The Oxford Handbook of International Antitrust Economics 9780199859191, 0199859191

Citation preview

T H E O X F O R D HA N D B O O K O F

I N T E R NAT IONA L A N T I T RU ST E C ON OM IC S Volume 1

CONSULTING EDITORS Michael Szenberg

Lubin School of Business, Pace University Lall Ramrattan

University of California, Berkeley Extension

THE OXFORD HANDBOOK OF

INTERNATIONAL ANTITRUST ECONOMICS Volume 1 Edited by

ROGER D. BLAIR and

D. DANIEL SOKOL

1

3 Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford New  York Auckland  Cape Town  Dar es Salaam  Hong Kong  Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trademark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New  York, NY 10016 © Oxford University Press 2015 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. CIP data is on file at the Library of Congress ISBN 978–0–19–985919–1

1 3 5 7 9 8 6 4 2 Printed in the United States of America on acid-free paper

Contents

List of Contributors Introduction

ix xiii

PA RT I   I N T ROD U C T ION A N D I N S T I T U T IONA L I S SU E S 1. Rationales for Antitrust: Economics and Other Bases Daniel A. Crane

3

2. Antitrust Enforcement Regimes: Fundamental Differences Keith N. Hylton

17

3. Economic Analysis of Antitrust Exemptions Peter Carstensen

33

4. Healthcare Provider and Payer Markets Cory S. Capps and David Dranove

63

5. International Antitrust Institutions Oliver Budzinski

119

6. Competition Policy in Public Choice Perspective Fred S. McChesney, Michael Reksulak, and William F. Shughart II

147

7. Antitrust Settlements Daniel L. Rubinfeld

172

8. The Economics of Antitrust Class Actions Roger D. Blair and Christine Piette Durrance

187

9. Behavioral Economics and Antitrust Mark Armstrong and Steffen Huck

205

vi   Contents

10. Experimental Economics in Antitrust Wieland Müller and Hans-Theo Normann

229

11. Optimal Antitrust Remedies: A Synthesis William H. Page

254

12. Private Antitrust Enforcement in the United States and the European Union: Standing and Antitrust Injury 281 Jeffrey L. Harrison 13. Freedom to Trade and the Competitive Process Aaron Edlin and Joseph Farrell

298

PA RT I I   M ON OP OLY:  S T RU C T U R A L C ON SI DE R AT ION S 14. Monopoly and Dominant Firms: Antitrust Economics and Policy Approaches Lawrence J. White

313

15. Market Definition Louis Kaplow

345

16. Bilateral Monopoly: Economic Analysis and Antitrust Policy Roger D. Blair and Christina DePasquale

364

17. Antitrust, the Internet, and the Economics of Networks Daniel F. Spulber and Christopher S. Yoo

380

18. The Antitrust Analysis of Multisided Platform Businesses David S. Evans and Richard Schmalensee

404

PA RT I I I   M E RG E R S 19. Efficiency Claims and Antitrust Enforcement Howard Shelanski

451

20. Unilateral Effects Bryan Keating and Robert D. Willig

466

21. Coordinated Effects: Evolution of Practice and Theory Jith Jayaratne and Janusz Ordover

509

Contents   vii

22. Buyer Power in Merger Review Dennis W. Carlton, Mary Coleman, and Mark Israel

529

23. Vertical Mergers Michael A. Salinger

551

Index

587

List of Contributors

Mark Armstrong, Professor of Economics and Director of Graduate Studies, University of Oxford Roger D.  Blair, Walter J.  Matherly Professor and Chair, Department of Economics, Warrington College of Business, University of Florida Oliver Budzinski, Professor of Economic Theory, Institute of Economics, Ilmenau University of Technology Cory S. Capps, Partner, Bates White Economic Consulting; formerly Staff Economist, Antitrust Division, Department of Justice Dennis W.  Carlton, David McDaniel Keller Professor of Economics, University of Chicago Booth School of Business Peter Carstensen, Professor of Law, University of Wisconsin Mary Coleman, Executive Vice President, Compass Lexecon Daniel A. Crane, Associate Dean for Faculty and Research and Frederick Paul Furth, Sr. Professor of Law, University of Michigan Christina DePasquale, Assistant Professor, Department of Economics at Emory University David Dranove, Walter McNerney Professor of Health Industry Management, Professor of Management and Strategy, Kellogg School of Management, Northwestern University Christine Piette Durrance, Assistant Professor, Department of Public Policy, University of North Carolina at Chapel Hill Aaron Edlin, Richard Jennings Endowed Chair, Professor of Economics, Professor of Law, University of California, Berkeley David S. Evans, Lecturer, University of Chicago Law School; Executive Director and Visiting Professor, Jevons Institute for Competition Law and Economics, University College London Faculty of Laws; Chairman, Global Economics Group Joseph Farrell, Professor of Economics, University of California, Berkeley Jeffrey L.  Harrison, Stephen C.  O’Connell Professor of Law, Levin College of Law, University of Florida

x   List of Contributors Steffen Huck, Professor of Economics, University College London Keith N. Hylton, William Fairfield Warren Distinguished Professor and Professor of Law, Boston University. Mark Israel, Executive Vice President, Compass Lexecon Jith Jayaratne, Executive Vice President, Compass Lexecon Louis Kaplow, Finn M.W. Caspersen & Household International Professor of Law and Economics, Harvard Law School; Associate Director, John M.  Olin Center for Law, Economics, and Business, Harvard Law School; Research Associate, National Bureau of Economic Research Bryan Keating, Senior Vice President, Compass Lexecon Fred S. McChesney, de la Cruz–Mentschikoff Endowed Chair in Law and Economics, University of Miami School of Law Wieland Müller, Professor of Economics, Department of Economics, University of Vienna Hans-Theo Normann, Professor of Economics, Dean of the Graduate School, Düsseldorf Institute for Competition Economics (DICE), Heinrich-Heine-Universität Düsseldorf Janusz Ordover, Professor of Economics, New York University and Senior Consultant, Compass Lexecon William H.  Page, Professor of Law and Marshall M.  Criser Eminent Scholar, Levin College of Law, University of Florida Michael Reksulak, Director, Social Sciences, Arts and Humanities, Qatar Foundation, Qatar National Research Fund Daniel L. Rubinfeld, Robert L. Bridges Professor of Law and Professor of Economics, Emeritus, University of California, Berkeley and Professor of Law, New York University Michael A. Salinger, Jacqueline J. and Arthur S. Bahr Professorship in Management Markets, Public Policy and Law, School of Management, Boston University Richard Schmalensee, Howard W.  Johnson Professor of Management Emeritus and Professor of Economics Emeritus, Sloan School of Management, Massachusetts Institute of Technology Howard Shelanski, Professor of Law, Georgetown University William F. Shughart II, J. Fish Smith Professor in Public Choice, Jon M. Huntsman School of Business, Utah State University D. Daniel Sokol, Professor, Levin College of Law, University of Florida and Senior Research Fellow, George Washington University Law School Competition Law Center

List of Contributors   xi

Daniel F.  Spulber, Elinor Hobbs Distinguished Professor of International Business and Professor of Management Strategy, Kellogg School of Management, Northwestern University Lawrence J. White, Robert Kavesh Professor of Economics, Department of Economics, Stern School of Business, New York University Robert D. Willig, Professor of Economics and Public Affairs, Woodrow Wilson School of Public and International Affairs, Princeton University Christopher S. Yoo, John H. Chestnut Professor of Law, Communication, and Computer & Information Science; Director, Center for Technology, Innovation & Competition; University of Pennsylvania Law School

Introduction RO GER D. BLAIR AND D. DANIEL SOKOL

Antitrust economics is a subset of industrial organization economics. What makes antitrust economics rather unique is the centrality of economic analysis to the development of antitrust law and policy. In the United States antitrust economics guides all antitrust analysis by government enforcers (at the federal level the Department of Justice Antitrust Division and the Federal Trade Commission) and courts. In other systems, the centrality of antitrust economics to antitrust law (typically called competition law) and policy has not been established. Instead, cutting-edge antitrust economic analysis competes with non-antitrust economics goals. Nevertheless, across the major non-US jurisdictions, antitrust economics is far more utilized now than previously. With global mergers and various types of conduct, increased coordination across agencies, practitioner lawyers and economists around the world trained in the latest theories of antitrust economics, and a rise of economic analysis in decision-making by adjudicators, the increasing role of international antitrust economics seems somewhat inevitable. The desire to provide scholars and policymakers across jurisdictions a reference tool to understand the most important developments in antitrust economics motivates this handbook. We have assembled many of the most important scholars in the field to provide overviews and analysis of the core issuers in antitrust economics. Although no handbook can be exhaustive, we have attempted to cover all of what we believe to be the major topics in the field. The developments in economic analysis across these areas that the handbook covers will shape policy and legal issues in the field for some time. We hope that the handbook will provide inspiration for new avenues of theoretical and empirical research in the field. Many people deserve thanks for this book. The project took a number of years to complete. Our editors at Oxford University Press deserve our gratitude for their patience and excellent editing. Coordinating production across so many chapters was not always easy. We particularly thank those authors who turned in their work in a timely manner.

T H E O X F O R D HA N D B O O K O F

I N T E R NAT IONA L A N T I T RU ST E C ON OM IC S Volume 1

PA R T I

I N T RODU C T ION A N D I N ST I T U T IONA L I S SU E S

CHAPTER 1

R AT IONA L E S F OR A N T I T RU ST Economics and Other Bases DANIEL A. CRANE

What are the purposes or objectives of the antitrust laws? There are several ways to answer this question. To answer it descriptively, one could look at the history of antitrust enforcement and ask who the antitrust laws have actually benefited or what they have achieved over time as they have been enforced. As the economist George Stigler wrote, “The announced goals of a policy are sometimes unrelated or perversely related to its actual effect and the truly intended effects should be deduced from the actual effects” (Stigler 1975, 140). The actual beneficiaries of antitrust enforcement may be quite different from those intended by the legislative framers of the antitrust statutes or the ones proposed by normative theorists. The answer to the descriptive question is contested. Public choice scholars have claimed that the history of US antitrust enforcers shows that small, inefficient businesses are the primary beneficiaries of antitrust enforcement, often at the expense of consumers and economic efficiency (McChesney and Shughart 1995). Many antitrust advocates would contest this claim and argue that consumers have been the primary beneficiaries of antitrust enforcement. In this chapter, I will not attempt to answer the empirical question. One could also attempt to answer the rationale question as a matter of legislative intent by looking to the intentions of the framers of the Sherman Act (1890) in the United States, the Treaty of Rome (1957), reborn in 2009 as the Treaty on the Functioning of the European Union (“TFEU”), or the foundational antitrust statute of any of the more than 100 nations that currently have antitrust laws. Such an approach is highly contestable too. Claims about the intentions of the framers of the Sherman Act, for example, have been controversial. The Chicago School theorist Robert Bork claimed to locate an economic efficiency objective in the congressional minds of 1890 (Bork 1978). Subsequent scholarship refuted Bork’s claim, finding the avoidance of wealth transfers from consumers to producers or protection of small, inefficient businesses to be the predominant legislative purposes (Lande 1982, DiLorenzo 1985, Hovenkamp 1988, Hazlett 1992).

4   Daniel A. Crane A third approach to determining antitrust rationales is to engage in a free-form normative debate about the role of monopoly, competition, law, and regulation in modern society. Should we have antitrust laws and, if so, why? This is where most of the action is today. Most scholars and antitrust practitioners—at least in the established antitrust regimes, particularly the United States and Europe—tend to assume that the foundational legal instruments creating antitrust law are sufficiently open-textured and subject to interpretation to accommodate a variety of different and sometimes conflicting normative objectives. Hence, the existential purposes of antitrust law are open for continued creation and recreation by litigants, scholars, judges, and government officials. Further, whoever have been the actual beneficiaries of antitrust law in the past, most antitrust practitioners assume that the proper beneficiaries can be reached if the right substantive norms, procedural rules, and institutional constraints are enacted. Thus, objections based on the past record of antitrust enforcement have largely fallen on deaf ears. For purposes of this introductory chapter, I  will approach the rationales question from a largely normative perspective. What are the leading contemporary views on what should be the purposes of antitrust law? To begin this discussion, let us consider the foundational question of intended beneficiaries: for whose benefit should antitrust law exist?

1.1.  Who Are the Intended Beneficiaries? Antitrust differs from most fields of law in one important respect: it is not instantly clear for whose benefit the antitrust laws exist. It is readily apparent that securities laws are intended to protect investors, antidiscrimination laws are intended to protect unpopular minorities, and labor laws are intended to balance the interests of employers and employees. But who are the intended beneficiaries of antitrust law? When one poses this question to a group of students, a multiplicity of answers usually flies back toward the chalkboard: consumers, employees, competitors, independent businesspeople, shareholders, taxpayers, voters, society at large. The list goes on and on. However long the list ultimately grows, it is subject to the challenge that many of the names on the list have clearly conflicting interests with those of other names on the list. Some of these conflicts are fairly obvious. In the founding era of US antitrust law, many people assumed that controlling the monopolistic power of the corporate trusts would benefit both shareholders and consumers. But shareholders and consumers generally have opposite interests in monopoly in competition. Even if the shareholders of “victim” companies are sometimes harmed by anticompetitive conduct, as a class, shareholders are often benefited by cartelization and many other antitrust violations, whereas consumers are harmed. Some of the conflicts are less obvious. Are working-class employees generally benefited or harmed if their employers obtain market power? Some empirical work

Rationales for Antitrust   5

suggests that employees tend to earn higher wages in more concentrated markets (Oi and Idson 1999), whereas consumers pay higher prices in those markets, thus creating a clash between the interests of employees and consumers on the intensity of antitrust enforcement. In most modern antitrust systems, the interests of consumers are placed at or near the top of the beneficiaries list. Thus, the US Supreme Court has described the Sherman Act as a “consumer welfare prescription,”1 and the European Commission has adopted consumer welfare as at least a primary purpose of competition law enforcement.2 Countries that have recently added or updated antitrust laws generally list consumer welfare high on their list of objectives. Article 1 of China’s 2007 Anti-Monopoly Law lists a number of objectives but identifies only one category of beneficiary—“safeguarding the interests of consumers.” India’s 2002 Competition Law speaks in its preamble of “protect[ing] the interests of consumers,” although it also lists the “freedom of other market participants” as another class of intended beneficiary. Although most antitrust systems recognize consumer welfare as either paramount or at least coequal with any other objectives, relatively little attention is paid in foundational antitrust statutes or other instruments to how to resolve a clash between intended beneficiaries should one arise. Even where some balancing of interests is suggested, it is hard to find applications in practice. For example, Article 101 of the TFEU, which generally prohibits agreements that unduly restrict competition, allows some such restrictive agreements that “contribute[] to improving the production or distribution of goods or to promoting technical or economic progress, while allowing consumers a fair share of the resulting benefit.” However, the European Commission has interpreted this to mean that the effect of the restraint on consumers must be at least neutral—meaning that a restriction may be allowed if, although impairing competition somewhat, it advances efficiency and makes consumers no worse off. This is not really a balancing of competing interests approach so much as “first do no harm” principle.

1.2.  Efficiency and Fairness Although the normative justification for antitrust law may appear controversial to lawyers, it is relatively straightforward to economists. Economists generally begin their account with an assumption that in a competitive market, competition forces firms to price at marginal cost—the cost of creating the marginal unit of production. Marginal

1  Reiter v. Sonotone Corp., 442 U.S. 330, 343 (1979).

2  See, e.g., Philip Lowe, Director General, European Commission Directorate General for Competition, Consumer Welfare and Efficiency—New Guidelines and Principles of Competition Policy (Mar. 27, 2007), available at http://ec.europa.eu/competition/speeches/text/sp2007_02_en.pdf; Neelie Kroes, European Commission for Competition Policy, Consumer Welfare: More than a Slogan (Oct. 21, 2009), available at http://europa.eu/rapid/pressReleasesAction.do?reference=SPEECH/09/486&format​=pdf.

6   Daniel A. Crane cost pricing (putting aside the tremendous complexities of defining and identifying it) thus becomes the benchmark against which competitive effects are judged. When firms engage in anticompetitive conduct—such as collusion, monopolization, or merger—that disrupts the market’s competitiveness, the producer firms are able to raise prices above competitive levels. Depending on the elasticity of demand, price increases can produce one or both of two price effects. First, customers with elastic demand forgo purchasing the relevant goods at the higher prices. Second, customers with inelastic demand continue to purchase the higher-priced goods, thus transferring some of their wealth to the producers. These two effects correspond with the two primarily economic theories of why anticompetitive conduct may be normatively undesirable. The first effect—that some customers cease buying at the higher price—creates an output reduction and deadweight loss. Producers have to cut production in order to sustain the price increase. Economists generally agree that this effect from anticompetitive conduct is allocatively inefficient. Allocative efficiency is that state of affairs where scarce social resources are put to their most valued uses. When the price of goods increases due to anticompetitive conduct and consumers with elastic demand buy less, allocations of scarce social resources that would have occurred in a competitive market no longer occur. Consumers and producers reallocate resources to second-best preferences, which diminishes net social utility. If antitrust law can prevent these anticompetitive effects, it can increase social utility by ensuring that output remains at the higher level determined by first-order consumer preferences. That normative account of antitrust law is relatively uncontroversial in the economics profession. The second effect—wealth transfers—is trickier to map normatively. Suppose that a horizontal merger between two luxury perfume companies creates market power that permits the merging firms to increase the price of luxury perfume 10 percent. Further assume that the consumers who buy luxury perfumes are demand inelastic and therefore continue to buy in roughly the same quantity at the higher prices. In this scenario, the merger produces no deadweight losses, but it does produce wealth transfers from consumers to producers. Is that problematic? The economics profession has no comparative advantage in answering this question. On the one hand, the reallocation of wealth from consumers to producers is not obviously inefficient in the way that the deadweight losses are. The relevant deployment of scarce social resources to the production and consumption of expensive perfumes continues to occur in equal measure. True, producers are somewhat richer and consumer somewhat poorer, but so what? The economics profession has relatively little to say about which interests groups in society deserve more or less wealth. It is true that very wealthy people obtain less utility out of the marginal dollar of wealth than do the comparatively poorer, but it is far from clear that antitrust systematically redistributes income from the wealthy to the poor. In the perfume example, the consumers could be quite rich and the producers benefited by the price increase could be the shareholders of the perfume companies who turn out to be widows and orphans who have invested their life savings in mutual funds. It is very difficult to generalize from the diminishing marginal utility of money to a normative theory of antitrust law.

Rationales for Antitrust   7

Although economic theory does not support the claim that wealth transfers from consumers to producers are allocatively inefficient, the avoidance of such wealth transfers plays a dominant role in the public conception of antitrust. The general public often perceives the appropriation of consumer wealth by producers as unfair and immoral. Government enforcers who bring criminal or civil proceedings against price-fixing cartels often make public statements to the effect that the cartelists are guilty of “stealing” from the consuming public. In this vision, antitrust law creates a normative baseline in which consumers are presumptively entitled to all of the gains of trade in consumer/producer transactions, which is what occurs when competition forces producers to price at marginal cost. Producers may only lawfully appropriate any surplus of trade if they obtain market power in a “fair” way, such as by building the proverbial better mousetrap. This moral conception of antitrust is controversial within the antitrust profession. Many practitioners and scholars view antitrust as a purely utilitarian endeavor geared at maximizing social value. Professor Herbert Hovenkamp, perhaps the leading contemporary antitrust scholar in the United States, has repeatedly argued that “antitrust has no moral content” (Hovenkamp 2005, 10, 54). Others accept that antitrust must be justified in fairness terms to make it politically palatable, but largely disregard wealth transfers and fairness considerations in their own work in the field. Still others believe that wealth transfers are legitimately a concern of antitrust law, that economics is useful in demonstrating the existence and incidence of these transfers, and that moral, political, or social theory then supplies the normative account for why wealth transfers produced by anticompetitive conduct should be avoided. An additional category of inefficiency identified by the highly influential Chicago School scholar and judge Richard Posner is that the allure of monopoly rents induces firms to invest resources in chasing down a monopoly position. As Posner explains, “an opportunity to obtain a lucrative transfer payment in the form of monopoly profits will attract real resources into the effects by sellers to monopolize and by consumers to avoid being charged monopoly prices” (Posner 2001, 13–14). If antitrust law can induce firms not to engage in this rent-seeking behavior, it can reduce a wasteful expenditure and hence make society more efficient. A final efficiency that may result from antitrust enforcement is producer efficiency. As a general matter, we might expect that firms will not deliberately engage in behavior that harms their own productive efficiency (unless in the rare case in which a self-inflicted wound raises the costs of rivals by an even greater amount than it harms the firm’s own efficiency). Yet there may be cases in which monopoly or highly concentrated oligopoly lulls firms into inefficiency and a more competitive market spurs them into greater productive efficiency. Two well-known quotes come to mind on this score. First, consider Nobel laureate John Hicks’s aphorism that “the best of all monopoly profits is a quiet life” (Hicks 1935, 8). Second, consider Judge Learned Hand’s statement in the landmark Alcoa case that “[m]‌any people believe that possession of unchallenged economic power deadens initiative, discourages thrift and depresses energy; that immunity form competition is a narcotic, and rivalry is a stimulant, to industrial progress; that the spur of constant stress is necessary to counteract an inevitable disposition to let well enough

8   Daniel A. Crane alone.”3 Antitrust-induced competition may spur slothful firms to become leaner and meaner, to the benefit of the firms themselves and society as a whole.

1.3.  Two Occasional Clashes Much of the time, it does not matter whether one views all of these potential evils—deadweight losses, wealth transfers, rent-seeking behavior, and industrial sloth—as within the scope of antitrust’s purposes. An anticompetitive act will often produce all of these effects simultaneously and thus should be enjoined or sanctioned (or else the act isn’t anticompetitive at all and thus produces none of the effects and should be permitted). However, in at least two common patterns, the effects cut in different directions. In those cases, the choice between competing existential visions of antitrust may become important. The first instance involves conduct that generates wealth transfers from consumers to producers even while lessening deadweight losses by expanding output. The classic example of conduct arguably producing such an effect is price discrimination. Under at least some conditions, charging different prices to different customers based on the customers’ different willingness to pay may increase total output as compared to charging all customers the same price. At the same time, however, price discrimination results in wealth transfers from consumers to producers. Assuming that both effects were present in a particular case, how should antitrust law proceed? If one believes that antitrust law should only be concerned with allocative efficiency and that price discrimination is generally output increasing (a controversial or at least overgeneralized assumption), then conduct enabling price discrimination should be permitted. However, if one believes that antitrust law should be normatively committed to preventing wealth transfers from consumers to producers, then conduct facilitating price discrimination should be prevented even if it increases allocative efficiency. Or, at least, the interest in avoiding wealth transfers should be balanced against the interest in increasing allocative efficiency. The second type of clash arises in cases where an act—typically a merger or joint venture—simultaneously increases productive efficiency and reduces consumer welfare. Nobel laureate Oliver Williamson famously described this tension in 1968 (Williamson 1968). Suppose that a merger between two competitive petroleum firms allows the firms to build a superefficient new refinery that lowers their production costs by 20 percent. Far from deadening industrial zest as hypothesized above, this merger will dramatically increase productive efficiency. Yet suppose that because of the merger the two firms also obtain some incremental market power and are able to raise prices 5 percent. This entails both deadweight losses and wealth transfers. Should the merger be allowed? Under a total social welfare standard, the merger should probably be allowed since the producer efficiency gains far exceed the consumer welfare and allocative efficiency losses (Blair

3  U.S. v. Aluminum Co. of America, 148 F.2d 416, 427 (2d Cir. 1945).

Rationales for Antitrust   9

and Sokol 2012, Carlton 2007, Elzinga 1977, Farrell and Katz 2006, Meese 2010, Motta 2004, Posner 1979, Turner 1969). However, if only consumer welfare counts, then the merger should be enjoined (Brodley 1987, Kirkwood and Lande 2008, Pittman 2007, Salop 2010, Pitofsky 1979). Under the current views of the US and EU antitrust enforcement authorities, such a merger should be blocked because producer gains that are not passed onto consumers cannot justify an anticompetitive merger that harms consumer welfare. Not all jurisdictions share this view. Canada, for example, has explicitly built a total welfare standard into its merger law, reflecting the need to consider all efficiency effects in considering the legality of a merger. Whether a jurisdiction adopts a consumer welfare or total welfare standard may depend in substantial part on the country’s economic developmental stage and geographic advantages and constraints. Many developing countries may find it expedient to allow conduct that improves productive efficiency even if harms short-run consumer welfare, on the theory that the country’s long-run welfare requires some short-run reductions in competition that enhance the productive strength of infrastructure or otherwise important industries.

1.4.  Competitors and Competition “Competition, not competitors” is a maxim frequently used to express core values in antitrust discourse. The US cases, at least, are replete with dicta to the following effect: “It is axiomatic that the antitrust laws were passed for ‘the protection of competition, not competitors.’ ”4 US commentators and sometimes even antitrust enforcers have been known rudely to accuse the European Commission or the European Court of Justice or General Court of violating this axiom by protecting competitors without regard to the implications for consumer welfare. The Europeans often rejoin (and not always less rudely) that the Americans misunderstand—that competitors are instruments of consumer welfare and that without competitors, there is no competition. In the midst of this sometimes heated discourse emerges an implicit consensus that the welfare of consumers should be the primary motivation of antitrust policy and that the disagreements are primarily over implementation. But some in the antitrust community take a different view—that economic rivalry is inherently good for society and that forcing antitrust regimes to justify their own existence with reference to narrow objectives like allocative efficiency or even consumer welfare impoverishes and indeed endangers the antitrust enterprise. In this view, competition should be encouraged on the belief that it produces a number of benefits to society, even though the relationship between competition and those benefits cannot always be demonstrated immediately or to the standards of scientific scrutiny. When courts or

4  Brooke Group, Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 224 (1993).

10   Daniel A. Crane antitrust authorities demand proof of actual harm to consumer welfare or other relevant interests, they will systematically dismiss meritorious cases simply because empirical proof of anticompetitive effects is lacking. In this view, antitrust law should encourage economic rivalry almost for rivalry’s own sake, taking it on faith or informed belief that, systemically observed, rivalry will produce desirable outcomes. This view was perhaps more popular in the past than it is today. Modern antitrust authorities tend to be populated by economists or lawyers trained in economic traditions, who tend to demand greater theoretical precision and empirical proof of relationships between allegedly anticompetitive acts or market concentration and specific harms in order to sustain an interventionist regulatory decision. Also, many in the antitrust community suspect that an unverified conviction in the unspecified benefits of economic rivalry may become a pretext for an antitrust policy that protects rent-seeking businesses, inefficient competitors, or greedy lawyers rather than any more winsome constituency. Hence, contemporary antitrust law generally seeks to explain economic rivalry as desirable because of its beneficial effects on the welfare of consumers or others, and does not rest on the claim that rivalry is inherently desirable.

