Conceptualising Procedural Fairness in EU Competition Law 9781509935413, 9781509935444, 9781509935437

Citation preview

CONCEPTUALISING PROCEDURAL FAIRNESS IN EU COMPETITION LAW What constitutes a fair procedure when it comes to EU competition law? This innovative book seeks to understand the philosophical considerations at the core of conflicting procedural fairness arguments in EU competition law practice. The author argues for a conceptualisation of procedural fairness as a distributional issue that can be solved by a practical fairness theory and a comprehensive methodology. To illustrate the usefulness of the conceptualisation, three procedural fairness problems from recent EU competition law practice are analysed: • the KME–Chalkor cases; • the Groupe Gascogne cases; • the regulatory question about using a collective redress mechanism for private enforcement of EU competition law. This unique approach provides a robust philosophical and methodological foundation for arguing about a wide range of procedural fairness dilemmas. The book is a must-read for academics and practitioners seeking an imaginative perspective on the philosophical foundations of arguments about procedural fairness in EU competition law and beyond. Volume 26 in the series Hart Studies in Competition Law

Hart Studies in Competition Law Media Ownership and Control: Law, Economics and Policy in an Indian and International Context Suzanne Rab and Alison Sprague The Interface between Competition and the Internal Market: Market Separation under Article 102 TFEU Vasiliki Brisimi Anti-Cartel Enforcement in a Contemporary Age: Leniency Religion Edited by Caron Beaton-Wells and Christopher Tran Public Procurement and the EU Competition Rules Albert Sánchez Graells The Concept of Abuse in EU Competition Law: Law and Economic Approaches Pınar Akman The Competitive Effects of Minority Shareholdings: Legal and Economic Issues Panagiotis Fotis and Nikolaos Zevgolis The More Economic Approach to EU Antitrust Law Anne C Witt Private Power, Online Information Flows and EU Law Angela Daly The Role of Competitors in the Enforcement of State Aid Law Fernando Pastor-Merchante The Legality of Bailouts and Buy Nationals: International Trade Law in a Crisis Kamala Dawar A Critical Account of Article 106(2) TFEU: Government Failure in Public Service Provision Jarleth Burke Dawn Raids Under Challenge Helene Andersson A Framework for European Competition Law: Co-ordinated Diversity Christopher Townley Evidence Standards in EU Competition Enforcement: The EU Approach Andriani Kalintiri The Metaphysics of Market Power: The Zero-sum Competition and Market Manipulation Model George Raitt Competition Law’s Innovation Factor: The Relevant Market in Dynamic Contexts in the EU and US Viktoria HSE Robertson Competition, Effects and Predictability: Rule of Law and the Economic Approach to Competition Bruce Wardhaugh Fighting Cross-Border Cartels: The Perspective of the Young and Small Competition Authorities Pierre Horna

Conceptualising Procedural Fairness in EU Competition Law Haukur Logi Karlsson

HART PUBLISHING Bloomsbury Publishing Plc Kemp House, Chawley Park, Cumnor Hill, Oxford, OX2 9PH, UK 1385 Broadway, New York, NY 10018, USA HART PUBLISHING, the Hart/Stag logo, BLOOMSBURY and the Diana logo are trademarks of Bloomsbury Publishing Plc First published in Great Britain 2020 Copyright © Haukur Logi Karlsson, 2020 Haukur Logi Karlsson has asserted his right under the Copyright, Designs and Patents Act 1988 to be identified as Author of this work. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage or retrieval system, without prior permission in writing from the publishers. While every care has been taken to ensure the accuracy of this work, no responsibility for loss or damage occasioned to any person acting or refraining from action as a result of any statement in it can be accepted by the authors, editors or publishers. All UK Government legislation and other public sector information used in the work is Crown Copyright ©. All House of Lords and House of Commons information used in the work is Parliamentary Copyright ©. This information is reused under the terms of the Open Government Licence v3.0 (http://www.nationalarchives.gov.uk/doc/ open-government-licence/version/3) except where otherwise stated. All Eur-lex material used in the work is © European Union, http://eur-lex.europa.eu/, 1998–2020. A catalogue record for this book is available from the British Library. Library of Congress Cataloging-in-Publication data Names: Haukur Logi Karlsson, 1979– author. Title: Conceptualising procedural fairness in EU Competition Law / Haukur Logi Karlsson. Description: Oxford, UK ; New York, NY : Hart, 2020. | Series: Hart studies in competition Law ; volume 26  |  Based on author’s thesis (doctoral - European University Institute, 2017) issued under title: A quantitative quest for philosophical fairness in EU’s competition procedure,  |  Includes bibliographical references and index. Identifiers: LCCN 2020016740 (print)  |  LCCN 2020016741 (ebook)  |  ISBN 9781509935413 (hardcover)  |  ISBN 9781509935437 (ePDF)  |  ISBN 9781509935420 (Epub) Subjects: LCSH: Antitrust law—European Union countries. Classification: LCC KJE6456 .H379 2020 (print)  |  LCC KJE6456 (ebook)  |  DDC 343.2407/21—dc23 LC record available at https://lccn.loc.gov/2020016740 LC ebook record available at https://lccn.loc.gov/2020016741 ISBN: HB: 978-1-50993-541-3 ePDF: 978-1-50993-543-7 ePub: 978-1-50993-542-0 Typeset by Compuscript Ltd, Shannon To find out more about our authors and books visit www.hartpublishing.co.uk. Here you will find extracts, author information, details of forthcoming events and the option to sign up for our newsletters.

ACKNOWLEDGEMENTS This book is based on the PhD thesis I wrote at the European University Institute (EUI) between 2012 and 2017. Since then my thinking has evolved on several issues, which is reflected in numerous changes and amendments. The book is a different work from the thesis in terms of structure and presentation, and it contains new research elements in chapters two, six and seven. During the EUI years my thoughts about the main problems benefited from the teachings of professors Giovanni Sartor, Joseph Weiler, Dennis Patterson and, most notably, my supervisor, Giorgio Monti, who stimulated my thinking and disciplined my writing through his thoughtful comments. The book also benefits from suggestions made by the examining board of the PhD thesis, who consisted of professors Davíð Þór Björgvinsson, Dennis Patterson and Ioannis Lianos. Professor Oddný Mjöll Arnardóttir provided invaluable assistance during the early stages by kindly helping to secure research grants to finance the PhD work and by co-supervising it. Judge Páll Hreinsson was generous to invite me to the EFTA Court for a few months in early 2016, which helped me develop a practical understanding of some of the key issues. The impact of Zane Rasnaca, Alastair Maciver and Juha Tuovinen should also be noted. As fellow researchers at the EUI, they provided moral encouragement and companionship during the years of research and writing in the Badia Fiesolana complex of the EUI. The work on the research was funded by the Icelandic Research Fund (Project Grants no 141274-051, 141274-052 and 141274-053, and Postdoctoral Fellowship Grants no 184905-051 and 184905-052). The book is dedicated to my wife, Áslaug Dögg Karlsdóttir, to our children, Aría and Kári, and to the boy we are expecting in the spring. Haukur Logi Karlsson Reykjavík, 31 January 2020

vi

TABLE OF CONTENTS Acknowledgements��������������������������������������������������������������������������������������������������������v 1. The Problem of Procedural Fairness������������������������������������������������������������������1 I. Procedural Fairness in EU Competition Law���������������������������������������������2 A. The KME/Chalkor Argument���������������������������������������������������������������4 B. The Elements of KME/Chalkor Dilemma�������������������������������������������7 C. The KME/Chalkor Dilemma in Context�������������������������������������������10 II. Towards a Conceptualisation of Procedural Fairness������������������������������17 2. The Essence of a Procedure���������������������������������������������������������������������������������20 I. Differentiating Procedural Law and Substantive Law�����������������������������20 II. Procedures as a Social Institution��������������������������������������������������������������23 III. The Institutional Elements of an Adjudicative Procedure����������������������26 A. Instrumental Structure������������������������������������������������������������������������26 i. Constitutive Rules�����������������������������������������������������������������������27 ii. Culture������������������������������������������������������������������������������������������29 iii. Sanctions��������������������������������������������������������������������������������������33 B. Normative Function����������������������������������������������������������������������������36 i. The Deontological View�������������������������������������������������������������37 ii. The Consequential View������������������������������������������������������������39 iii. The Functional Essence of a Procedure������������������������������������43 IV. Towards a Conceptualisation of a Fair Adjudicative Procedure������������44 3. The Structure and Function of EU Competition Procedures����������������������46 I. The Normative Purpose of EU Competition Law������������������������������������47 II. Stakeholders and Stakes in Competition Proceedings����������������������������51 A. The Main Stakeholders������������������������������������������������������������������������52 B. The Main Stakes�����������������������������������������������������������������������������������54 III. The Structure of the EU’s Competition Law Procedure��������������������������56 IV. Instrumental Structure and Normative Function������������������������������������61 A. The Constitutive Rules of EU Competition Law Procedure����������62 B. The Institutional Culture of DG Competition���������������������������������66 C. The Procedure before the CJEU���������������������������������������������������������68 D. Institutional Sanctions������������������������������������������������������������������������69 V. Arguments about Procedural Fairness in EU Competition Law�����������71

viii  Table of Contents 4. Ethics for Procedural Architects�����������������������������������������������������������������������75 I. Ethics and Legal Theory������������������������������������������������������������������������������75 II. Ethics beyond Legal Theory������������������������������������������������������������������������80 A. The Social Contract Narrative������������������������������������������������������������81 B. Hobbesian Ethics and Game Theory�������������������������������������������������86 C. Contractarian Ethics and Democratic Institutions�������������������������90 III. Ethics as the Dark Matter of the Law���������������������������������������������������������94 5. The Model of Fair Rules��������������������������������������������������������������������������������������96 I. The Role of the Rule Architect��������������������������������������������������������������������97 II. Rulemaking as a Decision Problem���������������������������������������������������������100 III. The Elements of the Model of Fair Rules������������������������������������������������101 A. The Policy Objective of a Rule���������������������������������������������������������102 B. Unification of Moral and Efficiency Claims�����������������������������������103 C. Interpersonal Comparison of Stakeholder Claims������������������������105 D. Registration of Preferences into Pay-off Matrices�������������������������111 E. Identifying Fair and Efficient Rules�������������������������������������������������112 IV. The Utility of a Model for Identifying Fair Rules�����������������������������������115 6. Analysis of Three Fairness Dilemmas������������������������������������������������������������118 I. The KME/Chalkor Cases���������������������������������������������������������������������������120 A. Analysis of the Cases�������������������������������������������������������������������������122 B. Subsequent Developments���������������������������������������������������������������126 C. The Court’s Approach to the Fairness Issue�����������������������������������127 II. The Groupe Gascogne Cases����������������������������������������������������������������������128 A. Analysis of the Cases�������������������������������������������������������������������������130 B. The Court’s Approach to the Fairness Issue�����������������������������������133 III. Collective Redress for Private Enforcement of Competition Law����������������������������������������������������������������������������������������135 A. Analysis of the Legislative Dilemma�����������������������������������������������137 B. The Commission’s Approach to the Fairness Issue������������������������140 IV. Symptoms of an Underdeveloped Theory of Procedural Fairness��������������������������������������������������������������������������������������������������������144 7. An Argument for a Data-based Democracy�������������������������������������������������146 I. Conceptualisation of Procedural Fairness in EU Competition Law����������������������������������������������������������������������������������������146 II. Solving Fairness Issues in EU Competition Procedure�������������������������149 A. Applying the Model of Fair Rules����������������������������������������������������151 i. Step 1: Identifying the Primary Policy Objective������������������152 ii. Step 2: Preference Function�����������������������������������������������������154 iii. Step 3: Preference Index�����������������������������������������������������������156

Table of Contents  ix iv. Step 4: Preference Matrices������������������������������������������������������159 v. Step 5: Identifying Fairness������������������������������������������������������160 B. Assessing the Utility of the Model���������������������������������������������������163 III. Technocracy and the Making of Laws�����������������������������������������������������164 IV. Data-based Democracy for Regulating EU Competition Procedure����������������������������������������������������������������������������������������������������167 Bibliography���������������������������������������������������������������������������������������������������������������172 Index��������������������������������������������������������������������������������������������������������������������������177

x

1 The Problem of Procedural Fairness Legal practitioners often overlook the importance of the conceptual link between laws and the reigning political philosophy of the society they serve. When entangled in the process of solving legal problems as private practitioners, or as official decision-makers of public institutions, the law may seem a closed system that abides exclusively to its own internal logic, unaffected by external social and political realities. Nevertheless, laws undeniably tend to mirror the human societies they serve. Novel technologies prompt legislative changes, major societal events are dealt with by amending the law, and gradual cultural progressions only become permanently cemented through alterations in society’s legal fabric. Laws do not only mirror what society is, they also reflect what society wants to be. The law can thus depart from the actualities of reality, both due to a failure to adapt to what is, and due to societies’ failure to adapt to what ought to be, according to the desired reality prescribed by their laws. Human society is by its very nature subject to constant change, fuelled by the fluidity of social interactions and the randomness of their consequences. The black letter laws are comparatively static; reminiscent of a snapshot at a given moment in time that mirrors what society once was, or a reflection of what society once wanted to be. The tension between the dynamic nature of society and the static character of the law creates multiple dilemmas and paradoxes for practitioners. This tension can be ignored by considering the domain of law as being a closed system of logical arguments that is self-sufficient in providing solutions to any question of law. By ignoring the temporal tension, laws will however fail to reach their full potential in their mission to guide and reflect the society that they are meant to serve. Acknowledgment of the temporal tension reveals weaknesses in the traditional juridical method and suggests an opportunity for methodological improvements. It is particularly difficult to reconcile the two-pronged purpose of the law, ie providing a sturdy framework for optimal behaviour whilst adjusting flexibly to the contemporary mood. A lax attitude towards the former risks complacency with socially suboptimal behaviour, while rigidness towards the later risks imposition of behavioural standards no longer deemed socially desirable. This problem can take many forms in legal practice and has been written about extensively in legal theory, often implicitly in the context of the doctrines of legal positivism (ie the law is a closed system), legal naturalism (ie the law is a part of a larger system of ethics), and legal realism (ie the law is what it is).

2  The Problem of Procedural Fairness The problem of fairness in the law is a variant of the temporal problem. Fairness can both be perceived as a fixed standard of optimality that society seeks to achieve through its laws and as a flexible standard whose substantive meaning is sensitive towards the current social temperament. It should be safe to presume a consensus about the normative requirement that the law ought to be fair. Determining objectively whether a specific act of law, or a specific legal action constitutes fairness is a harder task. The purpose of this book is to explore the general problem of fairness in EU laws, from the special point of view of the EU’s competition law procedure. The requirement of procedural fairness in Union law has legal basis in the Charter of Fundamental Rights (CFR), for example in Articles 41, 47 and 48. The Court of Justice of the European Union (CJEU) has also on many occasions referred to Article 6 of the European Convention on Human Rights (ECHR) as a standard for assessing compliance with procedural fairness. The approach to the question of fairness is initially posed as a practical problem within EU competition law, but will steadily gravitate towards an abstract conceptualisation using insights from legal and political philosophy, and from the theory of economics. In the end several practical problems from EU competition law enforcement are analysed and approached based on a theoretical conceptualisation. The basic idea of the book is to explore the idea of procedural fairness as a ­philosophical concept and see if such an understanding is capable of bringing clarity to contemporary problems of procedural fairness in EU competition law practice, some of which have been inconclusively debated for a long time. The book is thus not intended as a strictly legal doctrinal approach to problems of procedural fairness, but rather as a philosophical basis for such doctrinal approaches. As a starting point, elements of the academic discussion about procedural fairness in EU competition law will be drawn up, before engaging with the more fundamental philosophical issues in the background.

I.  Procedural Fairness in EU Competition Law Practitioners of EU law have for decades criticised the institutional scheme for the enforcement of EU competition law. Originally designed as an administrative procedure in the 1960s for the enforcement of a seemingly obscure field of Union law, the procedural regime contained in Regulation 17 and the succeeding Regulation 1/2003 has since been the backdrop for many of the most epic court cases in the brief history of the EU. In the early 2010s, litigators tested an argument about the compatibility of the institutional scheme with regards to recent developments towards a gradual criminalisation of competition law breaches and the increased prominence of individual procedural rights following the elevation of the CFR to a Union constitutional status.

Procedural Fairness in EU Competition Law  3 The KME/Chalkor litigation raised an issue about the compatibility of the EU’s institutional approach to competition law enforcement with human rights obligations to ensure procedural fairness.1 The academic community identified a potential problem prior to the litigation, scrutinised it closely once it got to court, and has reflected upon it afterwards in what has become a sizable literature.2

1 See three judgments delivered on 8 December 2011: Case C-272/09 P KME Germany and Others v Commission, EU:C:2011:810; C-386/10 P Chalkor v Commission, EU:C:2011:815; C-389/10 P KME Germany and Others v Commission, EU:C:2011:816. 2 See some of the main contributions: Denis Waelbroeck and Denis Fosselard, ‘Should the decisionmaking power in EC antitrust procedures be left to an independent judge? – The impact of the European Convention of Human Rights on EC antitrust procedures’ (1994) 14 Yearbook of European Law 111; Frank Montage, ‘The case for a radical reform of the infringement procedure under regulation 17’ (1996) 17 European Competition Law Review 428; Koen Lenaerts and Jan Vanhamme, ‘Procedural rights of private parties in the community administrative process (1997) 34 Common Market Law Review 531; Wouter PJ Wils, ‘The combination of the Investigative and prosecutorial function and the adjudicative function in EC antitrust enforcement: a legal and economic analysis’ (2004) 27 World Competition 201; Ólafur Jóhannes Einarsson, ‘EC competition law and the right to a fair trial’ (2006) 25 Yearbook of European Law 555; Eric Barbier de La Serre, ‘Procedural justice in the European Community case-law concerning the rights of the defence: essentialist and instrumental trends (2006) 12 European Public Law 225; Donald Slater, Sébastien Thomas and Denis Waelbroeck, ‘Competition law proceedings before the European Commission and the right to a fair trial: no need for reform?’ (2008) The Global Competition Law Centre Working Paper Series – Working Paper No. 04/08 (also published as Donald Slater, Sébastien Thomas and Denis Waelbroeck, ‘Competition law proceedings before the European Commission and the right to a fair trial: no need for reform?’ (2009) 5 European Competition Journal 97); Ian S Forrester, ‘Due process in EC competition cases: a distinguished institution with flawed procedures’ (2009) 34 European Law Review 817; Fernando Castillo de la Torre, ‘Evidence, proof and judicial review in cartel cases’ (2009) 32 World Competition 505; Wouter PJ Wils, ‘The increased level of EU antitrust fines, judicial review and the ECHR’ (2010) 33 World Competition 5; Jaime Flattery, ‘Balancing efficiency and justice in EU competition law: elements of procedural fairness and their impact on the right to a fair hearing’ (2010) 7 The Competition Law Review 53; Wouter PJ Wils, ‘EU anti-trust enforcement powers and procedural rights and guarantees: the interplay between EU law, national law, the Charter of Fundamental Rights of the EU and the European Convention on Human Rights’ (2011) 34 World Competition 189; Marc Jaeger, ‘The standard of review in competition cases involving complex economic assessments: towards the marginalisation of the marginal review? (2011) 2 Journal of European Competition Law & Practice 295; Peter Oliver, ‘”Diagnostics” – a judgment applying the Convention of Human Rights to the field of competition’ (2012) 3 Journal of European Competition Law & Practice 163; Anne-Lise Sibony, ‘Annotation of the judgment of the Court in Case C-272/09 P KME Germany and others v. Commission’ (2012) 49 Common Market Law Review 1977; Marco Bronckers and Anne Vallery, ‘Fair and effective competition policy in the EU: which role for authorities and which role for the courts after Menarini?’ (2012) 8 European Competition Journal 283; Renato Nazzini, ‘Administrative enforcement, judicial review and fundamental rights in EU competition law: a comparative contextual-functionalist perspective’ (2012) 49 Common Market Law Review 971; Heike Schweitzer, ‘Judicial review in EU competition law’ in Damien Geradin and Ioannis Lianos (eds), Handbook on European competition law (Cheltenham, Edward Elgar Publishing, 2013) 491; Ingrid Vandenborre and Thorsten Goetz, ‘EU competition law procedures’ (2012) 3 Journal of European Competition Law & Practice 578; Nils FW Hauger and Christoph Palzer, ‘Investigator, prosecutor, judge … and now plaintiff? The Leviathanian role of the European Commission in the light of fundamental rights’ (2013) 36 World Competition 565; Eric Barbier de la Serre, ‘Standard of review in competition law cases: Posten Norge and beyond’ in Carl Baudenbacher, Philipp Speitler and Bryndís Pálmarsdóttir (eds), The EEA and the EFTA Court: decentred integration: to mark the 20th anniversary of the EFTA Court (Oxford, Hart Publishing, 2014); Wouter PJ Wils, ‘The compatibility with fundamental rights of the EU antitrust enforcement system in which the European Commission acts both as investigator and as a first-instance decision maker’ (2014) 37 World Competition 5.

4  The Problem of Procedural Fairness The views in the literature mirror the opposing claims of the parties to the case before the CJEU. Parallel to the KME/Chalkor litigation, similar issues were tried before the other EEA court on the Kirchberg plateau in Luxemburg and before the European Court of Human Rights (ECoHR) in Strasbourg. The EFTA Court’s judgment in Posten Norge was delivered five months after the CJEU’s judgment, but the ECoHR’s judgment in Menarini was delivered two months prior to the CJEU’s judgment in KME/Chalkor.3

A. The KME/Chalkor Argument Leading up the KME-Chalkor litigation, parallel developments had occurred in two unrelated fields of EU law that, when taken together, gave rise to doubts about the compatibility of EU’s competition procedure with obligations to ensure procedural fairness. More specifically, these doubts concerned the procedure for imposing fines in cartel cases and the way in which the Court exercised its power to review such decisions. The first development arose following the modernisation of the competition enforcement regime with the entering into force of Regulation 1/2003, and is well explained by data published by the Directorate General (DG) Competition.4 Table 1.1  Cartel cases where fines were imposed by the Commission 1990–2019 (amounts adjusted for Court judgments – updated 7 November 2019)5 Period

Fine amounts (€)

Number of cases

Average fine per case (€)

1990–94

344,282,550

10

34,428,255

1995–99

270,963,500

9

30,107,056

2000–04

3,157,348,710

29

108,874,093

2005–09

7,863,307,787

33

238,282,054

2010–14

7,604,840,879

30

253,494,696

2015–19

8,234,322,023

27

304,974,890

27,475,065,449

138

199,094,677

Total

3 See judgment of the EFTA Court of 18 April 2012 in Case E-15/10 Posten Norge v EFTA Surveillance Authority [2012] EFTA Ct Rep 246; and judgment of the ECoHR of 27 September 2011 in A Menarini Diagnostics Srl v Italy App no 43509/08, CE:ECHR:2011:0927JUD004350908. 4 Council Regulation (EC) No 1/2003 on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty [2003] OJ L 1/1. 5 Source: http://ec.europa.eu/competition/cartels/statistics/statistics.pdf (updated 7 November 2019). This table merges tables 1.4 and 1.10 in DG Competition’s document and adds a calculation of average fine per case. The amounts concern fines imposed for infringements against Art 101 TFEU and its predecessors, ie ex Art 81 EC and ex Art 85 EC. In this table, a cartel case concerns a single proceeding against various undertakings concerned and may involve more than one infringement. Only those cartel cases where a fine was imposed were considered for the purpose of this table.

Procedural Fairness in EU Competition Law  5 In the 1990s, the average cartel fine amounted to approximately €30 million. By the early–mid 2000s, the fines had increased substantially and had by the late 2010s reached an average of €305 million.6 This increase coincided with the modernisation programme of the late 1990s and early 2000s, and was supported by a conscious use by DG Competition of condemning rhetoric to raise the social stigma of engaging in a conduct prohibited by the EU’s competition provisions.7 An example of this can be seen in a speech given by Mario Monti, then Commissioner for Competition, in Stockholm in September 2000: Cartels are cancers on the open market economy, which forms the very basis of our Community. By destroying competition they cause serious harm to our economies and consumers. […] In the words of Adam Smith there is a ‘tendency for competitors to conspire’. This tendency is of course driven by the increased profits that follow from colluding rather than competing. We can only reverse this tendency through tough enforcement that creates effective deterrence. The risk of being uncovered and punished must be higher than the probability of earning extra profits from successful collusion.8

Describing the conduct of breaching Article 101 of the Treaty on the Functioning of the EU (TFEU) as the infliction of a cancer on the economy and accusing the entities engaged in such activity of conspiracy against the general public was probably not what the original signatories of the EEC Treaty of Rome envisioned when they delegated the supposedly minor administrative issue of enforcing the competition provisions of the Treaty of Rome to the Commission in the early 1960s, but is consistent with a contemporary trend in competition law enforcement in many jurisdictions.9 The later development, which raised worries about the EU’s competition procedure’s compliance with fairness standards, concerned the constitutional project of the EU. In line with ordoliberal ideas popular at the time, the original EEC Treaty 6 Note that the numbers say nothing about the seriousness of the underlying breach, or the size of the firms involved, which will influence the size of the fine and thus influence how comparable each instance is with another. Commission officials have nonetheless conceded that there has been a noticeable increase in the size of cartel fines in recent decades. See Wils, ‘The compatibility with fundamental rights of the EU antitrust enforcement’ (2014) 8; Wils, ‘The increased level of EU antitrust fines’ (2010) 10–12. See also Castillo de la Torre, ‘Evidence, proof and judicial review’ (2009) 506. Wils however maintains that the increase is irrelevant for the purposes of assessing compliance with Art 6(1) ECHR, since the Engel criteria assesses the maximum permissible fine, the level of which has remained at 10% of annual corporate turnover since the implementation of Reg 17 in the 1960s. 7 For examples of this rhetoric see quotes listed in Slater, Thomas and Waelbroeck, ‘Competition law proceedings before the European Commission’ (2008) 14–15. 8 Mario Monti, ‘Fighting Cartels Why and How? Why should we be concerned with cartels and collusive behaviour?’ (2000) 3rd Nordic Competition Policy Conference, Stockholm, available at http://europa.eu/rapid/press-release_SPEECH-00-295_en.htm. 9 See further discussion on the intentions of the Treaty signatories with regards to the competition provisions in David J Gerber, ‘The Transformation of the European Community Competition Law?’ (1994) 35 Harvard International Law Journal 97, 103–07; and more generally on the history of EU’s competition law in David Gerber, Law and Competition in the Twentieth Century Europe: Protecting Prometheus (Oxford, Oxford University Press, 2001). On the existence of a worldwide trend of criminalising cartel offences see Peter Whelan, ‘Cartel criminalization and the challenge of “moral wrongfulness”’ (2013) 33 OJLS 535, 536.

6  The Problem of Procedural Fairness of Rome was often viewed as an economic constitution, referring to the imperativeness of the four free movement principles and the ancillary provisions on competition. Later, the Court gradually recognised more traditional rights principles as being part of the EU’s constitutional framework, despite their absence in the Treaty of Rome. This was in part due to a necessity following the establishment of the supremacy doctrine in Costa v Enel, which created the potential for EU laws to override constitutional rights in the Member States, including traditional rights provisions.10 In Internationale Handelsgesellschaft of 1972, the Court claimed that respect for fundamental rights forms an integral part of the general principles of law protected by the Court of Justice. The protection of such rights, whilst inspired by the constitutional traditions common to the Member States, must be ensured within the framework of the structure and objectives of the Community.11

From then on, it has been assumed that traditional rights form part of the constitutional framework of the EU, and since the judgment in Rutili in 1975, the rights under the ECHR have been explicitly acknowledged in the Court’s case law.12 Although tacitly recognised as being part of the primary law of the EU from the 1970s through the Court’s case law, an effort was not made to codify a clear list of rights for the EU until in 1999. Then the European Council decided to commission the drafting of a CFR to a senior body of representatives that adopted the name ‘the European Convention’. The Commission, the European Parliament and the Council of Ministers proclaimed the draft as the CFR in the Nice summit of 2000, but simultaneously decided that ‘the question of the Charter’s force [should] be considered later’.13 An updated version of the Charter was supposed to become part of the European Constitution of 2004, which failed in the ratification process.14 A reference was also made to an updated version of the Charter in the Lisbon Treaty of 2007, with the intention of giving it a comparable hierarchal legal status as the founding Treaties of the Union. When the Lisbon Treaty entered into force on 1 December 2009, the CFR thus acquired a Treaty status within the EU system of laws, which as a consequence made traditional rights for the first time an explicit part of the codified constitutional framework of the EU.15 For the purposes of the KME/Chalkor litigation, this gradual constitutional development with regards to traditional rights, gave rise to an argument that the 10 Case 6/64 Costa v ENEL, EU:C:1964:66. 11 Case 11/70 Internationale Handelsgesellschaft mbH v Einfuhr- und Vorratsstelle für Getreide und Futtermittel, EU:C:1970:114. 12 Case 36/75 Roland Rutili v Ministre de l’intérieur, EU:C:1975:137; the ECHR is explicitly mentioned at [32]. See further on the early development of fundamental rights protection within the EU in Einarsson, ‘EC competition law and the right to a fair trial’ (2006) 556–59. 13 European Council – Nice 7–10 December 2000: Conclusions of the presidency, available at www.europarl.europa.eu/summits/nice1_en.htm. 14 Treaty establishing a Constitution for Europe [2004] OJ C310/1. 15 Charter of Fundamental Rights of the European Union [2010] OJ C 83/391.

Procedural Fairness in EU Competition Law  7 institutional arrangement of competition enforcement, which was instituted by the Treaty of Rome in 1957 and by Regulation 17 in 1962, was no longer compatible with the recognition of the right to a fair procedure enshrined in the newly codified Article 47 CFR.16 Independently, these two separate but parallel developments – of criminalisation of competition law breaches and the constitutionalising of traditional rights – could each warrant a reconsideration of whether the institutional architecture of competition law enforcement was compatible with the current norms of procedural fairness. When combined, these distinct developments formed a powerful argument that required close attention by the stakeholders of the competition law enforcement regime. Arguably, the threshold of rights protection had risen over time: first following the implicit recognition of rights in the EU system of law; and later through an explicit codification. At the same time, the protective interests had also increased: on one hand as the result of increased social stigma against competition law breaches; and on the other hand, as the result of increased economic consequences for those caught committing such breaches. Thinking about the standard of criminality and the standard of a moral entitlement to a fair process as two separate constants, against which the factual context of a case is assessed, the argument of the KME/Chalkor litigation was that the substance of the rights constant had changed and that the factual circumstances with regards to the standard of criminality had also changed, both of which undermined the constitutionality of EU’s procedural apparatus for enforcing competition law.

B.  The Elements of KME/Chalkor Dilemma The narrative about the gradual criminalisation of competition law breaches fits uncomfortably with the traditional categorisation of the adjudicative processes for dealing with potential breaches against public law, which normally are categorised as either criminal or administrative. Usually matters of lesser importance are dealt with administratively, which is cheaper and more efficient. Matters of greater importance for the state or the individuals involved are hence dealt with through the more expensive and cumbersome criminal procedure. A narrative that suggest a gradual change within a system of binary categorisation needs to explain precisely when and how an object ceases to be one thing and becomes the other thing, ie in this case, a criminal or administrative proceeding. In the abstract, the punishment for a breach against legally protected public interests is usually rationalised through one of two modes of arguing: either through a deontological argument, in which a criminal procedure would be warranted if the moral stigma of being found guilty of a breach is great (ie retributive rationale);

16 Council Regulation No 17 (EEC): First Regulation Implementing Articles 85 and 86 of the Treaty [1962] OJ No 013.

8  The Problem of Procedural Fairness or through a consequentialist argument, in which a criminal procedure would be warranted if the consequences at stake would be great for any of the stakeholders (ie deterrence rationale).17 The level of the seriousness of the punishment for being found in breach thus usually correlates with the level of the stigma, or alternatively with the level of the consequences of the breach. Importantly, this typical correlation is not a necessary relation. If a punishment is designed based on deterrence theory, it is possible that the most efficient level of deterrence is achieved through a low-level punishment, irrespective of the importance of the interests at stake for the relevant stakeholders.18 Leaving such extraordinary circumstance aside, it should in most cases be safe to assume that breaches against public interests that are subject to severe punishments should ideally be dealt with through a criminal procedure, rather than an administrative procedure. The two main arguments underpinning the KME/Chalkor cartel litigation, ie the gradual criminalisation and the rise of fundamental rights, suggest that a cartel is a breach against legally protected public interests. Breach against these public interests is both considered morally reprehensible (eg DG Competition’s rhetoric) and is subject to a sizable pecuniary penalty. Irrespective of whether the enforcement rationale is based on the theory of retribution or deterrence, it seems that in the abstract these kinds of breaches should fall within the criminal sphere of public law enforcement, rather than the administrative sphere. The exception noted above with regards to a deterrence rationale of a punishment seems inapplicable. The fine is high rather than low, which will always create the potential for grave consequences that require a cautious high-quality approach to prosecution and adjudication. The legal landscape in KME/Chalkor was, however, complicated by the historical origins of the enforcement procedure, which had initially been designed in the 1960s as an administrative procedure. Even after the Commission had sharpened its rhetoric against competition law breaches and multiplied the level of fines in the early–mid 2000s, the legislator explicitly maintained the non-criminal definition of the procedure for imposing fines in competition cases when regulation 17 was modernised. This can be seen in Article 23(5) of the successor Regulation 1/2003, which restates the repelled Article  15(4) of Regulation 17/62: ‘Decisions taken pursuant to paragraphs 1 and 2 shall not be of a criminal law nature.’19 The situation was thus that although the actualities of competition enforcement seemed to 17 See generally on the theory of punishment in criminal law in HLA Hart, ‘The Presidential Address: Prolegomenon to the Principles of Punishment’ (1959) 60 Proceedings of the Aristotelian Society 1. 18 Usually a severe punishment would warrant a criminal procedure due to the potential consequences for the accused, but it is also possible that a morally reprehensible crime does not need a hefty punishment for achieving a deterrence effect, in which case the argument for a criminal procedure would need to rest on a deontological rationale, rather than on a consequentialist rationale. Further on this point see Whelan, ‘Cartel criminalization’ (2013) 540–43. 19 See Council Regulation (EC) No 1/2003 on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty [2003] OJ L 1/1.

Procedural Fairness in EU Competition Law  9 warrant a criminal categorisation, the law was explicit on the point that it should not be considered as such. Developments in the interpretation of the ECoHR of how a distinction should be made between a criminal procedure and an administrative procedure brought further complications. For the purposes of the ECHR the distinction mattered. Article  6 exclusively sets the standard of procedural fairness for criminal and civil procedures, not for administrative procedures.20 This had gradually led the ECoHR to expand the concept of a criminal procedure to increase the applicability of the procedural guarantees of the ECHR. This policy had, however, expanded further than intended. Citizens had started to demand their day in court to argue over parking fines and other minor issues based on the provisions of the ECHR about procedural guarantees in criminal cases, thus forcing the use of expensive procedural mechanism in cases that hardly warranted such expenditures of public resources. The ECoHR attempted to strike a balance in this sense with the Jussila judgment.21 There it abandoned the binary approach to the categorisation of criminal and administrative procedures in favour of a gradual approach in which procedures could be somewhat criminal in nature, but not so that they necessitated the full set of procedural guarantees provided by Article 6 ECHR: There are clearly ‘criminal charges’ of differing weight. What is more, the autonomous interpretation adopted by the Convention institutions of the notion of a ‘criminal charge’ by applying the Engel criteria have underpinned a gradual broadening of the criminal head to cases not strictly belonging to the traditional categories of the criminal law, […]. Tax surcharges differ from the hard core of criminal law; consequently, the criminal-head guarantees will not necessarily apply with their full stringency.22

The third element in the problem in the KME/Chalkor litigation, in addition to gradual criminalisation of competition law and gradual rise of fundamental rights, was thus that the actual procedure for imposing fines in competition cases had been designed to comply with the more relaxed procedural fairness standard for administrative procedures. Additionally, it was not entirely clear how the ECoHR would categorise the EU’s enforcement procedures in the field of competition law and thus whether, or to what extent, the procedural guarantees stipulated in Article 6(1) ECHR and the accompanying case law would apply. In the KME/Chalkor litigation, the Court was confronted with elements that pointed towards contradictory solutions. Observing the temporal problem described earlier, there had been changes in how the Commission wanted competition law breaches to be perceived by the public over the decades since the enactment of Regulation 17. The gradual recognition, and later the codification of fundamental rights into the EU legal system, can be viewed as a response to

20 Arguably the distinction is less important for the purposes of EU law, since procedural fairness of administrative procedures is protected under Art 41 CFR. 21 Jussila v Finland App no 73053/01, CE:ECHR:2006:1123JUD007305301. 22 ibid [43].

10  The Problem of Procedural Fairness changes in the public perception and expectations about rights and ethics in the law, and thus how rigidly the standard of procedural fairness should be interpreted in the context of competition law enforcement. The social stigma, the hefty fines and the gradual rise and explicit recognition of the right to a fair procedure, did by the time of the KME/Chalkor litigation suggest that breaches to the competition law regime should be perceived as of a criminal nature and accordingly be dealt with based on the criminal procedural modality. At the same time, however, the written law of the procedural regulation explicitly ordered that the criminal procedural modality should not be used for imposing fines for breaches against the competition provisions of the TFEU, suggesting that such breaches were not serious enough to warrant the procedural guarantees reserved for those accused of the most serious breaches against public order. The actual problem of legal interpretation, which the Court was confronted with in the KME/Chalkor cases, was more specific and more detailed than is indicated by the description of the broad elements above. The three elements formed a legal-factual texture, which dictated the range of possibilities to answer the more specific challenge made against the Court’s doctrine regarding the standard of review of the Commission’s competition decisions. The specific issue of the KME/Chalkor cases was the question whether the CJEU’s doctrine of providing the Commission with a deference to assess complex economic facts in competition cases and thus limit its judicial review of the Commission’s decisions, was compatible with the right to a fair trial enshrined in Article 47 CFR. These more specific issues will be analysed further in chapter six. For the purposes of providing context to the book, it suffices to summarise the problem, the Court’s solution and to review the relevant academic literature.

C. The KME/Chalkor Dilemma in Context Doubts about the compatibility of the EU’s competition law regime with principles of procedural fairness can be traced back to the 1970s and the 1980s.23 For the purposes of the KME/Chalkor litigation the first important contribution to the academic literature was Waelbroeck and Fosselard’s article from 1994 in the Yearbook of European Law.24 The arguments raised in this article set the tone for the ensuing debate about procedural fairness in the EU’s competition law regime and its compatibility with the procedural safeguards articulated in Article  6 ECHR and later Article  47 CFR. Notably, the argument was raised in the article about the incompatibility of the CJEU’s doctrine of granting the Commission with deference in assessing complex technical and economic facts,



23 Forrester,

‘Due process in EC competition cases’ (2009) 819–20. and Fosselard, ‘EC antitrust procedures’ (1994).

24 Waelbroeck

Procedural Fairness in EU Competition Law  11 with the requirement of the ECHR that a tribunal with full jurisdiction should review criminal charges.25 Another important contribution from the initial phase of the debate was an article by Montage, where he articulated three problems from the point of view of the accused with the enforcement regime under Regulation 17 and suggested that the best remedy would be that the Court of First Instance (now the General Court) should take the initial prohibition decision in competition cases. Accordingly, he argued that DG Competition’s role in competition proceedings should end with the issuance of a statement of objections (SO).26 In an article from 1997 in the Common Market Law Review, a judge and a legal secretary at the Court of First Instance jointly responded to the charge that the alleged deference doctrine was incompatible with the requirements of Article 6 ECHR. They conceded that the Court had repeatedly stated in its judgments that there was a certain discretion enjoyed by the Commission, but they argued that the Court did nonetheless exercise a full jurisdiction within the meaning of the ECHR on the account of what the Court effectively did in such cases. The system could thus be viewed as compatible with the ECHR due to a paradox in how the Court expresses itself and in how it actually acts.27 In an article from 2006, Einarsson reaches a similar conclusion based on a detailed review of the case law of the ECoHR, the General Court and the CJEU. Einarsson argues that if EU courts did in the past not review sufficiently the Commission’s administrative process in competition cases, the current case law (ie in 2006) indicated that the level of scrutiny was in line with the requirements of Article  6 ECHR as interpreted by the ECoHR.28 By the time Einarson’s article was published in 2006, the academic debate had more or less settled on the view that: (i) the Commission’s competition law enforcement procedure was criminal in nature within the autonomous meaning of Article 6 ECHR; (ii) the EU courts would need to exercise full judicial review jurisdiction over the result of the administrative procedure before the Commission in order to comply with the requirements of the ECHR as interpreted in the case law of the ECoHR; and (iii) the actual review conducted by the EU courts was no longer subject to the deference doctrine originating in Consten and Grundig of 1966.29 A sequence of events in late 2006 and in 2007 gave rise to a reconsideration of whether the consensus still held. On 23 November 2006, the Grand Chamber of the ECoHR handed down the judgment in Jussila v Finland, where it reopened the question of which procedural guarantees should be applicable to procedures

25 ibid 125–33. 26 Montage, ‘Infringement procedure under regulation 17’ (1996). 27 Lenaerts and Vanhamme (1997) 560–62. The co-author of the article, Judge Lenaerts, is from 2015 the President of the CJEU. 28 Einarsson (n 2) 612–14. 29 Joined Cases 56 and 58/64 Consten and Grundig v Commission, EU:C:1966:41, p 347.

12  The Problem of Procedural Fairness that did not belong to the traditional categories of criminal law, but which did fall under the criminal heading of Article  6 ECHR based on the Engel criteria. The judgment did not concern a competition procedure, but the logic of the finding indicated that it could be applicable to EU’s competition procedure as well. On 17  September 2007, the Court of First Instance handed down the Microsoft judgment. Interestingly, for the purposes of compatibility with Article 6 ECHR, the deference doctrine for complex economic and technical matters was restated subject to certain conditions.30 Finally, on 13 December 2007 the Member States signed the Lisbon Treaty and thus signalling the eventual 1 December 2009 entering into force of the CFR, which codified procedural rights equivalent to those of Article 6 ECHR into the constitutional structure of the EU. In a much-cited article from 2008, Slater, Thomas and Waelbroeck restated many of the arguments initially raised by Waelbroeck and Fosselard in 1994.31 They argued that the seriousness of competition law breaches warranted an initial decision taken by an independent tribunal within the meaning of Article 6 ECHR and that in any case the judicial review conducted by the Court of First Instance was not a full jurisdiction review within the meaning of the ECHR. Forrester, Microsoft’s counsel in the Microsoft case and later a judge at the General Court, also criticised the Commission’s competition procedure in an article from 2009 in the European Law Review, arguing that the power over the result of a public prosecution should not be in the hands of the politically appointed Commissioners of the Commission, due to a risk of a political bias.32 The Commission’s officials were quick to respond to these renewed speculations and doubts about the fairness of the competition procedure. In 2009, Torre wrote an article in defence of the system in place and in 2010 Wils did the same.33 Wils’ main argument was made with reference to the Jussila judgment of the ECoHR. The competition procedure should not be considered to be of a hardcore criminal nature and thus the procedural guarantees of Article  6(1) ECHR should not apply with their full stringency. This, Wils argued, permits the first decision in competition cases to be reached through an administrative procedure. Addressing the issue of deference in light of the Microsoft case, Wils resorted to the same argument as Lenaerts and Vanhamme used in their Common Market Law Review article of 1997; the CFI in Microsoft first grants deference to the Commission in paragraphs 87 and 88, then takes most of it back in paragraph 89, and finally in practice executes a full judicial review leaving no deference to the Commission.34 Jaeger argued along similar lines in his 2011 article in the Journal of European Competition Law & Practice.35 30 Case T-201/04 Microsoft v Commission, EU:T:2007:289 [87]–[89]. The actual formulation is somewhat strange; the deference doctrine is articulated in [87] and [88], but then reduced and conditioned in [89] with reference to para 39 of Case C-12/03 P Commission v Tetra Laval, EU:C:2005:87. 31 Slater, Thomas and Waelbroeck (2008, n 2). 32 Forrester (n 2) 831–33. Forrester was also Chalkor’s counsel in the KME/Chalkor litigation. 33 Castillo de la Torre (n 2) 505; Wils (2010, n 2). 34 Wils (2010, n 2) 18 and 26–28. 35 Jaeger, ‘The standard of review in competition cases’ (2011) 313.

Procedural Fairness in EU Competition Law  13 The entering into force of the Lisbon Treaty by late 2009 had the practical implication that it became easier to plead the points raised in the academic debate through litigations before the EU courts with reference to specific articles of the CFR, which supposedly provided at least equivalent procedural guarantees as were available under the ECHR. Already by 2010 the KME/Chalkor appeal was underway with the promise of settling the longstanding disagreement about the compatibility of the infringement procedure that resulted in a fine in competition cases, with the procedural fairness standards resulting from Article 6 ECHR and the accompanying case law of the ECoHR, and now also the corresponding standard in Article 47 CFR.36 Shortly before the CJEU could reach its decision in the KME/Chalkor cases, the ECoHR had the opportunity to clarify its position on the applicability of Article 6(1) ECHR with regards to competition law proceedings in the judgment in the Menarini case, which was delivered on 27 September 2011.37 In Menarini the ECoHR found the Italian competition law procedure, which like EU’s procedure was an administrative procedure at the first instance, to be criminal in nature according to the Engel criteria. The ECoHR nonetheless found the Italian competition procedure compatible with Article 6(1) ECHR, since Italian courts had full jurisdiction to review the administrative decision.38 In Menarini the ECoHR thus found competition proceedings to be criminal in nature, but that they should irrespectively not be subject to the strict interpretation of the procedural conditions of Article 6(1) ECHR. A competition decision imposing a penalty of a criminal nature could thus be taken through an administrative procedure, if an independent and impartial court or a tribunal had full jurisdiction to review the finding. This meant that the Jussila distinction between different degrees of criminality was extended to competition proceedings, in the sense that they did not belong to the traditional hardcore criminal types of proceedings, but were nevertheless somewhat criminal. The KME/Chalkor litigation came to an end with the CJEU’s judgments on 8 December 2011. The Court avoided the question of criminal or non-criminal nature of the proceedings, which had been one of the main points of the preceding academic debate and one of the central points discussed during the court proceedings. The Court also chose to approach the problem exclusively from the point of view of Article  47 CFR; it did not mention Article  6(1) ECHR.39 The Court concluded that the General Court had full jurisdiction to review the Commission’s fining decisions in cartel cases both in terms of facts and law, and thus the procedure was compatible with the standard of procedural fairness 36 The General Court handed down its judgment in the cartel cases against Chalkor and KME (Cases T-21/05 and T-25/05) on 19 May 2010, and the appeals were filed on 30 July and 3 August 2010. 37 A Menarini Diagnostics Srl v Italy App no 43509/08, CE:ECHR:2011:0927JUD004350908. 38 ibid [59] and [63]–[64]. See also a case note on Menarini in Oliver, ‘Diagnostics’ (2012). 39 Sibony points out that the Court could avoid answering the question about the criminal nature of the proceeding by only referring to Art 47 CFR, since the classification only has relevance in terms of Ar 6 ECHR; Art 6 ECHR only applies to civil and criminal proceedings, while Art 47 CFR is not conditioned on any specific type of a proceeding. See Sibony, ‘Annotation’ (2012) 1989–95.

14  The Problem of Procedural Fairness articulated in Article 47 CFR. To be able to do so, the Court quashed the deference doctrine that had been reinstituted by the Microsoft judgment and argued along the lines of Leanarts and Vanhamme in their 1997 article in the Common Market Law Review, and Wils in his more recent article of 2010 in World Competition. The Court argued that irrespective of the deference doctrine that the General Court says it was using, in practice, it was conducting a full judicial review. On 18 April 2012, a few months after the result in the KME/Chalkor litigation, the EFTA Court delivered a judgment in Posten Norge, which dealt with similar topics to Menarini and KME/Chalkor.40 In its judgment, the EFTA Court approached the problem from the point of view of Article 6 ECHR and was quick to conclude with reference to Menarini and AG Sharpston’s opinion in KME that the said procedure before the EFTA Surveillance Authority was criminal in terms of Article 6 ECHR.41 However, the EFTA Court noted, referring to Jussila, that the extent of the procedural guarantees provided by Article 6 ECHR ‘must be determined with regard to the weight of the criminal charge at issue’.42 The procedural guarantees were thus not to be considered as absolute, but rather as a balancing exercise in which the nature of the crime at stake was a factor. On an inference of this rationale, the Court concluded that the competition procedure at stake could be compatible with Article 6(1) ECHR, even if the initial criminal penalty decision was taken at an administrative stage, if the administrative decision could later be appealed to an impartial and an independent tribunal with full jurisdiction.43 The EFTA Court also discussed the deference doctrine with regards to complex economic and technical facts. The EFTA Court argued that its review of economic and technical facts was limited to a legality review, but that this limitation did not restrict the EFTA Court in conducting a full jurisdiction review in terms of Article 6 ECHR. Interestingly, the EFTA Court explained that the limitation inherent in a legality review barred the EFTA Court from substituting its own assessment for the EFTA Surveillance Authority’s (ESA) assessment because it considered its own assessment of the facts to be more correct; the EFTA Court could only annul ESA’s assessment if the inference drawn by ESA could not be substantiated by the relevant facts and was thus in breach of the principle of legality.44 While the judgments in the KME/Chalkor were rather cryptic in terms of providing answers to the main points of argument in the foregoing academic debate about the procedural fairness of EU’s competition procedure, the Menarini and Posten Norge judgments brought more clarity. These two judgments are explicit in that the Jussila doctrine of differing degrees of criminality applies to 40 Case E-15/10 Posten Norge v EFTA Surveillance Authority [2012] EFTA Ct Rep 246. 41 AG Sharpston was more explicit in her opinion than the CJEU in concluding that EU’s competition procedure should be considered criminal in terms of Art 6 ECHR. See Opinion of AG Sharpston in Case C-272/09 P KME Germany and Others v Commission, EU:C:2011:63. 42 Case E-15/10 Posten Norge v EFTA Surveillance Authority [2012] EFTA Ct Rep 246 [87]–[88]. 43 ibid [91]. 44 ibid [98]–[101].

Procedural Fairness in EU Competition Law  15 the field of competition law procedure and that consequently the procedural guarantees of Article 6(1) ECHR do not apply with their full stringency in that field of law due to the lacking level of hardcore criminality. The KME/Chalkor judgments follow the same rationale, but do so implicitly and without reference to Article 6(1) ECHR and the Jussila doctrine, and in fact without a reference to any rights theory.45 Several authors have taken stock of the legal situation after this trio of judgments.46 Sibony notes the absence of a discussion about the nature of competition proceedings in terms of procedural typology and the choice of the Court not to mention Article  6 ECHR in the KME/Chalkor judgments. She suggests several potential explanations that concern inter- and intra-institutional politics. Ultimately, she considers the result a disappointment. The guardian of the Treaty did ‘not expose a theory of guardianship’.47 Barbier de la Serre considers the EFTA Court in Posten Norge going beyond what the CJEU did in KME/Chalkor in terms of claiming jurisdiction to control the assessment of facts. He also notes a problem associated with a mismatch between what the CJEU says it does and what it effectively does in terms of judicial review. If the Court’s actions are inconsistent with its rhetoric, there is no way for an outsider to determine when the judicial review is in line with the required standards of procedural fairness.48 Bronckers and Vallery point out a paradox in the CJEU’s argument in KME/ Chalkor. The General Court had in its judgment under appeal repeatedly referred to a discretion it granted the Commission with, but the CJEU puzzlingly stated that this was no hindrance for the purposes of full judicial review: ‘In our view, it is unfortunate that the appearance of a less than full review by the General Court was not criticised by the Court of Justice. Justice must not only be done, but also be seen to be done.’49 Bronckers and Vallery note that the EFTA Court with its Posten Norge judgment explicitly embraced the lessons of Jussila and Menarini, but that the CJEU still has to confront the concept of judicial review more directly. They suggest that ‘perhaps the proper solution is to distinguish Commission decisions imposing fines, which fall within the criminal sphere protected by Article 6 ECHR, from other decisions’. By this, the Commission’s discretion could be eliminated for a particular type of decision, without having to eliminate it across the board; ‘Fining decisions must be deemed criminal cases from an ECHR perspective; other competition law decisions more resemble classic administrative law’.50

45 Sibony notes this as a missed opportunity for the CJEU. See Sibony (n 2) 2002. 46 See for example Oliver (n 2); Sibony (n 2); Bronckers and Vallery, ‘Menarini?’ (2012); Nazzini, ‘Administrative enforcement, judicial review and fundamental rights’ (2012); Schweitzer, ‘Judicial review in EU competition law’ (2013); Vandenborre and Goetz, ‘EU competition law procedures’ (2012) 578; Hauger and Palzer, ‘Investigator, prosecutor, judge’ (2013); Barbier de la Serre, ‘Standard of review in competition law cases’ (2014); Wils (2014, n 2) 5. 47 Sibony (n 2) 2002. 48 Barbier de la Serre (2014, n 2) 418 and 427–28. 49 Bronckers and Vallery (n 2) 292. 50 ibid 294–96.

16  The Problem of Procedural Fairness Bronckers and Vallery warn that pushing the fairness claim, of moving the initial decision power to a court, could imply inefficiencies in some Member States: ‘In other words, for a transitional period an approximation of fairness may have to be accepted, with improvements being made to the administrative process, in order not to jeopardise the effectiveness of competition law.’51 Wils considers that, following Jussila and Menarini, the main debate about the compatibility of the EU competition enforcement system with Article 6 ECHR, in which the Commission both investigates and decides, is over: The only question which is still open to debate is whether the General Court, when reviewing European Commission decisions, exercises sufficient jurisdiction to meet the ‘full jurisdiction’ standard laid down in the case law of the European Court of Human Rights. Either the General Court does exercise ‘full jurisdiction’, and there is no problem, or it does not, in which case the solution is for the court to modify its practice.52

Addressing the issue of the mismatch between the rhetoric on deference and the actual actions of the General Court, Wils notes that: What is decisive is whether the General Court in fact exercises full jurisdiction, not any general statements which the Courts may make as to its powers. It is nevertheless also important that the General Court is seen to exercise full jurisdiction. For this reason, potentially misleading general statements should be avoided.53

Nazzini considers the CJEU falling short of adopting a full correctness standard of judicial review in KME/Chalkor, implying that some elements of deference remain. He also considers the argument for saving the current system advocated in the literature by the Commission’s officials, such as Wils, which apparently was used by the CJEU in KME/Chalkor, to be ‘far from satisfactory’.54 Nazzini suggests that KME/Chalkor signals the demise of the deference standard of review in competition cases in favour of the correctness standard. He does, however, offer an alternative to retain some of the elements of the deference doctrine: ‘The alternative is to preserve, when appropriate, a deferential standard of review but with the necessary procedural counterbalance of functional separation between the prosecutor and the decision-maker within the Commission.’55 The minimal fix, suggested by Nazzini, would require the establishment of a separate adjudicative unit outside of DG Competition, which would report directly to the Commissioner responsible for the competition portfolio. This would thus be a functional separation, rather than a structural separation of prosecutorial and adjudicative powers.56 51 ibid 297. 52 Wils (2014, n 2) 11. 53 ibid 24. 54 Nazzini (n 2) 995–97. The argument he is referring to concerns the contention that the General Court supposedly conducts a full review irrespective of what he explicitly states in his judgments with regards to deference granted to the Commission. 55 ibid 999. 56 ibid 1002–04.

Towards a Conceptualisation of Procedural Fairness  17 To summarise the current state of play in the KME/Chalkor saga, it is now generally accepted that institutionally it is permissible in terms of Article 6 ECHR and Article  47 CFR to organise competition law enforcement based on a twotier adjudicative system, in which the former is an administrative-level authority. This is, however, only permissible if the adjudicative function at the later tier is conducted by an impartial and independent tribunal with full jurisdiction to review the results of the administrative tier. The court practice of the EFTA Court, the CJEU and the ECoHR is towards reducing or eliminating any deference granted to the administrative authorities, but the academic debate has not yet reached a consensus on how to optimise procedural fairness in competition proceedings. Some want to retain some elements of the deference doctrine, while others want to sever the adjudicative function from the administrative function through changes in the institutional structure. The consensus about the current legislative landscape thus only extents to what currently is; a consensus about what ideally ought to be in terms of procedural fairness in competition proceedings is still not in sight. Notably, the current consensus of the courts on the permissibility of deciding a competition procedure at the first instance through an administrative procedure rests on the assumption that the criminal nature of the acts that are the objects of such procedures exist in different degrees of severity, and that the associated procedural guarantees granted to defendants should take note of the nature of the act on a scale of criminal gravity. The implication of this apparent consensus of the three European courts is that the binary system of categorising procedures within the sphere of public law as either administrative or criminal is effectively obsolete for accurate descriptive purposes. The revelation that procedures can be somewhat criminal in nature, but not entirely, and that procedural design should somehow reflect these varying degrees of criminality, signals a rupture from a tradition of categorical absolutes in procedural design and a succession of a procedural architectural regime based on a notion of balancing. A methodological approach to this balancing is still underdeveloped in the academic literature. The absence of categorical procedural imperatives for determining the object of procedural fairness, and the lack of a comprehensive balancing methodology for achieving procedural fairness through procedural design, makes the notion of procedural fairness ambiguous. The legal concept of procedural fairness is thus in need of a sound philosophical rationale and a comprehensive methodology of execution to escape the perils of ambiguity.

