stands, firms may only be required to change their business model and adopt an alternative monetization strategy (for instance, by asking them to license their technology to third parties) in the exceptional circumstances defined in the case law. Intervention challenging firms’ business models – so that dominant firms are required to develop an alternative monetization strategy (or to forego a monetization strategy altogether) – defies an easy categorization. Several of the ‘ideal types’ described above may be relevant when applying competition rules in this context. For instance, a business model may be implemented in a manner that involves something akin to a tying strategy. The Android decision in the EU is a clear example in this sense.97 This case is notable – and different from precedents such as Microsoft, let alone traditional tying cases – in that the European Commission questioned the core of the firm’s business model. In accordance with the dominant player’s monetization strategy, its applications for smartphones were licensed for free to its customers. However, customers accepted, as a condition, the pre-installation of a whole suite of applications.98 The ostensible purpose of the joint licensing of applications was to allow the firm to monetize the value of its ecosystem via the advertising revenue generated by some of its products. In this sense, the practice is not fundamentally different from the activities of free-to-air broadcasters, which bundle content and advertising (the latter financing the former). By declaring the unlawfulness of Google’s business model, the Android decision forced the firm to develop an alternative monetization strategy (or to forego the strategy). Insofar as it did, the nature and

92 FTC, ‘Google Agrees to Change Its Business Practices to Resolve FTC Competition Concerns in the Markets for Devices like Smart Phones, Games and Tablets, and in Online Search’ (Washington DC, 3 January 2013). 93 Google Shopping (n 87) para 341. 94 Case C-165/19 P Slovak Telekom, as v Commission EU:C:2021:239, paras 46−51. 95 Case T-612/17 Google LLC and Alphabet Inc v Commission, pending. 96 In particular, competition law does not mandate vertical integration, nor does it prohibit it. Similarly, the EU competition law system does not mandate the least restrictive set of vertical restraints to attain a pro-competitive objective. Only in exceptional circumstances, defined in Magill (n 70) and Bronner (n 62), can a business model be altered. 97 Android (n 78). 98 ibid paras 172−91.

100  Research handbook on abuse of dominance and monopolization scale of intervention in the case cannot be compared with that of the relevant tying precedents, which left the firms’ business models untouched. Microsoft II, for instance, did not impact on the dominant player’s monetization strategy, which was (and remained) reliant on licensing fees. For the same reason, the remedy that is commonplace in traditional tying cases – a cease-and-desist order – would not be sufficient in a context like the one at stake in Android. Altering or foregoing a monetization strategy demands, by definition, proactive intervention.99

V. CONCLUSIONS Controversies surrounding the notion of abuse are unlikely to disappear in the short to medium term. This chapter shows, however, that there are some core aspects about the concept that are no longer disputed in the EU and US systems. It seems clear, first, that not all practices are inherently abusive, in the sense that they do not all entail an unequivocal departure from competition on the merits. The vast majority of potentially problematic conduct demands a case-by-case assessment to evaluate its object and/or impact in the specific economic and legal context of which it is a part. Second, the provisions on abuse are only triggered where a practice is at least capable of having anticompetitive effects. Third, it is, in principle, possible for dominant firms to advance a defence to escape the prohibition. Against this background, it seems possible to provide a tentative definition of the notion of abuse, which can be conceptualized as a practice that is at least capable of restricting competition and that either has no plausible pro-competitive rationale or the anticompetitive effects of which weigh more than any gains resulting from it. Other questions remain open. A recurrent issue concerns the mismatch between the legal status of some practices, on the one hand, and their nature and potential effects, on the other. Another question relates to the appropriate categorization of potentially abusive conduct. It is not always easy to tell one strategy apart from another. Insofar as the labelling exercise is not purely formal and has an impact on the ease with which the infringement can be established, frictions and controversies seem inevitable. These frictions have become more frequent and intense as competition authorities redirected their enforcement efforts to digital markets. The change in priorities has resulted in agencies questioning the design of products and business models, practices that defy an easy classification and which demand intervention that is more intrusive and far-reaching than in more traditional cases.

In the specific context of the Android case, for instance, the dominant firm introduced a ‘ballot box’ placing its products and those of rivals on a level playing field and allowing end-users to choose their preferred option. See Hiroshi Lockheimer, ‘Complying with the EC’s Android Decision’ (The Keyword, 16 October 2018), https://​www​.blog​.google/​around​-the​-globe/​google​-europe/​complying​-ecs​ -android​-decision/​, accessed 1 November 2022; and Oliver Bethell, ‘Changes to the Android Choice Screen in Europe’ (The Keyword, 8 June 2021), https://​blog​.google/​around​-the​-globe/​google​-europe/​ changes​-android​-choice​-screen​-europe/​, accessed 1 November 2022. 99

6. Exploitative abuses: recent trends and comparative perspectives Marco Botta

I.

INTRODUCTION: THE REVIVAL OF EXPLOITATIVE ABUSES

Since the Treaty of Rome,1 Article 102 of the Treaty on the Functioning of the European Union (TFEU) sanctions three types of exploitative abuses: ‘unfair purchase or selling prices’ (ie excessive pricing), ‘other unfair trading conditions’2 and ‘discrimination’.3 By studying the travaux préparatoires of the Rome Treaty, Akman concluded that the initial intention of the EU’s founding fathers was to sanction under Article 102 TFEU primarily exploitative conducts that directly harm the customers of the dominant firm.4 Nevertheless, due to the inability to set clear guidelines on what constitutes exploitative abuses and the possible overlap with sector-specific regulation, the European Commission has seldom investigated exploitative abuses over the past 60 years.5 The European Commission rather gave priority to investigations concerning exclusionary practices (eg refusal to deal, predatory pricing, fidelity rebates), where dominant firms indirectly harm final consumers by excluding competitors from the market.6 The ‘non-enforcement paradigm’ followed by the European Commission has been supported by the majority of economists, especially in relation to excessive pricing cases.7 First, if the price of the product is indeed ‘too high’, either consumers would stop buying the product or new firms would enter the market, thus forcing the dominant firm to lower its prices.8 Therefore, in a normal competitive scenario, the market would self-adjust and thus competition policy intervention would be unnecessary. Secondly, in network industries (eg electricity, gas, railways), price regulation could prevent excessive prices:9 if the competent National Regulatory Authority (NRA) has already regulated the access/retail price, the National Consolidated Version of the Treaty on the Functioning of the European Union (TFEU) [2012] OJ C326/47. 2 Article 102(a) TFEU. 3 Article 102(c) TFEU. 4 Pinar Akman, ‘Searching for the Long-Lost Soul of Article 82 EC’ (2009) 29 Oxford Journal of Legal Studies 267. 5 Michal Gal, ‘Abuse of Dominance – Exploitative Abuses’ in Ioannis Lianos and Damien Gerardin (eds), Handbook on European Competition Law (Edward Elgar Publishing, 2013). 6 ibid. 7 See, for instance, John Davies and Jorge Padilla, ‘Another Look at the Role of Barriers to Entry in Excessive Pricing Cases’ (SSRN, 2019), https://​papers​.ssrn​.com/​sol3/​papers​.cfm​?abstract​_id​=​3364881, accessed 2 November 2022. 8 ibid. 9 In relation to the overlap between competition and regulation in dealing with excessive pricing cases, see Giorgio Monti, ‘Excessive Pricing: Competition Law in Shared Regulatory Space’ (2019) 1

101

102  Research handbook on abuse of dominance and monopolization Competition Authority (NCA) should not intervene under Article 102(a) TFEU.10 Thirdly, firms develop new and innovative products if they can benefit from a monopoly price for a temporary period of time: by sanctioning excessive prices, an NCA would thus discourage the dominant firm to invest in innovation.11 While the wording of Article 102 TFEU has remained the same during the past 60 years, the ‘non-enforcement paradigm’ has progressively changed. In recent years, the European Commission and a number of NCAs have increasingly investigated exploitative abuses in a number of industries, such as gas, electricity and the pharmaceutical sector.12 In addition, a number of competition authorities have investigated exploitative conducts in digital markets. The ongoing European Commission investigations in the Apple Store case,13 the decision of the German Bundeskartellamt in the Facebook case14 as well as the decisions of the Autorité de la Concurrence in Google Ads Rules15 and Google News16 cases represent good examples of this emerging trend. In the US, antitrust enforcers have never imposed any sanction on exploitative conducts either under Section 2 Sherman Act17 or under Section 5 Federal Trade Commission Act.18 Unlike the EU, this is not due to a policy choice of the Department of Justice (DoJ) and the Federal Trade Commission (FTC), but rather due to the interpretation of antitrust law. According to well-established case law of the US Supreme Court, US antitrust law only sanctions a market behaviour that causes an ‘antitrust injury’:19 a distortion of the competitive process that harms the consumers’ welfare.20 In its landmark ruling in Trinko, the US Supreme Court pointed out that:

TILEC Working Paper, available at SSRN, https://​www​.tilburguniversity​.edu/​sites/​tiu/​files/​download/​ Monti​%20Excessive​%20pricing​.pdf, accessed 2 November 2022. 10 Massimo Motta and Alexandre De Streel, ‘Excessive Pricing in Competition Law: Never Say Never?’ in Konkurrensverket (Swedish Competition Authority) (ed), The Pros and Cons of High Prices (Lenanders Grafika 2007). 11 Patrick Hubert and Marie-Laure Combet, ‘Exploitative Abuse: The End of the Paradox?’ (2011) 1 Concurrences 51. 12 Marco Botta, ‘Sanctioning Unfair Pricing under Article 102(a) TFEU: Yes, We Can!’ (2021) 17 European Competition Journal 156. 13 European Commission Press Release, ‘Commission Sends Statement of Objections to Apple on App Store Rules for Music Streaming Providers’, IP/21/2061 (30 April 2021), https://​ec​.europa​.eu/​ commission/​presscorner/​detail/​en/​ip​_21​_2061, accessed 2 November 2022. 14 Bundeskartellamt Decision B6/22/16 (Facebook), 6 February 2019, https://​www​.bundeskartellamt​ .de/​SharedDocs/​Entscheidung/​DE/​Entscheidungen/​Missbrauchsaufsicht/​2019/​B6​-22​-16​.html;jsessionid​ =​C8​4AB98AE612​D2797BCB51​42ED3D0A5B​.2​_cid362​?nn​=​3591568, accessed 2 November 2022. 15 Autorité de la Concurrence Decision 19-D-26, 19 December 2019 (Google Ads Rules), https://​ www​.aut​oritedelac​oncurrence​.fr/​en/​decision/​regarding​-practices​-implemented​-sector​-online​-search​ -advertising​-sector, accessed 2 November 2022. 16 Autorité de la Concurrence Decision 21-D-17, 12 July 2021 (Google News), available at https://​ www​.aut​oritedelac​oncurrence​.fr/​fr/​decision/​relative​-au​-respect​-des​-injonctions​-prononcees​-lencontre​ -de​-google​-dans​-la​-decision​-ndeg, accessed 2 November 2022. 17 15 USC Code § 2. 18 15 USC Code § 45. 19 Brunswick Corp v Pueblo Bowl-O-Mat, Inc, 429 US 477, 488 (1977). 20 In NYNEX the US Supreme Court ruled that in a Sherman Act case the ‘plaintiff … must allege and prove harm, not just to a single competitor, but to the competitive process, i.e., to competition itself’. See NYNEX Corp v Discon, 525 US 128, 135 (1998).

Exploitative abuses: recent trends and comparative perspectives  103 The opportunity to charge monopoly prices is what attracts ‘business acumen’ … the possession of monopoly power will not be found unlawful unless it is accompanied by an element of anti-competitive conduct.21

The recent Qualcomm case represents a good example of the application of this case law by the US federal courts. In 2017, the FTC accused Qualcomm of monopolizing the market of microchips via its ‘no license, no chip policy’: manufacturers of electronic devices had to conclude a licence agreement with Qualcomm if they wished to purchase the Qualcomm chips.22 According to the FTC, Qualcomm licensing policy caused an ‘anticompetitive surcharge’ on its customers, who had to pay both the cost of the microchips as well as a royalty rate to Qualcomm. While the District Court of California upheld the FTC complaint, the Court of Appeal for the Ninth Circuit reversed the first instance ruling in August 2020.23 The Court of Appeal ruled that Qualcomm’s conduct did not cause a distortion of competitive dynamics in the downstream market – that is, no evidence of antitrust injury. The excessive pricing claim is therefore a ‘side’ argument in US antitrust law; without evidence of antitrust injury, the US federal courts would not uphold the claim concerning breach of antitrust law. Since exploitative conducts are a peculiarity of the EU competition law regime and represent one of the major differences with US monopolization cases, this chapter will focus on enforcement trends concerning exploitative abuses prohibited under Article 102 TFEU. In particular, the chapter will discuss the legal test and recent enforcement trends concerning excessive pricing (Section II), unfair trading conditions (Section III) and discriminatory pricing (Section IV). Section V concludes, discussing whether and to what extent we are witnessing a revival of exploitative conduct in Europe.

II.

EXCESSIVE PRICING

A.

The Legal Test

Article 102(a) TFEU prohibits a dominant firm from ‘imposing unfair purchase or selling price’ on its customers. In General Motors, the Court of Justice (CoJ) clarified for the first time the meaning of this provision of the Treaty.24 According to the Court, ‘an abuse might lie, inter alia, in the imposition (by a dominant firm) of a price which is excessive in relation to the economic value of the service provided’.25 Nevertheless, only in United Brands, the Court developed a test to assess when the price of the product exceeds its economic value, and it could thus be considered ‘unfair’. According to the Court, the price charged by a dominant firm is

Verizon Communications Inc v Law Offices of Curtis V Trinko, LLP, 540 US 398, 407 (2004). FTC Press Release, ‘FTC Charges Qualcomm with Monopolizing Key Semiconductor Device Used in Cell Phones’ (17 January 2017), available at https://​www​.ftc​.gov/​news​-events/​press​-releases/​ 2017/​01/​ftc​-charges​-qualcomm​-monopolizing​-key​-semiconductor​-device​-used, accessed 2 November 2022. 23 Federal Trade Commission v Qualcomm Incorporated, No 19-16122 (9th Cir 2020). 24 Case C-26/75 General Motors Continental NV v Commission ECLI:EU:C:1975:150. 25 ibid para 12. 21 22

104  Research handbook on abuse of dominance and monopolization in breach of Article 102(a) when: (i) it is ‘excessive’ in comparison to the ‘economic value’ of the product, and (ii) ‘unfair’ either ‘in itself’ or ‘when compared to competing products’.26 The United Brands test is difficult to apply, since it includes a number of vague concepts: in its jurisprudence, the CoJ has neither introduced a minimum threshold to assess when a price is ‘excessive’, under the first limb of the test, nor has it clarified the meaning of ‘unfair in itself’, under the second limb. Secondly, it is uncertain whether a competition agency should take into consideration the economic value of the product for consumers to assess the excessiveness of the price under the first limb. As argued in the introduction, the unclear aspects of the United Brands test have contributed to the non-enforcement paradigm that has characterized Article 102(a) in the past decades. Nevertheless, it is worth bearing in mind that the United Brands test is not the only method to assess when the price charged by a dominant firm is unfair, and thus in breach of Article 102(a) TFEU. In United Brands, the CoJ recognized that ‘other ways may be devised’.27 Benchmarking is the main alternative test to assess an unfair pricing case under Article 102(a).28 The logic is rather simple: rather than comparing the costs and the price charged by the dominant firm, the NCA should compare the allegedly unfair price with a benchmark price. In its jurisprudence, the CoJ has endorsed a number of benchmark methods, such as: price comparison either with competitors of the dominant firm in the same relevant market29 or with firms selling the same product in a different geographic market;30 comparison of the price charged by the dominant firm for the same product either in different geographic markets31 or to different customers;32 comparison of the price charged by the dominant firm over time.33 After having identified a suitable benchmark method(s), the NCA should assess whether the price of the dominant firm is, indeed, unfair in comparison to the benchmark price. In AKKA-LAA, the CoJ did not introduce a minimum threshold in this regard; the Court pointed out that the price of the dominant firm must be ‘appreciably high’, ‘significant’ and ‘persistent’ in comparison to the benchmark price to breach Article 102(a) TFEU.34 As the UK Court of Appeal of England and Wales recognized in Pfizer-Flynn,35 the competition agency should verify its findings of price unfairness under ‘multiple’ tests. When multiple methods are available, the convergence of the results would strengthen the relevance of the findings of the NCA. Similarly to the decision of the UK Office of Fair Trading (OFT)

Case C-27/76 United Brands Company and United Brands Continentaal BV v Commission ECLI:EU:C:1978:22, para 252. 27 ibid para 253. 28 Botta (n 12). 29 In United Brands, for instance, the European Commission noticed that the average price of bananas sold by United Brands was 7 per cent higher than the price charged in Europe by its competitors. See United Brands (n 26) para 266. 30 Case C-177/16 Autortiesību un komunicēšanās konsultāciju aģentūra v Latvijas Autoru apvienība v Konkurences padome ECLI:EU:C:2017:689, para 38. 31 Case C-78/70 Deutsche Grammophon Gesellschaft mbH v Metro-SB-Großmärkte GmbH & Co KG ECLI:EU:C:1971:59, para 19. 32 Case C-226/84 British Leyland Public Limited Company v Commission ECLI:EU:C:1986:421, para 29. 33 ibid. 34 AKKA-LAA (n 30) para 55. 35 Competition and Markets Authority v Flynn Pharma Limited and Pfizer Limited [2001] EWCA Civ 339. 26

Exploitative abuses: recent trends and comparative perspectives  105 in Napp,36 the NCA would not be required to carry out the United Brands test in order to assess the unfairness of the price charged by the dominant firm; the NCA could verify the findings of price unfairness under multiple benchmarking methods. Finally, the dominant firm could put forward some objective justifications. Both in Sirena37 and in Deutsche Grammophon,38 the CoJ recognized that excessive pricing is not an abusive behaviour if it is justified by ‘objective criteria’. In AKKA-LAA, the CoJ ruled that the unfair fees could be justified by the fact that the Latvian Copyright Society granted a higher royalty rate to its copyright holders, and thus it faced higher administrative costs in comparison to the copyright societies of other EU Member States.39 Therefore, the CoJ has accepted, in principle, that a dominant firm may justify its unfair pricing strategy due to objective justifications. However, in practice, the Court has never upheld such arguments in specific cases. B.

Recent Enforcement Trends: The Complexities of Regulated and Digital Markets

Most of the recent cases sanctioning excessive prices have taken place in the energy and pharmaceutical sectors. In the electricity industry, a number of NCAs have fined incumbent generators under Article 102(a) TFEU: by withholding capacity during the period of peak demand, incumbent electricity generators have been able to charge excessive prices in the wholesale electricity markets.40 Similarly, in the Gazprom case, the European Commission has sanctioned the excessive prices charged by Gazprom and accepted Gazprom’s commitment to remove a number of contractual clauses in the agreements concluded with gas distributors in Central and Eastern Europe: unfair clauses that limited the free flow of gas in the region.41 In the pharmaceutical sector, a number of NCAs have investigated cases concerning the unfair pricing of off-patent drugs: the Aspen decision of the Autorità Garante della Concorrenza e il Mercato (AGCM, Italian Competition Authority)42 and Pfizer-Flynn43 and Auden Mckenzie – Actavis,44 decisions by the UK Competition and Markets Authority (CMA), show that excessive prices of medicines charged by generics manufacturers on final consumers may be subject to antitrust scrutiny. In addition, the recent Aspen45 commitment decision shows that even the

36 The method of assessment followed by the OFT in Napp was later upheld by the UK Competition Appeal Tribunal (CAT). See Napp Pharmaceutical Holdings Limited and Subsidiaries v Director of Fair Trading, [2002] CAT n 1001/1/1/01, para 397. 37 Case C-40/70 Sirena Srl v Eda Srl and others ECLI:EU:C:1971:18, para 71. 38 Deutsche Grammophon (n 31) para 19. 39 AKKA-LAA (n 30) paras 58−9. 40 Rozeta Karova and Marco Botta, ‘Sanctioning Excessive Energy Prices as Abuse of Dominance’ in Pier Luigi Parcu and others (eds), Abuse of Dominance in EU Competition Law: Emerging Trends (Edward Elgar Publishing 2017) 169. 41 Upstream Gas Supplies in Central and Eastern Europe (AT.39816) Commission Decision of 24 May 2018. 42 Autorità Garante della Concorrenza e del Mercato Decision No 26185, 29 September 2016 (Aspen). 43 UK Competition and Markets Authority Decision CE/9742-13 Unfair Pricing in Respect of the Supply of Phenytoin Sodium Capsules in the UK, 7 December 2016. 44 CMA Press Release, ‘CMA Finds Drug Companies Overcharged NHS’ (15 July 2021), https://​www​ .gov​.uk/​government/​news/​cma​-finds​-drug​-companies​-overcharged​-nhs, accessed 2 November 2022. 45 Aspen (AT.40394) Commission Decision of 10 February 2021, OJ C-233/7.

106  Research handbook on abuse of dominance and monopolization European Commission is currently reconsidering its traditional non-enforcement paradigm vis-à-vis unfair pricing cases concerning off-patent drugs. In terms of applicable legal text, the competition agencies have generally opted for the United Brands test, rather than the benchmarking approach.46 On a number of occasions, the competition agencies have reached a commitment decision with the dominant firm, thus preventing judicial review.47 By contrast, infringement decisions have been subject to intense judicial review by national courts, which has led to ambivalent outcomes. A good example in this regard is represented by the Aspen and Pfizer-Flynn decisions. Both cases were investigated in parallel by the Italian and the UK competition authorities and concerned the same type of abuse: the excessive price increase introduced by a pharmaceutical company after the patent expiration and the rebranding of the medicine (ie when the drug became ‘generic’). Both authorities followed the United Brands test and collected comparable evidence to support their decisions. Nevertheless, the outcome of the judicial review was radically different: while TAR Lazio fully upheld the legality of the AGCM decision,48 the UK Competition Appeal Tribunal (CAT)49 and the Court of Appeal of England and Wales quashed the CMA decision in Pfizer-Flynn.50 In particular, the CAT and the Court of Appeal criticized the CMA for not having verified the findings of excessive pricing under multiple tests, besides United Brands.51 While in the pharmaceutical sector NCAs have been reluctant to investigate excessive pricing cases concerning medicines covered by patents, in the energy and pharmaceutical industries the NCAs have generally investigated excessive pricing cases falling outside the scope of sector regulation. In Aspen, for instance, the pharmaceutical company asked the Italian health care agency to delete Cosmos drugs (ie off-patent drugs used to cure certain types of cancer) from the list of medicines reimbursed by the Italian health care system, and thus subject to price regulation.52 Aspen forced the Italian health care agency to agree on price increases between 300 per cent and 1500 per cent; the pharmaceutical company threatened the health care agency that it would withdraw the drugs from the Italian market if it did not accept the proposed price increase. A second common feature of the recent excessive pricing cases is that the sanctioned conducts did not affect the innovation incentives of the dominant firm. In the pharmaceutical industry, for instance, the investigations concerned exclusively off-patent drugs – that is, medicines not covered by any patent right. Generic drugs manufacturers do not invest in R&D and do not develop new medicines. In other words, generic drugs manufacturers do not invest in research and innovation. In theory, the entry of generic drugs manufacturers into the market

For instance, United Brands test was applied both by the AGCM in Aspen as well as by the CMA in Pfizer-Flynn. 47 See, for instance, the Aspen commitment decision by the European Commission (n 45). 48 Aspen Pharma Trading Limited v Autorità Garante per la Concorrenza e il Mercato, TAR Lazio n 8945, 26 July 2017. 49 Flynn Pharma Limited and Pfizer Inc v Competition and Markets Authority, CAT n 1275-1276/1/12/17, 7 June 2018. 50 Pfizer-Flynn (n 35). 51 In its judgment, CAT asked the CMA ‘to consider a range of possible analyses, reflecting market conditions and the extent and quality of the data that can be obtained, to establish a benchmark price, or range, that reflects the price that would pertain under conditions of normal and sufficiently effective competition’. Pfizer-Flynn (n 49) para 443(1). 52 Aspen (n 42). 46

Exploitative abuses: recent trends and comparative perspectives  107 after the patent expiration should lower the medicines price. However, the recent excessive pricing cases show that this is not always the market outcome: due to the lack of substitutability between different drugs, high entry barriers in the market and the lack of price regulation after the patent expiration, the entry of generic drugs manufacturers might actually increase the price of medicines.53 In particular, rather than engaging in price competition with the former patent holder, the generic drugs manufacturers often enter into the market by purchasing the production rights of the drug from the former patent holder; after having re-branded the generics drug, the new manufacturers substantially increase the medicine price.54 By contrast, no NCA has ever investigated a case concerning the excessive price of drugs covered by patent protection. Although a number of authors have argued in favour of broadening the scope of excessive pricing investigations in the pharmaceutical sector,55 NCAs have generally been reluctant in this regard, for fear of discouraging innovation by the patent holder and of engaging in price regulation. A similar enforcement pattern is also visible in other industries characterized by a high degree of innovation. In Rambus, the European Commission sanctioned ‘patent ambush’: the lack of patent disclosure by the Standard Essential Patent (SEP) holder during the standardization process within a Standard Development Organization (SDO).56 According to the Commission, patent ambush is an abusive conduct since it could facilitate ‘patent hold-up’ – that is, allow the SEP holder to charge excessive royalty rates to the potential licensees. The Rambus case, however, represents a rather unique case, concluded by the European Commission via a commitment, rather than an infringement, decision. In contrast, no NCA in Europe has ever directly sanctioned a case of unfair royalty rates requested by the patent holder under Article 102(a) TFEU.57 Following a complaint submitted by a group of smartphone producers in 2007, the European Commission opened an investigation against Qualcomm for alleged excessive royalty rates relating to the licensing of its relevant patents for the 3G standard.58 Two years later, the implementers concluded a licence agreement with Qualcomm, and they withdrew their complaint; the Commission, thus, closed the investigation without taking any formal decision on the case.59 In a number of cases, sector regulation has prevented excessive prices, making antitrust enforcement ‘redundant’. A number of SDOs, for instance, have introduced internal rules

Margherita Colangelo and Claudia Desogus, ‘Antitrust Scrutiny of Excessive Prices in the Pharmaceutical Sector: A Comparative Study of the Italian and UK Experiences’ (2018) 41 World Competition 225. 54 ibid. 55 Chris Fonteijn and others, ‘Reconciling Competition and IP Law: The Case of Patented Pharmaceuticals and Dominance Abuse’ in Gabriella Muscolo and Marina Tavassi (eds), The Interplay between Competition Law and Intellectual Property: An International Perspective (Wouters Kluwer 2019). 56 RAMBUS (Case COMP/38.636) Commission Decision of 9 December 2009, OJ C-30/17. 57 Marco Botta, ‘The Challenge of Sanctioning Unfair Royalty Rate by the SEP Holder: When, How and What’ (2021) 44 World Competition 3. 58 European Commission Press Release, ‘Antitrust: Commission Initiates Formal Proceedings against Qualcomm’, MEMO/07/389 (1 October 2007), https://​ec​.europa​.eu/​commission/​presscorner/​ detail/​en/​MEMO​_07​_389, accessed 2 November 2022. 59 European Commission Press Release, ‘Antitrust: Commission Closes Formal Proceedings against Qualcomm’, MEMO/09/516 (24 November 2009), https://​ec​.europa​.eu/​commission/​presscorner/​detail/​ en/​MEMO​_09​_516, accessed 2 November 2022. 53

108  Research handbook on abuse of dominance and monopolization to encourage patent disclosure in order to prevent patent ambush.60 The latter is an example of industry self-regulation, aiming at preventing excessive pricing investigations by the competition authorities. In other cases, sector regulation has been introduced at the EU level. The EU Regulation on Wholesale Energy Market Integrity and Transparency (REMIT), for instance, aims at increasing market transparency and preventing forms of market manipulation in the electricity wholesale sector.61 Following the entry into force of REMIT, no NCA has sanctioned cases of excessive prices caused by capacity withdrawal in the electricity wholesale market.62 Similarly, following the European Commission decision in MasterCard,63 sanctioning the excessive multilateral interchange fees requested by credit cards, the EU has adopted a regulation on interchange fees for card-based payment transactions.64 REMIT and the Regulation on the interchange fees represent examples of ‘reactive regulation’: following a number of antitrust decisions sanctioning excessive pricing cases in the industry, the EU legislator ‘reacted’ to solve the market failures at the root of the previous antitrust investigations.65 The last enforcement trend concerns the emergence of excessive pricing cases in the digital sector. The most interesting case in this regard is represented by the Apple Store case, currently being investigated by the European Commission. In April 2021, the European Commission sent a Statement of Objection to Apple.66 According to the European Commission, Apple abused its dominant position via the Apple Store rules imposed on music streaming providers. In particular, the Statement of Objections considered excessive the 30 per cent commission fee requested by Apple from app developers using the Apple Store to sell their products. Secondly, the European Commission considered abusive the anti-steering rules imposed by Apple on app developers. Such rules limit the ability of app developers to inform users of alternative purchasing possibilities outside of the Apple Store. At the time of writing, it is unclear whether the European Commission will sanction both conducts in its final decision. However, it is worth noticing that the two conducts challenged by the European Commission in Apple Store represent two separate categories of abuse falling within the scope of Article 102(a): while the excessive fee requested by Apple from app developers would fall within the prohibition of excessive pricing, the anti-steering clauses should be assessed as unfair trading conditions, as further discussed in the next section. The application of the United Brands test and the benchmarking approach in the context of the digital economy is rather problematic. First of all, the digital economy is characterized

In 2005, for instance, the European Telecommunications Standardization Institute (ETSI) adopted new internal rules to encourage patents disclosure during the process of standard negotiation, in order to avoid patent ambush. See European Commission Press Release, ‘Commission Welcomes Changes in ETSI IPR Rules to Prevent “Patent Ambush”’, IP/05/1565 (12 November 2005), https://​ec​.europa​ .eu/​commission/​presscorner/​api/​files/​document/​print/​en/​ip​_05​_1565/​IP​_05​_1565​_EN​.pdf, accessed 2 November 2022. 61 Regulation (EU) No 1227/2011 of the European Parliament and of the Council of 25 October 2011 on wholesale energy market integrity and transparency, [2011] OJ L-326/1. 62 Karova and Botta (n 40). 63 MasterCard (COMP/34.579), EuroCommerce (COMP/36.518), Commercial Cards (COMP/38.580) Commission Decision of 19 December 2007, OJ C-264/8. 64 Regulation (EU) 2015/751 of the European Parliament and of the Council of 29 April 2015 on interchange fees for card-based payment transactions, [2015] OJ L-123/1. 65 Monti (n 9). 66 Apple Store (n 13). 60

Exploitative abuses: recent trends and comparative perspectives  109 by ‘close-to-zero’ marginal costs:67 the marginal cost faced by a digital platform to provide a digital service to an additional user is insignificant. In such a context, it is difficult to apply the cost-price test as the basis of United Brands. Secondly, digital markets are generally characterized by ‘winner takes all’ dynamics:68 the market for apps, for instance, is characterized by a duopoly – that is, Apple Store and Google Play. In such a contest, it is quite difficult to identify a suitable benchmark to measure the ‘excessiveness’ of the fee requested by the platform from its users. Finally, digital markets are often zero-price markets: final users ‘pay’ digital services either by transferring their personal data to the platform or via their attention to online advertising.69 Both the United Brands test and the benchmarking approach are based on price assessment; an approach not feasible in zero-price markets. The European Commission investigations in Apple Store are currently ongoing. If the European Commission concluded the investigations via an infringement/commitment decision under Article 102(a), the former would represent an important precedent for other online platforms that control a digital ecosystem. However, the European Commission will face a high burden to prove the case under the existing United Brands test and benchmarking approach. In particular, the EU General Court and the CoJ will have to review the findings of the European Commission in light of the existing case law in case the European Commission concluded the investigations via an infringement decision.

III.

UNFAIR TRADING CONDITIONS

A.

The Legal Test

Besides excessive prices, Article 102(a) TFEU also prohibits ‘unfair trading conditions’ imposed by a dominant firm on its customers. The expression refers to non-price conditions, especially contractual clauses. In Sabam I, the Advocate General defined unfair trading conditions as ‘obligations which are not absolutely necessary for the attainment of the object (of the contract) and which encroach unfairly on the customer’s freedom’.70 Such a general definition, however, was not endorsed by the CoJ in its ruling in Sabam I.71 The Court rather opted for a case-by-case approach to assess the unfairness of the challenged contractual clauses. In its case law, the CoJ has recognized in breach of Article 102(a) TFEU a number of contractual clauses, such as the automatic increase of the tariff by the dominant firm after the contract expiration.72 Secondly, the imposition of a contract valid for an indef-

67 OECD, ‘Big Data: Bringing Competition Policy to the Digital Era’ DAF/COMP(2016)14, https://​ one​.oecd​.org/​document/​DAF/​COMP(2016)14/​en/​pdf, accessed 2 November 2022. 68 ibid. 69 OECD, ‘Quality Considerations in Digital Zero-Price Markets’, DAF/COMP(2018)14, https://​one​ .oecd​.org/​document/​DAF/​COMP(2018)14/​en/​pdf, accessed 2 November 2022. 70 Opinion of Advocate General Mayras in Case C-127/73 Belgische Radio en Televisie and Société Belge des Auteurs, Compositeurs et Editeurs v SV SABAM and NV Fonior ECLI:EU:C:1974:11, para II.B.1. 71 Case C-127/73 Belgische Radio en Televisie and Société Belge des Auteurs, Compositeurs et Editeurs v SV SABAM and NV Fonior ECLI:EU:C:1974:25. 72 Case C-247/86 Société Alsacienne et Lorraine de Télécommunications et d’Electronique (Alsatel) v SA Novasam ECLI:EU:C:1988:469.

110  Research handbook on abuse of dominance and monopolization inite period/for an unlimited number of transactions has also been considered ‘unfair’ by the CoJ.73 Thirdly, the CoJ considered abusive the fact that the dominant firm forced its customers either to buy a bundle of different services,74 to accept services not previously requested;75 or to pay a flat tariff independently of the service received.76 Finally, a dominant firm would breach Article 102(a) TFEU if it limited the use of the product by the customer without any objective justification.77 Secondly, the CoJ has recognized as breach of Article 102(a) TFEU unfair contractual clauses ‘unilaterally imposed’ by the dominant firm on its customers.78 In other words, the terms enjoined under Article 102(a) TFEU were often standard clauses applied by the dominant firm on every customer, rather than the result of a negotiation process. Similarly to excessive pricing cases, the legal test developed by the CoJ to assess unfair trading practices is based on a ‘legalistic’ type of enforcement of competition rules: if the challenged contractual clause fell within one of the trading conditions prohibited under the CoJ case law, it was per se prohibited. In other words, the CoJ has never assessed the distortive effects of the unfair trading conditions on the downstream competition among the customers of the dominant company. Finally, dominant firms may put forward objective justifications. In United Brands, the CoJ pointed out that the dominant firm may protect its commercial interests, but in a ‘proportional manner’.79 However, in practice, no firm has managed to rebut the finding of abuse by putting forward objective justifications.80 B.

Recent Enforcement Trends: The Revival of Unfair Trading Conditions in the Context of the Digital Economy

Historically, Article 102(a) TFEU has rarely been relied upon by competition authorities to prohibit unfair trading conditions: the number of cases sanctioning non-price trading conditions under Article 102(a) TFEU is even lower than the number of excessive pricing cases. This non-enforcement trend is explained by two main reasons: first of all, the CoJ has never

Sabam I (n 71). ibid. 75 Case C-179/90 Merci Convenzionali Porto di Genova SpA v Siderurgica Gabrielli SpA ECLI:EU:C:1991:464, paras 8−24. 76 See, in particular, Case C-143/19 P Der Grüne Punkt – Duales System Deutschland GmbH v European Union Intellectual Property Office ECLI:EU:C:2019:1076; Case C-372/19 Belgische Vereniging van Auteurs, Componisten en Uitgevers CVBA (SABAM) v Weareone World BVBA and Wecandance NV ECLI:EU:C:2020:959. 77 United Brands (n 26) paras 130−62. 78 Alsatel (n 72). 79 United Brands (n 26) para 189. 80 For instance, in AAMS, the Italian monopoly in charge of the distribution of cigarettes in the country tried to justify the contractual clauses limiting the ability of foreign suppliers to sell cigarettes in Italy. AAMS argued that these clauses were necessary in view of the limited capacity of its distribution network. The EU General Court did not accept these justifications, ruling that: ‘AAMS has not proved to the requisite legal standard that the clauses mentioned above were necessary to protect its commercial interests and to avoid … the risk of its distribution network becoming overloaded’. Case T-139/98 Amministrazione Autonoma dei Monopoli di Stato (AAMS) v Commission ECLI:EU:T:2001:272, para 79. 73 74

Exploitative abuses: recent trends and comparative perspectives  111 developed a general test to assess unfair trading conditions comparable to the United Brands test. Secondly, unfair trading conditions may also be prohibited as unfair commercial practices under consumer law.81 A number of NCAs in Europe have the power to enforce both competition and consumer law.82 Agencies that can do both will generally sanction unfair contractual clauses under consumer, rather than competition, law.83 Unlike Article 102 TFEU, in fact, consumer law does not require the NCA either to define the relevant market or to assess market dominance and the foreclosing effect of the contested behaviour. Due to these reasons, the cases of enforcement of Article 102(a) vis-à-vis unfair trading conditions have been rare, at both the national and the EU level. This non-enforcement trend, however, is currently changing in the context of the digital economy. As mentioned in the previous section, in Apple Store, the European Commission is currently investigating the compatibility of the anti-steering provisions imposed by Apple on app developers.84 Secondly, in their recent decisions in Facebook,85 Google Ads Rules86 and Google News,87 the Bundeskartellamt and the Autorité de la Concurrence have sanctioned as abuse of a dominant position a number of unfair trading conditions imposed by a dominant digital platform on its users. In February 2019, the Bundeskartellamt condemned Facebook for abuse of a dominant position. The German NCA considered abusive the fact that Facebook illegally merged the personal data of its users collected ‘on Facebook’ (ie provided by users either at the moment of the registration or while using Facebook) and ‘off Facebook’ (ie collected by Facebook from third-party websites or from affiliated apps, such as Instagram and WhatsApp).88 Interestingly, the Bundeskartellamt sanctioned the abuse under Article 19(1) of the Gesetz gegen Wettbewerbsbeschränkungen (GWB, German Act against the Restraints of Competition).89 In particular, the Bundeskartellamt relied on the case law of the Bundesgerichtshof (BGH, German Federal Court of Justice),90 whereby unlawful contractual terms imposed by the dom-

81 See, in particular, Directive 2005/29/EC of the European Parliament and of the Council of 11 May 2005 concerning unfair business-to-consumer commercial practices in the internal market and amending Council Directive 84/450/EEC, Directives 97/7/EC, 98/27/EC and 2002/65/EC of the European Parliament and of the Council and Regulation (EC) No 2006/2004 of the European Parliament and of the Council (‘Unfair Commercial Practices Directive’), [2005] OJ L-149/22. 82 For instance, after the implementation of Directive 2005/29/EC, the Italian NCA is in charge of sanctioning unfair commercial practices. The NCA can impose fines of up to €5 million. See https://​en​ .agcm​.it/​en/​scope​-of​-activity/​consumer​-protection/​, accessed 2 November 2022. 83 In 2019, for instance, the Italian NCA carried out 34 investigations concerning the breach of competition rules and 89 investigations concerning unfair commercial practices. AGCM, ‘Relazione Annuale sull’Attività Svolta’ (2019). The 2019 AGCM annual report is available in Italian language at https://​ www​.agcm​.it/​pubblicazioni/​dettaglio​?id​=​80612a0d​-f4df​-4092​-af02​-59ac309a8814​&​parent​=​Relazioni​ %20annuali​&​parentUrl​=/​pubblicazioni/​relazioni​-annuali, 2 November 2022. 84 Apple Store (n 13). 85 Facebook (Bundeskartellamt) (n 14). 86 Google Ads Rules (Autorité de la Concurrence) (n 15). 87 Google News (Autorité de la Concurrence) (n 16). 88 Facebook (Bundeskartellamt) (n 14) 43−4. 89 ibid 287. The official English translation of the GWB is available at: https://​www​.gesetze​-im​ -internet​.de/​englisch​_gwb/​, accessed 2 November 2022. 90 Bundesgerichtshof, VBL Gegenwert II, KZR 47/14 (24 January 2017). Bundesgerichtshof, Claudia Pechstein, KZR 6/15 (7 June 2016).

112  Research handbook on abuse of dominance and monopolization inant firm on its customers would represent a breach of Article 19(1) GWB.91 In particular, the German NCA relied on the General Data Protection Regulation (GDPR)92 to conclude that the merging of ‘on-’ and ‘off-’Facebook data was unlawful.93 First of all, the merging breached the GDPR, since Facebook users had not given their explicit consent to the data processing.94 Secondly, the merger was not necessary for the performance of the contract – that is, the collection of off-Facebook data was not indispensable for the functioning of the social network.95 As a remedy, the Bundeskartellamt imposed the so-called ‘data silos’ obligation: Facebook could not merge users’ data collected within and outside its ecosystem.96 The Facebook decision is currently subject to a lengthy legal battle in German courts: in August 2019, the Düsseldorf Higher Regional Court issued an interim ruling to stop the enforcement of the Bundeskartellamt decision.97 While the Düsseldorf Court did not carry out a full review of the decision, it blocked its enforcement since it had ‘serious doubts’ in relation to its legality.98 In June 2020, the BGH reversed the previous interim ruling, and it sent the case back to the Düsseldorf Court to assess the substance of the case.99 In view of the diverging views of the BGH and Düsseldorf Court as to the legality of the decision, in March 2021, the Düsseldorf Court submitted a request for preliminary ruling to the CoJ, where the case is pending at the moment of writing.100 The decision of the Bundeskartellamt to assess the Facebook case under national competition law, rather than Article 102(a) TFEU, has been criticized by several authors.101 NCAs, in fact, do not have discretion to decide the legal basis of an antitrust investigation. Under Article 3(1) of Regulation 1/2003, NCAs ‘shall’ assess a conduct under Articles 101−102 TFEU if the conduct has an impact on intra-community trade.102 Although in the Facebook decision the Bundeskartellamt defined the geographic market as ‘Germany’ due to language barriers,103 social networks are cross-border services that clearly have an impact on intra-community trade.

Facebook (Bundeskartellamt) (n 14) 274−5. 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), [2016] OJ L-119/1. 93 Facebook (Bundeskartellamt) (n 14) 169−203. 94 GDPR (n 92) Article 6(1)(a). 95 ibid Article 6(1)(b). 96 Facebook (Bundeskartellamt) (n 14) 288−304. 97 Oberlandesgericht Düsseldorf, Facebook, VI-Kart 1/19 (V) (26 August 2019). 98 ‘Es bestehen ernstliche Zweifel an der Rechtmäßigkeit dieser kartell behördlichen Anordnungen’ (‘There are serious doubts about the legality of the antitrust decision’); ibid 6. 99 Bundesgerichtshof, Facebook, KVR 69/19 (23 June 2020). 100 Request for preliminary ruling submitted on 22 April by Oberlandesgericht Düsseldorf in the Case C-252/21 Facebook v Bundeskartellamt. 101 See, in particular, Marco Botta and Klaus Wiedemann, ‘The Interaction of EU Competition, Consumer, and Data Protection Law in the Digital Economy: The Regulatory Dilemma in the Facebook Odyssey’ (2019) 64 Antitrust Bulletin 426; Wouter Wils, ‘The Obligation for the Competition Authorities of the EU Member States to Apply EU Antitrust Law and the Facebook Decision of the Bundeskartellamt’ (2019) 3 Concurrences 58. 102 Council Regulation (EC) No 1/2003 of 16 December 2002 on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty, [2003] OJ L-1/1, art 3(1). 103 Facebook (Bundeskartellamt) (n 14) 110. 91 92

Exploitative abuses: recent trends and comparative perspectives  113 While the Bundeskartellamt erred to assess the Facebook case under Article 19(1) GWB, the Autorité de la Concurrence has recently adopted two important decisions sanctioning Google under Article 102(a) TFEU. In December 2019, the French NCA imposed a fine of €150 million on Google due to the unfair trading conditions imposed on advertisers (ie Google Ads Rules).104 According to the French NCA, a dominant firm can impose unfair trading conditions when it holds extremely high market shares … in such a case, the products offered by the dominant undertaking represent all or almost all of what the market has to offer; customers wishing to acquire them will have no choice but to accept the transaction terms determined by the dominant undertaking, however unfair they may be …105

In other words, the dominant firm is an ‘unavoidable trading partner’ for its customers. The Autorité defined unfair trading conditions under Article 102(a) TFEU as terms that ‘went beyond their ostensible purpose and were intended to strengthen (the firm) dominant position by reinforcing its customers’ economic dependence on it’.106 In that case, the French NCA ruled that Google was an unavoidable trading partner for digital advertisers. Secondly, Google unilaterally imposed the rules; digital advertisers ‘had no choice but to accept the Rules’ due to their dependence on Google.107 Thirdly, a number of rules were ‘unfair’, since they had been modified several times by Google, causing uncertainty for advertisers.108 In addition, Google applied the rules in a discriminatory manner among advertisers.109 Fourthly, the Autorité did not accept any objective justification put forward by Google.110 Finally, unlike the CoJ case law discussed in the previous section, the Autorité also assessed the potential anti-competitive effects of the rules: they increased uncertainty for the advertisers and distorted competition among the websites selling digital services.111 The decision of the Autorité in Google Ads Rules is a landmark decision, since it represents the first case sanctioning unfair trading conditions under Article 102(a) TFEU in the context of the digital economy. Contrary to the Bundeskartellamt decision in Facebook, the Autorité referred to the CoJ case law and it clarified the existing jurisprudence on Article 102(a) TFEU, by providing a definition of unavoidable trading partner, a general definition of what represents an unfair trading condition, and by adding an assessment of the anti-competitive effects of Google Ads Rules on the downstream market, in accordance with the recent effects-based approach in EU competition law.

Autorité de la Concurrence Press Release, ‘The Autorité de la Concurrence Hands Down a €150M Fine for Abuse of a Dominant Position’ (20 December 2019), https://​www​.aut​oritedelac​oncurrence​ .fr/​en/​press​-release/​autorite​-de​-la​-concurrence​-hands​-down​-eu150m​-fine​-abuse​-dominant​-position, accessed 2 November 2022. 105 Google Ads Rules (Autorité de la Concurrence) (n 15) para 357. 106 ibid para 370. 107 ibid para 431. 108 ibid para 409. 109 ibid para 420. 110 ibid paras 495−9. 111 ibid paras 439−66. 104

114  Research handbook on abuse of dominance and monopolization In April 2020, the Autorité adopted a second decision sanctioning Google for unfair trading conditions under Article 102(a) TFEU (Google News case).112 This is an interim, rather than a final, decision. The case started after the entry into force of the French Law 2019/775,113 which implemented Article 15 of the EU Copyright Directive.114 According to the new legislation, online websites had to obtain a licence from press publishers to reproduce newspaper articles. Google Search, Google News and Google Discover display ‘snippets’ – that is, short extracts of newspaper articles.115 After the entry into force of Law 2019/775, Google requested from the French press publishers a free licence and it stopped displaying article snippets from newspapers that refused to grant a free licence to Google.116 The newspapers affected by the Google ban noticed a substantial reduction in the number of visitors to their websites, and thus a decrease in their revenues from online advertising.117 In its interim decision in Google News, the French NCA followed the same steps of analysis as in Google Ads Rules. First of all, the Autorité concluded that Google was an unavoidable trading partner: French publishers had to accept the conclusion of a free licence agreement, otherwise their visibility and advertising revenues would substantially drop.118 Secondly, the imposition of a ‘free’ licence was unfair, since it went against the spirit of Article 15 of the Copyright Directive and Law 2019/775, which rather aimed at granting a fair compensation to publishers.119 Thirdly, Google did not provide any objective justification,120 while the conduct had potential anti-competitive effects: newspapers banned by Google were placed in a disadvantageous competitive position in comparison to the newspapers that concluded a free licensing agreement with Google.121 In its interim decision, the Autorité obliged Google to start negotiations with the Agence France Press (ie the association of French newspapers), in order to conclude a general licensing agreement within the industry.122 In July 2021, in view of the lack of progress in the negotiations, the Autorité imposed a fine of €400 million on Google for not having complied with the previous interim decision.123 Google News is an interim decision – that is, the Autorité still

112 Autorité de la Concurrence Press Release, ‘Droits Voisins: L’Autorité Fait Droit aux Demandes de Mesures Conservatoires Présentées par les Éditeurs de Presse et l’AFP’ (9 April 2020), https://​www​ .aut​oritedelac​oncurrence​.fr/​fr/​communiques​-de​-presse/​droits​-voisins​-lautorite​-fait​-droit​-aux​-demandes​ -de​-mesures​-conservatoires, accessed 2 November 2022. 113 Law 2019-775 tendant à créer un droit voisin au profit des agences de presse et des éditeurs de presse, 24 July 2019. The official text of the legislation in French language is available at https://​www​ .legifrance​.gouv​.fr/​jorf/​id/​JORFTEXT000038821358, accessed 2 November 2022. 114 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, [2019] OJ L-130/92. 115 Google News (Autorité de la Concurrence) (n 16) para 44. 116 ibid paras 90−95. 117 ibid paras 121−3. 118 ibid para 200. 119 ibid paras 203−8. 120 ibid paras 255−64. 121 ibid paras 218−33. 122 ibid paras 292−316. 123 Autorité de la Concurrence Press Release, ‘Rémunération des Droits Voisins: L’Autorité Sanctionne Google à Hauteur de 500 Millions d’euros pour le Non-Respect De Plusieurs Injonctions’ (13 July 2021), https://​www​.aut​oritedelac​oncurrence​.fr/​fr/​communiques​-de​-presse/​remuneration​-des​-droits​ -voisins​-lautorite​-sanctionne​-google​-hauteur​-de​-500, accessed 2 November 2022.

Exploitative abuses: recent trends and comparative perspectives  115 has to adopt a final decision on the case. However, it is a relevant case, since the French NCA applied the same framework of assessment as the Google Ads Rules case. Facebook, Google Ads Rules and Google News represent a good example of the revival of exploitative abuses in the digital economy. As discussed in the previous section, it is hard to imagine that an NCA may sanction an excessive pricing case in the digital economy. By contrast, unfair trading conditions imposed by dominant online platforms (ie digital gatekeepers) on their users might be prohibited as unfair trading conditions under Article 102(a) TFEU. The pending CoJ preliminary ruling in Facebook and the growing national jurisprudence in this field will hopefully clarify the standard of assessment of unfair trading conditions under Article 102(a) TFEU.124 A question that remains open is the future interaction between the Article 102(a) prohibition and the Digital Markets Act (DMA), recently proposed by the European Commission.125 The DMA proposal may be considered an example of sector regulation, aiming at preventing a number of market failures at the root of exploitative conduct prohibited under EU competition rules. In particular, Article 5(a) of the DMA proposal prohibits gatekeepers from merging data obtained from core platform services with data obtained from third-party services. The latter is de facto a ‘data silos’ obligation similar to the behavioural remedy imposed by the Bundeskartellamt in Facebook. If the DMA proposal was adopted in its current form, a case similar to Facebook would likely not be investigated in the future by any NCA in Europe.

IV.

DISCRIMINATORY PRICING

A.

The Legal Test

A dominant firm violates Article 102(c) TFEU when it applies ‘dissimilar conditions to equivalent transactions with other trading parties, thereby placing them at a competitive disadvantage’. Price discrimination clearly falls within the scope of ‘dissimilar conditions’.126 The discrimination can take different forms: besides classical discrimination in the form of different retail/wholesale prices, the dominant company can discriminate against its customers via selective price cuts and targeted rebates. The concept of ‘equivalent transactions’ was first interpreted by the CoJ in United Brands. According to the Court, to determine whether transactions involving the same product were indeed ‘equivalent’, the European Commission should analyse the ‘differences in transport costs, taxation, customs duties, the wages of the labour force, the conditions of marketing, the differences in the parity of currencies, the density of competition …’.127 On the other hand, the Court pointed out that the different levels of demand for bananas in different EU Member States could not be sufficient to justify a persistent prices disparity within the EU common

Request for preliminary ruling, Facebook (n 100). European Commission, Proposal for a Regulation of the European Parliament and the Council on contestable and fair markets in the digital sector (Digital Markets Act), COM/2020/842 final. 126 Marco Botta and Klaus Wiedemann, ‘To Discriminate or not to Discriminate? Personalised Pricing in Online Markets as Exploitative Abuse of Dominance’ (2020) 50 European Journal of Law and Economics 381. 127 United Brands (n 26) para 228. 124 125

116  Research handbook on abuse of dominance and monopolization market.128 United Brands case law has been consistently upheld in the subsequent CoJ case law: the Court has generally looked at the nature of the product/service sold by the dominant company to its customers and assessed whether the different supply costs faced by the dominant company made the transactions ‘equivalent’. For instance, in British Airways, the CoJ concluded that the sale of airline tickets by British Airways to different travel agents in the UK represented equivalent transactions.129 The concept of ‘competitive disadvantage’ has also been interpreted by CoJ case law. Traditionally, the Court has ‘presumed’ that price discrimination places the customer who pays the higher price for the same product/service at a competitive disadvantage in comparison to the ‘other trading partners’ (ie its competitors). In particular, the CoJ ruled in British Airways that the European Commission was not required to prove that the price discrimination caused ‘an actual quantifiable deterioration in the competitive position’ of the discriminated customer.130 Similarly, in Clearstream, the General Court did not assess whether the price discrimination had resulted in a loss of market share for the discriminated customers.131 Consequently, the General Court concluded that the strategy of price discrimination implemented by the dominant firm caused a ‘competitive disadvantage’ to its customers. The case law on competitive disadvantage has been revised by the CoJ in MEO.132 Similarly to British Airways, the Court ruled that Article 102(c) TFEU does not require the European Commission/NCA to quantify the competitive disadvantage suffered by the discriminated customer.133 On the other hand, in line with the effects-based approach to Article 102 TFEU followed by the CoJ in Intel,134 the Court added that the competition enforcer should take into consideration ‘all the relevant circumstances’ to determine whether price discrimination could produce a competitive disadvantage.135 By analogy to Intel, the CoJ ruled that the NCA should take into consideration the negotiating power of the customer of the dominant firm as regards the tariffs. Secondly, the NCA should analyse the conditions for charging those tariffs, as well as the duration and amount of the tariffs. Finally, the NCA should assess whether the dominant firm implemented a strategy aiming at excluding from the downstream market one of the trading partners ‘which is at least as efficient as its competitors’.136 Finally, while the European Commission/NCA faces the burden of proof concerning the existence of equivalent transactions and the competitive disadvantage suffered by the discriminated customers, the dominant firm can put forward objective justifications and argue that the price disparity is indeed legal.137 While objective justifications are possible in theory, they have not been accepted by the Court in practice.

ibid para 229. Case C-95/04 P British Airways plc v Commission ECLI:EU:C:2007:166. 130 ibid para 145. 131 Case T-301/04 Clearstream Banking AG and Clearstream International SA v Commission ECLI:EU:T:2009:317, para 194. 132 Case C-525/16 MEO – Serviços de Comunicações e Multimédia SA v Autoridade da Concorrência ECLI:EU:C:2018:270. 133 ibid para 27. 134 Case C-413/14 P Intel Corp v Commission ECLI:EU:C:2017:632. 135 MEO (n 132) para 28. 136 ibid para 31. 137 ibid. 128 129

Exploitative abuses: recent trends and comparative perspectives  117 B.

Recent Enforcement Trends: The Non-Enforcement Continues

This sub-section will be kept brief as, to date, there are no cases of discriminatory pricing successfully prohibited under Article 102(c) TFEU. In MEO, the CoJ extended the effect-based analysis of Article 102 TFEU to exploitative abuses. In particular, the references to the assessment of the ‘relevant circumstances’ is in line with the wording of the Intel ruling.138 On the one hand, MEO should be welcome, since it aligns the CoJ case law in relation to the analysis of different categories of abuses under Article 102 TFEU. On the other hand, MEO has substantially increased the burden of proof faced by competition enforcers to sanction forms of price discrimination under Article 102(c) TFEU. In particular, in order to assess the existence of a discriminatory strategy, the NCA would need to prove that the price discrimination is a repeated conduct; a strategy systematically implemented by the dominant firm vis-à-vis certain customers. Nevertheless, customers are often not aware of having been the subject of discrimination. Secondly, geographic discrimination could affect customers based in different countries, even outside of the NCA’s jurisdiction. Finally, the dominant firm could put forward some objective justifications. For instance, it could argue that the discriminatory pricing increased the product affordability, thereby increasing consumer welfare. Price discrimination has a mixed effect on consumer welfare, and sometimes it can increase the welfare of poorer consumers.139 In view of these considerations, the proof of a competitive disadvantage under the recent MEO case law would be a major challenge for any NCA committed to investigating a case of price discrimination under Article 102(c) TFEU. Therefore, it is not surprising that NCAs are generally reluctant to investigate this type of abuse; it seems unlikely that the coming years will witness a revival of this type of exploitative abuse.

V. CONCLUSIONS The chapter has analysed recent trends in exploitative abuses falling within the scope of Article 102(a) and Article 102(c) TFEU. In particular, the chapter has discussed the legal test developed by the CoJ case law and recent enforcement trends in relation to excessive and discriminatory pricing, as well as unfair trading conditions. While these conducts breach Article 102, since the Treaty of Rome, the European Commission and NCAs have rather focussed their enforcement priorities on exclusionary abuses. However, the recent years have witnessed a revival of exploitative abuses. By reviewing the recent enforcement trends in Europe, we can conclude that the revival of exploitative abuses is taking place only for excessive prices and unfair trading conditions under Article 102(a) TFEU. By contrast, the cases sanctioning discriminatory pricing under Article 102(c) TFEU remain extremely rare, due to the high burden of proof introduced by the CoJ in MEO.140

Intel (n 134). Dirk Bergemann and others, ‘The Limits of Price Discrimination’ (2015) 105 American Economic Review 921. 140 MEO (n 132). 138 139

118  Research handbook on abuse of dominance and monopolization Secondly, the revival of cases sanctioning excessive prices and unfair trading conditions has taken place within a limited number of industries, mostly in the energy, pharmaceutical and digital markets. Even within these industries, the revival concerns specific cases (eg the excessive prices charged by generic drugs manufacturers), rather than being a general enforcement trend in the industry. In digital markets, the revival of exploitative abuses has mostly concerned unfair trading conditions, rather than excessive pricing cases. The peculiarities of digital markets (ie close-to-zero marginal costs, winner-takes-all dynamics and zero-price markets) make it hard for an NCA to assess an excessive pricing case under the United Brands test141 and the benchmarking approach. By contrast, the recent decisions by the Autorité de la Concurrence in Google Ads Rules142 and Google News143 show that Article 102(a) TFEU may be relied on by the antitrust enforcers to prohibit the unfair trading conditions imposed by dominant online platforms on their users. Similarly, though based on Article 19(1) GWB rather than Article 102(a) TFEU, the Bundeskartellamt decision in Facebook may also be considered an example of this new enforcement trend in digital markets.144 The pending CoJ preliminary ruling in Facebook,145 and the increasing number of NCA decisions and national court rulings will hopefully clarify the standard of assessment of unfair trading conditions under Article 102(a) TFEU. Finally, it is worth mentioning that the revival of cases sanctioning excessive pricing and unfair trading conditions is closely connected to the presence of sector regulation. Sector regulation may solve the market failures causing exploitative abuses, making competition law enforcement ‘redundant’. A good example of this dynamic is represented by the REMIT146 and EU Regulation on interchange fees:147 following their entry into force, no NCA has sanctioned excessive pricing cases in the electricity wholesale and in the credit cards markets. Similarly, the pending DMA proposal might sanction a number of unfair trading practices by digital gatekeepers, making redundant the enforcement of Article 102(a) TFEU vis-à-vis unfair trading conditions in digital markets.148 The revival of exploitative abuses has taken place only in specific markets, characterized by high and stable entry barriers (ie when the dominant firm is an unavoidable trading partner), a lack of innovation incentives and a lack of sector regulation. Exploitative abuses, thus, remain an exceptional type of abuse of dominance; the revival has taken place only in the presence of very specific circumstances.

143 144 145 146 147 148 141 142

United Brands (n 26). Google Ads Rules (Autorité de la Concurrence) (n 15). Google News (Autorité de la Concurrence) (n 16). Facebook (Bundeskartellamt) (n 14). Request for preliminary ruling, Facebook (n 100). Regulation (EU) No 1227/2011 (n 61). Regulation (EU) No 2015/751 (n 64). European Commission DMA proposal (n 125).

