Research Handbook on Cartels 1839102861, 9781839102868

Table of contents :
Front Matter
Copyright
Contents
Figures
Tables
Contributors
Introduction to the Research Handbook on Cartels
PART I Fundamental concepts
1. The practical requirements of a successful cartel
2. The prevalence and injuriousness of cartels worldwide
3. An historical account of anti-cartel enforcement
4. The morality of cartel activity
PART II Substantive issues
5. The legal concept of a cartel
6. Cartels and the concept of the firm
7. The concept of a Single and Complex Continuous Infringement
8. Lawful cartels
9. Cartel facilitation
10. The concept of a ‘hub-and-spoke conspiracy’
11. Algorithmic tacit collusion
12. Smart contracts and cartel law
13. Cartels and the exchange of information
14. Buyer cartels
15. Crisis cartels
PART III Procedural issues
16. The difficulties of proving an unlawful cartel
17. Leniency programmes
18. Cartels and fines
19. Cartel activity and recidivism
20. The criminalization of cartel activity
21. Calculating cartel damages
22. Cross-border cartels and international cooperation
PART IV Cartel law in practice
23. The European Union
24. India and Pakistan
25. Hong Kong and Mainland China
26. Association of Southeast Asian Nations
27. Australia and New Zealand
28. South America
29. North America
Index

Citation preview

RESEARCH HANDBOOK ON CARTELS

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 Private Enforcement of Competition Law in the EU Edited by Barry J. Rodger, Miguel Sousa Ferro and Francisco Marcos Research Handbook on Cartels Edited by Peter Whelan

Research Handbook on Cartels Edited by

Peter Whelan Professor of Law, School of Law, University of Leeds, UK

RESEARCH HANDBOOKS IN COMPETITION LAW

Cheltenham, UK • Northampton, MA, USA

© Editor 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: 2023930160 This book is available electronically in the Law subject collection http://dx.doi.org/10.4337/9781839102875

ISBN 978 1 83910 286 8 (cased) ISBN 978 1 83910 287 5 (eBook)

EEP BoX

Contents

List of figuresviii List of tablesix List of contributorsx Introduction to the Research Handbook on Cartels1 Peter Whelan PART I

FUNDAMENTAL CONCEPTS

1

The practical requirements of a successful cartel Joseph E. Harrington, Jr

2

The prevalence and injuriousness of cartels worldwide John M. Connor and Robert H. Lande

22

3

An historical account of anti-cartel enforcement Susanna Fellman and Martin Shanahan

45

4

The morality of cartel activity Andreas Stephan

73

PART II

5

SUBSTANTIVE ISSUES

5

The legal concept of a cartel Okeoghene Odudu

90

6

Cartels and the concept of the firm Christopher Sagers

102

7

The concept of a Single and Complex Continuous Infringement Stefan Thomas

118

8

Lawful cartels Or Brook

135

9

Cartel facilitation Christopher Harding

156

10

The concept of a ‘hub-and-spoke conspiracy’ Mark Furse

169

11

Algorithmic tacit collusion Ariel Ezrachi and Maurice E. Stucke

187 v

vi  Research handbook on cartels 12

Smart contracts and cartel law Salil K. Mehra

213

13

Cartels and the exchange of information Daniel A. Crane

221

14

Buyer cartels Peter C. Carstensen

234

15

Crisis cartels Éric Barbier de La Serre

251

PART III PROCEDURAL ISSUES 16

The difficulties of proving an unlawful cartel Fernando Castillo de la Torre

271

17

Leniency programmes Cristina A. Volpin and Peerapat Chokesuwattanaskul

288

18

Cartels and fines Florian Smuda

308

19

Cartel activity and recidivism Catarina Marvão

332

20

The criminalization of cartel activity Bruce Wardhaugh

351

21

Calculating cartel damages Cento Veljanovski

369

22

Cross-border cartels and international cooperation Pierre Horna and Sophie Hunter

388

PART IV CARTEL LAW IN PRACTICE 23

The European Union Paolisa Nebbia

401

24

India and Pakistan Amber Darr

419

25

Hong Kong and Mainland China Thomas K. Cheng

438

26

Association of Southeast Asian Nations Rachel Burgess

454

27

Australia and New Zealand Julie Clarke

476

Contents  vii 28

South America Juan David Gutiérrez

500

29

North America Steven F. Cherry, Claire Bergeron, Lauren Ige, Mark Katz, Dajena Pechersky, Ernesto Duhne and Ivan Szymanski

521

Index550

Figures

2.1

Trend in cumulative cartel discoveries, worldwide, 1989–2018 (with 2026 projection)

25

2.2

Locations of price-fixing (with global cartels’ locations distributed)

31

2.3

Median affected sales, all guilty cartels, region of operation, 1990–2019

31

2.4

Mean recoveries, 332 international cartels, by type and region, 1990–2019

34

4.1

The relationship between cartel conduct and moral opprobrium 

84

11.1

‘Q-Learning to Cooperate’

209

17.1

The adoption of leniency programmes globally

291

18.1

Number of cases and undertakings over time

316

18.2

Gravity factor (left) and additional amount (right) over time

317

18.3

Average final fine per case (left) and per undertaking (right) over time

318

18.4

Percentage change between base fine and provisional fine (left) and between base fine and final fine (right)

320

18.5

Average leniency discount (%) per undertaking (left) and per case (right)

320

18.6

Histograms and kernel density estimates for provisional fine-sales ratios (left) and final fine-sales ratios (right)

323

18.7

Fraction of optimal deterrent fines depending on the ratio between overcharge rate and probability of detection

326

19.1

Average leniency reduction per year, for single (‘SO’) or multiple (‘MO’) offending firms (EU cartels, 1998–2020)

342

24.1

Comparing the enforcement strategies of CCI and CCP 2007–20

429

28.1

Number of cartels fined in South America, 2000–2020

512

viii

Tables

2.1

Mean average monetary penalties, international cartels, 1990–2019

32

2.2

Median average overcharges, by year and type

41

4.1

Proportion of respondents who felt cartels are harmful and should be prohibited 82

4.2

Do members of the public support the imprisonment of those responsible?

18.1

Process of fine calculation applied by the European Commission

315

18.2

Summary statistics

317

18.3

Share of deterrent fines (in %) for different values of overcharge level and probability of detection

325

19.1

Summary of empirical studies accounting for recidivism

346

19.2

List of multiple offending firms in EU cartels with LP applications (EU cartels, 1998–2020)

346

19.3

Set of all pairs of multiple offending firms with four or more interactions (EU cartels, 1998–2020)

349

24.1

Breakdown of orders passed by the CCI 2009–20

425

24.2

Breakdown of orders passed by the CCP 2007–20

426

26.1

ASEAN cartel provisions

464

26.2

Other ‘cartel’ provisions

469

28.1

Competition authorities in South America

503

28.2

The category of ‘agreements’ in South America’s antitrust laws

505

28.3

Cartel prohibitions clauses in South America’s antitrust laws

505

28.4

Types of horizontal agreements explicitly prohibited in South America’s antitrust laws 

506

28A.1 Competition agencies that enforce cartel laws in South America

ix

82

520

Contributors

Éric Barbier de La Serre, Member of the Paris Bar and Partner at Jones Day Claire Bergeron, Counsel, WilmerHale Or Brook, Associate Professor of Competition Law and Policy, School of Law, University of Leeds Rachel Burgess, Lecturer in Competition and Consumer Law, University of Southern Queensland Peter C. Carstensen, Fred W. & Vi Miller Chair in Law Emeritus, University of Wisconsin Law School Fernando Castillo de la Torre, Director of the Competition and Mergers Team, Legal Service, European Commission Thomas K. Cheng, Professor, Faculty of Law, University of Hong Kong Steven F. Cherry, Partner, WilmerHale Peerapat Chokesuwattanaskul, Lecturer in Law and Economics at the Faculty of Law, Chulalongkorn University Julie Clarke, Professor in Competition Law, Melbourne Law School, University of Melbourne John M. Connor, Professor Emeritus, Purdue University Daniel A. Crane, Frederick Paul Furth, Sr. Professor of Law, University of Michigan Amber Darr, Lecturer in Competition Law, University of Manchester and Senior Research Fellow, Centre for Law Economics and Society, University College London Ernesto Duhne, Partner, Santamarina + Steta Ariel Ezrachi, Slaughter and May Professor of Competition Law, University of Oxford and Director, Oxford University Centre for Competition Law and Policy Susanna Fellman, Professor of Business History at the School of Business Economics and Law at the University of Gothenburg Mark Furse, formerly Professor of Competition Law and Policy, School of Law, University of Glasgow Juan David Gutiérrez, Associate Professor at Universidad del Rosario and Co-Director of the Latin American chapter of Academic Society for Competition Law (‘ASCOLA’) Christopher Harding, Professor of Law, Aberystwyth University (RIP) Joseph E. Harrington, Jr., Patrick T. Harker Professor at The Wharton School of the University of Pennsylvania x

Contributors  xi Pierre Horna, Legal Affairs Officer, Competition and Consumer Policies Branch, Division on International Trade, United Nations Conference on Trade and Development (‘UNCTAD’) Sophie Hunter, Research & Project Assistant, Competition and Consumer Policies Branch, Division on International Trade, United Nations Conference on Trade and Development (‘UNCTAD’) Lauren Ige, Senior Associate, WilmerHale Mark Katz, Partner, Davies Ward Phillips & Vineberg LLP Robert H. Lande, Venable Professor of Law, University of Baltimore Catarina Marvão, Lecturer, Technological University Dublin and Affiliated Researcher, Stockholm School of Economics Salil K. Mehra, Charles Klein Professor of Law and Government, Temple University, James E. Beasley School of Law, Philadelphia Paolisa Nebbia, Référendaire, Court of Justice of the European Union Okeoghene Odudu, Professor of Competition Law, Faculty of Law, University of Cambridge Dajena Pechersky, Associate, Davies Ward Phillips & Vineberg LLP Christopher Sagers, James A. Thomas Distinguished Professor of Law at the Cleveland State University Martin Shanahan, Professor of Economic and Business History in the School of Business, University of South Australia and Elof Hansson Visiting Professor in International Business and Trade at Gothenburg University, Sweden Florian Smuda, Professor of Statistics and Empirical Economic and Social Research, University of Applied Sciences Koblenz Andreas Stephan, Professor of Competition Law, Centre for Competition Policy and School of Law, University of East Anglia, UK Maurice E. Stucke, Douglas A. Blaze Distinguished Professor of Law, University of Tennessee College of Law Ivan Szymanski, Competition & Compliance Associate, Santamarina + Steta Stefan Thomas, Professor of Law, Holder of the Chair in Private Law, Commercial Law, Competition and Insurance Law, Eberhard Karls University, Tübingen and Director of the Tübingen Research Institute on the Determinants of Economic Activity Cento Veljanovski, Managing Partner of Case Associates, IEA Law & Economics Fellow and Visiting Lecturer at the University of Buckingham Cristina A. Volpin, Competition Policy Expert at the Organisation for Economic Co-operation and Development (‘OECD’) Bruce Wardhaugh, Professor of Competition Law, Durham Law School

xii  Research handbook on cartels Peter Whelan, Professor of Law and Director of the Centre for Business Law and Practice, School of Law, University of Leeds

Introduction to the Research Handbook on Cartels Peter Whelan

When I was approached by Edward Elgar Publishing and asked to develop and edit a Handbook dedicated solely to the topic of cartels, I was more than happy to oblige. One of my first academic projects involved my acting as an editorial assistant on a Handbook published by Edward Elgar that focused on trans-Atlantic antitrust issues – and I enjoyed the process immensely. Almost 15 years later, I found myself at that point of my career when I was seriously contemplating devising my own Handbook with an ambition similar to that displayed by the collection that introduced me to the editing process. Edward Elgar thus came knocking at an opportune time. It was much appreciated that they were fully supportive of the concept that I had been forming in my mind: the creation of a collection that would aim to be an accessible yet comprehensive Handbook that appeals to competition law stakeholders across the globe, by providing in-depth analysis, from a variety of perspectives, of the most significant issues pertaining to the legal regulation of cartels. The current book represents the realisation of that concept. In order to achieve its aim, the Handbook (a) focuses on analysing the major economic, substantive and procedural issues that are encountered in cartel law; (b) contains chapters that, taken together, adopt theoretical, legal, economic, political and comparative outlooks on the issues under discussion; (c) provides insights into the experiences of various jurisdictions (with different degrees of maturity) in their dealing with cartels; and (d) captures the views of different stakeholders, including competition law academics, competition law economists, practising lawyers and competition law enforcers. The Handbook comprises 29 chapters and is divided into four parts. The first part is dedicated to fundamental concepts. The second part focuses on the analysis of various challenging substantive issues. The third part considers various procedural issues that are particularly relevant to cartel law. The final part of the book evaluates cartel law in practice; it presents a detailed evaluation of the experiences of various representative cartel law jurisdictions worldwide. The first part of the book (‘Fundamental Concepts’) comprises four chapters. Chapter 1 was written by Professor Joseph Harrington and focuses on the practical requirements for a successful cartel. Specifically, it analyses a variety of collusive practices in order to determine how they vary in their abilities to ensure success for cartelists and in their respective creation of legal risk. Chapter 2 was co-authored by Professor John Connor and Professor Robert Lande. Focusing on the economic injuries caused by price fixing, it outlines empirical evidence on the magnitude of such injuries globally and evaluates whether the deterrence objective of competition law enforcement is being pursued effectively. Chapter 3 provides an historical account of anti-cartel enforcement. Written by Professor Susanna Fellman and Professor Martin Shanahan, this chapter outlines and analyses the broad enforcement trends of cartel law over time, mostly across western countries. Chapter 4 was written by Professor Andreas Stephan and focuses on the moral wrongfulness of cartel activity. It identifies push and pull factors that influence whether cartel activity displays moral wrongfulness in any given jurisdiction. 1

2  Research handbook on cartels It is argued that these factors can be used to identify the types of cartel activity that would be viewed as morally offensive, as well as the policy tools that could potentially be employed to strengthen the moral opprobrium associated with such activity. The second part of the book (‘Substantive Issues’) comprises 11 chapters. Chapter 5, authored by Professor Okeoghene Odudu, explores the legal concept of a cartel. It analyses two problematic aspects of legal attempts to control cartelisation: determining what is objectionable with cartel conduct (consequentially or deontologically) and demonstrating that firms have conducted themselves other than independently. Chapter 6 was written by Professor Christopher Sagers. With reference to US and EU jurisprudence, it examines the issue of distinguishing cartels from the concept of the ‘firm’. Chapter 7 focuses on the concept of a ‘single and complex continuous infringement’ that has been developed by the EU courts. Written by Professor Stefan Thomas, this chapter explores the nature of the said concept, how such an infringement can be established and its implications for both public and private enforcement of EU cartel law. Chapter 8 examines ‘lawful’ cartels. That chapter, authored by Dr Or Brook, establishes that not every agreement among horizontal competitors would be unlawful under a given regime’s competition law and argues that a common framework is lacking in the assessment of the lawfulness of ‘cartels’. Chapter 9 focuses on the issue of cartel facilitation. Written by Professor Christopher Harding, this chapter evaluates the various ways in which cartel facilitation may occur and how cartel facilitation can be viewed as a species of delinquency that should be punished under cartel law. Chapter 10 was authored by Professor Mark Furse; it examines the issue of ‘hub-and-spoke’ cartels. With reference to key case law from the US and the EU, it critically evaluates the various ways in which a ‘hub-and-spoke’ arrangement may be dealt with for the purposes of competition law. Chapter 11 considers the issue of algorithmic tacit collusion. Written by Professor Ariel Ezrachi and Professor Maurice Stucke, the chapter argues that algorithmic tacit collusion is indeed possible and that antitrust enforcers should pay it increasing attention in practice. Chapter 12 is a short chapter that explores the evolving issue of smart contracts in cartel law. In that chapter, Professor Salil Mehra posits that smart contracts have some potential to unsettle cartel law and that such law must evolve if it is to serve consumers and the public interest. In Chapter 13, Professor Daniel Crane explores the issue of the exchange of information between competitors. Noting that the exchange of information between competitors can have both pro- and anti-competitive effects, it argues that such an exchange is not likely to cause antitrust issues when it is properly structured and that antitrust enforcers should pay careful attention to the relatively clear guidance that exists with respect to evaluating the lawfulness or otherwise of information exchange between competitors. Written by Professor Peter Carstensen, Chapter 14 considers the issue of ‘buyer cartels’. Its central thesis is that buyer cartels generate significant risks for the maintenance of a competitive market process and cannot be justified in a persuasive manner. In concluding the second part of the book, Chapter 15 moves the book’s focus to the issue of ‘crisis cartels’. In that chapter, Eric Barbier de La Serre argues that a lenient approach to crisis cartels on behalf of antitrust enforcers is very difficult to rationalise and that a structural crisis is best dealt with through mergers and specialisation agreements. The third part of the book (‘Procedural Issues’) comprises seven chapters. In Chapter 16, Fernando Castillo de la Torre explores the difficulties involved in proving an unlawful cartel. With a focus on the EU, the chapter examines issues such as the burden and standard of proof, the admissibility of evidence, evidence assessment, the reliance upon direct versus circumstantial or indirect evidence and the evidential value that one should give to leniency statements.

Introduction  3 Chapter 17 considers leniency applications. In that chapter, Dr Cristina Volpin and Dr Peerapat Chokesuwattanaskul emphasise the interdependence that exists between a well-functioning enforcement system and an effective leniency policy and identify the (six) common features that successful leniency programmes share. Chapter 18 focuses on cartel fines, and specifically their deterrent function. There Professor Florian Smuda compares the cartel fines imposed by the European Commission with those dictated by economic deterrence theory, finding that there is in fact a ‘deterrence gap’ in EU cartel law. In Chapter 19, Dr Catarina Marvão examines the issue of recidivism in anti-cartel enforcement. With reference to the empirical data, it argues that while in the US ‘true recidivism’ appears to have been eliminated by the antitrust authorities, it seems to be rising within the European Union. Chapter 20 was written by Professor Bruce Wardhaugh and focuses on the criminalisation of cartel activity. Its central argument is that the pragmatic rationales for criminal cartel sanctions are dependent upon their normative counterparts and thus, when there is no normative basis for cartel criminalisation, a practical one should also be lacking. With Chapter 21, Dr Cento Veljanovski moves the book’s focus to private enforcement, in particular the issue of calculating cartel damages. With a focus on the EU and the UK, it considers a multitude of relevant issues such as: the types of damages that may be sought by the victims of cartel activity; causation; passing on; and how to determine cartel overcharges, volume effects and the duration of cartels. Chapter 22 completes the third part of the book with an analysis of cross-border cartels and international cooperation. Co-authored by Dr Pierre Horna and Sophie Hunter, that chapter examines how game theory represents a useful analytical tool for antitrust authorities in developing countries to evaluate the obstacles they may face in ensuring international cooperation in the context of anti-cartel enforcement. The fourth and final part of the book considers how competition law has been implemented in practice in a cross-section of jurisdictions globally (‘Cartel Law in Practice’). Comprising seven chapters, it examines in turn the cartel law practice of the European Union (Chapter 23 by Dr Paolisa Nebbia), India and Pakistan (Chapter 24 by Dr Amber Darr), Hong Kong and Mainland China (Chapter 25 by Professor Thomas Cheng), the Association of Southeast Asian Nations (Chapter 26 by Rachel Burgess), Australia and New Zealand (Chapter 27 by Professor Julie Clarke), South America (Chapter 28 by Dr Juan David Gutiérrez) and North America (Chapter 29 by Steven Cherry, Claire Bergeron, Lauren Ige, Mark Katz, Dajena Pechersky, Ernesto Duhne and Ivan Szymanski). Read together, those particular chapters provide important practice-focused insights that, no doubt, will be appreciated by antitrust specialists across the globe. I would like to finish this introduction to the book on a personal note. Sadly, not long before the manuscript was finalised and sent to the publisher, one of its contributors, Professor Christopher Harding, passed away. I knew Chris personally and for many years have been a huge fan of his work on cartels. I am honoured that this collection contains what proved to be his last academic piece. This book is dedicated to his memory.

PART I FUNDAMENTAL CONCEPTS

1. The practical requirements of a successful cartel Joseph E. Harrington, Jr

I. INTRODUCTION In the context of a market, collusion is coordinated conduct among firms with the intent to constrain competition for their mutual benefit. A cartel is what we call that collection of firms. This chapter examines what it takes for firms to succeed in their collective endeavour to circumvent competition. As just defined, collusion is an economic phenomenon which existed before the adoption of antitrust or competition laws.1 Indeed, it was the egregious displays of collusion that inspired those laws. If one is to understand the practices that make for a successful cartel, it is necessary to appreciate how those practices vary in their efficacy in establishing and sustaining a supracompetitive market outcome but also in their legal exposure, with the commensurate risk of shutdown by the authorities and the imposition of financial penalties. In the US, EU and many other jurisdictions, firms engage in unlawful collusion when they have an agreement to unreasonably restrain trade. According to US jurisprudence, firms have an agreement when they have a ‘unity of purpose or a common design and understanding, or a meeting of minds’2 or ‘a conscious commitment to a common scheme designed to achieve an unlawful objective’.3 This perspective is echoed by the EU Courts, which have defined an agreement as ‘joint intention’4 or a ‘concurrence of wills’.5 In short, an agreement is a mutual understanding among firms to constrain competition. Through the lens of economic theory, an agreement corresponds well with an equilibrium which sustains a supracompetitive outcome.6 In order to prove there is an agreement, a common requirement is that ‘there must be evidence that tends to exclude the possibility that the [firms] were acting independently’.7 While many avenues have been pursued to argue that firms’ conduct could not have been reached independently, ‘few courts have found a conspiracy without some evidence of communication tending to show an agreement’.8 When firms communicate in a manner pertinent to future One of the earliest recorded cartels was formed in 1470 to control the price of alum (A. Günster and S. Martin, ‘A Holy Alliance: Collusion in the Renaissance Europe Alum Market’ (2015) 47 Review of Industrial Organization 1). 2 American Tobacco Co. v United States, 328 US 781 (1946). 3 Monsanto Co. v Spray-Rite Serv., 465 US 752 (1984). 4 Case 41-69 ACF Chemiefarma NV v Commission [1970] ECR 661, 663. 5 Case T-41/96 Bayer AG v Commission [2000] ECR II-3383, para 69. 6 D. Yao and S. DeSanti, ‘Game Theory and the Legal Analysis of Tacit Collusion’ (1993) 38 Antitrust Bulletin 113; G. Werden, ‘Economic Evidence on the Existence of Collusion: Reconciling Antitrust Law with Oligopoly Theory’ (2004) 3 Antitrust Law Journal 719. 7 Monsanto Co. v Spray-Rite Serv. Corp., 465 US 752 (1984), 768. 8 H. Hovenkamp, Federal Antitrust Policy: The Law of Competition and Its Practice (5th edn, West Academic Publishing, 2016) 243. 1

5

6  Research handbook on cartels conduct (either expressing intentions or conveying information relevant to intentions), they create the legitimate concern that they have influenced each other’s conduct and, therefore, their behaviour was not reached independently.9 With regard to liability and the attendant implications of possible shutdown and penalization, it is useful to recognize three broad categories of collusive practices. First, there is collusion so egregious that, once discovered, conviction is reasonably assured. Firms who pursue that brand of collusion know it is essential not to be discovered. Second, there is collusion that is legally problematic in that, once discovered, there may or may not be prosecution, and conviction is highly uncertain should a case be brought. Less legal exposure is the appeal of these practices, but they are generally less effective as a means of collusion. Finally, there is collusion that is grudgingly accepted by courts as lawful because there is no remedy.10 An example is price signalling, whereby a firm conveys an invitation to collude through its price. Another form of lawful collusion arises when government-mandated price caps act as a focal point.11 An episode during a German spectrum auction offers an example of lawful collusion.12 The German government was selling ten blocks of spectrum with the rule that any new bid had to be at least 10 per cent higher than the prevailing bid. The bidders were Mannesman and T-Mobile. Mannesman began with a bid of 20 million (Deutsche Marks per megahertz) on blocks 1–5 and a bid of 18.18 million on blocks 6–10. Why 18.18? With the minimum 10 per cent rule, the next highest bid that T-Mobile could submit on blocks 6–10 would be 20 million. A candidate explanation is that Mannesman was signalling to T-Mobile that each company should win five blocks at 20 million. Subsequent behaviour substantiates that intent and that T-Mobile correctly interpreted the signal. In the next round of bidding, T-Mobile bid 20 million on blocks 6–10 and did not submit a bid on blocks 1–5. There were no subsequent bids and each of them ended up with five blocks at a price of 20 million. These two bidders lawfully colluded. If, prior to the auction, the two companies had verbally shared the plan that was implicit in Mannesman’s bid, they would have unlawfully colluded. The point is that collusion is an economic phenomenon whose legal status lies on a continuum, ranging from clearly illegal to clearly legal, with grey areas between. How firms communicate is determinative of their legal exposure – the likelihood they are detected, prosecuted, and convicted – and whether they are able to coordinate on and sustain a supracompetitive

When they have the object or effect of distorting competition, so-called concerted practices are also prohibited by EU competition law. Such concerted practices involve a form of coordination between firms that falls short of an agreement properly understood. This chapter focuses solely on practices that align with the notion of an agreement. 10 ‘Courts have noted that […] individual pricing decisions (even when each firm rests its own decisions upon its belief that competitors do the same) do not constitute an unlawful agreement under section 1 of the Sherman Act […] [T]hat is not because such pricing is desirable (it is not), but because it is close to impossible to devise a judicially enforceable remedy for “interdependent” pricing. How does one order a firm to set its prices without regard to the likely reactions of its competitors?’: Clamp-All Corp., 851 F.2d 478, 484 (1st Cir. 1988). 11 See C. Knittel and V. Stango, ‘Price Ceilings as Focal Points for Tacit Collusion: Evidence from Credit Cards’ (2003) 93 American Economic Review 1703 (credit card markets); and C. Genakos, P. Koutroumpis and M. Pagliero, ‘The Impact of Maximum Markup Regulation on Prices’ (2018) 66 Journal of Industrial Economics 239 (fruit and vegetable markets). 12 P. Klemperer, Auctions: Theory and Practice (Princeton University Press, 2004). 9

The practical requirements of a successful cartel  7 outcome. Both the legal and the economic dimensions are crucial in determining a cartel’s success.13

II.

COLLUSIVE OUTCOMES

Collusion ultimately manifests itself in terms of a market outcome that raises firms’ profits and, almost universally, lowers consumer welfare. In principle, it is possible that collusion can be beneficial to both firms and consumers. If firms coordinate on higher prices but continue to make independent decisions on other strategic variables (such as service and advertising) then, given the higher profit from selling another unit (due to charging the collusive price), competition will intensify with respect to those other variables.14 It is possible that the intensification of competition on these other variables could more than offset the harm from higher prices so that consumers are better off and, at the same time, not all collusive rents are competed away so that firms remain better off.15 Theoretical possibilities aside, firms agreeing to a common price or a market allocation scheme (whereby each firm serves a segment of the market without competition) are considered sufficiently unlikely to produce substantial mitigating benefits that they are a per se (or by object) offence, rather than subject to the rule of reason (or by effect). In this section, we review some of the supracompetitive outcomes that cartels have produced. Let’s start with the canonical one: firms directly coordinate on the prices that consumers pay (that is, transaction prices). For example, members of the citric acid cartel agreed to a higher price for all customers except a firm’s largest customers who received a discount of three percent.16 Rather than coordinate on a particular price, they may instead agree to a range of prices. For example, trade associations have conspired with its members to impose a minimum price that all firms should charge.17 Even though higher transaction prices are the goal, some cartels have chosen instead to coordinate on prices that influence transaction prices. In industrial markets where firms set list prices and buyers routinely negotiate discounts off of them, cartels in urethane and cement markets agreed to a higher list price but not to discounts and, therefore, not to transaction

13 For surveys of collusive theories and empirical evidence relevant to cartel success, see M. Levenstein and V. Suslow, ‘What Determines Cartel Success?’ (2006) 44 Journal of Economic Literature 43; and J. Asker and V. Nocke, ‘Collusion, Mergers, and Related Antitrust Issues’, in K. Ho, A. Hortascu and A. Lizzeri (eds), Handbook of Industrial Organization: Volume 5 (Elsevier, 2021). 14 When firms coordinate with respect to some, but not all, of the ways in which they compete, it is referred to as semi-collusion; see F. Steen and L. Sørgard, ‘Semicollusion’ (2009) 5 Foundations and Trends in Microeconomics 153. 15 This possibility is shown in C. Fershtman and A. Pakes, ‘A Dynamic Oligopoly with Collusion and Price Wars’ (2000) 31 RAND Journal of Economics 207, for when firms collude in prices and compete in product quality. 16 This and other cases can be found in J. Harrington, ‘How Do Cartels Operate?’ (2006) 2 Foundations and Trends in Microeconomics 1. Also see J. Connor, Global Price Fixing (2nd edn, Springer, 2008), and R. Marshall and L. Marx, The Economics of Collusion: Cartels and Bidding Rings (The MIT Press, 2012). 17 For cases, see J. Harrington, ‘Heterogeneous Firms Can Always Collude on a Minimum Price’ (2016) 138 Economics Letters 46.

8  Research handbook on cartels prices.18 The presumption was that these higher list prices would ultimately result in higher negotiated prices. Given that buyers observed list prices, collusion could then affect bargaining and thereby transaction prices. However, in the EU trucks cartel, manufacturers coordinated on an internal list price which was never observed by customers.19 A higher internal list price has been argued to raise transaction prices by affecting a firm’s internal pricing process determining dealer prices. Some cartels agreed to restrict discounts, such as eliminating a cash discount for industrial purchases of salt20 and double coupons in grocery retail21 or capping discounts on travel bookings to 3 per cent.22 Yet another outcome is adopting a surcharge, while leaving cartel members with full discretion as to the other components that make up the final price. Firms coordinated on a fuel surcharge in the air cargo, air passenger and rail industries and a lead surcharge in a battery market. In some of these cases, it is not clear how it was an effective form of collusion. If firms coordinate on only list prices, why wouldn’t firms undermine the agreement by offering larger discounts in order to gain more market share? If they only agreed on a surcharge, why not try to gain market share by lowering the price of other components of the invoice? Nevertheless, collusion was effective, as evidenced, for example, in the air cargo case with customer damages exceeding US$1.2 billion. Some progress has been made on understanding how and when coordination on list prices is effective.23 The point to be underscored is that collusion intended to raise transaction prices can be achieved by coordinating on other variables which influence those transaction prices. Rather than coordinate on specific prices – whether transaction prices, list prices, discounts or surcharges – firms may agree to pricing rules or conventions that again will result in higher transaction prices. In the market for turbine generators, the only two suppliers coordinated on replacing a policy of negotiation with one of non-negotiable posted prices.24 In a stock exchange, market makers coordinated on not quoting prices ending in odd-eighths which, by raising the minimum bid-ask spread, resulted in higher price-cost margins.25 In an online market, poster sellers on Amazon Marketplace coordinated on pricing algorithms rather than individual prices.26 By doing so, they eliminated competition with each other while still competing with other sellers. Another path to higher transaction prices is to coordinate on restricting supply. Firms have used public announcements in the markets for chicken and pork to promote a collective Unless otherwise noted, J. Harrington and L. Ye, ‘Collusion through Coordination of Announcements’ (2019) 67 Journal of Industrial Economics 209, provide references for cases pertaining to coordination on list prices and surcharges. 19 Case AT.39824, Trucks, Commission decision, 19 July 2016, C(2016) 4673 final. 20 Morton Salt Company v United States, 235 F. 2d 573 (10th Cir. 1956). 21 United States v The Stop and Shop Companies, Inc., US District Court of Connecticut, No. CRE B-84-51 (9 November 1984). 22 Case C-74/14 ‘Eturas’ UAB and others v Lietuvos Respublikos konkurencijos taryba, ECLI:​EU:​C:​ 2016:​42, 21 January 2016. 23 Harrington and Ye (n 18). 24 J. Harrington, ‘Posted Pricing as a Plus Factor’ (2011) 7 Journal of Competition Law and Economics 1. 25 W. Christie and P. Schultz, ‘Why Do NASDAQ Market Makers Avoid Odd-Eighth Quotes?’ (1994) 49 Journal of Finance 1813. 26 Oxera, ‘When Algorithms Set Prices: Winners and Losers’, Discussion Paper, 19 June 2017. 18

The practical requirements of a successful cartel  9 curtaining of supply.27 Alternatively, collusion could have firms allocating the market so only a subset of firms supplies a particular collection of customers. This could take the form of exclusive territories whereby each firm is given monopoly power over a particular geographic market; or, when customers are large, allocating them among cartel members. With a customer allocation scheme, a firm is not supposed to solicit the business of a customer assigned to another firm and, should the customer approach the firm, it should offer a very high price or simply decline making an offer (‘Sorry but we do not have any available inventory’). The takeaway from Section II is that there are many ways in which firms can enact a collective increase in their prices and, more generally, constrain competition in a mutually beneficial manner.28 However, so as to make for a more coherent analysis, from hereon we will generally discuss collusion in terms of directly coordinating on transaction prices. A cartel is more successful when the profits earned while colluding are higher, which will be the case when price (or, equivalently, the overcharge which measures the percentage increase in price due to collusion) is higher. While a higher price reduces how much cartel members sell, the amount of reduction in sales depends on how much demand is diverted to non-cartel suppliers. The more the cartel can limit that diversion, the more profit it will earn. A cartel is also more successful when collusion has longer duration, which requires maintaining the cartel’s stability and avoiding detection that might cause the cartel’s shutdown. Finally, a cartel is more successful when it avoids or limits penalties in the event of conviction or settlement. The remainder of this chapter will discuss the challenges faced by a cartel in seeking to be successful and how firms have met those challenges.29

III.

OVERCOMING THE CHALLENGES TO SUCCESSFUL CARTEL ACTIVITY

Towards understanding cartel success, let us begin by contemplating why collusion might not succeed or achieve only modest success. First, firms may fail to coordinate. That is, firms do not agree to constrain competition or do not agree on a collusive arrangement. Second, firms may coordinate but the collusive arrangement is internally unstable. That is, there is a lack of compliance among cartel members which then reduces efficacy – overcharges are low, collusion periodically breaks down – and might result in the cartel’s collapse. Third, firms may coordinate on an internally stable collusive scheme but it is externally disrupted by non-cartel suppliers. As a result, the cartel sells fewer units and limits the overcharge in order to stem non-cartel supply, and its duration may be shortened because external instability spills

J. Harrington, ‘Collusion in Plain Sight: Firms’ Use of Public Announcements to Restrain Competition’ (2022) 84 Antitrust Law Journal 521. 28 A class of collusive outcomes not covered is when firms coordinate to perform exclusionary activities against some class of competitors. For some examples, see J. Kwoka, ‘Agreeing to Exclude: The North Carolina State Board of Dental Examiners’, in J. Kwoka and L. White (eds), The Antitrust Revolution: Economics, Competition, and Policy (7th edn, Oxford University Press, 2019); and L. Garrod, J. Harrington and M. Olczak, Hub-and-Spoke Cartels: Why They Form, How They Operate, and How to Prosecute Them (The MIT Press, 2021) Chapter 5. 29 Market conditions that contribute to a successful cartel are not reviewed here. The interested reader is referred to M. Motta, Competition Policy: Theory and Practice (Cambridge University Press, 2004). 27

10  Research handbook on cartels over to internal instability. Fourth, the cartel may fail because it is caught by public or private enforcers. We will examine each of these four challenges in turn: (a) coordination; (b) internal stability; (c) external stability; and (d) avoiding detection and penalization (enforcement). A. Coordination Though collusion is a joint activity, it begins with one firm (or, more exactly, one firm’s executives) deciding to shift rival firms’ beliefs that they are competing to a common belief not to compete, along with decisions as to how they are to constrain competition. To make this happen, a firm engages in a coordinating practice, which is communication intended to coordinate on a collusive arrangement.30 As we describe later, a cartel will communicate for other purposes but here it is done in order to agree not to compete and how not to compete. In seeking to coordinate, a firm communicates a message – either through an announcement or an action – which is conveyed privately (only among the prospective cartel members) or publicly (easily accessed by other market participants). An announcement is the expression of language, broadly defined. An action is an act that has consequences independent of the information that it delivers, while an announcement has consequences only because of the information that it delivers. For example, the announcement of a firm that it will raise price by 10 per cent in 30 days has consequences only if it conveys information that changes what other agents do, whether it induces customers to buy earlier (in order to avoid the anticipated price increase) or rival firms to match the price increase (so it is a coordinating practice). In contrast, the action of raising price by 10 per cent has consequences irrespective of the information that it delivers. Whether or not rival firms interpret a firm’s price increase as an invitation to collude, the firm’s revenue and profit will change, because fewer consumers will buy and those that do buy pay more. Let us begin by considering the use of announcements for firms to coordinate. The canonical coordinating practice is express communication, in which firms openly share their intentions and plans using natural language. For example, in the fine arts auction houses cartel, the Chairman of Christie’s conveyed to Sotheby’s Chairman: ‘We’re getting killed on our bottom line. I feel it’s time to increase pricing.’ Sotheby’s Chairman responded: ‘I agree. But it’s your turn to go first this time.’31 Here we have an unambiguous invitation to collude and acceptance of that invitation, which did indeed work, as both auction houses raised their commission rates. The history of cartels is replete with such cases. While such direct, unvarnished language is the most effective means to coordinate, it also delivers the most compelling evidence for proving a violation of competition law. Consequently, less transparent announcements may be used which, while less effective, leave less of an evidentiary trail. This category includes unilateral announcements whereby a firm expressly invites other firms to collude and, rather than soliciting an express acceptance, counts on them to act accordingly. This could take the form of a written letter describing

30 For a comprehensive discussion of communication in the context of cartels, see L. Kaplow, Competition Policy and Price Fixing (Princeton University Press, 2013) Chapter 3. 31 C. Mason, The Art of the Steal: Inside the Sotheby’s-Christie’s Auction House Scandal (Berkley, 2004) 121.

The practical requirements of a successful cartel  11 a collusive plan which is sent to all firms with the letter noting it has been sent to all firms32 or announcing a plan to increase price at a private meeting with other firms.33 The coordinating practices mentioned thus far all share the features of private communication with a reasonably explicit invitation to collude. Coordinating practices can also encompass public announcements such as advance price announcements. Consider a firm contemplating a joint increase in firms’ prices. It could raise its price hoping that other firms would interpret it as an invitation to collude. However, that is somewhat risky, for other firms may not follow and, in the meantime, the firm is losing sales. Alternatively, it could publicly announce a future price increase (specifying the amount and date) and wait for other firms to respond. If they make the same announcement, then firms have a ‘meeting of minds’ and implement the proposed price increase. If not, then the original firm retracts its announcement. This was the collusive theory used by the US Department of Justice against several airlines.34 Of course, firms will claim the announcements are intended to inform customers of future prices, which may well be true. That customers are a possible audience can make these cases difficult to prosecute but also make advance price announcements less effective as a means of collusion, since other firms may be uncertain as to whether there is an invitation to collude. Firms have also coordinated by using public announcements that refer to rival firms’ conduct rather than a firm’s own conduct. Three classes of messages have been identified.35 First, a firm describes how its future conduct is contingent on rival firms’ conduct. An example is AirTran stating during an earnings call that it would adopt a first-bag fee but only after its main competitor, Delta Airlines, did so. Shortly thereafter, internal documents show that Delta changed its decision and adopted a first-bag fee. Through further public announcements, the two airlines initiated the same fee on the same date. In this case, public announcements were used to form an agreement between firms to have a leader–follower arrangement which then resulted in a supracompetitive outcome. A second class of messages is when a firm announces how rival firms or the industry at large should behave. Examples are episodes of ‘capacity discipline’ in the airline and steel industries. Through earnings calls and statements at industry meetings, senior executives criticized past conduct as having been too aggressive and encouraged all firms to reduce capacity and supply in order to raise prices. Market data and their own subsequent announcements showed they succeeded. The third class has a firm announce how rival firms or the industry at large will behave. This forecast of future conduct could be an invitation to firms to act consistently with that forecast. As of yet, there are no documented episodes. A firm can also use actions to coordinate. In the turbine generator market, it has been argued that General Electric’s shift from negotiating with buyers to setting a non-negotiable price was an invitation to Westinghouse to collude because such a pricing policy could only be in GE’s self-interest if Westinghouse were to match that policy and coordinate its prices.36 Westinghouse did match it and supracompetitive prices followed as GE acted as a price leader.

Interstate Circuit, Inc., et al. v United States, 306 US 208 (1939). United States v Foley, 598 F.2d 1323 (4th Cir. 1979). 34 S. Borenstein, ‘Rapid Price Communication and Coordination: The Airline Tariff Publishing Case’, in J. Kwoka and L. White (eds), The Antitrust Revolution: Economics, Competition, and Policy (4th edn, Oxford University Press, 1994). 35 Harrington (n 27). 36 Harrington (n 24). 32 33

12  Research handbook on cartels Another avenue is price signalling, which, it has been argued, could be used to form an agreement, though not one that would be unlawful: ‘If a firm raises price in the expectation that its competitors will do likewise, and they do, the firm’s behavior can be conceptualized as the offer of a unilateral contract that the offerees accept by raising their prices.’37 In a meticulously documented analysis, price signalling was shown to be an effective form of collusion in an Australian gasoline market.38 In the same spirit is signalling through bids, as occurred in Federal Communications Commission (‘FCC’) spectrum auctions.39 As part of a market allocation scheme, a bidder used the last three digits of a multi-million dollar bid to signal to another bidder not to bid on a particular licence (which the FCC had numerated using three digits). As has been described, there is a wide array of coordinating practices. We next turn to exploring why they might fail in achieving a ‘meeting of minds’. One obvious source of failure is miscommunication; that is, the sending of a message fails to result in the receiver drawing the intended inference. Miscommunication may occur due to messages lacking clarity or veracity. Clarity refers to the ease with which the receiver can infer the meaning that the sender would like the receiver to infer. Veracity refers to the degree to which the meaning that the sender would like the receiver to infer reflects the sender’s true intentions. Clarity can be challenging when announcements do not involve the express use of natural language. For example, will rival firms interpret an advance price announcement as an invitation to collude? Will a higher price be so interpreted? Regardless of the mode of communication, veracity can be problematic because, even when the intended content of a message is clear, there may be a lack of trust. Ironically, the illegality of inviting a competitor to collude could actually enhance its veracity, as it may only be optimal for a firm to incur the risk of penalties if it intended to coordinate on higher prices.40 A second reason that coordinating practices may fail to deliver an agreement is due to bargaining breakdown. Even if messages are correctly interpreted and believed, there may be disagreement as to whether to collude or with regard to the collusive outcome or arrangement. In crystal clear language, the CEO of American Airlines invited Braniff Airlines to collude: ‘Raise your goddamn fares twenty percent. I’ll raise mine the next morning. You’ll make more money and I will too.’ However, Braniff’s CEO declined the invitation (and, having recorded the conversation, shared it with the US Department of Justice).41 Even if firms agree to collude, they may disagree on the collusive arrangement. There could be different views on the best collusive price. In a Quebec gasoline cartel, it took about 65 phone calls for several cartel members to agree to a price increase.42 Given that firms were heterogeneous – varying in their cost, number of stations, sale of complementary products – they probably had different views as to the most appropriate common price. Even if they In re High Fructose Corn Syrup Antitrust Litig., 295 F.3d 651, 654 (7th Cir. 2002) (Posner J). D. Bryne and N. de Roos, ‘Learning to Coordinate: A Study in Retail Gasoline’ (2019) 109 American Economic Review 591. 39 P. Cramton and J. Schwartz, ‘Collusive Bidding: Lessons from the FCC Spectrum Auctions’ (2000) 17 Journal of Regulatory Economics 229. 40 H. Aghadadashli and P. Legros, ‘Let’s Collude’, Working Paper, Universite Libré de Bruxelles, 4 September 2020. 41 United States v American Airlines, 743 F.2d 1114 (5th Cir. 1984). 42 R. Clark and J.F. Houde, ‘Collusion with Asymmetric Retailers: Evidence from a Gasoline Price-Fixing Case’ (2013) 5 American Economic Journal: Microeconomics 97. 37 38

The practical requirements of a successful cartel  13 do agree on price, they are likely to disagree on the allocation of demand. Market allocation is a zero-sum game, for the more sales one cartel member receives, the fewer sales there are to go to other members. A bromine cartel often experienced price wars due to bargaining breakdown as some cartel members sought to renegotiate the collusive outcome.43 The lysine cartel had no difficulty agreeing to a common price but struggled with finding sales quotas to satisfy all members.44 Though initially able to raise price without having settled upon a market allocation, the arrangement soon unravelled with a series of price cuts as firms sought to gain more market share. Only after coming back to the bargaining table and settling on an allocation was the cartel successful. Critical to the selection of coordinating practices is the efficacy–exposure trade-off. Coordinating practices vary in terms of their efficacy in achieving mutual understanding that firms will not compete and how they will collude. They also differ in terms of their legal exposure and, more specifically, the likelihood of detection and penalization. There is generally a trade-off because coordinating practices which are more effective for collusion also tend to come with more exposure. Private, express and unfettered communication is strong on efficacy but comes with considerable exposure. Though detection may be low by virtue of being private, conviction is high conditional on detection. In comparison, public announcements about rival firms’ conduct (for example, encouraging the industry to restrict supply) and advance price announcements are less effective and easier to detect but more difficult to prosecute. They are less effective and less prosecutable for the same reason: they could be intended for other market participants such as customers, input suppliers and investors. Using actions to coordinate – such as price signalling – is, compared to those public announcements, even less effective and, in most instances, comes with minimal or no legal exposure. Public announcements and actions are also less effective because they constrain the type of collusive arrangement. For example, advance price announcements may be able to coordinate on price but not on a market allocation scheme. Finally, it is worth emphasizing that leniency programmes, which have been quite useful in detecting and prosecuting cartels, are generally relevant only when firms engage in express, unfettered communication so that a leniency applicant has direct evidence of an agreement. Firms wary of a leniency programme may then opt for a subtler coordinating practice. B.

Internal Stability

Let us suppose firms agree to raise their prices to some common level (or by some common percentage) above the existing competitive price. This agreement comes with a source of potential instability in that there is a temptation to cheat. The competitive price has the property that it maximizes a firm’s current profit given other firms are charging the competitive price; in other words, each firm is doing the best it can given the prices set by competitors. Should rival firms raise their prices by, say, 10 per cent, the firm would also want to raise its price, for it now faces stronger demand given that customers find the alternative – rival firms’ products – to be priced higher. However, in maximizing current profit, the firm would raise its price by less than 10 per cent as it seeks to pick up more market share at this supracompetitive M. Levenstein, ‘Price Wars and the Stability of Collusion: A Study of the Pre-World War I Bromine Industry’ (1997) 45 Journal of Industrial Economics 117. 44 Case COMP/36.545/F3, Amino Acids, Commission decision, 7 June 2000, [2001] OJ L 152/24. 43

14  Research handbook on cartels price-cost margin. Thus, if the collusive agreement calls for it to raise price by 10 per cent along with all other firms, a firm is foregoing some current profits in doing so. There is then a temptation to cheat by undercutting the collusive price set by rival firms. In its simplest terms, this is the challenge for internal cartel stability. In practice, this temptation to cheat is neutralized by firms adopting strategies that embody a reward–punishment scheme. This strategy makes a firm’s current conduct (what price to charge) contingent on firms’ past conduct (what prices were charged). If a firm abides by the collusive outcome – which could involve high prices, sales quotas, customer allocation, etc. – then it is rewarded in the future by rival firms continuing to abide by the collusive outcome (for example, charging high prices, not selling more than their quotas, staying away from other firms’ customers), while if it departs from the collusive outcome (for example, setting a low price, selling above its quota, serving another firm’s customers), then it is punished in the future by rival firms acting aggressively to reduce the deviating firm’s profits (for example, lowering prices, selling more, soliciting the deviating firm’s customers). Collusion is a common understanding among firms that ties future rewards and punishments to current behaviour and, when effective, induces compliance with the supracompetitive outcome. This understanding can be viewed as contractual, though, where the penalties for acting contrary to the terms of the contract take the form of rival firms’ future punishing behaviour. Embodying this reward–punishment scheme, a collusive strategy is composed of three elements, some of which may be implicit but still understood by cartel members. First, there is the collusive outcome, such as what common price is to be charged. Second, there is the monitoring protocol, which refers to how firms will monitor each other for compliance with the collusive outcome. Third, there is the punishment that occurs when there is evidence of noncompliance. A cartel is internally stable when, given the reward–punishment scheme embodied in the collusive strategy, it is in each firm’s best interests to comply by acting in accordance with the collusive outcome. This means that the short-run gain in profits by deviating (for example, undercutting the collusive price and gaining more sales) is exceeded by the expected foregone future profits from the deviation being detected and the resulting lower profits from the punishment. The internal stability condition can be represented as: Collusive Profit + Expected Value of Future Collusive Profits > Deviation Profit + Expected Value of Future Punishment Profits.

The Expected Value of Future Collusive Profits encompasses the likelihood of continued collusion and possible future punishments in response to evidence of noncompliance, as well as possible cartel collapse. The Expected Value of Future Punishment Profits encompasses the likelihood that a deviation is detected and punished, as well as the likelihood that the firm gets away with it and collusion continues uninterrupted. The internal stability condition can be rearranged to: Future Loss from Deviation > Current Gain from Deviation

The practical requirements of a successful cartel  15 where Future Loss from Deviation = Expected Value of Future Collusive Profits - Expected Value of Future Punishment Profits Current Gain from Deviation = Deviation Profit - Collusive Profit.

The internal stability condition is more likely to be satisfied when deviations are detected sooner (which implies the current gain from deviation is lower) and the punishment is more severe (so the future loss from deviation is greater). As noted earlier, the charging of a higher common price will intensify competition along non-price dimensions. The more that the cartel can control those other dimensions and thereby keep collusive profit high, the more likely the internal stability condition is satisfied.45 Finally, when a firm cares more about future profit relative to current profit, the future loss from a deviation will loom larger and that will enhance stability. Satisfaction of the internal stability condition is one of the determinants of the collusive price. As the collusive price is set higher, the current gain from deviation increases as a given rise in sales from undercutting the collusive price yields more profits. On the other hand, a higher collusive price raises collusive profits and thus increases the future loss from a deviation. Generally, economic theory has shown that the first effect is larger in magnitude so that a higher collusive price makes it less likely the internal stability condition is satisfied. A cartel will then be constrained as to how high it sets the collusive price, for it will want to balance higher collusive profits against shorter duration due to a less stable cartel. Colluding firms will also limit how fast they raise prices. Constraining forces include cartel detection (inexplicably large price increases may cause customers to suspect collusion), learning (firms may be uncertain how high a price is stable and thus need to experiment) and buyer resistance (industrial buyers may threaten delaying purchases). Consequently, cartels tend to gradually raise price, as evidenced, for example, by cartels in citric acid and vitamins46 and in graphite electrodes.47 Monitoring for compliance is crucial for internal stability. If monitoring is less effective then the probability of a deviation being detected will be lower, which makes deviation more attractive. With retail markets, a firm’s price is posted (either in stores or online), which makes monitoring fairly straightforward. If retailers form a cartel and agree on certain prices, each can monitor the others for compliance. Price monitoring is well documented for cartels involving grocery stores, drugstores and many other retailers.48 In some of these cartels, an upstream manufacturer assisted in collecting price information as part of a hub-and-spoke cartel (which are covered below, in Chapter 10). Monitoring is a very different exercise with industrial products because transaction prices are often not public information. Those markets are typically characterized by the setting of list prices with buyer-specific discounts. That means the price that a buyer pays is private 45 In addition to commission rates, the fine arts auction houses cartel coordinated on many non-price dimensions: Mason (n 31), 119. 46 Connor (n 16). 47 M. Levenstein and V. Suslow, ‘Contemporary International Cartels and Developing Countries: Economic Effects and Implications for Competition Policy’ (2004) 71 Antitrust Law Journal 801. 48 Garrod, Harrington and Olczak (n 28).

16  Research handbook on cartels information to it and the seller, and thus is not observed by other sellers. Though a firm’s sales representative may periodically learn from customers what prices other firms are offering, such information is sporadic and not always reliable. If a cartel coordinates on setting some common transaction price but those prices are difficult to observe, there is the concern that a deviation might not be detected, which will then encourage deviations and undermine cartel stability. One response is to coordinate on list prices, for they are observable. While we have noted that some cartels pursued that strategy, it is a less effective form of collusion because it does not control discounts. More common is for a cartel to coordinate on transaction prices and put in place a market allocation scheme and to monitor sales rather than prices.49 For example, the lysine cartel agreed to a price and a global sales quota for each of the five cartel members: Ajinomoto – 73,500 tons; Archer Daniels Midland (ADM) – 48,000 tons; Kyowa – 37,000 tons; Sewon – 20,500 tons; Cheil – 6,000 tons. Each month, sales were reported to one of the cartel members, which were then shared at the quarterly meetings. Firms were complying as long as they did not sell more than their quota. As an ADM executive announced at a cartel meeting: ‘If I’m assured that I’m gonna get 67,000 tons by the year’s end, we’re gonna sell it at the prices we agreed to and I frankly don’t care what you sell it for.’50 Sales monitoring is made easier when the market allocation scheme assigns geographic markets or customers to cartel members. If a firm is not supposed to sell in a particular territory or to a particular customer, deviation is easily detectable. When it comes to monitoring, there are two other advantages of these schemes over sales quotas. First, a firm may not have full control over its sales due to unpredictability in its customers’ demands (and not wanting to upset customers by limiting supply) or some customers choosing to switch suppliers. Thus, a firm may exceed its sales quota even when it sets the collusive price. Second, a challenge with the use of sales quotas is that it requires cartel members to accurately report their sales, but a firm could easily have an incentive to underreport when it has oversold. We will return to that issue – and how cartels solved it – when discussing punishments. While allocation by territories or customers allows for more effective monitoring than sales quotas, some industrial markets do not lend themselves to divvying up the market in such a manner. Furthermore, there is the concern that the lack of willingness to serve some customers might create suspicions that firms are colluding. Monitoring serves both an ex ante and ex post role. Ex post, detection of a deviation offers an opportunity to induce the noncompliant firm to get back in line with the cartel plan. The well-documented sugar cartel showed that cartel meetings were often used for this purpose.51 However, more critical for the internal stability of the cartel is the ex ante role of monitoring. If a cartel member anticipates that it will be monitored and subsequently punished should it deviate, it will then be more inclined to comply. Thus, an essential complement to monitoring is the threatened punishment, to which we now turn. A punishment results in the cartel member suspected of deviation earning lower profits than if there had there been no evidence supporting its deviation. Let us begin by making two

Many such cartels are described in Harrington (n 16), which is the source of the ensuing facts. 10 March 1994 meeting of the lysine cartel; see ‘The International Lysine Cartel at Work, 3/28/00’, Antitrust Division, US Department of Justice. 51 D. Genesove and W. Mullin, ‘Rules, Communication, and Collusion: Narrative Evidence from the Sugar Institute Case’ (2001) 91 American Economic Review 379. 49 50

The practical requirements of a successful cartel  17 general points about punishments. From the perspective of cartel stability, a punishment should be designed to incentivize compliance, not wreak vengeance. Thus, a punishment should be severe enough to deter deviations but not more. The second point is that punishments are not always under the control of the cartel. When episodes of noncompliance reach a sufficiently high level of frequency or severity, some cartel members may choose to discontinue colluding, thereby causing the cartel’s collapse. There is still a punishment – the deviating firm (as well as all other firms) are back to earning lower competitive profits – but it may be more draconian than the cartel would have liked. A symmetric punishment refers to when all cartel members are harmed. We just mentioned one example – a return (temporary or permanent) to setting competitive prices – while another could be a short, intense price war which could have prices below competitive levels (and even below cost). The downside to a symmetric punishment is that it harms non-deviating firms, which is not only unfair but is counterproductive because it reduces the profitability of collusion. As a result, a firm may be more inclined to deviate when collusion is less profitable. An asymmetric punishment is one that is designed to punish only the deviating firm, and it does so by transferring profits from the deviating firm to the other firms. An asymmetric punishment can serve the purpose of incentivizing compliance and also provide compensation for those firms that were harmed by the deviation. As opposed to a symmetric punishment – which transfers surplus from the cartel to consumers (through lower prices) – an asymmetric punishment moves surplus around among the cartel’s members, and thereby is more attractive to a cartel. The transfer of profits from a deviating firm to other firms can take the form of cash, current sales or future sales. The transfer of cash is uncommon because an inexplicable monetary transfer between competitors is likely to generate suspicions. However, it has been used where legality was not a concern. Consider the 1926 international steel agreement which specified a quota for each country. Article 6 stated: ‘If the quarterly production of a country exceeds the quota which was fixed for it, that country shall pay in respect of each ton in excess a fine of 4 dollars.’ Article 7 provided compensation: ‘If the production of any country has been below the quota allotted to it, that country shall receive in compensation … the sum of 2 dollars per ton short.’52 More recent cartels have instead used inter-firm purchases to enact a transfer.53 Referred to as a ‘guaranteed buy-in’ in the lysine cartel and as ‘buy-back’ in the citric acid cartel, a cartel member that sold more than its quota was required to buy output from those members who sold below their quotas. For example, at the 14 November 1991 meeting of the citric acid cartel, Haarmann & Reimer was told to buy 7,000 tons of citric acid from ADM.54 As there are legitimate reasons for such inter-firm sales (for example, a firm is short on supply and has a contract to fulfil), they need not create suspicions like a monetary transfer. Another transfer scheme is to adjust the future market allocation by adjusting the market share or sales assigned to a firm that oversold or undersold its quota. In the zinc phosphate cartel, a particularly large

A. Plummer, International Combines in Modern Industry (Pitman Publishing Corp., 1938) 248. C. Leslie, ‘Balancing the Conspiracy Books: Inter-Competitor Sales and Price-fixing Cartels’ (2018) 96 Washington University Law Review 1. 54 Harrington (n 16). 52 53

18  Research handbook on cartels customer was rotated among the cartel members but would be disproportionately allocated to a member that had been underselling its quota.55 As previously noted, a sales monitoring scheme requires firms to accurately report their sales to other cartel members. Such a collusive arrangement creates a double challenge for cartel stability: a cartel member must be incentivized to price at the collusive level and to truthfully report its sales. The latter is problematic when overselling one’s quota means having to buy output from other firms. Nevertheless, this scheme was effective in a number of cartels, including those in the markets for citric acid, lysine and vitamins. It has been shown that a two-tier punishment can explain its success.56 If a firm anticipates accurately reporting its sales, it will be in its best interests to set the collusive price in order to avoid overselling and incurring the (tier one) punishment of buying output from other firms. A firm is incentivized not to underreport sales because of the (tier two) punishment of returning to competition which is triggered when the sum of all firms’ sales reports is too low. A firm that underreports, while presuming other firms are accurately reporting, would depress the aggregate sales report and thus increase the chances of causing a shift back to competition. Some cartels have used a third party to shore up internal stability by having them aid in coordinating, monitoring and punishing. To assist in arriving at a collusive outcome, a bidding ring of stamp dealers used a taxi driver,57 and the organic peroxides cartel used the consultancy AC Treuhand.58 A Swiss accounting firm validated firms’ reported sales in the lysine cartel,59 and in several retailer cartels an upstream supplier threatened retailers who undercut the collusive price with a higher wholesale price or denial of supply.60 C.

External Stability

A cartel can be made less effective or result in collapse because of the expansion of non-cartel supply. That cartel members are pricing at a supracompetitive level will cause existing suppliers who are not members of the cartel to increase their supply, and may even cause entry into the market. Non-cartel supply is a serious concern whenever significant capacity remains outside of the cartel.61 The global citric acid cartel encompassed only 60 per cent of global production and 67 per cent of EU production. In particular, Chinese suppliers were excluded, as they were with cartels in vitamins B1, B2, and C. The EU industrial tubes cartel left about 20 per cent of capacity out of its control. As a general rule, a cartel would like all firms to be members, for it would be better to have a firm inside the cartel keeping price high and restricting its supply than outside the cartel undercutting the collusive price. In practice, a cartel may choose to exclude some members due to a lack of trust. For a variety of reasons – such as past aggressive conduct or foreign ownership – a firm may be viewed with suspicion when it comes to abiding by a collusive ibid. J. Harrington and A. Skrzypacz, ‘Private Monitoring and Communication in Cartels: Explaining Recent Collusive Practices’ (2011) 101 American Economic Review 2425. 57 J. Asker, ‘A Study of the Internal Organization of a Bidding Cartel’ (2010) 100 American Economic Review 724. 58 Marshall and Marx (n 16). 59 Case COMP/36.545/F3, Amino Acids, Commission decision, 7 June 2000, [2001] OJ L 152/24. 60 Garrod, Harrington and Olczak (n 28). 61 The ensuing examples are from Harrington (n 16). 55 56

The practical requirements of a successful cartel  19 arrangement, in which case their inclusion could undermine internal stability. A firm may also be excluded related to enforcement concerns.62 More firms mean more ways in which a competition authority can learn of a cartel. If there is a leniency programme then one more cartel member is one more firm to compete with for receiving leniency, should it come to that. Especially when a firm is small – so its presence outside of the cartel will not be too disruptive – firms may prefer to exclude it from the cartel because of these enforcement concerns. A cartel may also not be all-inclusive because some firms choose not to join.63 Large firms will have to join if the cartel is to be effective. Small firms prefer not to join because their inclusion has a small effect on raising the collusive price but as a cartel member they may be expected to substantively restrict how much they produce. Medium-sized firms may or may not want to join, and their participation could make the difference between a moderately successful cartel and a highly successful cartel. It is worth noting that, even if a firm prefers to be outside of the cartel, cartel members could coerce participation by threatening exclusionary actions. As an example of the havoc non-cartel supply can create, consider the global vitamins cartel, which comprised 16 vitamins with an overlapping set of suppliers. Initiated during 1990–91, six cartels internally collapsed over 1994–95, while the other ten shut down due to government investigations over 1998–99. With the objective of understanding the source of internal collapse, a study focused on four vitamin markets: vitamin C, which did internally collapse, and vitamins A and E and beta carotene, which did not.64 They estimated the internal stability condition and found it was first violated for vitamin C in 1995 – the year in fact the vitamin C cartel collapsed – and was never violated for A, E and beta carotene. The determining factor in the collapse of the vitamin C cartel was the growth in Chinese supply, which increased four-fold in the cartel’s first four years and reduced the cartel’s market share from 85–90 per cent to 60–65 per cent. When a cartel is not all-inclusive, there are four strategies for dealing with non-cartel suppliers: starvation, coercion, bribery and takeover.65 Starvation limits non-cartel supply by taking control of an essential input or technology. When several Chinese firms expressed a desire to enter, the members of the global sorbates cartel agreed not to share their technology with them. Coercion curtails non-cartel supply through aggressive practices, such as selective price cuts, with the intent of continuing these practices until the non-cartel supplier limits its supply, joins the cartel or exits the market. In the district heating pipes cartel, the firm Powerpipe declined an invitation to join the cartel and later filed a complaint with the European Commission on the grounds that the colluding firms had acted anticompetitively against it. For example, after Powerpipe was awarded a sizeable contract, the cartel organized a boycott of Powerpipe’s customers and suppliers.

I. Bos and J. Harrington, ‘Competition Policy and Cartel Size’ (2015) 56 International Economic Review 133. 63 I. Bos and J. Harrington, ‘Endogenous Cartel Formation with Heterogeneous Firms’ (2010) 41 RAND Journal of Economics 92. 64 M. Igami and T. Sugaya, ‘Measuring the Incentive to Collude: The Vitamin Cartels, 1990–1999’ (2022) 89 Review of Economic Studies 1460. 65 These terms and the ensuing discussion are from J. Harrington, K. Hüschelrath and U. Laitenberger, ‘Rent Sharing to Control Non-Cartel Supply in the German Cement Market’ (2018) 27 Journal of Economics and Management Strategy 149. 62

20  Research handbook on cartels While coercion uses the stick, bribery uses the carrot by sharing collusive rents with non-cartel suppliers who agree to limit their expansion of supply. In order to control Coors, which was a producer of vitamin B2 but not a member of the cartel, cartel member Roche agreed to purchase 115 tons of B2 from Coors (which represented half of Coors’s capacity) and BASF in turn agreed to purchase 43 tons from Roche. In this manner, they shared the burden of controlling Coors’ supply. Takeover curtails non-cartel supply by acquiring non-cartel suppliers or the assets used to provide that supply. The carbon and graphite products cartel struggled with non-cartel suppliers known as ‘cutters’ who would purchase carbon blocks from the cartel members and then produce final products which competed with the cartel’s supply. In response to the aggressiveness of cutter EKL, cartel members considered not supplying graphite to it (starvation) and undercutting EKL’s price for customers considering doing business with EKL (coercion). Ultimately, the takeover approach was pursued, as cartel member SGL Carbon acquired EKL. D. Enforcement Suppose coordinating practices succeed in firms coming to a mutual understanding to constrain competition. Furthermore, they implement a collusive arrangement that is both internally stable – all cartel members comply – and externally stable – non-cartel suppliers do not substantively expand supply. If all that is done, the cartel may still fail because it is detected and penalized.66 That firms may be subject to public prosecution and private litigation can affect their decision as to the mode and extent of communications. Communications are relevant to achieving mutual understanding, the type of collusive arrangement (for example, is it just with regard to price or is there also a market allocation scheme) and monitoring for compliance as well as the imposition of punishments. The more extensive communications are, the more effective collusion will be, but also the greater the chance of detection and the imposition of penalties. In deciding on the communications protocol, a cartel balances higher collusive profits against possibly shorter duration and a higher likelihood of penalties. Enforcement will also impact cartel success in terms of the collusive overcharge. When prices are increased at a faster pace and to a higher level, detection by savvy customers (especially when they are industrial buyers) and the competition authority becomes more likely. In choosing the price path, the cartel may then moderate its price increases.67 Higher prices, as well as parallel price movements, will contribute not only to detection but also to the likelihood of prosecution and conviction. While economic evidence is not sufficient by itself to prove a violation of competition law, it can be an important part of a case. Finally, some penalties are tied to the overcharge. Private litigation seeks customer damages, which are measured by the overcharge multiplied by the number of units purchased by a customer. 66 For surveys of cartel detection, see G. Hay and D. Kelly, ‘An Empirical Survey of Price Fixing Conspiracies’ (1974) 17 Journal of Law and Economics 13; and J. Harrington, ‘Detecting Cartels’, in P. Buccirossi (ed.), Handbook of Antitrust Economics (The MIT Press, 2008) 213. 67 This trade-off is examined in Chapter 3.3 of J. Harrington, The Theory of Collusion and Competition Policy (The MIT Press, 2017). For indirect evidence that enforcement constrains the prices that a cartel sets and reduces the profitability of collusion, see I. Bos, S. Davies, J. Harrington and P. Ormosi, ‘Does Enforcement Deter Cartels? A Tale of Two Tails’ (2018) 59 International Journal of Industrial Organization 372.

The practical requirements of a successful cartel  21 The US Department of Justice can impose a fine as high as twice the harm to consumers. Thus, a cartel may limit its price in order to reduce this liability.68 Obviously, enforcement affects cartel success when it causes a cartel to be shut down. While conviction will surely result in a cartel’s collapse, detection and opening an investigation can often be sufficient. The ensuing increase in the probability of paying penalties can cause the cartel to become internally unstable. Less well recognized is that enforcement can reduce cartel duration even when a cartel goes undiscovered. Returning to the internal stability condition, expected penalties will lower the value of future collusive profits, which will reduce the foregone profits from deviating. By making collusion less profitable, enforcement makes a cartel less stable, and that will reduce cartel duration.69

IV.

CONCLUDING REMARKS

The takeaway should be that collusion is difficult but manageable. A successful cartel must coordinate on an internally stable collusive arrangement while controlling non-cartel supply and avoiding detection and penalization by public and private enforcers. Though the challenges are numerous and substantial, many firms in various industries have managed to master them. Successful cartels exist for three simple reasons. First, there is a lot of money at stake. Notably, cartels are most common in markets with highly similar goods where price competition is intense.70 In such a setting, even a modest increase in price can have a significant impact on a firm’s bottom line. Second, firms’ executives are smart, and they are used to solving difficult problems. Once a sufficient number of executives have come to believe that the primary obstacle to sufficient profits is too much competition, they are sharp enough to find a solution (as exemplified by the German spectrum auction at the start of this chapter). Third, there are many ways to constrain competition; collusion does not have to be sophisticated to succeed. In sum, cartels exist and thrive because firms have the incentive to collude, they have the skills to collude and they have at their disposal many ways in which to collude.

See Chapters 18 and 21 respectively for coverage of fines and damages. The ways in which enforcement can reduce cartel duration are examined in Chapter 3.1 of Harrington (n 67). 70 See P. Grout and S. Sonderegger, ‘Predicted Cartels’, Office of Fair Trading, Economic Discussion Paper, March 2005; and J. Harrington, ‘Thoughts on Why Certain Markets Are More Susceptible to Collusion and Some Policy Suggestions for Dealing with Them’, OECD Background Paper, Global Forum on Competition, 19 October 2015. 68 69

2. The prevalence and injuriousness of cartels worldwide John M. Connor and Robert H. Lande

I. INTRODUCTION The primary purposes of this chapter are to present empirical evidence on the size of the economic injuries resulting from contemporary price-fixing cartels and to assess the effectiveness of antitrust and competition-law enforcement that is designed to deter these harms. First, in Section II, we gather and analyse summary statistics on a large number of topics that might reveal whether discovered cartels are responding to the increased worldwide effort to suppress them; we examine numbers and sizes of price-fixing cartels, the damages they generate and key antitrust and competition law enforcement statistics. Broad geographic and jurisdictional differences are highlighted. Second, in Section III, we address whether current price-fixing penalties are sufficient to deter the formation of new cartels. In doing so, we first present a theoretical framework for determining optimal cartel penalties and then apply that framework to data on US-penalized cartels. That is, we examine whether the multiple sanctions imposed on such cartels are high enough to provide optimal deterrence. Our empirical results demonstrate that cartels are almost always under-deterred even in the United States, the jurisdiction that imposes the most severe sanctions. A fortiori, the overall levels of cartel sanctions should be increased dramatically worldwide.

II.

SUMMARY OF GLOBAL STATISTICS

Section II.A initially lays out a set of aggregate statistics on the numbers, affected sales, damages, and corporate and individual penalties connected with contemporary illegal price-fixing worldwide.1 Then, in Section II.B we look for geographic and jurisdictional differences in these features of cartels across major jurisdictions. Collecting and organizing such data is a daunting exercise, though one made feasible by the availability of reports of antitrust and competition law authority decisions on the World Wide Web since the early 2000s.2 The most comprehensive data set on cartels is ‘Private Because of subtle differences across jurisdictions, the conduct characterized as illegal will vary. However, explicit horizontal price or quantity restraints tend to be treated fairly similarly across most nations. This chapter focuses on cartels discovered since 1989. For analyses of mostly older cartels, see M. Levenstein and V. Suslow, ‘What Determines Cartel Success?’ (2006) 64 Journal of Economic Literature 43 and citations therein. 2 Around 2015, some large multinational law firms began assembling and publishing annual reports on cartel enforcement. See, e.g., Morgan Lewis, Global Cartel Enforcement Report (January 2021) www​ .morganlewis​.com/​pubs/​2021/​01/​global​-cartel​-enforcement​-report​-year​-end​-2020 [accessed 10 March 2022]. In the past, such efforts have been discontinued after five or ten years. 1

22

The prevalence and injuriousness of cartels worldwide  23 International Cartels’ (‘PIC’), which spans 30 years of consistent data collection.3 PIC comprises the most comprehensive compilation of legal–economic information on private international ‘hard-core’ cartels.4 Examining only ‘private’ collusion is appropriate for this Handbook, because non-sovereign cartels are potentially subject to antitrust or competition law enforcement; and ‘international’ is a criterion that is a precise way of capturing virtually all large cartels.5 Of course, no compilation can include the vast majority of such cartels that remained clandestine throughout their lives. In short, these data will considerably undercount the number and dimensions of cartel characteristics. A.

World-Wide Cartel Statistics

1. Cartel numbers PIC contains information on 1423 suspected or convicted cartels.6 This number is the tip of the cartel iceberg, because it excludes public, domestic and undiscovered cartels. ‘Public’ cartels are open agreements enforcing suppliers’ decisions by sovereign authority; they include multilateral commodity agreements, agricultural marketing schemes and compulsory cartels typically imposed during wars or deep recessions. They are normally immune from antitrust challenges. Historically, hundreds of public cartels have been recorded, but they appear to be in decline in the past few decades. Domestic cartels have no members from outside the jurisdiction that discovered the collusion. Domestic cartels are generally more numerous than international ones, but they tend to receive less attention in the press and in reports by antitrust authorities.7 A compilation of overcharge estimates classifies the number from domestic-membership schemes at double that from international schemes.8 DOJ data also 3 The PIC data set was built beginning around the year 2000 solely by the first author. No corporations, law firms or similar commercial entities have ever materially supported the development of PIC. The edition employed in this chapter includes cartels discovered during the 30 years from 1990 to 2019. It is fully accessible to interested readers at https://​purr​.purdue​.edu/​publications/​2732/​2 [accessed 10 March 2022]. An alternative version of PIC data is available at the OECD: http://​bit​.ly/​OECD​-ICStats [accessed 10 March 2022]. ICStats has the advantage of direct reporting of detected cartel activity from 55 jurisdictions, but the OECD’s summaries report on fewer variables than PIC. See OECD, OECD Competition Trends 2020 (Paris, 2020) www​.oecd​.org/​competition/​oecd​-competition​-trends​.htm [accessed 10 March 2022]; and OECD, OECD Competition Trends 2021: Volume I: Global Competition Update 2015-2019 (Paris, 2021) www​.oecd​.org/​daf/​competition/​oecd​-competition​-trends​-2021​-vol1​.pdf [accessed 10 March 2022]. The figures appearing in this chapter are constructed from the PIC data set. 4 PIC generally follows the definition of international price-fixing agreements employed by the Antitrust Division of the US Department of Justice (‘DOJ’). However, PIC defers to the somewhat variable concepts of hard-core cartel conduct employed by other jurisdictions. The term international primarily follows from a cartel’s membership composition. Therefore, an international cartel is a conspiracy in restraint of trade that has or is alleged to have one or more corporate or individual participants with headquarters, residency or nationality outside the jurisdiction of the investigating antitrust authority. Alternatively, cartels with at least two members with different nationalities are international. International conspiracies are typically more difficult to prosecute. 5 International cartels may have small sales in one nation, but the large majority are large and highly injurious. Many aspire to control multi-jurisdictional or indeed global prices. 6 PIC 2019, Table A0. 7 See OECD, Hard Core Cartels: Recent Progress and Challenges Ahead (Paris, 2003), which in its Annex A highlights a mere 17 cases of cartels with overcharge estimates out of more than 266 cartel cases prosecuted by OECD-affiliated authorities during the period 1996–2000. 8 See J. Connor, ‘Cartel Overcharges’ (2014) 26 Research in Law and Economics 249.

24  Research handbook on cartels suggest that during the period 1990–2017 about two-thirds of penalized corporate cartelists engaged in international cartels.9 The issue of how many undiscovered cartels there are relative to those discovered (the ‘survival rate’) is extremely knotty. Cartelists typically go to great lengths to operate covertly. Harrington and Wei warn that purported detection rates derived from widely cited empirical studies of listed US and EU companies (which average about 15 per cent) are in fact death rates – the end of collusion from either detection or natural collapse.10 Thus, the survival rates of cartels almost certainly exceed 15 per cent; that is, the number of undiscovered cartels is more than about seven times the discovered numbers, but by how much one cannot know. Considering the factors that result in most illegal cartels remaining hidden all their lives, we believe that 40,000 is a conservative projection of the number of illegal price-fixing cartels since about 1990.11 That is, the survival rate of contemporary cartels is at least 96 per cent. Of the 1423 private international cartels detected, 121 (or 8 per cent) of the cartel investigations have been abandoned or closed. Put another way, once a formal investigation of price fixing is launched, the chance that a cartel participant will be found liable for an antitrust infraction is 92 per cent. Reasons vary, but most investigations are shut down because of a lack of sufficient evidence of a violation. Lack of evidence may either derive from the internal discretion of antitrust-authority prosecutors or from the decision of a judge to dismiss a private civil damages suit.12 The United States, Canada, the EU and EU Member States tend to apply highly similar standards for the prosecution of hard-core cartels.13 Besides the 8 per cent ‘not proven’, as of late 2019, 129 cartels (9 per cent of the sample) were still being investigated by government authorities or sued by plaintiffs in private damages litigation. It is likely that 90 per cent of these probes will result in convictions or settlements. The percentage of cartels under investigation rises at an increasing rate with every new edition of PIC. An average of 81 cartels per year were detected in 2010–16 (a record number). By 2026, the number of discovered international cartels is projected to rise to 2200 (Figure 2.1).14 That leaves 1302 cartels (91.5 per cent of those discovered) that have been penalized by one or more antitrust or competition law authority (loosely, ‘guilty’ cartels). Convictions may be sorted into five categories. First, by far the largest category is filled with 790 cartels (61

9 See J. Connor, The Private International Cartels (PIC) Data Set: Guide and Summary Statistics, 1990–2019. (2020) https://​purr​.purdue​.edu/​publications/​2732/​2 [accessed 10 March 2022]. 10 Natural collapse can occur because of sovereign wars or internal dissention among members of the cartel; J. Harrington and Y. Wei, ‘What Can the Duration of Discovered Cartels Tell Us about the Duration of All Cartels?’ (2017) 127 Economic Journal 1977, 1983–84. 11 We start with 1423, triple it to account for domestic membership cartels and multiply by ten to account for undiscovered cartels. The few thousand ‘public’ cartels are not illegal because they are state-supported. 12 The PIC data do not record appeals won by defendants, partly because appeal courts can spend many years or more on rendering verdicts. Very few appeals result in all defendants being exonerated on the merits. 13 There are a few differences. The EU tends to consider combinations of vertical and horizontal price fixing to be a form of hard-core cartel conduct, while the US is more lax in such cases. The same is the case for frequent sharing of sensitive strategic business information with the objective of facilitating market coordination, which is often sufficient evidence for conviction in the EU. 14 The data sample for Figure 2.1 includes all 1292 cartels discovered during the period 1990–2018. The data for 2017–18 are incomplete because many investigations begin in secret and take more than two years to complete.

The prevalence and injuriousness of cartels worldwide  25

Figure 2.1

Trend in cumulative cartel discoveries, worldwide, 1989–2018 (with 2026 projection)

per cent of the guilty cartels) that had only government fines imposed and no known civil judgments; however, about 20 of these were being sued by private plaintiffs who have not yet reached a settlement with the guilty cartelists. Second, 167 cartels (13 per cent) paid both fines and civil penalties from follow-on damages litigation.15 Third, there were 98 cartels (7.5 per cent) that paid only civil damages and no fines. In these cases, the prosecutors were plaintiffs’ counsel who obtained typically large settlements without the benefit of a prior government conviction.16 Nearly all of these suits were court-supervised class actions in North America.17 In sum, the 1055 cartels (81 per cent) in the first three categories paid monetary penalties of some kind. Fourth, 100 cartels (7.6 per cent) accepted ‘consent decrees’ or similar non-monetary warnings or orders from antitrust authorities that require the members of a cartel to curtail specified anti-competitive practices.18 Although they are not technically an admission of guilt, we regard consent decrees as indicators that the authorities ‘more likely than not’ observed illegal cartel conduct.19 Resolution on indictments by means of decrees alone is most common

However, in the United States, all criminal convictions are followed by civil damages litigation. However, a minority may have benefitted from information available from a closed government probe. 17 See J. Connor, ‘Private Recoveries in International Cartel Cases Worldwide: What Do the Data Show?’ (2012) 1 Antitrust Chronicle 2. 18 Seven ‘consent decrees’ were in fact bans imposed on bidding rings of suppliers of goods or services to the World Bank. 19 For a discussion of the merits of settlements involving cartels, arguing that most involve such high penalties that they would have been unlikely to have been agreed to by defendants if the case was merit15 16

26  Research handbook on cartels in the EU, which accounted for half of the decrees. In other cases (in Switzerland, Japan and South Korea), warnings were issued because of a legal limitation on the issuance of fines or because the required fines were considered too high to be politically acceptable. Four consent decrees were issued by US agencies with only civil authority (the Federal Trade Commission (‘FTC’) and the Department of Transportation).20 Fifth, three prosecutions of cartels were interrupted by statutes of time limitations. Our reading of these cases is that these were probable cartel infringements that were abandoned because of an error by the authority or invincible non-cooperation by defendants. 2. Affected sales of cartels The sizes of cartels can be measured in several ways,21 but the most revealing economic measure of size is sales made by a cartel during its collusive period.22 Affected sales are occasionally revealed in a published decision or a posted guilty plea. Until such decisions can be utilized, we infer affected sales from information on the most precise definition of an affected industry’s market boundaries and the duration of collusion.23 We call these inferences ‘known’ affected sales estimates. Sales of cartels under investigation are generally unknown. Estimated known affected sales for 976 penalized cartels are $67 trillion worldwide.24 About 85 per cent of the guilty cartels have sales estimates; for these, the mean average cartel’s sales size is roughly $69 billion. (However, the median average sales size is much lower, $2.06 billion.) If the average sales size of the remaining cartels with no sales estimates is similar, then the affected sales of all detected cartels will add up to roughly $90 trillion – an enormous number but one that is a small share of relevant commercial activity.25

less, see J. Davis and R. Lande, ‘Towards an Empirical and Theoretical Assessment of Private Antitrust Enforcement’ (2013) 36 Seattle L Rev 1278. 20 The US DOJ began to replace nolo contendere pleas with guilty pleas in the early 1960s and stopped using consent decrees for cartel conduct in 1996. 21 For example, about 12,000 companies were identified as defendants in the 1423 detected cartels, a median of five members per cartel. Eleven cartels had 100 or more participants. Other measures of size could be geographic extensiveness or penalties, which are discussed below. 22 The duration of collusion is from the adoption of an effective agreement to when prices return to more competitive levels (including lags after formal collusion ceases). Sales figures from damages suits are preferred to those announced by other antitrust authorities. If sales by the cartelists are not available, then the most precise estimates from business-marketing sources are employed. Sales are nominal dollars, and thus are underestimated for older cartels. 23 We do not present analyses of cartel duration. We simply note that the PIC and the PFO data sets each show duration at a remarkably steady level over the past few decades as well as the last three centuries. Median duration for all cartels is about five years, but global cartels average 90 months. 24 The affected sales of two finance-industry cartels (each above $100 trillion, including derivatives) are omitted because the numbers appear to refer to assets affected rather than revenue flows. 25 Of course, $90 trillion in global revenues is a huge number, but it needs to be put into perspective. Focusing on sales of international cartels in North America, including the North American part of global cartels’ sales, estimates range from $37 to $44 billion. Revenues of all US firms in the construction, manufacturing and wholesaling industry groups – the most cartel-prone industry sectors – are reported to be $17.7 trillion in the 2007 Economic Census. Making an adjustment for Canada and multiplying by 29 years, such revenues total $550 trillion. Thus, we hazard that cartelized North American markets most likely comprise at most 6.7 per cent to 8.0 per cent of all relevant commercial sales.

The prevalence and injuriousness of cartels worldwide  27 3. Industry distribution of cartels During the period 1990–2019, international cartels arose primarily (56 per cent) in the manufacturing sector, with chemicals being the most prominent industry group; most are intermediate inputs sold to other manufacturers. Financial services, wholesale-retail trade, transportation and construction each account for 6–9 per cent of the number of cartels; most are business-to-business services. Thus, the great majority of cartelized markets supply inputs to other businesses. Since the 1990s, the dominance of manufacturing-sector cartels has been shrinking. Reflecting the changing mix of the structures of post-industrial economies, the shares of cartels in the services sector have expanded. 4. Corporate and individual cartel participants Tens of thousands of firms have participated in detected international cartels, but most are anonymous. The PIC data set identifies the names of 10,158 ultimate controlling parents26 that were involved in cartel activity during the period 1990–2019. In addition, there are about 4000 subsidiaries or intermediate operating companies of these parents cited by name. However, the largest number of participants are unnamed firms – more than 70,000 – not identified by name because some jurisdictions strictly follow national privacy laws. Many of the unnamed participants are banks that owned payment-card joint ventures such as Visa, Inc.27 Moreover, in several jurisdictions with strong privacy laws for government documents, cartel decisions often reveal the number of defendants only; for example, the Bundeskartellamt decision in the Waste-Packaging Collection prosecution simply states that there were 500 companies involved, and the German business press did not give the participants’ names. It turns out that the large number of anonymous cartelists implies only a small loss of information. One indicator is that anonymous parents accounted for a negligible share of total monetary penalties. The PIC data set lists unique monetary penalties for each cartel participation. By far the largest share of cartel penalties (63 per cent) was imposed on subsidiaries of large companies (the lion’s share are multinationals) that were identified by antitrust authorities as principally responsible for carrying out cartel activities. That is, the ultimate owners had passive roles in collusion, and their degree of responsibility for cartel conduct varies by jurisdiction.28 The ultimate parents themselves (or companies with simple corporate structures – single ownership layers) were legally responsible for 31 per cent of the penalties; in these cases the antitrust authorities held the top (or only) level of a company responsible for collusion. Thus, anonymous cartelists, while large in number, accounted for only 6 per cent of total cartel penalties.

Most are corporations, but a minor portion are holding companies, trusts, governments or families. Some of the parents were cited as jointly and severally liable by the antitrust or competition law authority, while in some other jurisdictions they were not mentioned in the decisions. PIC traces ownership of both. There is some double counting because of serial collusion. Thus, the numbers quoted are technically participations, not separate companies. 27 Thus, some of the unnamed defendants are not truly anonymous, because lists of the 22,000 US card-issuing banks could be constructed but doing so adds little information to the data set. And the portioning of penalties is unobtainable. In the PIC data set, there are about a dozen similar banking cases with large numbers of defendants. 28 Higher-level owners are generally not held responsible in the United States, but the reverse is true in the EU. 26

28  Research handbook on cartels In jurisdictions with criminal antitrust sanctions, individual cartel managers can be punished for price fixing with fines, incarceration and/or debarment. Prosecutors typically punish individuals along with their companies for the most severe egregious conduct. A total of 1382 executives were indicted for international price fixing during the period 1990–2019,29 of which 1164 (84 per cent) were penalized.30 Penalized individuals are concentrated among a small number of cartels and corporate cartelists. Only about 1 in 50 of all fined companies have had an employee or director penalized.31 Similarly, most cartels had no employee penalized. Of those cartels that have had at least one executive penalized, the mean number of individuals penalized is five per cartel. 5. Monetary penalties Antitrust and competition law authorities found 1157 international cartels guilty of crimes or competition-rule violations during the period 1990–2019. Total penalties imposed amount to $236 billion.32 Of these ‘guilty’ cartels, 89 per cent of the corporate participants paid monetary penalties of some kind, mostly fines. Total fines imposed33 worldwide amounted to $141 billion, a mean average of $230 million per cartel (median average $108 million). Civil damages paid by members of 265 cartels total $94 billion, or an average of $354 million per cartel.34 Both fines and damages are highly skewed numbers. Most executives indicted for price fixing are not penalized, but among the minority that were fined worldwide the average fine is roughly $59,000 per employee (the median fine is $50,000).35 About 73 per cent of cartel fines on individuals worldwide have been imposed by US courts, but since 2010 authorities in ten other criminal jurisdictions have accounted for the majority of fines on individual cartelists. For US international cartel enforcement, extradition of foreign nationals is a large and growing problem.36 During 1990–2019, prison sentences totalling more than 22,000 months were imposed on 410 cartelist executives worldwide, of which 77 per cent originated in the United States. The median US prison sentence is 14 months per person, and the median non-US sentence is 17

29 Many indicted executives agree to testify for prosecutors, thereby escaping with no or light sentences. 30 The PIC data set faithfully records fines and incarceration penalties by an antitrust authority. Debarment, disqualification, and dismissal actions are not quantified. 31 In Brazil, nearly all federal antitrust convictions of cartels are accompanied by two substantial fines on employees of each corporate cartelist. 32 All monetary penalties are converted into US dollars using the date a decision or settlement is first announced. 33 Recall that ‘imposed’ means announced fines, not necessarily paid fines. Some cartels are allowed to pay in installments over about five years, some have their fines reduced by appeals courts, and in some newer jurisdictions, officials have difficulties in securing any payments for fines. 34 Damages litigation through class action typically takes five or six years to be completed. The PIC data set measures from the year of filing to the year the first defendant offers a monetary settlement; the remaining defendants usually settle within a year or two. Mean average time is 5.3 years. Opt-out cases are shorter, but settlements are underreported. 35 US residents generally pay these fines personally, but in many other jurisdictions they are paid by employers. The average of 310 US fines and mandatory restitutions are $1.446 million, the median $70,000. 36 See J. Connor, ‘Problems with Prison in International Cartel Cases’ (2011) 56 Antitrust Bulletin 311. (Since 2010, the DOJ has extradited fewer than ten individuals for price-fixing violations.)

The prevalence and injuriousness of cartels worldwide  29 months.37 The great majority of these individuals have held positions in upper management or on boards of directors.38 Up until about 2010, incarceration penalties overwhelmingly originated in US courts, but since then the majority have been handed down by Brazil’s courts (for example, 19 cartel managers involved in the infamous Lava Jato cartels39), Israel’s 8, Japan’s 28, Korea’s 10 and the UK’s 12. In Germany and some other EU jurisdictions, names of individuals fined are confidential; anonymous defendants account for about 10 per cent of individual cartel penalties, and they are much less severely punished than are named individuals. B.

Geographic Differences in Cartels

In the wake of popular discontent with monopolies and cartels, several US states and Canada enacted antitrust laws in the 1880s, but evidentiary and remedies problems hampered effective prosecution. Passage of the Sherman Act in 1890 gave the US federal government effective criminal enforcement powers to convict hundreds of price-fixing cartels, but few international cartels were pursued and penalties were quite light until the early 1990s.40 The EU adopted a civil-administrative antitrust implementing statute (or ‘Regulation’) in 1962, thereby providing the European Commission (‘EC’) with specific powers to enforce the EU competition law provisions;41 but the EC was very cautious in enforcing these provisions until the late 1980s. By the early 1990s, only five jurisdictions had modern, effective anti-cartel laws, but soon after the Member States of the EU and many other nations began penalizing international cartels (Figure 2.1). The main point is that what had been largely a two-authority transatlantic activity before about 1990 expanded geographically to become a global enforcement enterprise. Cartelists now have many eyes trained on them, and while cartel fines originate from many directions, one can certainly question whether this increased vigilance has significantly discouraged cartel activity. In this sub-section, we examine broad geographic categories of cartel data. These regions also correspond to groups of jurisdictions with antitrust authorities that vary in their institutional capacities, maturities and acceptance of an antitrust culture among citizens and

37 Outside the United States, incarceration is often commuted (to home arrest, community service or probationary sentences) or followed by extremely lengthy appeals that delay prison sentences being served. 38 In Germany and some other EU jurisdictions, the names of individuals who are fined are withheld; anonymous defendants account for about 10 per cent of individual cartel penalties, and they are much less severely punished than are individuals cited by name in a decision. 39 C. McConnell, ‘Kovacic: Lava Jato is the World’s Most Important Antitrust Case’, Global Competition Review Online (27 October 2017) https://​glo​balcompeti​tionreview​.com/​kovacic​-lava​-jato​ -the​-worlds​-most​-important​-antitrust​-case [accessed 10 March 2022]. 40 See J. Gallo, K. Dau-Schmidt, J. Craycraft and C. Parker, ‘Criminal Penalties under the Sherman Act: A Study of Law and Economics’ (1994) 16 Research in Law and Economics 25. 41 Regulation 17/62 of 6 February 1962: First Regulation Implementing Articles 85 and 86 of the Treaty [1962] OJ 13/204 (superseded in May 2004 by 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 L1/1). The EU competition law provisions themselves were introduced via the adoption of the Treaty of Rome (which came into effect in January 1958).

30  Research handbook on cartels businesspersons. We are looking for hints about how differences in degrees of enforcement assertiveness have influenced cartel effectiveness.42 1. Cartel numbers and size Almost half (46 per cent) of the sampled cartels operated exclusively in Europe, of which 81 per cent fixed prices in Western (or West-Central) Europe.43 Moreover, because global cartels44 also had sales territories in Europe, after allocating these territories, 40 per cent of all international cartels fixed prices in Western Europe (see Figure 2.2). By contrast, ignoring global schemes, only 13 per cent of all international cartels operated exclusively in North America; counting global cartels with North American operations would raise that share to 16 per cent. Given that markets in Western Europe and North America are similar in economic size, the propensity for cartelization in Europe is roughly triple the rate of that of North America. Perhaps this indicates that business cultures in the United States and Canada are more compatible with antitrust objectives and more averse to violations of antitrust laws than their European counterparts. Detected cartels confined to the other four continents (Rest of the World, or ‘ROW’) are much more numerous than are North American cartels and almost as numerous as Western European ones. With global cartels distributed, ROW cartels account for 36 per cent of all regional operations. Their share increased markedly in the past 30 years. However, because the antitrust authorities in ROW tend to be less experienced and have more lengthy prosecutions, a smaller proportion of discovered cartels has been prosecuted (33 per cent) in ROW than in the other two regions (76–77 per cent prosecuted). Global cartels have the highest prosecution rate (88 per cent). Affected sales data need to be treated with caution. Mean average cartel sales data are deceptively high because a few large figures skew the means.45 The median size cartel is a more accurate average, namely, $2.1 billion (Figure 2.3).46 (Had guilty global cartels been shown separately, their median average sales size would be the largest, $6.665 billion. Instead, We identify high overcharges and long duration as the principal dimensions of cartel effectiveness, which is inversely related to consumer (or customer) welfare; stability is sometimes an additional characteristic of effectiveness. Some scholars refer to cartel ‘success’, but we eschew this wording as it is a cartelist’s perspective. 43 Except for Switzerland and, since Brexit, the United Kingdom, every nation in Western Europe is in the European Economic Area. In this chapter, Western (or West-Central) Europe has Finland, Germany, the Czech Republic, Austria and Italy as its eastern border. Several EU members are classified by the United Nations’ standards in Eastern Europe. The high share of cartels found in Western Europe is largely due to the longer history of effective anti-cartel enforcement by the EC and a surge in international cases prosecuted by EU Member States since 2000. Data on penalties are heavily weighted by EC prosecutions. 44 Global cartels operated in two or more continents, usually Europe and North America and often East Asia. Thus, areas of operation are double counted for the 201 global cartels in this Figure. Generally, the corporate national headquarters of members of global cartels are so diverse that most cannot be classified to one continent. 45 In fact, the large mean is driven largely by four recently discovered global banking cartels (such as LIBOR, Foreign Currency Exchange) with tentative sales estimates of $1.4 trillion. Antitrust authorities have been notably reticent about reporting affected commerce of financial-sector cartels with enormous derivatives’ values. Without these four cartels, the mean average sales for all cartels becomes $66.94 billion and the median $2.02 billion. 46 Note that because PIC data end in 2019, the UK is classified as an EU Member State. 42

The prevalence and injuriousness of cartels worldwide  31

Figure 2.2

Locations of price-fixing (with global cartels’ locations distributed)

Figure 2.3

Median affected sales, all guilty cartels, region of operation, 1990–2019

Figure 2.3 distributes global sales across four regions.47) The US and EU-wide cartels are the largest, $2.5 and $2.7 billion, respectively. Single-nation EU cartels are about half the size, That is, each territory affected by a global cartel is treated as a separate unit. Sales are often revised downward after conviction as detailed decisions become available. Several of the recent banking scandals (most global) with convictions not yet completed have enormously affected sales estimates. 47

32  Research handbook on cartels Table 2.1 Type and jurisdiction

Mean average monetary penalties, international cartels, 1990–2019 Number of

Average fines Number of

fines

imposed ($

damages cases

mil.)

Average

Average

Average

damages paid

severity ratioa

recovery ratiob

($ mil.)

(%)

(%)

US fines

158

260.7

225

384.8

21.6

Canada fines

64

7.06

84

11.4

14.9

104.7c 54.3

EC fines

163

261.2

2

845.0

11.1

85.0 203.8

EU NCA fines

399

78.4

12

243.6

30.8

Other Eur. fines

36

36.1

0

0

NA

NA

ROW fines

288

83.6

9

197.5

14.0

55.0

All jurisdictions d

1102

127.7

332

282.5

19.4

139.9

All cartels d

930

151.3

265

354.0

19.4

173.9

Notes: Severity = all penalties/affected commerce. Recovery = penalties/estimated overcharges; 620 unique ratios for 332 cartels. c There are 253 non-US fines’ recovery ratios that average 111.4 per cent. If one adds private settlements, the US and non-US mean recovery ratios are 138 per cent and 111 per cent, respectively. d There are more penalties than cartels because of multi-jurisdictional violations. a

b

viz., $1.4 billion. Somewhat surprising is the large median average sales size of international cartels operating in ROW: $1.6 billion. Cartel penalties and severity 2. Penalties are precise money metrics of geographic patterns of enforcement effort (Table 2.1).48 A total of $141 billion in fines were imposed on 930 cartels, several of them by more than one jurisdiction.49 Total fines on international cartels in Europe (principally by the EC and National Competition Authorities (‘NCAs’) within the EU) are 59 per cent higher than the $47 billion in fines imposed by US and Canadian authorities, which is explained mainly by the larger number of EU cartels.50 Fines in ROW total $24 billion and are rising fast. In addition to fines, 264 cartels had members that paid $94 billion in damages, of which 97 (37 per cent) paid only damages.51 When civil monetary penalties are added to fines, North American penalties far outweigh those in Europe, $135 billion and $81 billion, respectively. North American civil settlements are 92 per cent of the world total. Damages compensation outside North America has remained low until recently.52 While the absolute values of monetary penalties are impressive, relative to sales they are not. The mean average severity of worldwide cartel penalties is 19 per cent of affected sales, and it averages 10–15 per cent in most years. Severity ranges generally from 8 per cent to 14

48 Moreover, for complicated reasons, amounts of US and EU penalties by PIC actually exceed what is officially reported by the DOJ and EC. (For the DOJ, see Connor (n 9).) We ignore changes in penalties from appeals. 49 The US, Canada, EC, EU NCAs, non-EU Europe and ROW jurisdictions imposed 158, 64, 163, 399, 30 and 288 fines, respectively – 1102 fines in total. 50 It is not explained by severity. Mean severity per cartel in North America is actually 68 per cent higher than EU fines. 51 The average award of damages in 167 follow-on cases is $315 million, and $424 million in the 97 non-follow-on-same-jurisdiction cases. The latter include US cases inspired by EC decisions. 52 J.-F. Laborde, ‘Cartel Damages Actions in Europe: How Courts Have Assessed Cartel Overcharges’ (2019) 4 Concurrences 1 (charting the exponential growth of European private cartel damages cases, which rose from zero in 1998, to 50 in 2012, and to 239 in 2019. Courts in 30 nations are involved).

The prevalence and injuriousness of cartels worldwide  33 per cent across jurisdictions, except for much higher percentages among the EU NCAs and ROW authorities. However, the median average severities of 882 fines are quite low: from 3 per cent to 6 per cent (except Canada’s 1.6 per cent and a mere 0.1 per cent in ROW); adding private damages raises the US median severity to about 5 per cent of sales. The weighted average worldwide cartel payment is a mere 2 per cent of affected sales; it is highest for North American cartel payments and lowest among the global cartels. As interesting as these benchmarks may be, severity calculations cannot inform enforcers about the deterrence power of penalties. For that one needs data-intensive recovery ratios. 3. Cartel injuries and recovery Estimated aggregated discovered-cartel overcharges since 1990 are substantial – at least $24 trillion worldwide. Interestingly, projected cartel overcharges in Europe and North America are almost equal, as both their GDPs and severity ratios are roughly equal.53 Cartels operating in ROW seem to be the least affected by cartel pricing relative to its GDP, even though only a few of ROW jurisdictions are well equipped to enforce their anti-cartel laws.54 In Figure 2.4, we have calculated 620 all-important recovery ratios for 332 international cartels.55 The weighted mean average recovery is 140 per cent.56 US fines on average tend to recover slightly over 100 per cent of overcharges in the jurisdiction on average; EC and Canadian recoveries are lower. Something of a surprise is the robust average recoveries of fines by the EU NCAs: 204 per cent, led by the Italian, French, Spanish, and Portuguese authorities. Private class actions add considerably to recoveries in North America, but are largely absent in other continents. 4. Geographic differences among corporate cartelists There are at least 7200 companies that have been caught or punished for international cartel infractions worldwide. This number does not include thousands of parents of large groups or intermediate operating companies that were identified as culpable but were not directly penalized. Because historically antitrust enforcement is largely located there, cartelists by and large hail from rich countries. Of these companies, 77 per cent are headquartered in Western Europe and North America, and another 10 per cent in Japan (6.0 per cent) and Korea (3.7 per cent).57 This phenomenon is largely explained by difference in the industry mix and duration of cartels. We primarily refer to Brazil, Korea, Japan, Australia and South Africa, which had demonstrated institutional capacity by the early 2000s. More recently, diverse authorities (India, Singapore, Chile, Mexico, and others) are in the running. 55 It is comforting to note that this subsample is 26 per cent of the roughly 1300 convicted cartels in the PIC data set. However, the subsample is not entirely representative of even discovered cartels, as overcharge estimates take time to develop and publish and North American cartels are over-represented. 56 An internal survey of 15 national members produced only nine estimates of recoveries by fines imposed during the period 1996–2000. The average was 74.6 per cent of harm, and three were above 100 per cent. See OECD (n 7) Annex A. 57 The location and size of penalized cartel activity and the location of parents do not perfectly overlay. Consequently, parent groups frequently transfer large funds to penalized subsidiaries in other continents. PIC data from 1990 to 2010 reveal that 82 per cent of companies fined for price-fixing by US authorities are non-US; for the EC, the comparable percentage is 32 per cent. EU fine regulations specifically require ultimate controlling parents to pay their cartel fines to the EU Treasury, and, while less explicit for guilty subsidiaries, the same requirements hold in North America. An analysis of the 53 54

34  Research handbook on cartels

Figure 2.4

Mean recoveries, 332 international cartels, by type and region, 1990–2019

In fact, only 3.3 per cent of the cartelists are from Africa, 0.7 per cent from Oceania, 1.9 per cent from Latin America, 2.6 per cent from Eastern Europe and 7.7 per cent from Asia (excluding Japan and Korea). The distribution of monetary penalties pretty much tracks company numbers. 5. Geographic differences among individual cartelists Like corporate cartelists, the 863 indicted cartel executives were overwhelmingly citizens of rich nations: Europe (33 per cent), North America (26 per cent), Japan (13.0 per cent) and Korea (3.4 per cent). Interestingly, 37 per cent of these individual cartelists are associated with global cartels. Fines imposed on these executives are typically slaps on the wrist: the mean average of $3.8 million is distorted by a handful of large penalties; the median fine is $75,000. European executives or their employers have paid the most. Most fines and nearly all prison sentences are meted out by US courts. Prison sentences totalling 7876 months have been imposed on international cartelists. The largest recipients of prison sentences have been imposed on US citizens (they account for about 30 per cent of the prison-months). Relatively few Europeans have been fined or imprisoned, mainly because few countries in Western Europe have criminal antitrust laws. The greatest numbers of penalized individuals from ROW are citizens of Japan, South Korea and Taiwan: a few were condemned by their national courts, but the great majority worldwide have been sentenced by US courts

US balance of payments flows shows that ‘[a] large share of […] outward and inward transfers result from the prosecution of international cartels’: C. Bach, ‘Fines and Penalties in the US International Transactions Accounts’ [July 2013] Survey of Current Business 55, 55.