II.  Towards a Conceptualisation of Procedural Fairness The question this book seeks to answer is how to know if a procedural rule is fair. The problem outlined above, based on the KME/Chalkor litigation, shows how a specific problem with procedural fairness in EU competition procedure has been

18  The Problem of Procedural Fairness dealt with in the literature and in the decisional practice of the CJEU and its sister courts in Luxemburg and Strasbourg. Over time, a consensus has emerged that the concept of procedural fairness, as articulated in the ECHR and the CFR, is not an absolute term but a term subject to some type of balancing. What remains underdeveloped, both in the academic literature and in the decisional practice of the courts, is to explain further the underlying normative elements and the methodology of this balancing. The hypothesis of this book is that a consensus about both things is required for a satisfactory answer to questions about the fairness of a procedural rule. A consensus about the conceptualisation of procedural fairness in EU competition law is thus needed to provide an uncontroversial answer to individual questions about the fairness of procedural rules. As exemplified in the description of the KME/Chalkor saga in this chapter, a simple declaration by a court about what the law is on a particular issue of procedural fairness does not necessarily suffice to settle the underlying philosophical question about whether that procedure is actually fair. It is probable that discontent about the legal expression of procedural fairness will keep brewing until the law is in sync with a broader consensus about what procedural fairness necessitates. In this sense, the Court of Justice failed to use the opportunity presented by the KME/Chalkor litigation to link the expression of the law with a generally accepted theory, or a conceptualisation of procedural fairness. In chapter two the first step towards providing a conceptualisation of procedural fairness are taken by analysing the fundamental elements of adjudicative procedures, both in an instrumental sense and in a normative sense. To this end, adjudicative procedures are modelled as social institutions that have certain instrumental elements and a normative rationale. The design of these elements ultimately decides the success of the procedure in executing its designated task and the fairness of its output. At the core of each procedural design there is, however, a balancing dilemma, the solution to which represents procedural fairness. In chapter three the modelling of adjudicative procedures as social institutions with a certain instrumental structure and normative rationale is measured against the procedural regime for enforcing EU competition law. The analysis of the different elements of EU’s competition procedure reveal that the paradigm of viewing such procedures as a social institution is beneficial in terms of identifying, locating and analysing problems in procedural design. It is also revealed how different elements of the procedure influence the calibration of the balancing problem and its contemporary equilibrium. If the relevant legislator has been successful, the contemporary equilibrium solution to the balancing problem should represent procedural fairness. Chapter four addresses the problem of knowing when an equilibrium solution to a legislative balancing problem is fair. The two main approaches to legal theory – positivism and naturalism – provide incompatible answers to this problem. This is problematic, since both approaches shape the views and work of many practitioners in the ranks of judges, lawyers and administrative officials. Incompatible philosophical view on the issue of fair balancing in legislative design is a recipe for

Towards a Conceptualisation of Procedural Fairness  19 numerous insoluble problems in legal practice. To provide a less controversial basis for approaching fairness in the law, it is suggested that the law must be understood in the context of the larger political society it serves. Based on this starting point, a novel conceptualisation of fairness in the law is presented and argued for. This novel approach should provide the procedural architect with a plausible theory of ethics on which he can base his regulatory designs. In chapter five a sketch of a methodology for practically implementing the preceding conceptualisation of procedural fairness is presented. The aim of the methodology is to provide consistency and accuracy in the balancing of the fairness consideration in legislative and regulatory design. The methodology is designed for a future where massive data collections about individual preferences of whole populations are available and the computational processing capacity to work the data is both swift and cheap. In chapter six several dilemmas of procedural fairness from EU competition law practice are identified. These dilemmas are analysed with reference to the conceptualisation presented in the preceding chapters. This analysis reveals how problems of procedural fairness have been dealt with in EU competition law practice, both by legal practitioners arguing and deciding cases before the Court of Justice, and by EU’s legislative and regulatory authorities. To conclude, chapter seven demonstrates how the conceptualisation and the methodology for implementing the conceptualisation of procedural fairness can bring clarity to the identified dilemmas from EU competition law practice. The book concludes with reflections on the wider implications of the technocratic approach to democracy, which the book applies to solving problems of procedural fairness in the context of EU competition law. Overall, the book thus aims at: (i) philosophically conceptualising procedural fairness; (ii) building a methodology for implementing that concept in legislative design; and (iii) showing how both the concept and the methodology can be applied to procedural design dilemmas in EU’s competition procedure. If successful, the concept and the methodology should provide a philosophical basis for answering why an adjudicative procedure is fair, and a methodological platform for establishing when it is fair.

2 The Essence of a Procedure Conceptualising procedural fairness requires a distinct concept of fairness and a distinct concept of procedures. A disagreement about an issue of procedural fairness can thus relate to a disagreement about how the concept of fairness is understood, about how the concept of a procedure is understood, or both. The chapter starts by addressing the problem of how the procedural should be distinguished from the substantive in the context of the law. The chapter’s primary focus is on conceptualising adjudicative procedures as social institutions in the instrumental sense and on laying out an argument that, irrespective of form or structure, adjudicative procedures serve a special procedural agenda in the normative sense. This should provide a conceptualisation of the essence of a procedure that can be used as an input for solving dilemmas of procedural fairness and a framework for assessing the quality and potential weaknesses of different procedural designs. The chapter concludes by reflecting on how this conceptualisation illuminates a balancing dilemma at the essence of any procedure, the solution to which forms the object of procedural fairness.

I.  Differentiating Procedural Law and Substantive Law The conceptual distinction between the substantive and the procedural is an important one in the context of the law. In the institutional terminology introduced above, the procedural can be viewed as the mechanism that produces a substantive outcome. Substantive law thus defines the object of desire and procedural law defines the mechanism for achieving this object. Using the example of EU competition law, Articles 101 and 102 TFEU clearly define in the substantive sense which business practices should be prohibited on the internal market, while Article 103 TFEU directs that procedural regulations, such as Regulation 1/2003 EC, shall be enacted to give effect to the principles laid out in the two preceding substantive provisions. Thinking in abstract terms, things can and will go wrong in the process of realising an object of desire. For example, we may fail to correctly follow a recipe when baking a cake, and there is a possibility that the recipe is flawed or incorrect. Both reasons are liable to influence the outcome of our baking. Similar reasons can influence the realisation of the plans laid out by a code of law, such as the

Differentiating Procedural Law and Substantive Law  21 criminal code: the procedural regime for implementing the plans can be ineffective or flawed and, even if perfectly designed, the officials responsible for executing the procedure can make mistakes. The procedural designer must thus strive for making the procedural regime flawless and, ideally, an apparatus should be in place that accounts for the possibility of human error by an internal or an external review arrangement, or both. Viewed instrumentally, a procedure simply refers to a specific means aimed at bringing about a specific end. The difference between a means and an end is often subtle. A decision to pursue a specific end can imply a process that is so obvious and uncontroversial that it does not need to be spelled out.1 Similarly, a decision to follow a process can imply the pursuance of obvious consequences that are taken equally for granted.2 Procedural rules thus sometimes acquire quasi-substantive character, and substantive rules become quasi-procedural.3 The blurred boundary between the procedural and the substantive in praxis requires a motivation for treating them as categorically separate instruments of law. In the absence of a specific reason, the optimal instrument of law would often be a synthesis of a means specifically designed to achieve a specific end, or the other way around. Many laws are designed without a clear distinction between rules that serve as ends and rules that serve as means. Often, the imposition of such a distinction would be superfluous; an end usually implies a specific means, and a means usually implies a specific end, which eliminates the purpose of making a distinction between the two.4 At a certain level of legislative complexity, the need for a distinction between the substantive and the procedural arises. The importance of this distinction progressively rises with the level of complexity. If there is only one way to do things, or when only one thing will result from doing things, a distinction between an action and a consequence is unimportant. If, however, several actions can lead to the same consequence, or if a variety of consequences can be achieved through a specific action, it becomes useful and meaningful to distinguish between the means and the end in the legislative context.

1 A 90 km/hour speed limit on a specific road could be used as an example of a pursuit of a substantive end that implies a specific process or a means of driving to achieve; if one drives carefully below the speed limit, one will achieve the end of staying below the articulated limit. The process is however so obvious that we do not need a specific procedural rule to facilitate the substantive end. 2 A 90 km/hour speed limit could also be considered as a rule of procedure to achieve the end of careful driving. The process implies the end pursued so that the latter does not need to be spelled out explicitly. 3 Solum discusses the blurred line between the procedural and the substantive in more detail. See Lawrence B Solum, ‘Procedural Justice’ (2004) 78 Southern California Law Review 181, 215–24. Sunstein also discusses the difficulty of engineering procedures that do not paternalistically affect the substantive ends. See Cass R Sunstein, ‘The Storrs Lectures: Behavioural Economics and Paternalism’ (2013) 122 Yale Law Journal 1826, 1855–58. 4 In philosophy, a reference is sometimes made to Ockham’s razor to underscore the impression that the simple is usually superior to the complex, unless there is a specific advantage to the complexity.

22  The Essence of a Procedure A legal system can be viewed as multiple layers of substantive ends and procedural means.5 Usually each end can also be viewed as a means, depending on the observer’s point of view. Observed very broadly, constitutional provisions are ends that lower-level laws and rules serve as means to implement. Any laws and rules thus possess a procedural character from a constitutional point of view. The criminal code, for example, is the procedure by which the constitutional principles of property rights and a right to life and liberty is implemented. Observed from a different standpoint, the substantive provisions of the criminal code are ends and the procedural provisions of the criminal code the means to implement these ends. Observed yet more broadly, the most abstract constitutional principles have a procedural in character in relation to the concept of the social contract, and further regressions are possible from the social contract, depending on where philosophically we choose to locate the ultimate deontic property. Using EU competition law as another example, Articles 101 and 102 TFEU can be viewed as constitutional ends and Regulation 1/2003 EC as the mean through which the constitutional ends of prohibiting restrictive practices is implemented. Viewed alternatively, the internal market objective can be viewed as a substantive end and Articles 101 and 102 TFEU as procedural means to implement that ultimate objective. The distinction of the procedural and the substantive in the context of the law is thus sensitive to the observer’s point of view. The complexity of the criminal justice system has for a very long time warranted an explicit distinction between the procedural and the substantive. Through history there have been criminal codes prohibiting various behaviours, and criminal trial procedures have been used as a means to determine guilt. The same is true with the history of EU competition law; soon after the Treaty of Rome established the substantive provisions prohibiting cartels and abuse of monopoly market power in 1957 (ie Articles 85 and 86), Regulation 17 of 1962 established a procedural regime through which the substantive rules where to be implemented.6 There are many ways to serve justice, and without a deliberate choice of means a consistent choice of the optimal approach may not happen. Observed from the point of view of the criminal system, the criminal code that describes which acts are criminal and which reprimanding measures should be taken when such acts occur should be conceived as substantive in character. The provisions of the criminal code that describes the mechanism for determining if a breach has occurred to one of the substantive provisions, and for determining which reprimanding measures should be applied to such an incident, should accordingly be viewed as

5 Miller considers a layered system of joint actions as a defining feature of organisations. Each layer then consists of an end and a means but, viewed more broadly, the layer is one part of an organisationwide action that can consist of many independent actions. See Seumas Miller, The moral foundations of social institutions (Cambridge, Cambridge University Press, 2010) 48. 6 Reference is made to Council Regulation No 17 (EEC): First Regulation Implementing Articles 85 and 86 of the Treaty [1962] OJ No 013.

Procedures as a Social Institution  23 procedural in character. The distinction between the substantive and the procedural may not always be explicit in the relevant code of laws, thus it becomes necessary to approach the distinction functionally from a defined standpoint. The same analysis can be applied to the various instruments of EU competition law, which range from Treaty provisions to procedural Regulations and various guidelines, including guidelines on fines. Conceptually, questions about the nature of EU competition law in the substantive sense are different from questions about the nature of EU competition law in the procedural sense. The normative focus of questions about the former is on what ought to be considered a restrictive or an abusive practice, while the normative focus of questions about the later focus on what ought to be done to determine whether a predefined prohibited practice has occurred or not. Viewed in terms of the conceptual framework of social institutions, it is possible to view the EU competition law system holistically as a social institution with many complex parts.7 It is also possible to narrow the scope and view the procedural parts of the legal regime as a separate social institution. The concept of a social institution is abstract and adjustable to various analytical needs. The focus of this chapter is on the procedural aspects of competition law; the concept of a social institution will thus be adapted accordingly. There is an important nuance between a procedural character of law and a substantive character. The implication of this nuance is that, on one hand, there is fairness in the substantive sense (eg when is it fair to punish someone?), which will not be dealt with here, and on the other hand there is fairness in the procedural sense (eg when is a trial fair?), which is the subject of this chapter.

II.  Procedures as a Social Institution Society can be viewed as a collection of institutions, each having a distinct task that serves the grand corpus of society. John Rawls’ influential account of justice theory considered several major social institutions in the basic structure of society and the governance of social cooperation through these institutions being the main object of justice.8 This, he considered, was due to the basic structure’s profound impact on people’s life prospects.9 Rawls did not specifically list competition 7 A variety of approaches can be taken to study the many aspects of the competition law system. It can be approached from the viewpoints of the discipline of economics, sociology, philosophy, criminology, law, etc. The different parts can also be studied in isolation, for example cartels, monopolies, mergers, state aid, administrative procedure of competition cases, court procedure of competition cases, private enforcement, public enforcement, EU enforcement, NCA enforcement, etc. 8 Rawls’ idea about the basic structure of social institutions that ought to be fair was a part of his grander scheme of building an ethical theory about society based on deontological (Kantian) and contractarian ethics, instead of the utilitarian ethics prevalent during the mid-20th century. 9 John Rawls, A theory of justice (Cambridge MA, The Belknap Press of Harvard University Press, 1971) 7 and 54.

24  The Essence of a Procedure law procedures among his examples of major social institutions, but he did list trials and competitive markets as separate examples.10 Procedures for realising competitive markets, including trial procedures, would thus fit well among the vital institutions of Rawls’ basic structure. Such institutions can be perceived in the abstract as ‘a possible form of conduct expressed by a system of rules’ and also as ‘the realisation in the thought and conduct of certain persons at a certain time and place’ of the actions prescribed by the abstract rules. An institution thus can both be what the rules say it ought to be and what it is, as realised through the actions of the institutional actors. Rawls argued that the object of justice of an institution was how it was realised as opposed to what it ought to be in the abstract.11 A joint action is the common starting point for the main theories in the philosophical literature about social institutions. The transcendence from the individualistic, over to the social is the joint action of two or more individuals. Simple forms of a joint action, such as conventions, rules and norms, do not qualify as social institutions; a level of complexity and permanency is required. A social institution can however constitute several simple joint actions that form a semi-permanent behavioural pattern. Social institutions are also distinguished from super complex patterns of joint actions, such as whole societies or cultures, but individual social institutions often form constituent parts of societies and cultures.12 Social institutions can be perceived in different ways. In the philosophical literature, the competing views distinguish between seeing the individual in terms of the joint action and seeing the joint action in terms of the constituent individuals.13 The main difference being the location of the deontic property. Either the joint action is an independent deontic property, or it is a derivative from the deontic properties of the constituent individuals. In a plainer language, either a joint action is of value in and of itself, or its value is a regression from what the participants in the joint action find valuable. It is thus possible to understand institutions in terms of what its constituents require or understand the constituents in terms of what the institution requires. The creation and maintenance of social institutions can be achieved through collective acceptance. This means that their existence depends on a collective belief or a collective treatment as such. On the collective acceptance view, the collective acceptance is not reduced to an aggregate of individual beliefs; the acceptance is considered a performative,14 ie the fact that something is collectively believed 10 Rawls listed several other examples of such institutions, eg legal protection of free thought and conscience, private property, and the monogamous family. See ibid. 11 ibid 55. 12 Miller, Moral foundations (2010) 22–23. 13 Miller labels these different approaches as atomistic, holistic, and molecular theories of institutions. See ibid 27–34. 14 A performative or a performative utterance is a concept in analytical philosophy, introduced by the philosopher JL Austin, which denotes a sentence or a statement that is not falsifiable and that is a part of an act or an action. For example, a judge writes in the operative part of a judgment that the defendant is

Procedures as a Social Institution  25 utters a meaning onto the phenomena at stake, eg a social institution, which in turn becomes a performative of that collective belief or acceptance.15 This type of linguistic philosophy fits the positivist, ie the descriptive legal context especially well, as was proficiently demonstrated by HLA Hart’s work on the concept of the law.16 On this account, a legal procedure is a social institution with deontic properties simply because it is collectively treated and accepted as such; it is a social fact. A specific legal procedure would thus be considered fair, so long as it is collectively treated and accepted as such, regardless of its actual properties. Viewed differently, joint actions of certain complexity and permanence become institutions and derive their deontic properties from intentional actions of numerous actors directed towards the realisation of a common end. Contrary to the collective acceptance accounts, collective ends are reducible to the ends of the constituent individuals. The collective end is thus an aggregate of individual ends, but is nonetheless of such a nature that it needs to be pursued collectively to be achieved.17 There is thus a collective norm that an institution should be pursuing, and this norm is something that the individuals are committed towards pursuing through a joint action, or a layered structure of several joint actions. Viewed in this way, an institution consists of an input, a mechanism for processing the input, and an output that results from the processing of the input. In the legal philosophical context, these views are compatible with natural law theory, in which the laws are a mechanism that serve a higher normative end. An institutional view of the laws in such a way requires something beyond the laws. A mechanism or an institution that produces an outcome is insufficient if an understanding of the elemental qualities desired from the joint action are missing. The laws as an institutional mechanism can thus only be adequately designed if the designer knows which qualities the outcome ought to have. For the purposes of constructing an analytical framework for assessing the quality of different designs of legal procedures, the teleological approach to social institutions is better suited. While the collective acceptance approach provides a strong foundation for a positivist (descriptive) analysis of procedural designs, it is less suitable for a normative analysis. The teleological approach is, however,

not guilty of a crime, or a legislator issues a law that states that vaping in public buildings is prohibited. See JL Austin, How to do things with words (Oxford, Clarendon Press, 1962) 4–7. 15 Collective acceptance theories of social institutions build on Austin’s insight about the utterance of performatives. Prominent examples of such theorists are John Searle (The construction of social ­reality (London, Penguin, 1995); Making of the social world: the structure of human civilization (Oxford, Oxford University Press, 2010)) and Raimo Toumela (The philosophy of social practices: a collective acceptance view (Cambridge, Cambridge University Press, 2002)). 16 JL Austin and Hart both taught philosophy at Oxford in the 1950s and they cite and credit each other in their work. Hart’s explanation of his key concept about the rule of recognition is essentially to explain it as a performative: ‘[the rule of recognition] can neither be valid nor invalid but is simply accepted as appropriated for us in this way’. See HLA Hart, The concept of law, 2nd edn (Oxford, Oxford University Press, 1994) 109. 17 Miller refers to this kind of theories as teleological accounts and I will follow his terminology. See Miller (n 5) 37–55).

26  The Essence of a Procedure ideal since it presumes a normative agenda which informs the purpose of the relevant institution. For the initial step of assessing the quality of procedural designs, an analytical framework that answers what ought to be is thus more useful than a framework that answers what is.18

III.  The Institutional Elements of an Adjudicative Procedure Instrumentally, social institutions are defined by constitutive rules that outline the various duties, rights and roles of the individual participants. There is also an informal dimension to the instrumental function.19 The various role occupants formulate their actions in accordance with the formal rules and the formal mission of the institution, but, at the same time, they may be tempted to use any discretion associated with their role occupancy to further their private ends. Formal and informal sanctions are a common feature of social institutions for countering corruption, which is liable to undermine the institutional mission. Instrumentally, the structure of a social institution consists thus of a formal element of constitutive rules, an informal element of institutional culture, and a formal or informal element of sanctions. Additionally, social institutions consist of a normative element that assigns them with a functional rationale, which is the benchmark for assessing the performance of the instrumental elements.20 Normatively, an institution ought to be doing x; instrumentally, an institution is successful if it is doing x.

A.  Instrumental Structure A clue to what a social institution of an adjudicative procedure ought to be doing is an observation of what it is doing. The different instruments of which the institution consists should be viewed as normatively neutral. The instruments do not represent intrinsic ends; they are merely a means towards the grand institutional task. Thus, if an institution is instrumentally doing something, the chances are that it is doing so towards realising the normative end of the institution.

18 After having defined what ought to be a comparison will of course have to be made with what is for completing the qualitative analysis. 19 Rawls noted that it was necessary to distinguish between constitutive rules of an institution and the strategies of the role occupants for how the institution (in the formal sense) could be used to advance various purposes. Ideally, an institution should be designed in a way so that private interests of the role occupants and the institutional mission are coordinated. See Rawls, Theory of justice (1971) 56–57. 20 Miller uses a slightly different terminology. He considers that in addition to a constitutive activity, institutions have four salient properties, namely structure, function, culture and sanctions. See Miller (n 5) 24.

The Institutional Elements of an Adjudicative Procedure  27 The institutional structure of an adjudicative procedure is moulded around the inputs it receives and the expected output it is supposed to provide. The inputs are factual events that have occurred and pre-existing codified legislative provisions (in the substantive sense) that may regulate these factual occurrences. The expected output of an adjudicative procedure is an answer to the question whether the predetermined substantive laws apply to the factual events as they occurred. In between the inputs and the output is an institutional machinery instituted by law (in the procedural sense), which is meant to process the inputs of facts and law into an output of an adjudicative decision. In the following sections, I will analyse the typical instrumental structure of the procedural machinery for taking adjudicative decisions. Such decisions can be taken in the context of a criminal trial, a private law dispute, or as an administrative decision subject to judicial review (such as the Commission’s decisions in competition law cases). I shall argue that, irrespective of context, adjudicative procedures can be viewed as a social institution that consist of three instrumental elements, namely: (i) constitutive rules; (ii) informal institutional culture; and (iii) a system of formal or informal sanctions or both.

i.  Constitutive Rules Studies of legal history inform how the constitutive rules of modern adjudicative procedures came about. The ancient Roman law distinction between harmful delictum acts against individuals and harmful crimen acts against society is still the conceptual framework for the main forms of adjudicative procedures.21 This conceptualisation identifies the main types of institutional actors and stakeholders involved. For criminal procedures the main stakeholders are the person (or persons) accused, on one hand, and society usually represented by a state apparatus, on the other hand. A potential private victim of a crime has an auxiliary role in a criminal procedure, compared with its principal role in a private law litigation. In private law litigations the main stakeholders are the private persons involved in the dispute at issue, with the state having a more passive role. The precise role of individual actors in an adjudicative procedure is sensitive to how the role of the state and state power is politically perceived. If, for example, the state’s role is primarily perceived as being reactive, the state will focus on the conflict solving aspect of the procedure and thus give more control over the process to the private parties involved. If, however, the role of the state is perceived as activist, the state will actively use its control over adjudicative procedures to implement its various policies.22 21 George Mousourakis, Roman Law and the Origins of the Civil Law Tradition (Cham, Springer, 2015) 159–60. 22 Damaska argues that the common law trial is a genus of the reactive conflict solving type of state power, while the civil law trial is a genus of the activist type of state power. See Mirijan R Damaska, The faces of justice and state authority (New Haven, Yale University Press, 1986) 71–96.

28  The Essence of a Procedure The basic institutional structure of the criminal and civil procedure of Roman law have survived into the modern era remarkably intact. Shifting political trends in the location and the concentration of sovereign power have had influences on the micro design of adjudicative procedures with regards to the role of different actors in the procedure, but the basic macro structural elements have endured through more than two millennials.23 The constitutive rules of modern adjudicative procedures outline several distinct phases, which each has its set of role occupants that ideally perform their roles in accordance with the constitutive rules. The central feature of such procedures is the neutral decision-maker that takes a decision at the end of each procedural cycle. Each cycle thus consists of a process of gathering and processing information to enable a neutral decision-maker to reach a decision on the issue at hand. The end of each procedural cycle can be referred to as a decision point or a decision phase that is preceded by a framing phase. The procedural details of civil and criminal trial procedures and administrative adjudicative procedures are preoccupied with the functional task of enabling the framing of information and the taking of decisions. Using the example of a criminal procedure, its constitutive rules would normally establish at least two decision points: the formal accusation; and the judgment. If there were a possibility to appeal the first judgment, which there normally is, each subsequent judgment would form an additional decision point. Depending on the profile of the case, there could also be a separate decision to initiate a formal investigation, prior to the decision to make an accusation. Leading up to these decision points, there are framing phases that aim at revealing the truth about the factual and the legal circumstances at hand to enable a correct decision at the subsequent decision point. The framing phases are structured differently to serve different aspects of the framing task. The phase leading up to the formal accusation empowers the state with comprehensive investigatory tools to extract the necessary evidences to reveal the factual truth behind the suspected criminal activity. During this phase, the role of the suspect is passive but nonetheless subject to certain procedural guarantees that restrain the extensive investigatory powers of the state. During the actual trial process, the more rigidly regulated framing phase aims at providing a level field for the accused to present factual and legal objections to the accusation. The framing phases thus instrumentally deal with the gathering and the representation of Socialist  civil procedure, as practised in the USSR, provided for a greater role of the state in civil procedure than is practised in the modern liberal democracy. The state could, for example, initiate claims on behalf of reluctant plaintiffs. See further in Mauro Cappelletti and Bryant G Garth, ‘Introduction – Policies, Trends and Ideas in Civil Procedure’ in Mauro Cappelletti (ed), International Encyclopaedia of Comparative Law – Volume XVI – Civil Procedure (Leiden, Martinus Nijhoff Publishers, 1987) 16. 23 See further on this historical development in Haukur Logi Karlsson, A quantitative quest for philosophical fairness in EU’s competition procedure (European University Institute, PhD thesis, 2017) 130–43, available at http://hdl.handle.net/1814/48005.

The Institutional Elements of an Adjudicative Procedure  29 the relevant inputs for making the actual decisions. The decisional phases deal with taking the actual decisions, both the intermediary and the final decisions, based on the factual data gathered during the framing phases. The final judicial decision represents the ultimate institutional output of the adjudicative procedure. The main aim at each decisional point is to take the right decision but, as shall be explained later, the meaning of ‘right’ in that context is qualified by a normative benchmark. Corresponding analysis can be applied to civil and administrative adjudicative procedures. This typical instrumental structure revealed by the constitutive rules of adjudicative procedures should not be surprising; it correlates with and mimics how the human mind takes decisions involving high stakes. According to Kahneman, such decisions are taken based on deliberations about the situation at hand, identifying the relevant information and organising that information comprehensively to enable informed decision-making.24 The formal constitutive rules of adjudicative procedures provide for an instrumental structure of one or more cycles of framing and deciding. Typically, the cycles become more numerous as the interest at stake are higher. A simple traffic ticket will thus normally not need many cycles of frame-and-decide processes to conclude. Serious crimes, involving important private and societal interests, will often go through several cycles of frame-and-decide at different court instances before an ultimate decision. The instrumental quality of these two procedural phases is liable to affect the functional quality of the procedural institution.

ii. Culture The formal instrumental structure of an institution of an adjudicative procedure, as represented through its constitutive rules, outlines numerous institutional roles and assigns them to different institutional actors or role occupants, eg investigators, prosecutors, lawyers, judges, etc. In addition to the formal structure, a complementary informal instrumental structure normally influences the quality of the institutional output. This informal structure shall be referred to as an ‘institutional culture’.25 The effects of institutional culture on the institutional output varies, but is likely to be compounded by the level of discretion made available by the formal constitutive rules. If the formal rules are detailed in describing the institutional roles and how they should be performed, the latitude for a culture to thrive is somewhat limited. If, however, the formal rules are less prescriptive, an informal culture has a greater potential to mark the institutional output. 24 Kahneman distinguishes between the process for taking instant automatic decisions, and decisions that are taken slowly based on a deliberate framing decision process. See further in Daniel Kahneman, Thinking fast and slow (London, Penguin Books, 2012) 19–109. 25 Miller refers to informal norms that govern the roles of actors within institutions as an institutional culture. See Miller (n 5) 49–50.

30  The Essence of a Procedure Social institutions of adjudicative procedures tend to be tightly regulated by formal constitutive rules. This would indicate a limited space for an institutional culture to shape the output of the institutions. However, despite the rigidness of the regulatory framework, the complexity of the institutional structure allows for some discretion at various points. Such discretion is liable to affect the institutional output. In line with the formal structure of adjudicative procedures into different decisional and framing phases, and the assignment of different roles to different role occupants, different aspects of a potential institutional culture can be assessed. Judicial discretion can provide a space for a judicial culture to grow, prosecutorial discretion can take a cultural form, and investigative discretion as well. These different aspects of procedures are thus liable to be influenced by institutional culture applicable to the different roles and their occupants. On a judicial bench of several judges, the judges often have discretion about intricate details of how to reach a conclusion. Should they all concur? How hard should they try to concur? When is it appropriate to register a dissenting vote? How much ownership should the judge rapporteur have over the drafting process and how engaged are the non-rapporteur judges in the deliberation process? Do individual judges preside over their own expert and administrative staff, or does staff that is also at the disposal of other judges support them? As far as these kinds of issues are not regulated by formal rules, the practical solutions to these issues can become an institutional convention and thus an aspect of an institutional culture. Many more cases of potential breaches against law are reported and investigated than end up being the object of a formal prosecution or a formal administrative complaint. When taking their decisions on whether to issue a formal accusation, prosecutors and officials often have discretion to decide which cases to close and which to formalise through an official prosecution or a complaint. The way in which they exercise their discretion is normally limited by economic and policy considerations. Prosecutors will normally not bring cases before court unless there are reasonable probabilities of the court upholding the formal accusation, or at least a part of it, by a conviction judgment. Making a formal accusation and bringing people to court would often be a waste of time and resources if the probabilities of conviction were low. Many good cases may also be brought to the attention of prosecutors and officials, but due to capacity restraints they may decide to use their resources on other cases and thus not bring such cases towards a formal court or administrative process. To make the most of the available institutional resources it would often make sense to prioritise cases based on an impact assessment, prioritising high-impact cases.26 Officials may also decide to pursue easy 26 By impact I mean impact towards achieving some policy goals implied by the relevant code of laws in the substantive sense, or more immediate policy goals of the enforcement authorities or the political authorities. Note, though, that in some criminal procedural systems prosecutors have a general duty to prosecute and do thus not explicitly have the leverage to priorities cases based on an impact assessment.

The Institutional Elements of an Adjudicative Procedure  31 cases over difficult cases in order to make efficient use of their resources and achieve a higher success rate in courts, with a higher volume of successful cases. Conversely, although difficult cases can be intensive on institutional resources, the potential reward of succeeding in difficult cases is high-impact. The way in which prosecutorial and administrative discretion is exercised can become an informal institutional convention and so may shape institutional output by influencing which cases reach the courts and which do not. Many things can happen in everyday life that could be considered in breach of some law provision. Some of these things are reported to the authorities that subsequently decide whether to investigate the breach and prepare for a prosecutorial or an administrative decision. In making such decisions, discretion rests with the responsible institutional actors. Investigators often have extensive powers that enable intrusive investigative tactics. They may also be equipped with sophisticated tools and expertise that enables accurate fact-finding. In high-stake crimes, the full range of the investigative capacity may be utilised to uncover facts for enabling high-quality prosecutorial decisions. In other instances, it may be obvious that a crime has been committed, eg bicycle theft, but the investigators nonetheless make no proactive effort to investigate, giving prosecutors no alternative but to close the case due to lack of evidence. The investigative discretion is normally subject to similar economic and policy influences as the prosecutorial discretion. Insignificant crimes do not set in motion the full range of investigatory powers, nor the use of expensive tools and expertise. This leaves discretion to the relevant institutional role occupants. How this discretion is exercised can become a convention that forms an aspect of an institutional culture of adjudicative proceedings. The way in which the investigative discretion is applied influences the institutional output, since the quality of the initial framing process influences all subsequent decision-making by either facilitating decisional quality or hampering such quality. The discretion of different institutional role occupants, especially those with decisional capacity, can provide the grounds for an institutional culture to lay roots in adjudicative procedures. The level of discretion and its influence on the outcome of the enforcement process can vary between different systems. Good professional culture that facilitates the ultimate task of the process is of course desirable and should normally be reflected in a positive public attitude towards maintaining a level of discretion for the different institutional role occupants. Good culture can thus facilitate a trust in experts and acceptance of a level of flexibility for them to carry out their institutional roles diligently. Bad professional culture that corrupts the institutional output is on the opposite spectrum. Members of the public would tend to distrust such an institution, which potentially would trigger a political reaction towards curbing the discretion of the role occupants by formalising the institutional roles through further constitutive rules or formal sanctions. Rigid formal rules can however shift the attitude of the role occupants from approaching their tasks through discretionary expertise to a legalistic or bureaucratic approach, which dogmatically abides to the

32  The Essence of a Procedure letter of the law.27 An institutional culture exemplified by a professional exercise of expert discretion can thus be beneficial for the quality of the institutional output, but if the culture is corrupted, a political reaction to limit expert discretion is to be expected. Such reaction would aim at limiting the possibility of a harmful institutional culture affecting the quality of the institutional output. The mode of reaching decisions is a salient feature of adjudicative institutions (such as criminal procedures). This mode can be implicitly obligatory and thus form a part of institutional culture, instead being part of the formal constitutive rules. The input and the object of adjudication are normally two kinds of facts: facts of law as established prior to the event in question; and physical facts relevant to the event as it occurred. The constitutive rules often contain an explicit obligation to establish the truth about the event at hand and determine how the laws apply to the truth. Implicated is an obligation that the truth should be established by a rational inference from various sources of permissible evidence.28 As a principle, role occupants in adjudicative institutions are not formally bound by a hierarchy of evidence nor any rules about how they should be assessed; they do in other words enjoy discretion in how they assess evidence for establishing the truth and reaching their conclusions.29 This flexibility in the assessment of evidence is for countering a reliance on overly formalistic means of providing proof. The downside is a risk of arbitrariness of evidence assessment. As a counterweight against arbitrariness, judges are thus required to explain the grounds for their conclusion in writing, including how they assessed the evidences. A duty to explain implies an obligation on the judge to reach a decision based on a rational inference from the legal and physical facts at hand. Perhaps, there is not much stopping a judge from reaching an inconsistent and irrational conclusion, but as a matter of institutional culture, judges in adjudicative proceedings try to approach their task rationally. A failure to follow such an institutional culture would degrade the institutional output.

27 Kagan argues that political trust and mistrust can influence institutional design and choices in the context of administrative law. I am suggesting that similar concerns may influence the level of discretion provided to role occupants in other types of adjudicative proceedings. See Robert A Kagan, ‘The Organisation of Administrative Justice Systems: The Role of Political Mistrust’ in Michael Adler (ed), Administrative Justice in Context (Oxford, Hart Publishing, 2009) 161–80. 28 Lon Fuller distinguished between three modes of taking social decisions: elections, contracts and adjudication. Decision-making through elections and contracts is predominantly based on the preferences of the parties to the respective processes, which do not need to be rational as seen by an external observer. One is at liberty to make an absurd contract and cast a vote any way one pleases. In contrast, rational inference from legal and physical facts is at the essence of adjudicative decision-making. See Lon L Fuller and Kenneth I Winston, ‘Forms and Limits of Adjudication’ (1978) 92 Harvard Law Review 353, 363–72. 29 In earlier times under Roman-canonical proceedings, evidences were ranked in order of importance and the judge was required to reach a conclusion concurrent with the prevailing evidences. This could lead investigators to seek all means to extract the cardinal evidence, the confession, through brutal means and even torture. See Heinrich Nagel and Jerry Norton, ‘Evidence’ (2011) Encyclopædia Britannica, available at www.britannica.com/topic/evidence-law.

The Institutional Elements of an Adjudicative Procedure  33 Other role occupants in an adjudicative procedure must act in accordance with the decisional mode of the primary decision-maker. Absurd and inconsistent arguments before a rational decision-maker are pointless, while facilitation of accurate legal and physical fact-finding by other role occupants will usually improve the quality of the primary decision and thus the institutional output.

iii. Sanctions Sanctions are an important feature of most social institutions. In theory, institutions can survive without a system of sanctions, but in practice they are normally required, or are at least beneficial towards the institutional agenda. While the average human is rational most of the time and would thus see the benefit of cooperating with others through social institutions, the temptation of selfishness is constantly present. Sometimes acting selfishly coincides with cooperation through social institutions, while at other times acting solo will undermine such institutions. There is also a temporal issue at stake. A rational long-term perspective and a rational short-term perspective can suggest contradictory actions for the individual. In the short-term, it can make sense to act selfishly and undermine institutional agendas, while in the long-term it could be rational to cooperate and abide to institutional agendas. Social institutions that are meant to last for generations, thus usually require a mechanism for convincing people to comply with the rules of the long game of its agenda. A simple method is the use of sanctions that raise the individual cost of non-compliance. In that way, it becomes rational to follow institutional agenda, even when acting selfishly in a short-term perspective. Sanctions can take both a positive and a negative form. In a negative form, non-compliance would be penalised, but in a positive form, compliance would be rewarded. A sanction can thus both take the form of a forfeiture of a reward, or through the imposition of a more direct penalty. Viewed broadly, the spectrum of social sanctions for non-compliance can range from subtle peer pressure at one end to the death penalty at the other. Viewed more narrowly in terms of an institution of an adjudicative procedure, the issue of sanctions focuses on non-compliance with institutional norms by the different role occupants involved. The function and efficacy of a proceeding is undermined if the role occupants do not follow the formal rules regarding their roles, or act in contradiction to the institutional culture, or both.30 The reasons for such noncompliance can vary, ranging from various intentional self-interested reasons, to unintentional incompetency. Whichever reason, incentives and mechanisms to help maintain the proper institutional function are to be expected. These could be both formal and informal sanctions, internally and externally enforced. 30 A distinction should be made between sanctions, such as prison sentences, that are the output of the institutional mechanism, and sanctions that apply intra institutionally to role occupants and which have as their aim to maintain compliance with institutional norms. For the purposes of this work, the focus is on the latter.

34  The Essence of a Procedure Distinguishing between ‘we’ and ‘I’ modes of participating in social institutions is one way to look at the issue of compliance with intra-institutional norms.31 Following Toumela’s argument, a level of cooperative ‘we’ attitude is needed among institutional role occupants for the institution to properly function. If a role occupant acts with a selfish ‘I’ attitude his or her actions will not always correspond with what would be expected by a role occupant acting cooperatively in accordance with the ‘we’ attitude. To minimise the institutional harm from those acting with an ‘I’ attitude, formal and informal sanctions can be used to incentivise conformity with institutional norms. Focusing on an inherent agency dilemma is another way to look at institutional norm compliance. For performing institutional tasks, institutions are reliant on role occupants, ie private individuals with their own private desires and agendas. Institutional corruption is an expression of this dilemma. Role occupants may, for example, choose to use their role occupancy to further their own wealth or power or prestige, instead of instigating the institutional agenda diligently. The primary problem of institutional design is thus overcoming the selfish urges of different role occupants. This problem is often approached through sanctions and surveillance mechanisms that make corrupt behaviour more costly and the probability of detection higher.32 When analysing the use of sanctions in an institution of adjudicative procedures the focus needs to be on the subjects of such sanctions, ie the different role occupants. The purpose of the sanctions is to ensure their compliance with formal and informal institutional norms and thus make it less likely that their actions corrupt the institutional output. In terms of the types of sanctions, there are both formal and informal types of sanctions. Taking and giving bribes in the context of a trial procedure is, for example, subject to criminal sanctions, and the same applies to making false accusations and obstructing justice. More subtle forms of institutional corruption and oversight can be subject to administrative sanctions, eg a lawyer who disrupts a trial. Informal sanctions concerning career progression and reputation among peers can also influence the actions of individual role occupants. It has already been explained that an institution of an adjudicative procedure is usually composed of several cycles of framing and decisional phases, where each framing phase is concluded with a decision that either takes the case to the next stage or concludes it. Using a typical criminal procedure as an example, it can be seen that during the initial investigative phase in such a procedure, the primary decision-maker is the prosecutor.33 He or she initiates and manages the 31 Toumela argues that social institutions are group affairs and that cooperative ‘we’ attitudes (as opposed to selfish ‘I’ attitudes) of the group members are prerequisite for the functional success of such institutions. See Toumela, The philosophy of social practices (2002) 17–39, 162–64 and 176. 32 Jon Elster, Nuts and bolts for the social sciences (Cambridge, Cambridge University Press 1989) 147–48 and 156–57. 33 Here I am referring to the person wielding the prosecutorial power in each case. Sometimes this would be a special prosecutorial office and sometimes, especially for cases of minor importance, the police would wield this power.

The Institutional Elements of an Adjudicative Procedure  35 investigation, usually in collaboration with the police, and is responsible for the prosecutorial decision to either drop the case or take it to court. Through managerial discretion, they should be able to ensure adequate investigative quality, have an oversight over the proper performance of the investigators, and potentially use informal sanctions for creating incentives for diligent work. Both the investigators and the prosecutor are subject to criminal sanctions if they intentionally disrupt an investigation or are in other ways found criminally corrupt in their exercise of their institutional role. During the investigation, the accusation of a potential victim is under scrutiny, including the potential to sanction wrongful accusations. The accused and their actions are also under scrutiny, if they try to obstruct the investigation into the case, they could face a separate criminal sanction for obstruction of justice. During the actual trial, the roles shift. The judge becomes the primary decisionmaker and to an extent presides over a new investigative phase of the case where a prosecutor and a lawyer representing the defendant present legal and factual arguments, in support and as a defence against the formal accusation.34 During the trial, witnesses may also have a role, either by revealing something they witnessed, or as experts about the meaning and interpretation of specific physical evidence. The judge normally has a power to sanction the participants in the trial for any actions that may disrupt it. Witnesses can be subject to criminal sanctions if they give false statements in court. Prosecutors and judges can also become subjects to criminal sanctions for corruption in their official capacity connected to a trial. Comparable analysis applies to other types of adjudicative procedures with regards to formal and informal sanctions. Despite various types of sanctions intended to provide incentives for institutional actors to refrain from pursuing self-interested agendas, instead of the institutional agenda, a weakness persists. The primary decision-makers enjoy great discretion in terms of the substance of their actual decisions. So long as there are no external indications of corruption, the primary decision-makers possess the discretion to decide what happened and what the law is. They do usually have to provide some explanations for their conclusions, but they will not be sanctioned, at least not formally, if their decisions are implausible.35 Depending on where the particular decision phase is in the institutional hierarchy, the only remedy is to contest the decision at higher levels, until all such

34 Note, though, the difference between the civil law systems, where the judge is perceived as having an active investigative role, and common law systems, where the judge is more passive and leaves the investigative role to the defendant and the prosecutor. 35 Damaska notes an aspect of informal sanctions in rigidly hierarchal institutional organisations: ‘The spectre of superior audits promotes an ethic of cooperation, and the dynamics of hierarchical promotion contribute to a spirit of “team playing.” Those officials whose desire to make a special impact obstructs the harmonious functioning of the organization are likely to be bypassed for advancement.’ See Mirijan R Damaska, The faces of justice and state authority (New Haven, Yale University Press, 1986) 21.

36  The Essence of a Procedure options are exhausted. The primary decision-maker is thus always at liberty to reach the wrong decision, so long as there are no verifiable indications of corruption. At the investigative level, the main risk is under enforcement due to false negative decisional errors, ie decisions not to investigate in detail and not to prosecute or take case forward despite there being ample reasons to do so. At the trial level, there is a risk of both a false positive and a false negative error. Appellate courts reduce the detrimental effects of decisional errors of lower-level courts, but are themselves exposed to the same risk.

B.  Normative Function The constitutive rules of an institution of an adjudicative procedure normally establish an instrumental structure of framing and decisional phases, usually consisting of several intervals. This structure is shaped by the institutional purpose of facilitating the normative task of the procedural institution and the quality of the framing and decisional instruments is determined by how successful they are at this task. The object of facilitation is a decision on whether an event has occurred that infringes the relevant code of laws. Seemingly, this is a simple task; the laws and the facts are the only types of inputs, and the output is an answer derived from those inputs. However, the binary nature of the supposed outcome of an adjudicative process does not reflect the incremental nature of the two types of inputs. The facts of a situation are often cloaked by uncertainty and the applicable law is often enigmatic. A question about the ideal outcome of an adjudicative process is thus not only about finding the right answer; it is no less about how right the right answer needs to be. Reality imposes restraints on the normative function of adjudicative institutions. Even if all the relevant information is available, it may still be impossible in practice to assess it for the purposes of reaching the correct decision and, even if it is possible to process all the relevant information, it may nonetheless be undesirable to do so for various practical reasons. It may be desirable to impose a normative condition on adjudicative institutions that they should reach the right decision but, in practice, the primary condition may have to be relaxed by imposing a supplementary normative condition on how right the right decisions needs to be. The literature on normativity in procedural law largely agrees that the primary aim of adjudicative institutions is to take the right decision. There are, however, different approaches to how the practical restraints, outlined above, should be dealt with. The two main strands of the literature can be associated with two of the main philosophical approaches to ethics: on one hand, there are authors who approach the question of how right the right answer should be based on a deontic rationale; and on the other hand, there are authors who take a consequential approach to the same question.

The Institutional Elements of an Adjudicative Procedure  37

i.  The Deontological View The deontological approach to explaining adjudicative procedures presupposes that there are some essential irreducible procedural elements to a just procedure. The deontological approach requires a distinction between substantive justice, which refers to the outcome of a procedure, and procedural justice, which refers to how the procedure proceeds.36 In contrast, the consequentialist approach does not rely on this distinction. John Rawls and Ronald Dworkin argued elegantly that the instrumental normativity of procedural law can be approached through a deontic ethical rationale, and their approaches are worthy illustrations of such arguments.37 John Rawls explained the role of the deontological argument by distinguishing between a procedure in the substantive sense and in the pure procedural sense.38 A procedure in the substantive sense relies on an independent a priori criterion of what is being pursued through the procedure. The normative task of the procedure thus becomes to facilitate and achieve this outcome. If the pursued end is defined clearly, and it is also possible to create a procedure that perfectly facilitates that end, there is a possibility to achieve a perfect procedural justice in the substantive sense. Rawls, however, argued that procedural justice in the substantive sense is often flawed due to the practical difficulties of building a procedure that flawlessly facilitates the desired end. In case the idealised perfect procedural justice cannot be achieved, Rawls referred to imperfect procedural justice as the common practical alternative. The criminal trial is an example of such an imperfection, he argued. In the criminal trial, the criteria for the desirable outcome exists a priori, but the trial procedure gives no guarantee that the result will not lead to a false outcome. Due to the practical difficulties of achieving the perfect, or near-perfect procedural justice, Rawls introduced the notion of pure procedural justice. Pure procedural justice does not use the outcome of the procedure as a reference point for its success. The reference point of success becomes the purity of the constitutive procedural elements to the procedure. If the procedure to determine how a procedure ought to be conducted is followed in a pure enough fashion, it becomes enough for finding a procedure procedurally just. Ronald Dworkin illustrated how the deontological approach can provide a rationale to civil and criminal procedures.39 His argument started by identifying

36 Michael Adler, ‘Understanding and Analysing Administrative Justice’ in Adler, Administrative Justice (2009) 131–33. 37 For a more detailed literature review on this issue see Lawrence B Solum, ‘Procedural Justice’ (2004) 78 Southern California Law Review 181. 38 Rawls (n 9) 83–90. 39 Ronald Dworkin, ‘Principle, Policy, Procedure’ in Ronald Dworkin, A Matter of Principle (Oxford, Clarendon Press 1986) 72–103.

38  The Essence of a Procedure a tension at the base of the well-known procedural maxim from criminal procedure: ‘no one should be convicted for a crime that he did not commit’. He asked whether it follows from this maxim that each citizen ‘has a right to the most accurate procedures possible to test his guilt or innocence, no matter how expensive these procedures might be to the community as a whole’.40 The identified tension is between the ideal state of perfect accuracy of a procedure, and the practical reality of society’s limited resources to invest in individual procedures and procedural instances. Dworkin dismissed the use of a cost–benefit analysis for finding a reasonable compromise between the accuracy and the cost of a procedure. In support of that dismissal, he referred to a distinction between bare harm that people feel through the imposition of a law enforcement penalty, and the additional moral harm that people feel when penalised unjustly. Dworkin claimed that the ‘injustice factor in a mistaken punishment will escape the net of any utilitarian calculation’ no matter how sophisticated.41 On his view, moral harm should be considered as an objective notion that occurs also, and especially, when no one knows or cares about it. This supposedly makes moral harm an impossible object of utilitarian quantification. Any efforts to utilise cost–benefit analysis to strike the right balance between procedural accuracy and procedural costs are thus bound to result in a morally suboptimal result. Importantly, Dworkin recognised the unattainability of committing to the ideal of the perfectly accurate procedure in world of scarce resources, where multiple types of potential moral harms exist in addition to the prospect of being unjustly convicted for a crime.42 The practical impossibility of achieving the perfectly accurate procedure, and the difficulties of making a morally acceptable balance between accuracy and costefficiency, led Dworkin towards an alternative approach. Inspired by Rawls’ pure procedural justice argument, he proposed two principles of ‘fair play in government’: firstly that ‘any political decision must treat all citizens as equals’; and secondly that ‘a later enforcement of that decision is not a fresh political decision that must also be equal in its impact’.43 The principles defer the balancing problem to the political process, which Dworkin believes is well suited for deciding the relative importance of different moral harms.44 From the two principles, Dworkin inferred that people have two rights with regards to a criminal procedure: the procedure should ‘attach the correct importance to the risk of moral harm’; and ensure ‘consistent weighting of the importance of moral harm’. These two rights, Dworkin claimed, provide ‘a middle ground between the denial of all procedural rights’ implicit in the utilitarian approaches and ‘the acceptance of a grand right to supreme accuracy’.45

40 ibid

72. 80–81. 42 ibid 84. 43 ibid 84–85. 44 ibid 87. 45 ibid 89. 41 ibid

The Institutional Elements of an Adjudicative Procedure  39 The structure of these arguments made by Rawls and Dworkin is somewhat problematic for the present context of analysing adjudicative procedures as social institutions. If the fundamental deontic task of the procedure is to answer a question, the way to optimise that task can hardly be to focus on optimising deontic properties different to the one at hand. If these alternative properties are independent of the primary task, the functional task of the procedural institution will shift towards a different primary task, ie the institution will no longer be about answering the question it is supposed to do. If these alternative properties are constitutive of the primary task, a focus on realising them will only be relevant insofar as it facilitates the primary task, ie it may be helpful if the procedure is pure or fair and equal, but that is conditional on how these deontic targets facilitate a right answer to the main adjudicative task. Either way, there is no guarantee that a focus on alternative deontic properties will facilitate the central institutional task of finding the right answer to the adjudicative question at stake. A focus on getting the central task right, will however necessarily facilitate the task of getting the central task right. The regression of the accuracy problem suggested by Rawls and Dworkin is thus unhelpful in determining how right the right answer provided by an adjudicative procedure needs to be. The move suggested by them begs an analogous question; how pure does the pure procedure need to be, or how fair and equal does the fair and equal procedure need to be? For the purposes of procedural design, Rawls and Dworkin have identified procedural accuracy and procedural cost as the primary consequential concerns. Although not explicitly intended as such by their authors, the discoveries of pure procedural justice and the two principles of government fair play, can be viewed as proxies for the ideal balance between theses primary consequential concerns. Any rights deriving from the overarching principles of procedural justice thus become balancing maxims aimed at facilitating the ideal balance of the primary consequences at stake.

ii.  The Consequential View By focusing on the consequences pursued through a procedural institution, the constitutive normative foundations are in full view. The different constitutive instrumental elements have different tasks that relate, either to the facilitation of the framing task of the procedure, or the decisional task of the procedure. Various techniques are possible for performing both the framing task and the decisional task, but to assess which technique is the most appropriate for a given task a benchmark of optimality is needed. The object of optimality forms the normative agenda of a procedure. A procedural institution ought to strive for optimality, however substantively defined. The quality of the outcome of a procedural institution stands in relation to the substance of the substantive law it aims to facilitate. If a procedure succeeds in facilitating the outcome pursued by the substantive law, it has performed optimally.

40  The Essence of a Procedure If a procedure is not successful in facilitating the substantive outcome desired, its performance is suboptimal. In simplified binary terms, an adjudicative procedural institution succeeds when it gets things legally and factually right, but fails when it gets things legally or factually wrong. Even if procedural institutions succeed most of the time in facilitating the correct substantive outcome, an absolute certainty thereof can nonetheless be hard, or impossible to achieve. The plainest circumstance can easily become subject to an insurmountable factual uncertainty due to a simple personal variance in the perception of factual events. What a procedure can hope to do in such circumstances is facilitate the substantive outcome pursued with optimistic chances of being right. A procedure with high probabilities of getting things right is thus closer to optimality than a procedure with low probabilities of getting things right, other things being equal. Scarce institutional resources further compound an occasional impossibility of establishing the accurate factual situation in a procedural scenario. Skilled investigators and qualified decision-makers can, if given enough time, increase the probabilities of reaching a correct decision. The time of such professionals is a scarce resource that is not in supply without an investment in professional training and institutional capacity. The substantive law that a procedure aims to facilitate, applies to several potential future instances that can be estimated by economic planners from year to year. Given that the resources available to economic planners are limited, the resources available for facilitating a substantive legislative plan are also limited. This fact curbs the quantity of resources that can be spent on getting each procedural instance right. In a world of unlimited resources, hordes of investigators and highly trained decision-makers can be hired to ensure the best available procedure in terms of facilitating the correct substantive result. In this world, the procedures that give equivalent chances of a correct result are considered equivalent in terms of the level of optimisation, notwithstanding different cost levels. In a world of limited resources, the intensive use of resources on a procedural instance means that less will be available to optimise other instances. This implies that procedures with analogous probabilities of a correct result can additionally be ranked in terms of optimality based on their level of resource drainage. Thus, in a world of unlimited resources only the accuracy of the result matters, but in a world of limited resources both accuracy of the result and the efficiency with which the accuracy is achieved matter for determining optimality. The leading consequential approaches, on the issue of normativity in procedural institutions, implicitly assume a world of limited resources and thus identify accuracy (ie cost of error) and procedural costs (ie cost of administrating the procedure) as the primary considerations in procedural design.46 Using these

46 Richard Posner identifies the objectives of civil and criminal procedure from an economical perspective as being the minimisation of the cost of error and the cost of operating the procedural

The Institutional Elements of an Adjudicative Procedure  41 approaches, the optimal procedural design can be identified by finding the efficient balance between error costs and administrative costs, using cost–benefit analysis tools.47 The maximisation of procedural accuracy relative to procedural costs provides a plausible optimisation standard for pursuance in procedural design – plausible enough to rival the alternative approach of using procedural maxims of the deontological type. In a world of scarce resources, the use of imperative procedural standards becomes problematic due to the absence of a specific target of optimisation. If a compromise must be made between the imperative standard of accuracy and the practical concern of procedural cost, the deontological approach struggles with identifying and explaining where the optimal balance should be struck between the two, while the consequential approach can refer to the point of maximum efficiency. In a world of scarce resources, procedural accuracy must be considered as a quantitative term regardless of whether it is used in the deontological sense or the consequential sense. In the absolute sense, accuracy only exists in a world of unlimited resources. Procedural standards or maxims in the deontological sense thus essentially communicate a degree of accuracy, not accuracy per se. The quantity of accuracy can be expressed informally on an ordinal scale through natural language (eg proof beyond reasonable doubt), or formally on a cardinal scale through the language of probabilities (eg 90 per cent changes of an accurate result). The consequential approaches have the benefit of being able to refer to the methodology of cost–benefit analysis in support of their choice of preferable accuracy level, ie the accuracy level should be cost beneficial. Approaches based on deontological ethics struggle with explaining why, for example, the accuracy standard of beyond all reasonable doubt should be used and not what seems to be the ethically superior standard of beyond all doubt. Given that perfect accuracy cannot be achieved due to material constraints, each procedure that ends with a decision implies a probabilistic possibility of a decisional error that can take either of two forms: a false positive and a false negative (ie type one and type two error). Logically, the only way to avoid making type one errors is to always reach a negative decision, and conversely the only way to avoid making type two errors is to always take a positive decision. The dilemma of every decision-maker, in a world where errors will be made, becomes to balance the risk of making an error of each kind.

system. See Richard A Posner, Economic Analysis of Law, 6th edn (New York, Aspen Publishers, 2003) 563. Gordon Tullock mentions that there ‘has been an assumption that we want to lower cost and raise accuracy’ of procedures. See Gordon Tullock, Trials on Trial: Pure Theory of Legal Procedure (New York, Columbia University Press, 1980) 70. Robert Cooter and Thomas Ulen assume that ‘the economic objective of procedural law is to minimize the sum of administrative cost and error cost’. See Robert Cooter and Thomas Ulen, Law & Economics, 6th edn (New York, Pearson, 2012) 385. 47 See for example Cooter and Ulen, Law & Economics (ibid) 384–86. See also a similar approach based on the so-called ‘Hand formula’; Posner, Economic Analysis of Law (2003) 563–64.