7. A re-evaluation of the abuse of excessive pricing Kathryn McMahon1

I. INTRODUCTION A firm in a monopoly position has the power to persistently charge a price that is above the marginal cost. This higher price, constrained by the downward-sloping demand curve of the market, means that less is purchased than if the market were perfectly competitive and leads to the deadweight loss of monopoly. It causes a wealth transfer from consumers to producers. With these welfare losses, it is perverse that the monopoly price is not subject to greater scrutiny under competition law. But price regulation is thought best undertaken by sector-specific regulators with specialist knowledge.2 Competition authorities and courts refrain from the ongoing supervision of commercial relationships. Competition law views the firm’s charging of the monopoly price as a consequence of being in a monopoly position. In the EU, Article 102 of the Treaty on the Functioning of the European Union (TFEU) does not punish firms for merely being in a dominant position. In the US, similarly, Section 2 of the Sherman Act only punishes firms which ‘monopolize’ or ‘attempt to monopolize’.3 Accordingly, in the US, excessive pricing is not punished because ‘high prices’ are generally considered to be ‘self-correcting’, attracting new entry to the market. As Justice Scalia stated in the Supreme Court in Trinko: [t]he mere possession of monopoly power, and the concomitant charging of monopoly prices, is not only not unlawful; it is an important element of the free-market system. The opportunity to charge monopoly prices – at least for a short period – is what attracts ‘business acumen’ in the first place; it induces risk taking that produces innovation and economic growth. To safeguard the incentive to innovate, the possession of monopoly power will not be found unlawful unless it is accompanied by an element of anticompetitive conduct.4

Easterbrook pointed out the error costs of condemning high prices by stating that:

Thank you to the editors of this volume for their helpful comments on earlier drafts of this chapter. OECD, ‘Excessive Prices in Pharmaceutical Markets: Background Note by the Secretariat’, DAF/ COMP(2018)12, 9. 3 For an unlawful monopolization under Section 2 of the Sherman Act a plaintiff must establish the test in Grinnell which includes ‘(1) the possession of monopoly power in the relevant market and (2) the wilful acquisition or maintenance of that power as distinguished from growth or development as a consequence of a superior product, business acumen, or historic accident’: United States v Grinnell Corp 384 US 563, 570−71 (1966); FTC v Qualcomm Inc 969 F3d 974, 990 (9th Cir 2020) (‘Qualcomm’). 4 Verizon Communications Inc, v Trinko 540 US 398, 407 (2004); cf Berkey Photo, Inc v Eastman Kodak Company 603 F2d 263, 294 (2nd Cir 1979). 1 2

119

120  Research handbook on abuse of dominance and monopolization [i]f the court errs by condemning a beneficial practice, the benefits may be lost for good … If the court errs by permitting a deleterious practice, though, the welfare loss decreases over time. Monopoly is self-destructive. Monopoly prices eventually attract entry … judicial errors that tolerate baleful practices are self-correcting, while erroneous condemnations are not.5

But this assessment of error costs and the decision in Trinko are based on simple assumptions about the contestability of markets.6 Monopolies are not always self-correcting and many of these markets have significant barriers to entry.7 Notwithstanding this approach in the US, Article 102 TFEU can apply to conduct where customers or suppliers are exploited directly without the need to establish exclusionary conduct or ‘practices that cause consumers harm through their impact on competition’.8 Article 102(a) TFEU lists ‘unfair pricing’ as an abuse: ‘directly or indirectly imposing unfair purchase or selling prices or other unfair trading conditions’.9 The European Commission has generally been reluctant, however, to bring infringement decisions for ‘excessive pricing’10 but should this be so? Empirical research supports claims that higher levels of industry concentration have coincided with higher profit margins, weakened competition and income inequality.11 Price-cost margins continue to rise in particular industries such as pharmaceuticals.12 The

Frank Easterbrook, ‘The Limits of Antitrust’ (1984) 63 Texas Law Review 1, 2−3. Robert Pitofsky (ed), How the Chicago School Overshot the Mark: The Effect of Conservative Economic Analysis on U.S. Antitrust (OUP 2008); Lina M Khan, ‘Amazon’s Antitrust Paradox’ (2016) 126 Yale Law Journal 710. 7 Ariel Ezrachi and David Gilo, ‘Are Excessive Prices Really Self-Correcting?’ (2009) 5 Journal of Competition Law and Economics 249. Non-dominant firms can also exploit market power through higher prices which may not be self-correcting. The recognition by Chief Judge Posner in Khan v State Oil 93 F3d 1358, 1362 (7th Cir 1996) that ‘[a] supplier might, however, fix a maximum resale price in order to prevent his dealers from exploiting a monopoly position’ led to the US Supreme Court in State Oil Company v Khan 118 S Ct 275 (1997) ruling that maximum resale price maintenance under Section 1 of the Sherman Act could be dealt with under the rule of reason, overruling the per se violation in Albrecht v Herald Co 390 US 145 (1968). Another question is raised: what if the price is not high enough? Dominant firms may engage in strategic profit sacrificing behaviour or ‘limit pricing’ and not raise prices to monopoly levels in order to deter new entry: Oliver E Williamson, ‘Predatory Pricing: A Strategic and Welfare Analysis’ (1977) 87 Yale Law Journal 284, 294−306. 8 Case C-209/10 Post Danmark A/S v Konkurrencerådet EU:C:2012:172, [20]. The CoJ has stated that a margin squeeze is an ‘independent form of abuse’ which merely focuses on the relationship between the prices and cost. This arguably blurs the distinction between exploitative and exclusionary abuses: Case C-52/09 Konkurrensverket v TeliaSonera Sverige AB ECLI:EU:C:2011:83, [55], cf 54−6 (‘TeliaSonera’). 9 The prohibition in Article 102 is not for ‘excessive pricing’ but one of ‘unfair pricing’ which includes unfairly low pricing or predatory pricing. The chapter is confined to a discussion of ‘excessive prices’. 10 Opinion of AG Wahl in Case C‑177/16 Biedrība ‘Autortiesību un komunicēšanās konsultāciju aģentūra – Latvijas Autoru apvienība’ v Konkurences padome ECLI:EU:C:2017:286, [3] (‘AG Wahl Opinion’). 11 Steven Berry and others, ‘Do Increasing Markups Matter? Lessons from Empirical Industrial Organization’ (2019) 33 Journal of Economic Perspectives 44; Suresh Naidu and others, ‘Antitrust Remedies for Labor Market Power’ (2018) 132 Harvard Law Review 536. 12 European Commission, ‘Commission accepts commitments by Aspen to reduce prices for six off-patent cancer medicines by 73% addressing excessive pricing concerns’ (10 February 2021), https://​ ec​.europa​.eu/​commission/​presscorner/​detail/​en/​ip​_21​_524, accessed 19 January 2023. 5 6

A re-evaluation of the abuse of excessive pricing  121 COVID-19 pandemic has led to shortages of medical equipment and supplies leading to ‘price gouging’.13 Excessive pricing cases in basic commodities, utility services, and pharmaceuticals are also actively pursued in developing countries where they play an important role in competition advocacy.14 The Court of Justice (CoJ) of the EU in Telia Sonera explicitly referred to the provision in Article 102(a) TFEU as a basis for the consideration of high prices in a margin squeeze.15 Owners of patents which are used in an industry standard (standard essential patents (SEPS)) are also not able to exploit their inventions through the charging of excessive royalties and are required to allow implementers of the standard to obtain a licence on fair, reasonable and non-discriminatory (FRAND) terms.16 Renewed interest in the abuse comes from concerns about the dominance of digital platforms. While these markets are often characterized by ‘zero prices’ and low prices to obtain market share,17 they have also founded new claims for exploitative abuses18 and price discrimination through the increased use of algorithms and personalized pricing. Claims of ‘excessive pricing’ have been made against charges for ‘in-app’ purchases in digital app stores.19 Digital platforms which exert gatekeeper control over ‘ecosystems’ and/or narrow proprietary ‘walled gardens’ with a locked-in customer base are in a position to exploit this power through excessive prices. The regulation of an ‘excessive price’ may be problematic, however, in industries where a high profit margin is required for further investment and innovation. In a dynamic market, it is argued, less weight should be given to market share and concentration in the assessment of market power and more weight on the entrepreneurial capacities of the firm to innovate.20 The evidence suggests, however, that digital platforms, in particular, do not exhibit temporary dominance or the serial displacement of monopolies. The attraction of new entry

Penelope Giosa, ‘Exploitative Pricing in the Time of Coronavirus: The Response of EU Competition Law and the Prospect of Price Regulation’ (2020) 11 Journal of European Competition Law & Practice 499. 14 Richard Murgatroyd and others, ‘Excessive Pricing Regulation in China, South Africa, and other BRICS Member States’ in Tembinkosi Bonakele and others (eds) Competition Policy for the New Era: Insights from the BRICS Countries (OUP 2017). 15 TeliaSonera (n 8) [25]; cf Case C-280/08P Deutsche Telekom AG v Commission [2010] ECR I-9555, [29]. 16 Case C-170/13 Huawei Technologies Co Ltd v ZTE Corp, ZTE Deutschland GmbH ECLI:EU:C:2015:477; Unwired Planet International Ltd v Huawei Technologies (UK) Co Ltd [2020] UKSC 37. In Qualcomm (n 3) 990, the 9th Circuit Panel reversed the Northern District Court of California’s (FTC v Qualcomm Inc 411 F Supp 3d 658 (ND Cal 2019)) finding that, in the context of a FRAND dispute, profit-seeking behaviour alone is insufficient to attract antitrust liability because, citing Trinko, the charging of a monopoly price is what attracts business acumen. 17 Khan (n 6). 18 The German Bundeskartellamt found that Facebook’s collection and combination of user data from different sources was an exploitative and exclusionary abuse of dominance: Case B6-22/16 Facebook 6 February 2019. 19 Damien Geradin and Dimitrios Katsifis, ‘The Antitrust Case against the Apple App Store (Revisited)’ TILEC Discussion Paper No DP2020-035 (2020), available at SSRN: https://​ssrn​.com/​ abstract​=​3744192, accessed 3 November 2022. 20 J Gregory Sidak and David J Teece, ‘Dynamic Competition in Antitrust Law’ (2009) 5 Journal of Competition Law and Economics 581, 588, 598–601, 610. This view is based on longstanding economic debates about the effect of market structure on the contestability of technology markets: Joseph A Schumpeter, Capitalism, Socialism, and Democracy (Harper and Brothers 1942) 83. 13

122  Research handbook on abuse of dominance and monopolization is particularly difficult where direct and indirect network effects, high switching costs and consumer inertia can lead to entrenchment of a digital platform. The chapter will begin with a brief history of the abuse. It will then evaluate where instances of excessive pricing have been found in a number of EU and national court decisions. Some decisions have cautioned that the test for ‘excessive pricing’ is ‘semantic and black letter’.21 It has been held that there is ‘no single adequate method’ for determining when a price is excessive.22 The chapter will conclude that clearer development of the legal rules and limiting principles need to be established if the abuse is to be successfully enforced.

II.

HISTORICAL FOUNDATIONS OF EXCESSIVE PRICING UNDER ARTICLE 102 TFEU

Excessive pricing is based on the notion of a fair price. The notion of a ‘fair price’ has deep historical and ideological roots in Aristotle’s theory of justice and, during the medieval period, it is found in the concept of ‘usury’, of not gaining an advantage through exploiting the economic weaknesses of others, especially the poor, and the notion of a ‘just price’.23 Drawing on Aristotle, scholastics defined a ‘just price’ as the result of a voluntary and just exchange where justice is rendering to each person their due. Its relationship to the ‘market price’ is that: the market price will in the long run more or less coincide with the just person price as long as 1) no coercion, deceit, or market manipulation is involved; and 2) externalities are inconsequential or are factored into the price.24

An important context for the abuse of excessive pricing in Article 102 TFEU is the influence of ordoliberalism, developed by a group of economists and lawyers working at the University of Freiburg in the 1920s and 1930s.25 The ideas of ordoliberalism originated within the very particular historical, political and economic circumstances of Germany and Europe in the early part of the twentieth century. During the period of industrialization under the Weimar Republic, multiple cartels had been permitted to operate with impunity. The ordoliberals thought that while the state should be prevented from discretionary and arbitrary interventions in the market,26 it also had an important role in controlling private economic power to preserve 21 CMA v Flynn Pharma Limited and Pfizer Inc [2020] EWCA Civ 339 (UK Court of Appeal), [256], per Sir Geoffrey Vos (‘Flynn Pharma CA’). 22 Case C-177/16 Autortiesību un komunicēšanās konsultāciju aģentūra/Latvijas Autoru apvienība (AKKA/LAA) v Konkurences padome ECLI:EU:C:2017:689, [49] (‘Latvian Copyright’). 23 Michal Gal, ‘Abuse of Dominance: Exploitative Abuses’ in Ioannis Lianos and Damien Geradin (eds), Handbook on European Competition Law: Substantive Aspects (Edward Elgar Publishing 2013) 422. 24 Daryl Koehn and Barry J Wilbratte, ‘A Defense of the Thomistic Concept of the Just Price’ (2012) 22 Business Ethics Quarterly 501, 514−15. 25 Gregory J Werden, ‘Exploitative Abuse of a Dominant Position: A Bad Idea that now Should Be Abandoned’ (2021), available at SSRN: https://​ssrn​.com/​abstract​=​3790472, accessed 3 November 2022; David J Gerber, Law and Competition in Twentieth Century Europe: Protecting Prometheus (OUP 2001); Pinar Akman, ‘Searching for the Long-Lost Soul of Article 82EC’ (2009) 29 OJLS 267. 26 Franz Böhm, ‘Rule of Law in a Market Economy’ in Alan Peacock and Hans Willgerodt (eds), Germany’s Social Market Economy: Origins and Evolution (Macmillan 1989) 54; Razeen Sally, ‘Ordoliberalism and the Social Market’ (1996) 1 New Political Economy 233, 236; Gerber (n 25) 246ff.

A re-evaluation of the abuse of excessive pricing  123 economic freedom.27 Economic freedom was thought to be threatened by an unregulated economy which would allow the emergence of monopolies and oligopolies.28 No one individually or as a group should be subjugated and exploited by power groups.29 Competition law should be applied to prevent monopoly pricing; monopolies should be broken up; and certain anti-competitive practices, such as boycotts, loyalty rebates and price discrimination, should be forbidden per se.30 There is debate whether ordoliberal ideas actually found their way into post-war German competition law or the Treaty of Rome.31 The debate is complicated because there is no univocal view among ordoliberalists as to its fundamental principles and practice. The competition rules in the Treaty of Rome were the result of a compromise. The negotiations took place against the background of the 1956 Spaak Report.32 The ordoliberal goals of restricting monopolies per se and breaking up of concentration were eschewed in favour of an abuse provision for Article 102 TFEU.33 The regulation of monopolies per se was incompatible with an industrial policy for the rebuilding of post-war Europe that was more lenient towards cartels and permitted concentrations that could achieve economies of scale and scope (national and European champions) that could better compete against large and efficient US corporations.34 The regulation against the abuse of dominance was directed at safeguarding economic freedom against ‘dishonest and unfair competition’.35 This notion of unfair competition has specific links to the wording of Article 102(a) TFEU. It is also linked to the notion of the ‘special responsibility’ of a dominant undertaking ‘not to allow its conduct to impair genuine undistorted competition’.36 As Akman points out, a ‘tension between “freedom” and “fairness” inevitably arises as a result of ordoliberal competition policy which has as its main aim the protection of “economic freedom”’.37 The link between ordoliberal principles and ‘excessive pricing’ was implicit in AG Pitruzzella’s recent Opinion in SABAM where he argued that excessive prices in markets such as healthcare go beyond the distortion of competition to attack the fundamental values of our society, such as social equality, ‘precisely because it is

27 Viktor J Vanberg, ‘Consumer Welfare, Total Welfare and Economic Freedom – on the Normative Foundations of Competition Policy’ in Josef Drexl and others (eds), Competition Policy and the Economic Approach: Foundations and Limitations (Edward Elgar Publishing 2011) 44−71. 28 Walter Eucken, Jahrbuch für die Ordnung von Wirtschaft und Gesellschaft, 1−99 (The Competitive Order and its Implementation), translated and republished in (2006) 2 Competition Policy International 219, 224. 29 Böhm (n 26) 55−6. 30 Sally (n 26) 241. 31 Akman (n 25). 32 Intergovernmental Committee of the Messina Conference, Report by the Heads of Delegations to the Foreign Ministers (‘Spaak Report’) (21 April 1956), http://​aei​.pitt​.edu/​995/​1/​Spaak​_report​.pdf, accessed 3 November 2022. 33 Akman (n 25) 284−5. 34 Angela Wigger, ‘Debunking the Ordoliberal Myth in Post-war Europe’ in Josef Hien and Christian Joerges (eds), Ordoliberalism, Law and the Rule of Economics (Hart 2017) 171−2. 35 Hans von der Gröben, ‘Competition in the Common Market’ (Speech made during the debate on the draft regulation pursuant to Articles 85 and 86 of the EEC Treaty in the European Parliament, 19 October 1961) 11, http://​aei​.pitt​.edu/​14786/​1/​S49​-50​.pdf, accessed 3 November 2022. 36 Case 322/81 Michelin v Commission [1983] ECR 3461, [57]. 37 Pinar Akman, The Concept of Abuse in EU Competition Law (Hart 2012) 151.

124  Research handbook on abuse of dominance and monopolization framed within a legal system and is engendered by an economic culture which makes reference to the “social market economy” (Article 3(3) TEU)’.38

III.

THE COURT OF JUSTICE DECISION IN UNITED BRANDS

In the foundational case of United Brands39 the CoJ found that the imposition by a dominant undertaking of unfair purchase or selling prices was an abuse under Article 102 TFEU.40 United Brands, through its brand Chiquita, was the largest seller of bananas in the world. The Commission claimed that United Brands had abused its dominant position by selling its bananas in EU Member States at unfair and excessive prices. The prices in Germany were found to be 100 per cent higher than those charged in Ireland and there was a 20−40 per cent difference between the price of Chiquita and unbranded bananas.41 In introducing a test for ‘unfair pricing’ the CoJ asked whether the undertaking ‘has made use of the opportunities arising out of its dominant position in such a way as to reap trading benefits which it would not have reaped if there had been normal and sufficiently effective competition’.42 The Court set out a two-limb test. The first limb asks if the price is ‘excessive’ and will examine this on the basis that it ‘has no reasonable relation to the economic value of the product supplied’.43 If the answer to the first limb is in the affirmative, the second limb of the test asks ‘whether a price has been imposed which is either unfair in itself or when compared to competing products’.44 The application of this test involves complex economic and econometric modelling, the efficacy of which is largely dependent on the reliability of, and access to, data. The test emphasizes the importance of economic theory. The Court in United Brands stated ‘[o]ther ways may be devised − and economic theorists have not failed to think up several – of selecting the rules for determining whether the price of a product is unfair’.45 A.

The First Limb: Is the Price ‘Excessive’?

The Court stated that whether the price is ‘excessive’ can be determined ‘objectively’ by ‘making a comparison between the selling price of the product in question and its cost of production’.46 Economists will determine this difference as an ‘overcharge’: a markup on the price (‘benchmark price’) that would have been charged in a competitive market. The initial difficulty with the United Brands test is that it does not coincide with how economists 38 Opinion of AG Pitruzzella in Case C‑372/19 Belgische Vereniging van Auteurs, Componisten en Uitgevers CVBA (SABAM) v Weareone.World BVBA, Wecandance NV ECLI:EU:C:2020:598, [26] (‘AG Pitruzzella Opinion’). 39 Case C-27/76 United Brands v Commission [1978] ECR 207 (‘United Brands’). 40 ibid [248]. 41 ibid [240]. 42 ibid [249]. This seems to require a proof of causality between the dominant position and the ‘excessive price’ but no link of causality is required under Article 102 and an abuse can be found ‘regardless of the means and procedure by which it is achieved’: Case 6/72 Europemballage Corp and Continental Can Co Inc v Commission EU:C:1973:22, [27]. 43 ibid [250]. 44 ibid [252]. 45 ibid [253]. 46 ibid [251].

A re-evaluation of the abuse of excessive pricing  125 approach the determination of an ‘overcharge’. The test uses a cost-price analysis and only then is it examined to determine if it is unfair ‘in itself’ or ‘when compared to competing products’. But economists will determine an ‘overcharge’ by a number of methods drawing on both cost-price and comparators.47 The examination of these complex issues is not unknown to competition law and forms the foundation for the determination of private damages. The Damages Directive states that the ‘overcharge’ is the difference between ‘the price actually paid and the price that would otherwise have prevailed in the absence of an infringement’.48 The Commission states that various methods can be used to construct a non-infringement scenario for the purposes of quantifying the harm in damages actions.49 A ‘cost-based’ method, often called ‘cost-plus’, uses production costs for the affected product and a markup for a ‘reasonable’ profit margin ‘to estimate the hypothetical non-infringement scenario or finance-based approaches that take the financial performance of the claimant or the defendant as a starting point’.50 The CoJ in United Brands appreciated that calculating production costs was difficult as it may have to take into account indirect costs, variable costs due to scale of the enterprise, the territorial area, the number of products it manufactures and its number of subsidiaries.51 This calculation is particularly difficult if technology or intellectual property rights are concerned or where direct and indirect network effects arise in multi-sided and platform markets. B.

The Second Limb: Unfair ‘In Itself’ or ‘When Compared to Competing Products’

The second limb of the test in United Brands has two parts. The first part asks whether the price is ‘unfair in itself’. In his Opinion in Latvian Copyright, AG Wahl stated that a price can be unfair in itself where the ‘unfairness of a price can be determined without the need to make any comparison with similar or competing products’.52 In General Motors,53 General Motors, which was found to be in a dominant position, charged a fee for a conformity with standard certification for vehicles imported into Belgium from other Member States. The CoJ stated that a price could be found to be excessive if it: has the effect of curbing parallel imports by neutralizing the possibly more favourable level of prices applying in other sales areas in the community.54

The fee did not have any other reasonable justification or relationship to the quantity of services provided. 47 See Cento Veljanovski, Cartel Damages: Principles, Measurement, and Economics (OUP 2020) 190−99, ch 13, 14 48 Damages Directive, Article 2(20). 49 European Commission, ‘Staff Working Document practical guide quantifying harm in actions for damages based on breaches of Article 101 or 102 of the Treaty on the Functioning of the European Union’, SWD (2013)205, [26] (‘Quantifying Damages’). 50 ibid [28]. 51 United Brands (n 39) [254]; Peter Davis and Vivek Mani, ‘The Law and Economics of Excessive and Unfair Pricing: A Review and a Proposal’ (2018) 63 The Antitrust Bulletin 399. 52 AG Wahl Opinion (n 10) [121]. 53 Case 26/75 General Motors Continental NV v Commission [1975] ECR 1367. 54 ibid [12].

126  Research handbook on abuse of dominance and monopolization In Grüne Punkt-Duales System the CoJ confirmed a decision of the Commission finding that Der Grüne Punkt-Duales System Deutschland had abused its dominant position in the market for the collection and recycling of sales packaging due to the unfair structure of its licence fee for use of its trademarked logo ‘Green Dot’.55 The payment of the fee did not vary with the actual use of the collection services and was imposed solely on the use of the logo.56 The test of ‘unfair in itself’ has therefore tended to have a limited application to infringements of the single market objective or where there is a request for payment for which no product or service is provided. The second part of the second limb asks whether the price is unfair when compared to competing products. A comparator-based method compares the price in the infringement scenario with a non-infringement scenario. A number of methods can be used including a time-period approach by looking at prices in the time periods (adjusted for different costs) before or after the infringement on the same market or in other markets that have not been affected by the infringement. These other markets can include different but similar geographic markets or a different but similar product market.57 This approach attempts to establish a benchmark for comparison purposes, ‘as if’ there had been effective competition. The CoJ ultimately did not find an abuse in United Brands, as the Commission had ‘not adduced adequate legal proof of the facts and evaluations which formed the foundation of its finding’.58 The Commission was under a duty to require United Brands to ‘produce particulars of all the constituent elements of its production costs’.59 The Court was critical of the Commission’s reliance on the substantial differences in prices charged in Germany, Denmark and the Netherlands that were up to two and a half times and 100 per cent higher than the prices charged in Ireland.60 The Commission stated that it was ‘arbitrary’ to rely on these prices when the Irish market only constituted 1.6 per cent of the total volume sold.61 United Brands argued that it had produced a loss in four out of the five previous years and the prices in Ireland were to achieve entry to a new market.62 The Court, although questioning some of the evidence presented by the defence, did not find that the 7 per cent higher price charged by Chiquita could be excessive.63 The Court also accepted that for nearly 20 years banana prices had not risen in real terms on the market.64

55 C-385/07 P Der Grüne Punkt- Duales System Deutschland GmbH v Commission ECLI:EU:C:2009:456; Commission Decision 2001/463/EC of 20 April 2001 (Case COMP D3/34493 – DSD). 56 ibid [32], citing recitals 111 to 115 of the Commission decision. 57 Quantifying Damages (n 49) [33]. 58 United Brands (n 39) [267]. 59 ibid [256]−[257]. 60 ibid [239]. 61 ibid [243]−[244]. 62 ibid [261]−[262]. 63 ibid [266]. 64 ibid [265].

A re-evaluation of the abuse of excessive pricing  127

IV.

EU DECISIONS AFTER UNITED BRANDS

Subsequent EU decisions have grappled with the interpretation of the test in United Brands. In particular, the meaning of ‘economic value’ and what is an appropriate and adequate comparator when examining prices for competing products. A.

The Meaning of ‘Economic Value’ and the Decision in Scandlines

The first limb of the test in United Brands will find a price to be ‘excessive’ if it ‘has no reasonable relation to the economic value of the product supplied’.65 ‘Economic value’ has not always been consistently interpreted in the case law according to its meaning in economic theory where it takes into account both demand-side (consumer surplus: a measure of the willingness of the consumer to pay above the price actually paid) and supply-side factors (producer surplus).66 In Scandlines Sverige AB v Port of Helsingborg, the Commission rejected a complaint by Scandlines that the fees charged by the Helsingborg port operator, Helsingborgs Hamn AB (‘HHAB’), were excessive and discriminatory.67 In its assessment of ‘economic value’, the Commission stated that the actual production costs incurred by the port were not the only costs to take into account. It was important to consider non-cost factors ‘which are not reflected in the audited profits and losses of HHAB’.68 These included sunk costs (including the costs needed to rebuild the facilities), intangible values such as the premier location of the port and the opportunity cost if the land was used for other purposes.69 But how should ‘economic value’ be determined in the context of a corporate entity, wholly owned and operated by the City of Helsingborg? Do these legacy public investments require different considerations so that the terms ‘sunk costs’, ‘opportunity costs’ and ‘intangible values’ such as the location acquire different meanings? As the Commission acknowledged, the port was able to cross-subsidize many of these costs. The Commission concluded that there was insufficient evidence that the port fees were unfair when compared to other ports because it could not draw a meaningful comparison due to the lack of equivalency in services provided and different cost structures.70 Some of the comparator ports were also state owned and did not depreciate assets, making a comparison of costs difficult.71 The Commission included within ‘economic value’ demand-side factors such the consumers’ willingness to pay. The value of the services to the consumer and intangible values such as the location to the ferry operators were considered.72 But unless an adequate counterfactual

United Brands (n 39) [250]. Davis and Mani (n 51) 421−3; Robert O’Donoghue and Jorge Padilla, The Law and Economics of Article 102 (Hart 2013) 746. 67 Scandlines Sverige AB v Port of Helsingborg (Case COMP/A-36.568/D3), 23 July 2004 (‘Scandlines’). 68 ibid [209]. 69 ibid [209]−[210], [224]. 70 ibid [162]−[207]. 71 ibid. In the context of a margin squeeze, the Commission in Telefónica stated that ex ante investment incentives were less important if the undertaking had benefited from special or exclusive rights that shielded it from competition: Wanadoo España v Telefónica (Case COMP/38.784) Commission Decision of 4 July 2007 [304]. 72 Scandlines (n 67) [226], [234]−[235]. 65 66

128  Research handbook on abuse of dominance and monopolization is considered it is unhelpful to conjecture about the value to the consumer when the port can be classified as an ‘essential facility’ and an ‘unavoidable trading partner’. It is also important to consider whether customers/consumers can exert any countervailing power. B.

The Use of ‘Comparator Data’ and the Decisions in Tournier, Latvian Copyright and SABAM

The second part of the second limb of the test in United Brands asks whether the price is unfair when compared to competing products. In the EU, the examination of competing prices often involves the consideration of the prices in other Member States. But how many Member States should be compared and how can different structural, cultural and economic circumstances be considered? In making these comparisons, the single market objective is also clearly of primary importance, but the achievement of price parity has never been an objective, in itself, of the EU. In a series of EU cases, excessive pricing claims have been made against copyright collecting societies. In Tournier,73 a reference was made for a preliminary ruling to the CoJ from the Cour d’appel d’Aix-en-Provence, France, on the amount of licence fees charged to discotheques by SACEM, a French copyright collecting society. The CoJ found that SACEM’s scale of fees were ‘appreciably higher’ when a comparison is made with other Member States on a ‘consistent basis’.74 SACEM tried to justify the differences by the high level of copyright protection in France and its different legislative basis.75 Their collection extended to small amounts for small users in response to the wishes of authors.76 The Court rejected these arguments and focused on the higher operating costs of SACEM as compared to those in other Member States. The Court found that the proportion of receipts taken up by ‘collection, administration and distribution expenses rather than by payments to copyright holders is considerably higher’77 and that this heavy burden of administration was due to the ‘lack of competition on the market’.78 This decision demonstrates the difficulties in the assessment of prices and costs for intangible values such as copyright. A comparison across Member States of administrative cost is not useful when there are different legislative frameworks, employment laws, taxation, pension costs, ‘social habits and historical traditions’.79 The decision in AKKA/LAA80 (Latvian Copyright) concerned another copyright collective management organization. A preliminary reference for a ruling was made to the CoJ from the Supreme Court of Latvia. The Latvian Competition Council had found that AKKA, the Latvian Copyright organization, had abused its dominant position by imposing excessive licence fees for the public performance of music played in retailers and service providers. 73 Case 395/87 Ministère public v Jean-Louis Tournier [1989] ECR 2521, ECLI:EU:C:1989:319. Joined cases 110/88, 241/88 and 242/88 Lucazeau and others v Société des Auteurs, Compositeurs et Editeurs de Musique (SACEM) and others [1989] ECR 2811, ECLI:EU:C:1989:326 (‘Tournier’). 74 ibid [38]. 75 ibid [39]. 76 ibid [41]. 77 ibid [42]. 78 ibid [42]. 79 ibid [37]. 80 Latvian Copyright (n 22).

A re-evaluation of the abuse of excessive pricing  129 The CoJ provided guidance on what are sufficient comparators and ‘appreciably higher’ fees. The fees charged by AKKA exceeded the EU average by 50−100 per cent. A comparison of rates in the neighbouring Member States, Lithuania and Estonia, was not considered ‘insufficiently representative’81 although AG Wahl stated in his Opinion that comparison with just two countries would not produce reliable results.82 It was enough that the Member States had been selected in accordance with ‘objective, appropriate and verifiable criteria and that the comparisons are made on a consistent basis’.83 In making this comparison, however, account needed to be taken of ‘consumption habits and other economic and sociocultural factors, such as gross domestic product per capita’.84 To take account of differences in living standards and purchasing power, figures should be adjusted in accordance with a purchasing power parity index (PPI) based on the gross domestic product.85 A comparison of rates will be ‘appreciably higher’ if the difference is ‘significant’ and ‘persists for a certain length of time and must not be temporary or episodic’.86 The copyright management organization may be able to justify the difference by relying on objective dissimilarities between the Member States compared.87 The amount paid to rightsholders was important.88 Unlike Tournier, the level of administrative fees did not seem unreasonable as evidence of ‘inefficient management’.89 The Court emphasized that it fell to the competition authority to define its framework and that in doing so it ‘has a certain margin of manoeuvre and that there is no single adequate method’.90 In his Opinion, AG Wahl also emphasized the importance of combining several methods in stating that: in the absence of an ubiquitous test and given the limitations inherent in all existing methods, it is … crucial that in order to avoid (or, more correctly, to minimise) the risk of errors, competition authorities should strive to examine a case by combining several methods.91

He cautioned that ‘the combined application of several imprecise methodologies … may not lead to a more reliable conclusion’ and the importance of examining ‘the convergence of results’.92 The assessment of comparators could allow a ‘sanity-check’ of the benchmark price.93

ibid [40]. AG Wahl Opinion (n 10) [67]. The Latvian Competition Council had also considered the rates applied in other Member States. 83 Latvian Copyright (n 22) [41], [51], approving AG Wahl Opinion (n 10) [61]. 84 ibid [42]. 85 ibid [46]. 86 ibid [56], [55], approving AG Wahl Opinion (n 10) [107]. 87 ibid [57], [61]. 88 ibid [58]. 89 ibid [59]. 90 ibid [49], approving in part AG Wahl Opinion (n 10). 91 AG Wahl Opinion (n 10) [43], [44], noting this was the approach of the UK Competition Appeals Tribunal (CAT) in Napp Pharmaceutical v DG of Fair Trading [2002] CAT 1, [56]−[69], [390]–[405] (‘Napp Pharmaceutical’). 92 ibid [45], references omitted. 93 ibid [124]. AG Wahl suggests rather confusingly that the ‘unfairness of a price can be determined without the need to make any comparison with similar or competing products’: ibid [121]. 81 82

130  Research handbook on abuse of dominance and monopolization SABAM was a reference to the CoJ for a preliminary ruling from the Ondernemingsrechtbank Antwerp (Companies Court, Antwerp) concerning the royalties collected by the Belgium collecting agency, Belgische Vereniging van Auteurs, Componisten en Uitgevers CVBA (‘SABAM’) for the use of musical works at festivals organized by two companies Weareone. World BVBA (‘W.W’) and Wecandance NV (‘WCD’).94 The method used by SABAM for the calculation of the fee used a staggered flat-rate system based on the gross profits of the festivals. The festival organizers argued that there were alternative methods of calculation, making use of software technologies, that permitted a more accurate and precise measurement of usage and that this method did not take account of their costs of staging and promotion. While such technologies and the increasing use of streaming services can mean that such ‘per use’ approaches to the licence fee may become more common, European collecting agencies do not generally permit ‘per use’ licences in addition to a blanket licence or flat fee. In Tourner the charging of a flat fee was not in itself an abuse and could only be criticized ‘if other methods might be capable of attaining the same legitimate aim’.95 The Court in SABAM stated that a percentage of the gross receipts for ticket sales, without deductions for staging costs, was an acceptable method of calculation as long as it was not excessive and was based on the economic value of the licensed works.96 A more precise calculation method, through the use of technologies, was appropriate as long as it did not confer excessive and disproportionate administrative costs on the collecting society. It was for the national court to decide on the appropriate method, taking into account all of the circumstances, including the availability and reliability of these technologies and the protection of the rights holders.

V.

EXCESSIVE PRICING OF PHARMACEUTICALS: THE DECISION OF THE UK COURT OF APPEAL IN FLYNN PHARMA

The UK Court of Appeal decision in Flynn Pharma provides a comprehensive examination of the test for excessive pricing in the context of the pharmaceutical industry.97 The Court made observations about the determination of the benchmark test, the measurement of ‘economic value’, the interpretation of the second limb of the United Brands test, and the duty of the regulatory authority. The UK’s Competition and Markets Authority (CMA) had found that Pfizer and Flynn were dominant in the markets for the manufacture and distribution of phenytoin sodium capsules, a drug used in the treatment of epilepsy. The CMA found they had charged excessive prices for these drugs98 and imposed a fine of £84.2 million on Pfizer and £5.2 million on Flynn and 94 Case C-372/19 Belgische Vereniging van Auteurs, Componisten en Uitgevers CVBA (SABAM) v Weareone.World BVBA, Wecandance NV ECLI:EU:C:2020:959 (English translation not yet available) (‘SABAM’). 95 Tournier (n 73) [45]. 96 SABAM (n 94). The Court seemed to suggest that a fee charged by a collecting agency could be regarded as excessive merely by a consideration of the relationship between prices and cost: citing Tournier (n 73); C-52/07 Kanal 5 and TV 4 EU:C:2008:703 and Case C-351/12 OSA EU:C:2014:110. 97 Flynn Pharma CA (n 21). The UK case was decided pre-Brexit on the basis of the test under Article 102. 98 Competition Act 1998, Section 18 (UK) and Article 102 TFEU.

A re-evaluation of the abuse of excessive pricing  131 ordered a reduction in prices.99 Both Flynn and Pfizer appealed to the UK Competition Appeal Tribunal (CAT).100 The CAT found that the CMA had made errors of law and therefore set aside the penalties and remitted the question of abuse to the CMA for redetermination. The CMA appealed to the Court of Appeal which rejected the majority of the grounds of appeal.101 Phenytoin sodium capsules, branded Epanutin, were manufactured by Pfizer and exclusively distributed in the UK by Flynn. Patent protection for the drug had expired but the drug was still used under clinical guidance in the UK by a number of epilepsy sufferers (about 48,000 patients). Another company, NRIM, also supplied the drugs to the UK market. The drug was de-branded (‘genericized’) by Flynn in September 2012 and sold as Phenytoin. As it was de-branded it was no longer subject to price regulation under the National Health Service (NHS) Pharmaceutical Price Regulation Scheme (PPRS) and the prices charged by Flynn to customers were between 25 and 27 times higher than previous sales by Pfizer (pre-September 2012).102 NHS expenditure on phenytoin sodium capsules increased from about £2 million a year in 2012 to about £50 million in 2013. A.

The Sale of ‘Tail-end’ Genericized Drugs

Flynn specialized in acquiring ‘tail-end’ or ‘end-of-life’ pharmaceutical products which were off-patent but still in demand. The UK Regulatory Agency’s practice of ‘Continuity of Supply’103 where a patient, once stabilized upon a particular manufacturer should not be changed to another, resulted in Pfizer and Flynn effectively having a captive customer base. The Court cited with approval an OECD Background Paper on excessive prices in pharmaceutical markets.104 While ex post intervention for excessive pricing was considered rare, there were circumstances which can arise in certain sectors, such as pharmaceuticals, where it was important to apply the prohibition. This includes business strategies where firms may identify ‘market segments where prices could be successfully increased’.105 The manufacturer is able to exercise market power because the supply is limited or the market is small and declining, so that no competitors will enter the market and inter-brand competition will not constrain prices.106 The CMA noted that an undertaking’s dominance makes it ‘an unavoidable trading partner’107 where ‘those seeking a particular product or service are placed in a position of

99 CMA, Case CE/9742-13, Unfair pricing in respect of the supply of phenytoin sodium capsules in the UK, 7 December 2016 (‘Flynn Pharma CMA’). 100 Flynn Pharma Limited, Flynn Pharma Holdings Limited v CMA [2018] CAT 11 (‘Flynn Pharma CAT’). 101 Flynn Pharma CA (n 21). Following the Court of Appeal’s judgment, the CMA, on 5 August 2021, issued a statement of objections provisionally finding that the parties have infringed competition law by charging unfairly high prices for phenytoin sodium capsules: https://​www​.gov​.uk/​government/​news/​cma​ -accuses​-pharma​-firms​-of​-illegal​-pricing, accessed 3 November 2022. 102 Flynn Pharma CA (n 21) [38]. 103 This principle was used by the UK’s Medicines and Healthcare Products Regulatory Agency (MHRA). 104 Flynn Pharma CA (n 21) [102], citing OECD (n 2). 105 ibid [103]. 106 ibid. 107 Flynn Pharma CMA (n 99) [4.203], citing Case C-179/16 F Hoffmann-La Roche ECLI:EU:C:2018:25, [41].

132  Research handbook on abuse of dominance and monopolization economic dependence’.108 Pfizer was able to avoid reputational risk from significant price increases by the strategic insertion of Flynn into the supply chain as the exclusive distributor ‘rather than to benefit patients or add value to the NHS’.109 Charging excessive prices is opportunistic behaviour for which the NHS has little recourse to defensive strategies. The NHS manages uncertainties about drug prices by regulation and price caps but once the drug became genericized it was no longer subject to price regulation and the NHS could be exploited.110 While the Department of Health/NHS as a near monopsony purchaser should have been in a position to exercise countervailing power, it had diminished bargaining power in this instance and was only able to lodge a complaint to the competition regulator.111 B.

The Second Limb in United Brands: Alternatives or Cumulative?

The decision is significant for its interpretation of the second limb of the United Brands test. The CMA had used a benchmark of a return on sales (ROS) of 6 per cent and return on capital (ROC) of 21 per cent for the drug and had noted that these price increases had not been introduced in other EU Member States. The CMA read the second limb of the United Brands test as ‘alternative and not cumulative’.112 Having found that the prices were unfair in themselves, it did not find it necessary to reach a conclusion about whether those prices were unfair when compared to competing products.113 The CMA did go on to consider a comparison ‘for completeness’ because the parties had submitted representations on these points.114 The CAT was highly critical of the CMA’s methodology for calculating the benchmark price. The CMA should have taken account of comparable products, in particular the phenytoin sodium tablets which were sold at 25 per cent higher prices than the price of the capsules.115 The CAT preferred the approach of AG Wahl in his Opinion in Latvian Copyright, which emphasized that, as there was not an ubiquitous test and that each test had limitations, it was necessary to apply a combination of different tests to avoid errors.116 The CAT preferred to use a number of methods in a ‘triangulation’.117 If there is reliance on one alternative the other should be used as a ‘sanity check’.118 Accordingly, the CAT had found the reliance on one test to be an error of law and remitted to the CMA for further consideration. On appeal, the Court of Appeal stated that the CMA’s reading of the second limb in United Brands as mutually exclusive alternatives was unhelpful and ‘unduly rigid and literal’ and

ibid [4.203]. Flynn Pharma CA (n 21) [131]; cf OECD (n 2) [52]. 110 This opportunistic behaviour is similar to the hold-up problem when a contract is incomplete: Oliver Hart, ‘Incomplete Contracts and Control’ (2017) 107 American Economic Review 1731. 111 Flynn Pharma CA (n 21) [36]. 112 Flynn Pharma CMA (n 99) [5.477]. 113 ibid [5.476]. 114 ibid [5.478]. 115 Flynn Pharma CAT (n 100) [464]. 116 ibid [293] [294], citing AG Wahl Opinion (n 10) [45]. The CAT noted the General Court in T-306/05 Scippacercola and Terezakis v Commission EU:T:2008:9 and the Commission in Scandlines (n 67) had found that the two alternatives in the second limb were not cumulative: ibid [366]. 117 ibid [294(9)]. 118 ibid [294(14)], [368] citing AG Wahl Opinion (n 10). 108 109

A re-evaluation of the abuse of excessive pricing  133 ‘invests far too much significance in the disjunctive “or”’.119 It was important to read the decision contextually as it had several ambiguities.120 Sir Geoffroy Vos stated: [T]his issue seemed to turn on semantic and black-letter reasoning. As I have said, CJEU decisions and AG opinions are not to be construed as deeds.121

The Court noted that the CoJ in Latvian Copyright had not agreed that a combination of approaches was mandatory122 but did state that ‘no single method’ had emerged from a reading of the legal and economic literature.123 If there was a need for a benchmark, it was for the competition authority to choose. As long as the benchmark is ‘capable of providing a “sufficient” indication that the prices charged are excessive and unfair’.124 It did not agree with the CAT that a hypothetical benchmark price was mandated in all cases.125 C.

The Determination of ‘Economic Value’

The Court of Appeal offered observations on the meaning of ‘economic value’ in the pharmaceutical context. Economic value is a legal test ‘but, at base, it is an economic concept which describes what it is that users and customers value and will reasonably pay for’.126 In his Opinion in Tournier, AG Jacobs had observed that the French discotheques had no choice but to pay whatever price was charged for the blanket licence because they were completely dependent on the supply of music and ‘because of the absence of competition [those users] must, in effect pay whatever price is required’.127 The CMA had argued, on the basis of AG Jacobs’ view, that the patient benefit should be nil because the nature of the relationship was one of dependency. The patient was unable to switch away from the drug because of clinical guidance. The NHS had no option but to pay the price demanded.128 The Court rejected the view that AG Jacobs was: laying down an absolute and immutable rule that whenever there is dependency there was no residual scope for any economic value to arise … Even if there is dependency there might still be some economic value but not necessarily reflecting the full price demanded.129

In his Opinion in Latvian Copyright, AG Wahl states that in the determination of ‘economic value’ it is relevant to take into account the capacity and willingness of the customers to Flynn Pharma CA (n 21) [57]; cf [108]. ibid [64] citing the UK Court of Appeal in Attheraces Limited v The British Horseracing Board Limited [2007] EWCA Civ 38, ‘it would be wrong to read this passage too literally’: [115]. 121 ibid [256]. 122 ibid [84]. The Court noted that the CAT in Napp Pharmaceutical (n 91) had used a combination of approaches but it did not address the legal issue of whether a combination was mandatory: ibid [87]−[94]. 123 ibid [84]. The Court noted that the CoJ in Latvian Copyright (n 22) [37] did not endorse all of AG Wahl’s Opinion but had endorsed his view that there is ‘no single method’: AG Wahl Opinion (n 10), [36]. 124 ibid [125]. Emphasis in the original. 125 ibid [125]. 126 ibid [171]. 127 AG Jacobs Opinion in Case 395/87 Tournier ECLI:EU:C:1989:215, [65]. 128 Flynn Pharma CMA (n 99) [5.254]. 129 Flynn Pharma CA (n 21) [167]. Emphasis in original. 119 120

134  Research handbook on abuse of dominance and monopolization pay for the service and the economic benefit to the customers.130 This value may be higher than the benchmark price, or underlying costs, due to perception of superior quality and ‘non-macroeconomic’ values such as the extent to which customers can increase their business activities by the music in their premises.131 The Court of Appeal in Flynn Pharma highlighted that the case ‘affects the sums paid by the public purse for drugs’132 and ‘the prices charged exerted a material and detrimental effect upon the NHS, costing approximately £50m pa with no improvement in patient care and leading to disinvestment in other medical services’.133 If, as AG Pitruzzella stated in his Opinion in SABAM, the availability of medicines has an effect on social cohesion and social inequality,134 should we be more willing to find that the price is excessive because the purchaser is a public enterprise with limited resources? If we should take account of the special status of the state as purchaser, we should also consider that the NHS as a monopsony purchaser will have countervailing market power even outside the role of the price regulatory agency. In this case, for example, the Department of Health was able to intervene and negotiate with the principal manufacturer a reduction in the price of the phenytoin sodium tablets.135 D.

The Duties of the Regulatory Authority

The Court of Appeal went on to consider the duties of the regulator in its assessment of the evidence. The CMA has a duty to conduct a fair evaluation of all the evidence before it but it has a margin of manoeuvre as to how to perform that duty and there is no general duty to perform a full investigation in all cases.136 The competition authority has the burden of proof and its jurisdiction is treated as quasi-criminal because of the vast fines which can be imposed.137 Competition authorities are subject to the duties of good administration.138 The extent of the duty may include the need to take proactive steps to request information from third parties.139 It cannot ignore prima facie relevant evidence adduced by an undertaking.140 The evidence must be reliable, credible and substantiated.141 To discharge its onus of proof and taking account of the presumption of innocence, it had a duty to properly and fairly evaluate the comparator evidence regarding the price of the phenytoin sodium tablets offered by the undertakings as part of their defences.142 AG Wahl Opinion (n 10) [63], [90]. ibid [93], [128]. 132 Flynn Pharma CA (n 21) [1]. 133 ibid [131]; cf Flynn Pharma CAT (n 100) [5]. 134 AG Pitruzzella Opinion (n 38) [26]. 135 Flynn Pharma CA (n 21) [22]. 136 ibid [113]. 137 ibid [115]. Competition law is treated as a species of criminal law: ibid [136] citing Case C-235/92 P Montecatini v EC Commission [1999] ECR I-4575, [175]−[176]. 138 ibid [115]. 139 ibid [116]. 140 ibid [84]−[85], [88]−[89] citing Case C-413/14 P Intel v Commission ECLI:EU:C:2017:632, [137]−[147]. 141 ibid [114], [265] citing the General Court in Case T-216/13 Telefonica v Commission ECLI:EU:T:2016:369. 142 ibid [117]; cf [270]. 130 131

A re-evaluation of the abuse of excessive pricing  135

VI.

THE SEARCH FOR LIMITING PRINCIPLES

This review of the case law of the EU and national courts reveals that there is no ‘one single adequate method’ for the determination of an ‘excessive price’ and the regulatory authority has a margin of manoeuvre to devise its own framework. Given this uncertainty, it becomes extremely difficult to identify clear legal standards for the abuse.143 If error costs are to be avoided and more clarity is to be provided to undertakings, limiting principles need to be established. Unfortunately, the Commission and the Courts have spent little time on defining these broader legal principles in the case law. A.

Is ‘Excessive Pricing’ an Exclusively Economic Test?

In many ways the CoJ in United Brands set the foundation for these specification difficulties by suggesting that the test would evolve as ‘other ways are devised’ and economic theorists would not fail to ‘think up several’.144 But if the test is one devised by ‘economic theorists’ very few of the decisions cite economic literature, with the exception of AG Wahl in his Opinion in Latvian Copyright and the Court of Appeal in Flynn Pharma: ‘it is proper and helpful in a case such as this to look to economic literature for insight’.145 The Court of Appeal in Flynn Pharma acknowledged that the lines of authority between law and economics were not always clear as ‘some of the literature was indicative of what the law “should be” as opposed to “what it was”’.146 The Court stated that the ‘economic literature supports the conclusion of law’.147 But what is the ‘legal test’ when the scope for economic determination is so broad? B.

Application to Regulated Markets and Those Protected by High Barriers to Entry

What limiting principles should be imposed? AG Wahl suggests that the abuse should only apply to those markets protected by high barriers to entry or expansion … [unfair prices] … can only exist in regulated markets, where the public authorities exert some form of control over the forces of supply and, consequently, the scope for free and open competition is reduced.148

Market power is generally enhanced in regulated markets but this is balanced against the pursuit of a public interest. Collecting societies, for example, pursue legitimate aims and safeguard the rights and interests of their members in circumstances where individual collection of copyright licensing fees on a per use basis would be administratively unworkable. This lowering of costs justifies a blanket licence and expands output. The US Supreme Court noted in Broadcast Music: OECD (n 2) [21], [22], cited by Flynn Pharma CA (n 21) [104]. United Brands (n 39) [253]. 145 Flynn Pharma CA (n 21) [101]. 146 ibid [101]. Emphasis in original. 147 ibid [107]. 148 AG Wahl Opinion (n 10) [48]. In regulated sectors these infringements should be confined to cases of error or regulatory failures: ibid [49]. 143 144

136  Research handbook on abuse of dominance and monopolization [t]he blanket license is composed of the individual compositions plus the aggregating service. Here, the whole is truly greater than the sum of its parts; it is, to some extent, a different product.149

These legal monopolies are often investigated under antitrust law for possible anti-competitive agreements under Article 101 TFEU or restraint of trade under Section 1 of the Sherman Act. Is the agreement indispensable to the pro-competitive or public purpose of the agreement and, if so, is it the least restrictive alternative? It was relevant in Broadcast Music that the Department of Justice had entered into a prior consent decree to restrain the exercise of market power so that members could only grant the collective agency nonexclusive rights to license their works for public performance.150 In SABAM an excessive flat fee for a blanket licence may still amount to an abuse where alternative methods exist that allow more precise calculation of the usage as long as the right holders are protected. In the EU these agreements and ‘appropriate renumeration’ are governed by the 2014 Directive on collective management of copyright.151 This is also true of the regulatory cap imposed on the interchange fees charged by credit card networks.152 These Directives balance consumer benefits against the potential for exploitation. Is this form of regulation a better approach than an excessive pricing claim under Article 102 TFEU? C.

The Concept of ‘Economic Dependency’ and the Monopolization of Aftermarkets

O’Donoghue and Padilla recommend that ‘excessive pricing’ investigations should be confined to markets where ‘consumers have no credible alternatives to the products of the dominant firm’.153 In Tournier and Latvian Copyright, the music venues had no alternative but to deal with the collecting societies. The same is true of the shipping companies and the port authority in Scandlines. In Flynn Pharma, NHS/patients were unable to switch due to the clinical guidance. The courts have identified these relationships as a form of economic dependency. Is the exploitation of this ‘installed base’ in very narrowly defined markets, which may or may not be protected by proprietary rights, a form of limiting principle for the abuse of ‘excessive pricing’? One possible constraint on the charging of an ‘excessive price’ is the reputational effect on the brand of the dominant firm. In Flynn Pharma any reputational effects on Pfizer were avoided because it was a ‘tail-end’ product and Flynn was inserted in the supply chain as distributor. The management of this reputational effect on future sales and the ability of the dominant firm to exploit an ‘installed base’ through the charging of monopoly prices are issues that have been considered in aftermarket cases. A majority of the US Supreme Court in

Broadcast Music, Inc v CBS (1979) 441 US 1, 22−3. ibid 19. 151 Directive of the European Parliament and of the Council of 26 February 2014 on collective management of copyright and related rights and multi-territorial licensing of rights in musical works for online use in the internal market [2014] OJ L84/72. 152 Regulation (EU) 2015/751 of the European Parliament and of the Council of 29 April 2015 in interchange fees for card-based payment transactions [2015] OJ L123/1. 153 O’Donoghue and Padilla (n 66) 776. They also include mature markets where investment and innovation play little role: ibid. 149 150

A re-evaluation of the abuse of excessive pricing  137 Eastman Kodak154 held that Kodak, which sold photocopiers and micrographic equipment in a competitive primary market, could monopolize the aftermarket for repair parts and service. A separate market was found to exist for the parts or service of a single brand of original equipment.155 The Court found that ‘[i]f the cost of switching is high, consumers who already have purchased the equipment, and are thus “locked in,” will tolerate some level of service-price increases before changing equipment brands’.156 The Kodak decision demonstrates that even Section 2 of the Sherman Act may be used to restrain the charging of higher prices in aftermarkets. While subsequent Federal Circuit courts have limited the application of Kodak in the US to circumstances where the aftermarket policy has been changed post-purchase of the original equipment,157 the EU has accepted the Kodak approach in a number of decisions under Article 102 TFEU.158 These decisions recognize that exploitation through the imposition of high prices in narrow/proprietary aftermarkets can be abusive and this may have an important application to digital platforms. D.

The Application of the Abuse to Digital Platforms

Digital platforms, depending on the business model, perform intermediation roles and control an ecosystem of interdependent products or services in multi-sided markets. Market power can arise through the control of narrow proprietary ‘walled gardens’ where there are direct and indirect network effects, high switching costs, little multi-homing, information asymmetries and a high degree of consumer loyalty or inertia. Like an aftermarket, lock-in on one side of the market can co-exist with a high degree of competition on the other side of the market.159 This can give rise to a form of ‘economic dependency’ which allows exploitation by excessive prices over an ‘installed base’.160 These questions are central to the current actions in the EU, US and elsewhere concerning the fees and restrictive conditions imposed for in-app purchasing on smart phones and tablet devices on both the Apple App Store using iOS (and iPadOS) operating systems, and Google Play using Google’s Android operating system. The European Commission has submitted a statement of objections against Apple for abuse of dominance in the market for the distribution of music streaming apps through its App Store in response to a complaint by the music streaming service Spotify.161 In the US, Epic, the creator of the popular online video

Eastman Kodak Co v Image Technical Services (1992) 504 US 451. ibid 481−2. 156 ibid 476. 157 PSI Repair Services v Honeywell 104 F3d 811 (6th Cir 1997); Newcal Industries, Inc v IKON Office Solution 513 F3d 1038 (9th Cir 2008). 158 Case T-30/89 Hilti AG v Commission [1991] ECR II-439; Commission decisions: Case No IV/34.330 Pelikan/Kyocera (1999); Digital Undertaking, Press Release IP/97/868 of 10 October 1997; IV/E 2/36.431 –Info-Lab/Ricoh (1999). 159 Jacques Crémer and others, Competition Policy for the Digital Era: Final Report (European Commission 2019) 49. 160 United States House Judiciary Committee, Subcommittee on Antitrust, Commercial and Administrative Law, Investigation of Competition in Digital Markets (2020) 390. 161 European Commission, ‘Commission sends Statement of Objections to Apple on App Store Rules for Music Streaming Providers’ (30 April 2021), https://​ec​.europa​.eu/​commission/​presscorner/​detail/​en/​ ip​_21​_2061, accessed 3 November 2022 (‘Statement of Objections’). 154 155

138  Research handbook on abuse of dominance and monopolization game Fortnite, sought an injunction against Apple in the Californian District Court.162 Epic claimed that Apple has illegally monopolized the iOS App distribution market and the market for in-app payment processing on iOS devices. Apple is the sole distributer of apps on iOS. It requires all in-app purchases to be made exclusively via Apple’s own proprietary In-App Purchase (IAP) system and charges app developers a 30 per cent commission. Apple also restricts developers from informing users of alternative (usually cheaper) purchasing possibilities outside of the app (anti-steering rules).163 Epic’s US claim draws on the idea of the app distribution market as an aftermarket, similar to the analysis outlined above.164 Once purchased, iOS device users face lock-in to the iOS ecosystem which operates as a ‘walled garden’.165 As we have seen from Kodak a single brand market can be established. The opacity of pricing for in-app purchasing and high switching costs can mean that there is little impact on reputation and loss of sales in the primary competitive market if excessive prices are imposed on an ‘installed base’ in an aftermarket.

VII. CONCLUSION Excessive pricing is not punished in the US under Section 2 of the Sherman Act. In the EU it has been rarely enforced but the abuse is being re-evaluated as an effective tool for the regulation of dominant firm conduct under Article 102 TFEU. Recent cases concerning lifesaving drugs suggest that these issues have importance beyond the distortion of competition and are linked to our fundamental values and ‘awareness of social inequality’.166 The EU case law states that there is no single adequate method for determining an excessive price and a competition authority may develop its own framework for which it has a margin of manoeuvre. The evaluation of a benchmark price becomes even more open-ended and elastic when opportunity cost, intangible value and the capacity and willingness to pay are used to determine ‘economic value’. In Flynn Pharma, the ‘capacity to pay’ was limited by the public Epic Games, Inc v Apple Inc, Case No 4:20-cv-05640-YGR (ND Cal), 10 September 2021 (‘Epic v Apple’). The decision is on appeal. See the statements filed by the parties: Epic Games, Inc v Apple, Findings of Fact and Conclusions of Law, Case No 4:20-cv-05640-YGR-TSH, filed 8 April 2021 (‘Epic Findings of Fact’), https://​storage​.courtlistener​.com/​recap/​gov​.uscourts​.cand​.364265/​gov​.uscourts​.cand​ .364265​.407​.0​_1​.pdf, accessed 3 November 2022; Apple’s response, Epic Games, Inc v Apple Inc, Findings of Fact and Conclusions of Law, Case No 4:20-cv-05640-YGR, filed 7 April 2021 (‘Apple Findings of Fact’), https://​www​.courtlistener​.com/​recap/​gov​.uscourts​.cand​.364265/​gov​.uscourts​.cand​ .364265​.405​.0​.pdf, accessed 3 November 2022. Cf Geradin and Katsifis (n 19). 163 In the US the antitrust claim is not on the basis of excessive pricing but rather the maintenance of a monopoly under Section 2 and tying under Section 1 of the Sherman Act. In the EU the action under Article 102 could be framed around ‘excessive pricing’ but seems to be focused on the mandatory use of the IAP and anti-steering provisions. 164 The aftermarket claim was ultimately rejected by the US District Court: Epic v Apple (n 162) 45. EVP Vestager seemed to adopt a similar approach in her Press Conference on the announcement of the claim against Apple in the music streaming market: see Dimitrios Katsifis, ‘EU Sends Apple Statement of Objections (Music Streaming Services)’ (The Platform Law Blog, 30 April 2021), https://​ theplatformlaw​.blog/​2021/​04/​30/​eu​-sends​-apple​-statement​-of​-objections​-music​-streaming​-services/​, accessed 3 November 2022. 165 ‘Walled gardens’ are common business models for digital platforms and can provide welfare benefits of security, privacy, interoperability and network effects. 166 AG Pitruzzella Opinion (n 38) [26]. 162

A re-evaluation of the abuse of excessive pricing  139 purse and the CMA classified the economic value of the drug as nil because the relationship was one of dependency. The retailers in Latvian Copyright were also in a relationship of dependency and yet AG Wahl argued that the ‘economic value’ may be higher than the benchmark price due to consumer perception of the superior quality of the product.167 This measurement of quality is particularly difficult, however, in digital markets where information asymmetries, behavioural inertia and attention manipulation can distort consumer perception. These complex economic assessments have meant that many decisions are resolved on their own facts, subject only to judicial review if the authority fails to adequately assess exculpatory evidence. This chapter suggests that clear legal rules and limiting principles need to be developed beyond the consideration of complex economic facts. Competition law, as opposed to sector-specific regulation, should determine general rules that apply to all industries. A finding of excessive pricing should be possible in any market, not just legal monopolies, where there are high barriers to entry and where the dominant firm is an unavoidable trading partner producing economic dependency which can be subject to exploitation. Intent could also be a relevant factor in cases such as Flynn Pharma, where Pfizer sought to avoid the reputational effects of the higher prices. Digital platforms, acting as gatekeepers, can also exploit intermediation power and impose excessive prices on locked-in customers within narrow proprietary markets. A failure to regulate these business models will permit the entrenchment of market power and the continued exploitation of data. The abuse of excessive pricing may therefore prove to be important weapon within the Commission’s enforcement arsenal as it undertakes the challenging task of regulating digital platforms.

AG Wahl Opinion (n 10) [63], [90], [93], [128].

167

8. Predation and discrimination John B. Kirkwood

I. INTRODUCTION Predation and discrimination are both tools of exclusion. A dominant firm can use either tactic to weaken a rival and enhance its market power, enabling it to make higher profits at the expense of its customers. While predation and discrimination can take many forms, this chapter focuses on the two types that have attracted the most attention: predatory pricing and platform discrimination. The chapter also briefly covers secondary-line price discrimination, the principal concern of the US Robinson-Patman Act (R-P Act 1936). Secondary-line price discrimination occurs when a supplier charges different prices to competing buyers, disadvantaging the disfavoured buyer. The US government has not enforced the R-P Act for many years, but some private litigation continues and under new leadership the Federal Trade Commission (FTC) may challenge secondary-line price discrimination induced by a powerful buyer like Amazon. Predatory pricing and platform discrimination represent fundamentally different ways of subverting competition. Indeed, the two practices typify the polar models of exclusion: predation and raising rivals’ costs. In the predation model, a dominant firm competes more aggressively in the short term, losing money in order to weaken or extinguish a rival. Once the rival is injured or eliminated, the dominant firm can exercise greater market power and recoup its earlier losses. In the best-known example – predatory pricing – the dominant firm prices below its incremental costs in order to force a competitor to price below its own costs. The competitor eventually retrenches or exits the market altogether, allowing the dominant firm to raise its price to a supracompetitive level, offsetting its earlier losses and harming consumers in the relevant market. In contrast, platform discrimination works by raising rivals’ costs. In the raising rivals’ costs model, a dominant firm disadvantages a competitor not by competing more aggressively itself (as in the predation model) but by forcing up the rival’s costs. The rival then raises its price or reduces its output, which permits the dominant firm to raise its own price. Platform discrimination follows this roadmap. Suppose a dominant firm (such as Amazon) operates an online platform and allows third parties to sell on its platform. And suppose the dominant firm also offers its own products in competition with some of those third-party products. By discriminating against the competing third parties – providing them with fewer services, charging them higher fees, or refusing to deal with them altogether – the dominant firm can induce them to raise prices or reduce output and thereby increase its own market power. In this model, the dominant firm uses the leverage of its platform to injure third parties and exploit consumers. As these models indicate, both forms of exclusion enhance the market power of the dominant firm. When they are unjustified, they also harm consumers. They restrict choice, raise price, and may diminish innovation. Despite these adverse consequences, US law imposes serious obstacles to challenging either practice. EU law, in contrast, is less restrictive. This 140

Predation and discrimination  141 chapter recommends changes to both sets of laws that would enhance their ability to promote competition and protect consumers.1 The impact of secondary-line price discrimination induced by a powerful buyer is more complex. In many situations, it may injure smaller buyers but benefit consumers. Suppose a powerful buyer like Amazon induces suppliers to grant it a significant discriminatory discount and passes that discount on to consumers. The lower retail price would enhance consumer welfare but take business from Amazon’s smaller competitors. If the FTC successfully sued to block this discrimination under the R-P Act, its enforcement action would protect small business but harm consumers. In contrast, some types of buyer-induced secondary-line discrimination would injure both smaller rivals and consumers. Suppose Amazon induces a discriminatory discount, uses it to drive out all its competitors, and then raises its retail prices substantially. This acquisition of monopoly power would harm competition, small buyers, and consumers. In several other scenarios, buyer-induced secondary-line discrimination would also reduce consumer welfare.2 A.

US Law

The US Supreme Court bars liability for predatory pricing unless the plaintiff proves that the defendant (1) priced below an appropriate measure of cost and (2) was reasonably likely to recoup its losses.3 These demanding requirements reflect two factors: the Court’s fear of chilling procompetitive price cuts and its sense that true predatory pricing is rare. In combination, these factors have caused the Court to be highly sceptical of predatory pricing claims. It has ruled against the plaintiff in every predatory pricing case – and every comparable case – it has considered.4 Recent economic research, however, has undermined the bases for the Court’s hostility, casting doubt on both the theoretical and empirical grounds for the Court’s posture.5 This research may have produced a shift among the lower courts, making them more willing to consider predatory pricing claims. The Supreme Court itself has not reviewed a predatory pricing case since Brooke Group. But recent lower court decisions have displayed a more even-handed approach, allowing a greater percentage of predatory pricing cases to go forward

The chapter focuses on sell-side cases – cases in which exclusionary conduct is employed by a dominant seller to enhance its market power at the expense of its customers. Parallel analysis applies to buy-side cases, in which a dominant buyer uses exclusionary conduct to create monopsony power and exploit its suppliers, such as workers or small farmers. 2 See John B Kirkwood, ‘Buyer Power and Exclusionary Conduct: Should Brooke Group Set the Standards for Buyer-Induced Price Discrimination and Predatory Bidding?’ (2005) 72 Antitrust LJ 625. 3 Brooke Group Ltd v Brown & Williamson Tobacco Corp 509 US 209 (1993). 4 See Brooke Group (n 3); Cargill, Inc v Monfort of Colo, Inc 479 US 104 (1986); Matsushita Elec Indus Co v Zenith Radio Corp 475 US 574 (1986). See also Weyerhaeuser Co v Ross-Simmons Hardwood Lumber Co, Inc 127 S Ct 1069 (2007) (rejecting liability in a predatory bidding case); Atlantic Richfield Co v USA Petroleum Co 495 US 328 (1990) (rejecting liability in a maximum resale price maintenance case). 5 See Christopher R Leslie, ‘Predatory Pricing and Recoupment’ (2013) 113 Colum L Rev 1695; Patrick Bolton, Joseph F Brodley and Michael H Riordan, ‘Predatory Pricing: Strategic Theory and Legal Policy’ (2000) 88 Geo LJ 2239. 1

142  Research handbook on abuse of dominance and monopolization and offering solid reasons for dismissing the others. The era of intense, knee-jerk scepticism may be ending.6 Platform discrimination is a newer issue. It has arisen because of repeated allegations that the tech giants – Amazon, Apple, Facebook, and Google – discriminate against third parties selling on their platforms. All the tech giants allow third parties to sell on their platforms. They also frequently offer their own products in competition with these third parties, and when they do, they sometimes disadvantage them, demoting them in search results, copying their most popular products, or even cutting them off from access to the platform.7 No US plaintiff, however, has brought a successful action against a tech giant for discriminating against third parties that sell on its platform. Here, the problem is not excessive judicial scepticism but the limits of the Sherman Act 1890, which prohibits unilateral exclusionary conduct only when the conduct is reasonably likely to create monopoly power or a dangerous probability of monopoly power. Rarely, however, does platform discrimination create actual or probable monopoly power in any relevant market, even if the conduct is unjustified, reduces competition, and harms consumers. What is needed is an amendment to the Sherman Act that would make competition-reducing behaviour illegal – and subject to significant sanctions – even if it does not result in monopoly power. Section IV.B sets forth a detailed proposal. The R-P Act addresses secondary-line price discrimination. In essence, this Depression-era statute prohibits a seller from charging different prices to different buyers if the price difference threatens to injure competition between the seller and other sellers (primary-line discrimination) or between the buyer and other buyers (secondary-line discrimination). The Act contains defences for cost justification, meeting competition, and changing conditions, as well as numerous technical requirements. The Act also addresses discrimination in promotional payments and services, and prohibits buyers from inducing a discrimination that violates the Act. This chapter focuses on secondary-line price discrimination.8 B.