The prevalence and injuriousness of cartels worldwide  35 because of their participation in global cartels.58 (The majority of European cartelists have also surrendered to US courts.) C.

Distributional Consequences of Cartel Injuries

Cartel overcharges act resemble a system of regressive taxes. That is, effective collusion in the great majority of markets transfers income from relatively low-income buyers to relatively high income-owners and managers of the companies that raise selling price above the price but-for the collusion.59 The monopoly profits created by effective collusion are rents paid to the perpetrators in the form of stock-price increases, stock dividends and managerial rewards and perquisites. This general outcome most clearly occurs when a consumer good is cartelized, because such goods must be price-inelastic ‘wage goods’ and tend to be standardized, homogeneous items. Foods and supermarkets fit this category well, as does bid rigging against governments.60 Things get more complex when the good is a minor intermediate input purchased by manufacturers – the most common type of cartelized good. In this instance, the immediate income transfer is between two sets of manufactures. There are many cartelized steel and plastic products ultimately used in construction of roads and buildings; similarly, collusive prices in DRAMs and LCD Screens are passed on mainly to households with wide-ranging income levels. In such cases, ultimate distributional consequences are difficult to trace. Perhaps most difficult to analyse are the numerous and very large financial instruments that have been discovered to be cartels in the past decade. Markets for foreign exchange, credit default swaps and a panoply of derivatives priced off of the LIBOR index affect the fees paid by households for mortgages, personal credit and student loans, which is clearly regressive; however, the principal buyers are likely to be pension funds and sovereign wealth funds that in principle benefit broader income strata. Cartelization in these markets may therefore be income-distribution neutral. D.

The Special Dimensions of Global Cartels

The PIC sample identifies 198 global international cartels. Global cartels have several distinct characteristics.61 Although representing only 17 per cent of the number of penalized cartels, due to the fact that they were on average about six times larger in sales than the others, global schemes generated more than half of all international cartels’ sales and attracted 50 per cent of worldwide monetary penalties. Global cartels also achieve considerably higher overcharge rates than The majority of cartel fugitives are from the same three nations. Ironically, even high-income stockholders are injured by the elevated prices on cartelized goods, but this offset is minimal. 60 That is, tenders issued by governments have such highly specified dimensions that they are virtually homogeneous goods; the same may be said for commercial tenders. 61 A significant portion of global cartels are also members of interrelated, overlapping conglomerations, entities that we have deemed ‘supercartels’; see J. Connor, ‘Is Auto Parts Evolving into a Supercartel?’ (AAI Working Paper No. 13-04, 28 August 2013) www​.antitrustinstitute​.org/​~antitrust/​ sites/​default/​files/​WorkingPaper13​-04​.pdf [accessed 10 March 2022]. The bulk vitamins, banking, and auto-parts industry groups were organized in this fashion. 58 59

36  Research handbook on cartels other international cartels – 30–40 per cent higher than the typical cartel. However, partly because the geographic scope of global cartels is inevitably wider than the jurisdictions62 that penalize them, the severity of penalties on global cartels is lower than the average severities of the other international cartels. Moreover, the known overcharges generated by global cartels are a large multiple of the world’s antitrust penalties. Global price fixing is pursued by an elite group of cartelists. Only about one penalized international cartelist in eight has joined a global cartel. Individuals who were managers of global cartels – especially European and East Asian nationals – are punished with greater frequency than managers of non-global cartels. In sum, global cartels are the worst of the worst: bigger in sales size, covering more territory and jurisdictions, and causing relatively greater injuries to their customers. Although enforcement is moving in the right direction, global schemes are penalized less severely on average by antitrust authorities than their greater injuriousness would warrant.

III. DETERRENCE A.

The Economic Framework for Optimal Penalties

To oversimplify, the optimal penalty for collusion is the expected losses caused by the illegal conduct, which is roughly equal to the overcharges paid by the cartel’s direct customers, divided by the cartelists’ expected probability of being detected and punished. Other measures – such as focusing on the cartel’s expected gains – are usually believed to be less likely to lead to optimal deterrence.63 This formula is correct if several conventional assumptions are made; for example, cartelists are risk neutral and rational in calculating the benefits and costs of risky illegal behaviour.64 Many sources place the probability that hidden cartels will be detected at 10–30 per cent,65 and the chance of conviction of discovered corporate cartels is very high, probably above 90 per cent.66 To be conservative about the size of optimal penalties, we will assume a high detection probability of 25–30 per cent. If cartelists on the whole are risk-seeking, then the

62 The maximum exposure of a global cartel is LIBOR, which has attracted investigations by at least ten antitrust authorities (of 140 worldwide). 63 For an extended discussion of these issues on a corporate and individual basis, see J. Connor and R. Lande, ‘Cartels as Rational Business Strategy: Crime Pays’ (2012) 34 Cardozo Law Review 427, 431–47. 64 Mostly to simplify calculations, this formula may require a number of additional assumptions. Those assumptions include that cartelists are well informed about their market conditions, such as the elasticities of demand and industry supply. If they are, then expected monopoly profits will be equal to actual incremental profits from collusion. However, the actual economic profits from cartelization are typically lower than the theoretical monopoly profits because of friction within the cartel; even in highly profitable cartels, there will be a tolerably small amount of cheating on the agreement. Experienced cartelists likely take some cheating into their expectations about collusive profits. If so, overcharges are a good proxy of economic profits. The same can be said for expected penalties. 65 See D. Ginsburg and J. Wright, ‘Who Should Be the Target of Cartel Sanctions?’ (2010) 6 Antitrust Sanctions Competition Pol’y Int’l 3. 66 See Connor and Lande (n 63) 466–68.

The prevalence and injuriousness of cartels worldwide  37 optimal penalties that will result from the calculations that rely upon these estimates will be understated.67 To apply optimal deterrence principles, both harm to victims and encompassing penalties must be relatively measurable. There are three penalties for which fairly precise money metrics are publicly available: (a)

Fines imposed by US courts on both companies and on individuals.68 These amounts are known precisely. (b) Settlements paid to direct and indirect customers from private damages suits in the US. These figures are very largely in the public record and verifiable. They often can be significant (for a sample of 71 US cartel cases the median average settlement was 37 per cent of single damages and the unweighted mean was 65 per cent).69 Most cases settle, and final verdicts are relatively unusual. (c) The disutility (or inferred wage and utility losses) arising from imprisonment and house arrest for culpable executives in criminal cartel prosecutions. After an extensive analysis of analogous situations, we assigned what we believe to be a generous (dis)value of $6 million per year of imprisonment or house arrest.70 On the one hand, there are other costs of collusion that, as a practical matter, usually cannot be measured by outsiders. These include: (i) legal costs of defendants (these are rarely revealed publicly); (ii) managerial costs of coordination of the legal defence, corporate distraction, public and investor relations, etc.; (iii) the size of smaller opt-out settlements that are subject to confidentiality agreements (by contrast, large settlements are often material to financial statements and are therefore published); and (iv) reputational losses.71 To the extent that the penalties we can measure are smaller than the actual total penalties and costs associated with collusion, our computations will lead to an optimal penalty higher than it needs to be. We believe, however, that these non-measured costs of collusion are usually relatively small and transitory compared to the three that we list above, but we know no way of proving their relative magnitudes.72

The simple model of deterrence assumes that cartel participants are ‘risk-neutral’ – neither risk-averse nor risk-loving. 68 On the definitions of fines, see J. Connor and R. Lande, ‘Fine’, in W. Kovacic, D. Healey, R. Whish and P. Trevisán (eds), Global Dictionary of Competition Law (Concurrences, Paris, 2022). 69 See J. Connor and R. Lande, ‘Not Treble Damages: Cartel Recoveries Are Mostly Less than Single Damages’ (2015) 100 Iowa L Rev 1997, 1997. 70 See Connor and Lande (n 63) 449–54, for the methodology used. 71 Studies using stock market prices show no long-lasting effects after two years. See C. Alexander, ‘On the Nature of the Reputational Penalty for Corporate Crime: Evidence’ (1999) 42(S1) The Journal of Law & Economics 489, Table 2 (showing that average stock-price movements of less than 1 per cent due to announcements of antitrust price-fixing fines). 72 A cartel is an illegal joint venture, a ‘virtual company’ with no assets other than the human capital and devotion of its managers and some historical records. The costs of the managers’ labour and expenses related to travel should be netted out from the cartel’s revenues, but are usually negligible relative to the profits generated. However, to keep the cartel secret, the number of managers that manage a cartel is kept small – typically, to only two or three executives, who meet and communicate sporadically. Thus, for all practical purposes, overcharges are equivalent to the aggregate monopoly profits from cartel conduct, and the quotas or profit-sharing rules of the cartel determine the monopoly profits of each corporate cartelist. Nearly all the overcharges imposed on cartel customers are economic and accounting profits, 67

38  Research handbook on cartels On the other hand, cartels cause losses of income to customers that require knowledge of participants’ internal transactions to be measured accurately.73 Full overcharges themselves rarely are recouped by victims. Moreover, several harms to society are not included in penalties paid by cartelists, neither fines nor overcharges paid to direct74 purchasers. These additional harms include deadweight losses, umbrella effects75 and the time value of money. Deadweight losses (the allocative inefficiency effects of cartel pricing) can reach up to 50 per cent of the overcharge, although our best estimate is that these losses usually are only about 10 per cent to 20 per cent of the overcharge.76 Because fines and payments to injured parties usually are made many years after the collusive period ended, all of these penalties ought to be restated in constant dollars to reflect general inflation and the time value of money. The victims lose the earnings that could have been made by them on the money illegally confiscated by cartelists (who get to keep such additional profits themselves). The adjustments necessary to compensate for the absence of prejudgment interest can be significant. Nevertheless, we are unaware

which mostly accrue to the owners, stockholders, workers and cartel executives whose compensation or promotions are tied to the profitability of their companies or divisions of the companies. (Dead-weight losses are damages to the whole economy. They are not captured by the cartelists.) High-cost participants may exhibit low or zero accounting profits from a cartel, while their economic profits are substantial. In other words, cartels can prevent or delay corporate bankruptcies. 73 Our results might be conservative for a methodological reason we have not yet discussed. Many of our affected sales figures are derived from decisions of antitrust authorities and might be overly small. Higher sales data would tend to lower recovery ratios, mainly because figures derived from plaintiffs or seemingly reliable third-party sources often are larger than the sum of corporate sales employed in DOJ sentencing memoranda. There may be legally defensible reasons for such understatement. For example, because of the high degree of reliability of evidence needed to convict corporations for crimes, the DOJ may narrow the true collusive periods, geographic region, or scope of products employed for calculating sales to that which can be proven ‘beyond a reasonable doubt’. On the other hand, prosecutors sometimes may uncritically accept arguments made by defendants that diminish the scope of the affected market because of time pressures in settling guilty plea agreements, or because the government lacks the resources necessary to disprove defendant assertions. An example is In re Ready-Mixed Concrete Antitrust Litig., 261 FRD 154 (SD Ind. 2009). Ready-mix concrete is a standardized product; the counties involved and the time period were not at issue. A sales figure of $680 million for all seven firms involved in the cartel was reported in the local press; all seven paid civil settlements. The sales information purportedly came from transcripts of a jury trial of two convicted executives and from the testimony of the plaintiffs’ class expert in fairness hearings (plaintiffs prevailed). Sales according to DOJ documents were much less. One participant was granted amnesty; two others were not charged, most likely because of cooperation agreements; and the DOJ used a smaller geographic market definition than civil plaintiffs. When one adds up the affected sales from the DOJ sentencing memoranda for the four companies that were criminally convicted of price fixing through plea agreements, its total is $391 million. Taking into account the fact that two of the smallest cartel members were not convicted because of bankruptcies, the DOJ’s total market affected sales is as much as 40 per cent lower than the affected sales proven by the private litigants. These numbers are updated and slightly revised from those shown in Table 5 of J. Connor, ‘Cartel Overcharges’, in J. Langenfeld (ed.), The Law and Economics of Class Actions, Research in Law and Economics (Emerald House Publishing Ltd, 2014). One should note that 93 (or 5.5 per cent) ‘ineffective’ (zero overcharge) episodes are included. 74 We are not considering harms to indirect purchasers. Since cartel overcharges can be marked up before being passed along in the distribution chain, this omission could be significant. 75 For a definition and analysis of umbrella effects, see Connor and Lande (n 63) 461–62. 76 See Connor and Lande (n 63) 457–61.

The prevalence and injuriousness of cartels worldwide  39 of deadweight losses, umbrella effects or prejudgment interest ever being awarded in a US antitrust case.77 B.

Optimality of Cartel Penalties

The true detection rate is a critical issue in assessing cartel penalties. As mentioned above, starting with Bryant and Eckard,78 multiple statistical analyses of the stock-price movements of penalized corporate cartelists have identified detection rates (more precisely, death rates) in the range of 10–20 per cent.79 Subsequent analyses of later periods and other geographic regions have reaffirmed this figure for similar public companies.80 Moreover, confidential surveys of experienced antitrust attorneys have also shown that these lawyers believe that the cartel detection rate is between 10 and 30 per cent.81 Consequently, several leading legal– economic scholars of cartels, including the present authors, have adopted a 25–30 per cent range as an assumed parameter for investigations into the deterrence power of cartel fines and other penalties.82 Under reasonable assumptions, these analyses have generally found significantly suboptimal cartel penalties.83 However, there are potentially at least two logical errors involved in asserting that cartel detection rates are at these levels. First, the statistical analyses of detection have been misinterpreted.84 They have in fact measured the likelihood of the termination of collusion from all sources, both from prosecution (‘antitrust’ death) and from the breakdown of collusion unrelated to antitrust enforcement (‘natural’ death). Consequently, detection is lower than the level or range adopted by many scholars, though how much lower is unknowable. Second, some scholars interpret the statistical studies as measuring the annual probability of the cessation of collusion, whereas the appropriate measure is the probability over the entire duration of a collusive episode (some call it ‘dynamic’ detection). If the likelihood of

We are aware of punitive interest rates being applied by a court in a Finnish damages case; see J. Connor and T. Kalliokoski, ‘The Finnish Asphalt Cartel Court Decision on Damages: An Important EU Precedent and Victory for Plaintiffs’ (2014) 2 Antitrust Chronicle 1. 78 P. Bryant and E. Eckard, ‘Price Fixing: The Probability of Getting Caught’ (1991) 73(3) The Review of Economics and Statistics 531. 79 See C. Veljanovski, ‘The Effectiveness of European Antitrust Fines’ (13 November 2020) https://​ssrn​.com/​abstract​=​3730361 [accessed 10 March 2022] 22–24. One problem with these studies is that most illegal collusion is carried out by either small companies unlisted on stock exchanges or the unlisted subsidiaries belonging to sprawling multinational conglomerates. Collusive conduct by unlisted companies could vary from that by listed ones. The other caveat is that the appropriate probability is that of detection and conviction. The chances of a guilty cartelist being convicted after a full investigation are certainly less than 90 per cent and in some jurisdictions possibly as low as 60 per cent. Thus, taking into account conviction likelihoods alone, the appropriate probability is lower than 15 per cent, somewhere in the range of 9 per cent to 14 per cent. Finally, there is a distinct possibility that the introduction of enhanced enforcement methods (leniency programmes, variance-screening algorithms, international data-sharing among authorities, and the like) could have raised the chances of detection; see R. Abrantes-Metz and P. Bajari, ‘Screen for Conspiracies and Their Multiple Applications’ (2009–10) 24 Antitrust 66. 80 Connor and Lande (n 63) Table 1. 81 Most worked in the EU or North America. 82 See Connor and Lande (n 63) 465–68. 83 ibid. 84 Harrington and Wei (n 10). 77

40  Research handbook on cartels detection is, say, 9–15 per cent in each year of collusion, then the likelihood of detection over a multiyear period of enforcement is much higher.85 The longer the duration, the higher the ‘dynamic’ rate. These higher episodic detection rates result in fines much closer to optimally deterrent levels.86 Two problems with relying solely on statistical models are that they often have arcane restrictive assumptions and that they are not the only sources of wisdom about detection rates. For example, Harrington and Wei do not differentiate between implicit and (illegal) explicit collusion. Legally this distinction is crucial because only explicit collusion is illegal in most jurisdictions. Moreover, confidential surveys of antitrust practitioners about detection rates are at odds with quantitative modelling. It is important to consider that practitioners can observe illegal conduct by their clients that cannot be observed by outsiders. These experts are in positions to have privileged, if limited, information about their clients’ participations in undiscovered, never-prosecuted cartels.87 Prosecutors and defence attorneys tend to place ‘dynamic’ detection rates at 10–30 per cent for an entire episode. Should detection rates be in this range, we are back to the lion’s share of cartel penalties being sub-optimal. In Section II.B, we noted that recovery ratios of cartel fines, private damages suits and other monetary penalties are highest in the world for the United States and Canada. Therefore, focusing on US cartels offers the most stringent test of optimality of penalties. The median percentage overcharges by cartels that ended their collusion between January 1990 and December 2018 are summarized, by type of cartel in top three rows of Table 2.2.88 The third row summarizes 1024 episodic overcharges available for cartels detected during the years 1990–2018. The ‘All Years’ row summarizes 1705 episodic overcharges for every estimate available for the entire time period (1700–2018). Utilizing a median average figure helps adjust for a high degree of statistical skewness in each cell of Table 2.2. Historical overcharges vary systematically by type of cartel and temporal enforcement characteristics.89 Seven types of cartel are distinguished. In most historical periods, international cartels have exhibited larger overcharge rates than cartels with only domestic companies participating.90 ‘Legal’ cartels – many of them operating openly, such as registered export cartels – generally have slightly higher returns than clandestine cartels detected and punished by antitrust authorities. Bid-rigging schemes and buyers’ cartels have distinctly lower overcharges than classic price-fixing cartels. However, none of the rows stray far from the median average

If p is the probability of detection in one year in ratio format and N is the number of years of collusion, then the episodic probability of detection is P = 1 – (1 – p)N. See Harrington and Wei (n 10) 1984. For a five-year episode and p = 0.09 to 0.15, P = 1 – (1 – p)5 or P = 0.38 to 0.56; similarly, a seven-year period of collusion implies P = 0.48 to 0.68. By this reasoning, an annual chance of detection of 9 per cent to 15 per cent is equivalent to a 38 per cent to 68 per cent likelihood of episodic detection, which is 2.5 to 4.5 times higher than the annual rate. 86 Velkanovski (n 79) 24–25. 87 For example, these antitrust specialists may know of leniency applications regarding cartels that the authorities declined to prosecute. 88 These numbers are updated and slightly revised from those shown in Table 5 of Connor (n 73). One should note that 93 (or 5.5 per cent) ‘ineffective’ (zero overcharge) episodes are included. 89 In Table 2.2, for example, effective corporate leniency programmes were implemented after 1999. 90 The 2000–18 period is an interesting exception that is attributable primarily to a large number of US pay-for-delay pharmaceutical cases. For details on these cases, see J. Connor, ‘Antitrust Developments in Food and Pharma’ (2015) 7(1) Annual Review of Resource Economics 375, Table 2. 85

The prevalence and injuriousness of cartels worldwide  41 Table 2.2 Cartel episode end date

Median average overcharges, by year and type Membership National

Legal status International

Found guilty

Conduct Legal

Bid rigging

Classic price fixing

Buyers’ cartels

ALL TYPES

Median percent a 1990–1999

16.65

25.00

24.60

20.75

19.00

24.90

22.05

24.00

2000–2018

24.00

20.00

20.00

27.50

17.05

24.00

17.00

20.10

1990–2018

20.70

22.50

22.20

25.60

17.95

24.25

16.90

22.30

ALL YEARS

18.30

25.00

22.00

27.50

18.30

25.00

18.10

23.00

Note: a Medians of the point estimates or, where appropriate, of the midpoint of range estimates. Excludes hundreds of ‘peak’ overcharge estimates that refer to maximal price effects over brief periods of time, rather than an entire, usually multi-year, episode of collusion. Includes 93 zero estimates. Source: J.M. Connor, Price-Fixing Overcharges Master Data Set, a spreadsheet dated 4 December 2018

for all types of cartels, of 23 per cent.91 As for time periods, unadjusted median overcharges tended to decline by about ten percentage points from before 1920 to the early twenty-first century, and this pattern has been confirmed by more formal econometric testing methods.92 Median average overcharges since 1999 appear to hover in the 20–23 per cent range. If one generously assumes a 25–30 per cent chance that a cartel will be detected, ceteris paribus, the optimal total penalties on cartels ought to be within the range of 55.5–110 per cent of affected commerce, depending on type of cartel and time period being examined.93 Variation in cartel overcharges is limited to a small number of cartel industries.94 Moreover, there are only small differences in overcharge rates due to the type of publication surveyed, method of analysis employed, or whether courts were the original sources of the overcharge information.95

91 The results of verdicts in private cartel overcharge cases are similar. We searched for every cartel overcharge we could find in a final verdict in a US antitrust case. We found 25. They yielded an average cartel overcharge of 31 per cent and a median overcharge of 22 per cent. See J. Connor and R. Lande, ‘How High Do Cartels Raise Prices? Implications for Optimal Cartel Fines’ (2005) 80 Tulane L Rev 515, 515 and 551–59. 92 Bolotova developed a meta-analysis of 406 cartels that colluded from 1700 to 2005: Y. Bolotova, ‘Cartel Overcharges: An Empirical Analysis’ (2009) 70 Journal of Economic Behavior & Organization 321. She showed that, controlling for many structural and conduct characteristics, overcharges declined by about 8.8 percentage points from the period before 1890 to 1990–2005. 93 The smallest median overcharge in the top two rows of Table 2.1 is 16.65 per cent, which if divided by 0.30 (a high estimate of the probability that the cartel will be detected and convicted) equals 55.5 per cent. Similarly, the largest median overcharge is for illegal cartels in 2000–18: 27.5 per cent, which if divided by 0.25 (the lower estimate) equals 110 per cent. 94 A meta-analysis of overcharges by Bolotova (see (n 92) Table 4) found that the food, timber and machinery industries had significantly lower overcharges (5 to 10 per cent lower than the reference group, chemicals); services were 11 percent higher. These are only 4 of about 30 broad industry groups tested. 95 Readers especially interested in these issues are urged to consult more formal statistical analyses, such as J. Connor and Y. Bolotova, ‘Cartel Overcharges: Survey and Meta-Analysis’ (2006) 24 International Journal of Industrial Organization 1109 or J. Connor and D. Werner, ‘Variation in Bid-Rigging Cartels’ Overcharges’ (27 October 2018) https://​ssrn​.com/​abstract​=​3273988 [accessed 10 March 2022].

42  Research handbook on cartels Table 2.2 reports median average overcharges, not the more familiar mean average overcharges. In every cell of Table 2.2, mean average overcharges are higher than medians, because in every category a few cartels overcharge by extremely large amounts.96 The mean overcharge by domestic cartels up to 2013 was 35 per cent, and the mean was 56 per cent for international cartels. The overall mean average was 49 per cent.97 Which averaging is the better metric for optimal deterrence purposes? We do not know of a clear answer as to whether means or medians are better. Another issue is whether deterrence should be based upon the expected harms generated by effective cartels only, or by all cartels. We included all cartel episodes in our sample, even though approximately 5.5 per cent did not effectively raise prices. If, for policy purposes, one should focus only on discouraging cartels that succeeded in raising prices, cartels with zero overcharges should be omitted from our sample. This would raise both the median and mean results.98 To address the issue of optimality more formally, we assembled and analysed a large sample of US-convicted cartels and the three most important sanctions: fines actually paid, payouts made in all the private cases filed against these cartels, and the deterrent value of prison sentences imposed.99 We undertook this analysis for every US-sanctioned cartel during 1990 to 2005 for which we were able to ascertain the necessary data, using the methodology summarized in the previous section. Our survey of experts’ opinions and economic analyses suggests that the probability of detecting hidden cartels is 10–30 per cent and the chance of conviction above 90 per cent. To be conservative about the size of optimal penalties, however, we assumed in our analysis a 25–30 per cent probability that hidden cartels will be discovered and convicted. (If, as seems likely, cartelists on the whole are risk seekers, however, then this formula will understate the optimal penalties.) The overall results for the 75 cartels show that on average the value of the imposed US sanctions has been much less than it should have been for society to obtain optimal deterrence against cartelization.100 If mean averages of overcharge figures are used, the total value of the imposed sanctions are only 16–21 per cent of their optimal level. If median figures are used, the imposed sanctions averaged only 9–12 per cent of optimality.101 We found only one unusual cartel for which the totality of sanctions was possibly larger than optimal. A second cartel was probably optimally sanctioned. The other 73 cartels, however, were sub-optimally sanctioned, and most of them substantially so (the median average was less than 10 per cent of optimality).102 Connor (n 8) addresses this issue. He concludes that the studies that produced the highest estimates were generally methodologically sound and fit theoretical expectations (that is, cartels with very few members in highly inelastic markets). 97 See top row of Table 7 in Connor (n 8) 294. 98 ibid third row. 99 We valued the disincentive effect of a year spent in prison or under house arrest by a corporate executive at $6 million. See Connor and Lande (n 63) passim. 100 See Connor and Lande (n 63) passim. 101 ibid. 102 ibid. One interesting factor that helped drive these conclusions is the relatively small effect of prison sentences (that is, they usually only constituted a modest portion of the imposed sanctions). Their mean value per case was a relatively modest $13.6 million, or 17 per cent of the average fine (the median is zero because for the majority of the cartels in the sample (48 out of 75) there was no imprisonment). See Connor and Lande (n 63) 475, footnote 239. Even though we valued the deterrence from a three-year sentence at $18 million (which is more than most estimates of the value of an entire life), this pales in 96

The prevalence and injuriousness of cartels worldwide  43 Because the recent historical level of monetary sanctions in the United States averages only 9–21 per cent of the optimal, it follows that United States’ sanction levels should be five to ten times their historical levels. A fortiori, the current overall level of sanctions imposed by the rest of the world also should at least be quintupled.

IV. CONCLUSIONS The sanctions imposed on cartels by the United States are, overall, the highest in the world. That point notwithstanding, perhaps the most important policy concern in this area is whether even these sanctions are high enough to provide optimal deterrence. If they are not, then a fortiori the cartel sanctions imposed by the rest of the world also should be increased. Thus, even if detection rates are well below 15 per cent, there is still significant underdeterrence of cartels worldwide. Whether this suboptimality of sanctions should be remedied by increased corporate fines, a heavier reliance on private lawsuits or increased use of prison for cartel violations is likely to depend upon factors unique to each country. We make no judgement as to which combination of increased sanctions is likely to be optimal for which country. To move in the correct direction, however, we propose five specific recommendations that can and should be implemented, at least in jurisdictions with tough anti-cartel enforcement.103 First, legislation should double the guidelines employed to calculate criminal antitrust fines. Alternatively, the US Sentencing Commission could double its current presumption that cartels raise prices by an average of 10 per cent (median cartel overcharges since 1990 have averaged 23 per cent), or most antitrust authorities can use their existing discretion to choose the higher starting points or maximal aggravating factors in existing fining guidelines.104 Ratcheting up penalties for international cartels is especially needed. Second, either legislation or judicial guidelines should add prejudgment interest to both private treble damage actions and criminal fines. Third, several studies conclude that the benefits of intensified detection and tough enforcement outweigh the costs; hence, upon requests to legislatures, the budgets of the antitrust authorities should be increased significantly. Fourth, antitrust authorities ought to impose more innovative non-monetary penalties, such as corporate debarment from government tenders, banning convicted cartel managers from employment in their company and their employer’s industry and industry restructuring through mandatory asset sales. Finally, comparison to the possible rewards from cartelization: ibid. Nevertheless, the absence of a criminal sanction correlates with an exceedingly small overall sanction. Almost all of the 15 cartels with actual sanctions that were less than 2 per cent of optimal penalties had no criminal sanctions imposed: ibid. The absence of a prior criminal conviction means that obtaining optimal sanctions in private damages actions is hampered by having to prove the fact of collusion. By contrast, the unusual E-Rate cartel case involved 626 months of prison, which constituted 85 per cent of the sanctions in that case; see ibid 451–55. 103 Although all five would be desirable if implemented in the United States, we emphasize that we make no judgement whether similar provisions should be implemented in other specific nations. See J. Connor and R. Lande, ‘Does Crime Pay? Cartel Penalties and Profits’ (2019) 33(2) Antitrust 29. 104 The DOJ currently starts fine calculations at the lower (10 per cent of affected commerce) end of the US Guidelines range instead of the allowable 20 per cent end of the range. The EC typically starts at 15 per cent or 16 per cent of EU affected turnover instead of an allowable 30 per cent. Neither uses their existing powers to raise fines for recidivism very often; perhaps mere serial collusion ought to be punished.

44  Research handbook on cartels the antitrust authorities in the United States, the EU and other jurisdictions should implement rewards or bounties for whistleblowers. Each of these recommendations has been analysed in much more depth insofar as they might be suitable for implementation in the United States.105 The evident international harmonization of anti-cartel policies and enforcement over the past several decades suggests that the said recommendations also would reduce injuries from cartelization in most other jurisdictions.

See Connor and Lande (n 103).

105

3. An historical account of anti-cartel enforcement Susanna Fellman and Martin Shanahan

I. INTRODUCTION This chapter deals with anti-cartel enforcement from a historical perspective. The discussion focuses on the appearance and distribution of legislative approaches aimed at the anti-competitive behaviour of cartels, trusts and other private sector forms designed to reduce competition between firms. We do not include all possible forms of legislation designed to promote market competition markets, nor the full range of possible enforcement mechanisms that authorities may use to counter anti-competitive behaviour. Our focus is on the legislation and policies directly addressing cartels, trusts and monopolies, but it is not possible to separate these completely from other policies aimed at prohibiting restrictive agreements or advancing competition. We also omit legislation limiting other types of restrictive practices, such as patent legislation and merger control.1 Policies against cartels can cover a wide spectrum of approaches. These can include legislation that prohibits certain types of cooperation; regulation that requires particular types of agreement to be registered; or legislation that proscribes some forms of company structures. Anti-cartel laws may also include statutes that require markets exhibit certain levels of competition, or that firms in those markets follow particular processes when operating or procuring tenders.2 Anti-cartel frameworks may not mention individual firm activities directly but create market conditions (such as a minimum number of operators or degree of openness to international trade) that have the effect of precluding successful cartel creation. Regardless Legislation designed to promote competition can include areas as diverse as merger and acquisition law, foreign investment controls, international trade and public sector regulation. In this chapter the terms anti-cartel and antitrust are used interchangeably, while our use of the phrase ‘competition legislation’ refers generally to statutes focused on groups of firms cooperating in a manner designed to reduce competition. For discussions and examples of the range of dimensions involved see: E. Alemani, C. Klein, I. Koske, C. Vitale and I. Wanner, New Indicators of Competition Law and Policy in 2013 for OECD and Non-OECD Countries, OECD Economics Department Working Paper No 1104, www​.oecd​ -ilibrary​.org/​economics/​new​-indicators​-of​-competition​-law​-and​-policy​-in​-2013​-for​-oecd​-and​-non​-oecd​ -countries​_5k3ttg4r657h​-en, 96 [accessed 23 February 2022]; A. Bradford, A.S. Chilton, C. Megaw and N. Sokol, ‘Competition Law Gone Global: Introducing the Comparative Competition Law and Enforcement Datasets’ (2019) 16(2) Journal of Empirical Legal Studies 411; and S. Fellman and M.P. Shanahan (eds), Regulating Competition: Cartel Registers in the Twentieth Century World (Routledge, 2016). 2 For example of the degree of overlap, see the United Nations Model Law on Competition, the purpose of which is ‘[t]o control or eliminate restrictive agreements of arrangements among enterprises, or mergers and acquisitions or abuse of dominant positions of market power, which limit access to markets or otherwise unduly restrain competition, adversely affecting domestic or international trade or economic development’: United Nations Conference on Trade and Development, Model Law on Competition (United Nations, 2007) 3. 1

45

46  Research handbook on cartels of the approach (or mix of approaches) adopted, the ultimate aim of all such legislation is to prevent or diminish those activities by firms that are perceived to disadvantage other suppliers ‘outside’ the cartel, or customers, or both, unfairly. In particular, our account examines the foundation and spread of anti-cartel legislation that was designed: to prohibit cartels and trusts completely (outright prohibition); to curb their worst effects (prevent abuse of market power); or to monitor and record their existence (establish a cartel register). Apart from the case of outright prohibition, every jurisdiction adopting legislation of this type also varied in whether their approach involved strict interpretations (an intolerant attitude to transgressions) or less strict interpretations (a tolerant approach).3 For example, a law to prevent abuse of market power may in one (tolerant) jurisdiction allow price fixing if in the national interest, while in another (intolerant) jurisdiction the same price fixing may be prohibited. This variation of anti-cartel enforcement legislation reflects variation in historical antecedents, legal influences and the economic and political circumstances when the legislation was introduced.4 In this chapter we describe the broad development trends in the enforcement of these anti-cartel practices, mostly across Western countries. Despite a diversity of attitudes towards cartels in the first decades of the twentieth century, in the latter half of the century there developed globally an increasing intolerance towards cartels and restrictive agreements. Today legislation to control and prohibit organized anti-competitive behaviour is more comparable, albeit still with clear differences in focus. Countries also differed in the route they travelled to reach contemporary anti-cartel policies, with their paths being influenced by their legal traditions and dependent on specific economic and political contexts.

II.

CONCEPTUAL FOUNDATIONS

At their heart, anti-cartel policies assume that market actors should be prevented from cooperating in a manner that disadvantages competitors or customers other than through open and transparent market mechanisms. This starting point has a long history, although there are only faint echoes that link modern concepts of competition and anti-cartel policy with pre-eighteenth-century writers’ efforts to explain what made a ‘just price’, or what made a monopoly’s behaviour unacceptable. While several early writers, such as Aristotle (384–322 bc), Thomas Aquinas (1225–74) and Leonard Lessius (1554–1623), explored these ideas, their work was based on social assumptions almost unrecognizable today. For example, Aristotle described ‘just’ prices in the context of isolated exchange between two people; Aquinas argued that for a price to be just, it should equal a good’s ‘worth’. Lessius saw the just price as one set by public authorities for the general good or by people’s general estimations. More obvious for this discussion, he condemned prices ‘set by a conspiracy of sellers’ and ‘where by force or fraud, import of goods from another source is prevented in order to maintain a higher

3 Another level of variation between jurisdictions is the degree to which the laws are actually enforced in practice. In several jurisdictions ‘strong’ antitrust legislation has little practical impact as the bureaucratic infrastructure is not adequate. For discussion, see e.g. M.P. Shanahan and S. Fellman, ‘Shifts in Government Business Relations: Assessing Change Using the Restrictive Business Registers in the OECD, 1945–1995’ (2019) 63(8) Business History 1253. 4 Fellman and Shanahan (n 1).

An historical account of anti-cartel enforcement  47 price in the state’.5 Although the religious and moral frame of reference of these earlier writers is dismissed in modern economic theory, they still resonate (albeit quietly) with motivations behind regulations prohibiting abuse of market power, or predatory pricing. Adam Smith’s A Theory of Moral Sentiments (1759) and The Wealth of Nations (1776) both built on these ethical foundations. Writing before the industrial revolution, Smith had nevertheless experienced markets with multiple buyers and sellers and arm’s-length exchange of goods, making his ideas on ‘natural’ and ‘market’ prices’, and on the damage caused by small numbers of sellers influencing the market, clearly recognizable to modern readers. Smith’s evocative warning against conspiracy is frequently cited in support of anti-cartel policy positions: ‘People of the same trade rarely meet together even for merriment and diversion, but the conversation ends in a conspiracy against the public, or in some contrivance to raise prices.’6 Theoretical discussion on the market impact of monopoly (and noncompeting groups) was relatively late to develop in the second half of the nineteenth century.7 Precursor insights into the problems of non-competitive groups (J.S. Mill, 1848, UK) and of monopoly and duopoly (A.A. Cournot, 1838, France; Charles Ellet Jr, 1839, USA) emerged separately in different countries at this time. Alfred Marshall, whose life spanned the gilded age and the rise of trusts, and whose work influenced later generations, reflects views on both the benefits of cooperation between small firms and its limits when applied to big business.8 While proto-economists debated the pros and cons of antitrust theory through the late nineteenth and early twentieth centuries, differences in nations’ antitrust practices were at first eroded and then swept away by political events that saw antitrust attitudes dominate industrial and competition policy after the Second World War. The earlier debates often reflected different social and cultural perspectives on individual and collective action and turned on the consequences of cooperative behaviour versus stricter individualistic, antitrust policies.9 Stigler laid out the modern, firm-centric economic arguments against collusion by cartels, establishing a narrower focus of economic research on the costs and benefits of cooperation within cartels.10 The structure–conduct–performance paradigm of antitrust policy linked to the Harvard School of economics came to be at least partially eclipsed by the Chicago School’s view on theory in the 1980s. Their approach, based on theoretical confidence in market outcomes that maximized consumer welfare and a focus on the price mechanisms to eliminate inefficiency, produced conservative policies that minimized government intervention.11 This theoretical approach has been subjected to scrutiny since the rise of ‘big tech’, the Global Financial Crisis B. Gordon, Economic Analysis before Adam Smith: Hesiod to Lessius (Macmillan, 1975) 267. A. Smith, An Inquiry into the Nature and Causes of the Wealth of Nations (first published 1776, W. Strahan and T. Cadell, London, 1990) Bk I, ch X. 7 R. Eklund and R. Herbert, ‘Retrospectives – The Origins of Neoclassical Microeconomics’ (2002) 16(3) Journal of Economic Perspectives 197. 8 J. Kerstenetzky, ‘Alfred Marshall on Big Business’ (2010) 34(3) Cambridge Journal of Economics 569. 9 S. Fellman and M.P. Shanahan, ‘Beyond the Market: Broader Perspectives on Cartel Research’ (2020) 68(3) Scandinavian Economic History Review 195. 10 G. Stigler, ‘A Theory of Oligopoly’ (1964) 72(1) Journal of Political Economy 44; and M. Levenstein and V. Suslow, ‘What Determines Cartel Success?’ (2006) 44(1) Journal of Economic Literature 43. 11 I.L.O. Schmidt and J. Rittaler, A Critical Evaluation of the Chicago School of Antitrust Analysis (Kluwer Academic Publishers, 1989). 5 6

48  Research handbook on cartels (2007–09) and more recently the COVID-19 crisis. Governments’ apparent inability to rein in the power of big companies in the new platform economy has brought forth a great deal of literature in recent years.12 Current theoretical views on the economic costs of cartels and the economic impetus for antitrust policy are not only comparatively recent, several remain contested.13 Areas of contestation include cartels’ impact on social welfare, their contribution to deadweight loss and productive inefficiency. It has also been argued that anticompetitive behaviour ultimately undermines trust in the market as an institution. Empirically what is clear, however, is that over the past century, cartels of all types have consistently manipulated markets to raise prices above competitive levels.14

III.

HISTORICAL FOUNDATIONS

Policies to curb cartels and monopolies and to prevent collusion and price fixing are often considered a phenomenon of the modern industrial capitalism and the result of the emergence of the large firm. Efforts to prevent private monopolies and entrepreneurs colluding to fix prices or restrict competition exist far back in history. For example, in Mesopotamia, cities banned price fixing, and in China during the Tang Dynasty in the seventh century condemned cartels.15 From the medieval and early modern periods, we have more information. In the Italian Commune, laws on ‘fair’ prices existed, especially on foodstuff. Scholars like Marco Corradi and Raymond de Roover emphasize that theological considerations tend to be overemphasized when looking at commercial laws in medieval times; local rulers needed to solve practical problems and scholastics were well able to reconcile theological, moral and commercial concerns with practical solutions to concrete problems.16 Most monopolies were owned by the state. Until the nineteenth century, anti-monopoly laws were primarily to safeguard rulers, not to protect consumers, small artisans or tradespeople. While rulers may have held negative views on monopolies, collusion and price-fixing, and occasionally enacted legislation to prevent such activities, this fact did not prevent those same rulers from granting privileges to key producers.17 12 See e.g. T. Philippon, The Great Reversal. How America Gave Up on the Free Markets (Belknap Press & Harvard University Press, 2019); T. Lenard, ‘Introduction to the RIO Special Issue on Antitrust and the Platform Economy’ (2019) 54 Review of Industrial Organization 617; and L. Phillips Sawyer and H. Hovenkamp, ‘New Perspectives in Regulatory History’ (2019) 93(4) Business History Review 659. 13 M. Motta, Competition Policy: Theory and Practice (Cambridge University Press, 2004); and D. Healey, M. Jacobs and R. Smith (eds), Research Handbook on Methods and Models of Competition Law (Edward Elgar Publishing, 2020). 14 J. Connor and Y. Bolotova, ‘Cartel Overcharges: Survey and Meta-Analysis’ (2006) 24 International Journal of Industrial Organization 1109. 15 B. Hawk, ‘Antitrust in History’ (2018) 63(3) Antitrust Bulletin 275. 16 M.C. Corradi, ‘Notes on Competition and Justum Pretium Theory and Practice in Medieval Italy: Lessons for Modern EU Competition Price Theory?’ (2018) 63(3) Antitrust Bulletin 330; and R. de Roover, ‘The Concept of the Just Price: Theory and Economic Policy’ (1958) 18(4) Journal of Economic History 418. 17 B. Kahn, ‘Antitrust and Innovation before the Sherman Act’ (2011) 77(3) Antitrust Law Journal 757.

An historical account of anti-cartel enforcement  49 After the Civil War (1642–51), English attitudes to concentrated power saw them adopt more negative views towards monopolies and a more positive outlook on free trade. Their common law tradition included three fundamental principles: freedom to trade, fair prices and prohibitions against monopolies.18 Nonetheless, prior to the twentieth century, English laws against market manipulation were few. A key tension that emerged, especially with the rise of larger enterprises, was the tension between the freedom to contract (and thus legally restrict competition) and principles of free trade (which were against manipulating markets). This tension still underpins many of the alternative views on the extent to which markets and agreements between firms should be regulated. The English common law tradition is often considered as influencing the first modern legislation emerging in the US, Canada and other Commonwealth nations; but the extent of such influence has been debated. More broadly, European nations’ attitudes and legal traditions were also transferred, to varying degree, to their colonies – a clear factor that would influence the path of subsequent antitrust legislation in those countries. Such influences and legal traditions are complex phenomena. Even within Europe, views on collaboration and restrictive practices varied and legal changes did not always have their intended effects. Overall, the existence of ‘modern’ anti-cartel or pro-competition legislation prior to modern twentieth-century capitalism is often over-emphasized and the existence of such laws should be understood in terms of responses to local conditions in a particular society and not as ‘linear stages in an historical march toward modernity’.19 The three major pillars of modern antitrust law – regulation or prohibition against restrictive agreements, unilateral firm conduct (monopolization) and merger control – cannot be found before the late nineteenth century.

IV.

EARLY MODERN ANTITRUST/ANTI-CARTEL/ COMPETITION LEGISLATION

The first modern antitrust and anti-cartel legislation was a response to the development of industrial and corporate capitalism after the industrial revolution. Rapid technological change, advances in communication technology, the opportunity to operate across borders and few legal restrictions enabled a rapid concentration in big business; the vertically and horizontally integrated firm emerged.20 As big industrial firms had high fixed costs, the temptation to fix prices with competitors was large. While developments in the US are usually given primacy, similar changes were occurring in other rapidly industrializing countries. For example, in Germany, cartelization in several economic sectors occurred swiftly in the late nineteenth century. The rapid industrialization process was followed by a severe recession in the 1870s which led to a downward spiral in prices, with devastating effects for a large number of entrepreneurs. Many firms rapidly organized themselves to prevent ‘cut-throat’ competition.

B. Hawk, ‘English Competition Law before 1900’ (2018) 63(3) Antitrust Bulletin 350. Hawk (n 15) 276. 20 A. Chandler, Scale and Scope: The Dynamics of Industrial Capitalism (Harvard University Press, 1990); N. Lamoreaux, The Great Merger Movement in American Business, 1895–1904 (Cambridge University Press, 2010); and P. O’Brien, ‘The Great Merger Wave’, in R. Parker and R. Whaples (eds), Routledge Handbook of Major Events in Economic History (Routledge, 2013). 18 19

50  Research handbook on cartels This established a view in Germany that cartelization was a necessity and led to Germany later being identified as the country of cartels.21 The famous labelling of cartels as ‘children of distress’ (Kinder der Not),22 because they emerged in response to recessions and the fear of ‘ruinous competition’, meant that German authorities, until the interwar period, often viewed cartels as critical to economic stability. Research shows, however, that cartels are not formed only during recessions and to avoid price falls, but for many reasons and also during booms.23 While traditional price or market-sharing cartels were common in many modernizing countries in the late nineteenth century, the trust form is especially associated with the US. Private agreements between firms to manipulate the market could not be legally enforced, so the risk of cheating by cartel members was ever-present.24 In the trust model, subsidiary companies gave control of their firm to a larger trust, in exchange for a share in the profits of the overall business. The famous large US trusts, such as those in steel, oil and railways, emerged within a short time. These combines rapidly gained a dominant position in their markets; and with economic wealth came political power. These changes to industry structure, conduct and performance led to the creation of the most famous modern legislation to combat restrictive practices and monopolization, the Sherman Act, which was passed in 1890. The first country to pass a modern antitrust legislation, however, was not the US but Canada, which in 1889 passed legislation to prohibit ‘combinations’ that engaged in conspiracies to restrain trade. Canada too had experienced the rise of powerful combinations and the problems of price-fixing rings.25 Canada’s legislation was passed more quickly than in the US, partly because of the US corporate lobby’s success in delaying the passage of the Sherman Act. In 1889, 14 months earlier than in the US, the Canadian Combinations in Restraint of Trade Act (often called the Wallace Act) was passed.26 This legislation stated that every person who conspires, combines, agrees or arranges to restrain competition has committed an offence.27 The Canadian law deemed the restraint of trade a criminal offence. The Sherman Act banned ‘every contract, combination or conspiracy in restraint of trade in interstate and foreign trade’. As with the Canadian law, it treated these violations as criminal 21 I.E. Schwartz, ‘Antitrust Legislation and Policy in Germany – A Comparative Study’ (1956–57) 105(5) University of Pennsylvania Law Review 617; and D. Gerber, Law and Competition in Twentieth Century Europe: Protecting Prometheus (Oxford University Press, 2001) 75–76. 22 F. Kleinwächter, Die Kartelle. Eine Beitrag zur Frage der Organisation der Volkswirtschaft (Wagner, Innsbruck, 1883) 245. 23 See e.g. M. Levenstein and V. Suslow, ‘Cartels and Collusion: Empirical Evidence’, in R. Blair and D.D. Sokol (eds), The Oxford Handbook of International Antitrust Economics, Volume 2 (Oxford University Press, 2014). This was a question discussed early on in the literature; see e.g. R. Liefmann, Cartels, Concerns and Trusts (Methuen & Co., Ltd, 1932) 46­–54. 24 O’Brien (n 20). Although many countries tolerated cartel agreements at this time, few had legislation enabling cartel participants to enforce their agreements. 25 T.W. Ross, ‘Introduction: The Evolution of Competition Law in Canada’ (1998) 13(1/2) Review of Industrial Organization 1; M. Bliss, ‘Another Anti-Trust Tradition: Canadian Anti-Combines Policy, 1889–1910’ (1973) 47(2) Business History Review 177; and E. Iacobucci and M. Trebilcock, ‘Canada. The Competition Law System and the Country’s Norms’, in E. Fox and M. Trebilcock (eds), The Design of Competition Law Institutions (Oxford University Press, 2013). 26 Formally, Act for the Prevention and Suppression of Combinations formed in restraint of Trade, 52 Vict. c. 41 (1889). 27 C. Halladay, ‘The Origins of Canada’s Cartel Laws’ (2012) 25 Canadian Competition Law Review 157.

An historical account of anti-cartel enforcement  51 offences. Antitrust legislation was not entirely new in the US, as there was state-level antitrust legislation in existence prior to the Sherman Act.28 The US Sherman Act, however, set the stage for the next century of antitrust legislation and was to serve as a model for many other countries.29 For example, in 1906, the newly federated nation of Australia passed antitrust legislation based on the Sherman Act. The resulting ‘Australian Industries Preservation Act’ (and subsequent amendments) was, however, more focused on external monopolies and trusts than local businesses.30 It would not be until the passage of the Trade Practices Act in 1974 that Australian antitrust legislation began to approximate that in other modern economies.31 With the rise of large industry in parts of Europe, there had also been efforts to introduce pro-competition legislation in countries there. As early as 1870, an Austrian statute held that agreements among businessmen ‘designed to raise the price of a commodity to the detriment of the public’ were to be declared null and void.32 There were debates in the Austrian Parliament in the late nineteenth century about a system of registration to prevent price fixing.33 The proposal was influenced by a group of Austrian economists, who warned about the problems of cartels, and the legal scholar Adolf Menzel helped prepare the 1897 proposal. The scheme was never realized, but it later influenced other legislative initiatives in Europe. Ultimately Austria only introduced anti-cartel legislation in 1951. There were similar debates in Germany, especially after the swift growth of cartels in the 1870s. A general statutory provision relating to cartel regulation existed in the Trade Regulation Act of 1869, which stated: ‘All trade is open to everyone, unless this statute provides exceptions from or limitations upon this rule.’34 The question arose whether the provision applied to restraints of trade caused by cartels. The German courts held that a cartel may be harmful if the underlying agreements and arrangements intended to create, or resulted in, a monopoly and the exploitation of consumers. However, a fear of the alternative extreme – ‘ruinous competition’ – meant the courts generally ruled in favour of the cartel. In the precedent-setting Wood Pulp case35 the court held that, while economic freedom had been limited by the companies’ agreements, as these were not aimed at creating an exploitative monopoly, they were allowable. The court also found that agreements that were only temporary were acceptable. According to Schwartz, the balance between freedom to contract and freedom to trade was primarily interpreted to mean economic freedom from government interference, not freedom from contractually binding competitors.36 G. Stigler, ‘The Origin of the Sherman Act’ (1985) 14(1) Journal of Legal Studies 1, Table 1. W. Kovacic and C. Shapiro, ‘Antitrust Policy: A Century of Economic and Legal Thinking’ (2000) 14(1) Journal of Economic Perspectives 43. 30 K.A. Round, M.P. Shanahan and D.K. Round, ‘Anti-Cartel or Anti-Foreign: Australian Attitudes to Anti-Competitive Behaviour before World War I’ (2010) 56(4) Australian Journal of Politics and History 540. 31 H. Qaqaya and G. Lipimile (eds), The Effects of Anti-Competitive Business Practices on Developing Countries and their Development Prospects (United Nations Conference on Trade and Development, 2008). 32 P. Abel, ‘Legislation on Cartels in Austria’ (1952) 42(2) The Trade-Mark Reporter 101. 33 G. Hoffman, ‘The Austrian Cartel Law: Principles and Background’ (1969) 14(3) Antitrust Bulletin 249. 34 Schwartz (n 21) 625. 35 B. v den Sichsischen Holzstoff-Fabrikanten-Verband, Reichsgericht (VI. Zivilsenat), 4 Febuary 1897, 38 R.G.Z. 155. 36 Schwartz (n 21). 28 29

52  Research handbook on cartels The passage of the Sherman Act did not lead to an immediate transition to a strongly enforced antitrust regime. As the legislation was formulated broadly, the first years were marked by legal challenges and deliberations on how to interpret the law and the intent behind it.37 An important issue was to distinguish collaborations that suppressed competition from collaborations that promoted growth, that is, rationalization, and cooperation that improved efficiency.38 Initially, the Act gave the Supreme Court and federal judges significant powers to draw lines ‘between acceptable cooperation and illegal collusion, between vigorous competition and unlawful monopolization’.39 Critics feared that the Supreme Court was softening the law, by focusing only on ‘unreasonable’ restraints and narrowing the statute’s application. This led, in 1914, to the Clayton Act and to the establishment of the Federal Trade Commission (‘FTC’).40 The Clayton Act reduced courts’ discretion by specifically prohibiting certain tying arrangements, exclusive dealing agreements, interlocking directorates and mergers achieved by buying stocks. As an administrative body specifically designed to implement antitrust policies, the FTC was a new tool intended to improve the enforcement of antitrust legislation. It had the authority to identify and condemn ‘[u]nfair methods of competition’.41 Initially, the Canadian law was also loosely enforced.42 The legislation was amended several times but in reality it had little effect until the 1920s. Both the Sherman and Wallace Acts demonstrate a common pattern found in most antitrust and competition legislation around the world: that even when finally passed, the original legislation is soon challenged and amended as opponents fight to lessen the constraints of regulation. Also, the novelty of the legislation effected weak implementations. Early cases challenging the Sherman Act sharpened the courts’ and experts’ interpretations of what types of behaviour were to be curbed and what had been the lawmakers’ intent. These included cases against Trans-Missouri Freight Association43 and Standard Oil.44 The Standard Oil case led to the trust being broken into 34 different companies, hindering both vertical and horizontal control of the market. The case set important precedents. Critically, the case strengthened the ‘rule of reason’ test, which established that the court should consider the reasons behind otherwise market controlling agreements. The result was that45 such agreements were not necessarily illegal per se. The question to answer was whether an agreement unreasonably restrained trade, or whether another reason (such as enhanced efficiency) may justify the apparently noncompetitive arrangement.46 37 M. Sklar, ‘Sherman Antitrust Act Jurisprudence and Federal Policy-Making in the Formative Period, 1890–1914’ (1990) 35(4) New York Law School Law Review 791; and Kovacic and Shapiro (n 29). 38 Kovacic and Shapiro (n 29). 39 ibid 43. 40 ibid. 41 H. Hovenkamp, ‘The Federal Trade Commission and the Sherman Act’ (2010) 62(4) Florida Law Review 871. 42 Halladay (n 27). 43 United States v Trans-Missouri Freight Association, 166 US 290 (1897). 44 Standard Oil Co. of New Jersey v United States, 221 US 1 (1911). 45 R. Bork, ‘Legislative Intent and the Policy of the Sherman Act’ (1966) 9(3) Journal of Law & Economics 7. 46 Kovacic and Shapiro (n 29).

An historical account of anti-cartel enforcement  53 Scholars have subsequently discussed whether the ultimate goal of the original law was market efficiency, protection of competitors or advancing consumer welfare. Bork argued consumer welfare was a core goal, although later scholars have convincingly contested this. Bork’s argument underpinned the Chicago School’s focus on consumer welfare in antitrust policy decades later.47 Other commentators have noted that the Sherman Act aimed to satisfy multiple goals.48 There is considerable literature on whether the Sherman Act was a return to the English common law tradition that contracts restraining trade in a manner harmful to the public were unlawful. According to Sklar, the Sherman Act went further than the common law principle as it made restraints of trade that contravened the public interest illegal, and the collaborators liable to punishment. Such agreements were also liable to private, civil suits for damages by third parties.49 Spurred on by the rapid expansion of big business in the late nineteenth century, the Canadian and American antitrust laws were passed before those in other countries. Differences in each nation’s economy, stage of development, politics and culture meant that even when citing the Sherman Act as inspiration, lawmakers fashioned policies and legislation to deal with the specific problems and structures in their own country. In several countries, this occurred decades into the twentieth century. While the North American ‘antitrust’ policy has been frequently cited as influential in shaping pro-competitive attitudes, especially after the Second World War, each country followed its own path towards developing antitrust policy.50

V.

ANTI-CARTEL ENFORCEMENT IN THE INTERWAR PERIOD

The period which led to the Sherman Act was not only one of rapid industrialization but also an era of global economic integration, growing international investment and trade flows. The First World War ended this first period of globalization, while the interwar years brought with them heightened nationalism, economic isolation and a retreat from global trade.51 The interwar period is often considered the peak period for cartels, especially in Europe, where relatively tolerant attitudes towards restrictive practices were common. Increasing cartelization during this period was partly a result of policies during the First World War, when many governments encouraged or even forced industries to collaborate and cartelize as part of the war economy. For some governments, strong cartels and big business were also considered important national symbols during the war.52 For example, in

C. Grandy, ‘Original Intent and the Sherman Antitrust Act: A Re-Examination of the Consumer-Welfare Hypothesis’ (1993) 53(2) Journal of Economic History 359. 48 T. Hazlett, ‘The Legislative History of The Sherman Act Re-Examined’ (1992) 30(2) Economic Inquiry 276. 49 Sklar (n 37); and Kahn (n 17). 50 H. Schröter, ‘Cartelization and Decartelization in Europe 1870–1995: Rise and Decline of an Economic Institution’ (1996) 25(1) Journal of European Economic History 129; and Fellman and Shanahan (n 1). 51 M. Bordo, A. Taylor and J. Williamson (eds), Globalization in Historical Perspective (University of Chicago Press, 2001). 52 Gerber (n 21) 116. 47

54  Research handbook on cartels Germany, the government relied on cartels for economic mobilization.53 In Italy, which had fewer cartels in the early twentieth century than, for example, Germany, government support to key industries promoted concentration during the First World War.54 Similar patterns were observed in other countries. Despite the Sherman and Clayton Acts, business concentration also increased in the US, as large corporations in many industries benefited economically from war expenditure.55 After the war, benign attitudes towards restrictive practices, cartels and overall government­– business collaboration in key industries continued in many European countries. In several nations, cartels were not only tolerated but actively promoted by national governments. These favoured participation in both international and domestic export cartels, as they were considered to be in the national interest. This was especially the case in small open economies where participation was viewed as the only way to compete on the world market. Complementing this situation, businesses’ self-regulation and freedom of contract were considered to override harm from collusion. Even the UK – commonly regarded as supporting free trade and a free market – did not introduce modern competition regulation until after the Second World War. Significant collusion and concentration occurred in the interwar UK, while their courts were often reluctant to rule against restraint of trade, leaving the cartels a free hand to collude.56 Nonetheless, many economic experts, some political parties and various interest groups understood the cost of cartels and identified the harmful effects of restrictive practices of powerful cartels and trusts. Immediately after the end of the First World War, efforts were made to return to free trade, and restrictive practices and international cartels could hinder these efforts. In 1927 the League of Nations also held a series of debates on cartels and their effects, having commissioned several experts’ reports prior to the meeting.57 The meeting, however, did little other than recommend countries pay attention to the cartel phenomenon. Strong business interest groups had been influential in many of the debates and vigorously argued the case against restricting cartels. A dominant contemporary narrative was that cartels (national and international) were key players in maintaining economic governance and economic stability.58 Other organizations that were critical of the growing powers of the cartels were, for example, the cooperative movement in their international meetings and the Inter-Parliamentary Union in their meeting in 1930.59

53 J. Wolff, ‘Business Monopolies: Three European Systems in their Bearing on American Law’ (1934–35) 9(3) Tulane Law Review 325. 54 C. Giordano and F. Giugliano, ‘A Tale of Two Fascisms: Labour Productivity Growth and Competition Policy in Italy, 1911–1951’ (2015) 55(1) Explorations in Economic History 25. 55 M. Stoller, Goliath: The 100-Year War between Monopoly Power and Democracy (Simon & Schuster, 2019) 29–31. 56 J.D. Gribbin, ‘Recent Antitrust Developments in the United Kingdom’ (1975) 20(2) Antitrust Bulletin 377. 57 See e.g. P. De Rousiers, Les Cartels et les Trusts et leur Évolution (League of Nations, 1927); and J. Hirsch, Monopoles Nationaux et Internationaux du Point de Vue des Travailleurs, des Consommateurs et de la Rationalization (League of Nations, 1927). 58 M. Bertilorenzi, ‘Legitimising Cartels: The Joint Roles of the League of Nations and of the International Chamber of Commerce’, in Fellman and Shanahan (n 1). 59 L. McGowan, The Antitrust Revolution in Europe: Exploring the European Commission’s Cartel Policy (Edward Elgar Publishing, 2010) 64; and M. Shanahan and S. Fellman, ‘Introduction’, in Fellman and Shanahan (n 1).

An historical account of anti-cartel enforcement  55 Cartels were also the subject of considerable scholarly attention in the interwar period, which saw the publication of several works on international cartels and combines, as well as works on nations’ domestic economies, anti-cartel legislation and policies.60 Critical analyses faded with the Great Depression, however, and a benign attitude towards cartels became more prominent as fears of ‘ruinous competition’ and disastrous price falls increased. As democratic governments sought economic stability through cooperative organizations, nationalist, fascist and autarchic governments used cartelization as a tool to implement their economic policies and control business. Despite, and in some cases because of, the rise of cartels, the first stirrings of ‘modern’ anti-cartel or antitrust legislation in Europe are also evident in the interwar period. The development in the US and Canada did not go unnoticed, of course. There had also been serious attempts to introduce anti-cartel legislation in Austria in the late nineteenth century and echoes of this influenced subsequent European efforts. For example, the Norwegian Temporary Price Act of 1920 introduced the compulsory notification of restrictive business agreements, dominant enterprises and subsidiaries of foreign cartels and monopolies. The Norwegians extended this in their 1926 Trust Law, which was quite intolerant of anticompetitive behaviour and introduced a register to record cartels. This legislation is considered by several scholars as the first real competition legislation in Europe.61 The German authorities had also passed the German Decree against the Abuse of Economic Power Position (the ‘Cartel Ordinance’) in 1923, therefore formally the first European legislation, but the Norwegian legislation was more efficient and effective. The Ordinance was partly a response to Germany’s hyperinflation and a means to combat price rises. Small and medium-sized industries were important supporters of the legislation, as were consumer and labour interests. The influence of the Austrian debates is evident in the German approach, but the new Ordinance had more substance and political prominence.62 The aim was to use administrative procedures, to constrain cartels from abusing their market power to manipulate prices and output. The German Ordinance did not seek to forbid or eliminate cartels, only to curb their excess. The approach later became known as the abuse principle; cartels were to be punished only in cases they abused their market power – such as preventing cartels from boycotting former members who had withdrawn from an agreement.63 Overall, however, the legislation was not particularly strong.64 The fear of ‘ruinous competition’ was a ghost that haunted the German courts’ deliberations, causing them to favour cartels in their decisions.65 Some contemporary analysts even argued that the Ordinance was not really an

60 Core works included: W. Notz, Representative International Cartels, Combines and Trusts (Department of Commerce, 1929); Liefmann (n 23); and A. Plummer, International Combines in Modern Industry (Pitman & Sons, 1932). 61 Gerber (n 21) 6 and 15; H. Espeli, ‘Perspectives on the Distinctiveness of Norwegian Price and Competition Policy in the XXth Century’ (2002) 31(3) Journal of European Economic History 621; and H. Espeli, ‘Transparency of Cartels and Cartel Registers: A Regulatory Innovation from Norway?’ Fellman and Shanahan (n 1). 62 Gerber (n 21) 67. 63 C. Harding and J. Joshua, Regulating Cartels in Europe: A Study of Legal Control of Corporate Delinquency (Oxford University Press, 2003) 75. 64 Gerber (n 21) 134–35. 65 Schwartz (n 21).

56  Research handbook on cartels anti-cartel law.66 With the benefit of hindsight, later scholars consider the Ordinance a step forward for anti-cartel enactment legislation in Europe, especially as its approach influenced other countries.67 The ‘abuse principle’ became the dominant concept underpinning European counties’ cartel legislation up until the 1980s.68 The Ordinance did not have a long life. In 1929, the economic depression changed everything, and cartels were again fully accepted. The legislation was also modified in the early 1930s and then abolished after the Nazis’ rise to power. Cartels became part of the regime as private firms were forced into cooperative arrangements. A similar approach was taken in Italy after the rise of Benito Mussolini. Mussolini initially adopted a laissez-faire economic policy, but with the Great Depression, the country’s economic policies increasingly favoured both mergers and cartelization. In 1932, regulations creating both compulsory cartels and voluntary cartels was made law.69 In 1930 cartel legislation emerged in several countries, including in Eastern and Central Europe (that is, Hungary in 1932, Czechoslovakia and Poland in 1933, Yugoslavia in 1934, Bulgaria in 1936 and finally Romania in 1937).70 Comparatively little is known about these Acts as they were phased out during and after the Second World War, although some contemporary comments do exist. In all these countries, the legislation was heavily influenced by the Austrian and German models and focused on the avoidance of abuse. They also followed the Norwegian model by adopting a register. The practical implementation of the laws was often weak, and the meanings of ‘abuse of market power’, ‘public interest’ or ‘cartels’ were often ill-defined. Hungary in 1920 had prohibited the abuse of market power (Laws No. XV/1920 and No. V/1923), but with little effect as there was no state authority to determine what was the ‘right’ or ‘just’ price, or what was an ‘immoral business practice’.71 Anti-cartel legislation was introduced in 1931 (Law No. XX/1931). The Cartel Act did not prevent cartelization or protect consumers, however, as the legislation lacked clear definitions of cartel activities and the register’s contents were kept secret. Cartel agreements were valid contracts protected by law if they were registered. Actions against them could only be taken where they were endangering the public welfare or the economy.72 In Bulgaria, the Control of Cartels and Monopolistic Prices Act (1931) aimed to improve the legal and administrative framework promoting competition. A Cartel Commission was formed to register cartels and prices, but again enforcement was weak.73 Czechoslovakian legislation was passed in 1933 (Act on Cartels and Monopolies, No. 141 on 12 July 1933). Cartels were to register all valid agreements. The legislation did not prevent

Wolff (n 53). Harding and Joshua (n 63) 80–81. 68 Schröter (n 50). 69 Giordano and Giugliano (n 54). 70 See e.g. S. Timberg, ‘Restrictive Business Practices: Comparative Legislation and the Problems That Lie Ahead’ (1953) 2(4) American Journal of Comparative Law 445. 71 N. Varga, ‘The Research of Hungarian Cartel Supervision’ (2020) 11(2) Journal on European History of Law 211. 72 Varga (n 71); and M. Hidvégi, ‘Machine Building Cartels in Hungary, 1919–1949’ (2019) 20(1) Enterprise & Society 89. 73 Y. Todorova, ‘Historical Review of the Development and Competition Law System in Bulgaria until 1944’ (2018) 9(1) De Jure 30. 66 67

An historical account of anti-cartel enforcement  57 abuse or cartelization, and instead protected the participants, as the register was secret to the public.74 Teichova found that the majority of records were of domestic cartels, but a large number were also international – primarily German firms, as these had invested extensively in Czechoslovakia.75 In Poland, a Cartel Act was issued on 28 March 1933. Cartel agreements were valid and binding provided they were made in accordance with the cartel law. The cases were inspected by a Cartel Court, and the Supreme Court had the jurisdiction to punish violations.76 This legislation lay dormant after the Second World War until the late 1980s, when it was gradually revived as Poland transitioned to a market economy.77 The Yugoslavian Cartel Decree from 1934 was less tolerant than other Central European Cartel Acts.78 The first paragraph of the decree stated that cartels were forbidden, while the second paragraph allowed the Minister of Economic Affairs to grant special authorizations for their establishment. As actual implementation was also initially weak, the result was an ineffectual regulation.79 Anti-cartel legislation was also on the agenda of other western European countries. Danish and Swedish decision-makers followed the Norwegian example, and in Denmark a Price Act and registration system were introduced in 1936. The Danish cartel authorities only became active in monitoring restrictive practices after the Second World War, when new legislation was passed in 1955. Sweden had passed an anti-cartel law in 1925, which granted authorities permission to investigate cartels and their ‘influence on prices and sales’. It did not, however, produce any tangible results. The first legislation that produced real control of some restrictive practices was from 1946, when a cartel register was also established.80 In 1935 in Belgium and the Netherlands, Cartel Acts were introduced which enabled forced cartelization under specific circumstances. After Nazi occupation, the Dutch Act was replaced by a Cartel Decree, vesting the state with almost unlimited powers to regulate and control the economy.81 France had a different approach to monopolies, but even in that jurisdiction cartel legislation was under consideration in the early 1930s. A draft bill was prepared (Projet de loi relatif au regime des ententes commerciales et industrielles), but nothing eventuated. France introduced an Ordonnance in 1945, whereby the state could control prices. It was renewed on several occasions, but was only weakly enforced and had little effect on cartel agreements.82

A. Teichova, An Economic Background to Munich: International Business and Czechoslovakia 1918–1938 (Cambridge University Press, 2008), 57. 75 ibid 59. 76 W. Sokalski, ‘Supreme Court in Poland’ (1941) 4 Polamerican Law Journal 111. 77 M. Wise, ‘Review of Polish Competition Law and Policy’ (2003) 5 OECD Competition Law and Policy Journal 83. 78 Decree No. II 27945 from 3 August 1934. See F. de Kirnaly’s note in ‘Cartels’ (1934) 38 International Law Association Reports of Conferences 190. 79 K. Heinrich and G. Leighton, ‘Cartel Control: A Record of Failure’ (1946) 55(2) Yale Law Journal 297. 80 P. Sandberg, ‘Cartel Registration in Sweden in the Post-War Period’, in Fellman and Shanahan (n 1). 81 L. Petit, ‘The Dutch Cartel Collection in the Twentieth Century: Facts and Figures’, in Fellman and Shanahan (n 1); and Timberg (n 70). 82 C. Freedeman, ‘Cartels and the Law in France before 1914’ (1988) 15(3) French Historical Studies 478; and D. Gerber and R. Azarnia, ‘Dirigisme and the Challenge of Competition Law in France’ (1995) 3(1) Cardozo Journal of International and Comparative Law 9. 74

58  Research handbook on cartels The North American laws were mostly directed specifically at trusts, combines and monopolies. In contrast, the ‘cartel problem’ in Europe was aligned with efforts to combat the restraint of trade.83 While the US and the Canadian legislation ostensibly banned cartels, in interwar Europe, the cartel ‘problem’ was an issue of debate. This difference reflects economic differences between US and European markets, the preponderance of family-related firms in Europe, and attitudes to cooperation between businesses. The ‘cartel question’ was at first seen as a particularly ‘German problem’ by, for example, foreign experts, but it was to become an issue in many other countries. In the interwar period, the power of international cartels and trusts was of special concern to smaller countries. As global economic problems continued, courts began to reject the argument that price cooperation was benign because cartel members set ‘reasonable’ prices or desired only to stop downward price spirals.84 Horizontal and vertical price restrictions were increasingly considered unreasonable and against the public interest. A few non-Western countries introduced cartel legislation prior to the Second World War. These were mainly countries with strongly nationalistic governments and the statutes were often intended to promote, rather than prevent, cartelization and so enable state control of big business and industrialization. In 1931 Japan imposed an Important Industries Control Law, which mandated cartelization in many industries.85 In cases where the cartel included more than two thirds of the industry, the state authorities could force outsiders to comply with the cartel’s rules. After the invasion of Manchuria in 1937, the legislation was amended to allow the government to intervene directly into a cartel’s activities and set minimum prices and quotas. During the war, the mandatory cartels were replaced with mandatory trade or ‘control’ associations that could intervene directly in economic activities. These associations were no longer self-determining organizations, but served the state.86 A number of Latin American countries also introduced legislation that affected restrictive agreements during this period. In Brazil, an antitrust Act was passed in the 1930s, but its rationale was not to promote competition. It was designed as part of a nationalistic economic policy implemented by the autarkic leader Getfilio Vargas, who sought to industrialize and centralize the economy.87 In Argentina, a competition law was enacted in 1923 and renewed in 1946. The first two articles closely resembled sections 1 and 2 of the Sherman Act. The legislation was intolerant of cartels, prohibiting the creation, maintenance or exploitation of monopoly power and containing penal sanctions. The Argentinian Acts, although different

C.D. Edwards, Control of Cartels and Monopolies: An International Comparison (Oceana Publications, New York, 1967); and H. Thorelli, ‘Antitrust in Europe: National Policies after 1945’ (1959) 26(2) University of Chicago Law Review 222. 84 M. Perelman, ‘Fixed Capital, Railroad Economics, and the Critique of the Market’ (1994) 8(3) Journal of Economic Perspectives 189; and Kovacic and Shapiro (n 29). 85 H. Levy, ‘Industrial Law and International Cartelization’ (1949) 3(3) Industrial Law Review 184; T. Ohata and T. Kurosawa, ‘Policy Transfer and Its Limits: Authorised Cartels in Twentieth-Century Japan’, in Fellman and Shanahan (n 1); and B. Gao, ‘The State and the Associational Order of the Economy: The Institutionalization of Cartels and Trade Associations in 1931–45’ (2001) 16(3) Sociological Forum 409. 86 Gao (n 85); and Ohata and Kurosawa (n 85). 87 F. Todorov and M. Filho, ‘History of Competition Policy in Brazil: 1930–2010’ (2012) 57(2) Antitrust Bulletin 207. 83

An historical account of anti-cartel enforcement  59 in attitude to those in Brazil, were rarely enforced. Between 1933 and 1980, only four cases resulted in sanctions under these laws.88 Although the US has the reputation as the ‘birthplace’ of modern antitrust law, the first years of the interwar period in that country saw still quite weak enforcement of antitrust legislation, and the courts treated firms benignly.89 Powerful tycoons exercised strong influence in government while the FTC and the Supreme Court were relatively ineffectual. Business interests prevented antitrust cases proceeding far. Business and government were often closely aligned on economic issues.90 Despite US antitrust oratory, the Webb–Pomerene Act allowed US firms to participate in export cartels on international markets. The National Industrial Recovery Act of 1933 initiated a more vigorous antitrust policy. President Franklin D. Roosevelt appointed more energetic trustbusters into the Department of Justice, the most famous being Thurman Arnold, who was appointed as head of the Antitrust Division in 1938. Economists increasingly advocated a more forceful antitrust policy.91 At the same time there was an increasing focus on price fixing and market allocation cartels. In Canada the rules against firms’ involvement with restrictive practices had also been enforced sporadically and with lenience. Concerned by the limited ability to combat monopolies and cartels, the Combines Investigation Act of 1910 was reinvigorated in 1927. It was not until the 1960s, however, that new laws with real effect on anticompetitive behaviour were implemented.92

VI.

POST-SECOND WORLD WAR: THE GLOBAL SPREAD OF ANTI-CARTEL LEGISLATION

In the decades after the Second World War, intolerance of restrictive practices gradually expanded; legislation was adopted and more dynamic measures to combat restrictive practices were taken in many countries. The development is often attributed to the spread of the American antitrust ideas. This was especially the case in occupied Japan and Germany, and in a more general sense in Western Europe; but with the benefit of a longer perspective, these developments were more complex than a simple transfer of attitudes. American administrations in Germany, Italy and Japan immediately after the war promoted legislation designed to break up cartels, monopolies and strong business groups. The US considered that powerful business groups and cartels had been important backbones of German and Japanese military power. In Japan, breaking up cartels and strong business groups, such as the family-centred zaibatsus and bank-centred keiretsus, was considered important in the transition to peace and to minimize future threats. These policies never really became imbedded and after the withdrawal of the foreign administration the legislation was amended. The Organization for Economic Co-operation and Development, Competition Law and Policy in Latin America. Peer Reviews of Argentina, Brazil, Chile, Mexico and Peru (Paris, 2006). Laws to counteract monopolies and restrictive practices existed also in some other countries not dealt with here. For example, the Philippines instigated anti-cartel legislation in 1925, Mexico in 1926 and Costa Rica in 1932. Their implementation and effects remain unclear. See e.g. Timberg (n 70). 89 Kovacic and Shapiro (n 29). 90 Stoller (n 55) 33 and 43; and Phillips Sawyer and Hovenkemp (n 12). 91 Kovacic and Shapiro (n 29); Stoller (n 55) 134–39. 92 Iacobucci and Trebilcock (n 25). 88

60  Research handbook on cartels American model was also frequently resisted at the local level elsewhere, and businesses in, for example, Italy were especially successful in avoiding a top-down implemented competition policy. Italy remained one of the countries without antitrust and pro-competition laws until the 1990s. Despite several attempts in the 1950s and 1960s, the only successful antitrust legislation in Italy involved regulation against unfair competition in the Civil Code.93 In Germany, the occupation also involved French, British and Soviet forces. The four powers had somewhat different priorities, and it was in no one’s interest to destroy completely German industry. Segreto and Wubs emphasize that the US also was ambivalent towards German cartels and vacillated between pragmatic attitudes and aggressive statements.94 The Allies gave responsibility for anti-cartel legislation to the West German decision-makers in 1949. Differences between those seeking to prohibit cartels (the Josten Commission drafts), those promoting a framework to minimize cartelization (the ordoliberals) and those seeking fully free markets (business interests and heavy industry) led to drawn-out debates between politicians and experts, oscillating between anti-abuse-based legislation and prohibition. Not until 1957 was legislation based on the prohibition principle implemented, but with numerous potential exemptions.95 The US also pressured other Western European countries to change their policies towards cartels and restrictive practices. Such efforts were included in the Marshall Plan.96 Reform of anti-cartel regulation together with technical assistance, new managerial models and capital expenditure was promoted by the Americans as the way to lift Europe after the Second World War. In the early 1950s, the Schuman Plan and the foundation of the European Coal and Steel Community (‘ECSC’) paved the way for new policies in Western Europe. As the situation changed in Europe, the need for stability and growth saw American efforts to direct change in one direction fade. Attitudes had been influenced, however, and a wave of anti-cartel and competition legislation surged in Western Europe after the 1950s. However, the results were quite different from the US antitrust legislation. In most Western European countries, cartel or agreement registration became a common way to regulate restrictive agreements, especially where legislation was based on anti-abuse principles. The system, which had started in Norway, was used in different legal contexts throughout the 1950s, 1960s and 1970s.97 These registers were especially common in Europe, but existed also in some non-European countries, such as Australia, New Zealand (for a very brief period), Japan, Israel and India. In a system with anti-abuse-based legislation, the authorities documented information about restrictive practices so as to monitor and control these agreements and intervene in cases of abuse of market dominance. A registered agreement was, as a rule, permitted. This model existed in countries which prohibited restrictive practices and cartels but allowed exemptions, such as the UK, Germany, Israel, Japan and later India, Pakistan and South Korea. In these cases, cartel agreements exempted from the ban were L. Segreto and B. Wubs, ‘Resistance of the Defeated: German and Italian Big Business and the American Antitrust Policy, 1945–1957’ (2014) 15(2) Enterprise & Society 307; and P. Bianchi and G. Gualtieri, ‘Mergers and Acquisitions in Italy and the Debate on Competition Policy’ (1989) 34(3) Antitrust Bulletin 601. 94 Segreto and Wubs (n 93). 95 J.O. Hesse and E.M Roelevink, ‘Cartel Law and the Cartel Register in German Twentieth-Century History’, in Fellman and Shanahan (n 1). 96 Thorelli (n 83). 97 Fellman and Shanahan (n 1); and Edwards (n 83). 93

An historical account of anti-cartel enforcement  61 included in the register. Even the cartels permitted under the Webb–Pomerene exemptions in the US were registered.98 The European Economic Community (‘EEC’) introduced a system of registration for organizations that were exempted from the general ban on cartels affecting intra-EEC trade. Although it was initiated in 1962, the number of applications for exemption was so numerous that the administration could not keep up. The EEC system was abandoned in the late 1990s as the implementation of the intolerant European competition policies of the common market ensured its obsolescence.99 The practical enactment of anti-cartel legislation was similarly thwarted in India, where the backlog of cases became overwhelming. This was especially a problem in countries with intolerance-based legislation, where it was critical to receive a court’s exemption from a total ban. In more tolerant jurisdictions, administrators held a pragmatic attitude and let many cases go unregistered. In most cases, at least some of the information that was recorded was public. A public register was thought to allow consumers and competitors to be aware of any restrictive agreements, to restrain entrepreneurs and business from exploiting their market power and even to ‘shame’ the participants.100 In an era when many jurisdictions had little experience with anti-cartel enforcement, the registration system was often considered a pragmatic way to advance legislative reform. Even those who were reluctant to restrict cartels did not usually have serious objections against notification of their existence.101 The registration system came to an end in Europe in the 1980s and 1990s when stronger anti-cartel legislation and common European competition polices were adopted within the EU. Countries including Sweden, UK, Finland, the Netherlands and Austria had all introduced anti-abuse legislation including registration procedures in the late 1940s and 1950s. The UK, with a long tradition of promoting laissez-faire policies, only adopted the Monopolies and Restrictive Practices (Inquiry and Control) Act in 1948 as part of post-war rebuilding. Initially administrative and ‘gentle’ in nature until its strengthening in 1953, the legislation, together with a major public report in 1955, was credited with creating a sea-change in business attitudes.102 From 1956, companies had to record proscribed forms of cooperation on a Registrar of Restrictive Trading Agreements. To avoid registration, many agreements were simply dissolved. The legislation also allowed six ‘gateways’ whereby agreements against the public interest, or those otherwise proscribed, were specifically allowed and not registered. When these gateways proved to be too narrow for many forms of cooperation, some behaviour was driven ‘underground’.103 This approach remained in place, with only minor modifications, until new competition laws were introduced in the mid- to late 1970s. Not all major European countries introduced a registration procedure. France followed a quite different path which was aligned with its tradition of strongly dirigiste policies. After the war the French government introduced a new Ordonnance to restrict monopolies, and For details, see Fellman and Shanahan (n 1). L. Warlouzet, ‘Competition Policy in the European Economic Community, 1957–1992: The Curse of Compulsory Registration?’ in Fellman and Shanahan (n 1). 100 Shanahan and Fellman (n 3). 101 Espeli (2016) (n 61); and Gerber (n 21) 161. 102 Gerber (n 21). 103 A. Scott, ‘The Evolution of Competition Law and Policy in the United Kingdom’, LSE Law, Society and Economy Working Papers 9/2009, London School of Economics and Political Science Law Department, http://​ssrn​.com/​abstract​=​1344807 [accessed 23 February 2022], 10. 98 99

62  Research handbook on cartels in 1953 it set out a decree on cartels (Prohibition of Unlawful Settlement Agreements and Establishment of a Technical Commission on Cartels). In 1963, legislation prohibiting firms from abusing their dominant position was passed and in 1977 a competition commission was established. Overall, however, the French followed a state-led economic model and no active competition policy was implemented. Only in 1986 was an autonomous competition policy (Ordonnance no. 86-1243 or the ‘Competition Statute’) implemented. Attitudes had changed, but this shift also aimed to bring the French economy into accord with the European integration process.104 Another cartel-friendly country that reluctantly introduced anti-cartel legislation was Switzerland. It only enacted its first legislation, a Cartel Act, in 1962. The principle of freedom of trade and freedom of contract had dominated attitudes toward market regulation. The law prohibited boycotts and the cartel commission made decisions based on the balance of positive and negative effects from a cartel’s conduct. In 1985 the Act was revised, but substantial amendments only occurred in 1995, in conjunction with other reforms designed to make the economy more competitive. The Swiss had voted against entering the European Union, but the new legislation drew heavily on European legislation.105 A general problem with anti-abuse-based legislation was that it was often vague: it was unclear what abuse or harm meant in practice, while enforcement authorities were often understaffed and had little power to prevent damage. For example, was it the intent that was to be regulated or the effect? How much ‘harm’ was permitted? Commonly, some sectors were not covered by the legislation – often key industries, including agriculture, commodities and services.106 Despite these shortcomings, the introduction of anti-cartel legislation founded on inhibiting abuse of power was significant in changing attitudes towards anti-competitive restrictive behaviour and misuse of monopoly power.

VII.

DEVELOPMENTS IN NON-WESTERN COUNTRIES

Some Latin American countries were early adopters of anti-cartel initiatives. In these nations US antitrust principles were influential, even though there were frequently obstacles caused by political instability, autarchic or nationalistic regimes and military governments. For example, in Brazil it was only in the 1990s that efficient antitrust legislation was finally implemented. Through the post-war decades, Brazil’s governments prioritized building strong and competitive companies in specific industries. This often benefited big business. In 1994, a political breakthrough saw the introduction of laws to promote liberalization and other economic reforms.107 In contrast, Argentina had by 1946 already introduced legislation prohibiting activities designed to achieve monopoly power, restrict prices or allocate markets. Nonetheless, implementation was weak.

Gerber and Azarnia (n 82); and M. Dechamps, ‘Competition Policy: France’, in A. Marciano and G. Battista Ramello (eds), Encyclopedia of Law and Economics (Springer, 2019). 105 P. Gugler, ‘Competition Law and Policy in Switzerland’ (2007) 9(2) OECD Journal of Competition Law and Policy 7. 106 S. Fellman and M. Shanahan, ‘Sectoral Influence on Competition Legislation: Evidence from the Cartel Registers, 1920–2000’ (2018) 92(4) Business History Review 1. 107 Todorov and Filho (n 87). 104

An historical account of anti-cartel enforcement  63 In Colombia, legislation which prohibited a broad range of restrictive and cartel-type agreements was introduced in 1959. Companies could, however, receive exemptions from the ban if the agreements ensured industry stability.108 Chile also introduced anti-cartel legislation in 1959. US influence is especially notable here, as an American consultancy group had recommended the enactment of a competition law, while at the same time an Antitrust Commission was established. The latter was never actively implemented. It was not until the late 1990s that demands to reform the Decree and implement efficient competition policy were raised by both government agencies and academics. In 2003, a total ban on all restrictive practices was implemented.109 As Chile’s history reveals, it is not enough to have strong anti-cartel legislation; the anti-monopoly authorities must also have the power to act. In other Latin American countries, such as Mexico, appropriate anti-cartel legislation was often introduced only as free trade agreements, such as the North American Free Trade Agreement (‘NAFTA’) or the Uruguay Round, were initiated. In 1992, the Federal Law on Economic Competition (Ley Federal de Competencia Económica (‘FLEC’)) was introduced in Mexico in preparation for the signing of the NAFTA.110 In Japan, the very strict antitrust legislation introduced by the US occupying forces was considerably weakened in 1953 and replaced with a new domestically developed legislation, the Antimonopoly Act. The Act was in principle intolerant of monopolies, coercive restrictions and unfair practices, but allowed exceptions for cartels triggered by economic depression and rationalization and specific resale price maintenance agreements. A system of registering the exempted cartels was introduced and the number of such cartels recorded grew significantly over the years. The Japanese antitrust policy experienced considerable backlash, as it was weakly enforced. An important reason was the existence of two competing administrations; the Fair Trade Commission (‘FTC’) and Ministry of International Trade and Industry (‘MITI’). The MITI, which administrated industrial policies, prioritized industrial advancement over competition policy. In the 1960s this started to change, partly due to international efforts to liberate trade and curb restrictive practices. Tensions in trade relations with the US were also a concern. Domestically, there was increasing awareness of the cost of developmental policies and MITI’s strong position was questioned. Several exempted cartels were removed from the register. Change was slow, and it was only in the 1990s that the strategy of exempting cartels was phased out and anti-cartel policies became more active.111 Nations with colonial antecedents adopted anti-cartel legislation at different times, depending on a constellation of factors, including the shape of their economy (for example, manufacturing or agriculturally based), stage of development, political stability and timing of integration into the world economy. For example, South Africa adopted anti-cartel legislation relatively early, in 1955. This was a broad law where the Board of Trade and Industry could investigate agreements considered to be against the public interest. The law allowed the agreements to be removed altogether or declared unlawful. Australia had passed legislation as early as 1906 in the Australian Industries Preservation Act 1906. Despite its general antitrust

Edwards (n 83) 340. F. Agüero, ‘Chilean Antitrust Policy: Some Lessons behind Its Success’ (2016) 79(4) Chilean Law and Contemporary Problems 123. 110 U. Aydin, ‘Competition Law and Policy in Mexico: Successes and Challenges’ (2016) 79(4) Law and Contemporary Problems 155. 111 Ohata and Kurosawa (n 85); and Gao (n 85) 108 109

64  Research handbook on cartels appearance, the law was mostly aimed at external American trusts, and after a series of legal cases it was found ineffectual. It was not until 1965 that a modest Trade Practice Act was adopted that included a secret register of anticompetitive cartel agreements and some specific prohibitions on price fixing.112 New Zealand, like Australia heavily dependent on agriculture, briefly adopted a register for cartel activity in the late 1950s, but it abandoned this for more comprehensive laws. The timing of cartel legislation in other former colonies is frequently associated with their economic development, the rise of free trade agreements and, later in the century, the rise of globalization.113

VIII. 1980s/2000 ONWARDS: THE NEW CHALLENGES AND NEW AREAS Following the lead of the Reagan and Thatcher governments, combined with the increasing influence of the Chicago School of economics in the US, businesses were given more freedom from government regulation during the 1980s. This continued in the US under President Clinton, when regulations controlling banking and insurance were weakened.114 These changes were linked to political rather than economic or welfare considerations.115 Some of the changes, together with increased civil and criminal penalties, had the effect of lowering prosecutions, as only the worst offenders were charged with serious offences. A further consequence of increasing the focus on consumer welfare, the rule of reason and the economic impact of cartel behaviour was that prosecuting criminal behaviour became more problematic.116 The Sherman Act of 1890 gave authorities the power to prosecute using both civil and criminal sanctions. In the US, until 1974, criminal behaviour was categorized as a misdemeanour (with a maximum one-year penalty) and such prosecutions were the exception.117 The Antitrust Procedures and Penalties Act (Pub L No 93-528, § 3, 88 Stat 1706, 1708 (1974)) increased civil penalties and recategorized criminal cartel behaviour as a felony, with a maximum penalty of three years. The penalties (civil and criminal) were increased in 1987, 1990 and 2004 and criminal penalties in the US may now include a maximum of ten years in prison. In 1900, only the US, Canada (which always had slightly higher criminal penalties than the US) and Australia had criminal sanctions for certain forms of cartel behaviour. Other countries were slower to adopt criminal sanctions – and always for serious misconduct. For example, Brazil introduced sanctions in 1990; in the UK it took until 2003; in Denmark 2013.118 By

M.P. Shanahan and K.A. Round, ‘Transforming Australian Business Attitudes to Competition: Responses to the Trade Practices Act 1965’ (2013) 56(3) Business History 434. 113 ibid. 114 W. Mueller, ‘Antitrust in the Reagan Administration’ (1994) 21/22 Revue Française d’Études Américaines 427. 115 D. Wood and J. Anderson, ‘The Politics of U.S. Antitrust Regulation’ (1993) 37(1) American Journal of Political Science 1. 116 D.D. Sokol, ‘Reinvigorating Criminal Antitrust’ (2019) 60 William & Mary Law Review 1545. 117 W. Kolasky, ‘Criminalising Cartel Activity: Lessons from the US Experience’ (2004) 12 Competition & Consumer Law Journal 207. 118 M. Simpson, ‘The Criminal Cartel Offence around the World’ in Norton Rose Fulbright, Competition World: A Global Survey of Recent Competition and Antitrust Law Developments with Practical Relevance, 2016, www​.nortonrosefulbright​.com/​-/​media/​files/​nrf/​nrfweb/​imported/​ 112

An historical account of anti-cartel enforcement  65 2020, more than 30 jurisdictions criminalized certain forms of cartel behaviour, especially bid rigging.119 The formation of the European Economic Community in 1957 had as a major objective the creation of a common market between European nations. This process, by necessity, included the coordination of competition policy and the alignment of regulations dealing with anticompetitive behaviour and cartels; this took time to occur.120 The Treaty of Rome came into force in 1958, and while the first antitrust violation was successfully prosecuted in 1964, it was not until 1966 that the basis of future EU competition policies began to emerge. Colomo and Kalintiri argue that after 1992 the main motivation of antitrust policy shifted from ‘trade-enabling’ considerations to a more ‘economics-based’ approach, with market integration always an underlying factor.121 This policy development path was clearly different from the US.122 Similarly, variation is seen in other parts of the world.

IX.

INTERNATIONAL EFFORTS TO PROMOTE PRO-COMPETITION POLICIES AROUND THE WORLD AFTER THE SECOND WORLD WAR

Beginning slowly after the Second World War, and gaining pace in the last quarter of the twentieth century, the most dominant factor affecting international and national markets has been globalization. This second globalization period realized increasing global coordination, removal of trade barriers, increasing capital flows and, since the 1960s, rapid growth in new economic regions. Economic integration and trade openness was promoted by international organizations such as the Organisation for European Economic Co-operation (‘OEEC’) and its successor the Organisation for European Economic Co-operation and Development (‘OECD’), the General Agreement on Tariffs and Trade (‘GATT’) and the World Trade Organization (‘WTO’), and with the creation of free trade areas. These organizations’ activities included the publication of anti-cartel and competition legislative initiatives. For example, the Havana Charter in 1948, within the International Trade Organization, included one chapter devoted to restrictive business practices. In 1961 the Council of the OECD officially established a Committee of Experts on Restrictive Business Practices, while the OEEC had an Expert Group focusing especially on European competition policy issues.123 In 1964 competition​ - world—q2-2016.pdf?la=en-gb&revision=1cbcdf34-ae03-4a23-8f4d-f1ddd3c84f7f [accessed 19 April 2022]. 119 Organisation for Economic Co-operation and Development Council, 2020 Working Party No. 3 on Co-operation and Enforcement Criminalisation of Cartels and Bid Rigging Conspiracies – Note by Australia (Paris, 2020) https://​one​.oecd​.org/​document/​DAF/​COMP/​WP3/​WD(2020)8/​en/​pdf [accessed 23 February 2022]. 120 L. Warlouzet, ‘The Centralization of EU Competition Policy: Historical Institutionalist Dynamics from Cartel Monitoring to Merger Control (1956–91)’ (2016) 54(3) Journal of Common Market Studies 725; and N. Rollings and W. Laurent, ‘Business History and European Integration: How EEC Competition Policy Affected Companies’ Strategies’ (2020) 62(5) Business History 717. 121 P. Ibáñez Colomo and A. Kalintiri, ‘The Evolution of EU Antitrust Policy: 1966–2017’ (2020) 83(2) Modern Law Review 321. 122 Fellman and Shanahan (n 1). 123 M. Palim, ‘The Worldwide Growth of Competition Law: An Empirical Analysis’ (1998) 43(1) Antitrust Bulletin 105.

66  Research handbook on cartels the United Nations established the United Nations Conference on Trade and Development, in part because of developing nations’ concerns with the power of multinational companies. The OECD has been an important forum for competition policy issues, both monitoring legislation and their implementation and giving advice. Dialogue with non-OECD countries concerning competition policy began in the late 1980s. Countries in the early stages of industrialization often supported specific industries and/or used trade protection as part of their developmental policies, but since the 1970s international expert organizations had requested countries to create well-functioning markets and deregulate to improve economic efficiency. Individual countries were often advised to implement more active anti-cartel and competition policies, both as a condition of entry into international markets and to facilitate growth in their own countries. In 1998 the Council of the OECD adopted recommendations for action against ‘hard-core’ cartels.124 Hard-core cartels were defined as ‘an anticompetitive agreement, anticompetitive concerted practice, or anticompetitive arrangement by competitors to fix prices, make rigged bids (collusive tenders), establish output restrictions or quotas, or share or divide markets by allocating customers, suppliers, territories, or lines of commerce’.125 The recommendations included proposals that all member countries ensure their competition laws, institutions, enforcement mechanisms and sanctions were sufficient to halt and deter hard-core cartels, and further, that they should coordinate their efforts against such cartels. These proposals aimed to strengthen and extend the existing communities of competition law already in existence, such as those in the Caribbean Community and European Union.126 The EU has also advocated implementation of policies to curb restrictive practices in neighbouring countries, especially in Eastern and Central European countries in advance of their becoming Member States of the EU. It has also advocated anti-cartel policies in countries that aimed to increase economic exchange with the Union, such as Turkey and Tunis.127 The increase in globalization in the last decades of the twentieth century saw national governments take more interest in the activities of international cartels. This was a period when regulatory authorities began to adopt a wider range of techniques to obtain successful prosecutions again anti-competitive behaviour. These included whistleblower and leniency provisions that were introduced in the US (as Corporate Leniency Policies and Individual Leniency Policies) in 1993, and in the EU (as a Leniency Notice) in 1996. Although empirical estimates of these provisions’ impact demonstrate mixed results, almost 60 per cent of cartels

124 Organisation for Economic Co-operation and Development Council, Recommendation of the Council Concerning Effective Action against Hard Core Cartels (Paris, 1998) www​.oecd​.org/​daf/​competition/​2350130​.pdf [accessed 23 February 2022]. 125 ibid Rec A2.a. 126 P. Wirz, ‘Imprisonment for Hard Core Cartel Participation: A Sanction with Considerable Potential’ (2016) 28(2) Bond Law Review 89. 127 T. Doleys, ‘Promoting Competition Policy Abroad: European Union Efforts in the Developing World’ (2012) 57(2) Antitrust Bulletin 337.

An historical account of anti-cartel enforcement  67 detected in the EU are the result of leniency applications.128 Whistleblower policies appear to be more successful in some jurisdictions than in others.129 In 2008, the 12th meeting of member nations in UNCTAD produced the Accra Accord, which outlined UNCTAD’s role in formulating competition policies as a component of assisting national economic development.130 By 2020, this organization had established a working group on cross-border cartels. The OECD now publishes regular reviews of more than 40 nations’ antitrust, anti-cartel, merger and pro-competition regulations.131 It also has a database of hard-core cartels that have been prosecuted by competition authorities since 2012 from almost 50 jurisdictions.132

X.

A SWIFT INTRODUCTION TO LEGISLATION OUTSIDE THE WESTERN WORLD

This hardening of anti-cartel enforcement has been visible in different regions and in individual countries in most parts of the world since the 1990s. One important step for European economic development was the transition to market economies of the former socialist countries after the demise of the Soviet Union. Competition policies or anti-cartel laws were mostly redundant during the socialist regimes. After the collapse of the Soviet Union, Eastern European economies began actively transitioning to market economies; that change included the adoption of anti-cartel legislation. An easing of the state-led system occurred in some countries prior to 1989. For example, some rudimentary rules against price agreements existed in Yugoslavia, which had more free market elements than the other socialist economies. In Hungary some economic reforms occurred in the early 1980s and legislation which could be considered as pro-competition had emerged by 1984. This was taken further in 1990. In 1981 and again in 1990 Poland introduced legislation that edged the economy towards freer markets and, as mentioned earlier, the country revived its dormant pre-Second World War law.133 In 1991 Bulgaria, the Czech Republic, Slovakia and Latvia all introduced anti-cartel legislation, and by 1995 all post-socialist countries in Europe (including Russia) had competition laws. These were especially focused on restrictive practices and on confronting misuse of a dominant position and monopolization.134 At first implementation was not always very active, but 128 S. Brenner, ‘An Empirical Study of the European Corporate Leniency Program’ (2009) 27 International Journal of Industrial Organization 639; J.D. Jaspers, ‘Leniency in Exchange for “Cartel” Confessions’ (2020) 17(1) European Journal of Criminology 106; and N.H. Miller, ‘Strategic Leniency and Cartel Enforcement’ (2009) 9 American Economic Review 750. 129 T. Nyreröd and G. Spagnolo, ‘Myths and Numbers on Whistleblower Rewards’ (2021) 15(2) Regulation & Governance 82. 130 Qaqaya and Lipimile (n 31). 131 Organisation for Economic Co-operation and Development Council, Country Reviews of Competition Policy Frameworks (Paris, 2021) www​.oecd​.org/​daf/​competition/​cou​ntryreview​sofcompeti​ tionpolicy​frameworks​.htm [accessed 23 February 2022]. 132 Organisation for Economic Co-operation and Development, The OECD International Cartel Database https://​qdd​.oecd​.org/​subject​.aspx​?Subject​=​OECD​_HIC [accessed 23 February 2022]. 133 T. Varady, ‘The Emergence of Competition Law in (Former) Socialist Countries’ (1999) 47(2) American Journal of Comparative Law 229; and R. Pittman, ‘Competition Law in Central and Eastern Europe: Five Years Later’ (1998) 43(2) Antitrust Bulletin 179. 134 Doleys (n 127).