42  The Essence of a Procedure If the objective baseline probabilities that a positive decision is correct are equal to the probabilities that a negative decision is correct, the risk of making a type one error and a type two error is equal. If, however, using the same baseline probabilities, a procedure requires that to reach a positive decision the probabilities of that being the correct decision need to be higher than 50 per cent, the risk of making type one error disappears, but the risk of making type two error increases by the same margin. The baseline risk of making an error is unchanged at 50 per cent, but the procedural requirement eliminates false positives by increasing false negatives. If, however, the baseline probabilities change so that the chances of a positive decision being correct are 75 per cent and the chances that a negative decision is correct are 25 per cent, the choice of the allocation of risk between type one and two errors becomes more urgent. To reduce the number of overall errors the logical strategy would be to always take a positive decision and thus purposely make a type one error in one out of four decisions, while three out of four would for certain be correct. If, however, a procedure would require that a positive decision should only be reached if the baseline probabilities exceed 90  per  cent the decision-maker would need to reach a negative decision on each occasion. Thus, type one errors would be eliminated, but the number of overall errors would rise to three out of four decisions, all of which would be type two errors. In criminal procedures, a common procedural maxim dictates that people should not be found guilty unless there are high probabilities (ie more than 50  per  cent) of guilt. This signals to the decision-maker that he or she should emphasise not making type one errors; if in doubt, the default option should be a negative decision. The logical result from this procedural design is an increased number of total errors, but there is a lower risk of making type one errors than type two errors. Such an increase in the numerical volume of errors can, however, be justified. A reduction in the risk of making few serious mistakes by increasing the chances of making more numerous minor blunders is often preferable, both from a moral and from a cost–benefit perspective. The allocation of risk between type one errors, and type two errors requires the weighing of importance of each type. The consequential approaches can simply identify the point of the lowest aggregated error cost from both types as the optimal risk balance. The deontological approaches typically refer to intuitive standards, such as the well-known maxim from criminal procedure that it is far worse to convict an innocent person then it is to acquit a guilty person, which implies that the error risk profile should be tuned to minimise the risk of type one errors.48

48 Larry Laudan argues that the criminal trial is strongly biased in favour of minimising type one errors; ‘While trials, in theory, are designed to find out the truth about an alleged crime, the vast majority of procedural and evidence rules are designed to protect innocent defendants from wrongful conviction. Almost invariably, the rules designed to shield the innocent from conviction and related harms have the unintended but undeniable consequence of preventing many of the guilty from receiving their just deserts.’ See Larry Laudan, ‘The Rules of Trial, Political Morality, and the

The Institutional Elements of an Adjudicative Procedure  43 Dworkin criticised the use of cost–benefit analysis to identify the optimal error risk profile due to the immeasurability of the moral harm associated with type one errors.49 However, the alternative of using intuitive standards for determining the error risk profile can also lead to counterintuitive consequences. It has, for example, been pointed out that the procedural maxim that people should be allowed to state their case during a proceeding can decrease the chances that a type one error is committed and thus increase accuracy. This maxim, however, also invites a guilty person to mislead and confuse the proceeding and thus increase the chances of type two errors.50 It has also been suggested that adherence to the criminal procedural maxim of proof beyond reasonable doubt can have adverse consequences. Larry Laudan, for example, argues on the basis of statistical data that in the United States, ‘in any given year, one is six-times more likely to be the victim of a violent crime committed by someone falsely acquitted then one is to be the victim of a false conviction for a violent crime’.51 The consequential approach to the normative function of procedural institutions relies on defining cost and benefits of different procedural options. If the functional task of adjudicative institutions is to provide answers to the questions posed to them, the ultimate normative objective becomes to do so accurately. However, in a world of limited resources procedural errors will occur, which adds an efficiency consideration into the quest for optimality. The efficiency consideration exists at two levels. On one hand, a balance needs to be reached between achievable accuracy considering the available resources; on the other hand, an efficient balance needs to be reached between the types of errors the material restraints impose.

iii.  The Functional Essence of a Procedure Both of the traditional philosophical approaches to explaining the normative essence of adjudicative procedures agree that the ultimate norm of a procedure is to facilitate the prescribed criteria for a desirable outcome. The perfect procedure executes that task with perfect precision. Both approaches also recognise that perfection is hard to achieve in a world of limited resources. The approaches disagree in how a compromise ought to be reached in the balancing between the material restraints imposed by practical circumstances and the primary agenda of a procedure to accurately execute its facilitating task. Cost of Error: Is Proof Beyond Reasonable Harm Doing More Harm than Good?’ in Leslie Green and Brian Leiter (eds), Oxford Studies in Philosophy of Law: Volume I (Oxford, Oxford University Press, 2011) 196. 49 Dworkin, ‘Principle, Policy, Procedure’ (1986) 81. 50 Gordon Tullock points out that in an adversarial proceeding ‘a great deal of the resources is put in by someone who is attempting to mislead.’ See Tullock, Trials on Trial (1980) 96. 51 Larry Laudan, ‘The Rules of Trial, Political Morality, and the Cost of Error: Is Proof Beyond Reasonable Harm Doing More Harm than Good?’ in Green and Leiter (eds), Oxford Studies in Philosophy of Law: Volume I (2011) 202.

44  The Essence of a Procedure Using cost–benefit analysis for locating the optimal balance between accuracy and cost-efficiency comes with the usual caveats of over emphasising on the monetarily quantifiable at the cost of intangible variables, which Dworkin refers to as the element of moral harm. Using procedural maxims of the deontological type to strike the balance has the advantage of being able to include the element of moral harm but comes with the risk of neglecting the important consideration of cost-efficiency, which can result in counterintuitive consequences.

IV.  Towards a Conceptualisation of a Fair Adjudicative Procedure The philosophy literature on social institutions provides an analytical framework for studying the design of adjudicative procedures. On that view, adjudicative procedures are social institutions that consist of formal and informal structures that are designed for fulfilling a normative task. This normative task has two facets: the primary consideration is the accurate resolution of the adjudicative task; and the secondary consideration is to make the most of the available resources. By its nature, a social institution of an adjudicative procedure is imperfect at its task. Information about physical facts can be incomplete and difficult or impossible to extract. Legal facts can be vague or partially undetermined. Time is a commodity in demand and institutional resources are definite. The best that can thus be hoped for is a procedural institution that makes the most of the resources at disposal towards the institutional task. There will be decisional errors, but through deliberate institutional design the risk and costs from errors can be somewhat managed. It may be hard to avoid errors completely, but it is often possible through thoughtful design to fortify procedural weaknesses and channel the risk of errors towards their less expensive expressions. Decisions about institutional design are also decisions about who should carry risk and costs from potential errors in the institutional output. There are multiple stakeholders in any adjudicative institution that have different interests in how the risk and costs of errors are allocated by the institution’s designers. If the institutional designer follows procedural fairness maxims, these maxims will influence the allocation of risk and costs of errors between the different stakeholders. Comprehensive procedural rights for the accused will lower his or her risk and the potential cost of procedural errors, but at the same time, the risk may shift to society that in turn must carry additional costs in cases of procedural errors. Procedural fairness in deontic terms will thus eventually have distributional consequences. The question whether an institution of an adjudicative procedure is fair or not is in the end about whether the distribution of the risk and costs of errors is fair. Assuming finite institutional resources, the different stakeholders will care about how they are spent towards minimising errors in the institutional output.

Towards a Conceptualisation of a Fair Adjudicative Procedure  45 Finding a fair balance in the distributional sense is thus the essence of procedural fairness. This optimal balance may be expressed through deontic procedural maxims, but the metadata on which such maxims rely is a distinct distribution of the risk and cost of errors. People thus accept and support the use of procedural maxims due to the distributional fairness they represent. The acceptance of the underlying premises is of course not always explicit. It should, however, be safe to assume that for procedural maxims to acquire the status of collectively accepted institutional facts, they must represent a universally palatable solution to the underlying distributional dilemma, ie how the risk and cost of procedural errors should be distributed. If the deontic commands of the procedural maxims would not to lead to a fair distributional equilibrium in the consequential sense, there would be resistance against their applicability. A person can have various rights with regards to the workings of an institution of an adjudicative procedure. There can be certain universal moral rights that apply a priori irrespective of the institutional context. A person should, for example, as a universal principle, not be subject to torture, and thus torture should not be used as a means to extract evidence within the institutional context of a criminal procedure. There can also be certain moral institutional rights that only apply intra institutionally, but which have an imperative status in that context. Various procedural rights in the context of a trial, criminal investigations and administrative investigations should be considered as moral institutional rights. For example, a person would not have an a priori universal right to be heard by an impartial judge outside the institutional setting of a trial. Moral institutional rights are contextspecific and take meaning, at least partially, from the normative purpose of the relevant institution and the fairness equilibrium it seeks to implement. There are also mere institutional rights, which may result from a formal institutional design or an informal convention, but which does not directly influence the normative purpose of the institution or its fairness equilibrium. Abolishment or changes to institutional rights do not have moral repercussions for their beneficiaries.52 In this chapter, adjudicative procedures have been conceptualised as social intuitions with a distinct instrumental structure and a distinct normative function. This analysis has revealed that at the theoretical level there is largely a consensus in the literature about the intended function and purpose of adjudicative procedures, but that there is a room for disagreement about the balance between procedural cost and procedural accuracy. This adds an important consideration to the scholarly debate about procedural fairness in EU competition law (described in chapter one), which does not identify and address this crucial element in its discussion. In the next chapter this conceptualisation of adjudicative procedures will be applied to the specific context of EU competition procedure.

52 Miller refers to the rules of chess as an example of institutional rights that do not possess any moral qualities that would create repercussions in case of significant alterations. See Miller (n 5) 65–66.

3 The Structure and Function of EU Competition Procedures A procedure stands in relation to a substantive objective of some sort. EU competition law procedures stand in relation to the substantive objective of EU’s competition law regime. Procedures are auxiliary to the substantive objective and have as a primary purpose the objective’s realisation. For the purposes of conceptualising procedural fairness in EU competition law, the substantive objective of EU’s competition law is a fact that relates to the problem of procedural fairness. The solution to the problem of procedural fairness affects how the substantive objective is realised. The stakeholders in a competition law procedure can disagree about what the objective of the competition law regime ought to be, and to what extent the objective ought to reflect their subjective interests over the interests of others. However, the disagreement currently at issue relates to the procedural aspect, ie how the objective ought to be realised, not what it ought to be.1 To proceed to the question of how, it is useful to establish first what the how relates to. This chapter begins by outlining the central issues in identifying the normative purpose of EU competition law, which enables an identification of the main stakes and the main stakeholders in EU’s competition law procedure. Having established these elements, the procedural regime of EU competition law will be analysed against the conceptualisation from chapter two of adjudicative procedures as institutions. This should reveal that viewed as a social institution, EU competition procedure consist of certain constitutive rules, is subject to an institutional culture, and stays on task with the assistance of a system of internal and external sanctions. All these structural elements further aim at fulfilling the ultimate normative task of the procedure, which is to prepare and to take adjudicative decisions in competition cases accurately and efficiently. The chapter concludes by showing that on the basis of the forgoing analysis, arguments about procedural fairness in EU competition law can be located at a certain place within the institutional structure and analysed as precise balancing problems between conflicting interests of different stakeholders in the procedural regime. Overall, the chapter

1 Reference is made to the discussion in chapter two on the essence of procedures, regarding the difference between a substantive provision and a procedural provision, and why it is sometimes useful to explicitly legislate a procedure, while at other times it might be unnecessary.

The Normative Purpose of EU Competition Law  47 shows how the conceptualisation of procedural fairness in general, applies to the specific context of EU competition procedure.

I.  The Normative Purpose of EU Competition Law The process, or the act, of competition is the key concept in establishing the objective of competition law – both universally for all systems of law, and specifically with reference to specific systems of law. If there were, a single ultimate objective that all competition law regimes ought to be pursuing it would necessarily, in the strict sense, have to be built on the premise that there is something intrinsic about the process of competition that makes it worth pursuing independently of any other policy objectives. Alternatively, the process of competition can be viewed as a facility for pursuing ulterior policy objectives. If perceived in that way, the prospect of a universal objective of competition law is contingent on the universality of that specific policy. These are the two fundamental perspectives on the normativity of competition law in general and thus in EU competition law as well. The current normative objective of EU competition law supposedly is more economic than it used to be, and the interests of consumers are of principal concern.2 This perception is derived from the modernisation initiative of the Commission in the 2000s, which through several instruments of law elevated the interest of consumers to prominence and which had the self-proclaimed objective of approaching the competition law provisions of the Treaties more economically.3 Notably, around the same time an interesting constitutional development occurred. Following the Laeken declaration of 2001, a new constitutional Treaty was drafted, which proposed a change to how the competition law regime is constitutionally perceived.4 Among the primary objectives of the EU in the proposed constitution of 2004, was that the Union should offer its citizens ‘an internal market where competition is free and undistorted’.5 This was in stark contrast with the previous layout in Article  3(f) of the Treaty establishing the

2 Anne C Witt analysed the more economic approach in detail in her recent book. See Anne C Witt, The more economic approach to EU antitrust law (Oxford, Hart Publishing, 2016). 3 Monti explains that the turn towards the more economic approach was gradual and was initiated by the Commission’s Green Paper on Vertical Restraints in 1996. See Giorgio Monti, EC Competition Law (Cambridge, Cambridge University Press, 2007) 82–83. See also Ioannis Lianos, ‘Some Reflections on the Question of the Goals of EU Competition Law’ (2013) UCL CLES Working Paper Series 3/2013, 32, available at www.ucl.ac.uk/cles/research-paper-series. 4 The Laeken Declaration on the Future of the European Union was issued as an annex to the press release following the European Council meeting in Laeken, Belgium, 14 and 15 December 2001, available at http://europa.eu/rapid/press-release_DOC-01-18_en.htm. 5 See Article  1-3(2) of the Treaty Establishing a Constitution for Europe [2004] OJ 310/47: ‘The Union shall offer its citizens an area of freedom, security and justice without internal frontiers, and an internal market where competition is free and undistorted.’

48  The Structure and Function of EU Competition Procedures European Economic Community (EEC)6 and Article 3(g) of the Treaty establishing the European Community (EC),7 which listed the competition policy as one of the primary instruments for achieving the EEC’s and the EC’s aims listed in Article 2 of the respective Treaties. The proposed constitution infamously never became law. A few years later, when the negotiation of the Lisbon Treaty had been completed, the elevation of the constitutional status of the competition policy to an objective of the EU was nowhere to be found. In fact, the competition policy had been demoted to a protocol status from its previous status among the primary instruments for achieving the objectives of the EEC and the EC.8 After the Lisbon Treaty, the constitutional status of the competition policy is clearly that of a means rather than an end. Constitutionally, the promotion of the competition policy to an independent objective of the EU was explicitly rejected in favour of an instrumental view for achieving broader Union ends, especially the internal market objective.9 The Commission’s modernisation initiative that aims at an economic approach to competition law for the benefit of the consumers, and the constitutional status of competition policy within the Lisbon Treaty as an instrument for the facilitation of the internal market, provide the outlines of the current consensus on the objectives of EU competition law – at least for the purposes of the positive law of the Union. This representation is, however, vague and thus requires further elaboration. The Commission’s focus through the modernisation initiative, on being more economic and prioritising consumer interests, hints at normative elements. Protection of consumer interests on the internal market could be explained in terms of economic theory; in particular, welfare economics. On the surface, this seems a plausible approach to giving consumer interest a specific meaning, especially when being more economic in the methodological sense forms a parallel 6 ie Article  3(f) of the Treaty of Rome [1957] (Treaty establishing the European Economic Community): ‘For the purposes set out in the preceding Article, the activities of the Community shall include … the establishment of a system ensuring that competition shall not be distorted in the Common Market.’ 7 ie Article 3(g) of the Treaty of Maastricht on European Union [1992] OJ C 191/1: ‘For the purposes set out in Article 2, the activities of the Community shall include … a system ensuring that competition in the internal market is not distorted.’ 8 ie Protocol 27 to the Treaty of Lisbon (Treaty of Lisbon amending the Treaty on European Union and the Treaty establishing the European Community [2007] OJ C 306/1): ‘THE HIGH CONTRACTING PARTIES, CONSIDERING that the internal market as set out in Article 3 of the Treaty on European Union includes a system ensuring that competition is not distorted, HAVE AGREED that: To this end, the Union shall, if necessary, take action under the provisions of the Treaties, including under Article 352 of the Treaty on the Functioning of the European Union. This protocol shall be annexed to the Treaty on European Union and to the Treaty on the Functioning of the European Union.’ Lianos attributes the relegation of the competition policy to the protocol section to the lobbying of Nickolas Sarkozy, the French president at the time. See Ioannis Lianos, ‘Some Reflections on the Question of the Goals of EU Competition Law’ (2013) UCL CLES Working Paper Series 3/2013, 37–40, available at www.ucl.ac.uk/cles/research-paper-series. Although the aim of Protocol 27 was to neutralise the effects of removing the competition policy from the list of the Union’s primary instruments, the instrumental prominence of the competition policy was reduced by the Lisbon Treaty. 9 Protocol 27 bracketed the competition policy within the broader category of the internal market.

The Normative Purpose of EU Competition Law  49 objective. Perceived in that way, the focus on consumer interest could be taken to represent the objective of a specific welfare standard within welfare economics, namely consumer welfare.10 From the perspective of economic theory in the strict sense, a focus on consumer welfare in public policy is, however, not the optimal objective.11 The theoretical problem with a rigid consumer welfare standard is its inability to respond to the temporal dimension of the market. If social welfare within the sphere of competition policy is composed of the aggregated welfare of producers and the relevant consumers in a specific market, an efficiency enhancement conditioned on consumer welfare is only allowed if it is neutral with regards to, or enhances, the welfare of consumers.12 This in effect means that producers are only allowed to implement Pareto improvements with regards to the consumers. The inability of producers to implement Kaldor–Hicks improvements in the short term with regards to consumers, disincentives innovation, which in the long term harms aggregated total social welfare. Based on this type of rationale, it could be claimed that the optimal normative standard of competition policy should be total social welfare, rather than consumer welfare.13 The persuasiveness of the total welfare argument in competition policy depends the plausibility of severing distributional issues from efficiency issues in accordance with the second theorem of welfare economics. If only economic efficiency matters in competition policy, the normative objective of total welfare reigns. As a matter of theory, it may be desirable to separate efficiency issues from distributional issues in competition policy. As a matter of a legal fact, the current competition law regime of the EU is not founded on that presumption. On the contrary, Article  101 TFEU, the central competition provisions of the Treaty, suggest that distributional issues are internal to EU competition policy.14

10 Ezrachi claims that the concept of consumer welfare in competition policy may be a universal benchmark of economic analysis in competition policy, but a universal agreement of the benchmark’s properties is lacking. An agreement on the importance of the term thus mask a disagreement about its underlying rationale. See Ariel Ezrachi, ‘Sponge’ (2015) Working Paper CCLP (L) 42, 18–19, available at www.law.ox.ac.uk/sites/files/oxlaw/cclpl42.pdf. 11 Nazzini explains comprehensively the basics of the welfare economics approach to competition law and concludes that a consumer welfare standard is theoretically inferior to a long-term total welfare standard. See Renato Nazzini, The Foundations of European Union Competition Law: The Objective and Principles of Article 102 (Oxford, Oxford University Press, 2011) 32–50. See also Monti, EC Competition Law (2007) 83–86. 12 Efficiency is measured on different scales. Allocative efficiency and productive efficiency measure statically how efficiently resources are allocated, and goods produced at a specific point in time, while dynamic efficiency incorporates allocative and productive efficiency on a temporal scale that also measures the potential for innovation. These different methods for measuring the level of efficiency are methodological tools, while the standards of Pareto and Kaldor–Hicks are normative in character. See further Monti (n 3) 45. 13 Nazzini claims that long-term social welfare with regards to competition policy is ‘theoretically superior to any other objective’, and in his mind this applies with regards to the objective of consumer welfare. See Nazzini, The Foundations of European Union Competition Law (2011) 50. 14 Monti points out that Art 81(3) (now Art 101(3)) explicitly contains a distributional clause which fits poorly with neoclassical economics theory’s understanding of competition policy. See Monti (n 3) 25–26.

50  The Structure and Function of EU Competition Procedures This explicit legal fact lessens the compatibility of EU competition policy with a purely economic approach, especially those minimalistic ones that focus entirely on economic welfare. The Commission’s focus on consumer interests on the internal market will thus not be properly explained based on a pure welfare economic theory. Such an explanation would additionally need to assume that the Commission’s approach is illogical, ie that the Commission is somehow mistaken in preferring the inferior consumer welfare standard to the normatively superior total welfare standard. A proper understanding of the normative objective of EU’s competition policy thus requires the inclusion of the distribution issue, along with the efficiency issue. Economic theory is not ideal for dealing with questions of welfare distribution. Utilitarian calculus is insensitive to moral sentiments that are liable to influence the proper understanding of an equitable distribution policy. Distribution is dealt with at a more advanced level within the domains of ethics and political philosophy. The standards of Pareto and Kaldor–Hicks do address the distribution question to an extent, but the answers are not in line with most people’s intuition about distributional fairness. Welfare distribution is normally the primary task of the political, especially under democratic systems of government. The focus on the consumer in competition policy, as far as it is a distributional issue, is thus best understood as being from the perspective of the political (ie in the sense of moral and political theory), rather than in the sense of economic theory. By viewing the objective of consumer welfare and the focus on consumer interest in EU competition policy as a political value, instead of as an efficiency standard, the objective starts to make more sense. It is of course preferable from a political point of view to strive for a competition policy that enhances total welfare in the economic sense, but regardless of utilitarian welfare calculus, it may be even more preferable from a political perspective to accept a suboptimal welfare standard that is superior in terms of distributional fairness. The more economic approach pursuing the normative objective of consumer welfare and consumer interest on the internal market should thus be seen as a compromise between a purely welfare enhancing objective in the sense of economic theory, and the objective of distributional fairness in the sense of moral and political philosophy.15 The equilibrium point between these two normative objectives, as settled by the political process that drafted the treaty provisions on competition and by the political process that decides the enforcement strategies of the Commission, is in fact a political compromise between the competing preferences of the 15 Ezrachi and Monti point out that the design of the institutions responsible for the enforcement of EU’s competition policy is a factor in understanding the system’s function. For example, the body of 28 politically appointed Commissioners is the ultimate decision body in individual cases of competition law enforcement. This composition of the decision body invites political rationale into the decisionmaking on top of the economic rationale that the case-handlers at the lower administrative level may be more inclined to follow. See Ezrachi, ‘Sponge’ (2015) 8; Monti (n 3) 2–6.

Stakeholders and Stakes in Competition Proceedings  51 stakeholders in EU’s competition policy. While it may be rational for some stakeholders to prefer the objective of total welfare, at the same time it may be equally rational for other stakeholders to oppose the objective of total welfare in the absence of a rule of distribution that guarantees their interests. The political process, if sufficiently robust and efficient, resolves these conflicting claims equitably so that each stakeholder gets what he deserves, based on what he and other stakeholders want. If the current consensus on the normative objective of EU’s competition policy is extracted from the relevant Treaty provisions and the Commission’s stated enforcement policy the normative elements of economic efficiency and consumer interests appear. If the primary stakeholders in the competition policy hold conflicting preferences with regards to the realisation of these normative objectives, neither can be fully realised without compromising the other. The political process has negotiated an equilibrium between these two competing elements, which in turn is the sole normative objective of EU’s competition law. The current equilibrium emphasises the maximisation of social welfare, subject to a condition of distributional fairness with regards to consumers. The distributional condition is not strict; it simply emphasises that efficiency enhancements should be Pareto with regards to consumers. Such a condition still leaves a room for Kaldor–Hicks improvements with regards to competitors. The welfare of individual firms can thus be increased to the detriment of the welfare of other firms, but not to the detriment of consumers. This is thus the normative purpose of EU competition law in the substantive sense, which EU competition law in the procedural sense seeks to facilitate and implement accurately and efficiently.

II.  Stakeholders and Stakes in Competition Proceedings If, as argued in chapter two, the conceptual dilemma of any procedural design is about balancing equitably the normative considerations of procedural accuracy and procedural efficiency, the distributional issue must relate to differences of opinions about the ideal calibration of this balance. There are usually differently situated stakeholders in any procedural regime, with different stakes in its function and output. Assuming consensus among the different stakeholders with regards to the substantive objective of EU competition law, a separate disagreement can nonetheless exist about the proper calibration of the procedural means through which the consumer welfare objective is achieved. Due to the unattainability absolute procedural accuracy, discrepancies are liable to form with regards to how accurately the mutually agreed objective should be pursed, and at what cost.16 Differently situated individuals with regards to the procedure are liable to hold diverging preferences with regard to this balance of accuracy and the efficient

16 These

issues are discussed in detail in ch 2.

52  The Structure and Function of EU Competition Procedures use of enforcement resources. Essentially, this is the question of how many errors should be conceded through the procedural regime, and of which type. The optimal balance in this regard is the object of procedural fairness. The procedural architect should assume that the stakeholder’s preferences towards the accuracy/error question is dictated by a rational inference from their interests and practical circumstances. Each stakeholder that shares interests and practical circumstances with another stakeholder should thus through practical reasoning hold the same, or very similar, preferences towards the outcome of a specific regulatory design. In the same manner, it would be rational for a differently situated stakeholder to hold different preferences towards the outcome of a specific regulatory design. Based on this assumption, the key stakeholders in a procedure can be defined and categorised based on their practical circumstances and the preferences they are likely to hold in respond to their practical position concerning the consequences of the regulatory design at hand. Preferences towards specific consequences can be empirically measured or observed in number of ways. They could, for example, be estimated based on past choices (ie revealed preferences), and they could be estimated based on a survey or a questionnaire (ie stated preferences). A diligent procedural architect would be well advised to spend some resources on establishing through an empirical research an estimation of the actual preferences of the population in question towards the accuracy/error question. However, this is beyond the scope of this book. An attempt to compile a dataset on the preferences of different stakeholders in an EU competition law procedure will thus not be made for the purposes of this book. It is still possible to discuss briefly how the preferences are likely to trend in cases of rational actors, using informed guesstimates. Many of the key variables are obvious to any observer and their impact on a rational choice can be approximated. Along these lines, an attempt will be made to map the key stakeholders in EU competition law procedure and discuss the probable composition of their preferences towards the question of procedural accuracy/ errors.

A.  The Main Stakeholders For the purposes of the current procedural regime in EU competition law, a distinction is primarily made between interested parties and non-interested parties. The group of interested parties can further be divided into undertakings under investigation for having breached the competition provisions, and legal and natural persons that can assume the position of a complainant. The category of complainants can further be divided into undertakings that are competitors of the accused undertakings, and natural or legal persons that have the position of a consumer with regards to the accused undertakings or the relevant market at stake. The category of non-interested parties still holds stakes in the procedure, but only in their capacity as general taxpaying members of the public that pay for the enforcement

Stakeholders and Stakes in Competition Proceedings  53 system and suffer for the more general macro-economic consequences of competition law breaches. Non-interested taxpayers can both be undertakings and natural persons. For the present purposes, it can be extracted from the above that the primary stakeholders in a competition law procedure are the firms liable to become the subject of such procedures. Their competitors, whose competitiveness relies on the lawfulness of the market behaviour of other firms, are also important stakeholders. Individual firms or members of the public, in their capacity as consumers form a third group of relevant stakeholders, and in their capacity as taxpayers form a forth group of relevant stakeholders.17 The first type can be referred to as the accused, the second type as the competitor, the third as the consumer and the forth as the taxpayer.18 The role of the state and the state’s enforcement institutions in the design of a procedure should not be viewed as a stakeholder role.19 The role of the state in the design should not be confused with the role of the state during the actual procedural process, where the state’s enforcement institutions may have the role of acting on behalf of specific interests or stakeholders. The general consumer and the non-interested taxpayers, for example, have a limited access to the enforcement process of specific competition law cases, but implicit in the prosecutorial role of the enforcement agencies is the pursuance of the consumer’s interests and more generally the interests of the general public. In general, however, the state’s decisional and framing role during a procedure is to be considered above the interests of specific stakeholders. It is mostly about ensuring an equitable balance between competing interests during the procedural course of specific cases which, when acting in the prosecutorial capacity, may at times require acting on behalf of specific relevant but procedurally unprivileged interests.

17 The definition of a consumer in the Commission notice on the application of Art 101 TFEU could be used for the former group; ‘The concept of “consumers” encompasses all direct or indirect users of the products covered by the agreement, including producers that use the products as an input, wholesalers, retailers and final consumers, ie natural persons who are acting for purposes which can be regarded as outside their trade or profession. In other words, consumers within the meaning of Article 81(3) are the customers of the parties to the agreement and subsequent purchasers’; see Guidelines on the application of Article 81(3) of the Treaty [2004] OJ C101, para 84. The latter group could then be defined as any firm or individual that is not a member of any of the other groups of stakeholders, but who still is concerned with how public resources are spent on the enforcement of competition law and has a general macro-economic interest in minimising anti-competitive practices. 18 In practice, individual stakeholders often have mixed stakes. They can for example be competitors and consumers at the same time. For simplifying purposes, an assumption needs to be made that a stakeholder belongs to one group and one group only. Note also that in cartel proceedings, competitors are absent as a stakeholder group if all the relevant undertakings are part of the cartel. 19 This is consistent with the social contractarian view on society; the state as an institution acts on behalf of the people, not on behalf of itself. Enforcement institutions do thus have an auxiliary role and must act on a mandate from the stakeholders; an autonomous accord of institutions would be inconsistent with the democratic order of governance.

54  The Structure and Function of EU Competition Procedures

B.  The Main Stakes The dominant legitimate consideration for the accused during a competition proceeding, is to avoid becoming the subject of a false positive error, ie being falsely found guilty. It could also be said, of course, that the dominant consideration is often not to be found guilty at all, regardless of true guilt. For the present purposes, the assumption must however be made that all stakeholders are genuine in their commitment and support for the ultimate objective of competition law, and thus the accused is also in support of reprimanding true offenders against that objective, even if it turns out to be him or herself.20 The accused will thus ultimately prefer not to be found guilty, unless he or she is guilty. This will lead the accused to prefer high public investment in accuracy of the proceeding which, considering his or her proportion of potentially substantial error costs and his or her proportion of the public costs of the enforcement procedure, will remain marginally cost-beneficial for him or her longer than for other types of stakeholders. This will also lead the accused to prefer distribution of errors that gives premium to the avoidance of false positive errors, over false negative errors. The competitors of companies entangled in competition law proceedings will have an interest in seeing offenders against the objectives of the competition law regime brought to justice. The main motive for holding such interest would be their own position on the relevant market; if some actors are cheating, it is liable to deteriorate the competitiveness of those who stay within the limits of the permissible. In their capacity as operators on the market in question, an undetected breach against the rules can have serious consequences, while their share of the enforcement cost is not proportionated to their risk of becoming subjected to an error cost. The threshold for additional investment in accuracy to remain costbeneficial will thus be high, although not as high as the threshold for the accused. The competitor will have a dominating interest in avoiding false negative errors (ie false acquittals), while being at least neutral towards false positive errors. The competitor wants to avoid seeing offenders escaping justice through an enforcement error, since that would deteriorate his competitiveness. A false negative error on the contrary would not inflict any cost on the competitor. The competitor would thus be neutral in his preferences towards such errors.21 Consumers have the most to gain if the objectives of the competition law regime are realised in an accurate manner. Accurate compliance implies competitive prices and frequent innovations that maintain a high level of social welfare in which the consumers receive an equitable share. The primary interest of the 20 A preference for escaping justice cannot be considered as a legitimate preference in terms of designing public enforcement policy, a preference for not becoming the subject of injustice is however always a legitimate public policy consideration. 21 It could be argued that the competitor would prefer that the accused would be found guilty irrespective of true guilt. Like before, such a bad faith preference cannot be accepted as a benchmark for designing public policy.

Stakeholders and Stakes in Competition Proceedings  55 consumers concerns the price and the quality of the output of the relevant market. In terms of the procedural regime, the consumer should be willing to spend on additional accuracy, so long as his or her share of the enforcement expenditure is cost-beneficial when compared with the consumer’s gain in quality or price of the market output. The consumer is thus willing to pay for more accuracy, if he or she will gain at least as much in terms of price or quality of the relevant goods or services. The consumer is not conditioned to prefer a specific type of error allocation; he or she will simply prefer whichever allocation is the most efficient in increasing consumer welfare. The general taxpayer who does not qualify as an interested party in a competition proceeding does nonetheless carry a share of the enforcement costs and of the error costs that result from failures to accurately implement the objectives of the competition law regime. The individual taxpayer is detached from individual proceedings and is thus primarily concerned with larger macro-economic issues such as the efficient use of public resources and how the accurate enforcement of competition law influences social welfare in general. From the perspective of the taxpayer, it will remain rational to invest in procedural accuracy of competition proceedings as long as it remains cost-beneficial in terms of total welfare, and as long as there are no other competing investment options that are likely to yield higher increase in total social welfare. This implies that the category of taxpayers would be quicker to reach the threshold where an additional investment in accuracy would not be cost-beneficial. The taxpayer is not preconditioned on a preference for the type of errors to concede and will thus prefer whichever allocation that yields the greatest benefit in terms of social welfare. We can summarise the preference trends of the main stakeholders in a competition law procedure as follows: Table 3.1  Preference trends of stakeholders in competition law procedures Total accuracy v level of investment The accused

Prefers high accuracy almost at any cost

Distribution of errors Prefers type II errors

The competitor Prefers high accuracy and is insensitive to cost

Prefers type I errors

The consumer

Cost-beneficial accuracy in a narrow consumer sense

Neutral

The taxpayer

Cost-beneficial accuracy in a macro-economic sense

Neutral

By examining the provisions of Regulation EC 1/2003 and the Implementation Regulation EC 773/2004, we can see that firms subject to competition law proceedings are recognised as having entitlement to several procedural guarantees. Translated into the language of preferences towards accuracy and distribution of accuracy errors, these provisions on one hand ensure a certain level of investment on behalf of the EU in the accuracy of competition law proceedings, and on the

56  The Structure and Function of EU Competition Procedures other hand give guidance on the distribution of errors that favour the avoidance of false positive errors. Additionally, complainants (usually competitors, but sometimes consumers)22 have standing that enables them to provide an input into the proceedings that facilitates accuracy selectively in accordance with their stakes.23 The category of general consumers that do not file complaints have lesser access to individual proceedings, but the current normative objective of EU’s competition law regime gives premium to their interest, which should be reflected in individual public enforcement actions. The stakeholder status of non-interested taxpayers is not as such recognised through the procedural rules. The public authorities governing budgetary decisions will however presumably be strongly influenced in their budgetary decisions by this category of stakeholders that, due to its proportional and numerical size, wields a considerable political power.

III.  The Structure of the EU’s Competition Law Procedure Before an infringement to the competition law provisions of the TFEU can be indef­ initely established, a series of intermediary decisions are made, each preceded by a subprocedure aimed at facilitating that specific decision. Each of the intermediary decisions form a part of a holistic procedure that cumulates in the final decision on the existence of an infringement or not. This constitutes the instrumental structure of EU’s competition procedure. The significance of the intermediary decision points is that, at each point leading up to an Article  7 prohibition decision, an intermediary decision can be made to end the procedure without pursuing any further enforcement actions. After the case reaches a court stage, the nature of the decision points alters slightly; the stake is still about ending further enforcement actions on behalf of the Commission, but the power to do so no longer rests within the Commission. The Commission has already reached its conclusion on the prohibition of a specific behaviour, but the subjects to that decision can attempt to have the prohibition annulled by seeking judicial review from the CJEU. An enforcement action in 101 and 102 prohibition cases starts with a complaint that notifies the Commission of an alleged breach, or through the Commission’s own initiative, which may be based on market information received from interested or non-interested parties. After a potential breach has reached the attention of the Commission, an initial screening takes place that aims at facilitating a decision on whether to open a formal investigation. During 22 Consumers can in principle show legitimate interest in a case and become complainants in accordance with Art  7(2) of Regulation EC 1/2003, eg Joined Cases T-213/01 and 214/01 Österreichische Postsparkasse v Commission, EU:T:2006:151. In practice, individual retail consumers rarely lodge effective complaints. 23 Probably, the competitors would mostly provide incriminating information, but would refrain from providing information that might undermine a prohibition decision.

The Structure of the EU’s Competition Law Procedure  57 this initial phase, the Commission may resort to inspections of company premises that aim at securing potential evidences that are at risk of being spoiled by the subjects of the investigation. Less intrusive measures can also be used, such as information requests that require the firms involved to provide the Commission with the requested information. The Commission can also interview people for the purposes of collecting information. Depending on the strength of the initial lead, a decision must be made within the Commission as to how thorough investigation is needed during the initial informal stage to reach a decision on whether to open a formal investigation. In some cases, the allegation made in a complaint might obviously be unfounded or concern an infringement that is not sufficiently important for the Commission to take interest in pursuing an enforcement action. In such cases the Commission will simply close the case without entering an expensive investigation, or in some cases allocate it to a national competition authority (NCA). If, however, the Commission resorts to inspections or information requests during the initial phase, it must be presumed that the Commission already has doubts, which it intends follow up on through the opening of a formal investigation, unless the results of the inspections or the information requests rebuts the doubts the Commission already has about the alleged infringement.24 The opening of proceedings signals a commitment on behalf of the Commission to invest its resources in a particular case with the view of determining whether an infringement has been made to the competition provisions of the TFEU.25 Such a commitment is made on the implicit presumption that an infringement is probable and that the Commission has sufficient interest in the potential infringement to focus its resources there instead of somewhere else.26 The discretion of the Commission to open a formal investigation is not curbed by any official infringement probability standard. The Commission does however carry the burden of proof for infringement, and so from the Commission’s perspective it must be probable from the start that it will be able to discharge that burden. The decision to focus on a specific case over others can be motivated by many practical reasons. It might an easy case that is easy to conclude with a cost-efficient enforcement decision and thus demonstrate enforcement results. The incentive might also be a policy focus aimed at a specific market behaviour on which the Commission wants to signal its standing. Sometimes the decision might even be motivated by a desire by the Commission to advance a new enforcement doctrine in which the circumstances of a specific case might form a suitable pretext.27 Irrespective of 24 See further on the initial assessment in Commission notice on best practices for the conduct of proceedings concerning Articles 101 and 102 TFEU [2011] OJ C308/03, paras 12–16. 25 [2011] OJ C308/03, para 18. 26 The Commission of course cannot be explicit about what it believes with regards to a probable infringement and thus uses neutral language in its communication until a final decision on infringement is reached. See [2011] OJ C308/03, para 22. 27 See further discussion and references on the selection of cases in the Notice on Best Practices, see [2011] OJ C308/03, para 13.

58  The Structure and Function of EU Competition Procedures specific practical reasons for advancing a case, the overall objective of accurate and efficient enforcement always forms the superior reason, towards which the casespecific reasons function. During the formal investigation, the Commission can resort to the same investigative tools, outlined in Chapter V of Regulation EC 1/2003, as during the informal stage. The Commission can thus take statements, make inspections and issue information requests for the purposes of facilitating the decision on whether to issue a statement of objections (SO). The Commission may also hold state of play meetings with the subjects of the proceedings, the complainants, or third parties to discuss the case, either through its own initiative or at the request of the parties.28 The SO forms the second major decision point in an EU competition law proceeding. Note, though, that due to the need for stealth during the investigation of cartel conspiracies, the opening of proceedings and the issuance of the SO usually coincides.29 In cartel proceedings, the Commission has already at the point of a formal opening of investigation gathered the needed info to reach a preliminary conclusion and can thus simultaneously issue an SO. At the SO decision point, the Commission must decide whether to reach a preliminary conclusion of an infringement, or whether to close the case for some or all the parties without issuing an SO. In cases where the Commission does not intend to impose fines, the parties subject to the proceedings can instead offer commitments that address the concerns of the Commission. If deemed appropriate remedies, the Commission can adopt a decision where it accepts these commitments without concluding about the past or present existence of an infringement. The question of an infringement is thus left inconclusive, but the Commission nonetheless pledges to take no further action if the commitments are kept.30 An SO is only issued after an in-depth investigation, and it signals the Commission’s confidence that an infringement has been made. The SO outlines the objections the Commission has towards the behaviour of the parties subject to the proceedings and invites them to respond to these objections. A non-confidential version of the SO is made public and the complainants and interested third parties are invited to submit their comments on the SO to the Commission.31 The issuance of a SO functions as an official accusation that an infringement has been made, and the parties subject to the proceedings are challenged to reply to that accusation through a written reply and through an oral hearing orchestrated by the Commission’s Hearing Officer. Additionally, the parties are invited to a state of play meeting to discuss the case. To assess properly the Commission’s accusation, the addresses of the SO are granted with access to the Commission’s case file prior to both the submission of a reply to the SO and the oral hearing. 28 ibid paras 32–69. 29 ibid paras 4 and 24. 30 Commitment decisions are taken pursuant to Art  9 of Regulation EC 1/2003. See further on Commitment decisions [2011] OJ C308/03 paras 115–33. 31 ibid paras 84–85 and 104–05.

The Structure of the EU’s Competition Law Procedure  59 Depending on the content of the written replies, and the results of the oral hearings or any other communication with the parties after the issuance of the SO, the Commission moves towards the final decision point on whether to confirm in part or whole the preliminary conclusion reached in the SO, or whether to drop the case without finding an infringement.32 If the Commission finds an infringement and adopts a prohibition decision in accordance with Article 7 of Regulation EC 1/2003, it signals the definite conclusion of the Commission. That conclusion can however be challenged before the CJEU. In reaching a conclusion about an infringement to Articles 101 or 102 TFEU the Commission goes through three major intervals of framing and decision phases.33 At each of the three decision points, the Commission can choose to drop a case or to move on towards a conclusion of an infringement. The three decision points are progressively demanding in terms of accuracy of the finding about the existence of an infringement, and each of the progressing framing intervals marks a procedural investment aimed at facilitating ever-increasing accuracy at the subsequent decision points. If already at the first decision point the Commission is confident that it will not be able to find an infringement, it can drop the case and thus save resources. The same applies at the SO decision point. The progression of increasingly demanding decision points thus facilities the efficient use of enforcement resources. At each of the three decision points, the major stakeholders in the procedure might have different preferences regarding the baseline accuracy to be pursued. They might also have different preferences about the types of errors conceded at each point. While the preference function of each stakeholder is likely to be persistent through each of the decision phases, the magnitude of the preferences towards a certain outcome increases at each interval, as the stakes get higher. For example, the accused on one hand might emphasise that a false positive error is not made during the decision to open an investigation, but he might feel much stronger about such mistakes at the final decision point where he might potentially receive a hefty fine without being guilty of an infringement. The competitor might, on the other hand, hold stable preferences about false negative errors as the proceeding progresses towards the final decision point. For him the stake is always the same at each point with the prospect of the case being falsely dropped. The consumer, being neutral about the types of errors conceded, at the same time might emphasise the cost-efficiency ratio of the overall accuracy achieved and the corresponding enforcement investment. In that case, the consumer would require that the continuation of a case at each decision point should be beneficial towards reaching the accurate outcome, given the additional resources needed to proceed. The taxpayer would base his or her preferences on similar cost-benefit analysis

32 ibid paras 77–114. 33 Note however cartel proceedings where the decisions to open investigation and to issue a SO are often taken simultaneously.

60  The Structure and Function of EU Competition Procedures considerations, but would consider the investment in a more macro-economic manner, rather than just from the perspective of consumer welfare. After the Commission has concluded the administrative procedure with the imposition of a penalty or a fine, parties to the case that can show ‘direct and individual concern’ with the result can apply to the Court of Justice to have the decision reviewed based on Article  31 of Regulation 1/2003 EC and Article 263 TFEU.34 The provision stipulates that the Court of Justice shall have ‘unlimited jurisdiction’ to review the Commission’s decisions of that type, and that it may ‘cancel, reduce or increase’ the imposed sanctions. It follows from Article 256 TFEU, that the General Court has a first instance jurisdiction to hear applications to review competition decisions of the Commission, and that the General Court’s findings in such cases can only be appealed on points of law to the Court of Justice. This effectively makes the General Court the last instance to argue facts in competition cases, but the Court of Justice is the ultimate instance for arguing on points of law. Article 281 TFEU stipulates that a special protocol should be made for the statute of the CJEU, which can be amended by the Council and the Parliament through the normal legislative procedure. Additionally, Article 253 stipulates that the Court shall establish its rules of procedure that should be approved by the Council. The Statute of the CJEU and the Rules of Procedure, lay down in details the workings of the procedures before the Court of Justice and the General Court.35 The nature of the decision mechanism in competition proceedings before the Court of Justice is different from its nature before the Commission. The role of the Commission is reduced to that of defending its own conclusion at the administrative stage and a separate judicial body overtakes the adjudicative function. The intervals of framing and deciding are however in place; before the General Court decides there is a framing stage that involves written submissions from the parties and an oral hearing before the panel of judges, and before the Court of Justice decides there is also the opportunity to submit written pleadings and to state arguments in an oral hearing. Additionally, an Advocate General (AG) normally submits an independent written opinion to the Court of Justice during the latter framing phase at the judicial stage. An important difference between the judicial and the administrative procedure is that an annulment decision by the General Court can be appealed to the Court of Justice on points of law; a decision finding no breach before the General Court does thus not end the procedure, unlike in the

34 This criterion, which is now stipulated in Art 263 TFEU (ex-Art 230 TEC), was further developed in Case 25/62 Plaumann v Commission, EU:C:1963:1 into what is now known as the Plaumann test. The test makes it very hard to judicially challenge a Commission decision unless one is a privileged party to the procedure leading up to it. 35 Protocol (No 3) on the Statute of the Court of Justice of the European Union, annexed to the Treaties, as amended by Regulation (EU, Euratom) No 741/2012 of the European Parliament and of the Council [2012] OJ L228; Rules of Procedure of the Court of Justice [2012] OJ L265, as amended by [2013] OJ L173.

Instrumental Structure and Normative Function  61 procedure before the Commission where the procedure could be terminated at any decision point through a no-breach finding. The preference function of the stakeholders is likely to correspond to the function during the administrative procedure before the Commission but, again, the intensity level is likely to increase as it draws closer to the final and the ultimate instance; before the Court of Justice on points of law, and before the General Court on points of fact.

IV.  Instrumental Structure and Normative Function The EU’s competition law procedures can be analysed based on their instrumental function. As explained in chapter two, the instrumental function of adjudicative procedures consists of two functional elements: the framing function; and the decision function. By analysing the constitutive rules that control the EU’s competition law procedures it should be possible to identify and categorise individual provisions based either on their function towards facilitating the framing phase of the procedure, or the decisional phase of the procedure, or both. Alternatively, adjudicative procedures can be analysed based on their normative function. As previously explained, a procedure seeks to facilitate its substantive agenda with as much accuracy as materially possible; the normative agenda thus becomes accuracy and efficiency. On this account, it should be possible to identify four types of procedural provisions: on one hand provisions that instrumentally focus on the framing phase and either seek to facilitate accuracy or efficiency of the framing; and on the other hand provisions that instrumentally focus on the decision phase and seek to facilitate accuracy or efficiency of that functional capacity. To make things a bit more complicated, some provisions might contain hybrid functions in either the instrumental capacity or the normative capacity, or both at the same time. However, the hybrid provisions should nonetheless be composed of any of these two instrumental and two normative elements. Some procedural provisions are on their face not obviously procedural in character, as opposed to substantive. This applies, for example, to provisions providing for reprimand powers and powers to take certain types of decisions, typically located among the procedural provisions. Such provisions can be viewed as being ends rather than means; ie a punishment or a specific decision type becomes the consequence the subjects of the procedure strive to avoid. Viewed more broadly, however, the tool of reprimand is properly seen as an instrument for achieving a superior objective; a specific type of punishment is thus not an end but instead a means towards achieving something of substantive importance. The same applies to provisions describing the availability of types of decisions; in the intermediate term, they seem substantive in character, but viewed holistically, their procedural character dominates. What matters in this sense, is that in their capacity as

62  The Structure and Function of EU Competition Procedures intermediary substantive ends, these provisions depend on the rationale of somehow being facilitators of the superior objective. In what follows, I will treat this kind of provisions as procedural towards the grand objective of the competition law regime, not as substantive in the narrow intermediate sense of the procedural regime.36

A.  The Constitutive Rules of EU Competition Law Procedure The enforcement of Articles 101 and 102 TFEU in typical infringement cases is, in accordance with Article 103 TFEU, regulated by Council Regulation EC 1/2003, which grants the Commission with enforcement powers in the field of competition law.37 Based on Article 33 of Regulation EC 1/2003, the Commission has further regulated its own enforcement procedures through Commission Regulation EC 773/2004.38 The competition procedure as laid down in the TFEU and Regulation EC 1/2003 evolves around the question of whether a specific behaviour constitutes a prohibited conduct with regards to Articles 101 and 102 TFEU. Article 7 of Regulation EC 1/2003 empowers the Commission to prohibit conduct that infringes the Treaty provisions. The Commission can also take commitment decisions, where the infringing firms offer behavioural or structural commitments that address the Commission’s preliminary concerns.39 Additionally the Commission may decide that Articles 101 and 102 TFEU are inapplicable due to ‘Community public interests’ and the Commission can by a decision order interim measures to bring urgent competition law infringements to an end.40 Through Regulation EC 1/2003, the Council grants the Commission with specific powers to enforce Articles  101 and 102 TFEU. Some of the provisions define the jurisdiction of the Commission both in and of itself and with regards to national competition authorities (NCA), national courts and the cooperation between the different bodies, ie Articles 3, 4, 5, 6, 11, 12, 13, 14, 15, 16, 22, 29 and 32. The transitional, amending and final provisions of the Regulation, ie Articles  34–45, have little relevance for this analysis with the exception of Article 35, which directs the Member States to designate a national authority with the task of enforcing Articles  101 and 102. Such designation signals a required investment in enforcement resources that increases the overall enforcement capacity, which should translate into an increased total accuracy of the enforcement regime. 36 For further discussion on the distinction between the substantive and the procedural a reference is made to chapter two. 37 Council Regulation (EC) No 1/2003 on the implementation of the rules on competition laid down in Articles 81 and 82 of the Treaty [2003] OJ L1/1. 38 Commission Regulation (EC) No 773/2004 relation to the conduct of proceedings by the Commission pursuant to Articles 81and 82 of the EC Treaty [2004] OJ L123/18. 39 Article 9 of Regulation EC 1/2003. 40 Articles 8 and 10 of Regulation EC 1/2003.

Instrumental Structure and Normative Function  63 The provisions of Regulation EC 1/2003, which control the EU’s competition procedure, are set out in Table 3.2. Their instrumental and normative function is also defined. Table 3.2  Instrumental and normative function of Council Regulation (EC) 1/2003 Article no

Description

Instrumental function

Normative function

2

Burden of proof: The party alleging Decisional infringement carries the burden, but reverses if the accused claims an exception

Accuracy/ efficiency

7

Prohibition decision: Decision type available to DG Competition

Decisional

Efficiency

8

Interim measures: Decision type available to DG Competition

Decisional

Efficiency

9

Commitments: Decision type available to DG Competition

Decisional

Efficiency

10

Inapplicability: Decision type available to DG Competition

Decisional

Efficiency

17

Sector inquiry: Investigative tool available to DG Competition

Framing

Accuracy/ efficiency

18

Information request: Investigative tool available to DG Competition

Framing

Accuracy/ efficiency

19

Statement: Investigative tool available to DG Competition

Framing

Accuracy/ efficiency

20

Inspection: Investigative tool available to DG Competition

Framing

Accuracy/ efficiency

21

Further inspection: Investigative tool available Framing to DG Competition

Accuracy/ efficiency

23

Fines: Reprimand available to DG Competition

Decisional

Efficiency

24

Penalties: Reprimand available to DG Competition

Decisional

Efficiency

25

Limitation of penalties: Curbs the reprimand powers of DG Competition

Decisional

Efficiency

26

Limitation of penalties enforcement: Curbs the reprimand powers of DG Competition

Decisional

Efficiency

27

Right to be heard: Investigative duty of DG Competition

Framing

Accuracy

28

Professional secrecy: Investigative duty of DG Competition

Framing

Efficiency (continued)

64  The Structure and Function of EU Competition Procedures Table 3.2  (Continued) Article no

Description

Instrumental function

Normative function

30

Publication of decisions: Decision duty of DG Competition

Decisional

Accuracy

31

Review by the CJEU: Limitation of DG Competition’s decision powers

Decisional

Accuracy

The provisions of Regulation EC 773/2004, which control the EU’s competition law enforcement procedure, are listed in Table 3.3. Their instrumental and normative function is also defined. Supplementary to Regulation EC 1/2003, the Regulation regulates in more details the initiation of Commission proceedings in Article 101 and 102 cases, and the handling of complaints and the organisation of hearings. Table 3.3  Instrumental and normative function of Commission Regulation (EC) 773/2004 Article no

Description

Instrumental function

Normative function

2

Initiation of proceedings: DG Competition can use investigative powers and reject complaints prior to opening of formal proceedings

Framing/ decisional

Efficiency/ accuracy

3

Power to take statements: Supplementary to powers to investigate in Reg 1/2003

Framing

Accuracy/ efficiency

4

Questions during inspections: Supplementary Framing to powers to investigate in Reg 1/2003

Accuracy/ efficiency

5

Admissibility of complaints: Criteria for submitting complaints

Framing

Accuracy/ efficiency

6

Procedural participation of complainants: Rights of participation

Framing

Accuracy

7

Rejection of complaints: Procedure for rejection of complaints

Decisional

Efficiency

8

Access to info: Access rights of the complainant

Framing

Accuracy

9

Overlapping complaints: Complaints can be rejected if being processed by another competition authority

Decisional

Efficiency

10

Replies to SO: Supplementary to rights in Reg 1/2003

Framing

Accuracy

11

Right to be heard: Supplementary to rights in Reg 1/2003

Framing

Accuracy (continued)

Instrumental Structure and Normative Function  65 Table 3.3   (Continued) Article no

Description

Instrumental function

Normative function

12

Right to an oral hearing: Supplementary to rights in Reg 1/2003

Framing

Accuracy

13

Hearing of others: Supplementary to rights in Reg 1/2003

Framing

Accuracy

14

Conduct of oral hearings: Supplementary to rights in Reg 1/2003

Framing

Accuracy/ efficiency

15

Access to file: Supplementary to rights in Reg 1/2003

Framing

Accuracy

16

Confidentiality: Supplementary to rights in Reg 1/2003

Framing

Efficiency

17

Time limits: The articulation of various deadlines in a proceeding

Framing/ decisional

Efficiency

The tables above show that many of the provisions of the two main procedural regulations for EU competition law procedures relate to the accuracy and the efficiency of the framing phases of the procedure by articulating various framing tools at the disposal of the Commission and by conditioning the access of various parties to different sections of the framing phases. Many of the provisions also relate to the efficiency of the decisional phases by defining the types of decisions that can be reached at various decision points, for the ease of the decision-makers and parties alike. Few of the provisions relate to the accuracy of the decisional phases, which still is of great importance for the overall success of the procedure. The provision on the burden of proof implies a premium on a specific type of accuracy but does not condition the level of overall accuracy. The exposure of the Commission’s decision to the public through publication creates an incentive for decisions to be consistent with common rationality, and the potential to appeal the decisions to the CJEU generates similar pressure of accuracy on the decision process. The relative absences of provisions regulating the accuracy of the decision points in the main procedural regulations can partly be explained with reference to two reasons: firstly, the procedural regulations mainly deal with what the Commission can decide in the context of competition law, but less with how it decides; and secondly, while individual provisions do perhaps not directly engage the problem of decisional accuracy, the aggregate effect of the system defined in the procedural regulations, organised in three intervals of framing and decisional phases, is towards decisional accuracy. But there is more to it. The provisions that seem absent from the formal constitutive rules of the procedural regime have their representation in the informal institutional culture of DG Competition.