EU Law

In the EU claims of predatory pricing and platform discrimination are more likely to result in liability and significant monetary penalties. Unlike the US Supreme Court, the Court of Justice of the European Union does not view predatory pricing claims with deep scepticism. To the contrary, the Court has decided all three of its predatory pricing cases against the alleged predator.9 And unlike the US Court, it does not require plaintiffs to establish a reasonable prospect of recoupment, a difference that matters considerably. As of 2013, summary judgment in the US was most often granted for defendants because the plaintiff could not supply adequate

See Section II.A. See Section III.A. This chapter refers to Facebook rather than Meta, the name of Facebook’s parent, because the dominant social network is still called Facebook. 8 Brooke Group declared that the conduct elements of a primary-line price discrimination case are identical to the conduct elements of a predatory pricing case under the Sherman Act. Specifically, a primary-line plaintiff must show both below-cost pricing and recoupment. See Brooke Group (n 3). For a more detailed introduction to the R-P Act, see Kirkwood (n 2). 9 Case 62/86 AKZO Chemie BV v Commission [1991] ECR I-3359; Case T-83/91 Tetra Pak International SA v Commission [1994] ECR II-755, on appeal Case C-333/94 P Tetra Pak International SA v Commission [1996] ECR I-595 (Tetra Pak II); Case T-340/03 France Telecom SA v Commission [2007] ECR I-107, on appeal Case C-202/07 France Telecom SA v Commission [2009] ECR I-02369. 6 7

Predation and discrimination  143 evidence of recoupment.10 Both distinctions suggest that the European high court has shown no affinity for Chicago School analysis.11 To determine whether price cutting constitutes an abuse of dominance, the Court of Justice (CoJ) applies two tests, announced in AKZO and reiterated in Tetra Pak II and France Telecom. Rousseva summarizes them as follows: First, prices below AVC [average variable costs] must always be considered abusive because, in such a case, there is no conceivable purpose other than the elimination of a competitor. Secondly, prices below average total costs but above average variable costs are only to be considered abusive if an intention to eliminate can be shown.12

These tests may be overinclusive. The first test does not recognize any justification for pricing below average variable in cost when, in fact, there are justifications.13 The correctness of the second test depends on finding an eliminatory intent only when the price cutting is likely to result in long-run consumer harm. Yet, the Court has been quick to discern an eliminatory intent, even when the challenged price cutting may have benefited consumers in both the short run and the long run.14 On the other hand, the EU’s stern approach to predatory pricing may have produced significant consumer benefits by reducing episodes of predatory pricing. While the US courts continue to consider predatory pricing cases, neither the CoJ nor the General Court (GC) has had to resolve a predatory pricing case since 2013.15 The EU approach to platform discrimination has also been more aggressive to date than the American record. While government enforcement agencies on both continents have filed complaints, only in the EU has a complaint resulted in an order requiring a tech giant to pay a fine and change its behaviour.16 In the EU, secondary-line price discrimination can be an abuse of dominance. Article 102 of the Treaty on the Functioning of the European Union (TFEU) prohibits dominant firms from abusing their position and states that an abuse may consist in ‘applying dissimilar conditions to equivalent transactions with other trading parties thereby placing them at a competitive disadvantage’.17 In recent years, however, the European courts have rarely addressed secondary-line price discrimination. Their Article 102(c) cases have generally involved other forms of discrimination. C.

Organization and Overview

This chapter reviews recent developments in predatory pricing, platform discrimination, and secondary-line price discrimination and suggests reforms in each area. Section II covers case Leslie (n 5) 1740; Ekaterina Rousseva, Rethinking Exclusionary Abuses in EU Competition Law (Hart 2010) 134 (‘The difficulty of proving recoupment … normally results in rulings that favour the alleged predator’). 11 Rousseva (n 10) 4 (‘Yet at least as regards the case law of the Courts, Chicago school ideas have never found good soil in Europe’). 12 ibid 149–50. 13 See Section IV.A. 14 See Rousseva (n 10) 139, 165. 15 See Section II.B. 16 See Section III.B. 17 Article 102(c) TFEU. 10

144  Research handbook on abuse of dominance and monopolization law in the US and the EU on predatory pricing. The most important question is whether US courts have become less averse to predatory pricing claims and more open to liability. Put differently, has US law moved closer to EU law? Recent decisions suggest that it has, though there are few decisions and none litigated to final judgment. Without more cases, it is impossible to conclude that US law has reached an appropriate position on predatory pricing. Reform is still needed. In the EU, the European Commission (EC) continues to bring cases but the courts have not had to address predatory pricing in a decade. Section III addresses platform discrimination. Here, there is evidence, noted above and discussed below, that the tech giants have discriminated, without justification, against third parties selling on their platforms. At the same time, there have been no successful law enforcement actions in the US against such discrimination. In this area as well, reform is needed. In the EU, the law and the enforcement posture are more aggressive, and one case has proceeded to final judgment. Section IV briefly reviews recent decisions involving secondary-line price discrimination. In the US, private plaintiffs continue to bring a small number of R-P actions, about three a year. While no case has resulted in a final litigated judgment, plaintiffs do survive motions to dismiss and summary judgment motions. The reported decisions do not, however, involve powerful buyers like Amazon or Walmart and do not ask whether the alleged discrimination injured consumers. In the EU, secondary-line price discrimination may constitute an abuse of dominance, but in the last decade there have been even fewer cases than in the US. Most discrimination cases address other types of discrimination. Section V recommends changes in US and EU law that would better protect competition and consumers. It concludes that in the US it is time to take a step towards the EU approach and relieve predatory pricing plaintiffs of the obligation to show recoupment. Instead, once a plaintiff establishes that the defendant priced below an appropriate measure of its costs, the plaintiff should prevail if it rebuts the defendant’s asserted justifications for below-cost pricing. Reform is also needed in the US law governing platform discrimination. The Sherman Act should be amended to prohibit any exclusionary conduct – including platform discrimination – that reduces competition significantly, whether or not it results in monopoly power or a dangerous probability of monopoly power. Neither reform is necessary in the EU; both recommended changes are features of existing EU law. A predatory pricing plaintiff does not need to show recoupment in the EU, and a plaintiff challenging platform discrimination can establish dominance without proving the equivalent of US monopoly power. At the same time, modest changes in EU law (described below) would help reduce the likelihood that challenges to predatory pricing or platform discrimination would deter desirable behaviour. If US antitrust law ought to focus exclusively on protecting consumers (and small suppliers in monopsony cases), the R-P Act is badly in need of reform. Section V closes with recommended changes in the Act that would make consumer injury, rather than competitor injury, the test of secondary-line liability and would relax the cost justification and meeting competition defences so that discrimination that does injure consumers or small suppliers can be stopped.

Predation and discrimination  145

II.

PREDATORY PRICING

A.

US Developments

The critical issue in the US is whether courts have begun to move away from the hostility to predatory pricing claims expressed in three Supreme Court decisions from the 1980s and 1990s – Brooke Group, Cargill, and Matsushita.18 This hostility, often noted,19 rests on the fear that predatory pricing actions will deter procompetitive price cutting, and the assumption that true predation seldom occurs. Both conclusions, however, have been undermined by recent scholarship.20 As the Tenth Circuit recognized, this research shows that ‘price predation is not only plausible, but profitable, especially in a multi-market context where predation can occur in one market and recoupment can occur rapidly in other markets’.21 As a result, extreme scepticism is no longer warranted: ‘Although this court approaches the matter with caution, we do not do so with the incredulity that once prevailed’.22 A survey of recent decisions suggests that this greater receptivity to predatory pricing claims may be spreading to other US courts. Judges no longer question whether predatory pricing makes sense, they do not dismiss every case, and when they reject a claim, they offer straightforward, understandable reasons.23 Since 2013, when Christopher Leslie surveyed prior decisions,24 there have been rulings in seven predatory pricing cases, six on motions to dismiss and one on summary judgment.25 Together, they suggest greater openness to predatory pricing claims. Plaintiffs won almost 30 per cent of the decisions (two out of seven), a substantial improvement over their record at the Supreme Court, where they lost every case. In addition, the courts did not lace their opinions with anti-predation rhetoric. To the contrary, while a few judges emphasized the desirability of price cuts, no court suggested that allegations of predation should be viewed with scepticism, much less incredulity. Instead, the courts treated predatory pricing claims as legitimate so long as the plaintiffs had plausibly pled the two Brooke Group elements. Moreover, in deciding whether the plausibility test was met, the courts did not appear to be excessively demanding. The grounds they gave for rejecting a claim appeared to be reasonable. In one case, for example, the plaintiff refused to allege recoupment. Its complaint did not include a recoupment allegation and when given the opportunity to replead, the plaintiff See nn 3–4. See, for example, C Scott Hemphill and Philip J Weiser, ‘Beyond Brooke Group: Bringing Reality to the Law of Predatory Pricing’ (2018) 127 Yale LJ 2048, 2058 (‘The Court’s dicta in Brooke Group … reflects a deeply skeptical attitude toward predatory pricing claims’). 20 See n 5. 21 United States v AMR Corp 335 F3d 1109, 1115 (10th Cir 2003). 22 ibid. 23 Research conducted by Edlira Kuka, a law student at Seattle University. 24 See Leslie (n 5). 25 Energy Conversion Devices Liquidation Trust v Trina Solar Ltd 833 F3d 680 (6th Cir 2016); Superior Production Partnership v Gordon Auto Body Parts Co 784 F3d 311 (6th Cir 2015); Felder’s Collision Parts, Inc v All Star Advertising Agency, Inc 777 F3d 756 (5th Cir 2015); Affinity LLC v GfK Mediamark Research & Intelligence, LLC 547 Fed Appx 54 (2d Cir 2013); Dodge Data & Analytics LLC v iSqFt, Inc 183 F Supp 3d 855 (SD Ohio 2016); Vesta Corp v Amdocs Management Ltd 129 F Supp 3d 1012 (D Or 2015); Solyndra Residual Trust v Suntech Power Holdings Co, Ltd 62 F Supp 3d 1027 (ND Cal 2014). 18 19

146  Research handbook on abuse of dominance and monopolization refused to add a recoupment allegation.26 In these circumstances, dismissal was appropriate. Since Brooke Group is still the law, the court could not approve a complaint that disregarded one of its key requirements. Likewise, in Superior Production, the plaintiff presented essentially no evidence that the defendant had priced below cost. To the contrary, the plaintiff’s expert contended that the defendant’s pricing was predatory because the defendant had priced below its short-term profit-maximizing level. Since such above-cost pricing is protected under Brooke Group, the court properly granted summary judgment to the defendant.27 In a third case, the plaintiff could not show that the defendant had priced below cost unless the large rebates the defendant had received from its supplier were disregarded. When those rebates were taken into account, the defendant’s revenues exceeded its costs. If predatory pricing requires a short-term loss, the plaintiff could not establish one.28 In Affinity, the plaintiff’s problem was not that it failed to allege one of the elements of Brooke Group but that its allegations were implausible. The Second Circuit’s assessment of plausibility, however, was not overly strict. The court upheld the dismissal of the plaintiff’s below-cost allegation because the plaintiff had used its own costs to estimate the defendant’s costs when there were good reasons to believe that the defendant’s costs were lower: the defendant had much more experience in the industry than the plaintiff and its sales volume was significantly larger. Likewise, the plaintiff’s allegation of recoupment failed because the plaintiff’s ‘own declarations depicted remarkably low barriers to entry’.29 They indicated that a small, underfunded firm could establish a toehold in the market in a matter of months with relatively little upfront investment.30 In short, it appears that all four cases were properly decided. In these cases, predation as Brooke Group defined it (below-cost pricing plus recoupment) was highly unlikely. Their results can be explained without positing a Chicago School-style hostility to predatory pricing claims. Overall, the survey is too small and too limited to provide definitive conclusions. It covers only seven cases and does not reveal whether the plaintiffs ultimately won (through settlement or judgment) any of the cases that survived motions to dismiss. But the results do suggest grounds for cautious optimism. The courts viewed predatory pricing allegations as potentially valid, not inevitably frivolous, and did not reach for dubious arguments to dispose of them. An earlier survey covering the period between 1993, when Brooke Group was decided, and 2005 concluded that ‘predation cases are not hopeless’.31 This survey suggests a brighter conclusion – that US courts may be moving away from an automatic rejection of predatory pricing claims towards an objective consideration of the claims on their merits. B.

EU Developments

The most striking EU development has been the complete absence of predatory pricing cases in recent years. Since 2013 no predatory pricing case has been decided on the merits in either

See Energy Conversion Devices (n 25). See Superior Production Partnership (n 25). 28 See Felder’s Collision Parts (n 25). 29 Affinity (n 25) 56. 30 ibid. 31 Hemphill and Weiser (n 19) 2063. 26 27

Predation and discrimination  147 the CoJ or the GC.32 Unlike the US courts, which continue to evaluate a small number of predatory pricing cases each year, the European courts have not had to address the substance of a single predatory pricing case for at least eight years. Since the EC has continued to bring cases, this means that its approach has not been successfully challenged.

III.

PLATFORM DISCRIMINATION

The tech giants – Amazon, Apple, Facebook, and Google – dominate important products and services in both the US and EU. According to critics,33 they behave like the robber barons of the Gilded Age, crushing competitors and exploiting consumers. One of the principal charges against them is that they engage in platform discrimination: they favour their own products and services over products and services that third parties offer on their platforms. All the tech giants allow third parties to sell on their platforms. They also enter into competition with some of those third parties, and when they do, the evidence set forth below indicates that they sometimes discriminate against them, using various tactics to disadvantage the third-party products and boost their own offerings. This self-favouring injures the third parties, increases the tech giants’ market power and, when unjustified, reduces consumer welfare. A.

US Developments

There are three principal ways in which the tech giants allegedly engage in platform discrimination. They bias their search results to promote their own products and demote third-party products. They use the non-public data they collect about individual third parties to copy their most popular products. And they refuse to deal with third parties simply because they are competitors. The following sections describe each tactic and illustrate it with allegations from recent lawsuits and press reports.34 1. Search bias A Wall Street Journal investigation published in 2020 concluded that Google had engaged in search bias: ‘When choosing the best video clips to promote from around the web, Alphabet Inc.’s Google gives a secret advantage to one source in particular: itself. Or, more specifically, its giant online-video service, YouTube’.35 The Journal found that Google systematically favoured YouTube in its search results even when competitors like Facebook Watch and Amazon’s Twitch carried the same or similar videos and even when the number of their views or followers was greater. Google denied that it engaged in self-favouring but did not offer an 32 There was one procedural decision: Case T-201/11 Si.mobil telekomunikacijske storitive d.d. v Commission ECLI:EU:T:2014:1096, para 67 (ruling that the European Commission had properly refused to consider a predatory pricing complaint because the Slovenian competition authority was already dealing with it). 33 See, for example, Sean Moran, ‘Elizabeth Warren Proposes Breaking up Amazon, Facebook, Google’ (Breitbart, 18 March 2019), www​.breitbart​.com/​politics/​2019/​03/​08, accessed 10 November 2022. 34 These sections are drawn from John B Kirkwood, ‘Tech Giant Exclusion’ (2022) 74 Fla L Rev 63. 35 Sam Schechner and others, ‘Searching for Video? Google Pushes YouTube over Rivals’ Wall Street Journal (New York City, 14 July 2020).

148  Research handbook on abuse of dominance and monopolization explanation for the results. The Journal’s sources maintained that Google wanted to drive traffic its way and increase its bargaining leverage with content providers, reasons that hardly justify the discrimination. The Journal’s report strongly suggests search bias. The House Antitrust Subcommittee Report issued the same year presented additional evidence of self-favouring, indicating that Google has placed its services above competing sites even when its ranking algorithm would not warrant that priority. None of these accounts, however, suggests that Google’s search bias led to monopoly power or a dangerous probability of monopoly power. In contrast, a recent lawsuit by Colorado and 37 other states alleges that Google maintains monopoly power by biasing its search results against vertical search engines – those that specialize in a particular ‘vertical’ segment like hotels, restaurants, or local repair services. According to the complaint, this bias is evident because Google does not always discriminate against vertical search engines; it does so only when a particular vertical segment is lucrative for Google. In such cases, it restricts the advertising and/or the prominence of the competing service. The states allege that Google engages in this exclusionary conduct in order to bolster its monopoly power in general search. By suppressing the revenues of vertical search engines, Google allegedly prevents them from entering the general search market by becoming a general search engine themselves, banding together with other vertical services to create a new general search engine, or partnering with an existing general search engine. It is not clear, however, that any of these scenarios is reasonably likely. What is the chance that Yelp, Trip Advisor, or Hotels.com, individually or in combination, would overcome the daunting barriers that stand in the way of becoming a direct competitor to Google in general search? Unlike the vertical search engines, Google has indexed hundreds of billions of web pages and can draw on decades of experience responding to search queries. If Google has discriminated against vertical search engines, it is much more likely that it is trying to protect its market power in vertical search. Google sells advertising in these segments and operates competing services like Flights or Hotel Units. If Google could not suppress vertical rivals, its ability to charge supracompetitive advertising rates in these markets would be diminished. In short, the stronger theory is that Google’s search bias has enabled it to preserve market power in vertical search. Given the limits of Section 2, the states cannot pursue this theory now, but an amendment to the Sherman Act, discussed below, would enable them to do so.36 2. Product copying Critics also charge that the tech firms routinely undercut third parties that sell on their platforms by copying their most popular products. The tech firms allegedly identify those products by examining the confidential data they collect on individual third parties they host. In other words, they use non-public information about specific sellers to free ride on their product ideas, depriving them of business and undermining their incentive to develop new products.37

The states may still have a viable monopolization theory, since Google’s search bias would violate Section 2 so long as it made any significant contribution to the maintenance of Google’s monopoly power in general search. 37 This is discriminatory behaviour because it is one-sided. Amazon can use the confidential data it collects about third parties to identify and copy their most popular products, but Amazon does not make 36

Predation and discrimination  149 The tech firms compound the damage when they offer their own products at lower prices. Even the possibility of this behaviour may limit the funding available to start-ups. Further, the threat of copying a rival’s product can make it easier to acquire the rival at a bargain price. A 2014 study found that when Amazon first offered private label women’s clothing, its list of products included ‘25 percent of the top items first sold through [Amazon Marketplace] vendors’.38 Six years later, The Wall Street Journal interviewed Amazon employees who admitted they studied the sales data of specific third parties to determine which private label products to offer. Although Amazon had prohibited this conduct, the employees said they ignored the rules or found ways around them. They were willing to skirt the rules because individualized non-public data helped them determine ‘how to price an item, which features to copy or whether to enter a product segment based on its earning potential’.39 Similarly, several press investigations found that Apple had upgraded its apps with ‘the features of the most popular apps that other innovators built’.40 The antitrust analysis of product copying is complicated because mimicking a rival’s product can be procompetitive. When an entrant copies a dominant firm’s product and offers it at a lower price, consumers benefit. When Amazon enters a complementary product market, most consumers benefit because Amazon matches the quality of the third-party product but charges a lower delivered price, causing total market output to increase. Likewise, it is elementary business sense for a firm to scan its market to learn of product improvements it ought to copy. To be sure, intellectual property law often prohibits such mimicking in order to protect incentives to innovate, but here the third-party products were not patented and their distinctive features were not trade secrets. Generally, then, it does not make sense to prohibit a tech giant from copying third-party products. When a tech giant learns of a product improvement from publicly available information, it should be as free as any other firm to duplicate the improvement. But when it identifies the item by using non-public data about a specific third party, its copying poses a particularly direct threat to innovation. In that circumstance, the targeted firm may well be a pioneer – the first to develop an idea – and allowing a tech giant to take a pioneering idea is especially likely to undercut innovation. In contrast, when a platform uses other information – public information about popular products or non-public information that is aggregated across multiple competitors – there is less danger that the platform will free ride on a single seller’s new idea. No empirical studies address the issue – where to draw the line on tech giant product copying – and thus any change is necessarily tentative. But the lack of empirical research is a problem with tech giant exclusion generally, and should not stop courts or Congress from making reasonable judgements. This analysis suggests it would be appropriate to restrict platforms from using non-public data about specific third parties in deciding which products to copy. This restriction would prevent the worst instances of free riding while giving the tech

comparable data available to third parties. As a result, they cannot use Amazon data to identify and copy its most popular products. 38 George Anderson, ‘Is Amazon Undercutting Third-Party Sellers Using Their Own Data?’ Forbes (Jersey City, 30 October 2014). 39 Dana Mattioli, ‘Amazon Scooped Up Data From Its Own Sellers to Launch Competing Products’ Wall Street Journal (New York City, 23 April 2020). 40 Sally Hubbard, ‘Statement on Competition in Digital Technology Markets’ (Hearing before the Senate Subcommittee on Antitrust, Competition Policy and Consumer Rights, 10 March 2020) 14.

150  Research handbook on abuse of dominance and monopolization giants considerable latitude to enter complementary markets with cheaper or better products.41 Enforcing it would require internal information, but the Journal had no trouble obtaining such information from current Amazon employees. 3. Refusals to deal Critics have also charged the tech giants with a third form of platform discrimination – refusing to deal with certain firms because they are competitors. For instance, Amazon allegedly enters into exclusive distribution arrangements with suppliers that require it to remove competing suppliers from its platform. One critic contends that these expulsions amount to illegal monopolization. But she does not identify any markets that Amazon has monopolized.42 Moreover, when she explains why the suppliers want this exclusivity, the story she tells (if valid) is procompetitive. According to her, the suppliers sell products that require customer service in physical stores. They also sometimes offer their products through third parties on the Amazon Marketplace. Because the third parties frequently discount the products, free riding occurs: consumers visit the physical stores to take advantage of the in-store service but then purchase the products online. To prevent this free riding, the brands make Amazon their exclusive online outlet. In this account, exclusivity is a response to a market failure. The account may be incorrect, but even if it is, there is no evidence that Amazon’s exclusivity arrangements have resulted in monopoly power or a dangerous probability of monopoly power. The tech giants have also been accused of naked exclusion – refusing to deal with a firm solely because it is a competitor. In 2016, Apple allegedly blocked Spotify from access to the App Store simply because it posed a threat to Apple Music. Apple denies this, and in any event, the exclusion was temporary. Spotify returned to the App Store and consumer choice was restored. Recently, more serious allegations of naked exclusion were levelled against Facebook in complaints filed by the FTC and multiple states. They assert that Facebook denied access to its application programming interfaces (APIs) to app developers that competed with it or helped others compete with it. Specifically, Facebook allegedly adopted a policy that barred API access to apps that included a core functionality of Facebook or linked to competing social networks. These refusals to deal were assertedly so effective that they deterred any direct challenge to Facebook’s platform, thereby maintaining Facebook’s monopoly power. The complaints cite little evidence, however, that any of the affected apps would have developed into a competing social network. Moreover, Facebook asserts that it no longer engages in the practice: it dropped its policy in 2018 and has not cut off any apps since 2013. As a result, the district court ruled that neither the FTC nor the states were entitled to an injunction. The court also emphasized that neither plaintiff had alleged that Facebook’s conduct met the criteria for illegality articulated in Trinko.43 This is not surprising. When critics charge the tech giants with refusing to deal with competitors, they frequently fail to address whether the refusals ended a period of prior voluntary cooperation or entailed a profit sacrifice.44 41 Moreover, the use of seller-specific non-public data to copy a rival’s product would not be per se illegal. A plaintiff would have to show that the conduct was likely to reduce competition significantly – because, for example, its adverse effect on third-party innovation outweighed its beneficial effect on retail prices. This requirement would further protect product copying that is likely to enhance consumer welfare. 42 See Hubbard (n 40). 43 Verizon Communications Inc v Law Offices of Curtis V Trinko, LLP 540 US 398 (2004). 44 See ibid.

Predation and discrimination  151 4. Conclusion Overall, there is little doubt that the tech giants have engaged in platform discrimination. This ‘self-preferencing’ has not only injured third parties selling on their platforms, it has enhanced their market power and reduced consumer choice. There is limited evidence, however, that this unwarranted exclusion has resulted in monopoly power or a dangerous probability of monopoly power. While various government plaintiffs allege that Google’s search bias and Facebook’s refusals to deal have buttressed their monopoly power in core markets, most allegations of platform discrimination focus on complementary markets – the markets in which third parties sell goods and services. And in these markets, there is no evidence, to this author’s knowledge, that tech giant discrimination has created actual or probable monopoly power. As a result, when self-favouring distorts these markets, it is beyond the reach of the Sherman Act. That ought to be changed. B.

EU Developments

The EC moved more quickly than US authorities to challenge platform discrimination. The EC charged Google with search bias and Amazon with unjustified product copying. While the Amazon case is still under investigation, the case against Google resulted in a finding of abuse and a large fine. In both cases, however, there is reason to be concerned that the EC’s actions protected competitors at the expense of consumers. 1. Search bias In 2017, the EC concluded that Google had altered its search results so that its comparison shopping service, Google Shopping, was generally placed ahead of competing services.45 Ruling that this constituted an abuse of dominance, the EC fined Google €2.42 billion.46 The evidence indicated that Google had redesigned its search algorithm to favour its own products. In 2007 Google unveiled Universal Search, a new algorithm that gave ‘particular prominence to Google’s products’.47 Indeed, Universal Search placed Google Shopping ‘at or near the top of search results for comparative shopping services’.48 The issue was whether this priority was justified. The EC found that it was not,49 and concluded that Google’s conduct injured consumers as well as competitors. The EC did not find, however, that Google’s search bias resulted in monopoly power. While Google does not charge consumers for searches or complementary services like Google Shopping, it does charge advertisers to place messages on these products. But the EC did not conclude that Google’s new search design resulted in higher advertising rates. Although the new design reduced the sales of other comparative shopping services,50 the EC did not decide 45 See Eleanor M Fox, ‘Platforms, Power and the Antitrust Challenge: A Modest Proposal to Narrow the US-Europe Divide’ (2019) 98 Neb L Rev 101, 110–12 (summarizing Google Search (Shopping) (Case AT.39740) [2018] OJ C9/11 (Summary)). This discussion is also drawn from Kirkwood (n 34). 46 See Fox (n 45) 112. 47 Richard J Gilbert, ‘The US Federal Trade Commission Investigation of Google Search’ in John E Kwoka, Jr and Lawrence J White (eds), The Antitrust Revolution (7t h edn, OUP 2019) 490. 48 Fox (n 45) 111. 49 See ibid (noting that the EC found that the change had ‘no objective justification’). 50 See ibid (‘Google Shopping increased its share in all thirteen markets in the European Economic Area, in many by a large amount’).

152  Research handbook on abuse of dominance and monopolization that it enabled Google to elevate its ad rates to monopoly levels. Advertisers apparently had other choices.51 The GC affirmed the EC’s decision.52 In the US, the FTC investigated Google’s new search algorithm but decided not to issue a complaint.53 The staff attorneys who conducted the investigation agreed. While they objected to some aspects of Google’s behaviour, they did not recommend a complaint with respect to its search engine.54 Like the EC, moreover, the FTC found that the sites Google downgraded lost significant traffic but did not conclude that Google gained monopoly power.55 Unlike the EC, though, the FTC decided that Google’s new algorithm was justified. Richard Gilbert, an economist who consulted for the agency, noted that Universal Search produced a greater diversity of websites on the first results page and consumer responses indicated that they preferred that.56 In short, in contrast to the EC, the FTC found that Google’s new algorithm did not degrade its search results; it enhanced them.57 2. Product copying The EC recently charged Amazon with abuse of dominance because it had copied third-party products based on information it had gleaned from aggregated data about multiple sellers.58 This approach is troubling. If Amazon cannot use such aggregate data to decide which markets to enter, its ability to make profitable entry decisions will be reduced and consumers may be deprived of the inexpensive and high-quality products that Amazon generally provides. Notably, Commissioner Vestager did not assert that the EC’s action would benefit consumers; her focus was on competitors. She stated that Amazon’s conduct ‘marginalizes third-party sellers and caps their ability to grow’.59 While such an adverse effect on competitors could harm consumers, there is reason to believe that Amazon’s conduct increased consumer welfare.

See Gilbert (n 47) 499 (noting advertisers’ ability to place messages on ‘third-party websites’ and ‘alternative media such as social network sites, radio, and television’). 52 Case T-612/17 Google LLC, formerly Google Inc and Alphabet, Inc v European Commission ECLI:EU:T:2021:763. 53 See Statement of the Federal Trade Commission Regarding Google’s Search Practices, In the Matter of Google, Inc, FTC File No. 111-0163, 3 January 2013. 54 See Charles Duhigg, ‘The Case against Google’ New York Times Magazine (New York, 25 February 2018) 41. 55 See Gilbert (n 47) 502. 56 See ibid 502–3. 57 More than a decade later, the EC again fined Google for exclusionary behaviour, this time concluding that Google had refused to license its Android operating system to mobile phone makers unless they agreed to pre-install Google products like Google Search and Google Chrome. See European Commission, ‘Statement by Commissioner Vestager on Commission Decision to Fine Google €4.34 Billion for Illegal Practices regarding Android Mobile Devices to Strengthen Dominance of Google’s Search Engine’ (18 July 2018), https://​ec​.europa​.eu/​commission/​presscorner/​detail/​en/​STATEMENT​ _18​_4584, accessed 10 November 2022. See also Giorgio Monti and Alexandre Ruiz Feases, ‘The Case against Google: Has the US Department of Justice Become European?’ (2021) 35 Antitrust 26. 58 See Valentina Pop and Sam Schechner, ‘Amazon Violated Laws on Competition, EU Says’ Wall Street Journal (Paris, 11 November 2020) A1, A10 (Margrethe Vestager, the EC’s Executive Vice President, emphasized that the EC’s ‘case doesn’t focus on how Amazon gathers data about individual sellers’). 59 ibid. 51

Predation and discrimination  153

IV.

SECONDARY-LINE PRICE DISCRIMINATION

Although the R-P Act is widely criticized as a protectionist statute and is no longer enforced by the federal government, it continues to generate a modest but steady stream of private actions. No private action in recent years, however, appears to have involved secondary-line price discrimination induced by a powerful buyer like Amazon or Walmart. And no reported decision addressed whether the alleged discrimination was likely to have injured consumers. In the last decade, secondary-line price discrimination has not been a focus of judicial attention in the EU. Most decisions under Article 102(c) involved other forms of discrimination. Only two cases used the term secondary-line price discrimination and very few addressed it. A.

US Developments

Despite its disfavour in the US antitrust community, the R-P Act continues to be enforced by private plaintiffs. Over the last decade, approximately 30 decisions addressed R-P liability for secondary-line price discrimination, usually in response to motions to dismiss or motions for summary judgment filed by a defendant. Plaintiffs often prevailed in these decisions, defeating about half the motions to dismiss and more than a third of the summary judgment motions.60 None of the cases, however, resulted in a final judgment for the plaintiff. It appears, therefore, that secondary-line R-P enforcement continues to represent some litigation threat to defendants, though without settlement data it is difficult to say how large the threat is. No decision discussed the impact of the alleged discrimination on consumer welfare and no case involved an unmistakably powerful buyer like Amazon. Some cases, however, involved a favoured buyer that was commercially important to the allegedly discriminating seller.61 Rather than addressing consumer welfare, the courts asked whether the case presented ‘competitive injury’ – injury to one or more disfavoured buyers. In deciding whether it did, the courts continued to use the ‘Morton Salt inference’, which allows competitive injury to be inferred from a substantial and sustained price differential.62 In the absence of other evidence, this inference was enough to allow a case to go forward. But where a defendant attempted to rebut the Morton Salt inference by claiming that the plaintiff had not actually lost significant sales to a favoured buyer, the courts insisted on direct evidence of competitive injury. Specifically, the courts required the plaintiff to allege or supply evidence that it had lost more than a de minimis volume of sales.63 Some courts even required the plaintiff to identify the specific favoured buyer(s) and the specific lost sales, with a single lost sale being insufficient to demonstrate competitive injury.64 At the same time, no decision dismissed a case or granted summary judgment for the defendant on the ground that the defendant had established a cost justification or meeting

Research conducted by Nathan McCurtain, a law student at Seattle University. See, for example, Satnam Distributors LLC v Commonwealth-Altadis, Inc 140 F Supp 3d 405 (2015). In several cases the plaintiff claimed that the favoured buyer secured a discriminatory concession through bribery. For example, Sec Data Supply, LLC v Nortek Sec & Control LLC 2019 WL 3305628 (2019). 62 See FTC v Morton Salt Co 334 US 37, 50 (1948). 63 See, for example, Cash & Henderson Drugs, Inc v Johnson & Johnson 799 F 3d 202 (2015). 64 See, for example, Bendfeldt v Window World, Inc 2017 WL 4274191 (2017). 60 61

154  Research handbook on abuse of dominance and monopolization competition defence. Although these defences can halt R-P actions that are likely to harm consumers, the courts were unwilling to use the defences to shut down secondary-line cases before trial. Instead, the courts always ruled that the defendant had not presented enough evidence of these defences to avoid a full hearing.65 The courts’ decisions in buyer-liability cases were also mixed. The courts continued to insist that a plaintiff must prove not only that the defendant buyer had induced an illegal discrimination but that the buyer knew that the seller was aware that its discrimination was illegal.66 In some cases, the plaintiff surmounted that double hurdle with credible allegations that the buyer knew that the seller had not granted any competing buyer a comparable price preference.67 In other cases, the plaintiff could not show that the seller knew it had granted a discriminatory price.68 In no case was the cost justification or meeting competition defence decisive. While the plaintiff would have to establish at trial that the buyer knew that the seller had no defence, the courts were unwilling to resolve this issue at the pretrial stage. B.

EU Developments

Article 102(c) TFEU prohibits ‘applying dissimilar conditions to equivalent transactions with other trading parties, thereby placing them at a competitive disadvantage’. A plaintiff bringing an action under this section must show that the defendant is in a dominant position, that its conduct was capable of distorting competition, and that the distortion would place the plaintiff at a competitive disadvantage.69 In the last ten years, there were ten Article 102(c) cases. Eight were appeals from EC decisions and all ten were resolved in favour of the plaintiff. Although Article 102(c) covers secondary-line price discrimination, the European courts rarely used the term and seldom addressed cases of secondary-line price discrimination. One example was MEO.70 The Portuguese Competition Authority had brought a case following a complaint from MEO concerning GDA’s (the Cooperative for the Management of the Rights of Performing Artists, Portugal) use of discriminatory tariffs in copyright contracts with local Portuguese television broadcasters. The Authority alleged that as the sole manager of artist copyrights in Portugal, the GDA is dominant and that its tariffs discriminated between trade partners in a competitive relationship, which distorts competition in the internal market and is capable of hurting the competitiveness of MEO. On a request for a preliminary ruling, the CoJ decided that Article 102(c) does not require actual competitive harm. Instead, the Court stated: The concept of ‘competitive disadvantage’, for the purposes of subparagraph (c) of the second paragraph of Article 102 TFEU, must be interpreted to the effect that, where a dominant undertaking

65 See, for example, W Convenience Stores, Inc v Suncor Energy (USA) Inc 970 F Supp 2d 1162 (2013). 66 See, for example, Marjam Supply Co v Firestone Bldg Products Co LLC 2014 WL 5798383 (2014). 67 ibid. 68 See, for example, Gorlick Distrib Centers, LLC v Car Sound Exhaust System, Inc 723 F 3d 1019 (2013). 69 Case C-525/16 MEO-Servicos de Comunicacoes e Multimedia SA v Autoridade da Concorrencia ECLI:EU:C:2018:270, paras 13, 38. 70 MEO (n 69).

Predation and discrimination  155 applies discriminatory prices to trade partners on the downstream market, it covers a situation in which that behaviour is capable of distorting competition between those trade partners. A finding of such a ‘competitive disadvantage’ does not require proof of actual quantifiable deterioration in the competitive situation, but must be based on an analysis of all the relevant circumstances of the case leading to the conclusion that that behaviour has an effect on the costs, profits or any other relevant interest of one or more of those partners, so that that conduct is such as to affect that situation.71

Other decisions, such as Post Danmark II,72 involved primary-line discrimination. The Danish competition agency has brought an action following Konkurrenceradet’s complaint that Post Danmark had abused its dominant position in the Danish market by distributing unaddressed mail with discriminatory rates and rebates. On a request for a preliminary ruling, Post Danmark contended that its pricing practices violated EU law only if they could drive an equally efficient competitor from the market. The CoJ declined to apply this test and instead adhered to its more traditional case law, which holds that a dominant undertaking has a special responsibility not to allow its behaviour to impair genuine, undistorted competition on the internal market. Further: In order to determine whether a rebate scheme, such as that at issue in the main proceedings, implemented by a dominant undertaking is capable of having an exclusionary effect on the market contrary to Article [102], it is necessary to examine all the circumstances of the case, in particular, the criteria and rules governing the grant of the rebates, the extent of the dominant position of the undertaking concerned and the particular conditions of competition prevailing on the relevant market. The fact that the rebate scheme covers the majority of customers on the market may constitute a useful indication as to the extent of that practice and its impact on the market, which may bear out the likelihood of an anticompetitive exclusionary effect.73

The Court also stated that a dominant undertaking’s rebate scheme must have a probable anticompetitive effect. There is no need to show that the effect is of a serious or appreciable nature or that it exceeds a de minimis requirement.

V. REFORMS US law does not deal adequately with predatory pricing or platform discrimination. It discourages legitimate predatory pricing claims by requiring plaintiffs to show recoupment, and it prevents challenges to platform discrimination by requiring plaintiffs to prove actual or probable monopoly power. Both requirements ought to be dropped. Instead, a plaintiff in a predatory pricing case ought to be able to prevail if it shows that the defendant priced below cost and that its asserted justifications were invalid. In a platform discrimination case, a plaintiff ought to be able to prevail if it shows that the defendant’s discrimination was likely to reduce competition significantly, whether or not it created monopoly power or a dangerous risk of monopoly power. The R-P Act should also be reformed. A secondary-line plaintiff should have to show that the challenged discrimination was likely to harm consumers or small suppliers, not just dis-

ibid para 37. Case C-23/14 Post Danmark A/S v Konkurrenceradet ECLI:EU:C:2015:651. 73 ibid para 50.

71 72

156  Research handbook on abuse of dominance and monopolization favoured competitors. Further, if the plaintiff makes that showing, the cost justification and meeting competition defences should not stand in the way of relief. A.

Predatory Pricing

There are good reasons to abolish the recoupment requirement, as Leslie has recommended,74 and replace it with a better targeted and less onerous approach. In the past, the recoupment requirement has blocked many predatory pricing claims. As of the date of Leslie’s article, it was the single largest reason why these claims were dismissed. As scholars have shown, however, courts can infer recoupment from both the existence of below-cost pricing and the absence of a non-predatory explanation for the behaviour. Accordingly, instead of imposing a recoupment requirement, the law ought to allow a plaintiff to prevail if it proves that the defendant priced below cost and demonstrates that the defendant’s asserted justifications for that pricing were invalid.75 As Hemphill and Weiser note, the principal procompetitive reasons for a price below cost are ‘when a firm introduces a product with temporary low prices, or sets a low price in anticipation of later scale economies’.76 If a defendant advances one or both of these reasons, the burden would shift to the plaintiff to undercut the explanation. The plaintiff might show, for example, that an introductory pricing explanation was inapt because there had been no significant improvement in the defendant’s product or the defendant had priced below cost far longer than necessary to introduce a new product to consumers. Likewise, the plaintiff might rebut a scale economy explanation by demonstrating that there was no reason to expect that a sharp increase in the defendant’s market share would result in a substantial reduction in its costs. Whatever the purported justification, the plaintiff could establish liability by proving that the defendant priced below cost and that its justifications were inadequate. Even without these reforms, a plaintiff can take several steps to strengthen its case. For example: 1. If the defendant is a dominant firm, the plaintiff could assert that the Supreme Court’s scepticism about recoupment in an oligopoly would not apply. 2. The plaintiff could use any of the modern economic theories of rational predation since the Court did not critique any of them. 3. The plaintiff is free to pick the cost measure that works best for its case, since the Court was flexible on this issue.77 All these suggestions, however, assume that Brooke Group remains the law; they do not remove the need to adopt the reforms just described. The EU never imposed a recoupment requirement. Its leading decisions require below-cost pricing and, in certain cases, an intent to eliminate a competitor. Since the eliminatory intent element may sometimes be protectionist, the CoJ should replace it with a requirement that the Leslie (n 5). See Carl Shapiro, ‘Antitrust: What Went Wrong and How to Fix It’ (2021) 33 Antitrust 40 (‘My point is that the burden [should] rest on the dominant firm to explain its below-cost pricing not on the plaintiff to prove that entry barriers are sufficiently high that recoupment will be possible in the future’). 76 Hemphill and Weiser (n 19) 2055 n 28. 77 Ibid 2049. 74 75

Predation and discrimination  157 EC demonstrate that the alleged below-cost pricing was unjustified. In other words, in order to establish an abuse of dominance, the EC should have to prove that a dominant firm priced below cost and that its asserted justifications for the conduct were invalid. The presence or absence of an eliminatory intent should not be an issue. B.

Platform Discrimination78

The tech giants have excluded third parties selling on their platforms by demoting them in search results, using non-public seller-specific data to boost their own products, or refusing to deal with them simply because they are competitors. While this behaviour is not widespread, it appears to be unjustified and anticompetitive. It enhances the tech giants’ market power and injures their customers. Yet no one in the US has successfully challenged any of this conduct. The most likely reason is that the conduct did not violate the Sherman Act, since it is unilateral, not collusive, and did not result in actual or imminent monopoly power. This gap should be closed. The Sherman Act should be amended to reach unilateral exclusion by the tech giants that reduces competition significantly, even if it is unlikely to generate or maintain monopoly power. Further, the Department of Justice and the FTC should be authorized to obtain civil penalties if they establish a violation of this new section. This would couple public civil penalty enforcement with private treble damage actions, magnifying the deterrent effect of antitrust action.79 Congress can enhance the reach of the Sherman Act without unduly chilling procompetitive behaviour by restricting the scope of the amendment and imposing proof requirements that would make it almost impossible to attack procompetitive behaviour. Congress can restrict the scope of the amendment by stipulating that a defendant must operate a two-sided platform and earn annual revenues exceeding $70 billion, a threshold that would cover the tech giants but few other firms.80 Congress can further limit false positives by insisting that a plaintiff satisfy six proof requirements. 1. Market power The first requirement is significant market power. While the new statute would allow challenges to anticompetitive conduct that is unlikely to create monopoly power, it would not impose liability on conduct that is unlikely to result in significant market power. Conduct cannot impose significant harm on competition or consumers unless it creates or preserves significant market power.81

This section is taken from Kirkwood (n 34). See Harry First, ‘The Case for Antitrust Civil Penalties’ (2009) 76 Antitrust LJ 127, 162–3 (maintaining that the FTC and Justice Department should be empowered to impose civil penalties in monopolization cases). 80 Facebook, the smallest of the tech giants, had total revenues of $70 billion in 2019. US House of Representatives, Subcommittee on Antitrust, Commercial and Administrative Law of the Committee on the Judiciary, Majority Staff Report and Recommendations, Investigation of Competition in Digital Markets (2020) 133. The other tech giants were much larger; see ibid 175 (Google $160.7 billion); ibid 248 (Amazon $280 billion); ibid 331 (Apple $260 billion). 81 See John B Kirkwood, ‘Market Power and Antitrust Enforcement’ (2018) 98 Boston UL Rev 1169, 1173 (stating that market power is ‘central to antitrust because it distinguishes firms that can harm competition and consumers from those that cannot’). For the definition of ‘significant’, see n 83. 78 79

158  Research handbook on abuse of dominance and monopolization As this author has suggested, the best way to demonstrate significant market power is to prove that the challenged conduct has caused, or is likely to cause, significant anticompetitive effects.82 Inferring market power from actual or likely anticompetitive effects should be the primary way of establishing market power under the new statute.83 But it would not be enough. The plaintiff would also have to demonstrate significant market power through the traditional means – defining a relevant market and showing that the defendant has a substantial share of that market. Since Jefferson Parish,84 the minimum market share for establishing significant market power has generally been taken to be 30 per cent.85 The plaintiff would have to show that the defendant’s conduct enabled it to attain or maintain such a share. 2. Barriers to entry In addition, the plaintiff would have to establish significant barriers to entry. It is elementary that market power cannot last for any significant period of time without entry barriers.86 If the relevant market is not protected by significant obstacles to entry, the challenged conduct is unlikely to cause lasting anticompetitive effects and is more likely to be procompetitive. 3. Anticompetitive effects Proving anticompetitive effects is the central element of the plaintiff’s burden. It would help to establish market power but, more importantly, it is a necessary step in demonstrating that the challenged conduct is more likely to reduce competition than increase it. If the plaintiff cannot establish that the defendant’s exclusionary behaviour is likely to have significant anticompetitive effects – that it is likely not only to weaken or extinguish rivals but also to significantly raise price, reduce quality, restrict choice, or suppress innovation – the plaintiff’s case should be dismissed.87 Absent significant anticompetitive effects, the challenged conduct cannot possibly impose significant harm on consumers or workers, and an attack on such conduct may chill procompetitive activity. 4. Overall harm The plaintiff would have to show that the challenged conduct is likely to harm competition overall. This would follow directly from the plaintiff’s proof of significant anticompetitive effects when the defendant fails to establish a justification or the plaintiff shows that the justification could be achieved through a less restrictive alternative. But if the defendant proves

ibid. The plaintiff could also use direct economic evidence to show that the defendant’s conduct allowed it to charge a price significantly above cost. One reasonable measure of significance, based on the federal government’s traditional approach to merger analysis, is 5 per cent. If the challenged conduct likely enabled the defendant to maintain a price 5 per cent above the competitive level, the defendant has significant market power; see ibid 1218–20. 84 Jefferson Parish Hosp Dist No 2 v Hyde 466 US 2 (1984). 85 ibid 26–7. 86 See John B Kirkwood and Richard O Zerbe, Jr, ‘The Path to Profitability: Reinvigorating the Neglected Phase of Merger Analysis’ (2009) 17 George Mason L Rev 39, 40. A ‘significant’ entry barrier would permit a defendant to price 5 per cent above the competitive level for at least two years. 87 In a buy-side case, the plaintiff would have to show significant monopsony power, barriers to entry, and adverse effects on powerless suppliers such as workers. 82 83

Predation and discrimination  159 that the practice is likely to have procompetitive effects, the plaintiff would have to establish that its anticompetitive effects significantly exceed its procompetitive effects. Likelihood 5. The plaintiff would have to establish that the first four requirements were ‘likely’ to be met. It would not be enough to show, for example, a ‘reasonable possibility’ or ‘dangerous probability’ of significant market power. The plaintiff would have to prove that significant market power was likely. 6. Precise theory Finally, the plaintiff would have to describe its legal theory in terms that are clear enough that other business firms can understand it and precise enough that other firms can avoid the problem without eliminating significant procompetitive conduct. Before a court issues an injunction, awards treble damages, or imposes civil penalties, the plaintiff’s theory must pass these tests of clarity and precision. *** Together, these six requirements would erect a major barrier to an attack on procompetitive conduct. They would provide defendants with six substantive grounds on which to contest a plaintiff’s case, grounds that could be employed throughout the litigation (eg in motions to dismiss or motions for summary judgment) to defeat an assault on desirable behaviour. Together, they would sharply reduce the new statute’s adverse impact on procompetitive conduct. In sum, the US can effectively address platform discrimination by amending the Sherman Act in the ways just described. These changes would enable public and private plaintiffs to enjoin discrimination that is likely to reduce competition significantly, whether or not it results in actual or probable monopolization. Civil penalties and treble damages would augment the deterrent effect of these actions, reducing the future incidence of this behaviour.88 Given this solution, it is neither necessary nor desirable to break up the tech giants. Splitting them into smaller versions of themselves (horizontal restructuring) would likely result in higher prices

88 It may be desirable to supplement this approach with rulemaking. See Eleanor M Fox and Harry First, ‘We Need Rules to Rein in Big Tech’ (CPI Antitrust Chronicle, October 2020), https://​ssrn​.com/​ abstract​=​3724595, accessed 10 November 2022. Fox and First recommend that the FTC issue a rule prohibiting the tech giants from discriminating against third parties selling on their platforms. They recognize that such a major competition rule would be a new undertaking for the FTC. In its 108-year history, it has issued just one competition rule, a minor regulation intended to facilitate compliance with the Robinson-Patman Act in the Men’s and Boys’ Tailored Clothing Industry; see ibid 4 n 10. The FTC’s reluctance to rely more heavily on rulemaking is not surprising given the difficulties of formulating a good competition rule – one that is clear, easily administered, and neither too broad nor too narrow. For example, a rule that required platforms to offer ‘equal terms for equal service’ would be clear in concept but too narrow. A platform could avoid it by asserting that third parties often require more services than the platform’s own products (eg, assistance in managing their inventories or paying their bills). To address this problem, the FTC could expand the rule to specify the services a platform could provide and the charges it could impose for each. But that would involve the FTC in detailed regulation of the interactions between platforms and third parties. As a result, while a rule prohibiting platform discrimination may be a good idea, it should not be the sole means of policing tech giant discrimination.

160  Research handbook on abuse of dominance and monopolization or lower quality. Preventing them from selling their own products on their platforms (vertical restructuring) would deprive consumers of products they value.89 The EU does not require new legislation to implement this approach. Unlike Section 2 of the Sherman Act, Article 102 TFEU does not demand a showing of monopoly power or a dangerous probability of monopoly power. Its concept of dominance is much broader: a firm may be dominant even though its market share is significantly lower than the threshold typically employed by US law to indicate monopoly power.90 Moreover, the EC can already impose substantial fines on a firm that commits an abuse of dominance. As a result, the EC and the European courts are free to adopt the recommended approach to exclusionary conduct. While the proof requirements embedded in the approach would limit the EC’s discretion, they would reduce the incidence of false positives and enhance the net benefits for competition and consumers. C.

Secondary-Line Price Discrimination

The R-P Act should be amended in two ways.91 The first would shift the aim of the Act from protecting small competitors to protecting consumers or small suppliers. The simplest way to accomplish that reform is to delete the Act’s competitive injury language, forcing plaintiffs alleging discrimination to show harm to competition, not just harm to individual competitors. The second change would prevent the Act’s two principal defences – meeting competition and cost justification – from foreclosing liability when buyer-induced discrimination is likely to harm competition. One option is to eliminate the defences in purely equitable actions. In other words, if a plaintiff seeks only equitable relief, the defences would be unavailable.92 Alternatively, the defences could be eliminated in treble damage actions as well as equitable actions if the plaintiff proves by clear and convincing evidence that the challenged discrimination is like to harm competition.93 But whatever option is chosen, neither defence should stand in the way of relief when the plaintiff establishes that a powerful buyer has induced a concession likely to injure consumers or small suppliers.94

See Kirkwood (n 34) Part IV. See Rousseva (n 10) 67 (‘Although the factors indicating monopoly power are generally similar, the market share threshold which raises concern [under Section 2] is significantly higher than the one normally used as evidence of dominance’). 91 See John B Kirkwood, ‘Reforming the Robinson-Patman Act to Serve Consumers and Control Powerful Buyers’ (2015) 60 Antitrust Bull 358. 92 To maintain the incentive to bring a legitimate lawsuit, a successful plaintiff could still recover attorney’s fees. 93 The second option provides greater deterrence, while the first affords greater protection to a supplier that cannot realistically be expected to know, in the heat of negotiations, whether a concession is likely to reduce or promote competition. 94 Other changes in the Act should also be made: adjustments to the Act’s technical and jurisdictional requirements and conforming changes to the Act’s treatment of promotional discrimination. 89 90

Predation and discrimination  161

VI. CONCLUSION Predatory pricing and platform discrimination are legitimate targets of competition enforcement on both sides of the Atlantic. When they lack redeeming virtue, these exclusionary practices suppress rivals, enhance market power, and harm consumers. Although they achieve their anticompetitive ends in different ways, their adverse impact is the same. On both sides of the Atlantic, however, the current legal approach to these practices is off target. In the US, it is too permissive; in the EU, it may be too restrictive. In the US, a plaintiff cannot establish predatory pricing without showing not only that the defendant priced below cost but also that it was reasonably likely to recoup its losses. A plaintiff cannot successfully challenge platform discrimination, even if the discrimination is completely arbitrary, unless the plaintiff proves that it led to monopoly power or a dangerous probability of monopoly power. In the EU, neither of these obstacles exists. But the opposite problem may occur: the EC and courts may find predatory pricing or platform discrimination when the practice was justified and benefited consumers. This chapter proposes reforms to US and EU law that would remove these problems. The reforms would eliminate the recoupment requirement from US predatory pricing law and the monopoly power requirement from US unilateral exclusionary conduct law. The reforms would also include new requirements that would make it much more difficult for plaintiffs to challenge procompetitive behaviour in either area – requirements the EU should adopt. Together, the changes would enhance competition policy on two continents. At the same time, the US should amend the R-P Act so that it too promotes competition and consumer welfare.

9. The future of refusals to deal and margin squeezes in the face of sector-specific regulation Inge Graef

I. INTRODUCTION The concepts of refusal to deal and margin squeeze have played an important role in competition enforcement in the 1990s and 2000s. Various cases against these types of behaviour have taken place on both sides of the Atlantic in this period.1 While refusal to deal and margin squeeze cases were traditionally targeting physical infrastructures like bridges and ports as well as old economy industries like telecoms, energy and postal services, attention has now shifted towards questions about whether digital platforms and data can be subject to refusal to deal and margin squeeze claims. Even though this is indeed a frequently discussed topic in scholarship,2 the predominant focus of the applicable legal standard in recent EU competition cases regarding digital platforms lies elsewhere, namely on theories of harm such as discrimination and self-preferencing.3 This raises questions about how these newer theories of harm relate to the existing concepts of refusal to deal and margin squeeze, especially now that the EU Courts have clarified the relationship between these concepts in judgments involving access to telecoms and railway infrastructures.4 The distinction between outright and constructive refusals to deal lies at the heart of the relationship between the concepts of refusals to deal and margin squeezes. An outright refusal 1 See, in particular, the landmark US Trinko and EU Microsoft cases: Verizon Communications v Law Offices of Curtis V Trinko, LLP 540 US 398 (2004) and Case T-201/04 Microsoft Corp v Commission of the European Communities EU:T:2007:289. 2 See, for instance, Giuseppe Colangelo and Mariateresa Maggiolino, ‘Big Data as Misleading Facilities’ (2017) 13 European Competition Journal 249; Marixenia Davilla, ‘Is Big Data a Different Kind of Animal? The Treatment of Big Data under the EU Competition Rules’ (2017) 8 Journal of European Competition Law and Practice 370; Thomas Hoppner, ‘Duty to Treat Downstream Rivals Equally: (Merely) a Natural Remedy to Google’s Monopoly Leveraging Abuse’ (2017) 1 European Competition and Regulatory Law Review 208; Édouard Bruc, ‘Data as an Essential Facility in European Law: How to Define the “Target” Market and Divert the Data Pipeline?’ (2019) 15 European Competition Journal 177; Nikolas Guggenberger, ‘Essential Platforms’ (2021) 24 Stanford Technology Law Review 237. 3 Google Search (Shopping) (Case AT.39740) Commission Decision [2018] OJ C 9/11; European Commission, ‘Antitrust: Commission Sends Statement of Objections to Amazon for the Use of Non-Public Independent Seller Data and Opens Second Investigation into Its E-commerce Business Practices’ (Press release, 10 November 2020), https://​ec​.europa​.eu/​commission/​presscorner/​detail/​en/​ip​ _20​_2077, accessed 10 May 2021; European Commission, ‘Antitrust: Commission Opens Investigation into Possible Anticompetitive Conduct of Facebook’ (Press release, 4 June 2021), https://​ec​.europa​.eu/​ commission/​presscorner/​detail/​en/​ip​_21​_2848, accessed 4 June 2021. 4 Case T‑814/17 Lietuvos geležinkeliai AB v European Commission (Lithuanian Railways) ECLI:EU:T:2020:545; Case C-165/19P Slovak Telekom, as v European Commission ECLI:EU:C:2021:239.

162

The future of refusals to deal and margin squeezes  163 to deal occurs when a dominant firm does not grant access to the requested input at all. In case of a constructive refusal to deal, the dominant firm provides access to the requested input, but does so under conditions that, in fact, amount to a refusal to grant access. The extent of access offered by dominant firms in constructive refusal to deal cases does not allow for effective competition. Margin squeezes are the main example of constructive refusals to deal, where a dominant firm’s conduct in practice leads to a refusal to deal and excludes competitors from the market because it leaves an insufficient margin between the prices charged in the upstream and downstream markets. Despite the fact that the two types of behaviour have an equivalent effect, the legal standards to hold margin squeezes and refusals to deal abusive under Article 102 of the Treaty on the Functioning of the European Union (TFEU) are different – unlike the situation in US antitrust law where margin squeezes do not form a cause of antitrust harm independent of refusals to deal.5 In accordance with the case law, an outright refusal to deal is abusive under Article 102 TFEU only in exceptional circumstances, namely if it: (1) relates to an indispensable input; (2) excludes effective competition on a downstream market; (3) prevents the emergence of a new product (only if the input is protected by intellectual property law or trade secrets); and (4) cannot be objectively justified.6 However, these conditions, often referred to as the Bronner criteria – after the case from which they originate – do not apply to margin squeezes and other constructive refusals to deal.7 This means that margin squeezes can amount to an abuse of dominance even if the input is not indispensable and the dominant firm is not subject to a duty to deal. EU competition law and US antitrust law diverge on this issue, as Section 2 of the Sherman Act requires an antitrust duty to deal as a condition for establishing liability for a margin squeeze.8 Discussions about the relationship between refusals to deal and margin squeezes are becoming more prominent as newer theories of harm relating to self-preferencing have been created in EU competition law, which could be interpreted as constituting constructive refusals to deal. Against this background, this chapter explores what legal standards should apply to variations of refusals to deal in future EU competition enforcement. Insights are drawn from the literature, decisional practice, and case law to establish what legal conditions are capable of addressing the relevant competition concerns. References to US antitrust law are included by way of comparison. The chapter makes two findings. A first finding is that the distinction made in the case law between outright and constructive refusals to deal is not a workable and reliable way to determine whether the Bronner criteria should be applied, as the distinction often may be a matter of degree rather than black-or-white. Instead, the chapter argues that the existence of a regulatory duty under sector-specific regulation is a better indicator for deciding whether the Bronner criteria constitute the legal test for the assessment under Article 102 TFEU. With the adoption of the Digital Markets Act (DMA), this indicator will also become of relevance for assessing the behaviour of gatekeeping digital 5 See, for instance, Case C‑52/09 Konkurrensverket v TeliaSonera Sverige AB ECLI:EU:C:2011:83 and Pacific Bell Telephone Co v linkLine Communications, Inc (linkLine) 555 US 438 (2009). 6 Case C-7/97 Oscar Bronner GmbH & Co KG v Mediaprint Zeitungs- und Zeitschriftenverlag GmbH & Co KG, Mediaprint Zeitungsvertriebsgesellschaft mbH & Co KG and Mediaprint Anzeigengesellschaft mbH & Co KG ECLI:EU:C:1998:569; Case C-418/01 IMS Health GmbH & Co OHG v NDC Health GmbH & Co KG ECLI:EU:C:2004:257; Microsoft (n 1). 7 As made clear by the Court of Justice in TeliaSonera (n 5). 8 linkLine (n 5).

164  Research handbook on abuse of dominance and monopolization platforms. The DMA complements the existing EU competition framework by regulatory duties mandating access to data and essential platform services such as app stores.9 Another question is whether the right to data portability of the General Data Protection Regulation (GDPR)10 can constitute such a sector-specific regulatory duty when a competition case relates to restrictions of the portability of personal data. Sector-specific data access regimes also exist in the payment and energy sectors.11 Under the approach proposed in this chapter, their existence will have the effect that there is no need to apply the Bronner criteria to assess the abusive nature of refusals to give access to data under Article 102 TFEU. The complementary relationship between sector-specific regulation and competition enforcement is a feature of the EU system and differs from the situation in the US, where sector-specific regulation is considered to be a substitute for rather than complement to US antitrust enforcement. A second finding is that a nuanced approach is necessary to assess situations where the Bronner criteria are applicable. As the most debated condition, the indispensability requirement should not be construed as carrying such a high threshold that it can rarely be met in practice, nor should it be interpreted in a way that facilitates access to inputs that are not indispensable but merely ‘convenient’ for rivals to compete.12 The Bronner criteria have already been interpreted differently throughout the years depending on the factual circumstances at stake. This illustrates that the standards used for holding an input indispensable for competition can be tailored to the relevant market situation. To revitalize antitrust liability for refusals to deal in the US, a similar approach can be taken. Before elaborating on these two findings, Section II compares the case law in the EU and the US regarding the relationship between refusals to deal and margin squeezes. Section III explores the position of self-preferencing theories of harm based on an analysis of the Google Shopping case. Section IV discusses the relevance of the existence of a regulatory duty to determine whether the Bronner criteria apply as the legal test under Article 102 TFEU. Section V explores how the indispensability requirement can be interpreted in line with the market situation at stake in refusal to deal cases. Section VI concludes.

Regulation (EU) 2022/2065 of the European Parliament and of the Council of 14 September 2022 on contestable and fair markets in the digital sector (Digital Markets Act) [2022] OJ L 265/1, art 6(9), (10), (11) and (12) regarding, respectively, data portability for business users and end-users, continuous and real-time access for business users to data generated in the context of the use of platform services, access for third-party search engine providers to ranking, query, click and view data, and fair, reasonable and non-discriminatory (FRAND) conditions for business users to access app stores. 10 Regulation (EU) 2016/679 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 (GDPR) [2016] OJ L119/1, art 20. 11 Directive (EU) 2015/2366 of 25 November 2015 on payment services in the internal market (PSD2) [2015] OJ L337/35, arts 66 and 67; Directive (EU) 2019/944 of the European Parliament and of the Council of 5 June 2019 on common rules for the internal market for electricity (Electricity Directive) [2019] OJ L158/125, art 23(1). 12 The notion of ‘convenient facilities’ was used to criticize the lowering of the standards for holding refusals to deal abusive in the Microsoft case. See Derek Ridyard, ‘Compulsory Access under EC Competition Law: A New Doctrine of “Convenient Facilities” and the Case for Price Regulation’ (2004) 25 European Competition Law Review 669. 9

The future of refusals to deal and margin squeezes  165

II.

THE RELATIONSHIP BETWEEN REFUSALS TO DEAL AND MARGIN SQUEEZES: AN EU–US COMPARISON

A.