68  Research handbook on cartels according to Kovacic, transition economies quite vigorously adopted anti-cartel regulations in order to move from central planning to market economies.135 These laws were subsequently reformed; this was especially the case in those Eastern and Central countries that later aspired to enter the European Union and which committed to adopt EU competition policies. Other countries that transitioned to market economies during the last decades of the twentieth century also introduced legislative initiatives, most notably those in Latin America. The 1990s saw nations that did not have such legislation introduce it, while in those with existing legislation, the laws were more vigorously enforced. Competition laws in Latin America were often associated closely with political liberalization.136 In East and South East Asia, competition policies and cartel laws had been implemented quite early after the war. Outside Japan, India, Pakistan and South Korea introduced anti-cartel regulations in 1969, 1970 and 1980 respectively.137 India received its first legislation in 1969, the Monopolies and Restrictive Trade Practices Act (‘MRTP’) Act, in response to growing problems of concentrated business ownership, especially family-owned business groups. The MRTP Act was, in principle, prohibition-based legislation which required government permission for anticompetitive agreements and a registration of exempt cartels. The legislation was not vigorously enforced; it contained many ‘gateways’ and the administrators often focused on market problems other than cartelization and collusion. The system of registration suffered from overwhelming numbers of applications. Modern legislation intolerant of cartels and which prohibited restrictive practices was passed in 2002, but only implemented in 2009 when the MRTP Act was finally abolished.138 Taiwan initiated anti-cartel legislation in 1980, influenced by Japan and other Western countries and the expansion of free trade. Balancing pressures from strong domestic cartels, with policies enhancing trade only saw legislation enacted in 1991.139 Anti-cartel legislation was introduced in most East and South East Asian countries only in the 1990s and 2000. Many of these countries had pursued a developmental model which favoured active industrialization, protection and an active state. These strategies were pursued at the cost of vigorous competition policy. Moreover, the lack of a democratic political system in some countries also affected the active implementation of such a legislation. Nonetheless, the adoption of highly intolerant legislative frameworks has been rapid since the 1990s.140 China is in this respect a special case. Focused on internal political and economic problems after the Second World War, the most populous nation on the planet did not open its doors to outside economic markets until the late 1970s. The process proceeded relatively slowly,

W. Kovacic, ‘Lessons of Competition Policy Reform in Transition Economies for US Antitrust Policy’ (2000) 74(2) St John’s Law Review 361. 136 M. Palim, ‘The Worldwide Growth of Competition Law: An Empirical Analysis’ (1998) 43(1) Antitrust Bulletin 105. 137 F. Sayyeda, ‘Competition Law in Pakistan: Brief History, Aspirations and Characteristics’ (2012) 38(1) Commonwealth Law Bulletin 43; A. Bhattacharjea, ‘India’s New Antitrust Regime: The First Two Years of Enforcement’ (2012) 57(3) Antitrust Bulletin 449; and M. Yang, ‘Competition Law and Policy of the Republic of Korea’ (2009) 54(3) Antitrust Bulletin 621. 138 Battarcharjea (n 137). 139 See e.g. M. Williams, Competition Policy and Law in China, Hong Kong and Taiwan (Cambridge University Press, 2005); L. Liu, ‘All about Fair Trade. Competition Law in Taiwan and East Asian Economic Development’ (2012) 57(2) Antitrust Bulletin 259. 140 See e.g. Bradford et al (n 1). 135

An historical account of anti-cartel enforcement  69 with reforms to price controls, foreign investment, agriculture and entrepreneurial activity slowly being eased. More significant were changes in the early 1990s when many state-owned enterprises were at least partially privatized, although state monopolies remained in several sectors. While discussions about anti-monopoly legislation began in 1987, such laws were not formally adopted until 2007. The issue of how to regulate anticompetitive agreements became more urgent after China joined the WTO in 2001. The rapid growth of foreign firms in China and the country’s increasing integration in the global economy started to affect its own domestic market. The new legislation was broad in scope (including bans on restrictive practices, merger reviews and preventing abuse of market dominance), but involved balancing protection with a more open economy, creating regulations with considerable room for interpretation. Their enforcement has been flexible and adaptable.141 Under state capitalism, enforcement differs from systems with less government involvement. For example, the lack of an independent enforcement authority and the size and dominance of state-backed companies make separating economic and political behaviour difficult. Outside of the US and Europe, China is now the third major regulator of markets in the world. In several countries of the Middle East and North Africa (‘MENA’), competition legislation exists, but only since 2000. Most countries in the region are late comers in this field and some still lack legislation. The region should not be considered monolithic, however.142 For example, Israel has legislation dating from 1959, while among the nearby Islamic countries the process developed in different ways. Some early movers such as Tunisia, introduced legislation in 1991 to improve its trade relations with Europe. Its anti-cartel models came from the French Ordonnance of 1986 and later EU regulation.143 Complicating the introduction of such legislation is the belief that competition legislation is Western and to be resisted. Political instability has also affected several jurisdictions, and market intervention by the state is common. Since the 1990s, a number of African countries outside South Africa have prohibited price fixing, market allocation and other forms of anticompetitive behaviour. The effectiveness of these laws varies, and enforcement is often inadequate due to, among other things, political instability and occasionally corruption.144 Some African and MENA countries still lack any competition legislation.145 The technological and business changes in computing and communications in the 1980s, and their impact on markets, were resonant of the changes associated with steel, oil and railways in the nineteenth century. They saw new firms achieve dominant positions in a range of markets in many countries. The subsequent 40 years have seen many court cases with the new global players. The arguments, however still return to the question of freedom of contract versus freedom to trade, or whether market dominance was achieved through efficiency and techno-

Y. Svetiev and L. Wang, ‘Competition Law Enforcement in China between Technology and Industrial Policy’ (2016) 79(4) Law and Contemporary Problems 184. 142 M. Dabbah, Competition Law and Policy in the Middle East (Cambridge University Press, 2007). 143 P. Speelman, ‘Competition Law in the Middle East and North Africa: The Experiences of Tunisia, Jordan, and Egypt’ (2016) 48(4) New York University Journal of International Law and Politics 1227. 144 U. Aydin and T. Blüthe, ‘Competition Law & Policy in Developing Countries: Explaining Variations in Outcomes; Exploring Possibilities and Limits’ (2016) 79(1) Law and Contemporary Problems 1. 145 A. Bradford, A.S. Chilton, C. Megaw and N. Sokol, ‘Competition Law Gone Global: Introducing the Comparative Competition Law and Enforcement Datasets’ (2019) 16(2) Journal of Empirical Legal Studies 411. 141

70  Research handbook on cartels logical advance or abuse of power. With new technologies also came new versions of familiar forms of market manipulation: replacing ‘full-line forcing’ was ‘system lockout’; instead of ‘exclusive dealing’ came ‘network effects’. Unlike earlier cycles of market concentration and antitrust regulation, where the good or service was clear, there is today less transparency about what is actually being bought and sold (communication services to individuals or advertising space to business), or the price being paid (whether the price is paid via subscription or through unintentional transfer of personal attributes for on selling). The clearest example of regional difference in antitrust policies can be seen in differences in regulatory approach to the behaviour of the large computing and communications firms. Several of these companies, which either did not exist or were not particularly large in the 1980s, now have dominant positions in global, multibillion-dollar enterprises. Rollings and Waurlouzet suggest that the EU is now seen as acting more aggressively against such cartels than the US Department of Justice.146 Differences in attitude and approach are one explanation.147 While the US and EU are currently important jurisdictions influencing the global agenda on competition policy in this field, China’s attitude, especially to its own multinational firms, is also critical.148 The growth of China’s socialist economy has proved to be a test of how China adapts its own antitrust regulations to accommodate foreign companies and how the rest of the world adapts its antitrust regulations to accommodate China’s emerging state-owned enterprises.149 In an approach that echoes the public exposure of the cartel registers, China’s antitrust authority, the State Administration for Market Regulation, has required its ‘tech giants’ to publicly pledge their commitment not to engage in anticompetitive behaviour.150 Exactly how this framework will develop, and how the US and EU anti-cartel structures align with China, awaits to be seen.

XI.

CONCLUDING REMARKS

This chapter has provided a sweeping overview of the emergence and adoption of anti-cartel legislation over the past 130 years, combined with snapshots from specific countries. Modern anti-cartel legislation and policy measures to enforce competition first emerged in Canada and the US, were gradually adopted in parts of Europe and then spread to emerging Rollings and Laurent (n 120). M. Coppola and R. Nazzini, ‘The European and US Approaches to Antitrust and Tech: Setting the Record Straight – A Reply to Gregory J. Werden and Luke M. Froeb’s Antitrust and Tech: Europe and the United States Differ, and It Matters’ (Competition Policy International, 2020) www​.ftc​.gov/​system/​ files/​attachments/​key​-speeches​-presentations/​europe​-column​-may​-2020​-full​.pdf [accessed 23 February 2022]; and G. Werden and L. Froeb, ‘Antitrust and Tech: Europe and the United States Differ, and It Matters’ (CPI Antitrust Chronicle, 2019) www​.​competitio​npolicyint​ernational​.com/​antitrust​-and​-tech​ -europe​-and​-the​-united​-states​-differ​-and​-it​-matters​-2/​ [accessed 23 February 2022]. 148 A. Bradford, A. Chilton, K. Linos and A. Weaver, ‘The Global Dominance of European Competition Law over American Antitrust Law’ (2019) 17 Journal of Empirical Legal Studies 731. 149 E. Fox, ‘An Antimonopoly Law for China: Scaling the Walls of Government Restraints’ (2008) 75(1) Antitrust Law Journal 173; and A.H. Zhang, Chinese Antitrust Exceptionalism: How the Rise of China Challenges Global Regulation (Oxford University Press, 2021). 150 S. Yang, ‘China’s Tech Giants Vow, in Unison, to Play by Regulators’ Rules’ (The Wall Street Journal, 14 April 2021) www​.wsj​.com/​articles/​chinas​-tech​-giants​-vow​-in​-unison​-to​-play​-by​-regulators​ -rules​-11618402448 [accessed 23 February 2022]. 146 147

An historical account of anti-cartel enforcement  71 economies and developing countries. There are many reasons behind this global development: a breakthrough of anti-cartel and antitrust ideologies; economic theory-building; and increasing participation in international markets, economic integration and supranational efforts to promote free trade and removal of restrictive practices that hampered the free flow of goods and services. Factors including transnational learning and foreign advocacy to adopt new and more efficient legislation, as well as domestic economic modernization, have all contributed to the adoption of roughly aligned legislative approaches. The journey to adopt anti-cartel legislation has been complex, as countries and regions have followed their own historical path, and rarely has a country completely transformed from a jurisdiction tolerant of cartels to one that is completely non-tolerant. Since the 1990s global markets have seen a swift spread of legislative measures against cartels, including a rapid switch from a complete absence of legislation to legislation that is quite intolerant of cartels. Those countries that began earlier in the twentieth century, especially in Europe, often followed a more gradual, step-by-step transition towards less tolerant legislation. The development appears to imitate economic ‘catching up’ and is connected to this concept. Competition legislation is often linked to modernization of the economy, trade liberalization and, not infrequently, regime changes. Empirical evidence that such legislation is a prerequisite for economic growth or vice versa is not conclusive. Individual countries such as South Africa, Australia, New Zealand, Israel or Tunisia could, perhaps, have initiated competition acts earlier; however, in these cases, the colonial history and models derived from former colonial powers affected the implementation of their laws. In other countries with significant foreign influence, such as Germany and Japan, the introduction of anti-cartel legislation both triggered change and caused recoil. The effects of the various relevant pieces of legislations also varied. The existence of a particular piece of legislation is not enough. Its enforcement can be undermined by weak institutions or weak authorities. The need to foster a competition culture is frequently emphasized as a core for successful policies to curb anti-competitive behaviour. Countries that did not have formal competition legislation often had other laws that affected market behaviour and restrictive practices. Moreover, competition legislation includes a great variety of different types of restrictive practices and non-competitive behaviour; in some cases the law was narrow in scope, only controlling cartel agreements (horizontal or vertical), but it could also be broad and include a variety of restrictive practices and monopolistic behaviour, such as abuse of dominant market position and mergers and acquisitions. In some countries regulations focused on prices, unfair trade practices and monopolization, but not cartel agreements. The effect of specific legislation should always be viewed in its broader legislative context and as part of the economic and institutional situation. These anti-cartel laws always worked in conjunction with, and often complemented, other economic–political goals. Such important interactions, however, lie beyond the scope of this chapter. Attempts to balance the competing demands of free markets, freedom of trade, freedom of contracting and questions concerning economic efficiency and economic prosperity form a multifaceted problem that is not easy to resolve. As this chapter has shown, kings, dictators, autocrats and democratically elected officials have all grappled for decades with the problem of achieving equilibrium between these competing elements. Although circumstances change and technologies may advance, and the range of competing actors may vary from a local elite

72  Research handbook on cartels to global actors, the challenge of encouraging economic innovation and activity without facilitating exploitation and manipulation is a constant and ongoing one.

4. The morality of cartel activity Andreas Stephan

I. INTRODUCTION This chapter examines the extent to which cartel activity is morally wrongful and the resultant implications for cartel law more generally. It is important to begin by defining what is meant by morality and how it is different to ethics, as the two terms are closely related and are often used interchangeably. Morality concerns ‘a sense of right or wrong according to conscience’, while ethics concern ‘the rules and principles that ought to govern’ human behaviour.1 Thus, professional ethics generally prescribe how individuals are expected to behave in particular settings, regardless of whether those expectations align precisely with their sense of morality. Something unethical is not therefore necessarily something that is immoral. By contrast, morality is an instinctive emotion that is harder to pin down.2 While there are behaviours that would attract moral opprobrium from most or all in society, morality can also be highly subjective and capable of being wholly irrational depending on an individual’s particular set of beliefs.3 Any discussion of law and morality naturally gravitates towards the study of jurisprudence (or the philosophy of law), which consists of a wide spectrum of theories that explore this relationship. The most notable include: natural law theories that suggest laws are informed by a higher moral authority;4 positivism, which suggests there need be no connection at all between morality and law so long as the law is correctly formed;5 utilitarianism, which places an emphasis on the value of liberty over morality in promoting aggregate happiness for the greater good of all;6 and neoliberalism, which broadly argues that, to be moral, the law should not interfere with financial affairs (known as a ‘just holding’ in entitlement theory) through interventionist policies. A transfer is ‘just’ so long as the buyer pays a price they are willing to pay, although there is a contradiction in this argument in that it appears to assume the buyer has a choice and that markets are not therefore monopolized.7 The idea that there should be a strong link between morality and the law is weakened considerably by the fact we live in a liberal, pluralistic society in which there is a diversity of moral views and their relationship with the law is capable of changing significantly over time. In the UK, the publication of the 1957 Wolfenden Report reflected this relationship. It recommended

1 2

11.

Oxford English Dictionary (5th edn, HarperCollins, 2000). T.J. Horton, ‘Restoring American Antitrust’s Moral Arc’ (2017) 62(1) South Dakota Law Review

J. Coleman, ‘Competition and Cooperation’ (1987) 98(1) Ethics 76. T. Aquinas, Selected Political Writings (J.G. Dawson tr, Blackwell, 1959); J. Finnis, Natural Law and Natural Rights (Clarendon Press, 1980); and H.L.A. Hart, ‘Positivism and the Separation of Law and Morals’ (1958) 71 Harvard Law Review 593. 5 R. Dworkin, ‘Hard Cases’ (1975) 88 Harvard Law Review 1057. 6 J. Bentham, A Fragment on Government with an Introduction to the Principles of Morals and Legislation (first published 1776, Basil Blackwell 1948). 7 R. Nozick, Anarchy, State and Utopia (Basic Books, 1974). 3 4

73

74  Research handbook on cartels that homosexuality be decriminalized and pointed to glaring inconsistencies in the categories of behaviour that were unlawful at the time on the grounds of being ‘immoral’. They pointed, for example, to the fact that adultery between heterosexual couples was not a crime. There are also examples of where changes to the law have helped shape moral views – for example in relation to the perception of driving under the influence of alcohol. The law can therefore exist independently of morality and the relationship between the two can be highly subjective and capable of changing over time, as societal attitudes evolve. This is important because it opens up the possibility that certain types of cartel conduct might attract strong moral opprobrium while others might be viewed as morally ambiguous. What determines this is the characteristics of the cartel conduct, the circumstances that surround the conduct and the level of understanding an observer has about its nature and consequences. This is the central thesis of this chapter. We begin by setting out why morality is important to cartel law, despite competition law inhabiting a regulatory space that is largely void of moral considerations. This chapter argues that while the focus of morality lies mainly in the question of whether cartel behaviour should be treated as crime, it is actually important to the legitimacy of all cartel enforcement because such enforcement is often punitive and not merely regulatory in nature. It also speaks to the extent to which individuals are willing to comply with cartel law and report it to an employer or competition authority. The chapter then goes on to consider the extent to which cartels are considered morally wrongful. It is argued that as a fluid concept, what determine morally offensive cartel conduct are various pull factors that make it more likely that cartel conduct will attract moral opprobrium and push factors that have the opposite effect. Ultimately, these factors help us to identify both the types of cartel conduct that are most likely to be considered morally offensive and the potential policy tools for strengthening the moral opprobrium associated with cartel behaviour more generally.

II.

IS MORALITY IMPORTANT TO CARTEL LAW?

On the face of it, competition law inhabits a regulatory sphere that is principally concerned not with what is right or wrong, but with what is efficient. It is underpinned first and foremost by a set of economic theories that suggest interventions are necessary in markets to prevent monopoly outcomes that result in less being produced for a higher price, and a reduction in innovation and other benefits of rivalry. Yet in economics what is important is not the nature of the acts but rather their effect on the market. Academics belonging to the Chicago School of thought, for example, argue that behaviour should only be prohibited if it can be demonstrated to have a clear anticompetitive effect. They brand rules that are based on form (and especially those that ignore the effects of conduct altogether) as being blunt and inefficient in likely also prohibiting behaviour that is benign or beneficial.8 As Stones points out, their preference is

8 See for example: R.H. Bork, The Antitrust Paradox: A Policy at War with Itself (The Free Press, 1993); F.H. Easterbrook, ‘Workable Antitrust Policy’ (1986) 84(8) Michigan Law Review 1696; M.K. Block and J.G. Sidak, ‘The Cost of Antitrust Deterrence: Why Not Hang a Price Fixer Now and Then?’ (1980) 68(5) Georgetown Law Journal 1131.

The morality of cartel activity  75 to incorporate economics into an ex post effects-based analysis in a way that would combine rules, presumptions and structured tests, to ensure legal certainty and administrability.9 A case-by-case consideration of the effect on the market has undoubtedly become more important in competition law over the past 30 years. In the EU this has manifested itself in a more effects-based approach to merger control and aspects of abuse of dominance, and in the US there has been a gradual shift away from per se prohibitions of anti-competitive behaviour to the rule of reason approach that balances the likely pro- and anti-competitive effects.10 In justifying its decision to reclassify resale price maintenance as rule of reason, the US Supreme Court cited the fact that the ‘economics literature is replete with procompetitive justifications for a manufacturer’s use of resale price maintenance’.11 Much of the rest of the world (and many US state-level antitrust regimes) continue to treat minimum resale price maintenance as per se illegal. Even when it comes to the question of who should be protected by competition law, there is no consensus that might be somehow linked to morality. For example, there is a longstanding debate over whether competition policy should be focused on the consumer welfare standard or on total welfare.12 If we accept the contention of neo-classical economists that total welfare is preferable and that transfers between consumer to producer are welfare-neutral, then this very much shapes the extent to which such transfers can be characterized as wrongful, so long as total welfare is not depleted.13 Also, the neoclassical economic principles that underpin modern EU competition law hold very different values to the ordoliberal tradition that dominated before it.14 In this diverse regulatory landscape, the design of cartel laws does little to make them stand out as any less morally ambiguous than other areas of competition law regulation. In particular, prohibitions of the most serious forms of cartel conduct are not focused on effects and are deliberately designed to capture a very wide spectrum of anti-competitive arrangements, including those entered into out of ignorance and those involving only very minimal communication and exchange of information between competitors. Someone with limited knowledge of competition law would be forgiven for wondering how this type of prohibition – which could potentially engage two competitors in a very small market – could possibly have a closer link with morality than the exclusionary behaviour of a dominant firm. The wide scope of cartel laws is deliberate, to ensure that competitors cannot circumvent regulation by simply watering down the nature of their collusion. Yet these prohibitions are meant to capture the categories of cartel behaviour that are most likely to cause anti-competitive harm. In not considering the effects of these arrangements, these prohibitions do not generally distinguish

R. Stones, ‘The Chicago School and the Formal Rule of Law’ (2019) 14(40) Journal of Competition Law & Economics 527. 10 See D.D. Sokol, ‘Reinvigorating Criminal Antitrust’ (2019) 60(4) William & Mary Law Review 1545, discussing R.H. Bork, ‘Legislative Intent and the Policy of the Sherman Act’ (1966) 9 Journal of Law & Economics 7. See also S. Kimmel, ‘How and Why the Per Se Rule Against Price-Fixing Went Wrong’ [2011] Supreme Court Economic Review 19. 11 Leegin Creative Leather Prods., Inc. v PSKS, Inc., 551 US 877 (2007) 889 and 894–5. 12 See for example: M. Harker, ‘Antitrust Law and Administrability: Consumer versus Total Welfare’ (2011) 34(3) World Competition 433. 13 See e.g. R. van den Bergh and P. Camesasca, European Competition Law and Economics (Sweet & Maxwell, 2006). 14 P. Akman, ‘Searching for the Long-Lost Soul of Article 82EC’ (2009) 29(2) Oxford Journal of Legal Studies 267. 9

76  Research handbook on cartels between attempted and actual cartel infringements, or even require them to have been fully implemented. Indeed, any exchange of commercially sensitive information has the potential to fall into this category, including behaviour that is likely to have benign effects.15 It is notable that even in the United States, where (public) cartel enforcement is a criminal process, the scope of the Sherman Act is far-reaching and there is no strong mens rea element to signal the moral offensiveness of what is prohibited.16 This means that cartel prohibitions are not restricted to arrangements with clear characteristics of a hard-core cartel but also include what Whish and Bailey describe as ‘atypical’ cartels, by which they mean those on the fringes of the prohibition.17 In addition, there are key aspects of how cartel laws are enforced that further suggest morality is of little relevance. Stucke is critical of an enforcement system that he describes as ‘largely encamped in utilitarianism and the economic theory of optimal deterrence’.18 This is evident in the emphasis on punishing corporations over individuals and holding them vicariously liable for the actions of their employees. Stucke and others suggest the very high levels of corporate fines reflect a belief in the economic paradigm of deterrence (or deterrence theory), in which businesses make rational choices based on the benefits and costs of entering into a cartel. Under the logic of deterrence theory, it is only where the expected cost (represented by the likelihood of getting caught and the size of punishment) is significant enough that deterrence is achieved. Also, increasing the size of the sanction is thought to have the same pro-deterrence effect as increasing detection.19 In this theoretical economic space, morality is only relevant to the extent that any perceived threat of social stigma or loss of reputation adds to the size of the sanction.20 It is interesting to note that early writings in economics were concerned with rationalising the effects of morality in economic terms. In Theory of Moral Sentiments (1759), Adam Smith suggested that individual behaviour towards society was constrained by external law enforcement, external incentives for control of opportunistic behaviour (which he characterized as a market) and the internal structure of the individual’s utility function that performs self-monitoring against that behaviour.21 So while deterrence theory has no discernible link to morality, it does recognize the effect of moral opprobrium as a cost, meaning that the more morally offensive cartels are considered to be, the greater the cost of participating in a cartel.22 The second feature of cartel enforcement that suggests little space for morality is the strong reliance on unmistakably utilitarian tools such as leniency and settlements or plea bargains. Murphy argues that the use of leniency programmes undermine compliance by making the See the discussion of cover bids and review of the relevant literature and case law in A. Stephan and M. Hviid, ‘Cover Pricing and the Overreach of “Object” Liability Under Article 101 TFEU’ (2015) 38(4) World Competition 507. 16 United States v United States Gypsum Co., 438 US 422 (1978). 17 See R. Whish and D. Bailey, Competition Law (9th edn, Oxford University Press, 2018) Chapter 13. 18 M. Stucke, ‘Morality and Antitrust’ [2006] Columbia Business Law Review 442, 449. 19 G. Becker, ‘Crime and Punishment: An Economic Approach’ (1968) 76 Journal of Political Economy 169. 20 S. Shavell, ‘Law versus Morality as Regulators of Conduct’ (2002) 4 American Law & Economics Review 227. 21 See G.M. Anderson and R.D. Tollison, ‘Morality and Monopoly: The Constitutional Political Economy or Religious Rules’ (1992) 12(2) Cato Journal 373. 22 D.D. Sokol, ‘Cartels, Corporate Compliance, and what Practitioners Really Think about Enforcement’ (2012) 78(1) Antitrust Law Journal 201, 203. 15

The morality of cartel activity  77 enforcement environment distinctly void of morality. What matters in the leniency ‘game’ is who is first through the door, and not the fact that one should ensure that perpetrators are punished.23 While reduced punishment for cooperation is common in sentencing guidelines around the world, allowing a guilty party to escape punishment altogether does not sit well in most legal traditions outside of competition law, and it undermines attempts to establish a strong moral underpinning for the prohibition of cartels. Indeed, leniency and settlement procedures reward virtues like cheating, distrust and informing on others, in a way that many might consider distinctly immoral.24 These tools are made all the more controversial by the way they combine the fate of the firm with that of the individuals who are responsible for bringing about the cartel in the first place. Even where there are no sanctions against individuals, leniency programmes generally require undertakings to ensure the full and continued support of current and former employees. This can often undermine a business’ own attempts to discipline those responsible, as those individuals typically hold the key to applying successfully for leniency.25 Where sanctions against individuals do exist, there is some evidence that plea bargaining is being used strategically by some businesses to reduce their corporate fine by facilitating a guilty plea from one of their employees.26 For those who believe cartel enforcement is void of morality, the growing international criminalization of hard-core cartel conduct has drawn particular criticism. If, at best, cartel conduct is morally ambiguous, then the use of criminal law might be seen as entirely inappropriate, because it represents an extreme act of regulatory control rather than any public upswelling of moral outrage at their nature and/or effects. For example, Parker described cartel criminalization in Australia as driven by ‘blind’ economics: ‘Cartel conduct is not a criminal offence because of its moral or political ‘offence’-iveness per se. Rather, it is a criminal offence because achieving economic welfare through competition is thought to be so important that anti-competitive conduct should be prohibited and, once prohibited, subject to the toughest possible sanctions.’27 Similarly, Haines and Beaton-Wells describe cartel criminalization as the ‘poster child’ of competition policy, in that it ‘elevates competition to a collective moral universe’ and pro-

23 J. Murphy, ‘Combining Leniency Policies and Compliance Programmes to Prevent Cartels’, in C. Beaton-Wells and C. Tran (eds), Anti-Cartel Enforcement in the Contemporary Age: Leniency Religion (Hart, 2015) 321–3. See also M.E. Stucke, ‘Leniency, Whistle-Blowing and the Individual: Should We Create Another Race to the Competition Authority?’, in Beaton-Wells and Tran (n 23); and the discussion in M.E. Siltoaja and M.J. Vehkaperä, ‘Constructing Illegitimacy? Cartels and Cartel Agreements in Finish Business Media from Critical Discursive Perspective’ (2010) 92 Journal of Business Ethics 493. 24 A. Stephan, ‘Survey of Public Attitudes to Price-Fixing and Cartel Enforcement in Britain’ (2008) 5(1) Competition Law Review 123, 140; For a discussion of the role of distrust in competition law, see C.R. Leslie, ‘Trust, Distrust, and Antitrust’ (2004) 82(3) Texas Law Review 515. 25 P. Whelan, The Criminalization of European Cartel Enforcement: Theoretical, Legal, and Practical Challenges (Oxford University Press, 2014), 133. See also the discussion of AkzoNobel’s internal amnesty tool in G. De Stefano and A. Stephan, ‘Impact of Cartel Enforcement on Compliance in the Chemical Industry’ (2023) Journal of Antitrust Enforcement, forthcoming. 26 See for example H. Greimel, ‘Confessions of a Price Fixer’ Automotive News (16 November 2014), as discussed in A Stephan, ‘The Price Fixer: Compliance Perspectives from the Other Side’ in A. Riley, A. Stephan and A. Tubbs (eds), Perspectives on Antitrust Compliance (Concurrences, 2022). 27 C. Parker, ‘Economic Rationalities of Governance and Ambiguity in the Criminalization of Cartels’ (2012) 52 British Journal of Criminology 974, 976.

78  Research handbook on cartels motes the individual over collective efforts for prosperity.28 This sentiment is also reflected in the writings of some political scientists who are taken by the dominance of a relatively small field of economics in this far-reaching regulatory sphere. Wilks comments that it marks a shift that ‘privileges a conception of the market which is neo-liberal, or perhaps Anglo-American, in its basic assumptions about how firms and individuals behave’.29 Horton argues that neoclassical economics has ‘led antitrust regulation’ astray by treating it as amoral and that notions like ‘fairness’ should be as central to the discipline as welfare.30 Finally, Williams points out a major problem in relation to labelling morally offensive behaviour as criminal, in that there is often no consistency or coherence to which behaviour is labelled criminal and which is civil (albeit often with the civil law serving a similarly punitive role).31 The problem is that ‘credibility may be undermined if one activity is criminalised while another, apparently equally delinquent activity, is not’.32 These criticisms further reflect the perceived amoral underpinnings of cartel enforcement described above – in particular the reliance on deterrence theory. Whelan’s work explores whether criminal antitrust sanctions are appropriate in the absence of negative moral content. He observes that deterrence theory (the key justification for cartel criminalization) is seen as lacking legitimacy because of the negative consequences if it is applied to a morally neutral offence, as it weakens all criminal law.33 For many scholars the identification of a sufficient moral delinquency goes to the heart of whether it is justifiable to intervene using criminal law. It is what Williams describes as the ‘bootstrap problem’.34 This refers to the way criminal law is used by policy makers to change public opinion in relation to behaviour they may not instinctively consider to be morally delinquent enough to be labelled criminal. Where this fails, we are left with ‘sticky norms’ – where a gap exists between what the law treats as morally offensive and how it is actually perceived by society at large.35 Whelan counters that such arguments ignore the educative effects of sanctions: ‘they do not allow for the criminal law actually to create, and not just reflect, a moral opprobrium’ for what is considered undesirable behaviour. There is at least some reciprocal effect in this context.36 He nevertheless

28 F. Haines and C. Beaton-Wells, ‘Ambiguities in Criminalizing Cartels’ (2012) 52 British Journal of Criminology 953, 972. 29 S. Wilks, ‘Cartel Criminalisation as Juridification: Political and Regulatory Dangers’, in C. Beaton-Wells and A. Ezrachi (eds), Criminalising Cartels: Critical Studies of an International Regulatory Movement (Hart, 2011) 350–51, as discussed in C. Harding and J. Edwards, Cartel Criminality: The Mythology and Pathology of Business Collusion (Ashgate, 2015) 57. 30 Horton (n 2). 31 R. Williams, ‘Cartels in the Criminal Landscape’ in Beaton-Wells and Ezrachi (n 29) 309. 32 ibid. 33 Whelan (n 25), in particular 56–58, with reference to: F. Sayre, ‘Public Welfare Offences’ (1933) 33 Columbia Law Review 55; and H. Packer, The Limits of the Criminal Sanction (Oxford University Press, 1968). 34 Williams (n 31) 299. 35 S. Green, Lying, Cheating and Stealing: A Moral Theory of White Collar Crime (Oxford University Press, 2006) 24, citing D.M. Khan, ‘Gentle Nudges vs Hard Shoves: Solving the Sticky Norms Problem’ (2000) 67 University of Chicago Law Review 607. 36 Whelan (n 25) 57, with reference to: J. Coffee, ‘Does “Unlawful” Mean “Criminal”? Reflections on the Disappearing Tort/Crime Distinction in American Law’ (1991) 71 Boston University Law Review 193, 201; and H. Ball and L. Friedman, ‘The Use of Criminal Sanctions in the Enforcement of Economic Legislation’ (1964) 17 Stanford Law Review 197.

The morality of cartel activity  79 accepts that the lack of legitimacy is a concern that needs to be addressed if criminalization is ultimately to succeed.37 Although cartel laws appear to inhabit a distinctly amoral space, there are three reasons why the relationship between morality and cartels is of great importance. The first is in relation to the use of criminal sanctions discussed above. As well as the issue of legitimacy, there are practical considerations such as the willingness to bring prosecutions and the danger of jury nullification – where a jury acquits, regardless of how the law is designed or their instructions from the judge, because they do not feel cartel conduct should be treated as crime.38 The second is a broader issue of legitimacy regarding all cartel enforcement, regardless of whether it occurs through a criminal or administrative procedure. Indeed, there is a real question of what practical difference the labels ‘criminal’ and ‘civil’ have in modern law enforcement.39 The problem is that while competition law is principally an area of business regulation, the investigative tools and sanctions imposed are distinctly criminal in nature, as confirmed by rulings of the European Court of Human Rights.40 The purpose of cartel enforcement is not simply to deprive the perpetrators of any unjust enrichment, but to punish and deter. The public statements of competition authorities speak not of regulatory breaches, but of theft by ‘well-dressed thieves’.41 As Harding and Edwards note, the moral and political basis for strong censure remains uncertain and a matter of uneasy consensus. What is so bad about cartel activity so as to justify criminalisation? Despite a strong official rhetoric, the basis for strong condemnation remains unclear and contestable, and an effective and credible system of legal control requires a firmer foundation, and indeed legitimacy.42

Parker describes this as a key vulnerability and one that could ultimately undermine cartel enforcement altogether, as the lack of clear moral support could result in businesses lobbying for softer laws and less enforcement.43 It also makes cartel enforcement susceptible to other policy objectives that some believe are better served by a more collaborative approach to business practices. For example, a number of competition authorities signalled a weakening of enforcement to allow for greater coordination in supply chains in response to supermarket shortages caused by the COVID-19 crisis.44 Other pressures include moves to relax cartel 37 P. Whelan, ‘Morality and Its Restraining Influence on European Antitrust Criminalisation’ (2009) 12(1) Trinity College Law Review 40. See also B. Fisse, ‘Reconstructing Corporate Criminal Law: Deterrence, Retribution, Fault and Sanctions’ (1983) 56(6) Southern California Law Review 1141. 38 R v Stringer and Dean, Southwark Crown Court, 24 June 2015 (unreported); see also: P. Whelan, ‘Antitrust Criminalization as a Legitimate Deterrent’, in T. Tóth (ed), The Cambridge Handbook of Competition Law Sanctions (Cambridge University Press, 2022), 111–12. 39 For a discussion of the civil versus criminal issue, see Coffee (n 36) 193. 40 See, e.g., A. Menarini Diagnostics srl v Italy, Application No. 43509/08, Judgment, 27 September 2011. 41 US Department of Justice Assistant Attorney General, JI Klein, Speech at Spring Meeting of the ABA Antitrust Division, 6 April 2006. 42 Harding and Edwards (n 29) 5. 43 C. Parker, ‘The “Compliance” Trap: The Moral Message in Responsive Regulatory Enforcement’ (2006) 40 Law & Society Review 591. 44 See P. Ormosi and A. Stephan, ‘The Dangers of Allowing Greater Coordination between Competitors during the COVID-19 Crisis’ (2020) 8(2) Journal of Antitrust Enforcement 299; and M.P. Schinkel and A. d’Ailly, ‘Corona Crisis Cartels: Sense and Sensibility’, Amsterdam Centre for Law & Economics Working Paper No. 2020-03, 12 June 2020.

80  Research handbook on cartels enforcement to promote environmental sustainability, through the greater use of carve-outs or comfort guidance.45 Such initiatives carry a significant danger that competitors will develop new, possibly tacit, ways of suppressing competition and exerting monopoly power. Without a strong moral basis for prohibiting cartels, the growth of such pressures could lead to a significantly more permissive regulatory landscape. The third reason why morality is important concerns compliance with the law. If the ultimate aim of cartel law is to prevent cartel harm, then enforcement can only go some way in preventing future infringements. The idea that fines alone will achieve deterrence, for example, is fundamentally flawed because it assumes cartelists engage in (and have the information they need to weigh the relative costs and benefits of) cartel behaviour. Parker points out that ‘[s]imple deterrence will often fail to produce compliance commitment because it does not directly address business perceptions of the morality of regulated behaviour – it merely puts a price on noncompliance’.46 Also, truly optimal fines (if deterrence theory is to be closely followed) would likely bankrupt most firms and so the general level of cartel fines is likely to be less than the overcharges achieved by successful cartels.47 So enforcement needs to be backed by norms (whether derived from ethics, morality, education or a combination of all three) to encourage desistance and compliance. 48 As Hodges and others point out, a shared set of values, ‘based on an innate ability to differentiate between what is morally right and wrong’, is more effective than simply threats of deterrence-based sanctions.49 A strong sense of right and wrong (which we might describe as a ‘moral compass’) is also important to the efforts of in-house compliance teams working to ensure employees either avoid cartel conduct or are able to recognize and report it.50 It is important to recognize that industry-wide cultures of cartelization are likely to have given way (at least in relation to larger firms and markets) to infringements involving smaller groups of individuals operating within particular divisions and subsidiaries.

See for example European Commission, ‘Statement on ACM Public Consultation on Sustainability Guidelines’, 9 July 2020, https://​ec​.europa​.eu/​competition​-policy/​european​-competition​-network/​ documents​_en [accessed 24 February 2022]; Autoriteit Consument & Markt, Guidelines: Sustainability Agreements: Opportunities within Competition Law, 9 July 2020, www​.acm​.nl/​sites/​default/​files/​ documents/​2020​-07/​sustainability​-agreements​%5B1​%5D​.pdf [accessed 24 February 2022]. 46 Parker (n 43) 592. 47 See, for example, W. Wils, The Optimal Enforcement of EC Antitrust Law: Essays in Law and Economics (Kluwer Law International, 2002) 6.5.2; and P. Buccirossi and G. Spagnolo, ‘Optimal Fines in the Era of Whistleblowers: Should Price Fixers Still Go to Prison?’, in V. Ghosal and J. Stennek (eds), The Political Economy of Antitrust (Elsevier, 2007). 48 M. Stucke, ‘Am I a Price Fixer? A Behavioural Economics Analysis of Cartels’, in Beaton-Wells and Ezrachi (n 29). 49 C. Hodges, ‘Science-Based Regulation in Financial Services: From Deterrence to Culture’, Oxford Legal Studies Research Paper No 19/2020, 2020. 50 See for example S.S. Simpson and M. Rorie, ‘Motivating Compliance: Economic and Material Motives for Compliance’ in C. Parker and V. Lehmann Nielsen (eds), Explaining Compliance: Business Responses to Regulation (Edward Elgar Publishing, 2011) 69. 45

The morality of cartel activity  81

III.

ARE CARTELS MORALLY WRONGFUL?

Having established that morality is a subjective and fluid concept that is of great importance to cartel enforcement despite its morally ambiguous underpinnings, we now turn to the question of whether cartels can in fact be viewed as morally wrongful. In this section we argue that there are various pull and push factors that determine whether particular cartel conduct is likely to be viewed as morally offensive or morally neutral. Some readers may be surprised to know that the issue of cartels and moral wrongfulness pre-dates the discipline of economics. There is evidence dating back to ancient Athens of severe punishment for merchants acting to drive up prices by restricting the supply of food,51 and there are also examples of this throughout the Middle Ages, some of which were driven by the fear of popular revolt and civil unrest at the monopoly prices for food and other key goods.52 For much of the pre-industrialization era, production of key goods and commodities in Europe was controlled by guilds, or associations of craftsmen and merchants. These guilds often had a legal monopoly over their trade in a particular geographical region and acted in their collective interests, although there were merchants who were not guild members who typically could not command the same price.53 In a remarkable empirical study on guilds that includes data from the fourteenth to the nineteenth centuries, Ogilvie estimates that they achieved price overcharges and output restrictions that are comparable and possibly greater than those estimated in modern cartel studies.54 She also notes evidence that ‘customers complained about high prices and supply restrictions that made their daily lives difficult’ and that many would travel to other geographical locations to buy shoes and other products from rural suppliers who were not members of the urban guilds.55 However, in the absence of established and reliable supply chains, guilds played an important role in organizing pre-industrialization production and in some cases spurring innovation.56 The more recent history of cartels has been a journey from ambivalence to strict prohibition and increasing criminalization.57 With this historical context in mind, we turn to the survey research available into what members of the public think about cartel practices today. This is a growing body of literature that has so far shown fairly similar results between jurisdictions, despite differences in language and culture. As summarized in Table 4.1, all of it suggests that members of the public – without being provided with any information as to the effects of cartels or their legality – show a strong understanding that cartel practices lead to higher prices and should therefore be prohibited. While these results show a broad consensus in favour of cartel prohibition, they do not necessarily reflect how wrong cartel conduct is perceived to be or the level of punishment (if any) that is appropriate. When asked about punishment, respondents tend to gravitate towards fines (whether corporate or against individuals) and there is particular support for naming and L. Kotsiris, ‘An Antitrust Case in Ancient Greek Law’ (1988) 22 International Law 451. W. Letwin, ‘The English Common Law Concerning Monopolies’ (1954) 21 University of Chicago Law Review 355. 53 S. Ogilvie, The European Guilds: An Economic Analysis (Princeton University Press, 2019). 54 ibid 218–31. 55 ibid 219. 56 S.R. Epstein and M. Prak, Guilds, Innovation and the European Economy, 1400–1800 (Cambridge University Press, 2010). 57 On this history, see Chapter 3 in this volume. 51

52

82  Research handbook on cartels Table 4.1

Proportion of respondents who felt cartels are harmful and should be prohibited

United Kingdom

Germany

Italy

United States

France

Swedena

Australia

79%

78%

73%

66%

59%

80%

73%

Note: a The question in the Swedish survey was focused on harm and there was a separate question that presented different options for punishment. Source: A. Stephan, ‘An Empirical Evaluation of the Normative Justifications for Cartel Criminalisation’ (2017) 37(4) Legal Studies 621; Stephan (n 24); E. Combe and C. Monnier-Schlumberger, ‘Public Opinion on Cartels and Competition Policy in France: Analysis and Implications’ (2019) 42(3) World Competition 335; C Beaton-Wells et al, ‘The Cartel Project: Report on a Survey of the Australian Public Regarding Anti-Cartel Law and Enforcement’, University of Melbourne Legal Studies Research Paper No.519, 2011; Swedish Competition Authority Report, Kartellbrottslighetens moraliska ekonomi: Av Tage Alalehto och Daniel Larsson på uppdrag av Konkurrensverket, Uppdragsforskiningsrapport 2020:3 (in Swedish).

Table 4.2

Do members of the public support the imprisonment of those responsible?

United Kingdom

Germany

Italy

United States

France

Sweden

Australia

27%

28%

26%

36%

25%

17%

20%

shaming those responsible. Perhaps the strongest indicator of the level of moral opprobrium is in relation to whether the individuals responsible should be imprisoned. All of the said surveys asked this question, but, as Table 4.2 shows, only a minority of respondents in each jurisdiction felt this was an appropriate punishment. Further research is needed to understand why this is the case. It could, for example, be that business misconduct is associated more with a corporate entity rather than individual decisions made by its employees and agents. The higher support in the US could reflect the greater level of white-collar crime enforcement more generally there. Whatever the reasons, the results suggest that cartels do not attract the same level of moral opprobrium as traditional forms of crime. Education about the nature and effects of cartel conduct goes a long way in shaping perceptions and judgements as to how right or wrong the conduct is. In 2006, the Centre for Competition Policy conducted an experiment with members of the public as part of the BA Festival of Science.58 Around 50 participants were given a simple explanation of what a cartel was and asked if they felt it was deserving of punishment. They were then given a brief talk describing how cartels operate and shown excerpts from the FBI’s secret covert filming of meetings of the Lysine cartel in the 1990s (more recently turned into the 2009 film The Informant, starring Matt Damon). The effect of seeing competitors meeting in secret to carve up the market and increase prices, while also mocking their customers, was quite stark. The participants were asked the same set of questions at the end of the experiment, and it was apparent that attitudes to the conduct were transformed, despite no information being revealed about whether cartels were illegal or about the sanctions imposed on them. What the participants were primarily moved by was the nature of the conduct they were observing. Indeed, so powerful is the Lysine video that this author has stopped showing it to students enrolled on the competition law class, because it pushes them to be overwhelmingly pro-enforcement and therefore stifles balanced debate. The lysine video captured the era immediately preceding an explosion of cartel enforcement and continued proliferation of competition law internation58 ‘Public Perceptions of Collusion’, 2006 BA Festival of Science, University of East Anglia, 8 September 2006.

The morality of cartel activity  83 ally. Other covert recordings like it do exist (for example in the Marine Hoses case), but none are publicly available because cartel cases tend to be settled by businesses. Indeed, the effect of a greater reliance on streamlined procedures such as settlement is that less information enters the public domain about how cartels operate. It may also be less likely that businesses would risk holding physical meetings because of the risk of detection. Whether of the nature of the cartel conduct or of the harm that it causes, moral opprobrium is facilitated by knowledge. Survey work undertaken by the Competition and Markets Authority (‘CMA’) in 2014 suggests there is a striking level of ignorance among members of the UK business community. Almost half of respondents said they had never heard of competition law and a fifth admitted having recently made direct contact with their competitors.59 Similarly, although a high proportion of respondents in surveys have recognized cartels are harmful and should be prohibited, significantly lower proportions (53 per cent in the UK and Italy) actually knew that such practices were illegal. There are various explanations for this stark gap in awareness. The main issue is dissemination of enforcement outcomes, as media reporting tends to be limited to the business press, except on occasion when the businesses involved are household brands recognized by the wider public.60 There is also a conspicuous absence of a clear ‘victim’ in cartel cases, as the main harm (the overcharge) tends to be borne by other businesses, or passed down to a very large number of consumers with little in the way of a newsworthy focus. In relation to businesspeople specifically, there is the question of what constitutes a ‘fair price’, as cartel laws (with their drive for lower prices and innovation) are often in conflict with marketing strategies taught in business schools that seek to increase profitability and in some cases find lawful ways to suppress competition.61 The effect of greater knowledge on the extent to which cartels are associated with moral opprobrium depends crucially on the characteristics of the conduct, the law and how it is enforced. To explore these factors, we separate them into two categories: pull factors, which make the cartel conduct more likely to be viewed as morally offensive, and push factors, which have precisely the opposite effect, causing observers to view the conduct as morally ambiguous or neutral. This relationship is illustrated in Figure 4.1. A.

Pull Factors

To assist us in identifying pull factors, we are able to draw on the extensive published research exploring normative justifications for the criminalization of cartel conduct. For the purposes of this chapter we make only a brief reference to these justifications, as they are discussed more fully in Chapter 20 below. Delinquency/Secrecy: Arguably one of the strongest pull factors is clear evidence that the cartel participants knew they were defying a clear legal prohibition. Harding describes this as a form of ‘moral delinquency’ that is heightened by the determination and effort that goes

59 IFF Research, Report: UK Business’ Understanding of Competition Law Prepared for CMA, 26 March 2015. 60 See A. Stephan, ‘Cartel Criminalisation: The Role of the Media in the “Battle for Hearts and Minds”’, in Beaton-Wells and Ezrachi (n 29). 61 For a discussion of this, see S. Ennis, ‘Business Strategy and Antitrust Compliance’, in Riley et al (n 26).

84  Research handbook on cartels

Figure 4.1

The relationship between cartel conduct and moral opprobrium

into concealing what they are doing and obstructing enforcement by competition authorities.62 Indeed, the participants descend into a ‘spiral of delinquency’, as the greater their efforts to hide their behaviour, the more morally offensive it becomes. This is consistent with Green’s Moral Theory of White Collar Crime,63 where he identifies both disobedience and cheating as strong elements of moral delinquency, and with traditional crime theory, where what matters is the nature of the defendant’s ‘choice’ to engage in the morally reprehensible conduct.64 Very little cartel conduct occurs openly and it is interesting to note that cartelists were reluctant to talk about their conduct even during periods when it was lawful. De Jong’s description of the research efforts of Friedrich Kleinwächter, published in one of the earliest books on cartels, Die Kartelle (1883), should be emphasised here: little was known about their existence, their names, locations, activities or the persons that guided them […] He therefore organized a survey among the leading industrialists in Austria and Germany but was seriously disappointed: the information given was scarce, respondents were reticent with respect to pertinent facts and processes and nearly all insisted on secrecy.65

Deception/Cheating: Another strong pull factor is whether cartel behaviour involves a clear deception. This is most evident in bid-rigging, where the dishonest submission of false bids for C. Harding, ‘The Anti-Cartel Enforcement Industry: Criminological Perspectives on Cartel Criminalisation’ (2006) 14 Critical Criminology 181. 63 See Green (n 35). 64 ibid 22. 65 H.W. de Jong, ‘Market Theory in the Low Countries’, in H.W. de Jong and W.G. Shepherd (eds), Pioneers of Industrial Organization: How the Economics of Competition and Monopoly Took Shape (Edward Elgar Publishing, 2007) 62. 62

The morality of cartel activity  85 financial gain in an explicitly competitive process amounts to a clear fraud. The same is also true where there is evidence of false statements made to customers (for example, that prices are going up due to rising costs), where there is blatant market sharing (refusing to sell to each other’s customers) or where steps are taken to ensure the behaviour goes unnoticed by senior management or a parent firm. As Whelan explains in his book, this factor becomes problematic where cartelists do not expressly state they have colluded, but nonetheless fail to disclose that they have.66 Would most people consider that failing to be a deception? To help us answer this question we can look again at the relevant public survey findings. In the UK, Germany and Italy, two thirds of respondents (in the absence of any knowledge of cartels and before they were asked cartel-specific questions) indicated that they ‘expect each business they buy from to have set their prices independently of each other’.67 If this is the regular expectation of how business is conducted by most in society, then arguably a failure to disclose the existence of a cartel also amounts to a deception. These findings support various arguments made by Whelan, Wardhaugh and Beaton-Wells that cartel conduct can be conceptualized as theft or deception.68 Social harmfulness: This pull factor rests on the value that society places on the free market and the competitive process. Cartels can be characterized as an attack on this valued institution,69 or as ‘subverting the competitive process’.70 This can amount to a dishonest exploitation, creating strong parallels with other forms of corporate crime, such as insider trading and fraud.71 By harming one of our most fundamental interests, the harm is such that it deserves moral opprobrium and protection from the criminal law.72 This pull factor can be served by efforts the enforcement process can make to communicate precisely how the cartel is likely to have had a negative impact on the market and in particular on consumers and other businesses.73 B.

Push Factors

The relevant push factors draw partly on the earlier discussion in this chapter on whether morality is in fact relevant to cartel conduct. They consist of anything that blunts or muddies the idea that cartels amount to deliberate and harmful wrongdoing.

Whelan (n 25) 101–05. A. Stephan, ‘An Empirical Evaluation of the Normative Justifications for Cartel Criminalisation’ (2017) 37(4) Legal Studies 621, 638. 68 C. Beaton-Wells, ‘Capturing the Criminality of Hard Core Cartels: The Australian Proposal’ (2007) 31 Melbourne University Law Review 675; B. Wardhaugh, Cartels, Markets and Crime: A Normative Justification for the Criminalisation of Economic Collusion (Cambridge University Press, 2014) Chapter 1; and Whelan (n 25), Chapter 4. 69 See generally Wardhaugh (n 68). 70 A MacCulloch, ‘The Cartel Offence: Defining an Appropriate “Moral Space”’ (2015) 8(1) European Competition Journal 73. 71 Williams (n 31) 300–01, with reference to R v Hinks [2000] UKHL 53, [2001] 2 AC 241 (HL) and Fraud Act 2006, s.4. See also C. Beaton-Wells and B. Fisse, Australian Cartel Regulation: Law, Policy and Practice in an International Context (Cambridge University Press, 2011), Chapter 2. 72 Wardhaugh (n 68). 73 See A. MacCulloch, ‘The “Public” Wrong of Cartels and the Article 101 TFEU “Object Box”’ (2020) 65(3) Antitrust Bulletin 361; and A. Stephan, ‘Why Morality Should Be Excluded from the Cartel Criminalisation Debate’ (2012) 3(2) New Journal of European Criminal Law 127. 66 67

86  Research handbook on cartels Crisis: While greed is undeniably a key feature of cartel arrangements, many such arrangements are driven more precisely by crisis in the industry or in the wider economy.74 In economic terms the effects of these cartels is the same – prices are higher and output lower than would otherwise be the case. However, crisis creates space for a very different interpretation of the nature and motivation of cartel conduct, as is illustrated by the apparent weakening in enforcement in response to the COVID-19 crisis. Kovacic and Shapiro note how even the prohibition of hard-core cartel infringements wavered in the United States during the depths of the Great Depression.75 In the only UK criminal cartel case to be heard before a jury, two defendants were acquitted following suggestions of crisis that transpired from the Crown’s evidence. In cross-examining a witness, one of the defendant’s barristers described a situation of ‘ruinous competition’ in the market immediately preceding the cartel arrangement and the prospect of one of the three competitors in the industry becoming insolvent. A witness called to give evidence against the defendants described them as ‘heroes’ because their actions in forming the cartel had helped to safeguard jobs and the future of the company.76 The defendants were tried under the original UK cartel offence (which required the behaviour to have been dishonest), but after two weeks of evidence the jury took just two hours to acquit both defendants, even though a third individual had pled guilty and given evidence against his alleged co-conspirators. Ignorance and ‘atypical’ cartels: Survey results and interview evidence suggest there is still a real possibility of some cartels being formed out of ignorance.77 The problem here seems to be one of public awareness and education, but many will question whether cartel conduct can really attract moral opprobrium if some of those responsible do not even realize they are doing something illegal. This is especially a problem in relation to ‘atypical’ cartels or those at the fringes of cartel prohibitions, which display few if any of the pull factors described above.78 In particular, enforcement cases involving ‘atypical’ cartels may display neither the outcome nor the conduct that attracts moral opprobrium.79 Exemptions and carve-outs: Whether in pursuit of environmental sustainability or to deal with a crisis, exemptions and carve-outs from cartel laws will always be a strong push factor. Talk to the average person on the street about cartel enforcement and they are likely to ask why OPEC (Organization of the Petroleum Exporting Countries) is allowed to be in existence, with governments often engaging in precisely the sorts of conduct cartel law seeks to prohibit in relation to private business. It is interesting to note that although under Article 101(3) of the Treaty on the Functioning of the European Union exceptions on efficiency grounds are in principle very narrow and focused on economics arguments, a recent study by Brook suggests that this may be far from the case in practice.80

74 A. Stephan, ‘Price Fixing during a Recession: Implications of an Economic Downturn for Cartels and Enforcement’ (2012) 35(3) World Competition 511. 75 W. Kovacic and C. Shapiro, ‘Antitrust Policy: A Century of Economic and Legal Thinking’ (2000) 14 Journal of Economic Perspectives 43, 46. 76 R v Stringer and Dean, Southwark Crown Court, 24 June 2015 (unreported). 77 Stephan (2017) (n 67). 78 See Stephan and Hviid (n 15). 79 Harding and Edwards (n 29) 79. 80 O. Brook, ‘Priority Setting as a Double-Edged Sword: How Modernization Strengthened the Role of Public Policy (2020) 16(4) Journal of Competition Law and Economics 435.

The morality of cartel activity  87 Lack of individual sanctions and overreliance on leniency: These factors, among other features of cartel enforcement, reinforce a sense that cartels are a regulatory problem and therefore void of moral opprobrium. The failure to adequately punish the individuals responsible for cartels and a high frequency of cartelists being allowed to escape punishment altogether both serve to strengthen the idea that enforcement is simply a cost of doing business in a utilitarian regulatory landscape in which cheating – far from being the basis for moral opprobrium for the illegal behaviour – is actively rewarded by the law. As Gilchrist points out in discussing the broader problem of ignoring the individual within the sphere of punishing corporate misbehaviour, ‘[i]t sends a dangerous message of tolerance or even affirmance for their conduct’.81

IV.

CONCLUDING REMARKS

Morality concerns a person’s sense of right or wrong according to their conscience. While there are some behaviours that all or most in society would find morally offensive, morality is otherwise a highly subjective notion that is capable of changing over time. The law regulates many aspects of our lives and the behaviour it seeks to control does not need to coincide with a common sense of morality – especially as we live in a secular and pluralistic society. In this context, cartel law appears to exist in a landscape entirely void of morality. It is underpinned by neoclassical economic theory that treats morality (or at least its manifestations in terms of stigma or loss of reputation) as a cost and is so wide in its scope and design that it matters not whether the arrangement was intentional, whether it was implemented or had any actual harmful effect on the market. For these reasons (among others), the trend to criminalize cartel conduct has been strongly criticized as an illegitimate and damaging misuse of criminal law and of the label ‘criminal’. Nevertheless, morality is of great relevance and importance to the effectiveness of cartel enforcement. The crisis in legitimacy that exists within the criminalization debate extends to all cartel enforcement, by virtue of the punitive (and therefore criminal) nature of its investigative powers and sanctions. Corporate fines are not aimed at reversing any unjust enrichment, but are designed to punish and deter. Without strong moral underpinning reflected in public support, there is a danger that political support for cartel enforcement could be eroded by lobbying. This is already happening to some extent in how competition authorities responded to the COVID-19 crisis and the growing pressures to allow exemptions and carve-outs for cooperation that further environmental objectives such as sustainability. This is not to suggest that sustainability should not be a priority, but there are serious questions about how we ensure competitors’ cooperation is focused on that goal without unwanted side-effects, such as reduced post-cooperation competition through tacit or explicit collusive outcomes. Other challenges to cartel enforcement may come from arguments that collective solutions are needed to problems of distributional fairness and from the automation of pricing through the use of algorithms. This chapter explored what determines whether cartel conduct is deemed morally offensive. The central proposition is that as a subjective and fluid concept, morality will be influenced by the characteristics of particular cartel conduct and the circumstances surrounding the 81 G.M. Gilchrist, ‘Individual Accountability for Corporate Crime’ (2018) 34(2) Georgia State University Law Review 335, 387.

88  Research handbook on cartels behaviour. As such, we can identify various factors that either pull the cartel conduct towards moral opprobrium or push it away and make it morally ambiguous or neutral. This provides a useful guide to policy makers in their design and enforcement of cartel laws. It highlights the particular importance of case selection, the need to publicize ‘pull factors’ that arise from enforcement action (including anything that speaks to the harm caused), the need for individual sanctions that recognize the role of individuals (whether through criminalization or civil individual sanctions such as director disqualification) and, above all, the need for continued public education and business advocacy. There is already some evidence that attitudes to cartels may be hardening, as support for the imprisonment of cartelists increased from 11 per cent to 27 per cent in surveys conducted in the UK in 2007 and 2014.82 The need to continue building a moral base for cartel enforcement is made all the more important by the apparent drop in cartel cases being detected by competition authorities internationally. This could reflect increases in the level of compliance; but it could also reflect the changing nature of cartel behaviour and the strengthening of efforts to avoid detection.83

See Stephan (2008) (n 24); and Stephan (2017) (n 67). Harding and Edwards (n 29) 215–18.

82 83

PART II SUBSTANTIVE ISSUES

5. The legal concept of a cartel Okeoghene Odudu

I. INTRODUCTION The word ‘cartel’ does not have a legal definition but is used to describe conduct seen as inherently objectionable.1 Cartels are sometimes described as ‘manifest infringements [of competition law] which it is almost always impossible to exempt’2 and have memorably been called ‘cancers on the open market economy’.3 A central feature of the legal analysis of cartels in Europe is that courts do not ask competition authorities and complainants to demonstrate that the participants have caused (or are likely to cause) harm or obtain and exploit market power.4 The economic consequences of the action are ignored and condemnation is based ‘solely on whether certain conduct took place’.5 Since the conduct is inherently objectionable, and very high fines (and sometimes imprisonment) are likely to follow, those engaged in cartel conduct take great steps to conceal its existence. This explains why many of the ‘controversies in cartel cases relate, first and foremost, to whether the practice has actually taken place (this is the “fact-intensive” inquiry)’.6 A consequence of the inherently objectionable nature of the conduct is that the ‘law’ and disputes around cartels, at least since the promulgation of the OECD’s 1998 Recommendations on Cartels, are primarily focused on establishing the existence of the maligned conduct (as distinct from the effects such conduct might have). The ‘law’ on cartels is somewhat elusive, given that cartels are merely a species of horizontal agreement treated with enhanced opprobrium. This chapter seeks to identify what marks out this species of horizontal agreement for special condemnation and the implications of being marked out as such. The chapter outlines two problems associated with the legal concept of a cartel. The first problem concerns the definition of inherently objectionable conduct. What conduct is identified as such, and on what basis? This first problem has two strands. In Section II it is explained that there is a demand that certain decisions are taken independently. In Section III we specify the type of decisions that must be taken independently. In Section IV we explore consequential and deontological reasons why it might be thought objectionable not to take the specified decisions independently. The second problem concerns proving firms 1 DTI, A World Class Competition Regime (CM 5233), para 7.26. See also N. Dunne, ‘Characterizing Hard Core Cartels under Article 101 TFEU’ (2020) 65 The Antitrust Bulletin 376. 2 Commission of the European Communities Tenth Report on Competition Policy, para 115. 3 N. Kroes, ‘Declaring on the Crackdown: Recent Developments in the European Commission’s Campaign Against Cartels’ (SPEECH/06/595). 4 E. Gellhorn, W. Kovacic and S. Calkins, Antitrust Law and Economics in a Nutshell (5th edn, West, 2004) 191. 5 ibid 200. See also A. Gavil, W. Kovacic and J. Baker, Antitrust Law in Perspective: Cases, Concepts, and Problems in Competition Policy (West, 2002) 215–16 and W. Kovacic, R. Marshall, L. Marx and H. White, ‘Plus Factors and Agreement in Antitrust Law’ (2011) 110 Michigan Law Review 393, 404. 6 P. Ibáñez Colomo, The Shaping of EU Competition Law (Cambridge University Press, 2018) 17.

90

The legal concept of a cartel  91 have acted other than independently. In Section V the task of demonstrating the absence of independent action is set out. The first part is to specify what it means to act independently. The second part is to evidence the absence of independence, which may be done by direct or indirect evidence. Indirect evidence may be communication-based or economics-based.

II. INDEPENDENCE The standard account of cartels begins with the idea of concertation or ‘collusion’.7 In Monsanto v Spray-Rite, the US Supreme Court observed: The correct standard is that there must be evidence that tends to exclude the possibility of independent action by the [parties]. That is, there must be direct or circumstantial evidence that reasonably tends to prove that [the parties] had a conscious commitment to a common scheme designed to achieve an unlawful objective.8

Cartel conduct arises only when, among firms competing in a market, there is a ‘joint intention to conduct themselves on the market in a specific way’.9 While it is common to consider how the requirement of ‘conscious commitment to a common scheme’ is developed and applied, here we focus on the alternative offered, which is the requirement of ‘independent action’.10 In Suiker Unie the Court of Justice explained that competition law imposes an obligation of independence, which requires ‘each economic operator [to] determine independently the policy which it intends to adopt on the […] market including the choice of persons and undertakings to which he makes offers or sells’.11 The obligation of independence exists in relation to competing entities rather than non-competing entities. It is only ‘contrary to the rules on competition contained in the Treaty for a producer to cooperate with his competitors, in any way whatsoever’.12 The first element 7 The seminal papers include: D. Turner, ‘The Definition of Agreement under the Sherman Act: Conscious Parallelism and Refusals to Deal’ (1962) 75 Harv L Rev 655–706; J. Joshua, ‘Proof in Contested EEC Competition Cases: A Comparison with the Rules of Evidence in Common Law’ (1987) 12 European Law Review 315; M. Guerrin and G. Kyriazis, ‘Cartels: Proof and Procedural Issues’ (1992) 16 Fordham International Law Journal 268; G. Hay, ‘Horizontal Agreements: Concept and Proof’ (2006) 51 Antitrust Bulletin 877; Kovacic et al (n 5); V. Korah, ‘Concerted Practices’ (1973) 36 Modern Law Review 220; and G. Hay, ‘Oligopoly’ (1982) 67 Cornell Law Rev 439. 8 Monsanto Co. v Spray-Rite Svc. Corp. (1984) 465 US 752, 768 (emphasis added). 9 Case T-41/96 Bayer AG v Commission [2000] ECR II-3383, para 67, citing Case 41/69 ACF Chemiefarma v Commission [1970] ECR 661, para 112, Joined Cases 209/78 to 215/78 and 218/78 Van Landewyck and others v Commission [1980] ECR 3125, para 86, and Case T-7/89 Hercules Chemicals v Commission [1991] ECR II-1711, para 256. See also Case T-1/89 Rhône-Poulenc v Commission [1991] ECR II-867, para 120; Case T-2/89 Petrofina v Commission [1991] ECR II-1087, para 211;Case T-7/89 SA Hercules Chemicals NV v Commission [1991] ECR II-1711, para 256; and Case IV/F-3/33.708 – British Sugar plc [1999] OJ L76/1, para 66. 10 Monsanto Co. v Spray-Rite Svc. Corp. (1984) 465 US 752, 768. See also Case 48/69 ICI v Commission (Dyestuffs) [1972] ECR 619, para 101 (emphasis added). Cp Joined Cases 40 etc./73 Coöperatieve Vereniging ‘Suiker Unie’ UA and others v Commission [1975] ECR 1663, para 180. 11 Joined Cases 40 etc./73 Coöperatieve Vereniging ‘Suiker Unie’ UA and others v Commission [1975] ECR 1663, para 173. 12 Case 48/69 Imperial Chemical Industries v Commission [1972] ECR 619, para 118 (emphasis added).

92  Research handbook on cartels of cartel conduct is therefore that a relevant decision is not taken independently of a competitor. In the next section we set out which decisions must be taken independently of competitors.

III.

PROSCRIBED CONDUCT

Central to the understanding of cartel conduct is an idea that, in a market context, it is inherently objectionable if certain decisions are taken other than independently.13 The OECD’s 1998 Recommendation concerning Effective Action against Hard Core Cartels identifies as inherently objectionable ‘arrangements by actual or potential competitors to agree on prices, make rigged bids (collusive tenders), establish output restrictions or quotas, or share or divide markets by, for example, allocating customers, suppliers, territories, or lines of commerce’.14 In its Directive concerning Actions for Damages under National Law, ‘cartel’ refers to an agreement or concerted practice between two or more competitors aimed at coordinating their competitive behaviour on the market or influencing the relevant parameters of competition, through practices such as, but not limited to, the fixing or coordination of purchase or selling prices or other trading conditions, including in relation to intellectual property rights, the allocation of production or sales quotas, the sharing of markets and customers, including bid-rigging, restrictions of imports or exports or anti-competitive actions against other competitors.15

Similarly, the Commission’s leniency notice understands the requirement of independent action to centre on ‘the fixing of purchase or selling prices or other trading conditions, the allocation of production or sales quotas, the sharing of markets including bid-rigging, restrictions of imports or exports and/or anti-competitive actions against other competitors’.16 If the price to be charged, the customers or territories to be served and the price, terms and conditions submitted in an invitation to tender are not determined independently there is general consensus that it is inherently objectionable.17

13 D. Encaoua, ‘Kaplow, Louis: Competition Policy and Price Fixing’ (2014) 113 Journal of Economics 205, 207; and Dunne (n 1), 388–91. 14 See the now revised Recommendation of the Council concerning Effective Action against Hard Core Cartels (OECD/LEGAL/0452) https://​legalinstruments​.oecd​.org/​en/​instruments/​OECD​-LEGAL​ -0452 [accessed 13 July 2022]. Cp Gellhorn et al (n 4) 191. 15 Article 2(14) of Directive 2014/104/EU of the European Parliament and of the Council of 26 November 2014 on certain rules governing actions for damages under national law for infringements of the competition law provisions of the Member States and of the European Union (the ‘Damages Directive’). 16 Commission Notice on Immunity from Fines and Reduction of Fines in Cartel Cases [2006] OJ C298/11, para 1. 17 See International Competition Network, Defining Hard Core Cartel Conduct: Effective Institutions Effective Penalties, 10; and D. Baker, ‘The Use of Criminal Law Remedies to Deter and Punish Cartels and Bid-Rigging’ (2001) 69 George Washington Law Review 693. Though there is consensus, Kaplow raises the concern that ‘[c]onsensus has bred complacency’: L. Kaplow, Competition Policy and Price Fixing (Princeton University Press, 2013) xiii.

The legal concept of a cartel  93 A. Price Article 101(1)(a) of the Treaty on the Functioning of the European Union (‘TFEU’) identifies the fixing of purchase or selling prices as an action to be performed independently. In Dyestuffs the Court of Justice ruled that ‘it is contrary to the rules on competition contained in the Treaty for a producer to co-operate with his competitors, in any way whatsoever, in order to determine a co-ordinated course of action relating to a price increase and to ensure its success’.18 The expectation is that minimum prices, the range within prices are to be set, an amount or percentage price increase, discounts or allowances, transport charges, payments for additional services and any credit terms are established independently.19 And it is inherently objectionable if they are established other than independently. In Residential Estate Agency Services nine estate agents were condemned, having been found not to have determined independently their minimum commission rates for the sale of residential properties.20 In Passenger Fuel Surcharges two airlines were found not to have determined independently how to respond to increasing oil prices, instead agreeing to pass the cost on to the consumer.21 Arguments that price in the market is unaffected by the conduct have not succeeded because the objection is the failure to act independently, rather than the impact that such failure may have in the market.22 B.

Market Sharing

Article 101(1)(c) of the TFEU and many competition law jurisdictions require expressly that the territories and customers a firm will serve are determined independently. Not to do so is a – if not the – classic competition law infringement. Indeed, there are many infringement decisions that involve market-sharing cartels. For example, the European Commission imposed fines of €7 million on both Solvay and ICI in relation to their agreement that Solvay stay out of the United Kingdom and Ireland and ICI stay out of markets in continental Europe.23 The CMA condemned a market-sharing arrangement whereby Arriva agreed to withdraw from bus routes operating between Leeds and Holt Park in exchange for First Group withdrawing

Case 48/69 Imperial Chemical Industries v Commission [1972] ECR 619, para 118. Office of Fair Trading, Agreements and Concerted Practices, OFT401, paras 3.5–3.7. 20 Case 50235 Residential Estate Agency Services, Decision of the Competition and Markets Authority, 31 May 2017; and Case 50543 Residential Estate Agency Services in Berkshire, Decision of the Competition and Markets Authority, 17 December 2019. 21 CE/7691-06 Passenger Fuel Surcharges, Decision of the Office of Fair Trading, 19 April 2012. 22 In European Producers of Beams [1994] OJ L116/1, upheld in Case T-141/94 Thyssen Stahl AG v Commission [1999] ECR II-347, paras 254–56, the Commission and the General Court reject economic evidence demonstrating that market prices were no higher than could have been expected in normal conditions of competition. In Zinc Phosphate [2003] OJ L153/1, para 216, the Commission considered it irrelevant that the impact of the cartel is not what the cartel members expect. 23 Soda-Ash-Solvay/ICI [1991] OJ L152/1, annulled on procedural grounds in Cases T-30/91, T-31/91 and T-32/91 Solvay SA v Commission [2009] ECR II-4781 and Case T-36/91 ICI v Commission [1995] ECR II-1847. 18 19

94  Research handbook on cartels from bus routes operating between Leeds and Halifax.24 Firms should not agree to restrict their activities to their ‘home’ market or share geographic markets based on agreed quotas.25 C. Bid-Rigging It is expected that prospective suppliers independently determine and submit their best offer for the provision of goods or services specified in an invitation to tender. An objection would be raised where firms agree to take it in turns to submit the lowest bid (bid rotation); where one or more firms agree not to bid, or to withdraw their bids (bid suppression); and where bidders arrange for one or more of their competitors to submit an artificially high bid, distorting the procurer’s impression of the competitive price (cover bidding).26 The lack of independence has been identified and condemned by the European Commission since its early sugar investigation.27 The European Commission discovered that over a long period of time bids were not submitted independently in Gas Insulated Switchgear. With European firms having agreed not to win contracts in Japan, Japanese firms reciprocated by not trying to win contracts within the EU market.28 In Power Cables, producers of high voltage power cables were found to have allocated projects between themselves according to geographic region or customer, exchanging information to enable cover bidding to occur.29

IV.

THE BASIS OF THE INHERENT OBJECTION

A key difficulty with a legal conception of ‘cartels’ that captures inherently objectionable conduct is that the proscribed conduct is not always seen as inherently objectionable – context (if not effect) still seems to matter.30 For example, it is not objectionable for a number of competitors to fix the price they will jointly pay suppliers.31 Setting the maximum amount that

Case CP/1163-00 Market Sharing by Arriva plc and First Group plc, Decision of the Director General of Fair Trading No. CA98/9/2002, 30 January 2002. 25 Zinc Producer Group [1984] OJ L220/27; and Vitamins [2003] OJ L6/1. 26 See CMA advice for public sector procurers on bid rigging: www​.gov​.uk/​government/​publications/​ bid​-rigging​-advice​-for​-public​-sector​-procurers/​bid​-rigging​-advice​-for​-public​-sector​-procurers [accessed 13 July 2022] and Case AT.40098 Blocktrains [2015] OJ C351/5. 27 IV/26 918 European Sugar Industry [1973] OJ L140/17, upheld in Joined Cases 40 etc./73 Coöperatieve Vereniging ‘Suiker Unie’ UA and others v Commission [1975] ECR 1663. 28 Case COMP/F/38.899 Gas Insulated Switchgear C(2006) 6762 final, annulled in part on procedural grounds in T-122/07 to T-124/07 Siemens Österreich and VA Tech Transmission & Distribution v Commission [2011] ECR II-793. 29 Case AT.39610 Power Cables, C(2014) 2139 final. Cover-bidding was also a feature of Case 50366-1, Supply of Solid Fuel and Charcoal Products, Decision of the Competition and Markets Authority, 28 March 2018. 30 See Revised Version of 03/06/2015 Commission Staff Working Document Guidance on Restrictions of Competition ‘By Object’ for the Purpose of Defining which Agreements May Benefit from the De Minimis Notice SWD(2014) 198 final. See also C. Nagy, ‘EU Competition Law Devours Its Children: The Proliferation of Anti-Competitive Object and the Problem of False Positives’ (2021) 23 Cambridge Yearbook of European Legal Studies 290. 31 See, e.g., Guidelines on the Applicability of Article 101 of the Treaty on the Functioning of the European Union to Horizontal Cooperation Guidelines [2011] OJ C11/1), para 206. 24

The legal concept of a cartel  95 may be charged for a good or service is not seen as inherently objectionable.32 Nor is it objectionable to share markets when parties have engaged in joint development.33 The United States Export Trade Act 1918, commonly called the Webb–Pomerene Act, does not find conduct objectionable if it relates solely to the export trade.34 Further, during economic, environmental, social and political crises, otherwise objectionable conduct can be seen to be more palatable.35 Finally, some jurisdictions operate a de minimis doctrine and so consider that conduct is not objectionable if those involved have a low market share.36 For example, in Japan, ‘there are certain cases in which it is appropriate to allow SMEs to act in concert within a certain limit […] The Japanese Anti-Monopoly Act system has built into it an exemption system for certain activities undertaken by organisations made up of small scale enterprises.’37 Our initial account of cartel conduct is that there is an inherent objection to certain decisions being taken other than independently. It is however also seen that the context in which the decision is made seems to matter. Context matters if the need for independent decision is founded on consequential grounds, that is, that the conduct is harmful in economic terms. Condemnation does not automatically follow if either the economic harm does not (or is unlikely to) materialize in the specific context or the economic consequences are acceptable given the specific context. Context does not matter (or is relevant in different ways) if the need for independent decision is founded on deontological grounds, that is, that regardless of the economic harm, the conduct is normatively unappealing. It is possible simply to sketch out the two views below rather than give a comprehensive account of the merits of the consequential versus the deontological view, and it is not claimed that a single basis underpins the approach to cartel conduct in all jurisdictions.

Re Eurocheques [1985] OJ L35/43. In Re London Sugar Futures Market Ltd. [1985] OJ L369/25 it is said that agreeing that a good or service should not be provided free of charge can be viewed as permissible, so long as the price to be charged is set by each competing firm independently. 33 See ‘specialisation in the context of exploitation’ under Commission Regulation (EU) 1217/10, Article 5(b)(iii). See also COMP/C.2-37.398 Joint Selling of the Commercial Rights of the UEFA Champions League [2003] OJ L291/25; and Commission Press Release IP/03/1105 of 24 July 2003. 34 It should be noted that while ‘Associations under the Webb Pomerene Act exporting to the [EU] are not as such unlawful under [EU] law[, t]hey may, however, infringe Article [101] if they restrict competition within the common market, and are likely to affect trade between Member States to an appreciable extent’: IV/29.725 Wood Pulp [1985] OJ L85/1, para 123. See further E. Bond, ‘Trade Policy and Competition Policy: Conflict Vs. Mutual Support’, in M. Neumann and J. Weigand (eds), The International Handbook of Competition (Edward Elgar Publishing, 2004) 124–26. 35 T. Fisher, ‘Antitrust during National Emergencies: I & II’ (1942) 40 Michigan L. Rev. 969, 1161; D. Crane, ‘Did We Avoid Historical Failures of Antitrust Enforcement During the 2008-09 Financial Crisis?’ (2010) University of Michigan Law & Economics Working Papers 3; J. Fingleton, ‘Competition Policy in Troubled Times’ (Office of Fair Trading, 20 January 2009); and OECD Global Forum on Competition Policy Roundtable – Crisis Cartels (DAF/COMP/GF(2011)11, October 2011); and 36 ‘Background Note (by the German Delegation): Exemption for Horizontal Co-operation Agreements for Small and Medium-Sized Enterprises (SME) from General Cartel Ban’, in OECD, General Cartel Bans: Criteria for Exemption for Small and Medium-Sized Enterprises (1996) 6–17, 11 and 13; and ‘Canada: Criteria for Exemption for Small and Medium-Sized Enterprises from General Cartel Bans’, in OECD (n 36) 31–34. 37 ‘Japan’, in OECD (n 36) 35. 32

96  Research handbook on cartels A.

Consequential Objections

One reason why the proscribed conduct is objected to and therefore condemned may be that it causes (or is likely to cause) economic harm. For example, the independent setting of price is demanded because setting the price in this manner will ‘keep prices down to the lowest possible level’.38 The proscribed conduct is treated as such because the economic consequences are thought to be well known and reliably predictable – economic harm is ‘inevitable’.39 Objection to the proscribed conduct on consequentialist grounds can mean that the existence of the objectionable consequences is tested in each case. So, as an example, Advocate General Wahl writes that though ‘a horizontal agreement concerning the price of certain goods […] in general […] is highly harmful for competition, that conclusion is not inevitable where, for example, the undertakings concerned hold only a tiny share of the market concerned.’40 And Advocate General Bobek wishes to ensure that ‘there are no specific circumstances that may cast doubt on the presumed harmful nature of the agreement in question’.41 B.

Deontological Objections

A second approach is to consider that there are deontological objections to the proscribed conduct, and more recent scholarship pays attention to the inherent objection to cartel conduct.42 It is important to recognize that cartel conduct is exactly that – conduct: a way of behaving, rather than the consequences of any particular behaviour. An explanation given by Advocate General Kokott is that such conduct is comparable to the risk offences (Gefährdungsdelikte) known in criminal law: in most legal systems, a person who drives a vehicle when significantly under the influence of alcohol or drugs is liable to a criminal or administrative penalty, wholly irrespective of whether, in fact, he endangered another road user or was even responsible for an accident.43

What is said to be ‘wrong’ with cartel conduct is that it subverts a customers or consumers legitimate expectation of how a decision has been arrived at.44 Customers expect the relevant

Case 48/69 ICI v Commission (Dyestuffs) [1972] ECR 619, para 115. Case C-209/07 The Competition Authority v Beef Industry Development Society Ltd and Barry Brothers (Carrigmore) Meats Ltd [2008] ECR I-8637, AG Opinion, para 44; and Case C-228/18, Budapest Bank, ECLI:​EU:​C:​2019:​678, AG Opinion, para 40. 40 Case C‑67/13 Groupement des Cartes Bancaires v Commission, ECLI:​EU:​C:​2014:​1958, Opinion of AG Wahl, paras 42 and 81. 41 Case C-228/18, Budapest Bank, ECLI:​EU:​C:​2019:​678, Opinion of AG Bobek, paras 48–49 (emphasis added). 42 M. Stucke, ‘Morality and Antitrust’ [2006] Columbia Business Law Review 443. 43 Case C-8/08 T-Mobile Netherlands BV, KPN Mobile NV, Orange Nederland NV and Vodafone Libertel NV v Raad van bestuur van de Nederlandse Mededingingsautoriteit [2009] ECR I-04529, Opinion of AG Kokott, para 47. 44 A. Macculloch, ‘The Cartel Offence: Defining an Appropriate “Moral Space”’ (2012) 8 European Competition Journal 73, 85–89; A. Macculloch, ‘The “Public” Wrong of Cartels and the Article 101 TFEU “Object Box”’ (2020) 65 The Antitrust Bulletin 361, 366–69; P. Whelan, ‘Cartel Criminalization and the Challenge of “Moral Wrongfulness”’ (2013) 33 Oxford Journal of Legal Studies 535; Dunne (n 1) 396–99; Stucke (n 42) 495–503; and Chapter 4 in this volume. 38 39

The legal concept of a cartel  97 decisions to have been taken independently.45 It is the subversion of this expectation that causes ‘moral outrage’.46

V.

DEMONSTRATING INDEPENDENCE AND ITS ABSENCE

Certain decisions are required to be taken independently, either for consequential or deontological reasons. The central challenge faced by a competition authority or complainant is to show that a decision has been taken other than independently. Two issues emerge. The first is over what it means to take an independent decision. The second is how to demonstrate that the decision has been taken other than independently. A.

Independent Action and Its Absence

It is clear that firms do not act independently if they have agreed how they will act in the market. It has also been established that a firm does not act independently if it ‘disclose[s] […] the course of conduct which [it has] decided to adopt or contemplate adopting on the market’ to its competitors.47 Such a disclosure enables the firm to ‘influence the conduct on the market of an actual or potential competitor’.48 Finally, a firm is not seen as acting independently if it acts on information obtained from a competitor (directly or indirectly) that it could not otherwise have obtained.49 An important point to emphasize is that a firm’s decision remains independent even when it takes account of how it expects competing firms to react on the market – something often referred to as ‘intelligent adaptation’. The Court of Justice has been clear that such an approach remains independent, noting that ‘every producer is free to change his prices, taking into account in so doing the present or foreseeable conduct of competitors’.50 A firm is also taken to act independently (of its competitors) when it obtains information from others on which to base its decision; thus the Court of Appeals in the 8th Circuit, in Lomar Wholesale Grocery v Dieter’s Gourmet Foods, recognized that

45 A. Stephan, ‘An Empirical Evaluation of the Normative Justifications for Cartel Criminalisation’ (2017) 37 Legal Studies 621, 637–39. 46 Stucke (n 42) 507. Norris v Government of the United States [2008] UKHL 16; [2008] 1 A.C. 920 puts into serious doubt whether cartel conduct is inherently objectionable in the United Kingdom. Not only is it not clear that conduct is objectionable, but such conduct also has at certain times been actively encouraged by governments. See C. Harding and J. Joshua, Regulating Cartels in Europe (2nd ed., OUP, 2010) 53; and A. Falco, ‘The UK Cartel Offence: An Exploration into the Causes of the Underperformance Problem’ (2021) 42(12) ECLR 661, 674-675. 47 Joined Cases 40 etc./73 Coöperatieve Vereniging ‘Suiker Unie’ UA and others v Commission [1975] ECR 1663, para 174. 48 ibid. See also Case C-49/92 P Commission v Anic [1999] ECR I-4125, para 121. 49 Communication from the Commission – Guidelines on the Applicability of Article 101 of the Treaty on the Functioning of the European Union to Horizontal Co-operation Agreements [2011] OJ C 11/1, paras 92–94 and IV/26 918 European Sugar Industry [1973] OJ L140/17, 42. 50 Case 48/69 ICI v Commission (Dyestuffs) [1972] ECR 619, para 118. See also Communication from the Commission – Guidelines on the Applicability of Article 101 of the Treaty on the Functioning of the European Union to Horizontal Co-operation Agreements [2011] OJ C 11/1, paras 92–94 and IV/26 918 European Sugar Industry [1973] OJ L140/17, para 61.

98  Research handbook on cartels the ‘market strategy’ […] often will be the product of consultations between the supplier and his distributors. To apply a per se rule to the supplier’s solicitation of marketing advice from the distributor is to prohibit the supplier from drawing on the most important source of information in formulating and initially implementing a market strategy.51

Posner, an early proponent of treating intelligent adaptations as otherwise than independent action, revisits the idea in his review of Kaplow’s Competition Policy and Price Fixing.52 As the consequences of firms setting a price or staying out of markets in anticipation that their rivals will do the same are indistinguishable from the firms agreeing the same, it seemed illogical that the former was within the scope of independent action and therefore beyond the scope of the cartel prohibition. Following a deontological view, Kovacic et al, with whom Posner agrees, point out that intelligent adaptations leave courts ‘with a substantial conundrum because they cannot meaningfully instruct firms not to react to their rivals’ pricing’.53 The consequences are the same, from the deontological position the firms have done nothing wrong. B.

Demonstrating the Absence of Independence

The absence of independence may be shown directly, by examining the circumstances in which the decision was arrived at. In prosecuting cartel cases, competition officials prefer to have direct evidence of a decision being reached other than independently. Direct evidence is composed of documents (including email messages) expressing agreement and identifying parties to the agreement and oral or written statements from cartel participants describing the operation of the cartel and their participation in it.54 Gellhorn et al note that ‘[d]ocuments and testimony typically stand atop the hierarchy of proof because they tend to give the courts greater confidence that the defendants acted jointly’.55 Competition agencies are increasingly bestowed with enhanced investigatory powers, such as the ability to conduct unannounced inspections or conduct covert surveillance, which at times has enabled video or audio recordings that capture the details of the cartel’s operation contemporaneously.56 Leniency policies also encourage firms to provide agencies with direct evidence in exchange for grants of Lomar Wholesale Grocery, Inc. v Dieter’s Gourmet Foods, Inc., 824 F.2d 582, 593–94 (8th Cir. 1987). See also Case No: 2005/1071, 1074 and 1623 Argos Limited and Littlewoods Limited v Office of Fair Trading and JJB Sports PLC v Office of Fair Trading [2006] EWCA Civ 1318, para 99 and para 106; and Hasselblad [1982] OJ L161/18, para 49. 52 R. Posner, ‘Review of Kaplow, Competition Policy and Price Fixing’ (2014) 79 Antitrust Law Journal 761. See also Encaoua (n 13). On Posner’s earlier account, see R. Posner, Antitrust Law (2nd edn, University of Chicago Press, 2001) 55–60. 53 Kovacic et al (n 5) 410. See also R. Posner, ‘Review of Kaplow, Competition Policy and Price Fixing’ (2014) 79 Antitrust Law Journal 761, 765. 54 Case 65/86 Bayer AG and Maschinenfabrik Hennecke GmbH v Heinz Süllhöfer [1988] ECR 5249. 55 Gellhorn et al (n 4) 269. See also Judge Richard Posner in In re High Fructose Corn Syrup Antitrust Litig., 295 F.3d 651 (7th Cir. 2002). 56 The secret video tapes in Lysine and Galvanised Steel Tanks are the gold standard in cartel evidence, but are very rarely available. See S. Hammond, ‘Caught in the Act: Inside an International Cartel’, Speech, OECD Competition Committee Working Party No. 3 Public Prosecutors Program, www​ .justice​.gov/​atr/​speech/​caught​-act​-inside​-international​-cartel [accessed 13 July 2022]. In 2017, the CMA launched a free digital tool to that uses algorithms to spot unusual bidder behaviour and pricing patterns which may indicate that bid-rigging has taken place. See: www​.gov​.uk/​government/​news/​cma​-launches​ -digital​-tool​-to​-fight​-bid​-rigging [accessed 13 July 2022]. 51

The legal concept of a cartel  99 immunity or reduced penalties, though it remains possible to uncover cartel conduct without reliance on leniency tools.57 Direct evidence is not always available. Private litigants in particular, not having the enhanced enforcement tools afforded to public competition agencies, are often in the position of having to establish the absence of independence using indirect evidence. The greatest controversies in the law on cartels concern the extent to which it is possible to establish that decisions as to bid, market or price are taken other than independently, without there being direct evidence of how the decision was taken.58 In the absence of positive evidence as to how a decision was taken, it is possible to infer that a decision was taken other than independently when there is evidence that ‘tends to exclude the possibility that the alleged conspirators acted independently’.59 Evidence excluding the possibility of independent action may be communication-based or economics-based.60 1. Communication Communication evidence is that which shows the fact that communication or meetings have taken place but does not describe the substance of the communication or meetings.61 It is sometimes referred to as evidence of ‘contact’.62 The competition agency or complainant seeks to establish from the mere fact that contact has taken place, and although no records of what is discussed are available, that the attendees must have discussed bids, market allocation or prices. As an example, in Bananas the European Commission was able to establish that phone calls occurred once before prices were set and a second time shortly afterwards.63 As noted in the relevant decision, the parties had ‘informed the Commission that they had no notes or records concerning these communications’.64 The European Commission was able to establish that it was inconceivable that how the firms intended to conduct themselves on the market was not discussed in these calls. In doing so it relied on the conduct on the market following the calls, oral and written statements from

See as examples Case COMP/38.628 Nitrile Butadiene Rubber, C(2008)282 final; Case COMP/38.543 International Removal Services, C(2008) 926 final; and Case COMP/39.125 Car Glass, C(2008) 6815. 58 See W. Kopit, ‘Inferring Antitrust Conspiracies from Circumstantial Evidence: How Much Is Enough?’ (2007) 52 Antitrust Bulletin 417. 59 Matsushita Electric Indus. Co., Ltd. v Zenith Radio Corp., 475 US 574, 588 (1986). 60 In re High Fructose Corn Syrup Antitrust Litig., 295 F.3d 651, 655 (7th Cir. 2002). 61 Exchanges of correspondence, minutes of meetings, and other such documents, may be used to establish that ‘contact’ has occurred; see Case T-1/89 Rhône-Poulenc v Commission [1991] ECR II-867, AG Opinion, 955-957; and Case C-199/92 P Hüls v Commission [1999] ECR I-4287, paras 141–55. 62 Case T-15/99 Brugg Rohrsystem GmbH v Commission [2002] ECR II-1613, para 74; Case T-28/99, Sigma Tecnologie di rivestimento Srl v Commission [2002] ECR II-1845, paras 44–49; Joined Cases T-25/95 etc Cimenteries CBR v Commission [2000] ECR II-491, para 4098; Case COMP/36.571/D-1 Austrian Banks – ‘Lombard Club’ [2004] OJ L56/1; Joined Cases T-259/02 etc Raiffeisen Zentralbank Österreich AG and others v Commission [2006] ECR II-5169; and Joined Cases C-189/02 P etc Dansk Rÿrindustri. A/S, Isoplus Fernwarmetchnik Vertriebsgesellschaft mbH, Isoplus Ferwarmetchnik Gesellschaft mbtl, Isoplus Fernwarmetchnik GmbH, KE KEUT Kunststoftwerk GmbH, LR of 1998 A/S, Brugg Rohrsysteme gmbh, LR of 1998 and ABB Asea Brown Boveri LTD v Commission [2005] ECR I-05425. 63 Case COMP/39188 Bananas, C(2008) 5955 final, paras 51 and 62. 64 ibid para 147. 57

100  Research handbook on cartels the parties and contemporaneous documents that referred to the communications.65 Though it was argued ‘that the majority of the calls were under 10 minutes, which cannot demonstrate a pattern of any substantial discussions’, that argument was rejected.66 2. Economic Economic evidence is that which either shows an outcome to be implausible if there is independent action or demonstrates structural features of a market that tend to show that cartelization is feasible and so independent action unlikely. Certain outcomes are thought to be implausible if firms act independently. As an example, in Dyestuffs, ten producers of aniline dyes increased their prices by 15 per cent, then by 10 per cent and then by 8 per cent in January 1964, January 1965 and October 1967 respectively.67 The Commission concluded it was ‘not conceivable that without detailed prior agreement’ price increases could occur in this manner.68 The fact that the firms had not acted independently was inferred from: (i) the increases being applied to the same products, even though the ten firms each produced between 1500 and 3500 of some 6000 dyes; (ii) instructions to increase the price all being sent within a three-hour window; and (iii) the price increase notices being similarly worded, including identical drafting and spelling errors.69 By asking whether the conduct of the parties is contrary to their unilateral self-interest it is possible to infer that action is not unilateral. The question might be answered in the affirmative, and the absence of independence inferred, when firms: increase prices in time of oversupply; refuse to deal with a customer or supplier that appears to be an attractive business opportunity; or disclose information that rivals in a competitive market could use to their advantage. Showing that an outcome is implausible if there is independent action has fared more favourably than showing that cartelization is feasible and so independent action unlikely. This is because in Wood Pulp the Court of Justice made it clear that the feasibility of joint intention is insufficient when independent action remains as a plausible explanation for unlikely outcomes such as parallel conduct unprompted by, and unrelated to, a change in costs or demand, persistent and abnormally high profits and stable market shares.70 The Court of Justice had commissioned its own expert report and concluded the economic evidence did not rule out the prospect that the parties acted independently.71 Direct evidence is preferred and, in its absence, communication evidence is sought. To what extent will courts condemn conduct in the absence of communication evidence, that is, rely solely on economic evidence? In Wood Pulp the European Commission had attempted to

ibid paras 172–82. ibid para 80. 67 Case 48/69 ICI v Commission (Dyestuffs) [1972] ECR 619, paras 1–6 and 83–87. At para 58 there was some dispute as to the correct measure of price, the parties arguing that while the list price increased, customers were habitually given large discounts. 68 Re Cartel in Aniline Dyes [1969] OJ L195/11, para 7 (emphasis added). See also Case 48/69 ICI v Commission (Dyestuffs) [1972] ECR 619, para 54;Case 49/69 Badische Anilin-Und Soda-Fabrik AG v Commission [1972] ECR 713, para 35. 69 Re Cartel in Aniline Dyes [1969] OJ L195/11, paras 7–8; Korah (n 7) 220–21; Case 48/69 ICI v Commission (Dyestuffs) [1972] ECR 619, AG Opinion, 674. 70 Joined Cases C-89/85 etc.A. Ahlström Osakeyhtiö and others v Commission [1993] ECR I-1307, para 71. 71 ibid para 126. 65 66

The legal concept of a cartel  101 establish the absence of independence on the basis of economic evidence alone.72 As a result of its failure in Wood Pulp, in Cartonboard the European Commission placed greater emphasis on communication evidence to avoid the parties being able to ‘attribute […] the series of uniform, regular and industry-wide price increases in the carton board sector to the phenomenon of “oligopoly behaviour”’.73 It is generally considered that the reality of operating a cartel means that communication evidence will exist.74 This is not to discount the role of economic evidence in establishing that observed behaviour is determined other than independently. What it does suggest, however, is that economic evidence is more suited to the function of the case selection stage. In determining whether to investigate a market we need ‘to provide a plausible explanation’ of how a cartel can be said to exist and operate.75 The economic account may be used to cross that threshold and save considerable resources if the threshold is not crossed – a competition law investigation is resource intensive and a firm may suffer reputational damage merely by being the subject of a regulatory investigation.

VI. CONCLUSION Cartels are a species of horizontal agreement treated with enhanced opprobrium. This chapter has sought to identify what marks out this species of horizontal agreement for special condemnation. It does not attempt to map the features onto any specific legal regime, though such an exercise could be done. Instead it outlines two problems associated with legal attempts to control cartelization. The first is to specify what is wrong with cartel conduct, consequentially or deontologically. Independent action is expected in relation to price, market availability and bids submitted. To not act independently has negative economic consequences but is also objectionable in the absence of such consequences because it subverts a customer’s expectations of what a market offers. The deontological explanation is not often explored outside regimes that have sought to impose criminal sanctions, but criminal sanctions are not a necessary response to cartels. Identifying the objectionable conduct and explaining why it is objectionable paves the way for the second problem: proving that cartel conduct exists. How can we demonstrate that firms have acted other than independently? This involves an account of what it means to act independently, in situations when the reaction to that action can be readily observed or predicted and in which evidence can be concealed or destroyed. Although there are accounts of how it could be otherwise, the special opprobrium is triggered by the fact that firms have ‘communicated’ – it is communication that we are ultimately objecting to, and the task of cartel enforcement is to identify the ways in which the firms have communicated and why such communication is objectionable.

72 IV/29.725 Wood Pulp [1985] OJ L85/1. See also Joined Cases 29/83 and 30/83 Compagnie Royale Asturienne des Mines SA and Rheinzink GmbH v Commission [1984] ECR 1679, paras 16–17. 73 IV/C/33.833 Cartonboard [1994] OJ L243/1, para 73. See also Case T-304/94 Europa Carton AG v Commission [1998] ECR II-869, para 50. 74 Kovacic et al (n 5) 409–13. 75 ibid 408.

6. Cartels and the concept of the firm Christopher Sagers

I. INTRODUCTION A problem in cartel law is to distinguish between ‘cartels’ and other arrangements or entities that we think the law should treat differently. The law draws that distinction in various ways, but one important criterion is that, in many antitrust regimes, persons or components within a single ‘firm’ cannot conspire with one another. So, the basic law regulating agreements or multilateral action, which is the basis of cartel law, does not apply to such entities. This rule is an aspect of a more general tension, and like many problems in competition law, it is more difficult than it seems. Probably all competition laws need some rule to distinguish cooperation that is permitted from cooperation that is not. On the one hand, some cooperation is to be encouraged. Few goods or services can be produced by one natural individual alone, and even when cooperation is not strictly necessary to produce a given thing, it can generate extensive benefits. But on the other hand, the harms of concern to competition policies tend to be caused by cooperation in one way or another. Sometimes individual producers remain separate, but agree not to vie for the same customers or suppliers on terms of price or quality. That kind of cooperation is usually heavily disfavoured. But at other times, they combine more fully in some fashion – by merging, for example, or forming some joint venture – and combine so much production under common control that they can unilaterally restrain competition. That kind of cooperation is sometimes constrained too, but usually with much less vehemence. Implicit even in that basic example is that the distinction between good and bad cooperation can be difficult to draw and troubling to explain. Why does the law broadly prohibit separate ‘firms’ from agreeing on price or output, while remaining much more tolerant of mergers and other combinations that can have similar effects? When the same firms merge rather than conspire, the resulting entity will only make one price and output decision for the combined output, and the one-time competitors will be more fully, permanently combined. But so long as the merger itself was considered desirable, its decisions are likely legal. Why does law treat those outcomes so differently? Various explanations and distinctions might present themselves, but as we shall see, finding coherence within them has posed major challenges and led to sometimes regrettable policy. As leading scholars have explained, ‘[t]he courts have had to struggle’ in single-entity cases ‘with the unhappy dilemma of either drawing lines between different forms of conduct having virtually identical results, or treating different forms of conduct as being the same despite the

102

Cartels and the concept of the firm  103 differences’.1 Courts have been ‘bemused’ for decades by labels like ‘joint venture’,2 and so their decisions often read like ‘the use of a word rather than a rule’.3 This chapter examines this theoretical tension and the policy dilemmas it generates. It argues that the most basic source of the trouble is the apparent inevitability in single-entity cases of looking for objective answers by asking normative questions. Normative questions are more difficult, uncertain and perhaps ill-suited to law enforcement institutions, and they are particularly ill-suited to very preliminary decisions whether a law should even apply. On that most general level as well, we shall see that the challenge is made worse by a policy choice that is fundamental but not necessary: to focus mainly on conduct rather than structure. The chapter finally reviews the work of courts in the United States and the European Union, which both have developed substantial bodies of case law and critical commentary, and offers some comparative and reform-focused observations.

II. THEORY A.

Is the Distinction Necessary?

Whether and how to draw the distinction is a policy judgement. It balances administrative convenience and the practical needs of business against the risks of insulating harmful conduct without good policy reason. A first question might be whether the law really must distinguish joint and unilateral conduct at all. In principle, the law could empower officials to investigate any conduct or market situation, without preliminary distinctions such as the single-entity determination. Such an inquiry might ask whether a given situation involves net anticompetitive conduct or is organized in a net anticompetitive way. There is some precedent for such an approach. In the early US experience, some prominent observers doubted that unilateral and multilateral conduct should be treated differently.4 There also are some notable modern manifestations, such as the United Kingdom’s Market Investigation regime, which empowers authorities to investigate the competitiveness of given markets, with or without evidence of illegal conduct, and order structural changes.5 Whatever benefits it might have, such an approach probably poses several significant challenges, at least as the basis for day-to-day enforcement. A single entity distinction simplifies enforcement by quickly disposing of many unilateral action cases, since unilateral action is usually more difficult to challenge. That might be desirable if the threat posed by unilateral action is low. Without rules for summary disposal of cases where likelihood of harm is low, 1 D. Turner, ‘The Definition of Agreement under the Sherman Act: Conscious Parallelism and Refusals to Deal’ (1962) 75 Harv LR 655, 656. 2 R. Pitofsky, ‘Joint Ventures under the Antitrust Laws: Some Reflections on the Significance of Penn-Olin’ (1969) 82 Harv LR 1007, 1045–46. 3 A. Berle, ‘The Theory of Enterprise Entity’ (1947) 47 Columbia LR 343, 346. 4 Cp the discussion below in Section III. 5 See Competition and Markets Authority, Market Studies and Market Investigations: Supplemental Guidance on the CMA’s Approach (CMA3, July 2017) https://​assets​.publishing​.service​.gov​.uk/​ government/ ​ u ploads/​ s ystem/​ u ploads/​ a ttachment​ _ data/ ​ f ile/ ​ 6 24706/ ​ c ma3 ​ - markets ​ - supplemental​ -guidance​-updated​-june​-2017​.pdf [accessed 7 March 2022].

104  Research handbook on cartels enforcement could be much more costly. That does not in itself mean that a single-entity distinction is the right summary rule. Unilateral conduct as currently defined might not be so much less dangerous that affording it more deference is justified by the administrative savings. Any preliminary sorting rule will be arbitrary to some degree, and harmful if its factual premise is wrong. And in fact, this guiding premise of the single-entity distinction – that multilateral conduct is more dangerous – has been seriously questioned.6 But in any case, doing away with a single-entity distinction would also complicate business decision making. Managers must make internal decisions on price and output that could be much more difficult if their decisions were subject to legal review, even in principle. The problem is most severe in concentrated markets, in which firms might price supracompetitively and match one another’s elevated prices without any explicit agreement. While no one seriously doubts that that result is possible and undesirable, it could pose a special dilemma if it were made illegal. Firms would effectively have to determine the competitive price and charge it, even if other firms were charging more. But in practice, no firm can really say what the competitive price is. No firm knows even its own marginal costs on an ongoing basis, much less the marginal costs of other firms or in the market overall.7 Such a rule would pose a related problem of remedy. Imagine a court found a firm to violate the law by merely following other firms’ behaviour. The court could enjoin further violations or penalize the violation it found, but ensuring future compliance would require proceedings for contempt or subsequent damages as complex as the first. The effect could be ongoing entanglement of courts and enforcers in pricing and production. It would strain their resources and complicate the sensitivity of markets, dampening the function that is thought to be the most important benefit of market economies. B.

The Metaphysical Problem

Assuming some distinction is needed, however, there remain serious challenges in drawing it. To help with the explanation, the following describes two general approaches to the problem. Different approaches, however, tend to collapse into one another, as different faces of a general dilemma. They reflect the apparently irremediable need to find objective answers by asking normative questions, itself an artifact of the choice to follow a conduct-focused policy rather than one focused on structure. On the one hand, we could define some class of entities, based on common features we consider important, and call them ‘firms’. We could then treat cooperation that occurs within them more deferentially than conduct that occurs between them. To pick a term, we might call this the ‘legal approach’. On the other hand, we could ignore that essentially metaphysical question of legal doctrine, and remain agnostic about where firms’ boundaries lie. Instead, we could review any given example of cooperation and ask whether its participants maintain enough rivalry that the law should require them to compete, notwithstanding their relationship. We could call that the ‘economic approach’.