66  The Structure and Function of EU Competition Procedures

B.  The Institutional Culture of DG Competition The Commission’s internal decision-making mechanisms are of considerable importance for procedural accuracy in the context of competition law proceedings. The Commission acts in many different fields of EU law, but the application of the competition provisions is nonetheless in many ways procedurally unique in terms of how decisions are taken. The core Treaty provisions on how the Commission acts are found in Article 17 TEU and Articles 248–250 TFEU, and on their basis, the Commission has adopted the Rules of Procedure of the European Commission.41 Significantly, Article 17(6) TEU orders that the Commission shall act ‘as a collegiate body’, and Article 250 TFEU dictates that the ‘Commission shall act by a majority of its members’ conditioned on a quorum that in Article 7 of the Rules of Procedure is defined as the majority of the members of the Commission as defined in the Treaties (ie Articles 17 TEU and 244 TFEU). The basic structure of the decision mechanism of the Commission is a majority decision by the members of the Commission, ie the Commissioners, although in practice the Commission usually acts by consensus.42 The ultimate power to take decisions in the context of competition law proceedings thus rest with the College of the Commissioners of the Commission. The Commission is organised into different Directorates General (DGs) which are given different tasks for which the Commission is responsible. The Directorate General for Competition (DG Competition) is responsible for the enforcement of the competition provisions of the Treaties, including Articles 101 and 102 TFEU. The president of the Commission allocates one of the members of the Commission responsibility for the tasks carried out by DG Competition.43 This member prepares proposals and makes suggestions to the president of the Commission to put proposals relating to the enforcement of the competition provisions on the agenda of the Commission’s meetings.44 DG Competition is further organised into several directorates; these are composed of units which are further divided into case teams that are responsible for individual cases. The head of unit manages and is responsible for the work of the case team. Any proposed action on behalf of DG Competition that requires the approval of the Commission goes though the Director General, the most senior official within DG competition, which makes proposals to the Commissioner responsible for the competition portfolio. The Commissioner, if in agreement with the proposal, requests to get it on the agenda of the Commission’s regular meetings. 41 See The consolidated version of the Rules of Procedure of the Commission [2000] OJ L308/26. Note that the Rules were adopted before the reorganisation of the Treaties following the Lisbon Treaty in 2009 and were thus adopted with references the old Treaty numbers. 42 Article 8(2) of the Commission’s Rules of Procedure implies the practice of deciding by consensus. A ‘vote shall be taken if any Member so requests’. If no one asks for a vote, the Commission decides based on a consensus. 43 See Art 17(6)(a) TEU, Art 248 TFEU and Art 3 of the Rules of Procedure of the Commission. 44 See Ars 6 and 19 of the Rules of Procedure of the Commission.

Instrumental Structure and Normative Function  67 Before the Commission adopts a final decision in Article 101 and 102 TFEU cases, it has gone through several layers of internal framing mechanism that are intended to increase the accuracy of the eventual decision-making. The main object of accuracy is the work carried out by the case team and the proposed action the team makes. At the first instance, the case team must convince the Director General to make a proposal to the Commissioner. The Director General can seek advice from the Chief Competition Economist, who reports directly to him, on matters relating to the economics and the econometrics of the proposed action. He can also agree with the Commissioner to set up a peer review panel on selected cases either before or after the issuance of an SO. If peer reviewing is decided, a special team is assembled to scrutinise the case file and the proposed action by the case team, with the purpose of screening for errors and suggest improvements. The peer review team reports to the Director General and provides an input into his decision on whether, or what, to propose to the Commissioner. If the proposal of the case team survives the scrutiny of the Director General, it moves on to the second instance, where the Commissioner must decide on whether to propose it to the College of Commissioners. The Commissioner will normally take note of and base his decision on the input provided by the Chief Economist, the report of the peer reviewing team (if available), and the action proposed by the Director General.45 The third and the final threshold for adopting a Commission decision on matters of competition law is the College of Commissioners. In addition to the inputs already mentioned, the Commission seeks the advice of the Commission’s Legal Service in accordance with Article  23(3) of the Commission’s Procedural Rules and, in accordance with Article 14 of the Regulation EC 1/2003, an advice must be sought from an Advisory Committee composed of the representatives of the NCA of the Member States. The Hearing Officers of the Commission, institutionally organised to be independent of DG Competition, are responsible for monitoring the compliance of the Commission with procedural rights and guarantees throughout the proceedings in competition law cases.46 Before a final decision is taken, the Hearing Officer reports to the competent Commissioner (and several others) on the compliance with procedural standards and on whether the parties to the case have been given opportunity to exercise their right to be heard with regards to all of the objections raised in the proposed decision.47 In reaching a final decision in an infringement case of Article 101 or 102 TFEU, the College of Commissioners will thus have access to independent expert opinions on different 45 The workings of these internal checks and balances mechanisms is explained in a short document titled ‘Proceedings for the application of Articles 101 and 102 TFEU: Key actors and checks and balances’, available at http://ec.europa.eu/competition/antitrust/key_actors_en.pdf. 46 See further on the Hearing Officers in Decision C (2011) 5742 of the President of the European Commission on the function and the terms of reference of the hearing offices in certain competition proceedings [2011] OJ L275/29. 47 See Arts 14 and 16 of Decision C (2011) 5742 on the function and the terms of reference of the hearing offices in certain competition proceedings.

68  The Structure and Function of EU Competition Procedures aspects of the proposed decision, which should assist in facilitating an accurate outcome of their deliberation. The complex formal and informal institutional culture of DG Competition to an extent address and regulate the procedural aspect, concerning accuracy of the decision phase, that seem underrepresented in the formal constitutive rules found in the procedural regulations. Transparency is of essence with regards to the institutional culture of DG Competition and the College of Commissioners, otherwise these self-imposed restraints could become the subject of suspicion by parties external to the European Commission. In general, a healthy and professional institutional culture assists and facilitates the ultimate adjudicative task of procedural regimes, while corrupt or dysfunctional institutional culture could hamper the same task. The same will apply to the EU’s competition procedural regime.

C.  The Procedure before the CJEU The powers of the Court of Justice to review competition decisions of the Commission are based on Article 263 TFEU and more specifically on Article 31 of Regulation 1/2003 EC. The former grants the Court with general powers to review the legality of the Commission’s decisions, and the later grants the Court with a more specific unlimited jurisdiction to review competition decisions where a fine or a penalty has been imposed. The decisional function of the Court is conditioned by a requirement of independence of the Court’s members, stipulated in Article 19 TEU and Articles 253 and 254 TFEU. Essentially, this means that the Court should be impartial with regards to the stakeholders and their claims. The Court is further required by Article 19 TEU to ensure that ‘that the law is observed’ in the application of the Treaties and secondary legislation. This requirement methodologically conditions the Court to use the adjudicative mode of deciding; it should not decide based on its on preferences, but should observe the existing laws and use them to anchor its adjudication. The Treaties articulate the decisional jurisdiction of the Court, dictate the decisional modality the Court should use to reach its conclusions, and stress the prerequisite of the Court’s independence. The details of the decisional mechanism are further described in the Statute of the Court of Justice and in the Rules of Procedure of the Court of Justice. In many ways, these more detailed procedural regulations correspond to the procedural regulations of the Commission. The provisions that deal specifically with procedure can be categorised in the same way depending on whether they instrumentally deal with the framing or the decisional phase of the Court’s procedure, and whether they normatively deal with the accuracy or the efficiency of the two phases. Before the Court, the quality of the framing process is dependent on the parties to the case. If they provide high-quality written submissions and argue their case effectively at the oral hearings, the decisional task of the Court is facilitated. If, however, the parties fail to plead their arguments effectively during the written or

Instrumental Structure and Normative Function  69 the oral part of the procedure, it is likely to be reflected in the quality of the Court’s outcome, since the Court is to an extent bound by how the case is argued before it. The input of the Advocate General increases the quality of the framing phase; it provides the Court with impartial assessment of the case and the pleadings of the parties that it can choose to follow or, if in disagreement, seek an alternative approach. A two-tiered court procedure is an expensive measure to enhance the decisional accuracy of the procedural process in competition law, especially considering the preceding multi-layered administrative procedure before the Commission, which should already guarantee a high level of accuracy. Judges of a high quality are expensive experts and the necessary institutional infrastructure is also expensive. The Court of Justice is organised into chambers of judges (three, five, 15, or the Grand Chamber of all judges) that precede over individual cases, and each judge has a cabinet of supporting staff that assists them in composing draft judgments in cases where the judge acts as the judge-rapporteur, and assists them in taking positions in cases where other judges act as the judge-rapporteur. The organisation of the Court guarantees that each case is scrutinised by all the judges in the chamber and their respective assisting cabinet members. Many legal experts of the highest degree thus review important decisions that are assigned to the Grand Chamber. This organisation facilitates decisional accuracy, although at a high cost in terms of expert resources.

D.  Institutional Sanctions A simplified view of procedural institutions, such as the EU’s competition procedures, presumes a clever procedural design that facilitates the task of the institution performed by rational role occupants who accomplish their roles diligently in accordance with the constitutive rules and the prevailing institutional culture applicable to their respective roles. A more complete view will however need to account for the human element in any social institution. This human element makes every occupant of a role in a social institution susceptible to a persistent duality. On one hand is the institutional role, and on the other hand the role of a private human being. The ends pursued through these two different roles may or may not be compatible, which creates a risk that a person will pursue ulterior ends in his or her capacity as a role occupant in a particular social institution. In other words, there is a persistent temptation to adopt an ‘I’ attitude instead of a ‘we’ attitude.48 Institutional corruption occurs when individual role occupants intentionally or negligently perform their institutional roles towards their private agenda, rather

48 This is discussed further in chapter two. See also Raimo Toumela, The philosophy of social practices: a collective acceptance view (Cambridge, Cambridge University Press, 2002) 17–39, 162–64 and 176.

70  The Structure and Function of EU Competition Procedures than the institutional agenda. These private agendas can be of many different sorts, eg a private pursuance of power, prestige, wealth, etc. The archetypical example of an institutional corruption is the official that takes bribes in exchange for performing his or her duties in a specific way. But corruption in social institutions often takes more subtle forms. In the context of adjudicative institutions, this can, for example, happen when an investigator’s sense of justice induces him or her to conceal or ignore evidence that may be favourable to the accused. The investigator privately feels that the accused is guilty and allows that belief to influence how he or she performs their assigned institutional role and thus risks undermining the overall institutional agenda.49 Due to the risk of corruption, social institutions typically include a formal or an informal system of sanctions intended to induce its different role occupants to act towards the institutional agenda, instead of their private ends. The EU’s competition procedure includes such elements. For the more serious forms of institutional corruption, such as taking bribes, normal criminal sanctions apply. For the softer forms of corruption, which includes procedural mistakes performed with intention or by negligence, several counteractive procedural features can be identified. As mentioned above, one of the cultural aspects of DG Competition’s internal workings involves a peer review of the case file by an independent team of case handlers. This aspect of the procedure can be viewed as a countermeasure against a potentially biased view of case handlers on the facts of a case, which potentially would lead them to recommend moving forward with a case irrespective of exonerating facts. The mere possibility of having his or her view on the case file scrutinised should provide an incentive for the case handler to stay within his or her institutional role. It would reflect badly on their standing within DG Competition if their work could not survive such scrutiny and thus function as an informal sanction with a potential effect on their career prospects within the organisation.50 The peer reviewing process is primarily aimed at the internal workings of individual case teams and units within DG Competition. The role of the Hearing Officer can be viewed in a similar way. Among the Hearing Officer’s tasks is to verify whether the investigation and decisional processes within DG Competition have been undertaken in accordance with the applicable rules and that the procedural rights of the concerned undertakings have been duly respected. Institutionally independent of DG Competition, the Hearing Officer reports to the responsible Commissioner and thus provides an incentive for the top administrative layer of DG Competition to stay on the institutional task in their recommendations to the College of Commissioners. The possibility of a judicial review and an appeal of the findings of the General Court to the Court of Justice puts corresponding institutional pressures on the College 49 This concept of institutional corruption draws on Miller’s definition. See Seumas Miller, The moral foundations of social institutions (Cambridge, Cambridge University Press, 2010) 154–75. 50 Note Damaska’s discussion of informal sanctions within rigidly hierarchal organisations in Mirijan R Damaska, The faces of justice and state authority (New Haven, Yale University Press, 1986) 21.

Arguments about Procedural Fairness in EU Competition Law  71 of Commissioners and the judges of the General Court. If they wish their work to survive scrutiny, they need to adjudicate with accuracy and efficiency the competition case at issue.

V.  Arguments about Procedural Fairness in EU Competition Law The political characteristics of the College of Commissioners can be considered a source of contention with regards to its integrity as a decision-making body in legal adjudication.51 Although the Commissioners and the Commission are supposed to work independently of any national interests, according to Article 17(3) TEU and 245 TFEU, they are usually chosen from the top layer of the political elites of their respective Member States. As such, the Commissioners are usually apt at political decision-making, where interests and preferences reign supreme as motives for action. In many fields of the Commission’s work and decisions making a political background is very useful, since large parts of the Commission’s competences concern policy and policy-making. The process of legal adjudication and political decision-making is however distinct in important ways. While special interests and preferences are supposed to be inputs in political decisions with an ex-ante perspective, legal adjudication is supposed to be based primarily on a rational inference from the relevant facts and the applicable laws from an ex-post perspective.52 While the Commissioners of the Commission are often not specialist in the normal process of legal adjudication, and despite that all but one of them deal primarily with other issues than the competition portfolio, the institutional design of the competition law procedure still provides some guarantees that the outcome of individual competition law cases is decided based on the legal adjunction modality, rather than on the political modality. The key guarantee is the one provided for in Article 263 TFEU and Article 31 of Regulation EC 1/2003 regarding the possibility of having the competition law decisions of the Commission reviewed by the CJEU. The requirement in Article 31 of Regulation EC 1/2003 that the Commission’s decisions in competition law proceedings shall be published has a similar, but weaker, effect. Unlike the Commission, the CJEU is required by Article  19 TEU to use the decision modality of legal adjudication in its interpretation of the competition 51 Ian Forrester notes the risk of political bias in the Commission’s decision mechanism where the Commissioners, usually politicians, decide the result of a public prosecution. See Ian S Forrester, ‘Due process in EC competition cases: a distinguished institution with flawed procedures’ (2009) 34 European Law Review 817, 822 and 831–33. 52 In politics, we think about what we should do in the future based on the current desires of people, while in the court room we think about what happened in the past and how the existing laws apply to that happening. The distinction is in practice more nuanced and not always clear-cut, but as a rule of thumb this is a useful distinction, especially concerning what kind of thought processes ought to guide the different types of decision-making.

72  The Structure and Function of EU Competition Procedures provisions of the Treaties; ‘[the CJEU] shall ensure that in the interpretation and application of the Treaties the law is observed’. Naturally this requirement on the decision body at the final instance of a competition law proceeding has effects on all decision points leading up to the final instance. However, the effect is not uniform since only some results are caught by the effect during the administrative procedure before the Commission. As previously explained, the Commission proceedings in Article  101 and 102 TFEU cases are organised in three intervals of framing and decision phases. At each interval, there is a decision point where the Commission can decide to drop the case, or to continue towards an infringement decision. At each of these decision points, the potential of an eventual review by the CJEU forces a legal adjudicative modality of deciding if the Commission opts to progress towards an infringement decision. A decision to drop a case is not subject to the same pressure; the case is simply closed without any further possibility to have that decision reviewed.53 This has implications for the accuracy of the decisionmaking; while the framing phase might be perfectly executed in terms of accuracy by the Commission’s administrative services, the actual act of deciding might fail to produce the accurate result due to the final decision-maker’s call to drop a case on political grounds, instead of inferring a decision from the relevant facts and the applicable laws. There are, in other words, checks and balances that prevent the Commission from intentionally making false positive errors in competition cases, but less so if the Commission elects to make a false negative error. Viewed differently, but still within the accuracy/efficiency paradigm there might be, and usually are, valid efficiency reasons behind the Commission’s decision to drop cases. Even if the facts and the law recommend the pursuance of an infringement decision, the limits of the Commission’s enforcement budget might provide a valid reason for closing a case and concentrate the resources elsewhere. Such a decision can be taken validly based on the political modality of decisionmaking, rather than based on the legal adjudicative modality. A system can be envisaged, where the Commission would be required to take a formal reasoned decision irrespective of the finding of an infringement or not, which in both cases could be appealed to the CJEU.54 Such a system would force an adjudicative

53 By citing ‘community interests’ the Commission has discretion to prioritise its investigative efforts in competition cases and close cases that do not qualify as of interest for the EU community. Note the criteria for stating reasons articulated in the Automec case law, Case T-24/90 Automec v Commission, EU:T:1992:97 [85]–[86]; and more recently Case T-342/11 CEEES v Commission, EU:T:2014:60 [58]– [60]: ‘As regards the factors determining the exercise of the discretion enjoyed by the Commission under Article 9(2) of Regulation No 1/2003, it should be remembered that the latter has only limited resources, which it must use in taking action against a potentially wide range of behaviour which is contrary to competition law.’ 54 Once the Commission has opened a formal investigation into an alleged infringement to the state aid provisions of the Treaties, ie Arts 107–108 TFEU, it must close the case with a formal reasoned decision on the existence of an incompatible aid or not. Such decisions can be contested and annulled by the CJEU.

Arguments about Procedural Fairness in EU Competition Law  73 decision modality for both negative and positive infringement decisions, and thus lessen the probabilities of both false negatives and false positives. The choice of institutional design of not having such system in place can simply be viewed as a prioritisation of the available resources. The current institutional system thus emphasises the avoidance of false positives by requiring the legal adjudicative modality of taking decisions for reaching positive infringement decisions, while false negatives are prima facie permissible with reference to the political modality of deciding that permits political arguments to influence the conclusion. By examining the framing phases, prior to each of the decision points, an evolving perspective on the information that are the object of accuracy quickly reveals. The framing phase prior to the opening of a formal investigation emphasises the gathering of raw information data, uncontaminated by the agenda of the stakeholders involved. The investigative powers articulated in Regulation EC 1/2003 are meant to facilitate this initial fact-finding. The first decision point is thus primarily based on the investigator’s perspective on the relevant information. The intermediary decision point of the SO relies on the same investigative tools but allows for more input by the stakeholders and is thus not as concerned with the element of surprise as often is important during the very early stage of cases involving sophisticated conspiracies. The SO thus reflects the investigator’s perspective, slightly informed by the stakeholders. Before the final decision point, the Commission grants access to its preliminary findings and the sources underscoring that finding and invites the parties to the case and other interested stakeholders to exercise their right to be heard before the case is concluded with a final decision. The final decision at the administrative level should thus reflect both the investigator’s perspective and the stakeholder’s perspective on the relevant information. The court procedure subsequently takes a fresh, neutral view on the law and the facts that incorporates the arguments of those who participate in the court procedure at the first instance, and then again on the law at the last instance. It is important to note that while the right to be heard is potentially an important source of additional information for the facilitation of an accurate result, it can also be a venue for the parties to mislead and confuse the case. A guilty defendant facing a hefty anti-trust fine will have an enormous incentive to spend his or her resources on attempts to confuse and divert the decision-maker towards making a false positive error. Countering such attempts drains enforcement resources and leaves less for other enforcement uses. The appropriate timing and quantity of the execution of the right to be heard are thus important considerations, both in terms of ultimate accuracy of the result and the efficiency with which it is reached. The broad structure of EU’s competition law procedure has now been analysed and the main instrumental elements identified, which seek to facilitate the framing of relevant information for the taking of decisions that aim at producing accurate results as efficiently as materially possible. Some of the regulatory provisions are concerned with efficiency, while others are concerned with accuracy. Some are decisional instruments, while others are framing instruments. When combined in a holistic procedural regime, the hope of the regulator is that the aggregated

74  The Structure and Function of EU Competition Procedures average result from individual cases is at an equilibrium in terms of the result accuracy achieved and the cost with which that level is achieved. In complex procedures, such as EU competition procedures, an equilibrium is to an extent necessary at each major decision point. However, the equilibrium will not be the same at each point; as the procedure progresses towards a final finding of an infringement, the stakes get proportionally higher for some stakeholders, which justifies an increased weight of their preferences in the fair equilibrium solution. Arguments about procedural fairness essentially relate to the fair calibration of these equilibriums. Using this conception of procedural fairness in the context of EU competition law, means that it should be possible to analyse claims about procedural fairness in relation to a specific stakeholder preference, at a particular decision point in the procedure, in terms of increasing either the accuracy or the efficiency of the procedural output. Remedies to fairness claims that are well founded can be implemented by alteration of the constitutive rules of the procedure, by improvements to specific aspects of an institutional culture that corrupt a procedure, or by improving the system of sanctions that keeps individual role occupants of the procedural regime within the boundaries of their designated tasks. This conceptualisation of procedural fairness in EU competition law identifies a balancing dilemma at the core of any argument about procedural fairness. This dilemma requires a comprehensive ethical conception of fairness, which the procedural architect can use to identify the fair equilibrium solution to each balancing problem.

4 Ethics for Procedural Architects The ethical implication of the concept of procedural fairness is simply that a ­procedure should be fair. However, the precise meaning of ethics in the context of the law is difficult to specify. In legal theory, the role of ethics in the law is viewed very differently by the two mainstream approaches. On one hand, there is the naturalistic view, that ethics and law are intertwined. On the other hand, there is the positivistic view, which considers ethics and the law as separate entities. In the previous chapter a balancing dilemma was identified at the core of the concept of procedural fairness between the competing procedural considerations of accuracy and cost-efficiency. To resolve this balancing dilemma, the procedural architect must first know whether his or her positive actions determine the object of fairness in the procedural design, or whether his or her procedural design ought to comply with a pre-existing naturalistic standard of fairness. This chapter will analyse the two popular alternatives for explaining the role of ethics in the law. This analysis reveals a clue to how the procedural architect can approach the issue of ethics in his designs. Moving beyond the two main legal theory alternatives, an argument for an ethical standard will be made with reference to modern variants of Hobbesian social contract theory. The claim is that this argument provides a less controversial understanding of the link between the law and ethics than the orthodox naturalistic and the positivistic views. If successful, this argument provides a plausible narrative on which the procedural architect can base their approach to ethics in his or her legislative designs. This is a central piece in the puzzle of moving the conceptualisation of procedural fairness towards actual procedural designs in specific contexts, such as in EU competition law.

I.  Ethics and Legal Theory A mainstream view of legal philosophy holds that various types of codes of conduct, that collectively can be referred to as human-made positive law, can determine appropriate human behaviour. On this view, it is up to humans to decide how they behave, and for present purposes, to decide the standard or the object of fairness in the law. A competing view holds that although laws are human inventions or artefacts, their substance is subject to restraints deriving from a phenomenon that

76  Ethics for Procedural Architects is integrally part of the human condition, but at the same time beyond the reach of positive human intervention. Viewed in broader philosophical terms, this is a debate about the extent of philosophical determinism and to what degree humans have free will. In earlier times, natural law and divine law were considered as constants that restrained the extent to which humans could and ought to exercise their free will. Accordingly, human-made laws were to be anchored in the laws of the nature, or in the divine will. The modern version of the debate dates back to the birth of modern legal positivism in the monographs of Hans Kelsen from 1934 and HLA Hart from 1961, where they argued for a theory of positive law that is unbound by external ethical constraints.1 On the opposite spectrum, Lon Fuller, John Finnis and Ronald Dworkin contributed to a naturalistic view of the law during the second half of the twentieth century.2 The defining thesis of legal positivism, about the separation of ethics and law, needs to overcome the fact that judges occasionally base their decisions on considerations that, strictly speaking, cannot be considered as a part of the law in the positivist sense. This supposedly happens during the interpretation of legal principles, as opposed to the application of simple legal rules.3 The standard positivist answer to this usually takes one of two forms. The first builds on a move to include moral arguments as the positive sources of law. This is made possible by accepting moral considerations as part of the law on the condition that they have been properly validated by the legal system’s applicable processes for recognising rules. Since these processes must be a social constructions or artefacts, the law is ultimately decided by social facts alone. This approach is sometimes labelled as inclusive positivism because of the inclusion of moral considerations.4 The second answer concedes that judges frequently run out of legal sources. When this occurs, judges make new rules based on extra-legal norms that ultimately decide the case at hand. These norms could be moral norms, but since the judge is no longer engaged in application of legal sources, the separation of laws and morals holds. This approach is usually labelled as exclusive positivism, signalling the exclusion of moral considerations from the domain of the law.5

1 Hans Kelsen, Pure theory of Law (Berkley, University of California Press, 1967); HLA Hart, The concept of law, 2nd edn (Oxford, Oxford University Press, 1994). 2 Lon Fuller, The morality of law, rev edn (New Haven, Yale University Press, 1969); John Finnis, Natural law and natural rights (Oxford, Clarendon Press, 1980); Ronald Dworkin, Taking Rights Seriously (London, Duckworth, 1977); Ronald Dworkin, Law’s empire (Oxford, Hart Publishing, 1986); Ronald Dworkin, Justice for hedgehogs (Cambridge MA, Harvard University Press, 2011). 3 Dworkin famously pointed this out in his article ‘The Model of Rules’, reprinted in Dworkin, Taking Rights Seriously (1977) 14–45. 4 This approach is often associated with Jules Coleman and his followers. Hart endorsed this approach in The Postscript. For general discussion see Kenneth Einar Himma, ‘Inclusive Legal Positivism’ in Jules L Coleman and Scott Shapiro (eds), The Oxford Handbook of Jurisprudence and Philosophy of Law (Oxford, Oxford University Press, 2004) 125. 5 Joseph Raz is the key theorist associated with this approach. For general discussion see Andrei Marmor, ‘Exclusive Legal Positivism’ in Coleman and Shapiro (eds), Oxford Handbook (2004) 104.

Ethics and Legal Theory  77 Hart’s account of legal positivism was initially an attempt to reform legal positivism as described by Jeremy Bentham, and by John Austin in the Province of Jurisprudence Determined, published in 1832. This type of legal positivism was grounded on an assumption that laws were essentially the commands of the sovereign that could be implemented by force. The command and the threat of force was thus an integral part of the concept of law. Hart argued that the command theory could not explain why the orders of the armed bandit should not be considered instances of legally valid rules.6 In response to the insufficiency of the command theory of Bentham and Austin, Hart suggested the rule of recognition, which resembles Kelsen’s grundnorm, both of which became essential features of modern positivism.7 In short, in order for a norm to become a valid law within a legal system, it must have been validated by a superior rule (Hart’s version) or be compatible with the systems ultimate norm (Kelsen’s version). The principal purpose of the rule of recognition was to provide an authority for rules that made a distinction between the order of the gunman and the order of the legitimate official. This Hart managed to do without resorting to moral authority, by creating a master rule of other rules that legitimised orders of a predefined sort. If the order of the gunman were not sanctioned by the master rule as a source of law, it could not become an instance of a valid law. To emphasise the core element of legal positivism it is helpful to visualise the puzzle of legality as a question that needs to be answered based on certain facts.8 These facts can be termed ‘legal facts’, which distinguish them from other facts that are not relevant. The task is then to determine the composition of the legal facts to enable extraction of the answer to the question of legality. In its basics, positivist theory irreducibly maintains that all legal facts are determined by social facts alone.9 What is meant by ‘social facts’ remains flexible, but in a broad sense it refers to an entity that can be observed by the five human senses and subsequently described. A social fact thus observably is and refers to a ­posteriori knowledge. Consequentially, the question of legality on the positivistic view must be answered with reference to a knowledge that is observably present, while in contrast, the naturalistic approach would additionally consider moral facts that exist irrespective of what can be observed in the present. Moral facts refer to a priori knowledge that is intrinsically true and thus independently authoritative. Moral facts command how things ought to be, but do not describe how things presently are. The answer to the legality question in the naturalistic view is thus derived from a combination of how things are, and how they morally ought to be. 6 See Hart, The concept of law (1994) chs 2–4. See also HLA Hart, ‘Positivism and the Separation of Law and Morals’ (1958) 71 Harvard Law Review 593, 603; ‘Law surely is not the gunmen situation writ large, and legal order is surely not to be thus simply identified with compulsion.’ 7 Hart (n 1) 95. 8 Scott J Shapiro, Legality (Cambridge MA, Belknap Press, 2011) 25–27; Mark Greenberg, ‘How Facts Make Law’ (2004) 10 Legal Theory 157. 9 Shapiro, Legality (2011) 27.

78  Ethics for Procedural Architects The question of legality is a question of a priori knowledge, that is, what ought to validly direct an action. It is not a question of a posteriori knowledge. The point of reference is not what results of an action, but instead what directs an action. Hart’s concept of law, by ascribing to normative concepts like authority and obligation, is in this sense a description of causal effects. This, according to Shapiro, potentially puts positivism at odds with Hume’s fork.10 Hume formulated a maxim that dictates that an ‘ought’ cannot be derived from an ‘is’, meaning that for an output to be necessarily true, the input also has to be a priori true knowledge.11 If, however, the input is dependent on what is observable, the truth of output is contingent on the limits of the observation. To put this in context, the rule of recognition is an observable social rule that produces rules about how things ought to be. The output is thus formulated as universally true; it is based on the a priori knowledge that the rule of recognition is authoritative and that there is an obligation to abide to it without any reservation. The problem is that the rule of recognition itself is formulated as descriptive and observable, and is thus contingent upon what can be observed and described. The question then becomes whether it is possible to create something universally true, out of premises that are contingent upon what can be positively observed. Hume would deem that problematic. The output of contingent premises must be contingent as well. Hart explained that his master rule, that validates other rules, exists based on a social practice that is concordantly followed by the officials of the system. Its existence is thus a matter of a social fact.12 By this move, Hart attempted, under the influence of philosophers such as JL Austin, to defuse the concept of the rule of recognition of its normative charge as the ultimate source of legality.13 The absolute source of the rule that dictates legality is thus contingent on the social fact that the officials of the system follow it. Shapiro argues that this spin creates a new problem. If the rule of recognition is plainly an expression of a social practice, the question becomes about which social practices form the rule of recognition and which do not. It cannot be that everything that constitutes a social practice is an element of the law’s master rule. Another rule of recognition would be needed to distinguish the specific social practices that form the foundations of the first-level master rule. It would be possible to repeat the regression infinitively.14

10 ibid 47. 11 David Hume, Treaties of Human Nature (1738–40, David Fate Norton and Mary J Norton eds, Oxford, Oxford University Press, 2007) bk III, s I, last paragraph. 12 Hart (n 1) 110: ‘[T]he rule of recognition exists only as a complex, but normally concordant, practice of the courts, officials, and private persons in identifying the law by reference to certain criteria. Its existence is a matter of fact.’ See also pp 254–59 in the Postscript where Hart discusses Dworkin’s criticism of the practice theory. On p 259 Hart describes the core claim in the following way: ‘[T]he law of a system is identified by criteria provided by a rule of recognition accepted in the practice of the courts …’. 13 See discussion on the link between JL Austin’s and Hart’s work in section II of chapter 2. 14 Shapiro (n 8) ch 4, esp 95–96.

Ethics and Legal Theory  79 It thus seems difficult to give an entirely descriptive account of a normative claim without a reference to a normative authority of some kind, just as Hume suggested. One will always perceive through the senses more information than is relevant for describing a normative claim. To decide which information is relevant, a further normative judgement is required. What ought to be taken as an instance or element of a given normative claim, out of the total sum of instances that are perceived? The answer requires a reference to something that ultimately is the normative reason for the normative claim.15 The other alternative is to surrender to the notion of collective acceptance; if something is collectively accepted it becomes a social fact that can be described in the positivistic sense. There is no need to look any further for answers to what is; it just is. Collective acceptance thus becomes the ultimate deontic property. Hart’s idea to replace the command model with the rule of recognition was and is ingenious for many purposes. Upon closer philosophical scrutiny, however, the move is not immune to the logic of Hume’s fork. The rule of recognition initially created the impression that a normative claim about the law could be based on a descriptive premise. A closer look at the premise shows that to maintain the impression, one would need to repeat the move indefinitely. Attempts of legal positivists, after Hart, bear the mark of this hopeless regression. The arguments have become ever more abstract, narrower in scope of application, and increasingly detached from the practical reality of legal practitioners.16 Later, Dworkin deemed these attempts to save the positivist project as uninteresting, because they did not have anything interesting to contribute outside the narrowly defined discipline of legal philosophy; not to more abstract disciplines such as political philosophy, or to less abstract disciplines such as the practice of lawyers and judges.17 Perhaps that is the key to appreciating the merits of the legal positivist project: it may work within the narrow confines of the discipline of legal theory, but once the forces of the practical reality and the wider philosophical context start to pull, the project reaches its limits of utility.

15 Burazin suggests that laws and legal systems could be perceived as institutional artefacts that had an author, with an intention, and which had been collectively accepted. The normativity of the legal system, understood in this way, would come through the second condition about the author’s intention. See Luka Burazin, ‘Can there be an artefact theory of law?’ (2016) 29 Ratio Juris 385, 397–99. 16 Brian Leiter summarises the status of the debate in a recent article and proclaims (once again) a positivist victory over the Dworkinians and the naturalists. If it is a positivist victory, it may have been a Pyric one; the claim left standing is reduced to a narrow statement that positivism is the best theory to explain what the ordinary man thinks about the laws: ‘if we take seriously Hart’s explicit theoretical aim of doing justice to what the ordinary man understands about the modern municipal legal system, then we have no better theory than positivism’. See Brian Leiter, ‘Legal Positivism About the Artifact Law: A Retrospective Assessment’ in Luka Burazin, Kenneth Einar Himma and Corrado Roversi (eds), Law as an Artifact (Oxford, Oxford University Press, 2018) fn 39. 17 Ronald Dworkin, ‘Hart’s Postscript and the Character of Political Philosophy’ (2004) 24 OJLS 1, 36–37. Brian Leiter responded to this criticism on the legal positivist project in a critical article: Brian Leiter, ‘The end of Empire: Dworkin and the Jurisprudence in the 21st Century’ (2005) 36 Rutgers Law Journal 165, 178.

80  Ethics for Procedural Architects It should be noted that the other major branch of legal theory, the natural law tradition, is compatible with Hume’s fork and does thus not commit the sin of deriving an ‘ought from an is’ in the same way as legal positivism. The natural law tradition claims that the ultimate authority that supports any legal rule is its moral appropriateness. Morality, on this view, is a priori to any man-made law and is as such independently authoritative; one ought to follow the law because it is the ethically or morally right thing to do. Ethics standards, unlike the rule of recognition, are authoritative in the virtue of themselves. A person does not need any further reasons than ethical appropriateness to have a valid reason for any given action. Despite this quality, the natural law tradition has its own set of controversies. A typical objection capitalises on the indeterminacy of the concept of ethics and morality; a judgement of moral appropriateness often lies in the subjective eye of the beholder and is thus hard to establish objectively. Natural law theory is also often challenged with the paradox of the laws of Nazi Germany; if moral appropriateness is a function of legality, how can immoral laws exist? When pushed to their limits, the two main branches of legal theory rely on two incompatible assumptions. Natural law theory relies on the assumption of pre-existing normative ethics standards and positivist legal theory relies on the assumption of a normatively neutral collectively accepted social practices. In their absolute forms, these assumptions cannot co-exist. The debate about these positions has for the time being hit a dead end within the narrowly defined discipline of legal philosophy. One can simply decide on which side of the fence to work by embracing one of the two incompatible assumptions. One way to move forward would be to examine the underlying normative assumptions of the social institutions that produce and use the law. This is the task of political philosophy, the discipline that deals with the origin and being of political systems. Social contract theory is a branch of political philosophy that deals with the origins and being of political organisations.

II.  Ethics beyond Legal Theory The idea of the social contract is, in its abstract form, a simple metaphor. The metaphor of the contract tries to explain the being and function of political organisations, using insights from how actual contracts and agreements are understood. Over time understanding of contracts and bargaining has developed and with it the metaphor of the social contract. The perspective that supports my argument about the place of ethics in legal theory is a modern one that draws on rational choice theory and game theory. This genus of social contract theory can be traced back to the times of the Florentine Renaissance of the 1500s and the English Civil War of the mid-1600s. This variant is less well known than John Rawls’ seminal restatement of social contract theory in the Kantian tradition from the 1970s.

Ethics beyond Legal Theory  81

A.  The Social Contract Narrative John Rawls’ celebrated modern version of social contract theory draws on ideas developed during the enlightenment period. To understand his motives it is thus helpful to summarise the main ideas from the texts on which his restatement relied, and also ideas from texts he did not cite, but which are also important for understanding the idea of the social contract. Niccolò Machiavelli’s Il Principe (1513) was a radical text in many ways. Although not its main topic, Machiavelli discussed briefly the logic of power in a principality that was founded through the favour of the citizens. Machiavelli referred to it as the ‘civil principality’, which was not founded on, or held together through coercion or violence: ‘A man who becomes ruler through popular favour … must keep the people well disposed towards him. This will be easy, since they want only not to be oppressed.’18 For many centuries, Machiavelli was not held in high regard, due to his cynical outlook on the function of politics and power. By modern standards, however, his vision is remarkably current and builds on a presumption of a rational human mind responsive to rational self-preserving and incentives. Machiavelli recognised that in the absence of coercion, the political establishment of the civil principality must govern through the consent of the citizens. Thomas Hobbes argued in Leviathan (1651) that without an agreement between people, the inherent nature of humans would throw them into war of all against all. He observed that humans are homogeneous regarding physical and mental abilities, and that this should predispose them towards similar needs and desires, which ought to create high demand for specific materials. High demand in turn leads to low supply, and thus the war in the state of nature over the limited supplies becomes inevitable with its consequence of a ‘poor, nasty, brutish and short’ solitary life of men.19 Hobbes believed that these dreadful prospects and the potential of achieving greater prosperity should make humans preconditioned towards reaching an agreement with their fellow beings; once reason has compelled men to seek peace, they ‘should be willing (when others are too) to lay down [their] right to everything, and should be contented with as much liberty against other men as [they] would allow other men against [themselves]’.20 The primary motive for entering voluntarily into an agreement, according to Hobbes, is rationality from an individual point of view. The individual recognises the personal benefit of cooperation and he or she is prepared to make the effort of realising this potential on the condition that others do the same. 18 Niccolò Machiavelli, The Prince (Quentin Skinner and Russel Price (eds), Cambridge, Cambridge University Press, 1988) 36. 19 Thomas Hobbes, Leviathan, 2nd edn (Richard Tuck (ed), Cambridge, Cambridge University Press, 1996) pt I, ch 13. 20 ibid pt I, ch 14.

82  Ethics for Procedural Architects An important element in Hobbes’s elaboration of the social contract is an appreciation of the meaning of what is termed ‘cheap talk’ in modern game theory. Hobbes argues that words and agreements have no meaning in themselves. It is all about creating the right incentives, upon which the rational agent acts. To Hobbes, it was thus essential for the success of the social contract that it could be enforced by ‘some coercive power’.21 This insight about the role of incentives as the motor of the rational human being is still relevant. This sets him apart from many of his contemporaries who routinely used religiously based axioms in their theories, and from later theorists in the deontological tradition. John Locke in his Second Treatise of Government (1689) contributed to social contract theory differently. Locke’s starting point was the natural human condition. In that state – the state of nature – humans are free to act as they please so long as they abide to ‘the law of nature’.22 The law of nature burdens humans with the obligation to refrain from harming ‘anyone else in his life, health, liberty, or possessions’. Locke’s foundation for this law was based on religious doctrines. In short, he argued, that since humans were all made by God and are thus all the servants of Jesus, his deputy, they are by nature not allowed to harm themselves nor other equally ranked properties of God.23 The leap from the state of nature to a human society must be based on consent, according to Locke, and this consent is triggered by the person’s ‘intention of better preserving himself, his liberty and property’ to better ensure the enforcement of the law of nature.24 In Locke’s view, the power of society over men is limited by the scope of the powers they had themselves in the state of nature.25 This forms the initial moral obligation of society against its citizens. However, contrary to Hobbes, this obligation is not based on the willingness of a rational person to concede rights, but instead on the natural law that cannot be conceded by anyone due to its divine origins. Jean-Jacques Rousseau argued, in Du contrat social ou Principes du droit politique (1762), that self-preservation is the primary natural motive of humans. This motive, in turn, controls human logic in the state of nature.26 Hardship in the state of nature may however compel humans to take refuge in the company of others in a civil society, which can only happen through an agreement.27 In the state of nature the individual had natural liberty to act on impulses and instincts, while in the civil state he or sh has civil liberties that are limited by the moral obligation to conform to the general will.28 The power of the sovereign, which personifies the 21 ibid pt I, ch 14–15. 22 John Locke, Two Treaties of Government (Peter Laslett (ed), Cambridge, Cambridge University Press, 1988) para 4. 23 ibid para 6. 24 ibid para 95. 25 ibid paras 131, 134–35. 26 Jean-Jacques Rousseau, The Social Contract (Christopher Betts (tr), Oxford, Oxford University Press, 1994) bk I, ch 2. 27 ibid bk I, chs 4 and 6. 28 ibid bk I, ch 8.

Ethics beyond Legal Theory  83 general will, is in turn restricted by a special equality requirement, meaning that he or she cannot treat one of its parts more preferably than the others.29 Equality thus becomes the initial moral obligation of the sovereign towards the citizens. In this, Rousseau recognises that the act of cooperating on a mutual task, involves a priori commitment to behave in a specific way and, by doing so, one can expect others to be required to do so as well. According to Rousseau, the commitment of giving equality and the expectation of receiving equality is a prerequisite for forming a social contract and its primary output once established. Immanuel Kant’s categorical imperative is important with regards to the evolution of social contract theory. The categorical imperative was first described in Grundlegung zur Metaphysik der Sitten published in 1785. His argument starts as follows: Everything in nature works according to laws. Only a rational being has the capacity to act according to the representation of laws, i.e. according to principles, or a will. Since reason is required for deriving actions from laws, the will is nothing other than practical reason. If reason determines the will without fail, then the actions of such being that are recognized as objectively necessary are also subjectively necessary; i.e. the will is a capacity to choose only that which reason, independently of inclination, recognizes as practically necessary, i.e. as good. […] The representation of an objective principle in so far as it is necessitating for a will is called a command (of reason), and the formula of the command is called imperative. All imperatives are expressed by an ought, and by this indicate the relation of an objective law of reason to a will that according to its subjective constitution is not necessarily determined by it (a necessitation). They say that to do or to omit something would be good, but the say it to a will that does not always do something just because it is represented to it that it would be good to do it. Practically good, however, is what determines the will by means of representations of reason, hence not from subjective causes, but objectively, i.e. from grounds that are valid for every rational being, as such. It is distinguished from the agreeable, as that which influences the will only by means of sensation from merely subjective causes, which hold only for the senses of this or that one, and not as a principle of reason, which holds for everyone.30

Kant then continued to pronounce all imperatives as either categorical or hypothetical, meaning they either objectively or subjectively command the rational will of a person. Kant’s central thesis was that moral duties are categorical imperatives. They imply commands that apply to the will of any rational person, irrespective of the ends this person may decide to pursue. Kant contrasts the categorical imperative with the hypothetical imperative. The hypothetical imperative implies a command that is conditioned upon a will towards a certain end. The hypothetical

29 ibid bk II, ch 4. 30 Immanuel Kant, Groundwork for the Metaphysics of Morals, 2nd edn (Mary Gregor and Jens Timmermann (eds), Cambridge, Cambridge University Press, 2012) 26–27.

84  Ethics for Procedural Architects imperative thus commands that one must do something if a certain end is being pursued, while the categorical imperative does not rely on any specific end. It simply commands categorically in any case.31 Acting morally, according to Kant, is thus acting in a way that could be viewed universally rational. The rational will always commands a practically good action, and if the action can be considered universally good irrespective of specific ends, it is a morally good action. Or, as Kant put it: ‘act as if the maxim of your action were to become by your will a universal law of nature’.32 After Kant and following the rise of economics and utilitarianism in the eighteenth and nineteenth centuries, the social contract idea as a basis for theorising about the organisation of society became a marginal topic. By the mid-twentieth century, however, the idea of the social contract made a comeback through John Rawls’ 1971 monograph, A Theory of Justice. Rawls criticised utilitarian theories of justice and argued for an alternative by restating the social contract tradition as described by Locke, Rousseau and Kant.33 Rawls intentionally left out Hobbes’s Leviathan on his list of sources of inspiration, stating that ‘For all its greatness [it] leaves special problems’.34 The key ideas in Rawls’ social contract are the original position and the veil of ignorance. Rawls explained that his concept of ‘the original position of equality corresponds to the state of nature in the traditional theory of the social contract’. By rephrasing the description of the natural state, prior to the existence of the social contract, Rawls sought to emphasise the hypothetic nature of the contract. It was merely meant as a thought experiment to visualise what standard of justice persons would agree on, in a vacuum of equal bargaining power.35 To reach the level of equal bargaining power, Rawls realised that the parties to the agreement must be oblivious about their private position, when negotiating about the standard of justice. Thus, he devised the veil of ignorance, behind which the negotiation takes place. The core of Rawls’ argument of what happens behind the veil can be found in the following passage: Since all are similarly situated and no one is able to design principles to favour his particular condition, the principles of justice are the result of a fair agreement or bargain. For given the circumstances of the original position, the symmetry of every-one’s relations to each other, this initial situation is fair between individuals as moral persons, that is, as rational beings with their own ends and capable, I shall assume, of a sense of justice. The original position is, one might say, the appropriate initial status quo, and thus the fundamental agreements reached in it are fair.36 31 ibid 28–34. 32 ibid 34. 33 Rawls mentions David Hume, Adam Smith, Jeremy Bentham and John Stuart Mill as the main proponents of the utilitarian ideas he seeks to oppose. See John Rawls, A Theory of Justice, 2nd edn (Cambridge MA, Belknap Press, 1999) xvii–xviii. 34 ibid 10, fn 4. 35 ibid 11. 36 ibid 11.

Ethics beyond Legal Theory  85 Rawls’ social contract becomes fair if the negotiators are rational, morally bound, and in a symmetrical bargaining position. The moral requirement sets Rawls apart from Hobbes, who insisted on the cynical nature of humans. Rawls argued that utilitarian justice principles, which strive for maximising total utility irrespective of its distribution, are not compatible with his version of the social contract. He claimed that no rational person seeking their own means would agree to cooperate on the terms that only others should benefit from the cooperation. Rawls suspects that an agreement, between rational moral persons in the original position, would be reached on two principles of justice that he describes in the following way:37 First: each person is to have an equal right to the most extensive scheme of equal basic liberties compatible with a similar scheme of liberties for others. … Second: social and economic inequalities are to be arranged so that they are both (a) reasonably expected to be to everyone’s advantage, and (b) attached to positions and offices open to all.38

The content of the hypothetical social contract would accordingly become these two principles.39 The main challenge, however, is to explain why a rational deliberation would result in these two principles, rather than something entirely different. Rawls applied deductive reasoning and specific contingencies to support that point. He assumed that his rational persons would pursue their ends within the contingencies of some specific circumstances; at some point, a stable equilibrium is formed between the pursued ends and the specific circumstances. On Rawls’ view, this does not guarantee that the equilibrium is just. Therefore, his rational person must also be moral, ie with a sense of justice. This adds a layer to the specific circumstances and thus alters the feasible equilibrium point, by leaving out unjust situations. His original position thus eliminates all specific circumstances except for the moral condition. By doing this, he argued that all agreements reached in the original position are fair.40 His whole argument thus depended on the moral inclination of people. The moral character of people in Rawls’ model was expressed in two ways. Firstly, the rational behaviour of people was restricted by a condition of disinterest in the relative position of others. Rawls’ rational person thus only strives towards his or her own ends, but does not act on envious emotions by sabotaging the prospects of others. In this Rawls’ theory differs from Hobbes’s, which argued for the natural tendency of humans to compete in situations of scarcity. Secondly, the condition of a sense of justice compels individuals to honour agreements that are made: ‘in reaching an agreement, then, they know that their undertaking is not in vain: their capacity for a sense of justice insures that the principles chosen will be respected’.41 This again puts Rawls at odds with Hobbes, who remarked; ‘covenants

37 ibid

13. 53. 39 ibid 102. 40 ibid 103–04. 41 ibid 125. 38 ibid

86  Ethics for Procedural Architects without the sword are merely words, with no strength to secure a man at all’.42 In this sense, Rawls relied on a moral a priori in the same sense as Locke, but with a sophisticated Kantian spin. Through the veil of ignorance thought experiment, Rawls essentially to restated Kant’s universal law. Rousseau and Hobbes argue, however, that cooperative behaviour is dependent on a certain moral attitude that is triggered by the right incentives.

B.  Hobbesian Ethics and Game Theory David Gauthier was among the first authors to model the social contract in terms of game theory.43 He argued for a Hobbesian understanding of morality, where morality is the result of a rational choice. Thus, cooperation through a social contract did not require any a priori moral virtues. The object of moral appropriateness was simply the rational choice in the bargaining dilemma of cooperating with others. Gauthier modelled the bargaining problem of the social contract as a version of the prisoner’s dilemma, in which individual players are pinned against each other by the temptation of gaining by acting selfishly. Brian Skyrms and Kenneth Binmore have argued for a different modelling.44 Binmore’s version is sophisticated and warrants further elaboration. By portraying the problem of the social contract based on various games in the language of game theory it quickly emerges that cooperation is often an efficient equilibrium solution. Usually, however, there are also efficient equilibriums that are not cooperative. This leaves the question of why a cooperative equilibrium would be selected rather than an individualistic one. Deontological ethics seek to solve the issue with reference to a priori moral reasons, but in terms of rational choice and game theory, the solution needs to build on logic and rational incentives. Binmore builds a theory based on the stag hunt game that provides a solution to this issue.45 He claims that a ‘social contract is the set of common understandings that allow the citizens of a society to coordinate their efforts’. In his mind, positive law or moral sentiments do not keep societies together. The social contract 42 Hobbes, Leviathan (1996) pt II, ch 17. 43 David Gauthier, Morals by Agreement (Oxford, Oxford University Press, 1986). 44 Brian Skyrms, Evolution of the Social Contract, 2nd edn (Cambridge, Cambridge University Press, 2014); Brian Skyrms, The Stag Hunt and the Evolution of the Social Structure (Cambridge, Cambridge University Press, 2004); Ken Binmore, Game Theory and the Social Contract Vol I: Playing Fair (Cambridge MA, The MIT Press, 1994); Ken Binmore, Game Theory and the Social Contract Vol II: Just Playing (Cambridge MA, The MIT Press, 1998); Ken Binmore, Natural Justice (Oxford, Oxford University Press, 2005). 45 The stag hunt game concerns the dilemma of whether to cooperate with others on the hunt in order to catch the prized stag, or whether to rely solely on yourself and hunt for the less valued hare. The catch is that if you choose to cooperate and the others won’t, you end up with nothing, since cooperation is required to accomplish the great enterprise of hunting the stag. There are two Nash equilibriums in the stag hunt game. Either everyone hunts the stag, or everyone hunts the hare. The stag hunt game is derived from Rousseau’s story of the stag hunt. See Jean-Jacques Rousseau, A Discourse on the Basis and Origin of inequality Among Men (Boston, Bedford/St Martin’s, 2010) pt II.

Ethics beyond Legal Theory  87 is a rational solution to a coordination problem and ‘does not need any glue’ to exist. Binmore argues that three conditions ranked in an order of priority must be fulfilled to achieve a successful social contract. The first condition is stability. If not stable, a contract will not exist for long and is thus of negligible importance. To become stable, the common understanding and the coordinated behaviour must form a Nash equilibrium, in which each citizen’s strategy is the best response to the strategies of other citizens. Efficiency is the second condition. The stable equilibrium must be efficient in comparison with competing social contracts to survive. Efficiency helps select out of the available Nash equilibriums a viable set of efficient equilibriums. Several equilibriums can however result in a comparable efficiency, which leads us to the third condition. A test of fairness weeds out of the efficient equilibriums a single right equilibrium upon which the social group should coordinate. The social contract must thus be stable, efficient and fair.46 The question of why a cooperative equilibrium would be chosen, rather than an individualistic equilibrium, in a game is answered at the second tier of Binmore’s formulation. Binmore explains that instead of resorting to normatively anchored concepts like trust, duty and authority, game theory can explain the phenomenon based on the folk theorem. The folk theorem predicts that in a repeated game with an infinite timeframe, a cooperative efficient equilibrium can be maintained despite the potential of a one-off gain by cheating on the cooperation. If, however, the game has a definite end in sight, it will unravel in a race of opportunisms.47 This still only shows that it can be explained as rational to cooperate, despite an incentive not to cooperate in a repeated game, but this does not show that a cooperative equilibrium will be chosen. The nudge that eventually pushes humans towards cooperative social contract equilibriums can be explained, according to Binmore, based on the theory of natural selection.48 Inspired by Dawkins’ theory of the selfish gene, Binmore explains that individuals within a group seek to maximise their personal fitness by acting in accordance with the prevailing efficient equilibrium within the group.49 This group equilibrium is not necessarily settled on the most efficient equilibrium possible; other potentially more efficient group equilibriums might exist. If this group encounters another group that was settled on a more efficient equilibrium, it must quickly either amend its ways to withstand competition or face extinction from the game of life. This over time forces groups to settle on Pareto optimal equilibriums, where nothing goes to waste and nothing can be improved without undermining the group’s strategy against other competing groups. This is how the stag hunt 46 Binmore, Natural Justice (2005, n 44) 3–14. 47 ibid 10. It should be noted that the folk theorem is usually used to show that cooperation is possible even where there is a strong incentive to cheat like in the prisoner’s dilemma game. It will thus also apply to show the possibility to cooperate when the incentive not to cooperate is lesser than in the prisoner’s dilemma game, which is the case with the stag hunt game. 48 ibid 7–14. 49 Reference is made to the evolutionary theory argued for in Richard Dawkins, The Selfish Gene (Oxford, Oxford University Press, 1976).

88  Ethics for Procedural Architects will be chosen as a social contract, over the hare hunt in the stag hunt game. The group that settles on the less efficient hare hunt will not succeed in the game of life when competing with a group that hunts the stag. In principle, the same argument applies if the competing groups are playing the prisoners’ dilemma, or any other game that includes a Nash equilibrium that is not Pareto optimal. However, this is not enough to solve the equilibrium selection problem. Even if natural selection pushes towards Pareto optimality, many equilibrium solutions are still available. Pareto optimality can be achieved in various ways through different distribution of the pay-offs from the efficiency gain. A further tool is thus needed to coordinate on a single efficient equilibrium to avoid destabilisation due to incoherency. Binmore argues that this selection tool has the same deep roots in the human condition as the efficiency selection tool. When faced with a simple problem of distribution of an efficiency gain, people somehow by instinct feel that a certain distribution is right, while another is wrong. A fair division of a cake is an example. This intuitive feeling is a sense of fairness, which Binmore believes, is the ultimate arbitrator of equilibrium selection.50 According to Binmore, utilitarian and egalitarian theories of fairness are the main contenders for providing a tool to solve the selection problem.51 Utilitarian solution is at a point where the weighted sum of pay-offs is largest, while the egalitarian solution is ‘the efficient outcome at which each player’s weighted gain is equal’.52 To simplify we can assume that two utilitarians would agree on their highest combined pay-off, while two egalitarians would agree on the equilibrium that leaves them both equally off, irrespective of the total combined pay-off. Binmore uses Rawls’ theory of justice as fairness as a sophisticated example of egalitarianism and John Harsanyi’s utilitarian theory of the social contract as a sophisticated example of utilitarianism. Binmore thinks Rawls is right, but for the wrong reasons. He argues that the thought experiments of the original position and the veil of ignorance somehow intuitively make perfect sense, but that the reference to Kantian ethics fails to explain why the idea seems so right. Binmore’s thesis is that Rawls’ theory describes fairness norms that we use to solve countless everyday small-scale coordination problems – problems that we solve so effortlessly we hardly realise we are solving problems. These fairness norms are deeply embedded into our existence. Binmore even believes that they are written into our genes. He argues that evolutionary pressures could have created such a genetic prescription and influenced how this biological mechanism works together with our cultural heritage in choosing equilibriums in the game of life.53 Binmore summarises his position in the following way: [M]y theory of fairness … is the claim that all fairness norms in actual use share the deep structure of Rawls’ original position. This deep structure is biologically determined, and

50 Binmore

(2005, n 44) 14. 23. 52 ibid 29 and 31. 53 ibid 15–17. 51 ibid

Ethics beyond Legal Theory  89 hence universal in the human species, but the standards of interpersonal comparison that the original position needs as inputs are culturally determined, and hence vary with time and place.54

To explain why a utilitarian concept of fairness does not work, Binmore asks a rhetorical question: ‘why should I maximize the sum of utilities rather than my own?’55 This highlights a contingency on which utilitarian theory depends. In a world of rational agents, an equilibrium based on utilitarian distribution of payoffs, cannot be formed or maintained without external enforcement. Rational agents will always prefer to maximise their own utilities rather than the total utilities of the group, unless policed to do otherwise. Those who proportionally gain less from a utilitarian solution will have an incentive to team up with others in the same position and deviate from a utilitarian social contract. In theory, the utilitarian solution is the best solution in terms of maximising utility, but in practice external enforcement may be impractical or impossible, which reduces its feasibility and correspondingly elevates the feasibility of alternative equilibrium solutions that are not contingent on external enforcement.56 Binmore builds a simple formula to explain how an egalitarian fairness solution is not subject to this dependency on external enforcement. He assumes that in a coordination problem, two players would agree on a Nash bargaining solution, which is the maximum product of their gains from the point of disagreement. He then adjusts the social indexes of both players to match either a utilitarian or an egalitarian fairness solution, both of which are fixed at the point of the Nash bargaining solution. If then the available equilibriums of efficiencies are suddenly expanded and thus an incentive created to bargain a new social contract, the two standards of fairness lead to different equilibriums. The utilitarian solution seeks to maximise the total utilities irrespective of distributional concerns, which can only be achieved by external enforcement because some players will have an incentive to oppose the new distribution of utilities, which is potentially asymmetric with their current social indexes. The egalitarian solution is, however, always symmetric in a distributional sense, so that none of the players has an incentive to oppose the new distribution. Everyone receives an increase in pay-offs proportional to their current social indexes. The egalitarian solution is thus not dependent on external enforcement; no one has an incentive to deviate.57 The consequence of this formulation is that a social contract cooperation, based on an egalitarian distribution of pay-offs, can be formed and sustained spontaneously without external enforcement. However, a social contract cooperation based on utilitarian distribution requires external enforcement to stay in place. This provides the last piece in the puzzle of how a single equilibrium, out of



54 ibid

18. 149. 56 ibid 163. 57 ibid 158–59, 173–75. 55 ibid

90  Ethics for Procedural Architects the many available Pareto optimal equilibriums, can and will be selected. Without external enforcement, a rational person would seek an equilibrium solution that achieves egalitarian distribution of pay-offs. Other Pareto optimal equilibriums would need an external intervention to nudge in their favour, to override the spontaneous bargaining processes, which by default seeks an egalitarian distribution. By this Binmore explains the being of a social contract with reference to human rationality without having to rely on Hobbes’s external enforcement assumption; this also explains how Rawls’ Kantian-inspired social contract theory is compatible with rational choice.