The EU Perspective

A dominant firm can engage in a margin squeeze either by setting a very high price for competitors to access the input in the upstream market or by setting a very low price for the service provided to consumers in the downstream market.13 In Deutsche Telekom, the Court of Justice (CoJ) made clear that a margin squeeze can constitute an abuse of dominance in itself, irrespective of whether the wholesale or retail prices set by the dominant firm are abusive because of their excessive or predatory character.14 In other words, to hold a margin squeeze abusive under Article 102 TFEU it is not necessary for the dominant firm’s wholesale price as applied to competitors to be excessive or for its retail price targeted at consumers to be predatory. The spread between the wholesale and retail prices can be abusive in itself because of its exclusionary effect, absent any concerns regarding excessive and predatory pricing. Regarding the legal test, the CoJ was confronted in TeliaSonera with the question whether a margin squeeze can be abusive in the absence of the existence of a duty to supply under the Bronner criteria. TeliaSonera had argued before the Court that a dominant firm should be free to set its terms of trade, unless they are so disadvantageous that they qualify as a refusal to supply under the conditions laid down in Bronner, requiring in particular that the input to which access is sought is indispensable.15 The Court responded by explaining that TeliaSonera’s interpretation was based on a misunderstanding of the Bronner judgment. According to the Court, it cannot be inferred from Bronner that its conditions for establishing whether a refusal to deal is abusive must also be used to assess the abusive character of conduct consisting in the supply of goods or services on conditions that are disadvantageous.16 This implies that an outright refusal to deal is subject to a stricter legal test than a constructive refusal to deal in the form of a margin squeeze. The Court explicitly stated that the latter conduct can in itself form ‘an independent form of abuse distinct from that of refusal to supply’.17 Another interpretation, as put forward by TeliaSonera, would in the Court’s view ‘unduly reduce the effectiveness of Article 102 TFEU’ as every conduct relating to the terms of trade set by a dominant firm would otherwise have to meet the Bronner criteria in order to be regarded as abusive.18 The approach of the CoJ in TeliaSonera to treat outright and constructive refusals to deal differently can be criticized on the grounds that this provides incentives for dominant firms not to give access to non-indispensable inputs at all.19 Once a dominant firm does grant such access, it namely runs the risk of being held liable for imposing abusive access conditions even Richard Whish and David Bailey, Competition Law (Oxford University Press 2018) 771–2. Case C‑280/08P Deutsche Telekom AG v European Commission ECLI:EU:C:2010:603, para 183. 15 TeliaSonera (n 5) para 54. 16 ibid para 55. 17 ibid para 56. 18 ibid para 58. For a critical appraisal of this argument see, Niamh Dunne, ‘Dispensing with Indispensability’ (2020) 16 Journal of Competition Law and Economics 74, 96–7. 19 See also the reasoning by the Advocate General in TeliaSonera who argues that margin squeeze cases are analogous to refusal to deal cases so that the same underlying rationale should apply. Case C-52/09 Konkurrensverket v TeliaSonera Sverige AB ECLI:EU:C:2010:483, Opinion of Advocate General Mazák, para 23. 13

14

166  Research handbook on abuse of dominance and monopolization if the input is not indispensable and thus no duty to deal would otherwise exist under Article 102 TFEU. On the one hand, such disincentives for access resulting from the CoJ’s reasoning seem to go against the purpose of the competition rules to stimulate openness of markets and follow-on innovation. On the other hand, the CoJ’s reasoning can be regarded as a form of ‘estoppel abuse’, where a voluntary decision of a dominant firm to supply implies that it must do so on terms at which competitors can effectively compete.20 B.

Differences between the EU and US

The approach of the CoJ in Deutsche Telekom and TeliaSonera differs from the US approach on two accounts, because the latter requires an antitrust duty to deal as a condition for establishing liability for a margin squeeze and rules out the application of US antitrust law if a sector-specific regime already regulates the same conduct.21 In Trinko, the US Supreme Court took a restrictive approach towards refusals to deal. In particular, it held that the circumstances giving rise to antitrust liability in its 1985 Aspen Skiing judgment – the leading US case establishing antitrust liability for a refusal to deal – are ‘at or near the outer boundary’ of the scope of Section 2 of the Sherman Act.22 According to the Supreme Court in Trinko, Aspen Skiing constitutes a limited exception to the principle of freedom to contract, which only applies in situations where a monopolist terminates a voluntary and profitable prior course of dealing.23 The scenario at stake in Aspen Skiing concerned a monopolist unwilling to renew a joint ski ticket that it had voluntarily started to offer with a competitor a number of years earlier, to the extent that the monopolist was forgoing short-term profits by refusing to continue providing the joint ski ticket.24 The facts in the Trinko case differed, as it was the 1996 Telecommunications Act that obliged Verizon as the local incumbent to share its telephone network with rivals. Because Verizon did not voluntarily engage in a course of dealing with its rivals, the Supreme Court found that Verizon’s prior conduct could shed no light upon whether the refusal to deal was ‘prompted not by competitive zeal but by anticompetitive malice’.25 The Supreme Court confirmed this restrictive approach towards antitrust liability for refusals to deal in linkLine, which involved a margin squeeze resulting from the combination of high wholesale prices for internet service providers in the upstream market with low retail prices for consumers in the downstream market set by incumbent telephone companies. For a margin squeeze to violate Section 2 of the Sherman Act, the Supreme Court held that a monopolist must either refuse to deal under circumstances creating an antitrust duty to deal or engage in predatory pricing.26 In the view of the Supreme Court, it was clear from Trinko that ‘if a firm

20 Kevin Coates, ‘The Estoppel Abuse’ (21st Century Competition: Reflections on Modern Antitrust, 28 October 2013), http://​www​.twentyfir​stcenturyc​ompetition​.com/​2013/​10/​the​-estoppel​-abuse/​, acces-​ sed 10 May 2021. 21 For a comparison between these aspects of EU competition and US antitrust law, see Damien Geradin, ‘Limiting the Scope of Article 82 EC: What Can the EU Learn from the U.S. Supreme Court’s Judgment in Trinko in the Wake of Microsoft, IMS, and Deutsche Telekom?’ (2004) 41 Common Market Law Review 1519. 22 Trinko (n 1) para 409. 23 ibid para 409. 24 Aspen Skiing Co v Aspen Highlands Skiing Corp 472 US 585 (1985). 25 Trinko (n 1) para 409. 26 linkLine (n 5) para 448.

The future of refusals to deal and margin squeezes  167 has no antitrust duty to deal with its competitors at wholesale, it certainly has no duty to deal under terms and conditions that the rivals find commercially advantageous’.27 In other words, no violation of Section 2 of the Sherman Act can be established for a margin squeeze in the absence of an antitrust duty to deal. Unlike the situation in the EU where the CoJ held that margin squeezes constitute an independent form of abuse for which the Bronner criteria do not apply, US antitrust law uses the same legal standards to assess liability for refusals to deal and margin squeezes. As such, a duty to deal can only be imposed under current US antitrust law if: (1) the monopolist entered into a pre-existing voluntary course of dealing; and (2) the monopolist is willing to sacrifice short-term profits in order to achieve an anticompetitive end.28 Based on the facts at stake in the case, the Supreme Court concluded that the margin squeeze in linkLine did not violate Section 2 of the Sherman Act because the internet service providers had neither successfully identified an antitrust duty to deal nor established a predatory pricing claim.29 According to the Supreme Court, no antitrust duty to deal existed in linkLine for a similar reason as in Trinko, namely because the duty of the incumbent telephone companies to provide wholesale access to the internet service providers stemmed from regulations of the Federal Communications Commission and not from the Sherman Act.30 The US has thus taken the approach of ruling out any room for antitrust remedies once a sector-specific regime applies, while the CoJ made clear in Deutsche Telekom that EU competition law complements sector-specific regulation.31 Deutsche Telekom had attempted to justify its margin squeeze on the ground that it was subject to regulation imposed by the German telecoms authority obliging it to provide competitors with local access to its network. The wholesale prices Deutsche Telekom could charge to rivals were fixed by the authority and the retail prices for consumers were subject to a price cap. Nevertheless, the CoJ concluded that the margin squeeze could be attributed to Deutsche Telekom because the applicable regulation offered enough scope for Deutsche Telekom to limit or stop the margin squeeze by increasing its retail prices to end-users.32 The fact that a sector is regulated, therefore, does not imply that it is immune from parallel competition intervention in the EU. According to the Court in Deutsche Telekom, in situations where sector-specific regulation leaves room for autonomous conduct by market players, the restriction of competition can still be attributed to them and Articles 101 and 102 TFEU apply in parallel to the regulatory framework.33 In this regard, the EU approach diverges from the situation in the US where the existence of sector-specific regulation is considered to preclude parallel application of the US antitrust rules. As argued by the US Supreme Court in Trinko, the reasoning behind this choice is that additional antitrust intervention is considered to be costly while the additional benefit to competition provided by antitrust enforcement tends to be small ‘[w]here there exists a regulatory structure designed to deter and remedy anticompetitive harm’.34 This divergence between the EU and US will likely not disappear in the future, ibid para 450. See, for example, Novell, Inc v Microsoft Corp 731 F3d 1064 (10th Cir 2013), paras 1074–5, cert denied, 134 S Ct 1947 (2014). 29 linkLine (n 5) para 457. 30 ibid para 450. 31 Deutsche Telekom (n 14) para 80. 32 ibid para 183. 33 ibid para 80. 34 Trinko (n 1) para 412. 27 28

168  Research handbook on abuse of dominance and monopolization as the EU legislator is and has been complementing the existing competition framework with additional sector-specific regulation, such as in the form of the DMA, the Data Act, the Payment Services Directive 2 (PSD2) and the Electricity Directive – as further discussed in Section IV below.

III.

THE POSITION OF SELF-PREFERENCING THEORIES OF HARM

A.

The Legal Reasoning in the Google Shopping Case

The applicability of the Bronner criteria, and in particular the indispensability requirement, plays a key role in current debates about the legal test for holding self-preferencing abusive under Article 102 TFEU.35 Self-preferencing can be regarded as a constructive refusal to deal, as it concerns the conditions of access a dominant firm provides to downstream rivals. In its 2017 Google Shopping decision, the European Commission (‘Commission’) did not apply the indispensability requirement to hold Google’s self-preferencing in its general search results abusive.36 As such, the Commission’s legal reasoning raises questions about the position of self-preferencing as a self-standing theory of harm next to outright and constructive refusals to deal. In the case, Google was held liable for giving more prominent placement to its own comparison shopping service to the detriment of rival comparison shopping services. The Commission found that the more favourable positioning of Google Shopping in the general search results constituted an abuse of dominance, because it took traffic away from competing comparison shopping services and was thereby capable of having anticompetitive effects in the markets for comparison shopping services and general search services.37 In terms of the legal test, the Commission seems to refer to the notion of ‘leveraging’ by arguing that Article 102 TFEU does not only prohibit practices by a dominant undertaking tending to strengthen its dominance, but also covers conduct by which a dominant firm tends to extend its position ‘to a neighbouring but separate market by distorting competition’.38 According to the Commission, it is not novel to find an abuse for conduct consisting in the use of a dominant position in one market to extend that dominant position to one or more adjacent markets. In the words of the Commission, such conduct constitutes a ‘well-established, independent, form of abuse falling outside the scope of competition on the merits’.39 For this reason, the Commission did not apply the indispensability requirement and the other criteria laid down in Bronner. The General Court endorsed the approach of the Commission in its 2021 judgment by arguing that the lack of an express refusal to supply, consisting of a request or wish to be 35 Pablo Ibáñez Colomo, ‘Self-Preferencing: Yet Another Epithet in Need of Limiting Principles’ (2020) 43 World Competition 417; Pinar Akman, ‘The Theory of Abuse in Google Search: A Positive and Normative Assessment under EU Competition Law’ (2017) Journal of Law, Technology and Policy 301. 36 Shopping (n 3) para 651. Note that the Commission decision is under appeal. 37 ibid para 341. 38 ibid para 334. 39 ibid para 649.

The future of refusals to deal and margin squeezes  169 granted access and a consequential refusal, precludes practices from being described as a refusal to supply and analysed in accordance with the Bronner criteria.40 Even though such practices may result in an implicit refusal of access, they constitute an independent infringement of Article 102 TFEU ‘in view of their constituent elements which deviate, by their very nature, from competition on the merits’.41 According to the General Court, Google’s practices are an independent form of leveraging abuse and can thereby be distinguished from the conduct at issue in Bronner because the practices involve ‘active’ behaviour in the form of differential treatment in how Google promotes its own and degrades rival comparison shopping services instead of a ‘passive’ refusal of access to Google’s general results pages. For these reasons, the Commission was, indeed, not required to establish that the Bronner criteria were met in the view of the General Court.42 Ibáñez Colomo distinguishes between reactive and proactive remedies to determine whether indispensability is required as part of the legal test.43 In his view, the indispensability requirement should be applied in cases involving proactive remedies, namely structural remedies as well as behavioural remedies requiring a positive obligation and constant monitoring such as an access obligation. Indispensability would serve as a mechanism to limit competition intervention in these cases where the design, implementation and monitoring of remedies is complex. Such complexities are not present where reactive remedies are imposed, which consist of a one-off negative obligation to cease the conduct, so that there is no need in his view to use the indispensability requirement as a limiting principle in cases where reactive remedies suffice.44 However, the distinction between proactive and reactive remedies is not always clear-cut so that it may not be capable of serving as a useful mechanism to determine whether the indispensability requirement needs to be applied.45 This can be illustrated through the Google Shopping case. While the Commission argued in its decision that Google’s self-preferencing could be ended through a one-off cease-and-desist order to treat rival comparison shopping services no less favourably than its own comparison shopping service,46 the design and implementation of the remedy of equal treatment was in fact an intense process and required proactive measures.47 Nevertheless, Ibáñez Colomo has argued that the nature of the remedy can still be the guiding principle for determining whether indispensability is required by focusing on the substance of the competition intervention and remedy instead of on its form, as promulgated by the competition authority.48 This would mean that Google Shopping qualifies as a case involving a proactive remedy in which the indispensability requirement would need to be applied, despite the fact that the Commission presented the remedy as a reactive cease-and-desist order.

40 Case T-612/17 Google LLC, formerly Google Inc and Alphabet, Inc v European Commission ECLI:EU:T:2021:763, paras 232–3. 41 ibid para 233. 42 ibid para 240. 43 Ibáñez Colomo (n 35) 536. 44 ibid 536. 45 Dunne (n 18) 111 who also argues that there is no support in the case law and underlying policy concerns of refusals to deal to let the nature of the remedy instead of the characteristics of the violation determine whether the indispensability requirement has to be met. 46 Shopping (n 3) paras 699–700. 47 Dunne (n 18) 111. 48 Ibáñez Colomo (n 35) 544–5.

170  Research handbook on abuse of dominance and monopolization Considering that such a distinction between the substance and form of remedies is subject to interpretation and may sometimes only become clear at the stage of implementation of a competition decision (as illustrated by the Google Shopping case), it is submitted here that it is too uncertain to serve as the mechanism for deciding whether the indispensability requirement applies. The General Court, indeed, argued in its judgment that the applicability of the Bronner criteria should not depend on the measures ordered by the Commission to end an infringement.49 In the words of the General Court, there ‘can be no automatic link between the criteria for the legal classification of the abuse and the corrective measures enabling it to be remedied’.50 According to the General Court, there are elements in Google’s behaviour that are not directly linked to access to Google’s boxes on its general results page and do play a major role in the exclusionary effect, namely the degradation of competing comparison shopping services by means of adjustment algorithms and the promotion of Google’s own services in the search results. For these reasons, the General Court expressed the view that the fact that ‘one of the ways of ending the abusive conduct is to allow competitors to appear in the boxes displayed at the top of the Google results page’ does not mean that ‘the conditions for identifying the abuse must be defined having regard to that aspect alone’.51 A different approach to determining whether indispensability should be part of the legal test is proposed in Section IV below. The room to hold self-preferencing liable under Section 2 of the Sherman Act is quite limited under the current state of affairs, due to the US Supreme Court’s restrictive interpretation of the scope of antitrust liability for margin squeezes and refusals to deal. However, self-preferencing may become of relevance in the antitrust lawsuits started against Google at the end of 2020. For instance, one of the concerns raised in the antitrust lawsuit filed in December 2020 by the state of Texas, together with other states, relates to alleged preferential treatment by Google in the form of routing publisher inventory to its own advertising exchange and blocking competition from other exchanges on the same footing.52 In addition, the US House Judiciary Committee, Antitrust Subcommittee’s October 2020 report about the state of competition in digital markets (‘Antitrust Subcommittee’s Report’), called upon the US Congress to ‘consider overriding judicial decisions that have treated unfavorably essential facilities- and refusal to deal-based theories of harm’53 and to ‘consider establishing nondiscrimination rules to ensure fair competition and to promote innovation online’.54 As such, inspiration may be drawn from the EU approach towards self-preferencing to revitalize antitrust liability for refusals to deal in the US within and beyond digital markets.

Google (n 40) para 246. ibid para 244. 51 ibid para 245. 52 Case 4:20-cv-00957-SDJ, Google complaint led by the state of Texas (16 December 2020), paras 118–24, https://​www​.justice​.gov/​opa/​press​-release/​file/​1328941/​download, accessed 10 May 2021. 53 US House, Subcommittee on Antitrust, Commercial and Administrative Law of the Committee on the Judiciary, ‘Investigation of Competition in Digital Markets’ (October 2020), 397–8, https://​judiciary​ .house​.gov/​uploadedfiles/​competition​_in​_digital​_markets​.pdf​?utm​_campaign​=​4493​-519, accessed 10 May 2021. 54 ibid 382. 49 50

The future of refusals to deal and margin squeezes  171 B.

Google Shopping as a Constructive Refusal to Deal?

Despite the fact that the Commission did not apply the indispensability requirement in its Google Shopping decision, it did not refer to any constructive refusal to deal or margin squeeze cases to support this choice.55 Since the CoJ itself established that these practices do not require indispensability, the Commission could have used cases like Deutsche Telekom or TeliaSonera as relevant precedents. An analogy can indeed be drawn between self-preferencing by digital platforms and margin squeezes in the telecoms cases.56 Margin squeezes can lead to the exclusion from the market of downstream rivals due to the existence of an insufficient margin between the dominant firm’s upstream and downstream prices. Self-preferencing may similarly foreclose downstream competitors through the control the dominant platform holds over the conditions of access to the upstream market. More prominent placement of a dominant platform’s own downstream service in a ranking (such as the search ranking in Google Shopping) may harm downstream rivals to the extent they cannot effectively compete anymore with the vertically integrated dominant platform, because they receive less attention from consumers.57 Self-preferencing can also take the shape of a dominant platform’s preferential access to data about transactions conducted by downstream competitors.58 This is the concern expressed by the Commission in its ongoing Amazon and Facebook investigations.59 If a dominant platform holds preferential access to data providing information on, for instance, which products are successful or what could be upcoming trends among consumers, it may outcompete downstream retailers that cannot adapt their offerings to such insights to the same extent as the dominant platform. The key difference is that margin squeezes involve price-related conduct, while self-preferencing by digital platforms concerns non-price conduct. It is interesting to note in this regard that after the Commission’s publication of the Google Shopping decision, the EU Courts have clarified that the reasoning regarding the legal test set out in TeliaSonera applies not only to price-related terms but also to other terms of trade involving discriminatory conduct. In its December 2018 judgment in Slovak Telekom, the General Court pointed out that the CoJ in TeliaSonera did not refer to margin squeeze specifically ‘but rather to the supply of “services or selling goods on conditions which are disadvantageous or on which there might be no purchaser” and to “terms of trade” fixed by the dominant undertaking’.60 On these grounds, the General Court interpreted the wording of the CoJ in TeliaSonera as suggesting that the exclusionary practices referred to in the case did not solely concern a margin squeeze ‘but also other business practices capable of producing

In the context of the abusive nature of ‘leveraging’, see Shopping (n 3) para 334 and footnote 349. For an in-depth analysis of how the margin squeeze concept is relevant for vertically integrated online platforms, see Friso Bostoen, ‘Online Platforms and Vertical Integration: The Return of Margin Squeeze?’ (2018) 6 Journal of Antitrust Enforcement 355. 57 Inge Graef, ‘Differentiated Treatment in Platform-to-Business Relations: EU Competition Law and Economic Dependence’ (2019) 38 Yearbook of European Law 448, 478. 58 Vladya MK Reverdin, ‘Abuse of Dominance in Digital Markets: Can Amazon’s Collection and Use of Third-Party Sellers’ Data Constitute an Abuse of a Dominant Position under the Legal Standards Developed by the European Courts for Article 102 TFEU?’ (2021) 12 Journal of European Competition Law and Practice 181. 59 Amazon (n 3); Facebook (n 3). 60 Case T‑851/14 Slovak Telekom, as v European Commission ECLI:EU:T:2018:929, para 126. 55 56

172  Research handbook on abuse of dominance and monopolization unlawful exclusionary effects for current or potential competitors, like those classified by the Commission as an implicit refusal to supply access to the applicant’s local loop’.61 On appeal, the CoJ confirmed in Slovak Telekom that the General Court was right in how it had interpreted the TeliaSonera judgment. The CoJ clarified in Slovak Telekom that its statements in TeliaSonera referred not only to margin squeeze as a particular form of abuse ‘when it assessed the practices to which the conditions of the judgment in Bronner did not apply’.62 In other words, conduct of a dominant firm relating to non-price-related terms of access also does not have to meet the indispensability requirement of Bronner in order to be regarded as abusive. The CoJ explicitly stated in Slovak Telekom that the conditions laid down in Bronner do not apply ‘where a dominant undertaking gives access to its infrastructure but makes that access, provision of services or sale of products subject to unfair conditions’.63 According to the CoJ, such practices do not require the imposition of a duty on the dominant firm to give access to its infrastructure as a result of which a competition intervention will be ‘less detrimental to the freedom of contract of the dominant undertaking and to its right to property than forcing it to give access to its infrastructure where it has reserved that infrastructure for the needs of its own business’.64 As such, it is the distinction between outright and constructive refusals to deal that determines in the Court’s view whether the indispensability requirement needs to be met or not. Transposing this reasoning to Google Shopping would mean that the Commission was correct in its choice not to apply the indispensability requirement to determine whether Google’s self-preferencing was abusive. After all, Google did list rival comparison shopping services in its search results but in a way that did not allow them to effectively compete. Instead of claiming that leveraging constitutes an independent form of abuse, the Commission could thus have based its legal test on the reasoning of the CoJ in TeliaSonera regarding margin squeezes – assuming that the markets for general search and comparison shopping services can be regarded as vertically integrated.65 A basis for such an approach can be found in the General Court’s judgment. In support of its claim that Google’s self-preferencing formed an independent infringement of Article 102 TFEU distinct from that of refusal to supply, the General Court referred to the treatment of margin squeezing in TeliaSonera as an example of a case where indispensability was not applied even though it raised issues of access to a service.66 A further clarification along these lines by the CoJ on appeal in Google Shopping would be welcome to establish to what extent self-preferencing can, indeed, be regarded to fall within the existing box of constructive refusals to deal. Such a clarification would not only enhance the understanding of the scope for competition intervention against self-preferencing behaviour of digital platforms under Article 102 TFEU, but would also illustrate the future relationship between outright and constructive refusals more generally in the context of vertically integrated dominant firms that are widespread in other sectors too, like those for telecoms, postal, energy and railway services.

63 64 65 66 61 62

ibid para 126. Slovak Telekom (n 4) para 53. ibid para 50. ibid para 51. Disagreeing with this assumption and the relevance of margin squeeze, see Akman (n 35) 323–5. Google (n 40) paras 235 and 241.

The future of refusals to deal and margin squeezes  173 With the pressure rising in the US to strengthen antitrust enforcement, self-preferencing is also identified on that side of the Atlantic as a possibly anticompetitive practice under US antitrust law.67 Regarding the legal standard applicable to self-preferencing, the Antitrust Subcommittee’s Report recommended the US Congress to ‘consider whether making a design change that excludes competitors or otherwise undermines competition should be a violation of Section 2, regardless of whether the design change can be justified as an improvement for consumers’.68 This approach proposed by the US House Antitrust Committee would in fact go beyond the reasoning of the Commission in Google Shopping, where emphasis was placed on how Google’s self-preferencing created harm to consumers in the form of less choice and innovation in the market for online search. As such, the approach of the Commission arguably strikes a better balance between the interests of competitors and consumers by identifying self-preferencing behaviour that excludes competitors to the detriment of consumers.

IV.

THE RELEVANCE OF THE EXISTENCE OF A REGULATORY DUTY FOR THE LEGAL TEST UNDER ARTICLE 102 TFEU

A.

Rationale for the Existence of Sector-Specific Regulation as a Relevant Indicator

Even though the CoJ has clarified in Slovak Telekom that the indispensability requirement only applies to outright refusals to deal and not to constructive refusals to deal, a more normative point is whether this is a workable and logical distinction from a practical and policy perspective. In practice, the line between outright and constructive refusals to deal is not black and white. Whether access is given to an input may be a matter of degree.69 As noted by Dunne, in the context of its refusal to give access to interoperability information Microsoft argued before the General Court that ‘interoperability occurs along a continuum’ and that ‘it is not an absolute standard’.70 According to Microsoft, the minimum level of interoperability required for effective competition ‘is not difficult to achieve’.71 As such, Microsoft’s behaviour could also have been interpreted as a constructive instead of an outright refusal to deal because it was the extent of interoperability provided by Microsoft that was considered insufficient to enable effective competition.72 Similarly, the reason why the self-preferencing in Google Shopping raised competition problems was arguably because it resulted in rivals being displayed so low in the search results that the effect was the same as if they had not been listed at all.73 Beyond these issues of practical interpretation, there are also policy reasons why it is not logical to use the distinction between outright and constructive refusals to deal to determine whether the indispensability requirement is applicable. First, this choice leads to the situation

69 70 71 72 73 67 68

Subcommittee on Antitrust (n 57) 382. ibid 398. Dunne (n 18) 98. Microsoft (n 1) para 119, as noted by Dunne (n 18) 98 and footnote 161. Microsoft (n 1) para 120. ibid para 421. Ibáñez Colomo (n 35) 542.

174  Research handbook on abuse of dominance and monopolization that the room to hold outright refusals to deal abusive is more limited because the stricter Bronner criteria apply (including the indispensability requirement), even though this is a more serious type of abuse than constructive refusals to deal.74 As already discussed above, this provides dominant firms with incentives not to open up non-indispensable inputs at all. A possible consequence is that welcome forms of complementary innovations building on these non-indispensable inputs are not arising to the detriment of competition and consumers. Second, as stated by the General Court in Slovak Telekom, the seriousness of a refusal to deal should not solely depend on its form. Beyond the outright or constructive nature or form of the refusal, other factors are also of relevance.75 One of these other factors considered in Slovak Telekom is the regulatory context in which the refusal to deal takes place. It is submitted here that this is a more reliable and suitable factor to determine whether the indispensability requirement needs to be applied for assessing the abusive nature of a refusal to deal, whether outright or constructive. While the CoJ made clear in Deutsche Telekom that EU competition law applies in parallel to sector-specific regulation, it has been acknowledged in Lithuanian Railways (by the General Court) and Slovak Telekom (by the General Court and less explicitly by the CoJ) that the existence of a regulatory duty does influence the legal test for holding a refusal to deal abusive under Article 102 TFEU. The General Court held in Slovak Telekom that the applicable telecom legislation requiring Slovak Telekom as an operator with significant market power to provide unbundled access to its local loop constituted a relevant factor in the assessment of the abusive nature of its conduct. According to the General Court, such a regulatory duty defines the legal framework applicable to the case and ‘contributes to the determination of the competitive conditions under which a telecommunication undertaking carries on its business in the relevant markets’.76 The existence of a regulatory duty distinguishes the Slovak Telekom case from the Bronner case in the view of the General Court, where no regulatory obligation was at stake.77 Because the regulatory framework already acknowledged the need for access to Slovak Telekom’s local loop to enable effective competition in the Slovak market for high-speed internet services, the General Court argued that it was not necessary to demonstrate that such access was indispensable in line with the Bronner criteria.78 In Lithuanian Railways, the General Court similarly found that the indispensability requirement did not apply because sector-specific regulation already imposed a duty to deal on Lithuanian Railways.79 The General Court explained that the purpose of the exceptional circumstances laid down in Bronner is to ensure that the imposition of a duty to deal on a dominant firm ‘does not ultimately impede competition by reducing that undertaking’s initial incentive to build such infrastructure’.80 Where sector-specific regulation already imposes a duty to deal, such a requirement to protect the incentive of the dominant firm to invest in the creation of infrastructures is not present.81 In the words of the General Court: ‘Where there is a legal duty As put forward by Slovak Telekom in the appeal before the General Court, see Slovak Telekom (n 4) para 130. 75 ibid para 133. 76 ibid para 117. 77 ibid paras 118–20. 78 ibid para 121. 79 Lithuanian Railways (n 4) para 92. 80 ibid para 90. 81 ibid para 91. 74

The future of refusals to deal and margin squeezes  175 to supply, the necessary balancing of the economic incentives, the protection of which justifies the application of the exceptional circumstances developed in [Bronner], has already been carried out by the legislature at the point when such a duty was imposed’.82 Considering the underlying rationale of the Bronner criteria to find a balance between the interest in protecting short-term downstream competition and the interest in preserving the long-term incentives of dominant firms to invest in innovation,83 it is indeed logical not to require another balancing of whether access is indispensable for holding a refusal to deal abusive under Article 102 TFEU when a regulatory duty to deal already exists. While the analysis for establishing indispensability will not precisely overlap with the legislator’s considerations for imposing a regulatory duty to deal, the fact that a regulatory duty to deal applies would make a separate analysis of whether access is indeed also required under competition law unnecessary.84 In this regard, the current US approach is much more radical by precluding antitrust intervention where sector-specific regulation already exists.85 In a way, more attention for the existence of a regulatory duty to determine the applicable legal standard under Article 102 TFEU can bring EU competition law closer to US antitrust law. After all, the idea behind both approaches is that competition or antitrust enforcement should not redo the balancing already done by the legislator. However, the fundamental difference remains where the EU system relies on a complementary relationship between competition law and sector-specific regulation and the US system instead looks at antitrust law and sector-specific regulation as substitutes. B.

Relevance of Regulatory Duties in the Future

While the language of the General Court in Slovak Telekom and Lithuanian Railways was clear and outspoken on the relevance of the existence of a regulatory duty to deal, the CoJ was much more cautious in its judgment in Slovak Telekom. The CoJ acknowledged that a regulatory obligation can be relevant for assessing abusive conduct by referring to its judgment in Deutsche Telekom.86 However, after discussing the regulatory context, the CoJ continued by pointing at the fact that Slovak Telekom’s practices ‘did not constitute refusal of access to the appellant’s local loop but related to the conditions for such access’ as the reason why the Bronner criteria did not apply to the case.87 Even though the CoJ thus agreed with the General Court that no indispensability was required,88 it seems that the CoJ based this finding on the qualification of Slovak Telekom’s behaviour as a constructive refusal to deal rather than on the fact that a regulatory obligation existed.

ibid para 92. See Case C-7/97 Oscar Bronner GmbH & Co KG v Mediaprint Zeitungs- und Zeitschriftenverlag GmbH & Co KG, Mediaprint Zeitungsvertriebsgesellschaft mbH & Co KG and Mediaprint Anzeigengesellschaft mbH & Co KG ECLI:EU:C:1998:264, Opinion of Advocate General Jacobs, para 57 as repeated in a slightly different wording in Slovak Telekom (n 4) paras 47–9. 84 Dunne (n 18) 86–7. However, arguing that the existence of a regulatory duty to deal is not determinant based on a reading of the TeliaSonera judgment, see Ibáñez Colomo (n 35) 540 and 542. 85 linkLine (n 5). 86 Slovak Telekom (n 4) para 57. 87 ibid para 59. 88 ibid para 60. 82 83

176  Research handbook on abuse of dominance and monopolization However, the distinction between outright and constructive refusals to deal is often difficult to make. Google Shopping has already illustrated this and the same may be true for the ongoing Amazon and Facebook investigations regarding preferential access to data. The concern in these cases is that Amazon and Facebook have access to data about the activities of all retailers and advertisers, respectively, while the individual retailers and advertisers cannot see the full aggregate data.89 On the one hand, such behaviour can be qualified as a constructive refusal to deal because the retailers and advertisers do get access to part of the data. On the other hand, this conduct can also be seen as an outright refusal to give access to the full aggregate data required to understand the overall market trends. Depending on one’s interpretation, the legal standard will differ. For this reason, the existence of a precisely formulated regulatory duty seems a more objective and stable indicator to determine whether the indispensability requirement applies or not – even though other contextual factors not considered here may matter too. With the adoption of the DMA, certain gatekeeping platforms will be in a similar situation as the telecoms operators in TeliaSonera and Slovak Telekom where Article 102 TFEU complements the regulatory duties to provide access to search data (Art 6(11) DMA) and access to app stores (Art 6(12) DMA). This means that the Bronner criteria would not apply for assessing refusals to deal under Article 102 TFEU for which a regulatory duty already exists, because the legislator has already engaged in the necessary balancing of interests. Regarding the regulation of access to data in various industries including more traditional ones beyond digital platforms, several regimes already exist. As a horizontal framework, the GDPR provides data subjects with a general right to transfer personal data to another data controller irrespective of the sector or the purpose of the transfer.90 The PSD2 lays down an ‘access-to-account rule’ enabling third party providers to access a customer’s payment account information on the customer’s request in order to provide payment initiation or account information services.91 The Electricity Directive requires Member States to specify rules on access by eligible parties to metering and consumption data of the final customer.92 Additional rights to access and share data in the context of the Internet of Things are introduced in the Data Act.93 The rise of sector-specific duties to share data will reduce the need for EU competition law to intervene to open up datasets. However, where relevant, Article 102 TFEU can be invoked to complement the sector-specific frameworks. A question in this regard is to what extent the data access mandated under Article 102 TFEU can extend beyond the data access regulated in the sector-specific frameworks without applying the Bronner criteria. For instance, the scope of the GDPR’s right to data portability is limited to personal data ‘provided by the data subject’.94 The Article 29 Working Party (now the European Data Protection Board) has interpreted this term expansively as not only covering personal data knowingly and actively shared by data subjects (such as one’s contact details) but also personal data collected by observing the behaviour of data subjects and their use of a service or device (such as one’s interactions Amazon (n 3); Facebook (n 3). GDPR (n 10) art 20. 91 PSD2 (n 11) arts 66 and 67. For a discussion, see Oscar Borgogno and Giuseppe Colangelo, ‘Data, Innovation and Competition in Finance: The Case of the Access to Account Rule’ (2020) 31 European Business Law Review 573. 92 Electricity Directive (n 11) art 23(1). 93 Proposal for a Regulation of the European Parliament and of the Council on harmonised rules on fair access to and use of data (Data Act) [2022] COM(2022) 68 final, arts 4 and 5. 94 GDPR (n 10) art 20(1). 89 90

The future of refusals to deal and margin squeezes  177 on a social network). However, inferred or derived personal data is not included, while this data may carry competitive value, too, and could thus form the basis for an access request.95 As inferred or derived personal data falls outside the scope of the regulatory duty in the GDPR, the Bronner criteria will still need to be applied to impose a duty on a dominant firm to share such data under Article 102 TFEU. A similar situation would occur in the payment sector for requests by third parties to access payment data held by banks beyond the purposes of providing payment initiation and account information services as regulated by the PSD2. Considering the restrictive interpretation by the US Supreme Court of antitrust liability for refusals to deal and margin squeezes, the role of sector-specific regulation may become even more important than in the EU. Regarding self-preferencing practices, the US Antitrust Subcommittee’s Report already recommended the US Congress to adopt non-discrimination rules requiring ‘dominant platforms to offer equal terms for equal service’ applicable to price as well as to terms of access. Reference was made in this regard to previous experiences in relation to the net neutrality rules in the form of the – later repealed – 2015 Open Internet Order of the Federal Communications Commission, non-discrimination requirements for railroads designated as common carriers and for cable operators, as well as non-discrimination duties imposed as a complement to divestitures in antitrust enforcement such as in AT&T’s 1982 divestiture of Bell that was combined with an equal access obligation.96

V.

INTERPRETING THE INDISPENSABILITY REQUIREMENT

A.

The Standards of Indispensability

Where the Bronner criteria do apply to establish liability under Article 102 TFEU, a nuanced approach is necessary. The paradox here is that the Bronner criteria may be the least likely to be met in situations where they do apply, considering the high standards required to prove indispensability.97 However, the mere fact that the Bronner criteria are applicable should not prejudice the outcome of the competition analysis. Their mere applicability does not imply that the refusal to deal at stake is unlikely to be abusive under Article 102 TFEU. In this sense, the Bronner criteria should be read as a set of conditions balancing the interests at stake where the legislator has not already done so. The indispensability requirement is arguably the most debated among the Bronner criteria. It seeks to determine whether actual or potential substitutes for the input exist. Actual substitutes are readily available alternatives, for instance inputs sourced from other market players. Potential substitutes involve alternatives that the access seeker would itself be able to produce. The standards for applying indispensability have varied across cases. Regarding the availability of actual substitutes, the CoJ remarked in Bronner that access to Mediaprint’s nationwide newspaper home-delivery scheme was not indispensable because alternatives were available to Bronner for distributing its daily newspapers, such as delivery by post and sales in shops,

95 Article 29 Working Party, ‘Guidelines on the Right to Data Portability’ (2017) 16/EN WP 242 rev.01, 9–11. 96 Subcommittee on Antitrust (n 57) 382–3. 97 Dunne (n 18) 114.

178  Research handbook on abuse of dominance and monopolization even though these alternatives were less advantageous.98 In relation to the existence of potential substitutes, the Court referred to the need to establish whether ‘there are no technical, legal or even economic obstacles capable of making it impossible, or even unreasonably difficult’ for another publisher of daily newspapers to set up its own nationwide home-delivery scheme alone or together with others.99 The Court also made clear that the small circulation of the daily newspapers could not constitute a reason for arguing that the creation of an alternative scheme is not economically viable.100 For access to be regarded as indispensable, it would be necessary in the Court’s view to establish that it is not economically viable to set up a second home-delivery scheme with a circulation comparable to that of Mediaprint.101 As was later confirmed by the CoJ in IMS Health, in order to accept the existence of economic obstacles, it must be established that duplication of the requested input ‘is not economically viable for production on a scale comparable to that of the undertaking which controls the existing product or service’.102 While this is a strict standard, the presence of switching costs and dependencies of users on an existing input can be factors leading to a conclusion that access is indispensable. In IMS Health, the Court referred to ‘exceptional organisational and financial efforts’ that potential users would have to make in order to work with an alternative input.103 Because of the extent of participation by pharmaceutical companies in the development of the copyrighted brick structure of IMS, the CoJ held that it was for the national court to determine whether a supplier of an alternative structure would ‘be obliged to offer terms which are such as to rule out any economic viability of business on a scale comparable to that of [IMS]’.104 B.

Tailoring the Standards of Indispensability to the Market Situation at Stake

As the imposition of a duty to deal requires balancing the interest in protecting competition in the short term with the interest in protecting incentives to invest in innovation in the long term,105 the standards for qualifying a refusal to deal as abusive may vary depending on the circumstances of the case. While the General Court relied on the same legal test as used in earlier cases like Bronner and IMS Health,106 it applied lower standards in the Microsoft judgment including for the indispensability requirement. Regarding the indispensability of access to Microsoft’s interoperability information, the General Court namely explained that competitors needed to be able to interoperate with the Windows operating system on an equal footing,107 whereas the CoJ had stated in Bronner that access is not indispensable if alternatives are available even though they are less advantageous. As Microsoft held a quasi-monopoly position in the market for client personal computer operating systems protected by significant network effects, the General Court may have tailored the application of the legal test to the Bronner (n 6) para 43. ibid para 44. 100 ibid para 45. 101 ibid para 46. 102 IMS Health (n 6) para 28. 103 ibid para 29. 104 ibid para 29. 105 See Bronner (n 83) para 57. 106 Microsoft (n 1) paras 331–3. 107 ibid para 421. 98 99

The future of refusals to deal and margin squeezes  179 market situation in the case.108 In this regard, calls have also been made to lower the standards for imposing a duty to deal in data-driven markets where data is generated as a by-product of providing a service and incentives to produce such data would not be affected by duties to share data.109 Such tailoring of the applicable standards seems logical as the various interests and incentives involved will have a different weight depending on the market situation. As argued elsewhere, the presence of external market failures could be a key indicator to be taken into account.110 The need for competition intervention is arguably stronger in these cases as the self-correcting mechanism of the market will not work as effectively due to the market situation grown around the dominance of the incumbent. The presence of strong network effects, switching costs, and entry barriers may be an indication that external failures are present. The key issue would be to determine whether market characteristics enable an incumbent to artificially extend its dominance in time. In this regard, external market failures will be particularly likely to occur in the context of standardization or in cases where network effects and lock-in prevent rivals from successfully introducing alternative products due to the fact that consumers are not interested in switching to a new system.111 This is arguably what happened in IMS Health and Microsoft, as a result of which lower standards for the fulfilment of the Bronner criteria are desirable to open up the respective markets protected by external market failures. In this regard, the standards used for holding an input indispensable for competition can also be tailored to the relevant market situation. Although one may argue that this creates concerns regarding legal certainty, the status quo is hardly more predictable for dominant firms. For instance, Google Shopping would probably require a more flexible interpretation of indispensability as compared to the Bronner case, considering the strong position held by Google in the search engine market exceeding 90 per cent in most Member States and protected by high barriers to entry in the form of network effects, the lack of multi-homing by consumers and the scope of data held by Google.112 The General Court, indeed, noted in its judgment that Google’s general results page ‘has characteristics akin to those of an essential facility’ because there is no actual or potential substitute to replace the traffic comparison shopping services would otherwise gain through Google’s general results pages.113 According to the General Court, the Commission considered Google’s traffic as indispensable ‘by finding that the traffic generated by Google’s general search pages was not “effectively replaceable” and that other sources of traffic were not “economically viable”’.114 Such tailoring of the indispensability requirement and the other Bronner criteria does not imply that Article 102 TFEU is interpreted in a way that facilitates access to inputs that are not indispensable but merely ‘convenient’ for rivals to compete.115 Depending on the market situation, the interests involved

Inge Graef, ‘Rethinking the Essential Facilities Doctrine for the EU Digital Economy’ (2019) 53 La Revue juridique Thémis de l’Université de Montréal 33, 55. 109 For a discussion, see Jacques Crémer, Yves-Alexandre de Montjoye and Heike Schweitzer, ‘Competition Policy for the Digital Era’ (2019) Expert Report for Commissioner Vestager, 105–6. 110 Inge Graef, EU Competition Law, Data Protection and Online Platforms: Data as Essential Facility (Kluwer Law International 2016) 191. 111 ibid 191–2. 112 Shopping (n 3) paras 271–330. 113 Google (n 40) paras 224–6. 114 ibid para 227. 115 See Ridyard (n 12). 108

180  Research handbook on abuse of dominance and monopolization will have to be weighed differently and this needs to be reflected in the standards applied for the fulfilment of the Bronner criteria. Similarly, the standards applicable under US antitrust law to establish liability for refusals to deal could be tailored to the market situation at stake in order to revitalize antitrust enforcement in a way that reflects the market reality. While the notion of indispensability does not play a role in the legal test under Section 2 of the Sherman Act, the interpretation of the requirement of whether a monopolist is willing to sacrifice short-term profits in order to achieve an anticompetitive end may differ depending on the market situation. Where economic characteristics protect the position of a monopolist to the extent of allowing it to artificially extend its monopoly, there is arguably also a stronger need for US antitrust law to intervene.

VI. CONCLUSION Instead of using the distinction between outright and constructive refusals to deal for determining whether the strict Bronner criteria apply, the chapter submitted that the existence of a regulatory duty forms a more reliable and objective indicator for determining the legal test. The imposition of a duty to deal requires finding a balance between the interest in protecting competition in the short run and the interest in protecting innovation incentives in the long run. When a regulatory duty to provide access exists under a sector-specific framework, this balancing is already done by the legislator so that there is no need for an additional balancing under Article 102 TFEU by applying the Bronner criteria. As regulatory access duties are being developed for gatekeeping platforms in the DMA and for data in various sectors such as payment and energy, the chapter argues that these developments should be taken into account when assessing refusals to deal in these sectors under Article 102 TFEU. Even though more regulatory duties will likely be designed targeting concerns of access to digital platforms and access to data in various markets – including those beyond digital markets – in the near future, there is still a role for Article 102 TFEU to apply in parallel and complement the ex ante regulation by ex post competition enforcement – as is the case in telecoms. The fact that the indispensability requirement applies should not imply that the refusal to deal is unlikely to be abusive under Article 102 TFEU. The standards for the fulfilment of the Bronner criteria can be tailored to the market situation at stake, so that the weight of the different interests can be adjusted in order to address the underlying competition concerns. Convergence with the US approach seems unlikely, considering that US antitrust law, contrary to EU competition law, requires an antitrust duty to deal as a condition for establishing liability for a margin squeeze and rules out the application of US antitrust law if a sector-specific regime already regulates the same conduct. However, the pressure to strengthen the enforcement of the US antitrust rules is rising to the extent it may bring their interpretation closer to the more interventionist approach of EU competition law. To develop a legal test for self-preferencing and constructive refusals to deal more broadly, the US could draw inspiration from how the EU has been trying to balance the different interests at stake within the context of the enforcement of the EU competition rules as well as the design of sector-specific regulation in digital markets and beyond.

10. Tying and bundling A. Douglas Melamed

I. INTRODUCTION ‘Tying’ and ‘bundling’ denote practices in which two or more separate products are linked together in a sales transaction. They can take numerous forms.1 ‘Tying’, which is sometimes called ‘pure bundling’, occurs when a seller sells one product (the ‘tying product’) on the condition that the buyer also purchase the other product(s) (the ‘tied product(s)’).2 The sales of the tying and tied products might be simultaneous, or the buyer might purchase the tying product and commit to purchase the tied products(s) at a later date including in after-markets for parts and servicing for durable goods. The tying and tied products might be sold in fixed proportions or in variable proportions. Tying is commonly implemented by market conduct in which the seller refuses to sell the tying product unless the buyer purchases or agrees to purchase the tied product(s). Tying that is implemented by designing the tying product so that it is functionally compatible with complementary tied product(s), but not with substitutes for the tied product(s), is called ‘technical tying’. ‘Bundling’, or to be more precise ‘mixed bundling’, occurs when a seller provides incentives to induce the buyer to purchase multiple products but does not by market conduct or product design condition the sale of any one product on the buyer’s purchasing or committing to purchase any other product(s). The incentive is most often a discount so that the total price when the products are purchased in a bundle is less than the total price when they are purchased separately. Tying is discussed in Section II, with a summary of the basic economics followed by summaries of US and EU law. A similar structure is used for the discussion of bundling in Section III.

Because tying and bundling involve separate products, they should be distinguished from practices that are intended to require or induce the purchaser to buy from the seller a larger quantity of a single product. These practices include exclusive dealing and requirements agreements pursuant to which the purchaser agrees to buy all of its requirements of a single product during a specified time period from the seller, minimum quantity requirements, quantity discounts, and loyalty discounts pursuant to which the purchaser receives a discount if it buys from the seller a specified share of its purchases of a particular type of product during a specified time period. 2 The term ‘pure bundling’ is sometimes used more narrowly to denote selling products jointly in fixed proportions. See, eg, Commission, ‘Guidance on the Commission’s Enforcement Priorities in Applying Article 82 of the EC Treaty to Abusive Exclusionary Conduct by Dominant Undertakings’ [2009] OJ C45/2 (‘Article 102 Guidance’), para 48. 1

181

182  Research handbook on abuse of dominance and monopolization

II.

TYING (OR FORCED BUNDLING)

A.

The Basic Economics of Tying

Courts and commentators have in the past had extreme and simplified understandings of tying and its effects. The US Supreme Court was hostile to tying arrangements for many decades and repeatedly asserted that tying arrangements ‘serve hardly any purpose beyond the suppression of competition’ by leveraging the seller’s market power over the tying product into the market for the tied product.3 That hostility provoked a push back from the early Chicago School economists. In a seminal article, Ward Bowman argued that the idea that tying arrangements were intended to leverage tying product market power into the tied product market was implausible because there is only one monopoly profit to be made from the tying product, that profit could be realized by charging the monopoly price for the tying product, and the tying product monopolist thus could not increase its profits by using some of that power instead to coerce sales of the tied product. Bowman concluded that tying arrangements must further some other, presumably procompetitive, objectives.4 Neither the ‘one monopoly profit theory’ nor the sweeping concern about ‘leverage’ in the early tying cases is correct. The one monopoly profit theory is correct only in certain circumstances, the most important of which is that the tying and tied products must be used in fixed proportions.5 The leveraging concern ignores the several possible efficiency reasons for tying. Depending on the circumstances, tying can increase or decrease economic welfare. 1. Possible benefits from tying Tying can provide procompetitive benefits in some circumstances. First, tying can enhance product quality and protect the seller from undeserved reputational injury that could impair its competitive effectiveness.6 For example, the manufacturer of a complex machine might tie parts or other supplies to ensure that they meet certain quality standards and are well suited for use with its machines. The effect could be to improve the quality of the overall system and protect the manufacturers against customer complaints about problems with the machine that are really caused by inferior parts or supplies. Sometimes, these benefits can be achieved by the less restrictive means of providing better information to customers. Second, tying can result in lower production and distribution costs, especially where the tied and tying products have joint costs such that increased sales of one reduce the costs of both.7 Tying can also reduce purchasers’ search and transaction costs. Sometimes these benefits can

Northern Pacific Railway v United States 356 US 1, 6 (1958), quoting Standard Oil Co of California v United States 337 US 293, 305–6 (1949). 4 Ward S Bowman, Jr, ‘Tying Arrangements and the Leverage Problem’ (1957) 67 Yale Law Journal 19. 5 See Joseph Farrell and Philip J Weiser, ‘Modularity, Vertical Integration, and Open Access Policies: Towards a Convergence of Antitrust and Regulation in the Internet Age’ (2003) 17 Harvard Journal of Law and Technology 85. 6 Eg, Michael L Katz, ‘Vertical Contractual Relations’ in Richard Schmalensee and Robert Willig (eds), Handbook of Industrial Organization (Elsevier 1989). 7 Eg, Erik Hovenkamp and Herbert Hovenkamp, ‘Tying Arrangements and Antitrust Harm’ (2010) 52 Arizona Law Review 925. 3

Tying and bundling  183 be achieved by mixed bundling price incentives that do not require the customer to take the tied product.8 Third, tying can be used in two-sided platforms to optimize platform value by providing a subsidy to one side where negative prices are not feasible. For example, Google might bundle other services with its search services in order to attract users and thereby increase the value of the platform to advertisers.9 Fourth, tying can be used for price discrimination by, for example, selling a complex machine at a modest price and requiring customers to buy complementary supplies from it. The tying requirement enables a seller to charge more to those customers that make greater use of the machine and its supplies. It also enables the seller to charge a lower price for the machine itself and thus to expand output by selling to customers that are unwilling or unable to pay the higher price the firm would charge if it charged a uniform price to everyone. When used in aid of output-expanding price discrimination, tying can increase total welfare and, in some circumstances, consumer welfare.10 It benefits sellers that are unable to engage in perfect price discrimination when selling the tying product itself and are thus unable to capture all of the surplus potentially available from the tying product; that is another, common circumstance in which the one monopoly profit theory is inapplicable. Fifth, usage-based pricing like that described above, which is sometimes called ‘metering’, can be an efficient means of enabling the seller to share the risk that the buyers will not be able to use the tying product. If the risk is more efficiently borne by the seller, perhaps because it has a portfolio of heterogeneous customers all of whom are subject to the metering tie, the tying arrangement can reduce the total costs of the uncertainty and increase output.11 Sixth, tying can solve a Cournot complements problem when the seller has market power in both the tying and tied products.12 This problem, however, can usually be solved by mixed bundling, which would permit the buyer to purchase the tied product elsewhere if it thought that preferable to buying both products from the same seller. 2. Possible harms from tying Tying can also cause economic harms. These include both exclusion of rivals and exploitation of trading partners. The immediate effect of tying is to divert purchases from, and thus harm, existing or potential new competitors in the tied product market. It could divert purchases by satisfying the buyer’s demand for the tied product in whole or in part or by increasing the cost to the buyer of purchasing the tied product from a competitor. The latter might happen if, for example, the seller includes the tied product with the tying product and thus imposes on buyers the costs

Under some circumstances, tying can also enable more efficient allocation of limited distribution slots when sellers with different portfolios of products compete for the slots. See Doh-Shin Jeon and Domenico Menicucci, ‘Bundling and Competition for Slots’ (2012) 102 American Economic Review 1957. 9 Andrea Amelio and Bruno Jullien, ‘Tying and Freebies in Two-Sided Markets’ (2012) 30 International Journal of Industrial Organization 436; Jean Charles Rochet and Jean Tirole, ‘Tying in Two-Sided Markets and the Honor All Cards Rule’ (2008) 26 International Journal of Industrial Organization 1333. 10 Eg, Hovenkamp and Hovenkamp (n 7). 11 ibid. 12 ibid. 8

184  Research handbook on abuse of dominance and monopolization of disposing of or disconnecting the tied product if it wishes to use a tied product provided by a competitor. Diverting purchases from competitors in the tied product market can injure competition in two basic ways. First, exclusion of rivals in the tied product market can injure competition in the tied product market by enabling the seller of the tying product to obtain market power in the tied product market. The likelihood of this effect will depend, among other things, on the portion of the tied product market affected by the tying arrangement, whether the arrangement causes rivals to lose enough sales to drive them below efficient scale and thereby reduces their competitive effectiveness, and whether the tied product market includes differentiated products that might increase the resistance of buyers to the tying arrangement. If the one monopoly profit theory is inapplicable or if the seller is because of fringe competition or otherwise unable to capture the full monopoly profit from the tying product market, the seller could benefit directly from the creation of market power in the tied product market by being able to earn supracompetitive profits in that market.13 Even if the one monopoly profit theory does apply, the seller could benefit by charging supracompetitive prices to buyers of the tied product that do not also buy the tying product. Even if the one monopoly profit theory does apply in the short term, tying could be also harmful in a different way. If success in the tying product market requires that customers have access to suitable complements in the tied product market, exclusion of rivals in the latter could increase the costs of entry into the tying product market by forcing would-be entrants into that market also to enter the tied product market or to induce others to do so.14 Second, tying can entrench a monopoly in the tying product market, even if it does not create market power in the tied product market, if it weakens competitors in the tied product market that are themselves potential entrants into the tying product market or potentially important complements to such entrants. Excluding or weakening those competitors could reduce the likelihood of entry into the tying product market.15 Tying can also harm consumers in the tying product market, even if it has no exclusionary effect in the tied product market. Tying can, for example, enable the monopolist to evade a regulatory or contractual price constraint by tying an unregulated product to the regulated product, charging a nominally compliant price for the regulated product, and exploiting the market power of that product by charging a supracompetitive price for the tied, unregulated product.16 In this situation, the supracompetitive portion of the nominal price for the tied product, that is, the amount that exceeds the price at which the buyer could purchase the tied 13 Michael D Whinston, ‘Tying, Foreclosure, and Exclusion’ (1980) 80 American Economic Review 837. 14 Tying could also enable the seller to obtain rents available from future investments in the tied product market if tied product rivals would be able to obtain those rents in the absence of tying. Dennis W Carlton and Michael Waldman, ‘Upgrades, Switching Costs, and the Leverage Theory of Tying’ (2012) 122 The Economic Journal 675. 15 Dennis W Carlton and Michael Waldman, ‘The Strategic Use of Tying to Preserve and Create Market Power in Evolving Industries’ (2002) 33 Rand Journal of Economics 194. That was the basic theory in United States v Microsoft Corp 253 F3d 34 (DC Cir 2001) (en banc), in which the court found that Microsoft had engaged in unlawful monopoly maintenance by a course of conduct that weakened its rival in the Internet browser market and thus made it less likely that that rival would enter the operating system market in which Microsoft had monopoly power or aid other entrants into that market. 16 Holders of patents that are encumbered by fair, reasonable and non-discriminatory (FRAND) commitments sometimes seek to do this by licensing packages of patents that include patents not subject

Tying and bundling  185 product absent the tying arrangement, is a result of the market power of the regulated tying product and is in substance consideration for the tying product. This is a variation on the idea, which is sometimes called ‘Baxter’s Law’ or the ‘Bell Doctrine’, that regulated monopolies have an incentive to extend their power into unregulated markets.17 In addition, as explained above, tying can facilitate price discrimination among users of the tying product by enabling the seller to charge each according to its use of the complementary tied product. Depending on the details, such tying could reduce welfare of the heavy users that would otherwise have bought the tying product at a lower, uniform monopoly price more than it increases the welfare of the light users that buy the tying product with the tying arrangement but would not have bought the tying product at a uniform monopoly price. If so, the tying arrangement will reduce consumer welfare overall, even if it increases output and thus total welfare compared to the alternative of selling the tying product at a single monopoly price. If the demand for the tying product is largely independent of that for the tied product, tying can enable the tying product monopolist to increase profits even if it does not increase output.18 As a general matter, price discrimination reduces total welfare if it does not increase output, and it can reduce consumer welfare even if it increases output.19 B.

Governing Legal Principles

1. United States It has long been clear that alleged tying arrangements are unlawful if it is found after ‘thorough examination of the purposes and effects of the practices involved, that the general standards of the Sherman Act have been violated’.20 These general standards are embodied in what is known as the ‘rule of reason’, which prohibits only conduct that injures or is likely to injure competition and cannot be justified by efficiencies. Most tying arrangements, however, are assessed by special rules applicable only to tying. a) The per se rule In a series of cases beginning in the 1940s, the Supreme Court held that forced tying of two or more separate products is unlawful per se. In other words, tying arrangements could be found to be unlawful without proof that the particular tying at issue harmed competition.21

to the FRAND commitment and charging supracompetitive royalties for the tied patents or the package as a whole. 17 Jonathan E Nuechterlein and Philip J Weiser, Digital Crossroads (MIT 2005) 18–19; Paul L Joskow and Roger G Noll, ‘The Bell Doctrine: Applications in Telecommunications, Electricity, and Other Network Industries’ (1999) 51 Stanford Law Review 1249. 18 This can be illustrated by a simple example. Assume there are two products, X and Y, each of which costs $5 to manufacture and sell. Assume further that 100 buyers are willing to pay $10 for X and $8 for Y, and that a different 100 buyers are willing to pay $8 for X and $10 for Y. If each of the products were sold at separate, uniform prices, the profit maximizing price for each would be $8; and the monopolist’s profit would be $1200. If the monopolist can tie X and Y together, it can sell them for a total price of $18; and its profit would be $1600, with no increase in output. 19 These possible harms are discussed in detail in Einer Elhauge, ‘Tying, Bundled Discounts, and the Death of the Single Monopoly Profit Theory’ (2009) 123 Harvard Law Review 397. 20 Eg, Fortner Enterprises v United States Steel Corp 394 US 495, 500 (1969). 21 Eg, ibid; International Salt Co United States 332 US 392 (1947).

186  Research handbook on abuse of dominance and monopolization More recently, however, while courts have continued to say that tying is unlawful per se, they have made clear that tying is unlawful per se only under certain conditions. The Supreme Court held in 1977 that tying is unlawful per se only if the defendant is shown to have market power in the tying product.22 In the Jefferson Parish case in 1984, the Court explained that: the essential characteristic of an invalid tying arrangement lies in the seller’s exploitation of its control over the tying product to force the buyer into the purchase of a tied product that the buyer either did not want at all, or might have preferred to purchase elsewhere on different terms.23

The Court spelled out the current principles governing tying arrangements, which are sometimes called a ‘quasi per se rule’.24 Under this rule, tying is unlawful if four conditions are met: (i) the arrangement must link two separate product markets; (ii) the sale of one product is conditioned on the purchase of the other; (iii) the seller has sufficient market power in the tying product to distort competition in the market for the tied product; and (iv) the arrangement affects a ‘not insubstantial amount of commerce in the tied product market’. The separate product requirement (i) is intended to identify those situations in which the alleged tying is capable of foreclosing competition in a ‘product market distinct from the market for the tying item’. Thus, whether there are two separate products for tying purposes depends, not on the functional relation between the products or on whether they are complements, but rather on whether ‘there is a sufficient demand for the purchase of [the tied product] separate from [the tying product] to identify a distinct product market in which it is efficient to offer [the tied product] separately’.25 The courts have not spelled out the circumstances in which it will be deemed ‘efficient’ to offer the tied product separately. Possibilities include whether it is efficient for other firms that might have scope economies unavailable to the defendant to sell the tied product separately, whether it is efficient for standalone firms that might lack scope economies available to the defendant to sell the product separately, whether the price would exceed the defendant’s incremental or average total costs, or whether the price would exceed the defendant’s opportunity cost. As a practical matter, courts avoid refined analyses of such questions and assess the separate product market issue more impressionistically, often by inquiring simply whether there are substantial sales of the tied product, by the defendant or by others, separate from the tying product.26 Four of the nine Supreme Court Justices signed separate opinions in Jefferson Parish stating that they would abolish the per se rule for tying arrangements and assess alleged tying under the rule of reason. Depending on how the separate products test is applied, there might be little substantive difference. If a defendant can show that it would not be efficient to sell the tied product separately, its declining to do so would by definition be efficient; and its tying of the two products would be lawful under the rule of reason as long as it can show a legitimate reason for selling the tied product instead of relying on third parties to do so. Put differently, proof that it would make good business sense for the defendant to tie the two items without United States Steel Corp v Fortner Enterprises, Inc 429 US 610 (1977). Jefferson Parish Hosp Dist No 2 v Hyde 466 US 2, 12 (1984). 24 Eg, Einer Elhauge, ‘Rehabilitating Jefferson Parish: Why Ties without a Substantial Foreclosure Share Should Not Be Per Se Legal’ (2016) 80 Antitrust Law Journal 463. 25 Jefferson Parish (n 23) 21–2. 26 Eg, Kaufman v Time Warner 836 F3d 137, 142 (2d Cir 2016). 22 23

Tying and bundling  187 regard to any distortion of competition in the market in which the tied item might be sold separately should both suffice to demonstrate that there are not two markets sufficient to justify application of the quasi per se rule and enable the arrangement to survive a challenge under the rule of reason. The conditioning requirement (ii) is readily satisfied if, as is often the case, the seller literally conditions sale of the tying product on the purchase or commitment to purchase the tied product. Sometimes, however, it is not clear from the face of the transaction whether the conditioning requirement really was met. For example, the seller in a negotiated transaction might discuss only terms for the sale of both products without ever saying that the sale of one product is conditioned on purchase of another. In that case, the factfinder would have to determine whether the seller in effect made clear that it would not sell the tying product separately or, instead, the buyer chose to purchase both products from the seller without regard to any such conditioning. If the buyer never asked to purchase the tying product separately and never inquired about the terms on which the tying product would be sold separately, the factfinder is unlikely to find that the buyer was forced to buy the tied product from the defendant. Alternatively, the seller might offer the tying product separately but only at a price that is higher than the price charged for the tying product when the tying and tied product are both purchased. An antitrust plaintiff might claim that the price difference was coercive, and the defendant might argue that it reflected cost savings from the tying arrangement or other economies of scope. Lower court decisions have inquired whether the price charged when the tying product is sold separately is commercially reasonable and have found no conditioning if there are substantial sales of the tying product at the higher price.27 The market power (iii) and effect on commerce (iv) requirements are straightforward. The latter has been understood simply to require foreclosure of a sufficiently large amount of commerce in the tied product market to warrant antitrust concern.28 One court has repeatedly stated in dicta an additional requirement that the plaintiff show ‘anticompetitive effects’ in the tied product market;29 but none of the cases that have articulated this requirement explained what, if anything, must be shown to meet the requirement beyond the inference that might be drawn from proof of the other four elements.30 b) Justifications and other complications The Jefferson Parish articulation of a qualified per se rule would seem to suggest that there can be no defence of an arrangement that meets all four conditions for condemnation under the per se rule. The Supreme Court has, however, noted more recently that many tying arrangements ‘are fully consistent with a free, competitive market’,31 and several lower court cases decided

Ways and Means v IVAX Corp 506 F Supp 697 (ND Cal 1979); Nobel Scientific Indus v Beckman Industries 670 F Supp, 1313, 1324 (D MD 1968). 28 Jefferson Parish (n 23) 16. 29 Eg, Kaufman (n 26) 141; E & L Consulting, Ltd v Domas Indus, Ltd 472 F 23 (2d Cir 2006). 30 One case suggested that the injury to competition element requires only ‘some showing that the illegal tying arrangement results in actual foreclosure of competition in the tied product market’. Yentsch v Texaco, Inc 630 F2d 56, 59 (2d Cir 1980). That seems indistinguishable from element (iv) in the Jefferson Parish test. 31 Illinois Tool Works Inc v Independent Ink, Inc 547 US 28 (2006). The Court held that a patent does not necessarily give the patent holder market power and that the plaintiff must prove that the defendant has market power in the tying product in all tying cases. 27

188  Research handbook on abuse of dominance and monopolization both before and after Jefferson Parish have found otherwise unlawful tying arrangements to be justified by special circumstances.32 Because of the special attributes of intellectual property, including in particular uncertainty about the validity and scope of patents, tying involving such property is especially likely to escape per se condemnation.33 In one case, for example, the court rejected a claim that the tying of ‘essential’ and ‘nonessential’ patents constituted patent misuse. The court declined to treat the arrangement as unlawful per se ‘[i]n light of the efficiencies of package patent licensing and the important differences between product-to-patent tying arrangements and arrangements involving group licensing of patents …’.34 Technical tying is rarely found to be unlawful. Courts are reluctant to second-guess engineering and other product design decisions, and they generally find or assume that the product design feature that excluded competing alternatives to the tied product was intended to further some legitimate product improvement purpose.35 Some recent cases, however, have examined the technical justifications for the arrangement and have found them to be without factual support.36 Notably, these cases have been monopolization cases under Section 2 of the Sherman Act. A unilateral change in product specifications that has the effect of excluding rivals from the market for complementary products is not likely to involve the kind of agreement or coordination among separate entities necessary to subject the conduct to Section 1 of the Sherman Act. While a finding that the defendant has market power in the tying product will suffice for a case brought under Section 1, Section 2 has the more demanding requirement that the conduct at issue be shown to have had or to be likely to have the effect of creating or maintaining monopoly power for the defendant. The court held in Microsoft that the per se rule did not apply to plaintiff’s claim that Microsoft had unlawfully tied its personal computer (PC) operating system to its Internet browser. Microsoft argued that the operating system and browser were not separate products and that the integration of the two reflected continuing innovation that increased the functionality of the operating system. The court held that the separate demand test of the per se rule was unsuitable for determining whether the operating system and the browser were separate products, and that the test could not be used to find the alleged tying arrangement unlawful per se, because the test necessarily focused on past sales and other evidence of existing consumer preferences and did not take sufficient account of the possibility of new product improvements that might change consumer demand in the future. The court emphasized the 32 Eg, United States v Jerrold Elecs Corp 187 F Supp 545, 557–8 (ED Pa 1960), aff’d per curiam, 365 US 567 (1961) (concluding that a tie was justified for a limited time in a new industry to assure effective functioning of complex equipment); Mozart Co v Mercedes-Benz of N Am, Inc 833 F2d 1342, 1348–51 (9th Cir 1987) (upholding verdict for defendant on the ground that the tie may have been found to be the least expensive and most effective means of policing quality); Dehydrating Process Co v AO Smith Corp 292 F2d 653, 655–7 (1st Cir 1961) (affirming a judgment of a district court that directed a verdict in favour of the defendant on the ground that a tie was necessary to ensure utility of two products when separate sales led to malfunctions and widespread customer dissatisfaction). 33 See US Department of Justice and Federal Trade Commission, Antitrust Guidelines for the Licensing of Intellectual Property (2017) § 5.3–5.4. 34 US Philips Corp v Int’l Trade Comm’n 424 F3d 1179, 1193 (Fed Cir 2005). 35 Eg, Foremost v Eastman Kodak Co 703 F2d 534 (9th Cir 1983); Response of Carolina Inc v Leasco Response, Inc 537 F2d 1307 (5th Cir 1976). 36 Eg, Microsoft (n 15); Caldera, Inc v Microsoft Corp 72 F Supp 1295 (D Utah 1999).