See, e.g., J. Baker, ‘Exclusion as a Core Competition Concern’ (2013) 78 Antitr LJ 527. See generally R. Pittman, ‘Who Are You Calling Irrational? Marginal Costs and the Pricing Practices of Firms’ (June 2009) https://​papers​.ssrn​.com/​sol3/​papers​.cfm​?abstract​_id​=​1420942 [accessed 7 March 2022]. 6 7

Cartels and the concept of the firm  105 Neither has been very satisfactory. The legal approach begs elusive and perhaps untheorizable questions. While organizations may be real in some senses – it would seem wrong to say, for example, that a government agency does not exist because we cannot say for sure which of its contractors or outside partners are ‘in’ or ‘outside’ of it8 – that does not mean that anyone can answer the questions needed to use the concept to apply legal rules. It remains the case that no one can say objectively what organizations are, how we find their boundaries or how their shared attributes serve competition policy goals. Failure to find those answers has not been for lack of asking, and not just within competition policy. From time to time, metaphysical problems of this nature have been burning jurisprudential preoccupations in Europe and America, notably during the late nineteenth century, as the Second Industrial Revolution generated society-shaping economic and political changes.9 And yet, interest in them has usually ended with a fizzle,10 and nowadays most legal applications leave them unasked. A related social science literature is vast but inconclusive, and seems generally uninterested in metaphysical questions of a kind that could be used in a legal approach. Some real-world examples will illustrate the challenges. It is clear at a minimum that we cannot just use pre-existing legal categories as proxies for antitrust ‘firms’. Imagine a simple rule that agreements between formal legal entities are subject to a conspiracy law, while their internal decisions are not. That rule would be easy to apply, but it would get many situations wrong, and create easy opportunities for evasion. Most obviously, it would be undesirable if a group of fully distinct horizontal competitors could immunize a conspiracy just by incorporating it as a single business entity. Conspirators have often done just that, for example by setting up joint selling agencies as cartel coordinators and incorporating them as separate entities. If such an entity does no more than set a common price for the members’ goods, it is just a price-fixing conspiracy, especially if it is their exclusive sales agent. But conspirators have often argued that such arrangements can’t be conspiracies because they constitute single entities. While courts usually reject that argument in obvious cases, matters are usually more complex, and the mere fact of separate incorporation can often cause real confusion. An example might be the current effort of the Visa and Mastercard networks to evade the conspir-

8 See, e.g., J. Freeman, ‘The Private Role in Public Governance’ (2000) 75 NYU L Rev 543, 572 (decrying arguments that ‘there is no such thing as “public” and “private”’ or that ‘there is no such thing as [a government] agency,’ because ‘[t]here is clearly such a thing as the Environmental Protection Agency, the Securities and Exchange Commission, or the Internal Revenue Service. You can visit their headquarters in Washington’). 9 This period generated a vast legal and philosophical literature to consider business entities, and especially issues such as whether they should be legally ‘personified’, what role they played in the social and political order and their relation to the state. For an excellent review, see G. Mark, ‘The Personification of the Business Corporation in American Law’ (1987) 54 U Chi LR 1441. For more general treatment of the period and the upheaval it generated in thinking about business law and larger economic and political problems, see H. Hovenkamp, Enterprise and American Law, 1836–1937 (Harvard University Press, 1991); W. Hurst, The Legitimacy of the Business Corporation in Law of the United States, 1780–1970 (UVA Press, 1970); G. Gerstle, ‘The Protean Character of American Liberalism’ (1994) 99 Am Hist Rev 1043; and M. Sklar, The Corporate Reconstruction of American Capitalism, 1890–1916 (Cambridge University Press, 1988). 10 See Mark (n 9) (explaining how turn-of-the-century preoccupation with business entities dissipated after publication of the pragmatic point of view laid out in John Dewey’s influential paper, ‘The Historic Background of Corporate Legal Personality’ (1926) 35 Yale LJ 655.

106  Research handbook on cartels acy challenges they have faced for many years by separately incorporating the same decision making that the network’s member banks had traditionally done themselves.11 Other distinguishing features can be found for use in a legal approach, but they tend to face similar problems. For example, some see promise in asking how an organization communicates with or disciplines its components. Cartels usually operate by contract, and they often must police their own members with fines or punitive price responses.12 No such thing characterizes the firm as usually understood; firms usually make internal decisions by fiat. So perhaps the use of contracting or discipline could make a simple and administrable single-entity distinction.13 It might seem nicely attuned to antitrust rules, at least in that it follows their statutory language. Conspiracy law tends to apply only to ‘agreements’, and semantically the term implies that the parties were free to reject the agreement. But the problems with such an approach are serious and immediate. Combinations are easy to imagine that would operate through strong internal fiat directives, but that are nevertheless composed of economically distinct entities free to separate at will, and that are otherwise just ordinary cartels. Imagine that two horizontal competitors establish a joint venture as a free-standing, jointly owned corporation, and contribute all their productive assets to it, but without integrating the assets. That arrangement itself might be subject to merger challenge and pre-consummation review, but it might escape challenge, as the vast majority of them do.14 If it were then considered a single entity, it would be free of conspiracy law, and if it lacks the large market share required for monopolization challenge, it will be effectively free of antitrust altogether.15 Unfortunately, economic approaches have not really worked better. Superficially they seem more promising, because they seem already connected to goals that drive competition policy. However, while certain theories of ‘firms’ have flourished in economics, they offer no very useful means to distinguish kinds of cooperation that could be useful to competition policy, at least not by way of preliminary rules for whether the law even applies. One obvious candidate for economic theory of single entity might be the well-known ‘neoinstitutional’ or ‘transaction cost’ economics (‘TCE’). That theory asks why firms exist at all, and offers some arguments for how they will be structured in particular circumstances.

V. Fleischer, ‘The Mastercard IPO: Protecting the Priceless Brand’ (2007) 12 Harv Negot LR 137. See H. Hovenkamp and C. Leslie, ‘The Firm as Cartel Manager’ (2011) 64 Vand LR 811; and R. Posner, ‘Oligopoly and the Antitrust Laws: A Suggested Approach’ (1969) 21 Stan LR 1562, 1570. 13 See, e.g., Hovenkamp and Leslie (n 12) 861–62 (observing that this distinction might usefully be drawn in antitrust cases, but urging that it not be used as a single-entity rule); cf W. Wils, ‘The Undertaking as Subject of E.C. Competition Law and the Imputation of Infringements to Natural or Legal Persons’ (2000) 25 Eur LR 99 (arguing that a distinguishing feature of single entities should be control, as opposed to contract). 14 At least in the United States, where typically about 99 per cent of the thousands of mergers subject to pre-consummation review each year receive no meaningful review at all, and the many thousands more mergers not even subject to the filing obligation are almost never challenged by either government or private plaintiffs. 15 Admittedly, if the entity then seriously constrains trade, it might face retrospective merger challenge to its initial formation, but only within limitation periods, which in the United States will typically only be four years. Private merger challenges are subject to a strict statutory limitation period of four years, and while there is no statute of limitations on government civil challenges, the equitable doctrine of laches ordinarily limits their claims, often to roughly four years. See generally L. Sullivan, W. Grimes and C. Sagers, The Law of Antitrust: An Integrated Handbook (3rd edn, West Academic 2015). 11 12

Cartels and the concept of the firm  107 Its most basic insight is that in perfect competition, there would be no firms at all – all goods and services could be gotten through market transactions. Its aim is therefore to explain the market imperfection that requires them. Its answer has been that sometimes the difficulty of negotiating open-market sales makes them costlier than making a good in-house.16 As proof of how poorly suited TCE might be to single-entity determinations, one important offshoot still effectively holds that firms do not exist or are irrelevant.17 While many economists still consider firms to matter in various ways, they consider them mainly mechanisms to address normative problems, such as efficiency18 or agency cost,19 and seem generally uninterested in determining what firms ‘are’ or how we tell one from another. More importantly, even if such a theory could offer concrete answers, it would be ill-suited to the preliminary question whether or not a law even applies. As a doctrinal antitrust inquiry, it might help to decide that in a given case two or more individuals seem like a ‘firm’, because the cost or uncertainty of long-term relations among them would be so greatly reduced through cooperation. But that normative inquiry would pose complex and uncertain empirical questions at the very beginning of proceedings, at a time when likely little evidence has been adduced, and would put courts or enforcers in the generally disfavoured position of judging how economic activities are best organized. One might think an economic approach could avoid that challenging normativity by asking not whether entities should combine, but whether, given their incentives, competition among them would be feasible. That is, we could ask whether it would do any good for the law to treat people as capable of conspiring, given the likelihood they would compete if they did not conspire. Sometimes called an ‘identity’ or ‘unity’ of interests approach, this view has been adopted in the United States.20 Competition is thought to be feasible among entities that do not share ownership of the same profits, because if they do, they will have such an identity of interest that they would have no incentive to compete. In principle, persons who are in zero-sum adversity over the same profits will compete over price or quality to steal sales from one another, but persons who share the same profits will not. A variation is that certain collaborators should be exempt, even though they would not otherwise be single entities, because their 16 That idea was introduced in TCE’s seminal article, R. Coase, ‘The Nature of the Firm’ (1937) 4 Economica 386. The literature that followed is vast, notably including O. Williamson, Markets and Hierarchies: Analysis and Antitrust Implications (Free Press, 1975). 17 Namely, the ‘nexus of contracts’ model views ‘firms’ as no more than sets of contracts among capital and input suppliers. Several surprising consequences follow, including that the ‘power’ of discretionary management is not actually power at all, and that a corporate law is counterproductive if it is more than an off-the-shelf, standard-form contract that can be fully varied by a firm’s constituents. See, famously, A. Alchian and H. Demsetz, ‘Production, Information Costs, and Economic Organization’ (1972) 62 Am Econ Rev 777, 777 (‘[Firm managers have] no power of fiat, no authority, no disciplinary action any different in the slightest degree from ordinary market contracting between any two people. I can “punish” you only by withholding future business or by seeking redress in the courts for any failure to honor our exchange agreement. That is exactly all that any employer can do. He can fire or sue, just as I can fire my grocer by stopping purchases from him or sue him for delivering faulty products. What then is the content of the presumed power to manage and assign workers to various tasks? Exactly the same as one little consumer’s power to manage and assign his grocer to various tasks’). 18 See, e.g., Alchian and Demsetz (n 17) (conceiving firms as coordinators of more efficient production). 19 See, e.g., M. Jensen and W. Meckling, ‘Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure’ (1976) 3 J Fin Econ 305. 20 See text from n 37 onward.

108  Research handbook on cartels particular good requires cooperation to be produced at all. That argument has been especially common in sports antitrust cases, following a famous suggestion by Robert Bork.21 However plausible they might seem, such approaches offer little help in most cases, because incentives are in fact complicated in most organizations. On the one hand, decision-making within any business entity can raise conflicts and generate conflicting incentives, even as to decisions such as the pricing or design of products. Especially in large organizations, there may be no real ‘unity’ of interest at all.22 More seriously, substantial harmony of interest commonly characterizes collections of firms that are distinct, independent and capable of socially beneficial competition. That is likely true, for example, in league sports with separately organized teams, trade and professional organizations, purchasing or sales cooperatives and many other organizations. Indeed, the members of any cartel or oligopoly have substantially overlapping interests. While they also have adversarial profit interests, all of them benefit most by sustained and anticompetitive cooperation. In the end, all approaches seem to suffer the same central dilemma. There appears to be no means that is both coherent and administrable on a preliminary basis to make this decision. There is no pre-existing, natural line of separation among objectively distinct ‘firms’. There are only normative judgments about how different arrangements should be thought of. C.

Policy Dilemmas

The continued application of the distinction without a better basis for drawing it generates several policy problems. First, all single-entity rules will tend to leave some conduct beyond the law’s reach that might cause harm. Because multilateral conduct is usually treated more harshly, bad actors might insulate harmful conduct if they can claim to be single entities. If the single-entity distinction is poorly drawn, the conduct might go unaddressed for no good reason. The most important example was already discussed above – follow-the-leader behaviour in oligopoly. Competing firms in concentrated markets might not need to reach any agreement or combination in order to raise their prices or restrict their output. In principle, merely parallel conduct among them can cause as much harm as conspiracy or unilateral exclusion. However, with some important dissenting views,23 it is now generally presumed that independent conduct should be punished only if done with monopoly power, and that rules for multilateral restraints should not be applied without at least some basic proof of agreement.24 R. Bork, The Antitrust Paradox: A Policy at War with Itself (Free Press, 1993) 278. For the influence of Bork’s observation on sports antitrust cases, see Natl Collegiate Athletic Assn v Alston 141 S. Ct. 2141 (2021) 2156; American Needle Inc v Natl Football League 538 F.3d 736 (7th Cir. 2008) 743, revd 560 US 183 (2010); and Natl Collegiate Athletic Assn v Bd of Regents of Univ of Oklahoma 468 US 85 (1984) 101. 22 Even courts very sympathetic to single-entity arguments so recognize; see Chi. Prof’1 Sports Ltd. P’ship v NBA 95 F.3d 593 (7th Cir. 1996) 598 (refusing to require proof of complete unity of interest before applying single-entity treatment, because ‘[c]onflicts are endemic in any multi-stage firm, such as General Motors or IBM, […] [but] these wrangles […] do not demonstrate that [such] firms are cartels, or subject to scrutiny under the [§ 1] [for] their decisions’). 23 Notably, Posner (n 12). 24 A seminal statement of this concern is Turner (n 1). There have been important differing views, and above all a more aggressive application of § 1 to tacit agreement was urged by Richard Posner. See R. Posner, Antitrust Law: An Economic Perspective (University of Chicago 1976) 71 (‘If the economic evidence […] warrants an inference of collusive pricing, there is neither legal nor practical justification 21

Cartels and the concept of the firm  109 So, in most antitrust regimes, actual horizontal agreement on price or output can be punished, while harmful parallelism with the same effects goes unaddressed. Next, the rule for drawing the distinction effectively apportions market power, and it might do so capriciously or inconsistently with policy goals. If ‘firms’ are effectively immunized by a rule that reserves harsh treatment only for agreements among separate entities, then actors that are able to establish large ‘firms’ might be able to build market power unavailable to actors that remain separate. A poignant and much-discussed example is the problem of labour. Workers who try to bargain collectively with their employers would, in most legal systems, need special legal clemency or they would constitute illegal price-fixing conspiracies. For reasons of policy, legal systems give them that clemency, but labour groups must comply with the terms of their special exemptions or face liability. The effect can be to give much more power to the owners of capital, for no better reason than that the people and resources combined within an employer are treated as a firm. Moreover, there often is no good substantive distinction between labour groups and other producers who sell to powerful buyers but that enjoy no exemption at all. For example, family farmers and fishers often sell to powerful agricultural intermediaries. Independent doctors or dentists might have to negotiate with powerful insurers, and news organs negotiate with powerful online media platforms over ad revenues. At least under US law, none of these entities can bargain collectively with their powerful counterparts, because they are considered ‘independent contractors’ rather than ‘employees’.25 And so, as a practical matter, the law preferences combinations that markets find conducive. That might be desirable, but only if markets in fact favour consolidations whose social benefits outweigh their anticompetitive costs. The gain might or might not be net positive, but the preference seems to have been adopted with no serious consideration of the problem or evidence one way or the other. Those considerations lead finally to the most significant policy problem, which has been mentioned several times already: that single-entity dilemmas are an artifact of the choice to prioritize conduct over structure. While it may be obvious that entities cannot be expected to compete when they have identical interests, the question remains how large or powerful entities come to have identical interests in the first place, and whether it is wise to let it happen. If single-entity determinations seem troublesome or leave large enforcement gaps, it may counsel a more aggressive no-fault, structure-focused merger policy, or adoption of rules for no-fault monopoly or affirmative deconcentration, or some other renewed concern for structure in its own right. At a minimum, the problems posed by single-entity doctrines are the price we pay when we deemphasize structure.26 Evidently, a society must choose between

for requiring evidence that will support the further inference that the collusion was explicit rather than tacit’). 25 See generally H. Hafiz, ‘Labor’s Antitrust Paradox’ (2019) 86 Chi LR 381; S. Paul, ‘Antitrust as Allocator of Coordination Rights’ (2020) 67 UCLA LR. 378. For cases distinguishing contractors from ‘employees’ entitled to labour exemption, see Los Angeles Meat and Provision Drivers Union Loc 626 v United States 371 US 94 (1962); and Columbia River Packers Assn v Hinton 315 US 143 (1942). 26 P. Neal, W. Baxter, R. Bork and C. Fulda, ‘Report of the White House Task Force on Antitrust Policy’ (1968–69) 2 Antitr L & Econ R 11, 22–23 (recommending a new structural approach, because ‘[u]nder […] conditions [of oligopoly], it does not suffice for antitrust law to attempt to reach anticompetitive behavior; it cannot order the several firms to ignore each other’s existence. The alternatives, other than accepting the undesirable economic consequences, are either regulation of price (and other decisions) or improving the competitive structure of the market’); Turner (n 1) 671 (‘[T]o fall back on

110  Research handbook on cartels having some structural remedies or living with the cost of a conduct law that cannot reach some substantial injuries.27

III.

LEGAL APPLICATIONS

Several jurisdictions have established single-entity doctrines, and their experience bears out the challenges discussed. The case law is especially well developed in the United States, where the law for some decades harshly punished a range of multilateral conduct but makes unilateral conduct hard to challenge. The EU courts have also developed a substantial case law, and commentators there have examined it at length. The primary American antitrust law, the Sherman Act,28 seems to draw the distinction by its very language. The Act makes two things illegal: § 1 prohibits some conspiracies and agreements, while § 2 prohibits ‘monopoliz[ation]’, as well as attempts and conspiracies to monopolize. Section 1’s language specifically prohibits any ‘contract, combination, […] or conspiracy’ that unreasonably restrains trade. American courts and commentators have almost always taken those words to convey that one legal person, acting alone, cannot violate § 1. The conduct of a single entity therefore can only violate § 2. Interestingly, while American lawyers now take this principle as undoubted, there is substantial reason to question whether the Congress of 1890 intended it. Semantically, ‘combination’ could easily include things we now consider single entities, and in the legal and popular usage of 1890 the word likely included such things – for example, the ‘trusts’ of that period, for which antitrust law is named.29 Even the modern Supreme Court has acknowledged as much.30 There also is policy reason to revisit the distinction in American law. When the Supreme Court adopted its modern approach, it did so against a backdrop of harsh substantive conduct rules. Throughout the mid-twentieth century, American courts held a large range of agreements automatically illegal, and if those rules could apply to the internal arrangements of corporate families or other ordinary combinations, they could seriously complicate mundane [a structural] remedy is virtually to concede that the finding of liability on the ground of conspiracy is dubious at best. If effective and workable relief requires a radical structural reformation of the industry, this indicates that it was the structural situation, not the behavior of the industry members, which was fundamentally responsible for the unsatisfactory results’). 27 See generally C. Sagers, ‘#LOLNothingMatters’ (2018) 63 Antitr Bull 7; and Wils (n 13) 105. 28 15 USC §§ 1–7. 29 As Justice Stevens pointed out in his dissent in Copperweld Corp v Independence Tube Corp 467 US 752 (1984), the law of criminal conspiracy as it existed in 1890 recognized that affiliated corporations could conspire and that corporate agents could conspire with one another or with their corporations: ibid 785–86 (Stevens J). (Justice Stevens’s citations are all to post-1890 cases, but there is plenty of authority to the same effect predating 1890.) None less than Oliver Wendell Holmes believed the two sections of the Sherman Act were essentially the same: Northern Secs Co v United States 193 US 197 (1904) 404 (Holmes J) (‘All that is added to the 1st section by § 2 is that like penalties are imposed upon every single person who, without combination, monopolizes […] It is more important as an aid to the construction of § 1 than it is on its own account. It shows that whatever is criminal when done by way of combination is equally criminal if done by a single man’). For a more elaborate history of the issue, see C. Sagers, ‘Why Copperweld Was Actually Kind of Dumb: Sound, Fury, and the Once and Still Missing Antitrust Theory of the Firm’ (2011) 18 Vill Sp & Ent LJ 377, 381–88. 30 Copperweld Corp v Independence Tube Corp (n 29) 769 (‘Nothing in the literal meaning of th[e] terms [of § 1] excludes coordinated conduct among officers or employees of the same company’).

Cartels and the concept of the firm  111 organizational choices. But per se rules have been significantly limited and it has grown much harder to prove § 1 claims in general, so perhaps there is room to consider how much of the uncertainty and policy cost of single-entity inquiry is worthwhile. In any event, the issue came to a fairly urgent head in America in the early 1980s, against that history of mid-century rigour.31 Over some decades there had grown a controversial case law in which the Supreme Court suggested that members of the same corporate family could conspire, at least if they were separately organized.32 That particular result came to be known as the ‘intra-enterprise conspiracy’ rule, and it was the target of decades of vehement criticism,33 even by the federal enforcement agencies.34 Because many kinds of agreements remained per se illegal, it was possible for a corporate parent to break the law, and in principle to risk criminal penalties and treble private damages, merely by directing the sales operations of a wholly owned subsidiary. Critics were especially bothered that two different companies might be treated very differently, though they are identical in substance, if one of them were organized through internal divisions and the other through separately organized subsidiaries. Critics observed that the reason to choose one organization or the other might just be to secure tax advantages, comply with different local regulation, limit liability or increase efficiency, and in any event that the choice had no relevance to the competitiveness of markets. The issue was to some degree put to rest in two key decisions, the seminal Copperweld Corp v Independence Tube Corp of 198435 and a follow-up decision of much later called American Needle Inc v NFL.36 Tellingly, however, despite the elaborate reasoning laid out in those opinions, the several subsidiary doctrinal rules subsequently established and the presumption of many lawyers that the law had been substantially clarified, the law remains more or less as uncertain as before. Not only is the law not obviously clearer or better after Copperweld, it seems not even materially different.

31 See, e.g., Hovenkamp and Leslie (n 12) 815 (noting that the issue seemed more urgent because of the strictness of substantive rules of the time). 32 The intraenterprise conspiracy rule was based almost entirely on three Supreme Court opinions, each of them written by Justice Hugo Black, over about 20 years: Perma Life Mufflers Inc v Intl Parts Corp 392 US 134 (1968); Timken Roller Bearing Co v United States 341 US 593 (1951); Kiefer-Stewart Co v Joseph E Seagram & Sons Inc 340 US 211 (1951). Justice Black also claimed that an intraenterprise rule was applied in United States v Yellow Cab Co 332 US 218 (1947). See Kiefer-Stewart (ibid) 215 (citing Yellow Cab). His interpretation of Yellow Cab has been disputed, and critics have noted that even the three decisions adopting the doctrine could have been decided on other grounds, since there was a basis under existing law either for § 2 liability or for § 1 conspiracies with parties other than the defendants’ own corporate relatives; see Sagers (n 29) 381–84. 33 See, e.g., M. Adelman, ‘Integration and Antitrust Policy’ (1949) 63 Harv LR 27, 50–53; P. Areeda, ‘Intraenterprise Conspiracy in Decline’ (1983) 97 Harv LR 451; M. Handler and T. Smart, ‘The Present Status of the Intracorporate Conspiracy Doctrine’ (1981) 3 Cardozo LR 23 (1981); D. Kempf, ‘Bathtub Conspiracies: Has Seagram Distilled a More Potent Brew?’ (1968) 24 Bus Lyr 173; L. McQuade, ‘Conspiracy, Multicorporate Enterprises, and Section 1 of the Sherman Act’ (1955) 41 Va LR 183; J. Rahl, ‘Conspiracy and the Anti-Trust Laws’ (1950) 44 Ill LR 743; G. Stengel, ‘Intra-Enterprise Conspiracy Under Section 1 of the Sherman Act’ (1963) 35 Miss LJ 5; A. Jones, ‘Intraenterprise Antitrust Conspiracy: A Decisionmaking Approach’ (1983) 71 Cal LR 1732; and Note, ‘Intra-Enterprise Conspiracy Under Section 1 of the Sherman Act: A Suggested Standard’ (1977) 75 Mich LR 717. 34 See, e.g., D. Turner, ‘Address before the American Bar Association’ (1965) 10 Antitr Bull 685, 687 (speech by Assistant Attorney General for Antitrust). 35 Copperweld (n 29). 36 560 US 183 (2010).

112  Research handbook on cartels Copperweld’s primary holding was to reverse the intra-enterprise conspiracy doctrine and to hold that a parent and a 100 per cent subsidiary cannot conspire, as a matter of law. Technically the decision was limited to those particular facts,37 but in the course of its opinion the Court laid out several economic arguments favouring exemption of corporate families from conspiracy law. They would effectively become the reasoning of all subsequent single-entity determinations. The core of this reasoning is the ‘unity of interest’ approach. The Court explained that entities sharing the same profits would not compete even if the law forbade them from conspiring, so they should be treated as single entities.38 Importantly, the Court also stressed another value, which was key to the dominant conservative economics of the day, and claimed that it was a value chosen by the 1890 Congress: that multilateral conduct is in fact much more dangerous that unilateral conduct, and thus that only clearly, plainly multilateral agreement could be exposed to the tougher liability standards of § 1.39 Single-entity issues were thereafter still litigated and could be uncertain, notwithstanding the apparent new clarity. Defendants attempted to use Copperweld’s economic reasoning to expand the category of exempt arrangements, with mixed results.40 The campaign was driven above all by professional sports leagues and their supporters, who stressed that some cooperation between separate teams seemed necessary for the product to exist.41 At length that effort too was nominally put to rest, in American Needle. The unanimous opinion – written, interestingly enough, by a Copperweld dissenter – rejected single-entity treatment for a professional sports league made up of separately incorporated teams, with respect to their participation in a joint licensing entity they created to license their trademarks. The Court reaffirmed Copperweld and its unity-of-interest approach, but found that the teams’ interests in their jointly produced sports product were not so unified that they couldn’t compete in a range of respects. In some particulars, Copperweld and American Needle may have clarified the policy. It is now clear as a matter of law that a corporation cannot conspire with a subsidiary, whether wholly- or majority-owned. Section 1 likewise cannot be violated by officers or employees of the same firm,42 by a corporation’s unincorporated internal divisions43 or by agreement among a firm and other entities that are its ‘agents’.44 There also is some reassurance since American Copperweld (n 29) 767. The Court stressed that ‘[a] parent and its wholly owned subsidiary have a complete unity of interest’. The parent’s and subsidiary’s ‘objectives are common, not disparate; their general corporate actions are guided or determined not by two separate corporate consciousnesses, but one. They are not unlike a multiple team of horses drawing a vehicle under the control of a single driver. With or without a formal “agreement,” the subsidiary acts for the benefit of the parent, its sole shareholder’: ibid 771–72. 39 ibid 775–76. 40 See, e.g., Freeman v San Diego Bd of Realtors 322 F3d 1133 (9th Cir 2003) (rejecting single-entity treatment for professional association of realtors); Oksanen v Page Mem Hosp 954 F2d 696 (4th Cir 1991) (holding hospital and doctors with privileges to practice there one single entity); and City of Mt Pleasant v Assoc Elec Coop Inc 838 F2d 268 (8th Cir 1988) (finding cooperative of separately incorporated electric utilities a single entity). 41 See, e.g., Fraser v Major League Soccer 284 F3d 47 (1st Cir 2002) (refusing to reach question of single-entity treatment for professional sports league); Chi Prof1 Sports Ltd. P’ship (n 22) 598 (finding that professional sports league could be single entity, but remanding for further consideration). 42 Copperweld (n 29) 769 and n 15 (affirming lower-court caselaw that had so held). 43 ibid 770 and n 16 (so stating, and affirming lower-court caselaw). 44 This in fact was an old rule, adopted in United States v General Electric Co 272 US 476 (1926). As later applied, courts required substantial proof of actual, substantive agency, as opposed to arm’s-length relation pretending to be agency; see Simpson v Union Oil Co of Cal 377 US 13 (1964). But the rule has 37 38

Cartels and the concept of the firm  113 Needle that the modern Court will hesitate to extend single-entity treatment very far beyond those categories. It remains the law that price-fixers cannot simply incorporate their cartels or label them ‘joint ventures’. Likewise, nothing depends on whether the arrangement in question is ‘legitimate’ or legal in other respects. And the same entity might be single for some purposes but not others, so that its conduct can be considered unilateral in some cases but multilateral in others.45 And yet, the law may not really have been clarified so much at all. Even the specific holding for which Copperweld is celebrated – that companies cannot conspire with wholly-owned subsidiaries – was arguably not very significant. Despite all the controversy, the intra-enterprise conspiracy rule was almost never actually applied – arguably only twice in the law’s entire history, one occasion being the lower court decision in Copperweld.46 While courts (including the Supreme Court) obligingly reaffirmed the rule in the abstract, they almost always found reasons to reject it in application, and the Supreme Court often wrote of it in ways at odds with the doctrine.47 So, Copperweld arguably just reaffirmed the existing law, more or less. And that specific issue had much, much less policy significance once the several per se rules of mid-century antitrust had softened. The intra-enterprise conspiracy rule could indeed pose serious problems in the days when a whole variety of agreements could be per se illegal and generate criminal penalties and treble damages, including vertical agreements. But as the law stood even by the mid-1980s, the likelihood of substantive liability for a parent’s directions to its subsidiary was small. As it stands now, that likelihood is essentially nothing, as any lawsuit on such a basis is likely to be summarily dismissed and no government or private enforcer would likely waste the time and money. The other black-letter rules now clear under Copperweld were likewise already established before Copperweld reaffirmed them.48 Copperweld’s holding also happens to be the easiest possible issue in single-entity law, and resolving it sheds little light on most disputes, which usually are harder. The law remains difficult to apply, and the cases that remain difficult are the ones that were difficult before Copperweld; post-Copperweld courts have observed that the law remains a quagmire.49 Most telling of all, the post-Copperweld decisions are not really any different than pre-Copperweld cases. As before, the courts apply largely unguided, untheorized, impressionistic hunches, though Copperweld itself explicitly criticized the varying lower court approaches existing

been reaffirmed and appears to be alive and well. See, e.g., Valuepest.com of Charlotte Inc v Bayer Corp 561 F3d 282 (4th Cir 2009). 45 See Hovenkamp and Leslie (n 12) 823–34. 46 The other was Photovest Corp v Fotomat Corp, 606 F2d 704 (7th Cir 1979) 725–27. As explained, and as Copperweld itself stressed, even the Supreme Court opinions on which the intraenterprise conspiracy doctrine was based could all have been resolved on other grounds. 47 See Areeda (n 33) 462–70 (so noting and citing cases). 48 United States v Gen Electric Co 272 US 476 (1926) (firm cannot conspire with its own agent, acting qua agent); Joseph E Seagram & Sons Inc v Hawaiian Oke & Liquors Ltd 416 F2d 71 (9th Cir1969) (unincorporated divisions of the same firm cannot conspire with one another); and Kempf (n 33) 173–74 (discussing caselaw holding that officers of the same firm cannot conspire with one another or the firm). 49 See, e.g., Freeman v San Diego Bd of Realtors 322 F3d 1133 (9th Cir 2003); and Fraser v Major League Soccer 284 F3d 47 (1st Cir 2002) (‘The criteria suggested in the [single entity] cases are so general and so various (unity of interest, lack of existing competition, extent of control), as to emphasize the lack of any developed body of law’).

114  Research handbook on cartels before that decision, and which the courts have essentially preserved.50 In other words, as a practical matter, the seemingly sensible and influential Copperweld resolved virtually no questions at all, except the simplest and most obvious. The US experience also shows how a single-entity rule can damage policy. As mentioned, mere parallel oligopoly pricing, without some agreement on price or output, cannot be illegal, unless the individual oligopolists committed violations of § 2. But since the mid-twentieth century, the courts have made § 2 much more difficult to enforce than § 1, mainly by requiring very substantial, entry-protected market share as proof of ‘monopoly’. The problem has grown significantly more severe since the Supreme Court also began making conspiracy very difficult to prove, in its controversial opinions in Bell Atl Corp v Twombly51 and Ashcroft v Iqbal.52 This so-called Copperweld gap,53 in which harmful conduct cannot be challenged because it is unilateral but not monopolistic, has thus grown wide. As the law currently stands, effectively no conduct is illegal unless there is either elaborate, detailed proof of conspiracy, or monopoly power shown by market share of roughly 70 per cent or more and substantial entry protection. That point was reached in some large part because courts and commentators feared risking antitrust exposure of single entities, and a major contribution to the rhetoric limiting that exposure was Copperweld’s extended argument that multilateral conduct is much more dangerous than unilateral conduct.54 European Union law deploys an apparently very similar distinction, and it too is set out in a substantial case law. Article 101 of the Treaty on the Functioning of the European Union (‘TFEU’) prohibits ‘agreements’ between ‘undertakings’ whose ‘object or effect’ is the ‘prevention, restriction or distortion of competition’. As with Sherman Act § 1, the EU courts have interpreted Article 101 TFEU to require that there be more than one ‘undertaking’ for there to be ‘agreement’. The EU Merger Regulation likewise implies a single-entity notion, since it applies only to combinations of separate entities.55 Unilateral action can therefore violate only Article 102 TFEU, which more closely resembles Sherman Act § 2, in prohibiting ‘abuse […] of a dominant position’. One consequence is an effect like the American Copperweld gap, because Article 102 liability requires that an undertaking be ‘dominant’ and act ‘abusively’, and so that liability is more difficult to establish.56 There are some technical differences of doctrine, but overall the single-entity rules developed in Europe and the United States seem similar and serve similar policy commitments. In any number of decisions, the EU courts have held that there can be no conspiracy between

Sagers (n 29) 390–93 (summarizing cases). 550 US 544 (2007). 52 556 US 662 (2009). 53 See Baker (n 1). 54 Copperweld (n 29) 775–76. 55 Council Regulation (EC) No 139/2004 of 20 January 2004 on the Control of Concentrations between Undertakings (the EC Merger Regulation) [2004] OJ L24/1. 56 See Alison Jones, ‘The Boundaries of an Undertaking in EU Competition Law’ (2012) 8 Eur Comp J 301, 301–02; and Wils (n 13) 105. 50 51

Cartels and the concept of the firm  115 a parent and a subsidiary that it controls,57 among commonly owned corporate siblings,58 between an employer and its employees59 or between a firm and its non-employee agents.60 As in US law, the EU doctrine may contain some theoretical tension or confusion, in that its rationale is explained in varying terms. Some decisions explain it in terms of economic substance, similarly to Copperweld, saying that ‘[i]n competition law, the term “undertaking” must be understood as designating an economic unit […] even if in law [it] consists of several persons, natural or legal’, on the reasoning that ‘competition between [such] persons […] is impossible’.61 But in some other cases, the reasoning seems more legally formalistic, and rests on the view that commonly controlled entities are legally incapable of entering into ‘agreements’ for purposes of Article 101 TFEU, because a controlled entity has no discretion whether or not to ‘agree’ to anything.62 There also has been a fair bit of criticism of the EU doctrine, especially of one consequence it can impose at the remedy phase. Unlike US enforcers, the European Commission may impose fines on antitrust violators.63 Controversy arises because the maximum fine is set at 10 per cent of the firm’s global annual turnover, and that can include the entire turnover of a corporate group if the parent of the group exercised ‘decisive influence’ over a law-breaking subsidiary. Under certain conditions (viz., where it actually exercised decisive influence over the commercial policy of the violating company), a parent company and/or a sibling subsidiary thus can be vicariously liable for another company’s antitrust violation, despite an emphasis elsewhere in EU law on personal fault for imposition of liability,64 and despite some peculiar

57 Cases 56 and 58/64, Etablissements Consten SdRL and Grundigverkaufs-GmbH v Commission [1966] ECR 299, 340. 58 Case T-102/92, Viho v Commission [1995] ECR 11-17, para 47. 59 Joined Cases 40–48 etc, Cooperatieve Vereniging Suiker Unie v Commission [1975] ECR 1663, para 539. 60 Note that, under specific language in the Treaty itself, Article 101 TFEU does apply to an ‘association’ of undertakings; see Joined Cases 96/82 to 102/82etc, IAZ International Belgium and others v Commission [1983] ECR 3369, para 20 (finding that an association could not avoid Article 101 liability for conspiracy among its members, on grounds that it itself was a single entity). 61 Case 170/83, Hydrotherm v Compact [1985] ECR 3016, paras 10–11; and Case T-11/89 Shell v Commission [1992] ECR II-884, para 312. 62 Critics have noticed this tension, observing that the two separate rationales for finding a single entity – that the participants in the entity might be legally incapable of agreement and that the participants might be incapable of competing – are not the same, and won’t always both be true in the same case; see O. Odudu and D. Bailey, ‘The Single Economic Entity Doctrine in EU Competition Law’ (2014) 51 Common Market LR 1721, 1738–42. 63 In civil matters, the American enforcement agencies can ordinarily only seek injunctive remedies. The Justice Department can recover money when it prosecutes antitrust matters criminally (in which case it can secure criminal fines), but only a very narrow range of conduct is now prosecuted as criminal. Likewise, it can secure disgorgement of ill-gotten gains in some narrow circumstances (and the Federal Trade Commission’s (‘FTC’) disgorgement power was recently curtailed in AMG Capital Mgt LLC v FTC 141 S. Ct. 1341 (2021)). 64 See Jones (n 56) 304 and n 16 (stating this criticism and citing other critics). There is a very roughly equivalent doctrine in US law, under which principles are responsible for antitrust violations of their agents; see Am Soc of Mech Engineers Inc v Hydrolevel Corp 456 US 556 (1982). That doctrine is applied, however, only according to the common law of agency, under which corporate subsidiaries are not agents merely by the fact of the parent’s ownership. And even in cases of actual agency, the amount of any money damages or penalty would not be increased by considering the revenues of a larger corpo-

116  Research handbook on cartels policy effects. For example, we would not usually impute liability for a parent company’s wrongdoing on even wholly-owned subsidiaries.65

IV.

CONCLUSION: COMPARATIVE AND REFORM CONSIDERATIONS

Even though they already seem so similar, comparison of US and EU law suggests at least one important lesson. EU courts and critics have recently sought to import learning from the American approach, and above all the economic reasoning of the Copperweld doctrine.66 That is understandable, as was the hope of mid-century American lawyers for a more coherent doctrine. European lawyers are also keen to avoid specific policy problems, not unlike the peculiarities of intra-enterprise conspiracy that Americans found so frustrating. But the American experience really adds little support to a desire for a doctrine like Copperweld. For one thing, the policy problems of intra-enterprise conspiracy were problems much more in principle than in application, since intra-enterprise conspiracy was almost never found in practice. It also seems unlikely that the conjectural possibility of conspiracy liability actually dampened American companies’ organizational choices, even though it meant exposure to the per se rules and criminal liability risks that were much broader at that time. American corporations were actively acquisitive throughout the period, especially during the notorious conglomerate merger wave of the 1960s, and ownership of separately incorporated subsidiaries was common. So perhaps one lesson is that academic observers and others should be sure that problems are real before they solve them, not merely conceptual or intellectual. Second, Copperweld may have brought undesirable policy outcomes of its own. In the uncertainty that remained, defendants in all kinds of different arrangements sought single-entity treatment on vague claims that their interests were aligned in some way, and they not infrequently succeeded, even when they consisted of loosely affiliated, legally discrete entities.67 More importantly, the emphasis in Copperweld on the much greater danger of multilateral conduct seems to have contributed to America’s serious Copperweld gap. Admittedly, the problem may be less acute in Europe and other jurisdictions; in Europe, for example, Article 102 TFEU prohibits abuse of dominance by ‘one or more undertakings’ (emphasis added). So the gap may not be so large. But the point remains that in the American experience, academic concern over problems that turned out to be much more conjectural than real resulted in a serious worsening of what is now an acute policy problem. But in the end, maybe the most important lesson is just that, to whatever extent there were real problems to be solved, Copperweld’s celebrated ‘economic’ approach has not helped solve them. Its reasoning, seemingly so commonsensical, in fact resolves no difficult questions at all. By announcing its reasoning in a case whose facts happened to be the simplest possible

rate organization. Damages and fines are calculated according to actual injury caused or the nature of the conduct, without reference to any wrongdoer’s value. 65 Jones (n 56) 320 (making this point). 66 ibid 308; Odudu and Bailey (n 62) 1729–30; and Wils (n 13) 105. 67 See, e.g., City of Mt Pleasant (n 40) (cooperative of separately incorporated electric utilities was a ‘single entity’).

Cartels and the concept of the firm  117 iteration of a single-entity problem, the Court gave quite a misleading impression that the law would be improved by common-sense economics. But for all that, the direction of reform in both America and Europe might actually be towards increased application of single-entity protection. There is some indication in both US and European law that internal decisions of joint ventures might be exempt from conspiracy law where the creation of the joint venture itself could have been challenged under merger law. In the US, the much criticized Texaco Inc v Dagher68 seemed to hint as much without reaching the question, and some EU cases indicate that Art 101 TFEU should not apply to the ‘core’ decisions of combinations that were ‘legal at their inception’.69 All the evidence in this study suggests that any such change would just worsen the problems of a policy that may have been unnecessary to begin with. In summary, the basic lesson is that the seemingly commonsense need to let businesses make their own decisions might be fairly misleading. While legitimately individual businesses no doubt must make internal decisions hierarchically and without legal interference, the insularity they need absolutely should not be extended so far as many have assumed. Looser combinations of entities can easily withstand some legal exposure, and not so much is lost by considering them capable of conspiracy as to justify the cost of immunizing them from the law. If the comparatively mature experience of the American law teaches anything, it is that the gains of broader single-entity immunity are probably limited and come at significant cost.

547 US 1 (2006). Hovenkamp and Leslie (n 12) 866 (laying out criticisms of Dagher); and Jones (n 56) 324–25 (discussing European cases). 68 69

7. The concept of a Single and Complex Continuous Infringement Stefan Thomas

I. INTRODUCTION The concept of a Single and Complex Continuous Infringement (‘SCCI’)1 in EU competition law allows the European Commission (‘the Commission’) to treat different elements of conspiracy between firms at different times, and even relating to two or more distinct markets, as one unitary infringement. The doctrine can have huge implications in cartel cases. Depending on the context it can help the Commission build its case, plug evidence gaps, and prevent limitation periods from running. At the same time, it can benefit the perpetrator in that one unitary fine substitutes for several individual fines, so that the 10 per cent cap of Article 23 of Regulation 1/2003 is only applied once.2 Due to the ambiguity of the effects it may precipitate for the alleged perpetrator, the finding of an SCCI often becomes a matter of legal disputes in fine proceedings before the Commission and the EU courts. This chapter focuses on the concept of an SCCI. It examines in turn: the nature of the concept (Section II); how one establishes an SCCI (Section III); the implications of an SCCI for public enforcement (Section IV); and the implications of an SCCI for private enforcement (Section V).

1 On the concept of an SCCI, see generally P. Alexiadis, D. Swanson and A. Guerrero Perez, ‘Raising the EU Evidentiary Bar for the “Single and Continuous Infringement” Doctrine’ (2016) No. 4-2016 Concurrences 1; P. Biolan, ‘Limited Awareness of Cartel Participants: Any Consequence for the Single Infringement in EU Competition Law’ (2015) 6 J. Eur. Comp. L. & Prac. 383; J. Joshua, ‘Single Continuous Infringement of Article 81 EC: Has the Commission Stretched the Concept Beyond the Limit of Its Logic?’ (2009) 5 Eur. Competition J. 451; J. Kallaugher and A. Weitbrecht, ‘Microsoft and More – Developments under Articles 81 and 82 EC in 2007’ (2008) 29 ECLR 418; D. Riley, ‘Revisiting the Single and Continuous Infringement of Article 101: The Significance of Anic in a New Era of Cartel Detection and Analysis’ (2014) 37 World Competition 293; M. Romić, ‘Particularities of Proving a Single and Continuous Infringement of EU Competition Rules’ (2020) 22 YARS 169; K. Seifert, ‘The Single Complex and Continuous Infringement – “Effet” Utilitarism?’ (2008) 29 ECLR 546; P. Studt, ‘Case C-615/15 P Samsung SDI v Commission: The Concept of “Single and Continuous Infringement” and Cartels. Continuity or Change?’ (2017) 8 J. Eur. Comp. L. & Prac. 644; and M. van der Woude, ‘Judicial Control in Complex Economic Matters’ (2019) 10 J. Eur. Comp. L. & Prac. 415. 2 Case COMP/39092 Bathroom Fittings and Fixtures, Commission decision, 23 June 2010, para 1262 and footnote 1744.

118

The concept of a Single and Complex Continuous Infringement  119

II.

NATURE OF THE CONCEPT

A.

Different Types of SCCI

The concept of SCCI is actually a ‘legal characterization’3 that covers a range of situations in which a multitude of actions and/or effects on one or more markets are compressed into one unitary finding of an infringement in terms of law enforcement. First of all, SCCI can describe the fact that a single cartel on a particular market is composed of several agreements, concerted practices or decisions of associations of undertakings.4 In those instances, the notion of SCCI merely describes the fact that the cartelists have relied on a range of different means and forms of interaction when coordinating their behaviour. Also, it can be used to combine two or more concomitant infringements among the same group of cartelists relating to two or more distinct product markets or geographical markets. Sometimes, the Commission speaks of a ‘market’ when making reference to the entirety of product categories that fall within the scope of an SCCI;5 yet this is not meant as a market definition in a technical sense. In Pre-Insulated Pipes, for example, the Commission applied the SCCI concept to a geographical area while leaving the question open as to whether the agreement actually related to segments of the same market or to different markets.6 The jurisprudence has ruled that a market definition is not necessary in SCCI cases since ‘the fact that the infringement extended over distinct product and geographic markets does not in any event preclude the finding of a single infringement’.7 The Commission initially showed reluctance to combine different product markets if not all addressees were involved in the cartelization of all these markets. In PO/Thread, the Commission stated that three distinct infringements had existed, inter alia based on the fact that only some of the firms had been active on all three markets, and others merely on one or two of them.8 In Methacrylates, however, the Commission found an SCCI9 even though not all conspirators had been active on all affected markets.10 An SCCI may therefore be found in

3 Joined Cases T-101/05 and T-111/05, BASF AG and UCB SA v Commission ECLI:​EU:​T:​2007:​ 380, para 159. 4 Joined Cases T‑305/94 etc. Limburgse Maatschappij and others v Commission ECLI:​EU:​T:​1999:​ 80, para 697; Case T‑1/89 Rhône-Poulenc v Commission ECLI:​EU:​T:​1991:​56, para 126; Cases T‑9/99 HFB and others v Commission ECLI:​EU:​T:​2002:​70, para 186; and BASF (n 3) para 159. 5 Bathroom Fittings and Fixtures (n 2) para 1211. 6 Case IV/35.691/E-4 Pre-Insulated Pipe Cartel, Commission decision, 21 October 1998, para 140. See also Case COMP/36.571/D-1 Austrian Banks – ‘Lombard Club’, Commission decision, 11 June 2002, para 425; Joined Cases T‑259/02 to T‑264/02 and T‑271/02 Raiffeisen Zentralbank Österreich AG and others v Commission ECLI:​EU:​T:​2006:​396, paras 111 et seq, especially para 121; and Case COMP/F-1/38.338 PO/Needles, Commission decision, 26 October 2004, where three distinct products markets in the field of haberdashery and needles were compressed into one SCCI (see paras 47 and 256). 7 Case C-625/13 P Villeroy & Boch ECLI:​EU:​C:​2017:​52, para 165. 8 Case COMP 38.337 PO/Threads, Commission decision,14 September 2005, paras 264 et seq. 9 Case COMP/F/38.645 Methacrylates, Commission decision, 31 May 2006, paras 6 and 222 et seq. 10 ibid para 226. See also Case T-28/99 Sigma Tecnologie di rivestimento Srl v Commission ECLI:​EU:​T:​2002:​76, para 40; Case COMP/E-2/37.857 Organic Peroxides, Commission decision, 10 December 2003, para 310; Joined Cases C-204/00 P etc. Aalborg Portland A/S and others v Commission ECLI:​EU:​C:​2004:​6, para 258; and Case T-279/02 Degussa AG v Commission ECLI:​EU:​T:​2006:​103, para 155.

120  Research handbook on cartels cases where some of the firms have supported, through their collusive action, the implementation of the cartel on markets on which they were not personally active.11 Some argue that it may suffice that only one cartelist is responsible for the entire scope of the SCCI, while all other firms engaged in mere fractions of it, yet with no overlaps between each other.12 It is possible that two or more SCCIs constitute one combined SCCI if there are sufficiently strong links between them.13 In BASF, however, the Court of First Instance (now known as the General Court (‘GC’)) denied sufficiently strong bounds between the anticompetitive actions on a global level, which had already expired, and the subsequent actions that were taken in the same business field in Europe.14 The SCCI doctrine is mostly relevant in horizontal cases, yet the Commission also applies it to vertical restraints relating to different product markets, as in Nintendo.15 SCCI usually becomes important for infringements of a long duration, while some scholars advance that it can, as a matter of principle, also gain relevance in complex cartel infringements of a shorter lifespan.16 B.

The Doctrinal Approach behind SCCI

The doctrinal background behind the SCCI concept has never been extensively discussed in the decisional practice. There is no specific provision in EU law concretizing it. The underlying idea buoys in the second sentence of Article 25(2) of Regulation 1/2003, which states that in ‘the case of continuing or repeated infringements, time shall begin to run on the day on which the infringement ceases’. The GC has argued that SCCI ‘has its origin in a widespread conception in the legal orders of the Member States concerning the attribution of responsibility for infringements committed by several perpetrators according to their participation in the infringement as a whole’.17 Accordingly, the jurisprudence finds that the concept reconciles with the principle of personal responsibility and the rights of defence.18 The Commission argues that it would be ‘artificial’ to ‘break down’ the behaviour in a number of separate infringements ‘while the relevant conduct in reality constituted a single infringement’.19 However, one should be careful not to succumb to the circular argument that something is ‘treated as a single infringement because it is a single infringement’. As is often the case with unwritten rules in EU antitrust law, the reasoning oscillates between arguments based on substantive law and mere enforcement postulates. In the given context this means that it is unclear whether, in the case of an SCCI, there is actually one

See BASF (n 3) paras 177–79; and HFB and others (n 4) para 231. A. Colombani, J. Kloub and E. Sakkers, ‘Cartels’, in J. Faull and A. Nikpay (eds), The EU Law of Competition (3rd edn, Oxford University Press, 2014) para 8.461. 13 BASF (n 3) para 159. 14 ibid para 179 et seq. See also Kallaugher and Weitbrecht (n 1) 423. 15 COMP/35.587 PO Video Games; COMP/35.706 PO Nintendo Distribution; and COMP/36.321 Omega – Nintendo, Commission decision, 30 October 2002, paras 246 et seq, especially para 295. 16 Colombani et al (n 12) 3.99 et seq and 3.154. 17 BASF (n 3) para 160. See also Case C-49/92 P Commission v Anic Partecipazioni SpA ECLI:​EU:​ C:​1999:​356, para 84. 18 BASF (n 3) para 160; Anic Partecipazioni SpA (n 17) para 89. 19 Bathroom Fittings and Fixtures (n 2) para 794. See also Rhône-Poulenc (n 4) para 126; and HFB and others (n 4) para 186. 11 12

The concept of a Single and Complex Continuous Infringement  121 infringement in terms of Article 101 TFEU or whether two or more infringements are merely treated as one for the purpose of law enforcement. On the one hand, the GC refers to several ‘infringements’ for which liability is attributed to the perpetrators, yet at the same time these ‘infringements’ are then coined as one identical ‘infringement’ if the SCCI doctrine applies.20 One might conceive of SCCI as an emanation of the concept of an ‘agreement’ or a ‘restriction’ in terms of Article 101 TFEU. This would fit with the judgment in Team Relocations, where it was held that the concept of SCCI reconciled with the law of Article 101 TFEU.21 Yet competition takes place on markets so that, technically spoken, every market constitutes its own realm in which Article 101 TFEU can grow in significance. Therefore it is hard to see why several cartels on distinct markets in terms of Article 101 TFEU can, at the same time, be one unitary cartel in the spirit of the very same provision. Therefore, the SCCI doctrine should rather be considered as a matter of law enforcement. It is inspired by the criminal law concept of accessory and common purpose. Accessories can become liable for an infringement committed by a principal due to the contributions they rendered to this infringement. Similarly, the concept of common purpose allows one to ascribe liability to all participants in an infringement by an act of imputation. All national EU jurisdictions appear to know these concepts in the realm of criminal law and tort law, albeit with slightly different definitions. This view is reflected in the jurisprudence of the Court of Justice, where an SCCI is considered a type of ‘indirect participation’ in an infringement if a market concerned is one on which the SCCI perpetrator was not active personally.22 At the same time, the concept of SCCI is relevant for the sentencing of an individual addressee if and to the extent that it has committed multiple offences through one unitary action. Here, SCCI leads to the agency’s imposing one unitary fine capped at 10 per cent instead of a multitude of fines.23 This dimension of the SCCI doctrine is best described as a type of aggregate sentencing that ensures that the overall fine remains proportionate.

III.

ESTABLISHING AN SCCI

A.

Common Objective, Contribution and Awareness

From Polypropylene, over BASF, to the jurisprudence in Bathroom Fittings, the decisional practice on SCCI has evolved. As a starting point, the finding of an SCCI is not governed by a single metric. Rather, a comprehensive analysis must account for the nature of the overall conduct.24 According to the judgment of the Court of Justice in Team Relocations, the SCCI concept can be compressed into three main criteria: that an overall plan existed by which the participants pursued a common objective; that the firms made an intentional contribution to

BASF (n 3) para 160–61. Case C‑444/11 P Team Relocations NV and others v Commission ECLI:​EU:​C:​2013:​464, para 49. 22 Villeroy & Boch (n 7) para 167. 23 On that aspect of an SCCI, see Case T‑446/05 Amann & Söhne GmbH & Co. KG and others v Commission ECLI:​EU:​T:​2010:​165, para 150. See also Joshua (n 1) 465 et seq. 24 See e.g. Bathroom Fittings and Fixtures (n 2) paras 793–96. 20 21

122  Research handbook on cartels this objective; and that firms were aware or ought to have been aware of the anticompetitive conduct of the other cartelists.25 Mere knowledge about other cartelists restricting competition cannot suffice for a firm to become responsible for those actions. Rather, the jurisprudence requires that, beyond actual or potential awareness of these restrictions on other markets, a firm must have contributed to these restrictions through its participation in an overarching scheme. In Enichem Anic the Court of Justice alluded to a situation in which ‘the agreements and concerted practices found to exist, formed part of systems of regular meetings, target-price fixing and quota-fixing, and that those schemes were part of a series of efforts made by the undertakings in question in pursuit of a single economic aim, namely to distort the normal movement of prices’.26 Under such circumstances, it is possible, according to the Court, to consider that an undertaking that had taken part in such an infringement through conduct of its own which formed an agreement or concerted practice having an anti-competitive object for the purposes of Article [101(1) TFEU] and which was intended to help bring about the infringement as a whole was also responsible, throughout the entire period of its participation in that infringement, for conduct put into effect by other undertakings in the context of the same infringement.27

That is the case, in the Court’s view, ‘where it is established that the undertaking in question was aware of the offending conduct of the other participants or that it could reasonably have foreseen it and that it was prepared to take the risk’.28 In this line of reasoning, the GC stated in HFB29 that: According to the case-law, an undertaking may be held responsible for an overall cartel even though it is shown that it participated directly only in one or some of the constituent elements of that cartel, if it is shown that it knew, or must have known, that the collusion in which it participated was part of an overall plan and that the overall plan included all the constituent elements of the cartel.

The GC added that this legal argument ‘is not at odds with the principle that responsibility for such infringements is personal in nature, nor does it neglect individual analysis of the evidence adduced, in disregard of the applicable rules of evidence, or infringe the rights of defence of the undertakings involved’.30 When applying these principles, a distinction should be made between two types of cases. If all participants were active on the same markets, potential benefits of the illicit conduct will, most likely, have accrued to each participant in equal measure, since all perpetrators derived profits on the same affected markets. That provides an economic rationale to the SCCI allegation, if the SCCI requirements are met. The situation is different, however, if not all of the addressees were active on all markets covered by the SCCI allegation. In those instances, it becomes questionable in what way a firm, which was active on market A and has engaged

25 Team Relocations NV and others (n 21) para 50 et seq; and Joined Cases T‑204/08 and T‑212/08 Team Relocations NV et al. v Commission ECLI:​EU:​T:​2011:​286, para 35. On the requirement of awareness, see also Biolan (n 1) 383 et seq. 26 Anic Partecipazioni SpA (n 17) para 82. 27 ibid para 83. 28 ibid. 29 HFB and others (n 4) para 231. 30 ibid para 231.

The concept of a Single and Complex Continuous Infringement  123 in conspiracy there, can, at the same time through its participation in the ‘A-market cartel’, have contributed to the implementation of a restriction of competition relating to market B, on which it was not active, while other ‘A cartelists’ might have done business there and have coordinated their ‘B activities’. Nevertheless, the SCCI doctrine is also applied in this latter category. In that respect, the finding of a common objective requires thorough assessment. In Methacrylates, where the Commission for the first time combined several markets despite not all cartelists being active on each of them, it pointed to a set of facts on which the allegation of a common objective was based:31 The anticompetitive arrangements for the three relevant product categories showed ‘a number of common features’. A ‘core group of the same undertakings’ were involved in the anticompetitive arrangements. The ‘three major European producers’ participated in the arrangements for all three products. Also, there was a ‘direct link’ between the three products in that they shared the same chemical basis. Additionally, the three major European producers were ‘fully integrated’ so that their product portfolio covered all three segments. The Commission concluded that, therefore, these companies ‘paid great attention to the spill-over effects of the anti-competitive arrangements concluded for each of the products’. Moreover, ‘the cartelisation on one product automatically influenced the cost structure and/or prices of the other products’. The Commission also highlighted that meetings and contacts ‘were occasionally dedicated to more than one of the three’ products ‘with the venue for the meetings often being the same’. Besides, it was mentioned that ‘a number of representatives of the undertakings involved in the anti-competitive arrangements had responsibility for more than one product under investigation and were therefore aware or should have been aware of the existence of anti-competitive arrangements covering several products’. In Lombard Club the GC applied similar criteria in that it alluded to the fact that the restrictions were thoroughly organized by a high-level group of company representatives,32 and that organizational committees relating to different product segments held some of their meetings jointly.33 Moreover, the GC found it relevant that a top-level body was involved in the taking of fundamental decisions.34 The assessment of a ‘single objective’, therefore, involves a comprehensive assessment of such criteria.35 Some authors argue that the finding of a single objective can be based on a combination of several such criteria without dependence on a specific single criterion.36 In any event, the Commission requires some degree of knowledge or potential awareness of a firm about the effects that an SCCI could have on another market for it to become liable under the SCCI doctrine for those effects.37 In the absence of such (actual or constructive) awareness, a firm must only answer for the part of the infringement in which it was personally engaged.38

Methacrylates (n 9) para 223. Raiffeisen Zentralbank Österreich AG (n 6) para 123. 33 ibid para 120. 34 ibid paras 114–18. 35 On such comprehensive appraisal, see Case T‑357/06 Koninklijke Wegenbouw Stevin BV v Commission ECLI:​EU:​T:​2012:​488, paras 33–71. See also Romić (n 1) 171 et seq. 36 Colombani et al (n 12) para 8.462. 37 See e.g. Bathroom Fittings and Fixtures (n 2) paras 789, 861. 38 Joined Cases C‑293/13 P and C‑294/13 P Fresh Del Monte Produce Inc. v Commission ECLI:​EU:​ C:​2015:​416, paras 156 et seq and para 159; and Sigma Tecnologie di rivestimento Srl (n 10) para 45. See 31 32

124  Research handbook on cartels B.

On the Requirement of Complementarity

Besides these criteria, it is questionable whether the different parts of a collusive conduct must complement each other in an objective, functional sense for an SCCI to arise. After all, it is one thing to be aware of a competitive distortion on market A, and a quite different matter whether a conspiracy relating to market B or an overarching agreement lent itself to promoting this competitive distortion on market A. In order to establish sufficient links between two or more restrictions of competition, the GC in BASF therefore applied a complementarity test.39 According to the GC, an SCCI ‘cannot be determined by a general reference to the distortion of competition’, since such distortion is a consubstantial element of any relevant conduct under Article 101 TFEU.40 Rather, for complementarity to be given, the GC required that each conduct ‘was intended to deal with one or more consequences of the normal pattern of competition and, by interacting, contributed to the realisation of the set of anti-competitive effects intended by those responsible, within the framework of a global plan having a single objective’.41 The same complementarity test was applied in Almamet.42 The complementarity test ensured that mere knowledge about a third-party infringement could not suffice to incur personal liability. Yet the more recent jurisprudence has jettisoned this requirement and instead focused on the mere finding of a ‘single objective’. While the Court of Justice, in Villeroy & Boch, maintained that a ‘general reference to a distortion of competition on the markets concerned’ cannot suffice for an SCCI,43 the Commission can nevertheless, in the Court’s view, establish a single objective if ‘the various instances of conduct complained of pursued the same goal’, namely ‘coordinating’ the cartelists conduct with respect to several product categories to wholesalers.44 The allegation of such a single objective may, according to the Court of Justice, be substantiated by reference to various objective factors, such as the central role played by the wholesalers in the distribution chain, the features of that chain, the existence of umbrella associations and cross-product associations, the similarities in the way the collusive arrangements were implemented and the material, geographic and temporal overlap between the practices concerned.45

The Court of Justice held that ‘in those circumstances’ there is ‘no need to establish a link of complementarity between the practices complained of, given that a single and continuous infringement may be imputed to undertakings that are not in competition and does not require the relevant markets to be systematically defined’.46 If such a single objective can be established, it suffices, in the Court’s view, for a firm to become liable for the entirety of market also CFI of 14 May 1998, Case T-295/94 Buchmann ECLI:​EU:​T:​1998:​88, para 119; and Joined Cases T-25/95 etc. Cimenteries CBR and others v Commission ECLI:​EU:​T:​2000:​77, paras 4041 et seq. 39 BASF (n 3) especially paras 179, 181 and 209. 40 ibid para 180. 41 ibid para 179. 42 Case T‑410/09 Almamet GmbH v Commission ECLI:​EU:​T:​2012:​676, para 154. 43 Villeroy & Boch (n 7) para 166. The same was emphasized by the Court in Team Relocations NV and others (n 21) para 57. 44 Villeroy & Boch (n 7) paras 163 et seq. 45 ibid para 166. 46 ibid para 167.

The concept of a Single and Complex Continuous Infringement  125 effects if ‘it was aware of all the offending conduct planned or implemented by the other cartel members in pursuit of the same objectives’, or if it ‘could reasonably have foreseen that conduct and was prepared to take the risk’.47 This jurisprudence, however, if pushed to its extreme, may lead to a firm becoming liable for distortions on markets where it was not active, of which it was not aware and which it could not have influenced or prevented under any circumstance. Eventually, such type of liability can raise issues under the principle of personal responsibility. Therefore, it must be applied cautiously. The mere fact that, albeit in different markets, similar product categories were affected should not be considered an argument in itself that a firm had the objective to help other cartelists implement restrictions on markets on which this firm was not active. C.

The Requirement of Individual Assessment of Each Participant

SCCI liability must be assessed individually for each participant.48 While the scope of individual involvement can therefore deviate between SCCI participants, it is not required for an SCCI to exist that each participant was active within the entire scope of the unitary infringement. Accordingly, the Court in Team Relocations concretized that it is not necessary to establish that an ‘undertaking was or should have been aware of the offending conduct of the initial participants in the infringement or that it adhered to that infringement from the outset’.49 Also, it is not the case that the ‘condition of awareness can be established only if that undertaking contributed to the single and continuous infringement in a way identical to that initially put in place’.50 Rather, according to the Court, ‘an undertaking may be held responsible for a single and continuous infringement even if it did not participate in all of the offending conduct of which it is made up’.51 About the intentional contribution requirement, the Court stated that ‘it does not mean that the intentional contribution to those common objectives can be established only where the undertaking concerned has contributed to those common objectives since the start of the infringement or on condition that it pursued those objectives in ways identical to those put into effect when the infringement commenced’.52 Also, finding an SCCI is not prevented if the structure of the cartel was altered during its lifetime,53 or if the types of contribution made by each cartelist differ from each other.54 On the other hand, the GC in Quinn Barlo stated that the mere fact that an infringement related to an economic field in which an SCCI was found does not suffice to hold a firm liable for the entire SCCI. Mere ‘objective links between that infringement and the agreement’ in which an undertaking participated, ‘such as belonging to the same economic sector’, will not suffice if it has not been established that the firm ‘was aware of the existence of such a single

ibid para 167. J. Faull, L. Kjølbye, H. Leupold and A. Nikpay, ‘Article 101’, in Faull and Nikpay (n 12) para 3.95. 49 Team Relocations NV and others (n 21) para 54. 50 ibid. 51 ibid, referencing Case C‑441/11 P Commission v Verhuizingen Coppens ECLI:​EU:​C:​2012:​778, paras 43 to 45. 52 Team Relocations NV and others (n 21) para 56. 53 Methacrylates (n 9) para 218. 54 Anic Partecipazioni SpA (n 17) para 79. 47 48

126  Research handbook on cartels infringement or that it could reasonably have foreseen it and was prepared to take that risk’,55 and that the further requirements for an SCCI are fulfilled.

IV.

IMPLICATIONS OF AN SCCI FOR PUBLIC ENFORCEMENT

A.

On Discretion in Applying the SCCI Doctrine

The Commission speaks of being ‘entitled’ to apply the SCCI doctrine.56 Some authors claim the Commission has the ‘discretion’ to invoke it.57 It is expedient to become more specific here. The assessment of complex economic facts leaves the Commission some room for its own appraisal,58 and this also holds true for the assessment of an SCCI. Yet the legal SCCI concept, and the legal requirements underpinning it, are not matters which the Commission or the courts could possibly choose to deny in a given case. The union organs, therefore, do not have a discretion whether to apply the SCCI doctrine or to leave it aside. This is echoed by the judgment of the GC in Tokai Carbon et al. The GC assessed whether the imposition of three fines rested on ‘objective grounds’59 in that the requirements for the finding of an SCCI were absent. This part of the judgment was based on the premise that the Commission would have been entitled to impose three different fines ‘provided that the applicant had committed three separate infringements of Article [101(1) TFEU]’.60 Eventually, the GC concluded that the facts of the case did not give rise to an SCCI so that the triple fining was upheld. This demonstrates that the Commission is bound by the concept if the requirements for an SCCI are fulfilled. The GC therefore concluded that ‘the Commission does not have unlimited discretion in finding that the rules on competition have been infringed, or in determining whether the various unlawful acts constitute a single continuous infringement or a number of separate infringements, or in setting fines for those infringements’.61 B.

Evidence and Limitation Periods

The SCCI doctrine can help the Commission to ‘bypass’ evidence gaps in long-term multi-market offences.62 Where infringements on several interrelated markets have taken place, a lack of evidence for a certain period in relation to one market can be bridged by suffi-

Case T‑208/06 Quinn Barlo v Commission ECLI:​EU:​T:​2011:​701, paras 149 et seq. Bathroom Fittings and Fixtures (n 2) para 784. 57 Colombani et al (n 12) para 8.462. 58 Case C-413/06 P Bertelsmann and Sony Corporation of America v Impala ECLI:​EU:​C:​2008:​392, para 144; Case C-12/03 P Commission v Tetra Laval ECLI:​EU:​C:​2005:​87, para 38; and Joined Cases C‑68/94 and C‑30/95 France and others v Commission (Kali & Salz) ECLI:​EU:​C:​1998:​148, paras 223, 224. See also Case T-342/07 Ryanair v Commission ECLI:​EU:​T:​2010:​280, para 136; and van der Woude (n 1). 59 Joined Cases T-71/03 etc. Tokai Carbon Co. Ltd and others v Commission ECLI:​EU:​T:​2005:​220, paras 124. 60 ibid para 118. 61 Amann & Söhne GmbH & Co. KG (n 23) para 130. 62 See also Alexiadis et al (n 1) 3. 55 56

The concept of a Single and Complex Continuous Infringement  127 cient evidence relating to that period of an infringement on another market, provided that both markets are part of the same SCCI. In a similar vein, the application of the SCCI doctrine can prevent infringements on particular markets from becoming time barred if the restriction of competition was continued on other markets. Conversely, to deny an SCCI can benefit the addressee to the extent that some of the potentially affected infringements will be time barred if treated separately.63 According to the jurisprudence, a cartelist – despite its joining and leaving the cartel ‘from time to time’ – will not necessarily be treated as entering into a new agreement ‘with each change in participation’.64 It is necessary, therefore, to determine whether the infringement was continuous, repeated or terminated. C.

Calculation of the Fine

1. Affected markets The finding of an SCCI leads to the calculation of one unitary sanction for the entirety of distortions on all markets that are covered. This means that the restrictions on all markets must be treated as one unitary infringement for the determination of the basic amount.65 The 10 per cent cap of Article 23 of Regulation 1/2003 is not multiplied.66 Effectively, this is an emanation of the principle of aggregate sanctioning. If an SCCI cannot be established, the Commission will therefore impose several fines, as in Tokai Carbon.67 The Court of Justice has stated that the amount of the fine must reflect the degree of direct involvement of a firm in an SCCI infringement: The fact that an undertaking has not taken part in all aspects of an anti-competitive scheme or that it played only a minor role in the aspects in which it did participate must be taken into consideration when the gravity of the infringement is assessed and if and when it comes to determining the fine.68

Where a firm is held responsible for infringements on markets on which it was not active, the limited role is accounted for when applying the Fining Guidelines. Under the 1998 Guidelines,69 the Commission based its calculation on the turnover with the cartelized good in order to apply differential treatment, so that the non-activity of a cartelist on a cartelized market, that was part of an SCCI, would have translated into a lower fine for this addressee.70 Also, the Commission, under the 1998 Fining Guidelines, granted a reduction of 25 per cent

BASF (n 3) para. 158; and Case C-235/92 P Montecatini SpA v Commission ECLI:​EU:​C:​1999:​362, para 196. 64 HFB and others (n 4) para 234. 65 Bathroom Fittings and Fixtures (n 2) paras 1194–1209. 66 ibid paras 1182–84 and 1192. 67 Tokai Carbon Co. Ltd (n 59) paras 117–24. 68 Anic Partecipazioni SpA (n 17) para 90; see also Case T-59/99 Ventouris Group Enterprises SA v Commission ECLI:​EU:​T:​2003:​334, para 219. On the assessment of facts relating to the role and extent to which a firm engaged in a cartel, see also Case C‑440/11 P Commission v Stichting Administratiekantoor Portielje ECLI:​EU:​C:​2013:​514, paras 115 et seq. 69 Guidelines on the Method of Setting Fines Imposed Pursuant to Article 15(2) of Regulation 17 and Art. 65(5) of the ECSC Treaty [1998] OJ C9/3. 70 To that effect, Methacrylates (n 9) para 333. 63

128  Research handbook on cartels on the basic amount to a firm due to its limited role in the cartel if it was involved only in one of three cartelized markets.71 Under the 2006 Fining Guidelines,72 the basic amount is directly based on ‘the value of the undertaking’s sales of goods or services to which the infringement directly or indirectly relates’, so that whether an SCCI cartelist was active on all or merely on some of the interrelated markets will directly bear on the basic amount.73 In a subsequent step, the Commission determines a percentage of the turnover affected in order to account for the gravity of the infringement.74 It has been suggested that, where an SCCI cartelist is not active on all markets, the Commission will ‘likely’ consider the fact that this cartelization was part of an SCCI scheme comprising other markets as a circumstance that justifies an increase in the percentage, since the overall infringement ‘looks worse than any individual violation committed therein’.75 In addition, it is argued that the Commission has ‘leeway’ to take an SCCI into consideration when determining the gravity of the infringement.76 2. Duration Another factor that bears on the severity of the fine is the duration of the infringement. In BASF, the GC stated that an ‘anti-competitive agreement cannot, in principle, be regarded as a means of implementing another agreement which has already come to an end’.77 The GC highlights that the cartel period is not exclusively determined by the period ‘during which an agreement was in force’, but by reference to the period in which the firms acted in an anticompetitive way.78 This must be assessed individually for each cartelist. Such was the Commission’s practice under the 1998 Fining Guidelines.79 In a similar spirit, the 2006 Fining Guidelines stipulate that the basic amount is multiplied by the number of years of the infringement for each addressee. The reliance on the SCCI concept can therefore extend the overall cartel period in which a participant was involved. It must be recalled that an SCCI can be used by the Commission to bridge periods in which a cartelist was inactive so as to compress all contributions, despite interruptions, into one unitary participation. Despite the impact on limitation periods, this can increase the factor by which the basic amount will be multiplied and therefore inflate the fine. In that regard, the more recent decisional practice makes a distinction between ‘continuous’ infringements and ‘repeated’ infringements.80 While both variants constitute one unitary SCCI, the distinction bears on the determination of the individual participation of a cartelist in the infringement. In a continuous infringement, the undertaking is assumed to have engaged

ibid para 335. Guidelines on the Method of Setting Fines Imposed Pursuant to Article 23(2)(a) of Regulation No 1/2003, OJ 2006 C 210/2 (hereinafter ‘2006 Fining Guidelines’) para 13. 73 Seifert (n 1) 552. 74 Fining Guidelines 2006 (n 72) para 19. 75 Seifert (n 1) 552. 76 ibid. 77 BASF (n 3) para 191. 78 ibid para 187. 79 Methacrylates (n 9) paras 306–13; and PO Video Games, PO Nintendo Distribution and Omega – Nintendo (n 15) para 109. 80 Joined Cases T‑147/09 and T‑148/09 Trelleborg Industrie SAS and others v Commission ECLI:​ EU:​T:​2013:​259, paras 72–94, especially 87 et seq. 71 72

The concept of a Single and Complex Continuous Infringement  129 uninterruptedly for the entire period, while in a repeated infringement the firm is assumed to have paused, with these periods of inactivity not being counted for the determination of the duration. The category of ‘repeated’ infringement, therefore, effectively means that interruptions do not lead to a termination of previous engagements so that limitation periods will not run, while the total amount of the fine must nonetheless reflect the inactive periods adequately. In Coppens, the Court confirmed that the fact that an undertaking did not commit to the full extent of the SCCI warrants a reduction of the fine.81 In conclusion, there are three distinct ways in which passive interim periods can impact on the fine: the involvement may nonetheless have been ‘continuous’, it may have been ‘interrupted’ or it may have been terminated completely.82 Limitation periods will only start in the latter case. The GC emphasizes the importance of considering the facts of the matter in order to distinguish between these three types.83 On a conceptual level, what the continuous and the repeated infringement have in common is that the undertakings involved have the same common objective during the entire period irrespective of whether they continuously or repeatedly contributed to the pursuit of this goal.84 That is the constitutional element in finding a ‘single’ infringement.85 To establish the existence of a single objective in terms of duration, a variety of factors must be taken into consideration, such as ‘the identical nature of the objectives of the practices at issue, of the goods concerned, of the undertakings which participated in the collusion, of the main rules for its implementation, of the natural persons involved on behalf of the undertakings and, lastly, of the geographical scope of those practices’.86 If such is established, the nature of the participation – provided the other requirements for an SCCI are fulfilled – is either continuous or repeated. In order to distinguish between the continuous or repeated nature of the SCCI, it must be assessed whether there were periods of ‘interruptions’ in the participation of a cartelist. This mandates ‘an overall evaluation of all the relevant evidence and [indicia]’.87 As a starting point, the jurisprudence requires ‘objective and consistent indicia’ underpinning the allegation that an infringement extended over a longer period despite intertemporal periods for which evidence is lacking.88 At the same time, however, the jurisprudence has stated that there is a presumption of continuous participation, so that it is upon the undertaking to demonstrate that it was inactive during a certain period.89 Accordingly, mere lack of evidence for a participation during a certain period will not exonerate the undertaking from the allegation of

81 See Case T‑587/08 Fresh Del Monte Produce, Inc. v Commission ECLI:​EU:​T:​2013:​129, para 649, and paras 814–17; Fresh Del Monte Produce Inc. v Commission (n 38) para 152; and Commission v Verhuizingen Coppens (n 51) para 45. See also Colombani et al (n 12) para 8.459. 82 On the requirements that must be fulfilled for a firm to terminate its participation in a cartel, see Case C‑68/12 Protimonopolný úrad Slovenskej republiky v Slovenská sporiteľňa a.s. ECLI:​EU:​C:​2013:​ 71, para 27; Case C‑290/11 P Comap SA v Commission ECLI:​EU:​C:​2012:​271, paras 47 et seqq; and Colombani et al (n 12) para 8.465. 83 Trelleborg Industrie SAS (n 80) para 94. 84 ibid para 88. 85 On the notion of ‘single’ in that regard, see Almamet GmbH (n 42) para 152. 86 Trelleborg Industrie SAS (n 80) para 88. 87 Case C‑113/04 P Technische Unie BV v Commission ECLI:​EU:​C:​2006:​593, para 167. 88 ibid para 169. 89 Trelleborg Industrie SAS (n 80) para 87.

130  Research handbook on cartels a continuous infringement according to the case law.90 This does not mean, however, that the addressee must give full evidence of its non-involvement during a certain period. Mere ‘indicia’ can suffice to deny a continuous infringement.91 Depending on the facts of the case, especially on the relation between periods for which evidence exists and those for which it does not, even gaps of nine months or longer can be acceptable when finding a continuous infringement.92 An undertaking may even be considered to have conspired after the Commission has made inspections, if there is sufficient evidence to assume the engagement in an SCCI to that effect.93 The nature of the agreement must also be taken into account. While a price-fixing cartel, according to the jurisprudence, usually depends on a more closely knit collusion fabric, for a market-sharing agreement it might suffice to convene less frequently so that inactive intermediate periods will not necessary prevent from finding a continuous infringement.94 The Commission, in its decisional practice, explicitly honours the principle of in dubio pro reo if, for certain periods, the evidence and indicia do not support a continuous infringement, so that the dates most favourable for the undertaking will be assumed.95 If sufficient evidence and indicia are absent that would allow one to find a continuous infringement despite a gap period; a repeated infringement can be given, provided that the re-engagement after the interruption is an expression of the same infringement that was committed before. It speaks for a repeated infringement and against a continuous one if the gap period was much longer than the intervals were that usually laid between cartel meetings. The GC in IMI stated that a 16-month period of cartel silence was more than one year longer than the usual terms between contacts, which sufficed to deny a continuous infringement during that period.96 A repeated infringement can relate to the entire SCCI, that is, all participants interrupted their engagement. In addition, it may relate to an individual participant that interrupted its participation, while the other cartelists engaged in a continuous infringement.97 D.

Leniency Applications

The SCCI doctrine can affect leniency applications. If an applicant falls short of providing the requested amount of information about one of the relevant markets affected by a cartel, this

ibid. ibid. 92 Case T‑83/08 Denki Kagaku Kogyo Kabushiki Kaisha v Commission ECLI:​EU:​T:​2012:​48, para 124. 93 Case T‑385/06 Aalberts Industries NV v Commission ECLI:​EU:​T:​2011:​114, paras 83–106, where such was denied in the given case; confirmed upon appeal in Case C‑287/11 P Commission v Aalberts Industries NV and others ECLI:​EU:​C:​2013:​445. 94 Case T‑439/07 Coats Holdings Ltd v Commission ECLI:​EU:​T:​2012:​320, paras 138–55, where, based on this line of argumentation, a gap period of more than 11 years did not hinder a continuous infringement. 95 Case COMP/38.899 Gas Insulated Switchgear, Commission decision, 24 January 2007, para 448. 96 See the IMI judgment, where the GC changed the Commission’s finding of a continuous infringement despite a 16-month gap into a repeated infringement so that the period of inactivity was excluded from the calculation of the fine: Case T‑18/05 IMI plc and others v Commission ECLI:​EU:​T:​2010:​202, para 96. 97 See Gas Insulated Switchgear (n 95) paras 425 et seq and 448. See also Colombani et al (n 12) para 8.462. 90 91

The concept of a Single and Complex Continuous Infringement  131 can lead to the loss of the immunity/leniency benefit for the entire SCCI. Leniency applicants, therefore, will usually relate their submissions to all potentially affected markets.98 It can however be hard to predict which markets the Commission will eventually compress into an SCCI. Therefore, firms should assess the potential scope of an SCCI thoroughly before making a leniency submission. If firms are not active on a market that falls within the scope of an SCCI, it must suffice, however, for them to submit all the information they have about these third-market restrictions, even if that is not very much due to their own absence. E.

Rights of Defence

The addressees of a Statement of Objections (‘SO’) based on an SCCI can find themselves in a predicament if they want to prove that they were not involved in an infringement during a certain time period or on a market on which they were not active.99 Effectively, they are precluded from any argument that their conduct did not have, as its object or effect, a restriction of competition there, for the finding of an SCCI substitutes for their direct involvement in the infringement in that area of the SCCI. The jurisprudence has nonetheless ruled that the concept of SCCI is compatible with Article 101 TFEU and Article 53 of the EEA Agreement and the principle of due process,100 the principle of personal responsibility and the rights of defence of the undertakings involved.101 Also, the GC found the SCCI doctrine to comply with Article 7 (nulla poena sine lege) of the European Convention on Human Rights (‘ECHR’), even though the EU antitrust legislation remains silent as to the legal foundation and criteria of an SCCI.102 The GC found a sufficient legal underpinning in the published case law on SCCI.103 According to the jurisprudence, for the rights of defence to be observed it suffices if the firms can exercise those rights in respect of the elements that give rise to an SCCI, namely the ‘common objective, their intention to contribute to the infringement as a whole by their own conduct and their knowledge of the conduct of other participants or its foreseeability and the acceptance of the related risk’.104 In order to observe the addressee’s rights of defence, the Commission must, in its SO, therefore address whether it intends to rely on the SCCI doctrine. That point was raised by the applicant in BASF.105 Even though the GC did not elaborate on this argument in its judgment, such a duty to disclose the intention of invoking the SCCI doctrine is mandatory, since the applicability can be detrimental to the addressee, namely with respect to limitation periods. In Villeroy & Boch, the Court of Justice held that pieces of evidence can support the finding of a cartelist to have committed an SCCI infringement on a relevant market even though the exact same piece of evidence, viz. a leniency statement, was considered insufficient proof for an infringement on that market by the same chamber of the GC upon application of another cartelist.106 That line of reasoning, however, is objectionable under the in dubio pro reo prin See e.g. Bathroom Fittings and Fixtures (n 2) paras 126–128; and Seifert (n 1) 553. Seifert (n 1) 553. 100 Villeroy & Boch (n 7) para 163. 101 Anic Partecipazioni SpA (n 17) para 89; and HFB and others (n 4) para 231. 102 Amann & Söhne GmbH & Co. KG (n 23) paras 125–33. 103 ibid para 133. 104 Anic Partecipazioni SpA (n 17) para 89. 105 BASF (n 3) paras 133, 157 and 211. 106 Villeroy & Boch (n 7) para 168. 98 99

132  Research handbook on cartels ciple. It is hard to see how a piece of evidence can be sufficiently robust to give evidence for an SCCI infringement on a market in relation to one alleged perpetrator, if the same piece of evidence was considered insufficient by the same court in the same case with respect to another cartelist. If the Court has doubts as to the probative value of this piece of evidence in an appeal lodged by one addressee of an SCCI infringement, the Court cannot, as a matter of logic, be sufficiently convinced of the probative value with respect to the same allegations in relation to another cartelist of the same SCCI infringement.

V.

IMPLICATIONS OF AN SCCI FOR PRIVATE ENFORCEMENT

A.

Scope of Liability for Damages on Non-Contested Markets

To the extent that an SCCI imputes liability for product markets, geographic markets or time periods in which a firm did not sell goods, the question arises about civil liability vis-à-vis customers on those markets or during these periods. Some authors argue that such liability for third-market effects/inactive time periods can arise if, and to the extent that, the SCCI covers those areas and the relevant firm was held responsible for this SCCI.107 Yet even though a firm is accountable under Article 23 of Regulation 1/2003 for an SCCI when receiving a fine, it is questionable if this translates into a corresponding civil liability without more. Since SCCI should be conceived of as a specific sanctioning concept rooted in the realm of law enforcement rather than a matter of the substantive notions of Article 101 TFEU, the answer to the question hinges on whether the laws on antitrust damages provide an identical scope of liability as those sanctioning concepts used under Regulation 1/2003. Several arguments deserve consideration. In Skanska, the Court of Justice ruled that a successor in a business can become liable for infringements committed by the previous owner of the business in analogy to the successor liability rules governing the imposition of fines.108 One might want to argue from here that any entity that incurs responsibility for an antitrust fine is equally liable for the entirety of damages precipitated by the infringement which was sanctioned. Yet such construal of the Skanska judgment would overstretch the ratio decidendi. Skanska was about the notion of an undertaking and the effects of legal changes on its identity. The SCCI doctrine, however, deals with concepts that relate to accessory and common purpose in terms of criminal law as well as aggregate sanctioning. These matters do not relate to the wording and interpretation of Article 101 TFEU. They concern the sanctioning of infringements of Article 101 TFEU. It is very clear, however, that the imposition of fines is governed by principles that are not, by definition, identical with those that govern damages liability. If it were different, compensation could be awarded on grounds of Regulation 1/2003, and the EU Damages Directive109 would not exist. The Court of Justice has emphasized that, as a general rule, damages liability is shaped by the

Seifert (n 1) 553. Case C‑724/17 Skanska ECLI:​EU:​C:​2019:​204. 109 Directive 2014/104/EU of the European Parliament and of the Council of 26 November 2014 on certain rules governing actions for damages under national law for infringements of the competition law provisions of the Member States and of the European Union [2014] OJ L349/1. 107 108

The concept of a Single and Complex Continuous Infringement  133 national legislator110 within the boundaries of effet utile and the Damages Directive. Neither of them requires, however, to establish a rule that provides for a civil liability in analogy to the SCCI doctrine. Not even the national laws on antitrust sanctions for infringements of Article 101 TFEU are required to provide an SCCI rule. The ECN+ Directive,111 while making extensive stipulations about the national enforcement regime, does not mention this concept either. One might want to rely on the binding effect of a Commission decision for a damages case in order to establish civil liability in analogy to the scope of an SCCI fine decision. Yet the binding effect which a Commission decision precipitates is confined to the fact that the addressee of the decision participated in a cartel. It does not determine the amount of damages that a firm must compensate. This is not to say that an SCCI perpetrator must never be held liable for damages on such markets on which it was not active. Rather, this depends on the national principles governing damages claims, notably in the field of tort law and antitrust damages. It is conceivable that some EU member states will apply principles akin to those established by the Commission in its SCCI doctrine under Article 23 of Regulation 1/2003. Yet it is equally conceivable that the national provisions deviate from this doctrine. In Germany, for example, for someone to become jointly and severally liable for cartel damage to which this person has not contributed a physical cause, it is relevant whether this person acted in concert with a direct perpetrator.112 That, in turn, requires more than mere knowledge about the other person’s infringement. Rather, it is necessary to find some type of contribution that was directed at enabling or facilitating the infringement. Whether such a link can be established depends on the facts of a case. A mere reference to an SCCI decision cannot substitute for an assessment of these private law principles. B.

Recourse Claims for Joint and Several Liability

In any event, the fact that the cartelist was not active on a market can bear on the amount of recovery to be contributed by the other cartelists. Article 11(5) of the EU Damages Directive stipulates that the amount to be received by cartelists from their co-conspirators ‘shall be determined in the light of their relative responsibility for the harm caused by the infringement of compe­tition law’. This ought to allow for a consideration of the amount of harm precipitated on each relevant market. If customers on a market on which the cartelist was not active can, depending on the relevant private law principles and the circumstances, seek compensation from this cartelist, the latter carries a lesser ‘relative responsibility’ for their loss than his co-conspirators who were active on this market.

See Case C‑557/12 Kone ECLI:​EU:​C:​2014:​1317, para 24. Directive (EU) 2019/1 of the European Parliament and of the Council of 11 December 2018 to empower the competition authorities of the Member States to be more effective enforcers and to ensure the proper functioning of the internal market [2019] OJ L11/3. 112 See § 830 of the German Civil Code. 110 111

134  Research handbook on cartels

VI. CONCLUSION This chapter evaluated the EU-level doctrine of the ‘Single and Complex Continuous Infringement’. It explained the nature of the concept of SCCI, emphasizing in the process that the concept covers a range of situations in which a multitude of actions and/or effects on one market or more markets are compressed into one unitary finding of an infringement in terms of law enforcement. The chapter considered how one goes about establishing an SCCI, focusing on issues such as common objective, contribution and awareness and complementarity. The need to assess SCCI liability individually for each participant was also highlighted. Finally, the chapter outlined various implications of an SCCI for public enforcement and for private enforcement.

8. Lawful cartels Or Brook

I. INTRODUCTION The prohibition on hard-core cartels is one of the rare areas of consensus unifying competition law regimes across the globe. Cartels are dubbed as ‘the supreme evil of antitrust’1 and the ‘most egregious violations of competition law’.2 Nevertheless, not every agreement among competitors is deemed to be unlawful. In fact, for many years, various types of cartels have been perceived as a legitimate business strategy and as an engine for economic growth. Even today, a fine line exists between lawful and unlawful cartels. This chapter begins by tracing the historical shift in the international approach towards cartels (Section II). It demonstrates that the prohibition on cartels is not just a question of form, but also one of substance;3 namely, that even after the emergence of international scepticism towards cartels from the 1940s and especially following the 1990s, there is a wide array of acceptable forms of cartel or cartel-like agreements that are presumed to create efficiencies or promote industrial and social policies.4 Although some arrangements are accepted by multiple jurisdictions, to date there is no common framework that guides or can explain the assessment of lawfulness (Section III). Competition authorities, courts, governments and parliaments have often adopted rules incrementally, by responding to case-specific challenges. A notable divergence persists not only in the justifications invoked for lawfulness, but also in the legal, economic or political tests used to assess similar arrangements. To structure the discussion, this chapter distinguishes between two main types of justification.5 On the one end of the scale, there are arrangements that are presumed to create economic benefits, that is, cost and qualitative efficiencies enjoyed by the direct or indirect users of the relevant product or service. These benefits either directly affect prices or provide additional non-price value for consumers, such as new products, better quality or greater variety. On the other end of the scale, there are arrangements that are warranted by non-economic benefits. Those non-pecuniary benefits are more subjective and are not directly related to the characteristics of the relevant product or service. Accepting cartels on the basis of Verizon Communications v Law Offices of Curtis V. Trinko, 540 US 398 (2004) 408. See OECD Council, ‘Recommendation Concerning Effective Action against Hard Core Cartels’ (1998) OECD/LEGAL/0294; and OECD Council, ‘Recommendation Concerning Effective Action against Hard Core Cartels’ (2019) C(2019)87. 3 N. Dunne and I. Maher, ‘The “Acceptable” Cartel? Horizontal Agreements Within EU Competition Law: Introduction’ (2002) 5(3) Antitrust Bulletin 335, 335. 4 J. Fear, ‘Cartels’ in G. Jones and J. Zeitlin (eds), The Oxford Handbook of Business History (Oxford University Press, 2008) 273. 5 This distinction is based on the definitions presented by a group of experts in Office of Fair Trading, Article 101(3): A Discussion of Narrow versus Broad Definition of Benefits. Discussion Note for OFT Breakfast Roundtable (12 May 2010) paras 3.2–3.17. 1 2

135

136  Research handbook on cartels non-economic benefits entails balancing the harm caused to consumers against benefits created for society as a whole – such as promoting regional, national or international industrial policies; fostering employment; addressing situations of crisis; or protecting the environment. This balancing exercise is more contentious, often meriting socio-political value judgements and carrying distributional justice effects. This chapter uncovers a disconnect between the underlying economic and non-economic benefits invoked to justify cartels and the legal, economic or political strategies employed to assess their lawfulness (Section IV). It calls to rationalize the assessment of lawfulness by shifting the focus away from the types of cartels (for example, horizontal cooperation, export or crisis cartels) towards ensuring an alignment between the economic and non-economic justifications invoked and the assessment strategies. Such alignment, the chapter concludes, could enhance the effectiveness, transparency, legal certainty, democratic legitimacy and political accountability of the competition law regimes (Section V).

II.

BRIEF HISTORY OF LAWFUL CARTELS

Although attempts to collude on market prices and outputs are probably as old as markets themselves,6 the overwhelming opposition to cartels is a relatively recent phenomenon. This section illustrates that up until the mid-1940s, certain forms of cartels were viewed by members of industries and governments alike as a legitimate business strategy having a positive function on the organization of markets and generating societal benefits. The shift in the international approach towards cartels was mostly the result of strong American influence following the Second World War, which was later amplified and adjusted by the growing importance of EU competition law from the mid-1990s. A.

From the Late 1800s to the First World War: Cartels as a Social-Economic Institution

Cartels seeking to directly restrain outputs and sales of specific products first emerged as a common industrial organization model in the 1870s. During a time of falling world prices for agricultural and industrial products, most industrialized nations imposed high tariffs to protect their markets against foreign competition. Those tariffs, in turn, induced producers to enter into national cartels to maintain prices at profitable levels and avoid price swings.7 Banks too colluded to safeguard credit and local employment, many times receiving the support of state administration.8 Gradually, the cartel business model gained momentum, spreading from Europe and the US and becoming a defining feature of the global economy.9

6 In the words of R. Piotrowski, Cartels and Trusts: Their Origin and Historical Development from the Economic and Legal Aspects (G. Allen & Unwin, 1933). 7 C. Freedeman, ‘Cartels and the Law in France before 1914’ (1988) 15(3) French Historical Studies 462, 462; and S. Webb, ‘Tariffs, Cartels, Technology, and Growth in the German Steel Industry, 1879 to 1914’ (1980) 40(2) Journal of Economic History 309, 310. 8 H. Schröter, ‘Cartelization and Decartelization in Europe, 1870–1995: Rise and Decline of an Economic Institution’ (1996) 25(1) Journal of European Economic History 129, 133. 9 ibid.

Lawful cartels  137 The first blanket opposition to cartels appeared in the US, where the Sherman Act of 1890 stated that ‘every contract, combination, or conspiracy in restraint of trade […] is declared to be illegal’.10 Despite this broad wording, prohibiting all restraints to trade, the Supreme Court of Justice interpreted the prohibition as being limited to ‘unreasonable’ restraints. Accordingly, US antitrust law distinguishes between agreements that are per se restrictive and are irrebuttably presumed to be unlawful, and other agreements whose legality is determined under the rule of reason test.11 In its early years the US rule of reason was based on a legal proportionality test, examining whether a restraint was reasonably necessary to attain a legitimate agreement, or whether less restrictive alternatives were permissible. Yet in 1918 the Supreme Court embraced a cost–benefit approach, by which the legality of an agreement hinges on its net competitive effect.12 The per se category has been progressively narrowed, currently extending to naked price-fixing and market division agreements and to limited types of boycotts, concerted refusals to deal and tying arrangements.13 Nonetheless, the American approach still represents the principle of prohibition in the assessment of cartels, that is to say, a total ban that is subject to only a few exceptions.14 In parallel, up until the mid-1940s, most other countries continued to perceive cartels as generating a host of economic and non-economic benefits.15 Cartelization was not only accepted, but was regarded as a legitimate form of a capitalistic economy,16 carrying stabilizing effects on the public interest and economic life.17 In some European countries, cartels were also linked to a national cooperative ethos, positioning them as an adequate response to the threat of class conflict and social disintegration.18 This approach was fostered during the First World War, at the time when industrial production was first recognized as a key element of military success.19 Cartels were endorsed by governments, which preferred to control the production of a limited number of cartels rather than supervising numerous independent firms.20 They were also encouraged for industrial policy purposes, as a means to promote foreign trade.21 Even the US, which was guided by the

26 Stat. 209 (1890), sec 1. Standard Oil Co. v United States, 221 US 502 (1911). 12 G. Feldman, ‘The Demise of the Rule of Reason’ (2020) 24 Lewis & Clark L Rev 951, 954–61. 13 H. Hovenkamp, ‘The Rule of Reason’ (2018) 70 Fla L Rev 81, 83. 14 L. Federico Pace and K. Seidel, ‘The Drafting and the Role of Regulation 17: A Hard-Fought Compromise’, in K. Klaus Patel and H. Schweitzer (eds), The Historical Foundations of EU Competition Law (Oxford University Press, 2013) 60–61. 15 Fear (n 4) 268; H. Schröter, ‘Small European States and Cooperative Capitalism, 1920–1960’, in A. Chandler, F. Amatori and T. Hikino (eds), Big Business and the Wealth of Nations (Cambridge University Press, 1997) 189–96. 16 Schröter (n 8) 131. 17 A. Kuenzler and L. Warlouzet, ‘National Traditions of Competition Law: A Belated Europeanization through Convergence?’, in Klaus Patel and Schweitzer (n 14) 92–93. 18 D. Gerber, Law and Competition in Twentieth Century Europe: Protecting Prometheus (Oxford University Press, 1998) 75–76. 19 ibid; and W. Notz, ‘Cartels during the War’ (1919) 27(1) Journal of Political Economy 1. 20 Gerber (n 18) 116. 21 Notz (n 19) 2–6. 10 11

138  Research handbook on cartels principle of prohibition of cartels, allowed domestic firms to participate in international cartels as long as the affected goods were not sold in the domestic market.22 The economic and social merits attributed to cartels during the War are reflected in a report of the British Committee on Commercial and Industrial Policy, warning of manufacturers ‘acting independently and usually in acute competition one with another […] in many cases unable to pay adequate wages or to obtain sufficient capital for necessary improvements or developments’.23 Consequently, the report recommended that the principle of competition should be ‘supplemented or entirely replaced by co-operation and co-ordination of effort in respect of (1) the securing of supplies of materials, (2) production in which we include standardisation and scientific and industrial research, and (3) marketing’.24 B.

The Interwar Period: Early Competition Rules

Early ideas advocating the value of competition began to spread in Europe during the interwar period. Several European countries prohibited specific forms of competitive restraints as a response to the hyperinflation of the early 1920s, when cartel members shifted to consumers the risks associated with currency devaluation.25 The 1923 German Ordinance against the Misuse of Economic Power, for instance, was introduced as an emergency decree to combat hyperinflation. It did not prohibit cartels as such, but empowered the Minister of Economic Affairs to control abusive practices where a cartel ‘endanger[s] the national economy or the public interest’.26 A 1926 amendment to the French Penal Code had distinguished between ‘good’ and ‘bad’ cartels, by which the negative effects emerging from a cartel were balanced against its positive impact on economic development on a case-by-case basis.27 Notably, those early rules did not impose a blanket prohibition on cartels. Rather, they reflected a principle of abuse, prohibiting only cartels presumed to have harmful effects.28 In parallel, governments continued to encourage domestic cartelization as an instrument of industrial policy to protect their national economies against the economic crisis of 1929–33.29 Some governments have further concluded international cartels to shield their markets from expropriation, secure access to raw materials or markets and influence the host country’s domestic policy.30 See Section III(C) below. Committee on Commercial and Industrial Policy after the War, ‘Final Report’ (London, 1918) para 160. 24 ibid para 161. 25 United Nations Department of Economic Affairs, ‘International Cartels: A League of Nations Memorandum’ (1947, Lake Success) 10–11; Gerber (n 18) Chapter V; and Kuenzler and Warlouzet (n 17) 95. 26 RGBl. 1923 I, 1090 (English translation from R. Liefmann, Cartels, Concerns and Trusts (Arno Press, 1977) Appendix I). Also see Kuenzler and Warlouzet (n 17) 94. 27 Articles 419 of the Code Pénal, as added by the 1926 law. Also see Freedeman (n 7) 476–77; and Kuenzler and Warlouzet (n 17) 94. 28 Kuenzler and Warlouzet (n 17) 92–94. 29 United Nations Department of Economic Affairs (n 25) v and 5–10; Fear (n 4) 276; E. Hadley, Antitrust in Japan (Princeton University Press, 1970); and Schröter (n 8) 134–41. 30 H. Schröter, ‘Risk and Control in Multinational Enterprise: German Businesses in Scandinavia, 1918–1939’ (1988) 62(3) Business History Review 420. 22 23

Lawful cartels  139 Schröter has recorded the national approaches to cartels during the interwar period by distinguishing between four groups of countries:31 the first group – including many developed nations, such as Austria, Belgium, Czechoslovakia, Finland, France, Germany, the Netherlands, Norway, Sweden and Switzerland – expressed a strong positive approach towards cartels; the second group, consisting of industrializing countries such as Hungary, Italy, Japan, Poland, and Spain, largely demonstrated a positive view towards cartels, yet introduced supervisory agencies; the third group, including Bulgaria, Canada, Denmark, South Africa and the UK, took an ambivalent position; and the fourth group – which included Argentina, Australia, New Zealand and Yugoslavia – followed the US approach, taking an active stance against cartels. It was estimated that by the advent of the Second World War, about 40 per cent of the worldwide trade was subject to cartelization.32 C.

The Post-War Period: A Shift in the Manière de Voir

The modern opposition to cartels emerged following the Second World War. Driven by strong American influence, cartels were positioned as the enemies of competition and liberalism, and of democracy itself.33 This has gradually shifted the international manière de voir, namely the collective approach to what constitutes a legitimate way of conducting business, the social consensus over the organization of economic life and the paradigm of how the question of the lawfulness of cartels should be approached.34 For example, cartels were first prohibited in both Japan and Germany in 1947, as part of an American attempt to implement democratic practices and reform their economic structures. Inspired by the Sherman Act’s principle of prohibition, those laws included a strict prohibition on cartels.35 The American manière de voir was further extended to the rest of Europe as the US military government in Germany strongly promoted the inclusion of an anti-cartel provision in the Treaty of Paris of 1951 establishing the European Coal and Steel Community (‘ECSC’). In fact, this provision was a condition for terminating the American control over the German coal and steel industry under the allied de-concentration legislation.36 Despite the above, the prohibition on anti-competitive agreements that was included in the European Economic Community Treaty of Rome of 1957 (‘EEC’) leaves greater room for lawful cartels in comparison to the Sherman Act. The drafting of Article 101 TFEU (ex. Article 85 EEC) was strongly influenced by the French competition law at force during that time. As mentioned, the French law did not unequivocally prohibit all anti-competitive agreements, but reflected the principle of abuse warranting a case-by-case analysis to determine if

31 Schröter (n 8). For the EU member states’ approaches, see United Nations Department of Economic Affairs (n 25) 32–36; and Kuenzler and Warlouzet (n 17) 103–08. 32 United Nations Department of Economic Affairs (n 25) 2. Also see Fear (n 4) 276 and 278. 33 Fear (n 4) 279; Schröter (n 30) 197; and Schröter (n 8) 142. 34 Schröter (n 8) 130–31. 35 M. Ariga and L. Rieke, ‘The Antimonopoly Law of Japan and Its Enforcement’ (1964) 39 Wash L Rev 437, 437–40; T. Nakazawa and L. Weiss, ‘The Legal Cartels of Japan’ (1989) 34 Antitrust Bull 641, 641–42; and Kuenzler and Warlouzet (n 17) 96–98. Such strict prohibitions, nevertheless, were short-lived. Quickly after the American occupation ended, the Japanese and German competition laws were amended to permit various types of lawful cartels. 36 Article 65 ECSC. See E.-J. Mestmäcker, ‘Towards a Concept of a Workable European Competition Law’, in Klaus Patel and Schweitzer (n 14) 192.