1.5.  Dynamic and Static Efficiency Even keeping with a relatively narrow economic view that only efficiency should count in antitrust, clashes often arise over the relative importance of static and dynamic efficiency. Static efficiency refers to the efficiency consequences of price effects—particularly deadweight losses that occur when firms have to reduce output in order to raise prices. Dynamic efficiency refers to the gains that result from innovation and the introduction of new technologies over time. Obviously, both static and dynamic efficiency are desirable. Further, the types of antitrust interventions that simulate static efficiency sometimes also stimulate dynamic efficiency. But the opposite may also be true. In some instances, antitrust interventions may improve economic efficiency in the short run by stimulating competition and bringing down prices, but stymie innovation in the long run because firms have insufficient margins or are otherwise disincentivized to invest in research and development. This brings us to one of the most important conceptual controversies in antitrust policy: Is there inherently a clash between dynamic and static efficiency and, if so, which one should antitrust prioritize? To introduce this debate briefly, the two poles in the controversy are often stylized around two personalities—the Austrian economist and public intellectual Joseph Schumpeter and the US economist and Nobel laureate Kenneth Arrow. The Schumpterian view, ascribed to passages in Schumpeter’s 1942 classic Capitalism, Socialism and Democracy, is that capitalism is inherently characterized by transformative industrial innovations and that antitrust law foolishly tries to stand in the way of these “perennial gales of creative destruction” (Schumpeter 1942, 84). In this stylized Schumpeterian narrative, dynamic efficiency brings far more welfare to society than

Rationales for Antitrust   11

short-run static efficiency. Innovator firms typically wrest control of markets from incumbent monopolists and reign until another innovator displaces them. In this vision, there is little need for antitrust enforcement. The Arrowian perspective, ascribed to Arrow’s 1962 article “Economic Welfare and the Allocation of Resources for Invention,” claims just the opposite—that competitive markets generate innovation while monopoly markets stifle it (Arrow 1962). Arrow’s central observation was that a monopolist has a reduced incentive to innovate since the innovation will simply cannibalize sales of the monopolist’s existing products. By contrast, in a competitive market, many firms have incentives to innovate. The upshot of Arrow’s perspective is that what’s good for static efficiency is good for dynamic efficiency. Both sorts of efficiency are maximized by an antitrust policy that keeps markets competitive. There is a cottage industry of “Schumpeter versus Arrow” literature, some of it theoretical, some of it empirical; some taking one side or the other, some attempting to reconcile the two camps. Surveying that literature is beyond the scope of this chapter. For present purposes, note that one’s view of the rationales of antitrust may be importantly affected by one’s view of the relative importance of static and dynamic efficiency and of the industrial conditions that produce or retard them. A related point concerns the time horizon considered by antitrust enforcement agencies when making decisions about the lawfulness of competitive behavior. Sometimes conduct may produce short-run welfare increases but long-run welfare decreases. An example of this is predatory pricing, which involves drastic price reductions for a period of time followed by drastic price elevations once competitors are driven from the market. In the short run, consumers may be benefited, while in the long run they are harmed. A converse example is a joint venture that gives the joint venturers market power for the two years while they are developing a new product and not aggressively competing against each other. However, when the new product is finally introduced, consumers benefit greatly from the innovation and the firms revert to full competition. Both of these circumstances require antitrust decision-makers to balance short-run versus long-run considerations. Some decision-makers are willing to give significant weight to long-run effects, appropriately discounted based on time value and probability considerations. Other decision-makers believe that antitrust law should be largely contained to short-run considerations, on the view that long-run effects are inherently speculative.

1.6.  Monopsony and Buyer Power Existential questions sometimes arise in antitrust cases when the usual positions of buyers and sellers are inverted and the buyers, not the sellers, acquire or possess market power (Blair and Harrison 2010). In such cases of monopsonization or oligopsonization, the question may arise whether antitrust policy should be solely concerned with the welfare of consumers or other buyers, or whether it should be independently concerned with the welfare of sellers.

12   Daniel A. Crane Take, for example, the classic example of a buyer’s cartel. Buyer cartels, like seller cartels, are per se illegal under most antitrust laws.5 This might not be obvious at first blush. If one assumes that the purposes of the antitrust laws are to promote consumer welfare, then how does an agreement by buyers to lower the price of goods sold fit within the purposes of the antitrust laws? This is question is not limited to the cartel context. Similar questions arise about monopsonization as an offense and about mergers that create overly dominant buyers. One way to address this issue is to push back on the assumption that the antitrust laws normatively protect only consumer interests. Indeed, those who want to argue for a broader view of antitrust’s purposes often point to buyer-side cases as evidence that antitrust law in fact cares independently about seller welfare, which proves that the “consumer welfare prescription” language is overly narrow. On the other hand, even if one is concerned solely about consumer welfare, it should be noted that anticompetitive acts by buyers can sometimes harm consumer welfare. Assume, for example, that two firms compete in the sale of widgets. An essential component of a widget is a gidget. The two firms buy gidgets in competition with one another. The firms then secretly agree to depress the prices of gidgets by bidding less aggressively against each other. Suppose that the price of gidgets goes down 10 percent as a result. Assume, as in most markets, that the supply of gidgets is somewhat elastic—meaning that as the price of gidgets goes down, gidget sellers produce and sell fewer of them. As a result of colluding on input prices, the widget makers will have to decrease the amount of gidgets they purchase. And since gidgets are an essential component of widgets, this means they will have to decrease the number of widgets they sell too. Thus, the effect of the buyer collusion will be to reduce output and hence likely raise prices of widgets, to the detriment of consumers. The foregoing example assumed some factors—elastic supply, invariant ratio between input and output—that may not be true in many cases. The question of antitrust policy is whether a plaintiff complaining about buyer-side market power should have to prove harm to consumer welfare, or whether merely proving anticompetitive harm to sellers is sufficient. Although some recent US Supreme Court cases could be read as requiring a showing of threats to consumer welfare even in monopsonization cases,6 this issue remains largely unresolved as a matter of positive law.

1.7.  “Noneconomic” Objectives A final question is to what degree antitrust has, or should have, “noneconomic” objectives. “Noneconomic” is pointedly in quotation marks for a reason. In current discourse,

5  Mandeville Island Farms, Inc. v. Am. Crystal Sugar Co., 334 U.S. 219 (1948).

6  Weyerhaeuser Co. v. Ross-Simmons Hardwood Lumber Co., 549 U.S. 312 (2007).

Rationales for Antitrust   13

commentators often assume that an “economic” approach to antitrust law began with the Harvard School structuralists in the 1950s or the Chicago School free enterprisers in the 1960s and 1970s. They further assume that the schools of thought that preceded the post–World War II era were largely focused on ideological, political, or social values and not economic objectives. But this view is misguided. As Herbert Hovenkamp has observed, “Antitrust in the United States has seldom suffered from a shortage of economic theories suggesting why certain behavior should be unlawful” (Hovenkamp 2001, 259). Although neoclassical microeconomic theory did not dominate antitrust practice in the early years, economic theories of many varieties have always been part of antitrust discourse (Crane and Hovenkamp 2012). Even the intellectual traditions in competition policy most closely identified with an “noneconomic” orientation rely heavily on economic ideas and arguments to advance their normative claims. For example, the “Brandeisian” tradition, associated with US Supreme Court justice Louis Brandeis, is often described as a social or political theory supporting atomistic competition because of its beneficial effects on personal liberty and autonomy. However, in defending his vision for an aggressive antitrust policy, Brandeis engaged in explicit economic reasoning, challenging the theory of natural monopoly in high fixed cost industries and contrasting the short-term efficiencies of monopoly with its long-run waste (Brandeis 1934). On the other side of the Atlantic, the German ordoliberals of the Freiburg School advocated a similar policy of atomistic competition as a fundamental requirement of social justice and a well-functioning society. Free economic participation by all citizens was the primary channel for achieving liberal and humane values. Market power by individuals or firms stood in the way of this goal. Accordingly, the state had an affirmative obligation to promote market competition in order to secure liberal goals. The ordoliberals admitted that implementation of their vision might reduce economic efficiency, but thought this necessary to achieve a just society. Even so, the ordoliberals, like Brandeis, engaged in argumentation about the formation of economic knowledge and the place of efficiency in the hierarchy of social values, which makes it difficult to sustain the belief that theirs was a “noneconomic” view of antitrust law. To be sure, antitrust law has often received support because of its value in advancing objectives more accurately labeled political than economic. Many of the developments in antitrust law in the wake of World War II were reactions to prewar economic failures that contributed to catastrophic political failures. The 1950 Celler-Kefauver Act in the United States, which strengthened the US merger statute, was in large part a reaction to a congressional concern that undue aggregations of economic power during the 1930s in Nazi Germany, and to some extent in imperial Japan as well, facilitated the rise of fascist states (Pitofsky 1979). The United States pushed an antitrust law on Japan during the Occupation period, in large part to dismantle the power of the zaibatsu (First 2000). The treaties containing the foundational notions of European competition law—the European Coal and Steel Community of 1952, which morphed into the Treaty of Rome in 1957—were motivated as much by achieving European economic integration and a common market in order to lay the conditions for peace and stability

14   Daniel A. Crane as by securing economic efficiency. More recently, South Africa’s postapartheid antitrust regime has been motivated in substantial part by the need to break down the vestiges of the apartheid-era power structures and establish a more just social order. Beyond these obvious examples of macro political objectives, antitrust authorities sometimes justify specific antitrust interventions because of the social costs of the exercise of market power in the relevant industry. For example, the concentration of media outlets in a few hands might raise concerns that the surviving media empires will eschew the publication of diverse social and political viewpoints. Even if the case did not fit a traditional output reduction model, it might be appropriate for the antitrust authorities to stop a media merger in order to foster a more pluralistic infrastructure for the dissemination of competing viewpoints. Many in the antitrust community worry that considering these sorts of noneconomic factors will confuse and dilute antitrust enforcement. There are many examples of social harms that antitrust authorities could seek to control but might lack the expertise and political legitimacy to tackle. For example, consider antitrust enforcement in the tobacco industry. The governments of most industrialized countries today take the view that tobacco consumption is, on balance, socially harmful and should be reduced as much as possible. Many governments employ Pigouvian taxes on cigarettes, regulations, and other legal interventions to reduce smoking. Yet the very same governments will challenge mergers or other anticompetitive conduct by tobacco companies. If demand for tobacco products is somewhat elastic and the challenged acts have the effect of increasing tobacco prices, then antitrust interventions that preserve competitive markets and bring lower tobacco prices actually harm social welfare by increasing tobacco consumption (Crane 2005). Still, most antitrust practitioners and academics probably believe that it would be inadvisable for government enforcers or courts to begin making such trade-off evaluations in the antitrust domain. Antitrust law does not contain the analytical instruments to decide when a product is harmful or beneficial or what the optimal level of consumption should be. The prevailing view is that antitrust must rely on exogenously formed consumer preferences to dictate values and that the control of externalities (other than those caused by the exercise of market power) should be left to other regulatory instruments and governmental policy decisions. In the mainstream, then, antitrust law is a relatively specialized regulatory instrument. It is intended to preserve competitive markets, largely for the benefit of consumers, whose preferences are held sovereign. To the extent that antitrust law is designed to protect interests apart from those of consumers, any further classes of beneficiary must be able to explain how market competition advances their legitimate interests. Thus, even if antitrust law is normatively founded on both economic and noneconomic objectives, the practice and implementation of antitrust law generally requires justification of interventions using the vernacular of modern economic theory.

Rationales for Antitrust   15

References Arrow, Kenneth J. 1962. Economic Welfare and the Allocation of Resources for Invention. The Rate and Direction of Inventive Activity 609. Cambridge, MA: National Bureau of Economic Research. Blair, Roger D., and Jefferey L.  Harrison. 2010. Monopsony in Law and Economics. New York: Cambridge University Press. Blair, Roger D., and Daniel Sokol. 2012. The Rule of Reason and the Goals of Antitrust:  An Economic Approach. Antitrust Law Journal 78: 471. Brandeis, Louis D. 1934. Shall We Abandon the Policy of Competition? Osmond K. Fraenkel, ed., The Curse of Bigness: Miscellaneous Papers of Louis D. Brandeis. New York: Viking Press. Brodley, Joseph F. 1987. The Economic Goals of Antitrust: Efficiency, Consumer Welfare, and Technological Progress. New York University Law Review 62: 1020. Bork, Robert H. 1978. The Antitrust Paradox: A Policy at War with Itself. New York: Basic Books. Carlton, Dennis W. 2007. Does Antitrust Need to Be Modernized? Journal of Economic Perspectives 21: 155. Crane, Daniel A. 2005. Harmful Output in the Antitrust Domain: Lessons from the Tobacco Industry. Georgia Law Review 39: 321. Crane, Daniel A. and Herbert Hovenkamp. 2012. The Making of Competition Policy: Selected Sources. New York: Oxford University Press. DiLorenzo, Thomas J. 1985. The Origins of Antitrust: An Interest-Group Perspective. International Review of Law and Economics 5: 73. Elzinga, Kenneth G. 1977. The Goals of Antitrust: Other Than Competition and Efficiency, What Else Counts? University of Pennsylvania Law Review 125: 1191. Farrell, Joseph, and Michael L. Katz. 2006. The Economics of Welfare Standards in Antitrust. Competition Policy International 2: 3. First, Harry. 2000. Antitrust in Japan: The Original Intent. Pacific Rim Law and Policy Journal 9: 1. Hazlett, Thomas W. 1992. The Legislative History of the Sherman Act Re-examined. Economic Inquiry 30: 263. Hicks, John R. 1935. Annual Survey of Economic Theory: The Theory of Monopoly. Econometrica 3. Hovenkamp, Herbert. 1988. Antitrust’s Protected Classes. Michigan Law Review 88: 1. Hovenkamp, Herbert. 2001. Post-Chicago Antitrust: A Review and Critique. Columbia Business Law Review 2001: 257. Hovenkamp, Herbert. 2005. The Antitrust Enterprise:  Principle and Execution. Cambridge, MA: Harvard University Press. Kirkwood, John B., and Robert H. Lande. 2008. The Fundamental Goal of Antitrust: Protecting Consumers, Not Increasing Efficiency. Notre Dame Law Review 84: 191. Lande, Robert H. 1982. Wealth Transfers as the Original and Primary Concern of Antitrust: The Efficiency Interpretation Challenged. Hastings Law Journal 34: 65. McChesney, Fred S., and William E.  Shughart II. 1995. The Causes and Consequences of Antitrust: The Public-Choice Perspective. Chicago: University of Chicago Press. Meese, Alan J. 2010. Debunking the Purchaser Welfare Account of Section 2 of the Sherman Act: How Harvard Brought Us a Total Welfare Standard and Why We Should Keep It. New York University Law Review 85: 659.

16   Daniel A. Crane Motta, Massimo. 2004. Competition Policy, Theory and Practice. New  York:  Cambridge University Press. Oi, Walter Y., and Todd L. Idson. 1999. Firm Size and Wages. Orly Ashenfelter and David E. Card, eds., Handbook of Labor Economics 3: 2165–214. New York: North-Holland. Pitofsky, Robert. 1979. The Political Content of Antitrust. University of Pennsylvania Law Review 127: 1051. Pittman, Russell. 2007. Consumer Surplus as the Appropriate Standard for Antitrust Enforcement. Competition Policy International 3: 205. Posner, Richard A. 1979. The Chicago School of Antitrust. University of Pennsylvania Law Review 127: 925. Posner, Richard A. 2001. Antitrust Law. 2nd ed. Chicago: University of Chicago Press. Salop, Steven C. 2010. Question:  What Is the Real and Proper Antitrust Welfare Standard? Answer: The True Consumer Welfare Standard. Loyola Consumer Law Review 22: 336. Schumpeter, Joseph. 1942. Capitalism, Socialism, and Democracy. New York: Harper Bros. Stigler, George S. 1975. The Citizen and the State: Essays of Regulation. Chicago: University of Chicago Press. Turner, Donald F. 1969. The Scope of Antitrust and Other Economic Regulatory Policies. Harvard Law Review 82: 1207. Williamson, Oliver E. 1968. Economies as an Antitrust Defense: The Welfare Tradeoffs. American Economic Review 58: 18.

CHAPTER 2

A N T I T RU S T E N F OR C E M E N T REGIMES Fundamental Differences KEITH N. HYLTON

There are more than 100 antitrust enforcement regimes around the world (see, e.g., Hylton n.d.). Because of this, it is difficult to say much about antitrust statutes globally without running into the difficulty that such large numbers immediately imply for comparative projects. Comparisons are possible on a global scale, but they are necessarily limited to statistical summaries of the major features of the enforcement regimes.1 In spite of the large number of antitrust enforcement regimes, there are three that are recognized as extremely important in global commerce: the United States, the EU, and China.2 Moreover, China has modeled its antitrust law regime on that of the EU (Farmer 2010, 35–36). Given this, there are essentially two antitrust regime types that dominate global commerce: the United States and the EU. This chapter is a brief comparative study of the two major antitrust regimes. Even limiting my study to two competition regime types, there are many details in which the two types diverge. I will not compare the regimes in terms of all of the details. Instead, I will focus on three major areas in which fundamental differences are observed: enforcement, legal standards, and procedure. Within each of the three categories, I narrow the focus to a specific illustrative feature. In the enforcement category, I discuss penalty provisions. In the legal standards category, I examine predatory pricing law as a central feature illustrating fundamental

1  For an empirical study of competition law enforcement regimes around the world, see Hylton and Deng 2007. For comparative analysis, Hylton and Deng group countries into regions (Europe, North America, etc.) to compare the general stance of antitrust enforcement across regions. 2  The primacy of the US, EU, and Chinese antitrust regimes is evidenced by the fact that news stories discussing major mergers often focus on the approval processes in these three regimes. The coverage of the Google and Motorola Mobility merger exemplifies this trend (Bartz and Chee 2012; Whitney 2012).

18   Keith N. Hylton differences between the two regimes. Finally, I summarize broad differences in procedure under the regimes. With respect to enforcement, the EU and US regimes differ in that the EU imposes fines that are based on the violator’s gain, while the United States imposes harm-based penalties. In predation law, the United States has adopted a marginal cost standard and the EU has adopted an average cost standard. With respect to procedure, the United States is a thoroughly common-law system, while the EU’s procedure is closer to the civil-law system in its allocation of power between the courts and the enforcement agency. These differences have profound implications for the welfare consequences of global antitrust law enforcement.

2.1. Enforcement The economic theory of enforcement prescribes punishment schemes that maximize society’s welfare, by reducing the sum of the costs of offensive conduct and the costs of enforcement. In this part, I briefly review the theory, and use it to address core differences in the antitrust enforcement policies of the United States and the EU.

2.2.  Optimal Enforcement Policy The theory of optimal antitrust enforcement is traceable to Gary Becker’s article on the economics of punishment (Becker 1968). Becker argued that an efficient system of punishment would seek to internalize the social costs associated with offensive conduct. Internalization is accomplished by shifting the costs suffered by victims to the offender in the form of a penalty. Becker considered the implications of his argument for antitrust, arguing that the social costs arising from antitrust violations should be internalized by those engaging in anticompetitive conduct (Becker 1968, 198–99). Later, Landes (1983) provided a more detailed application of Becker’s analysis to antitrust. In the antitrust context, internalization requires the punishment authority to shift the costs suffered by consumers, in terms of monopolistic overcharges or restrictions in supply, to the monopolizing firm in the form of a penalty. Consider the case of a firm that takes some action that enables it to gain monopoly pricing power, and at the same time generates efficiencies in production or sale. For example, the action could be a merger that creates or enhances monopoly power and at the same generates efficiencies—say, by cutting redundant worksites. Alternatively, the action could be an exclusive dealing arrangement that forecloses a rival firm and at the same time reduces costs in the supply chain. The economic effects of the firm’s conduct can be examined in figure 2.1. Before the firm takes the action, the market is competitive, with price equal to marginal cost

Antitrust Enforcement Regimes   19

P

W T

p1 D

E

c0 = p0 c1 q

FIGURE  2.1  Monopolization

with Cost Reduction

(p0  =  c0). After the firm takes the action, it gains the power to raise its price to the monopoly level p1. However, costs fall, as a result of the firm’s action, from c0 to c1. The optimal enforcement policy in this case is to impose a penalty on the firm equal to the sum of the wealth transfer from consumers and the forgone consumer surplus, which is represented by the sum of areas T and D in figure 2.1 (Landes 1983; Hylton and Lin 2010). Why would setting the penalty equal to the sum of the consumer wealth transfer and the forgone consumer surplus (T + D ) be optimal? The reason is that it aligns the firm’s incentives with society’s incentives. If the penalty is equal to the sum of the consumer wealth transfer and the forgone consumer surplus, then the firm will choose to take the monopolizing action when and only when it enhances society’s wealth. Suppose the monopolizing action generates an efficiency gain, shown by the area E in figure 2.1. The firm’s action is welfare enhancing for society as long as the efficiency gain is greater than the forgone consumer surplus, that is, E > D . The firm will choose the monopolizing action if the gain the firm gets from monopolizing is greater than the expected penalty for monopolization. The firm’s gain is the sum of the transfer from consumers and the efficiency gain, T + E . Thus, if the penalty is equal to the sum of the transfer and the forgone consumer surplus (T + D ), the firm will take the monopolizing action only when it is welfare enhancing ( E > D ) .3 Although I  have used the terms “monopolization” and “monopolizing firm,” this analysis applies equally to cartels. If a cartel has an efficiency basis, then it will lead to an increase in price, generating a wealth transfer from consumers, and a reduction in supply costs, generating an efficiency gain. For simplicity, I will use the term “monopolizing firm” for the remainder of this chapter, even in instances in which the monopolizing act is the decision by a group of firms to create a cartel.

3 

To see this, note that the firm will take the monopolizing action when T + E > penalty. If the penalty is equal to T + D, then the firm will monopolize when T + E > T + D, or when E > D.

20   Keith N. Hylton This analysis implies that the optimal antitrust enforcement policy internalizes, to the monopolizing firm, the harm suffered by consumers. To be sure, this is the optimal policy because there is an efficiency gain resulting from the firm’s monopolizing action. If there were no efficiency gain ( E = 0) , then the optimal policy would set the penalty in order to completely deter the violator’s conduct (Becker 1968, 180; Hylton 1998, 197– 98). Such a complete-deterrence penalty would have to be at least as large as the wealth transfer from consumers (Becker 1968, 198–99; Landes 1983, 656). In other words, when the firm’s monopolizing action does not generate an efficiency gain, the optimal punishment policy is complete deterrence, which is accomplished by ensuring that the firm cannot profit from monopolization.4 Any penalty greater than the wealth transfer (T) satisfies the complete deterrence—or, equivalently, gain elimination—objective. Thus, there are two general approaches a punishment authority can take under an optimal punishment regime. One is to internalize consumer harm. The other is to deter completely by eliminating the expected profits from anticompetitive conduct. The internalization approach is appropriate for conduct that is either efficient or has a significant chance of being efficient. The complete deterrence approach is appropriate for conduct that is unambiguously inefficient. I have already provided examples of monopolizing conduct that may be efficient: mergers and exclusive dealing. As for conduct that is unambiguously inefficient, the obvious example is the price-fixing agreement (Posner 2001, 39). The standard price-fixing agreement involves no efficiency motivation; it is simply an arrangement to transfer wealth from consumers to producers (see, e.g., Leslie 1993).5 Under these conditions, the optimal punishment policy is to set a fine sufficient to eliminate the prospect of gain from the price-fixing cartel’s actions. In the case of a price-fixing cartel, the internalization policy still satisfies the optimal punishment goal, because the consumer harm is the same as the producers’ gain from price-fixing. For this reason, Becker concluded that the optimal punishment policy for antitrust is one that internalizes the consumer harm (Becker 1968, 199). While it is true that the two policies suggested by enforcement theory, harm internalization and complete deterrence, can be satisfied in the antitrust setting by a penalty that internalizes consumer harm, the different policy goals of the internalization and deterrence approaches should be kept in view. The reason is that there may be instances in which the internalization approach is administratively infeasible. For example, the internalization approach requires the punishment authority to produce a precise estimate of the consumer wealth transfer and the forgone consumer surplus. The data

4 

On the distinction between internalization and complete deterrence policies, see Hylton 1998, 425–33. 5  It is possible for a price-fixing agreement to be efficient. If the agreement is efficient, and the efficiency gains are enjoyed by the cartel, the cartel may have an incentive to continue the agreement even when faced with a damages remedy. See Becker 1968, 199 (“If . . . certain constraints of trade raise the level of economic welfare, fines could fully compensate society for the harm done, and yet some constraints would not cease, because the gain to participants would exceed the harm to others”).

Antitrust Enforcement Regimes   21

necessary to generate such an estimate may be unavailable. In such a case, where the risk of error is substantial, it is important to identify the precise policy basis for punishment.

2.3.  Actual Punishment Policies The actual punishment policies of the US and EU systems do not closely follow the prescriptions of optimal punishment theory. However, their key elements are distinctive and can be associated with the goals of harm internalization and complete deterrence. Consider US enforcement policy. The Sherman Act sets out a fixed maximum penalty of $100 million (Federal Trade Commission n.d., 1). In addition, US law permits courts to deviate from the fixed maximum penalty in the statute by imposing a penalty equal to twice the loss imposed by the violator on consumers (18 U.S.C. §3571; Hylton 2003, 49). In addition, private lawsuits enable plaintiffs to sue for treble damages (see, e.g., Hylton 2003, 48–49). Private lawsuits outnumber government lawsuits by a 10 to 1 ratio (Salop and White 1986, 1003). An optimal consumer harm-based penalty would divide the consumer harm, which is the sum of the wealth transfer and the forgone consumer surplus, by the probability that the monopolizing firm will be punished. Thus, going back to figure 2.1, the optimal penalty is equal to T + D, divided by the probability of punishment. The reciprocal of the probability of punishment is therefore the “optimal multiplier” that should be applied to the consumer harm penalty. American law, in contrast, imposes the transfer, T, as a penalty on the monopolizing firm, and employs either a multiplier of two under the fines enforcement law, or a multiplier of three for private lawsuits. These multipliers may or may not be optimal, depending on the relationship between the probability of punishment and the statutory multipliers of two and three. For example, if the probability of punishment—say because detection is difficult—is 1/10, then the multipliers on fines and damages would be too low to efficiently internalize the consumer harm. Although the statutory multipliers may not be optimal, the American approach is broadly consistent with the harm internalization approach to punishment. The penalty for an antitrust violation varies directly with the magnitude of the consumer harm. Thus, the American antitrust punishment system is one in which violators pay a penalty that is proportional to consumer harm, where the proportionality factor is greater than one. Now consider EU antitrust enforcement policy. The European Community Treaty (“EC Treaty”) provides that the penalty for an infringement will be based on the total sales of the violator, with a maximum penalty equal to 10 percent of the total revenue of the violator (Kaczorowska 2008, 873). Article 83(2) of the EC Treaty instructs that penalties should “ensure compliance with the prohibitions laid down in Article 81(1) and Article 82” (European Community Treaty 2002, art. 83). The relevant treaty provisions have been interpreted by enforcement officials and commentators to support the

22   Keith N. Hylton deterrence rather than the internalization objective (Showa Denko v. Commission, para. 58, cited in Wils 2007, 205 n. 57; Archer Daniels Midland v. Commission, para. 49).6 The lesson suggested is that the EU penalties are designed to deter prohibited conduct completely by eliminating the prospect of earning profits through conduct that violates the antitrust laws. The EU penalties are not designed to vary directly with the amount of consumer harm; they are structured to vary directly with the amount of profit that a firm gains from conduct deemed to violate EU competition law. Thus, the penalty provisions under US and under EU antitrust law reflect the major economic theories of punishment, complete deterrence and internalization. The EU has embraced complete deterrence as the objective of enforcement. The United States has adopted the harm internalization approach. One might be inclined to conclude that the EU has the relatively inefficient enforcement system. Penalties, such as those under the EU system, that run proportional to revenues punish efficient as well as inefficient monopolizing conduct with equal severity. The US system enables firms that intend to engage in efficient monopolizing acts to go forward and reap the rewards, as long as those rewards are sufficiently greater than the harm to consumers. On closer inspection, the comparison, on efficiency grounds, of enforcement provisions in the United States and EU requires consideration of more minute features. First, consider the distinction between cartelization (anticompetitive agreements among firms, such as price-fixing) and monopolization (anticompetitive single-firm conduct). Linking monetary fines to revenues, as the EU system does, may be efficient overall when applied to cartelization. If the cartels do not have an efficiency basis, as the enforcement structures in the United States and in the EU assume, then the complete deterrence objective is optimal. It happens that a policy of imposing penalties that are a multiple of consumer harm will achieve the complete deterrence objective just as effectively as a policy that aims to eliminate gains. But the policy of complete deterrence, rather than harm internalization, remains the correct policy on social welfare grounds. If a cartel does have an efficiency basis, then the EU system would impose inefficiently large penalties—assessing a penalty based on the sum of the wealth transfer and the efficiency gain (T + E ) instead of limiting the penalty to the wealth transfer (T ) . However, both punishment systems are based on the assumption that the vast majority of cartelization cases are inefficient wealth transfers. If this assumption is correct, then the EU system’s average degree of inefficiency in cartel punishment cases may be less than that of the United States. The reason is that the EU would impose the correct level of the fine in the vast majority of cases, and an inefficiently large fine in a minority of cases. Under the same assumptions, the US fine, limited to the transfer, might fall short of providing optimal deterrence because it fails to fully internalize consumer harm.

6  EU competition policy’s emphasis on deterrence was also apparent when the EU Commission imposed double the basic fine against Microsoft in the EU-Microsoft cases (see Economides and Lianos 2010, 372).