C.  Contractarian Ethics and Democratic Institutions The social contract thought experiment, in its various forms, usually tries to explain the logic of the move from the pre-cooperative stage of humans to the cooperative stage. Hobbes saw cooperation as a rational efficiency-enhancing step, away from the gruesome prospects in the state of nature. Locke saw cooperation as a natural consequence of a morally bound human nature, using divinity as the ultimate moral authority. Rousseau acknowledged the principle of equality as being interlinked with human cooperation, both as its prerequisite and its consequence, but that egalitarian attitude was subject to a deliberate choice, and thus not a preordained human behaviour. Kant explained morality as a categorical imperative, which secularised morality as an ultimate authority for human action. Rawls built on Locke’s morally bound human nature, but with morality founded on Kant’s categorical imperative instead of theology. His morally bound person would subsequently form a social contract based on Rousseau’s principle of egalitarian fairness. From this, two distinct ideas emerge about the origins of morality in cooperation. On one hand, human morality can be seen as an a priori to all human action, which implies that any product of a human action is morally bound. This is the approach of Locke and Rawls. On the other hand, morality can be seen as subject to free will and created by deliberate reciprocal human interaction, which accordingly is not an a priori to all human action. This is the selfish reciprocity of Hobbes and the egalitarian nature of Rousseau’s social cooperation. On this later view, morality only exists when humans so choose, eg by interacting in a cooperative way. Acting morally is thus the act of the rational mind, not a predefined part of the human condition. These two approaches to morality in cooperation face different challenges. The former has a similar obstacle as the natural law theorist: if humans are inherently moral, then how are immoral actions possible? The later faces a problem of stability; if humans are only subject to their free will, then how is moral cooperation possible without quickly faltering to opportunistic behaviour? Binmore gives us further clues about how to approach this problem. He argues, through game theory modelling, that it is rational to act morally. Additionally,

Ethics beyond Legal Theory  91 he uses anthropological and biological theories to depict a plausible evolutionary narrative about how this could have occurred. By this he tries to prove that there are scientific explanations for the phenomenon that philosophers have through the centuries referred to as morality. He shows mathematically that egalitarian distribution of pay-offs is a superior equilibrium solution in the game of life. The crucial point is, however, that other equilibriums can be maintained, but only artificially through external enforcement. Hobbes was on to something when he realised that humans need a plausible reason to cooperate. Rousseau realised that an egalitarian attitude was required within the confines of a workable social contract. Locke and Kant knew that a moral constant would enable successful cooperation. Rawls melded these ideas together into a sophisticated theory but was reluctant to acknowledge Hobbes’s cynical outlook on the human condition. What can be drawn from this is that there certainly is a moral constant, which is imprinted into the rational mind of humans. There is a single right way to solve coordination problems in human cooperation spontaneously, which involves egalitarian distribution of the gain from the efficient solution. Egalitarian fairness is thus a good candidate for the basis of morality and moral behaviour in the company of others. Egalitarian fairness can be reduced to a singular entitlement, and a corresponding singular duty; each has the moral entitlement to be treated with relative equality, and each has the duty to treat others with relative equality. Interaction based on these principles results in a stable Pareto efficient cooperative equilibrium, where no one has an incentive to withdraw his cooperativeness. The primary issue that emerges from the social contract literature is the question of what compels the free human mind to act morally. Hobbes’s insight was that the motor was a matter of simple logic: either cooperate, or risk perishing in the state of nature. He sensed that it was ultimately a choice whether to cooperate and thus whether to act morally, and that cooperative attitude depended on a proper incentive for the rational mind. Rousseau advanced Hobbes’s intuition by predicting that an egalitarian attitude was required to form and maintain a social contract. Rawls’ reliance on Kantian ethics struggles to explain why the exercise of free will would result in a moral choice. Even if the morally right attitude is identified, it does not necessarily follow that the agent will always choose to act morally. Assuming free will, Kant’s insight was that the human has a choice between acting on the hypothetical and the categorical imperative; ie, acting right from a universal perspective, or acting right from a selfish perspective. The alternative to the moral attitude in interpersonal relations, which Binmore showed through the cooperative solution of the stag hunt game and through the folk theorem, can be explained using the example of another game. The hawk– dove game is about competition for scarce resources in which the worst outcome results in a mutually destructive conflict between two hawks. If one yields and acts as the dove, the hawk receives a larger share than the dove, but the pay-offs are nonetheless Pareto optimal. If both yield, the result is not Pareto optimal since higher total pay-offs can be reached through another solution. In the hawk–dove

92  Ethics for Procedural Architects game, there are three Nash equilibriums: two pure strategies, where each acts the opposite role to the other; and one mixed strategy where the players randomly choose which role to assume. The random strategy only makes sense if the players cannot decide who assumes which role, but this strategy will not lead to optimal efficiency because sometimes both will act as hawks and sometimes both will act as doves. The pure strategies, on the other hand, rely on the capability of any player to maintain a hawk position and convince the other of their destiny as a dove. The optimal equilibrium solution to this game thus requires an attitude that is not morally compatible; one player would need to subdue the other. Political organs that are structured in the same manner as the optimal equilibrium solution of this game can be easily imagined. However, as the Pareto optimal solution suggests, such organs rely on coercion to remain stable. The importance of the attitude with which interpersonal relations are approached also becomes apparent if the situation is depicted as the prisoner’s dilemma game. The Pareto optimal result of the prisoner’s dilemma is achieved by a cooperative attitude on staying silent and thus getting away with the crime. The prisoner’s dilemma involves a strong individual incentive to confess and to frame the others; but the others recognise this incentive as well and thus have an incentive to do the same. The sensible strategy is thus for all actors to confess and blame the others in order not to be the only non-confessing culprit. This forms the Nash equilibrium for the prisoner’s dilemma, and it is not Pareto optimal. The problem for a group that settles on the confess–confess equilibrium is not internal instability, but instead competitive disadvantage with external groups that have found a way to cooperate on the Pareto optimal solution of the game by approaching the task with an egalitarian cooperative attitude. Morality thus enters the social contract through the attitude with which interpersonal interaction is approached. The display of different game scenarios with different behavioural incentives confirms this. If the interaction is approached competitively, like in the hawk–dove game, optimal stability requires coercion, while the stag hunt and the prisoner’s dilemma acquire stable optimality through cooperative egalitarian attitude. Morality in cooperation is thus simply a question about people’s strategic attitude. If a group decides to cooperate without an external enforcement mechanism, it must do so based on an egalitarian principle of distributive fairness. Otherwise, a stable Pareto optimal equilibrium cannot be maintained spontaneously. The output of egalitarian cooperation will necessarily bear the mark of the foundational principle. Any attempts to distribute pay-offs differently will either lead to destabilisation of the cooperation or call for a coercive enforcement to maintain stability. The social contract thus creates a moral commitment upon formation, which is the requirement of a cooperative attitude, which implies a compliance with an egalitarian principle of distribution. The view on the social contract and morality that has been articulated above shows how the idea of morality as an a priori concept can be harmonised with the concepts of free will and rationality. Cooperation approached on an egalitarian fairness principle is a priori the morally optimal approach. This optimality

Ethics beyond Legal Theory  93 is achieved through the long-term efficiency prospects in the game of life. The rational long-term strategy on an individual level is thus to use this principle of morality for guidance when exercising free will (ie be cooperative and flourish). By exercising free will opportunistically, long-term efficiency is put at risk for a shortterm gain. Such a behaviour is irrational in the game of life and thus immoral (ie be egoistic and perish). In Kantian terms, this is the difference between acting on the categorical imperative and the hypothetical imperative. Acting in accordance with Kant‘s universal law is thus a superior strategy in the game of life compared with acting on more opportunistic short-term urges. Systems of public governance can be seen as devices to coordinate the selection of social contract equilibriums. The social group’s balance of power is transcribed through the selection of governance structure and is thus decisive in determining the distribution of pay-offs. In groups where the balance of power is decentralised, the ability to externally enforce equilibrium solutions is limited and thus they tend to organise on a solution that is fair in the egalitarian distributional sense. In groups where the power is consolidated with elites, or a tyrant, the selection of an equilibrium can be sustained irrespective of fairness concerns, if it remains rational not to oppose those in power due to the risk of punishment. Democracy, seen in this context, is a system of governance that seeks to decentralise power by giving every member of the group an equal worth when it comes to coordinating on social contract equilibriums. The fundamental difference between democracy and other systems of governance rests in the ultimate normative reason for action within the system. Democracy must rely on an equitable system of pay-off distribution, because of its intrinsic decentralised power structure. If the system ignores the egalitarian a priori, it risks destabilisation or an authoritarian mutation. Authoritarian systems of government do not need such an a priori, because through consolidation of power and enforcement capabilities, they are able to unilaterally dictate actions. Democracy is thus the practical application of the abstract notion of the egalitarian social contract. The institution of democracy is about avoiding consolidation of power. In that way, democracy is the operationalisation and the deliberate maintenance of a spontaneous equilibrium selection device. It is an institution established to maintain and foster the moral egalitarian attitude in interpersonal interaction. The idea of the social contract explains how democracy is possible without ultimate reliance on coercive authority. The decision to act cooperatively implies compliance with a specific method for distributing the potential pay-offs that the cooperation ripens. The specific distribution method necessarily prescribes relatively equal shares to all equity-holders in the relevant social practice. Complex social organs and institutions that are designed on the premise of democracy are thus ultimately striving for equal distribution of the efficiency gain that the collective organ or the institution enables. Translated into the language of ethics, this is the ultimate ethical obligation of the organ, its institutions, its members and its laws.

94  Ethics for Procedural Architects

III.  Ethics as the Dark Matter of the Law Social contract theory shows that simple social and coordination practices can be created based on different equilibrium solutions and that the specific solution that is practised depends on the attitude with which the participants approach the interaction. The problem for legal positivism built on Hart’s premises, is that social practices are necessarily contingent facts. The contingency on which they rely is the attitude with which the practice was created and is maintained. A rule of recognition that is based on contingent social facts, such as social practices that are the result of a certain type of social interaction, are dependent on the same contingency as its foundational premises. Hart’s social practices are simply coordination equilibriums aimed at efficiency enhancements. Mysterious social practices that appear randomly or simply exist for no a priori reason do not exist in a world that assumes the existence of free will. Social practices and conventions are the output of a competitive or a cooperative interaction of rational human beings. The function of a social practice depends on the attitude with which it is maintained. Social practices can be maintained based on competitive capacity to force obedience through the threat of violence. Social practices can also be maintained through a cooperative attitude. Seen in this way, the rule of recognition is simply a stable equilibrium solution to a certain coordination problem within a certain system of governance. The system of governance dictates how equilibriums within that system are selected. Authoritarian systems have a greater choice over the selection due to their ability and willingness to apply force to maintain stability. Democratic systems have decentralised power structures and must thus select equilibriums that can be maintained with minimal enforcement effort. The only efficient equilibriums that can be maintained without enforcement are those who respect the principle of egalitarian distributional fairness. If the rule of recognition is the product of a social practice within a democratic system of governance, the moral a priori of egalitarian fairness ultimately binds it. If, however, the rule of recognition is the product of a system of governance that does not rely on the egalitarian a priori as the ultimate source of authority, the necessary embeddedness of this particular a priori in the governance system is no longer required. Legal positivism is defined by the refusal to recognise a necessary connection between law and moral norms. For what it is worth, this thesis holds when speaking about law in an unqualified sense. It even, and perhaps especially, holds in marginal cases like that of the laws of Nazi Germany, where the organ that produced the laws was not conditioned by normal standards of ethics. However, when talking about laws strictly in the context of organs that are organised on the principle of democracy, the separation thesis of legal positivism collapses. Laws that are created by an organ, that maintains stability due to a specific moral attitude of its members, cannot permit the creation of an output that contradicts the principal reason for the organ’s stable being. A rule of recognition and derivative

Ethics as the Dark Matter of the Law  95 laws that contradict the principle of egalitarian fairness thus cannot survive in a democratic system of governance because they do not form a stable equilibrium solution to the coordination problem they are meant to solve. Anomalies can appear, but they will be gradually corrected or ironed out to comply with the cooperative attitude. There is thus a necessary connection between the law and the moral principle of egalitarian fairness in every political organ that builds on the principle of democracy. The task of this chapter was to explore whether fairness in the law was subject to a deliberate decision, or whether it was determined by an a priori that independently controlled whether laws are fair or not. The traditional approaches to legal philosophy, under the headings of legal positivism and natural law theory, cannot answer this question because they both assume the answer in their foundational premises. Legal positivism assumes that fairness in the law is decided by the laws, while natural law theory assumes binding pre-existing morality that dictates what is fair in the contexts of the law. Social contract theory shows that the nature of the concept of law is sensitive towards the foundational attitude with which the political organ which the law serves was founded upon. If the law serves an organ that was founded upon a cooperative social contract (ie democracy), the ultimate a priori of the law is the egalitarian fairness principle. If the law serves a political organ that permits egoistic social equilibriums, the ultimate a priori can be a threat of violence, which then permissibly can provide the law with its ultimate authority. The task of this chapter was to determine how a procedural architect should approach the issue of ethics in their legislative designs. In particular whether their legislative choices would automatically represent a fair approach, or whether their designs were subject to a pre-existing criterion of fairness. The foundational assumptions of legal positivism and legal naturalism do not settle this issue, but instead assume contradicting answers. But with reference to the social contract branch of political philosophy, it has now been argued that a procedural architect that serves an institution based on the ideal of democracy must abide to the rationale of cooperative human interaction. This implies a pre-existing condition of egalitarian distributional fairness, against which their designs will be measured. Only if the procedural design passes this test can it be considered to represent procedural fairness. Egalitarian fairness is thus the dark matter of the law within the democratic state and its primary standard of ethics. In the context of solving the balancing dilemma at the core of any procedural design identified in the previous chapter, the procedural architect can approach that problem with reference to the standard of egalitarian distributional fairness. To do this they would need a practical methodology, which enables legislators to solve practical balancing problems in legislative design, such as in the design of EU competition procedure. An argument for such a methodology will be made in the next chapter.

5 The Model of Fair Rules Assuming a democratic contractarian view on society, the ultimate task of the architect of rules is to make rules that are supported by its stakeholders. Different groups of stakeholders hold potentially diverging preferences, which implies a balancing dilemma for the rule architect. The optimal solution to that dilemma is the object of the fair rule. A failure to identify and implement fair rules risks discontent among the citizens at the losing side of the rule, which can escalate into resistance and rebellion against the social equilibrium established by the unfair rule. From a macro perspective, the rise of political populism and growing economic disparities can be interpreted as symptoms of a failure to identify the egalitarian equilibrium solutions to regulatory dilemmas.1 Seen in this way, it could be argued that Marx’s spectre of inequality is still alive and well. If people generally find the idea of inequality appalling, a society organised around the idea of democracy should, in theory, gradually settle on a societal equilibrium where each citizen both gives and receives equality from his fellow citizens. Yet twenty-first-century democracy, in a methodological sense, seems unable to resist an accumulation of wealth and power within increasingly detached elites. It has been argued that the mechanics of the electoral process facilitates this development, but undeniably it also provides an outlet for a non-violent resistance; vote for the absurd, anti-intellectual and populist, to undermine the hegemony of political elites.2 Laws and rules are the blueprint of society. Through them society’s institutions are built and its economic accomplishments allocated. Assuming that fair laws in 1 Piketty has reported a trend of growing inequalities in recent decades, which in his view are the result of an empirical fact that the rate of return on capital is consistently higher than the rate of growth of income and output (ie owners of capital gradually accumulate proportionally more of the total wealth in society). See Thomas Piketty, Capital in the twenty-first century (Arthur Goldhammer tr, Cambridge MA, Belknap Press of Harvard University Press, 2014) 571. 2 David Van Reybrouck, Against elections: the case for democracy (Liz Waters tr, London, The Bodley Head, 2016) 1–57. See also Bernard Manin, The principles of representative government (Cambridge, Cambridge University Press, 1997) 132–60. Manin interestingly shows how the election of representatives changed from being perceived as an aristocratic form of government, over to being perceived a democratic form following the universal suffrage movement in the 19th and early 20th centuries. Arguably, this change of perception was mistaken given the actual outcomes of the representative democratic system, which in many ways facilitates greater influence over state affairs to the modern capitalist versions of the aristocrats. One way to view the Trump presidency, Brexit and similar electoral outcomes in other European countries is to see it as a reaction, or a rebellion, against a gradual societal slide towards Piketty’s inequality.

The Role of the Rule Architect  97 an egalitarian sense correlate with an egalitarian society, the task of this chapter is to outline a model for assisting rule-makers in identifying fair rules. It builds on contractarian ethics in the Hobbesian tradition argued for in the previous chapter and uses methodological components and concepts from decision and rational choice theory. Plainly, the argument is that if we can agree on an ethical standard for building society, it should also be possible to use modern decisional and computational methods for identifying the optimal rules for such a society. The key move of the chapter will be to argue for a comprehensive methodology, the model of fair rules, for identifying legislative options that comply with the ethical standard of the social contract introduced in chapter four. The model is based on several assumptions about the role of the rule-maker, which will be briefly outlined first. Having identified the key assumptions, the main elements of the model and its intended function will be described. The model is general in its scope of application for solving balancing dilemmas in legislative design and is thus applicable in the context of identifying fair procedural designs.

I.  The Role of the Rule Architect All the things a person may want in life can be represented as a preference towards a particular outcome. Living together in families, communities, societies and civilisations requires a person’s ability to cope with other people’s preferences. The glue that enables people to live together in small and large organisations is a blend of self-restraint and empathy for the needs of others. The idea of the social contract is a concise starting point for identifying the key ethical duties of the rule architect within a democratic society.3 When acting in isolation from other persons, the individual can indulge his or her preferences as they see fit. When acting within the confines of a cooperative social group, the individual must adjust his or her behaviour so that their preferences do not unduly burden the others. The individual has an incentive to restrain his or her desires if the aggregated impact of cooperation on their satisfaction is greater than the aggregated impact from acting alone. Social contracts can form spontaneously so long as the private benefit from cooperating outweighs the private cost of self-restraint. To remain spontaneously stable, the social contract must also be more efficient than competing contracts, and it must distribute the benefit equally among its members (based on an egalitarian distributional principle) to avoid rebellions and division of the social group. This is the account of the social contract introduced in chapter four; spontaneous social cooperation, kept together by the prospect of personal gain, for as long as nothing better comes around.



3 The

architect of the rules is the one responsible for how rules or laws are designed.

98  The Model of Fair Rules The rational member of the social contract must make an essential compromise between his own unrestrained preferences and the competing preferences of others. By following two simple bargaining maxims this compromise can form a stable equilibrium solution: firstly, the efficiency maxim dictates that the proposed action should have efficient consequences; and secondly the fairness maxim dictates that the consequences of the proposed action must be equally preferable to each member of the group.4 This is both the ethical guidance and the ethical mission of the social contract in consequentialist terms. If a cooperative action were not anticipated as efficiency-enhancing, the members of the social group would not have any incentive to undertake it. If the rewards of the joint effort were not to be distributed equally, some would have a weaker incentive than others to participate in the action. Disunity undermines efficiency through lower total production of the members. This creates the incentive to abandon the group for a competing group organised on egalitarian principles that would, due to better incentivised members, yield higher total rewards. The stable spontaneous social contract thus relies on the bargaining maxims of efficiency and fairness.5 The legislative process, in a modern democracy, can be viewed as a caricature of the primitive bargaining process that is frequently used for solving everyday problems of cooperation. In a setting of few people, bargaining can be concluded through a short discussion based on the two bargaining maxims (ie by finding an efficient and a fair solution). Within larger contexts, the simple process of bargaining can be simulated with the aid of simplifying assumptions.6 The idea of democracy can be considered an institutionalisation of the bargaining process on a societal scale. A democratic decision on a societal scale thus should ideally be efficient and fair with regards to the society’s members. In the modern democracy, rules are to be considered as the product of a democratic decision. Following this premise, each rule should accordingly strive for efficiency and fairness in line with the maxims of primitive bargaining. The role of the rule-maker in a democracy, premised in this way, is to identify and implement rules that optimise this normative agenda. By doing so the rule-maker is simulating a society-wide bargaining process, between the competing preferences of the society’s members; notably in a simplified way to ensure practicality and expediency.7 4 Kenneth Binmore, Natural Justice (Oxford, Oxford University Press, 2005) 1–19. Binmore argues that a successful social contract needs to satisfy three conditions: stability; efficiency; and fairness. His formulation assumes that stability will result from optimising efficiency and fairness, thus it should suffice to focus on those two. 5 ibid. 6 The most notable simplifying assumption is the principle of the majoritarian rule, which simplifies the task of reaching bargaining consensus within large groups and thus reduces the cost of reaching decisions. 7 Note that the simplification is usually layered, with a more detailed simulation for important decisions. This ranges from decisions with constitutional ramifications that often require special amendment processes to mundane administrative decisions that can be unilaterally issued by plain officials. This layering averts a prohibitive cost of simulating in details decisions with low societal importance.

The Role of the Rule Architect  99 For the rule-maker that follows the maxims of the social contract, the justification for a regulatory intervention could be based on either of the two maxims. If the current status quo lacks fairness or efficiency, the rule-maker has a valid reason to intervene. If, however, the current status quo is already at a stable equilibrium, optimised with regards to fairness and efficiency, a reason for intervention is missing. This applies regardless of whether the optimal situation resulted from a previous regulatory intervention or came about spontaneously. Viewed in this way, regulatory interventions have the primary purpose of maintaining an optimal efficiency level by responding to external changes, and to fairly rearrange the internal allocation of the efficiency output to fit the current mixture of preferences.8 Assuming individuals generally act to bring about specially targeted preferable consequences, it can also be presumed that individual participation in cooperative actions is based on the same rationale. Each cooperative action is thus aimed at satisfying a set of individual preferences. The rationale for participation does not need to be the same for all, and the intensity of gain that each action brings can vary between individuals. To achieve his or her task, the rule-maker thus needs to map these reasons and find a way to quantify them correctly and identify the optimal regulatory solution based on the fairness and efficiency maxims. The role of the rule architect is to take the best decisions as he or she can; he or she strives for the optimal decision exemplified through the optimal rule. His or her primary preference is to excel at taking decisions that bring about consequences that he or she personally prefers. If he or she is conditioned by the maxims of the social contract, he or she will satisfy their own preferences by optimising efficiency and fairness of the preference satisfaction of his or her peoples. The rule architect, on this view, is like an external observer that takes his or her sole pleasure out of tuning to perfection the realisation of the desires and pleasures of those he or she observes. An architect of rules that takes his or her task seriously would want to perform their task consistently and accurately. A standardised method for assessing competing rule options would thus be useful. It should be accurate, yet practical, and it would need to identify and compare the level of efficiency and the level of fairness inherited in different regulatory options. For the purposes of this chapter, the efficiency maxim refers to total efficiency of preference satisfaction in line with expected utility theory, and the fairness maxim refers to a condition of an egalitarian distribution of preference satisfaction. A fair rule is thus as efficient, as egalitarian distribution allows. This is an efficiency standard that contains a distributional condition that is stricter than the Pareto 8 The discovery of new efficient techniques for satisfying preferences could count as external circumstances, which the legislator might want to respond to through a regulation to improve efficiency. New ideas about what is preferable within a group, such as increased tolerance towards alternative lifestyles, may similarly prompt the rule-maker to intervene so that the satisfaction of diverse preferences is fairly facilitated through the rules.

100  The Model of Fair Rules and the Kaldor–Hicks efficiency standards. While those two allow private gain if others do not lose, or potentially do not lose; the egalitarian standard demands distributional equality of all societal output. This is the ethical standard that the social contract, which is argued for in more detail in chapter four, presented in consequential terms.

II.  Rulemaking as a Decision Problem The identification of the optimal regulatory decision requires a qualitative comparison and ranking of the available options. It needs to be established which alternatives are comparatively inefficient or unfair towards achieving the regulatory consequences pursued. The proposed model uses selected microeconomic tools from decision theory (ie expected utility theory) to achieve this comparison. In line with expected utility theory, the object of comparison are the preferences of the relevant population. Each available alternative has a certain utility value towards satisfying preferences of the population and the task is to find a comprehensive method for making such comparison and thus identify the optimal rule. A decision matrix can be used to frame decision problems in a clear schematic way. Potential rules are ranked in rows of a table, against one or more potential scenarios of future events, which are listed in the columns of the same table. The conceivable outcome of each rule in each potential scenario is then described as a distinct consequence in the cells at the relevant intersection of the table. By using Von Neumann and Morgenstern’s rationality assumptions it is possible to infer cardinal numerical values on the preference utility of competing options, given that the decision is taken under conditions of probabilistic certainty.9 This is the classic expected utility inference method.10 Based on further rationality assumptions argued for successively by Savage, Harsanyi and Aumann the inference of cardinal numerical utility rankings can be extended to decision options for groups of people under decision conditions of risk and uncertainty.11

9 Von Neumann and Morgenstern’s method is conditioned on the acceptance of four assumptions about the rationality of the decision-maker. Their rational decision-maker must have complete and transitive preferences, that are independent of irrelevant alternatives and which satisfy the axiom of continuity. See John von Neumann and Oskar Morgenstern, The Theory of Games and Economic Behavior, 3rd edn (Oxford, Oxford University Press, 1954) 15–31. 10 A simplified version of the method works in the following way: start by assigning the value zero (0) to the worst option and one (1) to the best option. To find out the utility value of a third option for a given person, he or she should be offered the choice of a gamble between winning the best option and the worst option, or definitely receiving the third option. Start by offering low odds of winning the best option and then gradually improve the odds until the person would be willing to take a chance on the gamble, instead of receiving the third option for sure. If the person shifts at 30% odds of winning, the utility value of the third option is 30% of 1, that is, 0.3. If the person won’t shift until the odds of winning the best option are 90%, the utility value of the third option is 0.9. 11 Savage argued for the Bayesian inference of probabilities (subjective probabilities) over frequentist inference of probabilities (objective probabilities), which can be used to expand the utility inference

The Elements of the Model of Fair Rules  101 The prospect theory of Tversky and Kahneman relaxes some of the rationality assumptions to accommodate for certain consistent consistency violations, which are not caught by the expected utility model.12 This comes at the cost of mathematical complexity, but the benefit is a more accurate description of preferences and thus a more accurate basis for finding the optimal decision. The key insights for the purposes of this chapter are the phenomena of reference point dependence and loss aversion. The former implies that people tend to assess preference gains and losses in relation to a status quo reference point, instead of with reference to absolute levels, as normally would be assumed to be rational.13 The later implies that losses are conceived as weighing more than corresponding gains, instead of having symmetrical effects on preferences, as would be assumed in the rational choice model.

III.  The Elements of the Model of Fair Rules In order to compare and identify the best rule-making options, as measured against the ethical maxims of the social contract, the rule-maker would need to: i) identify a rule-making policy; ii) map the relevant stakeholders and their preference function towards the policy; iii) rank in order of importance the desires of the different stakeholders; iv) describe in cardinal numerical terms the probable consequences of each potential rule option on the preferences of the stakeholders involved; and v) process the gathered information to extract the optimally fair rule-making option. The basic assumptions and the tools of decision theory, outlined above, provide the means to achieve these tasks.

model to probabilistic situations of risk and uncertainty. See Leonard J Savage, The Foundation of Statistics, 2nd edn (New York, Dover Publications, 1972). Harsanyi suggested the assumption of the common priori so that Bayesian probability inference could be extended to groups of people. His argument was that by assuming the same priori beliefs and knowledge, different rational people should make the same probabilistic inference. This significantly enables interpersonal comparison of utility. See John Harsanyi, ‘Games with Incomplete Information Played by Bayesian Players I–III’ (1967–1968) 14 Management Science (I)159; (II) 320; (III) 486. Aumann elaborated further on the consequences of Harsanyi’s common priori by showing that ‘people with the same priors cannot agree to disagree’ if their beliefs are common knowledge. See Robert J Aumann, ‘Agreeing to Disagree’ (1976) 4 Annals of Statistics 1236, 1236. 12 Daniel Kahneman and Amos Tversky, ‘Prospect Theory: An Analysis of Decision under Risk’ (1979) 47 Econometrica 263; Daniel Kahneman and Amos Tversky, ‘Advances in Prospect Theory: Cumulative Representation of Uncertainty’ (1992) 5 Journal of Risk and Uncertainty 297. 13 Studies have further shown that people perceive the status quo reference point in terms of their expectations of how things will go, rather than in terms of what it ends up being. In that way, the factual reality may not change, but nonetheless people can experience a sensation of loss or gain if they had expected something to happen. See further on this point, Botond Köszegi and Matthew Rabin, ‘A model of reference-dependent preferences’ (2006) CXXI The Quarterly Journal of Economics 1133; Botond Köszegi and Matthew Rabin, ‘Reference-dependent Risk Attitudes’ (2007) 97 American Economic Review 1047; Botond Köszegi and Matthew Rabin, ‘Reference-dependent Consumption Plans’ (2009) 99 American Economic Review 909.

102  The Model of Fair Rules

A.  The Policy Objective of a Rule Every rule must start with an idea of what ought to be achieved through the rule. If the grand objective of any moderation, or initiation, of a rule is to improve society’s equilibrium of fairness and efficiency, a more detailed guidance is needed on how to achieve such an equilibrium in specific instances. To these ends, it is helpful to think of the regulatory regime as a hierarchal structure. At the bottom are the numerous rules that provide a specific guidance on conduct, the rules are a part of a regulatory field that is further supported by policy objectives. These policy objectives are reinforced by the abstract notion of the social contract.14 Following this line of thinking, the ethical guidance emitted from the two maxims of the social contract become the leitmotif of a subsequent policy objective, which further influences the ethical rationale of specific regulatory fields and individual rules. A society of people, bound together by a mutual understanding that society ought to be both fair and efficient, will not tolerate a policy objective that contradicts these fundamental norms. The same applies to regulations and individual rules. The task of the architect of the rules is thus to interpret which requirements the norms lay in specific regulatory instances. A society might, for example, agree that certain acts should be prohibited and defined as criminal. The underlying rationale of such an agreement would be supported by reference to the need to maintain a fair and efficient equilibrium in society, ie the norms of the social contract. The prohibition would work towards optimising the equilibrium. This kind of a general agreement would count as a policy objective that would then be executed through a criminal code and through specific rules in the criminal code. When designing the criminal code and its individual rules, the rule architect ought to strive for achieving the optimal equilibrium, both at the level of the entire code and at the level of individual rules. Success for the architect is designing a criminal code, or a rule in the criminal code, that is in compliance with the fundamental norms of the social contract and does thus qualify as a fair criminal code, or a fair rule in the criminal code. By thinking about rules in this layered hierarchal manner, a chronology of what comes first is presupposed. Society must first agree on the foundational norms of the social contract, then an agreement on policy objectives needs to be reached, before an agreement on implementation through regulation can be finalised. Agreement on the content of specific rules can subsequently only be reached when an agreement on the general scheme of the regulation of which it forms a part is in place.15 14 ie the hierarchical relation is as follows, starting from the general, ending with the specific: social contract > policy objective > regulatory field > individual rules. 15 In a parliamentary context, these processes often occur simultaneously; the political compromise on public policy plan is reached at the same time as the technical implementation is decided. The distinction between disagreeing on policy and disagreeing on the best way to implement a policy is thus often blurred in parliamentary practice.

The Elements of the Model of Fair Rules  103 While agreeing on a policy objective, the relevant stakeholders in a regulation can still hold diverging preferences towards the type of regulation needed to implement the objective, and they can also agree on the type of regulation, but disagree on the best type of a rule to achieve the general regulatory objective. A disagreement about rule design can thus exist at different levels and the rule architect must always solve the disagreement at the higher level before engaging with a disagreement at a lower level. This is due to the input a higher lever agreement has on the solution of a lower-level disagreement. There is no guarantee that a rule is fair if it is not compatible with the general regulatory scheme, or the policy objective derived from the maxims of the social contract. The rule architect thus needs to start by defining the normative policy agreement at the level immediately above the one where they are going to make a new rule.

B.  Unification of Moral and Efficiency Claims The stakeholders in a rule should be defined based on their prior circumstances during the rule-making process with regards to the policy objective at stake. By inference from Harsanyi’s insight about the common priori, it can be assumed that stakeholders that do not share common priori, are liable to hold different preferences towards the options that are available for implementing the policy objective. Based on the idea of the common priori it should be possible, supposing rationality, to assume what each type of a stakeholder ought to want with regards to the policy objective at hand, and thus it should be possible categorise them based on their preference function. It is important to note that the preferences of different groups or types of stakeholders are not necessarily equally important. Ronald Dworkin famously talked about rights as trumps, to emphasise that certain preferences of individuals should weigh more than certain collective preferences.16 This neatly expresses an important truth about the preferences of different individuals. In deontological ethics, this idea is expressed through Kant’s notions of the categorical imperative and the principle of universal law. Reasons compatible with the categorical imperative are, from the perspective of deontological ethics, superior to reasons based on hypothetical imperatives.17 Rights are reasons that claim to be compatible with the categorical imperative and such reasons thus demand to be considered lexically superior to non-right reasons. The categorisation of reasons for actions, into rights and non-rights may be helpful in some cases. It is, however, a crude

16 Ronald Dworkin, ‘Rights as Trumps’ in Jeremy Waldron (ed), Theories of Rights (Oxford, Oxford University Press, 1984) 153–67. 17 Immanuel Kant, Groundwork of the metaphysics of morals, rev edn (Mary Gregor and Jens Timmerman tr, Cambridge, Cambridge University Press, 2012) 26–34; Richard Burnor and Yvonne Raley, Ethical Choices: an introduction to moral philosophy with cases, 2nd edn (Oxford, Oxford University Press, 2017) 155–64.

104  The Model of Fair Rules binary categorisation, which fails to account for the varying degrees of importance attached to individual rights within the wider category of numerous rights. The progressive inflation of the number and types of rights also highlights a problem at the margins of what counts as a right. Often the grounds for including or excluding a reason at the margin are not obvious when approached by a deontological rationale. For the purposes of identifying the optimally fair rule among several choices, the deontological approach to ethics is thus overly crude. Binary categorisation into rights and non-rights insufficiently reflects the complex hierarchy of reasons that dictate human conduct. Once it is accepted that rights may exist in different shades of importance, the deontological approach to rights as expressions of the categorically imperative collapses. How to decide which of two incompatible rights should prevail, if both claim to be universally applicable expressions of the categorically imperative? Consequential ethics provide a simple solution to the dilemma of competing rights. By focusing on the consequences of each proposed action, it is simple to decide what should prevail in each instance. The most preferable consequences should prevail, regardless of labelling as rights or imperatives. Consequentialist approaches to ethics thus allow for the ranking of rights in terms of importance on a cardinal scale. A specific right signals a preference towards specific consequences, and the intensity of the preference can be measured and compared with the preference for other rights. Thinking of rights as simplified proxies for a specific preference offers the possibility of more descriptive accuracy than the binary deontological scale of rights and non-rights can achieve.18 It is less controversial to apply consequential measuring techniques for determining the priority of regulatory choices that do not involve rights. On the orthodox variant, different considerations are simply ranked in terms costs and benefits denoted in monetary terms.19 Thus, if it were possible to quantify the preference utility of different rights considerations, the non-rights considerations should not pose a problem. On this view, every potential argument for or against a specific rule, including both ethical and non-ethical arguments, can be reduced to

18 In legal adjudication the proportionality test is frequently used to compare competing rights claims. The sophisticated variant of the test put forward by Alexy (see Robert Alexy, A Theory of Constitutional Rights (Julian Rivers tr, Oxford, Oxford University Press, 2002) 44–110) focuses on the consequences of the competing claims and applies a Kaldor–Hicks efficiency standard in order to determine whether a given restriction of a right is proportional or not. 19 The orthodox cost–benefit analysis is nowadays widely used by all kinds of regulators. The drawbacks of the method are well documents. Ethical considerations (such as various rights) are, for example, resistant to quantification through monetary means, and monetisation favours those with the means to pay the highest price. See for example Amartya Sen, ‘The Discipline of Cost–Benefit Analysis’ (2000) 29 The Journal of Legal Studies 931; Robert H Frank ‘Why is cost–benefit analysis so controversial?’ (2000) 29 The Journal of Legal Studies 913; Matthew D Adler and Eric A Posner, New Foundations of Cost–Benefit Analysis (Cambridge MA, Harvard University Press, 2006); Lewis A Kornhauser, ‘On Justifying Cost–Benefit Analysis’ (2000) 29 The Journal of Legal Studies 1037; Cass Sunstein, ‘The Limits of Quantification’ (2014) 102 California Law Review 1369.

The Elements of the Model of Fair Rules  105 a preference towards specific consequences. By attaching a high preference value to the consequences that individual rights promote, the importance of rights can be conveyed through the language of preferences. The shift from a deontological conception of rights, over to a consequential conception, is appealing in terms of practicality. It is challenging from the outset to explain the idea of multiple imperative rights in strict deontological terms. How do multiple imperative rights make any sense in social organs where compromises are integral to the organs function? It feels more reasonable to think of rights as strong preferences towards some substantive consequences that promote something essential to the human condition. For example, most people have a strong preference towards remaining alive. This preference can be branded as the right to life, but it is not imperative in the sense that it is conceivable that some other preference might override this preference given the right circumstance, eg using a lethal force to stop a school shooting. The shift from deontological ethical thinking to a consequentialist ethical thinking, facilitates an explanation of how conflicting ethical claims can be reconciled. A difficult argument about the prevalence of incompatible moral rights is replaced by a practical resolution of conflicting preferences. The role of marginal non-rights in relation to marginal rights can also be explained in a more satisfying manner based on consequential thinking about preferences. By eliminating the deontological rights distinction, a uniform set of intra compatible preferences appears that can be ranked in a cardinal order of importance. However, this only extends to stakeholders with the same common priori and thus the same preference function. To rank and compare the preferences of different types of stakeholders, which do not share common priori, an additional device is needed that achieves interpersonal comparison and importance ranking of what each stakeholder type wants.

C.  Interpersonal Comparison of Stakeholder Claims It should be noted that in the economics literature, the interpersonal comparison of utilities was long deemed impossible. Kenneth Arrow, for example, wrote: If we look away from the mathematical aspects of the matter, it seems to me no sense to add the utility of one individual, a psychic magnitude in his mind, with the utility of another individual. Even Bentham has doubts on this point.20

For the context of a rule-making practice, this view is overly negative. Most people have the capacity for empathy; some would even argue that the sophistication of human empathy contributes to their evolutionary success as a species.21 20 Kenneth J Arrow, Social choice and individual value, 2nd edn (New York, John Wiley & Sons, 1963) 11. 21 Robert Boyd and Peter J Richerson, ‘Culture and the evolution of human cooperation’ (2009) 364 Philosophical Transactions of the Royal Society B: Biological Sciences 3281.

106  The Model of Fair Rules Arrow’s impossibility theorem may of course hold in a strict mathematical sense, but the actual practice of being a human still demands such comparison. For rule-makers it is a standard procedure to assess the impact of regulations on those who must live by them. A prerequisite for such an assessment is an interpersonal comparison of some kind, often a cost–benefit analysis. The question is thus not whether interpersonal comparison of utility is possible; it is merely a matter of how accurate we would like the comparison to be.22 Arrow also warned that interpersonal comparison of utility would always require a value judgement.23 For the purposes of rule-making, this objection does not pose a problem: rules normally seek to implement a normative value; the task is simply to optimise the rule with regards to the preference utility of the relevant population towards that value.24 Branding some preferences as rights can be viewed as a rudimentary approach to interpersonal comparison of conflicting preference claims. The one that can claim a right should prevail against those who cannot claim a right. As previously explained, the simple binary scale is however insensitive to the small variations at the opposite margins of what counts as a right and has difficulties with prioritising conflicting claims within the category of rights. Interpersonal comparison based exclusively on rights also underestimates the importance of efficiency considerations. The traditional consequentialist method – of reducing claims to a monetary value and thus achieving interpersonal comparison of competing claims through the universal medium of money – faces the inverse problem. It has difficulties with incorporating ethical claims due to their lack of practical utility that can be translated by the medium of money.25 A first step towards solving these methodological difficulties is to think of preferences as a universal medium that incorporates both ethical claims for specific consequences and efficiency claims for specific consequences. Once a unity has been established with regards to the medium of measurement for the personal preferences of likeminded stakeholders, a method is needed to enable objective comparison of the importance of the personal preferences of different types of stakeholders towards a specific issue. If one group of stakeholders prefers a different solution to another group, a way is needed to objectively assess which group has a more important claim. The method suggested here is a preference index (PI) that interpersonally compares and ranks on a cardinal scale the objective importance of competing preference claims of different stakeholders. 22 In formal language, this is a question which of Arrow’s axioms should be relaxed. See Nolan McCarty and Adam Meirowitz, Political Game Theory (Cambridge, Cambridge University Press, 2007) 67. 23 Arrow, Social choice (1963) 11. 24 For our purposes this value is the ethical standard of the social contract as previously explained. The very existence of society thus presumes a foundational value (ie an ethical choice) that individual members can hold diverging preferences towards how is best attained. 25 See, for example, Sen, ‘Discipline’ (2000); Frank ‘Cost–benefit analysis’ (2000); Adler and Posner, New Foundations (2006); Kornhauser, ‘Justifying’ (2000); Sunstein, ‘Limits of Quantification’ (2014).

The Elements of the Model of Fair Rules  107 The PI can be used as an exchange rate between competing interests. In the same way as it can be established, based on market prices, how many US dollars are in a euro, it should be possible to establish a price ratio between the preference units of stakeholder A and stakeholder B. To build the PI, there are two practical requirements: first, a system to measure magnitudes of preferences; and second, a way to compare the importance of similar magnitudes of different preferences. A simple device can be used to solve the problem of magnitudes. Magnitudes of different precious metals can be described in comparable terms through the metric system; a system using the same measuring principles is possible for describing magnitudes of different preferences. A numerical value for the maximum fulfilment of a specific preference in consequential terms is simply assigned, and another numerical value for the complete deprivation of the same preference, ie the best and the worst consequences in relation to this specific preference.26 If  the assigned numbers are zero (0) and one (1), different consequences with regards to that specific preference can be assigned a magnitude value between 0 and 1, depending on the specific intensity of preference fulfilment. The specific level in each case can be found through the normal utility inference method of Von Neumann and Morgenstern. In this way, 10 per cent fulfilment of a specific preference can be talked about in relation to 10 per cent fulfilment of another preference. The advantage this brings is like the advantage of being able to talk about 100g of silver and 100g of gold. A uniform medium for comparing magnitudes (eg grams), enables the comparison of an objective value in terms of a price per magnitude unit (eg price per gram). An objective value of the preferences of different stakeholder types can be achieved (ie the objective price of each preference unit), either by reference to a universally applicable importance value, or by reference to a case-specific importance value. The former method is easier to apply due to its nature of generalization, but loses descriptive accuracy that can be achieved through a custom-made standard of importance valuation. A mixed comparison approach can also be imagined, which takes some of the case-specific details into consideration, but nonetheless tries to achieve results that can be applied universally to a range of circumstances. The key to achieving objective comparison can be explained based on the veil of ignorance thought experiment of John Rawls.27 It must be imagined that the 26 The exact point of complete deprivation or saturation of a specific preference can create an interpretive problem. For example, the preference for free expression is saturated and deprived at uncontroversial points (total prohibition or total liberty), while the preference for owning money is not as obviously saturated. One way to solve this would be to refer to the average lifetime use of a specific resource as a saturation point and the average use over some specific short timespan (a day, a month, or a year) as the point of scarcity. If we again refer to the preference for money, it would be saturated at the amount an average person needs during his or her lifetime, and deprived if the amount would only last the average person for one day. The preference for other types of resources could be bracketed in a similar way. 27 John Rawls, A theory of justice (Cambridge MA, The Belknap Press of Harvard University Press, 1971) 136–42. Note also Kant’s formulation of the universal law, which essentially is a device for

108  The Model of Fair Rules decision on the level of importance for the competing preferences is taken from behind the veil of ignorance; the decision on the importance weighing will apply to the decision-maker, but it does not yet know which type of stakeholder it is. Towards the end of achieving objectivity of importance inference, it is not enough to make a simple survey of what people prefer. Such results would be exposed to a bias in favour of the actual position of the respondents. A survey is thus needed, which measures preference importance from a state where the respondent is oblivious about his or her actual position. For the simpler comparison of generalised importance, the following method can be used. It focuses on absolute importance values, as opposed to the more nuanced importance values that are relative to the current status quo. If comparing two preferences (p1) and (p2), a simple survey using comparable magnitudes of both can be used to determine which is more important. Once an importance relation in the ordinal sense has been established, the Von Neumann and Morgenstern utility inference method can be applied to find the location of the inferior preference on a scale between the superior preference (1) and a neutral consequence (0). This would provide a cardinal ranking of importance for the stakeholder preferences under assessment. This method could provide a good general indication of preference importance, but is exposed to inaccuracies in case-specific circumstances since it ignores the insight about reference point dependency, described by Kahneman and Tversky.28 The more complex comparison method seeks to incorporate the reference point dependency element from prospect theory for achieving more descriptive accuracy in case-specific circumstances. A method that internalises this insight is thus proposed, based on the rationale that a subjective level of preferableness, attached to different types of preferences in a given situation, can be inferred from a comparison of the effects on each type from a positive and a negative magnitude change measured from the status quo. The main difference between the two variants is that the simpler variant measures preferences in absolute terms, while the more complex variant measures preferences in relative terms, ie relative to the status quo. The latter method can be sketched as follows: Comparing two preferences,29 1 (p ) and (p2), their range is bracket by describing a deprivation scenario (–pn) and a saturation scenario (+pn) for both. This provides four variables (–p1, +p1, –p2 and  +p2) that can be bundled into a best case (+p1+2) and a worst case (–p1+2) scenario. The status quo reference points (Qp1 and Qp2) form the fifth and the sixth establishing the objective point of view. See Immanuel Kant, Groundwork of the metaphysics of morals, rev edn (Mary Gregor and Jens Timmerman tr, Cambridge, Cambridge University Press, 2012) 34. 28 The reference point dependence is the key insight of Kahneman’s and Tversky’s prospect theory and it has been advanced by the insight of the expected status quo as opposed to the current status quo. See Köszegi and Rabin, ‘Reference-dependent preferences’ (2006); Köszegi and Rabin, ‘Reference-dependent Risk Attitudes’ (2007); Köszegi and Rabin, ‘Reference-dependent Consumption Plans’ (2009). 29 The method is not restricted to the comparison of two alternatives. In theory, any number of preferences could be compared and ranked using this method.

The Elements of the Model of Fair Rules  109 variables that can also be bundled into a general status quo (Qp1+p2). Where –p1+2 is the worst possible scenario and +p1+2 is the best possible scenario, the other points should fall on a line between these two extremes. The relative preferableness of each point could then be extracted and located on the line by using Von Neumann and Morgenstern’s utility inference method. That is how a given magnitude of a given preference can be compared with a corresponding magnitude of another preference to determine their relative impact on the scale between the worst and the best possible consequences. The use of a reference point adds nuance. The impact of a given magnitude is not being measured in absolute terms, but measured by counting from the currently deposited magnitude. To achieve this, the bundled reference point needs to be located on the overall scale between the best and the worst. The location is potentially not be in the middle, thus creating a problem of asymmetrical effects of corresponding magnitudes of positive and negative consequences. To overcome this, the scale needs to be split into a negative and a positive subscale, breaking at the location of the bundled status quo reference point. The split enables an assessment of the impact of a given loss or a gain magnitude for a specific preference, relative to the bundled status quo, and the bundled best or the worst-case scenario. To measure how preferable a specific preference type is compared with another type, measuring points are simply dropped on the defined scales of negative and positive consequences, using the same magnitude for both types (eg 20 per cent increase). The number of measuring points affects the accuracy of the description. Dropping only one measuring point risks neglecting different effects of losses and gains for different types of preferences.30 Using measuring points on both the positive and the negative side of the status quo requires symmetry in terms of distance from the reference point, or alternatively a special weighing to ensure that the outcome for gains and losses weigh equally in the aggregate result. By using the location of the bundled status quo on the overall scale, the registered impact on the two subscales can be translated into a loss or a gain on the overall scale relative to the current level of preference satisfaction. Comparison of the effects of corresponding magnitudes of different preferences shows the relative impact of each, and thus the relative preferableness of each. By using Von Neumann and Morgenstern’s utility inference method the difference in impact can be expressed in numbers. A simple start to applying this method would be to use the magnitudes already defined during the bracketing of the total range for each preference as measuring points on the subscales; status quo is 0, total deprivation is an 100 per cent loss on the negative scale, and complete saturation is an 100 per cent gain on the positive scale. Assuming completeness, the aggregate preferableness of a bundle of preferences equals the sum of its parts. Each part, ie each preference type, can however 30 The key element in some types of preferences may be that we care a lot about maintaining at least the current magnitude, but we would not care so much about an increase. For other types this effect can be in reverse.

110  The Model of Fair Rules carry different weight in the bundle. Using two preference types (p1 and p2) as an example and the magnitude of an 100 per cent gain as a measuring point, the sum of an 100 per cent gain for p1 and p2 should equal a corresponding magnitude for a bundle of the two. However, a given magnitude of p1 can potentially weigh more, or less, in the bundle p1+2 than an equivalent magnitude of p2.31 The difference in weight indicates a difference in relative preferableness of the two preference types. The weight of each relative to the bundle and the status quo can be extracted in numerical terms by applying the utility inference method of Von Neumann and Morgenstern. This can be repeated for any number of measuring points to increase accuracy. Subsequently these values on the subscales are translated into values on the overall scale. The aggregate weight of the measuring points for each of the preference types on the overall scale provides a comparison of impact from which relative preferableness of the different types can be inferred. We can call the aggregate result for each preference type a preference index and the ratio between different types a preference ratio.32 Table 5.1  Step-by-step guide to the composition of a PI Action

Result

Step 1 Define best- and worst-case scenarios for each stakeholder with regards to the policy objective at stake.

100% and 0% magnitudes of preference fulfilment established for each stakeholder with regards to the policy objective at stake.

Step 2 Identify the status quo for each stakeholder between the extremes of the best- and the worst-case scenarios (eg how preferable is the current situation with regards to the policy objective?).

Information about the status quo preference level for each stakeholder.

Step 3 Compare effect of a total loss and a complete gain on each stakeholder counting from the status quo as observed by a neutral bystander (eg how would a 100 per cent gain and a 100 per cent loss objectively affect the different stakeholders?).

Information about the relative importance of each stakeholder’s claim.

By establishing a PI for each of the relevant stakeholders with regards to a policy or a regulatory objective, the legislator has means to compare the relative importance of differing stakeholder preference claims. He can, for example, see that a 31 Think of a bundle of precious metals worth €1000. The bundle contains 100g of silver worth €100 and 100g of gold worth €900. 32 We can think of this as an exchange rate between different preferences: How many units of one preference would equal a single unit of another. The function would be like exchange rates of different currencies.

The Elements of the Model of Fair Rules  111 preference claim of stakeholder A is three times more important than the competing preference claim of stakeholder B. Depending on whether the insight about reference point dependence is utilised, this comparison refers to absolute levels of preference satisfaction (simple method), or the change in preference satisfaction measured from the current status quo (complex method).

D.  Registration of Preferences into Pay-off Matrices Having provided the means to interpersonally compare preference utility with the  PI, the next step is to describe how the different types of stakeholders feel about specific rule alternatives for realising the policy objective at stake. The PI required an objective assessment perspective, due to the contaminating potential of information regarding actual prior circumstances. The description of actual preferences is not subject to the same requirement. On the contrary, the rule architect should make a point of describing the subjective point of view. Each subjective point of view represents a separate preference function. Each person sharing common priori for a specific policy objective shares a preference function for that specific objective. By listing rule alternatives in the rows of a matrix and the potential states of the world in the columns of the same matrix, the pay-off in terms of preference satisfaction can be described in the cells of the matrix, representing how satisfied a stakeholder is with each proposed rule, given a specific state of the world scenario. The precise utility of each rule proposal could be inferred in a cardinal numerical format by using Von Neumann and Morgenstern’s inference methodology. A separate pay-off matrix could be assembled for each stakeholder type to get an overview of how the proposed legislative actions are received by different stakeholder types. To account for uncertain future events, different state of the world scenarios could be accounted for and assessed based on probability of occurrence (eg 30 per cent probability that that the state of the world will be a, but 70 per cent probability that it will be b). Table 5.2  Example of a preference matrix for policy objective X Proposed rule Future state of the world

Rule 1

Rule 2

Rule 3

a – World – b

a – World – b

a – World – b

Stakeholder A preference Stakeholder B preference Stakeholder C preference Stakeholder D preference

Once information has been gathered about how strongly different stakeholders prefer different rule options (ie pay-off matrices), and information about how

112  The Model of Fair Rules important from an objective point of view the preferences of different stakeholders are (ie PIs), information about the preference utility of different rules can be derived.