140  Research handbook on cartels positive effects on economic development or on society could counterbalance the agreement’s negative effects on competition. Therefore, while anti-competitive agreements are prohibited by Article 101(1) TFEU, Article 101(3) TFEU accepts agreements that contribute ‘to improving the production or distribution of goods or to promoting technical or economic progress, while allowing consumers a fair share of the resulting benefit’, provided they do not impose indispensable restrictions and do not afford firms the possibility of eliminating competition in respect of a substantial part of the relevant market. Furthermore, Article 101(3) TFEU does not directly distinguish between hard-core cartels and other types of restrictive agreements. Unlike the US regime, in which naked cartels are considered as per se violations that cannot be justified, all types of anti-competitive agreements can be accepted in the EU.37 The nature of cartels accepted under Article 101(3) TFEU has changed over the years. During the first decades of EU competition law enforcement, the European Commission (‘Commission’) was openly inspired by Keynesian economic theories suggesting that free markets were not the only source for economic development. It called to limit market forces in favour of promoting other industrial and social policies.38 The Commission’s annual report of 1970, for example, stated that the trend towards industrial cooperation and concentration became more marked. Accordingly, the Commission’s policy necessarily consisted in preserving effective competition in the common market while at the same time encouraging schemes for cooperation, reorganisation and combination calculated to render Community enterprises as competitive as possible both inside and outside the Common Market.39

As I explore empirically elsewhere, until the late 1970s the Commission was willing to accept cartels and other serious restrictions of competition in favour of supporting a host of industrial and social policies, and particularly the EU’s market integration imperative.40 The Commission, moreover, had not focused its enforcement efforts on fighting cartels but rather on vertical restraints that threatened the single market. The few horizontal agreements examined during that period were prohibited only when they impeded market integration in addition to harming competition (for example, in situations of geographic market-sharing). The Commission exempted cartels pertaining to fostering growth, regional cohesion and employment if they did not impede market integration, by means of accepting voluntary commitments or attaching conditions to the exemptions.41

37 However, as elaborated below, naked cartels are less likely to be justified. For an empirical overview of the application of Article 101(3) TFEU to by-object restrictions, see Or Brook, Non-Competition Interests in EU Antitrust Law: An Empirical Study of Article 101 TFEU (CUP, 2022) Figure 3.7. 38 H. Buch-Hansen and A. Wigger, ‘Revisiting 50 Years of Market-Making: The Neoliberal Transformation of European Competition Policy’ (2010) 17(1) Review of International Political Economy 20, 26–32; S. Ramírez Pérez and S. van de Scheur, ‘The Evolution of the Law on Articles 85 and 86 EEC: Ordoliberalism and Its Keynesian Challenge’, in Klaus Patel and Schweitzer (n 14) 21. 39 European Coal and Steel Community, European Economic Community, European Atomic Energy Community, Forth General Report on the Activities of the Communities 1970 (Brussels and Luxembourg, February 1971) para 21 (emphasis added). 40 Brook (n 37), 102–5, 121–3. 41 ibid, 123–30.

Lawful cartels  141 The possibility of exempting restrictive practices under Article 101(3) TFEU was further broadened in 1977, when the European Court of Justice held that EU competition law is guided by the notion of workable competition – that is, the degree of competition ‘necessary to ensure the observance of the basic requirements and the attainment of the objectives of the Treaty’.42 The Court opened the door for accepting cartels and other anticompetitive agreements on the basis of environmental benefits, development of sports and allocation and supply of scarce national resources among states. No strict test guided this assessment. The Commission enjoyed a wide margin of discretion to take into account almost all types of benefits.43 The tolerance towards cartels had also characterized the national approaches of many of the European Member States.44 In France, the above-mentioned distinction between ‘good’ and ‘bad’ cartels had remained in force even after the Ordinance on Competition and Freedom of Prices of 1986 had established modern French competition policy and was later codified in Book IV of the Commercial Code.45 Similarly, in the UK, prior to the adoption of the Fair Trading Act of 1973, the prohibition on anticompetitive agreements was subject to a public interest test. There was no presumption that competition was the most effective model, and cartels were accepted if they fostered efficient production and distribution of goods, encouraged new enterprises or secured efficient regional distribution of employment and resources.46 Even the German competition law, which was relatively strict in prohibiting cartels, included various exceptions up until the turn of the millennium. Warranted by a mix of economic, industrial policy and social considerations, the prohibition did not apply to cartels involving setting standard terms for business, delivery or payments; rebates cartels; crisis cartels; rationalization cartels; specialization cartels; cooperation cartels; and export and import cartels (until 1988).47 The law also shielded a number of industries from the competition rules,48 and entrusted the Federal Minister of Economics to approve cartels in exceptional cases for reasons concerning the general economy and the common welfare. Corresponding to the above, Haucap et al found that, between 1958 and 2004, the German competition authority registered 864 cartels based on those provisions, of which around two-thirds were rationalization cartels.49 D.

The 1990s Onwards: Modernization and ‘Europeanization’

The European approach towards cartels had started to shift by the mid-1990s, when the Commission identified the fight against cartels as an enforcement priority,50 adopted a leni Case 26/76 Metro v Commission [1977] ECR 1875, paras 20–21. O. Brook, ‘Struggling with Article 101(3) TFEU: Diverging Approaches of the Commission, EU Courts, and five Competition Authorities’ (2019) 56(1) Common Market Law Review 121, 133. 44 Schröter (n 8) 142–52; and Kuenzler and Warlouzet (n 17) 95–103. 45 French Ordinance 86-1243 of 1986 on Competition and Freedom of Prices, Articles 7 and 10. 46 D. Elliott and J. Gribbin, ‘The Abolition of Cartels and Structural Change in the United Kingdom’, in A. Jacquemin and H. De Jong (eds), Welfare Aspects of Industrial Markets (Springer, 1977) 346–47. 47 German Act against Restraints of Competition, Sections 2–8. 48 E.g. transportation and postal, agriculture and forestry, banking and insurance, copyright associations, supply of public utilities including electricity, and gas and water (ibid Sections 99–103). 49 J. Haucap, U. Heimeshoff and L.M. Schultz, ‘Legal and Illegal Cartels in Germany between 1958 and 2004’ (University of Düsseldorf DICE Discussion Paper, 2010) www​.econstor​.eu/​bitstream/​10419/​ 41423/​1/​638076714​.pdf [accessed 3 March 2022] 7–8. 50 European Commission, XXXIInd Report on Competition Policy 2002 (Brussels and Luxembourg, 2003) 32. Also see European Commission, XXXIIIrd Report on Competition Policy 2003 (Brussels and 42 43

142  Research handbook on cartels ency programme incentivizing firms to reveal secret cartels51 and declared that cartels are the most harmful agreements to consumers and the European economy in general.52 This transition was complemented by the substantive modernization of EU competition law of the early 2000s, when the Commission called to replace the workable competition notion with a short-term narrow consumer welfare standard.53 According to the Commission’s new interpretation, restrictions of competition will only be accepted under Article 101(3) TFEU if they create quantifiable economic benefits enjoyed by direct or indirect consumers.54 Moreover, the Commission announced that although Article 101(3) TFEU does not a priori exclude certain types of agreements, hard-core restrictions are ‘unlikely’ to fulfil the conditions of the Article.55 The Commission’s new approach has thus considerably limited the type of cartels and other anti-competitive agreements that can be deemed lawful. The Commission’s new approach carried spill-over effects to the European national regimes. As the modernization decentralized the application of Article 101(3) TFEU – and the decision of what amounts to a lawful cartel – to national competition authorities,56 it facilitated a process of soft convergence and voluntary harmonization to the Commission’s approach.57 Although some national divergence still persists,58 many Member States have adopted a consumer welfare-centric approach and a relatively strict opposition to cartels. The Member States’ commitment to fighting hard-core cartels was also pronounced in the ECN+ Directive of 2019, harmonizing certain provisions related to the conditions for granting leniency for secret cartels59 and declaring that commitments decisions are ‘in principle’ inappropriate to deal with such infringements.60 Despite the shift in the Commission’s approach, EU competition law still leaves greater room for accepting some cartel arrangements in comparison to the US. This is often explained by the multitude of goals pursued by EU competition law. EU competition law is aimed not only at economic efficiency, but also at the promotion of economic freedom and the integra-

Luxembourg, 2004) 5; European Commission, Annual Report on Competition Policy 2004: Volume I (Luxembourg, 2005) 4. 51 European Commission, ‘Notice on the Non-Imposition or Reduction of Fines in Cartel Cases’ [1996] OJ C207/4. 52 European Commission, XXVIth Report on Competition Policy 1996 (Brussels and Luxembourg, 1997) 27; European Commission, XXIXth Report on Competition Policy 1999 (Brussels and Luxembourg, 2000) 31; Commission, XXXth Report on Competition Policy 2000 (Brussels and Luxembourg, 2001) 35; and European Commission, XXXIth Report on Competition Policy 2001 (Brussels and Luxembourg, 2002) 24 and 30. 53 European Commission, ‘Guidelines on the Application of Article 81(3) of the Treaty’ [2004] OJ C101/97 (‘Article 101(3) Guidelines’) para 33. 54 Brook (n 37) 67–73. 55 Article 101(3) Guidelines (n 53) para 46. Also see Commission Staff Working Document, ‘Guidance on Restrictions of Competition “By Object” for the Purpose of Defining which Agreements May Benefit from the De Minimis Notice’ SWD(2014) 198 final, 4. 56 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 L1/1, Articles 3 and 5. 57 Brook (n 37) 75–92, 149–85. 58 ibid. 59 Directive (EU) 2019/1 of the European Parliament and of the Council of 11 December 2018 to Empower the Competition Authorities of the Member States to Be More Effective Enforcers and to Ensure the Proper Functioning of the Internal Market [2019] OJ L11/3, Articles 17–23 and Recital 50. 60 ibid Recital 39.

Lawful cartels  143 tion of the common market. Moreover, the competition law provisions form part of the EU Treaties, which pursue various economic and social objectives. Accordingly, the EU courts have long insisted that the competition rules must be read in the context of those objectives, which ought to be balanced against each other.61 The EU approach to lawful and unlawful cartels bears a significant global influence. Bradford et al point to the process of ‘Europeanization’ of worldwide competition law regimes, by which the majority of jurisdictions with competition law provisions adopt laws that more closely resemble the approach of EU competition law than that of US antitrust. In particular, they attribute some of the popularity of the EU model to its plurality of goals and its tendency to defer less to markets and more to governments’ ability to correct market failures.62

III.

TYPES OF LAWFUL CARTEL: BETWEEN ECONOMIC AND NON-ECONOMIC BENEFITS

Even after the global shift in attitudes towards cartels, it is clear that not all horizontal agreements between competitors are deemed unlawful. Various competition law regimes accept a host of arrangements that are presumed to create economic benefits to consumers or non-economic benefits for the general public. This section demonstrates that although there are some practices that are accepted by a number of jurisdictions, there is no common framework that guides or can explain the dividing line between lawful and unlawful cartels. In many instances, the assessment of lawfulness differs according to the type of cartel rather than the degree of harm to competition or the balance of interests. Competition authorities, courts, governments and parliaments have mostly refrained from adopting a systematic approach, and have responded to case-specific challenges emerging from certain types of cartels. This section, therefore, presents the most common types of lawful cartel, drawing attention to the typical benefits invoked to justify them. While it does not attempt to provide a comprehensive list of all types of lawful cartels, identifying the types of benefits offers a useful analytical tool that can inform the strategies for assessing such practices, which will be discussed in the next section. A.

Horizontal Cooperation Agreements

Cooperative strategies, by which firms combine complementary activities, skills or assets, are a prevailing and growing business practice.63 They include agreements on joint research and development; production, purchasing or commercialization; and technical or quality standardization. Prompted by global competition and rapid technological developments, such

G. Monti, ‘Article 81 EC and Public Policy’ (2002) 39 Common Market Law Review 1057; and N. Dunne, ‘Public Interest and EU Competition Law’ (2020) 65(2) Antitrust Bulletin 256, 260–62. 62 A. Bradford, A. Chilton, K. Linos and A. Weaver, ‘The Global Dominance of European Competition Law over American Antitrust Law’ (2019) 16(4) Journal of Empirical Legal Studies 731, 735. 63 T. Hemphill, ‘Cooperative Strategy, Technology Innovation and Competition Policy in the United States and the European Union’ (2003) 15(1) Technology Analysis & Strategic Management 93, 93–94. 61

144  Research handbook on cartels arrangements are viewed as lawful when they are likely to lead to substantial economic benefits, such as risk-sharing, cost-saving, increased investments or pooled know-how, or when they enhance product quality, variety and innovation. They are subject to special rules and are protected by safe harbours in many jurisdictions, including in the US and EU.64 Although the lawfulness of such horizontal cooperation agreements is mostly explained in economic efficiency terms, at times they were also justified by broader industrial policy considerations. For example, cooperative strategies were seen as a means to protect the survival of SMEs and offset structural disadvantages, and a number of jurisdictions exempt or limit SMEs from the prohibitions on anti-competitive agreements even in cases of serious restrictions of competition.65 Moreover, in the EU, some horizontal cooperation agreements were accepted by noting their function in enhancing the competitiveness of the European industries as an additional justification to the main economic benefits.66 B.

Crisis Cartels and Restructuring Arrangements

The term ‘crisis cartel’ has been coined to describe both lawful and unlawful cartels adopted during times of industry-specific downturns or general economic distress.67 While those agreements are typically highly restrictive and would not be accepted as a rule, they are vindicated by a host of economic and non-economic benefits. In a line of cases during the 1980s, for example, the Commission accepted industrial-restructuring agreements directed at an orderly reduction of excess capacity following the 1970s economic crisis, as long as they did not set prices or quotas. The Commission mostly based its decisions on market-wide pro-competitive benefits and industrial policy justifications, noting that free-market forces have failed to achieve the necessary capacity reductions to re-establish and maintain a long-term, effective and competitive structure.68 These conclusions were reinforced by non-economic benefits, observing that the coordinated closure ‘helps to mitigate, spread and stagger their impact on employment’.69

ibid; European Commission, ‘Guidelines on the Applicability of Article 101 of the Treaty on the Functioning of the European Union to Horizontal Co-operation Agreements’ [2011] OJ C11/1 para 2; Federal Trade Commission and the US Department of Justice, ‘Antitrust Guidelines for Collaborations Among Competitors’ (2000) Preamble. 65 OECD, ‘General Cartel Bans: Criteria for Exemption for Small and Medium-sized Enterprises’ (OCDE/GD(97)53, 1996) www​.oecd​.org/​competition/​cartels/​1920345​.pdf [accessed 3 March 2022] 8 and 11–15. Also see Brook (n 37) 296–301. 66 See e.g. Case IV/30.320, Optical Fibres, Commission Decision 86/405/EEC [1986] OJ L236/30 para 59; and Case IV/32306, Olivetti/Canon, Commission Decision 88/88/EEC [1987] OJ L52/60 para 54. Also see Brook (n 37) 113–15, 130–32, 193–4. 67 OECD, ‘Policy Roundtable – Crisis Cartels’ ((2011) DAF/COMP/GF(2011)11, 18 October 2011) www​.oecd​.org/​daf/​competition/​cartels/​48948847​.pdf [accessed 3 March 2022]. On crisis cartels, see Chapter 15 in this volume. 68 Case IV/30.810, Synthetic Fibres, Commission Decision 84/380/EEC [1984] OJ L207/17, paras 28–52; Case IV/30.863, BPCL/ICI, Commission Decision 84/387/EEC [1984] OJ L212/1, paras 33–40; and Case V/31.055, ENI/Montedison, Commission Decision 87/3/EEC [1986] OJ L5/13, paras 26–41. 69 European Commission, Twelfth Report on Competition Policy (Brussels and Luxembourg, 1983) para 39. Also see European Commission, XXIVth Report on Competition 1994 (Brussels and Luxembourg, 1995) para 85. 64

Lawful cartels  145 The economic value of crisis cartels has been more contested in recent years. However, non-economic benefits – such as securing employment, stabilizing prices and protecting domestic firms – still sometimes tilt the balance in favour of accepting such practices.70 In particular, such public policies have been invoked as a justification for a relaxation of the competition rules as a response to the COVID-19 crisis. The ICN, for example, declared that the unprecedented health, social and economic challenges emerging from the pandemic ‘may trigger the need for competitors to cooperate temporarily in order to ensure the supply and distribution of scarce products and services that protect the health and safety of all consumers’.71 C.

Export Cartels

Export cartels involve the coordination of commercial behaviour of domestic producers in foreign markets.72 As mentioned, they were prevalent from the late 1800s, and are still a common practice today. The US Webb–Pomerene Act of 1918 and the Export Trading Company Act of 1982, for example, permit exporters to exercise their collective bargaining power in foreign markets as long as such conduct does not lead to a restriction on imports into the US and does not restrict competition on the domestic market.73 Similar explicit and implicit rules tolerating export cartels are also found in many other jurisdictions, in particular in developing countries.74 The lawfulness of export cartels is often explained with reference to economic benefits and industrial policies. They are said to facilitate new market entry, bringing innovation or lower prices and allowing SMEs to counter the economic power of foreign buying cartels. Nevertheless, empirical research challenges such justifications, demonstrating that in practice, large international companies, rather than local SMEs, are taking advantage of such rules.75 D.

Sustainability Agreements

The lawfulness of agreements banning the use of environmentally unfriendly products or production means and setting environmental standards and fair trade labels has been subject to much debate in recent years. Whereas many of those agreements could be analysed under the rules applicable to horizontal cooperation agreements or rules of trade and professional associations, they are frequently treated as a stand-alone category benefiting from a more relaxed

OECD (n 67) 24. International Competition Network Steering Group, ‘Statement: Competition During and After the Covid-19 Pandemic’ (2020) www​.i​nternation​alcompetit​ionnetwork​.org/​featured/​statement​-competition​ -and​-covid19/​[accessed 3 March 2022]. See also O. Odudu, ‘Feeding the Nation in Times of Crisis: The Relaxation of Competition Law in the United Kingdom’ (2020) 19(2) Competition Law Journal 68. 72 F. Jenny, ‘Export Cartels in Primary Products: The Potash Case in Perspective’ [2012] Trade, Competition, and the Pricing of Commodities 99. 73 Webb–Pomerene Act, 15 USC; Export Trading Company Act, 15 USC. Also see Jenny (n 72) 105. 74 M. Levenstein and V. Suslow, ‘The Changing International Status of Export Cartel Exemptions’ (2004) 20 Am U Int’l L Rev 785, 800–06 and 819. 75 World Trade Organization, Report of the Working Group on the Interaction Between Trade and Competition Policy to the General Council (WT/WGTCP/7, 17 July 2003) 14. 70 71

146  Research handbook on cartels application of the competition rules.76 A number of competition authorities have adopted dedicated policy papers setting out their approaches to the assessment of such initiatives.77 The lawfulness of such arrangements is justified not only by the economic benefits created to direct and indirect consumers, but also by non-economic benefits for the public and future generations. Given the unprecedented climate emergency, and as environmental protection receives constitutional status in various legal systems, there is a growing acceptance that environmental legislation may not be sufficient to address those concerns alone and that sustainability concerns should be supplemented by private initiatives.78 E.

State Sovereignty

A multitude of cartel and cartel-like agreements are either backed or run by a state. Such agreements may be the direct result of a state measure (for example, required by law or as part of the administrative use of public power), be influenced by a state, or delegate to private firms a task that was previously performed by the state. The legal status of such agreements is highly controversial. On the one hand, they may bear significant anti-competitive effects, akin to private cartels. In fact, cartel members are likely to prefer government-backed restraints that do not need to be kept secret and are more stable because the state power protects against cheating.79 On the other hand, such arrangements may create benefits to society, for example when they aim to correct market failures or ensure distributional objectives traditionally safeguarded by state measures.80 Consequently, state-run or backed arrangements are protected in many jurisdictions as a representation of state sovereignty or for political reasons.81 The US state action defence, for example, provides that states are shielded from federal antitrust laws when they engage in a bona fide exercise of their sovereign regulatory powers and that firms are immune from antitrust liability when they act in furtherance of an articulated state policy and under active supervision.82 In the EU, there is no clear state action defence as the principle of primacy of EU law requires Member States not to introduce or maintain in force legislative or regulatory

Monti (n 61) 1073–75. See e.g. Dutch Authority for Consumers and Markets, ‘Draft Guidelines on Sustainability Agreements’ (2020), www​.acm​.nl/​sites/​default/​files/​documents/​2020​-07/​sustainability​-agreements​ %5B1​%5D​.pdf [accessed 3 March 2022]; Hellenic Competition Commission, ‘Staff Discussion Paper on Sustainability Issues and Competition Law’ (2020) www​.epant​.gr/​en/​enimerosi/​competition​-law​ -sustainability​ .html [accessed 3 March 2022]; OECD Competition Committee Discussion Paper, ‘Sustainability and Competition’ (2020) www​.oecd​.org/​daf/​competition/​sustainability​-and​-competition​ -2020​.pdf [accessed 3 March 2022]. 78 S. Kingston, ‘Competition Law in an Environmental Crisis’ (2019) 10(9) Journal of European Competition Law & Practice 517; G. Monti, ‘Four Options for a Greener Competition Law’ (2020) 11(3) Journal of European Competition Law & Practice 124; and S. Holmes, ‘Climate Change, Sustainability, and Competition Law’ (2020) 8(2) Journal of Antitrust Enforcement 354. 79 T. Muris, ‘State Intervention/State Action – A US Perspective’ (George Mason Law & Economics Research Paper, No. 04-18, October 2003) 2. 80 OECD, ‘Regulated Conduct Defence’ (DAF/COMP(2011)3, 1 September 2011) www​.oecd​.org/​ regreform/​sectors/​48606639​.pdf [accessed 3 March 2022] 9. 81 ibid. 82 Parker v Brown, 317 US 341 (1943). Also see OECD (n 80) 188–89. 76 77

Lawful cartels  147 measures that could render the competition rules ineffective.83 Yet in certain cases, the Court of Justice accepted national measures that limited the full application of the prohibition on anti-competitive agreements, recognizing that a state must be allowed to exercise its sovereign power to adopt measures that it deems justified for economic or social reasons. The EU case law on this matter is complex and unclear, giving rise to much controversy on the lawfulness of cartels operating under the protection of state sovereignty.84 Similar questions surround the legal status of arrangements made by a number of states, most famously the operation of the Organization of Petroleum Exporting Countries (‘OPEC’). OPEC was created in 1960 as an intergovernmental organization to coordinate and unify oil policies among its Member States to uphold income from taxes and royalties payment against the backdrop of declining oil prices.85 While there is little doubt that such arrangement would have been deemed unlawful if adopted by private entities, OPEC was not given any sovereign powers and adopts resolutions that are subject to the approval of the national governments. This raises many questions as to the power of national courts and competition authorities to assert jurisdiction and effectively bring a case against such practices.86 F.

Regulated Professions and Trade Associations

Similarly to the questions surrounding the lawfulness of cartels that are backed or run by a state, the conduct of regulated professions and trade associations may reflect a clash between harm to competition and the protection of the public interest. In many countries, the admission to and practice of such professions are subject to state or self-regulation codifying ethical and safety rules. While such rules inherently restrict the freedom of action – and when adopted in the context of a trade or professional association can be seen as a cartel – they may be deemed necessary to bridge asymmetry of information gaps between customers and service providers in areas of a high level of technical knowledge; reduce externalities such services may bear on third parties (for example, inaccurate audit reports misleading creditors or poorly constructed building jeopardizing public safety); and regulate services that produce public goods for the benefit of society in general (such as the administration of justice or urban development).87 In the US, regulated professions and trade associations are only immune from antitrust law subject to the conditions of the state action defence.88 In the EU, however, the Court of Justice introduced a case-specific exemption. In Wouters, it held that professional regulations that are inherent and necessary for the proper practice of a profession as organized by the Member OECD (n 80) 195. For an empirical overview and analysis of the EU jurisprudence, see Brook (n 37) 230–53. 85 OPEC, ‘Statute’ (2021) www​.opec​.org/​opec​_web/​static​_files​_project/​media/​downloads/​ publications/​OPEC​_Statute​.pdf [accessed 3 March 2022] Articles 1–3. 86 M. Joelson and J. Griffin, ‘The Legal Status of Nation-State Cartels Under United States Antitrust and Public International Law’ (1975) 9(4) The International Lawyer 617; I.. Grossack, ‘OPEC and the Antitrust Laws’ (1986) 20(3) Journal of Economic Issues 725, 725; and S.W. Waller, ‘Suing OPEC’ (2002) 64(1) University of Pittsburgh Law Review 105. 87 European Commission, ‘Communication from the Commission – Report on Competition in Professional Services’ COM(2004) 83 final (2004) paras 22–30. See also I. Wendt, EU Competition Law and Liberal Professions: An Uneasy Relationship? (Martinus Nijhoff Publishers, 2012) 1–10. 88 A. Edlin and R. Haw, ‘Cartels by Another Name: Should Licensed Occupations Face Antitrust Scrutiny?’ (2013) 162 U Pa L Rev 1093; R. Newman Knake, ‘The Legal Monopoly’ (2018) 93(3) Wash L Rev 1293. 83 84

148  Research handbook on cartels States will not be deemed as unlawful despite their restrictive effects on competition.89 In Meca-Medina, the Court extended this reasoning to anti-doping rules adopted by sporting associations, referring to their function in ensuring that competitive sport is conducted fairly and safeguards the equal chances for athletes, athletes’ health, the integrity and objectivity of competitive sport and ethical values.90 G.

Labour and Trade Unions

Collective agreements concluded by labour and trade unions may be seen as a form of collusion between employees and employers, coordinating provisions related to wages, working hours and conditions, employment security and health and safety standards.91 Nevertheless, EU competition law affords special protection to collective bargaining agreements between employers and employees, which are mostly justified with reference to non-economic benefits. The Court of Justice explained that social security systems, negotiated through collective bargaining, serve an important social objective of improving working and employment conditions for the public interest and are protected by the EU Treaty and Charter of Fundamental Rights.92 Because those social objectives would be undermined if such agreements were made subject to the competition rules, such agreements ought to be regarded ‘by virtue of their nature and purpose’ as falling outside the EU cartel prohibition.93 The rise of the gig economy has given birth to new questions regarding the lawfulness of collective bargaining. The EU’s special protection does not currently extend to the self-employed and to platform workers that do not enjoy traditional worker status.94 Yet, between 2020 and 2021 the Commission launched two public consultations aimed to ensure that ‘EU competition law does not stand in the way of collective agreements that aim to improve the working conditions of solo self-employed people (that is, self-employed without employees)’.95 In parallel, some Member States have taken local initiatives to facilitate such practices. An amendment to the Irish Competition Act, for example, allows to exclude certain categories of self-employed

89 Case C-309/99 J.C.J. Wouters v Algemene Raad van de Nederlandse Orde van Advocaten [2002] ECR I-1577, para 97. 90 Case C-519/04P Meca-Medina and Majcen v Commission [2006] ECR I-6991, paras 42–43. 91 T. Boeri and J. Van Ours, The Economics of Imperfect Labor Markets (Princeton University Press, 2013) 51–80; and OECD, ‘Competition in Labour Markets’ (2020) www​.oecd​.org/​daf/​competition/​ competition​-in​-labour​-markets​-2020​.pdf [accessed 3 March 2022]. 92 Article 151 and 152 TFEU and EU Charter on Fundamental Rights, Article 28. 93 Case C-67/96 Albany International BV v Stichting Bedrijfspensioenfonds Textielindustrie [1999] ECR I-5751, paras 59–60; Case C-219/97 Maatschappij Drijvende Bokken BV v Stichting Pensioenfonds voor de Vervoer- en Havenbedrijven [1999] ECR I-6121, paras 56–57; and Joined Cases C-115/97 to C-117/97 Brentjens’ Handelsonderneming BV v Stichting Bedrijfspensioenfonds voor de Handel in Bouwmaterialen [1999] ECR I-6025, paras 56–57. 94 Case C‑413/13 FNV Kunsten Informatie en Media v The Netherlands EU:​C:​2014:​2411. Also see D. Schiek and A. Gideon, ‘Outsmarting the Gig-Economy Through Collective Bargaining – EU Competition Law as a Barrier to Smart Cities?’ (2018) 32(3) International Review of Law, Computers & Technology 275, 281–82. 95 European Commission, ‘Collective Bargaining Agreements for Self-Employed – Scope of Application EU Competition Rules’ (2021) https://​ec​.europa​.eu/​info/​law/​better​-regulation/​have​-your​ -say/​initiatives/​12483​-Collective​-bargaining​-agreements​-for​-self​-employed​-scope​-of​-application​-EU​ -competition​-rules [accessed 3 March 2022].

Lawful cartels  149 workers from the act,96 and guidelines issued by the Dutch competition authority interpret EU and national competition laws as allowing self-employed and platform workers to coordinate their rates and other conditions among each other and to collectively negotiate with their clients.97

IV.

ASSESSMENT STRATEGIES

This section points to a disconnect between the underlying economic and non-economic justifications accepted for cartels – as presented in the previous section – and the legal, economic and political tests used to assess their lawfulness. It advocates rationalizing the assessment of lawfulness by shifting the focus away from the types of the arrangement towards ensuring an alignment between the justifications for cartels and the assessment strategies used to evaluate their lawfulness. As elaborated below, there are four main strategies that guide the assessment of cartels: economic analysis, legal analysis, exclusion rules and the exercise of enforcement discretion.98 The following sub-sections discuss those strategies in turn, highlighting the types of benefit they are suitable to address. A.

Economic Analysis

One common strategy for the assessment of cartels is the use of economic principles to compare the impact of an agreement on competition against its quantifiable effects on other public policies. This process is based on economic cost–benefit analysis, aiming to ensure the maximization of consumer welfare, or of an alternative economic concept. A prominent example of economic analysis can be found in the application of Article 101(3) TFEU following the modernization of EU competition law. As mentioned, the Commission shifted the application of the Article towards a greater focus on consumer welfare. Accordingly, its Guidelines explain that ‘[w]hen the pro-competitive effects of an agreement outweigh its anti-competitive effects the agreement is on balance pro-competitive and compatible with the objectives of the Community competition rules’.99 Similar ‘efficiency defences’ can be found all around the globe. Bradford et al show that approximately 75 per cent of the competition regimes include such a defence.100 A cost–benefit analysis also guides the application of the US rule of reason, at least in its classic version.101 Unlike the efficiency defence of Article 101(3) TFEU, the US rule deter Irish Competition (Amendment) Act 2017 (Act 12 of 2017). Dutch Authority for Consumer and Markets, ‘Guidelines on Price Arrangements between Self-employed Workers’ (2019) www​.acm​.nl/​sites/​default/​files/​documents/​2020​-07/​guidelines​-on​-price​ -arrangements​-between​-self​-employed​-workers​.pdf [accessed 3 March 2022]. 98 Brook (n 37) 21–3. This classification of the first three strategies is inspired by C. Townley, Article 81 EC and Public Policy (Hart Publishing, 2009) 6–7 and 28–29, which uses slightly different terminology, referring to market balancing, mere balancing and exclusion. 99 Emphasis added. Article 101(3) TFEU Guidelines (n 53) para 33. 100 Bradford et al (n 62) 741. 101 Feldman (n 12) 989, however, illustrates that the rule of reason was interpreted as a mix of economic and legal analysis, incorporating both cost–benefit analysis and the assessment of whether less restrictive means exist. 96 97

150  Research handbook on cartels mines whether an anticompetitive agreement existed in the first place. It offers a binary test: net pro-competitive restraints are permitted, and net anti-competitive effects are prohibited. Using economic analysis as a strategy for assessing the lawfulness of cartels is hailed as an objective, clear and consistent approach, which is divorced from protectionist or political influences. It is particularly suitable to assess efficiencies generated to the benefit of the users of the relevant product or service, such as those typically invoked with respect to horizontal cooperation agreements that were described in Section III(A) above. The assessment of such agreements requires one to reconcile two relatively comparable economic interests (harm to competition versus efficiencies to consumers), which can ultimately be measured by the impact on consumer welfare. However, an economic analysis also carries a number of limitations. First, as many have already observed, the objectivity of the economic assessment should not be exaggerated. The calculation of consumer welfare is not an exact science. It requires some policy choices, and may provide merely an ‘illusion of certainty’.102 Second, economic analysis is particularly problematic when it attempts to take into account non-economic benefits, by quantifying or using subjective proxies to integrate broader political and social factors within a cost–benefit analysis. In the Chicken of Tomorrow case, for example, the Dutch competition authority pronounced in monetary terms the animal welfare and environmental benefits resulting from a minimum sustainability standard agreed upon between supermarkets and chicken-meat producers.103 It based the application of Article 101(3) TFEU and the national equivalent provision on a consumer survey measuring willingness to pay for such benefits. This methodology was criticized as failing to capture the social value attached to animal welfare initiatives that is not directly enjoyed by consumers of chicken meat as well as long-term environmental effects.104 Applying economic analysis to balance irreconcilable economic and non-economic benefits is therefore highly contentious. This exercise is dependent on socio-political values and preferences that cannot be solved by a cost–benefit analysis. For this reason, a sole reliance on economic analysis may not be suitable for competition law regimes that have as their goals multiple objectives beyond economic efficiency and consumer welfare, and that do not offer a clear hierarchy among the different objectives. B.

Legal Analysis

Assessing the lawfulness of cartels may also hinge on a legal proportionality test, considering whether an anti-competitive agreement is ‘justified’ or ‘reasonable’ for attaining a legitimate economic or public policy aim. This test is not based on economic cost–benefit analysis, but on a more abstract legal weighing of interests. A competition authority or court must determine whether the restriction did not go beyond what is necessary and whether the claimed benefits did not exceed the harm to competition. 102 R. Pitofsky, ‘Political Content of Antitrust’ (1979) 127 U Pa L Rev 1051, 1065. See also T. Wu, ‘After Consumer Welfare, Now What? The “Protection of Competition”’ (Competition Policy International, 2018) https://​papers​.ssrn​.com/​sol3/​papers​.cfm​?abstract​_id​=​3249173 [accessed 3 March 2022]; and Townley (n 98) 38. 103 Dutch Authority for Consumers and Markets, ‘Analysis of the Sustainability Arrangements Concerning the Chicken of Tomorrow’ (2015) acm​.nl/​en/​publications/​publication/​13789/​ACMs​-analysis​ -of​-the​-sustainability​-arrangements​-concerning​-the​-Chicken​-of​-Tomorrow [accessed 3 March 2022]. 104 Holmes (n 78).

Lawful cartels  151 Legal analysis has guided the assessment of many cartels that were justified with reference to non-economic benefits. In the EU, legal assessment informed the legality of cartels operating under the protection of state sovereignty, regulated professions and associations, and the special protection awarded to unions and collective bargaining. Moreover, it had characterized the application of Article 101(3) TFEU prior to the modernization of EU competition law, allowing the Commission to exempt a broad array of crisis restructuring arrangements and sustainability agreements.105 Legal analysis was also used to assess economic benefits. In the EU, it directed the application of the ancillary restraints doctrine, examining if a clause restricting rivalry between undertakings was necessary for the realization of an overall legitimate commercial purpose of an agreement.106 The EU courts held that the doctrine does not entail an economic assessment weighing the pro- and anti-competitive effects, but rather a legal assessment in which the necessity of the restriction for the main operation is being assessed in the ‘abstract’.107 In the US too, the rule of reason was sometimes interpreted as entailing a form of the less-restrictive-alternative analysis to examine arrangements creating economic benefits, in place or in addition to the traditional cost–benefit analysis.108 Legal analysis provides competition authorities and courts with a degree of flexibility to disapply the prohibition of cartels to arrangements creating non-economic benefits that cannot be accurately captured by an economic analysis or to exempt agreements without undertaking a full cost–benefit analysis. This flexibility, however, is also the main source of criticism against it. Legal analysis introduces a degree of subjectivity to competition law enforcement as it follows a more abstract analysis, especially when it calls for striking a balance between irreconcilable economic and non-economic interests. Such balancing raises the concern that competition is used as a regulatory market-building tool, going well beyond the competition authorities’ mandate and goals of protecting competition. This, in turn, also raises questions as to the democratic legitimacy and political accountability of administrative competition authorities to perform such balancing exercise and creates legal predictability challenges. Such risks can be moderated if legal analysis is confined to assessing non-economic benefits that are acknowledged as legitimate in a specific legal system. In the context of the EU, for example, it can be used to strike a balance between harm to competition and the objectives of EU competition policy (efficiency, market integration and economic freedom) or non-economic interest explicitly protected by the cross-sectional clauses of the Treaties. Consequently, while the assessment of the existence of an interest, as well as of its weight, would remain at the discretion of the competition authority, the identification of the interest as legitimate would be left to the national legislator or judiciary.

See Section III above. See e.g. Article 101(3) Guidelines (n 53) paras 28–31 and the references there. 107 Case T-112/99 M6 and Others v Commission [2001] ECR II-2459, paras 107–109. Also see Case C-179/16 Hoffmann-La Roche v Autorità Garante della Concorrenza e del Mercato ECLI:​EU:​C:​2018:​ 25, paras 70–71. 108 Feldman (n 12). 105 106

152  Research handbook on cartels C. Exclusion Economic and legal analyses rest on the measuring and weighing of interests on a case-by-case basis. However, cartels can also be deemed lawful pursuant to rules of exclusion, limiting the application of the competition rules in specific practices or sectors. In all, 49 per cent of the worldwide competition law regimes include at least one partial or complete industry exemption; agriculture, transportation, insurance, banking, and fishing are the most commonly exempted industries.109 Such exclusions are common in the EU. In addition to the special sector-wide rules applicable to the supply and price variations in nuclear materials, military equipment and the agriculture sector,110 several Block Exemption Regulations (‘BERs’) automatically discharge certain categories of agreements without the need to undertake a detailed analysis. Most BERs offer a presumption of legality for relatively unproblematic agreements where there are few or no competition concerns or due to overriding economic benefits. A smaller number of BERs also protect highly restrictive agreements in specific sectors in favour of promoting EU industrial policy and the functioning of such sectors, even if there are limited economic efficiencies.111 The motor vehicles BER, for example, were criticized as awarding disproportionate concern for market integration and social interest over the competitive process.112 Similarly, in the UK, the Secretary of State may adopt an order excluding the application of the prohibition on anti-competitive agreements for exceptional and compelling reasons of public policy.113 Such exclusion orders were issued with respect to the defence industry, to supply of oil and petroleum products and as a response to the COVID-19 pandemic.114 Exclusion rules may also be case-specific. A Hungarian act, for example, authorizes the Minister for Rural Development to exclude from the cartel prohibition agreements in the agricultural sector that provide a fair income for producers, as far as they do not preclude all

109 A. Bradford, A. Chilton, C. Megaw and N. Sokol, ‘Competition Law Gone Global: Introducing the Comparative Competition Law and Enforcement Datasets’ (2019) 16(2) Journal of Empirical Legal Studies 411, Figure 5; O. Brook, ‘Block Exemption Regulations and Public Policy: In the Defence of BERS’ (2022) 23 Cambridge Yearbook of European Legal Studies 1. 110 See the Treaty Establishing the European Atomic Energy Community, Art 106a(3); Article 346(1) (b) TFEU; and Articles 38–44 TFEU, respectively. 111 Brook (n 37) 193–5. 112 S. Marco Colino, ‘Recent Changes in the Regulation of Motor Vehicle Distribution in Europe – Questioning the Logic of Sector-Specific Rules for the Car Industry’ (2010) 6(2) Competition Law Review 203, 210. 113 Competition Act 1998, Schedule 3(7). 114 Competition Act 1998 (Public Policy Exclusion) Order 2006 (maintenance and repair of warships); Competition Act 1998 (Public Policy Exclusion) Order 2007 (strategic and tactical weapons and their supporting technology) (repealed by the Competition Act 1998 (Public Policy Exclusion) (Revocation) Order 2011 (Team Complex Weapons); Competition Act 1998 (Public Policy Exclusion) Order 2008 (design, construction, maintenance and disposal of nuclear submarines); The Competition Act 1998 (Public Policy Exclusion) Order 2012 (supply of oil fuels in an emergency); Competition Act 1998 (Groceries) (Coronavirus) (Public Policy Exclusion) Order 2020; Competition Act 1998 (Health Services for Patients in England) (Coronavirus) (Public Policy Exclusion) Order 2020; Competition Act 1998 (Solent Maritime Crossings) (Coronavirus) (Public Policy Exclusion) Order 2020; Competition Act 1998 (Health Services for Patients in Wales) (Coronavirus) (Public Policy Exclusion) Order 2020; and Competition Act 1998 (Dairy Produce) (Coronavirus) (Public Policy Exclusion) Order 2020.

Lawful cartels  153 market actors from joining the agreement.115 Such an exception is ultimately based on the Minister’s judgement, leaving a political actor considerable discretion to accept otherwise unlawful cartels. Exclusion rules may be susceptible to the creep of protectionist and political influences. They are also less case-sensitive, as they resolve clashes between the protection of competition and other public policies by promoting one set of interests and ignoring the other.116 At the same time, exclusion rules deliver a number of unique advantages when they are carefully used to assess cartels that are justified on the basis of public policies that are external to the competition law regime and merit a political choice. First, such rules set clear, transparent and predictable safe havens that may increase political accountability, because unlike the economic and legal analyses, the decision to set aside the competition rules is entrusted into the hands of the parliament or government. This may be of particular value in jurisdictions with an independent administrative competition authority, lacking the political legitimacy to balance between competition and other public policies. Exclusion rules invite an open and transparent legal, economic and public debate on the lawfulness of certain arrangements, instead of disguising them under the pretence of economic or legal analysis. Second, since exclusion rules are not bound to the regular competition law analysis, they allow one to exempt directly selected practices without the risk of spill-over effects and creation of precedent. This may be of particular value when such exceptions are foreseen as applying for a limited period of time (for example, the UK COVID-19 exclusion orders) or are created as a tool of experimental governance when the emergence of new technologies or markets gives rise to uncertainties as to whether competition law should prohibit certain types of arrangements (such as in digital markets or the gig economy). D.

Enforcement Discretion

Finally, a cartel may be tolerated by means of exercising enforcement discretion. Differing from the three previous strategies, a decision not to enforce the competition rules against a specific cartel does not render it to be lawful. Nonetheless, such a decision may bear a similar de facto effect, especially in countries where private enforcement is underdeveloped or unlikely. At times, competition authorities have declared in advance that they will refrain from enforcement of certain practices. The Dutch competition authority, for example, announced that it would seize to exercise its enforcement powers ‘against sustainability arrangements that enjoy broad social support if all parties involved such as the government, citizen representatives, and businesses are positive about the arrangements’.117 This statement means that sustainability agreements resembling those which have previously been prohibited by the authority – such as the above-mentioned Chicken of Tomorrow case – will no longer be

115 Hungarian Act No. CLXXVI of 2012 on inter-branch organisations and on certain issues of the regulation of agricultural markets, Art 18/A(1). 116 Townley (n 98) 32. 117 Dutch Authority for Consumers and Markets, ‘Basic Principles for Oversight of Sustainability Agreements’ (2016) www​.acm​.nl/​en/​publications/​publication/​16726/​ACM​-sets​-basic​-principles​-for​ -oversight​-of​-sustainability​-arrangements [accessed 3 March 2022].

154  Research handbook on cartels actively pursued.118 Similarly, shortly after the outbreak of the COVID-19 pandemic, various competition authorities and international stakeholders declared that they would temporarily tolerate a host of anticompetitive agreements. The European Competition Network (‘ECN’) stated that it ‘will not actively intervene against necessary and temporary measures put in place in order to avoid a shortage of supply’, and the Commission announced that exchange of commercially sensitive information to adapt production, stock management, and distribution of medicine ‘would not give rise to an enforcement priority’.119 In other situations, enforcement discretion can be used to tolerate cartels on a case-by-case basis. A competition authority can choose not to pursue infringements due to underlying economic or non-economic benefits, even if the cartel is unlikely to be deemed lawful under a legal or economic analysis or exclusion rules. Alternatively, a competition authority may decide to accept commitments to close the investigation into an arrangement that promotes interests that are not directly related to the economic consequences of the anticompetitive behaviour of the relevant firms. Competition authorities should avoid relying systematically on their enforcement discretion to refrain from taking a stance on the lawfulness of cartels. The exercise of enforcement discretion is subject neither to the analytical framework applicable to the economic and legal analyses, nor to the legitimacy and accountability embedded in exclusion rules. In many competition law systems, it is limited by almost no procedural or substantive constraints and the reasons behind the authority’s prioritization decision are often not published or detailed.120 Consequently, implicitly accepting cartels by means of enforcement discretion hampers the effectiveness of the competition rules and raises issues to do with legal certainty, accountability and legitimacy. It encompasses many of the shortcomings of the previous strategies, without offering their advantages and safeguards.

V. CONCLUSION This chapter traced the changing international manière de voir towards lawfulness of cartels. It demonstrated that even after the advent of an international consensus on the harm caused by hard-core cartels, there is no common framework that can explain the dividing line between lawful and unlawful cartels. Similar types of arrangement are subject to different rules and are assessed pursuant to various legal and economic tests or subject to political exceptions. In many cases, the assessment of lawfulness differs according to the type of cartel rather than the degree of harm to competition or the balance of interests, leading to a disconnect between the underlying economic and non-economic benefits justifying some practices and the legal, economic or political strategies used to assess their lawfulness.

Brook (n 37) 156–65. European Competition Network, ‘Statement on Application of Competition Law during the Corona Crisis’ (2020) https://​ec​.europa​.eu/​competition/​ecn/​202003​_joint​-statement​_ecn​_corona​-crisis​ .pdf [accessed 3 March 2022]; and European Commission, ‘Communication from the Commission – Temporary Framework for State Aid Measures to Support the Economy in the Current COVID-19 Outbreak’ C(2020) 1863 final. Also see Brook (n 37) 341–5. 120 Brook (n 37), 310–13. 118 119

Lawful cartels  155 The assessment of cartels could be rationalized by shifting the focus away from the type of agreement towards ensuring an alignment between the economic and non-economic justifications for cartels and the assessment strategies used to evaluate their lawfulness. In particular, the chapter suggested distinguishing between three types of underlining justifications: economic benefits enjoyed by the users of the relevant product or service should be assessed on the basis of a cost–benefit test; non-economic benefits that are recognized as legitimate by the relevant competition law regime should be examined by a legal proportionality test; and non-economic benefits that are external to the relevant competition law regime should be handled by a limited number of transparent and explicit exclusion rules. In parallel, competition authorities should avoid relying on their enforcement discretion to avoid systematically taking decisions on the lawfulness of certain types of practices. Aligning the justifications with the assessment strategies can increase the effectiveness, transparency and accountability of the competition law regimes. Such alignment limits and streamlines the discretion of independent competition authorities, enhances the democratic legitimacy and political accountability of administrative competition law authorities to perform such balancing exercise and provides greater legal predictability for firms.

9. Cartel facilitation Christopher Harding

I.

THE CONCEPT OF CARTEL FACILITATION

A convenient starting point would be the observation that the legal world seems to lack a precise definition of an anti-competitive business cartel, as distinct from descriptive listings of activities and strategies that may be counted as such cartels. The term ‘cartel’ itself, with its complex etymology, is fraught with some ambiguity in popular use, where it might more readily suggest to many an instrument of organized crime in a mainly Latin American context (as for instance, in ‘Columbian drug cartel’1). Put another way, the word lacks specificity and contextual clarity and distinction, which raises problems for the legal significance it has now acquired. Put in yet another way: where can we find a clear legal definition of the concept of the anti-competitive business cartel which we have in mind in this discussion? In the United States, lawyers and legislation refer to ‘trusts’, ‘combinations’ and ‘conspiracies’;2 in Europe (and elsewhere in the world) there are references to ‘concerted practices’ and ‘infringements’ of certain Treaty or legislative provisions. Such formal terms are frequently more conveniently talked about as ‘cartels’, but not formally defined as such. Tellingly, UK criminal legislation has used the term ‘cartel offence’, but only as a section heading, the text of that provision then going on to provide an exemplary listing of the offending conduct.3 Underlying this practice and theory of legal uncertainty there is indeed an ontological puzzle – in what sense more precisely does a business cartel exist?4 To describe the matter as an ontological issue does not mean that it is only a philosophical conundrum or theoretical debating point. It is also an important practical and empirical challenge in attempts at legal control and regulation, and leads to issues of evidence and prosecutorial strategy. Unless we are clear about what we are objecting to in this kind of business behaviour, how can we proceed efficiently in efforts of legal control and regulation? How do we ‘frame the charge’, or decide who should be the subject of legal process and sanctions? This uncertainty has led to a great deal of legal argument and, as a consequence, legal business. To aid the discussion, it might be argued that there are three main senses in which something may reasonably be referred to as a business cartel.5 First, the term may be applied to the instrument which implements the arrangement, as in the expression ‘entering into a cartel’, whatever the form taken, such as an agreement, secret contract or verbal understanding. The cartel in this sense will be a crucial element in legal proceedings as essential evidence of something which is prohibited and unlawful – it will In a very simple internet search, drug cartels will figure prominently in the list. The Sherman Antitrust Act of 1890, 26 Stat 209, 15 USC. 3 The Enterprise Act 2002, Section 188. 4 See the discussion of cartel ontology in C. Harding and J. Edwards, Cartel Criminality: The Mythology and Pathology of Business Collusion (Ashgate Publishing, 2015) 25 et seq. 5 For further discussion of this analysis, see ibid 26. 1 2

156

Cartel facilitation  157 provide the substance and details of an illegal arrangement. This may be referred to as a cartel in the instrumental sense. Second, we might refer to the collection of legal persons involved in the arrangement, whether they be human individuals or corporate, as the cartel in the sense of a collectivity with a certain membership, as in the phrase ‘this company is part of the “X” cartel’. This is another crucial application of the term for the purposes of legal liability, at least when the intention is to investigate the actions of such persons and render them subject to legal sanctions. The cartel in other senses may be commonly referred to as the subject of investigation and legal control (for instance in reporting of case law), but fines or prison terms will be imposed on individual members of the cartel, not the cartel as a whole. This may be referred to as a cartel in the personal sense. Third, it is possible to talk about the cartel as an organization embodying an ongoing arrangement, in that way having a selected moment of beginning and ending, but comprising three essential elements – temporal (for how long), personal (who was involved and when) and spatial (in relation to what goods, services and markets): all important matters of scope. This may be referred to a cartel in the organizational sense. As such this conveys the idea of an infrastructure of anticompetitive activity, as a continuing series of meetings, communications, discussions and implementing actions (what in EU enforcement language has been described as a ‘single continuing infringement’6). This rather more sophisticated sense of the term may, for the purposes of legal analysis, be represented by the device of the ‘cartel as a box’, used by Harding and Joshua,7 which enables any one instance of an actual cartel to be precisely described in the form of a three-dimensional box image, with its three axes exactly determined in terms of time (beginning and end dates), membership (the parties involved) and space (the market within which the cartel operated). As such, the box works as a prosecutorial tool, showing what must be proven for the purposes of legal liability. The above three senses of the word ‘cartel’, all of which validly derive from linguistic practice, policy and legal process, can then help to explain the difficulty in drafting an agreed and succinct legal definition. But it may be observed that the third or organizational sense tends to include the previous two senses, and that is the sense in which the term will be used in the discussion here. But also, in view of the linguistic complexity of the term and the concept, it is then unsurprising that neat legal definition has been evaded in practice throughout many legal systems. The basic question of business cartel existence and definition has been summarized by Harding and Edwards in the following terms: In short, cartel ontology is difficult and problematic: in what sense, more exactly, does a cartel exist? It is important to appreciate that cartels arise from a complex situation, comprising elements of geography, market behaviour and market analysis, organisation, business culture, and collective and individual determination and decision-making.8

On this concept, see Chapter 7 in this volume. C. Harding and J. Joshua, Regulating Cartels in Europe (2nd edn, Oxford University Press, 2010) 173. 8 Harding and Edwards (n 4) 12. 6 7

158  Research handbook on cartels

II.

THE SIGNIFICANCE OF COLLUSION

Key to the understanding of the nature and operation of business cartels is the idea of collusion. As in many other contexts, human action is a choice between working together or working against each other, between co-operation and contest (or, put more dramatically, between war and peace). Again, much may follow from the descriptive language that may be employed. Whereas many might assume that ‘peace’ is more benign and positive than ‘war’, there may be somewhat different connotations in using words such as ‘co-operation’ (shading into a sense of suspicious collusion) and ‘competition’ (shading into a sense of healthy and progressive advance). Much depends of course on circumstance and context and is historically contingent, and in the economic domain the world in the past hundred years or so has moved towards a general preference for the competitive over the co-operative, so much so that there is now a significant global edifice of competition law that is broadly protective of that economic phenomenon.9 Business cartels are now part of that policy and legal discourse and indeed have assumed a totemic role, undergoing transformation from an early twentieth-century acceptance and justification to a late twentieth-century degree of censure expressed in language of an intemperate nature.10 This is not the place to debate either the economics or the policies of cartel prohibition. Rather, the topic is the more specific aspects and scope of the normative approach to cartels. In general terms, competition law as a system of legal protection proceeds from a fear and distrust of the use of economic power, certainly and justifiably from a reaction against economic dominance and monopoly power. Yet, wherein lies the objection to cartel activity as a species of economic strategy? It has not always been thus, and not all cartel arrangements are necessarily prohibited and unlawful. The prohibition has settled upon the category of cartels often now described as ‘hard-core’ as blatantly and unjustifiably anticompetitive, although it has to be admitted that the precise borderline between what is ‘nakedly’ illegal and beyond the pale and what might be tolerable may be open to argument. This observation returns the discussion to the heart of the matter – it is not only a question of what the cartel may comprise, but also one of what makes it objectionable. Partly this has come about as a problem of economic analysis and assessment of anticompetitive impact, and that is something which is notoriously open to argument and contestable. Partly it has come about through the gradual, incremental and empirical development of competition law, as it has been worked out as much by competition authorities and courts working on the hoof as through detailed legislative statement. But while there may be a tricky penumbra, there is also now, in a large number of legal orders, a clear sense and awareness of the illegality of the ‘hard-core cartel’ – an agreement to fix prices or other conditions of supply, to share markets and bid rigging.11 To that extent the unlawfulness of a certain kind of business activity has become manifest. And indeed, that very process in itself, whereby the condemnation of the ‘hard-core’ as unacceptable and unlawful as a matter of official policy, emerging in Europe from the end of the 1960s, has contributed another important element of delinquency,

See generally the historical overview presented by Harding and Joshua (n 7). Cp. J. Fear, ‘Cartels and Competition: Neither Markets Nor Hierarchies’ (Harvard University Working Papers, October 2006); and M. Monti, ‘Fighting Cartels – Why and How?’ (Speech, Stockholm, 11 September 2000) (noting that cartels are ‘cancers on the open market economy’). 11 For more detail, see Harding and Joshua (n 7). 9

10

Cartel facilitation  159 by pushing persistent ‘hard-core’ infringement underground and encouraging measures of subterfuge on the part of traders, what may be characterized as a ‘spiral of delinquency’,12 adding a kind of obstruction of justice to the pure economic offending. An important part of the cartel ‘offence’ or ‘infringement’ has become the determination to persist in conduct on the market known to be objectionable as a matter of policy and to take measures to disguise that determination. And this may be seen as the equivalent to the mindset of ‘conspiracy’ which had long been embedded in the American condemnation of anticompetitive conduct contained in the Sherman Act.13 Without this second layer of ‘bad attitude’, anticompetitive activities in themselves need not attract such a strong degree of condemnation, strong punitive and deterrent sanctions and even criminalization. Just as with anti-competitive activity, the matter could be dealt with by administrative control, more consensually, via negotiated resolution rather than confrontational measures of legal control. And, of course, this was the traditional European approach prior to the later twentieth century.14 But it is also in the nature of the beast. Fixing a price or sharing a market is rarely a single, momentary action, over and completed in a single transaction. It is something that needs to be planned, calculated and maintained in operation; in short, something that requires organization. Following this analysis, it may be helpful to present the cartel infringement almost in universal terms and across legal orders globally15 – and, whatever the precise legal category of the offence (criminal, administrative or civil), as comprising three crucial elements: ● market conduct which in substance is anti-competitive in its effect and strongly and consistently prohibited as a matter of official policy; ● requiring ongoing organization; and ● undertaken knowingly, usually subversively, in defiance of the legal prohibition. Those three elements may be encapuslated by talking about each cartel infringement as a matter of anticompetitive collusion. This unifying idea of collusion is strongly evident for instance in sanctioning practice, on the part of the European Commission and EU courts for instance in the calculation and confirmation of fines,16 and in the practice of national courts and bodies in the determination of administrative and criminal law penalties.

III.

THE ABSENCE OF VERY CLOSE FRIENDS

Returning for a moment to the more restricted idea of a cartel as a group of businesses coming together to decide on something anticompetitive, we can play around with the image of

Or ‘delinquency inflation’; see Harding and Joshua (n 7) 149. See further the discussion of the Norris case infra. 14 D. Gerber, Law and Competition in Twentieth Century Europe: Protecting Prometheus (Clarendon Press, 1998, reprint 2001). 15 And let us remind ourselves that it has not been possible to agree on an international definition or legal concept of a business cartel, still falling back on the varying national vocabulary of, for instance, ‘conspiracy’, ‘concerted practice’ or ‘infringement’. 16 C. Harding, ‘Forging the European Cartel Offence: The Supranational Regulation of Business Conspiracy’ (2004) 12 European Journal of Crime, Criminal Law and Criminal Justice 275. 12 13

160  Research handbook on cartels a party. Who attends the party? For many economists (and also lawyers) the answer seems almost self-evident – market players; competing suppliers of whatever goods or services. That description might be extended in a vicarious way to connected parties, typically within the same corporate group and operating broadly in the same market, not actually present at the party but similar to those who allow the family home to be used by delinquent young party-goers. That might be done in order to convey a compliance-inspired signal (‘keep your house in order’) or, more cynically, to justify the imposition of a larger fine via an alleged wider involvement. But there may also have been others present at the event, partying in an intense and significant way – a special kind of friend, there to help the party go well, although not directly involved in the party games. Less attention has been paid to this kind of guest at the party. Yet a number of ‘cartel parties’ have included such ‘guests’, and this has been happening for some time. Such attendance has been a matter of function and role, and forms may vary – sometimes the role has been undertaken by an individual company or an individual working for a company trading as a member of the cartel, sometimes by an outside third party actor with specialist skills, such as the ‘consulting companies’ who have occasionally been dealt with by the EU courts, as occurred in Treuhand or Icap (see further below). These ‘guests’ are the cartel facilitators and they may carry out a number of roles, which are sometimes significant for the operation and health of the cartel. They may be present at meetings or via communications; be involved in organizing meetings or maintaining contact; be responsible for the management and safe-keeping of any sensitive or confidential material; advise on strategy as the cartel moves forward; negotiate disputes and tensions between the cartel members; and oversee compliance with the internal cartel aims and measures. Generally, such actors perform an important management role, are very well informed and certainly have a clear awareness of the illegality or otherwise of the cartel’s operation. Yet they are less frequently disaggregated from that overall operation, even though the ‘trading’ members will be disaggregated in the most significant way for purposes of legal liability and the imposition of sanctions.17 What may be seen here historically as a failure of disaggregation has largely reflected the legal nature of anti-cartel proceedings in different systems. United States federal law, applying the Sherman Act, has happily bundled most things together as a conspiracy, involving corporate and individual human actors working with each other at the same time, eschewing any theoretical distinctions of legal personality (a company may conspire with its own alter ego representative at a cartel meeting, the former then sanctioned through a financial penalty and the latter through a prison term).18 Much of this follows from the construction of legal liability under the Sherman Act, combined also with the fact that many of these US criminal cases are resolved through the use of guilty pleas so that some legal niceties are not publicly debated in open trials before a jury. This approach differs in many respects from that adopted in EU cases and in fact in a large number of other national systems of competition law. Under such

Compare for instance the persecution of the individual cartelist Romano Pisciotti (Marine Hoses Cartel) with the gentle and tentative treatment of the consulting firm Treuhand. See L. Crofts and L. Nylen, ‘MLex Interview with Romano Pisciotti – Il Capitano’ (MLex, December 2015). 18 Thus, in the case of the Marine Hose Cartel, the Italian company Manuli was subject to a $2 million fine, while three of its representatives (including Pisciotti) received prison terms and fines under US law, although in economic terms they were part of the same organization and constituted one member of the cartel as a grouping. For the details of this cartel, see Harding and Edwards (n 4) 87. 17

Cartel facilitation  161 a conspiracy-based system of prosecution, cartelist friends, whether corporate or individual,19 may or may not be disaggregated for liability and sanctioning, depending very much on the outcome of investigations, the strength of evidence, defence strategies regarding the use of guilty pleas20 and the good or bad luck of fugitive evasion. There is both an indifference towards sanction accumulation and a determination to achieve maximum legal deterrence which effectively obscures the facilitators’ role. Under the EU rules, and then in many other legal systems which have adopted a similar model, the legal liability of cartelists has been constructed differently. The European-level rules, contained in successive Treaty provisions from Article 85(1) of the Treaty of Rome onwards, have consistently prohibited ‘undertakings’ from infringing the European Community/Union competition policy, whatever the precise form (usually larger companies) or personality of such undertakings. In practice, the EU sanctioning process imposes administrative fines on corporate actors. Only occasionally has the European Commission tuned its attention to a particular kind of corporate actor as a discrete participant in a cartel, as a consulting company, as in the handful of Treuhand cases discussed below. A third route of legal disaggregation in relation to cartel friends or facilitators became possible through the criminalization of cartel conduct in some legal systems, inspired by although not always precisely matching the early criminalization under the US Sherman Act. Fostered by the aggressive enforcement policy of the US Department of Justice, a number of European and other national systems jumped on a cartel criminalization bandwagon in the later part of the twentieth century, replacing or supplementing ‘administrative’ legal regulation of cartels with the introduction of criminal offences and the high-profile prosecution of individuals working typically for international companies, intended to scare businessmen with the prospect of the ‘inferno’ of prison terms. Interestingly, this brought some national criminal courts into the picture, and, insofar as many of the cases involved a cross-border element, also rules relating to extradition and international criminal law transfer, and then in turn sometimes basic legal protection arguments involving jurisdictions, such as that of the European Court on Human Rights. Very occasionally, cartel facilitators found themselves netted into this environment of legal enforcement.

IV.

CARTEL FACILITATORS AS MANAGERS AND MASTERMINDS

It may be helpful at this stage to provide a more detailed account, from some of the legal proceedings involving facilitators, of the kind of role carried out by such participants. On closer examination, this may belie their earlier characterization as merely friends or guests. It should be appreciated that cartel facilitation in this sense encompasses in practice a range of persons and statuses. Business cartels vary in their internal organization and operation, or in what may be described broadly as their internal governance. At one end of the spectrum,

And indeed, in the case of the Marine Hose Cartel, the Manuli ‘friend’ – Peter Whittle, aka the company PW Consulting – was one and the same, and apparently the directing force of Manuli’s involvement. 20 In the Marine Hoses case, two individuals were convicted, then acquitted under Sherman Act proceedings; see Harding and Edwards (n 4) 138. 19

162  Research handbook on cartels facilitation may shade into the internal direction and governance of the cartel organization by a powerful producer member, virtually as a kind of gang boss. Such a model of autocratic direct rule is exemplified by the Pre-Insulated Pipes Cartel, a significant European-wide cartel of the 1990s, notable for its highly aggressive strategies.21 The cartel’s strategies were determined and strictly enforced by a dominating ‘ringleader’ company, the large Swedish-Swiss conglomerate Asea Brown Boveri (‘ABB’), acting in this particular market through its Danish subsidiary. In the analysis of the Commission: There can be no doubt that ABB was the ringleader and main instigator of the cartel. Domination of the market via a cartel in which it played a leading role was a clearly stated strategic objective of the company. The whole enterprise was conceived, authorised approved and guided at the most senior corporate level. Throughout the whole five-year period the initiatives to consolidate, reinforce and extend the cartel came from ABB […] Both the cartel and the measures to deny its existence were conceived, directed and actively supported at a high level in ABB group management.22

ABB’s governing and facilitating role was recognized in the size of the fine imposed on the company by the Commission (70 million euros). But this was a corporate penalty, of the kind that ABB was no stranger to, and there was no further disaggregation of sanctions in relation to individual company executives (which might have occurred if the American market had been affected and the cartel had been subject to investigation by the US authorities). Three points may be made in relation to this particular example. First, in relation to international cartels, legal outcomes may depend very much on the happenstance of the scope of the cartel and the market involved. Thus, factors such as the product and the geographical scope of the arrangement will determine the matter of legal jurisdiction, and the outcomes may then vary regarding policies of prosecution and the application of sanctions. A quick comparison of the legal outcomes for members of the Marine Hose and Pre-Insulated Pipes Cartels, both serious international breaches, serves to demonstrate that point: a range of different sanctions were scattered around the world in the first case, but the second was dealt with just by European financial (and non-criminal law) measures.23 Second, major cartels cannot be set up and left to run their own course; they require a considerable effort of continuing attention, organization and management, as business systems which demand internal governance. That will be a matter for some member or members of the cartel, or some specialist directors or administrators – the ‘facilitators’, as we may call such persons. Their role is likely to prove crucial to the survival and success of the cartel and their wits will be pitted against those of external regulators and legal advisers, and such wit will depend upon experience, expert knowledge and strategic sense. As individuals, such persons may be company executives or employees of ‘consulting firms’ – no matter, the role is the same. Third, large cartels vary considerably not only in terms of product market, geographical scope and duration, as will be seen from both empirical survey and case law, but also in their membership. The latter is not only a question of number (literally two firms or legal persons upwards) but also of their kind and degree of involvement, as consistent or ‘in-and-out’ partic See Case T-21/99, Dansk Rorindustrie and others v Commission (2002) ECR II-168. Case No IV/35.691/E-4, Pre-Insulated Pipe Cartel, Commission decision, 21 October 1998 [1999] OJ L24/1, paras 121 and 155. 23 For a detailed account, see Harding and Edwards (n 4) Chapter 5. 21 22

Cartel facilitation  163 ipants, as committed and loyal or nervous and untrustworthy, as passive or active and enthusiastic, and this will depend on factors such as their own economic interests and business culture and provenance. That is something which requires management, whether through the blatant bullying of a dominant member such as ABB in Pre-Insulated Pipes or the diplomatic skills of a consulting firm such as Treuhand. Insofar as any facilitator liability is to be disaggregated, much will then depend on the evidence of each facilitator’s role. Another model of facilitation can be seen in action in the case of the Marine Hose Cartel – a one-man consulting company, responsible for orchestrating these tricky matters of internal governance and operation. So much was revealed by the prosecution and conviction of the individual Peter Whittle, following the successful raid on one of the Cartel’s meetings by the US Department of Justice in Houston in Texas in May 2007. Three British executives subsequently negotiated guilty pleas and transfers to serve prison sentences in the UK, and information regarding the cartel’s internal operation emerges from this legal process.24 The Cartel had operated since the mid-1980s, but was faltering somewhat by 1999 when Whittle, a former marine hose executive, was recruited as a new ‘member’ to act, through his company PW Consulting, as the Cartel’s co-ordinator. Whittle’s role as co-ordinator was described later by British judge Geoffrey Rivlin when passing sentence in Southwark Crown Court in 2008:25 This was a full-time job. The cartel was run as it had to be, with meticulous attention to detail. Code names were used, clandestine meetings were organised and held, agreements were reached, both in relation to the market share and bogus contract bids […] You were very deeply involved in all this dishonesty; indeed it formed the basis of your whole working life […] You were the co-ordinator, and you did that job very efficiently […] this was a serious crime and you played a leading role in it.26

Moreover, an internal cartel view of the co-ordinator’s role was provided by a fellow cartel member, Bryan Allison: All communications were filtered to me. I never had any direct access to any of these people really. I never spoke to them on the phone or met them. Everything I got was filtered through Peter Whittle who was the coordinator of the cartel. He would receive the tenders from companies and would then administer the cartel by ensuring that particular companies won the tender through a rigged bidding process. In addition he coordinated the creation of global price lists […] would tell people largely what they wanted to hear […] He was basically running the cartel for his own ends rather than the benefit of the members.27

There is a stark contrast in this last example between active participation and coordination and passive membership, which serves to complicate judgement of responsibility and liability arising from cartel participation, both at the corporate and individual levels. Another example of such facilitator modelling brings together a senior executive and facilitator role in the person of one individual, and then raises somewhat different legal argument

Harding and Edwards (n 4) 125 et seq. ibid. 26 R v Whittle, Brammar and Allison, Unreported Judgment, Southwark Crown Court, 10 June 2008. Whittle was sentenced to a three-year prison term, and was disqualified under the Companies Directors Disqualification Act 1986 for a period of seven years. 27 M. O’Kane, ‘Does Prison Work for Cartelists – The View from Behind the Bars’ (2011) 56 Antitrust Bulletin 483, 490. 24 25

164  Research handbook on cartels regarding liability and sanctions. British businessman Ian Norris had been the chief executive of the company Morgan Crucible, a participant in the Carbon Products Cartel, and there was evidence of his clear role in a ‘cover-up’ operation: setting up a task force to destroy or conceal incriminating evidence and preparing a script to be used in the event of verbal questions.28 Attempts to secure his extradition to stand trial for the Sherman Act offence in the US were frustrated by an absence of double criminality as between that offence and the only applicable British offence at the time of the cartel, conspiracy to defraud.29 The US Department of Justice neatly changed tack and charged Norris with obstruction of justice, and this was held to be sufficiently serious to justify the extradition (leading eventually to conviction and a prison term for Norris in the US).30 The Norris example demonstrates the possibility of some clear awareness and involvement in a cartel activity at senior management level (‘boss facilitation and direction’), but also how a distinction may be drawn between measures such as evidence destruction on the one hand and briefing which might be cast as legal or even compliance training on the other hand. Moreover, it takes the discussion back to collusive conduct, comprising antitrust awareness and defiance of the rules as a central element in the offending conduct. The same may be said of the role of consulting companies such as Treuhand, whose role is to actively promote and sustain the cartel and should not be cast in any way as the provision of legal or financial advice of the kind supplied by defence lawyers or financial institutions. Evidence from the EU case law relating to successive proceedings involving the Swiss consultancy firm Treuhand (and its antecedent, Fides Trust AG) from 1980 through to 201531 has shown that this facilitating role was consistent over time in relation to a number of cartels: the organization of meetings, often safely outside the EC/EU jurisdiction in Zurich, and the supply of logistical assistance. The firm carried out calculations for market shares (sometimes referred to as ‘pink’ and ‘red’ papers) identifying the shares of the producer companies and which were kept securely at Treuhand’s office. It was involved in the supply of data to cartel members, and would sometimes carry out a diplomatic role, offering to moderate in the event of tensions and disputes regarding shares and prices and to work out compromise solutions. Treuhand would arrange for the reimbursement of travel expenses for individual participants at meetings, in order to disguise the purpose of such travel and meetings. In summary, the role of Treuhand was ‘to ensure the effective operation of the cartel, while also working to ensure its secret character and as such was consciously law-defying’.32 This amounts to an active and decisive role in the organization of cartels, far removed from ‘passive’ or ‘bystander’ involvement. For a summary of the Norris legal saga, see Harding and Joshua (n 7) 295–96. See generally P. Whelan, ‘Resisting the Long Arm of Criminal Antitrust Laws: Norris v. US’ (2009) 72(2) Modern Law Review 272. 30 The cartel was in operation prior to the enactment (via the Enterprise Act 2002) of the statutory UK Cartel Offence. See Norris v United States (2008) UKHL 15; and Harding and Joshua (n 7). 31 Thus surely raising the question whether the firm may be characterized as a ‘recidivist’ player in this respect. The Commission has found evidence of Treuhand’s facilitation in this way in relation to three separate cartels: Italian Cast Glass [1980] OJ L383/19; Organic Peroxides [2005] OJ L110/44; Case COMP/38589, Heat Stabilisers, Commission decision, 11 November 2009, C(2009)8682 final. Official legal sources are the two reported judicial proceedings: Case T-99/04, AC-Treuhand AG v Commission [2008] ECR II-1501; and Case C-194/14P, AC-Treuhand v Commission ECLI:​EU:​C:​2015:​ 717. 32 C. Harding, ‘Capturing the Cartel’s Friends: Cartel Facilitation and the Idea of the Joint Criminal Enterprise’, (2009) 34 European Law Review 298, 301. 28 29

Cartel facilitation  165 Nor is Treuhand the only such consultancy firm engaged in such activity. More recently, in 2015, the Commission fined the consultancy firm Icap Management Services almost £15 million in relation to its ‘coordination’ via six instances of a daily spreadsheet of Yen LIBOR rates in the Yen LIBOR Cartel distributed to participants in the cash deposit Japanese Yen active market, which had a decisive influence on the conduct of the banks. Although Icap’s fine was later annulled by the General Court and then also the Court of Justice,33 for insufficient reasoning regarding the calculation of the fine, that did not serve to deprive Icap, as an interdealer broker and provider of post-trade services, of its facilitator role.34 Certainly, in a more practical sense, cartel facilitation, whatever the form and status of the persons involved, is at the heart of cartel activity. Insofar as ‘facilitation’ might suggest an ancillary role, it may give a misleading impression – indeed, ‘management’ or even ‘direction’ in some instances may be more accurate descriptors.

V.

CARTEL FACILITATION AS A SPECIES OF DELINQUENCY AND THE CASE FOR DISAGGREGATION

Why has the subject so often been consigned to the periphery of discussion? Insofar as this has not been true, it has usually been in the case of individual cartel defendants, such as Norris and Whittle, who have been castigated by criminal courts as serious offenders deserving to be imprisoned and pursued by determined attempts at extradition or other rendition. This has also in practice been part of the American tradition of personalizing and individualizing anticompetitive behaviour through the Sherman Act concept of conspiracy. On the other hand, in the context of the EU system of regulation, and that of some other national systems still wedded to the idea of administrative offending on the part of corporate actors (for instance, Germany), the approach to disaggregation of the facilitator’s role has been more tentative, as is evident from the line of Treuhand cases. On this view, it is to some extent an issue of agency, whereby it has proven easier to conceptualize and censure as a scheming mastermind an individual (and human) businessperson than a consulting firm. This has led to some odd legal outcomes. On the one hand, a cartel defendant such as Norris was motivated to invoke legal argument arising from the European Human Rights Convention to try to defeat the determined legal pursuit on the part of the US authorities. And among the Marine Hose cartelists, Allison and Pisciotti may have felt sore regarding their treatment and legal sanctioning compared to that of their fellow cartelist Whittle. On the other hand, consulting firms have been able to exploit a European enforcement and appellate system which has encouraged an expectation of relatively lenient treatment (nominal or quite small financial penalties) and the repetitive deployment of unconvincing legal defence arguments based

33 See Case T-180/15, Commission v Icap Management Services ECLI:​EU:​T:​2017:​795; and Case C-39/18P, Commission v Icap ECLI:​EU:​C:​2019:​584. 34 Icap’s role as an information supplier within the arrangement was clear enough, whatever the outcome of long argument before the EU Courts regarding fine calculation and its explanation in the context of the cartel being dealt with largely through the settlement procedure rather than the imposition of sanctions.

166  Research handbook on cartels upon principles of legal certainty, nullum crimen sine lege and the protection of legitimate expectations.35 To some extent the EU judiciary has contributed to the tentative and uncertain outcome by itself justifying facilitator liability as a passive activity. In its judgment in the second Treuhand case in 2008, the General Court confirmed facilitator liability in these terms: although not active in the market of the relevant product, the consulting firm’s role was that of a passive ‘co-perpetrator’ within the cartel and so was caught by the prohibition. By its conduct it had tacitly approved an unlawful initiative, without publicly distancing itself from the content of that initiative or reporting it to the administrative authorities. The effect of its behaviour was then to encourage the continuation of the infringement and to compromise its discovery. This was to engage in a passive form of participation.36 But this analysis understates the delinquent nature of much that passes under the heading of facilitation and verges on a kind of bystander liability. Admittedly, drawing a more exact dividing line between ‘active’ and ‘passive’ contributions to the operation and success of the cartel may not always be easy.37 It is a familiar problem of organizational liability and substantial and relevant cause and effect within the operation of joint criminal enterprise.38 It would be useful, as a matter of both policy and law, to address the issue of cartel facilitation in those terms. If policy-makers, enforcement agencies, lawyers and economists are serious about the problem of resolving the perceived ill of business cartels, there is an underlying dilemma of agency to be addressed, as indicated in the discussion above. Such cartels need to be understood as complex and continuing organizations, often possessing both personal and corporate structures and roles which may be difficult to penetrate and evidence from the outside. Their operation, globally, may vary very much according to markets, business culture, the political and legal context of enforcement and personal elements. As a matter of legal control, one size does not fit all, yet this has been the global assumption, driven by a faith in strong deterrence and a perception of business cartels as a standard and consistent form of rational actor.39 On the contrary, it may be argued that an effective legal response to the malaise of cartel activity depends upon a more nuanced approach40 which seeks to unpick the organizational complexity of the phenomenon. Having made that point, it is not wholly beyond the wit of legal expertise to attempt some disaggregation of business cartels, even though in some cases issues of evidence may remain challenging and productive of prolonged argument if legal systems tolerate that possibility, as

See generally the arguments of the parties in the Treuhand cases (n 31), and in particular Case C-194/14P, AC-Treuhand v Commision ECLI:​EU:​C:​2015:​350, Opinion of AG Wahl. 36 Case T-99/04, AC-Treuhand AG v Commission [2008] ECR II-1501, para 130. 37 McCulloch for instance, in trying to link offensive cartel conduct with unlawful anticompetitive gain, reduces the matter to an investigation of what the facilitator gains from the cartel in a manner similar to the anticompetitive producer; see A. McCulloch, ‘The “Public Wrong” of Cartels and the Article 101TFEU “Object Box”’ (2020) 65(3) Antitrust Bulletin 361. But to focus on facilitator gain is to state the obvious while at the same time skirting over the surface of substantial contribution to illegal objectives. 38 See generally C. Harding, Criminal Enterprise: Individuals, Organisations and Criminal Responsibility (Willan Publishing, 2007); Harding and Edwards (n 4) 29 et seq; and Harding (n 32). 39 See generally the discussion in Harding and Joshua (n 7). 40 See the argument and conclusions in Harding and Edwards (n 4). 35

Cartel facilitation  167 in the case of the EU system of regulation. Moreover, the advent of leniency programmes, as tools to cartel exposure, has to some extent exacerbated the complications of internal agency, by encouraging cartel participants to ‘game’ with leniency and immunity opportunities.41 And to some extent this remains an obscure aspect of the subject – for instance, to understand how Treuhand-type advising or intervention may relate to a spoken or otherwise awareness that cartel producer members may be constantly weighing up the advantage or disadvantage of cheating on each other in this way. But at the very least, a framework for disaggregation is possible. Different roles and degrees of commitment within cartels may be envisioned and categorized and used as a basis for ordering legal liability. To some extent that is already an established practice in relation to the determination of sanctions applied to cartel members in relation to the proven length of their involvement or any tendency to enter, leave, or re-enter cartels, although, as just noted, successful leniency applications can then disturb (sometimes it might be said, violently disturb) such calculations. But as a general point, as is true of criminal and other forms of legal liability, sanctions may be used to reflect proven perpetrator participation. It is also important, as a matter of both policy and justice, that efforts of legal control take into account participator awareness of wrongdoing as an element of that party’s significant and relevant contribution to harmful outcomes.42 In that respect, the established contrast between ‘light’ treatment of ‘knowing’ advisers and ‘heavy’ treatment of ‘unknowing’ senior levels within large corporate structures (typical of the EU approach)43 should be seen as a matter of concern. Often this has been driven by a deterrence-induced desire to increase the level of market share and fines based on the latter, rather than being clearly justified by any principle of vicarious or supervisory responsibility or removal of opportunistic gain. Due diligence defences should at least be considered in such cases, since there is sometimes clear evidence of ‘senior’ surprise and dismay regarding the existence of cartel activity in other parts and levels of corporate structures and organizations.44 Anti-cartel enforcement has become riddled with inconsistency in this respect – for instance, by rewarding established corporate recidivists who have become expert at leniency gambling,45 or in the determined international hounding of certain individuals, such as Romano Pisciotti,46 as ‘poster-boys’ for a proclaimed policy and faith in deterrence. Criteria of proximity and relevance may also be more fully considered as aids to the disaggregation of core involvement in cartel activity. For instance, it is almost a matter of common-sense reasoning to separate consulting firm awareness and management of cartel strategies from a vague and uncertain ‘bystander’ suspicion in the case of the example some-

41 See in particular: C. Harding, C. Beaton-Wells and J. Edwards, ‘Leniency and Criminal Sanctions in Anti-Cartel Enforcement: Happily Married or Uneasy Bedfellows?’, in C. Beaton-Wells and C. Tran (eds), Enforcement in a Contemporary Age: The Leniency Anti-Cartel Religion (Hart Publishing, 2015). 42 Applying such basic principles as legal certainty, protection of legitimate expectations and nullum crimen sine lege. 43 See the discussion in Harding (n 32). 44 For instance, in the case of the Art House Auctions Cartel (Sotheby’s/Christie’s), the infringement was very much in the nature of a ‘frolic’ on the part of a small group of top executives and once it had come to light other members of the two organizations were very quick to disown the cartel. See Harding and Edwards (n 4) 167 et seq. And see the examples given by Harding (n 32) 308–09. 45 See Harding, Beaton-Wells and Edwards (n 41). 46 Crofts and Nylen (n 17).

168  Research handbook on cartels times given of hotel owners or airport authorities who may idly muse about business guests or travellers talking about cartel plans while on their premises. There is now enough usable precedent in relation to atrocity offending (such as the commission of war crimes and crimes against humanity) and joint criminal enterprise to guide the development of concepts of proximate and relevant activity to count as core offending in the context of business delinquency.47

VI.

CARTEL FACILITATION AS A MATTER FOR FURTHER INVESTIGATION AND DEBATE

In summary, this may be presented as a subject which merits some reconfiguration and new insights, to be brought from the fringe and into the mainstream of reflection of what it means to talk about business cartels and their internal dynamic and organization. Some of those described as facilitators have been key players and it would be appropriate to describe their role as one of management, even sometimes direction. As such, it behoves an effective system of legal control to recognize that aspect of internal cartel dynamic and consider the possibilities of disaggregation within the complex agency of cartel operation. A closer study of ‘facilitation’ reveals the extent to which the Sherman Act device of conspiracy enables different outcomes of legal liability compared to a European pursuit of corporate offending and the pursuit of large corporate fines. Put another way, while the former system, by criminalizing in that way, emphasizes a behavioural offence and elements of ‘antitrust awareness’, the latter is more concerned with anticompetitive market effect, however that may have come about. This suggests a need to reconsider policy and enforcement objectives – to examine more closely the role of the person who hides away the spreadsheet rather than that of an unknowing senior executive in a head office in another country. But at the same time, that will require a determined research effort to penetrate activities and interrelations within sometimes large and complex organizational structures.

See Harding (n 32) 306–07.

47

10. The concept of a ‘hub-and-spoke conspiracy’ Mark Furse

I. INTRODUCTION The term ‘hub-and-spoke conspiracy’ can be contested. Although seductive in its simplicity, ‘homespun metaphors for complex economic activities only go so far’.1 Hub-and-spoke cartels (or, in the United States, ‘hub-spoke-and-rim conspiracies’2) embrace a melange of legal issues, complex factual analyses and evidentiary complexities which merit discussion. These complexities flow from the fact that hub-and-spoke cartels achieve unlawful horizontal coordination by a series of vertical agreements or understandings: they ‘sit at the crossroads of various theories of harm’.3 In a pure form, horizontal coordination is achieved without direct communication at the horizontal level, such that the arrangement may also be characterized as a species of information exchange,4 or a series of vertical agreements. In much of the enforcement discussed below, the form is not pure, with at least indirect contact (which is to be distinguished from no contact) at the horizontal level. Where contact between the parties is purely vertical, with no direct contact between the horizontal competitors, it should not be assumed that harm has been created, although if the hub is in possession of market power this assumption may not hold.5 As Botteman puts it: ‘[a]dvising and investigating potentially illicit hub-and-spoke exchanges is not for the faint hearted.’6

In In re Musical Instruments and Equipment Antitrust Litigation (Guitar Center) 798 F.3d 1186 (9th Cir. 2015) 1192. 2 See ABA Section of Antitrust Law, Antitrust Law Developments (8th edn, 2017) 19. 3 G.L. Zampa and P. Buccirossi, ‘Hub and Spoke Practices: Law and Economics of the New Antitrust Frontier?’ (2013) 9 Competition L Int’l 91, 92. 4 See P. Whelan, ‘Trading Negotiations Between Retailers and Suppliers: A Fertile Ground for Anti-Competitive Horizontal Information Exchange?’ (2019) 5 European Competition Journal 823, 833 (arguing that ‘the exchange of information between retailer and supplier may indeed be conducted without creating a “hub-and-spoke” arrangement, in certain limited circumstances such an exchange can be conceptualised as an exchange of information that is horizontal in nature (ie between the retailer and her competitor). The challenge here is to identify the detail of those circumstances’). 5 See A. Overd, ‘Effects Analysis in Hub and Spoke Cartels’ (2010) www​.crai​.com/​insights​-events/​ publications/​effects​-analysis​-hub​-and​-spoke​-cartels [accessed 28 February 2022]. Amore argues that a hub-and-spoke theory of harm is conditional upon the exercise of market power, which is ‘necessary, very important, yet probably not sufficient’: R. Amore, ‘Three (or More) Is a Magic Number: Hub & Spoke Collusion as a Way to Reduce Downstream Competition’ (2016) 12 European Competition Journal 28, 43. Space does not permit the economic effects of hub-and-spoke arrangements to be dealt with in detail in this chapter. An outline of the issues is provided in: OECD, Roundtable on Hub-and-Spoke Arrangements – Background Note (DAF/COMP(2019)14) www​.oecd​.org/​daf/​competition/​hub​-and​ -spoke​-arrangements​.htm [accessed 28 February 2022] Section 2.2 (paras 11–42). 6 Y. Botteman, ‘National Competition Authorities’ Investigations in Hub-and-Spoke Arrangements: A Critical Review’, in D. Gerard, M. Merola and B. Meyring (eds), The Notion of Restriction of Competition: Revisiting the Foundations of Antitrust Enforcement in Europe (Bruylant, 2017) 164. 1

169

170  Research handbook on cartels There is a substantial literature about hub-and-spoke cartels, much of which is referenced in this chapter.7 Notwithstanding that enforcement action and cases involving hub-and-spoke arrangements have been limited in both the United States and in the European Union,8 the Organisation for Economic Cooperation and Development (‘OECD’) held a roundtable on the subject in December 2019.9 A hub-and-spoke cartel is one in which there exist vertical relationships between a firm at level A and firms which are competitors at level B, in which, through a series of bilateral vertical contacts, the competitors at level B are brought into the position of operating as a cartel (typically, but not necessarily, by price-fixing). In the language of the US Court of Appeals, Seventh Circuit, the initiating firm at level A acts as a ‘ringmaster’, corralling the firms at level B, which need not have contact with each other.10 The benefits to the participants of such an arrangement are clear: the ‘hub creates collusive efficiency by reducing the need for horizontal coordination’, and ‘vertical agreements entered to effectuate the horizontal agreement may be harder for authorities to detect’.11 Where relationships between horizontal competitors attract antitrust scrutiny, vertical relationships are less likely to do so. The direction of flow in a hub-and-spoke arrangement will determine whether the restraints ensuing are inter- or intra-brand in nature. While legal risks of operating a hub-and-spoke conspiracy may be lower than is the case in a purely horizontal conspiracy, negotiating the terms of the conspiracy may be made more complex. Not only do the interests of the horizontal competitors have to align, but so too does the interest of the hub participant.12 This may account for the relatively small number of cases, although an explanation may equally lie in the difficulty in detecting such cases.

See, e.g., B. Klein, ‘Inferring Agreement in Hub-and-Spoke Conspiracies’ (2020) 83 Antitrust LJ 127; P. Ray, ‘Vertical Restraints and Collusion: Issues and Challenges’ (2020) 83 Antitrust LJ 1; P. Perinetto, ‘Hub-and-Spoke Arrangements: Future Challenges within Article 101 TFEU Assessment’ (2019) 15 European Competition Journal 281; B. Orbach, ‘Hub-and-Spoke Conspiracies’ (2016) The Antitrust Source 1; N. Sahuguet and A. Walckiers, ‘Hub-and-Spoke Conspiracies: The Vertical Expression of a Horizontal Desire?’ (2014) 5(10) Journal of European Competition Law & Practice 711; O. Odudu, ‘Indirect Information Exchange: The Constituent Elements of Hub and Spoke Collusion’ (2011) 7 European Competition Journal 205; E. Garrity, ‘A New Chapter in Antitrust Law: The Second Circuit’s Decision in United States v Apple Determines Hub-and-Spoke Conspiracy per se Illegal’ (2016) 57 BC L Rev 84; B. Klein, ‘The Hub-and-Spoke Conspiracy that Created the Standard Oil Monopoly’ (2012) 85 S Cal L Rev 459; and E. Prewitt and G. Fails, ‘Indirect Information Exchanges to Hub-and-Spoke Cartels: Enforcement and Litigation Trends in the United States and Europe’ (2015) 1 CLPD 63. For a very full discussion see L. Garrod, J. Harrington and M. Olczak, Hub-and-Spoke Cartels: Why They Form, How They Operate, and How to Prosecute Them (MIT Press, 2021). 8 See Perinetto (n 7) 280, noting investigation ‘only cursorily and occasionally at EU level’. 9 OECD (n 5). A substantial set of materials, including the valuable OECD, Background Note (DAF/COMP(2019)12), is accessible. 10 Toys “R” Us Inc v Federal Trade Commission 221 F.3d 928 (7th Cir. 2000) 934. Here the horizontal conspiracy found expression in a collective boycott, and the arrangement flowed upstream, from a leading distributor to toy manufacturers. 11 ‘Hub-and-Spoke Arrangements – Note by the United States’ (OECD, DAF/COMP/WD(2019)88) para 4. 12 See, e.g., P. Van Cayseele, ‘Hub-and-Spoke Collusion: Some Nagging Questions Raised by Economists’ (2014) 5 Journal of European Competition Law & Practice 164, 165: ‘one should never forget that in most cases the incentives of the supplier and his buyers are not aligned’. 7

The concept of a ‘hub-and-spoke conspiracy’  171 A good example of such an arrangement is found in the United Kingdom case of Argos/ Littlewoods.13 Here, applying the UK’s equivalent of Article 101 of the Treaty on the Functioning of the European Union (‘TFEU’) – the Chapter I Prohibition found in s 2 of the Competition Act 1998 – the Office of Fair Trading (‘OFT’) found that the toys and games manufacturer Hasbro had entered into an agreement and/or concerted practice with two major high-street retailers, Argos and Littlewoods, to fix prices of certain Hasbro products. Argos and Littlewoods had not been in contact with each other; business accounts managers at Hasbro had been in contact separately with Argos and Littlewoods, confirming to the latter the prices which Argos had agreed with Hasbro to adhere to.14 While the terminology may be relatively recent, being widely adopted only in the late 1990s, hub-and-spoke cartels are not new. The case of Interstate Circuit,15 heard by the Supreme Court in 1939, would now be described as a hub-and-spoke cartel, as might the 1983 EU case of AEG.16 Such an arrangement may be dealt with in a number of different ways for the purposes of competition law: (1) as a series of (potentially unlawful) vertical agreements;17 (2) as a horizontal agreement, or conspiracy, between the firms at level B; or (3) as an agreement or conspiracy in which all the firms, at both levels A and B, are involved. Each of these approaches has its difficulties, and the clear distinction offered here between vertical and horizontal restraints may not always be encountered in practice. Thus, in United States v Apple, Inc. the Court noted that ‘although this distinction is sharp in theory, determining the operation of

Decision of the Office of Fair Trading CA98/8/2003, Hasbro UK Ltd, Argos Limited and Littlewoods Ltd Fixing the Price of Hasbro Toys and Games (21 November 2003); upheld as to liability on appeal in Argos Limited and Littlewoods Limited v Office of Fair Trading (formerly the Director General of Fair Trading) [2004] CAT 24. 14 Although securing the appropriate evidence to confirm a horizontal arrangement may be challenging, here the OFT’s job was made easier by an extensive email chain which the investigation uncovered. Among these was an email from one of Hasbro’s sales directors which, in part, was in the following terms: ‘Ian … This is a great initiative that you and Neil have instigated!!!!!!!!! However, a word to the wise, never ever put anything in writing, its (sic) highly illegal and it could bite you right in the arse!!!!’ (Decision of the Office of Fair Trading (n 13) para 73). 15 Interstate Circuit, Inc v United States 306 US 208 (1939). Here, faced with reduced revenues, Interstate Circuit facilitated the agreement of movie distributors to impose minimum pricing on second-run theatres. The Supreme Court found a conspiracy extending to the distributors, holding in part that ‘[a]cceptance by competitors […] without previous agreement, of an invitation to participate in a plan the necessary consequence of which, if carried out, is restraint of interstate commerce is sufficient to establish an unlawful conspiracy’ (ibid 227). See Klein (n 7) 151–62. A later case, United States v General Motors Corp., 384 US 127 (1966), is also significant. 16 Case 107/82 Allgemeine Elektricitats-Gesellschaft AEG-Telefunken AG v Commission [1983] ECR 3151. 17 There is an increasing recognition that a legally and economically formalistic approach of treating vertical and horizontal restraints separately, with different presumptions as to efficiencies generated, anti-competitive injury and hence legality, may be flawed. See, e.g., M. Levenstein and V. Suslow, ‘How Do Cartels Use Vertical Restraints? Horizontal and Vertical Working in Tandem’ (2020) 83 Antitrust LJ 15, 16: ‘[o]bservers often miss the role of vertical relationships because both the economic and legal frameworks encourage us to characterise interfirm relationships as either horizontal or vertical.’ See also M. Levenstein and V. Suslow, ‘How Do Cartels Use Vertical Restraints? Reflections on Bork’s The Antitrust Paradox (2014) 57 J L & Econ S33. 13

172  Research handbook on cartels an agreement can be difficult as a matter of fact and turns on more than simply identifying whether the participants are at the same level of the market structure’.18 Discussing ‘the problem of ambiguity’, Orbach addresses the difficulties which flow when ‘[f]irms often use similar vertical restraints in their relationships with each multiple upstream or downstream trading partners, and those trading partners often respond to the restraints in a similar manner’.19 Distinguishing between legal and illegal arrangements in such situations raises significant evidentiary hurdles, the height of which may differ between regimes: in the US the hurdle is higher than is the case in the EU. In respect of option (1), the law relating to vertical agreements is more permissive in the US than it is in the EU. There are no vertical restraints which fall under the per se illegality of s 1 of the Sherman Act,20 and the Colgate doctrine21 gives manufacturers considerable latitude to set terms and conditions, and to refuse to supply to those who do not comply with them. This may incentivize hub-and-spoke claims, encouraging plaintiffs to seek to demonstrate horizontal conspiracy.22 That the rule of reason is permissive of vertical agreements may not be of significance: Posner suggested in 1981 that the presence of vertical restraints was of no consequence to horizontal illegality.23 In the EU, vertical agreements may be restrictions either by object or effect under Article 101 of the TFEU. The block exemption regulation (‘BER’)24 provides a limited safe harbour where market share thresholds are not exceeded, although certain restrictions, including resale price maintenance (‘RPM’), are not tolerated. Option (2) requires the relevant enforcer or court to establish to the degree required that there is in fact a horizontal agreement or conspiracy between the firms at level B, rather than a series of individual bilateral agreements. This may not be straightforward and is discussed further below. Option (3) provides perhaps the ‘easiest’ route to attack the arrangement on the basis that all the relevant firms are participants, operating a cartel which gives rise to a per se infringement in the US, or which is ‘restrictive by object’ in the EU. This approach avoids the need to flirt with legally diffuse boundaries, but, as is the position with option (2), relies upon evidence as to contact and understanding/agreement being available.

791 F.3d 290 (2d Cir. 2015) 313. Orbach (n 7) 2. 20 Section 5 of the Federal Trade Commission Act (‘the FTCA’) is also relevant to this chapter but will not be treated separately. In Toys “R” Us, Inc v Federal Trade Commission (n 10), discussed below, the court held that the FTCA ‘for present purposes tracks the prohibitions of the Sherman Act’ (ibid 933). 21 United States v Colgate & Co 250 US 300 (1919). 22 See Klein (n 7) 130: ‘as the antitrust analysis of vertical restraints has moved to a rule of reason standard, there is an increased incentive for plaintiffs to convert vertical restraint claims into per se hub-and-spoke claims.’ 23 R. Posner, ‘The Next Step in the Antitrust Treatment of Restricted Distribution: Per Se Legality’ (1981) 48 The University of Chicago Law Review 6. Posner argues that in ‘cases in which dealers or distributors collude to eliminate competition among themselves and bring in the manufacturer to enforce their cartel, or in which vertical restrictions are used to enforce a cartel among manufacturers, can be dealt with under the conventional rules applicable to horizontal price-fixing conspiracies. They are not purely vertical cases, and they would be decided the same way even if purely vertical restrictions were legal per se’ (ibid 22). 24 Commission Regulation 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. 18 19

The concept of a ‘hub-and-spoke conspiracy’  173

II.