Antitrust Enforcement Regimes   23

To clarify this argument, consider the following hypothetical. Suppose a firm that is (correctly) found guilty of participating in an inefficient price-fixing conspiracy has a total annual revenue of $20 billion. Suppose the firm’s gross gain from the conspiracy (T) is $150 million, and the forgone consumer surplus (D) is $75 million. The penalty under US law could be as high as $300 million—since the rules permit courts to deviate from the $100 million Sherman Act limit up to twice the gross gain from the conspiracy. The penalty under EU law could be as high as $2 billion (10 percent of revenue). If the likelihood of being punished for the conspiracy is 1/10, the amount necessary to internalize consumer harm is $2.25 billion, and the amount necessary to eliminate the prospect of gain is $1.5 billion. The Sherman Act penalty of $300 million would be too low, from either the internalization or the complete deterrence perspective. The EU rules, however, could result in an optimal penalty for deterrence purposes. If the EU penalty were set at a level between $1.5 and $2 billion, it would eliminate the gain prospectively, which is sufficient to meet the optimal punishment goal in this scenario. If the EU penalty were set at its maximum of $2 billion, it would come close to internalizing the consumer harm (the wealth transfer and the foregone consumer surplus). Moreover, the US enforcement system imposes imprisonment as a punishment on relevant officials in price-fixing cases, while the EU system limits punishment to the imposition of monetary fines (Connor 2001, 89). The imprisonment of firm officials may offset the tendency toward underdeterrence suggested in the numerical example just offered, but it depends on many factors that are shrouded in uncertainty. Imprisonment concentrates the risk of punishment on the minority of actors who are responsible for the price-fixing agreement. If those actors are not aware of the risk of imprisonment, they will not be deterred. If they are aware of the risk, they are likely to be deterred, but even this conclusion is uncertain because it depends on the actors’ risk preferences and psychological attachments. For the relevant actors, the expected cost of imprisonment, given the low probability, may appear to be small in comparison to the rewards. In addition, imprisonment generates excessive litigation, as firms and targeted officials spend huge sums to avoid the punishment. Some of the executives imprisoned have long productive records as employees in their industries; they are not common criminals. Society forfeits the value of their services by locking them up. These points of comparison suggest that the broad-brush, gain-based penalties under the EU system may be superior on social welfare grounds to the penalty system enacted under the US antitrust laws for cartelization cases. At the least, it is unclear a priori whether the United States or the EU has the socially preferable system for punishing cartel activity. With respect to monopolization, a different assessment seems appropriate. In the monopolization context, the EU fine system appears to be inferior on welfare grounds to the US punishment system. The EU system aims to strip the gains from monopolizing conduct, which deters both efficient as well as inefficient conduct. The US punishment system, in contrast, discriminates between efficient and inefficient monopolizing conduct. The discrimination process is not perfectly optimal, but it is probably superior on social welfare grounds to the EU punishment system.

24   Keith N. Hylton To clarify this argument, return to figure 2.1. Under the US system, the monopolizing firm would be required to pay a penalty that is likely to be a multiple of the consumer harm—a multiple of two under public enforcement, and three under private enforcement. If the efficiency gain from its conduct is sufficiently large, the US punishment system will not deter a firm from engaging in efficient monopolizing conduct. For example, suppose, in a particular instance of monopolization, the wealth transfer is $150 million, the forgone consumer surplus is $75 million, and the efficiency gain is $200 million. Assume also that the likelihood of being punished is 50 percent—which reflects the greater likelihood of detection and enforcement in monopolization cases. Thus, the firm’s gain from monopolization, which is equal to the sum of the wealth transfer and the efficiency gain, is $350 million. Moreover, since the efficiency gain exceeds the forgone consumer surplus, this is a case of efficient monopolization. The penalty imposed under the US system would be either $300 million (twice the transfer) under public enforcement or $450 million (three times the transfer) under private enforcement. In either case (public or private enforcement), the monopolizing firm would not be deterred under the US punishment system. The reason is that the expected penalty, whether under public or under private enforcement, would be less than the gain from monopolization: under public enforcement, the expected penalty would be $150 million, which is less than the $350 million gain from monopolization; and under private enforcement the expected penalty would be $225 million, which is also less than the $350 million gain from monopolization. The monopolizing firm would not be deterred by the threat of penalization under the US system, and this is the efficient result. Under the EU system, the monopolizing firm would be required to pay a penalty that is likely to be a multiple of the sum of the transfer and the efficiency gain (T + E ) . Thus, in a case of efficient monopolizing conduct, the firm definitely would lose rather than gain after taking the penalty into account. For example, suppose a firm with annual revenue of $20 billion engages in a monopolizing act in some part of its business. The transfer from consumers is $150 million, the forgone consumer surplus is $75 million, and the efficiency gain is $200 million. The firm’s total gain from monopolization is therefore $350 million. As in the previous example, this is a case of efficient monopolization. The EU could impose a fine on the firm as high as $2 billion.7 If it chooses a fine greater than $700 million, though still well below the penalty ceiling, it would completely deter the monopolizing act, an inefficient outcome. The general picture that emerges from this comparison is that the US antitrust law enforcement system is compatible with the efficiency goal while EU law is not. EU law has followed Bentham by setting fines sufficiently large to wipe out the gains from conduct deemed unlawful. US law has followed Becker by keeping damages closely tied to the harms suffered by consumers.

7  I have assumed that the firm is large (revenue $20 billion). Obviously, if the firm’s revenue is not much more than the amount it receives from the monopolizing activity, the EU fine may be far less than the amount required to deter.

Antitrust Enforcement Regimes   25

The inefficiency of the EU enforcement system is not limited to the EU. China’s competition law enforcement system is modeled on the EU’s provisions (Wei 2011, 812–15), as are the enforcement rules of several other competition enforcement authorities, including Pakistan and Singapore (Wilson 2011, 111; Ong 2007, 109–10). The prevalence of the EU enforcement model indicates a strong preference for enforcement provisions that threaten fines tied to gain, as a form of trade tariff, rather than fines tied to consumer harm.

2.4.  Substantive Standards: Predation The second important area in which to examine the differences between the major competition regimes is substantive law. There are many parts of substantive law to examine. At the most general level, however, the antitrust laws fall into one of two categories: prohibitions on cartelization, or prohibitions on monopolization. Since the substantive policies with respect to cartelization are virtually the same in all competition regimes, I will focus on monopolization law; specifically, predatory pricing. A predatory pricing claim is an assertion by one firm, the predation target or victim, that it has been injured by the low prices of another, the incumbent dominant firm. The victim typically argues that the dominant firm cut its price during a predatory campaign, in order to force the victim to sustain losses that would compel it to leave the market. After the victim leaves the market, the dominant firm, if all goes according to the theory, raises its price to the monopoly level. During the period in which the dominant firm prices at the monopoly level, it recoups the losses that it suffered during the predatory pricing campaign. Predatory pricing claims in the United States fall under Section 2 of the Sherman Act. The law under Section 2 puts high hurdles in the way of predation plaintiffs. In order to avoid a summary judgment in a predation lawsuit, the plaintiff must present evidence that the defendant set its price below some reasonable proxy for marginal cost, and that the market structure is such that it would permit the defendant to recoup its losses from the predation campaign, after the victim has been forced out of the market (Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 222–24). Both requirements are difficult to meet. The EU law on predatory pricing imposes a lighter burden on the plaintiff. Under EU law, unlawful predation is established if the evidence shows that the defendant sets its price below average variable cost (see, e.g., AKZO Chemie BV v. Commission, para. 71). If the defendant set its price below average cost but above average variable cost, then predation can be established if the evidence suggests that it was accompanied by an intention to exclude the plaintiff (AKZO Chemie BV v. Commission, para. 72). The evidence required to prove predatory intent includes objective factors, such as the duration of the predatory period and the number of units sold at the allegedly predatory price. The evidentiary requirements suggest that, in practice, a significant burden falls on the

26   Keith N. Hylton defendant to disprove predatory intent when price is below average cost and above average variable cost. The difference between the US and EU standards on price predation reduces to this: the United States uses a marginal cost test and the EU uses an average cost test. Figure 2.2 illustrates the fundamental differences between the United States and EU with respect to predation. Price cuts below the marginal cost curve (MC) are predatory in the United States, provided market structure evidence shows the plausibility of recoupment (Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 210). In the EU, price cuts below the average cost curve (AC) are predatory. Since marginal cost is nearly impossible to measure precisely, the US law encourages courts to examine reasonable proxies to marginal cost (Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 223). However, the goal is to use a measure that approximates marginal cost (United States v. AMR Corp., 1115–16). The US marginal cost test, as Areeda and Turner (1975, 701–2) argued, is the efficient standard. When price is above marginal cost, a cut in price moving it in the direction of marginal cost, along with an associated increase in consumption, enhances social welfare. Conversely, when price is below marginal cost, a cut in price moving it away from marginal cost reduces social welfare. The reason is that price, under ideal conditions, reflects the marginal benefit to society from producing an extra unit of a good. Marginal cost, under ideal conditions, reflects the resource cost to society of supplying an additional unit of a good to the market. As long as price exceeds marginal costs, society’s welfare can be enhanced by expanding consumption. Hence, a marginal cost test for predation is consistent with a policy of enhancing society’s welfare. The average cost test of the EU disregards the efficiency principle and creates a price umbrella based on the dominant firm’s average total cost. As long as a rival firm can match the average total cost of the dominant firm, it is shielded from additional price pressure under the EU law. In addition, given the uncertainties in measuring cost, and the amount and duration of losses a predatory target firm must experience before it is financially compelled to leave the market, the EU’s average-cost standard effectively shields relatively inefficient firms from vigorous price competition.

S/q

MC AC AVC

q FIGURE  2.2  Marginal

Cost, Average Cost, and Average Variable Cost

Antitrust Enforcement Regimes   27

The precise function or objective of the EU predation standard has never been set out clearly. One possible justification is that an average cost pricing standard enhances the set of options to consumers, by preventing efficient price predation. If this is indeed the purpose of the standard, it is unlikely that it enhances consumer welfare. It preserves relatively inefficient rivals and forces consumers to pay higher prices. The consumers themselves probably would have chosen to pay lower prices for fewer retail options, as they often do when price-cutting firms displace higher-priced rivals. Another possible justification for the EU average-cost standard is that it preserves employment, by reducing the frequency with which price competition leads to the exit of firms that are the victims of predatory pricing campaigns. Less efficient firms are more likely to survive under the EU standard. Perhaps in a state with a generous welfare system, the policy of preserving less efficient firms is less expensive for taxpayers than a policy that allows them to be driven out of business by efficient price predation. In other words, the average-cost standard may be, in essence, a public welfare policy. The efficiency of such a policy cannot be determined without taking into account the relative inefficiencies of the administrative state.

2.5. Procedure The third important difference between U.S and EU antitrust law is procedure. I refer to procedure in the broadest sense; from the processes by which the legal standards evolve to the methods used to determine the validity of evidence.

2.6.  Development of Law The United States is unique among competition law regimes in that its law is developed through the common-law process. The Sherman Act says relatively little (Sherman Act 2004, §§ 1 & 2). It can be accurately summarized by saying that it prohibits price-fixing and monopolization. The detailed rules that have developed in American antitrust law have almost all come out of the courts. The EU law has been set out in a relatively sparse treaty. The EC Treaty, however, is more detailed in its statement of prohibitions than is the Sherman Act (European Community Treaty 2002, art. 81). Moreover, the precise meanings of the key competition provisions of the EC Treaty, Article 81 (now Article 101 of the Treaty on the Functioning of the European Union (“TFEU”)) and Article 82 (now Article 102 of the TFEU), have been developed for the most part by the European Commission. For example, the EU Commission has been instrumental in clarifying “Market Definition” under EU Competition Law, which has ramifications for both Article 81 and 82 enforcement (see Report Prepared for the Competition Directorate-General of the European

28   Keith N. Hylton Commission 2005, 5). The EU courts defer to the European Commission on issues concerning the interpretation of the competition rules (Marsden 2009, 27).8 As a result, the system of EU courts is not the primary body that interprets the meaning of the competition provisions of the EC Treaty. These differences in the processes under which competition law develops have important implications. The American process essentially grafts onto the common law an additional branch called antitrust law without changing the process by which cases are litigated or decided in any substantial manner. Under the common-law process, courts independently develop a framework for applying legal rules and modifying them in light of facts or policy arguments. The common-law process permits both sides in litigation, plaintiff and defendant, to present their positions on the meaning of the law and the state of the evidence to an impartial observer, the court. The court inevitably has some degree of discretion in both matters. Cases that are easily decided on the basis of the statutory text, or on the basis of earlier decisions, do not continue for a long time in court. Judges dismiss them, or decide them quickly, or the parties settle. The vast majority of legal disputes that spend enough time in court to come before judges have sufficient uncertainty surrounding them that the judge inevitably has discretion to decide what the law requires as between the litigating parties. The discretion that judges have had under the US antitrust laws is equivalent to the discretion they have had under the common law for centuries. The discretion given to judges under the American process has permitted courts to consider the social consequences of their decisions, and to issue judgments that effectively reduce the social costs of the rules they administer. Common-law judges, in the course of examining the consequences and implications of their decisions, trade off the social costs of false convictions and false acquittals. Judges do not have to discover the relevant social costs on their own; the litigating parties have strong incentives to bring this information directly to the judges’ attention. Most likely because of this constant process of weighing cost trade-offs, American antitrust law has tended toward adopting efficiency-based legal standards, such as the marginal-cost-based test in predatory pricing law. The dominant role of the enforcement agent imparts a different tendency to the law’s evolution in EU. The enforcement agent will tend to interpret the law not in a manner that impartially weighs or trades off social costs from false convictions and false acquittals, but in a manner that minimizes its own enforcement costs. The dominance of per se standards based on relatively simple and abstract rules in the EU can be explained by their utility to the enforcement agent (Evans and Ahlborn 2009, 29). 8  The European Court of First Instance reviews Commission decisions by a “manifest error of assessment” standard that considers “whether the facts on which the Commission’s assessment was based were correct, whether the conclusions drawn from those facts were not clearly mistaken or inconsistent and whether all the relevant factors had been taken into account.” This limited standard of review is deferential to the Commission.

Antitrust Enforcement Regimes   29

The importance of administrative facility and the relative detachment of courts from the law-generation process put the EU system closer to a civil-law model rather than the common-law model of the United States. Given that EU courts are staffed largely with judges drawn from civil-law countries the tendency for a civil-law system to develop is natural (Eurofound 2011; Apple and Deyling 1995, 1). The EU rule on predatory pricing, for example, is an administratively simple rule. It does not require the complainant to generate a reasonable proxy of the defendant’s marginal cost, which is a difficult undertaking, both for the enforcement agent and for the court. It is not efficient as a rule governing competition. However, it is efficient on administrative grounds in comparison to the American rule. Although I have referred to American and EU law, the EU pattern has been replicated in China and other countries. Thus, one could say that competition law outside of the United States is largely shaped by demands of the enforcement agent. Competition law in the United States is shaped by the traditions of common-law courts.

2.7.  Evidence and Procedure The deference policy that EU courts have adopted with respect to the EC’s competition enforcement decisions implies important differences in the assessment of evidence and findings of fact. In the United States, antitrust enforcement agencies and private plaintiffs have to present evidence to a court and attempt to persuade the court that a conspiracy has occurred, or that a defendant should be deemed a monopolist. In the EU, the complainants face the same requirements but in the presence of the enforcement agent, rather than a court. If the enforcement agency has made a decision or is inclined to pursue a case, all evidence and policy arguments will be examined under the influence of that decision. The distinction between the US and EU systems is one between a court-centered process and an agency-centered process. In a court-centered process, as in the United States, the plaintiff, whether a private plaintiff or an enforcement agency, knows that it will have to persuade a skeptical court of the validity of its arguments, and must prepare its case with this in mind. Under an agency-centered process, the agency has a relatively small likelihood of ending up in a court in a particular case. A complainant approaching the enforcement agency will have to persuade the agency that it has a colorable claim, which is less than the burden, to prove that a violation of the law has occurred, borne by a private complainant in a court. Moreover, there is an important difference between the agency enforcement processes in the United States and in the EU with respect to the degree of separation between prosecutorial and adjudicative functions. The agency enforcement process in the United States is observed in practice when the FTC brings an enforcement action against an entity. Firms often approach the FTC with antitrust complaints against market-dominating competitors. If the FTC chooses to proceed with a complaint, it will

30   Keith N. Hylton either go to a federal court or to an administrative law judge, depending on the statute it seeks to enforce. While the FTC must bring enforcement actions for Clayton Act violations in federal court, the FTC may enforce Section 5 of the FTC Act through internal administrative litigation before an administrative law judge, a process known as Part III proceedings (Antitrust Modernization Commission 2007, 129). In either case, the FTC must prepare its arguments and evidence to withstand questioning by an independent official. Moreover, the FTC General Counsel’s office is independent of the FTC commissioners, the final decision-making body within the agency process (Coate and Kleit 1995, 1–2). There are substantial walls of separation between the prosecutorial and adjudicative arms of the FTC’s enforcement process. The agency process in the EU is distinguishable in the sense that the European Commission considers complaints and conducts investigations, but there is no separation between the prosecutorial arm and the adjudicative arm within the EC agency process. Evidence and arguments are not subjected to an independent assessment until they are appealed to the EU courts, which have adopted a deference policy with respect to such matters. And since the EU courts have adopted a deference policy, there really is no point at which the EC’s evidence and arguments are subjected to a rigorous and independent evaluation of merit. The enforcement system in China is based on the EU model. Moreover, the absence of democracy and of basic “rule of law” norms prevent the court system in China from becoming reliably independent of the interests of the government. The upshot is that outside of the United States, competition law procedure effectively combines prosecutorial and adjudicative functions. This is a very substantial chasm between the enforcement regimes. The procedures adopted under the EU model are inconsistent with fundamental requirements of due process in American law. The differences in substantive law probably pale in importance when compared to the differences in procedure. If the EU adopted efficiency-based legal standards, or an efficiency-based approach to punishment, the potential improvements in welfare could easily be vacated through the absence of reasonable due process safeguards in the enforcement process.

2.8. Conclusion With more than 100 competition law regimes, there are countless ways in which antitrust law regimes can vary around the world. However, two models are dominant in terms of their effects on global commerce: the United States and the EU. This chapter has compared those models in terms of enforcement, substantive law, and procedure. With respect to enforcement and substantive law, the United States has evolved toward an efficiency-based system while the EU has not. The procedural differences are perhaps more important than the differences in substantive law. The EU process of law development is closer to the civil-law model, while the American process is safely within the

Antitrust Enforcement Regimes   31

common-law tradition. Retail-level procedural issues, such as the treatment of evidence, reveal stark differences, the most significant of which being the relatively weak separation of prosecutorial and judicial functions within the EU enforcement process.

Acknowledgments I thank Danny Sokol for helpful comments. Adam Mayle provided excellent research assistance.

References Antitrust Modernization Commission. 2007. Report and Recommendations. Retrieved from http://govinfo.library.unt.edu/amc/report_recommendation/amc_final_report.pdf. Apple, James G. and Robert P. Deyling. 1995. A Primer on the Civil-Law System. Federal Judicial Center. Retrieved from http://www.fjc.gov/public/pdf.nsf/lookup/civillaw.pdf/$file/civil​ law.pdf. Areeda, Phillip, and Donald F. Turner. 1975. Predatory Pricing and Related Acts under Section 2 of the Sherman Act. Harvard Law Review 88: 697–733. Bartz, Diane, and Foo Yun Chee. 2012. Google Gets U.S., EU Nod to Buy Motorola Mobility. Reuters. Retrieved from http://www.reuters.com/article/2012/02/14/​us-google-motorolaeu-idUSTRE81C1HE20120214. Becker, Gary S. 1968. Crime and Punishment:  An Economic Approach. Journal of Political Economy 76: 169–217. Coate, Malcolm B., and Andrew N. Kleit. 1995. The Political Economy of Federal Trade Commission Administrative Decision Making in Merger Enforcement. Federal Trade Commission Working Paper 210. Retrieved from http://www.ftc.gov/be/workpapers/wp210.pdf. Connor, John. 2001. Global Price Fixing: Our Customers Are the Enemy. Boston, MA: Kluwer Academic Publishers. Economides, Nicholas, and Ioannis Lianos. 2010. A Critical Appraisal of Remedies in the E.U. Microsoft Cases. Columbia Business Law Review 2: 346–420. Eurofound. 2011. European Court of Justice. Retrieved from http://www.eurofound.europa.eu/​ areas/industrialrelations/dictionary/definitions/EUROPEANCOURTOFJUSTICE.htm. European Community Treaty. 2002. Official Journal of the European Communities C 325/39. Retrieved from http://eur-lex.europa.eu/en/treaties/dat/12002E/pdf/12002E_EN.pdf. Evans, David, and Christian Ahlborn. 2009. The Microsoft Judgment and Its Implications for Competition Policy towards Dominant Firms in Europe. Antitrust Law Journal 75: 887–932. Farmer, Susan B. 2010. The Impact of China’s Antitrust Law and Other Competition Policies on U.S. Companies. Loyola Consumer Law Review 23: 34–53. Federal Sentencing Guidelines. 198c. U.S. Code. Title 18. Federal Trade Commission. N.d. An FTC Guide to the Antitrust Laws. Retrieved from http://ftc.​ gov/bc/antitrust/factsheets/antitrustlawsguide.pdf. Hylton, Keith N. N.d. Antitrust World Reports. Retrieved from http://antitrustworldwiki.com.

32   Peter Carstensen Hylton, Keith N. 1998. Punitive Damages and the Economic Theory of Penalties. Georgetown Law Journal 87: 421–471. Hylton, Keith N. 2003. Antitrust Law:  Economic Theory and Common Law Evolution. New York: Cambridge University Press. Hylton, Keith N., and Fei Deng. 2007. Antitrust around the World: An Empirical Analysis of the Scope of Competition Laws and Their Effects. Antitrust Law Journal 74: 271–341. Hylton, Keith N., and Haizhen Lin. 2010. Optimal Antitrust Enforcement, Dynamic Competition, and Changing Economic Conditions. Antitrust Law Journal 77: 247–276. Kaczorowska, Alina. 2008. European Union Law. New York: Routledge-Cavendish. Landes, William M. 1983. Optimal Sanctions for Antitrust Violations. University of Chicago Law Review 50: 652–78. Leslie, Christopher R. 1993. Achieving Efficiency through Collusion: A Market Failure Defense to Horizontal Price-Fixing. California Law Review 81: 243–292. Marsden, Phillip. 2009. Checks and Balances:  EU Competition Law and the Rule of Law Competition Law International 5: 24–29. Ong, Burton. 2007. Competition Law Takes Off in Singapore: An Analysis of Two Recent Decisions. Competition Policy International 3: 101–31. Posner, Richard A. 2001. Antitrust Law. Chicago: University of Chicago Press. Report Prepared for the Competition Directorate-General of the European Commission. 2005. Market Definitions in the Media Sector: Comparative Legal Analysis. Retrieved from http://​ ec.europa.eu/competition/sectors/media/documents/chapter_1_ec.pdf. Salop, Steven C., and Lawrence J. White. 1986. Economic Analysis of Private Antitrust Litigation. Georgetown Law Journal 74: 1001–64. Sherman Act. 2004. 15 U.S.C. § 1 et seq. Wei, Dan. 2011. China’s Anti-Monopoly Law and Its Merger Enforcement:  Convergence and Flexibility. Journal of International Economic Law 14: 807–44. Whitney, Lance. 2012. Chinese Regulators Drag Heels on Google-Motorola Deal. CNET. Retrieved from http://news.cnet.com/8301-13578_3-57400661-38/​chinese-regulators-dragheels-on-google-motorola-deal/. Wils, Wouter P. J. 2007. The European Commission’s 2006 Guidelines on Antitrust Fines: A Legal and Economic Analysis. World Competition 30: 197–231. Wilson, Joseph. 2011. Crossing the Crossroads: Making Competition Law Effective in Pakistan. Loyola University Chicago International Law Review 8: 105–25.

Cases Cited Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209 (1993). Case 62/86, AKZO Chemie BV v. Commission [1991] ECR 3359. Case 289/04, Showa Denko v. Commission [2006], Opinion of Advocate General Geelhoed, ECR 05859. Judgment of the Court of First Instance, Case 329/01, Archer Daniels Midland v. Commission [2006] ECR 03255). United States v. AMR Corp., 335 F.3d 1109 (10th Cir. 2003).

CHAPTER 3

E C ON OM IC A NA LYSI S OF A N T I T RU S T E X E M P T ION S PETER CARSTENSEN

3.1. Introduction This chapter surveys potential economic theories for conferring statutory exemptions from the competition laws.12 Its goal is to provide a reasonably comprehensive survey of the economic theories for modifying or exempting classes of enterprises from competition law. The analysis does not extensively evaluate the theoretical limits or empirical validity of these arguments, but it does identify points that raise questions about the applicability of specific theories to specific exemptions. Most theories require very special facts to justify an exemption and those facts are unlikely to occur in a large and dynamic economy such as that of the United States. Thus, objective economic analyses would appear likely to suggest that only in very limited circumstances would an exemption or modification of competition law be an economically or socially optimal strategy to achieve a legitimate policy goal. In smaller economies or in situations where public policy is self-consciously seeking to transform an industry from state ownership or direct command and control regulation, there may be stronger economic arguments for temporary exemptions from or modifications of competition law. Finally, there is an emerging economic argument against judicial enforcement of 1  This chapter draws extensively on the economic analysis in c ­ hapter 3 of the ABA 2007. This work is used with the permission of its authors, Peter Carstensen, Christopher Sagers, and Ramsey Shehadeh. 2  Neither the advocates for particular exemptions nor the legislators who adopted them have paid close attention to economic analyses of their effects. Commentators have made similar observations about economic regulation legislation generally. See, e.g., Breyer, 1982, 189–314. In general, economic analysis does not appear to have guided or informed Congress’s crafting of specific exemption legislation.

34   Peter Carstensen competition policy in some industries even though the fundamental policy goal is a fully competitive market. The economic justifications for statutory exemptions fall into five main types.







1. If an industry has a natural monopoly that prevents durable competition, then competition rules may be preempted in favor of some other method of overseeing the market. 2. A  number of specific market and institutional failures can create a tension between the conventional model of robust competition that is the fundamental goal of competition law and the needs of market participants and stakeholders to have efficient market operations. 3. A preference for a specific socially desired activity or an intent to transfer wealth to a socially desirable group in society can be advanced to justify limiting the application of competition law in order to create exploitable market power for that group. 4. When public policy toward an industry or an overall economy shifts from either state ownership or direct regulation of price and output, even though market competition is the ultimate goal, a market-facilitating regulatory regime governing some or even all aspects of the industry distinct from competition law may be the more efficient transition policy. 5. In industries subject to extensive market regulation enforced through agency rules that are intended to facilitate a workably competitive market, the ad hoc intervention of courts in that process may impose serious inefficiencies in the market-facilitating regulatory process necessitating an exemption from competition law.

The second category, as the following discussion will demonstrate, incorporates a number of quite distinct economic theories that in turn call for specific kinds of market or institutional failures to be relevant. They share, however, the common characteristic of a generalized acceptance of having as competitive a market as is feasible and in theory, at least, using the exemption or modification of competition law as a means to achieve enhanced market performance. The third category reflects the positive framing of rent seeking by special interests. Because the goal of this chapter is to identify the various economic analyses that might justify, in the public interest, restricting or exempting business from the commands of competition law, the negative economic explanation for such exemptions and modifications will not receive any analysis here. Others have examined rent seeking in detail (see, e.g., Buchanan and Tullock, 1962; Tullock, 2005). As will also be evident from the following discussion, the first three categories have called forth the most explicit economic theories and models to explain when and why exemptions from competition law could be appropriate, but the final two categories may well reflect areas where the economic theories when developed will be even more compelling.

Economic Analysis of Antitrust Exemptions   35

In addition, a number of observed exemptions can find support in two or more economic theories. The theories themselves are not necessarily mutually exclusive, and because they address different aspects of market failure, they can be combined to provide a more elaborate economic justification for some exemptions.