Tying and bundling  189 ‘pervasively innovative character of platform software markets’ such as the market for PC operating systems and remanded the case to the district court for assessment of the tying claim under the rule of reason.37 It is unclear whether and how frequently other courts will reject application of the per se test on similar grounds when assessing new forms of tying or tying in other unfamiliar contexts.38 Tying arrangements that are not condemned under the per se rule can, in principle, be found to be unlawful under the rule of reason. Under this test, tying is unlawful if it is found to have injured competition in the tied product market. Such a finding ordinarily requires the plaintiff to define a tied product market and to prove that the harm to competition in that market outweighs any procompetitive justification for the tying arrangement. Few tying arrangements have been found under the rule of reason to violate Section 1 of the Sherman Act, perhaps because the evolution of the per se rule applicable to tying has resulted in its differing mostly in degree, rather than in kind, from the rule of reason.39 Tying arrangements have been found to violate Section 2 when they contribute to the creation or maintenance of monopoly power in either the tying or the tied product market and cannot be justified by procompetitive benefits.40 In those cases, however, the analysis was not affected by characterizing the arrangements as tying or by any special considerations applicable to tying. It is far from clear that tying can be unlawful when there is no exclusion of competitors in the tied product market. The Court in Jefferson Parish said that forcing a purchaser to buy a product ‘he would not have otherwise bought even from another seller in the tied-product market’ cannot injure competition ‘because no portion of the market which would otherwise have been available to other sellers has been foreclosed’, and it reiterated that it had condemned tying arrangements only when they foreclosed a substantial volume of commerce.41 It seems clear, therefore, that the quasi per se rule applies only when competitors in the tied product market are foreclosed. The Microsoft court held that tying can be unlawful under the rule of reason if it harms competition in the tying product market, even if it does not create market power in the tied product market; but the conduct in that case did exclude defendant’s competitors in the tied product market.42 As explained above, tying can be harmful in some circumstances in which there is no such foreclosure, and some commentators have argued that such harms should suffice for finding a tying arrangement to be unlawful.43 Professor Elhauge argues that, because the per se rule condemns tying even absent proof of harm to competition in the tied product market, it makes sense only if the law is intended to encompass the price discrimination effects discussed above, which Elhauge calls the ‘power effects’.44 No court has embraced this argument, and

Microsoft (n 15) 93. The Supreme Court has cautioned in general that ‘[i]t is only after considerable experience with certain business relationships that courts classify them as per se violations’. Broadcast Music, Inc v Columbia Broadcasting System, Inc 441 US 1, 9–10 (1979). 39 Suture Express, Inc v Owens & Minor Distrib, Inc 851 F3d 1029, 1038 (10th Cir 2017). 40 Microsoft (n 15) 41 Jefferson Parish (n 23) 16. 42 Microsoft (n 15). 43 Nicholas Economides, ‘Tying, Bundling and Loyalty/Requirement Rebates’ in Einer Elhauge (ed), Research Handbook on the Economics of Antitrust Law (Edward Elgar Publishing 2012); Elhauge (n 19). 44 Elhauge (n 19) 421. 37 38

190  Research handbook on abuse of dominance and monopolization a recent appellate court decision held that bundling of cable television channels into mandatory packages that increased prices and reduced consumer choice was not unlawful because it did not exclude competing channels.45 2. European Union Articles 101 and 102 of the Treaty on the Functioning of the European Union (TFEU) provide the authority for the European Union’s competition law. Article 101 prohibits agreements whose ‘object or effect’ is to restrict or distort competition ‘and in particular those’ that ‘make the conclusion of contracts subject to supplementary obligations which, by their nature or according to commercial usage, have no connection to the subject of such contracts’. This prohibition ‘may’, however, ‘be declared inapplicable’ in the case of agreements that are ‘indispensable’ to improving the ‘production or distribution of goods’ or to promoting ‘technical or economic progress’, provided that consumers get ‘a fair share’ of the benefits and the agreements do not ‘eliminat[e] competition in respect of a substantial part of the products in question’. Article 102 prohibits ‘abuse’ by firms in a ‘dominant position’, including ‘in particular’ making contracts subject to the kinds of ‘supplementary obligations’ described in Article 101. Article 102 does not by its terms include a defence when such conduct provides economic benefits such as set forth in Article 101. The language in Articles 101 and 102 is much more specific than that of their rough analogues, Sections 1 and 2 of the Sherman Act. Neither the courts nor the European Commission (‘Commission’), however, has been closely constrained by the specific language at least in part because, as the Court of Justice of the European Union emphasized, the particular agreements or abuses described by those Articles do not exhaust the universe of prohibited conduct.46 a) Elements of the tying offence There have been relatively few decisions of the Commission in tying cases and even fewer court decisions. Still, it seems clear from both the decided cases and the Commission’s Guidance documents that there are four requirements in order for tying to violate EU law.47 First, the tying and tied products must be separate products when the infringement begins.48 As in the US, the separate product issue turns largely on whether there is demand for the tied product separate from the tying product. The tying and tied products will be regarded as distinct if ‘a substantial number of customers would purchase or would have purchased the tying product without also buying the tied product from the same supplier, thereby allowing stand-alone production for both the tying and the tied product’. This might be proven by evidence of customer purchases when tying was not required, evidence of undertakings offering the tied product without the tying product, or evidence of undertakings without market power offering the tying product without the tied product.49 There can be separate products even if the products are complements that are used together or as part of an overall system.50 The issue in Brantley v NBC Universal, Inc (2012) 675 F3d 1192 (9th Cir). Case C-333/94P Tetra Pak International SA v Commission [1996] ECR I-05951. 47 Case T-201/04 Microsoft Corp v Commission [2007] ECR II-3601, paras 850–69; Article 102 Guidance (n 2) paras 47–58, 62. 48 Microsoft (n 47) paras 842, 912–44; Tetra Pak (n 46). 49 Microsoft (n 47) paras 919, 927; European Commission Guidelines on Vertical Restraints (2010/C 130/01) ‘Guidelines on Vertical Restraints’, para 51. 50 Case T-30/89 Hilti AG v Commission [1991] ECR II-1439, paras 102–19. 45 46

Tying and bundling  191 a tying case is not whether a buyer of the tying product needs a product sold in the tied product market, but whether it needs the particular tied product sold by the defendant.51 The Commission Guidance on Article 102 does not suggest any direct investigation of the economic efficiency or profitability of producing or selling the tying or tied products separately. While efficiency might be inferred from observed commercial conduct, it is not clear how the Commission would treat evidence offered to show that separate sales are not profitable and cannot be expected to persist even in the absence of tying. Second, the undertaking must have a dominant position in the tying product market.52 This requirement suggests a difference between US law and EU law. Most tying cases in the US arise under Section 1 of the Sherman Act and require, among other things, that the defendant has market power in the tying product market. In the EU, almost all of the few tying cases have arisen under Article 102, which has the more demanding requirement of a dominant position.53 Under EU law, a dominant position requires substantial market power but less than that required for monopoly power under US law.54 Third, the customer does not have the option of buying or leasing the tying product without the tied product.55 The Guidance defines tying as ‘situations where customers that purchase one product (the tying product) are required also to purchase another product from the dominant undertaking (the tied product)’ and states that tying can be ‘technical or contractual’.56 The buyer will be deemed not to have had the option of obtaining the tying product separately if the seller refuses to sell the tying product unless the customer also buys the tied product, offers only a bundle that includes both the tying and the tied product, or offers the tying product separately only on commercial terms that are unrealistic as a practical matter.57

51 The court acknowledged in the Microsoft case, as had the court in the US Microsoft case (n 15), that in evolving industries items that appear to be or had been separate products might not actually be separate products (para 913), but it found separate products in the case before it in part because Microsoft sold the tied product separately (paras 941–2). 52 Microsoft (n 47) para 870. 53 The Commission takes the position that tying can violate Article 101 in the narrow circumstance ‘where it results in a single branding type of obligation for the tied product’. Guidelines on Vertical Restraints (n 49) para 214. This infringement is subject to a block exemption if the share of the defendant in both the tying and tied markets is less than 30 per cent; ibid para 218. 54 An undertaking is regarded as having a dominant position if it is able to act ‘to an appreciable extent independently of its competitors, customers and ultimately consumers’; Case 27/76 United Brands and United Brands Continental v Commission [1978] ECR 207. While a very low market share is generally regarded as inconsistent with a finding of dominance, a share of 40 per cent has been found to be sufficient, at least when combined with other factors. Case 219/99 British Airways v Commission [2003] ECR II-5925. In the US, by contrast, monopoly power is often said to require a market share of 70 per cent or more, eg, Exxon Corporation v Berwick Bay Real Estates Partners 748 F2d 937, 940 (5th Cir 1984), and no case has found a firm with a share of less than 50 per cent to have monopoly power. 55 Microsoft (n 47) paras 960–61; Tetra Pak (n 46) para 31; Article 102 Guidance (n 2) para 48. 56 ibid para 48. 57 Case T 65-98 Van den Bergh Foods Ltd v Commission [2003] ECR II-4653 (free freezer only if used only for VdB ice cream). In Novo Nordisk, the Commission objected to a refusal by a dominant supplier of insulin to assume any responsibility or to honour any guarantee as regards its products when they were used together with compatible components (such as needles) of other manufacturers. Twenty-sixth Report on Competition Policy (1996) para 62.

192  Research handbook on abuse of dominance and monopolization Fourth, by contrast to the per se rule in US tying law, tying is unlawful in the EU only if it is found to have harmed competition or to be capable of doing so.58 The Guidance and the cases seem to contemplate a particular kind of harm to competition. The Guidance explains that, while tying is a common practice that often provides better products or lower costs, when used by an undertaking that is dominant in the tying product market, tying can harm consumers by foreclosing competing products from the tied product market59 and can ‘protect the dominant undertaking’s position in the tying market … indirectly through foreclosing the tied market’.60 Anticompetitive foreclosure is more likely when the tying strategy is long-lasting;61 when it includes multiple products and the undertaking has a dominant position in more than one of them;62 if foreclosure of sales in the tied product market reduces the number of producers, and thus leads to higher prices, in that market;63 if it prevents substitution of the tied product for the tying product, or enables evasion of price regulation, and thus leads to higher prices in the tying product market;64 or if, by reducing the number of suppliers of a complementary tied product, it makes entry into the tying product market more difficult.65 The Guidance makes no reference to harms from price discrimination or to other harms unrelated to foreclosure in the tied product market. By contrast to exclusive dealing and requirements contracts, tying does not preclude the buyer from acquiring products from competitors of the defendant. Instead, it forecloses competition by reducing demand for substitutes for the tied product. A tying arrangement can thus be unlawful even if it does not require the buyer to use the tied products it purchases.66 Tying can foreclose competition even if inclusion of the tied product does not increase the price paid by the buyer, especially if the tying increases competitors’ costs by denying them preferred modes of distribution or imposes costs on customers when they try to use substitutes for the tied product.67 Proof that the tying arrangement actually harmed competition in the tied product market is not necessary; a violation can be established by proof of ‘a reasonable likelihood that [the tie] … would lead to a lessening of competition so that the maintenance of an effective competition structure would not be ensured in the foreseeable future’.68 In assessing the likelihood of

Microsoft (n 47) paras 1031–90; Article 102 Guidance (n 2) paras 19–22, 52–58. ibid para 49. 60 ibid para 52. 61 ibid para 53. 62 ibid para 54. 63 ibid para 55. 64 ibid paras 56, 57; see also Guidelines on Vertical Restraints (n 49) para 217 (‘Tying may also directly lead to prices that are above the competitive level’). Although these reference to higher prices might suggest that increased prices provide a sufficient basis for finding harm necessary for a violation of EU law, they are more likely intended simply as a description of the possible consequences of tying. The Article 102 Guidance states explicitly that anticompetitive foreclosure is required for a violation, and the Guidelines on Vertical Restraints are limited to situations in which tying results in a single branding obligation. 65 Article 102 Guidance (n 2) para 58. 66 Microsoft (n 47) para 970; Hilti (n 50). 67 ibid paras 969, 1041–5. 68 Microsoft (n 47) para 867; Case T-203-01 Michelin v Commission [2003] ECR II-4071, para 239 (‘it is “sufficient to show” that the conduct “tends to restrict competition … or is capable of having that effect”’). 58 59

Tying and bundling  193 harm to competition, the courts will consider the undertaking’s overall course of conduct and will not necessarily require that the tying is sufficient by itself to harm competition.69 b) Possible justifications As noted above, Article 101 explicitly provides that certain otherwise unlawful tying arrangements might be justified if they are necessary to realize certain efficiencies that benefit consumers. Most tying cases, however, arise under Article 102. The Commission’s Article 102 Guidance states that it will consider as possible justifications for tying arrangements reductions in production or distribution costs that would benefit customers and reductions in transaction costs incurred by customers.70 It is especially likely to find such benefits when separate products are combined ‘into a new, single product’.71 But, as with any justifications under Article 102, the tying must be ‘indispensable’ to realizing the efficiencies, and the efficiencies must outweigh the anticompetitive effects on consumers so that no ‘net harm’ to consumers is likely.72 Thus, for example, while tying might reduce transaction costs, that benefit is unlikely to justify refusing to sell the tying product separately for those customers that prefer that option as long as there is sufficient demand for separate sales to satisfy the separate product test. Similarly, while development of a standardized platform for complementary products can provide real benefits, the complementary products may not be imposed unilaterally by tying but must instead be a consequence of uncoerced customer choice.73 Otherwise unlawful tying cannot be justified by efficiencies if the tying eliminates ‘effective competition, by removing all or most existing sources of actual or potential competition’, and conduct that creates or maintains a position ‘approaching that of a monopoly’ normally cannot be justified on efficiency grounds.74 This limitation echoes language in Article 101 and seems to embody both a concern about the magnitude of harm from monopolies and an expectation that new or increased competition is less likely in markets dominated by a single firm than in markets that already contain evidence of viable competitors. By contrast, US cases focus on the magnitude of the benefits compared to the harm caused by the allegedly anticompetitive conduct and do not appear to apply different standards depending on the degree of concentration or market power in the relevant markets. The difference might reflect the analytical context in which the issue arises. The issue in the EU is whether problematic conduct can be justified by its market effects; in the US, the efficiency defence goes to the threshold question whether the conduct should be regarded as anticompetitive or as legitimate competition on the merits. The courts have not upheld a tying arrangement found to be unlawful by the Commission on the basis of such efficiencies. The court in the Microsoft case rejected a number of proposed Tetra Pak International v Commission (Tetra Pak II) [1994] ECR II-755, para 135. Article 102 Guidance (n 2) para 62. 71 DG Competition, Discussion Paper on the Application of Article 82 of the Treaty to Exclusionary Abuses (2005), para 205. 72 Article 102 Guidance (n 2) paras 28, 30. In cases decided after the Guidance was published, the Court of Justice has articulated a different and possibly less demanding requirement that the conduct at issue is ‘necessary’ to obtain the efficiencies. Eg, Case C-209/10 Post Danmark A/S v Konkurrencerådet [2012] ECR I-172, para 42. 73 See Microsoft (n 47) paras 1151–7. 74 Article 102 Guidance (n 2) para 30. 69

70

194  Research handbook on abuse of dominance and monopolization justifications, most of which were based on claimed technical efficiencies.75 Some commentators have understood the decision to reflect a broad hostility to efficiency defences in tying cases, but that case was unusual. The case involved a finding that Microsoft had acted illegally, not just in bundling the Windows operating system with the Windows Media Player, but also in taking steps to prevent purchasers who wished to do so from replacing the Windows Media Player with a substitute provided by a competitor.76 The Guidance also states that tying might be justified if it is ‘objectively necessary for health or safety reasons’, but they note that health and safety regulation are normally responsibilities of the government. Undertakings are not expected to act on their own judgement about such matters.77 In Hilti, the dominant undertaking sought to justify its tie on the ground that using only Hilti-brand nails with its nail guns and cartridge strips was consistent with public safety and that third party consumables were unsafe. The General Court rejected Hilti’s argument, noting that Hilti did not approach the competent authorities in the UK to express its safety concerns. Moreover, the court concluded that it was not the task of an undertaking ‘in a dominant position to take steps on its own initiative to eliminate products which, rightly or wrongly, it regards as dangerous or at least as inferior in quality to its own products’.78

III.

MIXED BUNDLING

Mixed bundling entails selling in a package two or more different products that are also available for sale separately. The total price of the products in the package is usually less than the sum of the prices of the individual products in the package when they are sold separately.79 The difference or discount could take the form of a lump sum payment or a reduced per unit price.80 Common examples would be burger, fries and beverage, and a season ticket to sport or concert events. If the price of one of the products when sold separately is not commercially realistic, that product is deemed not to be available separately and, instead, to be the tying product in a tying arrangement.81 For antitrust purposes, mixed bundling differs from tying in that its potentially harmful effects are a result, not of buyers’ desire to acquire a tying product in the bundle, but rather of the potentially distorting effects of the prices demanded by the seller.

Microsoft (n 47) paras 1147–67. ibid paras 1149, 1158–60, 1164–5. 77 ibid para 29, citing Case T-30/89 Hilti (n 50) paras 118–19; Tetra Pak II (n 69) paras 83, 84 and 138. 78 Hilti  (n 50) para 118. The Court of Justice upheld this judgment in Case C-53/92 Hilti v Commission [1994] ECR I-667. 79 Antitrust Modernization Commission, ‘Report and Recommendations’ (2007) 94; David S Evans and Michael Salinger, ‘Why Do Firms Bundle and Tie? Evidence from Competitive Markets and Implications for Tying Law’ (2005) 22 Yale Journal on Regulation 37, 54. 80 Chiara Fumagalli, Massimo Motta and Claudio Calcagno, Exclusionary Practices: The Economics of Monopolization and Abuse of Dominance (Cambridge University Press 2018) 389. 81 See n 27. 75 76

Tying and bundling  195 A.

The Benefits and Harms of Mixed Bundling

By inducing increased purchases of one or all of the bundled products, mixed bundling can aid the realization of cost savings from scale or scope economies.82 Scope economies can give a multi-product firm a cost advantage over firms that cannot offer all of the products in the package.83 Mixed bundling packages that are offered to all customers on the same terms are more likely to reflect the seller’s cost savings than mixed bundling that is customized to meet the needs of individual buyers. If the firm has market power in two or more complementary products, offering all of them in a package with a reduced price can solve a Cournot complements problem. Mixed bundling can also increase output by enabling price discrimination between buyers that want all of the products in the bundle and those that want only one, and by enabling metering-type price discrimination among buyers of one product in the bundle based on how much they use other, complementary products in the bundle.84 And mixed bundling can be used as temporary promotional pricing; a firm might, for example, introduce new product X by offering customers of established product Y a discount if they also buy X, or it might offer one of the products for free when the buyer purchases the other.85 If bundling achieves the intended result of increased sales of one or more of the bundled products, it could tend to exclude competitors in the markets in which those products are sold. This is most likely if the firm offering the bundles has substantial market power in one of the products in the bundle which, together with the discount for the bundle, induces increased sales of other products in the bundle. In that event, mixed bundling could potentially have the same harms as the harms from the exclusion of competitors by tying or forced bundling. These harms are more likely if the defendant has a large share of sales in the affected product market and entry into that market is difficult. Mixed bundling takes the form of a price discount offered to those that purchase all the products in the bundle. If the bundling does result in lower prices, it will increase consumer welfare and, unless the price is below cost, total welfare.86 However, if the price for one or more of the bundled products when sold separately is higher than the price that would be charged if no bundle were offered, the bundle might not actually reduce price. In that event, it could reduce consumer welfare and, unless it increases market-wide output, total welfare.87 B.

Governing Legal Principles

1. United States Early cases found bundled discounts to be unlawful on the ground that they restricted consumer choice or disadvantaged competitors unable to offer substitutes for all of the bundled Eg, David S Evans and A Jorge Padilla, ‘Designing Antitrust Rules for Assessing Unilateral Practices: A NeoChicago Approach’ (2005) 72 University of Chicago Law Review 73, 90. 83 If the firm has market power in all of the products in the bundle, mixed bundling can deter entry of competitors offering only one of the products. Barry Nalebuff, ‘Bundling as an Entry Barrier’ (2004) 119 Quarterly Journal of Economics 159. 84 Fumagalli, Motta and Calcagno (n 80) 358–62. 85 Antitrust Modernization Commission (n 79) 95. 86 ibid. 87 Eg, ibid 100; Barry Nalebuff, ’Exclusionary Bundling’ (2005) 50 Antitrust Bulletin 321, 339–40. 82

196  Research handbook on abuse of dominance and monopolization products.88 In the controversial LePage’s case, the court held that a firm with monopoly power in transparent tape violated Section 2 of the Sherman Act when it offered a discount on its branded tape product and a variety of other products in response to competition from providers of private label tape. The court noted that the discounts were customized for individual customers and that the discounts foreclosed rivals without broad product lines from portions of the transparent tape market.89 Discounts that are customized for individual customers are unlikely to reflect economies of scope or other efficiencies, but even efficient bundled discounts can exclude rivals that do not have a broad product line and the economies that breadth can provide. These cases have been criticized largely on the ground that they did not set forth criteria for distinguishing benign bundling from anticompetitive bundling.90 More recent cases have used more sophisticated approaches that appear to be motivated at least in part by the fact that bundled discounts are widely used and by a desire for clear standards that will not deter procompetitive price discounts. The most important of these cases is Cascade Health Solutions v PeaceHealth.91 The court in that case treated bundled discounts offered by firms that have or are on the verge of obtaining monopoly power as a form of potentially predatory pricing and adopted what it called a ‘discount attribution’ test to determine when a bundled discount violates Section 2 of the Sherman Act.92 Under this test, a bundled discount is unlawful if, after allocating all discounts and rebates for the entire bundle of products to the competitive product, the resulting constructed price of the competitive product is below the defendant’s average variable cost for that product.93 The court rejected the alternative of an aggregate discount rule that would compare the aggregate price of the bundle to its aggregate costs on the ground that that test would enable a multi-product defendant to exclude a single-product firm that was more efficient in producing the competitive product.94 The court also declined to follow an earlier decision that would find a bundled discount to be unlawful if the plaintiff is at least as efficient as the defendant but cannot compete because of the bundled discount;95 the court rejected that alternative on the ground that that standard would leave the defendant uncertain about permissible pricing, because it could not know the costs of its competitors, and vulnerable to multiple suits from different competitors with different costs. And the court rejected the proposal, of a government-created advisory board, that the plaintiff should be required to prove, not just below-cost prices, but also a likelihood of recoupment of the resulting losses and that the bundled discounts have had or are likely to have an adverse effect on competition.96 These Eg, Advance Business Systems & Supply Co v SCM Corp 415 F2d 55 (4th Cir 1969). LePage’s, Inc v 3M 324 F3d 141 (3d Cir 2003). 90 Eg, Antitrust Modernization Commission (n 79); Daniel Rubinfeld, ‘3M’s Bundled Rebates: An Economic Perspective’ (2005) 72 University of Chicago Law Review 243. 91 502 F3d 895 (9th Cir 2007). 92 ibid 907, 916. 93 ibid 920. 94 If the prices on products A and B for which the defendant has market power were well above cost, the defendant could reduce prices on those products, so that the incremental price for including competitive product C in the bundle was very small, and still keep the aggregate price for the bundle above the aggregate cost. In that way, the defendant could exclude a more efficient producer of the competitive product, even if that producer could profitably sell its product at a price below the defendant’s incremental cost for the competitive product. 95 Ortho Diagnostic Sys, Inc v Abbott Labs, Inc 920 F Supp 455 (SDNY 1996). 96 Antitrust Modernization Commission (n 79) 99–100. 88

89

Tying and bundling  197 proposed additional requirements were intended to guard against the possibility that common forms of benign bundling might be found to entail below-cost pricing for one of the products in the bundle if all of the discounts in the bundle are allocated to that product. A later decision by the same court made clear that the discount attribution test applies only when at least one competitor of the defendant does not sell all of the products in the bundle.97 Most but not all subsequent cases have adopted the discount attribution test,98 but it is not without critics. Depending on the circumstances, it can be both overinclusive and underinclusive. It can be overinclusive if used to prohibit a benign package, such as a burger and fries, without regard to whether it threatens harm to competition in the market as a whole.99 This risk is diminished if the test is limited to situations in which competitors are unable to offer all of the products in the bundle. The discount attribution test can be underinclusive if the standalone prices for one or more products in the bundle are higher than they would be in the absence of the bundled option. In that case, the bundle might not actually provide a discount and could thus exclude competitors without providing a benefit to consumers.100 This risk might, however, be of little practical significance. If there is no evidence of substantial market acceptance of the higher standalone prices, the bundle might be deemed to be a tying arrangement. Also, higher standalone prices increase the discount that is attributed to the competitive product and thus the likelihood that its constructed price will be below cost. The discount attribution test can also be underinclusive for the same reasons that price-cost tests applied to predatory pricing can be underinclusive. In the first place, price-cost tests protect only competition from firms that are at least as efficient as the defendant when the bundled discounts are offered. They therefore do not protect competition from firms that, while less efficient, might nevertheless provide an important competitive constraint on the defendant.101 Nor do they protect competition from firms that, with more experience and scale, could become more efficient in the future. Moreover, price-cost tests do not prohibit above-cost prices even when they entail a profit sacrifice when compared to the prices that would be charged in the absence of the bundle and might thus make no business sense except as a scheme to exclude rivals.102 Above-cost prices can exclude less efficient rivals that provide a competitive constraint on more efficient firms and thereby increase consumer welfare. European Union 2. As in the US, the Commission and the courts in the EU have had less experience with mixed bundling than with tying. Early cases were quite formalistic and condemned mixed bundling arrangements whose form was capable of injuring competition. In Hoffmann-La Roche, for example, the court found discounts conditioned on purchases of most or all of the customer’s

Aerotec Int’l, Inc v Honeywell Int’l, Inc 836 F3d 1171, 1186–7 (9th Cir 2016). Eg, Collins Inkjet Corp v Eastman Kodak Co 781 F3d 264 (6th Cir 2015). 99 Nalebuff (n 87) 342; Dennis W Carlton, Patrick Greenlee and Michael Waldman, ‘Assessing the Competitive Effects of Multiproduct Pricing’ (2008) 53 Antitrust Bulletin 587. 100 See n 87. 101 Steven C Salop, ‘The Raising Rivals’ Cost Foreclosure Paradigm, Conditional Pricing Practices, and the Flawed Incremental Price-Cost Test’ (2017) 81 Antitrust Law Journal 371, 407. 102 Brooke Group Ltd v Brown & Williamson Tobacco Corp 509 US 209, 223 (1993). 97 98

198  Research handbook on abuse of dominance and monopolization requirements of a bundle of products to be an infringement without analysing their effects.103 The Commission’s case in Coca-Cola, which involved rebates to customers purchasing a wide range of products, is to the same effect.104 The Commission’s Article 102 Guidance adopts a more effects-based approach. It states that a multi-product discount ‘may be anticompetitive … if it is so large that equally efficient competitors offering only some of the components cannot compete against the discounted bundle’.105 Thus, the Guidance explains, [i]f the incremental price that customers pay for each of the dominant undertaking’s products in the bundle remains above the [long-run average incremental cost] of the dominant undertaking from including that product in the bundle, the Commission will normally not intervene since an equally efficient competitor with only one product should in principle be able to compete profitably against the bundle.106

This in principle is the same as the discount attribution test used in the US. And, as with the Aerotec case in the US,107 if competitors are able to offer all of the products in the bundle, the relevant question for the Commission is whether ‘the price of the bundle as a whole is predatory’.108 The Commission will also consider efficiencies for which the bundling is necessary if the benefits are sufficiently shared with customers. The efficiencies can include savings in production or distribution costs and reduced transaction costs for customers.109 The more fact-based and effects-based approach suggested in the Guidance has also been embraced by the courts. The Intel case is instructive. The case involved loyalty discounts, which the Court of Justice characterized as ‘discounts conditional on the customer’s obtaining all or most of its requirements’ from the dominant firm.110 The Court rejected the General Court’s and Commission’s condemnation of the discounts ‘by object’ on the ground that they were a type of conduct that is capable of harming competition. Instead, the court ruled, the conduct should be assessed by asking whether, in light of the specific facts of the case, the particular conduct at issue is capable of injuring competition. Perhaps more important, the court repeatedly stated that the standard under Article 102 is whether the conduct at issue is capable of excluding competitors that are as efficient as the dominant firm.111 These and similar cases reflect a willingness by the courts to give weight to discount attribution tests that identify when equally efficient competitors might be excluded by bundling practices and a reluctance by the courts to condemn conduct, like mixed bundling, that is capable of providing efficiency benefits without proof that the conduct is capable of harming competition under the specific circumstances.

105 106 107 108 109 110 111 103 104

Case 85/76 Hoffmann-La Roche v Commission [1979] ECR 461, paras 100, 111. Case COMP/A.39.116/B2 Coca-Cola [2005] OJ L253/21. Article 102 Guidance (n 2) para 59. ibid para 60. See n 97. Article 102 Guidance (n 2) para 61. ibid para 62. Case C-413/14 P Intel Corp v Commission ECLI:EU:C:2017:632, para 137. ibid paras 133, 134, 136, 139, 140; see also Post Danmark (n 72) para 21.

Tying and bundling  199

IV. CONCLUSION Tying and bundling can be beneficial or harmful, depending on the circumstances. The law in both the US and the EU thus prohibits tying and bundling only in specific circumstances. Tying is generally unlawful in both the US and the EU if the firm has market power in the tying product and the tying forecloses from competitors a substantial amount of sales in the tied product market. Bundling is likely to be unlawful in both the US and the EU if the firm has market power in one or more products in the bundle and the price of any one product in the bundle, after allocating the entire discount for the bundle to that product, is below the firm’s average variable cost for producing and selling that product.

11. Rebates Viktoria H.S.E. Robertson

I.

REBATES AS A COMPETITION LAW ISSUE

When a company grants a rebate to its business partner, the latter pays a lower price for the goods or services received. As competition law generally strives for lower prices on the market, it can appear counter-intuitive to penalize rebates granted by companies with market power. However, on closer inspection the rationale that underpins the view that rebates can sometimes be anticompetitive emerges: where a rebate is only granted based on conditions that induce loyalty or (near-)exclusivity on the purchaser’s side, this can foreclose the supplier’s competitors from access to the market, possibly even leading to the competitor exiting the market, which ultimately benefits the dominant supplier.1 While loyalty discounts can help to align the supplier’s and the purchaser’s interests, they can also prevent entry or expansion, lead to market exits or involve below-cost pricing.2 The question, then, turns on how to distinguish between procompetitive rebates that competitors can and should compete on, and anticompetitive rebates that may constitute vehicles to foreclose competitors from a market. This chapter addresses this question by discussing the case law on rebates from the EU and the US. It broadly distinguishes between quantity or volume rebates, which are granted based on the volume of purchases, and fidelity or loyalty rebates, that is, ‘discount schemes [that] allow sellers to offer buyers a better price conditional on the buyer demonstrating loyalty in the purchases they make’.3 This can, for instance, be a market-share discount that is conditioned on the purchaser obtaining a certain percentage of its needs in a product category from the supplier (‘share of need rebate’). In the case of loyalty rebates, the discount is usually retroactive and applies to ‘all units of a single product’.4 Loyalty rebates may be framed as pricing abuses where the price is at the centre of the rebate scheme, or as non-pricing abuses where the exclusionary element of a rebate scheme constitutes its central element. Rebate schemes may also be part of other pricing abuses, such

1 For an analysis of the foreclosure effects of loyalty rebates, see also Alexander Morell, Andreas Glockner and Emanuel V Towfigh, ‘Sticky Rebates: Loyalty Rebates Impede Rational Switching of Consumers’ (2015) 11 Journal of Competition Law and Economics 431. 2 Christian Ahlborn and David Bailey, ‘Discounts, Rebates and Selective Pricing by Dominant Firms: A Trans-Atlantic Comparison’ (2006) 2 European Competition Journal 101, 108–10; Roger D Blair and Thomas Knight, ‘Bundled Discounts, Loyalty Discounts and Antitrust Policy’ (2020) 16 Rutgers Business Law Review 123, 130. 3 OECD, ‘Fidelity Rebates: Background Note by the Secretariat’ (11 March 2016), DAF/ COMP(2016) 5, 4. 4 US Department of Justice, ‘Competition and Monopoly: Single-Firm Conduct under Section 2 of the Sherman Act’ (Section 2 Report) (September 2008) 106, https://​www​.justice​.gov/​sites/​default/​files/​ atr/​legacy/​2009/​05/​11/​236681​.pdf, accessed 18 June 2021.

200

Rebates  201 as excessive pricing, predatory pricing or discriminatory pricing, while bundling rebates have a tying element and can therefore be likened to tying.5 In the following, the case law and soft law on rebates under EU competition law is discussed, with a view to tracing the evolving approach to rebates in the EU. Thereafter, the chapter analyses how rebates are treated under US antitrust law, noting the need for the US Supreme Court to weigh in on the divided circuits. The chapter concludes with a brief outlook.

II.

REBATES UNDER EU COMPETITION LAW

Article 102 of the Treaty on the Functioning of the European Union (TFEU) prohibits the abuse of a dominant position on the relevant market. One of the types of behaviour that Article 102 TFEU lists as abusive and thus anticompetitive is the application of dissimilar conditions to equivalent transactions.6 This prohibition can apply to individualized rebate schemes that are not applied across the board. Article 102 TFEU also contains a general prohibition on abusive behaviour that can be relied upon to outlaw commercial behaviour that excludes competitors from the market or exploits customers. In early cases, both the European Commission (Commission) and the EU Courts – that is, the General Court (GC) and the Court of Justice (CoJ) – adopted a form-based approach towards anticompetitive rebates, giving considerable attention to the form of the rebate scheme rather than, perhaps, its substance. This characterization of the early cases, however, is deprecating and not very helpful to understand what forms of rebates are seen as anticompetitive and why.7 This is even more so as these cases did indeed take a large number of economic circumstances into account, rather than just concentrating on the form of a certain rebate scheme. In more recent cases, a more effects-based type of analysis of rebates has been applied by the Commission and accepted by the CoJ. However, there is no definite legal benchmark to establish whether a rebate scheme indeed contravenes Article 102 TFEU. In particular, the importance of the as-efficient-competitor test (AEC test) that the Commission advanced in its soft law instruments requires further inquiry. This evolution in the case law as well as in the Commission’s soft law is discussed below with a view to setting out the status quo of anticompetitive rebates in Europe. A.

Quantity and Loyalty Rebates in the Early European Case Law

The settled case law of the CoJ makes abundantly clear that the Court is not concerned about volume-based, incremental quantity rebates.8 Pure quantity rebates are therefore not regarded as an abuse of a dominant position as long as they fulfil certain conditions. The Court has highlighted that ‘a quantity rebate [should be] exclusively linked with the volume of purchases

These types of abuses are dealt with in Chapters 5–10 in this Handbook. Article 102(c) TFEU. 7 Richard Whish, ‘Intel v Commission: Keep Calm and Carry On!’ (2015) 6 Journal of European Competition Law & Practice 1, 1. Whish also points out that the EU Courts’ form-based rules on rebates are not per se rules, as the dominant company can always submit an objective justification; ibid 2. 8 Case 85/76 Hoffmann-La Roche v Commission ECLI:EU:C:1979:36, para 90; Case 322/81 Michelin v Commission (Michelin I) ECLI:EU:C:1983:313, para 72. 5

6

202  Research handbook on abuse of dominance and monopolization from the producer concerned’9 and should be ‘granted in respect of each individual order, thus corresponding to the cost savings made by the supplier’.10 Quantity rebates also need to be granted on a non-discriminatory basis in order to not fall afoul of Article 102(c) TFEU.11 While the Court’s early case law thus clearly did not regard quantity rebates as anticompetitive, it has always been wary of loyalty rebates that lead to exclusive purchasing on the purchaser’s side. In a string of leading cases, the Court has regarded loyalty-inducing rebates of various kinds as an abuse of a dominant position under Article 102 TFEU. In the Court’s view as set out in Hoffmann-La Roche, share of need rebates – that is, market-share discounts that are only awarded if the purchaser obtains a certain percentage of its requirements from the dominant supplier – are anticompetitive because they are ‘designed to deprive the purchaser of or restrict his possible choices of sources of supply and to deny other producers access to the market’.12 It is this foreclosure effect that the GC has also taken issue with.13 The foreclosure effect does not have to have materialized; it is sufficient that the rebate ‘is capable of having that effect’.14 Even rebate schemes that do not contain an exclusivity criterion can be found to be anticompetitive where they induce the purchaser to remain loyal to the dominant supplier. In Michelin I, the Court listed a number of factors that need to be taken into account when making such an assessment, including the criteria and rules for granting the rebate, whether the rebate has a tendency to remove or restrict the purchaser’s freedom to choose its sources of supply and to exclude competitors from the market, and whether the rebate scheme applies dissimilar conditions to equivalent transactions or reinforces the supplier’s dominant position.15 In Irish Sugar, the GC found that a rebate scheme that specifically targets certain purchasers based on their location constitutes an abuse of a dominant position, as do fidelity rebates based on exclusivity obligations and selectively granted target rebates.16 In British Airways, the Court emphasized that in order to carry out the competitive assessment of a rebate that is neither a pure quantity rebate nor a fidelity rebate under the Hoffmann-La Roche rule, one needs to ascertain whether the rebate can have an exclusionary effect on competitors, as well as whether the rebate makes it more difficult or impossible for purchasers to choose between various sources of supply or commercial partners.17 A rebate scheme that, on its face, is volume based, can also be designed in a way to be caught out under Article 102 TFEU: as the Court emphasized in Portugal v Commission, it lies in the nature of quantity rebates that purchasers of larger volumes of a product benefit from lower Joined Cases 40 to 48, 50, 54 to 56, 111, 113 and 114/73 Suiker Unie v Commission ECLI:EU:C:1975:174, para 518. 10 Case C‑23/14 Post Danmark v Konkurrencerådet (Post Danmark II) ECLI:EU:C:2015:651, para 28. 11 Suiker Unie (n 9) paras 522–3; Georg-Klaus de Bronett, ‘§ 22 Das Verbot des Missbrauchs marktbeherrschender Stellungen im EU-Kartellrecht (Art. 102 AEUV)’ in Gerhard Wiedemann (ed), Handbuch des Kartellrechts (3rd edn, CH Beck 2016), Section 22, para 95. 12 Hoffmann-La Roche (n 8) paras 89–90 (direct quote at para 90). 13 Case T-203/01 Michelin v Commission (Michelin II) ECLI: EU:T:2003:250, para 57. 14 ibid para 239 (direct quote); Case T-219/99 British Airways v Commission ECLI:EU:T:2003:343, para 293. 15 Michelin I (n 8) para 73. 16 Case T-228/97 Irish Sugar v Commission ECLI:EU:T:1999:246, paras 173–225; upheld on appeal in Case C-497/99 P Irish Sugar v Commission ECLI:EU:C:2001:393. 17 Case C-95/04 P British Airways v Commission ECLI:EU:C:2007:166, para 68. 9

Rebates  203 average unit prices, and this in itself does not make a quantity rebate scheme discriminatory. However, where a quantity rebate scheme is structured in a way that awards some purchasers ‘an economic advantage which is not justified by the volume of business they bring or by any economies of scale they allow the supplier to make compared with their competitors’, it will be seen as applying dissimilar conditions to equivalent transactions as prohibited under Article 102(c) TFEU.18 Where a quantity rebate scheme has loyalty-inducing effects, for instance because of long reference periods or because it is calculated based on total turnover rather than by an individual tranche, the GC has held that it may not be regarded as a purely volume-based rebate scheme that does not infringe Article 102 TFEU.19 While the EU Courts’ early case law on rebates has been called formalistic,20 it is actually informed by the insight that loyalty-inducing rebates have a tendency to foreclose the market for competing suppliers by keeping purchasers loyal to the dominant supplier. It is for this reason that the CoJ developed the Hoffmann-La Roche rule on fidelity rebates. And as the capability to restrict competition is sufficient to come within the prohibition of Article 102 TFEU, loyalty rebate schemes will be caught quite quickly. Also in Hoffmann-La Roche, however, the Court accepted that, in exceptional circumstances, loyalty rebates may be objectively justified.21 Therefore, the case law on rebates cannot be said to establish a ‘by object’ prohibition for loyalty discounts, although the approach to fidelity rebates under the early case law comes close to such a rule. B.

Conditional Rebates in the European Commission’s Guidance Paper

Over a decade ago, the Commission entered a more effects-based era for abuse of dominance cases with its Guidance Paper on enforcement priorities under Article 102 TFEU.22 In a section dedicated to conditional rebates, the Guidance Paper sets out how the Commission intends to carry out the competitive assessment of ‘rebates granted to customers to reward them for a particular form of purchasing behaviour’, that is, loyalty or fidelity rebates.23 It sets out an AEC test that is meant to assess whether a rebate scheme forecloses the market for an equally efficient competitor. It transpired, however, that officials within the Commission were not entirely aligned on the question of the applicability and workability of the AEC test.24 The Guidance Paper sets the scene by differentiating between a contestable and a non-contestable share of demand for each customer. A customer’s non-contestable portion of demand refers to that amount which a purchaser would in any case buy from the dominant

Case C-163/99 Portugal v Commission ECLI:EU:C:2001:189, paras 51–2 (direct quote at para

18

52).

Michelin II (n 13) paras 67–95, especially para 95. See Sofia Oliveira Pais, ‘Os Descontos de Exclusividade Numa Encruzilhada’ in Maria Lúcia Amaral (ed), Estudos em Homenagem ao Conselheiro Presidente Rui Moura Ramos (vol I, Almedina 2016) 1221. 21 Hoffmann-La Roche (n 8) para 90. 22 European Commission, ‘Guidance on the Commission’s Enforcement Priorities in Applying Article 82 of the EC Treaty to Abusive Exclusionary Conduct by Dominant Undertakings’ [2009] OJ C45/7 (Guidance Paper). 23 ibid paras 37–46 (direct quote at para 37). 24 See on this Nicolas Petit, ‘Intel, Leveraging Rebates and the Goals of Article 102 TFEU’ (2015) 11 European Competition Journal 26, 27–8 (containing further references). 19 20

204  Research handbook on abuse of dominance and monopolization company. The contestable portion of demand relates to that amount which a purchaser may buy from another company. The Commission’s concern is that a dominant company may use conditional rebates in order to leverage purchases from the non-contestable portion of demand to the contestable one.25 For price-based exclusionary conduct more generally, the Commission declares that it will only intervene where a dominant company’s behaviour may dampen competition from competitors that are as efficient as the dominant company.26 Only in exceptional circumstances will it consider competition from less efficient competitors to merit intervention.27 This means that, for conditional rebates, the Commission assesses ‘whether the rebate system is capable of hindering expansion or entry even by competitors that are equally efficient by making it more difficult for them to supply part of the requirements of individual customers’.28 The pricing test inherent in the AEC test, together with the concept of anticompetitive foreclosure that is advanced in the Guidance Paper, has been lauded as a ‘defensible standard’ that allows for an assessment whether a certain behaviour is exclusionary and thus anticompetitive, or not.29 On the other hand, the introduction of an AEC test also raises the standard of proof to which the Commission needs to show a rebate scheme’s anticompetitiveness, as compared to the early settled case law.30 While the AEC test emphasizes the nature of loyalty rebates as a pricing abuse, the competition issue for loyalty rebates is not only the (low, and thus perhaps predatory) price but the rebates’ exclusionary character.31 In order to carry out the AEC test, the Commission defines a ‘relevant range’, that is, that part of demand which a purchaser could satisfy through purchases from the dominant company’s competitor. Then, the Commission calculates the effective price that the competitor would have to offer the purchaser in order to make up for the rebate that the purchaser loses by changing her purchasing behaviour.32 Having calculated this effective price, the Commission assesses whether it is above or below the dominant company’s long-run average incremental cost (LRAIC) or its average avoidable cost (AAC).33 While an effective price above LRAIC is thought to allow an as-efficient-competitor to compete profitably, an effective price below AAC is understood to be generally able to foreclose as-efficient-competitors. An effective price between those two benchmarks needs to be assessed in depth by the Commission.34 Further factors may also be taken into account by the Commission, such as whether a rebate is individualized rather than standardized.35 Retroactive rebates – like the ones at issue in

European Commission, Guidance Paper (n 22) para 39. ibid para 23. 27 ibid para 24. 28 ibid para 41. 29 Giorgio Monti, ‘Article 82 EC: What Future for the Effects-Based Approach?’ (2010) 1 Journal of European Competition Law and Practice 2, 4. 30 ibid 9. 31 Ahmet Fatih Özkan, ‘The Intel Judgment: The Commission Threw the First Stone but the EU Courts Will Throw the Last’ (2015) 11 European Competition Journal 69, 75; Whish (n 7) 2; Case T-286/09 Intel v Commission ECLI:EU:T:2014:547, para 99. 32 European Commission, Guidance Paper (n 22) para 41. 33 On these cost benchmarks, see ibid para 26. 34 ibid paras 43–4. 35 ibid para 45. 25 26

Rebates  205 Michelin I – may be particularly problematic due to the probability of foreclosure effects.36 Efficiencies will also be considered as possible justifications.37 As the Court has emphasized, the Guidance Paper merely sets out the Commission’s enforcement priorities as regards Article 102 TFEU, and is neither binding on national competition authorities nor on national courts or the EU Courts when applying that provision.38 Instead, only the Commission itself is bound by its Guidance Paper while also remaining bound by the EU Courts’ case law on Article 102 TFEU.39 Therefore, it is not surprising that in one of the Commission’s first cases applying the new framework, its Intel decision,40 it relied both on settled case law and on its AEC test to make its case for the anticompetitive nature of Intel’s rebate scheme. Within the Commission, some considered this case a successful marriage of established case law with an effects-based approach.41 C.

Effects-Based Analysis of Rebates in More Recent European Case Law

In a number of cases, the Commission tested its new, more effects-based approach to the competitive assessment of rebate schemes, with varying degrees of success before the EU Courts. 1. Loyalty rebates in Tomra and Velux In March 2006, the Commission adopted a decision on loyalty rebates in the case of Tomra, which concerned reverse vending machines.42 At that point in time, the Commission had already published a Discussion Paper on exclusionary conduct that preceded the Guidance Paper discussed above, but not yet the Guidance Paper in its final version.43 In its infringement decision, the Commission found that Tomra had engaged in a commercial strategy including a number of anticompetitive practices in order to maintain its dominant position on the market for reverse vending machines. In particular, it found that, in several Member States, Tomra had entered into exclusivity agreements and de facto exclusivity agreements containing individualized quantity commitments and retroactive rebate schemes, that is, rebates with an exclusivity-enhancing effect.44 The Commission found that these constituted an abuse of a dominant position under the Court’s settled case law, including cases such as Hoffmann-La Roche, Portugal v Commission and Michelin I.45 Going beyond establishing the required legal standard as set out in the case law, the Commission ‘completed its analysis in this case by considering the likely effects of Tomra’s practices’46 – that is, it applied its effects-based analysis

ibid para 40. ibid para 46. 38 Post Danmark II (n 10) para 52. 39 Opinion of AG Kokott in Case C-109/10 P Solvay v Commission ECLI:EU:C:2011:256, para 21. 40 Commission Decision (COMP/C-3/37.990) Intel [2009] OJ C227/13. 41 Nicholas Banasevic and Per Hellström, ‘When the Chips are Down: Some Reflections on the European Commission’s Intel Decision’ (2010) 1 Journal of European Competition Law & Practice 301. 42 Commission Decision (COMP/E-1/38.113) Prokent-Tomra [2008] OJ C219/11. 43 See European Commission, ‘DG Competition Discussion Paper on the Application of Article 82 of the Treaty to Exclusionary Abuses’ (2005) (on file with author). 44 Prokent-Tomra (n 42) paras 97, 109, 123 ff, 131 ff. 45 ibid paras 297 ff. 46 ibid para 332. 36 37

206  Research handbook on abuse of dominance and monopolization to the case, setting out how the rebate scheme at issue together with other strategies, such as the acquisition of competitors, allowed Tomra to maintain a stable market share.47 When Tomra appealed the Commission’s decision to the GC, the latter re-emphasized that settled case law did not require the showing of a rebate scheme’s actual anticompetitive effects – the mere capability to produce such effects was sufficient.48 The Commission’s additional effects analysis was not held to be the basis for its infringement decision, and the GC therefore considered that it did not need to review it.49 The capability to have an anticompetitive effect was therefore confirmed as the legal benchmark by which rebates are to be assessed, while any effects-based analysis going beyond this benchmark was considered irrelevant to the outcome of the case. Overall, this was a win for the traditional assessment of loyalty rebates.50 Upon further appeal to the CoJ, the capability to restrict competition was again confirmed as the applicable legal benchmark for assessing the anticompetitiveness of rebates.51 No particular quantitative economic test was required in order to show a rebate scheme’s anticompetitiveness.52 Following two complaints by a competitor, the first in 2007 and the second in 2012, the Commission investigated a rebate scheme operated by window manufacturer Velux. It found that the rebates were not loyalty-inducing or individualized, and therefore did not have foreclosure effects. As such, the Commission did not further pursue these complaints.53 From what is known about the Commission’s analysis, it appears that the Commission conceptually applied the AEC test, as it believed that competitors with a similar cost structure would be able to compete with Velux’s post-rebate prices, and the rebate scheme therefore could not have exclusionary effects.54 2. The AEC test in Post Danmark II and the General Court’s Intel judgment The introduction of the AEC test by the Guidance Paper of 2009 did not only lead to a change in assessment at the Commission, but also had repercussions on national courts that apply Article 102 TFEU, as they started to wonder whether such a test would now become part of the assessment of rebates. In the preliminary reference of Post Danmark II (2015), the referring court inquired about the importance of the AEC test, as outlined in the Guidance Paper, for assessing rebates under Article 102 TFEU. In its preliminary ruling, the CoJ replied that while this test may constitute a useful analysis within the context of Article 102 TFEU, it is merely ‘one tool amongst others’ and not, as such, required to meet the legal standard to prove an anticompetitive rebate under the abundant case law.55 The Court therefore continued its

ibid paras 331–46. Case T‑155/06 Tomra v Commission ECLI:EU:T:2010:370, para 289. 49 ibid paras 286, 290. 50 See also Gianluca Faella, ‘The Antitrust Assessment of Loyalty Discounts and Rebates’ (2008) 4 Journal of Competition Law & Economics 375, 396. 51 Case C-549/10 P Tomra v Commission ECLI:EU:C:2012:221, paras 68–71. 52 Viktoria HSE Robertson, ‘Rebates under EU Competition Law after the 2017 Intel Judgment: The Good, the Bad and the Ugly’ (2018) 2 Market and Competition Law Review 15. 53 Commission investigation (AT.39451) Velux; Commission decision of 14 June 2018 (AT.40026) Velux. On the first Velux case, see Svend Albaek and Adina Claici, ‘The Velux Case: An In-Depth Look at Rebates and More’ (2009/2) Competition Policy Newsletter 44. 54 See the account provided by Albaek and Claici (n 53). 55 Post Danmark II (n 10) paras 57–8, 61 (direct quote). 47 48

Rebates  207 rebates line of case law and refused to generally require an AEC test.56 This was in line with the GC’s stance as expressed in its Intel judgment only a year earlier,57 which had not regarded any ‘specific economic tool’ as a necessary requirement for proving an anticompetitive rebate scheme either.58 In that judgment, the GC had also attempted to systematize the Court’s case law on rebates by setting out three categories of rebates that required different degrees of attention during an antitrust assessment.59 This was followed by the CoJ in Post Danmark II, albeit in somewhat less explicit terms, when it set out the following three categories of rebates. The first category are pure quantity-based rebates, which are not regarded as anticompetitive if certain conditions are fulfilled; particularly if they are not designed in a way to induce customer loyalty. The second category are loyalty rebates, which are typically seen as an abuse of a dominant position, thus amounting to somewhat of a ‘by object’ prohibition – albeit with the possibility of an objective justification. The third category are rebates that require a more in-depth analysis in order to ascertain whether they are capable of restricting competition.60 For this third type of rebate, it must be shown that the rebate is at least capable of having an anticompetitive effect. The dominant supplier can objectively justify its rebate scheme by showing that any anticompetitive effects are counterbalanced or even outweighed by efficiencies.61 This categorization of rebates is not uncontroversial,62 particularly as it may lead to a quasi-per se prohibition of certain types of loyalty rebates that could be unwarranted in individual cases. In its most recent case, the CoJ neither confirmed nor rejected this three-prong categorization.63 3. The Court of Justice’s Intel judgment: a ‘clarified’ legal framework In its Intel decision of 2009, the Commission relied on the CoJ’s settled case law on rebates as well as on AEC tests in relation to each of Intel’s customers in order to find that Intel’s rebate schemes for its central processing unit chips represented an abuse of a dominant position.64 While the Commission emphasized that the Guidance Paper did not apply to the Intel case, as

Björn Lundqvist, ‘Post Danmark II, now Concluded by the ECJ: Clarification of the Rebate Abuse, but How Do We Marry Post Danmark I with Post Danmark II?’ (2015) 11 European Competition Journal 557. 57 Miroslava Marinova, ‘Should the Rejection of the “As Efficient Competitor” Test in the Intel and Post Danmark II Judgments Lead to Dismissal of the Effect-Based Approach?’ (2016) 12 European Competition Journal 387, 392. 58 Paul Nihoul, ‘The Ruling of the General Court in Intel: Towards the End of an Effect-Based Approach in European Competition Law?’ (2014) 5 Journal of European Competition Law & Practice 521, 523. 59 Intel (n 31) paras 74–8. 60 Post Danmark II (n 10) paras 27–9. 61 ibid paras 47–9. 62 See, for example, Damien Geradin, ‘The Opinion of AG Wahl in Intel: Bringing Coherence and Wisdom into the CJEU’s Pricing Abuses Case-Law’ (2016) TILEC Discussion Paper 2016-034, 5; Julie Clarke, ‘The Opinion of AG Wahl in the Intel Rebates Case: A Triumph of Substance over Form?’ (2017) 40 World Competition 241, 255–8. 63 Mark Friend, ‘Loyalty Rebates and Abuse of Dominance’ (2018) 77 Cambridge Law Journal 25, 27. 64 Intel (n 40) paras 920–21, 923, 925, 1002–640. 56

208  Research handbook on abuse of dominance and monopolization it was published after the Commission had confronted Intel with a statement of objections, it also believed its decision to be consistent with the Guidance Paper’s approach.65 Upon appeal to the GC, the Commission decision was upheld. On exclusivity rebates, the GC found that these were ‘by their very nature capable of restricting competition’, and that therefore their actual effects did not need to be scrutinized in a lengthy test.66 This judgment, which some criticized as being overly formalistic,67 was further appealed to the CoJ. In his Opinion, Advocate General Wahl urged the Court to clarify that a rebate’s capability to foreclose must go beyond the ‘merely … hypothetical or theoretically possible’.68 He also attached importance to the fact that the Commission did, at length, engage in an AEC test. As it had done so, the GC could not simply ignore it.69 The Advocate General’s Opinion was welcomed as an important step towards more effects-based arguments under EU competition law.70 Against this background, the CoJ’s Intel judgment, rendered in September 2017, was seen as something of a litmus test for the applicability of the AEC test in the field of loyalty rebates. In a rather brief section on the antitrust assessment of rebates, the Court set out to consolidate its earlier case law on rebates and gave it a more effects-based spin without abandoning its settled views on loyalty rebates. It started out by underlining that Article 102 TFEU does not protect less efficient competitors from exiting the market.71 But based on a dominant undertaking’s special responsibility to maintain competition in the market, ‘Article 102 TFEU prohibits a dominant undertaking from … adopting pricing practices that have an exclusionary effect on competitors considered to be as efficient as it is itself’.72 The Court did not refer to the three-prong categorization of rebates that it had endorsed in Post Danmark II, instead focusing on loyalty rebates alone. It then ‘clarified’ its Hoffmann-La Roche line of case law as follows.73 If a dominant company provides evidence that its rebates were not capable of restricting competition, the Commission needs to take into account a number of factors in order to assess whether this is so: the company’s market position, the market covered by the rebates, the specific circumstances surrounding the rebates, and the presence of a strategy to exclude as-efficient-competitors.74 The Commission can only carry out an analysis on objective justifications for a rebate scheme once it has assessed ‘the intrinsic capacity of that practice to foreclose competitors which are at least as efficient as the dominant undertaking’.75 And, importantly: ‘[i]f, in a decision finding a rebate scheme abusive, the Commission carries out such an analysis, the GC must examine all of the applicant’s arguments seeking to call into question the validity of the Commission’s findings concerning the foreclosure capability

65 ibid para 916. The AEC analysis was carried out ‘on top’ of the legal test as required under the case law; see Banasevic and Hellström (n 41) 304. It remains questionable to what extent settled case law can slowly be overturned by adding non-required analyses. 66 Intel (n 31) paras 85 (direct quote), 103. 67 Damien Geradin, ‘Loyalty Rebates after Intel: Time for the European Court of Justice to Overrule Hoffman-La Roche [sic!]’ (2015) 11 Journal of Competition Law & Economics 579, 580. 68 Opinion of AG Wahl in Case C-413/14 P Intel v Commission ECLI:EU:C:2016:788, para 114. 69 ibid paras 168–70. 70 Geradin (n 62). 71 Case C-413/14 P Intel v Commission ECLI:EU:C:2017:632, para 133. 72 ibid paras 135, 136 (direct quote). 73 ibid para 138. 74 ibid paras 138–9. 75 ibid para 140.