THE LAW AND PRACTICE IN THE UNITED STATES

The Supreme Court found a hub-and-spoke conspiracy to be in breach of s 1 of the Sherman Act as early as 1939, in Interstate Circuit.25 A number of factors compel the courts in the US to focus on proving a horizontal conspiracy where a hub-and-spoke conspiracy is alleged to exist. A corollary of the applicability of the rule of reason to vertical restraints, and the Colgate doctrine, is that information exchanges which are purely vertical in nature cannot be condemned, whatever the nature of the information being exchanged. Illegality in hub-and-spoke conspiracies exists in the US only where ‘there is some set of facts showing a connecting agreement among the horizontal competitors’.26 This position differs to that relating to ‘rimless wheel’ conspiracies in other areas of federal law, where it may not matter that different conspirators operate at different levels of a market27 – in antitrust cases market positioning may be crucial to determining illegality. While there is no Supreme Court judgment to this effect, this is the position that has been upheld in all recent circuit cases,28 and earlier cases tentatively suggesting the contrary are unlikely to be reliable. In Leegin the Fifth Circuit held that ‘[i]n the absence of an assertion that the retailers agreed to [resale price maintenance] among themselves, there is no wheel and

Interstate Circuit, Inc v United States 306 US 208 (1939). ABA Section of Antitrust Law (n 2) 20. 27 See Kotteakos et al v United States, 328 US 750 (1946), where the Supreme Court characterised a fraudulent conspiracy as one of ‘of separate spokes meeting in a common center […] without the rim of the wheel to enclose the spokes’ (ibid 755). 28 See, e.g., Impro Products v Herrick 715 F.2d 1267 (8th Cir. 1983), in which the court appeared to suggest that a case could proceed where the plaintiff alleged ‘a single hub-and-spoke conspiracy in which each corporate defendant conspired separately with Dr Herrick, but did not conspire amongst themselves’ (ibid 1272; emphasis added). At 1280 the court cited Elder-Beerman Stores Corp. v Federated Department Stores 459 F.2d 138 (6th Cir. 1972) as to the simple requirement of ‘an overall plan to suppress the plaintiff as a competitor or that each defendant had knowledge that others were involved in the conspiracy’, but upheld the District Court in granting summary judgment for the defendant on the grounds that any evidence submitted was insufficient to find a breach of s 1 of the Sherman Act. In Howard Hess Dental Labs. Inc, v Dentsply Int’l, Inc. 602 F.3d 237 (3d Cir. 2010), the court found that the plaintiffs alleged ‘a hybrid of both vertical and horizontal conspiracies […] sometimes dubbed a “hub-and-spoke” conspiracy’ and upheld the district court’s dismissal of the claim, on the basis in part ‘that the amended complaint lacks any allegation of an agreement among the Dealers themselves’. In In re Insurance Brokerage Antitrust Litigation 618 F.3d 300 (3d Cir. 2010) the court stated that: ‘In all hub-and-spoke conspiracies, the horizontal agreement among the spokes supports the agreements between the hub and each spoke, and vice versa’ (ibid 346). In Pepsico, Inc. v Coca-Cola Co. 315 F3d 101 (2d Cir. 2002) the Court of Appeal upheld summary judgment for the defendant in the district court, holding, at 110, that the plaintiff’s ‘“offer of proof” of an agreement was simply that Coca-Cola assured the IFDs that the loyalty policy would be uniformly enforced and encouraged them to report violations. We agree with the district court that this was insufficient evidence of a horizontal agreement to withstand summary judgment.’ In Total Benefits Planning Agency, Inc. v Anthem Blue Cross & Blue Shield 552 F.3d 430 (6th Cir. 2008) the court held that ‘[t]here is no special exception for applying per se status just because there is a hub and spoke conspiracy; the complaint still must show some horizontal relationship’ (emphasis in original). In In re Musical Instruments and Equipment Antitrust Litigation (Guitar Center) 798 F.3d 1186 (9th Cir. 2015), the Ninth Circuit upheld a finding that, in the absence of any evidence of communication among the guitar manufacturers constituting the spokes, the plaintiff had failed to state a cause of claim. 25 26

174  Research handbook on cartels therefore no hub-and-spoke conspiracy’.29 However, the distinction is not as sharp as the case law cited here suggests. Section 1 of the Sherman Act requires that there be a ‘conspiracy’ in restraint of trade or commerce for the contested arrangement to be unlawful.30 The standard for proving conspiracy was set out by the Supreme Court in Monsanto.31 There must be a ‘conscious commitment to a common scheme [as shown by] evidence that tends to exclude the possibility of independent action’ by the parties.32 The permissive approach to vertical arrangements in the US ensures that demonstrating the required ‘conscious commitment’ to the ‘common scheme’ when there may be no contact between the horizontal participants in an alleged hub-and-spoke conspiracy is not straightforward. For their participation in a hub-and-spoke cartel to be unlawful, defendants must be shown to have entered into the arrangement knowingly, or intentionally, and must intend to participate in an action prohibited under s 1. This latter test goes not to the unlawfulness of the act under antitrust law, but to ‘the perpetrator’s knowledge of the anticipated consequences’.33 The only way to ensure that a hub-and-spoke cartel is caught within the s 1 rubric is by securing the evidence demonstrating conspiracy at the horizontal level. Vertical arrangements which are otherwise unobjectionable may form part of this evidence34 but are unlikely, in and of themselves, to be sufficient to establish a violation. There is but a small number of cases in the US where a hub-and-spoke conspiracy has been found to constitute a conspiracy to restrain trade. The two leading cases discussed below relied on the DOJ’s civil enforcement of s 1 of the Sherman Act, and on the FTC’s enforcement of the FTCA. In US v All Star Industries35 the Department of Justice (DOJ) successfully prosecuted a bid-rigging conspiracy to fix the prices of speciality pipe supplied in relation to cost-plus contracts. The district court handed down substantial fines, imposed a restitution order and sentenced three defendants to jail terms. As stated by the Fifth Circuit, the case was unique in ‘that the pipe distributors did not, as in a conventional case, come together voluntarily to fix prices […] directions flowed the other way. It was the suppliers’ customer […] which directed its suppliers in the scheme.’36 The Court rejected the appellants’ argument that ‘they dealt only with [their customer] and not with each other’,37 finding that they were ‘fully aware of their role in the overall scheme’.38 Succinctly, the Court stated that the ‘defendants cannot escape the per se rule simply because their conspiracy depended upon the participation

PSKS, Inc v Leegin Creative Leather Products 615 F.3d 412 (5th Cir. 2010) 420. See also, inter alia, Dickson v Microsoft Corporation 309 F.3d 193 (4th Cir. 2002) 203: ‘a wheel without a rim is not a conspiracy.’ 30 See generally ABA Section of Antitrust Law, Proof of Conspiracy under Federal Antitrust Laws (2010). 31 Monsanto Co v Spray-Rite Service Corporation 465 US 752 (1984). 32 ibid 768. 33 See United States v United States Gypsum Co 438 US 422 (1978) 446: ‘Where carefully planned and calculated conduct is being scrutinized in the context of a criminal prosecution, the perpetrator’s knowledge of the anticipated consequences is a sufficient predicate for a finding of criminal intent.’ 34 See Orbach (n 7) 4: ‘attempts to draw a sharp distinction between the rim requirement and rimless wheel theories are misleading. Circumstantial evidence (plus factors) may be used to establish the existence of the rim, and vertical coordination is a critical aspect of that circumstantial evidence.’ 35 962 F.2d 465 (5th Cir. 1992). 36 ibid 469. 37 ibid 470. 38 ibid note 11. 29

The concept of a ‘hub-and-spoke conspiracy’  175 of a “middle-man”, even if that middleman conceptualised the conspiracy, orchestrated it by bringing the distributors together around contracts it had with its buyers, and collected most of the booty’.39 The leading case in recent years is one arising under s 5 of the FTCA (in which the term ‘hub-and-spoke conspiracy’ is never used), Toys “R” Us Inc v Federal Trade Commission,40 on appeal from the FTC’s ruling Toys “R” Us, Inc.41 The FTC’s complaint stated that Toys “R” Us (‘TRU’), the nation’s largest, but not cheapest, toy retailer, had responded to price competition from ‘warehouse clubs’ to secure ‘agreements or understandings’42 with suppliers which had the result of restricting their sales to warehouse clubs. In the face of reluctance by some manufacturers to surrender sales unless their competitors did likewise, ‘TRU facilitated understandings among competing manufacturers to achieve substantial unity of action among them’;43 in short, there was in place a hub-and-spoke conspiracy with a downstream powerful retailer as the hub, and upstream toy manufacturers and suppliers as the spokes, with a rim constructed of ‘understandings’.44 The administrative law judge found that TRU first began to discuss the operation of the clubs with its suppliers in 1989–91 and had in 1992 disseminated its ‘warehouse club policy’, under which it would not deal with suppliers to the clubs unless the clubs were carrying the manufacturer’s entire line. Manufacturers were required to offer ‘specials and exclusives’ to TRU first; special packing was required for old and basic products; and clearances and closeouts would be permitted only if TRU was given the first opportunity to buy the product. The announced position made it clear that there was to be no discussion about prices. In and of itself, this policy would be lawful, falling within the Colgate safe harbour.45 Judge Timony found that the ‘manufacturers did not want to give up sales and they were also concerned that their competitors would gain share at their expense’.46 TRU then ‘tried to obtain a coordinated response from manufacturers by assuring them that TRU was applying its policy to each of its competitors and by telling each of the major manufacturers that its competitors were only selling to the clubs because the other was’.47 TRU then went further than this, acting as a conduit between manufacturers, passing on ‘the quid pro quo (ie, I’ll stop if they stop) from manufacturer to manufacturer’.48 Further, TRU ‘used the acquiescence of certain manufacturers to obtain the acquiescence of others’.49 Complaints made by one manufacturer about another were passed on, ‘allowing TRU to monitor compliance with the agreements and [assuring] the manufacturers that their competitors were complying’.50 In 1992 or early 1993 TRU changed its policy again, telling manufacturers that it would simply discontinue purchases of any products sold to the clubs. Enforcement mechanisms consistent with those out-

ibid 473. 221 F.3d 928 (7th Cir. 2000). 41 126 FTC 415 (1998). 42 ibid 416. 43 ibid 417. 44 See Amore (n 5) 37–39. 45 That this would be the position is acknowledged by the judge at 498. 46 126 FTC 415 (1998) 430. 47 ibid 431. 48 ibid 432. 49 ibid. 50 ibid 434. 39 40

176  Research handbook on cartels lined above continued to be applied. After analysing the agreements with the 14 involved toy manufacturers,51 Judge Timony found that the ‘purpose of the agreements […] was to restrain competition among toy retailers and among toy manufacturers’, and that the ‘campaign had its intended effect’.52 TRU’s argument that its approach ‘was justified as an effort to protect against free-riding’53 were rejected by the judge. While, as noted above, the Colgate doctrine would protect any firm announcing and applying a policy which would terminate distributors or suppliers, the doctrine does not extend to permitting a firm taking ‘the affirmative action of asking for or inducing acquiescence to its policy’.54 The evidence was that there were vertical agreements in place, not simply the imposition of policies. Toys “R” Us was heard before the Supreme Court’s Leegin judgment (discussed further below). Judge Timony was clear, based on the law at the time, that non-price vertical restraints were governed by the rule of reason, and that this would apply were TRU’s conduct ‘solely vertical and not motivated by price competition’.55 Here however, in the words of the Court of Appeals, the agreements ‘flunked scrutiny’.56 The relevant case law was driven by concerns as to the power of inter-brand competition to discipline intra-brand restraints. In the present case, however, the restraints related to inter-brand competition, and the applicable case law ‘did not take conspiracy out of the antitrust laws’.57 A horizontal conspiracy was in place among TRU’s suppliers, and this was orchestrated by TRU.58 While this was not a price-fixing agreement, it restricted sales, amounted to a boycott and was to be treated as per se illegal. TRU appealed. In a refreshingly short judgment, the Seventh Circuit unanimously affirmed the FTC’s decision. The central question for the Court was that of whether the Commission was right to find that there was ‘a horizontal agreement among the toy manufacturers, with TRU in the centre as the ringmaster, to boycott the warehouse clubs’.59 The Court found that while the Commission had relied on Interstate Circuit, the present case ‘if anything presents a more compelling case for inferring horizontal agreement than did Interstate Circuit’.60 The Court was clear: ‘the only condition on which each toy manufacturer would agree to TRU’s demands was if it could be sure its competitors were doing the same thing. That is a horizontal agreement.’61 Economic efficiency justifications for the vertical restraints were set to one side, and, as Klein has stated, ‘[t]he question whether there is a procompetitive efficiency rationale for the TRU contract demands is distinct from the question whether a horizontal agreement was reached’.62 Toys “R” Us was cited by Judge Cote, of the Second Circuit, in United States v Apple, Inc,63 in which Apple was found to be in breach of s 1 of the Sherman Act in a case which, like

53 54 55 56 57 58 59 60 61 62 63 51 52

ibid 437–76. ibid 477. ibid 488. ibid 498. ibid 513. 221 F.3d 928 (7th Cir. 2000) 930. 126 FTC 415 (1998) 514. ibid 507. 221 F.3d 928 (7th Cir. 2000) 934. ibid 935. ibid 936. Klein (n 7) 144. United States v Apple Inc. 952 F.Supp. 2d 638 (SDNY 2013).

The concept of a ‘hub-and-spoke conspiracy’  177 Toys “R” Us, dealt with inter-brand competition.64 The characterization of the conduct in this case as a ‘hub-and-spoke’ conspiracy may be contested: Judge Cote argued that the conduct complained of ‘is not properly viewed as either a vertical price restraint or solely through the lens of traditional “hub-and-spoke” conspiracies’.65 The DOJ and others alleged that Apple had committed a per se violation of s 1 by conspiring with five publishers of e-books to raise prices, particularly of New York Times bestsellers. The publishers settled the case, signing consent decrees. Under the terms of the agreements reached between Apple and the publishers, the prices of e-books offered by Apple to consumers were higher than was the case with sales through Amazon (which had set a ‘wretched’ price of $9.9966), but the publishers received less per title sold. The agreements incorporated most favoured nation (‘MFN’) clauses, requiring the publishers to offer Apple a price no lower than that of any of its competitors in the e-books market. The publishers then took control of pricing on Amazon, raising the prices of books sold for Amazon’s Kindle.67 Judge Cote recognized that the challenge facing the plaintiffs in the case was that set out by the Supreme Court in Interstate Circuit:68 ‘[w]here a vertical actor is alleged to have participated in an unlawful horizontal agreement, plaintiffs must demonstrate both that a horizontal conspiracy existed, and that the vertical player was a knowing participant in that agreement and facilitated the scheme.’69 After an extensive discussion of the background to the case, including the evidence relating to the conduct of Apple and the publishers, Judge Cote turned to the applicable legal standard. Monsanto Co v Spray-Rite70 required that ‘proof of joint or concerted action’ was required to establish a conspiracy in violation of s 1.71 A short discussion of the place of the rule of reason and the per se rule places the emphasis quickly on the application of the per se rule to restraints ‘that would always or almost always tend to restrict competition and decrease output’.72 Judge Cote notes here that ‘price-fixing agreements or agreements to divide markets that are horizontal in nature – meaning that the parties to the agreement are “competitors at the same level of the market structure,” – are per se unlawful’.73 Judge Cote then turns to the ‘direct or circumstantial evidence’74 that may be required to establish a conspiracy. The judge held that to prove an antitrust conspiracy ‘the antitrust plaintiff should present direct or circumstantial evidence that reasonably tends to prove that the [defendant] and others had a conscious commitment to a common scheme designed to achieve an unlawful objective.’ The evidence must also ‘prove

ibid 690: ‘Per se price-fixing agreements may also include those where a vertical player participates in and facilitates a horizontal conspiracy.’ See Toys “R” Us, Inc v Federal Trade Commission (n 10). 65 ibid 707. See also OECD, ‘Summary of Discussion of the Roundtable on Hub-and-Spoke Arrangements’ (DAF/COMP/M(2019)2/ANN3/FINAL): ‘the definition of hub-and-spoke may not apply to the case’ (ibid 9). 66 United States v Apple Inc. (n 63) 691. 67 791 F.3d 290 (2d Cir. 2015) 296. 68 Interstate Circuit, Inc v United States (n 15). 69 952 F.Supp. 2d 638 (SDNY 2013) 690. 70 Monsanto Co v Spray-Rite Service Corporation (n 31). 71 952 F.Supp. 2d 638 (SDNY 2013) 687. 72 ibid 688, citing Leegin Creative Leather Products, Inc. v PSKS, Inc. 551 US 877 (2007) 886. 73 ibid (footnotes omitted). 74 ibid 689. 64

178  Research handbook on cartels defendants had the intent to adhere to an agreement that was designed to achieve an unlawful objective; specific intent to restrain trade is not required.’ […] the evidence must demonstrate a ‘meeting of the minds.’ […] Just as a conspiracy’s ‘failure to achieve its ends’ after an intended period may be ‘strong evidence’ that the conspiracy did not in fact exist, the success of the conspiracy in achieving its goals may confirm the very existence of the conspiracy.75

Relying on testimony and records of meetings, telephone calls, numerous emails and transcripts of interviews, Judge Cote determined that there was ‘compelling evidence that Apple violated s 1 of the Sherman Act’ and ‘overwhelming evidence that the publisher defendants joined with each other in a horizontal price-fixing conspiracy’;76 that the conspiracy ‘would not have succeeded without the active facilitation and encouragement of Apple’;77 and that the evidence overwhelmingly demonstrated that Apple did not act independently.78 The terms that Apple offered to each publisher in new ‘agency Agreements’ were, ‘simply put, “terrible”’,79 but provided the framework within which the publishers would, in concert, change the terms on which they did business with Amazon. This was something that the publishers had been trying to achieve for months, and with the push from Apple were able to achieve ‘in a matter of weeks’.80 The efficacy of the approach was summed up by the judge in these terms: ‘Working together, and equipped with Apple’s agency agreements, Apple and the publisher defendants moved the largest publishers of trade e-books and their distributors from a wholesale to agency model, eliminated retail price competition, and raised e-book prices.’81 Apple’s formalistic defence to the action – that the focus should be on the terms of each (vertical) agreement, and that there was ‘nothing inherently illegal in those terms or the contract as a whole’82 – was brushed aside by Judge Cote, who pointed to the extensive negotiations leading to the conclusion of the agreements ‘when the conspiracy and Apple’s participation in it took shape’, and ‘the weeks that followed, during which the import of the agreements became apparent’.83 Apple appealed, unsuccessfully.84 Accepting that vertical agreements could constitute evidence from which a conspiracy could be inferred, the court held that ‘the district court did not err in determining that Apple organised a price-fixing conspiracy among the publisher defendants’;85 Apple’s argument that the agreements should be analysed under the rule of reason was therefore bound to fail. As the Court noted, there are a number of cases which appear to have characteristics typical of hub-and-spoke conspiracies in which the characterization

ibid (footnotes omitted) (emphasis added). ibid 691. 77 ibid. 78 ibid 698. 79 ibid 692. 80 ibid 693. 81 ibid. Similar issues were addressed by the EU Commission, although only as part of a wider investigation into e-books, resulting in two decisions (COMP/AT.39847 E-Books, Commission decisions of 12 December 2012 [2013] OJ C73/20 and 25 July 2013 [2013] OJ C378/25). The first decision was addressed to four publishers and Apple, the second to a single publisher. 82 952 F.Supp. 2d 638 (SDNY 2013) 698. 83 ibid 699. 84 791 F.3d 290 (2d Cir. 2015). The appeals court was split 2–1, with Judge Jacobs dissenting. The Supreme Court subsequently denied certiorari (Apple v United States 136 S. Ct 1376 (2016)). 85 791 F.3d 290 (2d Cir. 2015) 321. 75 76

The concept of a ‘hub-and-spoke conspiracy’  179 of the arrangements was that of per se unreasonable restraints on trade with the focus at the horizontal level.86 The relevant agreement in this case was not, the court held, any one vertical agreement incorporating terms on price, but ‘the horizontal agreement that Apple organised among the publisher defendants to raise e-book prices’.87 In essence, the court brushed aside arguments about the categorization of the various vertical agreements where it was clear as a matter of fact, as determined by the circuit court, that a horizontal conspiracy existed: ‘the court need not consider whether the vertical agreements restrained trade because all participants agreed to the horizontal restraint, which is “and ought to be, per se unlawful”.’88 The key passage in this context is the following: In short, the relevant ‘agreement in restraint of trade’ in this case is the price-fixing conspiracy identified by the district court, not Apple’s vertical contracts with the publisher defendants. […] the question is whether the vertical organiser of a horizontal conspiracy designed to raise prices has agreed to a restraint that is any less anticompetitive than its co-conspirators, and can therefore escape per se liability.89

The approach of the appellate court has not escaped criticism.90 In his dissenting opinion in United States v Apple, Inc.91 Judge Jacobs suggested, contra the majority, that Leegin had changed the law in relation to hub-and-spoke conspiracies. His conclusion is based on the Supreme Court’s statement at 893: ‘To the extent a vertical agreement setting minimum resale prices is entered upon to facilitate either type of cartel, it, too, would need to be held unlawful under the rule of reason.’ This argument was dismissed in the majority judgment. The majority doubted that the Supreme Court had intended to overturn in this respect earlier case law,92 but more fundamentally to Apple took the view that the Court ‘need not worry about the possibility that Leegin covertly changed the law governing hub-and-spoke conspiracies, however, because the passage relied upon by the dissent is entirely consistent with holding the ‘hub’ in such a conspiracy liable for the horizontal agreement that it joins’.93

86 The Court of Appeals cites Klor’s, Inc v Broadway-Hale Stores, Inc 359 US 207 (1959), and United States v General Motors Corp. 384 US 127 (1966). 87 791 F.3d 290 (2d Cir. 2015) 323. 88 ibid 325. 89 ibid. 90 See Garrity (n 7) and S. Marietta, ‘An Apple a Day Doesn’t Keep Doctor Miles Away: The Second Circuit’s Misuse of the Per Se Rule in United States v Apple’ (2016) 69 Rutgers UL Rev 325. For a different perspective on this case see Z. Flood, ‘Antitrust Enforcement in the Developing E-Book Market: Apple, Amazon and the Future of the Publishing Industry’ (2016) 31 Berkeley Technology Law Journal 879. The case is also discussed in Office of Fair Trading, The Competitive Effects of Buyer Groups – Economic Discussion Paper (OFT 863, 2007) paras 6.76–6.84. 91 791 F.3d 290 (2d Cir. 2015). 92 ibid 324: ‘If the Supreme Court meant to overturn General Motors and Klor’s – precedents that it has consistently reaffirmed – this cryptic sentence was certainly an odd way to accomplish that result. The Supreme Court does not normally overturn, or so dramatically limit, earlier authority sub silentio’ (internal quotes and footnotes omitted). See also Orbach (n 7) 12: ‘[t]his passage […] does not bear the weight given it by proponents of using the rule of reason’. 93 ibid.

180  Research handbook on cartels An approach similar to that advanced by Judge Jacobs was taken in Toledo Mack94 by the Third Circuit, but has not since been followed in that circuit, or elsewhere.95 Garrity sides with Judge Jacobs, arguing that ‘vertical price agreements, even if embedded in hub-and-spoke arrangements, are treated under the rule of reason to give parties, especially in new industries, the opportunity to present any procompetitive justifications’,96 and argues that the Court’s decision ‘contradicted’97 the Supreme Court’s approach in Leegin.98 The better view is that this is not the case, and that the majority was correct in its interpretation of Leegin.99 Orbach is persuasive: the antitrust jurisprudence of vertical restraints is too rigid and inadequate for cartels facilitated through vertical relationships. In this sense, the analytic framework courts use to analyse the hub-and-spoke conspiracies, underscores some of the limits of the antitrust jurisprudence of vertical restraints.100

It is, in this light, unfortunate that the Supreme Court denied certiorari. The treatment of hub-and-spoke conspiracies in US jurisprudence is superficially straightforward – there must be sufficient evidence of the existence of a conspiracy at the level of the rim – but the analysis in practice remains awkwardly complex. A bifurcation of an overall arrangement into its vertical and horizontal components, with each treated separately for the purposes of legal analysis, prioritizes the convenience of legal formality over economic reality.101 It is this approach that the both the district and appeal courts eschewed in Apple. As has been noted, in an analysis critical of the Appeal Court’s approach, the court ‘steadfastly refused to focus on the formalistic positioning of firms in the marketplace for antitrust analysis’.102 That the totality of actions in Apple gave rise to horizontal coordinated action seems beyond doubt. If, as Posner has argued should be the case, the vertical elements of the overall arrangement were treated as per se legal, and stripped out of the analysis, what is left? Arguably there would be sufficient evidence of ‘plus factors’ to sustain a finding of conspiracy between the book publishers to act in concert in their efforts to alter the terms on which they did business with Amazon: the district court made clear its view that no one individually could successfully do so. This approach is not without its difficulties given the conditions set out in Monsanto. In the absence of robust evidence of conspiracy, such cases are fragile, and the question whether a series of legal vertical agreements could support a finding of illegal horizontal agreement

Toledo Mack Sales & Service, Inc. v Mack Trucks, Inc. 530 F.3d 204 (3d Cir. 2008). See, e.g., Total Benefits Planning Agency, Inc. v Anthem Blue Cross & Blue Shield 552 F.3d 430 (6th Cir. 2008). 96 Garrity (n 7) 87. 97 ibid 102. 98 Similar arguments are made by Marietta (n 90), who states, inter alia, that the ‘legal error stems not from the Second Circuit’s analysis but rather from its ignorance of the Leegin decision, which fundamentally altered the relationship between the per se rule and the rule of reason’ (ibid 372). 99 The judgment in Leegin has itself been the subject of much analysis. See, in particular, A. Jones, ‘Completion of the Revolution in Antitrust Doctrine on Restricted Distribution: Leegin and Its Implications for EC Competition Law’ (2008) 53 Antitrust Bulletin 903. 100 Orbach (n 7) 14. 101 See Toys “R” Us Inc v Federal Trade Commission (n 10) 930: ‘it can be hard as a matter of fact to be sure what kind of agreement is at issue.’ 102 Marietta (n 90) 362. 94 95

The concept of a ‘hub-and-spoke conspiracy’  181 remains open. The court’s view was that it was important that the vertical Apple pricing agreements made sense only on the basis that they formed the basis for a renegotiation of price terms with Amazon – as the court made clear, the publishers received less money per sale under the Apple contracts but would overall be better off in the rearranged market.

III.

THE LAW AND PRACTICE IN THE EUROPEAN UNION

Neither the EU Commission nor the Court of Justice of the EU have directly addressed hub-and-spoke cartels in such terms in enforcement or judgments, although, as noted above, AEG could be so categorized.103 There have been decisions at the national level, determined on the basis of the application of EU law or national equivalents,104 and it is possible to draw out some guiding principles from the relevant principles of EU competition law. In applying EU competition law to hub-and-spoke cartels the competition authorities and the courts do not have to perform a delicate soft-shoe shuffle on the hubcap of the rule of reason. Vertical agreements may be, depending on their terms, restrictions by object or by effect. Information exchange may also be in breach of Article 101 TFEU by object, as will be the case where it relates to future pricing or output. Both the Commission’s vertical and horizontal guidelines make reference, albeit limited, to hub-and-spoke arrangements. Beyond these two references, the key Commission guidelines are silent on the treatment of hub-and-spoke arrangements. In its vertical guidelines brief mention is made of the relationship between hub-and-spoke arrangements and RPM: ‘[s]trong or well organised distributors may be able to force or convince one or more suppliers to fix their resale price above the competitive level and thereby help them to reach or stabilise a collusive equilibrium.’105 As the Commission states in its horizontal guidelines, ‘[i]nformation exchange can take various forms […] data can be shared indirectly through a common agency […] or through the companies’ suppliers or retailer’.106 At para 59 of the guidelines, the Commission states that ‘communication of information among competitors may constitute an agreement, [or] a concerted practice […] with the object of fixing, in particular, prices or quantities’. As is made clear, communication of information among competitors may take place where the transmission mechanism is not competitor-to-competitor, but competitor-to-hub-to competitor. Such indirect transmission does not preclude illegality – it simply makes it almost inevitable that this is more likely to be found in a concerted practice than in an agreement.107

See Allgemeine Elektricitats-Gesellschaft AEG-Telefunken AG (n 16). See also: COMP/AT.40182 Pioneer [2018] OJ C338/19; COMP/AT.40181 Philips [2018] OJ C340/10; and COMP/AT.40469 Denon & Marantz [2018] OJ C335/5. In these latter cases ‘there were indications (but no clear-cut evidence) that, in some instances RPM may have been driven by retailers who informed their supplier about low prices applied by other retailers and requested it to intervene to ensure a certain price level’: Note by the European Union (OECD, DAF/COMP/WD(2019)89) para 7. 104 At the OECD roundtable held in December 2019 (n 5), national contributions were made by Austria, Belgium, Germany, Greece, Hungary, Latvia, Portugal, Sweden and the UK. 105 European Commission, Guidelines on Vertical Restraints [2010] OJ C130/1, para 224. 106 European Commission, Guidelines on the Applicability of Article 101 of the Treaty on the Functioning of the European Union to Horizontal Co-operation Agreements [2011] OJ C 11/1. 107 The approach under EU competition law to the exchange of information is rightly more robust than is the case in the US. See e.g. Whelan (n 4) 824: ‘the act of exchanging information is capable of having 103

182  Research handbook on cartels Under Article 101 TFEU an agreement or ‘concerted practice’ is unlawful when, by object or effect, it prevents, restricts or distorts competition, and affects trade between member states.108 A concerted practice is ‘a form of co-ordination between undertakings which, without having reached the stage where an agreement properly so-called has been concluded, knowingly substitutes practical co-operation between them for the risks of competition’.109 This definition does not require the relevant undertakings to operate as horizontal competitors. In Suiker Unie the Court of Justice expanded on this, stating that ‘each economic operator must determine independently the policy which he intends to adopt on the common market including the choice of the persons and undertakings to which he makes offers or sells’.110 In Anic the Court stated that Article 101 TFEU applies ‘to all collusion between undertakings, whatever the form it takes’ and that the ‘only essential thing is the distinction between independent conduct, which is allowed, and collusion, which is not, regardless of any distinction between types of collusion’.111 As with Colgate in the US, purely unilateral acts are not prohibited; it is an essential requirement of Article 101 TFEU that an agreement or concerted practice involve ‘two or more undertakings’. The key judgment in this respect is that of Bayer, in which the Court of First Instance held that a distinction should be drawn between cases in which an undertaking has adopted a genuinely unilateral measure […] and those in which the unilateral character of the measure is merely apparent […] the latter must be regarded as revealing an agreement between undertakings and may therefore fall within [Article 101]. That is the case, in particular, with practices and measures in restraint of competition which, though apparently adopted unilaterally by the manufacturer in the context of its contractual relationship with its dealers, nevertheless receive at least the tacit acquiescence of those dealers.112

In VM Remonts,113 relying in part on Anic, the Court stated that an undertaking may, in principle, be held liable for a concerted practice on account of the acts of an independent service provider supplying it with services only if one of the following conditions is met:

a welfare-reducing effect and […] it should not be exempt a priori from the operation of antitrust law, particularly if consumer welfare is to the primary objective of this law.’ 108 It is not necessary for an infringement to be characterized definitively as either an agreement or a concerted practice and it is common to find phrases such as ‘whether by an agreement or by a concerted practice’ in enforcement decisions. See Cases T-305/94 etc NV Limburgse Vinyl Maatschappij v Commission [1999] ECR II-931, paras 696–98. Botteman (n 6) notes that the UK Court of Appeal, ‘one of the most cited authorities on hub-and-spoke exchanges, does not appear to get too bogged down by having to distinguish between an agreement and concerted practice’ (ibid 150). 109 Case 48/69 ICI v Commission [1972] ECR 619, para 64. 110 Cases 40/73 etc Suiker Unie v Commission [1975] ECR 1663, para 173. 111 Case C-49/92P Commission v Anic Partecipazioni [1999] ECR I-4125, para 108 (emphasis added). 112 Case T-41/96 Bayer v Commission [2000] ECR II-3383, para 71 (references omitted) (emphasis added). 113 Case C-542/14, SIA ‘VM Remonts’ (formerly SIA ‘DIV un KO’) and others v Konkurences padome ECLI:​EU:​C:​2016:​578, para 33 (emphasis added). See I. Apostolakis, ‘Antitrust Liability in Cases of Indirect Contacts between Competitors: VM Remonts’ (2017) 54 Common Market Law Review 605. If adopted, the opinion of Advocate General Wathelet would have created a rebuttable presumption of liability on the part of the spokes of a hub-and-spoke arrangement, where they use services provided by a hub engaged in anti-competitive acts.

The concept of a ‘hub-and-spoke conspiracy’  183 […] that undertaking could reasonably have foreseen the anti-competitive acts of its competitors and the service provider and was prepared to accept the risk which they entailed.

In other words, an undertaking may be party to a concerted practice under Article 101 without being in possession of direct knowledge of illegal activity. The case is not without its ‘analytical challenges’; the judgment is ‘brief and puzzling at times’ but has the potential to be ‘established as the authority for the attribution of antitrust liability to the principal for the acts of an external contractor in cases of hub-and-spoke cartels’.114 In Eturas,115 which has the appearance in part of a hub-and-spoke arrangement, the Court of Justice dealt with a request for a preliminary ruling from the Supreme Administrative Court of Lithuania. The Lithuanian Competition Council had imposed fines on a number of travel agencies and the E-TURAS booking platform, for entering into a concerted practice relating to a cap on discounting. The concerted practice was facilitated by E-TURAS by way of an amendment to the terms and conditions for engagement with the platform which was communicated to participants in the platform by way of an administrator message which, like an email, required opening, and by the subsequent imposition of a technical restriction on discounting.116 Appellants against the infringement decision argued inter alia that they were not aware of the content of the administrator message. The Court held that the presumption of innocence does not preclude the referring court from considering that the dispatch of the message at issue in the main proceedings may, in the light of other objective and consistent indicia, justify the presumption that the travel agencies concerned were aware of the content of that message as from the date of its dispatch.117

If the challenge in acting against a hub-and-spoke arrangement is to prove that there is a concerted practice operating at the horizontal level, the approaches of the Court of Justice outlined above appear to set a bar lower than that which applies in the US. A concerted practice cannot exist in the abstract. In Anic it is made clear that there must be ‘conduct on the market pursuant to those collusive practices’.118 This condition is fulfilled merely by the undertakings’ continued participation in the market. A final condition, however, requires that there is a link between the concerted practice and subsequent conduct, although ‘subject to proof to the contrary […] there must be a presumption that the undertakings participating in concerting arrangements and remaining active on the market take account of the information exchanged with their competitors when determining their conduct on that market’.119 While the bar does not appear to be set very high to establish the existence of a concerted practice, there remain evidentiary difficulties in establishing the intent to form a concerted practice between the spokes, but competition agencies and courts can surmount these. Prior to

Apostolakis (n 113) 606–07. Case C-74/14 ‘Eturas’ UAB and others v Lietuvos Respublikos konkirencijos taryva ECLI:​EU:​C:​ 2016:​42. 116 The message stated that the purpose of the capping policy was ‘to normalise the conditions of competition’ (ibid para 10). 117 ibid para 40. 118 Commission v Anic Partecipazioni (n 111) para 118. 119 ‘Eturas’ UAB (n 115) para 121 (emphasis added). 114 115

184  Research handbook on cartels Brexit, in Argos and Littlewoods (clearly a hub-and-spoke case, even though the term is never used) the UK Competition Appeal Tribunal (‘the CAT’) sums up perfectly the approach to be taken under Article 101 to such arrangements: broad assertions such as ‘it is lawful for a supplier to seek to persuade a retailer to follow RRPs’ or that Hasbro’s pricing initiative was ‘lawful in itself’ do not seem to us to be helpful ways of analysing the facts in the present case. The question at issue is whether there has, as a matter of fact, come into being ‘an agreement or concerted practice having as its object or effect the prevention, restriction or distortion of competition’ […] The issue then is whether what occurred can properly be characterised as purely unilateral action on the part of the supplier, or whether the facts disclose a sufficient degree of consensus to give rise to a relevant agreement or concerted practice […] That issue is a question of fact in each case.120

The case was appealed to the Court of Appeal.121 The subsequent judgment has received significant attention.122 At para 91 the Court held that: The Tribunal may have gone too far if it intended [illegality] to extend to cases in which A did not, in fact, foresee that B would make use of the pricing information to influence market conditions in which C did not, in fact, appreciate that the information was being passed to him with A’s concurrence.123

The Court upheld the CAT’s judgment but moderated the approach to be applied, to that where ‘A may be taken to intend that B will make use of that information’.124 An emphasis on intent is, arguably, at variance with EU jurisprudence, and certainly does not sit well with the later judgment of Remonts. The UK is not alone in having applied competition law to hub-and-spoke arrangements in Europe; in April 2019 the Spanish Competition Authority fined four undertakings engaged in the production and sale of tobacco €57.7 million for their engagement in a hub-and-spoke concerted practice,125 and in December 2020 the Portuguese Competition Authority gifted an early Christmas present in the form of fines totalling €304m to ten parties in two linked decisions finding that there was a hub-and-spoke conspiracy relating to the prices charged for certain beverages. As noted in the press release, ‘through a common supplier, companies ensured the alignment of sales prices to the public.’126 Eturas, discussed briefly above, may transpire to be one of the first in a new chain of cases. It has been recognized that Argos Limited and Littlewoods Limited v Office of Fair Trading (formerly the Director General of Fair Trading) [2004] CAT 24, para 713. The CAT also considered hub-and-spoke arrangements in JJB Sports plc v Office of Fair Trading, and Allsports Limited v Office of Fair Trading [2004] CAT 17 and in Tesco Stores Limited v Office of Fair Trading [2012] CAT 31. 121 Argos Limited and Littlewoods Limited v Office of Fair Trading and JJB Sports Plc v Office of Fair Trading [2006] EWCA Civ 1318. 122 See, inter alia, A. McCabe, ‘The English Court of Appeal’s Legal Test for “Hub and Spoke” Cartels – Is It Compatible with EU Jurisprudence’ (2012) 33 European Competition Law Review 452. 123 Emphasis added. The use of ‘may’ here is not entirely helpful. 124 [2006] EWCA Civ 1318, para 141. 125 www​.cnmc​.es/​node/​374435 [accessed 28 February 2022]. 126 Autoridade da Concorrência, Press Release 22/2020 ‘The AdC Imposed Fines on Six Large Retail Food Chains and Two Suppliers for Price Fixing, Harmful to Consumers’ (21 December 2020) http://​ concorrencia​.pt/​vEN/​News​_Events/​Comunicados/​Pages/​PressRelease​_202022​.aspx​?lst​=​1​&​Cat​=​2020 [accessed 28 February 2022]. 120

The concept of a ‘hub-and-spoke conspiracy’  185 E-commerce and online price comparison tools can facilitate hub-and-spoke arrangements and RPM, in particular as regards the monitoring of an agreement, and speedy reaction to deviations. When sales platforms play a role, cross platform parity agreements can lead to a lessening of competition between horizontal competitors, and platforms could facilitate anti-competitive supplier/seller actions.127

IV. CONCLUSION Taking into account increasingly complex commercial environments, and the new technologies, Perinetto argues that hub-and-spoke infringements, ‘because of their features, may represent the future of [Article 101 TFEU]’.128 His challenging proposal is that enforcement in hub-and-spoke arrangements should always require ‘an in concreto specific assessment of the mental involvement of the parties, of intent’, although he suggests too that ‘enforcers must always beware of relying too much upon intent to reach their conclusions’.129 A chapter examining hub-and-spoke arrangements that is differently written than the present one could foreground intent, which is nearly always challenging in competition law. Were it possible to do so within the legal frameworks, the better approach to hub-and-spoke arrangements would be to set aside the classic approach of analysing each element of the arrangement on the basis of strict legal classifications.130 Following Amore, ‘we should look more broadly at the cumulative effect of such conducts’.131 Not only does the vertical/horizontal distinction fall away in hub-and-spoke arrangements; so too may the distinction between inter- and intra-brand competition. As the cases show, the hub may coordinate competitors at either the downstream or the upstream level.132 The new economy, and two-sided markets in particular, challenge traditional approaches to market analysis and hence to legal analysis – ‘the identification of the position that each player has in the vertical chain relies heavily on which side (ie market) we want to focus on’.133 This is not to suggest that the requirement to establish clearly an infringement of any relevant specific legal provision should be entirely set aside. As Odudu forcefully argues, difficulties arise when there is ‘a failure to anchor the scrutinised behaviour to the legal provision said to be infringed’.134 One cannot wish away the differences between antitrust law in the US and competition law in the EU; attempts to synthesize these beyond the hard-core infringements are likely to fail. Recognizing therefore that legal differences abound, the questions that should be asked of an alleged hub-and-spoke arrangement are the following: (1) Is there an upstream or downstream hub, with horizontal competitors forming the spokes? (2) Is information exchanged between the hub and spokes? (3) Is this information of the sort that, if exchanged between horizontal competitors, could give rise to coordination on future pricing, output, or anti-competitive OECD (n 5) 2. Perinetto (n 7) 315. 129 ibid 316. 130 This is consistent with an economic approach to competition law enforcement. See Sahuguet and Walckiers (n 7) 711 (‘From an economics perspective, there is not always an indisputable motive to classify hub-and-spoke agreements as mainly vertical with a horizontal effect, or mainly horizontal with the involvement of a supplier’). 131 Amore (n 5) 29. 132 ibid 48. 133 ibid. 134 Odudu (n 7) 215. 127 128

186  Research handbook on cartels strategies? (4) Did the hub merely direct the spokes without their agreement to be so directed? (5) Is there evidence to suggest that the spokes acted in the belief or knowledge that the other spokes would do likewise?

11. Algorithmic tacit collusion1 Ariel Ezrachi and Maurice E. Stucke

I. INTRODUCTION The development of advanced algorithms and artificial intelligence raises many challenging legal and ethical questions about the relationship between humans and computers, humans’ control – or lack of it – over computers, and accountability for the computers’ activities. One area which has attracted attention has been the possible effect algorithms may have on market dynamics, and more specifically the way they could be used, or abused, to foster collusion and alignment of price without triggering intervention under current laws. In our earlier writing, we outlined four key scenarios where algorithms may be used to facilitate collusion.2 In 2016 we provided further context and analysis in our book Virtual Competition: The Promise and Perils of the Algorithm-Driven Economy.3 We developed our themes further in submissions to the UK House of Lords4 and OECD.5 Broadly, we have gleaned general consensus over our first two scenarios: Messenger (where algorithms help humans expressly collude) and Hub-and-Spoke (where a common intermediary, which provides the algorithm and the pricing decision mechanism, could facilitate price-fixing).6 Indeed, the European Commission (‘the Commission’) and United States antitrust authorities, among others, raised concerns that algorithms could facilitate collusion,7 and 1 This chapter is based largely upon A. Ezrachi and M. Stucke, ‘Sustainable and Unchallenged Algorithmic Tacit Collusion’ (2020) 17 Nw. J. Tech. & Intell. Prop. 217. 2 A. Ezrachi and M. Stucke, ‘Artificial Intelligence & Collusion: When Computers Inhibit Competition’, Oxford Legal Studies Research Paper No. 18/2015, University of Tennessee Legal Studies Research Paper No. 267 https://​ssrn​.com/​abstract​=​2591874 [accessed 2 March 2022]. 3 A. Ezrachi and M. Stucke, Virtual Competition: The Promise and Perils of the Algorithm-Driven Economy (Harvard University Press, 2016). 4 A. Ezrachi and M. Stucke, ‘Online Platforms and the EU Digital Single Market’, prepared for the UK House of Lords, Internal Market Sub-Committee (16 October 2015). 5 See https://​one​.oecd​.org/​document/​DAF/​COMP/​WD(2017)25/​en/​pdf [accessed 2 March 2022]. 6 See, e.g., Case C-434/15 Asociación Profesional Elite Taxi v Uber Systems Spain SL, ECLI:​EU:​ C:​2017:​364, Opinion of AG Szpunar, n 23; and Meyer v Kalanick, 174 F. Supp. 3d 817 (SDNY 2016) 822–27, reconsideration denied in part, 185 F. Supp. 3d 448 (SDNY 2016). 7 See, e.g., ‘Deputy Assistant Attorney General Richard A. Powers Delivers Remarks At Cartel Working Group Plenary: Big Data and Cartelization, 2020 International Competition Network Annual Conference’ (Washington DC, 17 September 2020) www​.justice​.gov/​opa/​speech/​deputy​-assistant​ -attorney​-general​-richard​-powers​-delivers​-remarks​-cartel​-working​-group [accessed 2 March 2022]; ‘Opening Remarks of FTC Commissioner Rohit Chopra Future of Work Roundtable US House of Representatives Committee on Education & Labor’ (Washington DC, 16 October 2019) www​.ftc​.gov/​ system/​files/​documents/​public​_statements/​1552143/​chopra​_​-​_opening​_remarks​_before​_committee​_on​ _education​_labor​_future​_of​_work​_roundtable​_10​-16​-19​.pdf [accessed 3 March 2022] 2; ‘Algorithms and Collusion – Note from the European Union’ (DAF/COMP/WD(2017)12, OECD, 14 June 2017) 7; ‘Algorithms and Collusion – Note by the United States’ (DAF/COMP/WD(2017)41, OECD, 26 May 2017) 6. That said, it has been acknowledged that, if the competitors independently and unknowingly

187

188  Research handbook on cartels have opened investigations on these scenarios.8 Most policymakers recognize how ‘pricing algorithms may make price fixing attempts more frequent and potentially more difficult to detect’.9 Most say ‘with confidence […] that the rise of pricing algorithms and AI software will require changes in our enforcement practices’; and most would agree that enforcers ‘need to understand how algorithms and AI software work in particular markets’.10 What have sparked debate, however, are our third and fourth scenarios, namely Tacit Collusion on Steroids – The Predictable Agent and Artificial Intelligence, God View, and the Digital Eye. In our third scenario, we noted how, already today, companies could unilaterally use algorithms with the intent to facilitate conscious parallelism. In our fourth scenario we predicted that, in the future, algorithms may arrive at this anticompetitive outcome on their own. Our Predictable Agent and Digital Eye categories raise significant policy issues, including: ● Does our current antitrust policy towards conscious parallelism apply when price optimization algorithms enhance firms’ ability to tacitly collude? ● Is the legal concept of agreement outdated for computer algorithms? ● Are our current antitrust laws sufficient to deter and prevent tacit algorithmic collusion? ● How can the agencies identify when algorithmic collusion occurs, especially when pricing is dynamic? ● What additional measures should be considered to reduce the additional risks associated with the industry-wide use of price optimization algorithms? ● In what ways should firms be obligated to integrate ethics and legality into a computer program? ● Should companies have an affirmative duty to program the computers so as to not tacitly collude? While there is little controversy that tacit collusion is generally beyond the reach of the competition laws of many jurisdictions, including the United States and European Union, some have questioned the likelihood of tacit collusion in either the brick-and-mortar economy or the digital economy. They argue that tacit algorithmic collusion should not pose any concern because collusion is unsustainable without communications.11 The issue is whether companies adopted the same or similar pricing algorithms, this would be ‘unlikely to lead to antitrust liability even if it makes interdependent pricing more likely’: ‘Algorithms and Collusion – Note by the United States’ ibid 6. An interesting issue is whether the competitors would be liable if they intentionally but unilaterally adopted the same algorithm knowing that this would make interdependent pricing more likely. 8 See, e.g., D. Mandrescu, ‘When Algorithmic Pricing Meets Concerted Practices – The Case of Partneo’ (CoRe Blog, 7 June 2018) www​.lexxion​.eu/​en/​coreblogpost/​when​-algorithmic​-pricing​ -meets​-concerted​-practices​-the​-case​-of​-partneo/​[accessed 3 March 2022]; and T. Bergin and L. Frost, ‘Software and Stealth: How Carmakers Hike Spare Parts Prices’ (Reuters, 4 June 2018) www​.reuters​ .com/​article/​autos​-software​-pricing/​rpt​-insight​-software​-and​-stealth​-how​-carmakers​-hike​-spare​-parts​ -prices​-idUSL5N1T60H9 [accessed 3 March 2022]. No formal findings, however, have been found in these cases against the respective carmakers or Accenture. 9 T. McSweeny, Commissioner, US Federal Trade Commission, ‘Algorithms and Coordinated Effects’, University of Oxford Centre for Competition Law and Policy (22 May 2017) www​.ftc​.gov/​ system/​files/​documents/​public​_statements/​1220673/​mcsweeny​_​- _oxford_cclp_remarks_-_algorithms_ and_coordinated_effects_5-22-17.pdf [accessed 3 March 2022]. 10 ibid. 11 U. Schwalbe, ‘Algorithms, Machine Learning, and Collusion’ (1 June 2018) https://​ papers​ .ssrn​.com/​sol3/​papers​.cfm​?abstract​_id​=​3232631 [accessed 3 March 2022]; T. Schrepel, ‘Here’s

Algorithmic tacit collusion  189 in concentrated industries ripe for tacit collusion would have the incentive and ability to develop pricing algorithms for that purpose. Some economists have argued that tacit collusion with three or more rivals – whether by algorithms or humans – is unlikely, as the ‘coordination problems are hard to solve without communication, even in simple static games’.12 According to this view, since algorithms cannot communicate to resolve this coordination problem, they cannot tacitly collude. And because pricing algorithms cannot tacitly collude, the antitrust laws – developed in the old economy – suffice for the digital economy. In this chapter, we explore their criticism and examine why it has not slowed the enforcers’ interest and momentum to tackle the policy issues underlying tacit algorithmic collusion. Indeed, the criticism reveals the widening divide between the law and market realties confronted by enforcers and courts on the one hand, and this economic viewpoint on the other hand. In Section II we outline the theory and the way in which pricing algorithms, in specific market conditions, may foster conscious parallelism. Sections III and IV next consider two areas that have attracted attention. In Section III, we tackle the instability of tacit collusion. Some argue that, absent some communication, tacit collusion is inherently unsustainable. This belief is based on experimental economics and the difficulty of sustaining tacit collusion under certain laboratory conditions. According to this view, the model of tacit collusion will rarely manifest itself in the real world without some supporting communication. The argument goes that this reality should subject ‘facilitated tacit collusion’ to the key EU antitrust law – Article 101 of the Treaty on the Functioning of the European Union (‘TFEU’) – and the key US antitrust provision – Section 1 of the Sherman Act – and resolve concerns as to parallel behaviour that may seemingly escape antitrust scrutiny. Here we see how some economic observations diverge from antitrust law and enforcement policies. When observing the market reality, courts and enforcers on both sides of the Atlantic have seen in the brick-and-mortar economy durable tacit collusion that seemingly occurs without any human communication. Absent evidence of any agreement, the enforcers and courts say that the tacit collusion is legal. Because the conduct is otherwise legal, the primary mechanism to prevent tacit collusion is merger review. Enforcers, when appraising proposed acquisitions, may block mergers that significantly increase the risk of tacit collusion. They expect ‘industry awareness’ to allow conscious parallelism to materialize post-merger without any illicit communications. Put simply, enforcers and antitrust plaintiffs search hard for evidence of express collusion and communication. But they and courts ultimately recognize that anticompetitive parallel behaviour can arise without communications, and thus comfortably occur within the zone of legality. Indeed, the ‘pure’ model of tacit collusion happens with enough frequency that neither the EU nor the US law presume any illicit communication. If this debate was simmering before online commerce and the rise of pricing algorithms, the debate became quite heated over the feasibility of algorithmic tacit collusion. If the algorithms cannot communicate with each other, some argue that algorithmic tacit collusion is nothing to lose sleep over. But many enforcers are worried that tacit algorithmic collusion is likelier and

Why Algorithms Are NOT (Really) a Thing’ (May 2017) https://​ leconcurrentialiste​ .com/​ 2017/​ 05/​ 15/​algorithms​-based​-practices​-antitrust/​[accessed 3 March 2022]; and K.-U. Kühn and S. Tadelis, ‘Algorithm Collusion’ www​.cresse​.info/​wp​-content/​uploads/​2020/​02/​2017​_sps5​_pr2​_Algorithmic​ -Collusion​.pdf [accessed 3 March 2022]. 12 Kühn and Tadelis (n 11); see also Schwalbe (n 11).

190  Research handbook on cartels harder to detect. Section IV thus addresses the debate as to the added risk offered by algorithms (without express communication). We note how humans may program algorithms to reflect the logic behind conscious parallelism – punish deviations and follow price increases. We note how the use of similar algorithms by different firms, and the ability to identify the strategies employed by others, may further stabilize conscious parallelism. Importantly, we explain that when executed carefully and absent illicit communication, these unilateral strategies would not trigger antitrust intervention under current laws. As part of this discussion, we also consider possible future technologies and the capacity of self-learning algorithms to adopt a strategy, which may lead to price increases (absent illegal collusion). The question here is whether, in some future markets, tacit collusion could be sustained without human intervention. This issue is both timely and important. If an antitrust agency accepts the view that tacit collusion is impossible without human communications, then it need not assess the risk of algorithmic tacit collusion. This can play out in two ways. First, rather than keep a close eye on these technological developments and consider potential policy responses, the enforcer would, as some urge, do nothing. It won’t develop algorithmic tacit collusion incubators or conduct market inquiries. It won’t even distinguish between legitimate human tacit collusion and enhanced algorithmic tacit collusion. Nor would it consider what forms of enhancement may be caught as facilitating practices or signalling, or which action may qualify as collusion.13 In short, the agency would continue with its leniency programme for price fixers14 and sniff out cases where humans still conspire. Second, the agency’s merger review will remain incomplete. A primary way to deter tacit collusion is merger review. The agencies lack good predictive models of when a merger significantly increases the likelihood of tacit or express collusion. As one economist explained it to the entering Honors Program lawyers at the US Department of Justice Antitrust Division, ‘the merger occurs and s*** happens’. Not surprisingly, merger review in recent decades has primarily focused on unilateral effects,15 which are relatively easier to model and estimate. But as more markets become more concentrated and more susceptible to tacit collusion, the harm from ignoring (or downplaying) this risk in merger review increases.

II.

ALGORITHMIC TACIT COLLUSION – THE BASE CONDITIONS

Let us first consider the general consensus on tacit collusion. Everyone agrees that it is a challenging area for antitrust enforcement, as it leads to an anticompetitive outcome (namely higher prices, reduced output or allocated markets) without any illegal agreement among competitors.16 As the OECD noted, ‘Although there is great variance in how jurisdictions 13 J. Harrington, ‘Developing Competition Law for Collusion by Autonomous Price-Setting Agents’ (22 August 2017) https://​ssrn​.com/​abstract​=​3037818 [accessed 2 March 2022]. 14 US Department of Justice, Antitrust Division, Corporate Leniency Policy (10 August 1993) www​ .justice​.gov/​atr/​leniency​-program [accessed 2 March 2022]. 15 See, e.g., M. Coate, ‘The Merger Process in the Federal Trade Commission from 1989 to 2016’ (18 April 2017) https://​papers​.ssrn​.com/​sol3/​papers​.cfm​?abstract​_id​=​2955987 [accessed 3 March 2022]. 16 See, e.g., ‘Algorithms and Collusion – Background Note by the Secretariat’ (DAF/COMP/ WD(2017)4, OECD, 16 May 2017) 17; Brooke Group Ltd. v Brown & Williamson Tobacco Corp., 509

Algorithmic tacit collusion  191 interpret the notion of agreement, they traditionally require some sort of proof of direct or indirect contact showing that firms have not acted independently from each other (the so-called “meeting of the minds”)’.17 Tacit collusion has taken another dimension with the proliferation of pricing algorithms. Many competition authorities recognize the risk that algorithms can facilitate and enhance tacit collusion. The OECD in 2016, for example, commented that these strategies ‘may pose serious challenges to competition authorities in the future, as it may be very difficult, if not impossible, to prove an intention to coordinate prices, at least using current antitrust tools’.18 With the industry-wide use of computer algorithms and artificial intelligence, the concern is that algorithmic tacit collusion can arise in markets where such collusion previously would have been unstable. The OECD in 2017 reached the following two conclusions: Firstly, algorithms are fundamentally affecting market conditions, resulting in high price transparency and high-frequency trading that allows companies to react fast and aggressively. These changes in digital markets, if taken to a certain extent, could make collusive strategies stable in virtually any market structure. Secondly, by providing companies with powerful automated mechanisms to monitor prices, implement common policies, send market signals or optimise joint profits with deep learning techniques, algorithms might enable firms to achieve the same outcomes of traditional hard core cartels through tacit collusion.19

Similar concerns as to the possible use of algorithms to sustain tacit collusion have been raised by policymakers and competition agencies (among them, Germany, Italy, France, the United Kingdom, Russia, Israel and Australia).20 Algorithmic tacit collusion – that is, the use of algorithms to execute unilateral and rational reaction to market characteristics that reflect interdependence – will not affect every market, or even most. As Virtual Competition explores, one would expect it in markets with several important characteristics. First, algorithmic tacit collusion likely would arise in concentrated markets involving homogenous products where the algorithms can monitor to a sufficient degree the competitors’ pricing, other key terms of sale and any deviations from the current US 209 (1993); and R.S. Khemani and D.M. Shapiro, Glossary of Industrial Organisation Economics and Competition Law (OECD, 1993) www​.oecd​.org/​dataoecd/​8/​61/​2376087​.pdf [accessed 3 March 2022]. 17 ‘Algorithms and Collusion – Background Note by the Secretariat’ (n 16) 17. 18 ‘Big Data: Bringing Competition Policy to the Digital Era’ (DAF/COMP(2016)14, OECD, 27 October 2016) https://​one​.oecd​.org/​document/​DAF/​COMP(2016)14/​en/​pdf [accessed 3 March 2022] para 81. 19 ‘Algorithms and Collusion – Background Note by the Secretariat’ (n 16) 49–50. 20 See, e.g., German Monopolies Commission (Monopolkommission), Biennial Report XXII: Competition 2018 (3 July 2018); ‘The French Autorité de la Concurrence and the German Bundeskartellamt Launch a Joint Project on Algorithms and Their Implications on Competition’ (19 June 2018) www​.bundeskartellamt​.de/​SharedDocs/​Meldung/​EN/​Pressemitteilungen/​2018/​19​_06​ _2018​_Algorithmen​.html [accessed 3 March 2022]; ‘Algorithms and Collusion – Note from Italy’ (DAF/COMP/WD(2017)18, OECD, 2 June 2017) 2; and Competition and Markets Authority, ‘Pricing Algorithms – Economic Working Paper on the Use of Algorithms to Facilitate Collusion and Personalised Pricing’ (CMA94, 8 October 2018) https://​assets​.publishing​.service​.gov​.uk/​government/​uploads/​ system/​uploads/​attachment​_data/​file/​746353/​Algorithms​_econ​_report​.pdf [accessed 3 March 2022]; www​.oecd​.org/​competition/​algorithms​-and​-collusion​.htm [accessed 3 March 2022]; www​.accc​.gov​ .au/​speech/​the​-accc’s​-approach​-to​-colluding​-robots [accessed 3 March 2022]; and www​.publications​ .parliament​.uk/​pa/​ld201516/​ldselect/​ldeucom/​129/​12908​.htm [accessed 3 March 2022] paras 178–79.

192  Research handbook on cartels equilibrium.21 Software may be used to report and take independent action when faced with a rival’s deviation, be it from the supra-competitive or recommended retail price. Conscious parallelism would be facilitated and stabilized to the extent that: (i) these the rivals’ reactions are predictable; or (ii) through repeated interactions, the firms’ pricing algorithms ‘could come to “decode” each other, thus allowing each one to better anticipate the other’s reaction’.22 As the OECD observed: The increase of market transparency is not only a result of more data being available, but also of the ability of algorithms to make predictions and to reduce strategic uncertainty. Indeed, complex algorithms with powerful data mining capacity are in a better place to distinguish between intentional deviations from collusion and natural reactions to changes in market conditions or even mistakes, which may prevent unnecessary retaliations.23

A second important market condition is that once deviation (for example, discounting) is detected, a credible deterrent mechanism exists.24 Unique to an algorithmic environment is the speed of retaliation.25 Computers can rapidly police deviations and calculate the profit implications of myriad moves and counter-moves to punish deviations.26 The speed of calculated responses effectively deprives discounting rivals of any significant sales. The speed also means that the tacit collusion can be signalled in seconds. The greater the improbability that the first-mover will benefit from its discounting, the greater the likelihood of tacit collusion.27 Thus, if each algorithm can swiftly match a rival’s discount and eliminate its incentive to discount in the first place, the ‘threat of future retaliation keeps the coordination sustainable’.28 Noteworthy are observations made by the European Commission in its 2015–16 e-commerce sector inquiry: About half of the retailers track online prices of competitors. In addition to easily accessible online searches and price comparison tools, both retailers and manufacturers report about the use of specific price monitoring software, often referred to as ‘spiders’, created either by third party software specialists or by the companies themselves. This software crawls the internet and gathers large amounts of price related information. 67% of those retailers that track online prices use (also) automatic software programmes for that purpose. Larger companies have a tendency to track online prices of competing retailers more than smaller ones […] some software allows companies to monitor several hundred

Guidelines on the Assessment of Horizontal Mergers under the Council Regulation on the Control of Concentrations between Undertakings [2004] OJ C31/3 (‘the EU Merger Guidelines’) para 41; and ‘Algorithms and Collusion – Note from Singapore’ (DAF/COMP/WD(2017)24, OECD, 31 May 2017) 2. 22 ‘Algorithms and Collusion – Note from the European Union’ (n 6) para 33. 23 ‘Algorithms and Collusion – Background Note by the Secretariat’ (n 15) 20. 24 The EU Merger Guidelines (n 21) para 41; and ‘Algorithms and Collusion – Note from the European Union’ (n 7) 8. 25 Contrast this with EU Merger Guidelines (n 21) para 53. 26 J. Priluck, ‘When Bots Collude’, The New Yorker (25 April 2015) www​.newyorker​.com/​business/​ currency/​when​-bots​-collude [accessed 2 March 2022]. 27 S. Hwang and S. Kim, ‘Dynamic Pricing Algorithm for E-Commerce’, in T. Sobh and K. Elleithy (eds), Advances in Systems, Computing Sciences and Software Engineering (Springer, 2006) 149–55; N. Abe and T. Kamba, ‘A Web Marketing System with Automatic Pricing’ (2000) 33 Computer Networks 775; and L.M. Minga, Y.Q. Fend and Y.J. Li, ‘Dynamic Pricing: E-Commerce-Oriented Price Setting Algorithm’ (2003) 2 International Conference on Machine Learning and Cybernetic 893. 28 EU Merger Guidelines (n 21) para 52. 21

Algorithmic tacit collusion  193 online shops extremely rapidly, if not in real time […] Alert functionalities in price monitoring software allow companies to get alerted as soon as a retailer’s price is not in line with a predefined price.29

A third condition is that ‘the reactions of outsiders, such as current and future competitors not participating in the coordination, as well as customers, should not be able to jeopardise the results expected from the coordination’.30 Thus, algorithmic tacit collusion will likely arise in concentrated markets where buyers cannot exert buyer power (or entice sellers to defect), sales transactions tend to be ‘frequent, regular, and relatively small’,31 and the market in general is characterized by high entry barriers. A fourth condition is that tacit collusion is more profitable than competition. The algorithm, in maximizing profits, ‘would need to decide that it is a better course of action than competitive pricing, especially if competitive pricing leads to drastically larger sales volumes’.32 When these conditions are present, tacit collusion is likelier. The stability needed for algorithmic tacit collusion is enhanced by the fact that computer algorithms are unlikely to exhibit human biases.33 Human biases, of course, may be reflected in the programming code. But biases will not necessarily affect decisions on a case-by-case basis: a computer does not fear detection and possible financial penalties or incarceration; nor does it respond in anger.34 ‘We’re talking about a velocity of decision-making that isn’t really human’, said Terrell McSweeny, a former Commissioner with the US Federal Trade Commission. ‘All of the economic models are based on human incentives and what we think humans rationally will do. It’s entirely possible that not all of that learning is necessarily applicable in some of these markets.’35 To be clear, no bright line exists as to when an industry becomes sufficiently concentrated for either express or tacit collusion.36 Indeed, competition agencies often struggle in predicting when a merger may facilitate tacit collusion. In addition, it is important to stress that the above phenomenon will affect a select number of markets. Still, when the above conditions

29 European Commission, Commission Staff Working Document – Preliminary Report on the E-Commerce Sector Inquiry (SWD(2016) 312, 15 September 2016) http://​ec​.europa​.eu/​competition/​ antitrust/​sector​_inquiry​_preliminary​_report​_en​.pdf [accessed 3 March 2022] paras 550–51. 30 EU Merger Guidelines (n 21) para 41. 31 Federal Trade Commission and US Department of Justice, Commentary on the Horizontal Merger Guidelines (2006) www​.justice​.gov/​atr/​file/​801216/​download [accessed 3 March 2022]. 32 ‘Algorithms and Collusion – Note from the European Union’ (n 6) 8. As the OECD noted, ‘market stagnation characterised by declining demand and the existence of business cycles may hinder collusion. This is because firms have strong incentives to profitably deviate when demand is high and reducing the costs of retaliation in future periods when demand is low’: ‘Algorithms and Collusion – Background Note by the Secretariat’ (n 16) 20. 33 EU Merger Guidelines (n 21) para 44; and ‘Algorithms and Collusion – Note from Singapore’ (n 21) 2. 34 M. Stucke and A. Ezrachi, ‘How Pricing Bots Could Form Cartels and Make Things More Expensive’ (Harvard Business Review, 27 October 2016) https://​hbr​.org/​2016/​10/​how​-pricing​-bots​ -could​-form​-cartels​-and​-make​-things​-more​-expensive [accessed 3 March 2022]. 35 D. Lynch, ‘Policing the Digital Cartels’ (Financial Times, 9 January 2017) www​.pros​.com/​about​ -pros/​news/​financial​-times​-policing​-digital​-cartels/​ [accessed 3 March 2022]. 36 Note, for example, research by Levenstein and Suslow, who offer several explanations for the lack of a clear empirical relationship between industry concentration and cartels involving express collusion: M. Levenstein and V. Suslow, ‘What Determines Cartel Success?’ (2006) 44(1) Journal of Economic Literature 43.

194  Research handbook on cartels are present, the risk of tacit collusion is high. Importantly, the nature of electronic markets, the availability of data, the adoption of similar algorithms by key providers and the stability and transparency they foster will likely push some markets that were just outside the realm of tacit collusion into interdependence.37 Indeed, software vendors are promoting their price optimization algorithms as a way to avoid price wars and increase prices and margins. Boomerang, for example, promotes how its price optimization software can ‘put an end to price wars before they even begin’.38 As the Italian Competition Authority observed, ‘a number of specialized software developers offer solutions than allow even small companies to implement “strategic” dynamic pricing strategies, offering tools to “auto-detect pricing wars” as well as to “help drive prices back up across all competition”’.39 Ultimately, we may see more instances in which similar pricing is not the result of fierce competition, nor the result of cartel activity, but rather the result of algorithmic tacit collusion. In those affected markets, one may witness the same result as express collusion, namely higher prices (than in a competitive setting), without this triggering antitrust intervention.

III.

THE (IN)STABILITY OF TACIT COLLUSION ABSENT COMMUNICATION

Algorithmic tacit collusion should not be seen as pervading the entire digital economy. It will likely arise in markets with the characteristics discussed in Section II. Some critics, however, have questioned the likelihood of sustainable algorithmic tacit collusion even in these markets. As we explain below, their arguments, however, have failed to persuade enforcers and courts with respect to tacit collusion in the brick-and-mortar economy. Nor are they likely to gain traction in the digital economy. In discounting the possibility of tacit collusion – whether by humans or algorithms – several economists point to earlier scholarship, which highlights the important role of communications in stabilizing and optimizing collusion. They argue that, while collusion without communication may be possible, it is highly unlikely. To their mind, the increase in transparency, speed in retaliation and frequency in contacts are insufficient, even under these market conditions. In markets with more than two companies, some kind of explicit coordination (such as communications) is needed to enter and sustain collusion. In extending the consensus that communication facilitates alignment (the exact level of communication needed remains unclear),40 and that complex market realities would make collusion, and tacit collusion, diffi37 One would expect tacit collusion to be feasible with a larger number of participants than commonly assumed. On the common market assumptions, see generally R. Selten, ‘A Simple Model of Imperfect Competition, Where Four Are Few and Six Are Many’ (1973) 2 International Journal of Game Theory 141; S. Huck, H.-T. Normann and J. Oechssler, ‘Two Are Few and Four Are Many: Number Effects in Experimental Oligopolies’ (2004) 53(4) Journal of Economic Behavior and Organization 435. 38 A. Sathe, ‘How Retailers and Brands Can Avoid the Race to the Bottom in Online Pricing’ (Internet Retailer, 9 July 2018) www​.digitalcommerce360​.com/​2018/​07/​09/​how​-retailers​-and​-brands​ -can​-avoid​-the​-race​-to​-the​-bottom​-in​-online​-pricing/​ [accessed 2 March 2022]. 39 ‘Algorithms and Collusion – Note from Italy’ (n 20) 3. 40 K.-U. Kühn, ‘Fighting Collusion by Regulating Communication Between Firms’ (2001) 16 Economic Policy 167; and Y. Awaya and V. Krishnay, ‘On Tacit versus Explicit Collusion’ (Working Paper, Penn State University, 2014).

Algorithmic tacit collusion  195 cult,41 they argue that absent communication, tacit collusion is unlikely. Their argument is that ‘firms are unlikely to develop a mutual understanding over a collusive strategy absent direct communication in the initiation phase’.42 According to this view, even in simple markets that exhibit the characteristics outlined in Section II, a coordination problem exists in markets with more than two rivals. This is so since the number of collusive equilibria present in a repeated game defies the simple alignment of prices.43 Accordingly, to increase the likelihood of sustained tacit collusion, one would require some form of communication either to kickstart stable tacit collusion, or to sustain it.44 The issue is principal and goes beyond the discussion of algorithmic collusion. This body of scholarship suggests that many times, tacit coordination is unlikely absent some form of illicit communication or centralized orchestration, even in markets with three rivals.45 These findings are often based on empirical observations under laboratory conditions, with perfect control and transparency over communications. Permitting the human subjects to communicate, even briefly, increased their ability to enter into and sustain both coordination and higher prices with higher numbers of participants. Absent communications, in these experiments, collusion was difficult, if not impossible, to reach and sustain. With the above in mind, these critics have argued that the debate over algorithmic tacit collusion is unwarranted. According to them, the unavoidable need for communication among firms would bring the parallel behaviour into the realm of antitrust enforcement and enable agencies to condemn it as an anticompetitive agreement or concerted practice under well-established case law. Thus, if algorithms do not (or cannot) ‘communicate’ with one another, then tacit algorithmic collusion is unlikely. Indeed, the degree of coordination required to align the algorithms would increase the risk of exposure and civil (and potentially criminal) liability. So, when we observe what appears to be tacit collusion in these markets, it is likely the result of illegal human communications. So why have these criticisms failed to persuade enforcers and courts with respect to tacit collusion in the brick-and-mortar economy, and why are they unlikely to gain traction in the digital economy?

41 For instance, where the environment is dynamic, demand is uncertain and competition is not limited to price. See E. Green, R. Marshall and L. Marx, ‘Tacit Collusion in Oligopoly’, in R. Blair and D.D. Sokol (eds), The Oxford Handbook of International Antitrust Economics: Volume 2 (Oxford University Press, 2014). 42 D. Byrne and N. de Roos, ‘Learning to Coordinate: A Study in Retail Gasoline’ (23 July 2018) https://​ssrn​.com/​abstract​=​2570637 [accessed 3 March 2022]. 43 On the role of communications, see J. Harrington, R. Hernan Gonzalez and P. Kujalc, ‘The Relative Efficacy of Price Announcements and Express Communication for Collusion: Experimental Findings’ (2016) 128 Journal of Economic Behavior & Organization 251; M. Fonseca and H.-T. Normann, ‘Explicit vs. Tacit Collusion – The Impact of Communication in Oligopoly Experiments’ (2012) 56(8) European Economic Review 1759; D. Cooper and K.-U. Kuhn, ‘Communication, Renegotiation, and the Scope for Collusion’ [2009] American Economic Journal: Microeconomics 6; J. Farrell and M. Rabin, ‘Cheap Talk’ (1996) 10(3) Journal of Economic Perspectives 103; and V. Crawford and J. Sobel, ‘Strategic Information Transmission’ (1982) 50(6) Econometrica 1431. 44 Independent of the discussion here, it has been shown that ‘after a period of collusion supported by regular communication, firms are able to maintain collusive prices even when communication is no longer possible’: Fonseca and Normann (n 43). 45 See, for example, Schwalbe (n 11).

196  Research handbook on cartels When competition agencies or courts observe conscious parallelism that yields supra-competitive pricing, they do not assume that the competitors must be communicating with each other to jump start or sustain the tacit collusion. The law in both the US and EU recognizes that, under certain market conditions, companies can behave as rational agents and adjust to market characteristics without any communications. The classic example is one gas station in a remote town silently reacting to the pricing of its competitors across the street. Such a phenomenon, while dampening price competition, is legal, and will not trigger intervention. As the US Supreme Court held: Tacit collusion, sometimes called oligopolistic price coordination or conscious parallelism, describes the process, not in itself unlawful, by which firms in a concentrated market might in effect share monopoly power, setting their prices at a profit-maximizing, supracompetitive level by recognizing their shared economic interests and their interdependence with respect to price and output decisions.46

Both EU and US antitrust law recognize that anticompetitive ‘behavior can sometimes be coordinated without any communication or other observable and reprehensible behavior’.47 That is why ‘[t]acit coordination is feared by antitrust policy even more than express collusion, for tacit coordination, even when observed, cannot easily be controlled directly by the antitrust laws’.48 In recognizing this possibility, antitrust plaintiffs in the EU and US can only attack this tacit collusion indirectly. One way is for the Federal Trade Commission (‘FTC’) to attack practices that facilitate tacit collusion under its broader powers under Section 5 of the FTC Act. Another way is to target mergers that foster tacit collusion, precisely because it can be accomplished without any communications or agreement among rivals.49 We thus observe a gap between the law and the criticism according to which communication is needed to enter into, or stabilize, conscious parallelism, and industry awareness will not suffice to support a common strategy. The law posits that anticompetitive parallel behaviour among few firms can naturally occur in markets with the conditions outlined in Section II. Indeed, it can occur with sufficient frequency in these markets that the law will not presume any underlying communications. (For if the courts believed that communications often accompanied conscious parallelism, a legal presumption would likely arise.50) This notion affects both merger review and other antitrust enforcement. In the case of ex ante merger review, the realization that tacit collusion may emerge when the market conditions in Section II are present will justify careful scrutiny of proposed transactions that would foster conscious parallelism. In the case of ex post antitrust enforcement, the realization that tacit collusion may emerge when market conditions are present may provide an explanation to the parallel conduct and bring it outside the scope of Section 1 of the Sherman Act and Article 101 TFEU. Accordingly, even when private plaintiffs, the DOJ or European agencies have ample evidence of anticompetitive parallel behaviour, that in itself will not serve as proof of an agreement or illicit concerted practice, when the market conditions for tacit collusion are

46 Brooke Group v Brown & Williamson Tobacco Corp., 509 US 209 (1993) 224, 227. See also FTC v H.J. Heinz Co., 246 F.3d 708 (DC Circuit, 2001) 725. 47 City of Columbia v Omni Outdoor Advert., Inc., 499 US 365 (1991) 396 n 10. 48 Heinz, 246 F.3d 708 (2001) 725. 49 ibid. 50 Eastman Kodak Co. v Image Tech. Servs., Inc., 504 US 451 (1992), 466–67.

Algorithmic tacit collusion  197 present.51 Courts instead will assume that tacit collusion is likely and will require additional proof which include evidence of illicit communication. It is only when parallel behaviour cannot be explained as the outcome of tacit collusion (or due to other factors) that it may serve as proof of illegal collusion. As the European Court of Justice held: Although parallel behaviour may not by itself be identified with a concerted practice, it may however amount to strong evidence of such a practice if it leads to conditions of competition which do not correspond to the normal conditions of the market, having regard to the nature of the products, the size and number of the undertakings, and the volume of the said market.52

It is for the competition agency and private plaintiff to establish that no other explanation for the parallel behaviour is present. In doing so, it should also consider whether the market may display the conditions for tacit collusion which could explain the parallel behaviour. And so, the case law puts the onus on the antitrust plaintiff to prove the implausibility of rational unilateral reaction to market characteristics. One example is CISAC v Commission, where the EU General Court quashed a finding by the European Commission that parallel behaviour between collecting societies was the result of illegal collusion with the aim of dividing the market.53 The Court held that the Commission did not establish, to the requisite legal standard, the existence of collusion between the collecting societies to fix the national territorial limitations. The evidence relied upon by the Commission was not sufficient to render implausible the explanation that the national territorial limitations were the result of individual, carefully considered and rational decisions on a practical and economic level, given the specific conditions of the market, and not the result of a concerted practice. The Court held that ‘the Commission must show precise and consistent evidence in order to establish the existence of the infringement’.54 Indeed, it is settled case law that where the Commission’s reasoning is based on the supposition that the facts established in its decision cannot be explained other than by concentration between the undertakings, it is sufficient for the applicants to prove circumstances which cast the facts established by the Commission in a different light and thus allow another explanation of the facts to be substituted for the one adopted by the Commission.55

Thus, the case law accepts that, absent proof of express collusion or communication, parallel action and tacit collusion may be the only explanation of the market outcome. Consumers may be harmed due to higher prices, yet the law cannot condemn the parallelism as illegal. In other words, the courts and agencies accept that tacit collusion is not only legal, but likely and sustainable in concentrated industries. Absent proof of an agreement, the plaintiff cannot challenge the anticompetitive conduct.

51 See, e.g., Bell Atl. Corp. v Twombly, 550 US 544 (2007) 554; and Harlem River Consumers Co-op., Inc. v Associated Grocers of Harlem, Inc., 408 F. Supp. 1251 (SDNY, 1976) 1278. 52 Case 48/69 Imperial Chemical Industries (ICI) v Commission (Dyestuffs) [1972] ECR 619, para 66. 53 Case T‑442/08 CISAC v Commission ECLI:​EU:​T:​2013:​188. 54 ibid para 96. 55 ibid para 99, citing Joined Cases T‑305/94 etc Limburgse Vinyl Maatschappij and others v Commission [1999] ECR II‑931.

198  Research handbook on cartels One interesting example is the United States Court of Appeals for the Seventh Circuit, which in 2015 explored price alignment without any actual communications among the parties. The opinion is noteworthy as its author Judge Richard Posner, in his earlier writings, thought that it was ‘improbable that prices could long be maintained above cost in a market, even a highly oligopolistic one, without some explicit acts of communication and implementation’.56 Nonetheless, writing for the Seventh Circuit, Judge Posner accepted the notion that anticompetitive tacit collusion can occur without any such communication: As for the apparent anomaly of competitors’ raising prices in the face of falling costs, that is indeed evidence that they are not competing in the sense of trying to take sales from each other. However, this may be not because they’ve agreed not to compete but because all of them have determined independently that they may be better off with a higher price. That higher price, moreover – the consequence of parallel but independent decisions to raise prices – may generate even greater profits (compared to competitive pricing) if costs are falling, provided that consumers do not have attractive alternatives.57

In this case, the action taken by the companies was deemed unilateral and reflected an economic rationale, in light of each firm’s demand function. The Seventh Circuit recognized that anti-competitive pricing could arise from purely tacit collusion: ‘There isn’t even evidence that [an employee of the defendant] had ever communicated on any subject with any employee of any of the other defendants’.58 As the court noted: [T]he Sherman Act imposes no duty on firms to compete vigorously, or for that matter at all, in price. This troubles some antitrust experts, such as Harvard Law School Professor Louis Kaplow, whose book Competition Policy and Price Fixing (2013) argues that tacit collusion should be deemed a violation of the Sherman Act. That of course is not the law, and probably shouldn't be. A seller must decide on a price; and if tacit collusion is forbidden, how does a seller in a market in which conditions (such as few sellers, many buyers, and a homogeneous product, which may preclude nonprice competition) favor convergence by the sellers on a joint profit-maximizing price without their actually agreeing to charge that price, decide what price to charge?59

The courts thus assume that ‘competitors in concentrated markets watch each other like hawks’.60 Each competitor will copy or respond to competitive responses without necessarily communicating with one another. And ‘it is not a violation of antitrust law for a firm to raise its price, counting on its competitors to do likewise (but without any communication with them on the subject) and fearing the consequences if they do not’.61 So how does one reconcile the views of the courts and enforcers on the one hand and the discrete subset of economists on the other hand? One explanation is that the case law is simply wrong. Tacit collusion is unlikely and communications are occurring, but the colluders are effectively covering their tracks. We are presented with a case of a Type II error (false negative) where courts are reaching a negative result (dismissing cases) when they should be reach R. Posner, ‘Oligopoly and the Antitrust Laws: A Suggested Approach’ (1969) 21 Stan. L. Rev. 1562, 1574. 57 In re Text Messaging Antitrust Litig., 782 F.3d 867 (7th Circuit, 2015) 871–72. 58 ibid 872. 59 ibid 873–74. 60 ibid 875. 61 ibid 876. 56

Algorithmic tacit collusion  199 ing a positive one (finding liability). While economists doubt the ability to enter and sustain conscious parallelism, the law assumes that it is possible without illicit communications and does not intervene. The problem is that if one were to reject the prevailing legal viewpoint, we may quickly shift to a Type I error (false positive), where courts reach a positive result (finding the defendants liable for price-fixing) when they should reach a negative one (finding the defendants not liable because they never agreed with one another). All the plaintiff would have to show in markets with more than two competitors is an anticompetitive outcome – whether by tacit or express collusion. If anticompetitive conscious parallelism/tacit collusion is considered implausible without communication, the court would infer communications among the competitors. Once the court makes this inference, it is a small step to infer from the unobserved communications – along with the observed anticompetitive behaviour – an agreement among the rivals, and thus liability under EU and US law. Under such economic theory, the distinction between express and tacit collusion would fade as the agencies, antitrust plaintiffs and courts would assume an illegal agreement whenever observing conscious parallelism with anticompetitive outcomes.62 This, of course, would send shivers through the defence bar and their clients. Courts would presume that firms communicated, even when they haven’t. And how would they prove that they did not communicate? To avoid prosecution, firms will be required to operate irrationally in the market. A second explanation focuses on the misalignment between market realities and the experimental evidence upon which some economists rely. According to this explanation, economic experiments, carried out in laboratories with test subjects that interact over a period of a few hours (and with absolute control over communications), do not necessarily provide a good proxy for actual market behaviour, where awareness of interdependence exists absent illicit communications. The lab experiments do not reflect the interdependence of tacit collusion (and often discount the stability of actual collusion). In practice, firms can sustain tacit collusion without illicit communication as they operate with awareness that develops over time as to the market dynamics, and the benefit they may attain from parallelism and the avoidance of price wars. Firms that operate over long periods of time in these highly concentrated markets benefit from ‘industry awareness’ and understand the interdependence among their actions.63 That awareness emerges from a large number of abstract signals and observations, none of which triggers antitrust intervention, and can reduce uncertainty about future actions with long-lasting effects on coordination. This awareness may substitute for communication in laboratory settings and, at the very least, provide a plausible explanation to the durable conscious parallelism. 62 One potential rebuttal is that the antitrust plaintiffs should still have to hunt evidence of the communications. But why? When there is fire, why would the court require plaintiffs to prove independently the existence of smoke. If what you are observing – namely the anticompetitive coordination – is only possible with communications, then why would the courts require the plaintiff to expend time and resources to prove the communication? In the end, tacit collusion would always violate Section 1 of the Sherman Act and Article 101 TFEU. 63 See, for example, R. Dhalla and C. Oliver, ‘Industry Identity in an Oligopolistic Market and Firms’ Responses to Institutional Pressures’ (2013) 34(12) Organization Studies 1803; and M. Peteraf and M. Shanley, ‘Getting to Know You: A Theory of Strategic Group Identity’ (1997) 18(S1) Strategic Management Journal 165.

200  Research handbook on cartels Whichever explanation one favours, either way – when determining illegality – the law rejects the argument that communication is essential to establish tacit collusion. Quite the contrary, the law accepts that when market conditions for tacit collusion are present, conscious parallelism yielding anticompetitive outcomes may be sustained. Put simply, tacit coordination can exist ‘without any actual communication among competitors’.64 Returning to our discussion of algorithms, the same legal approach applies. So, when we raised our third and fourth scenarios of algorithmic tacit collusion, most enforcers, judges and lawyers recognized this possibility. It derived naturally from the law (and the market reality that they encountered over the decades). Moreover, other economists and game theorists accept tacit collusion without communications. But, if one assumes that the sceptics are right, then the gap between their beliefs and the law has widened. If the sceptics are right, humans have somehow successfully skirted antitrust liability for decades by disguising their communications. (One wonders why more cartels don’t adopt this stealth communication to avoid prosecution.) But because pricing algorithms cannot engage in this ‘stealth communication’, tacit algorithmic collusion should be impossible. If true, then whenever enforcers observe what appears to be conscious parallelism in markets dominated by pricing algorithms, they have a stronger case to argue that the humans must have communicated. For any other explanation is impossible. But the enforcers and courts, to date, have not adopted this presumption. They recognize the possibility of humans communicating (our first scenario), but also recognize humans and algorithms tacitly colluding (our third scenario) without explicit communications. Another anomaly emerges. If the critics are correct, in industries conducive to tacit collusion, firms would have little, if any, incentive to use pricing algorithms. These firms apparently have a golden ticket – they can charge supra-competitive prices through stealth human communications without the threat of antitrust liability. So, one would not expect industries characterized by such tacit collusion – like gas stations – to switch to pricing algorithms. For if they did, their prices and profits, without the stealth human communications, would likely drop. If the prices don’t drop, then one must assume, under this economic theory, that the firms, as in the Topkins case in the US65 and the Trod and GBE case in the UK,66 not only agreed to collude but also communicated with each other on the algorithms needed to implement and sustain their collusion. The level of communications should thus significantly increase as the firms switch to pricing algorithms. This assumption, however, requires enforcers and plaintiffs to hunt for communications that might not exist (if their theory is wrong).

64 Comment of the Federal Trade Commission, Proposed Confidentiality Determinations for Data Required Under the Mandatory Greenhouse Gas Reporting Rule and Proposed Amendment to Special Rules Governing Certain Information Obtained Under the Clean Air Act (30 September 2010, 2010 WL 9440202) 6; and In re High Fructose Corn Syrup Antitrust Litigation, 295 F.3d. 651 (7th Circuit, 2002) 654. 65 Plea agreement filed in United States v David Topkins (30 April 2015) www​.justice​.gov/​atr/​case​ -document/​file/​628891/​download [accessed 3 March 2022]. 66 The UK antitrust authority found in 2016 that Trod Limited and GB eye Limited infringed the competition law by agreeing that they would not, in certain specified circumstances, undercut each other’s prices for posters and frames sold on Amazon’s UK website and used pricing algorithms to facilitate their illegal agreement: www​.gov​.uk/​government/​news/​cma​-issues​-final​-decision​-in​-online​-cartel​ -case [accessed 3 March 2022].

Algorithmic tacit collusion  201 Thus, courts and competition authorities have largely marginalized the ‘tacit collusion is impossible without communications’ arguments. Indeed, as we discuss below, the emerging evidence justifies the courts’ and agencies’ scepticism of the sceptics.

IV.

THE (IM)PLAUSIBILITY OF ALGORITHMIC COLLUSION

Let us now move to a second, related issue which merits our attention – whether pricing algorithms can support anticompetitive conscious parallelism. If we accept the legal premise that conscious parallelism can occur without the communications that expose firms to antitrust liability, then the issue is whether algorithms can facilitate tacit collusion, and do so in a superior manner to that of humans. Using the criteria that the courts and agencies have applied for decades to explain why mergers may facilitate tacit collusion (and thus should be enjoined),67 the factors outlined in Section II, including the transparency of online markets and speed in responding to prevent the discounting firm from benefitting from the price cut, explain how the rational use of algorithms can increase instances in which tacit collusion is sustained. That use may provide a valid explanation for price alignment, even if that goes beyond the conditions of tacit collusion (for instance, additional sellers on the market). After all, under the current legal regime, a rational and permissible use of algorithms to unilaterally align prices may indeed establish parallelism in instances where humans may fail to do so. Importantly, this is not a revolution, but rather an evolution. It will not happen effortlessly, nor on all markets. Some contend that even if tacit collusion without communication is possible in the brick-and-mortar economy, that does not mean it is possible in industries where prices are set by algorithms (and perhaps for some firms by humans). The potentially large number of collusive equilibria presented by algorithms will likely decrease the likelihood of alignment in a repeated game – that is, algorithms are unlikely to obtain and sustain tacit collusion. In what follows, we consider this argument. Let us start by stating the obvious. The discussion does not concern ‘the rise of the machines’ or the creation of ‘evil’ algorithms that seek to profit at the expense of consumers. It is a somewhat less exciting debate about the possibility of algorithms, designed by humans, to offer a superior instrument for the optimization of pricing decisions, in markets that may support conscious parallelism. In that respect, one should note the limitations of the pricing algorithm. It will not necessarily change the basic characteristics of every market, nor will it overcome instability that results from lower barriers to entry, maverick companies or fierce competition. The tool at hand, at times, will amplify the power to monitor and punish in instances when humans see a benefit in sustaining parallel behaviour. When discussing the extension of the human will, it is helpful to distinguish between ‘simple’ adaptive algorithms that are programmed to monitor and ‘react’, and more sophisticated self-learning algorithms that rely on artificial intelligence to autonomously determine the optimal strategy. That simplified distinction is of value for our discussion, as it helps identify instances in which the human played a role in appreciating the benefit in parallelism (thus we have evidence of anticompetitive intent, but not necessarily communications), and instances in 67 See, e.g., US Department of Justice and Federal Trade Commission, Horizontal Merger Guidelines (19 August 2010) §7; and EU Merger Guidelines (n 21) 5–18.

202  Research handbook on cartels which human involvement does not include an attempt to stabilize parallelism on the market. Let us explore both categories. A.

Simple Algorithms

Humans can program adaptive algorithms to reflect a pricing strategy which assumes interdependence on the market or is geared to push toward such interdependence. Humans identify the desirability of parallelism, and the particular market being ripe for conscious parallelism. Humans program the algorithm to reflect the unilateral actions of a rational agent in this tight oligopoly. Detection and punishment of deviation are imbedded into the pricing decision-making and so is the upward price adjustment that follows the price leader. In essence, tacit collusion happens at the human level and leads humans to utilize technology in order to stabilize it. As we saw in Section III, the law in the US and EU accepts that when market conditions are apt, such conscious parallelism can be established unilaterally, as humans develop an awareness to market dynamics and appreciate the interdependence among the rivals. As a result, the enforcers and private plaintiffs cannot legally challenge the new equilibrium (absent evidence of express collusion). As the algorithms are used unilaterally and as an extension of the human will, they will not be viewed, under the current antitrust policy, as facilitating practices that taint the behaviour as illicit collusion. That much is granted; but how likely is tacit algorithmic collusion without communications? To test the dynamic described above, let us start in the lab. Professors Nan Zhou, Li Zhang, Shijian Li and Zhijian Wang devised a Linear Extortion to Collusion Algorithm (LECA) which can ‘enforce its human rival to collude’.68 They then designed an algorithm– human game, where a human competed against the LECA algorithm for 600 rounds. In each round, the human and algorithm could decide the quantity of a product to produce. Importantly for our purposes, they could not otherwise communicate with each other. Nor did the human know of the algorithm’s pricing strategy. After they each selected a quantity of products they wanted to produce, they were told the human subject’s and algorithm’s profits. Over the first 300 iterations of the Cournot competition game,69 the humans learned that reducing quantity to reach the almost fully collusive level would secure the greatest profits. After learning this, the humans kept their quantity at the collusive level thereafter. In their algorithm–human duopoly market, the degree of tacit collusion rose to nearly 100 per cent in rounds 300 to 400. What is interesting is that the time to establish tacit collusion (about 400 rounds) in the algorithm–human experiment was quicker than the human–human collusion (about 800 rounds) experiments. From their experiments, the study’s authors concluded that, first, algorithms can facilitate tacit collusion more quickly and, second, there exist incentives for firms to use such algorithms in the market. That experiment, as the authors recognized, involved a duopoly. Now let us consider tacit collusion in markets with multiple competitors.

N. Zhou, L. Zhang, S. Li and Z. Wang, ‘Algorithmic Collusion in Cournot Duopoly Market: Evidence from Experimental Economics’ (21 February 2018) https://​ arxiv​ .org/​ pdf/​ 1802​ .08061​ .pdf [accessed 3 March 2022]. 69 On this concept, see W. Kenton, ‘Cournot Competition’ (Investopedia, 12 January 2018) www​ .investopedia​.com/​terms/​c/​cournot​-competition​.asp. 68

Algorithmic tacit collusion  203 Suppose an oligopolistic gas station market with limited transparency, that is, prices are only visible when reaching each gas station. In this market, customers can mitigate the search costs by asking friends about any available deals, visit a few gas stations and then support the one with the lowest price. Here, a gas station, by discounting, may increase its profits and develop a reputation for having a low (if not the lowest) price. At times, competitors, aware of the price reductions and promotions, would respond with their own initiative. While the gas prices are transparent, there is a lag for rivals to discover the lower price. (The rivals, after all, have to drive around town to monitor gas prices.) Their delayed response is likely to benefit the station with the reputation as a discounter. Under these market conditions, conscious parallelism is harder to sustain. The firms will likely compete as expected. We see here how markets ‘need to be sufficiently transparent to allow the coordinating firms to monitor to a sufficient degree whether other firms are deviating, and thus know when to retaliate’.70 This would be especially the case where customers are aware of the price, while competitors are not (for example, when there are significant and frequent discounts). When transparency and the rivals’ speed in responding to competitive behaviour increase in concentrated markets with homogeneous goods, so too does the risk of tacit collusion. With computerized pricing, the process may be faster and more stable. To foster parallelism, companies may adopt a pricing strategy which would be easy to decipher by competitors. Let us briefly illustrate with two examples. In 2012, petrol stations in Chile were required to post their fuel prices on a government website and to keep prices updated as they changed at the pump. An economic study found that this Chilean regulation softened rather than increased competition.71 The petrol stations’ margins increased by 10 per cent on average following the prices being posted on the government website.72 In Germany, the government suspected that an oligopoly of five firms – BP (Aral), ConocoPhillips (Jet), ExxonMobil (Esso), Shell and Total – dominated the off-motorway petrol station business.73 To promote competition, the government required the petrol stations to report to its government’s transparency unit any price changes for gasoline or diesel fuel in ‘real time’.74 The government’s transparency unit then transmitted the price data to consumers, with the aim that they could easily find the cheapest petrol nearby. Rather than lowering prices, the enhanced market transparency, one economic study found, actually increased prices

EU Merger Guidelines (n 21) para 49. F. Luco, ‘Who Benefits from Information Disclosure? The Case of Retail Gasoline’ (2019) 11(2) American Economic Journal: Microeconomics 277. 72 The softening of competition was common across brands and was not limited to a single Chilean city. Interestingly, although the stations’ margins increased across Chile, the effect was not uniform, varying from high-income to low-income areas; see ibid 278. 73 Bundeskartellamt, Fuel Sector Inquiry: Final Report (May 2011) www​.bundeskartellamt​.de/​ SharedDocs/​Publikation/​EN/​Sector​%20Inquiries/​Fuel​%20Sector​%20Inquiry​%20​-​%20Final​%20Report​ .pdf​?​_​_blob​=​publicationFile​&​v​=​14 [accessed 3 March 2022]. 74 R. Dewenter, U. Heimeshoff and H. Lüth, ‘The Impact of the Market Transparency Unit for Fuels on Gasoline Prices in Germany’ (May 2016) www​.dice​.hhu​.de/​fileadmin/​redaktion/​Fakultaeten/​ Wirtscha​ftswissens​chaftliche​_Fakultaet/​DICE/​Discussion​_Paper/​220​_Dewenter​_Heimeshoff​_Lueth​ .pdf [accessed 3 March 2022]. 70 71

204  Research handbook on cartels further. Compared to the control group, retail petrol prices increased by about 1.2 to 3.3 euro cents, and diesel increased by about 2 euro cents.75 Other studies also suggest that an increase in transparency can facilitate tacit collusion.76 First, these outcomes, which make sense under the legal standard, are harder to explain under the ‘no collusion absent communications’ theory. Under that economic theory, the government’s increase in transparency should not have prompted the rivals to increase prices further. Because, under the economic theory, sustaining tacit collusion among five competitors is implausible, the oligopolists, under the economic theory, must have been actively communicating to sustain their supra-competitive pricing. They conceivably would have communicated their dissatisfaction to each other after their daily drive. Rather, the result is consistent with the legal acknowledgement of sustained tacit collusion where each competitor watches the others like a hawk. To monitor pricing, the petrol station owners in Germany would drive past specified competitor petrol stations several times a day and note their prices. The monitored prices were then fed into the respective company’s electronic system. Generally, when one competitor increased petrol price, rivals would respond between three to six hours later.77 Now, with increased transparency from online pricing, the rivals can monitor and punish instantaneously. So, the increase in fuel prices was not the likely result of ‘communications’. Instead it likely reflects tacit collusion, where firms, aware of their interdependence, recognize that they will profit by acceding to the higher price rather than discounting. With pricing algorithms, the retaliation time is further reduced. As each firm taps into its rivals’ real-time pricing, no gas station likely profits by discounting. Given the velocity with which the pricing algorithms can adjust, each gas station is less likely to develop a reputation as a price discounter among its customers. Accordingly, the competitors will have less incentive to discount. On the flip side, the algorithms’ velocity of pricing decisions can shorten the time period for signalling price increases in other industries. Firms would no longer have to rely on lengthy (for example, 30-day) advanced price announcements, where they must wait and see what the competitive response is, to decide whether to raise prices (and to what extent). Computers can have multiple rounds whereby one firm increases prices and the rival computers respond immediately and without the risk that the firm that initiates the price increase will lose many customers to rivals. Essentially, companies may now need only seconds, rather than days, to signal price increases to foster tacit collusion. As we shift from a world where rivals drive around town to see the price that their rivals charge to one where pricing algorithms can achieve this within milliseconds, the human logic to maximize profits remains. Importantly, the algorithms help effectuate this logic. Needless to say, algorithms will not immunize market participants from disruptive technologies, entrants ibid. More generally, we also note another interesting study on the impact of price matching guarantee as stabilizing tacit collusive mechanism in petrol markets: L. Cabral, N. Dürr, D. Schober and O. Woll, ‘Learning Collusion: Theory and Evidence from a Gasoline Market Price Matching Guarantee’ (March 2018). 76 See, e.g., Byrne and de Roos (n 42); Project Update DNRME 18018-Variation 1: The Impact of MyFueINT on Retail ULP Prices in the Northern Territory (Griffith University, May 2018) www​ .parliament​.qld​.gov​.au/​Documents/​TableOffice/​TabledPapers/​2018/​5618T565​.pdf [accessed 3 March 2022]. 77 Dewenter et al (n 74). 75

Algorithmic tacit collusion  205 or mavericks. But absent such threats, the market participants can use pricing algorithms to sustain tacit collusion with higher profits (and do so without entering into any illicit communication or concerted practice). Humans may use additional means to further stabilize the conscious parallelism. They could, for example, limit variations in the design of the algorithms, making it easier to follow. Such unilateral moves are unlikely to trigger antitrust liability. Companies may invest in better tools to observe and imitate pricing decisions executed by other algorithms. Companies may, for example, introduce price matching guarantees to further support monitoring as deterrent mechanisms.78 The unilateral nature of the actions may well leave them outside the realm of Article 101 of TFEU, Section 1 of the Sherman Act and even Section 5 of the FTC Act.79 Going a step further, humans may use algorithms in a more aggressive way to decode the strategy used by competing algorithms and adjust accordingly.80 Depending on the technology used, this might trigger intervention – although, if used unilaterally, it remains open to argument that the decoding of information is not to be viewed as a communication or facilitating practice (and thus potentially illegal) but rather as unilateral (and legal) behaviour under the current law.81 To avoid the need to invest in decoding the competing algorithms, companies may adopt a different approach and use the same provider for their pricing algorithm, or the same provider for their dynamic pricing strategies. Returning to our example, several gas stations operating on a given market could use the same company for pricing decision-making. When multiple players use the same algorithm, data points and values, the likelihood for alignment increases. One example is the market for petrol in Rotterdam, the Netherlands, where a number of petrol stations, according to the Wall Street Journal, used the same provider – the Danish company a2i Systems – for advanced analytics to determine petrol prices.82 Importantly, one should note that the provision by the same company of dynamic pricing services, and the creation of a possible hub-and-spoke relationship, does not clearly infringe the competition laws. On its website, the company a2i Systems provides a case study to illustrate how it helped OK Benzin, Denmark’s leading petrol station owner, avoid a price war: ‘Between 2007 and 2012 the market was characterized by fierce competition and high volatility. At the peak there were 10 to 20 price changes a day, and the spread between the highest and the lowest price of the day could be up to 15 eurocent.’83 In enlisting a2i Systems, the leading retail network of approximately 700 petrol stations (which accounted for 25 per cent of the Danish retail fuel market) sought ‘to improve the pricing anal-

78 Price match may create an incentive to follow price increases by the price leader. See Cabral et al (n 75). 79 See, e.g., Ethyl Corp. v Fed. Trade Comm’n, 729 F.2d 128 (2nd Circuit, 1984) 139–40. 80 See M. Gal, ‘Algorithms as Illegal Agreements’ (2019) 34 Berkeley Technology Law Journal 67. See also B. Salced, ‘Pricing Algorithms and Tacit Collusion’ (Pennsylvania State University, 2015) http://​brunosalcedo​.com/​docs/​collusion​.pdf [accessed 3 March 2022]. 81 ‘Algorithms and Collusion – Note from the European Union’ (n 7) 8. 82 S. Schechner, ‘Why Do Gas Station Prices Constantly Change? Blame the Algorithm’ (Wall Street Journal, 8 May 2017) www​.wsj​.com/​articles/​why​-do​-gas​-station​-prices​-constantly​-change​-blame​-the​ -algorithm​-1494262674 [accessed 3 March 2020]. 83 ‘PriceCast Fuel Case Story’ http://​a2isystems​.com/​files/​pdf/​PriceCast​%20Fuel %20Case%20 Story%20('15).pdf [accessed 3 March 2022].

206  Research handbook on cartels ysis and decision process and optimize pricing according to their overall strategy in order to lower the cost of price wars or better yet, to avoid them’.84 As the Wall Street Journal reported, the complex algorithm operated by a2i Systems was tested against a control group which did not use the system to determine price. The result? ‘The group using the software averaged 5% higher margins.’85 For the petrol company, a2i Systems notes, this ‘means millions of Euros’ more annually.86 In essence, the a2i pricing algorithm lowered the cost and likelihood of price wars. Of course, this is not a case of a2i marketing its ability to coordinate a price-fixing cartel. That would subject it and the petrol stations in Europe to civil liability. Rather, it is about the unilateral use of a decision-making algorithm to soften competition. It is about using the a2i pricing algorithm to service multiple clients. The sharing of the same focal point, in our opinion, should raise concerns in such instances and call for some form of intervention. The hub-and-spoke algorithmic structure brings us further away from typical tacit collusion, but is yet to be challenged by competition agencies. It is important to stress that it differs from a cartel being facilitated by a hub-and-spoke structure.87 (The head of the DOJ in 2018 intimated a potential criminal case that may inform the legality of this practice.88) Indeed, it is an ‘incidental’ hub-and-spoke, which, while not driven by a cartel agreement, may nonetheless facilitate alignment. The UK Competition and Markets Authority (‘CMA’) expressed the greatest concern over this algorithmic hub-and-spoke structure ‘because it simply requires firms to adopt the same algorithmic pricing model’.89 As we indicated in Virtual Competition, such incidental hub-and-spoke, while not indicative of a cartel agreement, could indeed undermine competition. Let us move beyond hub-and-spoke, and note how algorithms may be used to amplify the effects of anticompetitive agreements. One example involves resale price maintenance (‘RPM’), which is where the manufacturer/distributor agrees with the retailer on what the minimum price should be for the manufacturer’s product. Historically the manufacturer would monitor and individually punish retailers that sold the manufacturer’s product below its suggested retail price. For example, after punishing retailer A, the manufacturer would voice its displeasure to retailers B, C and D. Punishing each offending retailer increases the manufacturer’s potential risks of antitrust liability, especially in jurisdictions where RPM is per se (or presumptively) illegal.90 But pricing algorithms can change that dynamic, to the detriment of consumers. The European Commission found in its e-commerce sector inquiry ‘increased use of automatic

ibid. ibid. 86 ibid. 87 See Interstate Circuit, Inc. v United States, 306 US 208 (1939). On hub-and-spoke cartels, see Chapter 10 in this volume. 88 See www​.broadcastingcable​.com/​news/​delrahim​-criminal​-case​-against​-anti​-competitive​-search​ -algorithms​-coming [accessed 3 March 2022]. 89 Competition and Markets Authority (n 20) para 5.35. 90 RPM is presumptively illegal in Europe and in some states in the US. RPM was per se illegal for nearly a century under the Sherman Act, until the Supreme Court in a controversial 5–4 decision, subjected it to a more deferential rule of reason standard: Leegin Creative Leather Products, Inc. v PSKS, Inc., 551 US 877 (2007). 84 85

Algorithmic tacit collusion  207 software applied by retailers for price monitoring and price setting’.91 Many, including the biggest online retailers, are using ‘pricing algorithms which automatically adapt retail prices to those of competitors’.92 In this environment, the manufacturer need not punish every offending retailer. Instead, the manufacturer would only have to punish one or two significant retailers that are discounting, and whose prices the other retailers’ pricing algorithms are tracking and matching. Once these discounters raise their prices, the other retailers’ pricing algorithms will automatically follow. The exposure to antitrust enforcement is reduced, due to the more limited communications. The Commission observed this anticompetitive dynamic in a 2018 vertical price-fixing case. As the Commission found, because many, including the biggest online retailers, were using pricing algorithms which automatically adapted the retail prices to those of competitors, the resale ‘pricing restrictions imposed on low pricing online retailers typically had a broader impact on overall online prices for the respective consumer electronics products’.93 In effect, the consumer electronics manufacturer only had to punish a few online discounters, and could be assured that many other retailers would automatically increase their prices. Thus, even in industries not susceptible to tacit collusion, one can obtain the same effect when manufacturers vertically fix prices with one significant retailer, and the other retailers’ pricing algorithms automatically follow suit. Consequently, the emerging evidence suggests that enforcers will likely uncover evidence of anti-competitive human intent in using relatively ‘simple’ algorithms to sustain tacit collusion, without any evidence of actual communications. After all, tech firms promote how their price optimization software can put an end to price wars before they even begin. B.

Artificial Intelligence

Now, let us turn to our fourth scenario, Digital Eye, where we raise the question of whether conscious parallelism could be established by self-learning algorithms without them reflecting the humans’ intention. Could algorithms that are based on reinforced learning provide a superior tool to sustain tacit collusion? And if so, when left to their own devices, might the pricing algorithms identify conscious parallelism as a superior strategy? We should start by stressing that the issue is not about algorithms conspiring against humans, but rather whether a self-learning price algorithm, programmed to optimize profit by interacting in a dynamic environment, may identify conscious parallelism as an optimal strategy. The question is whether in future markets, where the majority of dynamic pricing decisions will involve minimal human intervention, market equilibrium may be established above competitive levels – not as a result of express collusion, nor as a result of humans appreciating the benefits of tacit collusion (and programming their pricing algorithms accordingly), but rather as the result of rational action by independent learning algorithms which take account of various data points.