E.  Identifying Fair and Efficient Rules Following Tversky and Kahneman’s terminology about decision-making under risk, the steps taken to establish the PI and the pay-off matrices belong to the framing phase of decision-making, in which the relevant information is gathered, processed and often simplified to enable a subsequent valuation of the available decision options during the valuation phase, which precedes the eventual selection of the optimal option. For the purposes of the model of fair rules, the objective is to identify prospective rules that promote the ethical rationale of the social contract. This rationale can be summarised as follows: individually we seek to be treated fairly vis-à-vis the others by claiming our share of the collective output, and as members of the collective we seek efficiency to maximise the joint output from which the individual share is derived. As previously explained, the maxims of fairness and efficiency thus form the ethical rationale of the social contract. When the rule architect wields the power to define and implement the details of the social contract and the policy objectives, through regulations and individual rules, he or she must carefully consider the duality of the contract’s rationale; their decisions must promote the general good, while simultaneously making sure that the right incentives are in place from the individual’s point of view. Different valuation methods are needed to assess these different optimisation aspects. Through the information gathered in the matrices, knowledge has been gathered about what makes each stakeholder type tick. It is known how each of the types perceives different rule options in terms of magnitudes of preference satisfaction. The magnitudes represent a subjective view, seen from the perspective of each stakeholder type. To make the subjective information useful for assessing efficiency, an ability to compare and quantify the different subjective perspectives is required, ie the ability to interpersonally compare and quantify the magnitudes of preference satisfaction for different types of stakeholders. A subjective representation of magnitudes can be turned into an objective representation by factoring in the relative importance of each of the subjective perspective through the PI. After adjustment, the numbers in each of the matrices would represent the change in magnitude of preference satisfaction, considering the objective importance of the preference function of that specific stakeholder type. After adjustment, the numbers in different matrices can be meaningfully compared in terms of objective magnitudes of preference fulfilment.33 33 Think of two collectors of coins: One collects gold coins and prefers them over other coins; the other collects silver coins and prefers them over other coins. Their subjective preferences are different

The Elements of the Model of Fair Rules  113 The size, or the relative size, of the stakeholder segment under assessment is of key importance to determine the overall efficiency of each legislative option under consideration. The size could either be determined in absolute numbers, or as a percentage of the whole population of relevant stakeholders. For the purposes of manageability, it is useful to assume completeness of the diverging preference functions. This means that all members of the population must be assigned to one of the defined stakeholder types and not more.34 The total efficiency of each decision option is derived from three variables: the applicable subjective preference value, the relevant PI and the stakeholder segment size. The PI is used to translate subjective preference values (SVs) into objective preference values (OVs), and the population segment size (P%) determines the number of individuals producing the said objective value of preference satisfaction. These variables enable the extraction of the total output of each stakeholder segment and, through aggregation with other stakeholder segments, the total output of all relevant stakeholders for each of the regulatory options that are being compared. These numbers in their final format can be called objective ­utility (OU). The regulatory option with the highest OU score accordingly ranks as the most efficient option. Valuation of the fairness criteria requires an alternation of perspectives. Fairness should be identified subjectively from the perspective of the individuals involved, which contrasts efficiency that is best approached objectively in utilitarian terms. An individual demand for equitable treatment vis-à-vis the others, constitutes the essence of the fairness criteria. The optimal regulatory option should affect the preference fulfilment of everyone by an equitable magnitude. However, this should not be taken to mean that each has a claim to receiving the same magnitude of his or her subjective preference fulfilment. As has already been established, a subjective preference of a person has an objective value when compared with the preferences of others. Some hold objectively important preferences, while others hold less important ones. When making a claim for an equitable share, the relative importance of the claim in the objective sense will determine the magnitude at which the individual feels that he or she is being treated fairly in the subjective sense.35 due to their commitment to their respective collections. A way to represent a general preference view over these two subjective preferences would be to find out the general exchange index of collections of gold and silver coins. By knowing the exchange index, we could perhaps establish that a certain magnitude of gold coins – one coin, for example – is worth five coins of silver. The subjective preferences tell us what each wants, but the objective preference tells us that a gold coin is five times more preferable than a silver coin, generally speaking. 34 All stakeholders (Sn) in the set X must also be members of one of the subsets: Xa, Xb, or Xc. 35 Think of two individuals with different tastes. One has an expensive taste when it comes to wardrobe composition, while the other is happy with cheaper options. Imagine they work comparable jobs at the same company that annually allocates a small common budget for buying work clothes. To fulfil their private (subjective) preferences for business attire to the same magnitude level, one would need expensive suits, while the other would be just as happy with cheap suits. Objectively speaking, the expensive suit costs at least twice as much as the cheap suit. Given this difference in objective value, the one with the expensive taste would need to relax his or her claim to a fair share of the clothing budget.

114  The Model of Fair Rules During the valuation of the fairness criteria, the rule architect needs to translate the SVs described in the matrices into subjective values adjusted for relative importance. To achieve this, the legislator can, once again, use the PI. The PI shows the relative importance of a single magnitude unit of the distinct types of preferences involved. This information can be used to establish a ratio between the relevant preferences. The ratio indicates how much, in terms of magnitudes, is needed of each preference to achieve an equitable outcome.36 In practice, the rule architect would need to be able to see, when comparing the same decision option across the matrices of different stakeholders, whether they perceive the consequences equitably. In terms of clarity, it would thus be ideal if the rule architect would simply need to find numerical pairs in corresponding consequence cells of the different decision matrices. The pairs would indicate an optimisation in terms of equitable preference fulfilment between the different population segments and simultaneously indicate a solution to the fairness criteria. To achieve this clarity of representation for the purposes of valuating fairness, the numbers representing the subjective preferences and the PI would need to be divided, instead of multiplied as when the most efficient option is identified. The result could be referred to as objective fairness score (OF) as opposed to OU for measuring efficiency of specific regulatory options. Notably, the size of the stakeholder segment is irrelevant when determining fairness. Fairness is perceived individually, while efficiency concerns the entire organ at stake collectively. The identification of the fair regulatory options and the efficient regulatory options can be summarised in the following way where (SV) is subjective preference value, (PI) is preference index and (S) is a stakeholder: Inferring efficiency: SV×PI×S% = OU SV Inferring fairness: — = OF PI The rule architect now has information about the objective total utility of each regulatory option, and they have information about how each option is perceived in terms of objective fairness from the point of view of each stakeholder. To satisfy the maxims of the social contract, the rule architect would simply need to identify the rule options that have the most equitable distribution among stakeholders of the objective fairness score, and pick the one that has the highest total objective utility.37 This ensures that the fair options are identified and the most efficient of them is selected. A claim of receiving the same preference magnitude of suits would not be perceived as fair by anyone, not even by the person with the expensive taste. A fair magnitude ratio might be one expensive suit for you for every two cheap suits for me. 36 If, for example, the preference ratio between two preferences P1 and P2 is 1:0.75, we know that a single magnitude of P1 is equally preferable to 0.75 magnitudes of P2. An equitable legislative option would satisfy the preference satisfaction of the population in these proportions; ie for every fulfilled magnitude of P1 a 0.75 magnitude of P2 would also need to be fulfilled. 37 In theory, the preferableness of a specific legislative option should rank equally across different groups factoring in the objective value of their preferences. In practice, it could become necessary to

The Utility of a Model for Identifying Fair Rules  115 Using this model of fair rules, it should be possible to identify regulatory options that achieve optimality in terms of fairness and efficiency. The fairness criterion ensures ethical plausibility of the model’s recommendation, while the efficiency criteria ensures maximisation of overall preference satisfaction within the boundaries of the ethically plausible.

IV.  The Utility of a Model for Identifying Fair Rules Optimisation methods are widely used by all kinds of regulators. Broadly, they can be split into approaches based on either a deontological rationale or a consequential rationale. The proportionality test is a commonly used example of the former and cost–benefit analysis of the later. The defining feature of deontological approaches is their focus on the morally imperative, which should be followed irrespective of consequences. In practice, the rigidness of the morally imperative makes the deontological approaches to the law unattainable in their pure form, thus the practical standard is commonly relaxed to comply with the proportional. However, this has a methodological cost, as can be illustrated by analysing Robert Alexy’s celebrated variant of the proportionality test.38 Alexy’s representation of the three tiered proportionality test can be formalised in the following way where action (a) has consequences (+) and (–) for the primary interest (P) and the secondary consideration (S):39 Table 5.3  Consequentialist representation of Alexy’s proportionality test Step 1: Suitability

Consequences of (a) must benefit (P)

Efficiency

Step 2: Necessity

Consequences of (a) must benefit (P) more than other available actions

Pareto efficiency

Step 3: The sum of (+) consequences to (P) and (–) Narrow proportionality consequences to (S) from (a) must ≥0

Kaldor–Hicks efficiency

As this shows, the purity and simplicity of a deontological argument is sacrificed through the application of Alexy’s proportionality test. Its actual application implies the use of a consequential methodology, although not a sophisticated one in terms of quantification methods, and on top of that its output is hedged on an ethical standard equivalent to the Kaldor–Hicks efficiency test. Despite the allow a small range of variation in order not to exclude options that result in only marginal inequality of preference satisfaction. 38 Alexy, Constitutional Rights (2002) 44–110. 39 Giovanni Sartor has shown elegantly how Alexy’s proportionality test can be quantitatively formalised using non-numerical magnitudes (see Giovanni Sartor, ‘The logic of proportionality: reasoning with non-numerical magnitudes’ (2013) 14 German Law Journal 1419). This representation draws inspiration from his work.

116  The Model of Fair Rules intuitive appeal of the proportionality test, the output of the test is thus potentially incompatible with the ethical maxims of the social contract. An action can be considered proportional, despite allowing a distribution of overall efficiency enhancements, which may deteriorate the position of some population segments, while improving the position of other population segments. This is an outcome that is in line with the Kaldor–Hicks efficiency standard. The cost–benefit analysis approach to law is faced with an inverse problem. While claiming immunity from the deontic properties by focusing on a neutral measuring of consequences, a reliance on an ultimate ethical choice is inescapable. Commonly, the Kaldor–Hicks or Pareto efficiency standards are used as normative benchmarks of optimal efficiency, but both alternatives imply a selection of a standard with ethical and distributional implications. The Kaldor–Hicks standard explicitly allows inequality in favour of total efficiency, while the Pareto efficiency standard fares slightly better in ethical terms due to its stricter distributional requirement. By using monetisation as a quantification tool, the cost–benefit analysis also risks neglecting qualitative considerations that are not easily quantified in terms of money, such as ethical considerations, and is subject to a problem resulting from the fact that people do materially not have equal means to show willingness to pay. The result is a recommendation that risks neglecting important qualitative concerns, is quantitatively eschewed in favour of those with material means and is based on a utilitarian ethical standard.40 The model of fair rules, consequential in approach, reduces ethics in the context of rule-making to a single imperative demand: the rules ought to be fair. For the purposes of the model, fairness should be understood as a quantifying term in relation to specific desires of the people subject to the rules. The rule architect ought to fulfil the desires of the people to the greatest extent possible (ie efficiently), but must do so in quantities perceived as fair from the individual point of view (ie fairly in an egalitarian distributional sense). The rough sketch of a theory and methodology provided in this chapter outlines the chief considerations for enabling quantification of the fairness concept of the social contract. The next step for the rule architect would be to gather data and design the computer algorithms for extracting regulatory recommendations out of the data set, using big data techniques. As tantalising as this may sound for those who are accustomed with equating democracy with the election of representatives, it is hard to imagine a future where demographic data-mining on a grand scale will not play a part in democratic rule design, both in a procedural context and more generally. The overall objective of this book is to conceptualise procedural fairness in the context of EU competition law – as explained in chapter one, the current state of the literature on EU competition law is unclear about how procedural fairness should be understood and dealt with in practice. In chapter two it was established

40 Sen

(n 19); Frank ‘(n 19); Adler and Posner (n 19); Kornhauser (n 19); Sunstein (n 19).

The Utility of a Model for Identifying Fair Rules  117 that the concept of procedural fairness in general concerns a dilemma of balancing between the two chief normative considerations of any procedural design, ie between the accuracy and cost-efficiency of the procedure’s output. In chapter three it was demonstrated how this analysis fits the context of EU’s formal and informal procedural regime for enforcing the competition provisions of the TFEU. The solution to this balancing dilemma revolves around the elusive concept of fairness. The procedural architect needs to know how this concept conditions his or her suggestions to how the balancing dilemmas should be solved. In chapter four an argument was made for an understanding of fairness in the context of laws that defines it as a distributional issue. Once procedural fairness has been defined as a distributional issue, a comprehensive and consistent methodology for identifying optimal solutions to these distributional dilemmas is required. This chapter has outlined such a methodology and completed the theoretical aspect of conceptualising procedural fairness. In the two final chapters several dilemmas of procedural fairness in the context of EU competition law will be identified and analysed against the background of this methodology and conceptualisation.

6 Analysis of Three Fairness Dilemmas Publicly accessible documents from competition law litigations before the Luxembourg courts provide an insight, on not only what the law is on specific issues, but also on what the parties to the cases want the law to be. This becomes apparent if the issue at hand is undetermined, or vaguely determined. In such cases, the parties have more discretion to argue how the respective rule ought to be, usually based on their respective preferences. The courts also assume the ex ante perspective and thus determine what the rule ought to be for the current and future instances, rather than identifying what the rule is per se.1 The Advocates General (AGs) of the Court of Justice, assume a similar role in their opinions on legally undetermined questions, and thus reveal their preferences for how a specific issue should be resolved in an ex ante sense. Publicly accessible documents from the EU’s legislative processes provide similar information about the preferences of different stakeholders in legislation, gathered through public consultations on policy papers and draft proposals. However, the arguments of the legislator for the choosing of specific regulatory alternatives are not as transparent as in the case of a court decision. The expectation of rationality and reasonableness on the political process is weaker than it is on the judicial process. The theory of procedural fairness used by the legislator is thus mostly revealed through its actual choice, without an elaborate reasoning about the selection of that choice. Some indications about the motivation for a choice can nonetheless be extracted from discussions and views expressed during the political and the administrative process of adopting new legislation. The conceptualisation of procedural fairness in EU competition law, established in the preceding chapters of this book, provides a framework for thinking about difficult issues of procedural fairness in competition litigation, such as when no obvious answer can be extracted from the existing body of positive law, and in the design of new procedural rules through the legislative process. Litigation records of the Court of Justice provide information about how the different stakeholders and role occupants in the procedure perceive the concept of procedural 1 Note though how the different schools of legal philosophy explain what happens in such instances. Hart, for example, talked about gaps in the law that judges had discretion to fill, while Dworkin argued that judges never act as if they were assuming the legislative role, but rather relied on alternative sources of law, such as principles of ethics. More recent scholarship in philosophy and legal philosophy, eg by Timothy Endicott, has explored this issue under the term ‘vagueness’ in language and law.

Analysis of Three Fairness Dilemmas  119 fairness and information about their suggested solutions to difficult dilemmas. Documents from public consultations and policy studies provide similar information from the legislative process. In this chapter, three procedural fairness problems in EU competition law, which have been dealt with through litigations before the Court of Justice and through legislative initiatives, will be analysed in order to demonstrate the clarity that a common understanding of the concept of procedural fairness brings to disputes about the ideal design of procedures in competition law litigation. To identify suitable court cases for analysis, the case register of the Court of Justice was searched for cases with the subject matter ‘competition’ and containing reference to Article  47 CFR and Article  48 CFR in the grounds of judgment. After the entering into force of the Lisbon Treaty, the primary source of reference for those claiming rights of the procedural fairness type has been Article 47 CFR and Article 48 CFR and is thus an appropriate context to look for arguments relating to preferences over the fair design of competition procedures.2 This search identified 77 cases that had been closed by November 2019.3 These cases were subsequently reviewed with the aim of identifying issues concerning procedural fairness that were undetermined or vaguely determined by the existing positive law. Two such issues in separate lines of cases were identified. The legislative dilemma was identified by reviewing recent legislative initiatives within the field of competition law procedure. The issue of collective redress as a mechanism for private enforcement of competition law quickly emerged as a suitable topic to examine in more depth. This was partly due to the wealth of publicly available documents about the issue and partly due to how controversial the issue became among the differently situated stakeholders. Additionally, considerable academic literature already existed on the issue. The arguments put forward by stakeholders and other role occupants for solving the fairness dilemmas in these case studies were analysed in detail against the main elements in the conceptualisation of procedural fairness introduced in previous chapters of this book. In line with that conceptualisation, it should be predictable how the parties argue in terms of emphasis on either accuracy or efficiency. Furthermore, the methodology for how the AGs and the Court approach these problems of balancing from a neutral perspective is also revealed, thus providing information about the present consensus about how such issues ought to be dealt with, ie to the extent such consensus currently exists. Finally, the studies provided a glimpse into how the political and administrative legislative process of the EU deals with contentions issue of balancing contradicting procedural fairness considerations. 2 Note that these articles of the CFR are not the only sources for claiming procedural rights in EU law. There are other such sources with constitutional status, eg Art 41 CFR. Article 6 ECHR is also still a valid source for claiming procedural fairness, although it overlaps somewhat with Arts 47 and 48 CFR. 3 The survey was based on a search on the Court’s website (http://curia.europa.eu) in November 2019. 66 of the cases had a reference to Art 47 and 11 to Art 48.

120  Analysis of Three Fairness Dilemmas

I. The KME/Chalkor Cases On 8 December 2011, the Court of Justice handed down judgments in three cases concerning a cartel in the market for copper plumbing tubes and in the market for copper industrial tubes.4 These cases are usually referred to as the KME/Chalkor cases.5 The applicants each argued that the General Court had failed to instigate a full judicial review of the Commission’s cartel decision, as made evident by the Court’s acknowledgment of a discretion the Commission had to assess certain economic facts. This, they argued, amounted to a breach of their procedural rights as defined by Article 6 ECHR and Article 47 CFR. One of the contested issues concerned the proper classification of the Commission’s competition proceedings. The applicants argued that it should be categorised as a criminal proceeding for the purposes of Article 6 ECHR, which accordingly would move the procedure within the scope of the procedural guarantees provided therein. The Commission, however, relied on a recent case law of the ECoHR, which seemed to categorise competition proceedings as being not fully criminal in character for the purposes of those procedural guarantees. Advocate General (AG) Sharpston supported the Commission’s reading of the ECoHR’s case law in her opinion. She further opined that while an administrative procedure might not satisfy the strict conditions of Article 6 ECHR when imposing substantial anti-trust fines, it might nonetheless be satisfactory for the purposes of the ECHR that all aspects of the result of such procedure could be reviewed by a body that complies with the strict adequacy conditions for criminal proceedings.6 The approach to the case taken by the Court of Justice was in line with AG Sharpston’s suggestion. The Court implicitly recognised that compliance with Article 6 ECHR and Article 47 CFR required the potential of a full judicial review at the request of the parties. On the surface this seems an uncontroversial finding, but it was at odds with a longstanding doctrine of deference, first established in the Consten and Grundig judgment of 1966. According to this doctrine, the Commission was considered to enjoy a level of discretion in establishing complex economic facts in competition proceedings, although the Tetra Laval judgment of 2005 had signalled a more conditional approach to that discretion.7 The deference doctrine was grounded on the following expression in Consten and Grundig,

4 The Commission split its initial proceeding against the copper tube cartel into three cases: Case COMP/E-1/38.069 (copper plumbing tubes); Case COMP/E-1/38.121 (fittings); and case COMP/E-1/38.240 (industrial tubes). See also the discussion on this litigation, its prelude, and a review of the relevant literature in ch 1. 5 The applicants were respectively KME Germany AG, KME France SAS, KME Italy SpA and Chalkor AE Expexergasias Metallon. The KME group appealed in the context of the Commission’s inquiry into the plumbing tubes cartel (Case C-389/10 P) and the industrial tubes cartel (Case C-272/09 P), but Chalkor appealed in the context of the plumbing tubes cartel (Case C-386/10 P). 6 Opinion of AG Sharpston in Case C-272/09 P KME Germany and Others v Commission, EU:C:2011:63 [67]. 7 See Case C-12/03 P Commission v Tetra Laval, EU:C:2005:87 [39].

The KME/Chalkor Cases  121 which was routinely cited in the subsequent case law up until the KME/Chalkor litigation: Furthermore, the exercise of the Commission’s powers necessarily implies complex evaluations on economic matters. A judicial review of these evaluations must take account of their nature by confining itself to an examination of the relevance of the facts and of the legal consequences which the Commission deduces therefrom.8

The Court’s earlier case law thus created an obstacle to upholding the constitutional legitimacy of the Union’s institutional architecture for penalising cartels, as measured against the constitutional standards of procedural fairness laid down in the ECHR and the CFR. The pleading of the applicants was partially designed to challenge this controversy. Unexpectedly, the Court solved the dilemma by simply acting as if the Commission never enjoyed the discretion implied by the long line of cases building on the Consten and Grundig deference doctrine: As regards the review of legality, the Court of Justice has held that whilst, in areas giving rise to complex economic assessments, the Commission has a margin of discretion with regard to economic matters, that does not mean that the Courts of the European Union must refrain from reviewing the Commission’s interpretation of information of an economic nature. Not only must those Courts establish, among other things, whether the evidence relied on is factually accurate, reliable and consistent but also whether that evidence contains all the information which must be taken into account in order to assess a complex situation and whether it is capable of substantiating the conclusions drawn from it.9

With the doctrine of deference out of the way, the Court could quickly fend off the challenge with a reference to the Treaties and Regulation 1/2003: The review provided for by the Treaties thus involves review by the Courts of the European Union of both the law and the facts, and means that they have the power to assess the evidence, to annul the contested decision and to alter the amount of a fine. The review of legality provided for under Article 263 TFEU, supplemented by the unlimited jurisdiction in respect of the amount of the fine, provided for under Article 31 of Regulation No 1/2003, is not therefore contrary to the requirements of the principle of effective judicial protection in Article 47 of the Charter.10

Sensing that the General Court had in the judgment under appeal assumed the validity of the deference doctrine, the Court made a brief detour in its reasoning to explain what its deputy court had really meant: It must be noted in that regard that although the General Court repeatedly referred to the ‘discretion’, the ‘substantial margin of discretion’ or the ‘wide discretion’ of the Commission, including in paragraphs 35 to 37, 92, 103, 115, 118, 129 and 141 of the

8 Joined Cases 56 and 58/64 Consten and Grundig v Commission, EU:C:1966:41, p 347. 9 Each of the three judgments contains a paragraph containing this text. See Case C-272/09 P [94]; Case C-386/10 P [54]; and Case C-389/10 P [121]. 10 Case C-272/09 P KME v Commission, EU:C:2011:810 [106].

122  Analysis of Three Fairness Dilemmas judgment under appeal, such references did not prevent the General Court from carrying out the full and unrestricted review, in law and in fact, required of it.11

Given that the deference doctrine of Consten and Grundig still existed in some capacity prior to the KME/Chalkor litigation, the judgment signalled a definite departure.12 Viewed in functional terms, this meant that the Court effectively amended the prevailing procedural system by reducing the deference doctrine to ensure compatibility with Article 47 CFR. The Court thus choose to modify the existing procedural regime under the cover of its interpretive powers, instead of finding it unconstitutional which would have referred the issue to the legislator.

A.  Analysis of the Cases The issue of balancing in KME/Chalkor reflected an argument on behalf of the accused to increase the procedural safeguards against a false finding of the Commission, especially in cases where a substantial fine had been imposed that amounted to a criminal sanction. The Commission did not engage with the argument in the abstract, but instead relied on a categorisation in Article 23(5) of Regulation 1/2003 EC that classifies such procedures as administrative rather than as criminal. As an alternative argument the Commission referred to ECoHR’s distinction between hardcore and softcore criminal proceedings, the latter of which did not require the same procedural safeguards at the first instance, given that at the later instance a full judicial review was available. The alternative argument concedes to the main challenge of the applicants, concerning the necessity of exposing decisions on substantial fines to a full judicial review, unrestrained by the deference doctrine concerning complex economic facts. Since a concurrent view could be reached on the legal aspect of the problem based on the Commissions alternative argument, the disagreement moved onto factual issues about the Commission’s alleged decisional discretion. The applicants maintained that the General Court had in its judgment explicitly acknowledged deference to the Commission’s finding. Disagreeing, the Commission maintained that the General Court had not granted such discretion and that the arguments of the applicants had simply been rejected through the General Court’s application of its unlimited jurisdiction to review the Commission’s decisions. The Court of 11 ibid [109]. Nazzini is critical of this approach of the Court: ‘Three arguments have been put forward to “save” the current system. The first is that, in certain cases, the Union courts do not in fact apply the “manifest error” test – although they say that they do – but carry out a comprehensive review of the evidence. This seems to be the approach adopted by the Court of Justice itself in KME, where the Court said that, although the General Court had set out a test of deferential review, it did in fact carry out a full review of the Commission decision. This approach is far from satisfactory.’ See Renato Nazzini, ‘Administrative enforcement, judicial review and fundamental rights in EU competition law: a comparative contextual-functionalist perspective’ (2012) 49 Common Market Law Review 971, 997. 12 Arguably, the Court’s retreat from the deference doctrine with regards to complex economic facts started earlier with a narrowing of its scope following Tetra Laval of 2005.

The KME/Chalkor Cases  123 Justice eventually sided with the Commission, effectively reducing the Consten and Grundig deference doctrine. By this, the Court tipped the balance of the equilibrium between procedural accuracy and procedural cost-efficiency, towards more accuracy. The new equilibrium now constitutes the judicially approved benchmark for procedural fairness with regards to this specific issue. Building on the assumption that the Commission was in reality granted with a level of deference in assessing certain facts prior to the finding in the KME/Chalkor cases, the elimination or the reduction of this deference implies a an additional enforcement costs for maintaining the same level of procedural accuracy or for increasing it. Firstly, the Commission now needs to spend time on arguing its decisions more thoroughly to withstand potential challenges on points that were previously non-challengeable. Secondly, the range of issues that can be challenged has been increased, which probabilistically translates into more frequent challenges and more extensive pleadings, both of which drain private and public enforcement resources. Thirdly, the discretion, which the Commission supposedly had related to an expertise that the courts did not possess in the past (ie the proficiency in fact-finding grounded in the theory of economics) must now also be acquired and maintained by the courts. The courts must be able to review confidently all aspects of the Commission’s findings with an output that is at least as accurate as that of the Commission. It seems apparent that in the absence of an increase of resources towards each of these additional costs, the overall accuracy of the procedure has a potential to decrease, or at the minimum fail to increase. Failure on behalf of the Commission to properly state reasons for its findings can result in false annulments by the courts. Increased case load on an underfunded enforcement mechanism is liable to increase errors. Absence of expertise at the courts is also likely to result in enforcement errors. The Court’s rulings in the KME/Chalkor cases were thus only probable to increase the accuracy of the enforcement process, if an increase in enforcement resources were to follow. The Court’s motivation for its finding was to adjust the enforcement process to the accuracy requirements of the Charter of Fundamental Rights, but if this was to succeed, it unavoidably imposed various costs on the system. It can be debated whether the Court was a well-placed decision-maker to take such an investment decision, or whether the legislator was better placed to efficiently rearrange the enforcement mechanism. Notably, the number of judges at the General Court has since been significantly increased by the legislator.13 The Court’s decision in KME/Chalkor rested on a distinction which was based on ECoHR’s finding in Jussila v Finland.14 The Grand Chamber of the ECoHR 13 On 3 December 2015 it was announced that the number of judges at the General Court would be increased by 21 between 2016 and 2019 and that the Civil Service Tribunal would, along with its seven judges, be merged with the General Court. See Council of the EU Press Release 886/15 of 3 December 2015, available at www.consilium.europa.eu/en/press/press-releases/2015/12/03/eucourt-of-justice-general-court-reform/pdf. 14 Jussila v Finland App no 73053/01, CE:ECHR:2006:1123JUD007305301 [43].

124  Analysis of Three Fairness Dilemmas had concluded that the difference between criminal and administrative proceedings was gradual, rather than absolute, and that in between the two there existed a category of proceedings that contained elements of criminal proceedings but not in the traditional hardcore sense. A distinction should thus be made between the two when assessing the applicability of the procedural guarantees of Article 6 ECHR and the level of protection granted for each class of criminal procedures.15 The Court’s ruling in KME/Chalkor thus moved the competition proceedings of the Commission from the pure administrative sphere, to the in-between sphere where some of the criminal procedural guarantees apply, but not with full stringency in terms of Article 6 ECHR.16 In the judgment, the ECoHR rejects a strictly deontological interpretation of the conditions of Article 6. The Court adopts a consequentialist approach by acknowledging that procedural guarantees are not absolute conditions, but rather indicators that should be interpreted in context with the protective interests at stake: Notwithstanding the consideration that a certain gravity attaches to criminal proceedings, which are concerned with the allocation of criminal responsibility and the imposition of a punitive and deterrent sanction, it is self-evident that there are criminal cases which do not carry any significant degree of stigma. There are clearly ‘criminal charges’ of differing weight.17

15 The Grand Chamber of the ECoHR was not unanimous on this distinction in Jussila v Finland. Judge Loucaides wrote a dissenting opinion which was joined by judges Zupancic and Spielmann: ‘I find it difficult, in the context of a fair trial, to distinguish, as the majority do in this case, between criminal offences belonging to the “hard core of criminal law” and others which fall outside that category. Where does one draw the line? In which category does one place those offences which on their face value do not appear severe, but if committed by a recidivist may lead to serious sanctions? I believe that the guarantees for a fair trial envisaged by Article 6 of the Convention apply to all criminal offences. Their application does not and cannot depend on whether the relevant offence is considered as being in “the hard core of the criminal law” or whether “it carries any significant stigma”. For the persons concerned, whom this provision of the Convention seeks to protect, all cases have their importance. No person accused of any criminal offence should be deprived of the possibility of examining witnesses against him or of any other of the safeguards attached to an oral hearing. Moreover, to accept such distinctions would open the way to abuse and arbitrariness. I firmly believe that judicial proceedings for the application of criminal law, in respect of any offence, by the omnipotent State against individuals require, more than any other judicial proceedings, strict compliance with the requirements of Article 6 of the Convention so as to protect the accused “against the administration of justice in secret with no public scrutiny.’ As rightly pointed out by Trechsel, ‘the principle of public trial in criminal cases has an importance which goes beyond personal interests’. Therefore, once it was found (correctly) that the relevant proceedings in this case were criminal, the requirement of a public hearing in respect of them became a sine qua non. The failure to fulfil that requirement amounts, in my opinion, to a breach of Article 6 of the Convention.’ 16 Andersson argues that the Court of Justice had, with regards to the principle of presumption of innocence, already in 1999 acknowledged the criminal nature of EU competition proceedings in the Hüls case (Case C-199/92 P Hüls AG v European Commission, EU:C:1999:358 [150]). See Helene Andersson, Dawn raids under challenge: due process aspects on the European Commission’s dawn raid practices (Oxford, Hart Publishing, 2018) 41–42. 17 Jussila v Finland App no 73053/01, CE:ECHR:2006:1123JUD007305301 [43].

The KME/Chalkor Cases  125 By taking this step the Court risks the cohesion of the system of protection under Article  6 ECHR, as was pointed out by judge Loucaides in his dissenting opinion: ‘[T]o accept such distinctions would open the way to abuse and arbitrariness.’ A consequentialist approach, in a system that relies on a deontological mode of arguing, invites confusion about the permitted types of arguments. Should the rights under the ECHR be argued for based on their deontological character, or can rights arguments be won or lost through a cost–benefit analysis? Jussila v Finland seems to suggest the later and the KME/Chalkor by analogy relies on such rationale. Despite the consequentialist character of the arguments for the main finding in Jussila v Finland, the typical tool for solving such balancing dilemmas in the rights jurisprudence, the proportionality test was not employed, at least not explicitly. When the Court of Justice embraced the argument in KME/Chalkor, it not only withheld from making an explicit reference to the underlying rationale of the Jussila finding, but also abstained from commenting on the nature of the distinction between a criminal and an administrative proceeding, which was the critical finding in Jussila. As a result, the motivation for the finding of compatibility with regards to Article 47 CFR was minimal, especially when considering that the Court had to slightly alter the system in place to fit it within the parameters of the procedural fairness criteria of the CFR and the ECHR. A proper motivation would have required a balancing exercise grounded on a consequentialist rationale. Without such an argument, the result lacks a connection to a plausible theory of procedural fairness. It is also important to note that the Court of Justice refrained from referring to Article 6 ECHR in its finding in KME/Chalkor; it referred exclusively to Article  47 CFR.18 This created an additional obstacle; it barred the Court from using the interpretations of the ECoHR as an anchor for its own finding. If the Court does not recognise the deontological force of the provisions of the ECHR and the interpretations of the ECoHR of these provisions, it becomes problematic for the Court of Justice to use these same legal sources as uncited inspirations for its own findings. In that case it would need to repeat in full the rationale on which the Court’s sources of inspiration rest. The Court of Justice did not support its finding in the KME/Chalkor judgments with elaborate arguments explaining the deontological character of the procedural guarantees of Article 47 CFR, nor did it refer to the jurisprudence under the ECHR.19 In the absence of any true argument about a theory of procedural fairness, it is challenging to identify the philosophical nature of the Court’s reasoning 18 Sibony discusses the reasons for this in her note on the case, and comments that the Court avoided the question of categorisation as criminal or not by refereeing exclusively to Art 47 CFR. See Anne-Lise Sibony, ‘Annotation of the judgment of the Court in Case C-272/09 P KME Germany and others v. Commission’ (2012) 49 Common Market Law Review 1977, 1989–95. 19 The CJEU later altered its approach in this regard and made an explicit reference to Art 6 ECHR and the ECoHR’s Menarini judgment. See Case C-501/11 P Schindler Holding and Others v Commission, EU:C:2013:522 [35]–[36].

126  Analysis of Three Fairness Dilemmas in KME/Chalkor. It is hard to know whether the judgment was the result of a careful quantitative balancing between the competing considerations of procedural accuracy and procedural efficiency, or whether it was the Court’s intuitive interpretation of the rights imperatives, implied by Article 47 CFR. Probably, it was a bit of both. The Court was clearly under the influence of the Jussila ruling, which recognised the gradual nature of the right to a fair proceeding, but the Court did not explicitly employ any quantitative tools to establish where the balance between the competing interests should be located. Instead, the Court concluded, presumably by intuition that a fair equilibrium is reached, if the traditional criminal procedural guarantees partially apply to competition proceedings. In terms of the analytical approach in the Jussila judgment, the essential balancing issue in the KME/Chalkor litigation can be extracted through a simple question: how strict procedural guarantees should be employed to competition proceedings, given the protective interest at stake? As noted in chapter three, the balancing of procedural fairness in the competition proceedings of the EU can be assessed at various balancing points in the procedure. The point in question in the KME/Chalkor cases related to the judicial review of the General Court of the Commission’s decision to impose a cartel fine. The point at issue, follows the three decision points in the Commission’s administrative procedure, but precedes the ultimate decision of the Court of Justice. The focus is thus at the fourth decision point and the question is how rigid it should be in terms of decisional accuracy.

B.  Subsequent Developments On 18 July 2013, the Court of Justice reinforced its conclusion from the KME/ Chalkor cases in a decision in case C-501/11 P Schindler Holding and Others v Commission.20 The applicant relied on a similar argument as the applicants had done in the KME/Chalkor cases. In its conclusion, the Court pointed to a development in the case law of the ECoHR, which reinforced its earlier approach: In paragraph 59 of its judgment in A. Menarini Diagnostics v. Italy, the European Court of Human Rights explained that, in administrative proceedings, the obligation to comply with Article 6 of the ECHR does not preclude a ‘penalty’ from being imposed by an administrative authority in the first instance. For this to be possible, however, decisions taken by administrative authorities which do not themselves satisfy the requirements laid down in Article 6(1) of the ECHR must be subject to subsequent review by a judicial body that has full jurisdiction. The characteristics of such a body include the power to quash in all respects, on questions of fact and law, the decision of the body below. The judicial body must in particular have jurisdiction to examine all questions of fact and law relevant to the dispute before it.



20 Case

C-501/11 P Schindler Holding and Others v Commission [2013] ECLI:EU:C:2013:522.

The KME/Chalkor Cases  127 Ruling on the principle of effective judicial protection, a general principle of European Union law to which expression is now given by Article 47 CFR and which corresponds, in European Union law, to Article 6(1) ECHR, the Court of Justice has held that, in addition to the review of legality provided for by the FEU Treaty, the European Union judicature has the unlimited jurisdiction which it is afforded by Article  31 of Regulation No 1/2003, in accordance with Article  261 TFEU, and which empowers it to substitute its own appraisal for the Commission’s and, consequently, to cancel, reduce or increase the fine or periodic penalty payment imposed (Chalkor v Commission, paragraph 63).21

Following the Menarini case in the ECoHR it was clear that the Commission could, in its capacity as an administrative body, impose substantial cartel fines of a criminal nature, as long as an independent and impartial tribunal, in terms of Article 6 ECHR, had the power to review all aspects of the fining decision.22 Following KME/Chalkor, the General Court was considered to have such unlimited jurisdiction to review the Commission’s cartel fining decisions based on Article 263 TFEU and Article 261 TFEU in conjunction with Article 31 of Regulation 1/2003. The Court did not concede explicitly that the Commission’s cartel fines were of a criminal nature; just short of such concession, it explained that even if they were criminal penalties, it would not in itself be a breach of Article 6 ECHR.23 In a judgment in case C-67/13 P CB v Commission,24 decided on 11 September 2014, the Court of Justice set the General Court’s judgment under appeal aside. The Court found that the General Court had failed to observe the standard of judicial review established in the KME/Chalkor case law by giving the Commission too much discretion with regards to the assessment of certain economic facts. In a critical tone, the Court found that the General Court had shown indications of ‘a general failure of analysis’ that ‘reveal[ed] the lack of a full and detailed examination of the arguments of the appellant and of the parties’.25 The case law on this issue has now settled along the lines of the KME/Chalkor cases of the Court of Justice and the Menarini case of the ECoHR.

C.  The Court’s Approach to the Fairness Issue In the KME/Chalkor cases the applicants argued that the limits on judicial review in competition cases, exemplified by the Commission’s discretion to establish and evaluate certain types of facts, was not compatible with the strict accuracy standards for procedures of such type articulated in Article 6 ECHR and Article 47 CFR. Although not argued by any of the parties, the Commission’s discretion could have been rationalised with reference to the cost and accuracy implications entailed

21 ibid

[35]–[36]. See also paras [33]–[34]. Menarini Diagnostics Srl v Italy App no 43509/08, CE:ECHR:2011:0927JUD004350908. 23 Case C-501/11 P Schindler Holding and Others v Commission, EU:C:2013:522 [33]–[34]. 24 Case C-67/13 P CB v Commission, EU:C:2014:2204. 25 Case C-67/13 P CB v Commission, EU:C:2014:2204 [89]. See also paras [42]–[46] and [90]–[92]. 22 A

128  Analysis of Three Fairness Dilemmas in allowing complex economic facts to be challenged before courts that do not possess expert personnel in economics. The Court’s eventual finding reduced the Commission’s discretion, allowing parties subject to a cartel fine to challenge the Commission’s findings on more factual grounds than before. The Court was silent about the logic, or the theory, behind its decision to limit the Commission’s discretion. Reading between the lines, it can be assumed that the Court considered the limitations to judicial review, derived from the Consten and Grundig case law, were unacceptable due to the obligations under Article 47 CFR. By taking that view, the Court was also stating that the standard of procedural accuracy would be inadequate if the Commission’s discretion with regards to the assessment of complex economic facts was upheld. By opening more grounds to appeal competition decisions of the Commission, the Court of Justice was effectively deciding that the jurisdiction of the courts was to be increased in comparison with the previous practice. A greater jurisdiction entails probabilities of increasing the volume and the complexity of litigation, which in turn requires the channelling of additional resources into taking and defending court actions, and to adjudicate an accurate result. The Court’s decision in KME/Chalkor is thus a suggestion that further resources should be used towards competition enforcement, not just by those willing to pay more like those with disproportionally high stakes in the result (eg the addressees of a prohibition decision of the Commission), but also the citizens of the EU through the funding of the Commission and the Luxembourg courts. In the absence of increased public resources, a greater jurisdiction of the Courts will enable private parties to increase litigation resources and thus exploit the high standard of proof in their favour, to the effect that the risk of false negative errors, ie false annulments of prohibition decisions, will increase. This raises questions about the Court’s methodology for reaching such a decision. The alternative approach would have been to find the enforcement regime unconstitutional and defer the issue of institutional adjustment or rearrangement to the legislator, which possesses institutional expertise in making such decisions and is subject to democratic accountability mechanism towards those who ultimately pay for the enforcement regime.

II. The Groupe Gascogne Cases On 26 November 2013, the Grand Chamber of the Court of Justice handed down three judgments in appeal cases relating to an earlier cartel decision of the Commission concerning the market for industrial bags.26 In each of the three cases – C-58/12 P Groupe Gascogne v Commission;27 C-50/12 P Kendrion

26 Commission 27 Case

Decision in Case COMP/38354 – Industrial Bags [2005]. C-58/12 P Groupe Gascogne v Commission, EU:C:2013:770.

The Groupe Gascogne Cases  129 v Commission;28 and C-40/12 P Gascogne Sack Deutschland v Commission29 – the applicants pleaded that an excessive delay in the General Court’s procedure had breached their procedural rights under Article 47 CFR. The procedure in each case had lasted for more than five years, and there was more than a three-year delay between the end of the written court procedure, and the opening of the oral court procedure. Without setting in stone the maximum permissible procedural time, the Court concluded that: The length of that period [ie between the written and the oral procedure] cannot be explained by the circumstances of the case, whether it be the complexity of the dispute, the conduct of the parties or the supervening procedural matters.30

Due to the obvious breach in the cases to the right to have a hearing within a reasonable time, as required by Article 47 CFR, the main issue became to resolve how such breaches should be dealt with. The Court started by noting, with reference to its earlier case law, that the judgments under appeal would not be set aside since the delay in the proceeding would not have changed the outcome of the cases.31 The question was thus how the applicants should be compensated for the delay, without altering the main result of the cartel case. The existing case law provided two alternatives. In case C-185/95 P Baustahlgewebe v Commission, the Court ‘for reasons of economy of procedure and in order to ensure an immediate and effective remedy’ decided in a comparable situation to reduce the imposed fine.32 In case C-385/07 P Der Grüne Punkt–Duales System Deutschland v Commission, which did not involve a fine, the Court had taken a different approach by suggesting that the applicants needed to file a claim for damages against the Community in a new separate case.33 The opinion of AG Sharpston in Groupe Gascogne v Commission contains information about an unusual juridical methodology used by the Court in reaching its conclusion. The following passage indicates how uncertain the Court was about which approach to take: The Court invited the 27 Member States, the European Parliament and the Council to indicate in writing their views on the approach taken in, respectively, Baustahlgewebe and Der Grüne Punkt. Seven Member States indicated a preference for the former, three favoured the latter and six Member States expressed no preference. The Council endorsed Baustahlgewebe whilst acknowledging that the two remedies coexist and neither is perfect. The European Parliament considered the Der Grüne Punkt approach to be better.34 28 Case C-50/12 P Kendrion v Commission, EU:C:2013:771. 29 Case C-40/12 P Gascogne Sack Deutschland v Commission, EU:C:2013:768. 30 ibid [98]. 31 Case C-58/12 P Groupe Gascogne v Commission, EU:C:2013:770 [74]–[75]. 32 Case C-185/95 P Baustahlgewebe v Commission, EU:C:1998:608 [48]. 33 Case C-385/07 P Der Grüne Punkt – Duales System Deutschland v Commission, EU:C:2009:456 [195]. 34 Opinion of AG Sharpston in Case C-58/12 P Groupe Gascogne v Commission, EU:C:2013:360 [119].

130  Analysis of Three Fairness Dilemmas In the end, the Court adopted the Der Grüne Punkt approach, thus rejecting the pleas of the applicants in the three cases and instructing them to seek damages in a new action before the General Court. The decisive reason was the universal character of Der Grüne Punkt approach: [A] claim for damages brought against the European Union pursuant to Article  268 TFEU and the second paragraph of Article 340 TFEU constitutes an effective remedy of general application for asserting and penalising such a breach, since such a claim can cover all the situations where a reasonable period of time has been exceeded in proceedings.35

In comparison, the approach in Baustahlgewebe is limited to instances where a fine has been imposed, and relies on the Court’s own statute as a legal basis, instead of articles of the Treaties. The argument about procedural economy of the Baustahlgewebe approach did thus not succeed. In the months and years following the Grand Chamber judgments in Groupe Gascogne and Gascogne Sack Deutschland, several other cases concerning lengthy procedure before the General Court were resolved using the same rationale.36 In one of these cases, C-580/12 P Guardian Industries and Guardian Europe v Commission decided on 12 November 2014, AG Wathelet in his opinion made a last effort to oppose the approach taken in Groupe Gascogne using the following argument: [T]he appropriate mechanism for remedying a breach by the General Court of the reasonable time principle in a case such as the present, would, for reasons of economy of procedure and also to ensure an immediate and effective remedy, be to reduce the fine rather than to leave it to the parties to bring an action for damages before the General Court which, necessarily, will have been found to have failed to observe that principle by being unable to deliver its judgment within a reasonable time.

This was, however, to no avail; the Court maintained its approach from the Gascogne Sack Deutschland case and has done so in many other cases since.37

A.  Analysis of the Cases The main fairness issue in the Groupe Gascogne cases was more explicitly argued by the parties in terms of the accuracy/efficiency dichotomy, than in the KME/ Chalkor cases. There was little doubt about the existence of undue delays, but the contested issue was how to deal with and compensate for such procedural faults.

35 Case C-58/12 P Groupe Gascogne v Commission, EU:C:2013:770 [82]. 36 See for example the judgments in: C-238/12 P FLSmidth v Commission, EU:C:2014:284; C-578/11 P Deltafina v Commission, EU:C:2014:1742; C-243/12 P FLS Plast v Commission, EU:C:2014:2006; C-580/12 P Guardian Industries and Guardian Europe v Commission, EU:C:2014:2363. Numerous recent cases also refer to the same rationale. 37 Case C-580/12 P Guardian Industries and Guardian Europe v Commission, EU:C:2014:2363 [17]–[21].

The Groupe Gascogne Cases  131 On one hand, it was argued that the Court of Justice should deal immediately with such claims and compensate for procedural faults by lowering the cartel fines that had been imposed in the cases. On the other hand, it was argued that the right approach was to initiate separate procedures for damages that would be dealt with by the General Court, which normally deals with questions of fact in competition cases, unlike the Court of Justice that primarily deals with questions of law. The applicants, in line with their interests, argued for the simpler and more immediate solution to this problem; they wanted the cartel fines to be reduced without having to initiate a separate case before the General Court. The AGs had different opinions on this issue; while AG Sharpston argued for a separate action for damages with reference to accuracy considerations, AG Wathelet argued for emphasis on procedural economy by lowering the fines immediately. The Court of Justice approached the problem more holistically and thus found that a separate action needed to be taken by the applicants to conform to the general procedure for claiming compensations for procedural irregularities. The underlying rationale of the key arguments at odds in the Groupe Gascogne case line is partly reviled in the opinions of the AGs. AG Sharpston at one point suggested that the ‘right to a fair trial within a reasonable time is a right that encompasses two key components, not one’. She then continued by contrasting these key components: ‘Cutting every last possible corner in a search for swifter case-handling would not be compatible with maintaining the overall fairness of the proceeding.’38 Sharpston identifies swiftness as being at odds with the concept of a fair proceeding, but it is not entirely clear which meaning she attaches to the normative objective of fairness. Her argument could be interpreted to mean that swiftness risks procedural errors that in turn undermine the fairness of the procedure; ie that fairness represents the accuracy of the procedure’s result. This reading is reinforced by Sharpston’s remarks about the responsibility of the Member States to provide the courts with sufficient resources to enable swift case-handling, and that ‘difficulties caused by case overload, however real they may be, should be disregarded when assessing whether there has been excessive delay in handling a particular case’.39 By this she seems to suggest that swift case handling is an entirely independent consideration that will not be excused with reference to limited availability of enforcement resources. Treating procedural swiftness and procedural fairness as separate deontic properties that both should be realised independently to perfection at the same time is highly problematic in a world of definitive resources. When forced to choose between an economic procedural route and a more accuracy oriented procedural route, Sharpston puts the priority on accuracy: I understand that it is tempting to opt for the “procedural economy” of reducing the fine …. I am not satisfied by the intellectual foundations of such approach … and it has, at worst, the potential to become almost entirely arbitrary.40

38 Opinion of AG Sharpston in Case C-58/12 P Groupe Gascogne v Commission, EU:C:2013:360 [87]. 39 ibid 40 ibid

[90]. [132].

132  Analysis of Three Fairness Dilemmas Swiftness should thus be secondary when at odds with procedural accuracy, or as already noted by Sharpston in the cited passage above, reduction of accuracy for the purposes of swift case-handling is not compatible with procedural fairness. Ironically, the intellectual foundations of Sharpston’s analysis are weak in these regards, if measured against the canons of procedural theory.41 AG Wathelet’s remarks in Guardian Industries v Commission indicate a different understanding of the relation between procedural fairness and the expediency of the enforcement process. Wathelet cites the ‘economy of procedure’ as a decisive reason for solving the problem of how to provide redress for undue procedural delays at the General Court. He further explains that ‘it would be paradoxical if the only way to obtain redress for excessively lengthy legal proceedings were to bring another legal action, which would necessarily entail additional costs (both for the parties and for the company) and further delay’.42 Although not explicitly argued, this kind of reasoning presupposes that the expediency of a procedure is a component of a fair procedure. Undue delays are thus liable to impair the procedure’s fairness. On this view, further delays to the procedure to obtain redress are unlikely to provide an efficient way to rectify a breach to the principle of procedural fairness, caused by a procedural delay. On Wathelet’s view, case-handling swiftness is an efficiency function integral to procedural fairness. Procedural delays produce costs for the parties, which in turn are unfair in the procedural sense, if not offset by other elements or components of procedural fairness, such as accuracy. In contrast to AG Sharpston’s deontological approach to procedural fairness, AG Wathelet’s approach is more consequential in character by approaching the question of procedural fairness as a cost–benefit question that is sensitive to the issue of efficient use of public and private resources. The Court of Justice, in its reasoning for the result, did not assess the costs or benefits of lowering cartel fines immediately in respond to undue delays in the procedure. The Court simply stated that the procedure provided for in Articles 340 and 268 TFEU provided a sufficiently effective remedy of general application to address the problem of procedural delays. The Court nonetheless took a view on the factual situation by declaring the delay at issue amounting to a breach of Article  47 CFR but abstained from quantifying the harm suffered by the applicants. The argument of the Court seems to weigh the benefit of having a single uniform procedure for claiming damages in the EU law system, against the cost of having specific procedures for specific types of damages claims. The argument is consequential rather than deontological, but instead of making the assessment of the costs and benefits of a single instance decisive, the Court concluded that the general procedure ‘constitutes an effective remedy’, implying that it is effective enough for the instance at issue, but without prejudice to how effective it could be 41 Chapter two provides an overview of the main views on procedural theory. 42 Opinion of AG Wathelet in Case C-580/12 P Guardian Industries and Guardian Europe v Commission EU:C:2014:272 [111]. See also [110].

The Groupe Gascogne Cases  133 using the competing approach. By shifting the weight of the consequential reasoning away from the specific, over to the general, the Court can reach a conclusion that implies that the inconvenience of starting a new damages procedure in this case should be outweighed by the convenience of having a single uniform universally applicable procedure for claiming damages caused by the actions of the EU institutions. Despite the apparent consequential character of the reasoning of the Court, there are no indications in the text of the judgment that any actual quantitative assessment was made of the advantage of having a uniform system for claiming damages, or whether there actually is any advantage to emphasise uniformity over a specifically applicable efficient procedural solution, such as the one argued for by the applicants and AG Wathelet in the context of the Groupe Gascogne cases and the Guardian Industries case. Gascogne Sack Deutschland GmbH and Gascogne SA later sought close to €4 million in material and non-material damages for the delay in a separate proceeding before the General Court, which was subsequently appealed to the Court of Justice. The case was largely unsuccessful and resulted in the award of a lump sum of €5,000 in non-material damages for each of the applicants.43 The different approaches to the procedural problem in the Groupe Gascogne cases reveal different methodologies to solving a problem that was predominantly legislative in character. AG Sharpston advocated a deontological methodology, where the accuracy consideration is above other normative considerations, but she failed to consider the reality of definite enforcement resources. AG Wathelet acknowledged the importance of procedural efficiency for the concept of procedural fairness. The Court also acknowledged the role of procedural efficiency but adopted a macro perspective. The balancing point affected by the problem in Groupe Gascogne relates to a delay in the court procedure before the General Court and how that delay should be alleviated to minimise the effects on the fairness of the whole procedure. The issue could only be brought up at the framing stage preceding the fifth and the final decision point, since the delay occurred after the conclusion of the framing stage before the fourth decision point before the General Court.

B.  The Court’s Approach to the Fairness Issue The issue in the Gascogne cases was focused on the applicant’s entitlement to a certain level of procedural efficiency, rather than procedural accuracy. While it was easy to establish that the procedure before the General Court had not complied with the standard of swiftness required by Article  47 CFR, the Court struggled 43 Case T-577/17 Gascogne Sack Deutschland and Gascogne v Union, EU:T:2017:1; Joined Cases C-138/17 P and C-146/17 P European Union v Gascogne Sack Deutschland and Gascogne, EU:C:2018:1013.

134  Analysis of Three Fairness Dilemmas with establishing how to best remedy such problems. Again, the applicants wanted an efficient remedy, preferably an immediate reduction of the fine imposed, and the General Court had agreed in the judgments under appeal. AG Sharpston opposed arguments based entirely on considerations of procedural economy: The right to a fair trial within a reasonable time is a right that encompasses two key components, not one. Cutting every last possible corner in a search for swifter case-handling would not be compatible with maintaining the overall fairness of the proceedings.44

In this, she acknowledges a link between swiftness and accuracy. If swiftness becomes a primary consideration, accuracy will suffer and thus the fairness of the procedure. Expanding on this point, she continued: ‘To the extent that the available resources of the General Court are inadequate to deal appropriately with the present and reasonably to be expected future case load, the responsibility must lie with the Member States.’45 By this, she seems to be taking the view that courts should not be overly concerned about producing timely results, procedural swiftness is an economic issue that should be deferred to the Member States. Courts should primarily focus on producing accurate results. By dismissing the economic rationale, AG Sharpston could conclude that the Baustahlgewebe approach was inferior to the Der Grüne Punkt approach, due to reasons of potential inaccuracy: I understand that it is tempting to opt for the ‘procedural economy’ of reducing the fine (by some unspecified percentage, calculated on some unknown basis) in the context of an appeal. I am not satisfied by the intellectual foundations of such an approach (relationship between fine and conduct; jurisdiction; transparency) and it has, at worst, the potential to become almost entirely arbitrary.46

In its judgment, the Court of Justice substantively reached the same conclusion as suggested by AG Sharpston, but based on a different argument. Instead of dismissing economic efficiency considerations, the Court embraced them by refereeing to the efficiency advantage of having a uniform system for compensating for procedural delays: Admittedly, the present case concerns a similar situation to that giving rise to the judgment in Baustahlgewebe v Commission. However, a claim for damages brought against the European Union pursuant to Article  268 TFEU and the second paragraph of Article 340 TFEU constitutes an effective remedy of general application for asserting and penalising such a breach, since such a claim can cover all the situations where a reasonable period of time has been exceeded in proceedings.47



44 Opinion of AG Sharpston in Case C-58/12 P Groupe Gascogne v Commission, EU:C:2013:360 [87]. 45 ibid

[88]. [132]. On this point see also [127], [129] and [131]. 47 Case C-58/12 P Groupe Gascogne v Commission, EU:C:2013:770 [82]. 46 ibid

Collective Redress for Private Enforcement of Competition Law  135 Not convinced by this economic reasoning, AG Wathelet persisted to no avail in his subsequent opinion in the Guardian Industries case by restating the economic argument that had lost in Gascogne: Indeed, it would be paradoxical if the only way to obtain redress for excessively lengthy legal proceedings were to bring another legal action, which would necessarily entail additional costs (both for the parties and for the company) and further delay.48

The opinions of the AGs and the judgment of the Court of Justice display how methodologically contentious the equilibrium between procedural accuracy and procedural efficiency can be. AG Sharpston wants to emphasise accuracy, but AG Wathelet economic efficiency. The Court gave priority to efficiency, but in a macro procedural economic sense that lead him to the same conclusion as AG Sharpston who had argued for accuracy, and to the opposite conclusion of AG Wathelet who subsequently argued for procedural economy in a more specific micro economic sense. This reveals three categorically different methodologies for solving an apparently simple problem of picking an optimal procedure for compensation for procedural delays. The differing lines of arguments that are revealed in the opinions of the AGs and in the judgment show a complete lack of consensus about the proper grounds for arguing about how a procedural rule should be designed with regards to the issue of fairness. It is also noteworthy that the Court of Justice asked for, and presumably analysed, the preferences of the Member States and several other stakeholders towards the two competing procedural alternatives. That suggests that the Court’s ultimate judgment might be sensitive towards preferences of stakeholders, as is normal with the politicised legislative process, but which is usually not a permissible argument in support of adjudicative decisions.

III.  Collective Redress for Private Enforcement of Competition Law Private enforcement of competition law has been on the agenda in EU law since the early 2000s. The Court of Justice acknowledged in the Crehan case of 2001 the entitlement of a private party to claim damages for harm suffered through another party’s breach of EU competition law.49 In the Ashurst report from 2004, the status of private enforcement of competition law in the Member States was found be in a state ‘of astonishing diversity and total underdevelopment’.50 The Commission 48 Opinion of AG Wathelet in Case C-580/12 P Guardian Industries and Guardian Europe v Commission, EU:C:2014:272 [111]. See also [110]. 49 Case C-453/99 Courage Ltd v Bernard Crehan and Bernard Crehan v Courage Ltd and Others, EU:C:2001:465 [26]–[29]. 50 Denis Waelbroeck, Donald Slater and Gil Even-Shoshan, ‘Study on the conditions of claims for damages in case of infringement of EC competition rules’ (Ashurts, 2004) 1, available at https:// ec.europa.eu/competition/antitrust/actionsdamages/comparative_report_clean_en.pdf.