3.2.  Natural Monopoly The theory of natural monopoly has provided a number of justifications for exemptions from competition law. In the abstract, a natural monopoly exists when only a single firm can serve the market efficiently.3 A natural monopoly arises when demand for a particular good is not large enough for a single firm to produce at minimum efficient scale (MES), the point at which it is minimizing its per-unit costs in both the short and long term. In such markets, if there are competing firms, the gains that any one firm can realize from increasing its volume of sales, enabling it to lower its per unit costs, are so significant that only one firm can survive in competitive equilibrium. (In figure 3.1 MES is the lowest point in the U-shaped average showing the relationship between price and demand.) A “natural monopoly,” therefore, is a consequence of the interaction of technology, supply, and demand under the prevailing economic conditions. Technology plays a central role in defining both the cost of production over increasing volumes and the extent of scale economies. Other supply factors also directly influence the costs of production and whether they will decline with increased volume. Changes in taste (demand) for goods affect the other side of the supply-demand interaction. As the volume demanded increases, the potential for multiple producers also increases even if there are substantial scale economies. Thus, the economic model of “natural monopoly” is highly contingent on the specific characteristics of the market in a particular time and place. When, at some time and place, such a monopoly exists, it provides a rationale for intervention in the market and regulation of the price and output of the monopolistic industry. Not all natural monopolies, for example, one barber, gas station, or banking office in a small town, trigger the adoption of regulation or a preemption of competition law. A genuine natural monopoly thus presents a policy issue of whether the government should regulate the market directly or should instead adopt a laissez-faire response. A regulatory regime that imposes conduct rules (e.g., prohibiting price discrimination lacking cost justification) or excludes willing entrants may rest on the perceived natural monopoly characteristics of the market (Kahn, 1970–71; Spulber, 1989; Posner,

3  Grossman, 2003, contends that no monopoly is inherently natural but rather regulation confers monopoly status.

36   Peter Carstensen Price AC (Industry)

Demand MES FIGURE  3.1  Natural

Firm

Quantity

Monopoly:  Minimum Efficient Scale Given Demand Requires Only One

1969). Historically, Congress adopted regulatory schemes for a number of industries considered to be “affected with a public interest” that made their services “essential” and limited or precluded the application of competition law to them (Rossi, 2006, 31–50). The resulting legislation usually erected legal barriers to entry into and exit from these industries, which foreclosed competition with the licensed firms (cf. Sidak and Spulber, 1997). However, this rationale for regulation does not necessarily warrant limitations on competition law. Competition law can play a role in policing how the monopolist uses its market power in related markets and in its dealings with its customers (see Otter Tail, 1973). It is important to be very cautious about the conclusion that an activity is naturally monopolistic. The claim necessarily rests on both the current state of technology (supply) and demand, so the durability of such a situation is vulnerable to change. Identifying a firm as a natural monopoly often leads to restrictions on entry and exit that may stifle technological change or prevent recognition of changes in demand. Technological developments (e.g., alternative ways to provide telecommunications or increased efficiency in electric generation capacity) have led to critical examinations of many industries presumed to be naturally monopolistic. These examinations generally have resulted in reversal of that conclusion (Spence, 2005). However, elements of some industries may retain monopolistic characteristics, as has been experienced in electric transmission and distribution (Pierce, 1994). The status of being a lawful monopoly can create expectations of protection even after the underlying technological rational for any monopoly has disappeared (Sidak and Spulber, 1997). Moreover, once an industry has evolved in a regulated monopoly mode, there are extremely difficult challenges in transforming it into a workably competitive one. The ongoing difficulties in telecommunications and electricity highlight the impact of past legislative decisions on the contemporary evolution of the market (Bush and Mayne, 2004, 208).

Economic Analysis of Antitrust Exemptions   37

3.3.  Market or Institutional Failure The majority of theories justifying exemption from or modification of competition law proceed from a hypothesis about market or institutional failure that makes standard competition policy inapplicable to the industrial context at issue. Because there are a variety of potential market and institutional failures there are also a number of failure specific theories to explain why a specific type of exemption or modification is necessary and appropriate to move the market to a more efficient outcome. These theories all involve a specific market characteristic or competition policy rule that implies that competition, though it exists, will not lead to efficiency. The claim is that unconstrained competition under those circumstances, the norm of competition law, would frustrate the efficiency ends that competition law should seek. Thus, these arguments all rest on a preference for some measure of economic efficiency over competition. Usually, the preference rests on some static measure of efficiency and so tends to ignore the potential for dynamic change in competitive markets that may well resolve the market or institutional failure. Indeed, adopting an exemption or modification can entrench a characteristic that would otherwise have disappeared through the process of dynamic change. Removing legislatively created exemptions is usually very difficult because of the vested economic interests that such rules usually create. There are six relatively distinct economic theories that advocates of exemptions and modifications have advanced. Each theory has a specific set of predicate factual conditions. There can be some overlap among these claims, but generally, each addresses a unique market or regulatory context. The first theory set forth is that of destructive competition based on the “empty core” hypothesis. This is one of most durable if least credible economic arguments for an exemption from antitrust law. The second theory rests on a claimed inconsistency between the scale of demand and the scale of supply that leads to inefficiency. Those who advance this view interpret competition law to condemn all horizontal arrangements even if they serve legitimate productive ends. They also question the utility of judicial oversight of such joint enterprises. Modern competition law principles would not recognize these as convincing substantive arguments for exemption.4 The third theory involves coordination among competitors that cannot be distinguished from the behavior of cartels that limit the scope, range, or character of their products. This theory in turn has three variations. First, by standardizing, competitors can create the necessary scale to induce efficient supply. Second, collective action among producers where their joint activities produce consumer value to capture more of that 4 

FTC and Department of Justice, Antitrust Division, Antitrust Guidelines for Collaborations among Competitors (Apr. 7, 2000). The transaction costs arising from reliance on antitrust litigation might still provide a rationale for such exemptions. See Carstensen, 2011 (many transportation industry exemptions may primarily protect legitimate joint ventures but serve to avoid the risks of costly antitrust litigation).

38   Peter Carstensen value can address potential disabling supply-side externalities. The third variant of this theory concerns the exchange of information among competitors. Such exchanges can also easily fit into the cartel model especially when related to forward-looking data, but the exchange may also have important efficiency-enhancing characteristics that serve the public interest in competition even as they may also standardize some elements of that competition. To avoid expecting courts to “set sail on a sea of doubt, and . . . assume[ ]‌ the power to say, in respect to contracts which have no other purpose and no other consideration on either side than the mutual restraint of the parties, how much restraint of competition is in the public interest, and how much is not” (Addyston Pipe, 284, 1898), legislation can define the exceptional cases that should be exempt from the general per se rule against naked restraints of competition as well as prescribe an appropriate alternative oversight process. The fourth market failure theory is a more generalized claim that some aspect of market relationships needs private regulation in ways that foreclose open competition in order to avoid market chaos. Hence, an exemption from the relevant mediates of competition law is warranted. The fifth theory involves the creation of private “countervailing power” to move markets toward a more overall competitive model. The sixth and final market failure theory applies a basic efficiency argument and involves the least economics because it only identifies prior failures either in competition law itself or in some regulatory regime where the best remedy from an efficiency standpoint is to move the market toward a fundamentally more competitive situation but which requires that one or more of competition law’s commands be modified.

3.3.1.  Destructive Competition and the Empty Core Historically, a number of justifications for competition law exemptions have invoked claims of “ruinous competition” or “destructive competition” in various forms. The central theme is that while the market has a number of competitors, absent control over their prices or output, they will compete in such a way as to bankrupt the industry. The historical record suggests that rapid market changes often lead to price wars, but that in time industry participants learn how to price in ways that provide reasonably stable long-run viability (see, generally, Lamoreaux, 1985). Nevertheless, claims of destructive competition recur in industries that have some unusual market characteristics. Various efforts have been made to develop economic models that might explain and justify the claim of destructive competition. The most developed of these justifications is the “null core” hypothesis. The theory is that an empty (null) core leads to destructive entry and market instability. If a market has a core, then no group of individuals can obtain an alternative outcome that all members of the group would prefer (Osborne and Rubinstein, 1994). If the core is “empty,” then no stable outcome is possible. Transportation industries are among the more prevalent applications of the

Economic Analysis of Antitrust Exemptions   39

theory, particularly ocean shipping conferences.5 A substantial body of analytical work has developed the theory of the core (or, in the current focus, its absence under competition) as it may apply to this industry (Telser, 1978). The following example provides the main insight of the theory. Assume an industry has three identical firms that provide shipping services, and a large number of identical customers who demand one unit of service. Entrants can enter the shipping industry by building a ship at cost F. Suppose each of the three firms has one ship of capacity 100, for a total industry capacity of 300. Market demand is 250 units, consisting of 250 customers each demanding one unit of shipping. If two ships operate, each earns economic rents because excess demand is 50 units and prices are elevated above the competitive level, the level at which economic profits are zero. As a result of fixed costs, each firm’s average cost declines the more it fills its ship to capacity. Thus, a ship using all 100 units operates most efficiently and has lower costs than any competitor operating at a lower capacity. Indeed, the third ship withdraws from the market because it cannot obtain sufficient revenue to cover its operating costs once its competitors have filled their ships. Given only two ships in the market and excess demand at the marginal cost of service, the two ships can now raise their prices to the monopoly level. Market instability occurs because the third, idle, ship can now undercut the price of the two other ships. This competition drives down price and displaces one of the incumbent firms, but as soon as the two survivors raise price back to the monopoly level, the excluded ship can undercut the price of the new set of dominant firms. This displacement continues indefinitely, making a stable outcome impossible.6 The model assumes that the idle firm can re-enter costlessly. This repeated re-entry leads to “ruinous” competition. A stable, competitive market price based on marginal cost is impossible; at least one market participant can disrupt any possible equilibrium. Because there is always a firm that can block any current allocation, leading to a sequence of market disruptions, the market has no competitive equilibrium. Moreover, because during each competitive price war, revenue falls to marginal cost, which is assumed to be substantially below average total cost (the level necessary for long-term survival), the results are both the bankruptcy of some of the firms and serious price volatility. Whether a null core might provide a basis for a competition law exemption depends initially on whether the theory is an accurate depiction of real markets. Under the economic theory of the null core, no level of output exists at which demand and supply are in equilibrium at prevailing prices, principally due to lumpiness in output capacity.

5  See Sjostrom, 1989. Telser, 1996, points out that a superadditive production function and nondecreasing returns to scale are necessary conditions for a nonempty core but that these alone are not sufficient to guarantee a nonempty core. 6  A key assumption is that the price that the remaining shippers will pay to have the 50 units of cargo transported is less than the marginal cost of providing such service. In a monopoly or duopoly situation, the shippers having the highest value for their shipments will take the service. Hence, the remaining shippers are assumed to place a lower value on having their goods moved. As discussed further in the text, when total demand increases relative to the capacity of the individual units, the chance that an empty core will occur declines rapidly because the marginal operator can still find sufficient business to stay in the market.

40   Peter Carstensen With excess capacity (i.e., if N + 1 firms are present in a market that can only profitably support N firms) no market mechanism will induce any particular seller to shut down as efficiency would require. Rather, any nonactive firm has an incentive to undercut the active firms. While entry is profitable at pre-entry prevailing prices, entry will leave some firms with negative economic profits. The negative economic profits will induce exit, which will return the industry to its prior unstable state, inducing entry once again. The empty core is a direct result of this “lumpiness” in costs.7 Hence, the main barrier to the existence of a core stems from a question of scale; if demand were sufficiently large, then the null core is unlikely to be a binding constraint in practice because the lumpiness becomes small relative to the scope of the industry.8 Periods of low demand or industry decline appear most likely to result in a null core problem, because demand may be small relative to the minimum efficient scale of supply and excess capacity may encourage “price wars.” If a firm or firms were willing to produce only a portion of the lumpy level of output, this would produce a competitive equilibrium; however, the avoidable fixed costs imply that no firm would choose unilaterally to do so. One way to achieve equilibrium might be through cartelization. An industry would coordinate to determine which firms will operate in the market and how much each will produce. A cartel can restrict output to Nq*, allowing excess demand to remain unmet. This could be achieved using measures such as deferred rebates,9 dual-rate contracts,10 or direct monitoring of customers to ensure they deal only with members of the cartel. Alternatively, the cartel could set production quotas above minimum efficient scale so that each firm equally shares responsibility for meeting the unmet demand. This could be accomplished by adopting technologies with higher unit costs but smaller capacity increments. Such measures

7  This occurs when, for example, investment in productive capacity involves fixed costs and variable costs are increasing. The resulting average cost curve of a “Viner firm” is U-shaped, which guarantees a unique cost-minimizing level of output, q*. The cost structure of these industries, and regularity conditions required for satisfying them, are described in Viner ([1931] 1952). Under general conditions, demand is continuous so that the supply curve and demand curve will intersect. In a Viner firm, the firm’s inverse supply curve is discontinuous at the price p *, equal to the minimum of its average cost curve. No production will occur at any price below p * as profits would be negative and firms could do better by exiting. At p *, a firm will produce q *. At any higher price, a firm would be willing to increase production, but would be undercut by potential entrants offering (q *, p *) to the market. This discontinuity introduces the lumpiness in supply, as production can only be efficient at some integer multiple of q *. If supply and demand intersect at some level of output between Nq * and ( N + 1) q *, then N is the number of firms that can profitably operate in the market, but there is unfilled demand that allows rent-seeking pricing and the resulting instability of the market. 8  As the size of the industry, N , grows large relative to the most efficient-sized firm, q *, the extent of the “integer problem” grows arbitrarily small. More formally, Nq * approximates market demand Q * arbitrarily well as N grows large. Since ( Nq *) / ( N + 1) < Q * / ( N + 1) < ( N + 1)(q *) / ( N + 1), a firm’s required production level to meet market demand converges to its minimum efficient scale as N grows large. For practical purposes, even when N is not large, excess demand is bounded above by q *. 9  Deferred rebates reward customers for exclusivity ex post. 10  Dual rate contracts are formal agreements to provide service at a discount in exchange for exclusivity enforced by imposing fines.

Economic Analysis of Antitrust Exemptions   41

require coordination among competitors and therefore provide justification for a statutory exemption from competition law. Cartelization is not the only way to overcome an empty core problem. One solution would be to manufacture for inventory, producing at the efficient scale but for only part of the time. However, in the transportation industries to which the concept of the null core principally has been applied, producing for inventory is not an option; excess capacity literally is lost when the ship sails, or the plane takes flight. Nevertheless, market institutions have emerged in transportation industries that mimic storage. Carriers can differentiate their service so that excess capacity serves a different market, such as transporting goods that are not time sensitive. The result is a slightly higher unit cost that stems from the inventory carrying costs, but this cost allows greater smoothness in supply that may solve the problem of the core. These strategies are well known from competition between airlines in the United States (Whinston and Collins, 1992). Similarly, natural gas pipelines have devised different levels of service: firm supply, interruptible supply, and so on. Each of these service types imposes different costs and risks, and contracting and secondary markets for capacity on natural gas pipelines has led to efficient allocations of capacity without cartelization (United Distribution Companies, 1996). The empty core model may also be too stylized for its predictions to be taken as a literal description of the behavior of any real-world firms. One reason is that the model rests on knife-edge results, and small changes in the parameters and market institutions may lead to substantially different predictions. For example, the probability that a market is in a competitive equilibrium depends on whether costs are heterogeneous. If all firms have identical costs, the probability of competitive equilibrium is zero, but even a modest variation in minimum average costs can increase the probability of equilibrium substantially (Sjostrom, 1989, 1164–65). Alternatively, if the increase in average variable costs just offsets the decrease in average fixed costs for production levels slightly above minimum efficient scale, the average cost curve will have a flat region instead of being U-shaped. Consequently, there would be a range of cost-minimizing output levels; and if demand is in that range, there is no indivisibility or integer problem. A more fundamental objection is that real market participants are more sophisticated and forward-looking than core theory supposes them to be. The core is a static concept and cannot capture concern for the effects of current behavior on future returns. Lumpiness and excess capacity are costly for market participants, but market turnover may be more costly. Any potential entrant knows it may be gaining only in the short run; it may in turn be forced out by the prices of others, and will not act myopically. Firms are not likely to base entry and exit decisions on market conditions in an instant of time. In addition, real-world negotiations are costly.11 The lessons from these examples are that market institutions are flexible, and adaptation often can resolve null core problems more efficiently than other means, such as competition

11  Sjostrom, 1989, presents evidence that if coalition formation is costly and the size of an industry is large, a competitive equilibrium is likely to occur.

42   Peter Carstensen law exemptions that encourage cartelization.12 More generally, the strict requirements of the empty core model show that destructive competition is unlikely in most real-world industries. Even if there is a risk of an empty core, authorized cartelization is only one possible response. In addition, it creates significant risks of anticompetitive outcomes.

3.3.2.  Achieving Efficient Scale for Component Production If some component in the production process has a high fixed cost, and the quantity of the finished product demanded is relatively low, production costs will create a bottleneck such that a firm dominating that segment can exercise significant market power over the end product market. If the downstream users of the component can collaborate to produce it jointly, they can avoid the bottleneck.13 Such customers or sellers may have an incentive to overinvest in that stage to ensure that there will be no shortage.14 These are legitimate joint ventures, so the need for an “exemption” or even “modification” of competition law may seem hard to explain on the basis of contemporary doctrine. Because joint production projects involve horizontal collaboration and in some circumstances may have to limit output, necessitating some kind of allocation process, they are vulnerable to competition law challenges.15 But in the event of a challenge, the relevant competition law doctrines including “essential facilities” are not well developed.16 As a result, challenges, even if not meritorious, will involve extensive and costly litigation expenses in order to resolve the merits of the claimed violation. In addition, the courts are not well adapted to the ongoing administration of the rights and obligations for such bottleneck facilities, especially when disputes on access or allocation arise.17 These factors in combination could provide a justification for an competition law exemption for the operation and management of such productive

12 

ABA, 2007, 161–92, examines the application of Core Theory to the shipping conferences under the Shipping Act. See also Sagers, 2006. 13  The same analysis applies when the actors are upstream sellers into a similar monopsonistic market. 14  By hypothesis, variable cost is low, so the investors largely confront one-time investments. See Carstensen, 1992, 25–26 (farmer-owned grain elevators are an example of building facilities at the largest scale). 15  Two institutional considerations may provide a link. First, some cases, e.g., Topco (1972), have been read very broadly to condemn all agreements having any restrictive characteristics among competitors. While most antitrust lawyers recognize that such a reading is inaccurate, risk-averse investors and executives may still be deterred from rational conduct. Second, defending an antitrust case has a significant potential transaction cost even when the defendant prevails. Hence, for firms with limited resources, the risk of litigation may itself deter rational collaboration. See Carstensen, 2011. 16  The Trinko decision has questioned the viability of the essential facilities doctrine. Trinko, 2004, 410–11. 17  The Trinko, 2004, decision rests in part on the Court’s belief that the FCC and state regulators are better positioned to manage the issues of access to shared capacity than an antitrust court. Id. 412–14.

Economic Analysis of Antitrust Exemptions   43

facilities based on the simultaneous adoption of an alternative regulatory system that would oversee the operation of such a collective enterprise. For example, high fixed costs in research and development (R & D) may preclude significant R & D activity by even large firms if the sunk R & D investments are significant, as they are for integrated chip technology and some other high-technology areas. Hence, smaller firms may find R & D costs daunting. However, minimum efficient scale in production may be small enough to enable the same small firms to compete for sales of products once the products are developed. Allowing these smaller firms to coordinate R & D to achieve economies of scale, while maintaining product market competition, may be preferable to exit by a large number of producers unable to survive due to high individual R & D costs.18 Such a collective, productive enterprise will not in itself violate competition law although its specific conduct might. The obstacle to more collaboration might reside in a risk-averse view on the part of potential participants who require greater assurance that they will not face significant litigation costs as a price of joining the venture. In other areas such as airline code sharing or railroad rate bureaus dealing with rail car exchanges and lease rates, the parties face similar concerns. This may explain why firms engaged in apparently legitimate joint productive activities seek an exemption that appears to imply that the goals of the arrangement are anticompetitive.19 But other routes exist to achieve some or all of the insulation provided by statutory exemption or modification.20 But where an industry has a history of exemption, the “path dependency” of the legal and social institutions may lead parties to want to retain vestigial exemptions to avoid some litigation risks. In a slightly different case, an exemption can permit collaboration in one market while maintaining competition in another. Competing internationally may involve a greater minimum efficient scale than competing domestically. For example, an individual company may not be able to maintain sales offices and distribution facilities throughout the world profitably. However, a group of firms may be able to achieve minimum efficient scale in international marketing and trade operations. Such collaboration may involve both market allocation among competitors and price agreements on the joint sales. Such agreements could therefore raise competition law concerns under traditional case law, although current law and policy on such legitimate collaborations makes this unlikely.21 As a result, an exemption from competition law with respect to international trade could enable a group of firms in an industry to compete

18  The National Cooperative Research and Production Act may illustrate this concern. See ABA, 2007, 263–77. 19  For an analysis of the transportation exemptions supporting this thesis, see Carstensen, 2011. 20  The business review process of the Antitrust Division is a prime example. 26 C.F.R. § 50.6. 21  See, e.g., Topco (1972), which many read as a prohibition on any territorial allocation among competitors; see, e.g., Meese (2004). In fact, the Supreme Court sustained a less restrictive allocation of territories after remand (Topco, 1973). Moreover, Topco never in fact used those restraints and has continued to prosper as a joint venture. See First and Carstensen (2007).

44   Peter Carstensen internationally by creating a joint international marketing effort, while maintaining interfirm competition domestically. But if the primary function is only to create an efficient scale for global marketing, an exemption is unnecessary as a matter of substantive competition law.22 This justification rests primarily on the institutional limits of competition law as practiced and the failure to develop a more active program of business review clearances. As a result, the real-world process of avoiding competition law challenges to legitimate joint productive activities involves a perception and perhaps a reality of significant costs. An exemption or modification of competition law to limit those costs therefore provides only a very weak economic rationale for the creation or retention of an exemption.

3.3.3.  Coordination among Competitors to Create an Efficient Market The production of a good for consumption may require, or be more efficient with, coordination among competing suppliers in some dimensions. In contrast to the preceding justification, competitors are not engaged in collaboration in the actual production or distribution of the good or service. Instead, the problem is that the most efficient market context requires some express regulation of the ways in which the firms compete. Often government provides such regulation; indeed, this is a basic function of government. However, in other market contexts, no specific rule-maker exists with the authority and skills necessary to provide the standards or criteria under which the market is to operate. Moreover, collective action by competitors to define the terms of competition is fraught with competition law risks. Naked restraints on competition often involve the use of such standards to limit or suppress competition (Trenton Potteries, 1927, 393; Fashion Originators Guild, 1941; see generally Lande and Marvel, 2000). Modern competition law in fact has sought to accommodate such regulation when it is legitimate, although the articulation of the criteria for these determinations remains illusive.23

22  The Department of Commerce has defended the Export Trade Certificate program as lowering expected litigation and transaction costs even though the joint ventures it approves may not violate antitrust law. Testimony of John J. Sullivan, General Counsel, U.S. Dep’t. of Commerce Before the Antitrust Modernization Commission, Dec. 1, 2005 available at http://govinfo.library.unt.edu/amc/​ commission_hearings/pdf/Sullivan_Statement.pdf 23  See, e.g., Oklahoma, 1984 (upholding some regulation of competition in college football, but holding regulation of television rights to be outside the scope of authorization), Allied Tube & Conduit, 1988 (rejecting a safety standard based on bad process); see generally, Carstensen and Roth, 2000 (contending that regulatory agreements among competitors are lawful if public policy, i.e., legislation or other indicia of public acceptance, has authorized such regulation of competition, the restraint is within the scope of that authorization, and the process used was appropriate to the situation).

Economic Analysis of Antitrust Exemptions   45

Once again, the limits of competition law doctrine and potential high costs of resolving such issues in a competition law case provide a justification for seeking either exemption or modification of competition law. The contexts can be further divided into demand externalities, supply externalities, and information pooling.

3.3.3.1.  Network (Demand) Externalities A network or demand externality exists if the utility that consumer A receives from consuming a product depends on whether others consume that same type of product (and, potentially, who else consumes it). If significant demand externalities exist, competitors may want or need to coordinate on standards with each other. Creating compatibility among rival networks increases the value of the product to consumers by increasing the total size of the network. This increases the set of consumers and so has a positive effect on the utility of each consumer who purchases this product.24 Standard setting in telecommunications is a prime example. Consumers ultimately demand phone conversations, not phones or network interchange. The demands for phones and related service equipment derive from the demand for phone conversations. As a result, while telecommunications companies compete to provide cell phones and cell phone service, they must coordinate with their competitors to ensure interoperability to provide the good consumers ultimately consume, the resulting calls, including calls with consumers on competing networks.25 Coordination among competitors is an alternative to achieving a network through a single, monopoly firm. If a single firm offered the complete product of a phone conversation under all circumstances, such a firm likely would be a monopolist over a relatively large area, in the nature of the original AT&T network. It is evident that it is more efficient to have competition in supply of the component products used to make the conversation while enabling the supply of the ultimate product, the communication, through standards setting. However, the collective action of setting standards and their subsequent enforcement raises competition law concerns if not done by a government agency. The case law shows that in fact competition law courts will allow legitimate standards setting. But the courts are not adapted to engage in finely tuned review of the merits of regulation. Hence, shared jurisdiction between competition law and an activity-oriented regulator is likely to be the more efficient response (Hovenkamp, 2004). In such contexts, a statutory exemption may be useful to specify more precisely the scope of authority of the two legal regimes. The central point is that classification and other rules can define and facilitate markets where strategic conduct by either buyer or seller can occur. This is a particular issue in the context of networks where the potential for strategic conduct is very great. As 24  There is an extensive literature on network effects. See, e.g., Shapiro and Varian, 1994; Katz and Shapiro, 1985. 25  Interoperatiblity through network standardization does not preclude individual providers from seeking through various means to create barriers to keep their customers from switching to other competing suppliers.

46   Peter Carstensen discussed below, market-facilitating regulations are important to economic efficiency in other areas as well.

3.3.3.2.  Supply Externalities Consumer demand for professional football provides an example of this problem. Consumers want games and seasons, not individual teams, even if they root for the home one. Thus, a football team as a firm cannot supply the product unilaterally; one team in isolation is not entertaining. The product, the football game, which is generated as a result of coordination between two teams, creates the entertainment for the spectator, and the need for multiple teams to interact on games, seasons, and playoffs creates a supply externality. A supply externality means that an increase in the supply by a competitor increases a firm’s own supply, in this example by lowering the overall costs of producing the final product.26 In particular, a joint venture can coordinate a “season” and joint sales of the resulting product while competition for inputs such as players and coaches continues. Moreover, the collective enterprise produces value for spectators that its participants may not be able to capture. The potential problem is that the rights of the individual teams to charge for the programs they create may not provide sufficient revenue to cover the costs of producing the overall program of games. This might be particularly true where teams have differently sized markets and so have divergent revenue streams directly available to them. In such a situation, the league might fail and the overall supply of games would be lost. One solution to this problem is to facilitate the capture of more of the value created by the collective enterprise. But to capture additional revenue may require collective action both to limit supply (thus creating and collecting monopoly profits) and to allocate the resulting revenue so as to subsidize the teams facing the greatest economic challenges. Manifestly, such conduct might be questionable under competition law and so rational teams would ask for an exemption from that law. The central issues here are the scope of lawful collaboration in marketing the collective product. To the extent courts have interpreted competition law to ban any collaboration in sales, the firms may have a justification to seek some relief from such a prohibition. Thus, it may be justifiable on economic grounds to permit professional football teams to restrict through collective action the sale of their supply of football games if a popular but individually insufficiently remunerative sport otherwise would cease to operate. Supply externalities exist because multiple entities need to coordinate certain economic activities but also continue to compete in other aspects of the market. The externalities can also be internalized if the group combines into a single firm. Indeed, it is possible that a single firm might be more efficient than a collective of firms because it could better control all aspects of production. A number of recently formed sports

26  Although only a few examples exist, it is possible to have professional team sports without leagues or other joint structures. The Harlem Globetrotters are one of the best-known examples.

Economic Analysis of Antitrust Exemptions   47

leagues, such as Major League Soccer in the United States, have been organized as single-firm entities, presumably to achieve this kind of internalization. As in the case of network effects, the claim of supply externality sufficient to cause a serious reduction in or loss of valuable goods or services requires a number of very specific facts. The football example is illustrative because in that case, the central economic consideration to achieving survival of professional football was sharing of revenue and not the enhancement of revenue. In addition, the authorization of such collective market control beyond that permitted by competition law itself creates significant risks of excessive exploitation and other strategic conduct. Only direct regulation of prices, locations, or terms of access can address such issues. Thus, creating special market power without a balancing regulatory oversight of its use is likely to result in suboptimal performance.