European Commission, ‘Antitrust: Commission Fines Four Consumer Electronics Manufacturers for Fixing Online Resale Prices’ (Press Release, IP/18/4601, 24 July 2018) http://​europa​.eu/​rapid/​press​ -release​_IP​-18​-4601​_en​.htm [accessed 3 March 2022]. 92 ibid. 93 ibid. 91

208  Research handbook on cartels No doubt much is still uncertain as to the capacity of future reinforced-learning or deep-learning algorithms to reach conscious parallelism with no human intervention. Doubts as to learning algorithms’ ability to sustain collusion refer to their increased sophistication, which would make alignment difficult. Doubts are also linked to the need and ability of algorithms to establish a hidden channel of communication which could address problems of entering and sustaining collusion.94 While acknowledging current uncertainty, competition agencies around the world are now looking into these developments. There is no need to be alarmed, but it is important to acknowledge that the tech industry is taking its first steps in this direction for its algorithms. From an enforcement perspective, and at a high level of simplification, one may envisage two outcomes. Outcome 1: If learning algorithms are incapable of autonomously reaching tacit collusion, humans in markets that tilt towards conscious parallelism would either train them to achieve that outcome or refrain from using them (as such use, absent any significant offsetting gains and efficiencies, would reduce profits). Accordingly, in a market where humans appreciate the benefits of interdependence, and can do so without infringing the competition laws, they would not introduce uncontrolled disruptors that could unleash a price war. They will continue using simple adaptive algorithms. Indeed, we have not found online any third party developer of pricing algorithms that promotes its algorithms’ ability to unleash and prevail in an all-out price war. If self-learning pricing algorithms reduced overall profits by destabilizing pre-existing tacit collusion, competitors would be unlikely to employ them. Thus, in industries already susceptible to tacit collusion, companies would ensure alignment of the learning algorithm with the overall strategy. They would be sure to exploit the freedom offered to them under the law and unilaterally use adaptive or simple algorithms. Under this scenario we return to our previous category of human-driven tacit collusion enhanced by algorithms. The question is whether such use should be condemned by competition law or remain unchallenged. Outcome 2: If, on the other hand, self-learning algorithms could solve the coordination problem, through trial and error, with no human intervention, then we face an additional complexity in the form of undetected and unchallenged conscious parallelism. In such a scenario algorithms will be able to learn through experimentation and shift, without the knowledge of the human executives, from competitive pricing rules to collusive pricing rules and sustain that new equilibrium. Several groups of economists and computer scientists are exploring this avenue. Research has already shown how, under certain conditions, reinforcement learning can sustain cooperation.95 Furthermore, learning algorithms were shown to gravitate towards conscious parallelism in simple oligopolistic settings.96 These observations support the possibility that self-learning algorithms may autonomously establish conscious parallelism, with no human For papers dismissing the possibility for algorithm-driven tacit collusion, see Schwalbe (n 11). See, for example, J. Crandall, M. Oudah, Tennom, F. Ishowo-Oloko, S. Abdallah, J.-F. Bonnefon, M. Cebrian, A. Shariff, M. Goodrich and I. Rahwan, ‘Cooperating with Machines’ (2018) 9 Nature Communications 233; and J. Leibo, V. Zambaldi, M. Lanctot, J. Marecki and T. Graepel, ‘Multi-Agent Reinforcement Learning in Sequential Social Dilemmas’ (2017) https://​arxiv​.org/​pdf/​1702​.03037​.pdf [accessed 3 March 2022]. 96 T. Klein, ‘Assessing Autonomous Algorithmic Collusion: Q-Learning under Sequential Pricing’ (Amsterdam Center for Law & Economics Working Paper No. 2018-05); and G. Tesauro and J. Kephart, 94 95

Algorithmic tacit collusion  209 input, in environments in which they operate in parallel (rather than only in a simplified environment in which they face a stable fixed-strategy opponent). With all the uncertainty and caveats in mind, let us briefly note observations by one group (Professors Calvano, Calzolari, Denicolò and Pastorello from the University of Bologna, European University Institute and Toulouse School of Economics) that have shown that self-learning algorithms can have the capacity to achieve coordination on the tacit collusive outcome.97 In experiments with two Q-learning pricing algorithms, tacit collusion emerged in more than 60 per cent of the cases, and at even higher levels following sufficient simulation. Importantly, these results were observed in a significantly rich environment with up to 100 price levels. As illustrated below (Figure 11.1), forcing a price deviation by one algorithm to the ‘Nash Price’ (the static equilibrium price which would emerge if there was no tacit coordination) led the other Q-learning algorithm to react. Subsequently both returned to the pre-existing price level which represents the tacit collusive equilibrium (which is above the competitive price, but below the monopolistic (cooperation) price).

Source: E. Calvano, G. Calzolari, V. Denicolò and S. Pastorello, ‘Q-Learning to Cooperate’ (presented at the NBER event on ‘Economics of Artificial Intelligence’, 13–14 September 2018).

Figure 11.1

‘Q-Learning to Cooperate’

In an extension of that experiment, the same group of scholars used three Q-learning algorithms (that is, more than what some argue is possible for tacit collusion without communica‘Pricing in Agent Economies Using Multi-Agent Q-Learning’ (2002) 5(3) Autonomous Agents and Multi-Agent Systems 289. 97 E. Calvano, G. Calzolari, V. Denicolò and S. Pastorello, ‘Algorithmic Pricing: What Implications for Competition Policy?’ (University of Bologna – CEPR, Toulouse School of Economics, 27 June 2018); and E. Calvano, G. Calzolari, V. Denicolò and S. Pastorello, ‘Artificial Intelligence, Algorithmic Pricing, and Collusion’ (2020) 110(10) American Economic Review 3267.

210  Research handbook on cartels tions) in a rich price environment. The group reported that it continued to observe conscious parallelism and increased profitability, with short learning times. In their experiments, the scholars observe how difficult it may be to detect such tacit algorithmic collusion: ‘What is most worrying is that the algorithms leave no trace of concerted action – they learn to collude purely by trial and error, with no prior knowledge of the environment in which they operate, without communicating with one another, and without being specifically designed or instructed to collude.’98 Researchers are continuing to experiment with the likelihood of tacit algorithmic collusion in even more complex environments – with increased and changing numbers of algorithms, increased sophistication of algorithms and increased price levels. Needless to say, these are still early days in the development of AI and its application to pricing decisions. Uncertainty still remains as to the operation of future markets, costs associated with the learning phase, the ability to simulate and operate in a multi-agent environment and the effects of increased algorithm complexity.99 Furthermore, developments in the ability of algorithms to signal,100 monitor, decode and communicate in stealth mode101 will affect any future equilibria. But we encourage researchers to continue to develop algorithmic tacit collusion incubators that model rich and realistic environments. What would be also interesting is if antitrust enforcers (or scholars) could use algorithms deployed in the field to see pricing levels (and margins) for particular products. It might be of interest if the agency could inquire into how algorithms responded when one competitor exited (or entered) the marketplace (perhaps informing future merger review).

V. CONCLUSIONS So why does this debate matter? Pricing algorithm suppliers tout as a benefit their clients avoiding price wars. If this is real, and not marketing hype, then there are significant potential profits from algorithms that can foster tacit collusion. This would represent an area ripe for further exploration by companies and developers of pricing algorithms, who at present benefit from an emerging gap in antitrust enforcement that may enable the attainment of higher profits (without the fear of antitrust liability, which includes, in the US, criminal fines, incarceration and trebled damages for the injured antitrust plaintiffs). That emerging gap merits closer consideration by competition agencies. But algorithmic tacit collusion can be even harder to detect – especially when the algorithms leave no trace of concerted action. As EU Commissioner Vestager noted, ‘[t]he trouble is, it’s not easy to know

98 E. Calvano, G. Calzolari, V. Denicolò and S. Pastorello, ‘Artificial Intelligence, Algorithmic Pricing, and Collusion’ (VOX, CEPR Policy Portal, 3 February 2019) https://​voxeu​.org/​article/​artificial​ -intelligence​-algorithmic​-pricing​-and​-collusion [accessed 3 March 2022]. 99 M. McGlohon and S. Sen, ‘Learning to Cooperate in Multi-Agent Systems by Combining Q-Learning and Evolutionary Strategy’ (2005) 1(2) International Journal on Lateral Computing 58. 100 Crandall et al (n 95). See also Gal (n 80). 101 M. Abadi and D. Andersen, ‘Learning to Protect Communications with Adversarial Neural Cryptography’ https://​arxiv​.org/​abs/​1610​.06918 [accessed 3 March 2022].

Algorithmic tacit collusion  211 exactly how those algorithms work. How they’ve decided what to show us, and what to hide. And yet the decisions they make affect us all.’102 If one accepts tacit coordination as a material risk in some susceptible industries, then the competition agencies must develop tools to assess (and deter) this risk.103 No doubt enforcement action at times will be challenging. After all, condemning rational reaction for market characteristics would, in itself, distort competition. Condemning it when it is assisted by bots may lead to a similar anomaly. Identifying, auditing or monitoring algorithms may be expensive and illusive. Using means to affect market transparency, undermine detection or delay reaction can undermine the essence of competition. These challenges should give us a pause. When considering any likely enforcement action, we must acknowledge the costs of over-intervention. Yet, the cost of under-intervention must also be acknowledged, especially when premised on the theory that tacit collusion is implausible without human communication. Efforts can focus on detection capacity. Market and sector investigations may be used to better understand industries and markets where algorithms may be used to align price.104 Another instrument, which we discuss elsewhere, may involve developing Algorithmic Collusion Incubators, which enable competition officials to test under what conditions tacit collusion occurs and the effects and likelihood of different counter-measures to destabilize this conscious parallelism.105 In addition, pre-emptive tools and regulations, such as the proposed Digital Market Act, may offer competition authorities more effective monitoring tools. In parallel to detection, there is room to consider updating current antitrust enforcement to capture tacit collusion which is facilitated by algorithms. A refinement of the approach to signalling may be a good place to start. Restrictions on certain market manipulations (through bots that underscore parallelism) may be another. The issue should be approached in a measured manner, as part of the everlasting adjustment of competition enforcement to market and technological reality. Failing to do so may well lead us to future markets where a competitive price is a mere illusion, and price optimization is used as code for tacit collusion’s supra-competitive profits. Increased awareness of the risks of algorithmic tacit collusion should also be addressed when assessing possible proposed merger transactions. As brick-and-mortar shops are closing at a faster rate, as sellers and buyers migrate to the online world and as technology, communications, big data and big analytics reach new highs, the effects of pricing algorithms will become more prominent. In the digitalized environment, tacit collusion might turn from being a mere outcome of market characteristics into a strategy. While the phenomenon of tacit collusion is limited to markets with given characteristics, it nonetheless is likely to exhibit greater durability in an algorithm-driven environment. So, with that risk in mind, we are encouraged that many policymakers and competition agencies are not only taking this risk seriously but are devoting resources to better understand ‘Algorithms and Competition’, Speech, Bundeskartellamt 18th Conference on Competition, Berlin, 16 March 2017. 103 ‘Algorithms and Collusion – Note from the European Union’ (n 7) 9; ‘Algorithms and Collusion – Note by the United States’ (n 7) 6; ‘Algorithms and Collusion – Note from the United Kingdom’ (DAF/ COMP/WD(2017)19, OECD, 30 May 2017) 11. 104 German Monopolies Commission (n 20). 105 A. Ezrachi and M. Stucke, ‘Algorithmic Collusion: Problems and Counter-Measures’ (DAF/ COMP/WD(2017)25, OECD, 31 May 2017) https://​one​.oecd​.org/​document/​DAF/​COMP/​WD(2017)25/​ en/​pdf [accessed 3 March 2022]. 102

212  Research handbook on cartels the implications of algorithmic collusion. While it might not be as glamorous as the dawn raid, their efforts might deter competitors from devising pricing algorithms that can better exploit us.

12. Smart contracts and cartel law Salil K. Mehra

I. INTRODUCTION Competition law still struggles with the definition of ‘agreement’. This is despite its long history grappling with the issue: some of the earliest prohibitions against price fixing had, by the nineteenth century, recognized the importance to the antitrust mission of agreement as a concept.1 In the 1990s, well before autonomous intelligent agents were possible, the question arose as to whether and how the law of commerce would be able to handle them.2 A core question is what to make of such agents’ negotiations – are they ‘agreements’ in the conventional contract law sense? This question has particular salience for competition law, since several jurisdictions have ‘agreement’ predicates before prohibitions attach. As this brief chapter discusses, the interplay of fast-moving technology and longstanding competition law doctrine makes this a thorny dilemma; direct interventions to promote competition law by design must be made to prevent the erosion of markets’ social benefits, and, by extension, consumers’ trust in them.

II.

BACKGROUND: THE RISE OF SMART CONTRACTS

The twenty-first century has seen a rapid rise in the development of intelligent agents. Siri, Alexa and Cortana – names inspired by Nordic legend,3 the TV show Star Trek4 and video games5 – are not just anthropomorphic female voices, but also household fixtures. But user-friendly technology does not exhaust the limits of software-powered agents. A series of developments at a deeper level of automated contracting has been occurring, including

1 See, e.g.,15 US Code Section 1 (Section 1 of the US Sherman Act of 1890) stating that ‘[e]very contract, combination in the form of trust or otherwise, or conspiracy, in restraint of trade of otherwise, or conspiracy [i.e., agreement] […] is declared to be illegal’); and An Act for the Prevention and Suppression of Combinations formed in Restraint of Trade (emphasis added), 52 Vict., Chapter 41, Act of the Parliament of the Domination of Canada (Brown Chamberlin, 1892) 489 (criminalizing offences in connection with such combinations). 2 I. Kerr, ‘Spirits in the Material World: Intelligent Agents as Intermediaries in Electronic Commerce’ (1999) 22 Dalhousie L.J. 190, 203 (observing that preexisting closed systems ‘have the commercial advantage of clarifying all legal rules in advance’, but that a ‘shift towards more open systems’ seems to be occurring). 3 ‘How Apple’s Siri Got Her Name’ (The Week, 8 January 2015) http://​theweek​.com/​articles/​ 476851/​how​-apples​-siri​-got​-name [accessed 3 March 2022]. 4 A. Alba, ‘Where Are All the Male AI Voice Assistants?’ (The Week, 18 March 2017) http://​ theweek​.com/​articles/​684606/​where​-are​-all​-male​-ai​- voice-assistants [accessed 3 March 2022]. 5 T. Warren, ‘The Story of Cortana, Microsoft’s Siri Killer’ (The Verge, 2 April 2014) www​.theverge​ .com/​2014/​4/​2/​5570866/​cortana​-windows​-phone​-8​-1​-digital​- assistant [accessed 3 March 2022].

213

214  Research handbook on cartels the development of ‘agreement technologies’, blockchain-driven ‘smart contracts’ and their potential successors. A.

Agreement Technologies

Since early in this century, computer scientists have worked on the problem of how to design autonomous software agents that can negotiate with each other to come to mutually acceptable agreements.6 Key factors include the degree of autonomy, interactivity and interoperability of such agents. Early in the past decade, combining insights from game theory, contract theory and computer science, Agreement Technologies (‘AT’) emerged as a field in its own right.7 Specialists in this area recognized that future applications would involve human users delegating increasingly complex tasks to software agents, which in turn would interact with each other in sophisticated ways to make and execute agreements automatically and, perhaps, autonomously.8 That future has come to pass. Google search involves an automated auction among advertisers for the user’s attention with respect to sponsored ads.9 Spotify contracts online with subscribers and artists, splitting up a royalty pool based on an automated process measuring an artist’s ‘streamshare’ – a figure analogous to market share based on the percentage of a rightsholder’s works out of the total music streamed.10 Observers predict that supply chains will be increasingly run by autonomous computerized applications.11 As a form of smart contracting, agreement technologies must necessarily interact with law and social science. The inputs and outputs of autonomous agents will not be limited to computer hard drives and screens. Instead, they will involve the pitfalls and externalities that have long been part and parcel of contract law. Where agreement technologies run into a glitch, or impose unreasonable costs on third parties, they will need to be dealt with by law and regulation. Among such concerns, competition law must come into play. Indeed, the subject of algorithmic collusion can be seen as an embryonic attempt to address potential competitive harm from smart contracting.12 The field of agreement technologies has already addressed issues of legal frameworks and regulation.13 Because the field focuses on the design of interoperable agents and plat6 See N. Jennings, ‘Agreement Technologies’ [2005] EEE/WIC/ACM International Conference on Intelligent Agent Technology 17. 7 See S. Ossowski (ed.), Agreement Technologies (Springer, 2012) (summarizing the European Cooperation in Science and Technology Action project on Agreement Technologies). 8 ibid. 9 J. Nicas, ‘How Google’s Ad Auctions Work’ (The Wall Street Journal, 19 January 2017) www​.wsj​ .com/​articles/​how​-googles​-ad​-auctions​-work​-1484827203 [accessed 3 March 2022]. 10 Spotify, ‘Loud and Clear’ https://​loudandclear​.byspotify​.com/​?question​=​how​-is​-stream​-share​ -calculated [accessed 3 March 2022]. 11 N. Duckworth, ‘Vibrant Network Ecosystems Are Turning Supply Chains into Competitive Weapons’ (KMWorld, 13 January 2020) www​.kmworld​.com/​Articles/​Editorial/​ViewPoints/​Vibrant​ -network​-ecosystems​-are​-turning​-supply​-chains​-into​-competitive​-weapons​-136019​.aspx [accessed 3 March 2022] (predicting increasing use of ‘intelligent agents’ to automate transactions along the supply chain). 12 On algorithmic collusion, see Chapter 11 in this volume. 13 T. Rubio, Z. Kokkinogenis, H. Lopes Cardoso, R. Rossetti and E. Oliveira, ‘Regulating Blockchain Smart Contracts with Agent-Based Markets’ in P. Moura Oliveira, P. Novais and L. Reis (eds), Progress in Artificial Intelligence (Springer, 2019).

Smart contracts and cartel law  215 forms, proposals include requiring regulation-enabled systems and hybrid approaches where markets themselves are active agents that ensure regulation even where contract execution is automated, as via blockchain.14 Essentially, competition law would have to be built into such systems by design. B. Blockchain Undoubtedly, the most famous example of a smart contracting technology is blockchain – even though the term itself is ambiguous. As Kevin Werbach notes, the term ‘the blockchain’ may be used for the universe of blockchains, the subset of public blockchains or just the public ledger for Bitcoin.15 In that last role, as the backbone of Bitcoin, among other cryptocurrencies, blockchain is probably the most well-known form of distributed ledger.16 Specifically, distributed ledgers represent a consensus of copied, shared and synchronized digital data spread out across geography and the internet, without a central administrator governing it. They need not be completely open to all; a class of systems called permissioned ledgers are designed for private groups to share information, including potentially information about transactions.17 Generally, blockchain in its myriad forms combines two powerful features. First, because of the way in which information is transformed via a cryptographic hash, the data encoded is very secure; an attempt at even a small change in the underlying data would make it possible via algorithmic comparison to see that a change has been made – the root hash is stored in the block header, and the comparison of changed information would generate a different hash. Additionally, the distributed nature of blockchain makes the data transparent to some set of users – varying from public to private access. For several years, blockchain has attracted significant attention for its cartel-fostering implications.18 As one prescient observer, Izabella Kaminska of the Financial Times, noted in 2015: Why are the great and the good of the banking and financial services world suddenly extolling the virtues of blockchain, the technology that underpins the artificial scarcity of bitcoin? Possibly because they’ve finally figured out what the technology really facilitates is cartel management for groups that don’t trust each other but which still need to work together if they’re to protect the value and stability of the markets they serve […] [S]uppose the rate of production could be algorithmically controlled to make it more difficult to mine commodities, money or financial assets as and when too many entities enter the system? And suppose, on top of that, such a mechanism was dubbed ‘technology’ instead of a cartel system outright? Would anti-trust authorities be inclined to look the other way then?19

ibid. K. Werbach, ‘Why the Blockchain Needs the Law’ (2018) 33 Berkeley Tech. L.J., 487, 489 n 1. 16 For example, Dogecoin is also based on blockchain, though apparently via a ‘find-and-paste’ job on Bitcoin’s blockchain; see M. Serrels, ‘Dogecoin: The Origin Story of the Elon Musk Supported Cryptocurrency’ (CNET, 8 May 2021) www​.cnet​.com/​news/​dogecoin​-the​-origin​-story​-of​-the​-elon​ -musk​-supported​-cryptocurrency/​[accessed 3 March 2022]. 17 Werbach (n 15) 498. 18 I. Kaminska, ‘Exposing the “If We Call It a Blockchain, Perhaps It Won’t Be Deemed a Cartel” Tactic’ (Financial Times, 11 May 2015) www​.ft​.com/​content/​bb7f42ec​-a049​-3739​-b74d​-131e9357694c [accessed 3 March 2022]. 19 ibid (emphasis in original). 14 15

216  Research handbook on cartels Thibault Schrepel, a leading commentator on competition law and blockchain, has raised the alarm concerning the disruptive potential that blockchain poses to existing competition law enforcement.20 In particular, he has focused on the usefulness of smart contracts for cartel members, whose price-fixing agreements are both punishable by competition law and unenforceable under contract law.21 The chief concern is that blockchain-based smart contracts will make it possible for cartel members to overcome their inherent trust dilemma, and simultaneously make it harder for enforcers to destabilize that trust with leniency programmes.22 Schrepel envisions smart contracts that cartel members can use to execute, for example, a market allocation agreement, and to automatically monitor compliance and make side payments among them.23 While concern may be warranted, there are reasons to doubt whether self-executing, self-monitoring and self-payment-making blockchain-based cartels are really on the horizon. Certainly, if detectable, a blockchain-based smart contract that automates collusion and the distribution of cartel profits to conspirators is forbidden by antitrust law; Professors Lin William Cong and Zhiguo He believe that such a mechanism would be ‘easy to detect’.24 It is worth observing that the distributed ledger potentially leaves an immutable transaction trail – one that may be easier to track than cash payments deposited in a bank secrecy jurisdiction. Moreover, as Cong and He observe, government regulators could be proactive in addressing blockchain-based smart contracts. We already see governments pushing back on some of the predicates for a blockchain-based cartel. First, to the extent that a system of automated side payments is required, governments around the world, including those of the US and China, are seeking to undercut private cryptocurrencies and to encourage government-backed cryptocurrencies in their stead.25 Presumably, if governments succeed in this effort, that would make secret, automated transactions via blockchain significantly harder. Additionally, government enforcers could proactively require identification methods that would reduce the ability to create a blockchain-based ‘secret meeting’. For example, cryptographic methods such as zero-knowledge proofs could make an account-based ledger susceptible to corporate identification; the government might be required to seek a warrant or a court order that would require the corporation to produce such proof of transaction, while otherwise preserving privacy. However, blockchain-based smart contracts may create a more difficult problem for antitrust enforcement due not to secrecy, but opacity. Cong and He model the potential due to blockchain for increased collusion, and thus lower output and higher prices.26 They point out that a distributed ledger improves a colluder’s accuracy at responding to defection from

20 See T. Schrepel, ‘Collusion by Blockchain and Smart Contracts’ (2019) 33 Harv. J.L. & Tech. 117, 119; T. Schrepel, ‘Is Blockchain the Death of Antitrust Law? The Blockchain Antitrust Paradox’ (2019) 3 Geo. L. Tech. Rev. 281, 285. 21 Schrepel, ‘Collusion by Blockchain’ (n 20) 141–43. 22 See H. Eenmaa-Dimitrieva and M.J. Schmidt-Kessen, ‘Creating Markets in No-Trust Environments: The Law and Economics of Smart Contracts’ (2019) 35 Computer L. & Secy. Rev. 69. 23 Schrepel, ‘Collusion by Blockchain’ (n 20) 141–43. 24 L. William Cong and Z. He, ‘Blockchain Disruption and Smart Contracts’ (2019) 32 Rev. of Fin. Studs. 1754, Section 2.3.2.1. 25 See G. Tett, ‘A Contest to Control Crypto Is Under Way’ (Financial Times, 25 June 2021) (predicting that private crypto will be ‘crushed’ in China). 26 Cong and He (n 24) (Section 2.3.2.2).

Smart contracts and cartel law  217 a cartel. In the absence of a distributed ledger, a cartel member considering whether to retaliate faces a dilemma that makes cartels relatively unstable: if the firm suddenly fails to receive buyers, is that because a defector is taking away buyers, or is it because buyers have left the market more generally? A distributed ledger reflecting transactions in the market may make it clear to the firm that buyers are indeed present in the market, a state that they can infer from being contacted for verification by the blockchain when new transactions occur. As a result, it becomes easier for cartel members to avoid unwarranted punishment of defectors due to imperfect information, and, all things being equal, collusion should become more stable.27 Note that this latter scenario is potentially a significant problem for antitrust law. Fortunately for the EU, existing law may be interpretable in such a manner as to prohibit a collusion-facilitating transparent blockchain of the type Cong and He describe as the adoption of a practice that facilitates joint dominance. Unfortunately, however, in the United States, antitrust law has long left a crack into which this scenario may fall.28 In the absence of a judicial finding of ‘agreement’ – either explicit or inferred from revelatory conduct – interdependent supracompetitive pricing is not punishable under Section 1 of the Sherman Act. It is unclear whether independent adoption by competing firms of blockchain or another system of distributed ledger would suffice for an inference of agreement. C.

Alternatives to Blockchain

The evolution of smart contracts is unlikely to stop with blockchain. First, there are questions about its continuing dependability in a future with quantum computing; faster computers could theoretically overcome its encryption.29 Moreover, doubts exist regarding the scalability of blockchain as it handles larger amounts of data.30 And even now, the US Department of Justice and FBI have apparently found methods to recover ransoms paid in Bitcoin, which suggests that blockchain will not always put violators beyond the reach of law enforcement.31 That said, the issue of smart contracts for competition law will doubtless not end with blockchain. The technology will continue to advance – witness the development of alternatives to blockchain. For example, Chia replaces the expensive computation that blockchain’s ‘proof

ibid. S. Mehra, ‘Antitrust and the Robo-Seller: Competition in the Time of Algorithms’ (2016) 100 Minn. L. Rev. 1323, 1340–41 (describing how algorithmic collusion may fall into the gap by which the Sherman Act allows interdependent supracompetitive parallel pricing). See also A. Ezrachi and M. Stucke, ‘Artificial Intelligence and Collusion: When Computers Inhibit Competition’ [2017] 5 Ill. L. Rev. 1785; and A. Ezrachi and M. Stucke, Virtual Competition: The Promise and Perils of the Algorithm-Driven Economy (Harvard University Press, 2016) 29 (arguing that ‘as pricing shifts from humans to computers, so too will the types of collusion and behavioral exploitation in which companies may engage’). 29 A. Fedorov, E. Kiktenko and A. Lvovsky, ‘Quantum Computers Put Blockchain Security at Risk’ (Nature, 19 November 2018) www​.nature​.com/​articles/​d41586​-018​-07449​-z [accessed 3 March 2022] (predicting that ‘within a decade, quantum computers will be able to break a blockchain’s cryptographic codes’). 30 A. Narayanan, J. Bonneau, E. Felten, A. Miller and S. Goldfeder, Bitcoin and Cryptocurrency Technologies: A Comprehensive Introduction (Princeton University Press, 2016) 48. 31 V. Romo, ‘How a New Team of Feds Hacked the Hackers and Got Colonial Pipeline’s Ransom Back’ (National Public Radio, 8 June 2021) (reporting that US enforcers obtained the encryption key linked to the Bitcoin account to which the ransom was delivered). 27 28

218  Research handbook on cartels of work’ requires with proof of alternative hard drive space; IOTA sets aside the blockchain model for a ‘tangle’, better suited for large-scale deployment on a system of nodes with limited computing power.32 The possibility that technology will advance further – for example, building on quantum computing to create even stronger encryption – cannot be discarded.33 As a result, a competition law approach to smart contracts cannot focus simply on blockchain. Instead, it must be proactive regarding challenges that are below the horizon, yet nonetheless foreseeable.

III.

THE CHALLENGE FOR COMPETITION LAW

A.

The ‘Agreement’ Dilemma, Algorithmic Collusion and Smart Contracts

As noted in connection with blockchain above, competition law, particularly in the United States, has had a longstanding dilemma concerning its requirement that an ‘agreement’ be found before punishing collusion. In the 1960s, Richard Posner and Donald Turner engaged in a famous debate on this point. Posner pushed for a broad definition of Section 1 of the Sherman Act that would cover interdependent pricing by oligopolists, even where they did not make an agreement.34 Turner argued that while the term ‘agreement’ should reach beyond only explicit agreements, punishing interdependent pricing was a mistake since business people were simply rationally optimizing their prices given market realities – and at any rate, such a rule would be difficult to administer, given that nearly all businesses price with a view to their competitors’ responses. In the past decade, this academic debate has reignited. Louis Kaplow has pointed out the paradox that the agreement requirement creates: interdependent supracompetitive pricing that does not require an agreement will tend to be more harmful and more stable than such pricing that does require agreement.35 The emergence of algorithmic collusion has spawned a modest literature focusing on the agreement requirement as a potential hazard to preventing consumer harm.36 This work has been triggered in part by the recognition from both theoretical modeling and empirical experience that algorithmic collusion is real and may expand.37

K. Werbach, The Blockchain and the New Architecture of Trust (MIT Press, 2019) 58. Federov et al (n 29). 34 R. Posner, ‘Oligopoly and the Antitrust Laws: A Suggested Approach’ (1969) 21 Stan. L. Rev. 1562, 1576–92. Later, wearing judicial robes, Posner walked back from the argument he had made as a law professor; see In re Text Messaging Antitrust Litigation, 782 F.3d 887 (7th Circuit, 2015) 874 (stating that ‘Harvard Law School Professor Louis Kaplow […] argues that tacit collusion should be deemed a violation of the Sherman Act. That of course is not the law, and probably shouldn’t be’). 35 L. Kaplow, ‘On the Meaning of Horizontal Agreements in Competition Law’ (2011) 99 Cal. L. Rev. 683, 815. 36 M. Gal, ‘Algorithms as Illegal Agreements’ (2019) 34 Berkeley Tech. L.J. 67; S. Thomas, ‘Harmful Signals: Cartel Prohibition and Oligopoly Theory in the Age of Machine Learning’ (2019) 15 J. of Comp. L. & Econ. 159; F. Beneke, ‘Remedies for Algorithmic Tacit Collusion’ (2021) 9 J. Antitrust Enf. 152; and S. Mehra, ‘Price Discrimination-Driven Algorithmic Collusion: Platforms for Durable Cartels’ (2021) 26 Stan. J.L. Bus. & Fin. 171. 37 A. Ezrachi and M. Stucke, ‘Sustainable and Unchallenged Algorithmic Tacit Collusion’ (2020) 17 Nw. J. Tech. & Intell. Prop. 217. See, however, U. Schwalbe, ‘Algorithms, Machine Learning and Collusion’ (2018) 14 J. of Comp. L & Econ. 568. 32 33

Smart contracts and cartel law  219 What remains undecided is whether and how competition law must change to address its agreement requirement. Competition law scholars are still at an embryonic stage in grappling with these questions.38 Algorithmic collusion tends to fall into gaps in the agreement requirement because competitors can use mass data collection, computerized processing and automated responses to achieve supracompetitive pricing without explicit agreement. Under the traditional competition law framework, the question is when the steps taken to employ such technology can be equated with an ‘agreement’. Smart contracts make this question still more urgent. Technologies such as the blockchain make possible interoperability, information sharing and automated response among very large numbers of players – those players being potentially economic competitors. B.

Consumer Welfare, the Traditional Competition Law Framework and Beyond

The traditional competition law framework focuses on whether firms have made an agreement that injures consumer welfare, the goal being to bring private agreements into alignment with social welfare. As a result, it is primarily a reactive tool whose ex post penalties create ex ante deterrence. Smart contracts and the technologies that may follow them raise questions about the future sufficiency of the ‘antitrust cop’ as a model for promoting and safeguarding the social benefits of competitive markets.39 In the twentieth century, traditional markets were the only way to gather widespread information, digest it and allocate resources in an acceptable manner.40 Digital platforms increasingly can digest massive amounts of information and match resources with those who need them. Whether resources are allocated based on markets or an alternative, such as closed firms, has long been a question of interest to law and economics;41 however, traditional antitrust law stops at the boundaries of the firm.42 The emerging forms of organization and allocation may require a different approach than competition law’s existing framework. Beyond antitrust law, Rory van Loo has pointed out that the delegation of consumer sovereignty to digital assistants may have significant new impacts that require different, proactive regulation.43 Similarly, competition law may have to take proactive steps to prevent markets being balkanized by technological developments into a series of walled gardens that do not serve consumer welfare or social interest. Competition has come to view interoperability as a potential remedy for consumer harm in the context of platforms and digital networks.44 Going forward, as technology may shift economic activity

See Gal (n 36) 101–13 (adapting the stepwise rule of reason approach to the context of algorithmic collusion). 39 E. Iacobucci and R. Winter, ‘Abuse of Joint Dominance in Canadian Competition Policy’ (2010) 60 U. Tor. L.J. 219 (stating that ‘[t]he central purpose of antitrust law in a market economy is to align firms’ incentives with the social interest by encouraging vigorous competition within markets’). 40 Friedrich Hayek reconceptualized markets as a system of information transfer between agents that assembled data in a manner ‘beyond human capacity’ otherwise: F. Hayek, Individualism and Economic Order (University of Chicago Press, 1948) 187. 41 R. Coase, ‘The Nature of the Firm’ (1937) 4 Economica 386. 42 S. Paul, ‘Antitrust as Allocator of Coordination Rights’ (2020) 67 UCLA L. Rev. 378. 43 R. van Loo, ‘Digital Market Perfection’ (2019) 117 Mich. L. Rev. 8. 44 M. Kades and F. Scott Morton, ‘Interoperability as a Competition Remedy for Digital Networks’ (Equitable Growth, September 2020) https://​equitablegrowth​.org/​wp​-content/​uploads/​2020/​09/​092320​ 38

220  Research handbook on cartels from traditional markets to closed or semi-closed digital networks, mandated interoperability and competition by design may be required to safeguard consumer welfare.45

IV. CONCLUSION Smart contracts have the potential to unsettle competition law. But they are a step in an ongoing process of technological evolution, from enterprise software-powered big-box store chains, to Web 1.0 e-commerce, to algorithmic competition, and ultimately to further stages we cannot yet imagine. As markets and their substitutes and complements evolve, so too must competition law, if it is to continue to serve consumer welfare and the public interest.

-WP​-Interoperability​-as​-a​-competition​-remedy​-for​-digital​-networks​-Kades​-and​-Scott​-Morton​.pdf [accessed 3 March 2022]. 45 S. Mehra, ‘What is an Antitrust Problem Anyway?’ (2003) 68 Antitrust Bulletin (forthcoming).

13. Cartels and the exchange of information Daniel A. Crane

I. INTRODUCTION This chapter focuses on the relationship between cartels and the exchange of information between competitors. Clearly, members of a cartel have to exchange information on price, output and other key ingredients of their collusive arrangement. But not all exchanges of information among competitors are part of cartel agreements. Competitors have legitimate reasons to exchange certain kinds of information, and information exchanges can have procompetitive effects and strengthen competition under certain circumstances.1 What, then, is the relationship between information exchanges and cartels, and how does the law treat this phenomenon? Historically, this has been a challenging question for courts – one largely dependent on prevailing attitudes towards associations of competitors and horizontal collaboration. US law, in particular, has evolved significantly over time towards greater solicitude towards competitor information exchanges. EU law employs similar categories, but remains somewhat more restrictive on information exchanges. Overall, as will become apparent below, competitor information exchange is tolerated if it serves to facilitate more effective competition and inform all market participants, but it otherwise poses significant antitrust risks.

II.

EVOLUTION OF US LEGAL DOCTRINE

A.

Associationalism and Its Discontents

The roots of US legal policy towards competitors’ exchanges of information can be found in the ‘Associationalist’ movement in the pre-First World War era.2 Associationalism provided an ostensible solution to the problem of ‘ruinous competition’ associated with classical economic thought.3 Jerome Eddy’s prominent 1912 book The New Competition suggested that competitor information exchange could solve this ruinous competition problem. Eddy argued that businesses in the same industry often announced prices at levels that would yield K.-U. Kühn and X. Vives, ‘Information Exchanges Among Firms and their Impact on Competition’ (December 1994) https://​blog​.iese​.edu/​xvives/​files/​2011/​09/​Information​-Exchanges​-and​-their​-Impact​ -on​-Competition​.pdf [accessed 3 March 2022] (reaching the ‘general conclusion that information sharing in itself cannot generally be regarded as detrimental to welfare and therefore a restriction of competition’); and A. Kirby, ‘Trade Associations as Information Exchange Mechanisms’ (1988) 19 Rand. J. of Econ. 138 (finding that where total cost functions are sufficiently convex, expected consumer surplus also always increases when information is shared). 2 The following paragraphs draw in substantial part from E. Fox and D. Crane, Cases and Materials on US Antitrust in Global Context (3rd edn, West Academic Publishing, 2020). 3 M. Winerman and W. Kovacic, ‘Outpost Years for a Start-Up Agency: The FTC from 1921–1925’ (2010) 77 Antitrust L.J. 145, 155. 1

221

222  Research handbook on cartels ‘fair’ profits but that then engaged in secret price shading that drove real prices down towards cost. This tendency could be halted by competitors’ exchanges of pricing information. If every sale were immediately publicized, firms would stop shading prices, since such shading would only lead to competitor retaliation. Edy proposed ‘open price associations’ that would supply members with data on industry-wide production, inventories and sales. Eddy’s ideas gained wide acceptance, and information-exchanging trade associations became common in American business. Although Edy’s ideas sound suspiciously like a form of soft cartelization, they found support in high places during the 1920s, particularly in the person of Secretary of Commerce and future President Herbert Hoover.4 Hoover was an engineer by training and a wartime production bureaucrat, who disliked cartels but supported industry associations as producing technocratic efficiency. Hoover saw associations as key components in a system of managerial capitalism that eliminated waste and maximized efficiency. He believed that spontaneous markets produced too little cooperation and information, and that economic decision makers therefore lacked the information necessary to make rational decisions. What was needed was a ‘plan of individualism and associational activities’ that would ‘preserve the initiative, the inventiveness […] the character of man’ and yet ‘enable us to socially and economically synchronize this gigantic machine that we have built out of applied science’.5 Trade associations were to be the ‘key component of the envisioned intelligence and coordinating apparatus’.6 In Hoover’s vision, trade associations would be sanctioned simultaneously by the Department of Commerce and the Chamber of Commerce – by both the public and private sectors. In 1921 Hoover’s associationalist vision took a hit in the first significant Supreme Court decision on information exchange – American Column & Lumber Co. v US.7 American Column involved an ‘open price plan’ of the Hardwood Manufacturers Association under which members of the association reported detailed daily sales and shipping reports, monthly production and stock reports, price lists and inspection reports to the association. The American Hardware Manufacturers’ Association consisted of 400 members operating 465 hardwood mills, one third of the total production in the United States.8 In 1918, the Association adopted an ‘Open Competition Plan’, with the purpose of disseminat[ing] among members accurate knowledge of production and market conditions so that each member may gauge the market intelligently instead of guessing at it; to make competition open and above board instead of secret and concealed; to substitute, in estimating market conditions, frank and full statements of our competitors for the frequently misleading and colored statements of the buyer.9

See E. Hawley, ‘Herbert Hoover and the Sherman Act, 1921–1933: An Early Phase of a Continuing Issue’ (1989) 74 Iowa L. Rev. 1067. On classical economic thought and the ruinous competition hypothesis, see generally D. Crane and H. Hovenkamp, The Making of Competition Policy: Legal and Economic Sources (Oxford University Press, 2013). 5 H. Hoover, ‘Address before the American Engineering Council, Washington, DC, on the Engineer’s Place in the World’ (10 January 1924) (available at Hoover Presidential Library, Public Statement File, Volume 14, Number 345A). 6 Hawley (n 4) 1072. 7 257 US 377 (1921). 8 ibid 391. 9 ibid 393. 4

Cartels and the exchange of information  223 The association served as a ‘central clearing house for information on prices, trade statistics and practices’ among its members.10 The association then compiled these data and transmitted them to its members, and the members then met to discuss the reports at monthly meetings. Prior to the monthly meetings, members shared information about their future price and output plans. The Supreme Court found this plan to be ‘simply an expansion of the gentlemen’s agreement of former days, skillfully devised to evade the law’.11 In the Court’s opinion, the fundamental purpose of the Plan was to procure ‘harmonious’ individual action among a large number of naturally competing dealers with respect to the volume of production and prices, without having any specific agreement with respect to them, and to rely for maintenance of concerted action in both respects, not upon fines and forfeitures as in earlier days, but upon what experience has shown to be the more potent and dependable restraints, of business honor and social penalties.12

In other words, the association’s information exchange agreement was nothing other than a disguised cartel agreement. Importantly, Justices Holmes and Brandeis filed dissenting opinions, finding the voluntary and non-coercive exchange of information critically important to the functioning of a free and egalitarian society.13 They emphasized that the plan was voluntary and open, that the information collected was available to buyers, sellers, and the general public and that meetings were open to the public – a point recalling Brandeis’s famous aphorism that ‘[s]unlight is said to be the best of disinfectants’.14 The information exchange might somewhat lessen competition, but Brandeis had already shown in his Chicago Board of Trade opinion that ‘the Sherman Act does not prohibit every lessening of competition’ and ‘it is lawful to regulate competition in some degree’.15 But, continued Brandeis, the purpose of the Open Competition Plan was not even to regulate competition, but rather ‘to make rational competition possible, by supplying data not otherwise available, and without which most of those engaged in the trade would be unable to trade intelligently’.16 The Supreme Court returned to information exchange four years later in Maple Flooring Manufacturers v US,17 this time with a more sympathetic attitude. Maple Flooring involved an unincorporated trade association of members selling maple, beech and birch flooring that distributed among its members statistics on the average cost to association members of all dimensions and grades of flooring and their average prices and amount of stock at hand. The association also held meetings of members to discuss the information exchanged. The Justice Department saw this as a cartel agreement akin to that condemned in American Column, but the Supreme Court disagreed. Although it recognized that ‘[e]xchange of price quotations of market commodities tends to produce product uniformity of prices’, it also believed that ibid. ibid 410–11. 12 ibid 411. 13 See generally Daniel A. Crane, ‘Collaboration and Competition in Information and News During Antitrust’s Formative Era’ (29 June 2020) https://​knightcolumbia​.org/​content/​collaboration​-and​ -competition​-in​-information​-and​-news​-during​-antitrusts​-formative​-era [accessed 3 March 2022]. 14 L. Brandeis, Other People’s Money and How the Bankers Use It (Frederick A. Stokes, 1914) 92 (‘Sunlight is said to be the best of disinfectants; electric light the most efficient policeman’). 15 257 US 377 (1921) 415, citing Chicago Board of Trade v United States, 246 US 231 (1918). 16 257 US 377 (1921) 415. 17 268 US 563 (1925). 10 11

224  Research handbook on cartels ‘[k]nowledge of the supplies of available merchandise tends to prevent over-production and to avoid the economic disturbances produced by business crises resulting from over-production’. The Court then expressed a broadly supportive view of information exchange: It is the consensus of opinion of economists and of many of the most important agencies of Government that the public interest is served by the gathering and dissemination, in the widest possible manner, of information with respect to the production and distribution, cost and prices in actual sales, of market commodities, because the making available of such information tends to stabilize trade and industry, to produce fairer price levels and to avoid the waste which inevitably attends the unintelligent conduct of economic enterprise. Free competition means a free and open market among both buyers and sellers for the sale and distribution of commodities. Competition does not become less free merely because the conduct of commercial operations becomes more intelligent through the free distribution of knowledge of all the essential factors entering into the commercial transaction. General knowledge that there is an accumulation of surplus of any market commodity would undoubtedly tend to diminish production, but the dissemination of that information cannot in itself be said to be restraint upon commerce in any legal sense. The manufacturer is free to produce, but prudence and business foresight based on that knowledge influence free choice in favor of more limited production. Restraint upon free competition begins when improper use is made of that information through any concerted action which operates to restrain the freedom of action of those who buy and sell.18

B.

The High Point of Judicial Hostility to Information Exchange

American Column and Maple Flooring left some doubt about the future direction of judicial attitudes toward competitors’ exchanges of information. During the Depression era, the courts initially retreated from antitrust scrutiny of collaborative efforts by competitors to stabilize industry prices. In Sugar Institute v United States,19 the Court ruled that an industry ‘plagued’ by secret price concessions granted by ‘unethical’ refiners in a market of declining demand for a standardized product could lawfully agree to make advance announcement of price changes, along with exchanges of statistical information. Defendants were enjoined only from agreeing with each other to adhere to their announced prices. As long as they each maintained pricing freedom, they were allowed to act cooperatively ‘to end abuses and to foster fair competitive opportunities’.20 The Court held that ‘the fact that the correction of abuses may tend to stabilize a business, or to produce fairer price levels [does not] require that abuses should go uncorrected or that an effort to correct them should for that reason alone be stamped as an unreasonable restraint of trade’.21 Depression-era permissiveness towards competitor collaborations came to a crashing halt after 1937 when the Antitrust Division became increasingly aggressive in challenging cartel-like behaviour and the New Deal Supreme Court shifted dramatically in favour of antitrust enforcement.22 While trade associations continued to flourish, they became considerably more cautious about exchanging sensitive pricing and output information. When the Supreme Court next decided an information exchange case, it underlined that such caution was warranted. ibid 582–83. 297 US 553 (1936). 20 ibid 598. 21 ibid. 22 D. Crane, ‘The Story of United States v Socony-Vacuum: Hot Oil and Antitrust in the Two New Deals’, in E. Fox and D. Crane (eds), Antitrust Stories (Foundation Press, 2007) 107. 18 19

Cartels and the exchange of information  225 United States v Container Corporation of America23 did not involve a trade association, but rather competitor-to-competitor inter-seller price verification. The defendants, accounting for about 90 per cent of corrugated container sales in the Southeastern United States, were in the habit of informally exchanging price quotations to specific customers, with the expectation of reciprocity. Interestingly, the Court noted that the general tendency of the information exchange was to stabilize prices at a ‘downward level’. Although this might have been thought to be beneficial for buyers in the market, the Court observed that any ‘interference with the setting of price by free market forces is unlawful per se’.24 The Court then noted a series of factors that made this inter-seller price verification particularly suspect: the industry was dominated by a few sellers; the product was fungible and the only competition was over price; and demand was inelastic. In condemning the arrangement as unlawful, the Court held: ‘Price is too critical, too sensitive a control to allow it to be used even in an informal manner to restrain competition.’25 Following Container Corp., it seemed that certain kinds of competitor information exchanges might not only raise antitrust scrutiny, but be adjudged as illegal per se. However, with the advent of the Chicago School’s growing influence, the US Supreme Court shut the door on this possibility in 1978 in United States v Gypsum Co.26 The Justice Department indicted gypsum board manufacturers for criminal price fixing. A key part of the evidence was that the defendants frequently called one another to verify prices to particular customers – the sort of inter-seller price verification condemned in Container Corp. There were between 9 and 14 producers in the industry, with the eight largest companies accounting for 94 per cent of the market. The Justice Department secured a criminal conviction, but the Supreme Court reversed, holding that the jury charge was inappropriate for allowing the jury to find liability based on the information exchange’s effect on pricing, regardless of whether defendants intended or were even aware of the price effect. The Court also noted, in passing, that mere information exchange is never per se illegal: The exchange of price data and other information among competitors does not invariably have anticompetitive effects; indeed such practices can in certain circumstances increase economic efficiency and render markets more, rather than less, competitive. For this reason, we have held that such exchanges of information do not constitute a per se violation of the Sherman Act.27

Gypsum left the law on information exchange in some degree of confusion. Container Corp. had stated unequivocally that information exchange among competitors could support a finding of per se illegality and Gypsum, without purporting to overrule Container Corp., had held the contrary. The United States Supreme Court has not returned to the question of information exchange since Gypsum, leaving it to the lower courts to harmonize the doctrine.

393 US 333 (1969). ibid 512. 25 ibid. 26 438 US 422 (1978). 27 ibid 441 n 16. 23 24

226  Research handbook on cartels C.

Contemporary Doctrinal Structure of US Law on Information Exchange

The most comprehensive and authoritative analysis of contemporary antitrust doctrine toward competitor information exchange can be found in then-Judge Sonia Sotomayor’s decision for the US Court of Appeals for the Second Circuit in Todd v Exxon Corp.28 The decision is significant both because of its scholarly harmonization of Supreme Court precedent into an operationalizable doctrinal package and because of Judge Sotomayor’s subsequent elevation to the US Supreme Court. Todd involved a class action lawsuit by managerial and technical employees of 14 major integrated and petrochemical companies that collectively accounted for 80–90 per cent of the industry’s revenues and employed an equal share of the industry’s workforce. The defendant oil companies instituted a system whereby they periodically conducted surveys comparing past and current employee salary information and benchmarking them to job categories at Chevron, one of the defendants. A third-party consultant, Towers Perrin, compiled the information, analysed and refined it and distributed it to the defendants. Human resources personnel from the defendants then met at least three times a year to discuss the salary information and their budgets for future salaries. The trial judge dismissed the plaintiffs’ complaint, but the Court of Appeals reversed and allowed the case to proceed. Judge Sotomayor began by helpfully distinguishing between two types of antitrust cases in which information exchange may play a role. The first type is hard-core price fixing or cartel cases in which the plaintiff must prove that the defendants actually agreed to fix prices. In such cases, direct or ‘smoking gun’ evidence of collusion may be missing, and the question then is whether the circumstantial evidence is sufficient to prove collusion. When firms exchange in parallel conduct accompanied by ‘plus factors’ – factors that tend to disprove the possibility that the parallel conduct was the result of independent decision making – the circumstantial evidence may be sufficient to prove per se illegal price fixing.29 Defendants’ use of facilitating factors – factors that make it easier for cartel agreements to form or to punish cheating – may serve as a plus factor. Information exchange may be a facilitating factor, and hence may serve (with other circumstantial evidence) to meet a plaintiff’s burden of proving that the defendants engaged in per se illegal price fixing. A ‘closely related but analytically distinct type of claim’ is that the ‘violation lies in the information exchange itself – as opposed to merely using the information exchange as evidence upon which to infer a price-fixing agreement’.30 Unlike the first category, ‘[t]his exchange of information is not illegal per se, but can be found unlawful under a rule of reason analysis’.31 This doctrinal synthesis made sense of the Supreme Court’s muddled case law on information exchange. Cases such as Container Corp. should be understood as ones where the information exchange helped to prove that the defendants were engaging in price fixing. However, where the plaintiff was not alleging actual price fixing, but rather that the information exchange had the effect of causing a price increase through conscious parallelism, the claim required analysis under the rule of reason, including proof of a relevant market, market 275 F.3d 191 (2d Cir. 2001). Bell Atlantic Corp v Twombly, 550 US 544 (2007). 30 275 F.3d 191 (2d Cir. 2001) 199. 31 ibid. 28 29

Cartels and the exchange of information  227 power and anticompetitive effects and consideration of defendants’ offsetting efficiencies justifications. In Todd, the plaintiff was not alleging a per se claim, and therefore did not have the burden of proving that the defendants went further than agreeing to exchange information and actually agreed to fix prices. Cases of that variety are still brought, and sometimes founder on the lack of clear evidence that the defendants agreed to more than information exchange.32 On the other hand, the Todd plaintiff faced burdens that the plaintiff in a per se case would not – namely, of showing market power in a properly defined relevant market and anticompetitive effects. Despite these differences in the two types of information exchange theories, there are significant commonalities regardless of which theory a plaintiff pursues. The Todd Court analysed two broad categories of factors relevant to a rule of reason claim, but they could also be relevant to determining the plausibility of a claim that information exchange facilitated price fixing. The first bucket of factors falls under the rubric of market structure. The more that a market is structurally conducive to anticompetitive effects from information exchange, the stronger the plaintiff’s claims. Three structural factors are significant: fungibility; elasticity; and concentration. Fungibility: As is generally true in cartel cases, it is easier for competitors to soften competition through coordinated action if they are selling a homogenous or fungible product, as opposed to one that is differentiated or heterogeneous. Accordingly, the more that a product is fungible, the more likely it is that an information exchange among competitors evidences a price-fixing agreement (in a per se case) or supports an inference that the information exchange causes an anticompetitive effect. Elasticity: Market-wide elasticity of demand is an important factor in determining whether the competitors sharing information have the collective market power to increase prices and reduce output. In a market with elastic demand, customers may substitute outside the market in response to a price increase, thereby making coordinated price increases less profitable and therefore less tempting.33 Concentration: The more highly concentrated a market, the more likely that oligopolists can coordinate on pricing or output, whether explicitly in a per se case or implicitly in a rule of reason case. Notably, in Todd the defendants argued that the petroleum market would be considered unconcentrated under the Horizontal Merger Guidelines’ Herfindahl Hirschman Index concentration categories, but Judge Sotomayor found that the market was nonetheless sufficiently concentrated to sustain the plaintiff’s claims. The second bucket of factors goes to the nature of the data exchanged, who has access to it and how it is exchanged. Four factors are significant in this facet of the analysis: timing; specificity; public dissemination; and other meetings. Timing: In Todd, the court noted that ‘exchanges of future price information are considered especially anticompetitive’.34 The same goes for exchanges of information about prices being

See, e.g., Blomkest Fertilizer, Inc. v Potash Corp. of Saskatchewan, 203 F.3d 1028 (8th Cir. 2000) (en banc). 33 Observe that this elasticity factor is closely related to market definition, which also considers consumer cross-elasticity of demand in determining which products are substitutes. 34 275 F.3d 191 (2d Cir. 2001) 212. 32

228  Research handbook on cartels currently quoted, since such information may serve to diminish the seller’s incentives to lower their prices. By contrast, exchanges of past information are less threatening to competition. Specificity: As a general matter, the more aggregated and general the information exchanged, the safer is its exchange from an antitrust perspective. On the other hand, granular or customer-specific information exchange is problematic: ‘Price exchanges that identify particular parties, transactions, and prices are seen as potentially anticompetitive because they may be used to police a secret or tacit conspiracy to stabilize prices.’35 Public dissemination: Picking up on a theme from Justice Holmes’s and Brandeis’s dissents in American Column, the courts today recognize that information exchanged by competitors can serve a salutary public purpose if it is made widely available, including to buyers or other transactional counterparties. Judge Sotomayor noted in Todd that ‘in the traditional oligopoly (seller-side) context, access to information may better equip buyers to compare products, rendering the market more efficient while diminishing the anticompetitive effects of the exchange’.36 Conversely, when the information exchanged is kept confined to the competitors exchanging it, there is a stronger inference that it is being used to subvert competition. Other meetings: Although it is not inherently unlawful for competitors to meet, when competitors exchange sensitive pricing or other commercial information and then get together to talk about it, the suspicion arises that the further discussions are efforts to use the exchanged information to fix prices. Until instructed not to do so by a federal court,37 the US Justice Department was in the habit of reading the following quotation from Adam Smith during its summation in criminal cartel cases: ‘People of the same trade seldom meet together even for merriment or diversion, but the conversation ends in a conspiracy against the public and in some contrivance to raise prices.’38 Although the law is clear that oligopolists cannot be held liable merely for the act of meeting together, their meetings following the exchange of competitive information can help to create an inference that they were price fixing (again, in a per se case) or that their information exchange had an anticompetitive effect (again, in a rule of reason case). Although these factors are cumulative and it is impossible to say how much weight each holds, the combination of these factors provides fairly clear guidance on the risks of information exchange under current US law. Exchanges of information by competitors operating in markets with differentiated goods, high demand elasticity and low seller concentration, and where the information is for past transactions, aggregated and widely disseminated and there are no further meetings, lie at the low-risk end of the spectrum. At the opposite end are cases of homogenous goods, low demand elasticity, high concentration and the dissemination of present or future data on a customer-specific basis, with the information only disseminated to the firms sharing the information and further meetings to discuss the information exchanged.

ibid. ibid 213. 37 US v Taubman, 297 F.3d 161 (2d Cir. 2002). 38 A. Smith, An Inquiry Into the Nature and Causes of the Wealth of Nations (First published in 1776, Great Books edition, 1952) 55. 35 36

Cartels and the exchange of information  229

III.

EU LEGAL DOCTRINES

Trade associations have a long and influential history throughout much of Europe. Despite this – or perhaps because of it – the EU takes a much more severe perspective on competitor information exchange than does US law. Unlike Gypsum, which as we have seen rejected per se illegality for information exchange itself, certain types of information exchange may be condemned as restrictions of competition by object under Article 101 of the TFEU. In the Thyssen-Stahl case, the Advocate General argued that [a]n agreement on the exchange of information is incompatible with the rules on competition, even where the relevant market is not a highly concentrated oligopolistic market, if it reduces or removes the degree of uncertainty as to the operation of that market with the result that competition between undertakings is restricted.39

The Court of Justice took a similar point of view in the 2009 T-Mobile Netherlands judgment,40 which involved exchanges of information among representatives of five large mobile telephone operators. The operators ‘held a meeting [on 13 June 2001 at which] they discussed […] the reduction of standard dealer renumerations for postpaid subscriptions, which was to take effect on or about 1 September 2001’.41 The defendant firms argued that they had not made any agreement within the purview of Article 101, and that, even if they had, it lacked any anticompetitive effect. The Dutch court made a preliminary reference to the Court of Justice, which in turn held that for concerted practice to have an anticompetitive object, ‘it is sufficient that it has the potential to have a negative impact on competition’.42 In the case of an anticompetitive object, it is not necessary that there be an anticompetitive effect, which ‘can only be of relevance for determining the amount of any fine and assessing any claim for damages’.43 The Court observed that ‘Article [101] of the Treaty on the Functioning of the European Union is designed to protect not only the immediate interests of individual competitors or consumers but also to protect the structure of the market and thus competition as such’.44 A detailed examination of the legality of information exchanges under Article 101 appears in the 2011 Guidelines on the Applicability of Article 101 of the Treaty on the Functioning of the European Union to Horizontal Co-operation Agreements.45 The Guidelines frame their discussion around three general propositions: Information exchange is a common feature of many competitive markets and may generate various types of efficiency gains. It may solve problems of information asymmetries, thereby making markets more efficient. Moreover, companies may improve their internal efficiency through benchmarking against each other's best practices. Sharing of information may also help companies to save costs by reducing their inventories, enabling quicker delivery of perishable products to consumers, or dealing

Case T-141/94 Thyssen Stahl v Commission [1999] ECR II-347. Case C-8/08 T-Mobile Netherlands BVand others v Raad van bestuur van de Nederlandse Mededingingsautoriteit [2009] ECR I-4529. 41 ibid para 12. 42 ibid para 31. 43 ibid paras 30 and 31. 44 ibid para 38. 45 Guidelines on the Applicability of Article 101 of the Treaty on the Functioning of the European Union to Horizontal Co-operation Agreements [2011] OJ C11/1. 39 40

230  Research handbook on cartels with unstable demand etc. Furthermore, information exchanges may directly benefit consumers by reducing their search costs and improving choice.46 However, the exchange of market information may also lead to restrictions of competition in particular in situations where it is liable to enable undertakings to be aware of market strategies of their competitors. The competitive outcome of information exchange depends on the characteristics of the market in which it takes place (such as concentration, transparency, stability, symmetry, complexity etc.) as well as on the type of information that is exchanged, which may modify the relevant market environment towards one liable to coordination.47 [C]ommunication of information among competitors may constitute an agreement, a concerted practice, or a decision by an association of undertakings with the object of fixing, in particular, prices or quantities. Those types of information exchanges will normally be considered and fined as cartels. Information exchange may also facilitate the implementation of a cartel by enabling companies to monitor whether the participants comply with the agreed terms. Those types of exchanges of information will be assessed as part of the cartel.48

Although articulated in the different vocabulary of Article 101, the guidelines present a similar set of criteria to those identified in US law. Both the structure of the market and the nature of the information exchanged are relevant criteria. On market structure, the guidelines find information exchange more likely to have anticompetitive effects ‘in markets which are sufficiently transparent, concentrated, non-complex, stable and symmetric’.49 Similarly, exchanges of individualized data are more concerning than those of aggregated data,50 exchanges of historic data are more benign than those of present or future data,51 highly frequent exchanges are more suspect,52 and ‘exchanges of genuinely public information are unlikely to constitute an infringement of Article 101’, but a greater suspicion arises when data are kept to the exchanging parties.53 One concern raised in the guidelines that is characteristic of the EU’s concern for fringe players in the market and is not shared by US antitrust concerns foreclosure effects from information exchange: An exclusive exchange of information can lead to anti-competitive foreclosure on the same market where the exchange takes place. This can occur when the exchange of commercially sensitive information places unaffiliated competitors at a significant competitive disadvantage as compared to the companies affiliated within the exchange system. This type of foreclosure is only possible if the information concerned is very strategic for competition and covers a significant part of the relevant market.54

48 49 50 51 52 53 54 46 47

ibid para 57. ibid para 58. ibid para 59. ibid para 77. ibid para 89. ibid para 90. ibid para 91. ibid para 92. ibid para 70.

Cartels and the exchange of information  231

IV.

OPEN QUESTIONS OF LAW AND POLICY

Today, the legal criteria governing information exchanges among competitors seem relatively settled and predictable in both the US and the EU. Nonetheless, there remain open questions about antitrust risk created by information exchanges that are not easily answered simply by surveying the applicable legal doctrines and agency guidance. These fall into two big buckets: (1) to what extent is information exchange, standing alone, sufficient to create an inference of price fixing?; (2) when there is no evidence of actual price fixing, to what extent does information exchange still pose a serious antitrust risk? As previously observed, under US law challenges to information exchanges can be brought either on the theory that the information exchange evidences the formation of a price-fixing conspiracy (in which event it is illegal per se) or on the theory that the information exchange itself harms competition by reducing competitors’ incentives to compete aggressively on price. Although both of these theories are available in principle, the former theory is the one more commonly invoked. Plaintiffs relatively rarely challenge information exchange as an antitrust violation itself, and more frequently challenge it as circumstantial proof of price-fixing collusion. Although it is common to see competitor information exchange as part of the evidence supporting price-fixing charges in criminal antitrust cases, stand-alone civil challenges to alleged collusion based principally on the existence of competitor information exchange have not been overwhelmingly successful.55 Courts frequently note, per Gypsum, that competitor information exchange is not per se illegal standing alone and that its existence does not necessarily raise an inference of collusion. Typically, circumstantial evidence of agreement on prices or output in addition to the information exchange is necessary to make out a collusion case. Turning to rule of reason challenges, there have been relatively few cases in which plaintiffs bring such claims. Although the Todd Court allowed a private challenge to information exchange proceed past the motion to dismiss stage, most plaintiffs have chosen to allege per se theories, which comes with the advantage of not having to prove a relevant market, market power or anticompetitive effects, but the disadvantage of having to prove actual collusion. In one recent enforcement action, the Justice Department did bring a civil suit against six broadcast television companies that had allegedly participated in information exchange regarding revenue pacing – which compares a broadcast station’s revenues booked for a certain time period to the revenues booked in the same point in the previous year – in advertising

55 Significant recent US Court of Appeals judgments include: In re Flat Glass Antitrust Litig., 385 F.3d 350 (3rd Cir. 2004) (relying on information exchange among competitors to find that factual issues precluded summary judgment for defendants on conspiracy claims); Stanilaus Food Prods. Co. v USS-POSCO Indus., 803 F.3d 1084 (9th Cir. 2015) (rejecting claim that inter-firm communications sufficed as plus factors to prove conspiracy); In re Chocolate Confectionary Antitrust Litig., 801 F.3d 383 (3rd Cir. 2015) (finding that information exchange not sufficient to create an inference of price fixing conspiracy); In re Text Messaging Antitrust Litig., 782 F.3d 867 (7th Cir. 2015) (same); Omnicare, Inc. v UnitedHealth Group, Inc., 629 F.3d 697 (7th Cir. 2011) (finding that exchange of even pricing data insufficient to support inference of conspiracy); Blomkest Fertilizer, Inc. v Potash Corp. of Saskatchewan, 203 F.3d 1028 (8th Cir. 2000) (en banc) (divided en banc court finding that information exchange among competitors insufficient to create a genuine issue of material fact showing the presence of a price-fixing conspiracy among potash manufacturers).

232  Research handbook on cartels sales.56 While not alleging that the defendants had engaged in actual price fixing, the Justice Department alleged that the ‘exchange of competitively sensitive information allowed these television broadcast companies to disrupt the normal competitive process of spot advertising in markets across the United States’.57 To the extent that anti-competitive effects and procompetitive justifications are important in rule of reason cases, there remain open questions about the application of those principles. One question concerns what evidence a plaintiff would need to show to establish that an information exchange had an anti-competitive effect on prices. Information exchange among competitors can have the effect of smoothing prices – meaning that it eliminates both peaks and valleys in prices. A plaintiff would seem to have the strongest case if the mean price increased because of the information exchange – although even a slight increase in the mean price might not be adverse to the interests of buyers if it made prices more transparent and predictable for buyers. A closer case would arise if mean prices remained at or below previous prices because of the effects of the information exchanged, but a buyer argued that the level prices were higher than the ones he himself would have paid in the but-for world, perhaps because of the timing or unique nature of his purchases. In such a case, a court would have to decide whether an information exchange producing a market-wide mean price at or below previous levels may still be unlawful if it leads to net price increases for some subset of buyers in the market. Questions might also arise concerning the scope of procompetitive justifications allowed to justify an information exchange. Early on in the COVID-19 pandemic, the US Justice Department blessed a proposed arrangement by a group of pharmaceutical companies to exchange information regarding ‘manufacturing facilities, raw materials, and supplies that could be used to produce COVID-19 mAb treatments, specifically global capacity that has been reserved internally or through third parties for the potential production of COVID-19 mAb treatments’.58 The Division noted that [g]iven the immediate need for effective therapeutics, and the likely demand, the Proposed Conduct, which is seeking to use production capacity efficiently and enable the rapid, large-scale production of mAbs, could offer Americans considerable benefits and provide [them] with products or services that might not be available otherwise.59

The Division thus recognized that competitive information exchange may be a catalyst for rapid innovation, particularly if it is subject to competitive safeguards of the types that the parties proposed.

56 Department of Justice, ‘Justice Department Requires Six Broadcast Television Companies to Terminate and Refrain from Unlawful Sharing of Competitively Sensitive Information’ (13 November 2018) www​.justice​.gov/​opa/​pr/​justice​-department​-requires​-six​-broadcast​-television​-companies​ -terminate​-and​-refrain​-unlawful [accessed 3 March 2022]. 57 ibid. 58 Justice Department Business Review Letter, Re: Eli Lilly and Company, AbCellera Biologics, Amgen, AstraZeneca, Genentech, and GSK Expedited Business Review Request Pursuant to COVID-19 Expedited Procedure (23 July 2020) www​.justice​.gov/​atr/​page/​file/​1297161/​download [accessed 3 March 2022]. 59 ibid.

Cartels and the exchange of information  233

V. CONCLUSION Information exchange among competitors can be procompetitive, but it can also soften competition or facilitate price fixing. For the purposes of this volume, which is concerned with cartel behaviour, perhaps the most important summative point is that information exchange can give rise to cartel liability, but it can also give rise to liability even outside of cartel behaviour. At the same time, properly structured information exchange is unlikely to cause antitrust problems; therefore careful attention to agency guidelines and case law – which provide fairly clear guidance on this issue – is advisable. Further, this chapter has shown that judicial and agency attitudes toward information exchange are contingent on underlying attitudes about whether market economies should be characterized more by robust atomistic competition or rather by collaboration among rival firms. As those underlying attitudes shift, so does the law on information exchange.

14. Buyer cartels Peter C. Carstensen

I. INTRODUCTION This chapter examines the law and policy governing horizontal agreements among buyers.1 While buyer cartels should be absolutely illegal unless expressly authorized and regulated, some scholars and a few courts have argued that even pure buyer cartels should be lawful in some instances.2 On the other hand, reasonable buying groups that pool the needs of buyers can make real contributions to economic efficiency. Buyer cartels and buying groups have received increasing notice in the past 15 years, with growing awareness of buyer power’s potential for harmful competitive effects.3 The most recent area of interest is labour markets, where buyer power appears to be much more pervasive than conventional labour market theory had postulated.4 The central thesis of this chapter is that buyer cartels pose serious risks to the market process and have no persuasive justification. Hence, unregulated buyer cartels should be unlawful under all circumstances, despite the potential negative effects of seller power on buyers. At the same time, buying groups have significant utility even when they engage in minimal joint activities. But those same groups can result in significant cartelistic competitive harm if they have sufficient power in the buying market. Overall, there remains significant under-appreciation of the prevalence and the impact of buyer cartels and buying groups on the competitive process. 1 This chapter draws substantially on my earlier work on this topic: P. Carstensen, Competition Policy and Buyer Power: A Global Issue (Edward Elgar Publishing, 2017) Chapter 7 (hereafter Competition Policy); and P. Carstensen, ‘Buyer Cartels versus Buying Groups: Legal Distinctions, Competitive Realities, and Antitrust Policy’ (2010) 1 William & Mary Bus. L. Rev. 1. 2 See, e.g., Balmoral Cinema, Inc. v Allied Artists Pictures Corp. 885 F.2d 313 (6th Cir. 1989) 316–17; A. Devlin, ‘Questioning the Per Se Standard in Cases of Concerted Monopsony’ (2007) 3 Hastings Bus. L.J. 223, 241–43 (arguing that reduced costs resulting from monopsony should be considered in assessing the reasonableness of such collusion); similarly, some scholars have argued for the legality of both buyer and seller cartels: see Christopher Leslie, ‘Achieving Efficiency through Collusion: A Market Failure Defense to Horizontal Price-Fixing’ (1993) 81 Cal. L. Rev. 243; and R. Blair and D.D. Sokol, ‘The Rule of Reason and the Goals of Antitrust: An Economic Approach’ (2012) 78 Antitrust L.J. 471, 487–90. 3 In addition to my own work (see n 1), other notable examples include: R. Blair and J. Harrison, Monopsony in Law and Economics (Cambridge University Press, 2012); R. Blair and J. Harrison, Monopsony: Antitrust Law and Economics (Princeton University Press, 1993); R. Noll, ‘“Buyer Power” and Economic Policy’ (2005) 72 Antitrust L.J. 589; J. Kirkwood, ‘Powerful Buyers and Merger Enforcement’ (2012) 92 Boston U. L. Rev. 1485 and; C. Scott Hemphill and N. Ross, ‘Mergers that Harm Sellers’ (2018) 127 Yale L. Rev. 2078. 4 See, e.g., A. Manning, ‘Monopsony in Labor Markets’ (2021) 74 ILR Rev. 3; T. Angerhofer and R. Blair, ‘Collusion in the Labor Market: Intended and Unintended Consequences’, CPI, 12 June 2020; S. Naidu, E. Posner and E. Weyl, ‘Antitrust Remedies for Labor Market Power’ (2018) 132 Harv. L. Rev. 537.

234

Buyer cartels  235 As an initial matter, it may seem difficult to distinguish between the bidders at an auction who allocate purchases as well as deciding what they will pay for items from a group of hospitals that pool their purchases and seek lower prices in return for the volume they offer.5 In the United States, the auction-buying ring (which is a cartel) is illegal per se and its members are committing a felony,6 while the hospital buying group is subject to a rule of reason analysis, with a strong presumption of legality.7 What differentiates these situations is the functional characterization of the buyer-entity. But a nominally lawful buying group can become a cartel akin to the auction conspiracy.8 Conversely, a group of bidders at an auction might lawfully collaborate to make a better (that is, higher) bid for an item, often a bundle of components, than any one individual bidder could offer.9 Because there is a continuum of functions from a buying group or a cartel, effective public policy must identify the factors that justify alternative characterization of the entity. Because buyer power can arise from much smaller market shares than are usually associated with seller power, there are important policy implications for both the identification of buyer cartels and the assessment of the competitive risks that legitimate buying groups can present. Moreover, buyer cartels can include more participants and face lower risks of defection because of the strong incentives to remain. Such agreements can be more informal, even tacit, for the same reason. Thus, buyer cartel issues require a fuller recognition of the incentives that motivate and facilitate such conduct, as well as the need to modify conventional criteria for inferring the existence of a cartel in light of these economic incentives. Finally, although legitimate buying groups can efficiently respond to the needs of their participants, they can pose real threats to the long-term competitiveness of both the supply and demand sides of the market. The prevailing assumption that even very large buying groups are competitively benign ignores the differences between buyer and seller power that result in competitive risks from buyer power at levels that arguably do not create serious antitrust concerns on the selling side.10 Competition policy needs to recognize and articulate, much more clearly than it does at present, the competitive risks that powerful buying groups can pose and, in light of those risks, adopt better standards for judging the reasonableness of such buying groups. Section II focuses on the differences between buyer cartels and buying groups. It also explains key differences between seller cartels and buyer cartels. Section III sets forth conflicting contemporary legal standards governing both types of activity and evaluates the policy arguments for and against legalizing some buyer cartels. Finally, it evaluates the competitive 5 See B. Meier and W. Walsh, ‘Senate Panel Criticizes Hospital Buying Groups’, New York Times, 1 May 2002, C1. 6 See, e.g., United States v Taubman, 297 F.3d 161 (2d Cir. 2002). 7 See, e.g., Nw. Wholesale Stationers v Pac. Stationery & Printing, 472 US 284 (1985) 297–98; see also J. Jacobson and G. Dorman, ‘Joint Purchasing, Monopsony and Antitrust’ (1991) 36 Antitrust Bull. 1, 4 (arguing that joint purchasing should be treated more leniently than joint ventures among sellers). 8 See In re Brand Name Prescription Drugs Antitrust Litig., 123 F.3d 599 (7th Cir. 1997) 606. 9 When items are sold in a bundle, there is a plausible basis for a joint buying agreement in order to facilitate the most efficient use of the goods by combining the highest value a set of buyers would give for each component of the bundle. See G. Stigler, ‘A Note on Block Booking’ [1963] Sup. Ct. Rev. 152 (movie bundling explained in terms of maximizing value to distributor). Such buying agreements can, however, create antitrust issues. See, e.g., United States v Seminole Fertilizer Corp., No. 97-1507-CIV-T-17E, 1997 WL 692953 (MD Florida, 19 September 1997). 10 See, e.g., Jacobson and Dorman (n 7).

236  Research handbook on cartels benefits and risks associated with legitimate buyer groups to argue that stricter scrutiny of such entities is necessary. The central policy conclusions that emerge from this analysis are set forth in Part IV. With respect to buyer cartels, first, there is good reason to believe that they are more prevalent than current enforcement actions would imply. Second, while it is difficult sometimes to distinguish cartels from legitimate buying groups, cartels not expressly authorized by public authority should remain per se illegal. Third, buyer cartels can involve larger and less well-organized groups than standard cartel theory postulates. Fourth, because buyer power raises from small market shares, cartels can also be less inclusive and more narrowly focused than conventional cartel theory would suggest. Finally, with respect to legitimate buying groups, such entities can create significant competitive risks. Hence, those likely to cause such harms merit stricter scrutiny.

II.

BUYER CARTELS AND BUYING GROUPS

Distinguishing between a buyer cartel and a buying group is easy at a conceptual level.11 But in application, the two activities exist on a continuum ranging from a pure cartel to a buying group that totally integrates the purchase of all inputs. The central problems in characterization are that form and function are not congruent and that buyers do not necessarily differentiate the bases on which they obtain lower input prices. Hence, a group that is functionally a buying group can appear to be little more than a cartel, but it is equally possible to ‘dress up’ a cartel to look like a buying group. Buyer cartels have different characteristics than conventional seller cartels that need to be recognized. A.

Buyer Cartels

A cartel is a group of competitors who have agreed to limit or eliminate their competition in some economically relevant dimension.12 The objective of such a combination is to create, allocate and exploit power in the market. While conventional monopsony theory posits primarily exploitative explanations, that is, collusive or unilateral conduct to reduce input prices, buyer cartels can also: (1) allocate customers or suppliers; (2) agree on the composition of the output or input ratios; or (3) share competitively significant information.13 Moreover, buyer cartels can use their collective power to exclude or restrain their competitors in the markets in which they buy goods or otherwise regulate the nonprice dimensions of the supply market as a means of entrenching their own dominance in both upstream and downstream markets.14 A buyer cartel focuses on eliminating competition for input purchases to reduce prices or otherwise control supplier conduct. Buyers may collude to drive down input prices even 11 See T. Piraino, ‘A Proposed Antitrust Approach to Buyers’ Competitive Conduct’ (2005) 56 Hastings L.J. 1121, 1131–32. 12 See N. Rosenfelt, ‘The Verdict on Monopsony’ (2008) 20 Loy. Consumer L. Rev. 402, 405–06. 13 See R. Lande and H. Marvel, ‘The Three Types of Collusion: Fixing Prices, Rivals, and Rules’ [2000] Wis. L. Rev. 941, 951–53. 14 See, e.g., Klor’s, Inc. v Broadway-Hale Stores, Inc., 359 US 207 (1959) 212–13; Montague v Lowry, 193 US 38 (1904) 45; Toys ‘R’ Us, Inc. v Fed. Trade Comm’n, 221 F.3d 928 (7th Cir. 2000) 936–38.

Buyer cartels  237 though they sell in highly competitive markets where cartelization is unlikely.15 Such buyer collusion may eventually reduce the total output in the market.16 A significant reduction in output can affect the downstream market by reducing the volume of the output produced, resulting in higher prices for the remaining production.17 Examples of buyer cartels abound. One of the earliest was Montague v Lowry, involving a challenge to a buyer cartel of retailers of mantels and fireplace tiles.18 The cartel insisted that manufacturers deal only with its members – thereby boycotting the cartel’s competitors – and adopt and enforce resale price maintenance. Having a supplier enforce the cartel’s commands is often a more effective way for retailers to police their cartel than trying to do so directly.19 Another early case involved use of buying power to compel lumber companies to refuse to deal with integrated wholesaler-retailers.20 As in Montague, the cartel used its buyer power to eliminate potentially more efficient competition. Other buyer cartels have focused on labour or agricultural inputs. Several cases have challenged the practices of hospitals and other employers of nurses, alleging that employers engaged in collective wage-setting outside a union contract.21 In another case, a court of appeals decision upheld a complaint challenging as a buyer cartel a scheme to coordinate job classifications and consequent wage rates for various types of professional workers in the energy industry.22 Similar cartels have been documented in agriculture with respect to blueberries in Maine23 and tobacco in the Southeast.24 Another classic case involved an agreement to set the percentage of durum wheat in macaroni and spaghetti at a time when durum wheat was in short supply.25 Finally there are auction cartels, such as those involving postage stamps.26 Use of cartel power varies depending on the goals and interests of the participants, but also, importantly, on the nature of the supply market. If supply is relatively price inelastic, then 15 A good example is the timber buyer’s cartel in Alaska that drove down the price of trees in the regions where those companies operated, but which had no effect on the overall price of the lumber produced from those trees because that lumber competed in a much broader geographic market; see Reid Bros. Logging Co. v Ketchikan Pulp Co., 699 F.2d 1292 (9th Cir. 1983) 1303. 16 The likelihood of this effect and its substantiality are functions of the elasticity of supply; see Carstensen (2017) (n 1) Chapter 3. 17 This effect is discussed in R. Blair and J. Harrison (1993) (n 3) 36–42. See also R. Sexton and M. Zhang, ‘An Assessment of the Impact of Food Industry Market Power on US Consumers’ (2001) 17 Agribusiness 59; C. Doyle and M. Han, ‘Expropriating Monopoly Rents through Stable Buyer Groups’ (Amsterdam Center for Law and Economics, Working Paper No. 2009-03); see also M. Han, ‘How Buyer Groups Can Effectively Operate as Stable Cartels’ (2009) 62 Aenorm 14. 18 Montague v Lowry, 193 US 38 (1904). 19 See J. Palamountain, The Politics of Distribution (Harvard University Press, 1955) 99–100; W. Bowman, ‘The Prerequisites and Effects of Resale Price Maintenance’ (1955) 22 U. Chi. L. Rev. 825, 834–35; L. Telser, ‘Why Should Manufacturers Want Fair Trade?’ (1960) 3. J.L. & Econ. 86; see also Doyle and Han (n 17). 20 E. States Retail Lumber Dealers’ Ass’n v United States, 234 US 600 (1914). 21 See, e.g., Fleischman v Albany Med. Ctr., 728 F. Supp. 2d 130 (NDNY 2010). 22 Todd v Exxon Corp., 275 F.3d 191 (2d Cir. 2001) 191. 23 See ‘Judge OK’s Award in Blueberry Lawsuit’, Boston Globe, 4 January 2004, www​.boston​.com/​ news/​local/​articles/​2004/​01/​04/​judge​_oks​_award​_in​_blueberry​_lawsuit/​ [accessed 25 February 2022]. 24 DeLoach v Lorillard Tobacco Co., 391 F.3d 551 (4th Cir. 2004) 554. 25 Nat’l Macaroni Mfrs. Ass’n v Fed. Trade Comm’n, 45 F.2d 421 (7th Cir. 1965) 426–27. 26 See J. Asker, ‘A Study of the Internal Organization of a Bidding Cartel’ (2010) 100 Am. Econ. Rev. 724; and R. Marshall and M. Meurer, ‘Bidder Collusion and Antitrust Law: Refining the Analysis of Price Fixing to Account for the Special Features of Auction Markets’ (2004) 72 Antitrust L.J. 83.

238  Research handbook on cartels colluding buyers have a strong incentive to drive down price because it will not significantly reduce the supply of the input. In other cases, supply may be more price elastic, and reducing price would result in an inadequate supply,27 but buyers can still use their collective power to compel their suppliers to discriminate in price or refuse to deal with new entrants or marginal buyers.28 Thus, supply elasticity will affect the methods used by a cartel but may not alter its goals.29 B.

Legitimate Buying Groups

A buying group is a set of potentially competing buyers that pool their purchase orders and, jointly or through an agent, negotiate for the inputs they seek even if the products obtained will be shipped directly to each buyer and billed separately.30 The fundamental distinction between a legitimate buying group and a cartel is that a buying group acts to gain the efficiencies of a joint enterprise. The buyer-participants have integrated some or all of their input acquisition function by creating or participating in the buying group. Efficiencies (passed through as lower prices) can result from longer production runs, reductions in transaction costs or lower costs per unit of quality control, and can include protection against defective or dangerous products, preferred status with shipping services based on high volume and improved ability to develop new products.31 A buying group can reduce transaction costs by acting as a single buyer for a given input. A centralized buying system must have a coordinated way to distribute the inputs among participants as well as assure payment for them. The scale of purchases and the range or variety of inputs buyers need will determine whether the participants find a buying group to be useful. A buying cooperative can also bargain for price. In the abstract model of competitive prices, all prices arise from a market process that results in a single price known to all buyers and sellers. But in a real-world, workably competitive market, sellers almost always have some latitude in pricing. In particular, a seller may find it attractive to lower prices slightly to gain a large order that will ensure more efficient volume in the production facility. A buyer seeking a large volume may make a more extensive search of the market for suppliers and generate a more active bidding process as a result. Therefore, in a market where competition is imperfect but workable, buyers can gain a price advantage by employing more sophisticated and effective searches for the inputs they need. At the same time, sellers can profit from making

27 If, however, buyers can obtain additional inputs from another, more competitive market, they may be able to exploit one set of suppliers while still filling their residual supply needs. They will then be somewhat less concerned about supplier reaction. 28 The collective volume of purchases gives a cartel significant leverage over any individual seller because that seller must engage in a costly and time-consuming search for other buyers. See P. Carstensen, ‘Buyer Power, Competition Policy, and Antitrust: The Competitive Effects of Discrimination among Suppliers’ (2008) 53 Antitrust Bull. 271, 281 and 284. 29 Doyle and Han (n 17) posit a buyer cartel that requires suppliers to raise their prices, thus causing some reduction in sales but generating significant profits for the sellers, which the sellers then rebate to the buyers in the form of slotting fees or other kickbacks. See Discon v NYNEX, 525 US 128 (1998); NicSand v 3M, 507 F.3d 442 (6th Cir. 2007). 30 See T. Piraino, ‘A Proposed Antitrust Approach to High Technology Competition’ (2002) 44 Wm. & Mary L. Rev. 65, 92–93; and Piraino (n 11) 1151 n 139. 31 See Piraino (n 30) 835–36.

Buyer cartels  239 larger sales at lower prices when the transaction reduces selling costs or when the resulting volume takes advantage of scale economies. The economic logic is that some functions are subject to significant economies of scale and scope that will generate gains for the participants. But other aspects of the same business may be subject to diseconomies of scale. If those functions are separated, the resulting partially integrated enterprises are more efficient at both levels than a fully integrated enterprise of the same size. The evidence from the American grocery market is that in the 1950s and 1960s, at least, national chains were less efficient than regional chains, in part because the regional chains used buying groups to mimic the buying efficiency of the national chains but avoided the inefficiencies of the national chains’ centralized retail management systems.32 Buying groups occur in a number of retail activities. Two prominent American examples are Topco, which provides a joint buying programme and house brands for its members, which are midsized grocery store chains,33 and Northwest Wholesale Stationers, which provides buying, warehousing and delivery services to retail stationery stores in the Pacific Northwest.34 In the European Union (EU), Euronics is a buying group for independent electrical appliance retailers, representing 11,000 stores.35 The ten largest European grocery buying groups had between 30 and 100 billion euros in annual turnover as of 2011.36 Some buying groups originate with an entrepreneurial actor that organizes the group and acts as its agent, sometimes using a franchise model.37 The participants in the downstream market benefit from the efforts of an upstream coordinator, a third party with more efficient skills in the provision of inputs. There are no fixed parameters for a buying group. A legitimate buying group involves some integration of activities, but the integration is limited to facilitating the functioning of the particular enterprise.38 One way to distinguish a buying group from a buyer cartel is to focus on its functional goals. If the participants make an investment in, and consolidate, coordinate and administer some aspects of, their buying activity, then they are a prima facie buying group. Conversely, if the group exists only to agree on how the parties will conduct their own purchases, it is prima facie a cartel.39

P. Carstensen and H. First, ‘Rambling through Economic Theory: Topco’s Closer Look’, in E. Fox and D. Crane (eds), Antitrust Stories (Foundation Press, 2007) 178­–79. 33 See United States v Topco Assocs. Inc., 405 US 596 (1972) 598; see also Carstensen and First (n 32). 34 See Nw. Wholesale Stationers, Inc. v Pac. Stationery & Printing Co., 472 US 284 (1985) 286–87. See also E. Brunet and D. Sweeney, ‘Integrating Antitrust Procedure and Substance after Northwest Wholesale Stationers: Evolving Approaches to Pleadings, Burden of Proof, and Boycotts’ (1986) 72 Va. L. Rev. 1015. 35 See www​.euronics​.com/​[accessed 25 February 2022]. 36 Institute of Grocery Distribution, ‘Grocery Buying Groups’ (4 February 2016) www​.igd​.com/​ articles/​article​-viewer/​t/​grocery​-buying​-groups/​i/​15517 [accessed 25 February 2022]. 37 In the United States, examples include the IGA and Piggly Wiggly grocery chains. See www​.iga​ .com/​about [accessed 25 February 2022]; and www​.pigglywiggly​.com/​about​-us [accessed 25 February 2022]. 38 See P. Dobson, ‘Exploiting Buyer Power: Lessons from the British Grocery Trade’ (2005) 72 Antitrust L.J. 529, 543–44. 39 See, e.g., Mandeville Island Farms, Inc. v Am. Crystal Sugar Co., 334 US 219 (1948) 223–24. 32

240  Research handbook on cartels C.

Distinguishing Buying Groups from Cartels

Because of the flexibility inherent in the organization of legitimate buying groups, there can be significant difficulty in distinguishing such groups from cartels. The hallmark of a buyer cartel is that the buyers have coordinated their individual buying only to exercise power over sellers. The key characteristic is that the buyers have not integrated their buying activity to any degree. In contrast, a buying group will have integrated some elements of its input buying process. This is the crucial distinction. Once a buying group has been organized, it may be able to impose other, cartelistic, restraints on its suppliers, such as requiring them to refuse to deal with nonmembers or requiring more favourable terms than nongroup buyers would enjoy.40 In addition, the group may coordinate downstream competition through various restraints, including resale price controls or territorial allocation.41 There are plausible theoretical arguments in support of the claim that many of these restraints are necessary to facilitate the legitimate interests of the group. Therefore, it is essential to evaluate whether the restraint serves to address a risk of opportunistic behaviour by participants or otherwise coordinates their internal relations. Buying groups are also more likely to be legitimate when input production is price elastic, both because economies of scale are likely and because attempts to depress input prices artificially will be met with reduced output levels. In these situations, a buying group’s volume purchase creates higher net profits (and increased output) for producers while at the same time holding the total cost (price plus transactional expenses) down for buyers. Again, a focus on the specific activities and functions of the buying group is important because, even when supply markets are elastic, a group with a sufficient share of the buying side might well impose its will on producers with respect to nonprice aspects of competition. D.

Differences between Buyer Cartels and Seller Cartels

There are four ways in which buyer cartels present different characteristics from conventional seller cartel models. The differences make buyer cartels more likely and harder to detect. First, buyer power arises with much lower market shares than is generally assumed to be the case on the selling side of the market.42 Thus, the EU has found that a buyer taking as little as 20 per cent of a grocery input has the capacity to distort competition.43 In the UK, power was found in groceries where the buyer took as little as 10 per cent of the branded goods.44 Toys “R” Us had only about a 20 per cent share of the toy retail market but was able to coerce its

‘Most favoured nation’ or even ‘most favoured nation plus’ contracts provide a frequent means to achieve this goal. Cf. United States v Blue Cross Blue Shield of Michigan, 2:10-cv-14155-DPH-MKM (ED Michigan) www​.justice​.gov/​atr/​case/​us​-and​-state​-michigan​-v​-blue​-cross​-blue​-shield​-michigan [accessed 25 February 2022]. 41 See, e.g., United States v Topco Assocs. Inc., 405 US 596 (1972) 601–04. 42 This analysis is elaborated in Carstensen (2017) (n 1) 65–75. 43 See European Commission, Guidelines on the Applicability of Article 101 of the Treaty on the Functioning of the European Union to Horizontal Co-operation Agreements [2011] OJ C11/1 para 204; see also Case No IV/M.1221, Rewe/Meinl, Commission decision, 3 February 1999, C(1999) 228 final, para 99. 44 See Dobson (n 38). 40

Buyer cartels  241 major suppliers into refusing to deal with third parties.45 Because the buyer decides whether to buy and from whom to buy, this creates significant leverage, especially when suppliers face significant costs in seeking alternative outlets. Second, because buyers share an interest in keeping input costs down, the risks of defection from a cartel agreement are low. The defector must bid up the price and process the input into a larger volume of goods that will expand output and so is likely to lower sales prices even as its costs increase. In contrast, in a seller cartel, defection can yield an immediate gain to the defector by increasing volume even if the price per unit is lower. Thus, buyer cartels are generally more durable and also require less explicit organization and monitoring. Third, taken together, the lower market share necessary for power and the greater durability of buyer cartels leads to such organizations being made more varied. One result is that they can be more inclusive. Hence even in markets with a substantial number of competing buyers, cartelistic agreements are possible because the mutual interest of the participants gives them a strong incentive to cooperate.46 But for the same reasons, it is also possible to have cartels that have limited participation from competitors, and which can focus on limited objectives. The no-poach agreements in employment are good examples of this. The participants agree not to compete for each other’s employees but remain free to compete for new entrants and for employees of non-participants.47 Fourth, again because of the nature of buyer power, tacit collusion is very significant risk in buyer markets. If buyers respect each other’s suppliers and so do not compete head-to-head for their services, each has somewhat more freedom to impose burdens on its suppliers because those suppliers will have few, if any, alternative outlets for their products or services. Taken together, these four factors make buyer cartels more likely but also harder to identify and more challenging to remedy. Indeed, where the restraints are established, tacit understandings can maintain them more easily on the buying side than on the selling side.

III.

THE LEGAL TREATMENT AND COMPETITION POLICY ANALYSIS OF BUYER CARTELS AND BUYING GROUPS

A.

Legal Treatment of Buyer Cartels and Buying Groups

Many competition laws expressly prohibit cartel agreements among buyers, and the oldest known cartel case involved a grain-buying conspiracy in ancient Athens.48 It appears that the conspirators were put to death – a stricter penalty than is currently available even among countries that use criminal sanctions.49 In the United States it is unlawful, as well as a basis for treble damage liability, for buyers to agree (1) on what they individually will pay for goods or services, (2) that they will not bid against each other for particular items at an auction or

See Toys ‘R’ Us v FTC, 221 F.3d 928 (7th Cir. 2000). Todd v Exxon, 275 F.3d 191 (2nd Cir. 2001). 47 See United States v Adobe Sys. Inc., No. 1:10-cv-01629 (DDC, Final Judgment, 18 March 2011) www​.justice​.gov/​atr/​case/​us​-v​-adobe​-systems​-inc​-et​-al [accessed 25 February 2022]. 48 See L. Kotsiris, ‘An Antitrust Case in Ancient Greek Law’ (1988) 22 Int’l Law 451. 49 ibid 457. 45 46

242  Research handbook on cartels (3) that they will restrict wage or employment competition.50 Thailand’s competition law expressly prohibits horizontal restraints on buying,51 although the law appears generally to be unenforced.52 South Africa similarly forbids buyer-side restraints.53 Other examples include cases from Canada.54 The EU has held such conduct to be absolutely illegal.55 EU state competition authorities have also been aggressive in challenging it.56 This is consistent with the express terms of Article 101(1) of the Treaty on the Functioning of the European Union (‘TFEU’).57 Indeed, globally, there is general recognition that buyer cartels are anticompetitive and should be prohibited.58 There is a tension between the policy goal of consumer welfare and challenges to buyer cartels that present no evidence of direct harm to consumers.59 European and American case law reflects a consistent policy that such cartels are unlawful. This is more consistent with the goal of facilitating the competitive process than with the narrow focus on consumer welfare only.60 At the same time, American enforcement agencies, the EU and other competition authorities rarely object to the creation of ‘buying groups’.61 The United States apparently presumes that such groups are lawful if they buy no more than 35 per cent of the total volume of purchases 50 See, e.g., United States v Crescent Amusement Co., 323 US 173 (1944); Swift & Co. v United States, 196 US 375 (1905); Knevelbaard Dairies v Kraft Foods, Inc., 232 F.3d 979 (9th Cir. 2000); United States v Romer, 148 F.3d 359 (4th Cir. 1998); Reid Bros. Logging Co. v Ketchikan Pulp Co., 699 F.2d 1292 (9th Cir. 1983); and Asker (n 26). For cases dealing with employment issues, see, e.g., Todd v Exxon Corp., 275 F.3d 191 (2d Cir. 2001). See also United States v Adobe Sys. Inc., No. 1:10-cv-01629 (DDC, Final Judgment, 18 March 2011) www​.justice​.gov/​atr/​case/​us​-v​-adobe​-systems​ -inc​-et​-al [accessed 25 February 2022]. 51 Trade Competition Act B.E. 2542 §§ 25(1) and 27(2) (1999) (Thailand). 52 See C. Poomipark, M. Intaranont and S. Thanapase, ‘Thailand to Overhaul Trade Competition Law’, Mondaq, 25 February 2010, www​.martindale​.com/​matter/​asr​-929360​.Thailand​.pdf [accessed 25 February 2022]. 53 Competition Act 98 of 1999, as amended, § 4(1)(b)(i) (S. Afr.) (an agreement among firms is prohibited ‘if it involves any of the following restrictive horizontal practices: (i) directly or indirectly fixing a purchase or selling price or any other trading condition’) (emphasis added). 54 See, e.g., R. v Abitibi Power & Paper Co., 36 CPR 1888 (Que. QB); and 321665 Alberta Ltd. v Mobil Oil Can. Ltd., [2011] ABQB 292 (Canada). 55 See, e.g., Case C-209/07, Competition Authority v. Beef Industry Dev. Soc’y Ltd. [2008] ECR 1-8637; Case COMP/C.38.238/B.2, Raw Tobacco Spain, Commission decision, 20 October 2004, C(2004) 4030. 56 See European Competition Network, ECN Activities in the Food Sector (2012) http://​ec​.europa​.eu/​ competition/​ecn/​food​_report​_en​.pdf [accessed 25 February 2022]. 57 ‘The following shall be prohibited […] all agreements [that] directly or indirectly fix purchase […] prices or any other trading conditions’: Article 101 TFEU. 58 In 2020, India proposed to add an explicit condemnation of buyer cartels. See P. Barik, ‘India: Buyer’s Cartels Have Now Taken The Driver’s Seat In India’, Mondaq, 20 April 2020, www​ .mondaq​.com/​india/​cartels​-monopolies/​919282/​buyer39s​-cartels​-have​-now​-taken​-the​-driver39s​-seat​-in​ -india [accessed 25 February 2022]. 59 See text at notes 66–69 infra for a further analysis of this issue. 60 V. Daskalova, ‘Consumer Welfare in EU Competition Law: What Is It (Not) About?’ (TILEC Discussion Paper 2015-011, 1 May 2015) https://​papers​.ssrn​.com/​sol3/​papers​.cfm​?abstract​_id​=​2605777 [accessed 25 February 2022] 20–22. 61 See, e.g., US Department of Justice and Federal Trade Commission, Statements of Antitrust Enforcement Policy in Health Care (August 1996) www​.justice​.gov/​atr/​page/​file/​1197731/​download [accessed 25 February 2022] 54–55.

Buyer cartels  243 in a particular market and that product is no more than 20 per cent of the resulting revenue.62 Some EU decisions suggest that adverse competitive effects will arise only if a buyer takes more than 22 per cent of the volume in a market, but its guidelines require review if a buying group takes more than 15 per cent of any commodity.63 Thus, there is recognition that buying groups can have adverse competitive impact, but the market share necessary to justify serious inquiry remains uncertain. B.

Buyer Cartels and the Per Se Illegal Standard

Conventional analysis says that buyer cartels cause economic harm similar to that caused by seller cartels.64 But as with seller cartels, some scholars defend buyer cartels as enhancing efficiency in some circumstances.65 The credibility of these arguments depends in large part on the goals of competition policy against which those arguments are measured. Although the conventional explanation of why buyer cartels are undesirable has serious limitations, as discussed below, there remain serious and substantial reasons to object to such cartels and the justifications appear to overestimate substantially the potential gains to the competitive process. 1. Theories of competitive harm The conventional economic welfare argument against buyer cartels rests on the assumptions that buyers purchase in discrete units and resell in a market in which they are the only sellers and that producers face increasing unit costs as volume increases. Under these assumptions, when the price paid for inputs is reduced, output of that commodity declines. As a result, the static comparison of a world with and without a buyer cartel shows that the cartel causes a reduction in production and a consequent increase in prices to consumers. The conclusion, therefore, is that buyer cartels harm consumers as well as aggregate welfare.66 The force of this economic theory is contingent on the validity of its assumptions. The argument loses force if the cartel uses all-or-nothing contracts. Such contracts compel producers to deliver approximately the competitive output but at a lower price.67 Further, if the cartel members compete with many other producers who have different sources of input, they would have no incentive or capacity to raise the prices of their output.68 In such a situation, the cartel’s output will be lower than it would have been, but overall prices will not be noticeably affected because volume in the downstream market will not change appreciably. In either case, there

See ibid 54–55; see also M. Botti, ‘Observations on and from the Antitrust Division’s Buyer-Side Cases: How Can “Lower” Prices Violate the Antitrust Laws?’, paper presented at the 2007 meeting of the ABA Antitrust Section meeting, 19 April 2007, www​.abanet​.org /antitrust/at-committees/at-hcic/ pdt7program-papers/Botti-Paper.pdf [accessed 25 February 2022]. 63 See European Commission (n 43) para 204; see also Case No IV/M.1221, Rewe/Meinl, Commission decision, 3 February 1999, C(1999) 228 final, para 99. 64 On the economic harms from seller cartels, see Chapter 2 in this volume. 65 See Devlin (n 2); Leslie (n 2); and Blair and Sokol (n 2). 66 See, e.g., H. Hovenkamp, ‘Is Antitrust’s Consumer Welfare Principle Imperiled?’ (2020) 45 J. Corp. L. 101, 115–17. 67 Carstensen (2017) (n 1) 42–46 and 80–81. 68 See, e.g., Reid Bros. Logging Co. v Ketchikou Pulp Co., 699 F.2d 1292 (9th Cir. 1983); see also Mandeville Island Farms v American Crystal Sugar, 334 US 219 (1948). 62

244  Research handbook on cartels would be no adverse effect on the static price-output parameters on which the conventional economic models are based. Thus, a focus on static measures of consumer welfare as the goal of competition law weakens the argument that buyer cartels are necessarily undesirable. But if the primary goal of competition law is the protection and advancement of economic competition as a process, two related policy arguments support a broader condemnation of buyer cartels. First, buyer cartels, like seller cartels, directly distort the market process through private agreement.69 By preempting the market, any cartel undermines the fundamental goal of ensuring the goods and services clear markets at competitively determined prices. Second, a buyer cartel distorts incentives and causes a misallocation of economic rewards over time, undermining the core goals of a competitive market system. Competition policy should focus on the dynamic interest of providing incentives to invest and develop new, innovative solutions to problems. Buyer cartels, however, diminish the rewards to a producer below the level that a competitive market would have provided,70 which sends the wrong signal to investors and innovators. Indeed, the strategic responses to cartels, such as vertical integration, although rational, can, in dynamic terms, result in further distortions of the market process away from better structural options.71 2. The argument for legalizing some buyer cartels The primary argument for buyer cartels rests on a notion of countervailing power.72 Small buyers facing a monopoly or oligopoly-seller market are individually powerless to bargain for lower prices;73 they are compelled to accept the monopoly or oligopoly price. However, if these buyers together can make a credible threat that they will withhold their purchases unless they receive lower prices, they might succeed.74 Independent pharmacies sought an antitrust exemption that would enable them to bargain collectively over prices with oligopolistic wholesalers that both engaged in price discrimination and imposed unreasonably low reimbursement prices on independent retailers.75 The pharmacies’ belief was that if they were allowed to band together, they could obtain lower prices and higher reimbursement rates for the prescription drugs they resell.76 This analysis is based on a competition policy goal of static consumer welfare. The theory is that the successful buyer cartel will induce the seller to reduce prices and increase output,

See, e.g., Fed. Trade Comm’n v Super. Ct. Trial Lawyers Ass’n, 493 US 411 (1990); and Fashion Originators Guild of Am. Inc. v Fed. Trade Comm’n, 312 US 457 (1941). 70 See Blair and Harrison (1993) (n 3) 156–63. 71 See Omega Envtl. Inc. v Gibarco, Inc., 127 F.3d 1157 (9th Cir. 1997). 72 J.K. Galbraith, American Capitalism: The Concept of Countervailing Power (Houghton Mifflin, 1952) 109–12; see also Blair and Sokol (n 2) 487–90. 73 See generally T. Campbell, ‘Bilateral Monopoly in Mergers’ (2007) 74 Antitrust L.J. 521 (2007). This article produced some strong disagreement; see J. Baker, J. Farrell and C. Shapiro, ‘Merger to Monopoly to Serve a Single Buyer: Comment’ (2008) 75 Antitrust L.J. 637, 637–46. 74 Baker, Farrell and Shapiro (n 72) 638–46. 75 See Impact of Our Antitrust Laws on Community Pharmacies and Their Patients: Hearing before the Task Force on Antitrust and Competition Policy of the H. Comm. on the Judiciary, House Hearing, 110th Congress, 18 April 2007; and ibid 82–83 (statement of David A. Balto, representing the pharmacies). 76 ibid 11 (statement of Rep. Weiner). 69

Buyer cartels  245 moving the market towards competitive price and output levels.77 The model assumes that the buyer cartel has sufficient power to do so and also passes on its gains to consumers. But this outcome is contingent on the relative options of each side, as well as on the bargaining skill and sophistication of the parties. Powerful sellers can disrupt the group by offering some participants secret discounts to defect. In the short run, the defector would gain a head start in competing with the other members of the group in processing the input and producing saleable output. Under these circumstances, cartels might prove ineffective. The case for buyer cartels is plausible in the short run only if no other reasonable alternative exists.78 Monopoly profits at the producer level ought, over time, to induce other responses, such as creating an entity to produce the input, sponsoring entry, finding substitute inputs or creating a legitimate buying group. Entry or innovation would reduce or avoid the need for the monopolized product. Therefore, unless there is a strong argument that the seller’s power is not likely to dissipate even in the long run, the justification for a buyer cartel is weak.79 However, if a buyer cartel is the least worst option for establishing equitable prices and services, then an appropriate public regulatory body should oversee the cartel process – the interests of private parties will not necessarily be congruent with the public interest. C.