136  Analysis of Three Fairness Dilemmas issued a Green Paper on the issue in 2005, a White Paper in 2008, and eventually the Parliament and the Council adopted a directive on private damages (the Damages Directive) in 2014.51 Among the central issues in relation to private enforcement of competition law is the issue of collective redress. Often the harm suffered from competition law breaches is spread over many small companies or individual consumers that lack the incentive or the resources to seek redress for their harm. In many cases, the individual harm suffered is much smaller than the expected cost of litigating. The concept of collective redress addresses this difficulty by enabling the aggregation of many identical claims into a single action, thus permitting claimants to share the litigation costs and simultaneously saving time and resources for the defendant and the judicial apparatus. There are, however, different ways to design the enforcement procedures with regards to this enablement. Different approaches can make it more, or less, difficult to seek redress collectively, depending on how high the regulator wants the threshold to be. With the Green Paper, the Commission asked if special procedures should be made available for collective redress and how they should be framed.52 Following a public consultation of the Green Paper, the White Paper suggested two complementary measures. On one hand, the possibility for qualified entities, such as consumer associations or other similar entities, to bring representative actions on behalf of identified or identifiable victims. On the other hand, the possibility for opt-in collective actions, where different claimants could decide to bring a single action jointly.53 Both of these measures were suggested following a balancing exercise where the ‘objective of improving access to justice’ for the victims of competition law infringements was to be achieved, while ‘avoiding imposing unnecessary costs on potential defendants and the judicial system’.54 The Commission thus identified a balancing dilemma in the procedural design. The need to enforce the competition law provisions as accurately as possible by enabling those harmed to seek redress effectively, weighted against the potential of creating social costs by the risk of enforcement errors associated with the enablement of collective redress. The private class action in US anti-trust enforcement was the main source of inspiration for the potential European counterpart. The US and EU legal cultures are different, however, which has made EU regulators wary of following certain aspects of the US model.55 There is a worry among the business community in 51 ‘Damages actions for breach of the EC antitrust rules’, Green Paper, COM(2005) 672; ‘Damages Actions for Breach of the EC antitrust rules’, White Paper, COM(2008) 165; 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 (damages directive) [2014] OJ L 349/1. 52 Question H in the Green Paper. 53 Section 2.1 of the White Paper. 54 Commission staff working paper accompanying the White Paper on damages actions for breach of the EC antitrust rules (SWP) [2008] SEC/2008/0404 final, para 43. 55 In the White Paper it was emphasised that the Commission’s policy proposals were genuinely European, ‘rooted in European legal culture and traditions’. See White Paper, 3.

Collective Redress for Private Enforcement of Competition Law  137 Europe of exposing companies to a wave of meritless litigations, which increases the cost and risk of doing business.56 These same aspects have made the US model an effective enforcement tool and in recent years some of the flaws have been ironed out by increased judicial scrutiny during the early stages of litigation.57 The main features of the US system that were not part of the White Paper recommendation of the Commission were: an opt-out mechanism, which makes it much easier to collect participants in a class action than an opt-in mechanism; treble damages, which increases the pay-out for successful claims; and contingency fees, which makes it easier to fund the litigation of collective actions. All of those measures facilitate collective redress, but may at the same time result in more risk of enforcement errors in the form of meritless litigations and forced settlement of baseless claims.58

A.  Analysis of the Legislative Dilemma With regards to the issue of collective redress, the dilemma of the procedural regulator fits within the accuracy/cost-efficiency paradigm. The EU could increase the accuracy of its procedural regime for enforcing competition law by giving a large class of victims the tools to seek redress for their harm. However, there would probably be additional error costs and enforcement costs involved. The task of the regulator was thus to find an optimal solution to the balancing of the positive effects of increasing enforcement accuracy, with the negative effects of increasing error and enforcements costs. While analysing this dilemma, the regulator would need to make a distinction between two kinds of potential claims made by those who will potentially become the object of a claim by a collective redress mechanism. On one hand, it is likely that potential subjects will seek to limit such claims altogether, irrespective of their true guilt or actual harm suffered by the claimants. On the other hand, more sophisticated subjects will acknowledge the value for businesses in upholding a stringent regime for enforcing competition law infringements and thus limit their arguments to the error cost and the additional enforcement costs. Only arguments made from the latter perspective should qualify as valid from

56 See for example discussion in Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions, ‘Towards a European Horizontal Framework for Collective Redress’, COM/2013/0401 final, 8. 57 For a comprehensive overview of the US class action model see Arianna Andreangeli, ‘Collective redress in EU competition law: an open question with many possible solutions (2012) 35 World Competition 529, 529–58; Damien Geradin, ‘Collective redress for antitrust damages in the European Union: is this reality now?’ (2015) 22 George Mason Law Review 1079, 1087–91; Spencer Weber Waller and Olivia Popal. ‘The fall and rise of the antitrust class action’ (2016) 39 World Competition 29, 29–40. 58 Waller and Popal consider the opt-out mechanism the single most important feature of a successful class action regime. They consider that ‘at a conceptual level, opt-ins are not class actions; they are merely a form of joinder’, see Waller and Popal. ‘The fall and rise’ (2016) 48–51.

138  Analysis of Three Fairness Dilemmas the regulator’s perspective. Arguments about procedural design that disregard the substantive objective pursued by the procedural design are invalid, since the procedure predominantly aims at facilitating this very objective.59 In the case of competition law, collective redress should be viewed as a means towards achieving the substantive ends of the competition law regime, which in a general sense is to achieve consumer welfare by enforcing the competition law provisions of the TFEU. Arguments about collective redress that manifestly undermine the general purpose of competition law should thus be disregarded as invalid by the procedural designer. Views of some of the stakeholders towards the issue of collective redress in EU competition law are revealed in the numerous submissions submitted during the two rounds of public consolation prior to the adoption of the Damages Directive. As mentioned above, the Commission explicitly sought comments on the idea during the drafting process of the directive by designating a question on the issue in the Green Paper and by suggesting two distinct mechanism in the White Paper. It can thus be assumed that the Commission was seriously considering whether to make a collective redress mechanism a part of the instruments required to be harmonised and adopted into national legislations based on the directive. The views of associations of industry and businesses on the issue of collective redress as introduced in the Green Paper were generally negative. A recurring theme in the comments was that there was no need at all for such a mechanism and that it would impose costs on businesses, enrich the intermediaries handling the collective actions, and in any case would not succeed in providing consumers with a comprehensive remedy for claiming compensation for their harm. Often the US class action system was mentioned as an example that should be avoided.60 The views of consumer representatives were expressed through the submission of the European Consumer Organisation (BEUC), which represents over 40 national consumer organisations in EU and EEA states. BEUC expressed support for the initiative to establish ‘some sort of collective action on behalf of consumers’. Furthermore, it noted that many of BEUC’s members had ‘underlined the limitations of an opt-in system’; they would thus ‘favour an opt-out system’.61 After the Commission had issued concrete proposals for a collective redress mechanism in the White Paper, a second round of public consultation followed.

59 See further discussion about this conceptual point in ch 2. 60 Reference is made to comments made during the public consultation of the Green Paper on damages actions for breach of the EC anti-trust rules, in particular comments made by the following parties: The Association Francaise des Entreprises Privées (AFEP); Federation of German Industries (BDI); The Central Chamber of Commerce of Finland; Confederation of Finnish Industries (EK); International Chamber of Commerce (ICC); UNICE. See https://ec.europa.eu/competition/antitrust/ actionsdamages/green_paper_comments.html. 61 BEUC, ‘Damages actions for breach of EC antitrust rules: BEUC position on the Commission’s Green paper’, BEUC/X/026/2006, 21 April 2006, available at https://ec.europa.eu/competition/antitrust/ actionsdamages/files_green_paper_comments/beuc.pdf.

Collective Redress for Private Enforcement of Competition Law  139 Associations of industry and businesses remained sceptical of the concept of collective redress. In their comments, most expressed satisfaction with the fact that the Commission suggested an opt-in system, instead of an opt-out system, which would reduce the number of potential claimants. There was however a worry concerning the definition of eligible claimants in the White Paper. The Commission suggested that representative actions could be brought on behalf of identified and identifiable claimants. The latter category was considered in many of the submissions to amount to an opt-out regime and was thus strongly opposed. Few of the submissions still rejected the prospect of a collective redress in any form and many stressed that the issue should be dealt with horizontally over more policy areas than just the anti-trust context.62 The views of consumer representatives, as expressed through the submission of BEUC, were positive with regards to the general idea of providing mechanisms for collective redress for those harmed by breaches to the competition rules. In the submission it is however stressed that in order to achieve the objective of compensating for the full damage caused by the relevant breach, the default consideration should be the total group of concerned consumers and that representative actions should be allowed on their behalf. The issue of the distribution of the damages to the concerned consumers should be dealt with separately later. This approach would allow victims to come forward and claim compensation once the actual damages had been obtained through a representative action by consumer associations or other representative entities. BEUC thus still advocated for an opt-out approach to representative actions.63 The Commission also consulted with the European Consumer Consultative Group (ECCG), which was set up by the Commission to represent the views of consumers on the conception and the implementation of policies that concern the interests of consumers. In ECCG’s opinion, the preference of consumer representatives for an opt-out system was again highlighted. Such a mechanism would most likely ‘lead to a greater number of consumers receiving compensation, either directly or indirectly’. This was considered important within the anti-trust context since often the consumers are ‘unaware that their rights have been infringed or lack the necessary information to prove their losses’. In ECCG’s opinion, a study by Professor Rachael Mulheron was cited which shows that participation of consumers in a recent opt-in procedure was less than one per cent, while recent

62 Reference is made to comments made during the public consultation of the Green Paper on damages actions for breach of the EC antitrust rules, in particular comments made by the following parties: The Association Francaise des Entreprises Privées (AFEP); American Chamber of Commerce to the European Union; Federation of German Industries (BDI); The Confederation of European Businesses; CBI; Confederation of Finnish Industries (EK); Confederation of Swedish Enterprise; The German Association of Chambers of Industry and Commerce (DIHK); The Federation of Private Enterprises (Företagarna); International Chamber of Commerce (ICC); The Hellenic Federation of Enterprises (SEV). See https://ec.europa.eu/competition/antitrust/actionsdamages/white_paper_comments.html. 63 BEUC, ‘Damages actions for breach of EC antitrust rules’ (2008).

140  Analysis of Three Fairness Dilemmas experiences from the Netherlands and Portugal showed 97%–100 per cent participation in an opt-out procedure.64 As reflected in the comments submitted during the public consultation, the European regulator was pressed in opposite directions by the main stakeholders. On one hand, representatives of businesses resisted efforts to establish collective redress mechanisms and, if such mechanism were to be established, they should be as cumbersome as possible to operate. On the other hand, consumer organisations welcomed the prospect of a more effective redress mechanism for consumers and wanted the threshold for initiating claims through such mechanism to be as low as possible. The dilemma of the legislator was thus to identify the optimal action with regards to the legislative task, taking note of the arguments put forward by the stakeholders.

B.  The Commission’s Approach to the Fairness Issue Almost a decade after the Ashurts report had found the state of private enforcement of competition law in the EU to be ‘astonishingly diverse’ and ‘totally underdeveloped’, the Damages Directive was finally adopted in 2014. However, the initially bold plans of the Commission expressed in the Green Paper and the White Paper for a collective redress mechanism did not materialise as a part of the binding legal instruments of the directive. Instead the Commission issued a horizontal policy recommendation about common principles for collective redress, with a view to encouraging, instead of requiring, individual Member States to provide such mechanism within their respective jurisdictions.65 The recommendation proposed an even stricter approach to the standing in a collective action than was proposed by the Commission’s White Paper. The action should be launched by a claimant party that ‘should be formed on the basis of express consent of the natural or legal persons claiming to have been harmed’ and any exception to the opt-in principle ‘should be duly justified by reasons of sound administration of justice’. The issue about the ‘identifiable’ claimants had thus been resolved by further narrowing the class of potential claimants to those strictly identified as having opted in by the time of the conclusion of the court proceeding. This approach by the Commission to the dilemma of how the interests of consumers and businesses should be calibrated with regards to the effectiveness of collective redress for harm caused by competition law breaches is interesting when

64 ECCG, ‘European consumer consultative group opinion on private damages actions’, 23 November 2010, 4, available at https://ec.europa.eu/consumers/archive/empowerment/docs/ECCG_opinion_on_ actions_for_damages_18112010.pdf. A reference to Mulheron’s study is in fn 6 of ECCG’s Opinion. 65 Commission Recommendation of 11 June 2013 on common principles for injunctive and compensatory collective redress mechanisms in the Member States concerning violations of rights granted under Union Law [2013] OJ L 201.

Collective Redress for Private Enforcement of Competition Law  141 considered in the light of the task at hand. The regulatory agenda of the Damages Directive was and is very clearly to further facilitate the effective enforcement of Articles 101 and 102 TFEU. For example, in paragraph three of the preamble to the directive it is stated that the full effectiveness of Articles 101 and 102 TFEU, and in particular the practical effect of the prohibitions laid down therein, requires that anyone … can claim compensation before national courts for the harm caused to them by an infringement of those provisions.

The leitmotif of the regulatory efforts of the Commission was thus explicitly acknowledged as being to enhance the procedural mechanism for enforcing the prohibitions integral to the competition provisions of the TFEU. Considering this, the eventual approach to the collective redress mechanism seems odd and not in compliance with the overall regulatory objective.66 A clue to the rationale of the Commission in its approach to the collective redress mechanism suggested in the White Paper can be found in the accompanying Commission staff working paper (SWP). In its discussion about collective redress it starts by acknowledging that ‘competition law is a field where collective redress mechanisms can significantly enhance the victims’ ability to obtain compensation and thus access to justice, and contribute to the overall efficiency in the administration of justice’.67 The risks associated with a US style of class actions are noted, but at the same time it is highlighted that ‘specific features may be designed in order to serve as effective safeguards against misuse of the system’. Such measures could ensure that the mechanisms achieved ‘the objective of improving access to justice and guaranteeing compensation, whilst avoiding imposing unnecessary costs on potential defendants and the judicial system’.68 The tone here suggests that although there may be drawbacks to collective redress mechanism for enforcing competition law, there are nonetheless ways to balance the error costs and the expected gain to enforcement through a careful regulatory design. A key argument made in the SWP addresses the Commission’s decision to suggest an opt-in design, instead of an opt-out design. The argument starts by making a strong case against the use of an opt-in regime: An opt-in collective action system would usually result in a smaller number of victims claiming damages than in an opt-out system, thereby limiting corrective justice, and would have as a consequence that some of the illicit gain may be retained by the infringers, thereby limiting the deterrent effect of the mechanism. By requiring the

66 Juska attributes the failure of the Commission’s efforts to implement a specific collective redress mechanism for the anti-trust context to an opposition by the member states, who favoured a horizontal approach to collective redress. See: Zygimantas Juska, The role of collective redress actions to achieve full compensation for violations of European Union competition law, PhD thesis (Leiden University, 2019) 118, available at http://hdl.handle.net/1887/71940. 67 Commission staff working paper accompanying the White Paper on damages actions for breach of the EC antitrust rules (SWP) [2008] SEC/2008/0404 final, para 40. 68 ibid para 43.

142  Analysis of Three Fairness Dilemmas identification of the claimants (and the specification of their alleged harm suffered), an opt-in collective action may also render the litigation in some way more complex since it increases the defendant(s) possibility to dispute each victim’s harm.69

Immediately following this argument, the Commission states that this kind of mechanism should nonetheless be selected for the competition law contexts, since the opt-out system could lead to litigation excesses, as has happened in other jurisdictions. In particular, there would be ‘an increased risk that the claimants lose control of the proceedings and that the agent seeks his own interests in pursuing the claim’.70 These arguments of the Commission, supporting their case for suggesting an opt-in mechanism instead of an opt-out mechanism, are unconvincing. By the structure and the logic of the argument, the natural conclusion of it would suggest the rejection of the opt-in system. It is as if there is a missing reason for why the opt-out system was not favoured instead.71 Putting the pieces together, it can be speculated that the lobbying argument of business representatives simply won the day against the lobbying argument of the consumer representatives. The argument was not won based on its logical robustness, but instead based on the relative political weight of the lobbying parties. The arguments of businesses were given more relative weight than the arguments of consumer representatives, irrespective of actual effect of the different mechanism towards the regulatory task at hand. The collapse of the consumer cause for an opt-out mechanism for claiming compensations for competition law breaches was subsequently completed by the Commission’s move towards a horizontal approach to collective redress. By then the regulatory rationale had changed, which altered the structure of the argument for or against an opt-in mechanism. The argument was no longer about efficient and effective enforcement of Articles 101 and 102 TFEU, but instead about providing effective means of compensating consumers for harm caused by various activities of businesses in different regulatory fields. The Commission highlighted this substitute task in its communication accompanying the recommendation on collective redress by explaining that ‘an opt-out system may not be consistent with the central aim of collective redress, which is to obtain compensation for harm suffered, since such persons are not identified, and so the award will not be distributed to them’.72 This argument of the Commission would have been inapplicable where the central aim of collective redress was the effective enforcement 69 ibid para 58. 70 ibid para 58. 71 Accompanying the White Paper there was an impact assessment report. In this report an opt-out and an opt-in mechanism were compared, which resulted in a more favourable result for the opt-in mechanism. However, the framing of the opt-out mechanism exaggerated the drawbacks of that approach by not accounting for simple measures that could minimise such negative effects. The actual impact assessment of the alternative options was also rather rudimentary. See Commission staff working document accompanying document to the White Paper on damages actions for breach of the EC antitrust rules: ‘Impact assessment’ [2008] SEC/2008/0405 final, 29–30, 38–43 and 52–53. 72 COM/2013/0401 final, 12.

Collective Redress for Private Enforcement of Competition Law  143 of Articles 101 and 102 TFEU, as envisaged during the preparatory stages of the Damages Directive.73 This case study shows how the EU regulator approached a balancing issue in the design of procedures for enforcing the competition provisions of the TFEU. Collective redress was initially identified by the Commission as an effective mechanism for facilitating the task at hand, but during the legislative process the mechanism was watered down and eventually abandoned for most practical purposes. The legislator was aware of the accuracy/efficiency balancing issue, but it is unclear based on which arguments or methodology it eventually decided against a sector-specific mechanism for collective redress for enforcing competition law. It is however revealed that this result was lobbied for on behalf of businesses, while consumer organisations consistently emphasised the importance of such mechanism for enforcing competition law. Notably, the balancing dilemma faced by the regulator in the design of the Damages Directive was explicitly acknowledged as being a procedural fairness dilemma. In paragraph four of the Preamble to the Directive it is emphasised that the right to effective remedies for claiming compensation for infringements of Union competition law is derived from Article 19(1) TEU and from Article 47(1) CFR. The legislator thus had an ample constitutional support for a robust collective redress mechanism, had he chosen to tip the fairness balance of the procedural design further towards the lobbying arguments of the consumer representatives. There are not, however, any indications in the publicly available documentation from the legislative process that the legislator explicitly framed the question about how the collective redress issue should be dealt with, by asking how the different approaches measured against the constitutional requirements for procedural fairness articulated in Articles 19 TEU and Article 47 CFR.74 Clearly, the result of the Damages Directive is less compatible with these constitutional requirements from the perspective of consumers than the initial idea of including a robust collective redress mechanism as an additional tool for private enforcement of EU competition law. It is also not obvious that any of the concerns put forward by the representatives of businesses truly justified, in a rational sense, the scrapping of the idea of a sector-specific collective redress mechanism. Undoubtedly, the result reflected an outcome of an extensive 10-year balancing exercise of the administrative and political processes, which dominate

73 ie where the primary aim is effective enforcement, it becomes a secondary issue to distribute the award so long as the primary aim is achieved. The Commission turned this rationale on its head in the Communication accompanying the Recommendation. 74 Accompanying the Commission’s proposal for the Damages Directive was an impact assessment report that clearly associated the impact of the proposal with the principle of effectiveness as articulated through Art 19 TEU and Art 47 CFR. By this time, the opt-out collective redress mechanism favoured by consumer representatives was no longer on the table and thus the White Paper opt-in proposal and the non-binding horizontal approach became the main alternative options. See: ‘Commission staff working document – Impact assessment report – Damages actions for breach of the EU antitrust rules’ (SWD) (2013) 203 final.

144  Analysis of Three Fairness Dilemmas regulatory decisions of this type. Whether the ultimate decision accurately reflects a careful calibration of the underlying variables of social preferences, as envisioned by the social contract ideal, is however questionable.

IV.  Symptoms of an Underdeveloped Theory of Procedural Fairness The case studies show a lack of consensus about how the concept of procedural fairness should be understood and applied in the context of EU competition law. This was also reflected in the views of academics explored in the literature review in chapter one. The court cases were picked due to the difficult balancing dilemma the Court of Justice was confronted with in each of them. The difficulty of the dilemma rested on a lack of clear legislative guidance on how it should be solved. The leading theories of how courts approach such dilemmas, of undecided or vaguely decided issues from a legislative point of view, suggest that they either seek interpretive answers with the help of alternative legal sources (Dworkin), or they fill in the gaps of the open textured law (Hart).75 As the case studies show, the difficulty is fundamental and can hardly be resolved with reference to the leading theories. Taking the former view, the problem becomes a lack of consensus about their content of the alternative legal sources. There is even disagreement about what fairness means. Taking the latter view, the gap in the law should be filled with reference to a methodology (ie rule of recognition) which is habitually followed. However, this solution collapses when there is a deep disagreement about the methodological premises for filling the gaps. The case studies also show that the actual EU legislator does not fare much better than the Court of Justice when faced with the regulatory task of improving the procedural fairness situation for private victims of competition law breaches. The significant finding of the case studies with regards to the judgments in the KME/Chalkor case line and in the Gascogne case line, is how underdeveloped the Court’s theory of procedural rights is. In the former case, the equilibrium of accuracy and economic efficiency was nudged towards accuracy without a proper motivation. In the latter case, the more accurate route of two choices was also chosen, but on the grounds of an underdeveloped economic argument.

75 Hart described the open texture of law and language: ‘uncertainty at the borderline is the price to be paid for the use of general classifying terms in any form of communication concerning matters of fact.’ HLA Hart, The concept of law, 3rd edn (Oxford, Clarendon Press, 2012) 128. Dworkin argued for the right answer thesis, meaning that vague concepts should be interpreted in context to extract the right answer with regards to an issue at hand. A summary of his position can be found in Ronald Dworkin, Justice for hedgehogs (Cambridge MA, Belknap Press, 2011) 123–88, and a more specific argument in Ronald Dworkin, ‘No right answer?’ (1978) 53 New York University Law Review 1. For a critique of the Dworkinian view see for example Ólafur Páll Jónsson, ‘Vagueness, interpretation, and the law’ (2009) 15 Legal Theory 193.

Symptoms of an Underdeveloped Theory of Procedural Fairness  145 These two case clusters provide a good example of how an actual decisionmaker has approached two different issues of procedural fairness in which the law as it stood was too vague to provide a definitive answer. In the absence of a definite ex-ante answer, the possibility of providing an ex-post answer based on a previous ex-ante legislative decision is removed. The decision-maker is thus forced into the ex-ante mode of deciding, which abides to a different rationale than the ex-post mode of deciding. Facing an ex-ante decision, the Court showed signs of inconsistency in its methodology for approaching such questions, which resulted in a de facto rule-making that was poorly motivated in terms of rights theory and in terms of the theory of economics. The opinions of the AGs showed similar signs of confusion about which methodology ought to be used for determining the content of a procedural rule in the context of competition proceedings, where the existing law does not provide a definite ex-ante answer. The third case study shows how an actual legislative decision-maker approached a procedural fairness issue in the design of the Damages Directive. During the legislative process, strong opposing lobbying views emerged with regards to a proposed collective redress mechanism for claiming private damages for harm caused by competition law infringements. In the end the legislator took a decision, but this decision seemed not entirely consistent with the stated legislative task and the apparent relative strength of the key arguments wielded by the stakeholders during the process. The case studies further show disagreement about whether the concept of procedural fairness should be understood as a deontological concept or a consequential concept. They show different opinions about the true normative properties of procedural fairness and about whether the concept should be approached from a macro or micro perspective. All of these issues are fundamental theory issues that need to be settled before individual procedural issues can be resolved.76 The main task of the preceding chapters of this book has been to conceptualise procedural fairness in order to provide an uncontroversial starting point for discussion about individual dilemmas of procedural fairness in the design of EU competition law procedures. In the final chapter I will approach the dilemmas identified through the case studies in this chapter based on this conceptualisation, and suggest solutions that are compatible with a robust theory of procedural fairness. I will also display how the model of fair rules could be applied to orderly solve the fairness dilemmas in a consistent and coherent way based on an ex-ante decision methodology.

76 The disagreements revealed through the case studies about the concept of procedural fairness extend beyond normal vagueness in the language of the law, as described by Hart and other authors (see for example Timothy Endicott, ‘Law is necessarily vague’ (2001) 7 Legal Theory 379).

7 An Argument for a Data-based Democracy Chapter one set out the fundamental concern of this book as follows: how do we know if a procedural rule is fair? This question was subsequently placed in the context of certain dilemmas of procedural fairness in EU competition law. The main finding of the book is that a useful understanding of the concept of procedural fairness, both generally and in the special context of EU competition law, is to perceive it as an issue of distributional fairness. The concept of procedural fairness has certain normative elements that need to be equitably balanced through a procedure’s instrumental structure in order to fulfil the desiderata of fairness. It is thus possible to ascertain whether a procedural rule, or a whole procedural regime, is fair by examining whether the underlying normative elements of the rule are calibrated at an equilibrium that is perceived as equitable by the stakeholders of the rule. Based on this conceptualisation of procedural fairness it is possible to build a data-driven methodology that can be computerised for quickly and accurately approximating the procedural fairness of competing procedural designs. This concluding chapter starts by summarising the main findings of the book. Next, the procedural fairness dilemmas identified in chapter six will be measured against the methodology sketched out for solving such dilemmas in chapter five. Finally, it concludes with reflections on the more general meaning and implications of the book’s main findings on the practice of democratic rulemaking.

I.  Conceptualisation of Procedural Fairness in EU Competition Law Chapter one explored an ongoing debate in academic circles about the ideal design of the EU’s regime for enforcing competition law with regard to ideas about procedural fairness. Practitioners of EU competition law have shown keen interest in this debate and some of the important contributions to the literature are from their ranks. The issue was at the centre of high-profile litigations in the KME/Chalkor cases before the CJEU in the early 2010s and at a similar time the same issue was raised before the EFTA Court through the Posten Norge case and before the

Conceptualisation of Procedural Fairness in EU Competition Law  147 ECoHR through the Menarini case. The argument in each of these cases was placed in a context where the nature of anti-trust breaches was said to have gravitated over time, away from its initial origins as an administrative offence and towards a more serious criminal classification. At the same time the acknowledgment of fundamental rights within the anti-trust context had become more explicit. Both developments were said to warrant a reconsideration of whether the administrative regime for penalising anti-trust violations were compatible with modern ideas and human rights standards of procedural fairness. In each of the three cases the constitutionality of the enforcement architecture was upheld. The factual context of gradual criminalisation was however acknowledged and a conception of procedural rights as a relative concept, sensitive to the precise gravity of the offence at issue, was adopted on the model of the ECoHR’s Jussila judgment. This was done implicitly in KME/Chalkor and explicitly in Posten Norge and Menarini. This signalled a departure from a more binary model, where an offence was either an administrative offence or a criminal offence, each with its fixed set of procedural rights and requirements. The comments in the academic literature show that the debate about the fairness of the procedural regime did not end with the judgments. Opinions are still voiced about possible improvements, or the state of insufficiency. Conversely, there are also opinions expressing satisfaction with the status quo. By examining the KME/Chalkor saga, a persistent problem with reaching a consensus on how a procedural regime should be structured to ensure compliance with procedural fairness standards is identified. A review of the subsequent literature shows that the dilemma is resistant to court decisions about how the laws should be interpreted. This indicates that the source of the dilemma is located at a conceptual level, rather than in different doctrinal readings of the applicable legal sources. This finding motivates the main task of the book to conceptualise procedural fairness in the context of EU competition law, for the purposes of supplying an answer to the question of how to know if a competition law procedure is fair. In chapter two the project of conceptualising procedural fairness was initiated with an analysis of the concept of an adjudicative procedure. To facilitate this analysis, the phenomenon of an adjudicative procedure was presented in the analytical terms of the philosophy literature on social intuitions. Viewed in these terms, several instrumental elements can be identified that are universal to adjudicative procedural institutions. Normally such institutions consist of one or more adjudicative cycles that are regulated by the constitutive rules of the institution. Each cycle consists of a framing phase that frames and organises the relevant information for the subsequent decisional phase, where the actual adjudicative decision is made. In addition to formal constitutive rules, procedural institutions are also shaped by an informal culture that can beneficially or adversely influence the output of the procedural process. Procedural institutions should also consist of a system of sanctions, which provides the different role occupants within the institution with incentives for acting in line with the institutional agenda.

148  An Argument for a Data-based Democracy In addition to these instrumental elements, adjudicative procedural institutions are widely considered to adhere to a specific normative rationale. This rationale consists of two competing elements. The first normative consideration of adjudicative procedures is to facilitate a correct outcome. Due to uncertainties and definitive adjudicative resources, the concept of the correct outcome is a probabilistic concept, where absolute certainty is impossible to ascertain. This leads to a second normative consideration, which is the cost-efficiency of reaching a probabilistically accurate result. The normative rationale of any adjudicative procedure is thus to make the most of the available resources to reach an accurate result, where a balance needs to be reached between the degree of accuracy and cost-efficiency. This represents a normative dilemma integral to any procedural design: how accurate does the output need to be, and at what cost? The ideal solution to the dilemma is represented by the concept of procedural fairness. The concept of procedural fairness perceived in this way is a solution to a balancing dilemma and thus essentially a question of equitable distribution. In chapter three the procedural regime for enforcing EU competition law was analysed in terms of the preceding conceptualisation of procedural institutions. First, it was explained how a distinction must be made between viewing the normative rationale of EU competition law in the substantive sense and in the procedural sense. While the current consensus is that EU competition law should in the substantive sense aim at the maximisation of social welfare, subject to a distributional condition favouring consumers, the normative objective of EU competition law in the procedural sense is still the same as for any other adjudicative procedural regime, ie to facilitate the substantive objective with accuracy and efficiency. In the procedural regime for enforcing EU competition law, several types of stakeholder can be identified. Each is uniquely situated with regards to the ideal balancing of the procedure’s accuracy and efficiency. The accused will not want the same equilibrium solution as the competitor, or the consumer, or the general taxpayer. The difficulty of achieving a procedurally fair design is about reconciling those differences of opinions. By examining the instrumental design of EU’s competition procedure, it is possible to identify how different aspects of the procedural regime are designed for calibrating the normative balance between accuracy and efficiency, both with regards to the framing and the decisional task of the procedure. This is evident in the formal constitutive rules of EU’s competition procedure and in the institutional culture of DG Competition and the College of Commissioners. Mechanism for sanctioning or discouraging institutional corruption are also in place, both formally through a criminal law mechanism and informally through peer reviewing of case files and the supervising role of the Hearing Officer. The conceptualisation of procedures as adjudicative institutions thus provides a robust framework for analysing and identifying possible balancing problems in the EU’s regime for enforcing competition law. However, this is not enough for solving the dilemmas – a conception of fairness in procedural design is required.

Solving Fairness Issues in EU Competition Procedure  149 Chapter four poses the issue of fairness in relation to the law as a question about the place of ethics in legal decision-making. Seeing the standard of fairness as an ethical standard reveals two incompatible views about the role of fairness in legal decision-making in the two mainstream approaches in legal philosophy. In simplified terms, legal positivism maintains that the existence of a balanced procedural design is subject to a positive decision about what constitutes a fair balance. Also, in simplified terms, legal naturalism argues that a procedural design is balanced fairly when it adheres to the ethical standard of fairness, which exists prior to, and independently of, positive legislative decisions. Instead of arguing for one view over the other, an alternative argument about the place of ethics in the law was offered. This view maintains that the place of ethics in the law is sensitive to the structure and logic of the political community that creates them. The idea of the social contract is a common way within political philosophy to explain the nature and being of modern liberal democracies, which are founded on an assumption of a cooperative humanity. Using the analogy of the contract, a modern variant of this idea provides a narrative for understanding how fair balancing of different preferences can be achieved on a societal scale without having to resort to coercive tactics of governance. By adjusting this theory to the legislative context, a plausible narrative about the place of ethics in legislative design appears. A procedural rule is thus fair if it abides to the rationale of cooperative human interaction. This means that a rule will be considered procedurally fair if it complies with the ethical standard of egalitarian distributional fairness, as defined in the social contract literature. In simplified terms, this means that a rule will be deemed fair if it is designed in a way that equitably respects the different preferences of the relevant types of stakeholders. The two main elements of a conceptualisation of procedural fairness are thus, on one hand, the dilemma of any procedural design to balance procedural accuracy and procedural efficiency, and on the other hand, the fair solution to the dilemma which is achieved by the distributional fairness principle. In chapter five, a methodology for identifying optimal solutions to such balancing dilemmas was sketched out and in chapter six several such procedural dilemmas in EU competition law were identified. In the following section, these competition law dilemmas are measured against the proposed methodology.

II.  Solving Fairness Issues in EU Competition Procedure The problem of fair balancing arises where decisions have the potential to inflict asymmetrical consequences on the stakeholders involved. Ideally, a procedure should not give rise to differences once the stakeholders involved have agreed on the objective pursued through the procedure. However, the practical reality of limited resources imposes a need to accept a probabilistic risk of procedural

150  An Argument for a Data-based Democracy errors. The concession of errors in a procedure gives rise to two types of balancing problems: firstly, the total level of error concession; and secondly, the composition of the error types conceded. Thus, even if the stakeholders have an agreement on the substance pursued through the procedure, a disagreement can remain on how to deal with errors. Accordingly, the stakeholders in an EU competition law procedure can disagree over two balancing questions at each of the major decision points in the procedure: how much should be spent on achieving the correct result; and the probabilistic distribution of the error cost that must be conceded. Disagreements in relation to the first question can be explained based on the different effects a public investment in a procedure will have on the different types of stakeholders. Assuming that each stakeholder is equally sensitive towards a general increase in the tax burden, linked to an additional public investment in enforcement of the competition law regime, it might still be rational for the different stakeholders to prefer different levels of total investment due to their differing circumstances and thus differing stakes. The non-interested taxpayer will prefer an investment that is cost-beneficial in the broader macro-economic sense; the consumer will prefer an investment that is cost-beneficial measured by the benchmark of consumer welfare; the competitor will prefer an investment that is cost-beneficial in terms of maintaining a level playing field in the marketplace; and the accused will prefer an investment that is cost-beneficial towards avoiding unwarranted enforcement actions. Due to the differing stakes, the marginal utility of an additional investment in procedural accuracy is liable to vary between the different types of stakeholders. In the same way, the breakeven cost point is likely to be located at different investment levels, depending on how important the consideration of accuracy is for the different stakeholders involved. All other things being equal, it would thus be rational for the different stakeholders to prefer different levels of public investment in the enforcement of competition law. Differences regarding the second question can be explained based on the same rationale. Different stakeholders have potentially varying preferences towards the concession of different types of errors. The concession of an error has general welfare implications, but in some cases also special implications that compound the general implications. Enforcement errors in competition law thus have general welfare implications that effect all non-interested taxpayers and interested consumers, but additionally the accused and the competitors are liable to suffer special consequences due to their special stakes. The special consequences will however vary between the accused and the competitor, depending on the type of the error. The accused will prefer a false negative error, while the competitor will prefer a false positive error, assuming both are rational. In situations where the stakeholders have different preferences in the solution of a balancing dilemma, the role of the regulator is to find an equitable solution in which those preferences are fairly reflected. The model of fair rules can be used to find this fair equilibrium solution. For the purposes of locating the fair equilibrium points, with regards to the identified questions about procedural accuracy in

Solving Fairness Issues in EU Competition Procedure  151 the context of the EU’s competition law procedure, the model could be applied at the final decision point, where all the procedural steps and procedural possibilities aggregate in a grand balance of procedural accuracy and efficiency. The model could also be applied at each of the intermediary decision points by aggregating the procedural steps and possibilities up to that point to identify the intermediate equilibrium of procedural fairness. In high-profile competition law cases, the balance at the intermediary decision points can matter a lot. It is, for example, of great importance for publicly listed companies, that a public accusation of an anti-trust or cartel infringement is not made, unless a certain level of accuracy is guaranteed before it becomes permissible for public organs to intervene through such decisions. The consequences of an error at the intermediate decision points are of course not as severe as it would be at the final instance, and thus the fair equilibrium point erects a lower accuracy threshold for making an accusation, than it would for finding an eventual infringement. To apply the model of fair rules on the fairness dilemmas identified in ­chapter six, data input regarding the preferences of the relevant stakeholders is required. In the absence of such empirical data, an explanation in few steps about how the model could be applied and few calculated guesses about the preferences of the main stakeholder types must suffice. The guesses are based on a likely rational response to the incentives at issue that are noticeable to any observer, and on information revealed through the case studies and the legislative study at issue.

A.  Applying the Model of Fair Rules The central claim in KME/Chalkor was that accuracy of the competition procedure should be increased by abolishing or reducing the discretionary power of the Commission to establish certain complex economic facts at the end of the administrative part of the competition procedure. This claim entails a presumption that the system in place was not accurate enough and that the proposed change would increase the accuracy. The counter-argument to this claim, although not raised during the proceeding, could focus on these presumptions, ie that the current system was accurate enough and that the change did not appreciably increase accuracy. Additionally, it could be argued that the proposed accuracy enhancement was only cost-beneficial for a narrow group of stakeholders, while for most of the others, the additional investment was not desirable when compared with the gain in the level of accuracy. The central claim in Groupe Gascogne was that the most efficient way to enhance and rectify the fairness of a competition procedure that had been excessively delayed before the General Court was to discount the cartel fine imposed immediately at the final instance before the Court of Justice. The counter-argument, which prevailed, was that the accuracy of the competition procedure and the overall efficiency of the procedural system would be better served by using the universally

152  An Argument for a Data-based Democracy applicable procedure for claiming damages caused by the EU institutions, instead of permitting the use of a special procedure in cases where a cartel fine had been imposed. When preparing the enactment of the Damages Directive, the Commission assessed the possibility of creating a designated mechanism for collective redress against those found in breach of competition law. The main argument in favour of the establishment of such a mechanism contended that this would enhance the enforcement of competition law by enabling more victims to be compensated for the harm caused by those in breach. Those in favour additionally wanted the threshold for operating such mechanism to be low to ensure its effectiveness. The argument against maintained that this kind of mechanism would be abused by opportunists, who would launch false claims with a view to force companies into unmerited settlement payments. The mechanism would thus lead to inaccuracies of enforcement and waste of resources. Those against emphasised that if such mechanism would be established, the threshold for its use should be high so that unmerited litigations could be kept at a minimum. This latter argument ultimately prevailed in the Commission’s proposal for the Damages Directive, which did not suggest making such a mechanism mandatory in the national legislations of the Member States. As previously noted, the problems in the two court cases were undecided from a legislative point of view, which barred the Court from deciding based on an ex-ante answer previously provided by a legislator and thus required the Court to provide its own ex-ante type of an answer for the case at hand and for future cases involving the same or similar problems. Based on this understanding the model of fair rules can be applied on all the identified problems.

i.  Step 1: Identifying the Primary Policy Objective The primary policy objective, in the context of EU competition law, is currently settled at an equilibrium point that aims at total social welfare, restricted by a Pareto distribution rule with regards to consumers. The primary instrument, with which EU’s competition law regime seeks to pursue this objective, is through the instrument of competition in the marketplace, the details of which are further articulated in the substantive provisions of the competition law regime. Supplementary to the substantive provisions of the EU’s competition law regime are the procedural provisions that are described in various EU legal instruments and in the enforcement practices of DG Competition, the College of Commissioners and the Court of Justice. These provisions and practices serve the same auxiliary agenda as any other procedural rules: to facilitate with accuracy and efficiency the grand agenda of the substantive regime to which they provide their auxiliary service. The main objective of the procedural regime is thus different from that of the substantive regime. The objective of the procedural regime refers to an equilibrium point between the normative considerations of accuracy and efficiency, while the plan of the substantive regime is currently fixed at an

Solving Fairness Issues in EU Competition Procedure  153 equilibrium point between a utilitarian efficiency principle and a political distribution principle. The object of preference for the stakeholders in the context of the EU’s competition proceedings is thus the balance between accuracy and efficiency. As previously explained, this balance can be influenced externally by the increase or decrease of the total available resources, or it can be influenced internally through the probabilistic distribution of the types of errors conceded. In both cases, the exact equilibrium point between the two variables becomes the object of preference for the stakeholders involved. If the regulator manages to find an equitable solution to the balance between efficiency and accuracy in a procedure, based on the preferences of the stakeholders involved, the equilibrium point represents procedural fairness. A claim about procedural fairness thus relates to the balance between accuracy and efficiency of a procedure. In practical terms, the claim is often that more should be spent on accuracy, or that the risk of committing a certain type of accuracy errors should be reduced. Alternatively, the practical claim can be that the procedural process should be more efficient, either through higher investment of public resources, or through the relaxation of the accuracy demand. Viewing the central claim in KME/Chalkor as a policy proposal that the Court in its capacity as a decision-maker had to decide on how to respond to, the objective behind the proposal is easily identified as towards increased accuracy of the competition procedure at the decision point before the General Court. By expanding the jurisdiction of the General Court to scrutinise decisions of the Commission, the total accuracy of the procedure is liable to increase if additional resources are committed to match the increase in workload. If, however, the jurisdiction is expanded without committing additional resources, the rules of proof will require that the existing resources will be prioritised towards reducing false conviction errors, which will leave fewer resources towards scrutinising acquittal judgments for errors. Viewing the competing claims in Groupe Gascogne in the same way, as policy proposals to adopt a rule on how to compensate for undue procedural delays before the General Court, the objective behind the proposals can quickly be revealed as, on one hand, being towards increased efficiency of the procedure and, on the other hand, towards increasing accuracy. By adopting a simple procedure for claiming damages due to procedural delays, total efficiency of the procedural process is enhanced and the incentive for the institutions to handle cases timely and efficiently is reinforced. The incentive that is reinforced, is however likely to undermine the accuracy of the procedure, unless additional resources are committed towards enhancing the case-handling time. The other claim is based on accuracy considerations, ie by mixing the compensation procedure with the cartel procedure, the accuracy of both is put at risk. The competing views on if and how a collective redress mechanism should be part of the Damages Directive can be viewed as an argument about the distribution of accuracy and costs associated with realising the enforcement of competition law. The policy proposal behind the Damages Directive was clearly stated in the

154  An Argument for a Data-based Democracy Directive’s preamble as being to enhance the enforcement of competition law. One of the questions during the legislative process was whether a collective redress mechanism would achieve that objective with accuracy and cost-efficiency. Lowering the threshold for claiming private damages would supposedly increase the risk associated with breaching the law, while such policy might also increase the risk of false positives by enabling unmerited claims to succeed.

ii.  Step 2: Preference Function Presuming that a given equilibrium point of accuracy and efficiency in competition proceedings will have different effects on people depending on their situation, it is logical that the preferences of people will differ with regards to where the equilibrium point should be located on the axis between efficiency and accuracy. It should thus be possible to categorise different stakeholders in a competition proceeding based on their role in the proceeding, which is determined by their situation relative to the proceeding. Regarding the EU’s competition law proceedings, four types of stakeholders have been defined: the accused; the competitor; the consumer; and the taxpayer. Each of these stakeholders are situated differently, which gives them an incentive to hold diverging preferences with regards to the question, where the optimal equilibrium point of procedural accuracy and procedural efficiency should be located. The accused and the competitor will prefer large public investment in accuracy, but they would hold diverging preferences regarding the types of errors that should be the object of emphasis within the confines of the currently available resources. The consumer will prefer public investment in more accuracy if his or her share of the cost would not exceed their expected increase in welfare and they would be neutral with regards to the allocation of the conceded error types. The general non-interested taxpayer will support investment in procedural accuracy if it remains cost-beneficial in a macro economic sense, which means that the investment should yield welfare returns that exceed the returns from other investment alternatives. The taxpayer is, like the consumer, not sensitive towards the typological allocation of conceded errors. The preference function of the stakeholders towards the possible solutions of the identified dilemmas of procedural design can be assumed, based on the effects the proposed changes are liable to have on the stakeholders and on the current status quo. The effects of changing the competition procedure along the lines of the main claim in KME/Chalkor can be roughly assumed to be as follows: the accused would prefer to reduce the discretion of the Commission to establish complex facts, since they would benefit from increased accuracy of the procedure or, in the case of no additional enforcement resources, to increase the emphasis on reducing false positive errors. The competitors would, however, be neutral towards reducing the discretion of the Commission because only a finding of a breach would be scrutinised, and they have an an active interest in this being as accurate as possible. If no additional resources would be committed alongside

Solving Fairness Issues in EU Competition Procedure  155 the change, the competitors would oppose it due to increased risk of false negative errors. The consumers would favour increased accuracy of the procedure so long as it could be shown that the cost of doing so would yield a net increase in their own welfare. They would also only favour more emphasis on reducing false positives if it could be shown to increase their welfare beyond the status quo. The taxpayers would support additional investment if it were to increase welfare more than competing investment options, and they would also only support increased emphasis of reducing false positives if it were to increase the taxpayers’ welfare beyond the status quo. The effects of adopting the procedures argued for in Groupe Gascogne involves comparing two alternative options, rather than comparing a reform claim with the status quo. A distinct problem of these cases involved ambiguity about the actual content of the status quo. In such cases, the best approach is to assess the preferences of each of the relevant stakeholders. The accused in the main competition proceeding would, as is indicated by the pleadings of the applicants, prefer the simple procedure of immediately lowering the ultimate cartel fines to compensate for undue procedural delays on behalf of the General Court. The channelling of public resources to a new procedure for determining damages would leave less to get things right in the original cartel procedure and would additionally impute significant additional direct and indirect costs on the accused. Even in the case of additional public investment, it is probable that the gain by the accused of having a separate damages procedure would be eclipsed by the additional private costs. The competitors would, on the other hand, prefer that the damages claim did not interfere with the cartel proceedings and is thus dealt with separately through a procedure that would be relatively accurate, given that additional public resources would be committed towards the task. In the case of no such investment commitment, the competitors would have to weigh the harm to accuracy on the cartel proceeding of including the damages claim, with the harm to accuracy of channelling some resources towards initiating a separate procedure. The consumers would choose the cheaper of the two procedural alternatives, unless the accuracy gains from choosing the expensive one would increase their own welfare. The same logic would apply if there were no additional public investment involved. The taxpayers would in the same way chose the cheaper option, unless an investment in the expensive option could be rationalised as a beneficial investment compared with alternative investment options. With regards to the implementation of a special collective redress mechanism for the competition law context several options could be weighed against the status quo of not making such a mechanism mandatory. The different options would range from a highly effective mechanism, which is prone to errors, to an ineffective mechanism that makes few errors. Assessing how different stakeholders in a competition proceeding would respond to these options, it can be assumed that the accused would prefer the status quo, where he or she will not have to defend against private litigants taking collective actions to claim damages for the harm caused by their violations. In any case, the accused would want a mechanism that

156  An Argument for a Data-based Democracy does not make false positive errors about his or her fault and they would thus prefer high precision with regards to the avoidance of such errors. On the other hand, the competitor would prefer an effective mechanism that would discourage any violations to the competition laws and thus make sure that the playing field is fair. He or she would not be concerned about potential enforcement errors. The consumer would also prefer an effective mechanism for collective redress and would not be concerned about errors. The taxpayer would want an effective mechanism, but would at the same time want to minimise the social cost of errors and enforcement. He or she would thus prefer a mechanism that is somewhat effective, but which would also be designed for minimising social costs. He or she would thus prefer an efficient equilibrium solution in a macro-economic sense between effectiveness of enforcement and social cost.

iii.  Step 3: Preference Index By establishing a preference index (PI), an attempt is made to compare the relative importance of what each stakeholder wants. Not aiming at impeccable precession, the method is meant to provide the regulator with a useful tool for comparing the relative importance of different lobbying arguments, which is methodologically more robust than instantly applied intuition or a reduction to welfare inputs digestible for a traditional cost–benefit analysis.1 The first step in building a PI is to define magnitudes of the different preferences at stake. A starting point is to define minimum and maximum magnitudes; ie if this stakeholder gets all of what they want, the consequence will be this, and if they get nothing of what they want, the consequence will be that. This provides two numerical values that represent specific consequences at the opposite extremes. By comparing two stakeholder types (or more) in this way, a pair (or pairs) of corresponding magnitudes appears, ie the consequence of a 0  per  cent and a 100  per  cent preference fulfilment for the different stakeholders. This information enables an objective comparison of the different profiles. Simply by assessing the relative importance of a specific level (eg 0 per cent or 100 per cent, or any other comparable magnitude) of preference fulfilment for two or more stakeholder types. The perspective of assessment must be neutral with regards to the stakeholders involved; their subjective interest must not influence how their preferences are objectively viewed. The aim is to establish how important the claim of a stakeholder is compared with the claim of another, as observed by a neutral bystander. Two variants of the PI were suggested. A simple one that compares comparable magnitudes of different preferences and determines their relative importance

1 As mentioned in ch 5, the PI is inspired by the utility inference method of Van Neumann and Morgenstern, Savage’s extension of the Bayesian probability model, the common priori assumption of Harsanyi, and the reference point dependence insight articulated by Kahneman and Tversky.

Solving Fairness Issues in EU Competition Procedure  157 through Van Neumann and Morgenstern’s utility inference method, and a more complex one that incorporates the status quo reference point to increase descriptive accuracy. Using the balancing problems in KME/Chalkor and Groupe Gascogne as examples, the comparison starts by defining a best- and a worst-case scenario for each of the stakeholders. It has already been established that the balancing problems relate to specifically located decision points in an EU competition procedure.2 A definition of the best and the worst outcomes in relation to these decision points would thus be established, ie the worst and the best outcome for each stakeholder with regards to decision point four in KME/Chalkor, and the fifth point in Groupe Gascogne. The collective redress legislative problem would be framed in similar terms by defining the possible outcomes of the different legislative options with regards to the relevant stakeholders. Once the best- and the worst-case scenarios have been defined, one option would be to assemble a simplified version of the PI by simply inferring objective utility values in a numerical form on the corresponding magnitudes and thus reveal their relative importance.3 The other option would be to assemble the more complex version. This would provide a more case-specific point of view by defining the current status quo in terms of prefence for the stakeholders involved and by locating this status quo on the axis between the worst- and the best-case scenarios.4 Irrespective of which variant is used, for simplifying purposes it would need to be assumed that the best possible consequence is at a point where the preferences of all stakeholders are completely fulfilled, and the worst possible consequence is where the preferences of all the stakeholders are completely ignored. All possible combinations of preference fulfilment of the stakeholders involved should thus fall on an axis between these two extremes, including the current status quo. Assuming this, all that needs to be done is to compare the impact of a change to one stakeholder type on the axis of overall preference to the impact of a change to another stakeholder type. By comparing the impact, an insight is gained on the objective importance of the arguments of different stakeholders with regards to a specific issue. The composition of the PI can be summarised as follows: In the context of a competition law procedure, a regulatory change towards increased accuracy could have a certain overall impact on the satisfaction of the population in question. But on closer inspection, some would be very happy (eg the accused), while others would be slightly positive (eg the consumer) or almost 2 In ch 3, five major decision points are identified in a complete competition procedure: three administrative decision points before the Commission (start formal investigation, SO and final decision); and two judicial decision points before the courts in Luxemburg (General Court and CJEU). 3 The question could simply be ‘how important is it to fulfil all of this stakeholder’s desires, compared with fulfilling all of another’s stakeholder’s desires?’ 4 The inclusion of the reference points adds a dimension into the comparison. Instead of comparing absolute levels of preference fulfilment at a single measuring point, the comparison considers the preferableness of changes to the current level of preference fulfilment.

158  An Argument for a Data-based Democracy Table 7.1  Step-by-step guide to the composition of a PI Step no

Action

Result

Step 1

Define best- and worst-case scenarios for each stakeholder with regards to the policy objective at stake.

100% and 0% magnitudes of preference fulfilment established for each stakeholder with regards to the policy objective at stake.

Step 2

Identify the status quo for each stakeholder between the extremes of the best- and worst-case scenarios (eg how preferable is the current situation with regards to the policy objective?).

Information about the status quo preference level for each stakeholder.

Step 3

Compare effect of a total loss and a complete gain on each stakeholder counting from the status quo as observed by a neutral bystander (eg how would a 100% gain and a 100% loss objectively affect the different stakeholders?).

Information about the relative importance of each stakeholder’s claim.

indifferent (eg the taxpayer) about the change. When the regulator is deciding whether the proposed rule is a sensible, he or she will notice conflicting lobbying arguments from the different stakeholders about its importance. An argument made by a certain stakeholder may express intensive concerns, but that does not necessarily equate the general importance of the argument. Thus, a tool that gives the regulator a neutral perspective on the importance of the various claims of the different stakeholders would be useful. By using the proposed method, a numerical value can be given on how important a specific consideration is in a general sense. A fixed ratio can then be established between the different considerations. For example, it could be that the argument of an accused stakeholder in a competition proceeding would objectively be found twice as important as an argument of a taxpayer stakeholder. The former would argue for accuracy of the proceeding while the latter would argue for the cost-efficiency of the proceeding. The PI of the taxpayer argument thus weighs less than the PI of the argument of the accused, at a preference ratio of 2:1, for this regulatory situation.5 By establishing systematically the relative importance of different lobbying arguments of stakeholders towards specific balancing problems, the legislator gains a tool that enables him or her to tune with more accuracy the optimal solution to regulatory problems.6 The PI for different stakeholder interests gives an

5 Note that these are not actual numbers. Data from an empirical survey is required to establish actual numbers. 6 This presumes that optimality is a specifically calibrated balance of the preferences of the relevant population.

Solving Fairness Issues in EU Competition Procedure  159 objective perspective on what matters in the balancing and which weight should be given to individual types of claims.

iv.  Step 4: Preference Matrices A range of actions can be proposed to achieve a specific legislative plan. These actions can be received differently by the various stakeholders involved, depending on their expected impact and consequences. By listing the proposed actions and the intensity with which individual types of stakeholders like or dislike the probable consequences of a proposed action in a matrix, it becomes easy to compare and process further the subjective value of an action, as seen from the perspective of each of the stakeholders. Using the examples previously mentioned, a matrix would simply be assembled for each stakeholder in a competition proceeding and the intensity of the effects on the preferences function of each would be measured with regards to the proposed actions. By surveying how preferable a proposed action is for each of the stakeholder types, an estimation is being made about how close to the point of optimisation the action is, from the perspective of each stakeholder type. If the probable consequences of the proposed action were close to the optimum for a stakeholder, the action would measure as preferable at a certain intensity in the preference matrix for that stakeholder. The probable consequences of the same action could at the same time suggest an outcome far from the point of optimisation for a different stakeholder, and thus measure on his preference matrix as un-preferable with a certain intensity. For the purposes of the balancing problems in KME/Chalkor and Groupe Gascogne, the probable consequences of limiting the discretion of the Commission to establish complex economic facts, and the consequences of the two competing options for resolving damages claims for procedural delays in cartel proceedings would need to be envisioned. This could be achieved through a survey among the relevant stakeholders about how preferable the different actions are, measured against the current status quo. The result of the survey would then be recorded in a numerical form into a preference matrix for each of the stakeholders, with a special provision for each of the proposed actions. The recorded data would represent the subjective value of each of the proposed action with regards to each of the relevant stakeholder type. Table 7.2  Preference matrix for the KME/Chalkor and the Groupe Gascogne problems Proposed action

Reduce COM discretion

Lower cartel fines

Separate damages proceeding

Level of resources

+ Resources =

+ Resources =

+ Resources =

The accused The competitor The consumer The taxpayer

160  An Argument for a Data-based Democracy The same can be done in relation to whether to establish a collective redress mechanisms for the private enforcement of competition law. The different regulative options can be compared in a matrix. Table 7.3  Preference matrix for the collective redress question Opt-out + Opt-in + US-style class measures to representative action prevent abuse actions

Opt-in + measures to prevent abuse

Level of resources + Resources = + Resources = + Resources =

+ Resources =

Proposed action The accused The competitor The consumer The taxpayer

v.  Step 5: Identifying Fairness The fifth and the final step in applying the model of fair rules evolves around interpreting and working the numbers gathered during steps three and four into a comprehensive solution to the problem of procedural fairness. During step four, the subjective preferences of the stakeholders involved in the problems in question were established, and during step three, the objective importance of the differing preference functions of the different stakeholders were established. It is thus clear what each prefers subjectively, and how important that preference is objectively. During the final step, these two variables are merged into a third variable, which seeks to represent an objective value of the preferences of the stakeholders. The formulation is as follows: The objective value (OV) of an individual preference claim is a derivative from the subjective value (SV) of an individual preference claim and the applicable individual PI: OV = PI × SV For public policy purposes, the aggregated OV of a policy choice should be maximised, conditioned on an equitable distribution of the individual SV. The overall efficiency of public policy should thus be maximised, but only to the extent that individual subjects of the policy would have their private preferences fulfilled fairly, in comparison to others. To establish the combined OV of a specific choice the numbers of the stakeholders holding a specific preference would need to be considered by multiplying their number with the individual OV (OV × (Stakeholder × n)) or by multiplying their proportion of the whole with the individual objective value (OV × (stakeholder × 1/n). In practice this means that not only the importance of the preference that an individual stakeholder holds matters (ie the PI), but also how many hold that particular preference of the relevant population.