3.3.3.3.  Information Pooling Sharing information that includes forward-looking projections can be essential to efficient pricing in some service industries, notably property and casualty insurance, where the cost of the service is primarily a function of estimated future expenses. Efficiency gains can be substantial if the pool of information is large. When the information comes from multiple firms the participants must both coordinate their data classification and share their projections. The ability to monitor the overall risk stems from access to information about the types of insured parties (their risk factors) and their historical risk performance. The coordination among competitors to gather and maintain information on the risk profiles of insured parties can reduce the total costs of insurance and mitigate market failure, in the form of a disintegrating risk pool, which can result from adverse selection and moral hazard. The existence of such an information pool also reduces the barriers to entry into the business of insurance since the new entrant does not have to develop its own database. Because the cost of information related to risk in an insurance pool may decline over reasonable ranges of output, an insurer acting unilaterally may not be able to organize such information efficiently. Moreover, more limited databases are likely to result in less well-informed individual pricing decisions. Information sharing is not necessarily illegal under competition law; however, when the information includes projections of future expenses (e.g., insurance claims) that are the primary element in the cost of the product being sold, the concerns grow. Information sharing of this sort could expose those engaged in the sharing to continued risks of competition law litigation. Those risks might, therefore, deter the most efficient and comprehensive data collection and analysis. The McCarran-Ferguson Act exempts insurance from federal competition law oversight and assigns oversight to the states, thereby permitting insurance companies to share information with less risk of competition law intervention (ABA, 2007, 133–60). As a general matter, this economic analysis supports only a narrow and carefully crafted exception to competition law that formulates a safe harbor for essential information exchange. Many markets have developed other ways of pooling information

48   Peter Carstensen without requiring direct interaction of producers. Commodity and securities markets are examples where demand for forward-looking information has called forth a variety of means of producing that data.

3.3.4.  Market Facilitation in General All these foregoing examples illustrate some aspects of market failure and in doing so illustrate the fact that markets are not inherently efficient or rational. It is unrealistic to think of firms existing outside some framework of law and regulation that constitutes the institutional context within which the firm transacts with its customers and suppliers. When viewed from the perspective of an industry or market, the challenge is to have a set of institutional conditions that facilitate rather than frustrate rational and efficient buyer and seller choices. A fundamental function of the legal system is to create a set of rules that provide this institutional context within which businesses operate. However, within specialized business activities, the generic rules may be insufficient to deal with inefficient practices that can only be remedied by collective action. Standard setting in the context of network effects is a good example. If producers and their customers can agree on specifications for products, that makes substitution and comparison easier, thereby stimulating competition along the remaining dimensions of the business, for example, quality or price. Some markets, particularly insurance, operate based on risk pooling where the members of the pool share common characteristics. The concept is that even though individual risks may vary, members of a reasonably defined pool can share the risks of that group in order to achieve the efficiencies, including risk spreading over a mass market for the services involved. Absent agreement and enforcement by collective action, a risk pool can disintegrate as a result of moral hazard and adverse selection by insureds and skimming the pool by insurers. Adverse selection involves hidden information: in the case of insurance, the insured party knows more about its risk profile than does the insurer. Hence, those with greater risk insure while those with low risk do not. Moral hazard involves hidden action or inaction:  the insurer cannot observe whether the insured party is taking steps (e.g., through behavioral changes) to reduce the risk of adverse outcomes. Conversely, insurers can sometimes observe that some members of the risk pool are better risks. If the insurer can selectively insure those parties and reject the members of the pool likely to have claims, it can earn excess income. Given competing insurers, regulations requiring nondiscrimination among members of the pool and even rate regulation can be responses intended to deal with these risks. However, the logical linkage between rates and pooling is tenuous at best since such regulation undermines the objective of using competition for the business of members of the pool to ensure efficient and fair prices. If such restraints come from direct industry action, they represent limits on competition of a sort that require express exemption from competition law to be lawful. But if

Economic Analysis of Antitrust Exemptions   49

the classifications and nondiscrimination rules come from a regulator, then the controls define the market and potentially facilitate competition.27 Standard setting has also been a recurrent source of anticompetitive harms. Carefully crafted limits on competition law might facilitate legitimate standards setting without providing an opportunity for undue market foreclosure. The Standards Setting Organization Development Act of 2004 provides a narrow exemption from treble damages for organizations with specific organization and operational characteristics, but does not exempt the members if they exploit the organization for anticompetitive purposes (see ABA, 2007, 263–76).

3.3.5.  Creating “Countervailing” Power to Facilitate Efficiency If an economic actor is perceived to exercise market power in an economic relationship and this market power could result in inefficient outcomes, then one possibility is to enable countervailing market power to offset the potential deleterious effects on allocative efficiency that competition between the asymmetric economic agents could yield (see Galbraith, 1952). Where, for example, a monopsonistic buyer can make an all or nothing offer to its suppliers, it can appropriate all of the economic rents that had accrued to its suppliers (Blair and Harrison, 1993, 73). This imbalance in turn affects the willingness of such suppliers to invest in and sustain their businesses. Over time this can have a serious negative effect on the efficiency of the industry despite the economic gains to the monopsonist. This economic justification could explain a number of competition law exemptions including those for labor union collective bargaining (Norris-LaGuardia Act) and exemptions for the formation of cooperatives among farmers and fisherman (Capper-Volstead Act, Fishermen’s Collective Marketing Act). However, except in the most general terms, no advocate of these statutes advanced this theory (Galbraith, 1952). Some of these same exemptions also could be justified on the basis that they reduce contracting costs. A group of sellers or buyers without market power can still collaborate to agree on prices and other terms of trade with a limited set of buyers. This reduces the costs of negotiation and may limit the ability of the more powerful actor to engage in strategic conduct. Reduction in such risks in turn may induce more efficient behavior and greater productivity on the part of the group that is able to protect itself. A third consideration is that collective agreements may provide protection for idiosyncratic investments by individuals that improve their productivity. Such investments would be made only if a reasonable return would exist. The opportunity for employers, for example, having induced such investments to play off workers against each other

27  Where the affected industry members agree on regulations to advocate to the regulator, no statutory exemption is necessary because the Noerr-Pennington doctrine will shield the activity. Eastern RR Presidents, 1961; Pennington, 1965.

50   Peter Carstensen could reduce incentives to invest in firm-specific training. Collective bargaining provides the potential for protection from being “held up” by the firm, that is, employees will not receive higher pay for their higher productivity resulting from the training. This claim assumes that the employer would require a substantial number of employees with similar skills, hence its post-training capacity to engage in strategic conduct. Further, individuals would have to recognize their collective interest. This has not always occurred, as experience with franchise investments and chicken grower contracts demonstrate. In many circumstances public regulation of the contractual process may provide better outcomes than attempting to authorize private countervailing power.

3.3.6.  Mitigating Prior Failures in Market Regulation Two types of mitigation of prior regulations might provide a basis for justifying an exemption. First, the deleterious effects of prior regulation might require that the industry be allowed to engage in conduct or mergers that could otherwise violate competition law in order to achieve a transformation in its structure or conduct consistent with long-run workable competition and efficiency. Second, the courts might interpret the competition law in ways that create inefficiency in particular industries or contexts. If the legislature is unwilling or unable to devise a general revision of the statutes to overcome the problem it perceived in the judicial interpretation, it can respond with a more focused exemption or modification of the laws to insulate a particular industry from the doctrinal problem.

3.3.6.1.  Mitigating Regulation Competition law has sometimes been interpreted to focus exclusively on consumer welfare rather than aggregate welfare (Salop, 2005). While this is a generally useful criterion when applied in well-functioning markets where the focus of concern is seller power, the exclusive focus on consumer welfare would be inefficient in certain settings, particularly when industry restructuring may be needed to mitigate the deleterious effects of historical regulation or when buyer power is the cause. The case of railroads is one example. Many railroads and shippers contended that regulation under the ICC had become so inefficient that the structure of railroad ownership and pricing was in dire need of change. However, many of the changes that were argued to be efficient in achieving increases in social welfare, in particular mergers, would impose welfare losses on some consumers in the short term, at least. Under conventional competition law merger analysis these reorganizations would be vulnerable to challenge because of these anticompetitive effects. Such justifications may explain why the Staggers Rail Act of 1980 and related legislative changes in regulation of the railroads, with the overall goal of deregulation, continued to exclude competition law review of mergers and instead assigned exclusive jurisdiction to the regulatory agency (ABA, 2007, 193–216). Bank merger law (Bank Merger Act, Bank Holding Company Act) with its “convenience and needs” defense for mergers provides another example of

Economic Analysis of Antitrust Exemptions   51

the potential application of this justification (ABA, 2007, 50; the defense has not proven to be relevant in any merger litigation). Such exemptions ought to be transitional interventions to allow the industry to be transformed. A concomitant requirement would seem to be some kind of regulatory oversight of the process to ensure that the changes in structure and conduct actually serve the underlying goal of developing a workably competitive context. In the case of airlines, Congress terminated the exemption from competition law review of mergers, but in the rest of public transportation these arguably transitional exemptions continue to exist, authorizing anticompetitive arrangements and foreclosing competition law review of specific harmful conduct.

3.3.6.2.  Mitigating Competition Law Doctrine Areas of competition law doctrine could be classified as improper if, under a modern rule-of-reason analysis, the behavior likely would not be considered a competition law violation but the courts for reasons of stare decisis failed to recognize the legitimate aspects of the behavior and instead continued to enforce the doctrine. A weaker version of this justification exists where a lower court has entered a judgment that legislators believe is an improper application of competition law and an exemption is proposed as a way to avoid further litigation costs even if the end result of that litigation would be rejection of the erroneous ruling. The economic argument therefore combines transaction cost savings with enhanced market efficiency resulting from the elimination of inefficient doctrines. For example, legislators justified the Soft Drink Interbrand Competition Act as a way to authorize vertical restrictions including exclusive territories to soft drink bottlers that the FTC had sought to condemn. The legislative majority contended that this organization was economically optimal. Experience showed the error of this assumption, but the exemption remains (ABA, 2007, 277–84). Indeed, a number of the statutory exemptions had their origins in reactions to specific lower-court decisions that a majority of Congress deemed incorrect applications of competition law. But in most of those instances, such as the Charitable Gift Annuity Antitrust Relief Act of 1995, Congress immunized conduct that ordinarily would violate competition law. Hence, the reason for the exemption was that the proper enforcement of competition law would condemn such conduct, but for some other reason Congress wished to protect it.

3.4.  Wealth Transfers and Protection from Competition Efficiency achieved through vigorous competition is only one of many goals that a government could pursue (Calabresi and Melamed, 1972). In any number of settings, some may claim that alternatives to vigorous competition might or will achieve social,

52   Peter Carstensen economic, or political goals. These goals can range from protection of specific socially desired groups (e.g., farmers or workers) to managing allocations of resources in times of extreme scarcity such as war. Illustrative are exemptions such as the Robinson-Patman Act’s permission of price discrimination favoring nonprofits and protection of rebates by cooperatives (15 U.S.C. § 13(b), Corgill, 2001). This economic argument rests on a cost-benefit analysis including a consideration of alternative ways to accomplish the wealth transfer. The central contention is that limiting or eliminating competition law’s application to some activity or industry will result in increased wealth that the recipients are inherently worthy of or will then be applied to serve a socially desirable purpose. Moreover, the economic argument should demonstrate that the method chosen to accomplish the wealth transfer is both appropriate in terms of who is paying as well as receiving the transfer, and efficient in the sense of imposing no greater cost than necessary to achieve the desired goal. This latter criterion implies that the limit on competition law is the least restrictive option available to the legislature. The negative characterization of these arguments is that they reflect “rent seeking” through manipulation of the legislature. The result is that special privileges are given to politically powerful interest groups. Such groups usually advance the arguments set forth below in various forms as the justification for the rights being sought. While a realistic view of much of the legislation of this sort is that its primary purpose is to reward such groups based on their political power and that no creditable public interest based justification exists, nevertheless it is important at this stage of this analysis to recognize that a plausible economic argument exists for rewarding specific groups when the reward in fact serves a legitimate public purpose. There are three variations on this economic argument.

3.4.1.  Conferring Market Power to Achieve Specific Goals There are several contexts in which competition law is modified or excluded from application where the economic argument is that such a change will assist in achieving some specific goal of public policy. While these types of exemptions all involve essentially the same economic argument, it is useful to separate them to show their core claims are similar.

3.4.1.1.  Social Goals Socially desirable results often require the investment of resources. Direct subsidy from government or indirect subsidies via modification of tax law are well-known examples. But if the beneficiaries of these subsidies are in competition they may dissipate the gains that they received. Hence, the legislature can seek to provide financial support for such activities by modifying the laws of competition to reduce the impact of robust competition on efforts to raise or protect funds for various kinds of favored activities. This model assumes that the activity involves some kind of market conduct that faces competition and that the impact of competition is to reduce the gain available to subsidize the desired activity.

Economic Analysis of Antitrust Exemptions   53

The analysis necessary to support this justification is that the protected actors are engaging in socially desirable activity that some segment of the population supports or ought to support, but the forces of the market impair the capacity of the actors to capture that gain. This is a balancing test in which the costs to the supporters of the activity are less than the resulting value to society of the activity being supported. Nonprofit entities intended to promote public welfare require charitable contributions for their funding. However, competition among such institutions for contributions can both raise the costs of fundraising and reduce the value of some types of contributions. The general encouragement of collective fundraising through United Way and similar joint solicitation programs responds to the need to reduce costly fundraising efforts. Such joint programs do not raise inherent competition law problems because they are examples of legitimate joint ventures. But if nonprofits sell services whose price includes a charitable contribution component and they face competition in the sale of such services, the result is likely to be a reduction in the charitable element. Because the charitable organization has inflated its price relative to the cost of the service in order to capture funds for its charitable goal, competition would be likely to reduce the charitable overcharge. Hence, allowing such competitors to collude to limit or eliminate price competition will, in theory, increase their ability to raise funds for their charitable objectives. One specific example is competition among charities to offer annuities under which a donor receives an annual payment and any funds remaining at the death of the donor revert to the charity. If charities compete with each other on the payment to donors, then Bertrand competition could leave charities bidding away all their contributions by offering annuities that leave no residual for the charity. This would defeat the objective of the charity offering the annuity and may well even frustrate the desires of the annuitant. If charities were permitted to collude on annuity rates, then the charities could set the rate in terms of benefit to society rather than individual benefit. Such a justification could support the Charitable Gift Annuity Antitrust Relief Act of 1995. Of course, simply allowing charities to fix the rates of annuities does not guarantee that the resulting gains will go to the intended charitable use. Thus, logically the charity would also have to ensure that the funds captured by this collusion go to a charitable use. Other potential examples of this justification exist. Colleges, especially private schools, compete for students in part based on offers of financial aid (i.e., a discount on the list price of education). However, if the college has decided that it will grant financial aid based strictly on need, it may find itself competing with other similar colleges that have made a more generous determination of a student’s need. If Congress concluded that competition based on the price of private college education did not serve the public interest, this could justify allowing such colleges to agree on the need of students for whom they were competing and thereby avoid price competition through financial aid (see, e.g., Need Based Financial Aid Act, 2007). This would increase cost to those students for whom colleges wanted to compete while either lowering the cost to those for whom they did not compete (i.e., their financial aid, tuition discounts, would be larger)

54   Peter Carstensen or reducing the expenditure on total financial aid (i.e., increasing the revenue of the college as a result of reduced price competition). Once again, authorizing collusion alone permits the increase in effective price, but without further oversight of how the resulting gains are used it is questionable whether the public interest goals for such an exemption would be served.

3.4.1.2.  Protecting National Champions If a country has a cost advantage in some global market, it may want to exploit that advantage by allowing a firm to monopolize the product so that it can maximize the gains it can achieve in the market. Such entities, thus, have the appearance of monopolies or cartels and require exemption from domestic competition law in order to operate. By controlling the amount and timing of exports, the national champion can extract the maximum economic rents from the global market. Those funds in turn are often returned to the government or to the participants in the venture, for example, wheat farmers in Canada or milk producers in New Zealand. The effect on domestic production is to depress output to some degree in order to control supply and manage price. Globally, the effect is to raise prices and transfer wealth from other countries to the national champion. The roots of this perspective are in a mercantilist view of the world in which each country tries to maximize its wealth. A similar economic logic explains the historic justification for exempting export cartels from competition law. By eliminating competition among exporters they could in theory increase their prices to foreign buyers. The Webb-Pomerene Export Act, 1918, authorized such cartels in the United States but only permitted them to eliminate competition for foreign sales. The economic reasoning behind such cartels is that if there were competition in export sales, the result would be lower prices to the foreign buyer and less profit to the domestic firm. This argument rests on a number of assumptions. Cartels can increase the wealth of the nation only if other countries do not outlaw such exploitation of their markets or retaliate with comparable cartels. It also assumes that the domestic producers have a cost advantage over all other producers in the world. If they did not, then assuming competition from other producers, it would be difficult, absent global collusion, to raise prices. In sum, the ability to create cartels to serve either the public or the national interest is very limited in fact. While an appealing argument in some circumstances, it is unlikely to produce the benefits claimed if implemented. Finally, there are some domestic national champions. In these cases, the argument is that to induce the enterprise to make socially desirable investments it must be given the ability to exploit those segments of the market that might otherwise be open to competition. For many decades, American Telephone and Telegraph’s (AT&T) monopoly of telecommunications services in the United States rested on a version of this argument. That is, it provided national long-distance service (subject to rate regulation) in return for being protected from competition. Early history shows that there was in fact more potential for competition in telecommunications services, but AT&T convinced regulators and legislators that its monopoly provided

Economic Analysis of Antitrust Exemptions   55

better service than any other option for many decades. Similarly, electric and natural gas providers have made similar arguments when some part of their activity is amenable to competition. Basically, the concept is that the low-cost (i.e., potentially competitive) classes of customers should pay a premium to support services to others. The argument for such cross subsidy is probably more creditable in contexts where there are serious limits to the capacity of government to provide such services and where the overall market is suitably transparent so that there is reasonable assurance that the wealth appropriated will be deployed to serve otherwise unservable needs.

3.4.1.3.  Promoting National Security The final economic argument in this group rests on concerns for national security. One variant restricts entry into specific types of business by foreign firms or nationals. The theory is that the economic gains to competition are not worth the risk of foreign control of important elements of national security. For example, foreigners are allowed to own only 25% of American commercial airlines and foreign airlines cannot provide service between domestic airports in the United States. The historic rational was that airlines were a potential part of the national defense and so control needed to rest with Americans. Similarly, domestic ocean shipping and passenger service is reserved for American operators, some of which today are indirectly owned by foreign companies. The economic argument was that the inefficient and costly domestic service ensured that more seamen would exist so that in the event of a war the navy could expand with experienced personal. These claims may seem antiquated and probably are, but in context they rest on a plausible theory of how to maintain specific skills and resources that might be valuable in time of war. There is a significant potential cost to the economy of these actions. Moreover, the parties paying for these services may not be representative of the community getting the benefit. Thus, there is a potential for significant distortion. Only careful examination of the allocation of costs and benefits could determine whether this method of maintaining reserve resources is the best option. A different, but related, security issue arises in the production of military equipment. On the one hand, governments want domestic production because that is the only secure source of equipment in times of war. But on the other hand, demand in peace may be low as the volume required is low. Thus, both restrictions on the identity of those enterprises that can compete and potential market allocation plans to preserve alternative suppliers in times of low demand may well exist. In these cases, government incurs substantial expense to obtain its necessary supplies, but in return maintains a domestic source or sources of supply. Once again, the cost-benefit analysis can provide the necessary economic theoretical support for the restraints. The risks to national security of a fully competitive, international market may be sufficient to justify restraints on the identity of the firms and level of competition within that market.

56   Peter Carstensen

3.4.2.  Subsidization of a Favored Activity A related justification for protecting particular actors or activities from the ordinary rules of the market is that the costs of protection are worth the benefit to some other socially desirable objective. As an economic justification, this implies a cost-benefit analysis in which the consumers (or sellers) paying the costs are appropriate parties to bear that cost, and the cost is reasonable in relation to the social benefit that will result. For example, maintaining competition in “points of view” for news while allowing monopoly pricing of newspapers and advertising to be shared between papers having different points of view may be preferable to reduced diversity in news (Newspaper Preservation Act, 1970). These limits on competition go beyond those that arise from joint publishing itself that would be lawful under standard competition law analysis. Once again, the exemption alone will not guarantee that the public interest goal will be served. Indeed, increased profits from eliminating competition might, in fact, induce less rather than more competition in points of view. Hence, to achieve the subsidy’s goal it is necessary to provide some mechanism to ensure that the funds are used for the desired goal. Professional sports leagues may provide another example of the same notion of selective subsidy intended to support a socially desired activity. If Congress concluded that professional football leagues were important social institutions but had inadequate income to compensate players and coaches sufficiently to induce the socially optimal level of performance, then it could provide an additional subsidy for these leagues by authorizing them to eliminate competition in the sale of broadcast rights to capture increased revenue (Sports Broadcasting Act, 1961). Once again, the exemption serves only to increase profits, and some other mechanism must exist to ensure the intended recipients receive the increased resources. The network effects justification for sports leagues having some immunity from competition law, already discussed in section 3.3 above, reinforces this economic argument. Commercial spots have, the argument runs, both a network effect and face the fact that individual teams cannot capture much of the value produced by league play. Hence, in combination, these two economic arguments provide support for exempting collective action by leagues from competition rules.

3.4.3.  Wealth Transfer to a Favored Class of Market Participants The final kind of subsidy involves a choice to protect some class of actors from the rigor of competitive markets because the individuals involved in the favored class are perceived as socially desirable. The economic premise is that by shielding some kind of anticompetitive activity by the members of this class, their economic position will be enhanced at the expense of other classes of actors. As in all other examples of wealth transfer, this implies a cost-benefit analysis that identifies who will gain and who will pay. To have effect, the rights created by law must have reality in the marketplace. Otherwise, the apparent subsidy becomes meaningless.

Economic Analysis of Antitrust Exemptions   57

Several exemptions might rely on this justification. In agriculture and fishing (Capper-Volstead, 1922; Fishermen’s Collective Marketing Act, 1933) a premise is that the participants are particularly socially desirable. Hence, allowing them to create cooperatives and consolidate those cooperatives to create bargaining power in the sale of their produce can be justified as transferring wealth to a particularly deserving group (see generally ABA, 2007, 87–132). A similar justification may exist for labor union exemptions. Individual workers may face a great deal of buyer power in labor markets. Hence, allowing workers to organize unions and engage in collective bargaining can be seen as a means of transferring wealth from employers to employees. The specific choices of the groups of workers to be so advantaged and the scope of their authority to engage in collective activity are the ways in which the grant of rights is fitted to the underlying justification. All three of these examples raise serious questions whether the exemption has accomplished its intended objective. First, there is the question of who pays this subsidy. For example, should families with many children subsidize dairy farmers? Second, the grant of an exemption facilitates the creation of useful economic power only if other conditions are satisfied. Given the relatively small share of the total workforce in labor unions, it is again an open question how this exemption has contributed to the general well-being of workers.28 Lastly, these exemptions do not in themselves direct the benefits they create to any specific set of beneficiaries. Much of any benefit resulting from these exemptions may be captured by parties other than the intended recipients. Earlier, these same exemptions were presented as examples of ways of resolving market failures. There are serious collective action and free rider problems in markets with many sellers and relatively few buyers. Hence, as suggested in section 3.3.5 above, these same exemptions can draw on an economic argument based on creating countervailing power to facilitate efficient market transactions.

3.5.  Exemptions Facilitating the Transition of Industry Structure from State Ownership or Direct Regulation to Market Orientation When national policy toward competition changes with respect to regulated industries, there is often a need for a substantial transition period in which it may be doubtful that

28 

Labor unions have also advocated for generally applicable protection legislation that has served the interests of all workers.

58   Peter Carstensen ordinary competition law should apply. Essentially, the goal is to restructure an industry either by some type of dissolution and transfer of ownership or by indirect means that open the market to competition and entry. Such transitions usually require an administrative agency to implement the legislative scheme. Intervention in the process by ad hoc judicial declarations can disrupt and complicate this process. Thus, the theory is that efficient transition requires a distinct process and administration. In addition, in many cases, there will remain significant market bottlenecks that require continued regulatory oversight. For example, in electricity, the transmission and distribution networks are inherently monopolistic. Moreover, the nature of electricity requires that an entity undertake a variety of tasks necessary to coordinate the flows over the system and balance supply and demand. Hence, while competition in generation and potentially in retailing electricity is possible and likely to promote efficiency, other parts of the system require centralized control. To ensure that the competitive elements are able to function efficiently, there needs to be an administrative oversight of the noncompetitive elements. This in turn creates a potential for conflict between the administrative agency and generalist courts administering competition law. One resolution is to confer exclusive jurisdiction on the regulatory agency even when it has a mandate to promote a competitive market. Essentially then, the economic analysis is one of administrative efficiency. In Tinko, 2004, for example, the Supreme Court seemed to embrace the idea that a administrative agencies, state and federal, should have a wide scope of authority to oversee the markets rather than allowing the courts to intervene. However, the opinion also suggests that Otter Tail, 1972, remains good law as well, which again suggests that there must be a balancing of when agency oversight is a more efficient solution with its lesser sanctions and when a court with broader remedial powers is to be preferred.

3.6.  Exemptions That Improve the Efficiency of the Enforcement of Competition Policy Closely related to the foregoing theory is a related one that focuses on the institutional choice among agencies that might enforce competition policies. The key difference is that there are some industries, including insurance, banking, securities, and agricultural commodities, that are inherently competitive but subject to substantial regulatory oversight. The goal of the regulation is to facilitate and advance economically desirable competition. This substantially duplicates what courts would do in enforcing competition law. The specific rules and regulations, however, may differ and create conflict. Moreover, the enterprises subject to such dual control may find the system unduly costly with respect to compliance. Again, an economic model of efficient market administration provides a basis for an argument for either exclusive jurisdiction in the regulatory

Economic Analysis of Antitrust Exemptions   59

agency based on its comparative expertise with respect to the industry or alternatively a policy of primary jurisdiction in which the generalist courts before making any decision refer the dispute to the agency to learn its views on the merits of the conduct at issue (Ricci, 1973). There are counterarguments that agencies are subject to the risk of capture by the regulated industries. Hence, an independent source of competition law authority it essential to protect the interest in obtaining a competitive market process. Certainly, in the United States, the tendency has been to allow dual enforcement in many such contexts (RCA, 1959; El Paso, 1964, Silver, 1963). Again, the economic efficiency test would be whether the cost savings from exclusive or primary jurisdiction are sufficient to offset the potential losses to competitive efficiency that may come from agency inaction or special interest influence.

3.7. Conclusion It should also be evident that there is a potential to combine network claims, domestic champion arguments, and partial natural monopoly models to justify a result. Indeed, it is generally the case that the specific arguments for exemptions from competition law and policy can be stacked on top of each other. A few are mutually exclusive, but most emphasize different aspects of a market issue and so can provide a cumulative economic argument for limiting the application of competition law. This effort to state sympathetically the economic theories supporting exemptions still results in a largely negative overview. Each of the various theories requires specific factual contexts. Moreover, those facts are themselves likely to be highly contentious in any particular context and are not necessarily long-run givens of any situation. As a result, while there are coherent economic theories for limiting the scope and application of competition law, they are not very powerful or pervasive.

References American Bar Association (ABA). 2007. Federal Statutory Exemptions from Antitrust Law. Monograph 24. Chicago: American Bar Association. Blair, Roger, and Jeffrey Harrison. 1993. Monopsony: Antitrust Law and Economics. Princeton, NJ: Princeton University Press. Breyer, Stephen. 1982. Regulation and Its Reform. Cambridge, MA: Harvard University Press. Buchanan, James M., and Gordon Tullock. 1962. The Calculus of Consent: Logical Foundations of Constitutional Democracy. Ann Arbor: University of Michigan Press. Bush, Darren and Carrie Mayne. 2004. In (Reluctant) Defense of Enron: Why Bad Regulation Is to Blame for California’s Power Woes (or Why Antitrust Law Fails to Protect Against Market Power When the Market Rules Encourage Its Use. Oregon Law Review 83: 207.