Rebates  209 of the rebate concerned’.76 The Court noted that, in its decision, the Commission had carried out an AEC test at length, despite holding that this was not required under the case law. As the test represented a cornerstone of the Commission’s analysis, the GC also had to review it upon appeal.77 As the GC had not done so, the CoJ set aside the GC’s judgment and referred the case back to the latter to review the rebate scheme’s capability to restrict competition in this light.78 The GC has yet to decide on that case.79 The Intel judgment was both welcomed as an important step towards a less formalistic and more effects-based approach towards rebates and criticized for failing to provide more clarity as regards the legal test for rebates under EU competition law.80 It attempted to find a middle way between the Court’s settled case law that it has relied upon for many decades and allowing more economics-based approaches to find their place. While no new case on rebates has reached the CoJ since Intel, the Commission has focused on loyalty rebates and other loyalty-inducing schemes in recent cases. 4. Qualcomm and Google Android: applying the ‘clarified’ legal framework In its Qualcomm decision of 2018, the Commission for the first time applied the CoJ’s ‘clarified’ or consolidated legal framework for assessing loyalty rebates.81 It emphasized that, as a matter of principle, ‘[e]xclusivity rebates or exclusivity payments are … presumed to constitute an abuse of a dominant position’.82 However, a dominant player can try to rebut this presumption by showing that its rebates were not even capable of producing anticompetitive effects. For this, the Commission needs to analyse a rebate scheme’s ‘intrinsic capacity’ to exclude an as-efficient-competitor from the market.83 In the case at hand, the Commission concluded that Qualcomm’s exclusivity rebates reduced Apple’s incentives to switch to another supplier of LTE (Long Term Evolution) chipsets.84 While the Commission did not carry out its own AEC test, the AEC test presented by Qualcomm relied, in the Commission’s view, on a series of wrong assumptions,85 highlighting that, while conceptually useful, the correct application of this economic test in practice is far from straightforward and can be used as the battleground for a proxy fight, as a replacement for the ‘by object’ abusive nature of loyalty rebates. The case is currently on appeal before the GC.86 It remains to be seen whether it will further clarify the place of the AEC test in EU competition law.

ibid para 141. ibid paras 142–4. 78 ibid paras 148–50. 79 Case T-286/09 RENV Intel v Commission (case pending). 80 Friend (n 63) 25. 81 Massimiliano Kadar, ‘Article 102 and Exclusivity Rebates in a Post-Intel World: Lessons from the Qualcomm and Google Android Cases’ (2019) 10 Journal of European Competition Law & Practice 439, 444–5. 82 Commission Decision (AT.40220) Qualcomm (Exclusivity payments) [2018] OJ C269/25, para 382. While decided on 24 January 2018, the decision was only published on 8 June 2020. 83 ibid para 285. 84 ibid paras 412 ff. 85 In particular, see ibid paras 489–95. 86 Case T-235/18 Qualcomm v Commission (case pending). 76 77

210  Research handbook on abuse of dominance and monopolization In a further case that is currently on appeal before the GC, the Commission fined Google for exclusivity clauses in relation to its Google search engine.87 While this case is not directly concerned with rebates, it will provide insights into exclusivity-enhancing commercial strategies that a dominant company engages in – insights that may also come to bear on exclusivity-inducing rebate schemes. Overall, the case law set out above shows that while the conceptual underpinnings of the AEC test may well have been accepted by the EU Courts, the Commission’s effects-based approach is still held in check by the Courts’ more principled approach to exclusivity rebates.88 Also, while the CoJ’s Post Danmark II and Intel judgments appear to approve of the use of an AEC test in order to ascertain the foreclosure effects of rebate schemes, that case law in no way requires that the Commission carry out such a test in its antitrust assessment of rebates.89 As far as the AEC test as such is concerned, it is questionable how well this cost test can assess the anticompetitiveness of a rebate scheme under settled case law – but also beyond.90 The case law on rebates still requires some consolidation in order to clarify the legal framework for exclusivity rebates and, particularly, for ‘third category’ rebates – if such a category is still considered useful by the CoJ. While the Commission’s enforcement priorities can be helpful, it is the CoJ’s guidance that is required – guidance that should be mindful of the practical applicability of the antitrust rules. While this guidance can and should rely on insights from economics, it is also true that – as then-Judge Breyer underlined in the US – ‘[r]ules that seek to embody every economic complexity and qualification may well, through the vagaries of administration, prove counter-productive, undercutting the very economic ends they seek to serve’.91 While EU competition law is still refining its approach to loyalty rebates, companies may want to rely on the Commission’s Guidance Paper to put in place procompetitive rebate schemes.92 Nevertheless, the legal uncertainty surrounding the proper legal test applicable to loyalty rebates and the soft law nature of the Guidance Paper remain a considerable challenge.

III.

REBATES UNDER US ANTITRUST LAW

Section 2 of the Sherman Act prohibits the monopolization of trade as well as attempted monopolization.93 While there are no specific rules on the antitrust treatment of loyalty rebates, US antitrust rules rely on ‘imperfect analogies’:94 existing theories of harm such as

Commission Decision (AT.40099) Google Android [2019] OJ C402/19; on appeal as Case T-604/18 Google and Alphabet v Commission (pending). 88 Similarly, see Ioannis Lianos, Valentine Korah and Paolo Siciliani, Competition Law: Analysis, Cases, & Materials (OUP 2019) 1109. 89 Robertson (n 52) 38. 90 ibid 42–3 (containing further references). 91 Barry Wright v ITT Grinnell 724 F2d 227, 234 (1st Cir 1983). 92 Hans Zenger, ‘Devising Loyalty Rebates that Comply with the As-Efficient-Competitor Test’ (2013) 3 Concurrences 16, 19. 93 15 USC § 2. 94 Fiona M Scott Morton and Zachary Abrahamson, ‘A Unifying Analytical Framework for Loyalty Rebates’ (2016) 81 Antitrust Law Journal 777, 806. See also Sean P Gates, ‘Antitrust by Analogy: Developing Rules for Loyalty Rebates and Bundled Discounts’ (2013) 79 Antitrust Law Journal 99, 102. 87

Rebates  211 exclusive dealing,95 predatory pricing96 or tying97 are used in order to assess rebate schemes under Section 2 of the Sherman Act. It is not settled, however, which of these theories of harm should prevail in the analysis of rebate schemes.98 Where the post-rebate price is below a certain cost measure, predatory pricing may be the applicable theory of harm; where the post-rebate price is above cost but leads to anticompetitive foreclosure, the theory of harm may be more oriented towards exclusive dealing.99 As rebate schemes often contain combinations of elements, a careful analysis is required in order to determine which theory of harm to apply.100 Bundled discounts, that is, ‘the practice of offering discounts or rebates contingent upon a buyer’s purchase of two or more different products’,101 can also be prohibited under the antitrust laws because they can lead to a tying arrangement.102 This latter theory of harm for rebates, however, will not be further explored in this chapter as it pertains to tying rather than to loyalty-inducing rebates.103 The US Supreme Court has not (yet) accepted a case that would allow it to develop a legal test applicable to loyalty-inducing rebates. In fact, in 2013 it did not accept to hear an appeal on the rebate scheme at issue in Meritor (see below).104 In this section, the extant case law on rebate schemes, mainly developed by the federal US Courts of Appeals, will be discussed with a view to distilling some general lessons from these cases. A.

Low Prices as a Procompetitive Feature

As rebates lead to lower prices, and low prices are both beneficial to customers and can intensify competition, the predominant view under US antitrust law is that rebate schemes are not normally anticompetitive. In fact, in Barry Wright v ITT Grinnell, the US Court of Appeals for the First Circuit emphasized that price cuts ‘are normally desirable’ as long as they remain above cost and that such a price cut ‘primarily injures only higher cost competitors’.105 Early on, the idea that less efficient rivals should not be protected by the antitrust laws therefore took a hold in US antitrust law. Similarly, in the rebate case of Virgin Atlantic Airways v British Airways, the US Court of Appeals for the Second Circuit considered that ‘as long as low prices

95 Joshua D Wright, ‘Simple but Wrong or Complex but More Accurate? The Case for an Exclusive Dealing-Based Approach to Evaluating Loyalty Discounts’ (3 June 2013), https://​www​.ftc​.gov/​sites/​ default/​files/​documents/​public​_statements/​simple​-wrong​-or​-complex​-more​-accurate​-case​-exclusive​ -dealing​-based​-approach​-evaluating​-loyalty/​130603bateswhite​.pdf, accessed 16 June 2021. 96 Herbert J Hovenkamp, ‘Discounts and Exclusions’ (2006) 3 Utah Law Review 841; referred to as the ‘Hovenkamp test’ in US Department of Justice, Section 2 Report (n 4) 111. 97 Jarod M Bona, ‘Loyalty Discounts and the FTC’s Lawsuit against Intel’ (2010) 19 Journal of the Antitrust and Unfair Competition Law Section of the State Bar of California 6, 10. 98 See Geradin (n 67) 581. 99 Similarly, see US Department of Justice, Section 2 Report (n 4) 107. 100 Gates (n 94) 119. 101 US Department of Justice, Section 2 Report (n 4) 91. 102 Smith Kline v Eli Lilly 575 F2d 1056 (3rd Cir 1978); Le Page’s v 3M 324 F3d 141 (3rd Cir 2003). 103 On tying, see Chapter 10 of this Handbook. 104 Eaton v ZF Meritor, cert denied, 133 SCt 2025 (2013). 105 Barry Wright (n 91) 235.

212  Research handbook on abuse of dominance and monopolization remain above predatory levels, they neither threaten competition nor give rise to an antitrust injury’.106 This reasoning is in line with the US Supreme Court’s settled case law.107 Quantity discounts that are only based on the volume of purchases and not attached to further conditions do not risk antitrust exposure under US antitrust law, for it has been established that ‘volume discount contracts are legal [b]ecause … they are not exclusive dealings contracts that preclude competition in violation of the Sherman Antitrust Act’.108 Instead, quantity discounts are seen as a procompetitive aspect that the Sherman Act does not intend to prohibit.109 B.

Single-Product Loyalty Discounts in the Department of Justice’s Report

In a 2008 Report on Section 2 of the Sherman Act, the US Department of Justice’s Antitrust Division spelled out – not unlike the European Commission’s Guidance Paper110 that was drafted around the same time – how it believed loyalty discounts should be assessed under the antitrust laws.111 While the Report was withdrawn by the Obama administration in 2009,112 it nevertheless allows for insights into how the Department envisaged assessing loyalty-inducing rebates. As it was so short-lived, however, it shall only briefly be discussed here. The Report acknowledges that single-product loyalty discounts, that is, discounts that are offered ‘on all units of a single product conditioned upon the level of purchases’,113 may be anticompetitive where a company with monopoly power offers them in return for certain exclusivity or where they lead to below-cost pricing.114 Overall, however, the Department regards both the cases litigated up until 2008 and the academic commentary on loyalty discounts to be rather favourable towards discount schemes.115 It sets out how predatory pricing and foreclosure analysis could be applied to loyalty discounts and concludes that the Department would probably apply a predatory pricing analysis in most cases of loyalty discounts.116 On the analysis of foreclosure effects, the Department stresses that ‘an approach requiring courts to determine whether a portion of a market is uncontestable and to quantify that portion, … would be difficult to administer’.117 This directly relates to the European Guidance Paper, which suggests exactly this type of cumbersome analysis.118 While the Report discusses the AEC test in more general sections and considers it an apt test to assess the exclusionary nature of pricing practices,119 the Department – unlike its European

Virgin Atlantic Airways v British Airways 257 F3d 256, 269 (2nd Cir 2001). Atlantic Richfield v USA Petroleum 495 US 328, 340 (1990); Pacific Bell Telephone v Linkline Communications 555 US 438, 451 (2009). 108 Western Parcel Express v United Parcel Service 190 F3d 974, 976 (9th Cir 1999). 109 Fedway Associates v US Treasury 976 F2d 1416, 1418 (DC Cir 1992). 110 European Commission, Guidance Paper (n 22). 111 US Department of Justice, Section 2 Report (n 4) 106–17. 112 US Department of Justice, ‘Justice Department Withdraws Report on Antitrust Monopoly Law’ (11 May 2009), https://​www​.justice​.gov/​opa/​pr/​justice​-department​-withdraws​-report​-antitrust​ -monopoly​-law, accessed 23 June 2021. 113 US Department of Justice, Section 2 Report (n 4) 106. 114 ibid 106–7. 115 ibid 110. 116 ibid 116. 117 ibid 117. 118 On this, see Section II.B. of this chapter. 119 ibid 43–5. 106 107

Rebates  213 counterpart – does not regard it as appropriate for assessing loyalty discounts. Where a plaintiff wants to rely on a theory of foreclosure, the Department urges it to show ‘that the discount forecloses a significant amount of the market and harms competition’.120 Similar as in the Qualcomm case, the ability of a competitor to stay in the market needs to be factored into the analysis. Furthermore, the Department considers that ‘a single-product loyalty discount should be illegal only when (1) it has no procompetitive benefits, or (2) if there are procompetitive benefits, the discount produces harms substantially disproportionate to those benefits’.121 C.

Rebates as Predatory Pricing or Exclusive Dealing

Loyalty rebates may be anticompetitive where they lead to below-cost pricing.122 In order to constitute predatory pricing, a rebate scheme must lead to prices below cost in addition to a prospect that the grantor of the rebate will recoup its sacrificed profit.123 The standard to which a predatory pricing claim based on a rebate scheme needs to be proven is the so-called Brooke standard, established by the US Supreme Court in 1993. There, the Court held that in order for a predatory pricing claim to succeed, the plaintiff must prove that the company holding monopoly power priced below ‘an appropriate measure of … cost’, and that there is a ‘dangerous probability’ that the predatory pricing will allow the company to recoup its losses.124 As US antitrust law is set to protect competition rather than competitors,125 less efficient competitors cannot expect relief under the Sherman Act. Exclusive dealing, as a second theory of harm applicable to rebates, can constitute an antitrust infringement even below the threshold of predatory pricing. As it does not matter whether the exclusivity is expressly contained in an agreement or whether an agreement leads to de facto exclusivity,126 loyalty-inducing rebate schemes can be caught under this theory of harm. Exclusive dealing is deemed anticompetitive where it leads to significant foreclosure. To date, not many rebate cases have been successfully litigated on this antitrust theory of harm.127 The following sub-sections briefly look at the British Airways case as an example of how predatory pricing could not be established for the rebate scheme at issue, and then move towards a number of cases in which the antitrust assessment of rebates as exclusive dealing was addressed under Section 2 of the Sherman Act and also under Section 5 of the Federal Trade Commission (FTC) Act.128 1. The US British Airways case: no predatory pricing The British Airways case was litigated in both the EU and the US. Against the background of this case, it is no overstatement to find that EU and US approaches to loyalty rebates have opposing starting points. In the European case, the Commission and the EU Courts regarded that carrier’s discount scheme for travel agents as abusive because the ‘performance reward ibid 117. ibid. 122 Cascade Health v Peacehealth 515 F3d 883 (9th Cir 2007). 123 Brooke Group v Brown Williamson Tobacco 509 US 209 (1993). 124 ibid; Virgin Atlantic (n 106) 266. 125 Brooke Group (n 123) 224–5. 126 ZF Meritor v Eaton 696 F3d 254, 270 (3rd Cir 2012). 127 See, in particular, Section 4. 128 15 USC § 45. 120 121

214  Research handbook on abuse of dominance and monopolization schemes … were both discriminatory against some of their beneficiaries in relation to others and had as their object and effect … the reward of the loyalty of those agents to BA’.129 In the US case, on the other hand, the US Court of Appeals for the Second Circuit viewed loyalty rebates as ‘kinds of agreements [that] allow firms to reward their most loyal customers [which in turn] promotes competition on the merits’.130 It also emphasized that the antitrust laws were supposed to safeguard ‘competitive conduct, not individual competitors’.131 British Airways’ rebate scheme, which contained performance targets and first-dollar provisions – through which a rebate applies retroactively once the target is reached132 – was not held to constitute an unreasonable restraint of trade (Section 1 of the Sherman Act), as the plaintiff had not shown that the scheme had anticompetitive effects under the rule of reason.133 Under Section 2 of the Sherman Act, the Court then assessed whether the rebate scheme could constitute predatory pricing. Virgin Atlantic argued that British Airways attempted to monopolize the market through ‘predatory foreclosure and the bundling of ticket sales in an attempt to foreclose transatlantic competition [on routes originating at Heathrow airport] by diverting passengers from Virgin and other airlines to itself’.134 Relying on the Brooke standard set out above, the Court found that predatory pricing was not proven based on British Airways’ costs and the possibility of recoupment.135 2. Assessing exclusivity in Concord Boat v Brunswick In Concord Boat v Brunswick, the question arose under what conditions a rebate scheme could constitute anticompetitive exclusive dealing. The case was brought by a number of boatbuilders against Brunswick, the market leader in the manufacturing of stern drive and inboard marine engines for motorboats. As part of its marketing strategy, Brunswick started to offer rebates to boatbuilders and dealers in 1984, including market-share discounts, additional rebates for long-term market-share agreements, and volume discounts. None of the rebate programmes required boatbuilders or dealers to buy engines from Brunswick, or not to buy from competitors.136 In their antitrust lawsuit, the boatbuilders accused Brunswick, as a dominant player, of using its discounts programme to restrain trade (Section 1 of the Sherman Act) and monopolize the market (Section 2 of the Sherman Act), in addition to anticompetitive acquisitions.137 This behaviour, in the plaintiffs’ view, allowed Brunswick to charge higher prices while at the same time excluding competitors from the market.138 They characterized the market-share discounts as effectively imposing a ‘tax’ on boatbuilders and dealers who wanted to buy engines from Brunswick’s competitors.139 While they regarded Brunswick’s

British Airways (n 14) para 299. Upheld on appeal in British Airways (n 18). Virgin Atlantic (n 106) 265. 131 ibid 258. 132 ibid 261. 133 ibid 264–5. 134 ibid 265. 135 ibid 265–72. 136 Concord Boat v Brunswick 207 F3d 1039, 1044 f (8th Cir 2000). 137 Note how in the European Tomra case, a combination of rebate schemes and acquisitions was equally engaged in by the dominant firm; Section II.C.1 above. 138 Concord Boat (n 136) 1045–6. 139 ibid 1046. 129

130

Rebates  215 various discounts as ‘de facto exclusive dealing’, Brunswick argued that they ‘represented pro competitive business conduct’.140 In a jury trial, the jury found for the boatbuilders on all counts of the antitrust claims and awarded them damages of about USD 133m after trebling.141 The District Court did not grant Brunswick’s renewed motion for judgment as a matter of law or its motion for a new trial.142 The US Court of Appeals for the Eighth Circuit reversed on appeal. It held that the discount scheme employed by Brunswick may have constituted de facto exclusive dealing, but not exclusivity contracts as such.143 In some instances, purchasers were only required to obtain 80 per cent of their requirements from Brunswick in order to qualify for the rebate – but chose to obtain 100 per cent from Brunswick. Had Brunswick charged supra-competitive prices, the purchasers would have bought their remaining requirements elsewhere. In addition, the purchasers’ reactions to changes to Brunswick’s discount programmes led to lower market-share requirements being imposed,144 showing that Brunswick would not have got away with a truly anticompetitive rebate scheme. Under Section 1 of the Sherman Act, the relevant factors to assess such de facto exclusive dealing include ‘the extent to which competition has been foreclosed in a substantial share of the relevant market, the duration of any exclusive arrangement, and the height of entry barriers’.145 The Court found that the plaintiffs had neither shown foreclosure of a substantial share of the market nor exclusivity of the contracts.146 Under Section 2 of the Sherman Act, the plaintiffs argued that Brunswick’s rebate scheme formed part of an exclusionary strategy that led to higher prices for them. The Court emphasized ‘the general rule that above cost discounting is not anticompetitive’147 and that cutting prices was ‘the “very essence of competition”’.148 Brunswick’s rebate scheme was not considered to constitute a case of predatory pricing either, as it was not shown that the discounted prices were below cost.149 3. The Intel case and rebates under the FTC Act In parallel to the European Commission’s Intel case, the Federal Trade Commission (FTC) also brought a case on anticompetitive rebates against Intel. While in the EU this case reached the CoJ and is still shaping the antitrust assessment of loyalty rebates, the US case was ultimately settled. FTC v Intel is notable insofar as the FTC relied on Section 5 of the FTC Act to make its case.150 The FTC accused Intel of anticompetitive strategies relating to its microchips that had exclusionary effects on its competitors and impinged on consumer choice as well as on innovation. Its rebate scheme was one aspect of these anticompetitive strategies.151 The

ibid 1054. Concord Boat v Brunswick, 21 F Supp 2d 923, 925 fn 2 (ED Ark 1998). 142 ibid 941. 143 Concord Boat (n 136) 1058. 144 ibid 1056. 145 ibid 1059. 146 ibid 1059–60. 147 ibid 1061. 148 ibid 1062; citing Matsushita v Zenith Radio 475 US 574, 594 (1986). 149 Concord Boat (n 136) 1062. 150 15 USC § 45. 151 FTC, Complaint in the Matter of Intel Corporation (16 December 2009) FTC Docket No 9431, https://​www​.ftc​.gov/​sites/​default/​files/​documents/​cases/​091216intelcmpt​.pdf, accessed 23 June 2021. 140 141

216  Research handbook on abuse of dominance and monopolization case was settled with a consent decree in 2010. Said decree ensured, amongst others, that Intel would no longer ‘condition … benefits to computer makers in exchange for their promise to buy chips from Intel exclusively or to refuse to buy chips from others’.152 The FTC’s Intel case has been accused of ‘confus[ing] the landscape’ of antitrust rules for loyalty rebates, as it was based on Section 5 of the FTC Act, a provision that private plaintiffs cannot invoke.153 While the US Supreme Court has affirmed that all violations of the Sherman Act also violate the FTC Act, the FTC Act has a broader reach than the Sherman Act.154 From a comparative perspective, the Intel cases were seen as a ‘fragile rapprochement’155 of EU and US approaches to loyalty rebates that could have led to the emergence of a more harmonized approach. The settlement, however, meant that the case was not given an opportunity to establish a court precedent. 4. Meritor and Eisai: applying the rule of reason to loyalty discounts In Meritor and Eisai, the US Court of Appeals for the Third Circuit had to decide whether a loyalty rebate scheme could be contrary to the Sherman Act even where post-rebate prices were above cost and the predatory pricing analysis would therefore not lead to the finding of an antitrust infringement. Meritor v Eaton was a case in the heavy-duty truck transmissions market. At issue were long-standing agreements between transmission supplier Eaton and truck manufacturers that contained target rebates, that is, rebates that retroactively applied once a certain sales target had been reached. Meritor, Eaton’s competitor in that market, characterized these rebates as de facto exclusive dealing arrangements.156 Eaton granted a conditional rebate to truck manufacturers if they obtained an individualized percentage of their requirements from Eaton, ranging between 70 and 97.5 per cent. While truck manufacturers were not required to buy from Eaton, Eaton could terminate long-term agreements if these goals were not met and it could require the manufacturer to repay any contractual savings made under the arrangement, which also included some up-front payments. Eaton’s transmissions had to be listed as the ‘standard offering’ in the manufacturers’ catalogues. Sometimes, manufacturers had to remove competitors’ products from their catalogue.157 Truck manufacturers had to grant preferential pricing to Eaton’s transmissions compared to those from competitors. Agreements with manufacturers also contained an English clause, allowing truck manufacturers to buy transmissions from a competitor, if the latter offered a lower price or higher quality, the truck manufacturer notified Eaton of that offer, and Eaton could not match the price or quality.158 According to

FTC, ‘FTC Settles Charges of Anticompetitive Conduct against Intel’ (4 August 2010), https://​ www​.ftc​.gov/​news​-events/​press​-releases/​2010/​08/​ftc​-settles​-charges​-anticompetitive​-conduct​-against​ -intel, accessed 21 June 2021 (direct quote); Intel, FTC File No 061-0247, 9. 153 Bona (n 97) 7. 154 Herbert J Hovenkamp, ‘The Federal Trade Commission and the Sherman Act’ (2010) 62 Florida Law Review 1, 3. 155 Daniel A Crane, ‘Formalism and Functionalism in the Antitrust Treatment of Loyalty Rebates: A Comparative Perspective’ (2016) 81 Antitrust Law Journal 209, 216. 156 ZF Meritor (n 126) 263. 157 ibid 265. 158 ibid 266. 152

Rebates  217 Meritor, these exclusive dealing arrangements meant that it could only access 8 per cent of the relevant market. It exited the market in 2007, just after bringing the antitrust lawsuit.159 The District Court found in favour of Meritor.160 The US Court of Appeals for the Third Circuit reviewed this judgment and highlighted that Meritor’s allegations of anticompetitive conduct could only hold if Eaton’s conduct could either be found to be anticompetitive under the price-cost test of predatory pricing (Brooke standard) or under the rule of reason standard that applied to exclusive dealing. For the latter, it emphasized that, based on Tampa Electric, ‘an exclusive dealing arrangement will be unlawful only if its “probable effect” is to substantially lessen competition in the relevant market’.161 To show this, it would be necessary to demonstrate ‘significant market power by the defendant, substantial foreclosure, contracts of sufficient duration to prevent meaningful competition by rivals, and an analysis of likely or actual anticompetitive effects considered in light of any procompetitive effects’.162 In addition, the Court named further factors to be taken into consideration, namely any coercive behaviour by the monopolist, the possibility to terminate the exclusive dealing agreement, and use of this strategy by competitors.163 Eaton contended that Meritor’s was essentially a predatory pricing claim and that the price-cost test should be applied to it.164 The Appeals Court disagreed, finding instead that as Meritor had complained about a number of aspects in Eaton’s agreements with purchasers, and price was not the ‘predominant mechanism of exclusion’, the case needed to be assessed under the rule of reason rather than based on a price-cost test.165 Exclusive dealing arrangements could also be unlawful where they were only partial and imposing de facto rather than express exclusivity.166 Based on the foreclosure produced by Eaton’s rebate scheme, the agreements’ long-term nature, the additional anticompetitive provisions included in the agreements and an overall balancing of procompetitive and anticompetitive effects, the Court ruled that Eaton’s rebate scheme constituted an unlawful exclusive dealing agreement.167 The US Supreme Court decided not to review the Meritor judgment,168 missing out on an important opportunity to clarify the US approach to rebates. Eisai v Sanofi concerned the market for four anticoagulant drugs in which Sanofi’s product Lovenox had a market share of 81.5 to 92.3 per cent in the relevant timeframe.169 Lovenox had an additional indication for severe heart attacks that other anticoagulants lacked. Eisai, which marketed Fragmin, one of the three other anticoagulant drugs with a market share of about 4.3 per cent to 8.2 per cent, complained that Sanofi engaged in a number of anticompetitive practices: in marketing Lovenox to hospitals, Sanofi offered market-share and volume discounts. Its offering also included an access clause that would limit a hospital’s ability to prioritize

161 162 163 164 165 166 167 168 169 159 160

ibid 267. ZF Meritor v Eaton 769 F Supp 2d 684 (D Del 2011). ZF Meritor (n 126) 268–9; citing Tampa Electric v Nashville Coal 365 US 320, 327–9 (1961). ZF Meritor (n 126) 271 (references to further cases omitted). ibid 272. ibid 273. ibid 277. ibid 282. ibid 286–9. Eaton v ZF Meritor, cert denied, 133 SCt 2025 (2013). Eisai v Sanofi-Aventis US 821 F3d 394, 399 (3rd Cir 2016).

218  Research handbook on abuse of dominance and monopolization other anticoagulant drugs in its formulary, that is, the list of hospital-approved medications. Non-compliance with these terms meant that the hospitals forfeited their discount.170 Upon appeal, the US Court of Appeals for the Third Circuit affirmed the District Court’s judgment in favour of Sanofi.171 It assessed the case under a rule of reason standard for de facto exclusive dealing, with a particular emphasis on establishing whether the arrangement led to foreclosure.172 It found that, although Sanofi’s drug had an additional indication compared to the other three anticoagulant drugs, the case could not be assessed as a bundled rebate case.173 It distinguished Eisai from its finding of an anticompetitive exclusive dealing arrangement in Meritor on the basis that, in Eisai, ‘[u]nlike in … ZF Meritor, Lovenox customers had the ability to switch to competing products. They simply chose not to do so’.174 Therefore, Sanofi’s rebate scheme was not considered to constitute an anticompetitive exclusive dealing arrangement. On the other hand, the Court was not persuaded by Sanofi’s argument that the price-cost test was the appropriate standard in this case, as the plaintiff had complained about more than just low prices.175 These two cases, both litigated before the US Court of Appeals for the Third Circuit within a relatively short period of time, allow us to draw a distinction between the applicability of the predatory pricing test and the more nuanced rule of reason standard for exclusive dealing. Despite the different outcomes in Meritor and Eisai, which can primarily be attributed to the factual evidence available to the courts,176 these cases present important precedent as US antitrust law further develops its approach to loyalty-inducing rebate schemes. Until the US Supreme Court accepts a loyalty rebate case, they will provide valuable guidance in this area of antitrust. As was emphasized in both cases, while ‘prices are unlikely to exclude equally efficient rivals unless they are below-cost, exclusive dealing arrangements can exclude equally efficient … rivals, and thereby harm competition, irrespective of below-cost pricing’.177 5. Theories of harm and real-life counterfactuals in FTC v Qualcomm More recently, in FTC v Qualcomm the District Court for the Northern District of California was asked to adjudicate on the anticompetitiveness of a range of business practices in modem chip markets.178 One of these practices included Qualcomm’s agreements with Apple that meant that ‘Apple received hundreds of millions in incentives from Qualcomm only if Apple purchased substantial volumes of Qualcomm modem chips’.179 Quoting Aerotec International, the Court recalled that, under specific circumstances, exclusivity rebates or market-share discounts can constitute de facto exclusive dealing ‘because they coerce buyers into pur-

ibid 400. Eisai v Sanofi-Aventis US Case No 08-4168 (MLC) (DNJ, 28 March 2014); aff’d in Eisai (n 169). 172 Eisai (n 169) 402–3. 173 ibid 404–7. 174 ibid 407–8. 175 ibid 408. 176 Chiara Fumigalli, Massimo Motta and Claudio Calcagno, Exclusionary Practices: The Economics of Monopolisation and Abuse of Dominance (CUP 2018) 199. 177 ZF Meritor (n 126) 281; referring to US v Dentsply International 399 F3d 181, 191 (3rd Cir 2005); Eisai v Sanofi-Aventis US, Case No 08-4168-MLC, p 55 (DNJ, 28 March 2014). 178 FTC v Qualcomm 411 F Supp 3d 658 (ND Cal 2019). 179 ibid 763. 170 171

Rebates  219 chasing a substantial amount of their needs from the seller’.180 A clawback provision meant that had Apple sold a single handset with a modem from a Qualcomm competitor, it would retroactively lose all its rebates, amounting to over half a million USD.181 The Court qualified Qualcomm’s agreements with Apple as de facto exclusive dealing arrangements. It held that, based on Tampa Electric, these infringed Section 2 of the Sherman Act because they substantially foreclosed competition by preventing other competitors from supplying Apple, stopping competitors to enter the market, preventing Apple to initiate patent litigation, and being long-term agreements.182 Qualcomm’s exclusivity deal with Apple was also part of a broader strategy pursued by Qualcomm, involving this type of exclusivity agreement with purchasers.183 Upon appeal, the US Court of Appeals for the Ninth Circuit unanimously reversed the ruling, arguing that the District Court had gone ‘beyond the scope of the Sherman Act’.184 It found that Qualcomm’s discounts did not have the ‘actual or practical effect of substantially foreclosing competition in the CDMA modem chip market’ because, after signing quasi-exclusive contracts with Qualcomm in 2013, Apple actually switched suppliers and no longer purchased modem chips from Qualcomm, but from Intel.185 This real-life counterfactual, in the eyes of the Court, demonstrated that Qualcomm’s discount scheme did not, in fact, have exclusionary effects. The FTC requested a rehearing en banc, which the Ninth Circuit denied.186 FTC v Qualcomm is reminiscent of the Ninth Circuit’s judgment in Allied Orthopedic, where it had also held that exclusive dealing agreements by sensor producer Tyco, which contained volume discounts and exclusivity bonuses, were lawful because they did not actually coerce purchasers to purchase from Tyco.187 As we move ahead in establishing justiciable standards for loyalty-inducing rebates, it remains an open question whether the factual requirements that the Third Circuit (Meritor, Eisai) and the Ninth Circuit (FTC v Qualcomm, Allied Orthopedic) have established in order to prove the exclusivity-inducing nature of a rebate scheme can be reconciled. It will be interesting to see whether the Department of Justice’s antitrust lawsuit against Google, involving the same exclusivity clauses that are currently before the GC in Europe, will be able to shed more light on such exclusivity-enhancing commercial strategies by dominant players.188 Here, future developments may also have important repercussions on the antitrust rules for loyalty rebates. Under US antitrust law, a loyalty rebate scheme may be assessed either under a theory of predatory pricing or as an exclusive dealing arrangement.189 Despite rebate schemes representing ever-present business practices, ‘the antitrust law governing these discounts is unclear,

180 ibid; referring to Aerotec International v Honeywell International 836 F3d 1171, 1182 (9th Cir 2016). 181 FTC v Qualcomm (n 178) 763. 182 ibid 766–70. 183 ibid 770–72. 184 FTC v Qualcomm No 5:17-cv-00220-LHK, p 9 (9th Cir, 11 August 2020). 185 ibid 54. 186 FTC v Qualcomm No 5:17-cv-00220-LHK, p 54 (9th Cir, 28 October 2020). 187 Allied Orthopedic Appliances v Tyco Health Care Group 592 F3d 991, 996 ff (9th Cir 2010). 188 US v Google No 1:20-cv-03010-APM (DDC, complaint filed 20 October 2020). On the European case, see text accompanying n 87 above. 189 Blair and Knight (n 2) 131.

220  Research handbook on abuse of dominance and monopolization confusing, and constantly changing’.190 In fact, the precise legal standard under either theory of harm is still being developed by the courts and has yet to reach the Supreme Court. Where the non-price nature of a loyalty rebate scheme is at the centre of a commercial strategy and thus of a complaint, there will be parallels drawn to exclusivity schemes that do not necessarily involve traditional rebate schemes. Until such rules become more settled, companies wishing to minimize their antitrust exposure may wish to opt for rebate schemes that the purchaser can readily terminate, prefer short-time over long-time rebate schemes191 and avoid ‘first dollar’ retroactive schemes.

IV.

THE ANTITRUST ASSESSMENT OF REBATES – AN OUTLOOK

Depending on the angle that one takes, rebates can be seen as a discriminatory practice where they are granted to some purchasers but not to others, as an exclusive dealing arrangement where they require a purchaser to obtain (nearly) all of its requirements from a dominant supplier, as a predatory pricing practice where they lead to below-cost prices (and recoupment), or as a tying practice where bundled rebates are at play. This chapter dealt with the first three theories of harm as they apply to single-product loyalty rebates in both the EU and the US. Pure quantity discounts, on the other hand, will not be found to be anticompetitive in either jurisdiction. From a comparative perspective, ‘the degree of divergence between the assessment of rebates under section 2 of the Sherman Act and Article 102 of the Treaty could hardly be larger’.192 Cases litigated on both sides of the Atlantic – such as British Airways, Intel or most recently Qualcomm – highlight these diverging approaches to loyalty rebates in the EU and the US.193 EU theories of harm primarily relate to discrimination and probable exclusionary effects brought about by loyalty-inducing rebates, while US theories of harm mainly rely on qualified exclusionary effects brought about by predation or exclusive dealing. The US approach has been praised as ‘more pragmatic’,194 whereas the EU approach has sometimes been criticized as overly formalistic.195 As Daniel Crane observed, ‘European law is more likely to draw on formal rules to prohibit loyalty rebates, U.S. law is more likely to draw on formal rules to permit them. Europe employs legal formalism, and the United States uses economic formalism’.196 In essence, both approaches are still developing, and this development can be observed in identical cases that are being simultaneously dealt with in both jurisdictions. In the near future, the Google Android case may provide a further impetus to clarify the

Bona (n 97) 6. ibid 19. 192 Hans Zenger, ‘Loyalty Rebates and the Competitive Process’ (2012) 8 Journal of Competition Law & Economics 717, 718–19. Also identifying loyalty rebates and predatory pricing as areas of significant divergence, see Brian A Facey and Dany H Assaf, ‘Monopolization and Abuse of Dominance in Canada, the United States, and the European Union: A Survey’ (2002) 70 Antitrust Law Journal 513, 549–55. 193 Zenger (n 192) 719. 194 Ahlborn and Bailey (n 2) 119. 195 Whish (n 7) 1. 196 Crane (n 155) 210. 190 191

Rebates  221 antitrust assessment of exclusivity agreements, which can also be useful for an assessment of loyalty-inducing rebates. Both the European Commission and the US Department of Justice have, through their respective soft law instruments, actively engaged in devising more effects-based but also workable rules for assessing loyalty rebates. To date, courts on both sides of the Atlantic have struggled to come up with a workable, sustainable legal benchmark for loyalty rebates. New cases are continuously adding to this endeavour. In this development, some degree of convergence can be seen from both sides: the EU is introducing more effects-based elements into its rule-based approach to loyalty rebates, while different theories of harm continue to be tested in the US in order to carve out the applicable legal standard for loyalty rebates. Against the different ideological underpinnings of EU competition and US antitrust law, it is not surprising that considerable challenges will remain in harmonizing the antitrust assessment of rebates.197 However, it is an endeavour that parties subject to antitrust exposure in both jurisdictions will certainly welcome. By tracing both the EU and US approaches to loyalty rebates, this chapter exposed some major differences in the general approach to rebates, which is much more sceptical in the EU, while in the US the procompetitive view of rebates largely prevails. Nevertheless, it is striking how emerging voices from the US are moving towards a recognition of the exclusionary effects that de facto exclusivity rebates can adduce. At this point, both the US Supreme Court and the European CoJ will need to weigh in on the future of the antitrust assessment of loyalty-inducing rebates.

ibid 220.

197

12. The role of intent in abuse of dominance and monopolization Mariateresa Maggiolino

I. INTRODUCTION Scholars in common law countries usually agree that each cause of action, be it criminal or civil, is made up of a set of elements which plaintiffs must prove to enforce their rights and seek judicial redress or relief against the defendant.1 Likewise, scholars of civil law countries are accustomed to state that both civil and criminal prohibitions create a legal world, which is made up of elements that must be proved to hold that the conduct described in those prohibitions has taken place in that legal world.2 Either way, in western legal orders each legal provision is conceptualized as the sum of its constituent elements. Among them, the so-called ‘subjective element’ refers to the different mental states under which defendants may perpetrate the prohibited conduct and firms’ intent may be characterized as one of these states of mind, together with many others, such as recklessness, knowledge, negligence3 or fault.4 However, not every legal prohibition must include a subjective element (or a specific form of it) as one of its building blocks, that is, as one of the requirements that plaintiffs must prove (or defendants disprove) to show (or escape) liability. Absent a clear indication in the letter of the law, ascertaining whether a legal prohibition contains such an element is not always an easy task. Indeed, still today antitrust experts do not know if firms’ intent is a necessary requirement of the offence described in Section 2 of the Sherman Act and Article 102 of the Treaty on the Functioning of the European Union (TFEU). Instead, today it is quite clear that under Section 2 and Article 102 antitrust decision-makers5 use firms’ intent for evidentiary purposes, that is, to understand if the conduct under their scrutiny is likely to harm competition. This practice raises two main concerns. First, one can question – as the international community has often done – if the analysis of firms’ intent may actually reveal the competitive impact of a given behaviour. Second, one may discuss – and this is the original contribution that the chapter offers – if and why antitrust decision-makers should ever be permitted to use firms’ intent to prove the occurrence of a conduct capable

Stroud Francis Charles Milsom, Historical Foundations of the Common Law (OUP 1981). Domenico Rubino, La fattispecie e gli effetti giuridici preliminari (Giuffré 1939). 3 See, eg, Andrew Ashworth and Jeremy Horder, Principles of Criminal Law (OUP 2013) 170–71. 4 See, eg, André Tunc, ‘The Proper Place of Fault in a Modern Law of Tort’ in Konrad Zweigert and Ulrich Drobnig (eds), International Encyclopedia of Comparative Law, Vol. 11: Torts (Martinus Nijhoff Publishers 1983) 63. 5 Different actors apply antitrust law in the US and in the EU. ‘Antitrust decision-makers’ is the imprecise, but quick phrase used here to refer to all of them at once, on the premise that while public and private plaintiffs are the ones who must prove a violation of antitrust prohibitions, judges are those who have to ascertain if the standard of proof has been met in the specific case. 1 2

222

The role of intent in abuse of dominance and monopolization  223 of harming competition under Section 2 and Article 102 (hereinafter, also ‘the behavioural requirement’ or ‘the prohibited conduct’). To answer these questions, the chapter elaborates on the economic-driven meaning that firms’ intent acquires within the antitrust domain6 and on the complex relationship existing between the case in which plaintiffs are required to prove the intent of defendants to show their liability and the case in which plaintiffs do not have to prove defendants’ intent, but can use it to show that defendants engaged in the prohibited conduct. The chapter is structured as follows. Sections II and III provide readers with a brief analysis of the US and EU experiences in dealing with firms’ intent while applying Section 2 and Article 102. Section IV focuses on the notion of firms’ intent in antitrust law, indicates why the economic conceptualization of firms’ intent helps to understand the competitive consequences of a given practice, and points out the interpretive difficulties in determining whether US and EU decision-makers fully endorse this economically driven notion of firms’ intent. Section V explains the usual reasons why legal scholars qualify firms’ intent as one of the building blocks of legal prohibitions, the enforcement consequences of such a choice, and the conditions under which antitrust decision-makers should be permitted to use firms’ intent for evidentiary purposes. Finally, Section VI explains why Section 2 and Article 102 should not give any room to firms’ intent for evidentiary or other purposes and the conditions under which scholars might argue that these two provisions already do so.

II.

THE US EXPERIENCE

Section 2 is silent on firms’ intent. Therefore, the definition of its role in monopolization and attempted monopolization cases has been left to US courts and can be summarized as follows. If the Department of Justice (DOJ) ever decided to prosecute a monopolistic practice criminally – something that is very unlikely today,7 and has rarely occurred in the past8 – it would have to prove the firm’s general intent ‘to maintain its monopoly or to eliminate competition’ in monopolization cases, and its ‘specific intent to achieve or build a monopoly’ in attempted monopolization cases.9 By contrast, if private or public plaintiffs bring a monopolization case in civil court, it is difficult to definitively determine whether or not they must satisfy the intent requirement. In some rulings, the Supreme Court has clearly decided to ignore questions of intent: the Court has not only assumed that no monopolist acts unconsciously,10 but has also held that a dominant firm’s motivation should not be analysed on its own, but understood in light of the circumstances.11 In other rulings, the Court has spoken extensively about defendants’ ‘wrongful For the sake of brevity, the chapter uses ‘antitrust law’ and ‘competition law’ synonymously. US Department of Justice, ‘Antitrust Division Manual’, Ch. III, C.5 (4th edn, 2008), http://​www​ .justice​.gov/​atr/​public/​divisionmanual/​index​.html, accessed 5 May 2021. 8 Kansas City Star Company v United States 240 F2d 643, 664 (8th Cir 1957) (‘Section 2 is not restricted to conspiracies or combinations to monopolize but also makes it a crime for any person to monopolize or to attempt to monopolize any part of interstate or foreign trade or commerce’). 9 Ibid 663–4. 10 American Tobacco Co v United States 328 US 781, 814 (1946) quoting United States v Aluminium Co (Alcoa) 148 F2d 416, 431–2 (2nd Cir 1945). 11 Verizon Communications Inc v Law Offices of Curtis V Trinko, LLP 540 US 398, 409 (2004). 6 7

224  Research handbook on abuse of dominance and monopolization intent to restrain trade’, their ‘anticompetitive purposes’, ‘intended effects’, ‘intended target’ or ‘intended victims’.12 However, despite such loose language,13 the focus was not on the mental state of the defendants, but on the practices at issue and their anticompetitive effects. Similarly, in another ruling, while stating that it considered the defendants’ general intent, the Supreme Court did not examine it, but inferred its existence from both the conduct in question and the relevant market in which it took place.14 Finally, even the Supreme Court that first spoke of the ‘wilful acquisition or maintenance’ of monopoly power as a building block of the monopolization offence neutralized the evil inherent in such an intent by identifying as permissible situations those in which monopoly power derives from a ‘superior product’ or ‘business acumen’, that is, situations in which firms’ actions are certainly no less intentional than monopolistic practices.15 In contrast, in some other rulings, US courts have actually taken into consideration the firms’ general intent to engage in the conduct at stake16 by inferring it from both direct and circumstantial evidence made of corporate documents, managers’ statements, facts and economic data. However, even in these cases, US courts have never analysed the culpability of the involved firms or the blameworthiness of their practices. They have used firms’ general intent to characterize the conduct in question.17 Their purpose has been to disambiguate situations where the analysis of the practice itself has not been deemed sufficient to assess the ability of the conduct to harm competition.18 For example, this happened in a famous predatory innovation case,19 decided by a jury,20 and in some refusal to deal cases where intent was used to tip the scales in favour of the plaintiffs.21 After all, more than a century ago, while explaining the rule of reason analysis that applies to both Section 1 and Section 2 of the Sherman Act Standard Oil Co v United States 221 US 1, 75 (1911); United States v American Tobacco Co 221 US 106, 182 (1911); Brooke Group Ltd v Brown & Williamson Tobacco Corp 509 US 209, 225 (1993); and Weyerhaeuser Co v Ross-Simmons Hardwood Lumber Co Inc 127 S Ct 1069, 1071 (2007). 13 Phillip Areeda and Herbert Hovenkamp, Antitrust Law (Aspen Publishers 2008), Vol III, paras 606b and 651. 14 United States v Griffith 334 US 100, 105 (1948); Eastman Kodak Co v Image Tech 504 US 483 (1992); Poster Exchange, Inc v Nat’l Screen Service Corp 431 F2d 334 (5th Cir 1970); and Six Twenty-Nine Prods, Inc v Rollins Telecasting, Inc 365 F2d 478 (5th Cir 1966). 15 United States v Grinnell Corp 384 US 563, 570–71 (1966). 16 Alcoa (n 10) 148 F2d, at 432. 17 William J Kolasky, Deputy Assistant Attorney General, Antitrust Division, US Department of Justice, ‘What Is Competition?’, Address before the Seminar on Convergence, Sponsored by the Netherlands Ministry of Economic Affairs, The Hague, Netherlands (28 October 2002), http://​www​ .justice​.gov/​atr/​public/​speeches/​200440​.htm​#N​_1​_ , accessed 5 May 2021. 18 The chapter maintains that in Professional Real Estate Investors, Inc v Columbia Pictures Indus, Inc 508 US 49, 60 (1993) the Court did not deem the defendant’s motivation to ‘interfere directly with the business relationship of a competitor’ as a constituent element of Section 2 of the Sherman Act, but of the definition of what a sham litigation is. Not by chance, the Court clarified that proof of a sham does not relieve the plaintiff of the obligation to demonstrate a substantive antitrust violation. 19 CR Bard, Inc v M3 Sys, Inc 157 F3d 1340 (Fed Cir 1998). 20 On this point, see William E Kovacic, ‘The Intellectual DNA of Modern US Competition Law for Dominant Firm Conduct: The Chicago/Harvard Double Helix’ [2007] Columbia Business Law Review 1, 53 (interestingly observing that the Harvard antitrust scholars ‘discouraged reliance on evidence of subjective intent in large part because consideration of intent evidence too often served to mislead juries’). 21 Aspen Skiing Co v Aspen Highlands Skiing Corp 472 US 585, 602 (1985) (holding that intent is ‘merely relevant to the question whether the challenged conduct is fairly characterized as “exclusionary” 12

The role of intent in abuse of dominance and monopolization  225 cases, the Supreme Court argued that ‘knowledge of intent may help the Court to interpret facts and predict consequences’.22 Likewise, in Microsoft, a much more recent case, the Court of Appeals of the District of Columbia has clarified that ‘evidence of the intent behind the conduct of a monopolist is relevant only to the extent it helps us understand the likely effect of the monopolist’s conduct’.23 In all of these cases, then, the US courts did not use corporate intent as a standalone element of the cause of action for monopolization. They used it as an evidentiary element to show that the behavioural requirement of Section 2 had been satisfied. In particular, they did so when they found that the other available evidence was not sufficiently compelling to demonstrate the anticompetitive impact of the practice at stake24 – a circumstance that happened especially when the conduct had legitimate business justifications that indicate its procompetitive effects.25 In attempted monopolization cases brought in civil court, however, the US Supreme Court clearly requires public and private plaintiffs to prove the existence of the defendant’s intent and, in particular, the defendant’s specific intent to prevail by improper means.26 Evidence of the generic intent to do the act or evidence of any specific intent to outperform rivals, ‘achieve monopoly power’, ‘exclude competition’ or ‘control prices’ would not be deemed relevant.27 Some courts – including the Supreme Court – have inferred the existence of such an intent from the improper conduct alone.28 Other courts have collected direct and circumstantial or “anticompetitive”’); United States v Colgate & Co 250 US 300, 307 (1919); and Associated Press v United States 326 US 1, 15 (1945). 22 Chicago Board of Trade v United States 246 US 231, 238 (1918). 23 United States v Microsoft Corp 253 F3d 34, 59 (DC Cir 2001). 24 Lehrman v Gulf Oil Corp 464 F2d 26, 38 n9 (5th Cir 1972) (‘when a firm displays an anti-competitive animus in the operation of an otherwise ambiguous business practice, what the firm seeks to accomplish provides as sure an indicator of the actual effect of the practice on competition as can be found in the shifting sands of antitrust litigation’). 25 General Indus Corp v Hartz Mountain Corp 810 F2d 795, 804 (8th Cir 1987); Illinois ex rel Burris v Panhandle e Pipe Line Co 935 F2d 1469, 1483 (7th Cir 1991); and SMS Sys Maint Serv, Inc v Digital Equip Corp 188 F3d 11 (1st Cir 1999). 26 Alcoa (n 10) 431–2; Spectrum Sprts, Inc v McQuillian 506 US 447, 454–5 (1993); Griffith (n 14) 105; Brunswick Corp v Riegel Textile Corp 752 F2d 261, 264 (7th Cir 1984); Illinois ex rel Burris v Panhandle e Pipe Line Co 935 F2d 1469, 1481 (7th Cir 1991); and Perington Wholesale, Inc v Burger King Corp 631 F2d 1369, 1377 (10th Cir 1979) (‘[T]he gravamen of [the attempt to monopolize] offense is the intent to achieve the unlawful result’). Furthermore, see Howard Hess Dental v Dentsply Intern 602 F3d 237 (3rd Cir 2010) (dismissing the complaint for insufficient allegations of specific intent); Gordon v Lewistown Hospital 423 F3d 184 (3rd Cir 2005) (establishing that the specific intent requirement was not met because the hospital’s purpose was to discipline a surgeon rather than create a monopoly). 27 General Indus Corp v Harts Mountain Corp 810 F2d 795, 801 (8th Cir 1987); American Football League v National Football League 205 F Supp 60, 64–5 (D Md 1962); Swift & Co v United States 196 US 375, 396 (1905); and Kearney & Trecker Corporation v Giddings & Lewis, Inc 452 F2d 579, 599 (7th Cir 1972). 28 Spectrum (n 26) 454–5 (‘[u]nfair or predatory conduct may be sufficient to prove the necessary intent to monopolize’); Tops Markets, Inc v Quality Markets, Inc 142 F3d 90 101 (2d Cir 1998); General Indus Corp v Harts Mountain Corp 810 F2d 795, 802 (8th Cir 1987); and GTE New Media Services Inc v Ameritech Corp 21 F Supp 2d 27, 38 (DDC 1998) (‘[i]n determining whether the plaintiff satisfies the specific intent to monopolize element, a court can infer intent from conduct that has no legitimate business justification but to destroy or damage competition’); and City of Moundridge, Ks v Exxon Mobil Corp 471 F Supp 2d 20 (DDC 2007) (‘Specific intent is sufficiently pled where it is otherwise apparent from the character of the defendants’ actions alleged, such as eliminating a viable means of competition or by channelling customers away from the competition’).

226  Research handbook on abuse of dominance and monopolization evidence to analyse the firm’s intent and, thus, characterize ambiguous practices. Sometimes they have done this in favour of the plaintiff,29 although in predatory pricing cases the Supreme Court has expressly limited this approach, holding that below-cost pricing along with intent to injure a rival are not, by themselves, sufficient to demonstrate the existence of an antitrust violation.30 In other cases, the US courts have used firms’ intent in favour of defendants, especially when the courts themselves had been rather imprecise in defining the relevant market in which the defendant might have been able to acquire monopoly power and the anticompetitive nature of the practice in question.31 In sum, in relation to the US experience we face a clash between what US courts declare and what they do. Often, they have said to include firms’ intent among the building blocks of monopolization and attempted monopolization, as if Section 2 of the Sherman Act governed a regime in which antitrust liability depended on the culpability of the defendants. However, not only have many US courts inferred the existence of corporate intent from the prohibited conduct in both monopolization and attempted monopolization cases, they have also hardly ever used the firms’ intent alone to establish defendants’ liability, regardless of the harmless or harmful nature of the conduct at stake. In other words, unlike in an ordinary personal liability regime, US antitrust judges have never been interested in acting in the name of business ethics and, thus, punishing the evil intent behind some practices. They have never focused on the intent of firms alone as if it were a standalone element of the prohibition; rather, they seemed concerned only with behaviours that involve strengthening a market position without offsetting benefits in terms of efficiency and innovation.32 Thus, there should be room to interpret Section 2 of the Sherman Act as establishing a liability regime in which firms’ intent is not a requirement of the offence, that is, a regime of absolute liability.33 However, giving proof of the complexity of the issue, US judges have also accepted that both plaintiffs and defendants use firms’ intent for evidentiary purposes,34 that is, to prove the anticompetitive or procompetitive effects of the practice at stake. As Section V will show, this practice is hardly compatible with the idea that Section 2 of the Sherman Act does not include firms’ intent as one of the elements of the prohibition.

Union Leader Corp v Newspaper of New England, Inc 180 F Supp 125, 140 (D Mass 1959); Gordon (n 26) 214–15 (holding that evidence of predatory intent is necessary to characterize as anticompetitive a practice for which the defendant had put forward a legitimate business justification, ie an ambiguous practice); and Conwood Co v United States Tobacco Co 290 F3d 768 (6th Cir 2002) (where the amount of evidence of bad intent induced the court to characterize as anticompetitive practices that, as a matter of fact, were only unfair). 30 Brooke Group Ltd v Brown & Williamson Tobacco Corp 509 US 209 (1993). 31 William Iglis & Sons Banking Co v ITT Continental Banking Co 668 F2d 1014, 1027 (9th Cir 1981) and Knutson, Inc v Daily Review 548 F2d 795, 814 (9th Cir 1976). But also consider Chicago Board of Trade v United States 246 US 231, 238 (1918) (admitting that in rule of reason cases intent may be helpful, but ‘not because a good intention will save an otherwise objectionable act’); and Griffith (n 14) 105–6. 32 Areeda and Hovenkamp (n 13) para 805. 33 Absolute liability provisions do not require proof of the subjective element in relation to either of the elements comprising the prohibited conduct. By contrast, strict liability provisions may require proof of the mental element in relation to at least one (but not all) of the elements of the prohibited conduct. See, in this regard, John Smith and Brian Hogan, Criminal Law (OUP 2008) 152. 34 Cole v Hughes Tool Co 215 F2d 924 (10th Cir 1954); Bepco, Inc v Allied Signal, Inc 106 F Supp 2d 814, 828 (MDNC 2000); and United States v Dentsply Int’l 399 F3d 181, 197 (3d Cir 2005). 29

The role of intent in abuse of dominance and monopolization  227

III.

THE EU EXPERIENCE

Article 102 TFEU is also silent on the defendants’ intent. However, the EU courts, together with the European Commission (Commission), have not completely disregarded firms’ intent in deciding cases of abuse of dominance and, in particular, in cases of exclusionary and anticompetitive abuses that are the functional equivalent of the practices prohibited under the offences of monopolization and attempted monopolization in US antitrust law.35 The approach of EU decision-makers can be summarized as follows. Since the first case dealing with Article 102 TFEU, the Court of Justice of the European Union (CoJ) has repeatedly observed that ‘the concept of abuse is an objective concept’.36 In other words, it has always been clear that antitrust plaintiffs, whether public or private, do not need to prove the defendants’ intent to show the occurrence of an abuse of dominance, because firms’ intent is not a constituent element of the prohibition. However, over the years, some judicial and administrative practices have called into question this interpretation of Article 102 TFEU and, consequently, the existence of a regime of absolute liability which was traditionally said to govern cases of abuse of a dominant position.37 Those practices can be grouped in three sets, as follows. First, for nearly four decades, the CoJ has upheld a test to detect predatory prices made of two prongs, loss-making sales and the predatory intent of the dominant firms. It determined that when monopolists set prices below their average variable cost (AVC), loss-making sales are in re ipsa and predatory intent is presumed. In contrast, when dominant firms charge prices between their AVC and their average total costs (ATC), loss-making sales cannot be assumed, so predatory intent must be proven, with the ultimate effect of making it the determining factor in distinguishing abusive practices from legitimate conduct.38 Although only applied in three cases,39 this intent-based test has given rise to two popular interpretations, neither of which is The chapter, hence, does not look at the use of intent in exploitative and purely discriminatory abuses. 36 Case C-85/76 Hoffmann-La Roche & Co AG v Commission of the European Communities [1979] ECR 461, para 91; Case C-62/86 AKZO v Commission (hereinafter AKZO) [1991] ECR I-03359, para 69; and Case C-170/13 Huawei Technologies Co Ltd v ZTE Corp and ZTE Deutschland GmbH EU:C:2015:477, para 45; Case C-165/19 P Slovak Telekom, as v European Commission (hereinafter Slovak Telekom) ECLI:EU:C:2021:239, para 41. 37 Pinar Akman, ‘The Role of Intent in the EU Case Law on Abuse of Dominance’ [2014] European Law Review 316; Maria Melicias, ‘The Use and Abuse of Intent Evidence in Antitrust Analysis’ [2010] World Competition 569; and Antonio Bavasso, ‘The Role of Intent under Article 82 EU: From “Flushing the Turkeys” to “Spotting Lionesses in Regent’s Park”’ [2005] European Competition Law Review 11. 38 Valentine Korah, ‘The Paucity of the Economic Analysis in the [EC] Decision on Competition: Tetra Pak II’ [1993] Current Legal Problems 148; Michal Gal, ‘Below-Cost Price Alignment: Meeting or Beating Competition? The France Telecom Case’ [2007] ECLR 382; Pablo Ibáñez Colomo, ‘Intel and Article 102 TFEU Case Law: Making Sense of a Perpetual Controversy’ [2014], https://​ssrn​.com/​ abstract​=​2530878, accessed 5 May 2021. 39 AKZO (n 36) paras 71–2 (introducing the two-prong test reported in the text – a test that, interestingly, in its first wording does not contain any express reference to the defendant’s intent, but to the object of its practices); and paras 79, 99, 102 and 115 (for the use of the word ‘intention’ to identify the goal of harming a specific rival). See, further, Case C-333/94P Tetra Pak International SA v Commission [1996] I-05951, paras 41–2 (where the formulation of the test acquires a more subjectivist nuance because the Court establishes that ‘prices below average total costs but above average variable costs are only to be considered abusive if an intention to eliminate can be shown’ and then clarifies that when 35

228  Research handbook on abuse of dominance and monopolization compatible with an absolute liability regime. The first reading stands for the idea that, through such tests, the CoJ has introduced the dominant firms’ intent as one of the building blocks of Article 102 TFEU.40 However, this interpretation does not explain whether the predatory intent should be proven in addition to, or in lieu of, the potential anticompetitive harm that, along with the material ongoing practice, is the behavioural requirement of Article 102 TFEU. Adding intent to what must be proven would not make the plaintiff’s job any easier. Rather, in such a scenario, a plaintiff who is able to prove the predominant anticompetitive effects of the practice but not the predatory intent of the dominant firm would lose the case. Replacing the proof of the competitive impact with the proof of intent would make the intent-based test not only a means of introducing business intent into the constituent elements of Article 102 TFEU, but would also change the behavioural requirement of Article 102 TFEU from ‘any dominant firm’s pricing practice capable of harming competition’ to ‘any dominant firm’s pricing practice falling below its ATCs’. The second interpretation of the intent-based test establishes that firms’ intent does not work in relation to prices that fall below dominant firms’ AVC because those prices amount to an object restriction,41 but it does function with respect to prices that fall between AVC and ATC as evidence necessary to disambiguate conduct that would otherwise be difficult to characterize and find illegal.42 Under this reading, then, the intent-based test developed for predatory pricing would be no more than a specification of the general idea that business intent can function as a functional piece of evidence to establish that the practice in question is capable of harming competition. However, this discussion of the intent-based test has lost momentum. In the most recent cases involving price-based conduct, EU decision-makers have refrained from using such a test, instead relying on the equally efficient competitor test and/or basing their analyses solely on the costs of dominant firms.43 Second, according to some scholars,44 the role of intent stands out in EU competition law, because EU enforcers often address the actions of dominant firms to achieve an anticompetitive object, goal, plan, strategy or purpose. For example, in some cases the Court has made clear that a dominant firm’s practice cannot be allowed ‘if its purpose is to strengthen that

‘prices [are] considerably lower than average variable costs[, p]roof of intention to eliminate competitors [is] not necessary’); and Case C-202/07 P France Telecom v Commission [2009] I-02369, para 110. 40 Colm O’Grady, ‘The Role of Exclusionary Intent in the Enforcement of Article 102 TFEU’ [2014] World Competition 459, 471 et seq. 41 Eirik Osterud, Identifying Exclusionary Abuses by Dominant Undertakings under EU Competition Law: The Spectrum of Tests (Kluwer Law International 2010) 145. 42 As a matter of economics, the only dominant firms’ prices that should always been deemed irrational and, hence, anticompetitive by object, are those that fall below marginal costs. All the other possible measures of costs dragged into legal rules are instead proxies of this invisible benchmark and, as a consequence, may result in being either over- or underdeterrent. 43 See, in this regard, Commission, ‘Guidance on the Commission’s Enforcement Priorities in Applying Article [102 TFEU] to Abusive Exclusionary Conduct by Dominant Undertakings’ [2009] OJ C45/7, paras 63–74; Case COMP/35.141 Deutsche Post AG [2001] OJ L125/27 (predatory pricing); Case C-209/10 Post Danmark v Konkurrenceradet ECLI:EU:C:2012:172, paras 27–8 and 31–9 (selective price cutting); Case C-413/14 P Intel Corp v European Commission ECLI:EU:C:2017:632, para 136 (rebates); Case C‑280/08 P Deutsche Telekom v Commission [2010] ECR I‑9555, para 177; Case C-50/09 TeliaSonera Sverige [2011] I-00527, para 64; Case C-295/12 P Telefonica v Commission EU:C:2014:2062, para 124; and Slovak Telekom (n 36) para 70 (all four on price squeeze). 44 Patrick Perinetto, ‘Intent and Competition Law Assessment: Useless or Useful Tool in the Quest for Legal Certainty?’ [2019] European Competition Journal 153 and O’Grady (n 40) 461–2 and 481–2.

The role of intent in abuse of dominance and monopolization  229 dominant position and thereby abuse it’45 or that ‘the practice in question takes place in the context of a plan by the dominant undertaking aimed at eliminating a competitor’.46 More importantly, in other cases the Court has characterized conduct as capable of harming competition precisely because it was part of a plan whose object was to harm competition. This is certainly the case with vexatious lawsuits, which are considered abusive when dominant firms undertake them as part of a broader strategy to harm competition.47 It is also the case with dominant firms’ prices falling between AVC and ATC, in relation to which the Court stated that ‘prices below average total costs […] but above average variable costs, must be regarded as abusive if they are determined as part of a plan for eliminating a competitor’.48 However, these arguments revolving around the purpose of dominant firms’ conduct should not be taken as blatant evidence of the inclusion of firms’ intent among the building blocks of Article 102,49 quite the contrary. When assessing the object of firms’ actions, EU courts together with the Commission are referring not to the mental state of the firms involved, but to the economic rationality behind their practices.50 Even when EU judges and officials expressly talk about intent, they infer its existence from facts and circumstances,51 sometimes on the basis of firms’ internal documents.52 As a matter of fact, in the great majority of cases in which intent was at stake, the Commission and the CoJ have substantiated it by looking at objective elements, such as prices and costs, sometimes derived from internal documents;53 the timing and other specific characteristics of the conduct in question, such as its selective or discriminatory nature;54 and the detailed steps taken by the dominant firm to put its overall strategy into practice.55 In other words, the Commission has rarely considered pure motivational messages devoid of any factual basis.56 The Commission has never delved into the psychology or the morality of the defendants to appreciate their culpability or social dangerousness. The only concern of the 45 Case 27/76 United Brands Company and United Brands Continentaal BV v Commission of the European Communities [1978] ECR 207, para 189; Case T-65/89 BPB Industries and British Gypsum v Commission [1993] ECR II-389, para 69; and Joined Cases T-24/93, T-25/93, T-26/93 and T/28/93 Compagnie Maritime Belge Transports and Others v Commission [1996] ECR II-1201, para 107. 46 Case T-228/97 Irish Sugar plc v Commission of the European Communities [1999] II-02969, para 114; and Compagnie Maritime (n 45) paras 147–8. 47 Case T-111/96 ITT Promedia NV v Commission of the European Communities [1998] II-02937, para 55. 48 AKZO (n 36) para 72. 49 Mario Siragusa, ‘Italy: New Forms of Abuse of Dominance and Abuse of Law’ in Pier Luigi Parcu and others (eds), Abuse of Dominance in EU Competition Law: Emerging Trends (Edward Elgar Publishing 2017) 134. 50 See next, Section IV. 51 For a similar opinion see Renato Nazzini, The Foundations of European Union Competition Law (OUP 2011) 58. 52 See, eg, Geoffrey Manne and Marcellus Williamson, ‘Hot Docs vs. Cold Economics: The Use and Misuse of Business Documents in Antitrust Enforcement and Adjudication’ [2005] Arizona Law Review 609. 53 Case T-83/91 Tetra Pak International SA v Commission of the European Communities ECLI:EU:T:1994:246, para 151. 54 AKZO (n 36) paras 101–10. 55 AKZO (n 36) paras 79–82 and Case T-340/03 France Télécom SA v Commission of the European Communities ECLI:EU:T:2007:22, paras 199–200. 56 Of this type could be the report of Tetra Pak that indicated that the directors of the Italian branch felt the desire to make greater financial sacrifices to better support the competition of Elopak – see Tetra Pak (n 53) para 151.