Solving Fairness Issues in EU Competition Procedure  161 The level of importance of a specific preference directly affects its combined OV, and the relative number of stakeholders holding that specific preference can either inflate or deflate its combined OV. Often there would be an inverse relation between the relative number of stakeholders holding a particular preference, and the individual OV of the preference, ie, the accused in a cartel proceeding might hold preferences towards the accuracy of the proceeding that would objectively be recognised as very important, but due to how small a proportion of the relevant population of stakeholders the accused represent, the combined OV of the accused’s preference might be much lower proportionally than the individual OV of his or her preference. The opposite could often be the case with consumers and taxpayers. They may hold comparatively unimportant individual preferences, but due to their vast numbers the combined value of their preferences becomes an important consideration for designing public policy. In the three balancing dilemmas the preference matrices created during the fourth step should be adapted by adjusting the SVs with the PI for each stakeholder and the relative numerical size of each stakeholder’s type. The matrices should be updated using the following formulation on each of the preference values: The combined objective value (COV) of the preference of a particular type of stakeholders is a function of the individual SV multiplied with the stakeholder’s PI and the relative size of the stakeholder type (1/n Stakeholder). COV = SV × PI × (1/n Stakeholder) Following this adaptation, the total utility of each proposed action for each stakeholder type is defined and so the total utility of each proposed action can be established by adding these together. If interested in maximising the utility of public policy, this would provide the information to enable the identification of the most efficient policy action. If interested in identifying the optimally fair policy option (ie that the optimum of maximising total welfare is restricted by a distribution condition), only those policy options that guarantee equity of preference utility among the stakeholder types are eligible for selection. The task is thus to identify among the eligible policy options the single option that provides the highest aggregated utility of preference fulfilment. In practice, it could prove difficult to compose a policy proposal that would guarantee absolute equity. It could thus be necessary to define a margin within which some inequality of preference fulfilment would be permitted, to enable the creation of more efficient policy options. The matrix created in the context of step four (ie subjective preferences of the stakeholders) is the object of analysis for comparing equity of preference fulfilment. To account for differences in the OV of the preferences, the PI is used to transform the SVs into OVs. However, contrary to where the OV was found for the purposes of establishing the combined objective value (COV), the SVs would need to be divided by the PI, not multiplied. This is to facilitate comparison of the numbers, so that equal numbers represent an equitable distribution of preference fulfilment between different stakeholders holding preferences that are unequally

162  An Argument for a Data-based Democracy important. Low intensity of unimportant preference would thus be required to match high intensity of an important preference to achieve equitable distribution of preference fulfilment. Using a cartel proceeding as an example, an objectively important preference of the accused would thus need to be fulfilled at a higher intensity than an objectively unimportant preference held by the taxpayer, to achieve equitable distribution of preference fulfilment.7 At the end of the fifth step, two decision matrices should thus have been established: one representing the COV of the preference fulfilment of the various stakeholders for each of the proposed actions, and another representing the relative level of objective preference fulfilment for each of the stakeholders involved with regards to each of the proposed actions. Using these two types of matrices, an efficient option can be identified that also fulfils the criteria of distributional fairness. For the purposes of procedural regulation, the legislative option that satisfies the criteria of equitable distribution of preference fulfilment among the relevant stakeholders, and does so with the utmost efficiency, can be branded as optimal in terms of procedural fairness. Considering the KME/Chalkor problem in these terms, it seems evident that the accused in the initial cartel proceeding have most to gain from limiting the Commission’s discretion to establish complex economic facts. The other stakeholders would seem to have less of an incentive to support such changes and would potentially stand to lose from them. The net effect of the change could thus easily reduce total welfare by reducing COV towards the procedure at stake. This however shifts the attention to the distribution side of the equation. If the current status quo is fixed at an equilibrium that is very differently preferred by the relevant stakeholders in the objective sense, the consideration of procedural fairness might very well bar the continued use of the status quo due to unequal distribution of preference fulfilment. In that case, the other stakeholders might have to accept a reduction in their own preference fulfilment and a reduction in the total combined output of preferableness to accommodate the fairness claim of the accused. In the case of KME/Chalkor, this might mean that even if all the stakeholders involved (except for the accused) would stand to lose from limiting the discretion of the Commission, the right thing to do considering the optimisation of procedural fairness could still be to adopt the change if the result of the preference survey would provide an empirical validation. The problem in Groupe Gascogne is slightly different. The status quo was unclear and thus the competing options were two new alternatives, rather than the status quo versus a change. A potential outcome of the comparison of these two options could be that one would be more efficient, while the other would be more equitable in terms of the distribution of preferableness. A decision would then have to be made about whether both options, or only one option, is sufficiently

7 ie, if something is relatively important, more of that needs to be done in order to reach an equitable distribution compared with something that is not as important.

Solving Fairness Issues in EU Competition Procedure  163 equitable. If both satisfy the criteria, the efficient option should be chosen, but if only the inefficient option satisfies the criteria, it would nonetheless have to be chosen as the only available option that satisfies the condition of procedural fairness. In Groupe Gascogne the applicants preferred an immediate reduction of the cartel fines imposed, rather than having to start a separate procedure for claiming damages. The balancing problem could thus be said to hinge on whether the less preferable option was, from the point of view of the accused, nonetheless preferable enough to satisfy the condition of equitable distribution of preference fulfilment among the relevant stakeholders. The Court of Justice decided that the later procedure was indeed fair enough, and that it would be more efficient to use a universal procedure for settling damages claims that arise in the context of competition procedures. Empirical data about the preferences of the stakeholders in the cases could show whether the intuition of the Court about the preferableness of the competing procedural alternatives was right or wrong, and whether it managed to identify the optimally fair procedure. The question about whether to establish a collective redress mechanism for the private enforcement of competition law was raised in a context where the status quo meant that most victims of anti-competitive behaviour had no realistic means to seek damages from those responsible. In that sense, the status quo was already at an equilibrium that was unfair to all stakeholders, except those who were guilty of a breach. During the legislative process of the Damages Directive, this was explicitly recognised as constituting problems with regards to obligations under Article 19 TEU and Article 47 CFR (see further discussion in chapter six). When assessing the different regulatory options for a collective redress mechanism, the option argued for by the largest group of stakeholders was quickly discarded on the basis of an impact assessment that framed the option in an unfavourable light. In the end, the decision was about choosing to maintain the status quo and implementing an ineffective system favoured by representatives of businesses. Ultimately the status quo prevailed and thus so did the situation that had at the onset been deemed problematic with regards to procedural fairness as articulated in Article 19 TEU and Article 47 CFR. It is evident that the legislator decided that the interests of one group of stakeholders should prevail to the detriment the stakeholder groups of competitors and consumers. A more robust methodology for assessing the different alternatives could have provided the legislator with a vital data input for either supporting the decision it took, or for supporting an alternative solution to the legislative task at hand.

B.  Assessing the Utility of the Model The results from this application reveal that the arguments used by the Court for its conclusion in Groupe Gascogne were consequential in nature and could have been supported with empirical evidence about the preferences of the relevant stakeholders, instead of relying on the intuition of the members of the Court about what was

164  An Argument for a Data-based Democracy sufficiently fair, or optimally efficient. The finding in KME/Chalkor was supported by a more traditional deontological type of an argument about procedural imperatives, but could also have benefited from a more consequential approach in which the preferences of the relevant stakeholders would have been analysed in more details. The approach of the Commission to the collective redress question in connection to the Damages Directive was consequential in character, but the weighing of the arguments in the final legislative proposal was compromised by inadequate framing of the alternative options and by an irrational inference from the underlying facts. The effects of these methodological shortcomings aligned perfectly with the outcome desired by representatives of businesses, as expressed through two rounds of public consultation. In the absence of actual data about the preferences of the stakeholders, it is hard to conclude whether the quasi-legislative decisions of the Court of Justice in KME/Chalkor and Groupe Gascogne where warranted, viewed in terms of the optimally fair balance between procedural accuracy and procedural cost-efficiency. It is however probable that the KME/Chalkor decision, to reduce the discretion of the Commission to establish complex economic facts, is liable to increase the cost of competition procedures in general. But it is uncertain whether doing so will appreciably increase the accuracy of such procedures. Also, by choosing the expensive procedural alternative in Groupe Gascogne, of starting a separate procedure for claiming damages for undue delays in competition procedures, there is a risk of an overinvestment in the accuracy of an easy legal and factual question. It can be doubted whether the additional investment is cost-beneficial for the relevant stakeholders involved, in terms of the balance of procedural accuracy and cost-efficiency. The decision to maintain the status quo with regards to collective redress in connection with the enactment of the Damages Directive seems to have maintained an unfair procedural situation, where large classes of anti-trust victims are unable to effectively claim compensation for the harm caused by businesses who violate competition rules. During the legislative process several viable improvements were suggested for improving the fairness of the procedural regime, but were not ultimately successful. By applying the model of fair rules to these kinds of problems, methodological uniformity and accuracy is achieved in dealing with hard questions of what the law ought to be, both for legislators in the traditional capacity and for the occasional legislators wearing the scarlet robes of the Court of Justice.

III.  Technocracy and the Making of Laws The project of conceptualising procedural fairness in EU competition law was designed to answer a simple question: how do we know if a competition law ­procedure is fair? By exploring the concept of procedural fairness philosophically,

Technocracy and the Making of Laws  165 it has been established that the element that causes a procedure to qualify as procedurally fair, is an equitable equilibrium solution to the problem of balancing procedural accuracy and procedural efficiency. The answer to why a procedure qualifies as fair is thus intertwined with how optimally the equilibrium of these two variables is calibrated. Optimal calibration correlates with a procedurally fair regulatory design. To answer objectively and consistently if a procedural design is optimally calibrated in this regard, a methodology of some kind is required. The main suggestion made in this book is that any question about procedural fairness in EU competition law has a definite answer which can be extracted or approximated by an adequate methodology. Treating the question of procedural fairness as a methodological issue has implications for regulators – in this case, the regulator responsible for the enforcement of EU competition law. Normally the task of regulators is perceived to be deciding how things ought to be in future instances, but when a regulatory issue is viewed as a methodological problem, the ‘ought’ already exists and the main issue is about how this ‘ought’ can be achieved methodologically. The solution to the methodological issue argued for in this book is to resort to decisional expertise in a technical sense. In other words, the ideal procedural design, with regards to fairness issues, should be extracted by the expert technocrats. Technocracy as an alternative to democracy in regulatory design has been discussed in the social sciences literature in various context over the past century. The technocratic ideal originated in the US during the first half of the ­twentieth  century.8 The central idea of the early proponents of technocracy was that the modern industrial state should be organised in accordance with the principles of human rationality. To achieve this, the administration of the state needed to be separated from politics and the regulation of industry and the economy placed in the hands of scientific experts and problem-solvers.9 If this could be achieved, ‘the exercise of decision-making power would be largely removed from the corruption of politics and transferred to expert public servants’, which could freely exercise their rationality to the problem at hand for the common good.10 Viewed in these absolute terms, technocracy is framed as an opposite to democracy as a form of government. The experts should take decisions about issues of governance, isolated from the populist pressures associated with democracy. In this sense, democracy is viewed as a threat to the rational order of technocracy. Viewed the other way around, from the perspective of democracy, there is also a concern that an orientation towards technocracy undermines the ideal 8 Alfred Moore, Critical elitism: deliberation, democracy, and the problem of expertise (Cambridge, Cambridge University Press, 2017) 37. 9 Daniel A Crane, The institutional structure of antitrust enforcement (Oxford, Oxford University Press 2011) 69–70. 10 William E Akin, Technocracy and the American dream: the technocrat movement, 1900–1941 (Berkley, University of California Press, 1977) 3.

166  An Argument for a Data-based Democracy of democracy as a form of government. The worry is that democratic discourse over policy options becomes dominated by a few experts who have a monopoly on declaring the scientific truth about a given issue.11 Defying the experts thus becomes a defiance of rationality, which greatly narrows the available spectrum for democratic deliberations of the citizens. It has been further argued that the asymmetry in knowledge between an expert and a non-expert is often complete. The expert knows and others do not know. The others do not even know if the expert is expressing his or her expert view. If decisional authority is delegated to experts on these terms of information asymmetry, any effort to second guess their decisions becomes meaningless.12 One way to address the problem of information asymmetry is to ensure communication and transparency between experts and non-experts. But taken to the extremes, this too can become meaningless if the experts assert the power to control and define the dialogue. In such situations, the experts define which communication is meaningful and which is not, thus keeping control over which claims are permissible in democratic deliberations. There is, in other words, a concern that democracy is at risk where the reality is defined by expert authority, since political deliberations are bound by terms dictated by the experts.13 A tactic to counter the supremacy of the experts is to politicise expert knowledge, so that each political preference is supported by a rationality claim of a supposed expert. The political discourse thus becomes dominated by special interests speaking through expert proxies.14 This can lead to confusion among the citizens about the truth and a distrust in true expertise, ultimately creating opportunities for populist political movements using anti-elitist and anti-intellectual rhetoric as a political platform.15 Rather than treating technocracy and democracy as opposites, they could be viewed as complementary. To understand how this could be, a more detailed understanding of the two terms is required. In line with how the idea of democracy

11 Habermas expressed this worry in his book Theory and Practice (originally published in German in 1963) and this worry links with his celebrated idea about communicative action, which emphasises public discourse over ideas and matters of social policy. See, for example, in Jürgen Habermas, Theory and practice (John Viertel tr, Boston, Beacon Press, 1973) 264–65. See also more generally in Jürgen Habermas, Theory of Communicative Action (Thomas McCarthy tr, Boston, Beacon Press, 1984). 12 Moore assigns this argument to Robert Dahl. See Moore, Critical elitism (2017) 39–40. See also Robert Dahl, Democracy and its critics (New Haven, Yale University Press, 1991). 13 Moore (n 8) 40. See also Frank Fisher, Citizens, experts, and the environment: the politics of local knowledge (Durham, Duke University Press, 2000) 24–28. 14 A potent contemporary example is the political discussion about climate change, which concerns a scientific issue that requires a global political action. For a discussion about politics and science see Roger A Pielke, ‘When scientist politicize science: making sense of controversy over The Sceptical Environmentalist’ (2004) 7 Environmental Science & Policy 405. 15 This development has been branded as post-truth politics and has been associated with electoral outcomes such as Brexit and the Trump presidency. See further Jayson Harsin, ‘Post-Truth and Critical Communication Studies’ (2018) Oxford Research Encyclopaedia of Communication, available at https://oxfordre.com/communication/view/10.1093/acrefore/9780190228613.001.0001/ acrefore-9780190228613-e-757.

Data-based Democracy for Regulating EU Competition Procedure  167 has been approached in this book, it can be viewed as a technique for taking social decisions where it is deemed important that the preferences of the relevant population are reflected in the decisional outcome. This is an appropriate decisional method, where the normative objective at stake is absent, uncertain, or contested. In such instances the normative objective is settled democratically based on the population’s preferences. Technocracy can also be viewed as a technique for taking social decisions, but unlike democracy, it ideally needs a clear normative objective as an input. If the normative objective is unsettled, it becomes the prerogative of the experts and the elites to impose their normative preferences on the general population, which are not necessarily in sync with the sentiments of this population. The technocratic method is well suited for solving decisional problems where the normative objective is well defined, so that the experts can concentrate on finding the optimal means towards the generally accepted normative end, using rational choice and scientific methods. Viewed as absolute forms of government, a technocracy would thus insist on the technocratic method for all government policy decisions, including decisions where the normative objective is vaguely decided.16 Democracy as an absolute would insist on the democratic method for all government policy decisions, including decisions which are predominantly about finding the best means for achieving a clear normative objective.17 Clearly, a complementary approach, where the strength of each decisional method is adequately used, is potentially superior to the pure forms of each. But the difficult problem is to design such a mixed system of governance, where each decision is addressed by an appropriate decisional modality to achieve the best of both worlds, instead of a system that exacerbates the flaws in both.18

IV.  Data-based Democracy for Regulating EU Competition Procedure In the modern democratic administrative state, experts and politicians interact and coexist in various ways in government. In practice, there is thus not a question 16 As a pure form of governance, Centeno defines technocracy as: ‘The administrative and political domination of a society by a state elite and allied institutions that seek to impose a single, exclusive policy paradigm based on the application of instrumental rational techniques’. See Miguel Angel Centeno, ‘The new Leviathan: the dynamics and limits of technocracy’ (1993) 22 Theory and Society 307, 314. 17 Pettit discusses three types of decisional scenarios, where the politics intrinsic to the democratic method obstruct the identification of the best decisional outcome. See Philip Pettit, ‘Depoliticising Democracy’ (2004) 17 Ratio Juris 52, 54–58. 18 Moore suggests that a division of labour between the two approaches to social decisions and explains how the place of expertise in a democracy is often said to be for informing democratic deliberation, instead of manipulating it. Moore (n 8) 34 and 58.

168  An Argument for a Data-based Democracy of either or in terms of democracy and technocracy, but rather a question of increments and emphasis on each in relation to the other. As to the enforcement of anti-trust policy, Daniel Crane has argued that while the original enforcement intuitions in the US were modelled in accordance with the democratic decisional modality, there has since been a gradual shift towards a technocratic modality. The symptoms of this shift are that initially, anti-trust policy was a political issue about how the economy should be organised and it was debated as such by politicians and public intellectuals. Gradually, the anti-trust economic argument gained acceptance and a status of orthodox in the law and administrative practice. Following this acceptance, the issue of enforcement in the popular consciousness diminished, while administrative experts and elite private lawyers and economist quietly assumed complete control of the task, away from the political fray.19 This shift further coincided with an increased political tolerance towards big corporations during the 1960s and 1970s, but intolerance towards economic concentrations was one of the main incentives for the passage of the original Sherman Act in 1890.20 Crane argues that in the US the political elites of both major political parties have struck an implicit bargain over the enforcement of anti-trust policy. The bargain entails that ‘antitrust is an important but largely technical matter that should be administered vigorously but without great public fanfare’. He attributes this bargain to three factors. First, a consensus has been reached about consumer welfare and economic efficiency as the normative goal of anti-trust policy. With this consensus in place, the issue of enforcement thus becomes about means, rather than ends. Second, the ideological issues at stake have also largely been settled. There is now a consensus about the legitimacy of the capitalist order, which was more contested in earlier times, and there is also a relative consensus that the freedom of property and contract does not permit corporations to engage in anticompetitive practices. Third, the enforcement of anti-trust policy is generally not viewed as a distributional issue that should be monitored closely by political interests. The economic distributional effects of anti-trust policy are less predictable than with many other policies.21 This narrative generally fits with the analysis that technocracy is a suitable decisional methodology, where the normative political agenda is uncontested and well defined. While the enforcement of anti-trust policy gradually assumed a technocratic character, the initial institutional structure remained intact. In the US, anti-trust is thus still mainly enforced through non-expert institutions, such as juries, generalist judges, generalist prosecutors and private litigants. The Federal Trade Commission could be viewed as a technocratic institution but, as explained by

19 Crane, Institutional structure (2011) 69–72. 20 Tim Wu reflects on the origins of the US trust and antitrust movements in Tim Wu, The curse of bigness: antitrust in the new gilded age (Durham, Columbia Global Reports, 2018) 24–32. 21 Crane (n 9) 87–88.

Data-based Democracy for Regulating EU Competition Procedure  169 Crane, its powers are subject to judicial review of generalist courts and it was designed to enforce norms created by statutory law, not to regulate competition.22 Crane views this as a ‘major glitch’ in the narrative of the technocratic shift, which has ‘created a mismatch between ends and means with major implications for the functioning’ of the US anti-trust system.23 EU competition law was initially modelled on US anti-trust law, but the institutional structure differed and the enforcement ethos was from the beginning more administrative and thus more expert-oriented.24 The enforcement is subject to judicial review, but due to the large share of competition cases in the total case log of the EU courts, the European judges can be assumed to be more familiar with competition law doctrines and issues than their US counterparts.25 This aura of technocracy in EU competition law enforcement has not been lessened following the modernisation of the early 2000s, where the place of economic thinking and economic argumentation was elevated as an enforcement philosophy, over more traditional legalistic approaches.26 As discussed in chapter three, a relative consensus currently exists over the normative objective of EU competition law, ie the consumer welfare paradigm, which lessens the need for a democratic oversight over the enforcement work of the technocrats. The issue of anti-trust and competition law enforcement has thus for many decades been more bent towards a technocratic modality than a democratic modality. But, as explained above, this is a matter of degrees, not of absolutes. Considering this analysis about regulation of competition law in the substantive sense, one could ask if the regulation of competition law in the procedural sense could also benefit from a more technocratic approach. The basic parameters for successfully making such an argument are that the normative objective of the procedural regime is uncontested and relatively clear, which would make the task about the technicalities of finding optimal means, rather than about the political task of defining ends. Viewed broadly, the normative objective of procedural

22 Crane explains the ideological conflict over how anti-trust should be regulated by comparing the views of Theodore Roosevelt and Woodrow Wilson. Roosevelt urged the replacement of the model created with the Sherman Act with an administrative model where specialised experts took complex decisions about trusts instead of generalist courts. Wilson, however, feared that corporations would infiltrate such administrative agencies and thus bend their output in their favour. Thus, he argued for an administrative agency that had prohibitory powers, not structural powers, which were under judicial supremacy. Wilson eventually won the 1912 presidential election, and the Federal Trade Commission was established in line with his vision. See Crane (n 9) 14–20. 23 Crane (n 9) 89. 24 For a detailed discussion about the origins of EU competition law see David Gerber, Law and Competition in the Twentieth Century Europe: Protecting Prometheus (Oxford, Oxford University Press, 2001) ch 9. See also Haukur Logi Karlsson, A quantitative quest for philosophical fairness in EU’s ­competition procedure, PhD thesis (European University Institute, 2017) 170–82. 25 When comparing the US and the EU systems of anti-trust enforcement, Crane concludes that while much of US anti-trust law ‘is made by generalist judges or juries …, European competition law is primarily the product of expert or semi-expert deliberation’. See Crane (n 9) 194–98. 26 Anne Witt discusses the implications of the modernisation program in Anne C Witt, The more economic approach to EU antitrust law (Oxford, Hart Publishing, 2016) 253–309.

170  An Argument for a Data-based Democracy fairness is relatively uncontested, but due to its lack of practical precision, differently situated stakeholders make different claims about what the concept entails in competition proceedings. This lack of normative clarity would initially recommend a democratic approach to the issue of regulating competition procedures. However, the conceptualisation presented in this book opens a possibility for integrating technocratic methods into the decisional process by extracting a democratic view on how the competing considerations should be balanced in the procedural design. Thinking of democracy as a method for establishing a normative agenda in society using popular preferences as an input does not exclude a more technocratic method for verifying and analysing the inputs. If democracy is an uncontested and welldefined end for taking certain decisions, the means of how democracy is practised does not necessarily exclude a technocratic method. Direct democracy, representative democracy and deliberative democracy are prominent means for achieving the normative end at the core of the democratic idea, ie to decide a normative social agenda based on popular preferences. The suggestion of this book is that technocratic means could be superior for extracting and processing popular preferences for enabling an eventual decision about a normative social agenda, in this case about the calibration of procedural accuracy and procedural efficiency in EU competition procedure. This would not be at odds with the ideal of democracy; it would be a contemporary manifestation of democracy, informed by modern data, statistical and computational techniques. These technocratic means, towards realising the democratic ideal in modern governance, could be referred to as data-based democracy. This book concerns the special case of procedural fairness in EU competition law, but the arguments made with regards to conceptualisation and methodology could potentially be scaled up to fit a wider context, such as general regulatory decision-making in the EU. Looking beyond the topic of book, a further topic of research could be to ask which is the best way to maintain the ideal of democracy in a society of around 500 million people, living in 27 Member States, each with its distinct political culture, and speaking 24 official EU languages? Is it representative democracy, which is vulnerable to the influences of organised lobbying of special interest, at the cost of the common good? Is it deliberative democracy, in which an informed discourse is established between the political elites and the diverse and vast populations of the EU, simultaneously in 24 official EU languages?27

27 Habermas has argued that a common European wide political public sphere is needed for the EU citizenship to succeed: ‘a communicative network extending across national boundaries and specializing in the relevant questions – is of central importance for the emergence of such a European identity …. It is not enough that common policies are institutionalized in Brussels and Strasbourg and that the European citizens could influence these policies through the election of a parliament with its own factions.’ See Jürgen Habermas, Europe: the faltering project (Ciaran Cronin tr, Cambridge, Polity Press, 2009) 87. See also Jürgen Habermas, The Crisis of the European Union: a response (Ciaran Cronin tr, Cambridge, Polity Press, 2012) 48.

Data-based Democracy for Regulating EU Competition Procedure  171 Or is it data-based democracy, in which modern techniques are used for establishing equitable equilibriums based on what the population truly wants? Perhaps, a data-based democracy could regenerate the democratic ideal in the twenty-first century, just as representative democracy did in the seventeenth century, and universal suffrage in the twentieth century.28

28 For a comprehensive overview of the development of how democracy has been practised and perceived through different eras, see Bernard Manin, The principles of representative government (Cambridge, Cambridge University Press, 1997); David Van Reybrouck, Against elections: the case for democracy (Liz Waters tr, London, The Bodley Head, 2016).

BIBLIOGRAPHY Adler, Michael, ‘Understanding and Analysing Administrative Justice’ in Michael Adler (ed), Administrative Justice in Context (Oxford, Oxford, Hart Publishing, 2009). Adler, Matthew D and Posner, Eric A, New Foundations of Cost–Benefit Analysis (Cambridge MA, Harvard University Press, 2006). Akin, William E, Technocracy and the American dream: the technocrat movement, 1900–1941 (Berkeley, University of California Press, 1977). Alexy, Robert, A Theory of Constitutional Rights (Julian Rivers tr, Oxford, Oxford University Press, 2002). Andersson, Helene, Dawn raids under challenge: due process aspects on the European Commission’s dawn raid practices (Oxford, Hart Publishing, 2018). Andreangeli, Arianna, ‘Collective redress in EU competition law: an open question with many possible solutions (2012) 35 World Competition 529. Arrow, Kenneth J, Social choice and individual value, 2nd edn (New York, John Wiley & Sons, 1963). Aumann, Robert J, ‘Agreeing to Disagree’ (1976) 4 Annals of Statistics 1236. Austin, JL, How to do things with words (Oxford, The Clarendon Press, 1962). Binmore, Ken, Game Theory and the Social Contract Vol II: Just Playing (Cambridge MA, The MIT Press, 1998). —— Game Theory and the Social Contract Vol. I: Playing Fair (Cambridge MA, The MIT Press, 1994). —— Natural Justice (Oxford, Oxford University Press, 2005). Boyd, Robert and Richerson, Peter J, ‘Culture and the evolution of human cooperation’ (2009) 364 Philosophical Transactions of the Royal Society B: Biological Sciences 3281. Bronckers, Marco and Vallery, Anne, ‘Fair and effective competition policy in the EU: which role for authorities and which role for the courts after Menarini?’ (2012) 8 European Competition Journal 283. Burazin, Luka, ‘Can there be an artefact theory of law?’ (2016) 29 Ratio Juris 385. Burnor, Richard and Raley, Yvonne, Ethical Choices: an introduction to moral philosophy with cases, 2nd edn (Oxford, Oxford University Press, 2017). Cappelletti, Mauro and Garth, Bryant G, ‘Introduction – Policies, Trends and Ideas in Civil Procedure’ in Mauro Cappelletti (ed), International Encyclopaedia of Comparative Law – Volume XVI – Civil Procedure (Leiden, Martinus Nijhoff Publishers, 1987). Castillo de la Torre, Fernando, ‘Evidence, proof and judicial review in cartel cases’ (2009) 32 World Competition 505. Centeno, Miguel Angel, ‘The new Leviathan: the dynamics and limits of technocracy’ (1993) 22 Theory and Society 307. Cooter, Robert and Ulen, Thomas, Law & Economics, 6th edn (New York, Pearson, 2012). Crane, Daniel A, The institutional structure of antitrust enforcement (Oxford, Oxford University Press, 2011). Dahl, Robert, Democracy and its critics (New Haven, Yale University Press, 1991). Damaska, Mirijan R, The faces of justice and state authority (New Haven, Yale University Press, 1986). Dawkins, Richard, The Selfish Gene (Oxford, Oxford University Press, 1976). de La Serre, Eric Barbier, ‘Procedural justice in the European Community case-law concerning the rights of the defence: essentialist and instrumental trends (2006) 12 European Public Law 225.

Bibliography  173 —— ‘Standard of review in competition law cases: Posten Norge and beyond’ in Carl Baudenbacher, Philipp Speitler and Bryndís Pálmarsdóttir (eds), The EEA and the EFTA Court: decentred integration: to mark the 20th anniversary of the EFTA Court (Oxford, Hart Publishing, 2014). Dworkin, Ronald, ‘Hart’s Postscript and the Character of Political Philosophy’ (2004) 24 Oxford Journal of Legal Studies 1. —— ‘No right answer?’ (1978) 53 New York University Law Review 1. —— ‘Principle, Policy, Procedure’ in Ronald Dworkin, A Matter of Principle (Oxford, Clarendon Press 1986). —— ‘Rights as Trumps’ in Jeremy Waldron (ed), Theories of Rights (Oxford, Oxford University Press 1984). —— Justice for hedgehogs (Cambridge MA, Harvard University Press, 2011). —— Law’s empire (Oxford, Hart Publishing, 1986). —— Taking Rights Seriously (London, Oxford, Duckworth, 1977). Einarsson, Ólafur Jóhannes, ‘EC competition law and the right to a fair trial’ (2006) 25 Yearbook of European Law 555. Elster, Jon, Nuts and bolts for the social sciences (Cambridge, Cambridge University Press, 1989). Endicott, Timothy, ‘Law is necessarily vague’ (2001) 7 Legal Theory 379. Ezrachi, Ariel, ‘Sponge’ (2015) Working Paper CCLP (L) 42. Finnis, John, Natural law and natural rights (Oxford, Clarendon Press, 1980). Fisher, Frank, Citizens, experts, and the environment: the politics of local knowledge (Durham, Duke University Press, 2000). Flattery, Jaime, ‘Balancing efficiency and justice in EU competition law: elements of procedural fairness and their impact on the right to a fair hearing’ (2010) 7 The Competition Law Review 53. Forrester, Ian S, ‘Due process in EC competition cases: a distinguished institution with flawed procedures’ (2009) 34 European Law Review 817. Frank, Robert H ‘Why is cost–benefit analysis so controversial?’ (2000) 29 The Journal of Legal Studies 913. Fuller, Lon L and Winston, Kenneth I, ‘Forms and Limits of Adjudication’ (1978) 92 Harvard Law Review 353. Fuller, Lon L and Winston, Kenneth I, ‘Forms and Limits of Adjudication’ (1978) 92 Harvard Law Review 353. Fuller, Lon, The morality of law, rev edn (New Haven, Yale University Press, 1969). Gauthier, David, Morals by Agreement (Oxford, Oxford University Press, 1986). Geradin, Damien, ‘Collective redress for antitrust damages in the European Union: is this reality now?’ (2015) 22 George Mason Law Review 1079. Gerber, David, ‘The Transformation of the European Community Competition Law?’ (1994) 35 Harvard International Law Journal 97. —— Law and Competition in the Twentieth Century Europe: Protecting Prometheus (Oxford, Oxford University Press, 2001). Greenberg, Mark, ‘How Facts Make Law’ (2004) 10 Legal Theory 157. Habermas, Jürgen, Europe: the faltering project (Ciaran Cronin tr, Cambridge, Polity Press, 2009). —— The Crisis of the European Union: a response (Ciaran Cronin tr, Cambridge, Polity Press, 2012). —— Theory and Practice (John Viertel tr, Boston, Beacon Press, 1973). —— Theory of Communicative Action (Thomas McCarthy tr, Boston, Beacon Press, 1984). Harsanyi, John, ‘Games with Incomplete Information Played by Bayesian Players I–III’ (1967–1968) 14 Management Science (I)159; (II) 320; (III) 486. Hart, HLA ‘The Presidential Address: Prolegomenon to the Principles of Punishment’ (1959) 60 Proceedings of the Aristotelian Society 1. —— ‘Positivism and the Separation of Law and Morals’ (1958) 71 Harvard Law Review 593. —— The concept of law, 2nd edn (Oxford, Oxford University Press, 1994). —— The concept of law, 3rd edn (Oxford, Clarendon Press, 2012).

174  Bibliography Hauger, Nils FW and Palzer, Christoph, ‘Investigator, prosecutor, judge … and now plaintiff? The Leviathanian role of the European Commission in the light of fundamental rights’ (2013) 36 World Competition 565. Himma, Kenneth Einar, ‘Inclusive Legal Positivism’ in Jules L Coleman and Scott Shapiro (eds), The Oxford Handbook of Jurisprudence and Philosophy of Law (Oxford, Oxford University Press, 2004). Hobbes, Thomas, Leviathan, 2nd edn (Richard Tuck ed, Cambridge, Cambridge University Press, 1996). Hume, David, Treaties of Human Nature (David Fate Norton and Mary J Norton eds, Oxford, Oxford University Press, 2007). Jaeger, Marc, ‘The standard of review in competition cases involving complex economic assessments: towards the marginalisation of the marginal review? (2011) 2 Journal of European Competition law & Practice 295. Jónsson, Ólafur Páll, ‘Vagueness, interpretation, and the law’ (2009) 15 Legal Theory 193. Juska, Zygimantas, The role of collective redress actions to achieve full compensation for violations of European Union competition law, PhD thesis (Leiden University, 2019). Kagan, Robert A., ‘The Organisation of Administrative Justice Systems: The Role of Political Mistrust’ in Michael Adler (ed), Administrative Justice in Context (Oxford, Hart Publishing, 2009). Kahneman, Daniel and Tversky, Amos, ‘Advances in Prospect Theory: Cumulative Representation of Uncertainty’ (1992) 5 Journal of Risk and Uncertainty 297. —— ‘Prospect Theory: An Analysis of Decision under Risk’ (1979) 47 Econometrica 263. Kahneman, Daniel, Thinking fast and slow (London, Penguin Books, 2012). Kant, Immanuel, Groundwork for the Metaphysics of Morals, 2nd edn (Mary Gregor and Jens Timmermann eds, Cambridge, Cambridge University Press, 2012). Kelsen, Hans, Pure theory of Law (Berkley, University of California Press, 1967). Kornhauser, Lewis A, ‘On Justifying Cost–Benefit Analysis’ (2000) 29 The Journal of Legal Studies 1037. Köszegi, Botond and Rabin, Matthew, ‘A model of reference-dependent preferences’ (2006) CXXI The Quarterly Journal of Economics 1133. —— ‘Reference-dependent Consumption Plans’ (2009) 99 American Economic Review 909. —— ‘Reference-dependent Risk Attitudes’ (2007) 97 American Economic Review 1047. Laudan, Larry, ‘The Rules of Trial, Political Morality, and the Cost of Error: Is Proof Beyond Reasonable Harm Doing More Harm than Good?’ in Leslie Green and Brian Leiter (eds), Oxford Studies in Philosophy of Law: Volume I (Oxford, Oxford University Press, 2011). Leiter, Brian, ‘Legal Positivism About the Artifact Law: A Retrospective Assessment’ in Luka Burazin, Kenneth Einar Himma and Corrado Roversi (eds), Law as an Artifact (Oxford, Oxford University Press, 2018). —— ‘The end of Empire: Dworkin and the Jurisprudence in the 21st Century’ (2005) 36 Rutgers Law Journal 165. Lenaerts, Koen and Vanhamme, Jan, ‘Procedural rights of private parties in the community administrative process (1997) 34 Common Market Law Review 531. Lianos, Ioannis, ‘Some Reflections on the Question of the Goals of EU Competition Law’ (2013) UCL CLES Working Paper Series 3/2013. Locke, John, Two Treaties of Government (Peter Laslett ed, Cambridge, Cambridge University Press, 1988). Machiavelli, Niccolò, The Prince (Quentin Skinner and Russel Price eds, Cambridge, Cambridge University Press, 1988). Manin, Bernard, The principles of representative government (Cambridge, Cambridge University Press, 1997). Marmor, Andrei, ‘Exclusive Legal Positivism’ in Jules L Coleman and Scott Shapiro (eds), The Oxford Handbook of Jurisprudence and Philosophy of Law (Oxford, Oxford University Press, 2004). McCarty, Nolan and Meirowitz, Adam, Political Game Theory (Cambridge, Cambridge University Press, 2007).

Bibliography  175 Miller, Seumas, The moral foundations of social institutions (Cambridge, Cambridge University Press, 2010). Montage, Frank, ‘The case for a radical reform of the infringement procedure under regulation 17’ (1996) 17 European Competition Law Review 428. Monti, Giorgio, EC Competition Law (Cambridge, Cambridge University Press, 2007). Monti, Mario, ‘Fighting Cartels Why and How? Why should we be concerned with cartels and collusive behaviour?’ (2000) 3rd Nordic Competition Policy Conference, Stockholm. Moore, Alfred, Critical elitism: deliberation, democracy, and the problem of expertise (Cambridge, Cambridge University Press, 2017). Mousourakis, George, Roman Law and the Origins of the Civil Law Tradition (Cham, Springer, 2015). Nazzini, Renato, ‘Administrative enforcement, judicial review and fundamental rights in EU competition law: a comparative contextual-functionalist perspective’ (2012) 49 Common Market Law Review 971. —— ‘Administrative enforcement, judicial review and fundamental rights in EU competition law: a comparative contextual-functionalist perspective’ (2012) 49 Common Market Law Review 971. —— The Foundations of European Union Competition Law: The Objective and Principles of Article 102 (Oxford, Oxford University Press, 2011) 32–50. Oliver, Peter, ‘“Diagnostics” – a judgment applying the Convention of Human Rights to the field of competition’ (2012) 3 Journal of European Competition Law & Practice 163. Pettit, Philip, ‘Depoliticising Democracy’ (2004) 17 Ratio Juris 52. Pielke, Roger A, ‘When scientist politicize science: making sense of controversy over The Sceptical Environmentalist’ (2004) 7 Environmental Science & Policy 405. Piketty, Thomas, Capital in the twenty-first century (Arthur Goldhammer tr, Cambridge MA, Belknap Press of Harvard University Press, 2014). Posner, Richard A, Economic Analysis of Law, 6th edn (New York, Aspen Publishers, 2003). Rawls, John, A theory of justice, 2nd edn (Cambridge MA, Belknap Press, 1999). —— A theory of justice (Cambridge MA, The Belknap Press of Harvard University Press, 1971). Rousseau, Jean-Jacques, A Discourse on the Basis and Origin of inequality Among Men (Boston, Bedford/St Martin’s, 2010). —— The Social Contract (Christopher Betts tr, Oxford, Oxford University Press, 1994). Sartor, Giovanni, ‘The logic of proportionality: reasoning with non-numerical magnitudes’ (2013) 14 German Law Journal 1419. Savage, Leonard J, The Foundation of Statistics, 2nd edn (New York, Dover Publications, 1972). Schweitzer, Heike, ‘Judicial review in EU competition law’ in Damien Geradin and Ioannis Lianos (eds), Handbook on European competition law (Cheltenham, Edward Elgar Publishing, 2013) 491. Searle, John, Making of the social world: the structure of human civilization (Oxford, Oxford University Press, 2010). —— The construction of social reality (London, Penguin, 1995). Sen, Amartya, ‘The Discipline of Cost–Benefit Analysis’ (2000) 29 The Journal of Legal Studies 931. Shapiro, Scott J, Legality (Cambridge MA, Belknap Press, 2011). Sibony, Anne-Lise, ‘Annotation of the judgment of the Court in Case C-272/09 P KME Germany and others v. Commission’ (2012) 49 Common Market Law Review 1977. Skyrms, Brian, Evolution of the Social Contract, 2nd edn (Cambridge, Cambridge University Press, 2014). —— The Stag Hunt and the Evolution of the Social Structure (Cambridge, Cambridge University Press, 2004). Slater, Donald, Thomas, Sébastien and Waelbroeck, Denis, ‘Competition law proceedings before the European Commission and the right to a fair trial: no need for reform?’ (2008) The Global Competition Law Centre Working Paper Series – Working Paper No 04/08. —— ‘Competition law proceedings before the European Commission and the right to a fair trial: no need for reform?’ (2009) 5 European Competition Journal 97. Solum, Lawrence B, ‘Procedural Justice’ (2004) 78 Southern California Law Review 181.

176  Bibliography Sunstein, Cass, ‘The Limits of Quantification’ (2014) 102 California Law Review 1369. —— ‘The Storrs Lectures: Behavioural Economics and Paternalism’ (2013) 122 Yale Law Journal 1826. Toumela, Raimo, The philosophy of social practices: a collective acceptance view (Cambridge, Cambridge University Press, 2002). Tullock, Gordon, Trials on Trial: Pure Theory of Legal Procedure (New York, Columbia University Press, 1980). Van Reybrouck, David, Against elections: the case for democracy (Liz Waters tr, London, The Bodley Head, 2016). Vandenborre, Ingrid and Goetz, Thorsten, ‘EU competition law procedures’ (2012) 3 Journal of European Competition Law & Practice 578. von Neumann, John and Morgenstern, Oskar, The Theory of Games and Economic Behavior, 3rd edn (Oxford, Oxford University Press, 1954). Waelbroeck, Denis and Fosselard, Denis, ‘Should the decision-making power in EC antitrust procedures be left to an independent judge? – The impact of the European Convention of Human Rights on EC antitrust procedures’ (1994) 14 Yearbook of European Law 111. Waelbroeck, Denis, Slater, Donald and Even-Shoshan, Gil, ‘Study on the conditions of claims for damages in case of infringement of EC competition rules’ (Ashurts, 2004). Waller, Spencer Weber and Popal, Olivia. ‘The fall and rise of the antitrust class action’ (2016) 39 World Competition 29. Whelan, Peter, ‘Cartel criminalization and the challenge of “moral wrongfulness”’ (2013) 33 OJLS 535. Wils, Wouter PJ, ‘EU anti-trust enforcement powers and procedural rights and guarantees: the interplay between EU law, national law, the Charter of Fundamental Rights of the EU and the European Convention on Human Rights’ (2011) 34 World Competition 189. —— ‘The combination of the Investigative and prosecutorial function and the adjudicative function in EC antitrust enforcement: a legal and economic analysis’ (2004) 27 World Competition 201. —— ‘The compatibility with fundamental rights of the EU antitrust enforcement system in which the European Commission acts both as investigator and as a first-instance decision maker’ (2014) 37 World Competition 5. —— ‘The increased level of EU antitrust fines, judicial review and the ECHR’ (2010) 33 World Competition 5. Witt, Anne C, The more economic approach to EU antitrust law (Oxford, Hart Publishing, 2016). Wu, Tim, The curse of bigness: antitrust in the new gilded age (New York, Columbia Global Reports, 2018).

INDEX NB–page locators in bold indicate information found in tables adjudicative procedure: balancing efficiency and justice, 18–19, 38–39, 40–41, 45, 51–52, 148 fairness, 44–45, 147 institutional structure, 26–27, 148 constitutive rules, 27–29 culture, 29–33 sanctions, 33–36 normative function, 36, 43–44, 148 consequential view, 39–43 deontological view, 37–39 punishment, 33–36 administrative enforcement, 2–3 Commission, 60, 61, 72 courts, 11, 17 Court of Justice of the EU, 60–61, 69 European Court of Human Rights, 11–12, 13 KME/Chalkor dilemma, 7–10, 120, 123–24, 125, 126, 151 administrative sanctions, 34 Alexy, Robert: proportionality test, 115 cost-benefit analysis, 115–16 balancing efficiency and justice, 74, 94–95, 116–17, 148–51 adjudicative procedure, 18–19, 38–39, 40–41, 45, 51–52, 148 collective redress, 136, 137, 141, 143–44 Groupe Gascogne cases, 133, 157, 159, 164 KME/Chalkor dilemma, 122–26, 157, 159, 164 preferences of stakeholders, 153, 158 Binmore, Kenneth: bargaining, 89–90 cooperative equilibriums, 86–87 Pareto’s optimal equilibriums, 91–92 game theory, 86–87 influences, 87–90 Pareto’s optimal equilibriums, 87–88, 90 justice as fairness, 88–89

cartel criminalisation, 2, 7–8, 121, 147 cartel fines, 4 Regulation 1/2003, impact of, 5 Charter of Fundamental Rights of the EU (CFREU), 2, 6–7, 12–14, 119 collective redress, 143, 163 Groupe Gascogne cases, 128–29 KME/Chalkor cases, 120–22, 125–26 right to a fair trial, 10, 17 collective redress for private enforcement, 119, 135–37, 153–54, 155–56, 163–64 accuracy/cost-efficiency dilemma, 137–40 balancing efficiency and justice, 136, 137, 141, 143–44 fairness, 140–44 Charter of Fundamental Rights of the EU, 143, 163 preference matrix, 160, 160 compatibility of competition law and procedural fairness, 10–11, 71–74 competition law procedures, 46–47, 148–49 Court of Justice of the EU, 68–69 instrumental function, 61–62 normative function, 47–51, 61–62 constitutive rules, 62–65 culture of DG Competition, 66–68 institutional sanctions, 69–71 procedural fairness arguments, 71–74 stakeholders, 51–53 stakes, 51–52, 54–56 structure, 56–61 constitutional project of the EU, 5–7 cost-benefit analyses, 41, 44, 54–56, 59–60, 150, 154 proportionality test, 115–16 Court of Justice of the EU (CJEU), 3–4 administrative enforcement, 60–61, 69 competition law procedures, 68–69 deference, 17 functional separation of powers, 16 KME/Chalkor case, see KME/Chalkor case

178  Index criminal sanctions, 34–35, 70, 122 criminalisation of competition law breaches: cartels, 2, 7–8, 121, 147 concerns about increased criminalisation policies, 7–10 European Court of Human Rights, 11–12 KME/Chalkor cases, 7–9 sanctions, 34–35, 70, 122 standard of criminality, 7 degrees of criminality, 14–15 Damages Directive 2014, 136, 138, 140–41, 143–44, 145, 152, 153–54, 163–64 deference doctrine, 11–12, 14, 16, 17, 120–23 democracy: data-based democracy, 167–71 democratic contractarian society, 93 see also social contract theory legal positivism, 94–95 legislative process, 98 technocracy as an alternative, 165–67 Dworkin, Ronald, 37–39, 43, 44, 76, 79, 103, 144 EFTA Court, 4 deference doctrine, 14, 17 Posten Norge case, see Posten Norge case egalitarian distributional fairness, 89–90, 91, 93, 94, 95, 97, 99–100, 116, 149 enforcement of competition law, 2–3, 146–47 administrative enforcement, 2–3 Commission, 60, 61, 72 courts, 11–12, 13, 17, 60–61, 69 KME/Chalkor dilemma, 7–10, 120, 123–24, 125, 126, 151 administrative law origins, 8, 17 collective redress for private enforcement, 119, 135–37, 153–54, 155–56, 163–64 accuracy/cost-efficiency dilemma, 137–40 fairness, 140–44 preference matrix, 160, 160 criticisms, 2 fines and enforcement procedures, 9–10 private enforcement, collective redress for, 119, 135–37, 153–54, 155–56, 163–64 accuracy/cost-efficiency dilemma, 137–40 fairness, 140–44 preference matrix, 160, 160 Regulation 1/2003, 4–5 Regulation 17/62/EEC, 7, 11, 22 Treaty of Rome, 6–7

ethics: ethical implications of procedural fairness, 75, 94–95 legal theory, 75–76 legal naturalism, 80 legal positivism, 76–79 social contract theory, 80–86 democracy, 93 game theory, 86–90 morality in cooperation, 90–93 European Convention on Human Rights (ECHR): deference doctrine, 11–12, 17 right to a fair trial, 2, 9, 10–17, 120, 124–27 European Court of Human Rights (ECtHR), 4 administrative enforcement, 11–12, 13 criminal and administrative procedure distinguished, 9–10 criminalisation of competition law breaches, 11–12 Jussila case, see Jussila case KME/Chalkor cases, 3–4 interpretation concerns, 7–10 Menarini case, see Menarini case fairness in law, 2 balancing, see balancing efficiency and justice fair rules, see model fair rules spending on fairness, 150 distribution of the error cost, 150 see also model fair rules fines and enforcement procedures, 9–10 see also enforcement of competition law fundamental rights and competition law, 2–3, 6, 146–47 KME/Chalkor cases, 8, 9–10, 123 game theory: Binmore, 86–87 bargaining, 89–90 cooperative equilibriums, 86–87 influences, 87–90 justice as fairness, 88–89 Pareto’s optimal equilibriums, 87–88, 90 Gauthier, David, 86 Hobbesian morality, 86 Gauthier, David, 86 Groupe Gascogne cases, 128–30, 151, 153, 155 analysis, 130–33

Index  179 balancing efficiency and justice, 133, 157, 159, 164 fairness, 133–35 Charter of Fundamental Rights of the EU, 128–29 Hart, H.L.A., 25, 76–79, 94, 144 hierarchy of laws, 6 Hobbes, Thomas, 81–82 game theory, relationship with, 86–90 Rawls’ theories distinguished, 84–85 selfish reciprocity, 90–91 Hume, David, 78–80 human cooperation: Hobbes: selfish reciprocity, 90–91 Rousseau, 90 human society: see also democratic contractarian society fair rules, 96–97 role in shaping law, 1–2 judicial discretion, 30 judicial review, 10, 11–12, 15, 70–71, 168–69 KME/Chalkor cases, 120–21, 122, 126–28 Jussila case, 9, 11–12, 123–26 degrees of criminality, 13–16 proportionality test, 125 Kaldor-Hicks efficiency, 49–51, 99–100, 115, 116 Kant, Immanuel, 83–84, 86, 88, 90–91, 93, 103 Kelsen, Hans, 76, 77 KME/Chalkor cases, 3–4 background, 4, 120 EU’s constitutional project, impact of, 5–7 Regulation 1/2003, impact of, 4–5 balancing administrative and criminal sanctions, 122–26 cartels as breaches against public interest, 8 Charter of Fundamental Rights of the EU, 120–22, 125–26 classification of competition proceedings, 120 criminalisation of competition law, 7–9 deference doctrine, 120–22 degrees of criminality, 14–15 European Court of Human Rights: interpretation concerns, 7–10 fairness, 127–28 fines and enforcement procedures, 9–10 judgment, 13–14

reinforcement: Schindler Holding and Others case, 126–27 rise of fundamental rights, 7–9 legal positivism, 1, 76–77, 79–80, 94–95, 149 legal naturalism, 1, 25, 76, 80, 95, 149 see also social contract theory legal realism, 1 legislative guidance on procedural fairness, lack of, 144–45 Lisbon Treaty, 6, 12, 13, 48, 119 Locke, John, 82, 84, 86, 90–91 Machiavelli, Niccolò, 81 Menarini case, 4, 13, 126–27, 146–47 degrees of criminality, 14–16 model fair rules, 101 application, 151–52 identification of policy objective, 152–54 identifying fairness, 160–63 preference function, 154–56 preference index, 156–59 preference matrices, 159–60 assessing utility, 115–17, 163–64 identification of fair and efficient rules, 112–15 interpersonal comparison of stakeholder claims, 105–11 policy objectives of a rule, 102–3 registration of preferences, 111–12 unification of moral and efficiency claims, 103–5 utility, 115–17 moral entitlement to fair process, 7, 91, 133–34 morality and moral behaviour, 90–91, 92–93, 95 morality of law, 76, 80 natural law, see legal naturalism Pareto efficiency, 49–51, 115, 116 optimal equilibriums, 87–88, 90, 91–92 positivism, see legal positivism Posten Norge case, 4, 14, 146–47 degrees of criminality, 14–15 preference indexes, 110 model of fair rules, 156–58, 158 preference matrices, 111 model of fair rules, 159–60, collective redress, 160 Groupe Gascogne cases, 159 KME/Chalkor cases, 159

180  Index private enforcement: collective redress, 119, 135–37, 153–54, 155–56, 163–64 accuracy/cost-efficiency dilemma, 137–40 fairness, 140–44 preference matrix, 160, 160 procedural law: substantive law distinguished, 20–22 competition law, 22–23 criminal justice system, 22–23 procedures as a social institution, 23–26 adjudicative procedure, see adjudicative procedure proportionality test: cost-benefit analysis, 115–16, 115 Jussila case, 125 prospect theory, 101, 108 punishment: adjudicative procedures, 33–36 administrative sanctions, 34 criminal sanctions, 34–35, 70, 122 deterrence rationale, 8 formal and informal sanctions, 26, 31, 33–36, 70 informal sanctions, 26 institutional sanctions, 147, 148 adjudicative procedures, 33–36 retributive rationale, 7 seriousness, 8 social sanctions, 33–34 Rawls, John, 23–24 Binmore, influence on, 88 deontology, 37–39 social contract theory, 80 Hobbes, 81–82, 91 influences, 81–86 Kant, 83, 91 Locke, 82, 90 Machiavelli, 81 Rousseau, 82–83 utilitarian theories of justice, criticisms of, 84–86 veil of ignorance, 107–8 Regulation 1/2003, 2 criminal law, 8, 122 deference doctrine, 121 entry into force, 4–5 instrumental and normative function, 63–64 review of competition decisions, 68, 127

Regulation 17/62/EEC, 2 competition enforcement, 7, 11, 22 criminal law, 8 Regulation 773/2004: instrumental and normative function, 64–65 right to a fair trial, 2, 9, 10–17, 120, 124–27 Rousseau, Jean-Jacques, 82–83, 84, 86, 90–91 rules and rule-making, 96–97 efficiency maximum, 99–100 egalitarian distributional principle, 97, 99–100 ethical duties of rule-makers, 97–100 justification for intervention, 99 model fair rules, 101 identification of fair and efficient rules, 112–15 interpersonal comparison of stakeholder claims, 105–11 policy objectives of a rule, 102–3 registration of preferences, 111–12 unification of moral and efficiency claims, 103–5 utility, 115–17 prospect theory, 101 ranking priorities and options, 100–1 sanctions, see punishment selfish gene, 87 selfish reciprocity, 90–91 Skyrms, Brian, 86 social contract theory: Binmore: bargaining, 89–90 cooperative equilibriums, 86–87 justice as fairness, 88–89 Pareto’s optimal equilibriums, 87–90 egalitarian distributional principle, 97 game theory: Binmore, 86–90 Hobbesian morality, 86 morality in cooperation: Binmore, 90–93 Rawls: Hobbes, 81–82 influences, 81–86 Kant, 83 Locke, 82 Machiavelli, 81 Rousseau, 82–83

Index  181 utilitarian theories of justice, criticisms of, 84–86 rules, 96–97 ethical duties of rule-makers, 97–100 social institutions: collective acceptance approach, 25 procedures as, 23–26 teleological approach, 25–26 social sanctions, 33–34 stakeholders in competition proceedings, 52–54 preference trends, 55 stakes in competition proceedings, 54–56 substantive law: procedural law distinguished, 20–22 competition law, 22–23 criminal justice system, 22–23 supremacy doctrine, 6 technocracy: rule/law-making, 164–67

Treaty of Rome, 5–7, 22 Treaty on European Union (TEU): Article 17, 66, 71 Article 19, 68, 71–72, 143, 163 Treaty on the Functioning of the EU (TFEU): Article 101, 5, 10, 20, 22, 49–50, 56–59, 62, 66–68, 72, 138, 141, 142–43 Article 102, 10, 20, 22, 56–59, 62, 66–68, 72, 138, 141, 142–43 Article 103, 20, 62 Article 245, 71 Article 250, 66 Article 253, 60, 68 Article 254, 68 Article 256, 60 Article 261, 127 Article 263, 60, 68, 71, 121, 127 Article 268, 130, 132, 134 Article 281, 60 Article 340, 130, 132, 134

182