60   Peter Carstensen Calabresi, Guido, and Douglas Melamed. 1972. Property Rules, Liability Rules, and Inalienability: One View of the Cathedral. Harvard Law Review 85: 1089. Carstensen, Peter C. 2011. Replacing Antitrust Exemptions for Transportation Industries: The Potential for a “Robust Business Review Clearance.” Oregon Law Review 89: 1059. Carstensen, Peter C. 1992. The Content of the Hollow Core of Antitrust: The Chicago Board of Trade Case and the Meaning of the “Rule of Reason” in Restraint of Trade Analysis. Research in Law and Economics 15: 1. Carstensen, Peter C., and Bette Roth. 2000. The Per Se Legality of Some Naked Restraints: A [Re] Conceptualization of the Antitrust Analysis of Cartelistic Organizations. Antitrust Bulletin 65: 349. Corgill, Dennis S. 2001. Distributing Products under the Nonprofit Institutions Act:  Price Discrimination, Arbitrage, and Fraud in the Pharmaceutical Industry. Brigham Young University Law Review 2001: 1383. First, Harry, and Peter Carstensen. 2007. Rambling Through Economic Theory: Topco’s Closer Look. In Dan Crane and Eleanor Fox, Antitrust Stories. New York: Foundation Press. Galbraith, John K. 1952. American Capitalism:  The Concept of Countervailing Power. Boston: Houghton Mifflin. Grossman, Peter. 2003. Is Anything Naturally a Monopoly? In Peter Grossman and Daniel Cole, The End of Natural Monopoly: Deregulation and Competition in the Electric Power Industry. Boston: JAI. Hovenkamp, Herbert. 2004. Antitrust and the Regulatory Enterprise. Columbia Business Law Review 2004: 335. Kahn, Alfred. 1970–71. The Economics of Regulation: Principles and Institutions. New York: Wiley. Katz, Michael and Carl Shapiro. 1985. Network Externalities, Competition and Compatibility. American Economic Review 75: 424. Lamoreaux, Naomi. 1985. The Great Merger Movement in American Business, 1895–1904. New York: Cambridge University Press. Lande, Robert, and Howard Marvel. 2000. The Three Types of Collusion: Fixing Prices, Rivals, and Rules. Wisconsin Law Review 2000: 941. Meese, Alan. 2004. Intrabrand Restraints and the Theory of the Firm. North Carolina Law Review 83: 6. Osborne, Martin, and Ariel Rubinstein. 1994. A Course in Game Theory. Cambridge, MA: MIT Press. Pierce, Richard. 1994. The State of Transition to Competitive Markets in Natural Gas and Electricity. Energy Law Journal 15: 323. Posner, Richard. 1969. Natural Monopoly and Its Regulation. Stanford Law Review 21: 548. Rossi, Jim. 2006. Regulatory Bargaining and Public Law. New York: Cambridge University Press. Sagers, Christopher. 2006. The Demise of Regulation in Ocean Shipping: A Study in the Evolution of Competition Policy and the Predictive Power of Microeconomics. Vanderbilt Journal of Transnational Law 39: 779. Salop, Steven. 2005. Anticompetitive Overbuying by Power Buyers. Antitrust Law Journal 72: 669. Shapiro, Carl and Hal Varian. 1994. Information Rules: A Strategic Guide to the Network Economy. Boston: Harvard Business School Press.

Economic Analysis of Antitrust Exemptions   61

Sidak, J. Gregory, and Daniel Spulber. 1997. Deregulatory Takings and the Regulatory Contract: The Transformation of Network Industries in the United States. New York: Cambridge University Press. Sjostrom, William. 1989. Collusion in Ocean Shipping:  A  Test of Monopoly and Empty Core Models. Journal of Political Economy 97: 1160. Spence, David. 2005. The Politics of Electricity Restructuring: Theory vs Practice. Wake Forest Law Review 40: 417. Spulber, Daniel. 1989. Regulation and Markets. Cambridge, MA: MIT Press. Telser, Lester. 1978. Economic Theory and the Core. Chicago: University of Chicago Press. Telser, Lester. 1996. Competition and the Core. Journal of Political Economy 104: 85. Tullock, Gordon. 2005. The Rent-Seeking Society. The Selected Works of Gordon Tullock, vol. 5. Indianapolis: Liberty Fund. Viner, Jacob. [1931] 1952. Cost Curves and Supply Curves. Zeitschrift fur Nationalokonomie 3: 23. Reprinted in George Stigler and Kenneth Boulding, Readings in Price Theory 6: 198. Chicago: R. D. Irwin. Whinston, Michael, and Scott Collins. 1992. Entry and Competitive Structure in Deregulated Airline Markets: An Events Study Analysis of People Express. Rand Journal of Economics 23: 445.

Cases: US Supreme Court Allied Tube & Conduit Corp. v. Indian Head, 486 U.S. 492 (1988). Eastern Railroad Presidents Conference v. Noerr Motor Ft. Co., 365 U.S. 127 (1961). Fashion Originators Guild of America v. FTC, 312 U.S. 457 (1941). NCAA v. Oklahoma, 468 US 85 (1984). Otter Tail Power Co. v. United States, 410 U.S. 366 (1973). Ricci v. Chicago Mercantile Exchange, 409 U.S. 289 (1973). Silver v. New York Stock Exchange 373 U.S. 341 (1963). United Mine Workers v. Pennington, 381 U.S. 657 (1965). United States v. El Paso Gas Co., 376 U.S. 651 (1964). United States v. RCA, 358 U.S. 334 (1959). United States v. Topco Associates, 405 U.S. 596 (1972). United States v. Topco Associates, 414 U.S. 801 (per curiam 1973). United States. v. Trenton Potteries, 273 U.S. 392, 393 (1927). Verizon v. Law Offices of Curtis V. Trinko, 540 U.S. 398 (2004).

Cases: US Courts of Appeal Addyston Pipe & Steel v. U.S., 85 Fed. 271 (6th Cir. 1898), aff ’d, 175 U.S. 211 (1899). United Distribution Companies v. FERC, 88 F.3d 1105 (D.C. Cir. 1996).

62   Peter Carstensen

Statutes Bank Holding Company Act, 1966, amending 12 U.S.C. § 1849(b). Bank Merger Act, 1966, amending 12 U.S.C. § 1828(c). Charitable Gift Annuity Antitrust Relief Act,1995, 15 U.S.C. §§ 37–37a. Capper-Volstead Act, 1922, 7 U.S.C. §§ 291–92. Fishermen’s Collective Marketing Act, 1933, 15 U.S.C. §§ 521–22. McCarren-Ferguson Act, 1945, 15 U.S.C. §§ 1011–15. National Cooperative Research and Production Act, 1993, 15 U.S.C. §§ 4001–4. Need Based Financial Aid Act, 2007, 15 U.S.C. § 1, Note. Newspaper Preservation Act, 1970, 15 U.S.C. §§ 1801–4. Norris LaGuardia Act, 1932, 47 Sta. 70 codified at 29 U.S.C. §§ 101–13. Robinson-Patman Act, 1936, 15 U.S.C. § 13. Soft Drink Interbrand Competition Act, 1980, 15 U.S.C. §§ 3501–3. Sports Broadcasting Act, 1961, 15 U.S.C. §§ 1291–95. Staggers Rail Act of 1980, 94 Stat. 1895 (1980) codified as amended in 49 U.S.C. §§ 701–706, et seq. Standards Setting Organization Development Act, 2004, amending 15 U.S.C. §§ 4301–4. Webb-Pomerene Export Act, 1918, 15 U.S.C. §§ 61–66.

CHAPTER 4

H E A LT HC A R E P R OV I DE R A N D PAY E R M A R K E T S CORY S. CAPPS AND DAVID DRANOVE

4.1.  Antitrust in the Provider and Payer Sectors Among developed nations, the US healthcare system is unique in its dependence on competition. For the past 75 years, private health insurers have competed for enrollees. For the past 25 years, medical providers have competed to be included in insurer networks. The Obama administration’s 2010 healthcare reform legislation, the Patient Protection and Affordable Care Act (ACA), while controversial in many respects, reaffirmed the important role of competition in health insurance and provider markets.1 Although the US healthcare system may be dependent on competition, a cursory glance at the system suggests a surprising dearth of competitors. Many health insurers, notably those operating under license of the Blue Cross and Blue Shield Association, have market shares of 50 percent or higher in many metropolitan areas.2 At the same time, many areas of the nation are dominated by just one or two large provider systems.3 The US Department of Justice (DOJ) and the Federal Trade Commission (FTC) have taken notice and increased their enforcement activities. 1 

Patient Protection and Affordable Care Act, Pub. L. No. 111-148, 124 Stat. 119–1025 (2010). Although precise and reliable estimates of health plan market shares are difficult to obtain (Dafny et al. 2011), multiple data sources and the authors’ experience indicate that this is indeed true in many areas. 3  Examples include Northern Virginia, where Inova accounts for about two-thirds of hospital beds and volume; the Bay Area, where Sutter has about a 40 percent share; and Pittsburgh, where UPMC has about a 55 percent market share (Capps and Dranove 2011; West Penn Allegheny Health System, Inc. v. UPMC, 627 F.3d 85 (3d Cir. 2010)). 2 

Table 4.1  Litigated Hospital Merger Cases Year

Case

Location

Blocked?

United States v. Rockford Mem. Hosp., 717 F. Supp. 1251 (N.D. Ill. 1989), aff’d, 898 F.2d 1278 (7th Cir.)

Rockford, IL

Yes

Augusta, GA

Yes

1

1989

2

1991 US Fed. Trade Comm’n v. University Health, 1991-1 Trade Cases ¶69,400 (S.D. Ga.) and 1991-1 Trade Cases ¶69,444 (S.D. Ga.), rev’d, 938 F.2d 1206 (11th Cir. 1991)

3

1994 In re Adventist Health Sys./West, FTC Docket No. Ukiah, CA 9234 (ALJ Dec. 9, 1992); In re Adventist Health Sys./ West, FTC Docket No. 9234 (Apr. 1, 1994); Ukiah Adventist Hospital v. FTC, No. 93-70387 (9th Cir. May 18, 1994)

No

4

1995 US Fed. Trade Comm’n v. Freeman Hospital, 911 Joplin, MO F. Supp. 1213 (W.D. Mo. 1995), aff’d, 69 F.3d 260 (8th Cir. 1995)

No

5

1995 United States v. Mercy Health Services, 902 F. Supp. 968 (N.D. Iowa 1995), vacated as moot, 107 F.3d 632 (8th Cir. 1997)

Dubuque, IA

No

6

1996 US Fed. Trade Comm’n v. Butterworth Health Corp., 946 F. Supp. 1285 (W.D. Mich. 1996), aff’d per curiam, No. 96-2440 (6th Cir. July 8, 1997)

Grand Rapids, MI No

7

1997 United States v. Long Island Jewish Med. Ctr., 983 F. Supp. 121 (E.D.N.Y. 1997)

Long Island, NY

No

8

1998 US Fed. Trade Comm’n v. Tenet Healthcare Corp., 17 F. Supp. 2d 937 (E.D. Mo. 1998), rev’d 186 F.3d 1045 (8th Cir. 1999)

Poplar Bluff, MO

No

9

2000 California v. Sutter Health Sys., 84 F. Supp. 2d 1057 (N.D. Cal.), aff’d mem., 2000-1 Trade Cas. (CCH) U 87,665 (9th Cir. 2000), revised, 130 F. Supp. 2d 1109 (N.D. Cal. 2001)

Oakland, CA

No

10

2004 In re Evanston Northwestern Healthcare Corp., No. 9315 (F.T.C. Aug. 6, 2007)

Evanston, IL

N/A

11

2008 In re Inova Health Sys. Found., No. 9326 (F.T.C. May 8, 2008)

Manassas, VA

Abandoned

12

2011

In re ProMedica Health Sys., No. 9346 (F.T.C. Mar. 28, Toledo, OH 2012), aff’d per curiam, No. 12-3583 (6th Cir. Apr. 22, 2014)

Yes

13

2011

Settled, no divestiture

14

2012

US Fed. Trade Comm’n v. Phoebe Putney Health Sys., Albany, GA No. 1:11-cv-00058-WLS (M.D. Ga. June 27, 2011), rev’d and remanded, No. 11-1160 (U.S. Feb. 19, 2013) US Fed. Trade Comm’n v. OSF Healthcare Sys., No. Rockford, IL 11 C 50344 (N.D. Ill. Apr. 5, 2012)

Source: Authors’ review of hospital merger cases.

Yes

Healthcare Provider and Payer Markets   65

Since 2008, the DOJ has challenged alleged anticompetitive practices by Blue Cross and Blue Shield of Michigan and by United Regional Health System in Wichita Falls, Texas; the DOJ has also challenged health plan mergers in Michigan and Montana.4 During that same period, the FTC successfully challenged a number of proposed hospital mergers (see table 4.1). There has also been a flurry of private antitrust activity, including (1) a class certification case involving a consummated hospital merger that the FTC concluded had lessened competition and increased prices, (2) challenges to alleged exclusionary practices by large hospital systems, and (3) several cases aimed at ending seven decades of territorial division by Blue Cross and Blue Shield plans.5 The effectiveness of the US market-based health system may be shaped by the outcomes of these and similar cases. In this chapter, we discuss the economics of antitrust in US healthcare markets.6 We begin by describing key institutional details and presenting some of the seminal research on healthcare competition. We then discuss antitrust issues in provider markets, including mergers and exclusionary practices. We then turn to antitrust issues in insurer markets. We conclude with a discussion of the antitrust implications of healthcare reform.

4.2.  How Healthcare Markets Work Understanding the application of antitrust law to healthcare requires a basic understanding of healthcare institutions. In this section, we describe the healthcare “value chain,” the role of the health insurer as a third-party intermediary, and the evolution of payer and provider markets.

4.2.1.  The Healthcare Value Chain Suppose that you woke up experiencing mild chest pain. You consider going to the emergency room but instead make an appointment to see your primary care physician (PCP). During your visit, the PCP asks about your health, takes your blood pressure,

4 

Respectively, these cases are summarized in US DOJ (2010a, 2011a, 2010b, and 2011b). In January 2012, the Court of Appeals for the Seventh Circuit overturned the district court’s decision to deny certification to a class of patients and self-funded plan sponsors suing Evanston Northwestern Healthcare, the hospital system that the FTC sued in 2004 (Messner v. Northshore Univ. HealthSystem, 669 F.3d 802 (7th Cir. 2012). Leibenluft (2010) reviews several private cases involving alleged monopolization by hospital systems. Cases recently brought against the “Blue” plans include Richards v. Blue Cross & Blue Shield of Ala., No. cv-01133 (N.D. Ala. filed Apr. 17, 2012); Cerven v. Blue Cross and Blue Shield of N.C., No. 5:12-cv-00017 (W.D.N.C. filed Feb. 7, 2012); and UPMC. v. Highmark, Inc., No. 2:12-cv-00692 (W.D. Pa. filed May. 23, 2012). 6  Antitrust is generally not an issue in other nations where provider prices are heavily regulated and the state is the dominant (if not only) insurer. There have been several antitrust cases in the Netherlands, 5 

66   Cory S. Capps and David Dranove and performs an electrocardiogram. On the basis of these test results, the PCP refers you to a cardiologist. The cardiologist performs additional tests and informs you that you have a blocked artery that is too advanced for treatment with statin drugs. The cardiologist refers you to a cardiothoracic surgeon who considers all the test results and concurs with the diagnosis. The surgeon discusses two alternative interventions:  coronary angioplasty, in which a stent is inserted to open the clogged artery, and coronary artery bypass surgery (CABG), in which the clogged passage is removed and replaced by arterial tissue taken from your thigh. Of these two options, the surgeon believes that CABG is the best choice. You go home, do some research on the Internet, and talk things over with your PCP and your cardiologist. Everyone agrees that you should have CABG surgery. The surgeon is affiliated with one of the best-known hospitals and offers to schedule the procedure for the upcoming week. Your medical care saves your life and also generates a lot of revenue for your providers. Your PCP will make about $100 and the cardiologist $500; the surgeon will earn $2,500. The hospital does best of all, netting $5,000 or more.7 These fees are approximate and depend on who is paying the bill. Private insurers tend to be more generous than government payers (i.e., Medicare for the elderly and disabled and Medicaid for the indigent). If you are among the 15 percent of Americans who do not have insurance, then your medical bills might be highest of all, although many providers have charity care policies and are also prepared to write off unpaid medical bills as bad debt.

4.2.2. Pricing Physicians sell thousands of different “products.” These are commonly referred to as procedures, and each procedure is associated with a Current Procedural Terminology (CPT) code. Hospitals also sell a wide range of products. These are sometimes sold on the basis of Diagnosis Related Groups, or DRGs. Alternatively, hospitals may set prices for each individual service provided during a hospital stay. These are specified in hospital “chargemasters.” The term “charges” is often used to describe the list prices set by providers. Unlike many other businesses, where customers normally pay the price that has been set by the seller, few patients actually pay charges. This partly reflects the fact that most patients have insurance and therefore are responsible for only a portion of the medical where the government recently began deregulating elements of its system and introducing market forces (Capps, Varkevisser, and Schut 2008). For a broader discussion of competition and antitrust outside the United States, see Gaynor and Town 2011). 7  Earnings are net of costs and do not include the opportunity cost of the physician’s time (Birkmeyer et al. 2010).

Healthcare Provider and Payer Markets   67

bill. But even factoring in the insurer’s payment, the total amount paid rarely equals the amount charged. Private health insurance proliferated in the United States after World War II. By 1960, well over half of all Americans had private health insurance. Initially, most private insurers paid full charges. This created an incentive for providers to increase prices, because their patients paid, at most, nominal copayments and therefore did not choose their provider on the basis of price. Insurers eventually capped payments at “Usual, Customary, and Reasonable” (UCR) fee levels. The “usual” fee is the fee that the provider most frequently charges for the procedure. The “customary” fee is based on a percentile of the fees charged for the same procedure by other providers in the community. “Reasonableness” is a loose concept used to handle exceptional billing circumstances. A  similar dynamic occurred in the hospital market. Initially, insurers paid hospitals their full charges but then moved to a cost-based reimbursement model analogous to regulated utility pricing. UCRs and cost-based reimbursement may limit exorbitant provider payments, but they do not give providers incentives to reduce prices. Indeed, cost-based reimbursement may have discouraged hospitals from becoming more efficient. Throughout the 1960s and 1970s, provider prices grew at a much faster pace than general inflation. The tide turned in the 1980s after the introduction of “selective contracting.” Firms providing administrative services (sometimes known as ASO firms) for self-insured employers saw an opportunity to hold down hospital prices.8 ASO firms made the following offer to local hospitals, many of which had excess capacity: “Reduce your prices to our clients, and we will steer more patients to your hospitals.” At the same time, ASO firms approached their clients with the following offer: “Establish a differential cost-sharing plan in which employees have lower copayments if they visit hospitals that offer discounts.” For employers looking to save money on health insurance and hospitals looking to fill empty beds, these offers were too good to refuse, and such plans grew rapidly over the ensuing years. Providers who agree to give discounts to an insurer or ASO are said to be part of that entity’s “provider network,” and insurance arrangements that rely on selective contracting as a major method for cost containment are known as “preferred provider organization” (PPOs). PPOs can create provider networks in two ways. First, they may negotiate with individual providers. Because most patients strongly prefer broad access to care, it is not surprising that most negotiations end successfully; providers agree to give discounts, and those providers are included in the PPO network. Second, because it can be impractical for PPOs to negotiate with small provider organizations, such as solo physician practices, PPOs often propose a fee schedule that providers either accept or decline. 8  Many employers in the United States, including nearly all those with over 500 employees, self-insure for their employees’ healthcare spending. They are fully responsible for employee health spending (although some purchase stop-loss policies). Most self-insured firms contract with a third party to handle claims processing and other administrative aspects of insurance. These third parties include noninsurer ASO firms and traditional health insurers.

68   Cory S. Capps and David Dranove

4.2.3.  The Evolving Nature of Competition in Healthcare Prior to selective contracting, patients had free choice of provider. Thanks to generous insurance coverage, price was rarely a primary consideration. Observing these perverse incentives, health economists argued that competition among healthcare providers was unlikely to have the salutary effects of traditional textbook competition, and they proposed an alternative theory of hospital competition known as the Medical Arms Race (MAR). According to MAR theory, hospitals competed for patients by using expensive medical technology and generous staff support to attract doctors. Hospitals passed the costs along to insurers through cost-based reimbursement. Insured patients followed their doctors’ referral decisions and were largely unconcerned about the cost of care. Many empirical studies that use data from the 1980s find evidence consistent with the theory of the MAR.9 In seminal research, James Robinson and coauthors find that hospitals with more competitors tend to have higher costs per patient (Robinson and Luft 1985) and more employees per patient (Robinson 1988), and they offer more high-tech services (Robinson, Garnick, and McPhee 1987; Luft et al. 1986). Noether (1988) finds that hospital costs are higher in less concentrated markets, which is consistent with Robinson’s findings. However, she also finds that prices (measured as full charges) are uncorrelated with the degree of competition, suggesting that consumers might not bear the burden of the MAR. Dranove, Shanley, and White (1993) argue that ad hoc measures of competition in reduced form MAR models might be correlated with unobservable patient severity and/or hospital quality, causing omitted variable bias. Adding additional variables to account for potential patient flows across regions diminishes the estimated magnitude of the MAR. Overall, these empirical studies of the MAR fed the conventional wisdom that healthcare competition was harmful. The growth of selective contracting over the course of the 1980s led economists to challenge the conventional wisdom. To test whether the effects of competition had changed, Zwanziger and Melnick (1988a) estimate separate Robinson-style reduced form regressions for consecutive years of data in California in the 1980s, a period of rapid PPO growth. They confirm that costs and concentration were negatively related at the beginning of their sample period but that, in the latter years, the relationship was gone. Zwanziger and Melnick (1988b) show that the rate of growth of costs in California during this time was much smaller in more competitive markets. Dranove, Shanley, and Simon (1992) use a similar empirical framework to examine prices and price/cost margins in California over a similar time period. They find that price/ cost margins were falling during their sample period of 1983–1988, presumably as a result 9  Studies of physicians show that utilization and spending are higher in markets that have more physicians per capita. This is offered as evidence of “supplier induced demand” whereby physicians respond to competitive pressures by convincing patients to receive more medical services. There is substantial disagreement in this literature on the quality of research methods used to demonstrate inducement (Dranove and Satterthwaite 2000).

Healthcare Provider and Payer Markets   69

of pricing pressure brought about by selective contracting. More importantly, margins were unrelated to market concentration in 1983, but they were significantly and positively related in 1988. By using national data through the early 1990s, Connor, Feldman, and Dowd (1998) also find a reversal of the cost-concentration relationship. The overall conclusion is that the growth of selective contracting eliminated or even reversed the MAR. Later research suggests that the beneficial effect of market forces on hospital pricing began to diminish by the turn of the century. Although precise causality has not been established, researchers attribute this to two mutually reinforcing market shifts that occurred at roughly the same time. First, consumers began to demand more inclusive hospital networks. Second, the hospital industry experienced a wave of consolidation in the mid-1990s. These likely combined to increase hospital systems’ negotiating leverage and resulted in higher prices.10 In summary, the growth of selective contracting injected competition into healthcare markets. Whereas insured patients (and their physician agents) were relatively insensitive to pricing, PPOs could steer enrollees to lower-priced providers. A series of reduced form regression studies confirm the traditional, positive correlation between price and concentration in markets dominated by managed care. Altogether, this research strongly suggests that, as in traditional markets, competition in healthcare markets tends to lower costs and improve welfare. In the remainder of this chapter, we discuss the economic literature, case law, and key enforcement actions of the DOJ and the FTC in three areas that, together, account for approximately two-thirds of healthcare spending and 12 percent of GDP (CMS 2010). Our discussion of hospitals focuses on merger policy and exclusionary conduct. Our discussion of physician services examines trade-offs between market power and efficiency and the agencies’ enforcement record against what are often alleged to be, in effect, naked cartels. Our discussion of health insurance explores issues of both monopsony power and monopoly power. We conclude with a discussion of the likely effects of healthcare reform on provider and insurer competition.

4.3.  Antitrust and Hospitals Most antitrust cases involving hospitals center on three main issues: hospital mergers; allegedly anticompetitive exclusionary conduct with respect to physicians; and allegedly anticompetitive exclusionary conduct with respect to horizontal rivals, such as specialty

10 

Town and Vogt (2006) surveyed dozens of studies of the relationship between hospital concentration and price, and they conclude that an 800-point increase in the Herfindahl-Hirschman Index (HHI) within a metropolitan area leads, on average, to about a 5 percent price increase. The HHI is computed as the sum of the squared market shares of the participants in a market. Thus, for example, an 800-point increase would result from a merger of two firms in a market with five equally sized firms (the HHI would increase from 202 + 202 + 202 + 202 + 202 = 2, 000 to 402 + 202 + 202 + 202 = 2, 800).

70   Cory S. Capps and David Dranove hospitals and ambulatory surgery centers. Each arena has its own history and invokes a distinct set of legal and economic paradigms.

4.3.1.  Hospital Mergers Through the end of the 1980s, the FTC and the DOJ lost only one litigated case in which they opposed a hospital merger.11 In the 1990s, litigation outcomes dramatically reversed, and the FTC and DOJ lost six consecutive cases. In 2001, the State of California attempted to block a hospital merger in Oakland and also lost.12 Following this string of seven losses, prospective hospital merger enforcement essentially ceased for a decade. During roughly the same time that state and federal agencies were losing hospital merger cases, the industry was undergoing a wave of mergers and acquisitions. That wave peaked in 1996, when there were approximately 150 hospital mergers and over 300 mergers and system acquisitions.13 As a result, by 2000, hospital ownership was substantially more concentrated than it had been in 1990. Since 2008, however, the FTC has, based on decisions to date, successfully blocked three of the four hospital mergers it opposed.14 The fourth case, Pheobe Putney, followed a very winding path. The FTC initially lost at both the district court level and in the Eleventh Circuit, not on the basis of a failure to establish a substantial risk of harm to competition from the merger but rather on the basis of state action immunity (i.e., the doctrine that states can supplant competition with regulation).15 The FTC appealed that ruling to the Supreme Court, which ultimately ruled unanimously in favor of the FTC.16 In a final twist, after its Supreme Court win, the FTC settled the case prior to remand and without a divestiture agreement. The FTC abandoned its efforts to obtain a divestiture because

11 

United States v. Carilion Health Sys., 707 F. Supp. 840 (W.D. Va. 1989). Litigated hospital merger cases dating back to 1989 are listed in table 4.1. 13  With respect to hospitals, it is common to distinguish between “mergers” and “acquisitions.” “Mergers” are transactions in which separate hospitals combine their operations under a shared license. “Acquisitions” are transactions in which hospitals retain their licenses but are owned by a common governing body (Cuellar and Gertler 2003). 14  In April of 2014, the Sixth Circuit upheld the FTC’s win in ProMedica. ProMedica Health Sys., Inc. v. Fed. Trade Comm’n, No. 12-3583 (6th Cir. Apr. 22, 2014). 15  Indeed, the Eleventh Circuit observed: “We agree with the [FTC] that, on the facts alleged, the joint operation of Memorial and Palmyra would substantially lessen competition or tend to create, if not create, a monopoly.” US Fed. Trade Comm’n v. Phoebe Putney Health Sys., No. 1:11-cv-00058-WLS (11th Cir. Dec. 14, 2011), 9. 16  “We hold that Georgia has not clearly articulated and affirmatively expressed a policy to allow hospital authorities to make acquisitions that substantially lessen competition. The judgment of the Court of Appeals is reversed, and the case is remanded for further proceedings consistent with this opinion.” US Fed. Trade Comm’n v. Phoebe Putney Health Sys., No. 11-1160, slip op. (U.S. Feb. 19, 2013), 19, available at http://www.supremecourt.gov/opinions/12pdf/11-1160_1824.pdf. 12 

Healthcare Provider and Payer Markets   71

Georgia’s certificate of need (CON) law made it “unlikely that any possible divestiture buyer could obtain the necessary CON approval to operate an independent hospital.”17 Putting aside the unique history of the Phoebe Putney case, two key factors underpin the sharp turnaround in federal hospital merger enforcement. First, the Evanston case—a consummated merger case stemming from Evanston Northwestern Healthcare’s (ENH) acquisition of Highland Park Hospital (HPH) in 2000—laid a strong foundation for future prospective merger enforcement. By using data and documents spanning four years after the closing of that acquisition, the FTC established that the merger had resulted in substantial postmerger price increases and that those increases were the result of market power ENH gained from the acquisition.18 Second, well before the FTC brought its case against ENH—and spurred in large part by the remarkable string of losses in the 1990s—a number of economists studied hospital competition more closely. The insights from that body of research, particularly as regards geographic market definition and competitive effects analysis, appear to have played a significant role in sparking the enforcement turnaround. Two landmark studies explore that research. One study, Improving Healthcare: A Dose of Competition, was released by the DOJ and the FTC in 2004 and examined all aspects of healthcare competition, including hospital mergers (US DOJ and FTC 2004). The second study was a survey sponsored by the Robert Wood Johnson Foundation titled “How Has Hospital Consolidation Affected the Price and Quality of Hospital Care?” (Town and Vogt 2006).19 The FTC/DOJ report summarizes the body of economic research on hospital competition as follows: Most studies of the relationship between competition and hospital prices have found that high hospital concentration is associated with increased prices, regardless of whether the hospitals are for-profit or nonprofit.