230  Research handbook on abuse of dominance and monopolization Commission has always been the competitive impact of the conduct subject to its scrutiny, and the CoJ affirmed this practice. The same conclusion holds true in relation to the fact that the CoJ accepts the ‘only plausible explanation’ (OPE) concept,57 and does so not only when dominant firms’ prices are below AVC, but also when dominant firms make an abusive use of administrative procedures. The OPE argument, which derives from post-Chicagoan economics,58 provides that, when in a given market the conduct at stake would be irrational but for its anticompetitive effects, the conduct is only capable of producing those anticompetitive effects. Thus, ‘[a] dominant undertaking has no interest in applying such prices except that of eliminating competitors so as to enable it subsequently to raise its prices by taking advantage of its monopolistic position, since each sale generates a loss’.59 For instance, it would be irrational for a dominant pharmaceutical company to incur the costs that an unnecessary change in the formulation of its monopolized drug entails, unless such a change triggered an administrative procedure that excludes generics from the market and, thus, maintains the market price at the monopoly level.60 Again, this focus on the inherent economic rationality of a practice has nothing to do with the state of mind of a specific legal person or the state of mind of those who act for that person, having the authority to bind/represent it. It does not concern their psychology, morality or social dangerousness; it concerns the facts and circumstances that make economic sense of the practice in question. Finally, quite often and regardless of the type of practice at issue, the CoJ has stated that plaintiffs may use firms’ intent as evidence of the anticompetitive effects of its business conduct. Namely, it declared that: while, for the purposes of application of Article 102 TFEU, there is no requirement to establish that the dominant undertaking has an anticompetitive intent, evidence of such an intent, while it cannot be sufficient in itself, constitutes a fact that may be taken into account in order to determine that a dominant position has been abused.61 57 Mariateresa Maggiolino, ‘Plausibility, Facts and Economics in Antitrust Law’ [2014] Yearbook of Antitrust and Regulatory Studies 107. 58 See Michael Salop, ‘Exclusionary Conduct, Effect on Consumers, and the Flawed Profit-Sacrifice Standard’ [2006] Antitrust Law Journal 355. 59 AKZO (n 36) para 71; Tetra Pak II (n 39) para 41 (explaining that, ‘prices below average variable costs must always be considered abusive. In such a case, there is no conceivable economic purpose other than the elimination of a competitor, since each item produced and sold entails a loss for the undertaking’). 60 This was indeed the second abuse contended to AstraZeneca – see Case COMP/A. 37.507/F3 AstraZeneca, affirmed in Case C-457/10 P AstraZeneca AB and AstraZeneca plc v European Commission ECLI:EU:C:2012:770. As to the first abuse ascertained in this decision, it is true that AstraZeneca’s fraud made its request for the Supplementary Protection Certificate (SPC) unlawful; but when the boundaries of a patent almost overlap with the boundaries of the relevant market, any request for an SPC – even a fully legitimate request – is an exclusionary action capable of increasing market prices, because it keeps generics out of the market with the final result of leaving prices unchanged. Thus, AstraZeneca’s fraud stripped away the immunity that, for the sake of innovation, patent holders must always – and do always – enjoy while asking for SPCs, but it did not change the competitive assessment of the practice at stake. In other words, such a subjective element was crucial in proving the unlawfulness of the conduct, ie, its opposition to the current legal order in which patents must coexist with competition law, but it was not crucial in proving that the request of a SPC was capable of harming competition. 61 Case C-307/18 Generics Ltd and Others v Competition and Markets Authority ECLI:EU:C:2020:52, paras 67–8 (verbatim); and Case T-712/14 CEAHR v Commission EU:T:2017:748, para 102.

The role of intent in abuse of dominance and monopolization  231 In other words, the Court held that the examination of the abusive nature of monopolistic conduct must be made by considering all of the specific circumstances of the case,62 and the firm’s intent may be one of them.63 Therefore, while refuting the idea that Article 102 TFEU includes a subjective element among its building blocks, the CoJ has upheld the decisions in which the Commission used a firm’s intent to support its other findings on the abusive nature of the practice at stake. In these cases, however, the EU decision-makers have never investigated the culpability of the involved firms: they have instead sought to craft an accurate characterization of the monopolistic practices at hand and avoided analysing the mental state of the dominant firms involved as if it were a standalone element of the infringement. Indeed, in a recent judgment, the CoJ has clearly stated that ‘[t]he existence of an intention to compete on the merits, even if it were established, [cannot] prove the absence of abuse’, and confirmed that it would be an error in law if the Commission relied ‘exclusively on the intention or policy of [the dominant firm] in order to substantiate its finding of an infringement of competition law’.64 To sum up, since the first abuse of dominance case, EU antitrust scholars have been taught to think that these violations of competition law exist independently from any inquiry about the intent of defendants. Indeed, the CoJ has expressly refuted the idea that Article 102 TFEU includes a subjective element among its building blocks; and, together with the Commission, it never investigated the culpability of dominant firms as if it were an autonomous element of the infringement. Despite that, the CoJ has maintained that the intent of dominant firms is one of the many factors that can contribute to show the potential anticompetitive impact of the practice at stake – in predatory pricing cases, a decisive factor that can tip the scales in favour of the plaintiff. Section V will explain why this evidentiary use of the intent of firms is not fully consistent with the idea that Article 102 TFEU does not encompass any subjective element, especially when – as is the case in the EU – only plaintiffs can use firms’ intent to show the anticompetitive impact of the conduct, but defendants are not actually permitted to escape antitrust liability by arguing that they acted without intent. However, before getting to this issue, one should question whether it is actually reasonable to assume that firms’ intent can reveal the competitive impact of their conduct. This question is what the chapter turns to next.

IV.

DOES THE INTENT OF FIRMS REVEAL THE COMPETITIVE IMPACT OF THEIR CONDUCT?

As seen in the previous sections, both US and EU decision-makers rely on firms’ intent to determine the potential or actual effects on competition of the conduct under their scrutiny, especially when detecting such effects is complex. But can firms’ intent actually reveal the competitive impact of a business practice? The answer to this question depends on what we mean by ‘firms’ intent’. 62 TeliaSonera Sverige (n 43) para 68; and C‑95/04 P British Airways v Commission [2007] ECR I‑2331, para 67. 63 C-549/10 P Tomra and Others v Commission ECLI:EU:C:2012:221, paras 20–21 (holding that ‘the existence of any anticompetitive intent constitutes only one of a number of facts which may be taken into account in order to determine that a dominant position has been abused. However, the Commission is under no obligation to establish the existence of such intent on the part of the dominant undertaking in order to render Article 82 EC applicable’). 64 Tomra (n 63), paras 22 and 24.

232  Research handbook on abuse of dominance and monopolization If we interpret the word ‘intent’ consistently with many other fields of law,65 that is, as the state of mind with which a firm, or those with the authority to bind/represent it, have undertaken the conduct under investigation, it is hard to believe that, by knowing firms’ intent, antitrust decision-makers are capable of predicting and appreciating the competitive impact of the conduct. In the real world, there is always a hiatus between what agents want and what agents make happen. Even when agents’ intent covers both a specific material practice and its consequences, it is by no means certain that such intent will materialize. For example, suppose a dominant firm wants to apply predatory prices. Despite the intent, if its prices fall above its ATC they cannot harm competition. On the other hand, firms’ practices can distort competition regardless of the state of mind with which firms undertake them. For example, suppose that a dominant firm does not want to foreclose its horizontal rivals in an anticompetitive way. If it concludes a ten-year exclusive dealing contract with a distributor holding 45 per cent of the relevant market, it still forecloses its horizontal rivals and avoids price competition in the medium to long run, regardless of its intent to do so.66 Thus, firms’ intent is unlikely to actually reveal the potential or actual effects of a practice. All the more so if such intent is inferred from the blustery statements of some managers (including those who have the authority to bind/represent the firm), or if it is shaped as a generic desire to do better than rivals, since this mental disposition is common to both procompetitive and anticompetitive conduct.67 Of course, one could counter that firms – and, in particular, directors and top managers – are in the best position to predict and evaluate the effects of their conduct, because they do not act by chance.68 Indeed, firms – and those who manage them – know what they do, the market scenario in which they compete, as well as the impact that their conduct may produce on certain competitive elements, such as output and market prices, the quality and variety of supply, and the rate of innovation in the industry. However, upon closer inspection, those who raise this (correct) argument are adopting an economic-driven notion of intent, which can easily find a place within the antitrust domain. Indeed, due to the influence that economics has on the interpretation of Section 2 of the Sherman Act and Article 102 TFEU, in competition law the notion of intent may acquire an objective meaning, which has nothing to do with the actual mental state of a single person, be

65 See, eg, Thomas Weigend, ‘Subjective Elements of Criminal Liability’ in Markus D Dubber and Tatjana Hörnle (eds), The Oxford Handbook of Criminal Law (OUP 2014); Peter Cane, ‘Mens Rea in Tort Law’ (2000) 20 Oxford Journal of Legal Studies 533–56; Kenneth W Simons, ‘Rethinking Mental States’ (1992) 72 Boston University Law Review 463; John Finnis, ‘Intention in Tort Law’ in David G Owen (ed), Philosophical Foundations of Tort Law (OUP 1995) 229–46; Aharon Barak, Purposive Interpretation in Law (Princeton University Press 2005); Hugh Beale and others (eds), Cases, Materials and Text on Contract Law: Ius Commune Casebooks for the Common Law of Europe (Hart Publishing 2010). 66 The readers will forgive this simple exemplification of the foreclosure theory. In this regard, see Patrick Rey and Jean Tirole, ‘A Primer on Foreclosure’ (2006), http://​idei​.fr/​sites/​default/​files/​medias/​ doc/​by/​tirole/​primer​.pdf, accessed 15 July 2021. 67 See, eg, Ronald Cass and Keith Hylton, ‘Antitrust Intent’ [2001] Southern California Law Review 657; Herbert Hovenkamp, ‘The Monopolization Offense’ [2000] Ohio State Law Journal 1039; Frank Easterbrook, ‘Monopolization: Past, Present and Future’ [1992] Antitrust Law Journal 99; and Richard Whish, Competition Law (OUP 2008) 731. 68 Maurice Stucke, ‘Is Intent Relevant?’ [2012] Journal of Law and Economics 801; Marina Lao, ‘Reclaiming a Role for Intent Evidence in Monopolization Analysis’ [2004] American University Law Review 151.

The role of intent in abuse of dominance and monopolization  233 it a natural or a legal one, but instead derives from the application of rational choice theory to businesses. Consider that, in the world designed by economists, businesses are rational economic agents that make profit-maximizing choices based on the specific market scenarios in which they compete, that is, based on the facts and circumstances that characterize those scenarios.69 They are never unaware or unconscious. They do not act negligently or make blameless mistakes. They always consciously and voluntarily choose to act in a certain way. In addition – and more importantly, from the perspective of this chapter – these rational firms make choices about their practices based on the consequences they expect those practices to produce. The reason why firms follow a specific course of action has nothing to do with personal attitudes, moods, values or what might be some sort of moral distinction between ‘good’ and ‘bad’. Rational firms adopt a specific course of action because, given the facts and circumstances of the market in which they compete, they want to produce the expected effects associated with that course of action. Therefore, only in such a context – which is a blatant fictio iuris directly derived from economics – the intent of firms (or of those who act on their behalf) coincides with the consequences of firms’ actions and depends on the facts and circumstances that, in the given scenario, explain why for a rational agent the practice at stake makes economic sense. In other words, when firms (and those who have the power to represent and bind them) are conceptualized as rational economic agents, their intent simply consists of what any rational agent (devoid of any personal nuance) would have wanted in the same facts and circumstances,70 irrespective of their psychology, morality or societal dangerousness. To be sure, one could counter that in reality firms, directors and top managers are not rational agents: they engage in complex decision-making processes, stumble in cognitive biases and bear several cognitive limits. However, this chapter does not stand for the idea that real firms are rational agents, although if they were considered so, they would be incentivized to behave rationally. This chapter maintains that only in a scenario in which firms are conceived of as rational agents we have grounds to support the evidentiary use of firms’ intent, because only in that scenario firms’ intent can actually function as a reliable proxy for the procompetitive or anticompetitive effects of firms’ practices.71 That said, it is difficult to determine whether and when US and EU courts have actually endorsed this economics-based understanding of firms’ intent. It is true, however, that, in some decisions, they have clearly referred to firms as if they were rational agents,72 and that, as said above, they have never addressed the culpability of the defendants as if they were interested in probing the psychology or moral attitude of the defendants or as if these were crucial

See, eg, N Gregory Mankiw, Principles of Economics (South-Western Publishing 2017). Interestingly enough, the economic-driven notion of intent here analysed is similar to the ‘objective notion’ of intent often used in contract law, which does not inquire the mental states of the parties involved, that is their specific expectations and anticipations, but grasps what a reasonable person would have wanted in a specific situation. See, in this regard, Catherine Mitchell, Interpretation of Contracts (Routledge-Cavendish 2007). 71 Katharine Kemp, ‘A Unifying Standard for Monopolization: “Objective Anticompetitive Purpose”’ [2017] Houston Journal of International Law 113. 72 American Tobacco (n 10) 814 quoting Alcoa (n 10) 431–2 (both of them pinpointing to the archetypal of the rational economic agent); and Irish Sugar (n 46) para 170; Case T-203/01 Michelin v Commission [2003] II-4082, para 241; and France Telecom (n 39) para 195 (all specifying that there is no difference between the object of the effects of dominant firms’ practices). 69 70

234  Research handbook on abuse of dominance and monopolization elements in understanding the blameworthiness and/or harmfulness of the practice at stake. Indeed, sometimes antitrust rulings and decisions are replete with terms such as ‘intention’ or ‘motives’, but they focus on the facts and circumstances that give economic sense to the business actions at stake.73 The explanation might be that US courts, the Commission and the EU courts suffer from a kind of subjectivist bias in approaching and referring to economic rationality.74 For example, in Bell v. Dow Chemical Co. the Fifth Circuit held that the District Court had to decide whether the conduct at stake ‘was actuated by innocent motives rather than by an intention and desire to perpetuate a monopoly’, by taking into consideration ‘legitimate business concerns […] such as cost savings, shortage of supplies, more efficient production […]’.75 Likewise, in a quite recent opinion, Advocate General Mazàk argued in front of the CoJ that ‘the evidence of intent is not altogether irrelevant insofar as it may actually be relevant to the assessment of the behaviour of a dominant undertaking, which requires an understanding of the economic rationale of that behaviour, its strategic aspects and its likely effects’.76 Indeed, it seems that when US and EU courts evaluate ‘intent’, ‘desire’ or ‘motive’, they are actually concerned with economic rationality. If this is the case, relying on firms’ intent to derive the economic impact of a business behaviour makes sense; but is this practice also worthwhile and legitimate? Section V elaborates on this question next.

V.

WHAT EXCLUDING OR INCLUDING THE SUBJECTIVE ELEMENT INTO A LEGAL PROVISION MEANS AND IMPLIES

In order to qualify firms’ intent as one of the building blocks of Section 2 and Article 102, one should consider: (i) the consequences that this decision entails; (ii) the usual reasons why legal scholars consider intent as a necessary element; and (iii) how this decision affects the use of business intent for evidentiary purposes. First, the introduction of the subjective element in legal provisions does not imply that decision-makers may refrain from proving the illicit conduct: it does not work as a short-

73 Consider a dominant firm producing ice-creams that concludes ten-year exclusive dealing contracts with several wholesalers together holding a market share of about 50 per cent. As a consequence of these contracts, the other ice-cream producers see an increase in the cost of distributing their product, because they now need to either offer very attractive contractual conditions to the wholesalers that are still viable, or create their own distribution channel. All things equal, the dominant firm will hence progressively be subject to lower competitive constraints and will be able to successfully keep its prices higher than they could otherwise have been. Decision-makers can describe this behaviour both in either a subjective or an objective way: they can state that the dominant firm intends to keep its market price as high as possible, by excluding its horizontal rivals from the market for ice-cream production; or they can say that the dominant firm engages in a conduct that raises rivals’ costs, which is capable to marginalize horizontal competitors into a niche of the primary market and to reduce the competitive pressure on market price. However, in both cases, the analysis focuses on the circumstances that occurred, the characteristics of the conduct, the relevant legal-economic context in which it took place, and on what makes economic sense. 74 Areeda and Hovenkamp (n 13) para 805. 75 847 F2d 1179, 1185 (5th Cir 1988) (emphasis added). 76 Opinion of AG Mazàk, 2 February 2012, Case C-549/10 P Tomra EU:C:2012:55, para 10 (emphasis added).

The role of intent in abuse of dominance and monopolization  235 cut. For example, if Section 2 and Article 102 actually encompassed firms’ intent, antitrust decision-makers could never argue that, because they have been successful in proving firms’ intent to harm competition, they can refrain from showing that the conduct in question is likely to enhance market power without producing any redeeming virtue in the relevant market. If they did so, they would fail to prove the behavioural element of Section 2 and Article 102, and this would be contrary to the general rule mentioned above that says plaintiffs must prove all the constituent elements of a given prohibition to show that defendants are liable.77 Thus, it is difficult to identify the benefits that antitrust plaintiffs would gain by considering corporate intent as one of the constituent elements of Section 2 and Article 102 or by reading the case law discussed above in this direction. Second, the choice to include defendants’ subjective element in a statutory provision produces two additional enforcement consequences. It allows defendants to get away with their wrongdoing by claiming (and proving) that they did not intend to engage in the practice in question; that they were unaware that the practice was capable of triggering adverse effects; or that they did not expect the latter as one of the possible implications of their practice. Moreover, within legal systems that distinguish the ‘mind’ of the firm involved from the mind(s) of the specific individual(s) who actually engaged in the alleged misconduct,78 corporate defendants can avoid the charges by claiming that the infringements were due to the conduct of somebody devoid of the authority to bind/represent the firm. When legal provisions encompass the reprehensibility of defendants’ conduct, all these justifications are fully legitimate, because they are meant to show that the plaintiffs have not met the subjective element requirement. Thus, those who support the inclusion of firms’ intent into Section 2 and Article 102 or who read US and EU case law as proof of this thesis should also clarify if they are aware of the further enforcement disadvantages that this choice implies. Moreover, usually the decision to place the defendant’s subjective element in a legal provision is not intended to satisfy mere evidentiary concerns; rather, it is intended to condition the existence of the violation of that provision on the precise identification of the defendants’ mental states. More clearly, traditionally, the choice to give space to defendants’ culpability is not intended to put that element at the service of proving the prohibited conduct; rather, it is intended to give autonomous credit to defendants’ state of mind. And this is on three general premises, which may coexist or not. First, no natural or legal person should be punished for a wrongful act without showing that they chose to do it or, if negligence is considered, without showing that they acted without due diligence.79 Second, society’s reaction to the same prohib-

See above n 40 and the accompanying text. The reasoning developed in relation to the first interpretation of the intent-based test is a good exemplification of what is said in the text. 78 This is not an issue in the US. However, prior to Brexit, it was a problem in at least one EU member state, namely the United Kingdom. In light of this, it could be argued that especially those legal systems that ‘result’ from the sum of multiple legal systems should consider absolute liability rules as a way to escape this kind of issue. 79 In criminal law, the principle that there is no such thing as a guilty person without a guilty mind (actus non facit reum nisi mens sit rea) almost always applies – see Ashworth (n 3) 28–9 and 158–9. In fact, criminal law considers the subjective element irrelevant only in very few specific cases, often named regulatory offences, where the protection of fundamental societal interests is at danger; ibid 164–73. Similarly, most torts consist of negligent or intentional conduct. Exceptions include product liability and abnormally dangerous activities. As a matter of social adjustment, in these cases defendants, as businesses, are deemed to be in a better position to assume the risks and shoulder the harm resulting 77

236  Research handbook on abuse of dominance and monopolization ited conduct may vary depending on how dangerous and, therefore, reprehensible the mindset of the perpetrator was, with the ultimate goal of taking the most severe approach against those who, by acting on purpose, proved to be more socially dangerous.80 Third, there may be practices that are legal to the extent that they are not undertaken with malice.81 Thus, for the sake of the internal coherence that legal orders should have, those who support the inclusion of firms’ intent in Section 2 and Article 102, or who read the US and EU case law as proof that firms’ intent is a requirement of the two provisions, should also explain why they argue this. Do they really believe that the existence of an antitrust violation should depend on the culpability and mindset of the defendants? Do they really assume that a dominant firm’s practice is lawful if undertaken without malice or negligence? Or do they give so much weight to firms’ intent for pure evidentiary reasons? Third, one could argue – as the CoJ has often done, but not the US courts – that using firms’ intent for evidentiary purposes does not imply considering it as one of the constituent elements of the relevant antitrust prohibitions. In other words, one could – as EU judges have often done – design a scenario in which, on the one hand, antitrust plaintiffs are not required to prove firms’ intent to show the defendants’ liability, but they are permitted to rely on firms’ intent to show the anticompetitive effects of firms’ conduct; and, on the other hand, defendants are not allowed to escape antitrust liability by arguing that they acted without intent, but they may be charged with an anticompetitive practice because of their intent to harm competition. However, this interpretation prompts two challenges. As a matter of logic, it is hard to understand why, once a provision does not encompass firms’ intent as one of its building blocks, the parties should be permitted to gather and use evidence on that element, notwithstanding the fact that, in the ‘legal world’ that the provision in question creates, that specific element does not even exist. Furthermore, for the sake of the principle of equality of arms, it should not be conceded that while plaintiffs are allowed to use companies’ intent to show that a practice is capable of harming competition, defendants are not allowed to get away with their wrongdoing by building their defence on their (lack of) intent. In other words, such a hybrid solution that makes corporate intent a sword in the hands of plaintiffs but not a shield in the hands of defendants is clearly unfair. As said above, the CoJ seems to permit such an imbalance, while US judges do not. The latter allow both plaintiffs and defendants to rely on firms’ intent, notwithstanding their declarations as to the role that firms’ intent should not have within Section 2 of the Sherman Act.

from their products/actions than innocent consumers – see William L Prosser, ‘The Assault upon the Citadel (Strict Liability to the Consumer)’ (1960) 69 Yale Law Journal 1099. 80 On this policy issue, see Areeda and Hovenkamp (n 13) para 805b.2 (noting that ‘In the case of an ordinary criminal offence […] the defendant’s intent bears significantly on whether society needs to fear [the defendant]. But there is no such an intent requirement in the […] monopolization offense. […] [A]ntitrust concerns itself not with the purity of an actor’s soul. […] What is in defendant’s heart tells us little about the consequences, actual or prospective, are the kind society need to fear’). 81 This, for example, was the issue underlying a recent European Free Trade Association (EFTA) judgment analysing Article 12, Reg UE n 596/2014, which prohibits market manipulation. See Case E-5/19 Criminal proceedings against F and G (2020), https://​eftacourt​.int/​download/​5​-19​-judgment/​ ?wpdmdl​=​6449, accessed 15 July 2021 and Chiara Picciau, ‘Recenti spunti giurisprudenziali sulla frammentaria nozione di manipolazione del mercato’ [2020] Nuove Leggi Civili Commentate 1286, 1302 et seq.

The role of intent in abuse of dominance and monopolization  237 In summary, as a matter of logic and to protect the principle of equality of arms, antitrust plaintiffs who are genuinely interested in using business intent for evidentiary purposes cannot continue to argue that Section 2 of the Sherman Act and Article 102 TFEU are absolute liability rules: they must concede that these provisions include firms’ intent among their building blocks. However, in light of what was said in Section IV about the probative value of firms’ intent and in this section about the practical disadvantages of this choice and the policy values it represents, one should seriously question whether it is really worthwhile to interpret Section 2 of the Sherman Act and Article 102 TFEU as to include a subjective element.

VI.

IT WOULD BE BETTER IF SECTION 2 AND ARTICLE 102 DID NOT GIVE ANY ROOM TO FIRMS’ INTENT. TO SOME EXTENT, THEY ALREADY DO NOT

Legal prohibitions do not include firms’ intent as one of their constituent elements when the potential harm of the prohibited conduct is considered dangerous enough to be unworthy of the leeway that is usually given to subjective liability. Think, for example, of regulatory offences or product liability: these behaviours are so hideous and dangerous that society cannot admit that defendants use their (lack of) intent as a shield against their liability. In other words, absolute liability provisions stand for the idea that society and those harmed by the prohibited conduct should not bear the costs of winning defences based on intent.82 Therefore, if Section 2 and Article 102 did not encompass firms’ intent as one of their building blocks, they would convey a simple policy message: that the protection of competition is so paramount that it cannot be undermined by any analysis of firms’ intent. After all, competitive distortions are a question of fact to be appreciated on a case-by-case basis: they happen irrespective of the state of mind of the involved firms. Thus, the decision to give room to defendants’ intent would stand for the idea that some competitive distortions must go unpunished when their perpetrators acted with good intentions. At present, it is hard to think of any reason why such a message would be appropriate, especially considering that the ‘inchoate mode’83 of Section 2 and Article 102 already shows how the drafters of the Sherman Act and the Treaty of Rome considered the interest in avoiding any competitive harm to be very significant. In other words, the fact that those two provisions may even apply when the competitive harm has not yet occurred84 demonstrates the legislative interest in protecting competition in the best way possible. Therefore, it would not be outrageous if scholars pursue the same interest by endorsing the idea that a competitive distortion must be prohibited regardless of any concern about the state of mind of the firms that engaged in the relevant anticompetitive conduct. Despite some references to intent as an evidentiary tool in both the US and EU case law, this chapter has clearly shown that the use of corporate intent for evidentiary purposes is only

See nn 79 and 80. Andrew Ashworth and Lucia Zedner, ‘Prevention and Criminalization: Justifications and Limits’ (2012) 15 New Criminal Law Review 542, 545 for the idea that wrongdoings defined in an ‘inchoate mode’ are ‘substantive offenses that are defined in such a way that they penalize conduct before it reaches the stage of causing harm’. 84 See, eg, American Tobacco (n 10) 810 and AstraZeneca (CoJ) (n 60) para 112. 82 83

238  Research handbook on abuse of dominance and monopolization legitimate if Section 2 and Article 102 treat such a state of mind as one of their requirements. Therefore, antitrust scholars face a trade-off between the use of firms’ intent to prove the anticompetitive effects of their practices and the enforcement and policy advantages that come together with an absolute liability regime. To be sure, there is an interpretive option that resolves this trade-off, explains why the statements of the CoJ are definitively correct, and makes the evidentiary use of corporate intent be meaningful. If firms’ intent coincides with the economic rationality of their practices, plaintiffs that rely on such intent do not focus on any mental state or attitude: they examine facts and circumstances to understand the procompetitive or anticompetitive conditions under which a given practice makes economic sense. Thus, first of all, such an analysis is able to reveal the competitive impact of the conduct (see Section IV). Second, this analysis remains objective and independent from any elaboration on firms’ culpability, as it would have been in a scenario in which firms’ intent was not at stake. In other words, if firms’ intent coincides with economic rationality all the policy issues and enforcement consequences connected to a regime of absolute liability (as discussed in Section V) do not come into question and, thus, one can keep on affirming that Section 2 and Article 102 do not include the subjective element among their building blocks. Therefore – and third – if firms’ intent coincides with economic rationality, the idea of the CoJ that firms’ intent may at the same time be an evidentiary but not a constituent element of Article 102 makes sense, not only as a matter of logic, but also in light of the principle of equality of arms. As a matter of logic, when the parties do not gather or use evidence on firms’ mental state, but collect evidence about facts and circumstances, the fact that in the ‘legal world’ that Article 102 creates the mental element does not exist is not called into question. In light of the principle of equality of arms, consider that under an absolute liability regime defendants are allowed to escape liability by relying on facts and circumstances. Within Section 2 and Article 102 these facts and circumstances are the business justifications that show the redeeming virtues of dominant firms’ monopolistic practices. Hence, by interpreting firms’ intent in an objective, economic-driven way, we give defendants the shield they were missing, while saving the idea that Section 2 and Article 102 do not include the subjective element as one of their building blocks. The further development of this approach in detail is left for future work.

13. The special responsibility of dominant undertakings Giorgio Monti and Ekaterina Rousseva1

I. INTRODUCTION The first time the European Court of Justice (hereinafter, the CoJ) spoke of a dominant firm’s special responsibility was in Michelin. The applicant complained about the Commission’s finding that the quality of its product and services were indicators that it had a dominant position. It took the view that this penalized a firm’s commercial success. While affirming that these factors were legally relevant, the Court conceded that A finding that an undertaking has a dominant position is not in itself a recrimination but simply means that, irrespective of the reasons for which it has such a dominant position, the undertaking concerned has a special responsibility not to allow its conduct to impair genuine undistorted competition on the common market.2

Given the context in which this passage is found (a discussion of dominance), the Court could not have intended to use this as an interpretative key about the notion of abuse. Instead, this statement proved useful for the Court to signal that dominance does not imply an infringement automatically.3 Ten years later, the Court of First Instance (now, General Court) and subsequently the CoJ began to refer to an abridged version of the quoted passage, focusing only on the special responsibility that a dominant firm has to not impair undistorted competition, and situated the Michelin dictum in the framework of its discussion of whether conduct is an abuse of dominance.4 This recast indicated a renewed importance of the concept. Being now related to the notion of abuse, an evolving element of Article 102 of the Treaty on the Functioning of the European Union (TFEU) than the notion of dominance, the concept of special responsibil1 The opinions expressed are strictly personal; they do not represent the views of the European Commission or any of its services. The law in this chapter is stated as of 15 December 2021. Just before publication, we were kindly allowed to introduce a reference to Case C-680/20 Unilever Italia Mkt. Operations EU:C:2023:33. 2 Case 322/81 NV Nederlandsche Banden Industrie Michelin v Commission EU:C:1983:313, para 57. 3 In the context of undertakings that are granted special or exclusive rights the Court came very close to recognizing automatic infringements when state regulation made an abuse of dominance inevitable. The high-water mark of this approach is Case C-320/91 Criminal Proceedings against Paul Corbeau EU:C:1993:198. This doctrine was used strategically to liberalize markets. 4 Case T-65/89 BPB Industries Plc and British Gypsum Ltd v Commission EU:T:1993:31, para 67. The first court case to use this passage in the context of abuse is Joined Cases C-395/96 P and C-396/96 P Compagnie maritime belge transports SA v Commission EU:C:2000:132, where para 57 of the ruling in Michelin (n 2) is quoted twice, first in a discussion of dominance (para 37) and second in a discussion of abuse (para 85) when examining if the dominant undertaking was responsible for the conduct or if this was required by regulation.

239

240  Research handbook on abuse of dominance and monopolization ity becomes more complex and intriguing. Moreover, this recast makes pertinent the question of how exactly dominance, which is the basis for the responsibility, is relevant for the responsibility to materialize. It therefore also transforms the static concept of dominance into a more dynamic concept, whereby the degree of market power matters. Contrary to what is sometimes assumed, the CoJ’s reference to the notion of special responsibility is not ubiquitous. It is only since 2009 that reference to the dominant firm’s special responsibility emerges with more frequency in the Court’s case law,5 but even then the CoJ only refers to it in 10 out of some 19 judgments that dealt with aspects of Article 102 TFEU.6 Two events might account for the increased use of the notion: first, the Member States, with the Lisbon Treaty, appeared to have demoted the role of competition policy by placing the notion of ‘undistorted competition’ in a Protocol, and second, the Commission issued a guidance paper on enforcement priorities for exclusionary abuse, suggesting a reinterpretation of the concept of abuse along the philosophy of the more economic approach that the Commission had embraced for other antitrust rules.7 These events shaped the Court’s evolving interpretation of Article 102 TFEU. The CoJ was quick to explain that the Treaty amendments had no impact.8 Although the Court has taken longer to rule on the implications of the new approach canvassed in the Guidance Paper (which sets out the Commission’s enforcement priorities), there is little doubt left that this document has had a certain impact on how judges developed the law on abuse.9 In developing the notion of exclusionary abuse since the release of this Guidance, the Court makes frequent references to special responsibility to assist its reasoning. Notably, the Court has not referred to the special responsibility in exploitative abuse cases. A simple explanation is that exploitative abuses are not about breaching the duty not to ‘impair genuine undistorted competition’ but about extracting benefits from an already distorted competition. Another possible explanation is that in view of the market power required

5 We focus on the CoJ here because it is the institution tasked with the final interpretation of the Treaty. 6 These are the judgments where the Court makes reference to special responsibility: Case C-202/07 P France Télécom v Commission ECLI:EU:C:2009:214, para 105; Case C‑280/08 P Deutsche Telekom v Commission ECLI:EU:C:2010:603 (hereinafter Deutsche Telekom I), para 83; Case C-52/09 Konkurrensverket v TeliaSonera Sverige AB ECLI:EU:C:2011:83, para 24; Case C-209/10 Post Danmark A/S v Konkurrencerådet ECLI:EU:C:2012:172 (hereinafter Post Danmark I), para 23; Case C‑457/10P Astra Zeneca AB and Astra Zeneca plc v Commission EU:C:2012:770, paras 134 and 149; Case C-23/14 Post Danmark A/S v Konkurrencerådet ECLI:EU:C:2015:651(hereinafter Post Danmark II), para 71; Case C‑413/14 P Intel v Commission EU:C:2017:632, para 135; Case C-307/18 Generics (UK) Ltd and Others v Competition and Markets Authority EU:C:2020:52, para 153; Case C-165/19 P Slovak Telekom v Commission EU:C:2021:239, para 40; Case C-152/19 P Deutsche Telekom v Commission EU:C:2021:238, para 40. 7 Treaty on the Functioning of the European Union (TFEU), Protocol 27 on the Internal Market and Competition; Guidance on the Commission’s enforcement priorities in applying Article 82 of the EC Treaty to abusive exclusionary conduct by dominant undertakings [2009] OJ C45/7 (hereinafter Guidance Paper). 8 TeliaSonera (n 6) paras 20–24. 9 In contrast the academic community has explored the possible implications of the Guidance Paper extensively. For some early discussions, see eg P Akman, ‘The European Commission’s Guidance on Article 102 TFEU: From Inferno to Paradiso?’ (2010) 73 Modern Law Review 605; G Monti, ‘Article 82 EC: What Future for the Effects-Based Approach?’ (2010) 1 Journal of Competition Law and Practice 2.

The special responsibility of dominant undertakings  241 for such abuses and the gravity of exploitation as such, there is no need for a reminder that such conduct is unacceptable.10 Some consider that emphasis on the role that the notion of special responsibility has is exaggerated. It is clear that the special responsibility is not the basis upon which one can find an abuse.11 The conclusion that the passage is a ‘statement of the obvious’ is hard to resist.12 However, there are certain aspects of the notion of abuse of dominance for which the dictum in Michelin proves helpful in interpreting the boundaries of Article 102 TFEU. In an earlier judgment (Hoffmann-La Roche) the Court suggested that dominant undertakings should be subject to stricter rules than non-dominant undertakings, because, in the presence of dominance, the degree of competition in the market is already weakened and therefore any further interference with the market structure is likely to distort competition.13 As we explain below this accounts for some fairly wide obligations that dominant undertakings have sometimes been asked to carry. The idea of subjecting a powerful undertaking to stricter rules is not alien to the US jurisprudence. The emblematic Alcoa case is often seen as a manifestation of this idea and has not remained without criticism.14 That said, the notion of special responsibility, much like other contentious phrases in EU antitrust law (eg competition on the merits), finds no ready twin in US antitrust. The closest the Supreme Court has come to the notion of special responsibility seems to be a dictum by Scalia J in Kodak: ‘Behavior that might otherwise not be of concern to the antitrust laws … can take on exclusionary connotations when practiced by a monopolist.’15 However this is simply a reaffirmation that the monopolization offence requires a finding of monopoly power before one moves to explore if the conduct infringes Section 2 of the Sherman Act. This chapter is structured in a manner that allows for a demonstration of how the various references the Court makes to the Michelin dictum concerning special responsibility have affected the evolution of the notion of exclusionary abuse. Section II discusses what the Court means by genuine undistorted competition. This allows the presentation of how the Court has used the Michelin obligation to move the interpretation of abuse away from a predominantly

For example, excessive pricing raises concerns only in markets with an entrenched dominant position where entry and expansion of competitors could not be expected to ensure effective competition in the foreseeable future. See OECD Policy Roundtables, Excessive Pricing, 2011 (EU report), available at https://​www​.oecd​.org/​competition/​abuse/​49604207​.pdf, accessed 19 January 2023. However, a reference to the special responsibility of dominant undertakings can also be found in some Commission decisions on exploitative abuses. See eg Case AT.40394 Aspen, decision of 10 February 2021, paras 78–9. 11 This is in line with the bulk of the literature, eg R Nazzini, The Foundations of European Union Competition Law (Oxford 2011) 174 noting that the Court ‘was not laying down a test for abuse’. 12 R Whish and D Bailey, Competition Law (9th edn, Oxford 2018) 198. I Lianos, ‘Categorical Thinking in Competition Law and the “Effects-Based” Approach in Article 82 EC’ in A Ezrachi (ed), Article 82 EC: Reflections on its Recent Evolution (Hart 2009) 24–30 goes further to discuss why such a special responsibility should exist in the first place since the same anticompetitive harm can also be caused by non-dominant undertakings. 13 Case 85/76 Hoffmann-La Roche v Commission EU:C:1979:36, paras 91 and 120. 14 United States v Aluminum Co of America 148 F2d 416 (2d Cir. 1945). See eg JE Lopatka and PE Godek, ‘Another Look at Alcoa: Raising Rivals’ Costs Does Not Improve the View’ (1992) 35 The Journal of Law & Economics 311. 15 Eastman Kodak Co v Image Technical Services, Inc 504 US 451, 488 (1992). 10

242  Research handbook on abuse of dominance and monopolization form-based approach to a state where today the default position is that an abuse requires a ‘qualified’ effects analysis, with a restricted category of conduct judged restrictive by object. Section III explores the link between dominance and abuse and, in particular, how the holding of a dominant position impacts the assessment of different types of conduct and what the interplay between market power and abusive behaviour is. Section IV discusses whether the Michelin obligation is proscriptive (a responsibility not to allow) or also prescriptive (a responsibility to behave in a particular way). Section V considers the impact that regulation has on the scope of the special responsibility.

II.

GENUINE UNDISTORTED COMPETITION

A.

Towards a Qualified Effects-Based Approach

The duty not to impair genuine undistorted competition is the essence of the special responsibility set out in Michelin. Read together with the definition of abuse in Hoffmann-La Roche, this duty means a dominant undertaking must not restrict the competition that remains in the market, nor its potential growth.16 However, the notion of competition that the EU Courts endorsed has been the subject of contestation. At a high level, the debate about the meaning of the notion of a restriction of competition for the purposes of exclusionary abuse is about the kind of competitive process that Article 102 TFEU is meant to protect. Broadly, and with the risk of simplification, two viewpoints have been shaping the debate. According to the first, the competitive process is hindered when the conduct alters the existing market structure, namely, it excludes or marginalizes one or more competitors regardless of the importance of those competitors for the effectiveness of competition. This approach is supported by the argument that, when there is dominance, any further restriction of competition becomes problematic and therefore there is no need for further analysis. On the second view, conduct is an abuse only if the harm to the competitive process is capable of leading to harm to consumers.17 This requires an assessment of whether the restriction of competition is significant enough to cause harm to consumers. Both views can probably find support in the seminal statement in Continental Can that Article 102 TFEU is ‘not only aimed at practices which may cause damage to consumers directly, but also at those which are detrimental to them through their impact on an effective competition structure’18 as well as in the General Court’s clarification in British Airways that ‘competition law concentrates upon protecting the market structure from artificial distortions because by doing so the interests of the consumer in the medium to long term are best protected’.19 The first view is often associated with a form-based approach and the second with an effects-based approach. It appears that over the years the CoJ has moved gradually towards the second position. For example, in some more recent case law, when describing the impact of the conduct, the

Hoffmann-La Roche (n 13) para 91. On these debates see generally L Lovdahl Gormsen, A Principled Approach to Abuse of Dominance (Cambridge University Press 2010); E Deutscher and S Makris, ‘Exploring the Ordoliberal Paradigm: The Competition-Democracy Nexus’ (2016) 11 Competition Law Review 181. 18 Case 6/72 Europemballage and Continental Can v Commission EU:C:1973:22, para 26. 19 Case T-219/99 British Airways v Commission EU:T:2003:343, para 264. 16 17

The special responsibility of dominant undertakings  243 Court, intentionally or not, has omitted the word ‘structure’, and referred only to conduct that is detrimental to consumers ‘through the impact on effective competition’.20 On one occasion, the Court even added to the classical definition of abuse in Hoffmann-La Roche – repeated in case law for decades – that the effects of hindering competition have to be ‘to [the] detriment of consumers’.21 Indications for this trend are likewise found in the much more inquisitive consideration by the Court of ‘all the circumstances’ of the investigated case,22 and in the acknowledgement that ‘not every exclusionary effect is necessarily detrimental to competition’.23 This development was flagged in the Commission’s Guidance on enforcement priorities, which suggested that ‘[a]ccording to the case-law, holding a dominant position confers a special responsibility on the undertaking concerned, the scope of which must be considered in the light of the specific circumstances of each case’.24 Not long ago, in Post Danmark II, the CoJ expressly held that the assessment of abuse ‘seeks to determine whether the conduct of the dominant undertaking produces an actual or likely exclusionary effect, to the detriment of competition and, thereby, of consumers’ interests’.25 This trend was confirmed by Intel, the ruling of the CoJ on exclusionary abuse which has attracted most attention in the present era. Here, the notion of special responsibility is cited to explain why Article 102 TFEU prohibits a dominant undertaking from, among other things, adopting pricing practices that have an exclusionary effect on competitors considered to be as efficient as it is itself and strengthening its dominant position by using methods other than those that are part of competition on the merits.26

The Court echoes Post Danmark I’s statement that the exclusion of inefficient rivals may not constitute an abuse of dominance.27 Shortly after the ruling in Intel, in MEO the Court applied a similar principle to secondary line price discrimination. It clarified that, for Article 102(c) TFEU to apply, the dominant undertaking’s conduct has not only to put a trading partner on a downstream market at a disadvantage, but the disadvantaged trading partner has to be at least as efficient as its competitors.28 While the practical implementation of this suggestion raises numerous questions – for example, which competitors should be used as a reference point for efficiency, and how a dominant undertaking can know which of its trading partners is efficient or not – the point that the Court seems to have wanted to make is that harm to a trading partner becomes a concern only if that trading partner is relevant for the effectiveness of the competitive process and ultimately to consumers.29 20 See Post Danmark I (n 6) para 20; TeliaSonera (n 6) para 24; Joined Cases C-468/06 to C-478/06 Sot Lélos kai Sia and Others EU:C:2008:504, para 68, and Deutsche Telekom I (n 6) para 176. 21 See Post Danmark I (n 6) para 24. 22 TeliaSonera (n 6) para 28; Post Danmark I (n 6) para 26. 23 Post Danmark I (n 6) para 22; TeliaSonera (n 5) para 43; Intel (n 6) para 134. 24 Guidance Paper (n 7) para 9. 25 Post Danmark II (n 6) para 69. 26 Intel (n 6) para 136. 27 Post Danmark I (n 6) para 22. 28 Case C-525/16 MEO – Serviços de Comunicações e Multimédia SA v Autoridade da Concorrência EU:C:2018:270, para 31. 29 See R Horvath, L Peeperkorn, E Rousseva, ‘The Preliminary Ruling in MEO: Closing the Circle of Article 102 TFEU’ (2020) 11 Journal of European Competition Law & Practice 35; R O’Donoghue QC and J Padilla, The Law and Economics of Article 102 TFEU (3rd edn, Hart Publishing 2020) 986.

244  Research handbook on abuse of dominance and monopolization The move to an effects-based approach seems to be supported also by the fact that some years ago the Court made space for an efficiency defence under Article 102 TFEU. The defence replicated the conditions of Article 101(3) TFEU and made it clear that the efficiency brought by the dominant undertaking has to benefit consumers.30 If the distortion of competition were unrelated to consumer harm, it would have made little sense to make the defence contingent upon consumers receiving benefits from the claimed efficiencies. There is little doubt that a shift to a qualified assessment of the competitive process implies an increased evidentiary burden for the party proving the abuse. However, this burden is mitigated by the fact that proving potential (as opposed to actual) effects is in principle sufficient for finding an abuse. The CoJ has made it clear that, although negative actual effects on competition are usually a strong indication for an abuse, showing such effects is not a prerequisite for the application of Article 102 TFEU.31 Already in British Airways the Court held that conduct can be regarded as being abusive as soon as the behaviour of the undertaking in a dominant position tends, having regard to all the circumstances of the case, to lead to a distortion of competition and that in such circumstances the Commission cannot be required in addition to adduce proof of an actual quantifiable deterioration in the competitive position of the business partners.32 In TeliaSonera and in Post Danmark II the Court expressly held that, in order to establish whether a practice is abusive, that practice must have an anticompetitive effect on the market, but the effect does not necessarily have to be concrete, and ‘it is sufficient to demonstrate that there is an anticompetitive effect which may potentially exclude competitors’ who are at least as efficient as the dominant undertaking.33 Indeed, requiring actual effects would have rendered Article 102 TFEU a toothless instrument, as a belated intervention, which waits for competition to be actually distorted can have limited remedial effect. It is worth noting that, unlike in Article 101 TFEU case law, where the EU Courts have referred to the ‘capability’ and ‘likelihood’ of an agreement to restrict competition in order to distinguish between the two categories of harmful practices (restriction by object and restrictions by effect) in the Article 102 TFEU case law the same terminology does not seem to have the same purpose. On occasion, the Courts have referred to the tendency or the capability of conduct to produce negative effects as being synonymous.34 In Post Danmark II the Court, after referring to the capability of the conduct to restrict competition, held that ‘only dominant undertakings whose conduct is likely to have an anticompetitive effect on the market fall within the scope of Article [102 TFEU]’,35 echoing the General Court’s view in British Airways that it is sufficient to demonstrate that the conduct ‘tends to restrict competition, or, in other words, that the conduct is capable of having, or likely to have, such an effect’.36 The upshot of the evolution of the case law described above is that the default setting for establishing an exclusionary abuse is what may be termed a qualified effects standard, that is, an assessment that identifies as problematic only restrictions of the competitive process that Case C-95/04 P British Airways v Commission EU:C:2007:166, para 86. See E Rousseva, ‘Efficiency Defence under Article 102 TFEU: Retrospectives and Prospectives’ Concurrences No 2-2014. 31 Case C-549/10P Tomra Systems ASA v Commission EU:C:2012:221, para 79. 32 British Airways (n 30) para 145. This was reiterated in MEO (n 28) para 27. 33 TeliaSonera (n 6) para 64; Post Danmark II (n 6) para 66. 34 British Airways (n 30); Tomra (n 31). 35 Post Danmark II (n 6) para 67. 36 British Airways (n 19) para 293. 30

The special responsibility of dominant undertakings  245 are (at least potentially) harmful to consumers. The Court does not require proof of actual effects, nor does it require a quantification of the effects on consumer welfare, let alone an application of a price-cost as efficient competitor test for all conduct which falls under Article 102 TFEU. Instead, the special responsibility of dominant undertakings means that they may not engage in conduct whose potential effect is the exclusion of efficient competitors, that is, competitors that contribute to the effectiveness of competition. The Receding Space of Abuses by Object

B.

In view of the development described in the previous section a question that merits discussion is whether there is conduct that can be considered abusive by its nature, similar to the restrictions by object under Article 101 TFEU. In particular because the effects-based approach has been applied to conduct at times considered abusive by object (eg tying and margin squeeze).37 This trend continued with Intel. In this case the CoJ developed the case law on exclusivity rebates – traditionally considered abusive by their nature – by suggesting that the defendant has two ways of countering the presumption of the abusive nature of these rebates. First, it may provide supporting evidence that the rebates are incapable of harming competition. If so, then the Commission must carry out an effects-based analysis. Second, the undertaking may show that the conduct benefits from the efficiency defence.38 More generally, the Court left the impression that, for Article 102 TFEU to apply, at least potential anticompetitive effects need to be proven.39 In the current authors’ view, despite these developments, there is still a range of practices that would not warrant the assessment of effects.40 For instance, one question with which the Court in Intel did not deal with is whether there are naked restrictions (eg prohibiting customers to even try a competitor’s product), which as described in the Commission’s Guidance Paper are forms of conduct that ‘can only raise obstacles to competition’.41 Drawing on the Guidance Paper, Nazzini has argued that the case law contains a category of instances of ‘naked exclusion’.42 Examples of this kind of conduct include a patent holder’s misleading a patent office or a dominant company removing a rival’s product by buying it from retailers.43 To this one must add conduct that partitions the internal market, where the Court issued a very strongly worded reminder as to why this is an abuse: there can be no escape from the prohibition laid down in Article [102 TFEU] for the practices of an undertaking in a dominant position which are aimed at avoiding all parallel exports from a Member See respectively Case T-201/04 Microsoft v Commission EU:T:2007:289 and Telia Sonera (n 6). Intel (n 6) paras 138–9 for the first approach and para 140 for the efficiency defence. 39 Intel (n 6) paras 133–7. 40 Cf Lianos (n 12) and P Ibanez Colomo, ‘Beyond the “More Economics-Based Approach”: A Legal Perspective on Article 102 TFEU Case Law’ (2016) 53 Common Market Law Review 709. 41 Guidance Paper (n 7) para 22. 42 R Nazzini, ‘Abuse of Dominance: Exclusionary Non-Pricing Abuses’ in I Lianos and D Geradin (eds), Handbook on European Competition Law: Substantive Aspects (Edward Elgar Publishing 2013) 482–4. 43 Respectively Astra Zeneca (n 6) and Case T-228/97 Irish Sugar v Commission EU:T:1999:246, paras 226–35. However, at least with respect to Astra Zeneca it is also possible to read the CoJ’s ruling as justifying a finding of abuse on the basis of the likely effects of the conduct, see Astra Zeneca (n 6) paras 106–9. 37 38

246  Research handbook on abuse of dominance and monopolization State to other Member States, practices which, by partitioning the national markets, neutralise the benefits of effective competition in terms of the supply and the prices that those exports would obtain for final consumers in the other Member States.44

Taking all these claims together, they seem to suggest that there remains conduct that can be found abusive without demonstrating its anticompetitive effects. As effects do not matter for finding an abuse, dominant undertakings cannot argue in their defence that the effect of their conduct is negligible. From this perspective, one can argue that not engaging in such conduct entails greater responsibility than the duty not to engage in conduct which can produce potential negative effects. C.

Increasing Space for Special Responsibility

Although the scope of abuses that are anticompetitive by their nature might have shrunk, this does not mean that there is no scope for novel abuses and thus for an expansion of the special responsibility. In fact, an approach that is more attentive to the economic context allows the Court to clarify the scope of abuse of dominance and when necessary to identify new types of abusive conduct in a measured manner. For example, in Slovak Telekom, the CoJ distinguished between an abuse consisting of a dominant company’s refusal to share the infrastructure that it has developed for its own business needs with its downstream competitors and an abuse consisting of a dominant company’s offering of its product on the downstream market under unfair conditions which dissuade competitors.45 The Court clarified that while in the first scenario it has to be shown that the refused access or product is indispensable for competition downstream, in the second scenario this is not a requirement. The second scenario can possibly include situations where the dominant undertaking has already de facto offered the product, where it is obliged by regulation to do so,46 or where its business model cannot be profitably run without sharing infrastructure it has developed with competitors downstream.47 Similarly in the context of the licensing of standard essential patents (SEPs), the Court identified a setting where a refusal to license intellectual property rights could constitute an abuse but where the test set out in IMS Health need not be followed.48 Finally, as discussed in section III.B below, in UK Generics,49 the Court suggested a new form of abusive behaviour which consists of a series of anticompetitive agreements connected by a common exclusionary strategy. Thus far, this chapter has argued that the special responsibility placed on dominant undertakings has undergone transformation at the hands of the EU Courts. First, it was shifted from being a consideration relating to dominance to one relevant to considering the notion of abuse

Sot Lélos (n 20) para 66. In contrast Advocate General Colomer considered that there are no per se abuses (para 76). 45 Slovak Telekom (n 6) paras 49–50. 46 Slovak Telekom (n 6) and Case T-814/17 Lietuvos geležinkeliai AB v Commission ECLI:EU:T:2020:545. An appeal is pending at the time of writing. 47 As it is the case with Case AT.39740 Google Search (Shopping) (27 June 2017), upheld in Case T-612/17 Google and Alphabet v Commission (Google Shopping) ECLI:EU:T:2021:763. 48 Case C-170/13 Huawei Technologies Co. Ltd v ZTE Corp and ZTE Deutschland GmbH EU:C:2015:477, para 53. The general approach to refusals to license is set out in Case C-418/01 IMS Health v NDC Health EU:C:2004:257. 49 Generics (n 6). 44

The special responsibility of dominant undertakings  247 in the 1990s. Second, during the last decade the CoJ has gradually moved away from stating that this responsibility is to refrain from certain types of conduct to a qualified effects standard. The effects which matter are the likely harm to consumers or a meaningful harm to the competitive process. In US parlance, Article 102 TFEU no longer protects competitors.50 As a result restrictions by object are now few and far between. This, however, does not mean that the special responsibility shrinks. New types of behaviour have been identified as problematic for competition and have led to the expansion of the material scope of the special responsibility.

III.

THE LINK BETWEEN DOMINANCE AND ABUSE

There is a widespread understanding that Article 102 TFEU does not require that abuse be causally connected to the defendant’s dominant position. Support for this position is usually sought in Continental Can and Hoffmann-La Roche where the CoJ suggested that the question of a causal connection was immaterial since the strengthening of a dominant position can be an abuse regardless of the methods and means used to attain it.51 However, these rulings do not mean that dominance and abuse can or should be delinked. If that were the case, imposing a special responsibility on dominant undertakings would make little sense. Rather, the explanation given by Vogelenzang several decades ago about the relationship between dominance and abuse is scrupulous and remains valid today.52 According to Vogelenzang, as the abuse consists of two elements – an act and its consequences – the link between dominance and abuse can be manifested in two ways: in some cases as dominance being the means for the conduct to take place, but in others, the link is manifested in that the dominance is necessary either for the negative effects to arise or for their strengthening. Continental Can and Hoffmann-La Roche fall in the second category. With a view to seeking the rationale for the special responsibility, this section first explores dominance as a necessary condition for practices to be considered anticompetitive. It then shows that with the evolution of the assessment under Article 102 TFEU, as outlined in Section II.A, the degree of dominance has become increasingly important for assessing the harmful effects of practices and therefore also for the special responsibility dominant undertakings bear. Finally, the section shows that, even in cases where the scrutinized conduct could fall under the prohibition of both Articles 101 and 102 TFEU, the Commission and the EU Courts have chosen to apply Article 102 TFEU. As will be argued below, although in these instances the role of dominance for finding an infringement is far less obvious, the choice of the Commission and the EU Courts to apply Article 102 TFEU has nevertheless been informed by the understanding that dominant undertakings bear a special responsibility and was guided by healthy pragmatism.

This paraphrases Brown Shoe v US 370 US 294, 320 (1962). For a comparative account predating the evolution discussed here, see EM Fox, ‘We Protect Competition, You Protect Competitors’ (2003) 26 World Competition 149. 51 Continental Can (n 18) para 27; Hoffmann-La Roche (n 13) para 91. 52 P Vogelenzang, ‘Abuse of a Dominant Position in Article 86: The Problem of Causality and Some Application’ (1976) 13 Common Market Law Review 61. See also E Rousseva, Rethinking Exclusionary Abuses in EU Competition Law (Hart 2010) 75–9. 50

248  Research handbook on abuse of dominance and monopolization A.

The Relevance of the Degree of Dominance for the Assessment of the Effects of Abusive Behaviour

Many practices that could constitute abuse if carried out by a dominant undertaking would be unobjectionable when carried out by non-dominant firms. This is because, in the absence of market power, these practices are not capable of excluding competitors in an anticompetitive way. If non-dominant companies nevertheless adopt such market behaviour, its effects are likely benevolent. The most obvious example is predatory pricing, that is, pricing at a level that entails a sacrifice of profits in the short run to eliminate competition and increase profits in the longer run.53 In principle, any company, dominant or not, can sell below cost. However, a non-dominant company would not have the capacity to sell below cost long enough to drive competitors out of the market. Even if it succeeds to do so, it would be unable to recoup the losses incurred when it raises its prices.54 Other examples of practices that are unobjectionable if carried out by non-dominant companies are refusal to supply and margin squeeze, as well as other forms of constructive refusals to deal which now the CoJ itself commonly refers to as imposing unfair conditions.55 Absent dominance, a company taking such actions is more likely to lose than gain anything from such conduct. In TeliaSonera the CoJ held that, as a general rule, the degree of dominance is not relevant for finding an abuse because Article 102 TFEU does not envisage any variation in form or degree in the concept of a dominant position but also clarified that this does not mean that the undertaking’s strength in the market is irrelevant.56 The risk of anticompetitive effects, and thus of an abuse, is greater the more dominant the undertaking is. For example, when an undertaking is an unavoidable trading partner it faces no competition for at least a large part of the demand. The non-contestable part of the demand confers a competitive advantage which allows the company to exclude competitors without sacrificing profits. The greater the degree of dominance, the larger the non-contestable share and the more difficult for competitors to compete. Similarly, the greater the dominance, the greater the possibility of the dominant undertaking’s conduct to cover a greater part of the market and to foreclose. In Intel the Court expressly pointed out the market share covered by the conduct as an important element of finding an abuse in the context of rebates when the Commission is required to examine their effects.57 Moreover, some conduct only constitutes an abuse when the degree of market power is significant and where mere dominance will not suffice. The most obvious category is refusals to deal where the abuse may only be committed where the dominant undertaking holds infrastructure or goods that are indispensable for market access. The same applies to above-cost prices. According to AG Mengozzi the reason the Court has confirmed as abusive the setting of prices above costs was that the dominant undertaking expressed an intent to exclude, its M Motta, Competition Policy (Cambridge University Press 2004) 412. In the US, courts have acted on allegations for predatory pricing by companies with small market shares but recoupment has not been established; see eg Brooke Group Ltd v Brown & Williamson Tobacco 509 US 209 (1993), the alleged predator, Brown & Williamson, held only 12 per cent of the relevant cigarette market. 55 Slovak Telekom (n 6) para 50. 56 TeliaSonera (n 6) paras 80–81. 57 Intel (n 6) para 139. 53 54

The special responsibility of dominant undertakings  249 dominance was close to monopoly and the above cost price cuts were part of a large range of abusive conduct whose cumulative effect was harmful.58 Finally, the degree of dominance may affect what tests must be applied to establish an abuse. In Post Danmark II, the Court held that having a large market share, combined with structural advantages resulting from a statutory monopoly on a related market, obviates the need for applying an as-efficient-competitor test because such structure of the market makes the emergence of an as-efficient competitor practically impossible.59 Although, after Post Danmark II, in Intel the Court affirmed the principle that Article 102 TFEU applies as long as a dominant company excludes an equally efficient competitor, this aspect of the ruling in Post Danmark II can hardly be considered overruled given the specific situation described for which the test would be unsuitable. Bringing all these observations together, the degree of dominance frequently determines the scope of the special responsibility. B.

Abuse of Dominance when Conduct also Infringes Article 101 TFEU

When an anticompetitive agreement is concluded and one of the undertakings occupies a dominant position, formally the conditions of Articles 101 and 102 TFEU would be satisfied and in principle both provisions can apply. Early on, the CoJ held that in such circumstances the Commission is entitled to decide whether to proceed under Article 101 or Article 102 TFEU by considering the nature of the reciprocal undertakings entered into and the competitive position of the various contracting parties on the market(s) in which they operate.60 However, this recommendation was somewhat vague. Since then, three patterns have emerged. The first results from the Court’s gradual alignment of the effects-based approach in Articles 101 and 102 TFEU. The consequence of this has been that in these settings it makes no difference whether one pursues such conduct as an anticompetitive agreement or as an abuse of dominance. However, as a matter of practice the Commission has opted for using Article 102 TFEU. The second pattern is derived from the Commission’s practice of borrowing from the notion of restriction by object in Article 101 TFEU to inform a finding of abuse of dominance. This has been applied in particular in cases of agreements containing territorial restraints. These two developments are discussed in sections III.B.1 and III.B.2. They pose a challenge to the notion of special responsibility because one could argue that it would also have been possible to pursue such conduct using Article 101 TFEU. The third pattern is the parallel application of Articles 101 and 102 TFEU, which is discussed in section III.B.3.