Buying Groups: Efficiency Gains and Competitive Risks

A number of efficiency arguments justify the organization of buying groups, but their competitive risks frequently go unrecognized or are unduly minimized. By marshalling a significant share of the market demand for an input, a buying group can also create a variety of competitive risks in both upstream supply markets and the downstream markets in which the participants compete. With increased power, the buying group can impose onerous conditions on sellers that exploit them, require restrictions that foreclose competitors in the buying market or use control of significant inputs to achieve coordination of competition in the downstream market.80 The most obvious risk is that the buying group will exploit its power to drive prices below a competitive level. The buying group gains economically whether the reason for the lower prices is transactional efficiency, productive efficiency or market power exploitation. A second, related risk, illustrated in a number of cases, is the use of buyer power to exclude competing producers from access to inputs. But some restraints on the freedom of suppliers might be reasonable and even essential to the efficient operation of legitimate buying groups. For example, if a buying group contracts for a differentiated good – especially if it has invested in its development – it should be entitled to retain the rights to that product to avoid the risks

See Devlin (n 2) 241–43; and Blair and Sokol (n 2) 487–90. Donald Baker has argued for limiting the rules that require access to networks, which he believes reduce the incentives for the creation of competing networks; see D. Baker, ‘Compulsory Access to Network Joint Ventures Under the Sherman Act: Rules or Roulette?’ [1993] Utah L. Rev. 999, 1127–28. The same argument applies to other issues in buyer-power contexts. 79 In addition, a buyer cartel can result in coordination on the downstream side of the market if the buyers constitute a substantial part of the resale market. If that occurs, the buyer cartel would morph into a seller cartel whose function would be to raise price and reduce output to the customers of its participants. 80 Carstensen (2017) (n 1) Chapter 4 (examining these competitive risks in some detail). 77 78

246  Research handbook on cartels of free-riding or other strategic conduct by either the producer or competing buyers.81 The factual–functional question is whether the restraint is reasonably necessary for the legitimate needs of the venture.82 The third risk, illustrated by Topco, is the use of the buying group to allocate downstream markets or set resale prices.83 Topco’s members used their annual meeting to allocate and reallocate territory to minimize intragroup competition.84 There was no justification for allocating territory to promote the legitimate activities of Topco, which provided its members with house brand groceries.85 Buying groups might use their power to discourage innovation in the input markets. New inputs might require buyer investment in a new plant or equipment and might alter the barriers to entry. Moreover, innovations might induce more competition among buyers in their downstream markets. Using their buyer power to insist on limits to the kinds of innovation that suppliers engage in would reduce the risk of disruptive innovation. Even if attempted, this use of buyer power might have no effect in many markets. If a small number of buyer groups dominate an input market, they can foreclose competition in the input market by selecting the same supplier. This will foreclose entry and indeed eliminate existing competition. The resulting monopoly price from the supplier will raise downstream prices, but the buyers may be able to use their buyer power to bargain for a substantial share of the profit. An illustrative example occurred in the automotive sandpaper market.86 A supplier with deep pockets made attractive long-term offers to buyers, forcing out the more poorly financed incumbent, which ultimately resulted in the entrant’s capture of nearly 100 per cent of the sales.87 The excluded competitor claimed that the entrant then shared a substantial part of its projected monopoly profits with the buyers, who collectively dominated the downstream retail market.88 This community of interest eliminated the buyers’ incentive to seek competitive supplies.89 The size of a buying group becomes competitively problematic when its purchase volume exceeds the quantity necessary for minimum efficient scale. For example, the EU expressly requires that buying groups be no more restrictive than is necessary to achieve their legitimate objectives.90 But it may be inappropriate to condemn a group that provides significant effi-

81 A group that invested in research and development to create an improved input could rationally insist that the contract producer provide that input only to the group that had made the investment and taken the risks. 82 See Carstensen (2017) (n 1) 126–29; see, e.g., United States v Visa, 344 F.3d 229 (2nd Cir. 2003). 83 See United States v Topco Assocs. Inc., 405 US 596 (1972). 84 ibid 601–03. 85 Carstensen and First (n 32) 182–85. 86 NicSand Inc. v 3M Co., 507 F.3d 442 (6th Cir. 2007). 87 ibid 447–49. 88 ibid. Nevertheless, the court held that there was no antitrust violation on these facts. 89 This conclusion assumes that the downstream market is not subject to easy entry. See J. Klish, ‘Serving Economic Efficiencies or Anticompetitive Purposes? The Future of Group Purchasing Organizations and the Antitrust Safety Zone’ (2005) 2 Ind. Health L. Rev. 173. In the sandpaper case, as in the case of hospital supplies discussed by Klish, it was not feasible to enter the downstream market only to market the specific product. A more general model of this conduct is found in Doyle and Han (n 17). 90 See Article 101(3) TFEU. See also European Commission (n 43), paras 189–224 (policy on buying groups).

Buyer cartels  247 ciency savings to its participants, unless it is possible to fashion two or more efficient entities from the participants and unaffiliated members. The EU has a safe harbour for buying groups that take no more than 15 per cent of the input and sell no more than 15 per cent of the relevant output.91 Larger groups must seek advance approval although, as illustrated in the examples in the guidelines, larger groups are allowed in some contexts.92 These policies seek to address the potential exclusionary and exploitive risks that buying groups present. In sum, buying groups can contribute to efficiency, but can also be used to exploit producers, retard innovation, facilitate tacit or express downstream collusion as well as exclude the group’s competitors from access to essential inputs. Risks to competition exist whenever a buying group acquires a substantial share of the inputs on which it is focused. Substantial should mean a share of 15 per cent or more of general inputs or even 10 per cent or more of a class of goods resold by retail members of the buying group. The EU policies, the UK studies and the American cases show that at such levels significant risks of buyer power exist.

IV.

RETHINKING COMPETITION POLICY FOR BUYER COMBINATIONS

Competition law relating to buyer combinations needs thoughtful reappraisal today. A.

Buyer Cartels: More Pervasive and Requiring Specific Standards

1. Likelihood The general paucity of buyer cartel cases stands in stark contrast to data which show that input prices are vulnerable to exploitation.93 This suggests that there is substantial potential for conspiracies among competing buyers to affect prices. The recent empirical work on labour markets showing the scope of monopsony and oligopsony power has highlighted the capacity of employers in these markets unilaterally or collusively to exploit the labor input in their products and service.94 The resulting emergence of litigation focused on such exploitation through ‘no-poach’ agreements underlines the scope of risk. The experience with labour markets and the underlying data on price effects in input markets generally strongly suggest that more enforcement resources should be invested in examination of buyer-side collusion. There are metrics that can provide clues, such as unexplained increased margins between input and output prices either in an entire market or in regional input markets.

European Commission (n 43), para 208. By way of comparison, American antitrust authorities use a 35 per cent share of inputs. See US Department of Justice and Federal Trade Commission (n 61) 53–54. 92 European Commission (n 43) paras 221–24. 93 C.E. Fee and S. Thomas, ‘Sources of Gains in Horizontal Mergers: Evidence from Customer, Supplier, and Rival Firms’ (2004) 74 J. Fin. Econ. 423, 424–27; and S. Bhattacharyya and A. Nain, ‘Horizontal Acquisitions and Buying Power: A Product Market Analysis’ (2011) 99 J. Fin. Econ. 97. Other studies have found substantial losses to sellers resulting from buyer cartels. See J. Kwoka, ‘The Price Effects of Bidding Conspiracies: Evidence from Real Estate Auction “Knockouts”’ (1997) 42 Antitrust Bull. 503, 503; and J.P. Nelson, ‘Comparative Antitrust Damages in Bid-Rigging Cases: Some Findings from a Used Vehicle Auction’ (1993) 38 Antitrust Bull. 369, 386 Table 4 and 392–94. 94 See references at n 4 above. 91

248  Research handbook on cartels 2. Buyer cartels should remain per se illegal Earlier sections of this chapter have evaluated the arguments for a general acceptance of buyer cartels when they provide countervailing power. The weakness of those arguments provides strong support for retaining a per se rule except when there is express authorization for a cartel. Any short-term gains are likely to be offset by longer-term harms. Moreover, the development and implementation of appropriate criteria for any general basis to allow such cartels would be extremely difficult, if not impossible. As William Howard Taft observed more than 100 years ago: ‘where the sole object […] is merely to restrain competition […] there […] is no measure of what is necessary […] except the vague and varying opinion of judges as to how much, on principles of political economy, men ought to be allowed to restrain competition.’95 Taft warned that courts would ‘set sail on sea of doubt’ if they undertook to decide with respect to agreements with ‘no other purpose and no other consideration on either side than the mutual restraint of the parties, how much restraint of competition is in the public interest, and how much is not’.96 Hence, a per se rule prohibiting unauthorized buyer cartels remains the best public policy. 3. Buyer cartels require buyer-side criteria The preceding analysis shows that there is need for criteria appropriate to the buyer side of the market. In particular, two factors deserve recognition. Buyer power arises at relatively low market shares in comparison to standard analysis of seller-side risks. Hence, a buyer cartel can incorporate a lower share of the overall market and still have the capacity to exploit suppliers or exclude some competitors from market access. This is most evident in the cases involving ‘no-poaching’ agreements. The conspiracy among some, but not all, of the leading Silicon Valley technology firms provides a very good example.97 The implication of this insight is that enforcers need to be more attentive to evidence of collusion among any set of buyers regardless of their overall position in the market. Another distinction between buyer and seller cartels is that buyer cartels can include more participants and are less likely to require detailed policing or overt understanding. The incentives to collude in many circumstances where supply is price inelastic, combined with the disincentives to defect, mean that there is less need for the kind of verification that is frequently looked for in seller cartels where the incentives to defect are greater. Hence, a low number of potential participants as well as overt policing mechanisms are less likely to be evident on the buyer side. This makes detection harder. But the absence of such indicia if other elements suggest a cartel should not dissuade enforcers from further investigation. Indeed, the criteria used to exclude ‘tacit’ collusion from condemnation themselves merit reconsideration on the buying side of the market. Rather, the central issue should be remedy and not the absence of more formal collusion. If a remedy can restore workable competition to the supply side of the market, that ought to justify intervention, given the inherent incentives to exploit buyer power.98

United States v Addyston Pipe Steel, 85 F. 271 (6th Cir. 1898) 282­–83. ibid 283–84. 97 See United States v Adobe Sys. Inc., No. 1:10-cv-01629 (DDC, Final Judgment, 18 March 2011) www​.justice​.gov/​atr/​case/​us​-v​-adobe​-systems​-inc​-et​-al [accessed 25 February 2022]. 98 In some situations tacit buyer collusion is unavoidable, but market conduct regulation can modify the adverse impact to make it more difficult for buyers to collude. 95 96

Buyer cartels  249 B.

Stricter Review of the Competitive Impact of Buying Groups

There should be closer scrutiny of buying groups. They have the potential to cause anticompetitive effects in both upstream and downstream markets. A strong presumption of legality for all groups with formal characteristics of a legitimate joint venture and a market share below a generous threshold creates a serious risk of false negatives. Buyer power arises at market shares below those currently considered problematic, but such organizations can and do serve legitimate functions. Thus, the challenge is to design screens and filters that more accurately identify the buying groups that are likely to raise competitive concerns. Three screens can help authorities identify those groups that merit further and more focused inquiry. The first screen should be based upon the proportion of the input market taken by the buying group and the structural concentration of the buying side of the market. In general, more than five buyers are usually necessary to ensure a workably competitive buying market.99 Hence, a buying group could be presumed lawful only if it takes less than 10 or even 15 per cent of a commodity, regardless of the overall concentration of the market.100 This should not, however, preclude investigation of cartelizing restraints.101 The second screen should be a rough estimate of the elasticities of supply and demand in the buying group market. The barriers to exit on the supply side are relevant, as is the capacity and willingness of other buyers to take the commodity if its price declines slightly.102 When these factors are present, the competitive risks both upstream and downstream are reduced. On the other hand, if supply is price inelastic and overall demand is relatively unresponsive to price changes, the potential for anti-competitive conduct that exploits suppliers increases. The third screen should be the scope and range of scale economies in the supply market. When such economies exist, there are both incentives for legitimate efficiency-enhancing transactions and for anti-competitive use of control over the volume of business necessary to achieve scale economy. Efficiency gains provide the primary justification for groups that fall into the presumptively illegal category after focused review using the first two screens. The burden of proving that the scale of the organization is necessary to achieve economies should rest on the buyer group and its members. Dissolution is the preferable remedy (assuming that two or more buying groups would be feasible) because it ensures a more robustly competitive buying market without the need for ongoing regulatory oversight. If dissolution is impractical due to economies of scale, careful examination of the policies and procedures of the buying group can ensure that it has no greater adverse effect on the competitive process than is absolutely necessary to achieve its legitimate function. Ultimately, the goal of the screening process is to focus inquiry on the kinds of buyer groups that create the greatest risk of anticompetitive conduct. The goal should not be to eliminate an See Carstensen (2017) (n 1) 76. The threshold could be lower where the product is a branded and differentiated good rather than a more generic input; see ibid 214–17. 101 See, e.g., United States v Topco, 405 US 596 (1972), where the downstream territorial restraints had no functional relationship to the purposes of the buying group. See Carstensen and First (n 32) 203. 102 See Blair and Harrison (2010) (n 3) 53–64. The authors propose a buyer-power index in which these factors are central to assessing whether the buyer or buyer group has effective power. This measure can show power at low market share or little power despite high market share. 99

100

250  Research handbook on cartels efficiency-enhancing buyer group merely because it creates risks of anticompetitive conduct, but rather to focus on ways of avoiding or reducing the risks of such conduct. Topco is an illustration.103 The venture was prohibited from engaging in downstream territorial allocations among its members but continued to survive and prosper because its members found the group’s other services valuable.104

V. CONCLUSION Both buyer cartels and legitimate buying groups present serious challenges to maintaining a strong and viable competitive process. Despite some plausible arguments that buyer cartels provide countervailing market power to a concentrated sellers’ market, the demands of a workable competitive process preclude accepting this justification as a defence. Naked restraints of competition by buyers should be absolutely illegal unless subject to direct public regulation. Enforcement authorities also ought to look more broadly at the impact of parallel buying practices – especially in markets with relatively few buyers and many sellers. In contrast to naked restraints of competition by buyers, legitimate buying groups provide transactional efficiency as well as negotiating capacity that individual buyers often lack. As such they can make positive contributions to the overall competitive process. However, the different ways in which such groups are constituted make it hard sometimes to differentiate between legitimate joint buying ventures and naked restraints of buyer competition. For this reason, it is not irrational to implement a screening process, beginning with market-share benchmarks, to help identify the kinds of arrangement – facially legitimate – that merit further review. Thus, the general thrust of the current enforcement policies in the United States and the EU is appropriate, but the market-share screen in the United States is too generous. Overall, competition authorities ought to be more focused on the competitive risks that buying groups pose to the market.

United States v Topco Assocs., Inc., 405 US 596 (1972). Carstensen and First (n 32) 201–02.

103 104

15. Crisis cartels Éric Barbier de La Serre1

I. INTRODUCTION Economists are often mocked for their inability to reach common conclusions.2 However, if one had to identify the most established consensus among competition economists, it would probably be that, as a whole, cartels are very harmful.3 Yet, in the presence of a harsh economic crisis, even this strong consensus may need to be revisited. This may be justified for instance if market forces do not appear sufficient to solve structural overcapacity issues caused by a downturn. In this case, in the absence of public intervention or private coordination aimed at reducing or reallocating production capacities, inefficient facilities may remain on the market, to the detriment of clients. In addition, in crisis times, it cannot be excluded that collaboration will avoid the disappearance of undertakings which contribute to the competitive dynamics of the market. As a result, under extreme circumstances, there may be a tradeoff to consider between an immediate loss of competition and the maintenance of a healthy competitive structure in the long-run, at least if entry on the market is not easy. Of course, this does not mean that cartels – or restructuring agreements more generally – should normally be deemed valid in crisis times. There are, on the contrary, multiple reasons to be wary of cartels even during such exceptional periods. First, various alternatives to horizontal coordination – whether public or private – may exist to settle overcapacity issues. Mergers in particular may constitute a less harmful alternative. While mergers may affect the competitive structure of a market on a permanent basis, since an independent competitor will generally disappear, they normally generate more efficiencies (in particular through cost reductions) than cartel-like behaviour. In addition, merger control is well-equipped to take into account overcapacity issues and the difficulties encountered by the sector, not only through the failing-firm defence but also, if the stringent conditions attached to this theory are not met, as part of the competitive analysis (for example, as part of the counterfactual).

The views and opinions set forth herein are the personal views or opinions of the author; they do not necessarily reflect views or opinions of the law firm with which he is associated. 2 See, e.g., George Bernard Shaw, who is reported to have said: ‘If all economists were laid end to end, they would not reach a conclusion’: B. Robertson, J. Johnson and R. Hummerstone, Little Giant Encyclopedia: Toasts and Quotes (Sterling Publishing Company, 2009) 195. 3 While views on the methodology may differ, economists generally conclude the existence of price overcharge inflicted by cartels. See, e.g., E. Combe and C. Monnier, ‘Fines against Hard Core Cartels in Europe: The Myth of Overenforcement’ (2011) 56 Antitrust Bulletin 235, for a panorama of various methodologies used to assess the harm caused by cartels. For a balanced view on such harm, see, e.g., I. Bos and E. Pot, ‘On the Possibility of Welfare-Enhancing Hard Core Cartels’ (2012) 107 Journal of Economics 199. 1

251

252  Research handbook on cartels Second, from a more empirical standpoint, there seems to be a broad consensus that these countries that allowed cartels in crisis times did not solve their problems, but rather made them worse.4 Academic research has shown for instance that, in the United States, the NIRA (which organized the industry through ‘codes of fair competition’) delayed rather than fostered economic recovery.5 As a result, the overwhelming view is that, as such, crisis periods should not justify the implementation of cartels. The fact remains however that competition authorities have not always been so opposed to crisis cartels. And even nowadays, some prominent competition authorities still appear ready to accept that, under genuinely exceptional circumstances, coordination among competitors on the most sensitive competitive topics (and in particular on an orderly decrease of capacity) may be contemplated to settle the problems raised by a prolonged crisis. In the EU, for instance, crisis cartels taking the form of restructuring agreements are one of the rare instances where an agreement with an anticompetitive object may be found to trigger efficiencies justifying an exemption under Article 101(3) of the Treaty on the Functioning of the European Union (‘TFEU’), although this is subject to a very strict analysis. While the gate is undeniably narrow, it remains open, at least in theory. This chapter will examine this question mostly from an EU law perspective. Many of the conclusions nonetheless remain valid for other jurisdictions, as EU law is quite representative of the view taken in a number of countries. Section II deals with the difficulties associated with the definition of a crisis cartel. Section III describes the evolution of the policy and thinking on crisis cartels, with a special focus on EU law. Finally, Section IV presents the current EU rules on liability and exemptions for crisis cartels.

II.

WHAT IS A CRISIS CARTEL?

The notion of a crisis cartel is not easy to grasp. There are two main reasons for this. First, in many jurisdictions, crisis cartels are a phenomenon rather than a legal concept. In EU law, for instance, there is no regulation, nor even case law, that refers to ‘crisis cartels’ as a commonly accepted (or precisely defined) legal notion. Second, the notion of a crisis cartel is polymorph, as it covers multiple forms of collusion that occur during economic downturns. Admittedly, a commonly accepted feature of crisis cartels is that they imply coordination among competitors in the context of a crisis. However, beyond this general characteristic, many terminological questions arise.

4 J. Rivas, ‘The Views of the GCLC’, in M. Merola, J. Derenne and J. Rivas (eds), Competition Law in Times of Economic Crisis: In Need of Adjustment? (Bruylant, 2013) 45. 5 OECD, Crisis Cartels (DAF/COMP/GF(2011)11, 18 October 2011) 151 (Ireland) and 174 (Norway). See also the references cited at ibid 238.

Crisis cartels  253 A.

Is State Sponsorship Necessary?

The first question is whether the notion of a crisis cartel supposes the existence of State-sponsored measures authorizing or fostering coordination among competitors to settle a crisis situation, possibly through legislation.6 In this chapter, we will cover crisis cartels that are either organized by private parties without State support or simply encouraged by the State.7 By contrast, we will not cover crisis cartels that are either imposed or authorized by public regulations on the basis of non-competition grounds. This is because State-sponsored cartels may be justified by a broad range of socio-political, economic concerns, such as maintaining employment, which are often extraneous to the competitive analysis. In addition, while – as a matter of economics – State-sponsored cartels are not necessarily more justified than purely private coordination, they enshrine a public choice which – as a matter of law or political theory – makes them inherently more acceptable than a private cartel (provided that they comply with the State’s constitutional rules and international obligations). As reflected by the law on Article 106 TFEU and Article 4(3) of the Treaty on European Union (‘TEU’), which does not prohibit all types of State interference with competition, one cannot always assess State measures as severely as private actions. This does not mean, however, that the existence of State measures is totally irrelevant in this context. First, if the formation of a crisis cartel is not imposed or authorized – but is nonetheless encouraged – by the State, this may constitute at least an attenuating circumstance when setting the amount of the fine.8 Second, the possibility of State action through public bailouts may be relevant to determine whether a private cartel is truly indispensable (subject to State aid rules where they exist), although such measures are also imperfect from a competition standpoint.9 B.

Does a Crisis Cartel Imply Secrecy?

The second question relates to the necessity of an element of secrecy. Nowadays, the word ‘cartel’ is commonly associated to covert collusion. However, the notion of a cartel does not necessarily involve secrecy. For instance, the EU Damages Directive does not refer to covert action when it defines cartels.10 And indeed, not requiring an element of secrecy appears particularly justified for crisis cartels, which encompass hard-core cartels caused by a crisis but also less covert forms of competitor collaboration designed to address the consequences

6 OECD (n 5) 20, according to which the notion may refer to ‘a cartel that was formed during a severe sectoral, national, or global downturn without State permission or encouragement’ or ‘situations where a government has permitted, in other cases fostered, the formation of a cartel among firms during several sectoral, national or global economic downturns’. 7 For instance in 2008, when the Greek Minister of Agriculture supported an agreement to set prices and restrict sales/output in the fish-farming sector in Greece; see Hellenic Competition Commission, Decision No. 492/Vi/2010. 8 See Section IV.C below. 9 L. Vitzilaiou, ‘Crisis Cartels: For Better or For Worse?’ [2011] 2 CPI Antitrust Chronicle 2. 10 See Article 2(14) of Directive 2014/104/EU of the European Parliament and of the Council of 26 November 2014 on Certain Rules Governing Actions for Damages under National Law for Infringements of the Competition Law Provisions of the Member States and of the European Union [2014] OJ L349/1.

254  Research handbook on cartels of a downturn and, in particular, some restructuring agreements designed to deal with overcapacity. C.

Which Type of Collusive Conduct?

During severe crisis times, competitors may be inclined to coordinate on various forms of capacity or output reduction, but also on prices, for instance, so as to maintain sufficient profitability on the market or to make output reduction more effective.11 However, the notion of a crisis cartel is also sometimes used to designate narrower forms of collusive action. It may refer to any type of horizontal collusion among competitors on the most sensitive aspects of competition (on capacity, output or prices, for instance) that occurs in the context of a crisis. But it may also refer, more restrictively, to agreements between a large number of (if not all) competitors in a given sector that are designed to settle excess capacity issues caused by a crisis.12 The latter definition covers the type of restructuring agreements that the European Commission (‘the Commission’) has sometimes authorized, although under strict conditions. By contrast, at least in the EU, the broader definition is less meaningful from a legal standpoint, as the sole fact that coordination on the most sensitive aspects of competition occurred in a crisis context normally triggers very limited consequences on the legality of the related action. There is a broad consensus to the effect that, for instance, price fixing designed to address the consequences of a crisis remains unjustifiable. As a result, in this chapter we will focus on crisis cartels meeting the narrow definition. However, on this aspect, legal categories appear to be quite blurred. For instance, it is not always easy to distinguish a potentially valid restructuring agreement from a cartel designed to limit output. D.

Which Type of Crisis?

Over the past 70 years the world economy has suffered a number of economic shocks that were prone to the formation of crisis cartels, including two oil shocks (in 1973 and 1979), the dotcom bubble (in 2000), the financial crisis (in 2008) and the sanitary crisis related to COVID-19 (from 2020). There have also been multiple downturns affecting specific sectors of the EU economy. However, assuming that there is room for an exemption applicable to crisis cartels, it cannot cover any crisis. Otherwise, any exemption that would be triggered by a crisis would risk becoming quasi-permanent. The problem is compounded by the fact that many cartels arise in the context of a crisis. On this point, EU case law has even taken the view that ‘as a general rule cartels come into being when a sector encounters problems’.13

See, e.g., the Greek crisis cartel concerning fish-farming reported by Vitzilaiou (n 9); and OECD (n 5) 131 (Greece). 12 Concerning the latter, see A. Fiebig, ‘Crisis Cartels and the Triumph of Industrial Policy Over Competition Law in Europe’ (1999) 25 Brooklyn Journal of International Law 607. If a restructuring agreement is bilateral rather than multilateral, it will often qualify as a specialisation agreement (see below). 13 See e.g., Joined Cases T-236/01 etc. Tokai Carbon v Commission [2004] ECR II-1181, para 345. 11

Crisis cartels  255 If not all types of crisis may justify cartels, where should one draw the line? As will be further detailed below, an important distinction must be made between structural and cyclical downturns. In the EU, the dominant view is that only the former type of crisis may justify cartel behaviour, as the overcapacity issues created by a cyclical crisis should normally be overcome through price adjustments which may lead the least efficient undertakings to disappear.14 This was also the view in Germany until 2005 (when the specific crisis cartel exemption enshrined in German legislation was suppressed and the conditions for a possible exemption were aligned on those of Article 101(3) TFEU).15 But then, what is a ‘structural crisis’? According to the Commission, structural over capacity exists where over a prolonged period all the undertakings concerned have been experiencing a significant reduction in their rates of capacity utilisation and a drop in output accompanied by substantial operating losses and where the information available does not indicate that any improvement can be expected in this situation in the medium-term.16

The decisional practice shows that, normally, an exemption may be considered only when there is a strong consensus on the existence of a durable overcapacity crisis. In Synthetic Fibres, for instance, the structural crisis had lasted more than ten years, and there was a strong institutional consensus to the effect that a reduction of overcapacity was necessary.17

III.

POLICY EVOLUTIONS ON CRISIS CARTELS UNDER EU LAW

While the Commission currently promotes a hard stance against cartels, including when they are implemented in crisis times, its position has not always been so strict. On the contrary, the Commission has sometimes been rather pragmatic in the presence of dire economic circumstances, although this has resulted in a very limited number of exemption decisions. To our knowledge, the Commission has never disowned these precedents. However, at the beginning of the millennium, it strived to limit the situations in which an exemption may apply. A.

From the 1970s to the Mid-1990s: The Commission’s Flexible Approach

In EU law the first signs of a legal debate on crisis cartels seem to date from the early 1970s, due to the harsh overcapacity crisis caused by the two oil shocks of this period, in particular in the EU textile and petrochemical sectors.

OECD (n 5) 122–23 (Germany). According to the same source, the exemption has very rarely been applied. 15 ibid (European Union) 110. 16 European Commission, Twelfth Report on Competition Policy (Brussels and Luxembourg, 1983) para 38. See also OECD (n 5) 110 (European Union). For a discussion of this definition, see Fiebig (n 12). 17 European Commission, Eighth Report on Competition Policy (Brussels and Luxembourg, 1979) para 42. See also Case IV/30.810, Synthetic Fibres, Commission decision, 4 July 1984 [1984] OJ L207/17, paras 8–10. 14

256  Research handbook on cartels In 1972, in the Cimbel case, a group of Belgian cement producers notified a cartel that included elements of market allocation and price fixing. In support of the notified agreement, the cartelists argued that it had been concluded during the Great Depression of the 1930s. The Commission refused to exempt the cartel. However, it did so not because, as such, a crisis could not justify a cartel, but because the crisis at stake in this case was no longer ongoing.18 This suggested that a case-by-case approach was needed.19 The Commission reaffirmed its cautious approach when it was asked to rule on a coordinated attempt by the European textile industry to decrease capacity in an orderly fashion. In the Second Report on Competition Policy of 1972, the Commission acknowledged that ‘the cut polyester fibre industry labour[ed] under heavy surplus capacity’ and that, ‘because the manufacturers endeavor[ed] to scale down their surpluses, pressure [was] brought to bear on prices’.20 The Commission nonetheless objected to the agreement because it covered ‘the participants’ production and selling policies’. However, the Commission took care to note that the question of whether the structural difficulties experienced by the industry could ‘be solved in a manner compatible with the rules of competition of the EEC Treaty was not settled in this case’.21 Six years later in 1978, in light of the hard crisis still faced by European industry, there were discussions within the Commission concerning the adoption of a regulation (pursuant to what is now Article 103 TFEU) declaring Article 85 EEC (now Article 101(1) TFEU) inapplicable to agreements between firms aimed at balancing excess capacity with a lasting decline in demand.22 Commissioner for Industrial Affairs Etienne Davignon and Commissioner for Competition Raymond Vouel produced a draft regulation according to which the Council of Ministers would decide, upon proposal by the Commission, whether a coordinated reduction in the capacity of a given sector was necessary. While this proposal was abandoned, this did not put an end to the debate on crisis cartels. On the contrary, in its Eighth Report on Competition Policy of 1979, the Commission reported a second attempt of the textile industry to agree on a decrease of capacity. The Commission noted that, since 1972, the continuing economic crisis and the commissioning of new production capacity in Europe had increased the precariousness of the situation in the European man-made fibres industry, where capacity utilization rates had dropped to an average of 70 per cent. There was therefore a broad consensus to the effect that capacity needed to be restricted. This context prompted eleven producers to notify an agreement organizing an orderly restriction of capacity. The Commission once again dismissed the application, as the notified agreement included a quota system.23 This is because, according to the Commission,

18 Case IV/243, 244, 245, Cimbel, Commission decision, 22 December 1972 [1972] OJ L303/72, para 21. 19 European Commission, Second Report on Competition Policy (Brussels and Luxembourg, 1973) para 30 (‘the agreement had not solved the overcapacity problem […] Moreover, the purely potential danger of cut-throat competition could not justify at the present time the elimination of competition in the field of prices’). See also ibid para 29 concerning another agreement (in the Netherlands) designed to settle overcapacity issues in the cement industry. 20 ibid para 31. 21 ibid. 22 See Fiebig (n 12); and I. Kokkoris, ‘Should Crisis Cartels Exist amid Crises?’ (2010) 55 Antitrust Bulletin 727. 23 European Commission (n 17) para 42.

Crisis cartels  257 an agreement to restrict capacity could be authorized only if it did not include provisions fixing prices, production or sales.24 However, once again the Commission refrained from finding that crisis cartels should necessarily be found anti-competitive. On the contrary, one year later, in view of the relentless economic recession and acute unemployment, the Commission decided to offer guidance on crisis cartels in its Twelfth Report on Competition Policy.25 These guidelines are still influential today. In particular, the report set a definition of ‘structural overcapacity’ that was used in recent Commission policy guidance.26 The report also stated an economic principle on which the Commission still relies, and according to which, although one should normally count on market forces to settle overcapacity issue, ‘economic circumstances do not necessarily guarantee a reduction of the least profitable surplus capacity’.27 As a result, the Commission recognized that certain agreements addressing a whole sector may be condoned if they are ‘aimed solely at achieving a coordinated reduction of overcapacity and do not otherwise restrict free decision-making by the firms involved’, which supposes in particular that companies do not resort to price fixing or quotas.28 The Commission also offered guidance on how the four conditions of Article 85 EEC (now Article 101(3) TFEU) could be satisfied. The Commission emphasized the need for a detailed programme of closures and positive effects on employment.29 It also noted that, as an alternative to sectoral agreements, it could envisage reciprocal specialization agreements between a small number of firms (which would enable them to close excess capacity).30 One year later, in its Thirteenth Report on Competition Policy of 1983, the Commission confirmed what it called its ‘flexible approach towards joint structural capacity reductions aimed at achieving a healthier structural situation in the sector concerned’.31 That same year, it examined a notification made by the six major Community zinc producers, who sought exemption of a ‘shutdown agreement’ under Article 85(3) EEC. In view of the heavy financial losses suffered by the European zinc industry and the fact that the agreement was to last for a fixed period, set in advance, the Commission considered exempting the agreement, although eventually the parties gave up their plans in view of the improvement of the situation on the zinc market.32 One year later, in 1984, the Commission finally adopted its first decision exempting an agreement to restrict capacity, which concerned the synthetic fibres industry. This agreement had been concluded between the ten biggest European companies active in this sector and 24 European Commission, Eleventh Report on Competition Policy (Brussels and Luxembourg, 1981) para 46. In 1980, the Commission refused to exempt a cartel agreement among Italian cast glass producers that included quantitative sharing under the form of sales quotas. The Commission’s decision is not based on the absolute impossibility of raising a crisis defense, but on the fact that it was ‘not possible to allow, in the guise of a crisis cartel, restrictions which are not indispensable’: Case IV/29.869, Italian Cast Glass, Commission decision, 17 December 1980 [1980] OJ L383/19, para 3. 25 European Commission, Twelfth Report on Competition Policy (Brussels and Luxembourg, 1983) para 38. 26 ibid para 38. 27 ibid para 38. 28 ibid para 39. 29 ibid paras 39–41. 30 ibid para 41. 31 European Commission, Thirteenth Report on Competition Policy (Brussels and Luxembourg, 1984) para 61. 32 ibid paras 58–59.

258  Research handbook on cartels aimed at closing down parts of their production capacity (on average 18 per cent) by specified dates. The actual implementation of the cuts was to be controlled through an internal reporting system, and failure to carry out the capacity reductions agreed upon would give rise to compensation payments. In spite of its clear anticompetitive object and effect, the agreement was exempted. Crucially, the Commission found that: (i) ‘market forces by themselves had failed to achieve the capacity reductions necessary’; (ii) the agreement ‘ensur[ed] that the shake-out of capacity w[ould] eliminate the non-viable and obsolete plant that could only have survived at the expense of the profitable plant through external subsidies or loss financing within a group, and w[ould] leave the competitive plants and business in operation’; (iii) fears about future price levels were alleviated by buyer power; and (iv) the agreement did not interfere with the parties’ freedom to determine their output or deliveries.33 And the same year, the Commission once again confirmed its flexible approach when it exempted a scheme among competitors in the petrochemical industry designed to settle overcapacity issues, which it assimilated to a specialization agreement.34 A few years later, in view of the acute recession of the early 1990s, the Commission deemed it necessary to reiterate its policy statements in the XXIIIrd Report on Competition Policy,35 and in 1994 it exempted for the second time a broad restructuring agreement, in the Dutch Bricks case.36 The notified agreement was designed to address a structural reduction in the demand for common bricks, which had been replaced by alternative building and finishing materials. According to the Commission, while the cyclical situation in the construction industry had exacerbated the reduction in demand, several indicators pointed to long-term underlying structural overcapacity. As a result, the Dutch producers notified an agreement pursuant to which the obsolete capacity production units would be concentrated in a more modern plant operating at higher capacity and productivity level. However, the agreement also contained provisions to fix production quotas and leading to the sharing of virtually the entire output of bricks in the Netherlands. The parties dropped these restrictions, which led the Commission to exempt the agreement. In sum, the Commission has not always been averse to horizontal cooperation aimed at reducing capacity, provided that such cooperation is indispensable to achieve productive efficiency and the measures are narrowly tailored to this objective. However, recent policy statements show that these precedents should now be taken with great caution.

33 Synthetic Fibres (n 16) paras 31–32, 39, 41 and 43. See also European Commission, Fourteenth Report on Competition Policy (Brussels and Luxembourg, 1985) paras 81–82. 34 Case IV/30.863, BPCL/ICI, Commission decision, 19 July 1984 [1984] OJ L212/1. See also: European Commission (n 33) paras 83–84 and para 85 (noting, with respect to Case No IV/30.778, Rovin, a ‘favourable attitude to another restructuring operation’). See two other specialization agreements in the petrochemical industry that were designed to settle overcapacity issues and were approved in spite of their anti-competitive object and effect: Case IV/31.055, ENI/Montedison, Commission decision, 4 December 1986 [1987] OJ L5/13; and Case IV/31.846, Enichem/ICI, Commission decision, 22 December 1987 [1988] OJ L50/18. In the latter decision, the Commission noted that ‘the imperative need to reduce overcapacity in this sector [was] in the general interest of the Community’ (ibid para 41). 35 European Commission, XXIIIrd Report on Competition Policy (Brussels and Luxembourg, 1993) para 82. 36 Case IV/34.456, Stichting Backsteen, Commission decision, 29 April 1994 [1994] OJ L131/15. See also European Commission (n 34) para 89.

Crisis cartels  259 B.

The New Millennium: A Hardened View

In 1999, a legal scholar complained that the Commission’s practice on crisis cartels constituted a ‘triumph of industrial policy over competition law’.37 This was an overstatement, since to our knowledge (and leaving aside specialization agreements) the Commission has exempted only two sectoral restructuring agreements due to a crisis. But in any event this fear is now outdated, as the Commission’s official policy on crisis cartels has become much more severe. This shift may be explained by two main reasons. First, the Commission’s general policy against cartels considerably hardened at the beginning of the millennium. In September 2000, Mario Monti, who at that time was the Commissioner in charge of Competition, stated that cartels were ‘cancers on the open market economy, which forms the very basis of our Community’.38 It therefore became increasingly difficult for the Commission to accept the possibility of exempting measures which, objectively, are very detrimental to competition in the short term. Second, the early 2000s also correspond to the period when the application of EU competition law was decentralized. National courts and authorities were granted the right – through Regulation No 1/2003 – to apply Article 101(3) TFEU in full. At that time, the Commission may have been concerned that this new right would result in an excessively loose application of the exemption. As a result, the Commission’s guidelines on Article 101(3) TFEU reflect a rather conservative view of the conditions for the application of this provision, and this strict view has even more reasons to apply in the presence of what normally constitutes a hard-core restriction of competition. Maybe for this reason, the guidelines do not address the issue of capacity reduction agreements directly, but only refer in general terms to agreements allowing for better planning of the production or better capacity utilization.39 Neither do the Commission’s guidelines on horizontal cooperation contain any explicit reference to restructuring agreements.40 Admittedly, the Commission seems to have never officially disowned its two precedents of the 1980s–1990s. However, at the very least, it has strived to limit their precedential value. First, although the financial crisis of 2008 constituted a major shock for the EU economy and the rest of the world, it has not triggered any relaxation of the fight against cartels.41 On the contrary, the Commission intervened in 2010 to prevent the implementation of the ‘Baltic Max Feeder’ scheme whereby European ‘feeder’ vessel owners coordinated to address the collapse of world trade following the financial crisis of 2008 and, as result, agreed to jointly cover the

Fiebig (n 12). M. Monti, ‘Fighting Cartels Why and How? Why Should We Be Concerned with Cartels and Collusive Behaviour?’ (SPEECH/00/295, 3rd Nordic Competition Policy Conference, Stockholm, 11–12 September 2000). 39 Guidelines on the Application of Article 81(3) of the Treaty [2004] OJ C101/97, para 68. 40 Guidelines on the applicability of Article 101 of the Treaty on the Functioning of the European Union to Horizontal Co-operation Agreements [2011] OJ C11/1. 41 According to Neelie Kroes, then Commissioner in charge of Competition, ‘encouraging cartelists and others would be guaranteeing disaster. It would drag down recovery, increase consumer harm and create more cartel and cartel cases into the future. No-one wins – today’s softness is tomorrow’s nightmare’: N. Kroes, ‘Tackling Cartels – A Never-Ending Task’ (SPEECH/09/454, Conference on Anti-Cartel Enforcement: Criminal and Administrative Policy – Panel Session, October 2009, Brazil). 37 38

260  Research handbook on cartels costs of removing vessels from service.42 The Commission’s press release does not contain any sign that it was somehow ready to maintain its flexible approach of the previous millennium. Second, in the Irish Beef case43 – which will be presented further in Section IV below – but also in its contribution to the OECD Report on crisis cartels of 2011,44 the Commission promoted a restrictive view on the validity of restructuring agreements. In substance, the Commission submitted that crisis cartels may still be exempted, but only in very special market failure situations, which will be further detailed below (that is, when there is structural overcapacity that does not result from a cyclical downturn, in combination with an inefficient war of attrition between competitors). This harder stance is in line with the progressive suppression or restriction of crisis cartel exemptions in many countries where they still existed at that time (for example, in Germany and Japan).45 In the United States, while legislation of the Great Depression era allowed the implementation of a number of crisis cartels through ‘codes of fair competition’, the US government’s official position is now crystal-clear: ‘[c]artels are illegal at any time and are subject to criminal prosecution, regardless of the economic climate.’46 The harsh economic crisis resulting from the COVID-19 sanitary crisis also confirms this hardened stance. In its temporary framework for assessing antitrust issues related to business cooperation in response to situations of urgency stemming from the current COVID-19 outbreak, the Commission pointed out that competition rules, far from constituting rigid barriers to the behaviour of operators, allow for adjustments to be made in situations of economic crisis.47 The Commission acknowledged that ‘[t]he exceptional circumstances of this time and its related challenges may trigger the need for undertakings to cooperate with each other in order to overcome or at least to mitigate the effects of the crisis to the ultimate benefit of citizens’.48 The Commission has therefore been rather open to competitor collaboration designed to address the shortage of essential products and services during the COVID-19 outbreak (notably medicines and medical equipment). It has also accepted that it may be necessary to provide undertakings with ad hoc feedback or comfort on the legality of specific cooperation initiatives.49 The Commission has accepted in particular that cooperation in the health sector might occur through a trade association (or an independent advisor, or independent service provider, or public body) so as to: (i) coordinate joint transport for input materials; (ii) contribute to identifying those essential medicines for which, in view of forecast production, there are risks of shortages; (iii) aggregate production and capacity information, without exchanging individual company information; (iv) work on a model to predict demand on a Member State

European Commission, ‘Antitrust: Commission Closes Investigation into “Baltic Max Feeder” Scheme’ (IP/10/374, 26 March 2010). 43 Case C-209/07 BIDS and Barry Brothers [2008] ECR I-8637. 44 OECD (n 5). 45 ibid 122–23 (Germany) and 155–56 (Japan). 46 OECD (n 5) 215 (United States). 47 European Commission, ‘Communication from the Commission – Temporary Framework for Assessing Antitrust Issues Related to Business Cooperation in Response to Situations of Urgency Stemming from the Current COVID-19 Outbreak’ [2020] OJ C116/7. 48 ibid para 3. 49 ibid paras 4–5. 42

Crisis cartels  261 level and identify supply gaps; and/or (v) share aggregate supply gap information and request participating undertakings, on an individual basis and without sharing that information with competitors, to indicate whether they can fill the supply gap to meet demand (either through existing stocks or increase of production).50 The Commission has also been open to measures designed to adapt production, stock management and, potentially, distribution in the industry, which may require exchanges of commercially sensitive information and a certain level of coordination of which site produces which medicines (so that not all undertakings focus on one or a few medicines, while others remain in under-production). According to the Commission, while such exchanges and coordination between undertakings are in normal circumstances problematic under EU competition rules, they would not raise issues or – in view of the emergency situation and temporary nature – give rise to an enforcement priority, to the extent that they would be: (i) designed and objectively necessary to actually increase output in the most efficient way to address or avoid a shortage of supply of essential products or services, such as those that are used to treat COVID-19 patients; (ii) temporary in nature; and (iii) not exceeding what is strictly necessary to achieve the objective of addressing or avoiding the shortage of supply.51 However, this pragmatic position does not reveal any kind of renewed openness to crisis cartels, since what is at stake here is not overcapacity or a shortage of demand (as is normally the case in crisis cartels), but on the contrary under-capacity (or poorly allocated capacity) and therefore shortages of supply. These are completely different situations: in the case of COVID-19, coordination is designed to meet an overriding requirement (that is, the efficient organization of productive capacities to address a health crisis); it is not designed, like in classical crisis cartels, to address overcapacity concerns caused by a structural downturn. In some Member States, competition authorities have authorized measures which are more oriented towards the economic consequences of the sanitary crisis and, in particular, the maintenance of a competitive structure. In Germany, the Bundeskartellamt stated that it would not (at this stage) take action against coordinated measures among car manufacturers that were designed to ensure the restart of automotive production.52 As noted by the Bundeskartellamt, as the automotive industry relies on complex supply chains, downtime at individual sub-suppliers can cause even more economic harm by considerably delaying the restart of production processes at many suppliers and manufacturers. As a result, the companies concerned were to set up stakeholder groups and exchange information within and between the stakeholder groups over a limited period of time. Owners, employees, customers, creditors and public authorities were to exchange information on the solvency, credits, aid measures or operational problems of a company and quickly develop effective restructuring measures together. However, the scheme included many safeguards. In particular, the relevant suppliers in the automotive industry remained free to choose when and in what way they would like to restart their activities. The scheme also included strict limitations on the exchange of sensitive

ibid paras 10–13. ibid paras 14–15. 52 Bundeskartellamt, ‘Crisis Management Measures in the Automotive Industry – Bundeskartellamt Supports the German Association of the Automotive Industry (VDA) in Developing Framework Conditions under Competition Law Aspects’ (9 June 2020) www​.bundeskartellamt​.de/​SharedDocs/​ Meldung/​EN/​Pressemitteilungen/​2020/​09​_06​_2020​_VDA​.html [accessed 6 March 2022]. 50 51

262  Research handbook on cartels data, and the relevant suppliers remained free to choose when and in what way they would like to restart their activities. This type of cooperation is closer to the restructuring measures that are generally covered by classical crisis cartels aimed at rationalizing output. However, this cooperation remains restricted in scope. In addition, it is under the control of public authorities and is limited to a short period (the restart of the automotive industry). As a result, the acceptance of such measures does not constitute a sign of renewed openness to crisis cartels.

IV.

LIABILITY AND EXEMPTIONS UNDER EU LAW

Under EU law, collusion to restrict capacity, and even more so to fix prices, will nearly always be found to be anticompetitive by object, even if it originates from a crisis. A more delicate question is whether the harsh economic conditions that may have triggered this behaviour justify an exemption under Article 101(3) TFEU. As is apparent from the Commission’s most recent policy statements, the existence of a crisis may justify an exemption only in genuinely exceptional circumstances. In addition, when an infringement is found and a fine is imposed, the existence of a crisis is increasingly less relevant as a mitigating circumstance. A.

Crisis Cartels as a Restriction by Object

According to the Court of Justice, in order to determine whether an agreement between undertakings or a decision by an association of undertakings reveals a sufficient degree of harm to competition that it may be considered a restriction of competition ‘by object’ within the meaning of Article [101(1) TFEU], regard must be had to the content of its provisions, its objectives and the economic and legal context of which it forms a part.53

The three decisive elements to qualify an anticompetitive object are therefore: (i) the content of the agreement; (ii) its objectives; and (iii) its economic and legal context. For crisis cartels, the economic context of the agreement (that is, the crisis) matters less than its content (for example, provisions decreasing and/or allocating capacity), as the sole existence of the agreement is very likely to result in a restriction of competition. As to the parties’ intention, they are in general less decisive than the content of the agreement and its context.54 As a result, even if the parties to a crisis cartel were well intentioned and aimed at preserving a healthy competitive structure in the medium or long term, this would normally not be sufficient to escape the finding of a restriction by object. The leading case on this issue is BIDS (also called ‘Irish Beef’).55 In this case, the major beef and veal producers in Ireland had put in place a capacity reduction agreement taking the form of an association of undertakings, the Beef Industry Development Society (‘BIDS’). According to the agreement, certain producers withdrew from the market in return for the Case C-67/13 P CB v Commission ECLI:​EU:​C:​2014:​2204, para 53; see also ibid (‘When determining that context, it is also necessary to take into consideration the nature of the goods or services affected, as well as the real conditions of the functioning and structure of the market or markets in question’). 54 ibid para 54. 55 Case C-209/07 BIDS and Barry Brothers [2008] ECR I-8637. 53

Crisis cartels  263 payment of compensation from the BIDS, which was funded by the producers remaining on the market. The Irish Competition Authority objected to this scheme and brought the case before the Irish courts. Upon request for a preliminary ruling, the Court of Justice held that the BIDS arrangements were intended to implement a common policy whose object was to encourage the withdrawal of certain players from the market.56 According to the Court, such arrangements ‘conflict patently with the concept inherent to the treaty provisions relating to competition, according to which each undertaking must act independently’.57 In addition, as stressed by Advocate General Trstenjak in her observations,58 the means put in place to attain the objectives of the BIDS agreements included restrictions by object. First, each remaining producer was subject to a contribution which was calculated on the basis of the additional cattle slaughtered beyond their usual volume of production. As a consequence, the companies staying on the market were tempted to freeze their production in order to limit their contributions to the exiting producers.59 Second, the exiting producers were subject to non-compete obligations preventing them from using or leasing their plants, then hindering the possible use of those capacities by new operators entering the market to compete with the remaining producers.60 The scheme was therefore found to be anticompetitive by object, and in our view this outcome is not affected by the view – later expressed by the Court of Justice – according to which the notion of anticompetitive object should be interpreted restrictively.61 The Commission’s observations in the BIDS case also provide interesting insight into its current policy on crisis cartels.62 In these observations, the Commission relied on the Weyl Beef Production63 and FNCBV64 cases to argue that a crisis context is not in itself sufficient to exclude the application of Article 101 TFEU.65 In Weyl Beef Production, the Commission had authorized a State aid restructuring scheme in the Dutch beef sector that was designed to settle overcapacity issues. The scheme provided for departure from the market in exchange of public compensation. One company challenged the decision authorizing the scheme based on the anti-competitive nature of the private reorganization agreements underlying the aid. On this point, the Court of First Instance (now the General Court) concluded that even if the agreements fell within the scope of Article 101 TFEU, they could not be assessed independently of the aid, as they were inextricably linked to it.66 The Commission inferred from this holding that a crisis context is not in itself sufficient to exclude the application of Article 101 TFEU

ibid para 33. ibid para 34. 58 Case C-209/07 BIDS and Barry Brothers [2008] ECR I-8637, Opinion of AG Trstenjak. 59 Case C-209/07 BIDS and Barry Brothers [2008] ECR I-8637, Judgment of the Court, para 37. 60 ibid para 38. 61 Case C-67/13 P CB v Commission, EU:​C:​2014:​2204, para 58. 62 European Commission, ‘In the Court of Justice of the European Communities – Written Observations of the Commission of the European Communities – Case C-209/07’ (14 August 2007) https://​ec​.europa​.eu/​dgs/​legal​_service/​submissions/​c2007​_209​_obs​_en​.pdf [accessed 5 March 2022]. 63 Joined Cases T-197/97 and T-198/97 Weyl Beef Products and others v Commission [2001] ECR II-303. 64 Joined Cases T-217/03 and T-245/03 FNCBV and others v Commission [2006] ECR II-4987. 65 European Commission (n 62) paras 53–55. 66 Joined Cases T-197/97 and T-198/97 Weyl Beef Products and others v Commission [2001] ECR II-303. 56 57

264  Research handbook on cartels to reorganization agreements.67 Similarly, in FNCBV, the Court of First Instance (now the General Court) stated that the existence of a crisis in a particular industry is not sufficient to exclude the application of Article 81 EC (now Article 101(1) TFEU).68 Building on these two precedents, the Commission argued in BIDS that, even if overcapacity was part of the context to be taken into account to assess the qualification of object, by signing the agreements, both outgoing and remaining BIDS members entered into a series of legal obligations to reduce capacity. The Commission agreed with the Irish Competition Authority that agreements constituting a reduction in capacity necessarily lead to a reduction in output.69 And since the agreement at stake restricted the freedom of the parties to decide on investments and competitive strategies or to leave the market,70 it did not matter whether the remaining producers retained their freedom to produce or not.71 B.

The Very Strict Conditions for the Exemption of Crisis Cartels

In EU law, even an agreement with a clear anticompetitive object may in theory be exempted under Article 101(3) TFEU.72 Crisis cartels offer a good illustration of this exceptional possibility, since in Synthetic Fibres and Dutch Bricks the agreements analysed by the Commission were exempted in spite of their anticompetitive object.73 However, the path to an exemption now appears to be very narrow. An exemption seems available only in presence of: (i) overcapacity that does not result from a cyclical downturn; and (ii) risks of an inefficient war of attrition between competitors having a similar cost structure. As a result, in the Commission’s own words, ‘it will be very difficult for parties to succeed with a defence under Article 101(3) TFEU’.74 This is further illustrated by the BIDS case. In this matter, the Irish Supreme Court had not requested the Court of Justice to determine the conditions under which an output reduction agreement may be exempted under Article 101(3) TFEU in spite of its anticompetitive object. This question was nonetheless raised at a later stage of the national proceedings, which prompted the Commission to state its views in an amicus curiae brief of 30 March 2010 (pursuant to Article 15(3) of Regulation No 1/2003).75 This brief provides useful insights on the Commission’s current understanding of how Article 101(3) TFEU may apply to crisis cartels, although it does not address the condition relating to the elimination of competition (and, on this point, simply refers to the Commission’s guidelines).76

European Commission (n 62) para 53. Joined Cases T-217/03 and T-245/03 FNCBV and others v Commission [2006] ECR II-4987, para

67 68

90.

European Commission (n 62) para 77. ibid para 80. 71 ibid para 88. 72 Case T-17/93 Matra Hachette v Commission [1994] ECR II-595, para 85. 73 Synthetic Fibres (n 17). 74 OECD (n 5) 120 (European Union). 75 European Commission, ‘Amicus Curiae Brief of 30 March 2010 before the Dublin High Court in Case No. 7764 P, The Competition Authority vs. BIDS’ (hereafter ‘Amicus Curiae’). 76 ibid para 16. See also K. Dekeyser, ‘The Views of the EU Enforcement Officers’, in Merola et al (n 4) 149. 69 70

Crisis cartels  265 Concerning the first condition of Article 101(3) TFEU, that is, the need for a contribution to economic progress, the Commission notes that capacity reduction agreements may trigger two types of efficiencies. First, the agreement may lead to the removal of inefficient capacity from the industry. However, according to the Commission, the parties must be able to (i) establish the prior existence of the inefficient capacity on the market, and (ii) give sufficient indication of what capacity will be removed, that is, which firms are to exit the market or which plants are to reduce their capacity, or at least the criteria according to which capacity will be reduced.77 Second, a capacity reduction agreement may lead to a better utilization rate of the remaining capacities. A restructuring agreement that would provide for limitations on production seems unlikely to benefit from the exemption, although the Commission does not rule out a case-by-case approach.78 The Commission then specifies the type of cost benefits which may arise from greater capacity utilization, in the hypothetical situation where there is no output limitation. According to the Commission, most of the time, a capacity reduction agreement will allow savings of fixed costs, as the remaining players will be able to increase their output while spreading their fixed costs over a larger quantity of products.79 The Commission stresses that a capacity reduction agreement would more rarely lead to a reduction in variable costs (that is, the costs which vary with output). This is an important distinction since, as will be seen below, reductions in variable costs are much more likely to meet the third condition of Article 101(3) TFEU than reductions in fixed costs. According to the Commission, a saving of variable costs may happen where higher production levels enable the use of more efficient technology, where larger purchases of raw materials decrease input costs, or in industries that lend themselves to learning economies.80 In its brief, the Commission does not identify employment protection as a possible efficiency. This is hardly surprising in view of the Commission’s recent focus on ‘objective economic efficiencies’ under Article 101(3) TFEU, that is, mostly cost and qualitative efficiencies.81 Yet, historically, the Commission and the EU Courts have already taken into account employment as an objective worthy of protection under Article 101(3) TFEU, not only in the context of restructuring agreements,82 but also more generally.83 Realistically, one should not expect a relaxation of the Commission’s position on this point in the coming years. However, this narrow interpretation remains to be tested before the EU courts. The second condition of Article 101(3) TFEU relates to the indispensability of the restriction. This condition consists of a twofold test. First, the agreement in general must be reasonably necessary to bring about efficiencies. Second, each of the individual restrictions arising from the agreement must be reasonably necessary to achieve the claimed efficiencies. In its observations, the Commission focuses on the first test, as the second one depends on a casuistic assessment of each agreement.84 Amicus Curiae (n 75) paras 20–23. ibid paras 24–25. 79 ibid para 26. 80 ibid para 27. 81 Guidelines on the Application of Article 81(3) of the Treaty [2004] OJ C101/97, paras 59–72. 82 European Commission (n 16) para 39; Synthetic Fibres (n 17) para 38, and Stichting Baksteen (n 36) para 27. 83 Case 26-76, Metro I [1977] ECR 1875, para 43. See also Rivas (n 4). 84 Amicus Curiae (n 75) para 29. See also OECD (n 5) 115 (European Union). 77 78

266  Research handbook on cartels The Commission notes in particular that it will verify whether ‘no other economically practicable and less restrictive means of achieving the efficiencies’ were possible. And the indispensability considered is not the indispensability to the existence of the agreement itself, but indispensability for the achievement of the benefits identified. In the case of restructuring agreements, this supposes answering the question of whether market forces alone could have solved the problem of overcapacity without the collective intervention of individual undertakings.85 The Commission underlines that, traditionally, when demand falls, it is for each undertaking to adapt its pricing policy and to decide whether and to what extent its overcapacity becomes economically unsustainable.86 Nevertheless, the Commission acknowledges that overcapacity problems cannot always be solved within a reasonable time period by the free interplay of market forces, in which case overcapacity is likely of a structural nature.87 The Commission gives the example of a ‘war of attrition’ scenario that leads to a prisoner’s dilemma: in certain circumstances, firms may refrain from releasing unused capacity in order to induce their rivals to leave the market first. In this case, no operator will take the risk of reducing its own capacity and suffer economic losses. Instead, each player will wait for another player to reduce capacity in order to benefit from the overall fall in capacity in the industry.88 The Commission stresses that firms are likely to participate in such a war of attrition in two situations: either when companies refuse to reduce capacity because of high fixed or sunk costs that normally decrease with output, or when firms cannot reduce capacity because they operate in a stable, transparent and symmetrical market structure in which firms of similar size and cost structure are sufficiently aligned to maintain capacity at excess level.89 While the Commission accepts that these two situations do not constitute an exhaustive list, it nonetheless assumes that in most cases merger and acquisitions and specialization agreements constitute less restrictive remedies to solve a problem of structural overcapacity in an industry.90 The Commission considers that, generally, these two alternatives tend to eliminate inefficient plants and therefore reduce overcapacity on the market.91 The Commission’s strict view is compounded by its decisional practice, which shows that reciprocal specialization agreements may indeed constitute valid alternatives to industry-wide restructuring schemes, provided that they do not incorporate any price-fixing, quota-fixing or market-allocation provisions. For instance, in 1984, the Commission exempted an agreement between two British manufacturers (British Petroleum Chemicals and Imperial Chemical Industries) that aimed at reducing capacity.92 The Commission stated that ‘given that the over-capacity in the industry is of a structural nature, market forces would have been too slow at bringing about the necessary radical changes. These agreements by their immediate closure of plants, accelerate the tendency to re-establish the equilibrium in supply and demand.’93

87 88 89 90 91 92 93 85 86

Amicus Curiae (n 75) para 32. ibid para 33. ibid para 35. OECD (n 5) 117 (European Union); and Amicus Curiae (n 75) para 36. OECD (n 5) 117 (European Union); and Amicus Curiae (n 75) para 37. Amicus Curiae (n 75) paras 37–39. OECD (n 5) 117 (European Union). BPCL/ICI (n 34). ibid para 35.

Crisis cartels  267 The Commission noted that the petrochemical sector was characterized by excess of capacity, and that the agreement would allow the two firms to reduce production costs. It also found that, even after the agreement and the elimination of one producer, consumers would still have the choice of supplier.94 Finally, concerning the third condition of Article 101(3) TFEU (according to which consumers must receive a fair share of the resulting benefits), the Commission recalls that pass-on benefits must at least compensate for any potential negative impact caused by the reduction of capacity.95 The Commission then specifies the nature of the cost benefits and the degree of competitive constraints on the market, which are the two other elements it will take into account in the assessment of the pass-on defence. First, as regards the nature of cost benefits, the Commission explains that reductions of variable costs are more likely to lead to a fair share of cost efficiencies benefiting to consumers than reductions of fixed costs. This is because, as a general rule, pricing decisions of profit-maximizing firms are not determined by fixed costs but by variable costs.96 In this context, the Commission will adopt a case-by-case approach in order to identify cost benefits for consumers, as capacity reduction agreements generally tend to reduce the fixed cost component of unit costs.97 Second, the Commission recalls that the degree of competitive constraints on market players will be assessed in view of (i) actual competition, (ii) potential competition and (iii) buyer power. In this regard, a restructuring agreement must not go beyond a simple reduction in capacity and directly lead to the withdrawal of certain players. In addition, in order not to hinder potential competition, a restructuring agreement must not increase barriers to entry in a sector which involves high fixed or sunk costs.98 Finally, on buyer power, the Commission notes that undertakings with excess capacity will generally be subject to greater competitive pressure from purchasers than undertakings with low overcapacity. As a result, a capacity-reducing agreement would generally constrain buyers because of the decrease in the supply on the market. However, the Commission concludes that this will depend on the nature of the market in question.99 C. Sanctions The existence of a crisis is an element that competition authorities may consider when setting the amount of fines. For instance, the Commission has often taken into consideration the economic situation of the industry where the infringement occurred.100 In the Seamless Steel Tubes decision, the Commission held that the crisis affecting the steel pipe and tube industry in the 1970s could be seen as an attenuating circumstance justifying a 10 per cent reduction in the See ENI/Montedison (n 34); and Enichem/ICI (n 34). Amicus Curiae (n 75) para 40. 96 ibid para 41. 97 OECD (n 5) 119 (European Union). 98 Amicus Curiae (n 75) paras 45–46. 99 ibid para 47. 100 See, e.g., Case IV/30.064, Cast Iron and Steel Rolls, Commission decision, 17 October 1983 [1983] OJ L317/1, paras 72–75; and Case IV/30.988, Agreements and Concerted Practices in the Flat-Glass Sector in the Benelux Countries, Commission decision, 23 July 1984 [1984] OJ L212/13, para 55. 94 95

268  Research handbook on cartels amount of the fine.101 More recently, in the French Beef case, the Commission applied a 30 per cent reduction due to the encouragement of the cartel by the French government and a further 60 per cent reduction linked to long-term effects of the mad cow crisis on the relevant market, which went beyond a straightforward collapse in price.102 However, according to the Court of Justice, the Commission is not required, when determining the amount of the fine, to take into account the poor financial situation of an undertaking, since recognition of such an obligation would be tantamount to giving unjustified competitive advantages to undertakings least well adapted to the market conditions.103

In our view, this stance is over-simplistic. By definition, in a crisis context, the parties’ competitors are subject to the same structural crisis, and the fact that they did not participate in the cartel does not mean that they thrive or that they are necessarily more adapted to the market conditions. The fact remains that, as such, the existence of a crisis does not constitute a mitigating circumstance. This is also true in other jurisdictions. For instance, in the US, while the US Sentencing Guidelines take into account the defendant’s ability to pay, this is linked to its particular situation and not to general economic conditions.104

V.

CONCLUSION – THE GRIM FUTURE OF CRISIS CARTELS

In 1985, the Commission held that a system whereby inefficient firms are kept afloat is against the public interest.105 Although competition authorities, including the Commission, have not always been adverse to crisis cartels, this behaviour has become extremely difficult to justify. In the EU, for instance, the Commission has adopted only two decisions exempting crisis cartels implemented in the form of restructuring agreements, and it is far from obvious that it would adopt the same decisions if it had to rule on these cases today. Mergers seem to constitute a more robust way to deal with a structural crisis and, in particular, overcapacity (provided that sufficient individual economic incentives exist to implement a concentration). In any event, crisis cartels appear to be increasingly difficult to implement even under the form of a restructuring agreement. First, globalization makes the successful implementation of a crisis cartel very difficult. Even if such a cartel is authorized in one jurisdiction, which as noted above has become an ambitious assumption, it may be prohibited in another jurisdiction 101 Case IV/E-1/35.860-B Seamless Steel Tubes, Commission decision, 8 December 1999 [2003] OJ L140/1, paras 168–69. 102 Case COMP/C.38.279/F3, French Beef, Commission decision, 2 April 2003 [2003] OJ L209/12, paras 176–85. 103 Joined Cases 96/82 etc IAZ International Belgium and others v Commission [1983] ECR 3369, paras 54–55; and Joined Cases C-189/02 P etc Dansk Rørindustri and others v Commission [2005] ECR I-5425, para 327. 104 OECD (n 5) 216 (United States), referring to US Sentencing Guidelines, Guidelines Manual (2010), § 8C3.3. See now Sentencing Guidelines, Guidelines Manual (2010), § 8C3.3, which enshrine the same principle. 105 Case IV/30.307, Fire Insurance (D), Commission decision, 5 December 1984 [1985] OJ L35/20, para 44.

Crisis cartels  269 where the cartel also has effects. Second, it is now undoubtedly riskier to implement a crisis cartel than in the 1980s, since: (i) the repression of cartels has become a clear enforcement priority, with less stability for cartelists due to the rise of leniency programmes and higher fines associated to a finding of infringement; and (ii) agreements between competitors (like any other agreements) can no longer be notified to the Commission for approval. One must never say never, but the exemption of crisis cartels has likely become a thing of the past.

PART III PROCEDURAL ISSUES

16. The difficulties of proving an unlawful cartel Fernando Castillo de la Torre1

I. INTRODUCTION Proving the existence of a cartel poses special problems for the competition law enforcer. Cartels are typically unlawful in all systems and national laws do not normally provide for any exceptions. Participants understand that their conduct is unlawful, and so they take pains to conceal it. If an investigation into their conduct is undertaken, the participants usually do not co-operate with it, except through a leniency programme. Cartels can take significant measures to conceal their crimes:2 discussions are conducted over payphones and untraceable burner phones; face-to-face meetings use offsite locations; participants may register in hotels under false names or assign code names to firms or individuals within the cartel. Probably the most common way in which illegal contacts are concealed is by organizing them in conjunction with trade association meetings. They require the destruction of incriminating evidence, and it is not uncommon to see communications that terminate with ‘destroy this email after reading’. Therefore, the task of finding the evidence can be daunting. Fact-finding nowadays may require the most advanced technological means to review the voluminous electronically stored information in the undertakings’ files. The two main challenges for the decision-maker in this area are obtaining evidence and proving the infringement to the requisite legal standard.3 The ways to obtain the evidence are regulated differently in the different legal systems, and may vary depending on whether the enquiry is carried out in a civil, administrative or criminal setting. In turn, how high the standard of proof is will determine how far the investigative efforts of the claimant must go. One crucial aspect will be whether case law in each jurisdiction requires direct evidence of

All views expressed are personal to the author and should not be attributed to the European Commission (‘the Commission’) or its Legal Service. 2 C. Leslie, ‘How to Hide a Price-Fixing Conspiracy’ [2021] 4 University of Illinois Law Review 1199. 3 See, for a comparative view, OECD, Prosecuting Cartels without Direct Evidence (DAF/ COMP/GF(2006)7, 11 September 2006); and OECD, ‘Prosecuting Cartels without Direct Evidence of Agreement’ (2007) 9(3) OECD Journal of Competition Law and Policy 49. On specific legal systems, ABA Antitrust Law Section, Proof of Conspiracy under Federal Antitrust Laws (2nd edn, 2019); F. Castillo de la Torre and E. Gippini Fournier, Evidence, Proof and Judicial Review in EU Competition Law (Edward Elgar Publishing, 2017); M. Brealey and K. George (eds), Competition Litigation: UK Practice and Procedure (2nd edn, Oxford University Press, 2019) Chapters 10, 12 and 13; K. Schmidt, ‘GWB § 57’, in U. Immenga and E.J. Mestmäcker (eds), Wettbewerbsrecht (6th edn, 2020) paras 1–38; L. Idot, ‘Réflexions sur l’Évolution de la Preuve des Pratiques Anticoncurrentielles Devant les Autorités de Concurrence’ [2017] 4 Concurrences 45; K. Arai, ‘Indirect Evidence in Japanese Cartel Control’ (2015) 46 IIC – International Review of Intellectual Property and Competition Law 340; and A. Lamadrid de Pablo and A. Balcells, ‘La Prueba de los Cárteles en Derecho Español’, in J. Porras, J.M. Beneyto and J. Maillo (eds), La Lucha Contra los Cárteles en España (Aranzadi, 2015).  1

271

272  Research handbook on cartels the infringement, and to what extent indirect or circumstantial evidence may be enough to convince, ultimately, the judge who will have to deal with the case. At the same time, proving a cartel in most jurisdictions is facilitated by the fact that provisions of competition laws prohibiting anticompetitive agreements apply not only to explicit agreements but also to other types of joint arrangements, variously identified as ‘arrangements’, ‘combinations’ or ‘concerted practices’. However, regardless of the precise terminology, liability for a competition law violation can be imposed only if it can be shown that the parties reached some ‘conscious commitment to a common scheme’.4 The focus of cartel litigation is generally whether or not the claimant can prove an agreement or unlawful coordination.5

II.

BURDENS OF PROOF: SHIFTS AND INFERENCES

Issues of burden and standard of proof typically arise in competition cases generally and in cartel cases in particular. The rules differ depending on whether the case concerns a civil claim or is of criminal or administrative nature. As regards the latter, insofar as, for cartels, administrative enforcement generally implies the imposition of penalties, the burden and standard tend to be quite similar to that applicable in a criminal context. The legal burden of proof is the obligation of a party to meet the requirement of a rule of law that a fact in issue be proved to the satisfaction of the court. The burden of proof is said to frequently act as a burden of persuasion: even if the law frequently states who should prove what (such as ‘the plaintiff will prove the facts representing the basis of his claim’), in fact either by way of case law or by way of legislation the rules on burden of proof are standards for the allocation, between the parties, of the negative effects of the final lack of proof of material facts. In other words, they are aimed at providing the court, which must always decide the case and may not end up with a non liquet decision, with criteria that the court is to apply in its final decision-making, at the point at which it finds out that some material facts have not been proved.

OECD (2007) (n 3) 60–61. In the EU, case law has softened such difficulties by embracing a wide concept of ‘agreement’ and ‘concerted practice’. An agreement within the meaning of Article 101(1) of the Treaty on the Functioning of the European Union (‘TFEU’) can be regarded as having been concluded where there is a concurrence of wills on the very principle of a restriction of competition, even if the specific features of the restriction envisaged are still under negotiation; see Case T-9/99, HFB and others v Commission, EU:​T:​2002:​70, paras 151–57 and 206; Case T-240/07, Heineken Nederland v Commission, EU:​T:​2011:​ 284, para 45; and Case T-444/14, Furukawa Electric v Commission, EU:​T:​2018:​454, para 54. There is no need to demonstrate the precise mechanism by which the restrictive object was sought: Joined Cases T-67/00 etc, JFE Engineering and others v Commission, EU:​T:​2004:​221, para 203; and there is no need to show effects as regards agreements or concerted practices which have as their object the restriction of competition. The concept of concerted practice does in fact imply the existence of reciprocal contacts, and the condition of reciprocity is met where one competitor discloses its future intentions or conduct on the market to another when the latter requests it or, at the very least, accepts it. See Joined Cases T-25/95 etc, Cimenteries CBR and others v Commission, EU:​T:​2000:​77, paras 1849 and 1887. The exchange does not need to be reciprocal, in the sense of both sides exchanging the information; see Case T-54/03, Lafarge v Commission, EU:​T:​2008:​255, para 458; and Case T-377/06, Comap v Commission, EU:​T:​ 2011:​108, para 70. 4 5

The difficulties of proving an unlawful cartel  273 Typically, in civil and criminal cases the burden is for the plaintiff or prosecutor. In the EU, the burden of proving an infringement was typically borne by the Commission: it is incumbent on the Commission to prove the infringements which it has found and to adduce evidence capable of demonstrating to the requisite legal standard the existence of circumstances constituting the infringement. […] In doing this, the Commission must establish in particular all the facts enabling the conclusion to be drawn that an undertaking participated in such an infringement and that it was responsible for the various aspects of it.6

This is now ‘codified’ in Article 2 of EC Regulation 1/2003,7 which places the burden of proof on the claimant who is asserting that a decision, agreement or concerted practice infringes Article 101(1) TFEU. This general regime applies both to public authorities and to private claimants. Even if it is quite uncommon to claim that a cartel is to be exempted under Article 101(3) TFEU, it must be recalled that the same Article 2 provides that the burden of proving the four conditions for exemption lies with the party claiming exemption.8 Whereas the legal burden of proof is normally determined by law, the different shifts in the ‘evidential’ one are normally refined in the case law. The evolution of EU law could not escape this phenomenon. However, this reality covers different types of situation, which are not to be treated in the same way. In the area of competition law, the leading case where this was made explicit is Aalborg, pre-dating the adoption of Regulation 1/2003: it should be for the party or the authority alleging an infringement of the competition rules to prove the existence thereof and it should be for the undertaking or association of undertakings invoking the benefit of a defence against a finding of an infringement to demonstrate that the conditions for applying such defence are satisfied, so that the authority will then have to resort to other evidence. Although according to those principles the legal burden of proof is borne either by the Commission or by the undertaking or association concerned, the factual evidence on which a party relies may be of such a kind as to require the other party to provide an explanation or justification, failing which it is permissible to conclude that the burden of proof has been discharged.9

These two paragraphs make a clear difference between ‘defences’, where the burden appears to be more squarely on the party invoking it (as later codified in Regulation 1/2003), and shifts of what is sometimes called the ‘evidential burden of proof’. This ‘evidential’ burden of proof, or the ‘burden of production’, is the basis for the use of a number of evidentiary tools, such as factual presumptions. Jurists approaching this issue from a continental tradition take a somewhat similar view, even if the language used is not always identical. They tend to distinguish between the objective and the subjective burden of proof.10 The allocation of the evidential (or

Case C-49/92 P, Commission v Anic, EU:​C:​1999:​356, para 86; see also Case T-132/07, Fuji Electric v Commission, EU:​T:​2011:​344, para 84. 7 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 L1/1. 8 On the burden and the standard of proof for the application of Article 101(3) TFEU, see Castillo de la Torre and Gippini Fournier (n 3) paras 3.093–3.110. 9 Joined Cases C-204/00 P etc, Aalborg Portland and others v Commission, EU:​C:​2004:​6, para 79; see also Case C-609/13 P, Duravit v Commission, EU:​C:​2017:​46, para 57. 10 On the burden of proof as an objective rule to decide in case of uncertainty, see A-L. Sibony, Le Juge et le Raisonnement Économique en Droit de la Concurrence (LGDJ, 2008) para 826. Advocate General Kokott has differentiated the categories at stake: ‘The burden of proof determines, first, which 6

274  Research handbook on cartels subjective) burden of proof leads to situations where certain facts may be said to be proven in the absence of alternative explanations by the undertaking. It is generally accepted that these shifts of the evidential burden on to the other party have to do more with the standard of proof than with the legal burden of proof.11 Accordingly, courts and competition authorities make inferences all the time. This role is particularly salient for cartels, as the authorities do not always possess a clear evidential record or documentary evidence of the infringements. Hence, they should be capable of supplementing the evidential record with inferences that allow the relevant circumstances to be reconstituted. Yet, inferences cannot go too far. The EU judicature has stressed in several judgments that ‘the participation of an undertaking in a cartel cannot be inferred from speculation based on imprecise evidence’.12 Certain inferences, amounting to a shift in the evidential burden of proof, have developed in the area of cartels. Some inferences are qualified as ‘presumptions’, even if that category is full of nuances.13 One important presumption in cartel cases concerns the link between participation in anticompetitive meetings and adherence to the anticompetitive practice. It is sufficient for the administrative authority to show that the undertaking concerned participated in meetings at which anticompetitive agreements were concluded, without manifestly opposing them, to prove its participation in the cartel. In a shift in the evidential burden of proof, it is then for that undertaking to put forward evidence to establish that its participation in those meetings was without any anticompetitive intention by demonstrating that it had indicated to its competitors that it was participating in those meetings in a spirit that was different from theirs.14 The reason underlying that principle is that, having participated in the meeting without publicly distancing itself from what was discussed, the undertaking has given the other participants to believe that it subscribed to what was decided there and would comply with it. A party which tacitly approves of an unlawful initiative effectively encourages the continuation of the

party must put forward the facts and, where necessary, adduce the related evidence (subjektive or formelle Beweislast, also known as the evidential burden); second, the allocation of that burden determines which party bears the risk of facts remaining unresolved or allegations unproven (objektive or materielle Beweislast)’ (Case C-8/08, T-Mobile Netherlands, EU:​C:​2009:​110, Opinion of AG Kokott, footnote 60). As she also explained in her Opinion in FEG: ‘before there is any need to allocate the burden of proof at all, each party bears the burden of adducing evidence in support of its respective assertions. A substantiated submission by the Commission can be overturned only by an at least equally substantiated submission by the parties. […] [I]f in its decision the Commission draws conclusions as to the conditions prevailing in a particular market on the basis of objectively verifiable evidence from stated sources, the undertakings concerned cannot refute the Commission’s findings simply by unsubstantiatedly disputing them. Rather, it falls to them to show in detail why the information used by the Commission is inaccurate, why it has no probative value, if that is the case, or why the conclusions drawn by the Commission are unsound. This requirement does not represent the reversal of the burden of proof … but the normal operation of the respective burdens of adducing evidence’ (Case C-105/04 P, FEG v Commission, EU:​C:​ 2005:​751, Opinion of AG Kokott, paras 73–74). 11 Case C-97/08, Akzo Nobel v Commission, EU:​C:​2009:​262, Opinion of AG Kokott, para 74. 12 See Case T-68/09, Soliver v Commission, EU:​T:​2014:​867, para 58; Case T-104/03, Toshiba v Commission, EU:​T:​2015:​610, para 50; and Case T-772/15, Quanta Storage v Commission, EU:​T:​2019:​ 519, para 93. 13 A. Kalintiri, ‘Analytical Shortcuts in EU Competition Enforcement: Proxies, Premises and Presumptions’ (2020) 16(3) Journal of Competition Law & Economics 392. 14 Aalborg Portland and others v Commission (n 9) para 81. For different cases of application, see Castillo de la Torre and Gippini Fournier (n 3) paras 2.072–2.074.

The difficulties of proving an unlawful cartel  275 infringement and jeopardizes its discovery.15 Other presumptions and inferences are typically present in cartel cases.16

III.

ADMISSIBILITY, EVIDENCE ASSESSMENT AND STANDARD OF PROOF

Before the judge must decide the case by applying the burden of proof, other concepts come into play. Proving a cartel implies first gathering admissible evidence. Evidence must be presented to the fact-finder or be gathered by the relevant authority; that is, it must find its way into the file. Evidence must also be admissible according to the rules of the relevant legal system. Exclusionary rules have been widely considered a hallmark of Anglo-American evidence, but they are less prominent in continental law systems and also in EU practice. ‘Extrinsic’ exclusionary rules, that is, rules rejecting probative information for the sake of values unrelated to the pursuit of truth (such as certain privileges, or the exclusion of illegally obtained evidence), exist in some form in EU law. However, the ‘intrinsic’ exclusionary rules, that is, rules which try to ensure accuracy of fact-finding (such as the hearsay rule), do not find a real analogue in continental trials and in EU law. This is not to say that the dangers related to some type of evidence will be ignored, but the objection will not be structured as an exclusionary rule: caution will be applied at the time of assessing the weight of the evidence.17 In principle, in the EU, the fact that the Commission is bound to respect the general principles of law, including fundamental rights, which are recognized as part of the Union legal order means, inter alia, that evidence that has been collected or processed in contravention of fundamental rights is in principle inadmissible.18 Yet, in practice it is not so clear whether

Aalborg Portland and others v Commission (n 9) paras 82, 84. See Castillo de la Torre and Gippini Fournier (n 3) paras 2.069–2.078 and 2.088–2.093. 17 For details, see M. Damaška, ‘Of Hearsay and Its Analogues’ (1992) 76 Minn L Rev 425, 444–56. The same applies to character evidence or similar information about a person’s past life. Continental evidentiary theory focuses only on whether information from a person’s past has probative value: if it is probative, it can and should be used in adjudication: M. Damaška, Evidence Law Adrift (Yale University Press, 2013) 17. 18 See Joined Cases 46/87 and 227/88, Hoechst v Commission, EU:​C:​1989:​337, para 34; Case C-94/00, Roquette Frères, EU:​C:​2002:​603, para 49; Case T-410/09, Almamet v Commission, EU:​T:​ 2012:​676, para 39; and Case T-54/14, Goldfish BV and others v Commission, EU:​T:​2016:​455, para 47. For further discussion see Castillo de la Torre and Gippini Fournier (n 3) paras 4.011–4.020. 15 16

276  Research handbook on cartels unlawfully obtained evidence must always be excluded,19 and the case law of the EU judicature has become more nuanced over time.20 If evidence is admitted, one must then examine the ‘strength’ of that evidence, where the focus is on how the decision-maker weighs the evidence in reaching the verdict. In this connection, two main properties of evidence must be distinguished.21 The first property is the probative value of the evidence. Evidence must first be assessed in order to determine its probative value. However, this assessment is not done in the abstract and it is relational to what the party intends to prove: evidentiary inferences and arguments must be developed (when it is a party) and assessed (by the decision-maker) to determine the evidentiary support to the conclusions. Evidence does not normally ‘speak by itself’, and therefore parties and the decision-maker must construct evidentiary arguments based on the items presented. Here courts have typically developed common principles which actually apply in different contexts, and they are not necessarily unique to competition law. In a comparative scheme, assessment criteria can be ordered according to the degree of freedom they confer upon the fact-finders, stipulating that a criterion is ‘objective’ if it decreases such freedom and is ‘subjective’ if it increases it. This is, however, not a categorical distinction, it is a matter of degree. In other words, objective criteria are evidence-centred since they focus on the evidence at the fact-finder’s disposal, while subjective criteria are evaluator-centred since they focus on the states of mind and attitudes of the fact-finder (on prudence, conscience, conviction, etc.).22 Whether such conceptualizations have a real impact on the practice of the courts remains to be seen. 19 This is not an outright consequence of the European Convention on Human Rights (‘ECHR’), which appears to focus more on how the evidence is used as opposed as how the evidence was obtained. This is the distinction between Beweiserhebung and Beweisverwertung, which is well developed in German legal doctrine. On the ECHR issue, see I. Vilsmeier, Tatsachenkontrolle und Beweisführung im EU-Kartellrecht auf dem Prüfstand der EMRK (Mohr Siebeck, 2014) 139–40. The relevant question is whether the proceedings as a whole, including the way in which the evidence was obtained, were fair. This involves an examination of the alleged unlawfulness in question and, where the violation of another Convention right is concerned, the nature of the violation found (Khan v United Kingdom, Application No. 35394/97, ECHR 2000-V, § 34; and Allan v United Kingdom, Application No. 48539/99, ECHR 2002-IX, § 42). As to the examination of the nature of the ECHR violation found, the question whether the use as evidence of information obtained in violation of Article 8 ECHR rendered a trial as a whole unfair contrary to Article 6 ECHR has to be determined with regard to all the circumstances of the case, and in particular to the question of respect for the applicant’s defence rights and the quality and importance of the evidence in question: Gäfgen v Germany, Application No. 22978/05, ECHR 2010 § 165. 20 It is apparent from the Court’s case law that in deciding whether to exclude information and evidence obtained in contravention of the requirements of EU law, regard must be had, in particular, to the risk of breach of the adversarial principle and, therefore, the right to a fair trial entailed by the admissibility of such information and evidence (see Goldfish BV and others v Commission, n 18, paras 50–87; Joined Cases C‑511/18 etc, La Quadrature du Net and others, EU:​C:​2020:​791, para 226; and Case C-276/01, Steffensen, EU:​C:​2003:​228, paras 76 and 77). 21 A third property is the ‘degree of completeness’. Not only must the evidence be assessed in a holistic way (see below), but, more fundamentally, the decision-maker has a duty to examine all relevant circumstances when adopting its decision. The weight of the allegedly incriminating evidence will be assessed not only against the potential exculpatory evidence in the file, but bearing in mind the evidence that the decision-maker could have obtained. 22 The criterion of the ‘intime conviction’ is not purely subjective. The relevant texts tend to underscore the need to fix that state of mind on the basis of the evidence. It does not boil down to a purely subjective and psychological feeling about the defendant’s guilt.

The difficulties of proving an unlawful cartel  277 While most civil law countries usually have in their respective codes some basic rules about evidence and proof, they do not tend to be very detailed. Most of the time these rules set evidence assessment criteria instead of standards of proof. Nowadays these criteria are rather relaxed, leaving room to judicial discretion. This trend is historically explained by the nowadays widely accepted principle of ‘free evaluation’ of the evidence. The EU judicature has held on a number of occasions that the prevailing principle in Community law is the ‘unfettered evaluation of evidence [‘libre administration de la preuve’ in French],23 the only relevant criterion for such evaluation being the reliability of the evidence’.24 The establishment of general rules is inimical to such a principle. The principle of unfettered evaluation implies precisely the scarcity of such rules.25 The credibility, and thus the probative value, of a document depends on its origin, the circumstances in which it was drawn up, the person to whom it is addressed and the soundness and reliable nature of its content.26 Accordingly, even a single piece of evidence can be sufficient to establish an infringement provided that the probative value is such that it attests definitively to the existence of the infraction.27 The freedom in the evaluation of the evidence frees the fact-finder from legal rules, but it does not imply licence to disregard extralegal canons of valid inference and is usually combined in domestic law with the obligation to duly reason the factual findings.28 As an essential guarantee of a fair trial, courts are under an obligation to explain in detail why and how their factual findings stem from the evidence presented during the proceedings.29

23 See, inter alia, Case C-407/14 P, Dalmine v Commission, EU:​C:​2007:​53, para 63; and Joined Cases T-426/10 etc, Moreda-Riviere Trefilerías and others v Commission, EU:​T:​2016:​335, para 110. 24 Case C-411/04 P, Salzgitter Mannesmann v Commission, EU:​C:​2007:​54, para 45; Joined Cases C-239/11 P etc, Siemens and others v Commission, EU:​C:​2013:​866, para 128; Case T-58/14, Stührk Delikatessen Import v Commission, EU:​T:​2018:​474, paras 71–72; and Case T-433/16, Pometon v Commission, EU:​T:​2019:​201, para 115. 25 Siemens and others v Commission (n 24) paras 190–91. The principle of free evaluation of the evidence appeared as a rebellion against the principle of legal assessment, which prescribed the ways and means in which each element was to be proven. In Continental Europe, the rebellion rested in important part on the belief that the probative weight of evidence is a matter too unruly to obey, and too contextual to be captured in a web of categorical legal norms. In its most pristine historical form, the principle of free evaluation of the evidence (‘free proof’) was conceived to be so radical that it demanded not only freedom from legal chains in analysing evidence but also freedom from all inter-subjectively ascertainable standards governing the fact-finding decision: the adjudicator’s inner persuasion (his ‘conviction intime’) was thought sufficient to justify the verdict. 26 Stührk Delikatessen Import (n 24) paras 72 and 73 and the case law cited. 27 Cimenteries CBR and others v Commission (n 5) para 1848. 28 The appellate review forces the adjudicator to justify its findings in a written opinion. The hierarchical supervision will typically give rise to authoritative statements on the adequacy of the evidentiary support for the factual findings of the court below, and although the prevailing legal doctrine refuses generally to acknowledge that these statements (expect at most those imposed by compliance with fundamental rights) can be the font of binding legal norms, the standards contained in these statements are actually observed. The fear of reversal exerts a degree of constraining influence on the first instance judge’s freedom. However, most of the time these standards are highly contextual and they seldom harden into sharp-edged rules. See Damaška (2013) (n 17) 21. 29 How the EU courts comply with this obligation depends on each case. Certain judgments are more open about qualifying certain pieces of evidence as having particularly high, high, moderate or low probative value, but this intrinsic value is always based, as any evidentiary issue, on generalizations. Other judgments stress the overall strength of the evidence as a whole, rather than that of the different pieces of evidence. When examined in isolation, each piece of evidence is included in a given ‘category’, to

278  Research handbook on cartels The second property relates to the ‘sufficiency’ of the evidence, generally articulated as the ‘standard of proof’. Fact-finders need to consider whether the evidence meets the relevant standard of proof. The standard of proof rules set the qualitative or quantitative thresholds that must be reached to have a decision in favour of the party which bears the burden of proof. This party bears the risk that the amount of evidence is insufficient with respect to the standard. So, standard of proof rules are, in the end, also decision rules. They justify decisions in favour of one party or the other, based on the factual claims and the arguments given the evidence.30 The standards of proof can be ordered according to the degree of belief on the claim at stake (qualitative reading) or according to the probability value that the claim is attributed (quantitative reading). Different degrees of belief would correspond to three theoretical basic standards: ‘beyond reasonable doubt’, ‘clear and convincing evidence’ and ‘preponderance of the evidence’. In civil cases, whereas the common law tradition applies the balance of probabilities,31 the continental tradition tends to be more demanding.32 In criminal law, the ‘beyond reasonable doubt’ is commonly used throughout the world. When sanctions are imposed outside criminal proceedings, the applicable standard tends to be close or identical to the one applied in criminal proceedings. Accordingly, in EU law, the ‘benefit of the doubt’ is now systematically recalled as regards the existence of the relevant conduct.33 The case law has also referred, and still refers, to the need to convince the judge, which is the ultimate aim which certain attributes are normally associated. As for any generalization based on experience, parties may provide indicia that the assumption on which the generalization is made does not apply in the instant case. Using case-specific evidence, one party can challenge the existence of the background conditions under which a generalization becomes operative. That is why, even when relying on generalizations, the EU judicature often leaves the door open for the existence of specific circumstances (Castillo de la Torre and Gippini Fournier (n 3) paras 2.102 and 4.023–4.024). 30 Standards of proof are not criteria of evidence assessment. Instead, they presuppose and require some criteria that determine the probative force, or probative value, of the evidence in play, in order to decide whether it meets the relevant threshold. Evidence must be assessed in order to check whether it satisfies a relevant standard of proof. G. Tuzet, ‘Evidence Assessment and Standards of Proof: A Messy Issue’ [2020] 2 Quaestio Facti 87, 90, explains that this difference is not always recognized and it is not uncommon that some legal scholars, practitioners and decision-makers miss the difference between assessment criteria and standards. The common law world shows a prominent concern for standards of proof, and this finds an explanation in the features of jury trial (ibid 104). 31 For the different burdens in different types of litigation within the same legal system, see, for example, for the United Kingdom, Brealey and George (n 3) Chapter 10. 32 Authors often assume that the standard applied in civil proceedings is that of balance of probabilities. However, this is not necessarily the standard applied in many legal systems of the member states of the EU. For a lively debate on this, see K. Clermont and E. Sherwin, ‘A Comparative View of Standards of Proof’ (2002) 50 American Journal of Comparative Law 243; M. Taruffo, ‘Rethinking the Standard of Proof’ (2003) 51 American Journal of Comparative Law 659; and K. Clermont, ‘Standards of Proof Revisited’ (2009) 33 Vermont Law Review 469. See also C. Engel, ‘Preponderance of Evidence versus Intime Conviction – A Behavioural Perspective on a Conflict between American and Continental European Law’ (2009) 33 Vermont Law Review 435, making a similar point: ‘in principle, continental law does not make a difference between civil law and criminal law.’ 33 Joined Cases C-403/04 P and C-404/04 P, Sumitomo Metal Industries and Nippon Steel v Commission, EU:​C:​2007:​52, para 52; Case T-110/07, Siemens AG v Commission, EU:​T:​2011:​68, para 44; Case T-348/08, Aragonesas v Commission, EU:​T:​2011:​621, paras 93–94; Case T-377/06, Comap v Commission, EU:​T:​2011:​108, para 56; Pometon v Commission (n 24) para 108; Case T-240/17, Campine v Commission, EU:​T:​2019:​778, para 108; and Case T‑105/17, HSBC Holdings v Commission, EU:​T:​ 2019:​675, 204, 265. On the evolution of the case law on this point, see Castillo de la Torre and Gippini Fournier (n 3) paras 2.021–2.031.

The difficulties of proving an unlawful cartel  279 of the evidence, and which is formulated with the expression frequently used of achieving a ‘firm conviction’ (‘intime conviction’ in the French version).34 The EU judicature has developed specific case law as regards proof of certain aspects. Beyond the proof of the cartel as such, many other aspects may have to be proven depending on the case. The burden of proof of duration lies with the claimant, even if case law has softened the burden in certain cases.35 The claimant may have to prove that a line of conduct by multiple participants may be considered to be a single and continuous infringement.36 The claimant may have to prove that the defendant is liable for the infringement as well.37

IV.

A HOLISTIC ASSESSMENT OF THE EVIDENCE

One important issue that affects how courts evaluate the different types of evidence, and in particular indirect or circumstantial evidence, is whether the court is willing to consider all evidence that is proffered as a whole, giving it cumulative effect, or whether it requires that each item unequivocally supports the hypothesis of agreement. In High Fructose Corn Syrup Judge Posner strongly adopted the holistic approach. The trial judge in the case had refused to consider the communication evidence that was offered because he thought its character was such as to ‘require that a substantial inference be drawn in order to have evidentiary significance’. This is correct in the sense that no single piece of the [communication] evidence […] is sufficient in itself to prove a price-fixing conspiracy. But that is not the question. The question is simply whether this evidence, considered as a whole and in combination with the economic evidence, is sufficient to defeat summary judgment.38

It should be noted, however, that there are different views even within the US. Other federal appeals courts have examined each item of circumstantial evidence independently as to whether it tended to exclude independent, unilateral action of competitors.39 In the EU, it is not necessary for every item of evidence produced by the Commission to satisfy those criteria in relation to every aspect of the infringement. It is sufficient if the body of evidence relied on by the Commission, viewed as a whole, meets that requirement.40 The items of evidence must not be assessed separately, but as a whole.41 The outcome of any evaluative process will be a complex one, as the specific factual matrix of each case will be JFE Engineering and others v Commission (n 5) para 179. Castillo de la Torre and Gippini Fournier (n 3) paras 3.022–3.085. 36 ibid paras 3.001–3.021. 37 ibid para 2.078, with analysis of all different scenarios dealt with by EU courts so far. See also, more recently, Case C-882/19, Sumal, EU:​C:​2021:​800. 38 In re High Fructose Corn Syrup Antitrust Litig., 295 F.3d 651 (7th Cir. 2002) 661. 39 See Williamson Oil Co. v Philip Morris USA, 346 F.3rd 1287 (11th Cir. 2003). 40 Case T-448/14, Hitachi Metals v Commission, EU:​T:​2018:​442, para 118 and the case law cited therein. 41 Goldfish and others v Commission (n 18) para 93 and the case law cited therein. In the cartel context, the holistic approach implies that the authority may rely on pieces of evidence dated before or after the period for which the Commission finally finds the infringement, insofar as the document is useful to interpret facts relating to that period; see Case T-655/11, FSL and others v Commission, EU:​T:​2015:​383, paras 178, 217; Lafarge v Commission (n 5) para 428; Pometon v Commission (n 24) para 113 and 263; Case T-8/16, Toshiba Samsung v Commission, EU:​T:​2019:​522, paras 270­–76, 364; 34 35

280  Research handbook on cartels necessarily different. Precedents need to be relied upon with the outmost caution. Documents with low intrinsic probative value may become important if they match other documents and the impact that each piece of evidence will have on the overall assessment crucially depends as well on factors external to it. The soundness and reliability of a piece of evidence is proportional to the extent to which that evidence is consistent with other known facts.

V.

DIRECT VS CIRCUMSTANTIAL OR INDIRECT EVIDENCE

A. General One of the traditional evidentiary debates regarding cartel conduct is whether direct evidence is necessary or if circumstantial evidence may instead be enough. While the debate has developed a bit differently in each jurisdiction, depending on the features of its own evidentiary rules as well as the specific type of enforcement at issue (criminal, administrative or civil), the issues tend to be the same conceptually. There is a trend in OECD countries towards building cases based on direct evidence. However, countries continue to bring cases employing mostly circumstantial evidence where it is appropriate.42 It is likely that countries just beginning an anti-cartel programme will have some difficulty generating direct evidence, and hence will have to rely more on circumstantial evidence in early cases. Further to the distinction between direct and circumstantial evidence, there is a second dichotomy between contemporaneous evidence (emails or documents drafted in close connection with the events) and ex post evidence (prepared for the purpose of the proceedings, typically statements by individuals or companies). Direct evidence tends to demonstrate directly the relevant fact, whereas indirect evidence tends to prove a fact with which the relevant fact bears a logical relationship. Common types of direct evidence include a document or documents (including email messages) essentially embodying the agreement, or parts of it, and identifying the parties to it, oral or written statements by co-operative cartel participants describing the operation of the cartel and their participation in it.43 Circumstantial evidence, in turn, consists of ‘communication’ evidence and ‘economic evidence’, which include firm conduct, market structure and evidence of facilitating practices. Indirect evidence can rarely be used in isolation and usually requires corroboration or is itself used to corroborate other (possibly also indirect) evidence. Yet, under certain circumstances, cartel participants can be prosecuted, even against the highest standards of proof, without direct evidence of the agreement or of their involvement in it.44 Indirect (circumstantial) evidence is used in most jurisdictions, including those Case T-475/14, Prysmian v Commission, EU:​T:​2018:​448, 207–16, upheld by Court of Justice in Case C-601/18 P, Prysmian v Commission, EU:​C:​2020:​751, paras 141–43. 42 OECD (2007) (n 3) 58–60. 43 ‘Direct evidence is evidence which requires no mental process on the part of the tribunal of fact in order to draw the conclusion sought by the proponent of the evidence, other than acceptance of the evidence itself. Circumstantial evidence is evidence from which the desired conclusion may be drawn, but which requires the tribunal of fact not only to accept the evidence presented, but also to draw an inference from it’: P. Murphy, Murphy on Evidence (8th ed., Oxford University Press, 2003) 23. 44 ‘Classic criminal proceedings allow for the use of circumstantial evidence and recourse to principles derived from experience’: Case C-8/08 P, T-Mobile Netherlands, EU:​C:​2009:​110, Opinion of AG Kokott, para 93.

The difficulties of proving an unlawful cartel  281 with the longest and most successful records of cartel prosecutions, where competition law enforcers now enjoy the virtuous circle of strong sanctions in past cases energizing leniency programmes, which generate direct evidence in new cases and stronger sanctions. Indirect evidence is particularly important for competition law enforcers in jurisdictions without such enforcement records, given that cartel conspiracies are cloaked in secrecy and direct evidence is not forthcoming.45 There are different types of circumstantial or indirect evidence typically used in cartel cases. One is evidence that cartel operators met or otherwise communicated, but which does not describe the substance of their communications. This is often referred to as ‘communication’ evidence, and includes records of telephone conversations between competitors (but not their substance), or of travel to a common destination or of participation in a meeting, for example during a trade conference. It also includes other evidence that the parties communicated about the subject, such as minutes or notes of a meeting showing that prices, demand or capacity utilization were discussed, or internal documents evidencing knowledge or understanding of a competitor’s pricing strategy, such as an awareness of a future price increase by a rival.46 A broader category of circumstantial evidence is often called ‘economic’ evidence. Economic evidence identifies primarily firm conduct that suggests that an agreement was reached, but also conduct of the industry as a whole, elements of market structure which suggest that secret price fixing was feasible and certain practices that can be used to sustain a cartel agreement. Insofar as it covers actual conduct in the market, the economic nature of this type of evidence comes from the fact that it requires inferences premised on economic reasoning, but conduct evidence does not have, as such, an economic nature. Reliance on that type of evidence may require advice from economic experts.47 Conduct evidence includes, first and foremost, parallel pricing, such as changes in prices by rivals that are identical, or nearly so, and simultaneous, or nearly so.48 Industry performance could also be described as conduct evidence. It includes abnormally high profits or stable market shares. Evidence related to market structure can be used primarily to make the finding of a cartel agreement more plausible, even though market structure factors do not prove the existence of such an agreement. Over the years, courts, competition authorities and competition experts have come to accept that ‘conscious parallelism’, which involves nothing more than identical pricing or other parallel behaviour deriving from independent observation and reaction by rivals in the marketplace, is not unlawful.49 This view is well grounded in economic theory. Economic theory and case law have made it clear that something more than conscious parallelism is required. Defining that ‘something more’ has proved difficult; courts in a few jurisdictions have wrestled with the problem for decades. Some type of conduct will be more difficult to explain by anything but concerted action; that is the case of parallel boycotts or individual bid-rigging actions that are repeated.50 OECD (2007) (n 3) 58. ibid 61–63. In the EU, some of this conduct is more than just circumstantial evidence and can be considered by itself a concerted practice; see n 78 below. 47 On economic expertise, see, for example, Brealey and George (n 3) Chapter 13; and Castillo de la Torre and Gippini Fournier (n 3) paras 4.105–4.124 and 5.059–5.074. 48 It includes other forms of parallel conduct, such as capacity reductions, adoption of standardized terms of sale, and suspicious bidding patterns, e.g., a predictable rotation of winning bidders. 49 OECD (2007) (n 3) 70. 50 Arai (n 3) 346–47. 45 46

282  Research handbook on cartels B.

The United States

The US Supreme Court has famously said that ‘“circumstantial evidence is the lifeblood of antitrust law” because direct evidence will rarely be available to prove the existence of a price-fixing conspiracy’.51 Although the Federal Rules of Evidence do not mention direct or circumstantial evidence, courts rely on the distinction in a variety of contexts, including in Section 1 conspiracy cases.52 The reason why the dichotomy of direct versus circumstantial evidence is important, as opposed to the sufficiency of the evidence, lies in the role of the court in evaluating the legal sufficiency of evidence, especially on motions for summary judgment.53 In the US, the use of circumstantial evidence may depend on whether the case is criminally prosecuted or not. Only in unusual circumstances will the Department of Justice (‘DOJ’) proceed with a criminal prosecution when direct evidence is lacking.54 US courts have defined direct evidence in the context of antitrust conspiracies as ‘evidence that is explicit and requires no inferences to establish the proposition or conclusion being asserted’.55 Beyond such descriptions of direct evidence, in the US many courts define the concept of direct evidence by listing examples of it.56 In reality, direct evidence is not a monolithic category in US law. Direct evidence exists along a continuum of strong to weak: from admissions of guilt to a recorded phone call between competitors in which the defendants agreed to fix prices, to hand-written notes detailing inter-competitor conversations. Moreover, the difference between direct and indirect evidence is not always clear.57 Many have pointed out that all evidence depends