17 

US Fed. Trade Comm’n, “Hospital Authority and Phoebe Putney Health System Settle FTC Charges That Acquisition of Palmyra Park Hospital Violated U.S. Antitrust Laws,” news release, August 22, 2013, available at http://www.ftc.gov/opa/2013/08/phoebe.shtm. 18  Regarding price increases, the Commission in Evanston stated as follows: There is no dispute that ENH substantially raised its prices shortly after the merging parties consummated the transaction. . . . [T]‌he evidence demonstrates that the transaction enabled the merged firm to exercise market power and that the resulting anticompetitive effects were not offset by merger-specific efficiencies. The record shows that senior officials at Evanston and Highland Park anticipated that the merger would give them greater leverage to raise prices, that the merged firm did raise its prices immediately and substantially after completion of the transaction, and that the same senior officials attributed the price increases in part to increased bargaining leverage produced by the merger. In re Evanston Northwestern Healthcare Corp., No. 9315, slip op. at 4–5 (F.T.C. Aug. 6, 2007), available at http://www.ftc.gov/os/adjpro/d9315/070806opinion.pdf. 19  In a follow-up survey, Gaynor and Town (2012) review post-2006 research and reiterate the prior conclusion that hospital competition tends to reduce prices and improves quality.

72   Cory S. Capps and David Dranove The Town and Vogt survey reached a similar conclusion: Research suggests that hospital consolidation in the 1990s raised prices by at least five percent and likely by significantly more.

The research surveyed in the DOJ/FTC and Town and Vogt reports has shown that many—though surely not all, and very likely not even a majority of—hospital mergers in the 1990s resulted in increased market power and higher prices for hospital services. The agencies targeted mergers that seemed most likely to increase prices. Why then did the agencies lose so many cases? Two factors explain the defeats. The first, and more significant, is that courts accepted patient-flow-based methods as appropriate for defining relevant geographic markets for hospitals. These flow-based methods tended to result in very large relevant geographic markets that included a large number of hospitals, which resulted in court decisions favorable to mergers. The second factor, which played a substantive role in only a few agency losses, is the nonprofit status of the merging hospitals. We review each of these in turn.

4.3.1.1.  Market Definition in Hospital Merger Litigation Most of the litigated hospital merger cases hinged upon the definition of the relevant geographic market. The FTC or DOJ usually alleged a geographic market in which there were no more than three or four competing hospitals (including the merging pair), while the merging hospitals advanced a much broader geographic market with many more hospitals. Under the agencies’ alleged geographic markets, the mergers would be “presumed to be likely to enhance market power” under the Herfindahl-Hirschman Index (HHI) thresholds set forth in the Horizontal Merger Guidelines. Under the hospitals’ alleged broader geographic markets, the challenged mergers would be “unlikely to have adverse competitive effects.”20 In these cases, economists testifying as expert witnesses commonly employed an approach to geographic market definition known as “Elzinga/Hogarty” (EH) analysis, which involves using patient origin and destination data to identify an area that has low inflows and outflows of patients. In most of these cases, both sides—the government and the merging hospitals—used EH in support of their claims. That is, the economic debate was not about whether EH should be used, but how. United States v. Rockford Memorial Hospital, one of the last two agency wins before the string of losses in the 1990s, is illustrative.21 In Rockford, experts for both sides employed 20 

HHI thresholds and the associated presumptions are described in section 5.1 of the Horizontal Merger Guidelines (US DOJ and FTC 2010). The HHI is calculated as the sum of the squared market shares of market participants. For example, in a market with three equally sized competitors, the HHI is 33.32 + 33.32 + 33.32 = 3, 327. Under the current guidelines, mergers that result in an HHI below 1,500 are presumed to be unlikely to have anticompetitive effects and mergers that result in an HHI above 2,500 are presumed likely to have anticompetitive effects. Under the merger guidelines in effect during the 1990s, those thresholds were 1,000 and 1,800. See section 1.51 in US DOJ and FTC (1997). 21  United States v. Rockford Mem. Hosp., 717 F. Supp. 1251 (N.D. Ill. 1989), aff ’d, 898 F.2d 1278 (7th Cir. 1990).

Healthcare Provider and Payer Markets   73

the EH method to define the relevant geographic market, though they identified significantly different relevant markets. The court offered extensive criticism of the defendant’s implementation of EH and comparatively modest criticisms of the government’s EH implementation; ultimately, the court conducted its own EH analysis.22 Based on that analysis, the court identified a relatively small geographic market, held that the proposed merger would “substantially increase the concentration of an already concentrated market,” and permanently enjoined the proposed merger.23 In so holding, the court noted as follows:24 [T]‌he court’s criticism of the defendants’ application of the Elzinga-Hogarty test was not meant as a blanket indictment of the test. On the contrary, the court subscribes to the Elzinga-Hogarty test as a useful tool for eliminating certain geographic areas from consideration as relevant markets [emphasis added]. The court will employ the defendants’ Elzinga-Hogarty information, such as it is, to help pare the geographic market to its appropriate size.

The DOJ’s win in Rockford established EH as appropriate for defining relevant geographic markets in hospital merger cases.25 As events unfolded, however, that win laid the groundwork for agency losses in six of the next seven hospital merger cases.26 For example, both sides offered EH analyses in Ukiah, which involved an attempt by the FTC to unwind a consummated merger of the only two hospitals in Ukiah, California.27 The outcome hinged upon whether Santa Rosa, located about 60 miles south of Ukiah, should be included in the relevant geographic market. With Santa Rosa included, the merger would have a minimal impact on concentration; if, however, the geographic market were confined to Ukiah, the merger would create a monopolist. After protracted litigation, the FTC commissioners ultimately dismissed the FTC staff ’s complaint, with FTC commissioner Azcuenaga writing, “The Elzinga-Hogarty statistics for the Ukiah-Willits-Lakeport area do not support finding a relevant geographic market

22 

Id. at 1264–78. Id. at 1280, 1292. 24  Id. at 1271. 25  The government won one more case after Rockford: US Fed. Trade Comm’n v. University Health, 938 F.2d 1206, 1210 (11th Cir. 1991). In this case, the relevant geographic market was not hotly contested. Instead, after the district court denied the FTC’s request for a preliminary injunction, noting that (1) the acquisition target was a weak competitor, (2) the merger would create significant efficiencies, and (3) University Health was a nonprofit entity, the Eleventh Circuit rejected the district court’s finding on each point and reversed. Id. at 1220, 1222, 1225. 26  One of the seven losses was by the State of California; of the remaining six, five were lost on the geographic market definition. The sixth, US Fed. Trade Comm’n v. Butterworth Health Corp., was lost on the basis of the merging hospitals’ nonprofit status. The history of these cases and the role of economic research in the reversal of litigation outcomes that began in 2008 are discussed in more detail in Capps (2014). 27  At the time of the merger, in 1988, there was a third hospital in Ukiah, but that hospital closed in 1990, well before the final decision. 23 

74   Cory S. Capps and David Dranove limited to those communities. . . . The Elzinga-Hogarty measure of outmigration . . . is only 74.57 percent.”28 Based on that finding, the Commission upheld the administrative law judge’s finding that Santa Rosa should be included in the relevant geographic market, and dismissed the complaint.29 Subsequent cases proceeded similarly: • United States v. Mercy Health Services. The DOJ offered an EH analysis showing inflows of 24 percent (76 percent of patients treated by hospitals in the government’s geographic market were from inside that area) and outflows of 12 percent (88  percent of residents of the 15-mile radius area selected one of the hospitals within the relevant geographic market).30 The court rejected the DOJ’s alleged market, viewing those inflows and outflows as evidence that patients could practicably go outside the government’s narrow alleged geographic market in response to an attempted exercise of market power. • US Fed. Trade Comm’n v.  Freeman Hospital. The court rejected the FTC’s 54-mile-diameter relevant geographic market in favor of the hospitals’ 13-county, 100-mile-diameter geographic market.31 • US Fed. Trade Comm’n v. Tenet Healthcare Corp. The FTC secured an initial win when the district court noted that “at some point, a hospital ceases to become a practical alternative for general acute care because of distance,” and issued a preliminary injunction.32 On appeal, however, the US Court of Appeals for the Eighth Circuit reversed, holding that “the district court improperly discounted the fact that over twenty-two percent of people in the most important zip codes already use hospitals outside the FTC’s proposed [54-mile radius] market . . . the FTC’s proposed market is too narrow.”33

4.3.1.2.  Conceptual Flaws of the Elzinga-Hogarty Method The EH method was based on two studies of beer and coal markets conducted in the 1970s (Elzinga and Hogarty 1973, 1978). Under EH, the relevant geographic market is defined as an area (1) that accounts for at least 75 percent of the sales of the relevant firms and (2) for which inflows and outflows are both low. Experts and courts in the 1990s debated whether to define “low” as less than 25 percent or as less than 10 percent. Either way, with respect to hospitals, it was and remains common for a meaningful fraction of 28 

In re Adventist Health Sys./West, FTC Docket No. 9234, 117 FTC 295 (F.T.C. Apr. 1, 1994). The ALJ concluded that “the extensive admission by Santa Rosa hospitals of Ukiah areas residents for routine inpatient care indicates that they should be included in the relevant geographic market” (In re Adventist Health Sys./West, FTC Docket No. 9234, 117 FTC 263 (ALJ Dec. 9, 1992)). 30  United States v. Mercy Health Services, 902 F. Supp. 968, 976, 978 (N.D. Iowa 1995). 31  US Fed. Trade Comm’n v. Freeman Hosp., 911 F. Supp. 1213, 1218–19, 1222 (W.D. Mo. 1995). 32  US Fed. Trade Comm’n v. Tenet Healthcare Corp., 17 F. Supp. 2d 937, 943 (E.D. Mo. 1998), rev’d 186 F.3d 1045 (8th Cir. 1999). 33  Tenet Healthcare, 186 F.3d 1045, 1053 (8th Cir. 1999). 29 

Healthcare Provider and Payer Markets   75

patients to travel a considerable distance for hospital care. The result is very large geographic markets, even though the majority of patients choose a hospital located close to their home.34 Economists have noted that EH is an inappropriate tool for defining the geographic boundaries of hospital markets, for several reasons (Capps et al. 2002). First, because most patients who travel do so for reasons other than price, such flows are not generally reliable indicators of the likely response to a price increase. This flaw is sometimes referred to as the “silent majority fallacy,” because the behavior of a relatively small set of travelers does not predict the behavior of the majority of nontraveling patients. Second, EH does not reflect how hospital prices are actually determined in a managed care environment; namely, via negotiations between hospitals and insurers. Instead, EH embeds the presumption that because some patients travel for care, many additional patients would also readily do so if hospitals were to raise prices (and therefore such price increases would not be profitable). Consider that when a hospital increases its prices in negotiations with an insurer but remains in-network, the financial incentives of patients and referring physicians to use that hospital are typically unchanged or minimally reduced.35 Thus, the assumption embedded in EH analysis that patients will travel considerable distances to obtain lower prices does not hold generally in hospital markets. Ken Elzinga, the cocreator of the EH methodology, labeled this the “payor problem” (Elzinga and Swisher 2011; Werden 1990).36 Taken together, the flaws of the EH approach make it prone to generating overly broad relevant geographic markets and, thereby, to identifying a large number of competitors. The result during the 1990s was more permissive hospital merger enforcement; indeed, the economic literature indicates that the courts were overly permissive (Town and Vogt 2006).37

34  The conclusions of an EH analysis can also depend on how the methodology is implemented. In particular, the unit of analysis (zip code or county) is important, as is the order in which the candidate market is expanded (Frech, Langenfeld, and McClure 2004). 35  In recent research, Gowrisankaran, Nevo, and Town (2013) find that patients in managed care plans are only slightly sensitive to hospital prices (estimated price elasticities are below 0.15). 36  As explained in Elzinga and Swisher:

[I]‌gnoring the Silent Majority Fallacy can make the geographic market for hospital services appear “too big” in circumstances where market definition is based on patient flow data. When that happens, the E-H test produces a perverse result: the boundaries, being too broad, may embrace nearby hospitals whose existence has no competitive consequence in disciplining the pricing discretion of the merging hospitals. . . . The Silent Majority Fallacy cautions against a rote, mechanical application of the E-H test in hospital mergers. . . . (137–144) 37  In several of the cases of the 1990s, economists used critical loss analysis to support claims of broad geographic markets (e.g., United States v. Mercy Health Serv., 902 F. Supp. 968 (N.D. Iowa 1995), vacated as moot, 107 F.3d 632 (8th Cir. 1997)). The first step in critical loss analysis is to identify of the percentage of customers who would have to depart in order to make a given price increase unprofitable (this is the critical loss)—the main input into this calculation is the firm’s margin. The second step is to estimate

76   Cory S. Capps and David Dranove

4.3.1.3.  The FTC’s Hospital Merger Retrospective and the Evanston Case In 2002, the FTC launched a hospital merger retrospective study and issued subpoenas for documents and pricing data to a number of recently formed hospital systems (FTC 2002).38 For some mergers, the FTC concluded that prices had not increased or had only increased to some health plans. However, from its review of the 2000 acquisition of Highland Park Hospital by Evanston Northwestern Healthcare (both located in Chicago’s North Shore area), the FTC concluded that the merger had resulted in substantial price increases, and, in 2004, the FTC sued to unwind that merger.39 Because Evanston involved a consummated merger, the case hinged largely upon whether the merger had caused significant price increases rather than on the nuances of market definition. Experts for both sides undertook a series of “difference-in-difference” analyses to evaluate whether Evanston Northwestern’s price increases exceeded those of comparable hospitals. Ultimately, both the administrative law judge and the commissioners concluded that Evanston Northwestern’s price increases were anticompetitive.40 Although the efficacy of the Commission’s conduct remedy has been questioned (the Commission overturned the administrative law judge’s divestiture order), Evanston

the actual loss and compare that to the critical loss. If the actual loss exceeds the critical loss, then the hypothesized price increase would not be profitable. (US DOJ and FTC 2010). In practice, however, the critical loss analyses advanced in the hospital merger cases of the 1990s simply added a rhetorical layer to what was fundamentally an EH analysis. That is, there was no formal calculation or estimation of the “actual loss” that would result from a price increase. Instead, the actual loss was evaluated by reference to outflows, so this approach was not substantively different than an EH analysis (Langenfeld and Li 2001; see also, Katz and Shapiro 2003; O’Brien and Wickelgren 2004; Danger and Frech 2001). 38 

A recently published collection of articles by FTC staff economists and experts in various hospital litigation matters describes the circumstances in which the FTC has found, and has not found, evidence of hospital market power. See “Special Issue: Hospital Mergers and Antitrust Policy,” International Journal of the Economics of Business 18, no. 1 (2011). For example, one article in that collection concludes that there were significant postmerger price increases in the case that the State of California litigated and lost in 2001 (Tenn 2011). 39  Initial Decision of Chief Administrative Law Judge Stephen J. McGuire, In re Evanston Northwestern Healthcare Corp., FTC Docket No. 9315, (ALJ Oct. 20, 2005); In re Evanston Northwestern Healthcare Corp., No. 9315, slip op. (F.T.C. Aug. 6, 2007), available at http://www.ftc.gov/os/adjpro/​ d9315/070806opinion.pdf. 40  The Commission stated: In summary, we find that the merger enabled ENH to exercise market power, and that ENH used this market power to increase its average net prices to MCOs for acute inpatient hospital services by a substantial amount—at least the 9% or 10% calculated by Baker. . . . We also find that because the merger enabled ENH to raise prices by a substantial amount (at least equal to a SSNIP) through the unilateral exercise of market power, the geographic triangle in which the three ENH hospitals are located constitutes a well-defined antitrust geographic market under Section 7. In re Evanston Northwestern Healthcare Corp., No. 9315, slip op. at 78 (F.T.C. Aug. 6, 2007), available at http://www.ftc.gov/os/adjpro/d9315/070806opinion.pdf.

Healthcare Provider and Payer Markets   77

helped to establish two important principles that would lay the groundwork for future prospective hospital merger enforcement: • The Elzinga-Hogarty methodology is likely to be misleading, at least when applied to hospitals. In reaching this conclusion, the Commission relied in part on testimony by the co-creator of the test, Ken Elzinga.41 Given Evanston Northwestern’s location on Chicago’s North Shore, the relevant geographic market defined by using EH would likely have included most of the Chicago area and would have implied a low likelihood of anticompetitive price increases. The direct evidence on postmerger price increases established that an EH-defined market would be overly broad and misleading as regards competitive effects. • Hospital merger analysis should focus on the effects of a merger on the outcomes of price negotiations between hospital systems and health plans rather than on the hospitalization decisions of individual patients. In reaching this conclusion, the Commission relied in large part on economic research that formally modeled competition among hospitals in a selective contracting environment.42

4.3.1.4.  Bargaining, Leverage, and Willingness-to-Pay The economic research referenced by the Commission in Evanston began in the late 1990s, motivated in large part by academic interest in the agencies’ mounting losses in hospital merger cases. Economists developed methods that were designed to more closely match industry conditions and, therefore, to more accurately identify relevant geographic markets and more reliably predict competitive effects. The main papers in the first wave of this line of research are Vistnes (2000), Town and Vistnes (2001), Capps, Dranove, and Satterthwaite (2003), and Gaynor and Vogt (2003).43 These papers and several more recent papers are thoroughly reviewed in Gaynor and Town (2011), so we address them only briefly. 41 

“Professor Elzinga testified that the E-H test was not an appropriate method to define geographic markets in the hospital sector because of the ‘silent majority fallacy’ and the ‘payor problem.’ ” Id. at 75–78. 42  The Commission explained: Contrary to [Evanston Northwestern’s] position, bargaining markets are quite common and fully consistent with unilateral effects theory. And most economists who have recently studied the issue have concluded that bargaining models are appropriate for hospital markets because bilateral negotiations between MCOs and hospitals determine prices that often are unique to the particular negotiation. The record in this case also demonstrates that hospital prices in the Chicago market are set through bilateral negotiations. Id. at 62 (citing Capps, Dranove, and Satterthwaite 2003, and Town and Vistnes 2001). See also, id. at 72 (citing Vistnes 2000). 43  Strictly speaking, Gaynor and Vogt do not adopt a bargaining framework; instead, they model equilibrium in a traditional differentiated Bertrand context and use that to simulate the effects of hospital mergers. However, the equations that result from that approach are similar to those in T&V and CDS. The empirical predictions are also similar (Gaynor, Kleiner, and Vogt 2013).

78   Cory S. Capps and David Dranove The key insight underpinning these papers is that prices and the composition of health plans’ networks are determined in bilateral negotiations between hospital systems and health plans.44 Economic theory indicates that, in a negotiating context, outcomes are determined by the outside options of the negotiating party, or the payoffs each would receive in the event that no agreement is reached (Binmore, Rubenstein, and Wolinsky 1986; Horn and Wolinsky 1988). One important implication is that bargaining leverage will increase in response to events that increase the cost to the counterparty of failing to reach an agreement. This comports with simple intuition: a party to a negotiation does better the more the other party needs a deal. Mergers of competing hospitals can enhance the bargaining leverage of the merged entity and facilitate price increases by eliminating health plans’ ability to play the hospitals off against each other. That is, premerger, health plans can turn to one hospital in the event that the other demands an excessively high price. Postmerger, however, health plans lose the ability to exclude one hospital and retain the other, making the outside option for health plans less attractive.45 Capps, Dranove, and Satterthwaite (2003) and Town and Vistnes (2001) derive formulae characterizing the value that a hospital or set of hospitals adds to a health plan’s provider network—this is now commonly referred to as “willingness-to-pay” (WTP)— and they show how to use WTP to predict the price effects of hospital mergers. Both sets of authors also validate the WTP metric by establishing that it is positively associated with hospital pricing and profits.46 Simulation results in both papers indicate that certain hospital mergers that would be deemed benign under the EH approach would likely result in significant price increases.47 Gowrisankaran, Nevo, and Town (2013) augment the WTP approach by adding and estimating a formal bargaining model. Beginning with Evanston, the FTC has couched its theory of harm in terms of the effects of the challenged merger on leverage in negotiations with health plans. In subsequent cases, the FTC has offered expert testimony that includes analyses of WTP and

44  Vistnes (2000) describes this as the “first stage” of two-stage competition, with the “second stage” consisting of competition among in-network hospitals for individual patients. Prices are the key variable in first-stage competition, whereas second-stage competition occurs primarily on the basis of nonprice characteristics (because patients face little or no variation in costs when choosing among in-network hospitals). 45  This framing assumes that the merged entity engages in all-or-nothing contracting for both hospitals. Even in the alternative of hospital-by-hospital negotiations, the merged entity’s leverage would increase. In the latter case, the mechanism is that each hospital recognizes that if it is excluded from a network, some of the lost volume will be captured by the merger partner—essentially, the outside option for the merged hospitals’ improves. This reasoning is closely analogous to the ideas underlying diversion and upward pricing pressure analysis (Farrell and Shapiro 2010; Werden 1996). 46  Ho (2006) provides further validation by showing that the demand for a health plan is greater when it offers a network of hospitals with higher WTP. 47  Using claims data from an insurer, Capps and Dranove (2004) study the effects of increased hospital concentration on pricing and find that many mergers that courts in the 1990s would likely have let through (had they been challenged) resulted in substantial price increases.

Healthcare Provider and Payer Markets   79

bargaining.48 Despite being relatively new (as judged by the slow evolution of the case law), this testimony has been accepted and relied upon at least in part by two federal district court judges, an FTC administrative law judge, and a majority of the commissioners of the FTC.49 WTP analysis has shifted from novel to routine in hospital merger cases and is very likely to continue to play a key role in hospital merger cases.

4.3.1.5.  Nonprofit Status Although market definition has taken precedence in many, if not most, hospital merger cases, the nonprofit status of most merging hospitals has also played a significant role in one case and a lesser role in others.50 Standard oligopoly theory underlying antitrust law and economics assumes that firms maximize profits. This assumption may seem inappropriate for healthcare markets, where the vast majority of hospitals, as well as many nursing homes and home care agencies, are owned by nonprofit or government entities. Many health insurance plans are also operated as nonprofits. Indeed, some observers have questioned the application of antitrust law to nonprofit hospitals. Kopit and McCann (1988) reason that nonprofit hospitals do not seek to maximize profits. Moreover, they reason, nonprofits face oversight from boards of trustees drawn from the local community and so would not increase prices even if they could.51 Even if nonprofits do not care about profits (a big supposition) economic theory suggests that antitrust agencies should maintain their scrutiny. For example, Newhouse (1970) suggests that managers of nonprofits seek to maximize “prestige,” which is loosely defined as some combination of size, complexity, and quality. Prestige-maximizing hospitals exploit market power by raising prices and using the resulting “profits” to fund potentially excessive facility growth and technology acquisitions. Pauly and Redisch (1973) model the hospital as a “physician’s cooperative” that is run to maximize the profits of its medical staff. In this model, a hospital that gains market power does not increase prices, but its medical staff does. In both models, patients may be harmed if nonprofits obtain market power.

48 

Although the FTC and DOJ both brought hospital merger cases in the 1990s, in the wake of Evanston, primary responsibility for hospital merger enforcement appears to have shifted to the FTC. 49  Specifically, WTP evidence was offered in Inova, ProMedica, and OSF Healthcare. The FTC commissioners’ decision to block ProMedica’s acquisition of St. Luke’s Hospital is currently on appeal to the Sixth Circuit. 50  Efficiencies are also important in hospital merger cases—even mergers that give hospitals enhanced pricing power may be procompetitive if they result in merger-specific cost savings that are likely to be passed on to consumers or in merger-specific quality increases. In general, the analysis of efficiencies in hospital merger cases is not unique or distinct from other industries, so we do not address this topic in detail. Although the literature on costs and quality is less extensive than the price literature, there is substantial evidence that hospital mergers generate neither consistent efficiencies nor quality improvements (though they can in specific instances) (Town and Vogt 2006; Gaynor and Town 2012). 51  Kopit and McCann also claim that “price typically is not an important element in the purchase of hospital services.” To support this claim, they cite a textbook from 1983 and references therein. However, 1983 predates the explosive growth of selective contracting.

80   Cory S. Capps and David Dranove Kopit and McCann also exaggerate the role of the nonprofit hospital board. Board members may have little or no influence over pricing decisions and are generally not involved in contractual negotiations. Even if they were involved, they would need to balance their duties as board members (which include assuring the long-term health of their hospital) against their private goals as members of the local community. Philipson and Posner (2009) offer a theoretical model of competition among nonprofit entities in which nonprofits are models as exhibiting some degree of “output preference”—that is, nonprofits’ maximize an objective function that is a weighted average of the institution’s profits and its output. They show that competition among such nonprofit firms will maximize social welfare as long as the social planner and the nonprofit firms place the same value on output. They also show that, although output preference on the part of nonprofits may result in lower prices, the effects of increases in market power are similar for nonprofit and for-profit institutions.52 For these reasons, Philipson and Posner argue against preferential antitrust treatment of nonprofit institutions. Ultimately, the theoretical literature does not offer a clear answer to the question of whether or not nonprofits will exercise market power. As a practical matter, this issue was not central to a hospital merger case until the 1996 case, Butterworth Health.53 In this case, the FTC filed an administrative challenge to the merger between Butterworth Hospital and Blodgett Memorial Medical Center. The two merging hospitals, both nonprofits, were the largest of four hospitals in Grand Rapids, Michigan. Despite siding with the FTC on market definition and concentration—a rarity for cases litigated around this time—the district court denied the FTC’s request for an injunction to block the merger, and the Court of Appeals for the Sixth Circuit affirmed.54 In denying the request for an injunction, the court relied heavily on testimony by William Lynk, the economist retained by the merging hospitals. In part, his testimony was based on a publication prepared in conjunction with this case (Lynk 1995). Lynk’s study employs a traditional regression of prices on market concentration using a single cross-section of hospitals in California in 1989. Based on the included indicator variable for nonprofit status, he finds that prices were positively correlated with market 52  That is, a nonprofit entity that values output will produce more and price below an otherwise similar for-profit entity. In that circumstance, however, the adverse effect of an increase in market power may well be greater for a nonprofit entity than for a for-profit entity. 53  US Fed. Trade Comm’n v. Butterworth Health, 946 F. Supp. 1285 (W.D. Mich. 1996). 54  The court stated:

[N]‌ onprofit hospitals operate differently in highly-concentrated markets than do profit-maximizing firms . . . the boards of these two hospitals are comprised of prominent community and business leaders whose employees depend on these facilities for services, and who have demonstrated their genuine commitment to serve the greater Grand Rapids community through their governance of the hospitals . . . both boards have given this commitment concrete form through the Community Commitment [to freeze prices for a period of time after the merger] and . . . substantial cost-savings and efficiencies would be realized as a result of the merger. Id. at 1302.