Post Danmark I (n 6) paras 90 to 93, drawing on Compagnie maritime belge transports (n 4) and Irish Sugar (n 43). In Google Shopping (n 47) para 183, the General Court made specific reference to Google’s ‘superdominant position’ and indicated that this degree of market power meant that ‘it was under a stronger obligation not to allow its behaviour to impair genuine, undistorted competition on the related market for specialised comparison shopping search services’. 59 Post Danmark II (n 6) para 59: ‘On the other hand, in a situation such as that in the main proceedings, characterised by the holding by the dominant undertaking of a very large market share and by structural advantages conferred, inter alia, by that undertaking’s statutory monopoly, which applied to 70% of mail on the relevant market, applying the as-efficient-competitor test is of no relevance inasmuch as the structure of the market makes the emergence of an as-efficient competitor practically impossible’. 60 Hoffmann-La Roche (n 13) para 116. 58

250  Research handbook on abuse of dominance and monopolization 1.

The extension of the effects-based assessments from Article 101 TFEU to Article 102 TFEU Exclusive dealing and tying agreements are good examples of the first trend. It is settled case law that exclusivity agreements are analysed by reference to their effects when considered under Article 101 TFEU.61 Moreover, they are block exempted when the supplier’s and buyer’s market share each does not exceed 30 per cent and the duration of the non-compete clause does not exceed five years.62 The factors relevant to assessing market power under Article 101 TFEU are the same factors that are considered when assessing dominance.63 Under Article 102 TFEU exclusivity agreements, with or without rebates, have been traditionally considered anticompetitive by their nature (corresponding to ‘by object’ infringements under Article 101).64 However, as discussed in section II.B above, after the ruling in Intel exclusivity rebates have become subject to an effects-based assessment on the condition that parties submit during the administrative proceedings evidence that their practice is incapable of foreclosing competition. After the more recent ruling in Unilever the same principle also applies to exclusivity obligations without rebates.65 If they do so, the analysis under Article 102 TFEU becomes essentially the same as that carried out under Article 101 TFEU. Tying agreements are block exempted from the application of Article 101 if each party’s market remains below the 30 per cent market share thresholds, and if not block exempted can be found harmful only if their anticompetitive effects can be shown. While initially tying was considered abusive by its nature under Article 102 TFEU,66 after the rulings in Microsoft67 and Intel,68 there is little doubt left that its effects need to be shown for an abuse to be found. In sum, there has been a convergence in the assessment of exclusivity rebates and tying carried out under Articles 101 and 102 TFEU. A quick overview of the Commission practice demonstrates that, whenever a dominant undertaking embarked on exclusivity or tying practices, the Commission has a clear preference to apply Article 102 TFEU.69 Indeed, the application of Article 101 TFEU to such practices appears rather exceptional.70 However, it does not appear that this choice, especially in view of the evolution of the case law, is necessarily determined by the position of the companies or the nature of the reciprocal undertakings entered into as suggested in Hoffmann-La Roche in the 1970s. This is because the market power, which determines the position that an undertaking occupies, is relevant under both provisions. The

Case C-234/89 Delimitis v Henninger Bräu EU:C:1991:91; Case C-345/14 Maxima Latvija EU:C:2015:784. 62 Commission Regulation (EU) No 330/2010 of 20 April 2010 on the application of Article 101(3) of the Treaty on the Functioning of the European Union to categories of vertical agreements and concerted practices [2010] OJ L102/1. 63 See Commission Guidelines on Vertical Restraints [2010] OJ C130/1, paras 134–7. 64 Hoffmann-La Roche (n 13); Intel (n 6) para 137. 65 See Case C-680/20 Unilever Italia Mkt Operations EU:C:2023:33, paras 47–50. 66 Case C-53/92 P Hilti AG v Commission EU:C:1994:77; Case C-333/94P Tetra Pak International SA v Commission EU:C:1996:436. 67 Case T-201/04 Microsoft Corp v Commission EU:T:2007:289. 68 Intel (n 6). 69 See eg Case AT.40153 E-book MFNs (4 May 2017); Case AT.40099 Google Android (18 July 2018); Case AT.40220 Qualcomm (24 January 2018). 70 Article 101 TFEU has been applied to tying practices in Case AT.39230 Rio Tinto Alcan (20 December 2012) and Case Nos IV/34.073, IV/34.395 and IV/35.436 Van den Bergh Foods Limited [1989] OJ L246/1, however, in parallel with Article 102 TFEU. 61

The special responsibility of dominant undertakings  251 nature of the reciprocal undertakings entered into does not appear a determinative factor either. Distributors are often under economic pressure exercised by their suppliers but this has not been a reason not to apply Article 101 TFEU.71 Conversely, the Court has on many occasions held that a dominant undertaking may be liable for an abuse, even if the contract constituting the abusive behaviour is requested by the dominant undertaking’s customers.72 The Commission’s choice for Article 102 TFEU in these cases is pragmatic and appears to make enforcement more rational. First, the Commission has indicated that ‘[t]he degree of market power normally required for a finding of an infringement under Article 101(1) of the Treaty is less than the degree of market power required for a finding of dominance under Article 102 of the Treaty’.73 Therefore, if dominance is proven it can be assumed that the market power needed for the application of Article 101 TFEU is also met. As has been pointed out by commentators, where dominance is established, Article 102 TFEU can be seen as lex specialis and take precedence over Article 101 TFEU which would be lex generalis.74 Furthermore, because the notion of dominance is well developed in case law and practice with sufficiently clear parameters, the finding of dominance is more difficult to contest than the finding of a significant degree of market power required for the application of Article 101 TFEU. Second, as the finding of dominance implies that the degree of competition is limited, the proof of any further restriction of competition may often be sufficient to justify the application of Article 102 TFEU. Third, if the contractual practice is one of a series of anticompetitive practices in which the dominant undertaking is engaged – some involving agreements, some not – it makes more sense to rely on one provision which can catch all aspects of the behaviour. In such circumstances the application of Article 102 TFEU can better capture the entirety of the anticompetitive strategy, especially if the practices constitute a single and continuous infringement.75 Finally, there are a number of procedural efficiencies if Article 102 TFEU is applied. They derive from the fact that there is only one investigated company and thus one undertaking as a party to the administrative proceedings. This means, for example, having only one statement of objections, one decision, fewer confidentiality claims during the proceedings and in the context of the publication of the decision. Moreover, in certain circumstances if the dominant company has concluded various agreements with many different customers, even their identification and association to the proceedings could be challenging and entail significant administrative resources. Applying Article 102 TFEU instead of Article 101 TFEU in by object restrictions A similar pattern emerges in recent Commission practice where it appears that the notion of restrictions by object is imported into Article 102 TFEU. In BEH, the Commission had concerns that BEH was abusing its dominant position on the unregulated wholesale market 2.

See the discussion on this point in Rousseva (n 52) 435–40. Intel (n 6) para 137; Hoffmann-La Roche (n 13) para 89. 73 Rio Tinto Alcan (n 70) para 97. 74 O’Donoghue and Padilla (n 29) 51. One can draw a similar conclusion from the ruling in Unilever (n 65) paras 26–32. 75 See eg Google Android (n 69); Case AT 37.990 Intel (13 May 2009). 71 72

252  Research handbook on abuse of dominance and monopolization for the supply of electricity in Bulgaria, by including in its wholesale supply contracts destination clauses which limited the buyers’ freedom to decide where to resell the electricity they purchased.76 Depending on the destination clause some wholesalers could export the electricity, others could sell it only in Bulgaria. The Commission had concerns that these were territorial restrictions on the resale of electricity which impeded trade in electricity between Bulgaria and neighbouring Member States, and hindered the development of a wider regional wholesale market within the EU.77 Similarly, in Gazprom, the concerns were that territorial restrictions might have prevented the free flow of gas across the investigated Central and Eastern European gas markets and fragmented and isolated these markets along national borders.78 The reasoning of the Commission in the decisions in the two cases – both resolved with binding commitments on the dominant undertakings – was very similar. The Commission first observed that the practices at hand could amount to territorial restrictions which were in principle prohibited under Article 101 TFEU as having an anticompetitive object. Second, the Commission recalled that Hoffmann-La Roche allows it to choose which provision to apply. Third, the Commission relied on case law such as United Brands and Suiker Unie to illustrate that Article 102 TFEU could be used to prohibit practices that partition markets along national borders.79 In both decisions, the Commission made the point that behaviour that raises barriers to trade between Member States, and consequently to the internal market, constitutes a restriction by object which obviated the need of showing even potential effects. While pointing out that it was not required by law to show effects, the Commission nonetheless outlined the potential effects on competition resulting from the territorial restrictions.80 This line of Commission decisions is an interesting trend of development but the relevance of the special responsibility is far from being obvious. This is because in these cases dominance is neither a precondition for the practice to amount to an infringement, nor for the strengthening of its harmful effects. More generally, the question is why these agreements are pursued under Article 102 and not Article 101 TFEU.81 These Commission decisions have not been challenged by the parties and have not been reviewed by the EU Courts. Nevertheless, some reflections are offered next in attempting to identify the rationale for the Commission’s choice for Article 102 TFEU. First, it can be argued that the approach taken in these decisions is a logical continuation, or even a consequence, of the alignment of the effects-based assessment under Articles 101 and 102 TFEU. There is no obvious reason why the alignment should be reserved only for restric Case AT.39767 BEH Electricity (10 December 2015). ibid, para 67. 78 Case AT.39816 Upstream gas supplies in Central and Eastern Europe (24 May 2018), para 61. 79 In Joined Cases 40 to 48, 50, 54 to 56, 111, 113 and 114 to 173 Coöperatieve Vereniging ‘Suiker Unie’ UA and others v Commission EU:C:1975:174, a dominant sugar refinery was found to have violated Article 102 TFEU by threatening to stop sugar supply unless the distributors complied with its restrictive export policy. In Case 27/76 United Brands v Commission, EU:C:1978:22, a contractual provision imposed by United Brands on wholesalers not to sell bananas while they were still green had the effect of partitioning the market along national lines. 80 More recently a similar approach was applied in InBev with respect to restrictions on the import of beer products between the Netherlands and Belgium, where the Commission noted that also for the purposes of Article 102 TFEU such conduct is by its very nature capable of restricting competition. Case AT.40134 InBev beer trade restrictions (13 May 2019), para 219. 81 For a critical view of the application of Article 102 instead of Article 101 to various practices, see O’Donoghue and Padilla (n 29) 49–51. 76 77

The special responsibility of dominant undertakings  253 tions by effects especially in the light of Budapest Bank where the CoJ held that the same anticompetitive conduct can be regarded as having both as its object and effect the restriction of competition, within the meaning of Article 101 TFEU.82 The Court recalled that restrictions by object do produce effects but because of the sufficient degree of harm they entail, it is unnecessary to analyse and prove these effects.83 From this perspective, one can argue that, when a dominant undertaking concludes an agreement that is restrictive by object, the harmful effect is even greater, and as effects do not need to be proven under Article 101 TFEU, a fortiori they do not need to be demonstrated under Article 102 TFEU. More importantly, from a practical point of view, it makes little sense to assess the effects of conduct under Article 102 TFEU if it is clear that the same restriction of competition under Article 101 TFEU will be considered to reveal a sufficient degree of harm even in the absence of dominance. Another, probably simpler explanation is that the Commission exercises its prosecutorial discretion to determine whether to pursue an infringement action against the dominant undertaking only or also against all its co-contractors.84 As mentioned above, there are numerous procedural efficiencies to gain if Article 102 TFEU is the chosen legal basis. 3.

The collective effect of individual by object agreements for an infringement of Article 102 TFEU The EU Courts have not excluded the parallel application of Articles 101 and 102 TFEU to the same anticompetitive practice, provided that the application of Article 102 TFEU is not based on a mere recycling of the facts that were used to find an infringement of Article 101 TFEU and that there is an additional element justifying its application.85 In UK Generics the CoJ considered a scenario where each pay for delay agreement between the originator and generics manufacturers was a discrete infringement of Article 101 TFEU for which both sides were to blame. However, the Court explained that it was also possible that, in addition to a number of agreements each having anticompetitive effects, the dominant undertaking who held the process patent could be found to have infringed Article 102 TFEU ‘for the possible additional damage that strategy may cause to the competitive structure of a market’.86 The Court observed that the possible cumulative effects of the various agreements and the fact that the conclusion of those agreements was part of an overall contract-oriented strategy may have a significant foreclosure effect on the market, depriving the consumer of the benefits of entry into that market of potential competitors. Put differently, one can apply Articles 101 and 102 TFEU together, but each provision challenges distinct, albeit related, conduct. In the operative part of the judgment, it seems as if the Court makes the application of Article 102 TFEU conditional upon the finding of such additional harm.87 This approach appears much more Case C-228/18 Gazdasági Versenyhivatal v Budapest Bank and Others ECLI:EU:C:2020:265. ibid, paras 33–5. 84 It is worth recalling that the EU Courts have made clear that it is legitimate for the Commission to find an infringement against only one undertaking amongst several undertakings that are in a similar position. See British Airways (n 30) para 66; Joined Cases C-89/85, C-104/85, C-114/85, C-116/85, C-117/85 and C-125/85 to C-129/85 Ahlström and Others v Commission EU:C:1993:120, para 146. 85 Joined Cases T-68/89, T-77/89 and T-78/89 SIV and others v Commission EU:T:1992:38, para 360; Case T-51/89 Tetra Pak v Commission ECLI:EU:T:1990:41, para 24. 86 Generics (n 6) para 147. 87 Generics (n 6) para 172, indicating that Article 102 applies ‘provided that that strategy has the capacity to restrict competition and, in particular, to have exclusionary effects, going beyond the specific anticompetitive effects of each of the settlement agreements that are part of that strategy’. 82 83

254  Research handbook on abuse of dominance and monopolization convincing than the approach taken years ago in Van den Bergh Foods. In the latter case, the Court of First Instance (now, General Court) upheld the parallel application of both Articles to the same exclusivity agreements, justifying the application of Article 102 TFEU on top of Article 101 TFEU with the argument that the dominant position enabled the undertaking to induce its retailers to agree to the exclusivity clauses.88 It was far from clear why and how such inducement could trigger the application of Article 102 TFEU but would not fall under Article 101 TFEU. Economic pressure, and the absence of interest in the agreement, are also characteristic of practices falling under Article 101 TFEU and ‘inducement’ has been a reason for an agreement to be found restrictive of Article 101 TFEU elsewhere.89 4. Towards a coherent relationship between Articles 101 and 102 TFEU On the one hand, it can be argued that the evolution charted above ensures greater consistency in the application of competition law. On the other hand, the similarity of the analytical framework for dominant and non-dominant firms in cases where the theory of harm is vertical foreclosure or market disintegration appears to call into question the ‘special’ responsibility of dominant undertakings, since all firms have a duty to avoid conduct of that nature. However, it appears that the Commission’s choice to apply Article 102 TFEU instead of Article 101 TFEU is made with some practical considerations in mind, in particular the presence of well-established criteria under Article 102 TFEU and the convenience of addressing a single actor. Furthermore, when it is clear that the dominant undertaking is orchestrating a set of commercial contracts whose cumulative effect is anticompetitive while the other parties to these agreements are often disadvantaged by this market design, it makes no sense to find them responsible for anticompetitive agreements.

IV.

A RESPONSIBILITY TO BEHAVE IN A PARTICULAR MANNER

The dictum in Michelin establishing a special responsibility imposes a requirement to forbear certain forms of conduct. Dominant undertakings may not price below cost, enter into exclusivity agreements or engage in tying, for example. A dominant undertaking seeking to defend itself against new entry is also limited in the steps that it may take to keep rivals out.90 In certain limited instances, however, the special responsibility means that the undertaking is required to act in a specific manner to avoid being found to have abused its dominant position. An extreme manifestation of this point is found in Baltic Rail. The dominant undertaking dismantled some railway track ostensibly because of defects and, contrary to its regulatory duties to ensure safe and interrupted traffic, did not proceed to repair it. This had the result

88 Case T-65/98 Van den Bergh Foods v Commission EU:T:2003:281, para 162. In this case, a dominant ice cream supplier required its retailers to use the freezers it supplied exclusively for its brand of ice cream. As retailers had no space for a second freezer in their shop, this clause amounted to exclusivity agreements. In a commitment decision (Rio Tinto Alcan, n 70) the Commission pursued both lines in parallel. However, given the preliminary nature of competition assessments in this type of procedure such duplication appears justified. 89 See Rousseva (n 52) 455. 90 The so-called meeting competition defence is very narrowly cast.

The special responsibility of dominant undertakings  255 of foreclosing market access to a rival provider of freight services. According to the General Court, the dominant undertaking should have taken into account its responsibility under Article 102 TFEU and avoided eliminating all prospect of the Track being returned to service in the short term by means of a staggered reconstruction, by complying with its duty to minimise disturbance on the rail network by restoring the normal situation following a disturbance.91

Another example of this category of obligation is the ‘duty to negotiate’ established by the CoJ in Huawei. This case concerns disputes between the holder of a standard essential patent (SEP) and implementers who use the patent to develop products downstream. Some context is necessary to explain the role of competition law. The SEP holder in this case had participated in a standardization procedure and made a promise to license its SEPs on fair, reasonable and non-discriminatory (FRAND) terms. This procedure is designed to ensure that patent holders receive a fair reward and that implementers are incentivized to take licences and develop innovative products downstream. Moreover, it is designed to avoid two risks. First, the risk of hold-up (ie the SEP owner refusing to license until the price is high enough). Second, the risk of hold-out (ie the would-be licensee dragging out negotiations on the right price). In an ideal world this system maximizes welfare by optimizing incentives to innovate up- and downstream, by increasing the number of licences, thereby facilitating market entry, and by reducing transaction costs. However, many imperfections remain when this system operates, not least that the validity of some patents is disputed. In Huawei the SEP holder sought to secure an injunction against the implementer who was using the SEP without a licence. The question was whether seeking an injunction for a patent infringement could, given the legal and economic context of the case, constitute an abuse of a dominant position. According to the CoJ the FRAND commitment that the dominant undertaking made created a legitimate expectation on would-be licensees that there would be negotiations on a licence. Consequently, the Court held that the dominant SEP holder seeking an injunction would abuse its dominant position unless, before seeking an injunction, it tried to negotiate a licence. It started from the premise that the participation in a standard-setting procedure leading to a FRAND commitment created legitimate expectations that licenses would be granted.92 Consequently, ‘in order to prevent an action for a prohibitory injunction or for the recall of products from being regarded as abusive, the proprietor of an SEP must comply with conditions which seek to ensure a fair balance between the interests concerned’.93 The Court went so far as to prescribe a model for how the SEP holder and implementers should behave.94 The common denominator in these two examples is that, when a dominant undertaking has engaged in certain forms of conduct which are not themselves abusive (eg dismantling rail track for safety reasons or promising to license on FRAND terms), it then has a responsibility to act in specific ways so as not to harm competition.

93 94 91 92

Lietuvos geležinkeliai AB (n 46) para 223. Huawei (n 48) para 52. ibid, para 55. ibid, paras 61 to 71.

256  Research handbook on abuse of dominance and monopolization

V.

THE ROLE OF REGULATION

In Michelin the CoJ intimated that Article 102 TFEU was neutral: all dominant undertakings have a special responsibility, irrespective of the reasons for which they hold a dominant position. Indeed, even state-owned undertakings are bound by Article 102 TFEU.95 Similarly, undertakings subject to regulation are not absolved from the duty to comply with the competition rules if the regulation leaves them the possibility to take autonomous commercial decisions.96 However, case law suggests that the nature of the special responsibility is affected by regulation. This is particularly clear in cases where utilities markets have been opened to competition.97 For the purposes of this chapter, the significance of these developments is twofold: first, there exist a number of former monopolies that achieved their dominant position as a result of state support, and second, EU regulation is designed to open markets to competition and reverse the state privileges that dominant undertakings received for many years. These past and present regulatory efforts combined have an effect on the application of competition law. In the context of refusals to deal, recall that in the absence of regulation a dominant undertaking is only under a duty to deal with rivals under very restrictive conditions, notably where access to the dominant undertaking’s assets is essential. In cases where the dominance is protected by intellectual property rights the party requiring the licence must further undertake to introduce a new product.98 These criteria are rough and ready ways to balance competition and the incentive to innovate. Matters change in a regulatory context. First, if legislation imposes a duty to deal then ‘the necessary balancing of the economic incentives … has already been carried out by the legislature at the point when such a duty was imposed’.99 Similar considerations apply when there is (or has been) a legal monopoly and ‘the undertaking has not invested in the construction of the infrastructure, which was built and developed with public funds’.100 In Baltic Rail, both of these features were present. The undertaking was dominant in the market for the management of railway infrastructure and in the downstream market for the provision of rail transport services for oil products. It used its dominant position upstream by removing portions of railway track so that a rival in the downstream market was unable to provide services. In this setting the General Court held that it was unnecessary to apply the normal refusal to deal test because the railway lines had been built using public funds and in managing railway infrastructure the dominant undertaking was under statutory duties to give

See eg Case C-351/12 OSA, EU:C:2014:110, para 86. Matters differ if the state obliges an undertaking to act in a manner which constitutes abuse. In this context, only the state is infringing EU competition law. 96 Deutsche Telekom I (n 6) paras 80–81, 92. 97 See generally A Manganelli and A Nicita, The Governance of Telecom Markets (Palgrave 2020) ch 1. 98 Or contribute to technical development, see Microsoft (n 37) para 647, by reference to Article 102(b) TFEU. For criticism, see J Killick, ‘IMS and Microsoft Judged in the Cold Light of IMS’ 2004 1 Competition Law Review 23. 99 Lietuvos geležinkeliai AB (n 46) para 92. 100 ibid, para 93. 95

The special responsibility of dominant undertakings  257 access to third parties.101 In such circumstances, the infringement of Article 102 TFEU only requires a showing that the conduct impedes market entry.102 These considerations also applied in the CoJ’s assessment of exclusionary conduct in Slovak Telekom.103 The Court held that there is no need to show that access to the local loop was indispensable for entry into the downstream market of broadband internet access in cases where the dominant undertaking had already given access and what was contested was whether the terms of access were exclusionary.104 This was particularly so in this instance because the Slovak telecom regulator had (on the basis of EU Law) imposed an obligation on the incumbent to give access to its upstream infrastructure so as to stimulate competition in the downstream market for retail broadband services.105 Furthermore, the Court recalled that Slovak Telekom had rolled out its infrastructure under the protection of the state: Slovak Telekom had exclusive rights and funded the construction of its infrastructure via monopoly rents.106 In these circumstances, there is no need for a competition law assessment that balances the dominant undertaking’s property rights and society’s incentives to invest on the one hand with the promotion of competition on the other.107 Thus far, the regulatory framework has influenced the dominant undertaking’s special responsibility when it engages in (or refuses to engage in) commercial relations with rivals. It remains to be seen if there are other settings where the regulatory framework is decisive to the scope of the special responsibility. For example, in Post Danmark I, the Court indicated that, ‘[w]hen the existence of a dominant position has its origins in a former legal monopoly, that fact has to be taken into account’.108 However, it is not clear how the past existence of a legal monopoly can inform an assessment of whether below-cost prices or other similar forms of exclusionary conduct are capable of harming competition. It may be that a former legal monopoly might have deeper pockets and therefore would be able to sustain a predatory pricing campaign for longer, or that its longer presence in the market makes some customers unwilling to switch to a new entrant. However, these are specific aspects that can be considered case-by-case: indeed, as shown above the special responsibility of dominant undertakings depends in part on the legal and economic context.

VI. CONCLUSION The special responsibility of dominant undertakings today is quite different from that which existed at the time of Michelin. In its original formulation, it was an affirmation that domi-

ibid, paras 94 and 95. ibid, para 99. This approach ratifies the Commission’s position in the Guidance Paper (n 7) para

101 102

82.

Slovak Telekom (n 6). ibid, para 59. 105 ibid, paras 56–8. 106 ibid, para 55. 107 But see N Dunne, ‘Competition Enforcement and Regulatory Alternatives’ (OECD 2021) DAF/ COMP/WP2/WD(2021)22 10 expressing concerns that this judgment opens up dominant undertakings to the risk of limitless liability. We consider this risk remote: the Court would not, for example, countenance a finding of abuse when a dominant company infringes environmental law. 108 Post Danmark I (n 5) para 23. 103 104

258  Research handbook on abuse of dominance and monopolization nance was not prohibited but only its abuse. Gradually, the Court utilized the notion of special responsibility to shape the concept of abuse itself. As a result the subsequent evolution of the concept of special responsibility has been intertwined with the evolution of the notion of abuse. For this reason this chapter first offered a snapshot of this evolution, showing that most forms of conduct are currently assessed for the likelihood of their impact on the effectiveness of the competitive process. Save for naked abuses, exclusive supply obligations without rebates, and practices that partition the internal market, for the majority of conduct that falls within the purview of Article 102 TFEU one must show at least potential negative effects. For some forms of conduct which are deemed restrictive by object the dominant undertaking is now empowered to bring evidence to challenge this conclusion and require that potential negative effects be shown. As a result, the concept of special responsibility is no longer identified with a duty to abstain from a blacklist of conduct. This is not to say that the notion of abuse is now settled. Even within the new formulation, new kinds of abusive conduct may be identified for which special responsibility arises. From this perspective, neither the scope, nor the importance of the concept of special responsibility is shrinking. Second, the chapter sought to clarify why special responsibility is borne only by dominant undertakings. For this reason it explored the possible links between dominance and abuse. It illustrated that the special responsibility arises when dominance is either a necessary condition for a restriction of competition to take place or for stronger anticompetitive effects to occur. The chapter also reflected on the possible overlapping application of Articles 101 and 102 TFEU to contractual practices that involve a dominant player, which raises questions about the very essence of the concept of special responsibility: how can it be special if non-dominant undertakings have the same duties? However, it appears that in cases of overlap the emerging practice is to apply Article 102 TFEU. The chapter flagged a range of reasons which justifies this choice: dominance is a well-established concept which more easily explains the negative effects of contractual practices; the application of Article 102 TFEU allows for procedural efficiencies; Article 102 TFEU appears a more suitable tool in instances where the dominant firm orchestrates a course of conduct involving a series of agreements and unilateral conduct whose cumulative effect is exclusionary. Finally, while the concept of special responsibility suggests a uniform set of principles applicable in all abuse cases, this chapter showed that there are circumstances where the obligation is more onerous. First, the degree of dominance is relevant to identifying certain practices as abusive. Second, firms whose dominance results from the state granting them special rights and dominant firms which are subject to sector-specific regulation both bear additional special responsibilities that other dominant firms do not have. Finally, in limited circumstances, the special responsibility does not just prohibit certain conduct but requires dominant undertakings to act in ways that facilitate competition.

14. Abuse without dominance and monopolization without monopoly Or Brook and Magali Eben

I. INTRODUCTION Antitrust law is said to be concerned with the harm caused by the use of market power.1 Monopolies can create losses to society – such as inefficiencies, deadweight loss, wealth transfers, and lost investment incentives – which would not have occurred in the absence of significant market power.2 Market power, therefore, creates ‘the potential for competitive harm’.3 Indeed, both the EU and the US antitrust provisions on unilateral conduct only apply when there is a significant degree of market power.4 The assumption is that, but for market power, the harmful effects of the practice could not have occurred.5 Thus, even though some EU case law has dismissed the need to prove causality between the existence of market power and the harmful conduct,6 the link between market power and harm to competition cannot truly be severed.7 Despite the importance of the notion of market power, there is no consensus as to the degree and type of economic power which translates into harm to competition:8 First, while many firms have some ability to influence the conditions under which a product is offered, it is not Alan J Meese, ‘Market Failure and Non-Standard Contracting: How the Ghost of Perfect Competition Still Haunts Antitrust’ (2005) 1 Journal of Competition Law and Economics 23; Robert O’Donoghue, and Jorge Padilla, The Law and Economics of Article 102 TFEU (3rd edn, Hart Publishing 2020) 3. 2 Keith Cowling and Dennis C Mueller, ‘The Social Costs of Monopoly Power’ (1978) 88 The Economic Journal 727; Chiara Fumagalli and others, The Economics of Monopolisation and Abuse of Dominance (CUP 2018) 7–8; O’Donoghue and Padilla (n 1) 318. 3 George A Hay, ‘Market Power in Antitrust’ (1992) 60 Antitrust Law Journal 807, 808; see also Baker, citing Indiana Federation of Dentists: ‘market power is “but a surrogate for detrimental effects”’ (Jonathan Baker, ‘Competitive Price Discrimination: The Exercise of Market Power without Anticompetitive Effects (Comment on Klein and Wiley)’ (2003) 70 Antitrust Law Journal 643, 650. 4 As clear from the wording of the provisions themselves: abuse of dominance (Article 102 TFEU) and monopolize (Section 2 of the Sherman Act). 5 O’Donoghue and Padilla (n 1) 318; Fumagalli and others (n 2) 7–8; Louis Kaplow, ‘On the Relevance of Market Power’ (2017) 130 Harvard Law Review 1329. 6 Case C-6/72 Europemballage Corp and Continental Can Co Inc v Commission EU:C:1973:22, para 27; Case 85/76 Hoffmann-La Roche & Co AG v Commission EU:C:1979:36, para 91; Case C-457/10P AstraZeneca AB and AstraZeneca plc v Commission EU:C:2012:770, para 267. 7 Case C-333/94P Tetra Pak International v Commission ECLI:EU:C:1996:436, para 27; Case 27/76 United Brands Continentaal BV v Commission EU:C:1978:22, para 249. 8 See for an overview of the foundations of unilateral conduct laws in different jurisdictions and their relationship to market power: Katharine K Kemp, Misuse of Market Power: Rationale and Reform (CUP 2018) 32, 39; Douglas Bernheim and Randal Heeb, ‘A Framework for the Economic Analysis of Exclusionary Conduct’ in Roger D Blair and D Daniel Sokol (eds), The Oxford Handbook of International Antitrust Economics (OUP 2014) 4. 1

259

260  Research handbook on abuse of dominance and monopolization evident at which point market power is sufficiently ‘substantial’ to be capable of having an impact on competition that merits intervention.9 Second, it is unclear whether the EU and US antitrust prohibitions on unilateral conduct should apply where a firm only possesses relative power over the other economic actors active on a different (downstream or upstream) market. Relative power may be (ab)used to limit the opportunities of those actors, and thus their ability to profitably operate on their own markets. For example, a firm enjoying relative power may price its products above the competitive price range and jeopardise its customers’ profitability, knowing that all or some of its customers have no other alternatives. Yet, it is unclear whether and when such harm to the individual businesses’ interests should be classified as harmful to competition. This is by no means an automatic conclusion. The line between protecting competition and protecting competitors is a hard one to draw.10 Moreover, the relationship between market power and competitive harm is not straightforward. Equating market power with harm to competition does not only raise questions about what is meant by market power, but also what is meant by harm. The harm described in much of the modern antitrust literature reflects a very specific notion of competitive harm. What is protected is effective competition,11 where effectiveness is measured by the degree of competitive constraint in the market and its potential for efficient outcomes. It is possible, however, to conceptualise the harm in a broader manner, focusing on the detriment of other firms’ participation in their markets, regardless of the extent to which they contribute to efficient market outcomes. There is no – or at most a fragile – consensus on the harm to be tackled through abuse of dominance and monopolization provisions.12 This lack of clarity about the notion of market power and its link to theories of harm is evidenced not only by provisions falling formally within antitrust law, but also by the existence of provisions on unilateral conduct outside the narrow antitrust framework. Both EU and US laws have considered lower thresholds of market power, relative power, and provisions on gatekeepers as extensions to the antitrust rules on unilateral conduct. The extent to which these rules should be considered as competition law is not always clear and, most importantly, there is a distinct lack of clarity on the link between power and harm, and whether there is a dividing line between harm to effective competition and harm to competitors.

9 Einer Elhauge, ‘Defining Better Monopolization Standards’ (2003) 56 Stanford Law Review 259, 330. 10 See Pinar Akman, The Concept of Abuse in EU Competition Law: Law and Economics Approaches (Hart Publishing 2012) 134–44; Eleanor Fox, ‘Abuse of Dominance and Monopolization: How to Protect Competition without Protecting Competitors’ in Isabella Atanasiu and Claus-Dieter Ehlermann (eds), European Competition Law Annual 2003: What Is an Abuse of a Dominant Position? (Hart Publishing 2006) 69; Adrian Künzler, ‘Economics Content of Competition Law: The Point of Regulating Preferences’ in Daniel Zimmer (ed), The Goals of Competition Law: ASCOLA (Edward Elgar Publishing 2012) 208. 11 For an interesting analysis on the meaning of competition in Article 102 TFEU, see Thomas Eilmansberger, ‘How to Distinguish Good from Bad Competition under Article 82: In Search of Clearer and More Coherent Standards for Anti-Competitive Abuses’ (2005) 42 Common Market Law Review 129. 12 See, Eleanor Fox, ‘What Is Harm to Competition? Exclusionary Practices and Anticompetitive Effect’ (2002) 70 Antitrust Law Journal 371; Akman (n 10) 25, noting, inter alia, the following possible goals: ‘the maximisation of consumer welfare, economic efficiency, the dispersion of economic and political power, the unclogging of markets, the protection of easy entry into business and the protection of small businesses’ welfare’.

Abuse without dominance and monopolization without monopoly  261 This chapter explores to what end the EU and US prohibitions on unilateral conduct apply or could apply to situations falling short of market power. In doing so, it makes two significant contributions. First, it is the first to comprehensively compare the EU and US approaches to abuse without dominance and monopolization without monopoly. It analyses existing and proposed unilateral conduct rules in the two jurisdictions, found both in and outside of competition laws, which are based on lower market power thresholds, relative power, and gatekeeping power. Second, it shows that no evaluation of abuse without dominance or monopolization without monopoly can be complete without querying what kind of harm the laws are meant to address. Nonetheless, the chapter shows that, in practice, there is a distinct lack of clarity on the link between market power and the theory of harm. As such, this chapter concludes by advancing that the degrees and types of power addressed by the provisions on unilateral conduct should correspond to the harm the EU and US laws intend to fight and interests they seek to protect. This is linked to the question on the objectives of competition law, which is related but distinct, and falls outside of the scope of this chapter. The remainder of this chapter is structured as follows: section II focuses on EU law. After presenting the notion of abuse of dominance in its meaning under Article 102 of the Treaty on the Functioning of the European Union (TFEU), it discusses three situations in which EU law may apply to abuse without dominance: lower degrees of market power, national rules on relative power, and a new EU proposed legislation on gatekeepers. Section III shifts the focus to US law. It explores the limits of applying Section 2 of the Sherman Act and Section 5 of the Federal Trade Commission Act to situations short of monopoly, and reviews proposals of extending antitrust law to include exploitation by firms with relative or gatekeeping power. Section IV concludes.

II.

EU: ABUSE WITHOUT DOMINANCE

A.

Abuse of Dominance under Article 102 TFEU

Article 102 TFEU, the EU prohibition on the abuse of a dominant position, includes two substantive constituent elements: dominance and abusive conduct. Market power, thus, plays a dual role. Market power is essential for establishing dominance and underpins the theories of harm that justify calling conduct ‘abusive’. Article 102 TFEU is, thus, premised on the existence of market power which weakens the conditions for competition on the market. The abusive conduct itself ought to be understood by reference to the existing market power.13 Unlike US antitrust law, the concept of abuse in the EU competition law encompasses both exploitative conduct,14 in which a dominant firm uses its market power to impose on its trading partners conditions it could not impose in the absence of market power, and exclusionary conduct, by which a dominant firm prevents or hinders competition on the market. As a Commission official put it: ‘the [EU] Founding Fathers’ faith in competition as a process of rivalry between competitors was not strong enough to tolerate customer/consumer exploita See Hoffmann-La Roche (n 6) para 91. Accordingly, the imposition of ‘unfair’ prices and conditions is expressly listed as an abuse in Article 102(a) TFEU. In Continental Can (n 6) paras 20–27, the CoJ held that Commission could not validly pursue a policy of refusing to prosecute such cases. 13 14

262  Research handbook on abuse of dominance and monopolization tion in the short run’.15 The theories of harm under Article 102 TFEU are notoriously varied,16 even if, since modernisation, the Commission expresses a focus on impairments of ‘effective competition’ with ‘an adverse impact on consumer welfare’.17 As elaborated below, the possibility to sanction abuse of market power which does not amount to a dominant position arises in three contexts in the EU: first, the Commission, EU Courts, and national competition authorities and courts were often criticised for finding that a firm holds a dominant position in the meaning of Article 102 TFEU even in situations where such firm held a low degree of market power. The Article had been applied to firms that did not hold substantial market power, particularly by finding dominance despite comparatively low market share. Second, EU law leaves some room for the parallel application of national laws on relative power (such as abuse of economic dependence or superior bargaining power). Those provisions regulate different types of economic powers, which might overlap or conflict with the provision of Article 102 TFEU. Third, a new EU initiative calls to regulate the operation of large online platforms. Such initiative views gatekeepers as a new type of market power, which differs from the traditional concept of dominance. Each of these alternatives will be discussed in turn, highlighting the uncertainty in the manner in which they reflect the relationship between market power and the competitive harm. B.

Abuse of Dominance in Cases of Low Degree of Market Power

As mentioned, Article 102 TFEU only applies to firms holding a dominant position. According to the Court of Justice (CoJ), such a position confers on the dominant firm ‘a special responsibility not to allow its conduct to impair genuine undistorted competition on the common market’.18 The EU prohibition does not restrict conduct by a non-dominant firm that will allow the firm to achieve dominance or that causes harm to consumers. Despite the importance of the notion of dominance, it is neither defined in the EU Treaties and secondary laws nor does it correspond to solid economics literature. Rather, it was developed on a case-by-case basis to determine the point at which unilateral conduct becomes subject to scrutiny under Article 102 TFEU.19 In United Brands, the CoJ famously defined a dominant position as ‘a position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market by giving it the power to behave to an appreciable extent independently of its competitors, customers and ultimately of its consumers’.20 The reference to ‘appreciable extent’ is vital here, as no company is ever truly and completely independent:

Luc Gyselen, ‘Rebates: Competition on the Merits or Exclusionary Practice?’ in Claus-Dieter Ehlermann and Isabella Atanasiu (eds), European Competition Law Annual 2003: What Is an Abuse of a Dominant Position? (Bloomsbury Publishing 2006) 290. Although Article 102 TFEU applies to both types of conduct, in practice, priority has been given to exclusionary conduct. See the Guidance on the Commission’s Enforcement Priorities in Applying Article 82 of the EC Treaty to Abusive Exclusionary Conduct by Dominant Undertakings [2009] OJ C45/2 (Commission Article 102 TFEU Guidance) para 7. 16 Akman (n 10) 130. 17 Commission Article 102 TFEU Guidance (n 15) para 19. 18 Case 322/81 Nederlandsche Banden Industrie Michelin v Commission ECR 3461, para 57. 19 Richard Whish and David Bailey, Competition Law (OUP 2018) 187. 20 Emphasis added. United Brands (n 7) para 65. Also see Hoffmann-La Roche (n 6) paras 38 and 91. 15

Abuse without dominance and monopolization without monopoly  263 what matters is whether there is a lack of appreciable competitive constraint.21 The Court emphasised that such a position does not entail absolute market power (perfect monopoly) or preclude the existence of some competition. Rather, a firm will be deemed dominant when it has ‘an appreciable influence on the conditions under which that competition will develop’.22 In establishing the required degree of market power, the EU Courts have often relied on market shares as a proxy. In Hoffmann-La Roche, for example, the CoJ held that, although the importance of the market shares may vary from one market to another, ‘very large shares are in themselves, and save in exceptional circumstances, evidence of the existence of a dominant position’.23 Later, the Court clarified that 50 per cent or more of the relevant market can be deemed as the ‘very high market share’ indicative of dominance, establishing the ‘AKZO threshold’.24 The Commission and EU Courts have been criticised for finding dominance even when firms held only a comparatively low market share.25 For example, they have found that a market share between 40–50 per cent could be an indication of dominance when taken together with other market conditions.26 Moreover, even when the market share was below 40 per cent, dominance was possible, even if unlikely.27 In Gøttrup Klim, accordingly, the CoJ did not dismiss the possibility that a firm holding 36 and 32 per cent market shares in the two relevant markets could be classified as dominant.28 A similar trend is also observed at the national level, when national competition authorities and courts apply Article 102 TFEU and/or the national equivalent provisions. Studies have shown that national rules heavily rely on market shares as a proxy for dominance and explicitly set very low market share thresholds above which dominance is presumed.29 While the above demonstrates that there is no single market share threshold above which dominance is a given, and indeed there is no uniform direct relationship between market share and the level of market power, there seems to be some doubt whether market power is sufficiently substantial to merit intervention when market shares are below 50 per cent.30 Below 21 See Alison Jones and others, Jones & Sufrin’s EU Competition Law: Text, Cases, and Materials (OUP 2019) 302. 22 ibid, para 39. 23 ibid, para 4. 24 C-62/86 AKZO Chemie BV v Commission EU:C:1991:286, para 60. Also see AstraZeneca (n 6) para 176. 25 John Vickers, ‘Market Power in Competition Cases’ (2006) 2(supp1) European Competition Journal 13; Niels Gunnar and Helen Jenkins, ‘Reform of Article 82: Where the Link between Dominance and Effects Breaks Down’ (2005) 26 European Competition Law Review 606. 26 For example, in United Brands (n 7) para 105, dominance was established with respect to 45 per cent market share, and in Case T-219/99 British Airways v Commission EU:T:2003:343, para 183, with respect to 39.7 per cent. 27 Commission Article 102 TFEU Guidance (n 15) para 14. 28 Case C-250/92 Gøttrup Klim v KLG EU:C:1994:413, para 48. 29 College of Europe, ‘Study on the Impact of National Rules on Unilateral Conduct that Diverge from Article 102 of the Treaty on the Functioning of the European Union’ (2012), https://​ec​.europa​.eu/​ competition/​calls/​tenders​_closed​.html, accessed 14 December 2022, 12–13 and Annex VII. 30 William Landes and Richard Posner, ‘Market Power in Antitrust Cases’ (1981) 94 Harvard Law Review 946; American Bar Association, Market Power Handbook: Competition Law and Economic Foundations (2nd edn, ABA Publishing 2012) 111, n 49; Eleanor Fox, ‘The Market Power Element of Abuse of Dominance: Parallels and Differences in Attitudes’ in Claus-Dieter Ehlermann and Mel Marquis (eds), European Competition Law Annual 2007: A Reformed Approach to Article 82 EC (Hart

264  Research handbook on abuse of dominance and monopolization this threshold, there are likely competitive forces constraining the conduct of the firm. If so, the EU practice of establishing dominance mostly on the basis of a low market share proxy may result in a disconnect between the harm alleged and the market power. Kaplow explained that ‘where market power is negligible (…) a practice cannot be anticompetitive’,31 at least in so far as the definition of ‘anticompetitive’ is not the exercise of market power itself. The United Brands definition of dominance was also criticised for offering merely vague and contentious criteria that do not clarify the type of economic power caught by the provision.32 Situations of dominance clearly arise where a firm can act independently from its competitors, not facing appreciable competitive constraints. Indeed, the dominance test often equates such independence with the economic concept of substantial market power, that is, the ability profitably to raise prices above the competitive level for a significant period of time.33 This is reflected in the General Court’s judgment in AstraZeneca,34 as well as by the Commission’s Article 102 TFEU Guidance.35 For the most part, the focus in practice and scholarship is therefore on the ability to behave independently of competitors. Yet, according to United Brands, a firm would be deemed dominant also where it can act independently from its customers. It is hard to imagine that a firm could act completely independently of its customers, as this would require demand to be entirely inelastic. No firm can, in a sustained way, appreciably ignore its demand curve. The economy validity of the ‘customer’ part of the definition is therefore doubtful.36 Yet, it can open the door for reflections on the power of a firm over its customers. For example, the Commission has previously based a finding of dominance principally on an undertaking’s power over customers. In ABG/Oil the Commission held that BP was dominant given its power over its customers, despite the existence of other oil companies.37 In fact, the Commission stated that the seven producers in the case, including BP, were dominant over their customers, who had become completely dependent on them for the supply of a scarce product.38 The application of this logic to other cases is unclear. ABG/Oil was decided against the exceptional backdrop of the 1973 oil crisis, with its peculiar economic circumstances, and later cases have not classified firms as having a dominant position solely due to the dependence of their cusPublishing 2008); John Vickers, ‘Market Power in Competition Cases’ (2006) 2 European Competition Journal 14. 31 Kaplow (n 5) 1329. 32 Jones and others (n 21) 303–4. 33 Commission Article 102 TFEU Guidance (n 15) paras 10–11. Despite the wording of the guidance, substantial market power is not only limited to power over prices. It also includes the power to limit output or innovation, to decrease the quality of production, or limit the choice of goods or services. See, for example, Google Search (Shopping) (AT.39740) Commission Decision C(2017) 4444 final [2017] OJ C9/17 593 2, paras 168, 321–4; Case T-691/14 Servier and Others v Commission EU:T:2018:922, paras 1567–86. 34 Case T-321/05 AstraZeneca v Commission EU:T:2010:266, para 267, as indirectly affirmed by the CoJ in appeal in AstraZeneca (n 6) paras 177–81. Also see Jones and others (n 21) 302–3. 35 Commission Article 102 TFEU Guidance (n 15) para 11. 36 Simon Bishop and Mike Walker, The Economics of EC Competition Law: Concepts, Application and Measurement (Sweet & Maxwell 2010) 228; O’Donoghue and Padilla (n 1) 186. 37 ABG/Oil companies operating in the Netherlands (IV/28.841) Commission Decision [1977] OJ L117/1, 9. The CoJ has overturned this Commission’s decision based on the conduct and remedy analysis rather than on the assessment of dominance: Case 77/77 Benzine en Petroleum Handelsmaatschappij BV and Others v Commission [1978] ECR 1513. 38 ABG/Oil (n 37) 8–9.

Abuse without dominance and monopolization without monopoly  265 tomers. In Metro I, for example, the CoJ rejected the claim that brand dependence can be an indication of dominance when a firm held only 5–10 per cent market share.39 At the same time, the Court has left some room for finding dominance by virtue of power over a customer. In Metro II, it clarified that ‘shares of the market as insignificant as that held by Saba preclude the existence of a dominant position save in exceptional circumstances’.40 This raises the question as to the possible harm when a firm has power over a customer. In the ABG Oil decision, the Commission was concerned because, as part of its response to the oil shortage, BP (and others) had reduced its deliveries to ABG, so that ABG was ‘threatened with a total lack of product’, putting it at a competitive disadvantage.41 The harm to competition as a whole was not clear, and the Court held that BP did not violate Article 102 TFEU by prioritising its long-term customers over ABG.42 The application of Article 102 TFEU to firms with low market shares, as well as to firms which exclusively have power over customers, raises questions about the harm fought and interests protected by Article 102 TFEU – what is the anticompetitive harm alleged: harm to competition as a whole or to individual firms? C.

National Laws on Abuse of Relative Power and Unfair Competition

National laws of some EU Member States contain provisions limiting the abuse of relative power. Such provisions are found in national competition and private laws, and result in significant fragmentation across the EU.43 This has given rise to interesting questions as to the relationship between EU competition law and national laws regulating a superior position compared to a commercial partner, or an advantageous market position short of dominance, which are explored in this section. The EU Treaties have not determined the relationship between EU competition law and national laws, leaving this question to be determined by secondary EU law.44 For many years, this question was left open, and was determined on an ad hoc, case-by-case basis. Beginning with the Walt Wilhelm judgment of 1969, the CoJ’s case law had reflected the principle of supremacy of the EU competition law provisions, while accepting the parallel application of national laws when they did not prejudice the full and uniform application of EU law.45

Case 26/76 Metro SB-Großmärkte GmbH & Co KG v Commission ECLI:EU:C:1977:167, paras 17–18. 40 Emphasis added. Case 75/84 Metro SB-Großmärkte GmbH & Co KG v Commission ECLI:EU:C:1986:399, para 86. 41 ABG/Oil (n 37) 3. 42 BP v Commission (n 37) para 32. 43 For a comprehensive overview of national practices, see College of Europe (n 29). 44 Article 103(2)(e) TFEU (ex Article 83 EC). 45 Case 14/68 Walt Wilhelm v Bundeskartellamt ECLI:EU:C:1969:4. While this principle was first adopted with respect to Article 101 TFEU, in further cases it also applied to conflicts with Article 102 TFEU; see, for example, C-253/78 Procureur de la République v Giry and Guerlain ECLI:EU:C:1980:188, para 15; C-67/91 Dirección General de Defensa de la Competencia v Asociación Española de Banca Privada and Others ECLI:EU:C:1992:330, para 11; C-550/07P Akzo Nobel Chemicals Ltd and Others v Commission and Others ECLI:EU:C:2010:512, para 103; C-17/10 Toshiba Corporation v Úřad pro ochranu hospodářské soutěže ECLI:EU:C:2012:72, para 81. 39

266  Research handbook on abuse of dominance and monopolization This relationship was first codified by Regulation 1/2003 (the ‘Regulation’), decentralising the enforcement of EU competition law to national competition authorities.46 The Commission’s initial proposal for the Regulation had called to alter the Walt Wilhelm rule, in favour of granting greater primacy to EU competition law. Accordingly, it suggested that, in situations of conflict, Articles 101 and 102 TFEU would apply ‘to the exclusion’ of conflicting national laws.47 This was justified by the need to level the playing field across the EU. The Commission explained that it would be inconsistent with the notion of a single market if practices that are capable of affecting cross-border trade would be subject to different standards.48 Nevertheless, this relationship turned out to be one of the most contentious points in the drafting of Regulation 1/2003.49 After years of negotiations between the Member States, the Regulation adopted a more nuanced approach: first, Article 3(1) of the Regulation expresses the principle of parallel application, noting that national competition authorities must apply the provisions of EU competition law when they apply the national competition law prohibitions on anticompetitive agreements or abuse of dominance. Second, Article 3(2) of the Regulation expresses a qualified principle of supremacy of EU law, stating that the application of national competition law may not lead to the prohibition of agreements that may affect trade between Member States and are not prohibited according to Article 101 TFEU. Yet, with respect to abuse of dominance, Article 3(2) opted for a more limited obligation. It specifies that Member States are not precluded from adopting ‘stricter national laws which prohibit or sanction unilateral conduct engaged in by undertakings’.50 The limits of the supremacy of Article 102 TFEU are not fully clear. Article 3(2) of the Regulation does not explain what ‘stricter’ national competition laws are. Are they stricter in the sense of imposing more severe sanctions on abuse of a dominant position? Are they stricter because they may prohibit a wider range of abusive conduct by dominant firms? Are they stricter in terms of the required market share or market power, prohibiting abusive conduct by firms that would not be classified as having a dominant position according to EU competition law? The latter interpretation leaves considerable room for conflicting national competition laws on relative power. This interpretation was supported by the Commission’s Staff Working Paper preceding the adoption of the Regulation summarising the political negotiations with the Member States. Departing from its initial proposal, the Commission explained that the proposed wording of Article 3 – suggesting it will give primacy to ‘the abuse of a dominant Council Regulation (EC) No 1/2003 of 16 December 2002 on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty [2003] OJ L1 (Regulation 1/2003). 47 Commission, ‘Proposal for a Council Regulation on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty and amending Regulations (EEC) No 1017/68, (EEC) No 2988/74, (EEC) No 4056/86 and (EEC) No 3975/87 (“Regulation implementing Articles 81 and 82 of the Treaty”)’ COM (2000) 582 final (Commission’s Proposal for a Council Regulation on the implementation of the rules on competition). 48 ibid, Recital 8 and Explanatory Memorandum, para 14. 49 As declared, in particular, in the Presidency Progress Report, ‘Proposal for a Council Regulation on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty and amending Regulations (EEC) No 1017/68, (EEC) No 2988/74, (EEC) No 4056/86 and (EEC) No 3975/87’ 11848/00 RC 13 – COM(2000) 582 final, para 10. 50 This statement refers to ‘stricter national laws’ rather than to ‘stricter national competition laws’. Yet, the wording of Recital 8 clarifies that it regulates situations of conflict between EU and national competition laws. 46

Abuse without dominance and monopolization without monopoly  267 position’51 – implies that the Regulation would only exclude the application of national rules that are equivalent to those contained in Article 102 TFEU. Hence, national competition laws could remain applicable to the extent they would either apply to unilateral conduct of a non-dominant firm or would regulate unilateral conduct of a dominant firm in a stricter way than Article 102 TFEU.52 According to the Commission, this more lenient approach to national rules on unilateral conduct was warranted because Article 102 TFEU only applies to dominant firms, which are likely to raise prices and reduce output in order to increase their profits.53 Article 3, therefore, merely aimed at introducing a common minimum standard for practices covered by Article 102 TFEU and certain cooperation mechanisms of the network of competition authorities. While this interpretation is not explicitly articulated in the final version of Regulation 1/2003, it may be supported by the unique phrasing used in Article 3(2); while Article 3(1) refers to an ‘abuse prohibited by Article [102]’ (namely, to an abuse of a dominant position), Article 3(2) applies more generally to any ‘unilateral conduct’. This interpretation is also supported by the wording of Recital 8, noting that ‘[t]hese stricter national laws may include provisions which prohibit or impose sanctions on abusive behaviour toward economically dependent undertakings’.54 Third, in addition to the room left by Article 3(2) to national competition rules, Article 3(3) governs the adoption and application of other national laws. This paragraph was agreed upon in the final stage of the negotiations, after the Council had pressured the delegations to reach a compromise.55 Article 3(3) adds that ‘[w]ithout prejudice to general principles and other provisions of Community law’, the principles of parallel application and supremacy of EU competition law expressed by Articles 3(1) and (2) do not preclude the ‘application of provisions of national law that predominantly pursue an objective different from that pursued by Articles [101] and [102]’. Recital 9 clarifies the scope of Article 3(3), noting that because Articles 101 and 102 TFEU ‘have as their objective the protection of competition on the market’, Member States may adopt national legislation that ‘pursues predominantly an objective different from that of protecting competition on the market’. The Recital refers to the example of national laws on unilateral or contractual unfair trading practices, explaining that such legislation

Emphasis in original. Commission, ‘Staff Working Paper: Commission’s proposal for a new Council Regulation implementing Articles 81 and 82 – Article 3’, 2000/0243(CNS) 15–16. 53 ibid. 54 This sentence was introduced by the delegations of the Member States in November 2002; see Council of the European Union, Competition Working Party, ‘Proposal for a Council Regulation on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty and amending Regulations (EEC) No. 1017/68, (EEC) No. 2988/74, (EEC) No. 4056/86 and (EEC) No. 3975/87’, ST 13983 2002 INIT, https://​data​.consilium​.europa​.eu/​doc/​document/​ST​-13983​-2002​-INIT/​en/​pdf, accessed 14 December 2022. 55 Presidency Conclusions, ‘Barcelona European Council 15–16 March 2002’, SN 100/1/02 REV 1 https://​www​.consilium​.europa​.eu/​uedocs/​cms​_data/​docs/​pressdata/​en/​ec/​69871​.pdf, accessed 14 December 2022, para 17; Council of the European Union, Competition Working Party, ‘Proposal for a Council Regulation on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty and amending Regulations (EEC) No. 1017/68, (EEC) No. 2988/74, (EEC) No. 4056/86 and (EEC) No. 3975/87’ ST 12998 2002 INIT https://​data​.consilium​.europa​.eu/​doc/​document/​ST​-12998​ -2002​-INIT/​en/​pdf, accessed 14 December 2022. 51 52

268  Research handbook on abuse of dominance and monopolization pursues a specific objective, irrespective of the actual or presumed effects of such acts on competition on the market. This is particularly the case of legislation which prohibits firms from imposing on their trading partners, obtaining or attempting to obtain from them terms and conditions that are unjustified, disproportionate or without consideration.

Drawing the borderline between practices that fall under Articles 3(1) and (2) of the Regulation and those covered under Article 3(3) is not straightforward. According to the Commission, a practice should be classified as national competition law governed by Articles 3(1) and (2), where it combats excessive market power or protects smaller firms against larger competitors.56 This combines two types of theories of harm: in addition to ensuring the competitiveness of the market, it includes national provisions on abuse of superior bargaining power or significant influence aiming to regulate disparities of bargaining power in distribution relationships, including where neither the supplier nor the distributor holds a dominant position. This entails that national rules on abuse without dominance would have to be applied alongside Article 102 TFEU. By comparison, a practice should be classified as ‘other national rule’ falling under Article 3(3) if its objective is to regulate contractual relationships between firms by stipulating the terms and conditions that, for instance, suppliers must offer to distributors (rather than firms’ competitive behaviour on the market).57 Citing De Smijter and Kjølbye, the Commission argued that the main distinction between national competition law and other national laws is whether the aim of the provision is limited to regulating a contractual relationship with a view to protecting a weaker party against a stronger party or whether competition on the market is taken into account in the elaboration or the application of the rule. Each individual provision of national law should be examined, rather than the overall statute in which it is contained.58 While this distinction may have its merits, it does not directly flow from the wording of the Regulation. Vaguely declaring that it aims at the ‘protection of competition on the market’,59 the Regulation does not fully explain what interests are governed by EU competition law in general, and Article 102 TFEU, in particular. It does not detail the legal or economic theory of competition it aims to protect or what its limits are. De Smijter and Kjølbye, for example, advocated interpreting the Commission’s ‘competition on the market’ requirement as protecting competition in situations of market power. They held that a national rule should be deemed as national competition law when one of the conditions of its application is that ‘one of the parties possesses market power’.60 The limits between national competition and other laws also cannot be inferred from Article 3 of the Regulation itself. Article 3(2) and Recital 8 classify as national competition law, legislation that prohibits or sanctions ‘unilateral conduct’, including ‘abusive behaviour toward economically dependent undertakings’. This wording does not explicitly refer to the possession of (relative or market) power. Nor does it specify that the unilateral conduct should

56 European Commission, ‘Commission Staff Working Paper accompanying the Communication from the Commission to the European Parliament and Council, Report on the functioning of Regulation 1/2003’ COM(2009) 206 final, para 181. 57 ibid, para 162. 58 ibid, n 213, citing Eddy De Smijter and Lars Kjølbye, ‘The Enforcement System under Regulation 1/2003’ in Jonathan Faull and Ali Nikpay (eds), The EC Law of Competition (OUP 2007) para 2.59. 59 Regulation 1/2003, Recital 9. 60 De Smijter and Kjølbye (n 58) para 2.63.

Abuse without dominance and monopolization without monopoly  269 have an effect on competition, or that the national law would take into account the process or structure of competition in the market. In similar vein, Article 3(3) of the Regulation and Recital 9 suggest that a rule may be classified as other national law when it is either contractual or unilateral in nature. Such rules, it is stated, pursue a different objective, ‘irrespective of the actual or presumed effects of such acts on competition’.61 Thus, it is conceivable that unfair trading practices which a dominant firm is able to impose in light of its (relative or market) power may well fall under Article 3(3). Although the rules governing national laws on abuse without dominance and their relationship with Article 102 TFEU are unclear, Article 3 has received very limited attention in EU jurisprudence and scholarship.62 In practice, many Member States adopted national laws on abuse of dominance, abuse of superior market power, abuse of relative power (economic dependence), and unfair commercial practices. It is uncertain when those rules are aimed at the protection of competition as such, and when they are directed at the protection of competitors. The relationship between these different types of economic power and the harm which they seek to avoid is not yet settled. D.

Gatekeeping Powers

The European Commission’s Digital Markets Act (the ‘Act’) offers an additional way to regulate market power of firms, in a manner distinct from dominance. The Act introduces a new type of power. This power is premised on access to markets rather than a dominant position on a specific market. While the prohibition on abuse of dominance of Article 102 TFEU will remain applicable to the conduct of gatekeepers, it will be limited to situations of market power over competitors in a relevant market. The Act suggests imposing special obligations on an online platform if it operates as a gatekeeper, namely where (a) it has a significant impact on the internal market; (b) it operates a core platform service which serves as an important gateway for business users to reach end users; and (c) it enjoys an entrenched and durable position in its operations or it is foreseeable that it will enjoy such a position in the near future.63

Whereas the premise of Article 102 TFEU is that the use of market power may create harm to competition, the Act assumes that gatekeeping power can make markets less contestable and less fair.64 According to the Act, the characteristics of gatekeepers are likely to lead to serious imbalances in bargaining power, and, consequently, to unfair practices and conditions for business and end users and to lower contestability of markets.65 Hence, the Act imposes obligations and prohibitions to allow businesses and users to ‘reap the full benefits of the platform

Regulation 1/2003, Recital 9. For example, see Wouter Wils, ‘The Obligation for the Competition Authorities of the EU Member States to Apply EU Antitrust Law and the Facebook Decision of the Bundeskartellamt’ (2019) 3 Concurrences 58. 63 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), art 3(1). 64 ibid, Recitals 5, 9–11. 65 ibid, Recital 4. 61 62

270  Research handbook on abuse of dominance and monopolization economy and the digital economy at large, in a contestable and fair environment’.66 These include, but are not limited to: a prohibition on the combination of personal data collected across services unless there has been consent in accordance with the General Data Protection Regulation; prohibitions on requirements that business users use one or more of their services (such as their identification service) for the use of the core platform service the users actually want; a prohibition of self-preferencing in ranking services; obligations of portability; obligations to provide fair and non-discriminatory access to products such as app stores; and obligations to provide access to tools and information. It is worth contrasting the Act with the reform of national competition law in Germany, in which a new section in the Act against Restraints on Competition targets digital undertakings ‘with paramount significance across markets’.67 Both the EU’s Act and the new German provision on competition target a type of gatekeeping power, by which the firm plays an important role for the access by others to one or more markets. The German parliamentary documentation indicates that this was seen as ‘market power of a new nature’. It is not based on market shares, but rather on the key position a firm holds in comparison to other market participants, allowing it to determine how they interact with each other.68 While the EU adopts new regulation at least formally distinct from the EU provisions on competition law, the German legislator included its new prohibition into the German competition law. Like other competition law prohibitions, the new German provision would be enforced ex post by the national competition authority, based on a prohibition of abuse, and will enable the gatekeepers to provide objective justifications for its conduct. This formal inclusion in competition law creates at least an expectation that the German provision aligns with the same harm and interests as other provisions of national competition law. Yet, its objectives sound quite similar to the descriptions of the objectives of the EU’s Act which appears, at least formally, not to be competition law. For the most part, the power and conduct covered by the provision seem predicated on the ability of firms to access and compete on a market, though it is not evident that efficiency plays a role in the expected harm. There are, however, references to ‘fair’ competition and ensuring equality in the parliamentary discussions.69 It is noteworthy that the new German provision, differently from other competition law prohibitions, does not require the German competition authority to prove anticompetitive effect. This means the decisional practice will likely not be very useful in clarifying what the harm and interests are which enforcement is meant to address. The introduction of a new type of market power and the corresponding obligations in the Act and the German provision raises questions relating to the harm and interests covered by Article 102 TFEU compared to these new rules on gatekeeping power. Both the EU Act and the German new provision of competition law seem less concerned with efficiency-focused theories of harm, moving towards contestability, access, and even fairness. In contrasting See wording in Commission, ‘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 final, 2–3. 67 German Act against Restraints of Competition: Gesetz gegen Wettbewerbsbeschränkungen (GWB), Section 19a. 68 Deutscher Bundestag, ‘Beschlussempfehlung und Bericht des Ausschusses für Wirtschaft und Energie’ (13 January 2021) Drucksache 19/25868, 7. 69 Deutscher Bundestag, ‘Wettbewerbsrecht 4.0 – Digitales Monopoly beenden’ (28 October 2020) Drucksache 19/23698(neu) 1; Deutscher Bundestag (n 68) 8. 66

Abuse without dominance and monopolization without monopoly  271 them, it is not only apparent that the harm fought and interests protected may be varied and unclear, but moreover that the distinction between what is competition law and what is not (at least formally) is far from evident. It has also not yet been clarified – and, given the lack of effects analysis, is unlikely to ever be clarified in decisional practice or jurisprudence – how gatekeeping power creates these harms and whether they would have occurred in the absence of this power. Similar to the dividing line between national laws on the abuse of relative power and the abuse of dominance under Article 102 TFEU that have been examined in the previous section, this raises the question as to where harm to competition ends, and when concerns with fairness and the interests of individual firms begin.

III.

US: MONOPOLIZATION WITHOUT MONOPOLY

A.

Monopolization under Section 2 of the Sherman Act

Section 2 of the Sherman Act prohibits monopolization, namely the offence of creating or maintaining a monopoly through anticompetitive exclusionary means.70 The Section is principally concerned about the process of creating a monopoly through nefarious means. As such, it implies fault or at least negligence on behalf of a firm: it is the means to reach market power which are frowned upon, not the position itself.71 Similar to the EU prohibition, the provision has two substantive constituent elements: monopoly power and harmful conduct. Market power is vital not only to the identification of a monopoly position, but also to the assessment of the harmful effects. Causality is inherent in the language of Section 2 (‘to monopolize’),72 making it even more evident than in the EU that conduct has to be considered in light of existing market power. Exclusionary strategies (including input or customer foreclosure) are often only harmful because the firm has a degree of power on the market.73 A crucial difference with the EU lies in the type of conduct that is considered problematic under the provision. Section 2 does not cover exploitative conduct. As noted, the focus of the monopolization offence is on the acquisition of monopoly position through unacceptable means, namely through exclusionary strategies. The exploitation of monopoly power (ie charging prices or setting terms that would have not been accepted in the absence of market power), accordingly, is not prohibited.74 As in the EU, market shares are often used as a proxy for market power. At first glance, it may seem that US courts require a different type, or at least a different degree, of market power. Where in the EU a 50 per cent market share creates a presumption of dominance,75 US courts expect at least a 70 per cent market share to establish monopoly, with a finding of monopoly

15 US Code § 2. Herbert Hovenkamp, Federal Antitrust Policy: The Law of Competition and Its Practice (5th edn, West Academic Publishing 2016) 24, 361. 72 Elhauge (n 9) 330. 73 Hovenkamp (n 71) 357. 74 Verizon Communications Inc v Law Offices of Curtis V Trinko, LLP (2004) 540 US 398, 407. 75 See Section I‎ I(‎B) above. 70 71

272  Research handbook on abuse of dominance and monopolization below 60 per cent being doubtful.76 Nevertheless, such practical differences do not necessarily point to different theoretical frameworks of market power. As Hovenkamp rightly notes, ‘the antitrust “monopolist” is a dominant firm’.77 Monopoly power, dominance, and market power are variations on the same theme.78 In both the EU and the US, such market power refers to the firm’s relative freedom from competitive constraints, which in the US is generally described as the ‘power to control prices’.79 As noted above, market power is unlikely to be absolute, since most firms will face some demand elasticity, potential entry, or countervailing power at some point in their lifespan.80 As such, the difference between the EU and US lies not in the expectation of an absolute monopoly position, but rather in the degree of market power which is sufficient to bring the conduct within the scope of the provisions. By relying on different market share thresholds as indirect evidence of market power, the two jurisdictions deploy different presumptions to find sufficient dominance to bring a firm and its conduct within the scope of the law. The possibility to sanction unilateral conduct in cases falling short of mon