Healthcare Provider and Payer Markets   81

concentration for for-profit hospitals but negatively correlated for nonprofits—that is, that nonprofits do not charge higher prices in more concentrated areas. Lynk’s provocative findings were critiqued almost immediately. Dranove and Ludwick (1999) test the robustness of Lynk’s methods and find that the results hinged on several critical and questionable assumptions. Keeler, Melnick, and Zwanziger (1999) respond to Lynk by noting that the market for hospital services was evolving as a result of the growth of selective contracting. Examining data from California in 1986–1994, they find that during the early years of their data, concentration and prices were negatively correlated for nonprofits; this confirms Lynk’s finding. However, this effect was reversed in later years: nonprofits charged higher prices in more concentrated markets. Lynk’s study and the two critiques rely on traditional reduced form regressions in which the key predictor variable is an ad hoc measure of market power based on the HHI. In their structural models of hospital pricing, Town and Vistnes (2001), Capps, Dranove, and Satterthwaite (2003), and Gaynor and Vogt (2003) all examine whether nonprofits exploit their market power. Neither Town and Vistnes nor Capps, Dranove, and Satterwaite find any difference in the extent to which nonprofits and for-profits exploit their bargaining leverage, as measured by WTP, by raising prices. Gaynor and Vogt find that nonprofits face less elastic demand than for-profits and that they set their prices accordingly (i.e., higher). These structural analyses provide further evidence against relaxing antitrust scrutiny of nonprofits. There have also been several case studies of nonprofit hospital mergers. Most compare price changes at the merging hospitals with price changes at other hospitals in the same market.55 Krishnan (2001) studies two mergers in Ohio and California and finds that prices at the merging hospitals increased more for those procedures in which the hospitals had the most market power. Vita and Sacher (2001) find that prices increased subsequent to a merger of two nonprofit hospitals in a concentrated market. Economists working for or at the FTC produced studies that examined price changes in the aftermath of three mergers, each a merger of nonprofits, that were not litigated by the federal agencies (Haas-Wilson and Garmon 2011; Thompson 2011; Tenn 2011). Prices unambiguously increased after two of the three mergers and price changes after the third merger were mixed. An alternative justification for lax antitrust treatment is that nonprofit hospitals might use their market power to provide services that benefit the community, such as uncompensated care. Upon inspection, however, this rationale is also not compelling. Funding social policy through market power is inefficient relative to direct funding through taxation and subsidy. Indeed, nonprofits receive preferential tax treatment—a form of direct subsidy—in exchange for providing a commensurate level of community benefits. Moreover, empirical research has not established that hospitals with more 55  Spang, Arnould, and Bazzoli (2009) perform “before and after” analyses of 125 mergers and 1,040 system acquisitions and find that for-profits raised prices after mergers but nonprofits did not, consistent with Lynk’s claims. However, their measure of price includes Medicare and Medicaid reimbursements, which should not be affected by market power. They also assume that all mergers, including system acquisitions, have an equal impact on price. Finally, their regressions appear to be prone to mean regression. It is not clear whether or how the resulting specification errors bias their results.

82   Cory S. Capps and David Dranove market power provide more uncompensated care; rather, there is evidence to the contrary (Garmon 2009; Capps, Carlton, and David 2010).

4.3.1.6.  State Action Immunity As recounted above, the FTC has prevailed in three of its last four attempts to block a hospital merger. The one loss arose from the acquisition by Pheobe Putney Health System in Albany, Georgia, of its only nearby competitor, Palmyra Park Hospital. Both the district court and, on appeal, the Eleventh Circuit appeared to conclude that the acquisition would result in a monopoly or near-monopoly.56 Nevertheless, the district court concluded that, on the basis of the state action immunity doctrine, the merger was not subject to the federal antitrust laws, and denied the FTC’s request for a preliminary injunction enjoining the merger. The state action immunity claim arose in this case because the transaction was structured so that the state-run Hospital Authority of Albany-Dougherty Counties acquired Palmyra Park Hospital and then leased it to the private nonprofit entity, Phoebe Putney Health System. That is, the state-run hospital authority was the formal acquiring entity. The FTC alleged that the hospital authority was not and is not sufficiently involved in the acquisition and that a private party, Phoebe Putney Health System, was for all intents and purposes the acquirer. The district court rejected the FTC’s characterization, and the Eleventh Circuit upheld, reasoning that state action immunity applied because anticompetitive effects were a foreseeable consequence of the state’s creation of the hospital authority.57 The FTC appealed this decision to the US Supreme Court and, in February 2013, the Court unanimously ruled in favor of the FTC.58 Although antitrust issues were not at the center of this case, the Court’s decision appears significant in several respects. First, the Court upheld the proposition that the US healthcare system relies on competition:59 [N]‌othing in the Law or any other provision of Georgia law clearly articulates a state policy to allow authorities to exercise their general corporate powers, including their 56  The district court did not dispute the FTC’s claim that “the acquisition . . . will create a virtual monopoly for inpatient general acute care services in Albany, Dougherty County, Georgia, by eliminating competition between PPMH and Palmyra, the only two major hospitals that service not only the Albany, Dougherty County community, but the communities of the surrounding six counties.” Order at 39–40, US Fed. Trade Comm’n v. Phoebe Putney Health Sys., No. 1:11-CV-58 (M.D. Ga. June 27, 2011), available at http://www.ftc.gov/os/caselist/1110067/110627phoebeputneyorder.pdf. The Eleventh Circuit went further, noting that it “agree[d]‌with the Commission that, on the facts alleged, the joint operation of [Phoebe Putney] Memorial and Palmyra would substantially lessen competition or tend to create, if not create, a monopoly.” US Fed. Trade Comm’n v. Phoebe Putney Health Sys., 663 F.3d 1375 (2011). 57  Id. at 1376. 58  “We hold that Georgia has not clearly articulated and affirmatively expressed a policy to allow hospital authorities to make acquisitions that substantially lessen competition. The judgment of the Court of Appeals is reversed, and the case is remanded for further proceedings consistent with this opinion.” US Fed. Trade Comm’n v. Phoebe Putney Health Sys., No. 11-1160 (U.S. Feb. 19, 2013), 19. 59  Id. at 16–17.

Healthcare Provider and Payer Markets   83

acquisition power, without regard to negative effects on competition. The state legislature’s objective of improving access to affordable health care does not logically suggest that the State intended that hospital authorities pursue that end through mergers that create monopolies. Nor do the restrictions imposed on hospital authorities, including the requirement that they operate on a nonprofit basis, reveal such a policy. Particularly in light of our national policy favoring competition, these restrictions should be read to reflect more modest aims.

Second, as the preceding quotation shows, the Court noted, in passing, that a requirement that hospital authorities “operate on a nonprofit basis” does not reveal a policy of “improving access to affordable health care . . . through mergers that create monopolies.” This is consistent with the research on nonprofit hospitals and market power that we summarized in the preceding section.60 Third, the Court ruling may resolve what had been an apparent split between the Eleventh Circuit and the Sixth Circuit. In the MFN case brought by DOJ in the Sixth Circuit, the district court rejected Blue Cross and Blue Shield of Michigan’s claim that, as a quasi-regulated entity, it was immune under the state action doctrine; the Sixth Circuit court refused to hear an appeal.61 Thus, the Sixth Circuit appears to have adopted a more exacting standard for state action immunity than the Eleventh Circuit. In the wake of the Supreme Court decision, the state action doctrine is likely to be more narrowly applied and of less use to defendants.

4.3.1.7.  Quality and Efficiency The FTC does not challenge all hospital mergers that would result in a substantial increase in concentration in a well-defined market. Both quality and efficiency considerations can mediate the FTC’s enforcement decisions. For example, in 2009, the FTC closed its investigation of a consummated hospital merger in Temple, Texas, explaining that, although the FTC was concerned that the merger would enhance market power, the acquired hospital’s financial position was precarious and there was no viable alternative buyer.62 As another example, the FTC’s expert on hospital quality in OSF-Rockford and ProMedica recently testified that, in at least one instance, he has concluded that a proposed hospital merger would likely increase quality.63 60 

A group of economics professors, including the authors of this chapter, submitted an amicus brief in this case arguing that (1) “economic theory provides no determinate conclusions regarding whether nonprofits will exploit market power if given the opportunity” and (2) “there is a strong consensus in empirical research that, in general, nonprofit hospitals do exploit their market power by raising prices.” Brief of Amici Curiae Economics Professors in Support of Petitioner at 6, US Fed. Trade Comm’n v. Phoebe Putney Health Sys., No. 11-1160 (U.S. Aug. 17, 2012). 61  Czernin (2012). We discuss this case in section 4.5.5.6., infra. 62  US Fed. Trade Comm’n, “Statement of the Bureau of Competition Director Richard Feinstein on the FTC’s Closure of Its Investigation of Consummated Hospital Merger in Temple, Texas,” December 23, 2009, available at http://www.ftc.gov/os/closings/091223scottwhitestmt.pdf. 63  Transcript of Proceedings at 93, US Fed. Trade Comm’n v. OFS Healthcare Systems, No. c 50344 (N.D. Ill. Feb. 1, 2012). The expert, Dr. Patrick Romano, did not identify the merger. Nevertheless, the

84   Cory S. Capps and David Dranove

4.3.2.  Exclusionary Conduct by Hospitals While hospital merger cases receive the lion’s share of attention, challenges to alleged anticompetitive exclusionary practices by hospitals are actually far more common. These usually arise in private antitrust actions in which physicians allege monopolization by hospitals. There are two main types of exclusionary conduct alleged by physicians. The first involves hospitals with closed medical staffs that grant only certain physicians privileges to admit and treat patients. Physicians who are excluded from the staff may argue that the hospital is using its power in the inpatient services market to restrict competition in the physician services market. The second type of case involves hospitals that secure contracts with insurers that grant exclusivity in the provision of certain services, such as outpatient surgery. Competing outpatient surgery providers, which are often physician-owned, allege that the hospital is using its power in the inpatient services market to restrict competition in the outpatient services market. Both types of cases involve allegations of tying and/or bundling.

4.3.2.1.  Staff Privileges Cases In order to treat patients at a hospital, a physician must have privileges there. A large body of antitrust cases stem from hospitals’ decisions to deny privileges to a physician or group of physicians. As the American Bar Association Section of Antitrust Law (2010) noted: “[T]‌he courts have rejected the vast majority of staff privilege cases.” The ABA and Hammer and Sage (2002) review the case law surrounding physician privileges cases in great detail, and so we address this topic only briefly. Antitrust cases brought against hospitals by physicians face an array of legal and economic challenges. Because hospitals and their physician staffs are not competitors, most courts have found that privilege denials, which benefit some members of the hospital’s physician staff at the expense of the denied physicians, are unilateral actions of the hospital, and so Section 1 of the Sherman Act rarely applies (tying, which we discuss below, is judged under Section 1). Under Section 2, the plaintiff will generally have to establish that the defendant possesses or is likely to attain substantial market power. This can be particularly challenging for the plaintiff physician in cases in which the hospital has not participated in the allegedly monopolized market (i.e., a particular physician service line).64 Moreover, there can clearly be valid and procompetitive reasons, such as maintaining quality, for a hospital to deny privileges to a

clear implications are that (1) the FTC had sufficient initial concern to ask Dr. Romano to conduct an analysis of quality and (2) after Dr. Romano conducted his analysis and identified likely quality improvements, the FTC did not challenge the merger. Other factors may have also played a role in that decision. 64  This characterization assumes that the hospital does not own one or more physician groups in the relevant specialty. An interesting question moving forward is whether the trend of increasing acquisition of physician practices by hospitals will change how courts view this element. Even so, the applicability of Section 1 may be limited by Copperweld Corp. v. Independence Tube Corp., 467 US 752 (1984), which holds that companies are logically incapable of conspiring with their own subsidiaries.

Healthcare Provider and Payer Markets   85

particular physician or group.65 And the excluded physician or group is often only one of many in an area, making harm to competition (rather than to competitors) difficult to establish and less likely to outweigh potential procompetitive benefits.66 While most physician specialties use hospitals only for part of their practice, certain “hospital-based” specialties render the bulk of their services in the hospital setting and provide those services primarily in an ancillary capacity. Radiology, anesthesiology, and pathology (RAP) are the main hospital-based specialties, and it is common for hospitals to contract with such hospital-based providers on an exclusive basis. A second line of cases, closely related to the privileges cases described above, involves a RAP physician who is denied privileges and then sues the hospital and the chosen exclusive provider of hospital-based services. Though other violations are usually alleged, the leading allegation is that the exclusive contracts constitute illegal tying between hospital services (the tying good) and the relevant RAP services (the tied good). The US Supreme Court appeared to settle the legal requirements for proving illegal tying in its decision in Jefferson Parish Hosp. Dist. v. Hyde.67 Hyde was an anesthesiologist who was denied privileges to practice at Jefferson Parish Hospital in New Orleans. Hyde argued that Jefferson Parish used its market power in inpatient surgery to force patients to purchase anesthesia from a physician group designated by the hospital. The Court ultimately ruled against Hyde and, in the process, established the elements required for a tie to violate Section 1 of the Sherman Act. Areeda and Hovenkamp (2012) describe these elements as follows: (1) two distinct products, (2) the two products are in fact tied together or customers are coerced, (3) the supplier possess substantial market power over the tying product, (4) injury to competition (rather than just to competitors), and (5) a not insubstantial volume of commerce is affected. For plaintiffs, proving that the hospital has substantial market power in the tying market has proven to be the biggest legal barrier. Hyde and many similar lawsuits have turned on geographic market definition, and, as discussed earlier, courts have historically accepted broad geographic market definitions in hospital cases. Recent hospital merger cases suggest that the tide could turn, because courts have increasingly accepted comparatively narrow geographic markets—that is, it may now be easier for plaintiffs

65  The point may not apply when the decision to deny privileges is motivated by economic rather than clinical considerations, such as a hospital denying privileges to physicians that own, practice at, or refer to one of the hospital’s competitors. 66  Although the other challenges remain, harm to competition may be less of a barrier when the hospital is excluding an entire class of practitioner. For example, Sage, Hyman, and Greenberg (2003) note that the FTC has successfully challenged professional oppositions to new modes of healthcare delivery, such as “nonphysician practitioners, hospital-sponsored clinics, and out-of-town brand-name providers.” The FTC has continued in this effort, most recently in North Carolina, where the commissioners unanimously ruled that the state Dental Board’s rules precluding nondentists from providing teeth-whitening services harmed competition. In re North Carolina Bd. of Dental Examiners, No. 9343 (F.T.C. Dec. 7, 2011), available at http://www.ftc.gov/os/adjpro/d9343/111207ncdentalopinion.​ pdf. 67  Jefferson Parish Hosp. Dist. v. Hyde, 466 U.S. 2 (1984).

86   Cory S. Capps and David Dranove to demonstrate that a hospital possesses substantial market power. Even so, physicians, whether hospital-based or not, are likely to continue facing substantial hurdles in challenging a hospital’s decisions over grants of privileges.

4.3.2.2.  Tying and Bundling In an increasing number of situations, especially those involving selective contracting, hospitals do not demand outright exclusivity but instead link pricing to a grant of partial or complete exclusivity. These cases are distinct from the privileges cases in that the excluded party is a clear horizontal competitor to the hospital—typically, a competitor for a subset of services. They are also distinct in that they more closely resemble bundling than tying. For example, in PeaceHealth, the defendant offered a full range of primary, secondary, and tertiary hospital services, but the plaintiff only offered primary and secondary services.68 PeaceHealth offered better terms to insurers that agreed to exclude the plaintiff and purchase the entire set of hospital services—primary, secondary, and tertiary care services—only from PeaceHealth. The plaintiff, McKenzie Hospital (now operating as Cascade Health), sued, alleging that PeaceHealth used its monopoly power in tertiary services to exclude McKenzie from the market for primary and secondary services. Antitrust analysis of potentially anticompetitive bundling, such as in PeaceHealth, remains an area of significant legal and economic ambiguity.69 Bundling is covered in more detail in ­chapter 19 of this Handbook, so we will focus on healthcare applications. Because insurers often exclude one or more local providers from a network and the excluded competitor may believe that it has been treated unfairly and seek a legal remedy, healthcare will likely remain fertile ground for lawsuits alleging anticompetitive exclusion through bundled discounts. Often, network exclusion is a natural outcome of competition: an insurer may reject a price that is too high or a provider may reject a price that is too low. Either way, the resulting network exclusion is a consequence, not a failure, of the competitive process. This highlights the importance of distinguishing between exclusion that results from competition and exclusion that lessens competition. Conditioning is a critical factor that distinguishes between the type of exclusion that is inherent to competition in a selective contracting environment and the type of exclusion that some courts have condemned and that DOJ has challenged in recent years. Specifically, where a firm’s pricing is conditioned on the terms under which its customers trade with the firm’s rivals, the potential for anticompetitive effects is greater.70 Bundled

68 

Cascade Health Solutions v. PeaceHealth, 515 F.3d 883 (9th Cir. 2008). On the legal front, LePage’s Inc. v. 3M, 324 F.3d 141 (3d Cir. 2003) (en banc), cert. denied, 124 S. Ct. 2932 (2004), offers a standard that is very distinct from that in PeaceHealth. LePage’s focuses on the ability of the excluded firm to offer an “equally diverse” set of products. This standard would seemingly condemn all instances of bundling by a hospital with respect to, for example, an ambulatory surgery center. On the economic front, although most economists now recognize that the conditions necessary for “Chicago School” results to hold do not apply generally, substantial debate remains over the conditions under which exclusionary conduct is likely to be anticompetitive. 70  In contrast, without such conditioning, the more likely explanation for observed exclusion is that the excluding firm made a better offer to customers and, in the process, filled its needs. This still leaves 69 

Healthcare Provider and Payer Markets   87

discounts of the sort at issue in PeaceHealth are an example of conditioning: the pricing of one product (tertiary services) is higher if the customer purchases other products (primary and secondary services) from the bundling firm’s rivals.71 Importantly, however, conditioning alone does not establish that a given instance of exclusion is or even might be anticompetitive. Instead, conditioning is more akin to a safe harbor: absent some form of conditioning, exclusion is unlikely to entail anticompetitive effects; where conditioning is present, further investigation may be warranted. Ultimately, these investigations are fact intensive and case specific. In recent cases, price-cost tests have been used by some economists and courts to differentiate between bundled discounts that have anticompetitive effects and ones that do not. PeaceHealth and United Regional are two notable examples.

4.3.2.2.1. PeaceHealth As noted above, PeaceHealth was the sole provider of tertiary hospital services in Lane County, Oregon. Both McKenzie and PeaceHealth offered primary and secondary hospital services. PeaceHealth offered bundled discounts to insurers that agreed to purchase all three levels of service exclusively from PeaceHealth. In evaluating the pattern of conduct, the Ninth Circuit first took heed of the Supreme Court’s Brooke Group rulings: “[T]‌he Supreme Court has forcefully suggested that we should not condemn prices that are above some measure of incremental costs” and “the Court’s reasoning poses a strong caution against condemning bundled discounts that result in prices above a relevant measure of costs.”72 At the same time, the Ninth Circuit also took account of the effects of linkages between PeaceHealth’s pricing of tertiary and nontertiary services. These considerations ultimately led to the Ninth Circuit’s discount-allocation rule and price-cost test:73 [T]‌o prove that a bundled discount was exclusionary or predatory for the purposes of a monopolization or attempted monopolization claim under §2 of the Sherman Act, the plaintiff must establish that, after allocating the discount given by the defendant on the entire bundle of products to the competitive product or products, the defendant sold the competitive product or products below its average variable cost of producing them.

Elsewhere, the Ninth Circuit made clear that this entire line of inquiry only applied to “firms holding or on the verge of gaining monopoly power in the relevant market.”74 predatory pricing, which does not entail conditioning, as a potential theory of harm; however, outright predation is difficult to establish economically and courts have long been skeptical of such allegations. 71  In a 2012 speech, Fiona Scott-Morton, then deputy assistant attorney general for economics at the Department of Justice, used the term “Contracts that Reference Rivals”—i.e., contracts with terms that depend on “information outside the buyer-seller relationship”—to make a similar distinction (Scott-Morton 2012). 72  PeaceHealth, 515 F.3d 883, 901 (9th Cir. 2008); Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209 (1993). 73  PeaceHealth, 515 F.3d 910. 74  Id. at 915.

88   Cory S. Capps and David Dranove Thus, the test is two-pronged: (1) does the defendant possess substantial market power and (2) is the effective price after the bundled discount is allocated to the competitive products below the defendant’s average variable cost?75 After the Ninth Circuit overturned the district court’s grant of summary judgment in favor of PeaceHealth, the case settled. However, the test has endured.76

4.3.2.2.2. United Regional In 2010, in its first monopolization case since Microsoft, the DOJ sued United Regional Health System in Wichita Falls, Texas, a system with an inpatient market share of 90 percent and an outpatient market share of 65 percent. The DOJ alleged that “United Regional has maintained its monopoly power in the relevant [inpatient and outpatient services] markets by entering into contracts with commercial health insurers that exclude United Regional’s competitors in the Wichita Falls area from the insurers’ health-care provider networks (‘exclusionary contracts’).”77 As the complaint further explained, United Regional’s bundled discounts would have caused health insurers to pay 13 percent to 27 percent more to United Regional had they contracted with United Regional’s only local inpatient rival, physician-owned Kell West Regional Hospital. In its filings, the DOJ adopted the general framework of PeaceHealth and focused its analysis on whether United Regional’s contracts would exclude equally or more efficient competitors. However, the DOJ clarified its view that, in implementing a price-cost test, discounts should only be allocated to the “contestable” volume, meaning output that, absent the bundled discount would likely shift to the rival:78 To accurately determine whether United Regional’s discounted prices are above cost, however, the entire discount should be attributed not to the entire volume of 75  This test is similar in nature to the “Ortho test,” except that it compares the allocated price with the defendant’s own costs rather than with the plaintiff ’s. Most firms are unlikely to know their rivals’ costs, particularly their marginal costs. This approach, therefore, provides more guidance to firms seeking to offer bundled discounts without inviting litigation. Additionally, evaluating whether a discount-allocated price is below a rival’s variable costs risks condemning discounted pricing when the effect is to exclude a less efficient rival. Although it is true that even inefficient rivals can provide a meaningful constraint on the pricing of firms with market power (e.g., if a monopolist has marginal costs of 5 and would set a monopoly price of 9, while the rival has marginal costs of 7), it may be difficult in practice to distinguish such instances from overprotection of inefficient firms. The Ninth Circuit also relied heavily on the Antitrust Modernization Commission (AMC 2007), though it disagreed with the AMC on several lesser points, most notably, the need to establish recoupment. PeaceHealth, 515 F.3d 910 n.21. 76  For example, the test was applied by the plaintiff ambulatory surgery center in Peoria Day Surgery Ctr. v. OSF Healthcare Sys., No. 06-1236, (C.D. Ill. Dec. 30, 2009). The case settled after the district court denied summary judgment on most of the antitrust claims. 77  Complaint at 2, United States v. United Reg’l Health Care System, No. 7:11-cv-00030 (N.D. Tex. Feb. 25, 2011), available at http://www.justice.gov/atr/cases/f267600/267651.pdf. 78  Competitive Impact Statement at 15, United States v. United Reg’l Health Care Sys., No. 7:11-cv00030 (N.D. Tex. Feb. 25, 2011), available at http://www.justice.gov/atr/cases/f267600/267653.pdf. The FTC adopted a similar approach in its case against Intel (Complaint at 9, In re Intel Corp., No. 9341 (F.T.C. filed Dec. 6, 2009)). In most circumstances, the total size of the discount is readily calculable, taking the bundled and à la carte prices as given. If that discount is allocated only to a subset of the

Healthcare Provider and Payer Markets   89

the “competitive product[s]‌,” as suggested by the court in PeaceHealth but rather to the patients that United Regional would actually be at risk of losing if an insurer were to choose non-exclusivity (the “contestable volume”). Under some factual circumstances, the contestable volume may consist of the entire volume of the overlap services. . . . Under other circumstances, however, such as in this case, the contestable volume is likely smaller than the entire volume of the “competitive product.” (Citations omitted.)

The case settled. The consent decree explicitly permits United Regional to (1) charge different prices to different insurers; (2) offer volume discounts that are based on the total volume of services purchases from United Regional (“market share discounts,” meaning discounts based on the percentage of a customer’s total requirements that are filled by the seller, are banned); (3) implement a nonlinear price schedule such that the price for units purchased above some threshold is lower than the price for units purchased below the threshold, so long as the lower price, after allocation of all discounts, is above the relevant measure of cost; and (4) participate in tiered or narrower network products.79 These provisions do not really mark a conceptual break from PeaceHealth; rather, they clarify the DOJ’s approach to questions that the Ninth Circuit did not directly address.

4.3.3. Summary During the past three decades, some of the highest profile healthcare antitrust cases have involved hospitals, and rightfully so. Waves of hospital consolidation have given many hospitals and hospital systems considerable negotiating leverage. Developments in the economic and legal analysis of hospital markets have created increased antitrust risk in both merger cases and unilateral conduct cases. At the same time, the steady shift of healthcare away from hospitals is leading to a reexamination of competition and antitrust in physician markets.

4.4.  Antitrust and Physicians Two aspects of antitrust law apply mainly to physician practices:  price-fixing and boycotts.80 In principle, all providers are capable of price-fixing. Yet we are unaware alleged monopolist’s output, rather than to all of its output, the result will be a discount-allocated net price that is lower and, therefore, more likely to be below the relevant measure of cost. 79  United States v. United Reg’l Health Care Sys., No. 7:11-cv-00030-O (N.D. Tex. Sept. 29, 2011).

80  In recent years, the FTC has also blocked a handful of physician acquisitions. See, e.g., US Federal Trade Commission, “FTC Bureau of Competition Director Issues Statement on Providence Health & Services’ Abandonment of Its Plan to Acquire Spokane Cardiology and Heart Clinics Northwest,” news

90   Cory S. Capps and David Dranove of any price-fixing allegations in modern healthcare cases that do not involve physicians or dentists.

4.4.1.  Price-Fixing and Fee Schedules As early as the nineteenth century, physicians routinely published “fee bills,” which listed their prices for common medical procedures such as removing blisters, dressing wounds, and delivering babies. Often several physicians in an area signed a joint fee bill. The 1848 “Charlottesville document” is a fee bill listing prices for nearly 100 services and signed by twelve local doctors (Savitt 2011). The Sherman Act outlawed such blatant price-fixing, but physicians still have many opportunities for tacit or explicit price collusion. Physicians often belong to medical societies organized by location and specialty. Annual meetings of these societies provide opportunities for physicians to exchange fee information, which was often a simple matter of citing their prices for their most common procedures (using the standardized Current Procedural Terminology codes). During the 1970s, the FTC took action against several medical societies that published “relative values” for CPT codes. Arizona v. Maricopa County Medical Society is perhaps the most famous of these cases.81 Maricopa County published a fee schedule that it touted as a recommended fee ceiling. The state successfully argued that the fee schedule was a facilitating practice that helped physicians gravitate towards a common fee. The rise of selective contracting has led many independent physicians to attempt to increase their bargaining leverage with insurers. For example, Preferred Health Services (PHS) represented 70 percent of the physicians in the Seneca, South Carolina, area.82 PHS established a fee schedule for its member physicians and refused to contract with health plans that did not accept the schedule. Another example is White Sands Health Care System (WSHCS), which included 80 percent of the physicians in Alamogordo, New Mexico. WSHCS used a consultant to execute a collective refusal to contract with area insurers except upon collectively agreed-upon price and nonprice terms.83

release, April 8, 2011, available at http://www.ftc.gov/opa/2011/04/providence.shtm, and US Federal Trade Commission, “FTC Order Will Restore Competition for Adult Cardiology Services in Reno, Nevada,” news release, August 6, 2012, available at http://ftc.gov/opa/2012/08/renownhealth.shtm. The FTC is currently in litigation seeking to unwind a consummated acquisition of a physician group by a hospital system in Idaho on the grounds that the combination of the acquired physicians and the system’s existing employed physicians will substantially lessen competition. US Fed. Trade Comm’n v. St. Luke’s Health System, No. 1:12-cv-00560-BLW-REB (D. Idaho filed Mar. 12, 2013), available at http://www.ftc.​ gov/os/caselist/1210069/index.shtm. 81  82  83 

457 U.S. 332 (1982). Preferred Health Services, Inc., 139 F.T.C. 266 (2005) (consent order). White Sands Health Care System, L.L.C., 139 F.T.C. 15 (2005) (consent order).

Healthcare Provider and Payer Markets   91

Since the early 1990s, the FTC and DOJ have brought actions against more than 30 ventures that brought independent physicians together for collective negotiations with payers. On the surface, these enforcement actions seem simple: a group of independent, competing sellers cannot exchange price information and/or collectively negotiate price and other terms with customers. However, these cases are not as straightforward as they seem. Many health policy experts believe that some form of provider integration is essential to reducing medical spending. Acting together, physicians can exchange information about best practices, develop clinical protocols, more efficiently share or direct referrals, and bear risk (e.g., accept incentive contracts rather than fee-for-service contracts) (Hough et al. 2010).84 In order to encourage integration without creating heightened antitrust risk, in 1996 the FTC and DOJ issued revised Statements of Antitrust Enforcement Policy in Health Care (DOJ and FTC 1996). The statements established safety zones for medical groups that wished to collectively bargain with insurers. Physicians could collectively negotiate with payers if they met any one of the following conditions: • A  market power safe harbor, applicable when a physician venture includes less than 20 percent of physicians in each affected specialty for exclusive joint ventures or 30 percent for nonexclusive joint ventures. • Physicians in the venture exchange price information through a “messenger model,” a mechanism intended to allow a physician joint venture to efficiently negotiate with a payer without facilitating tacit or overt price collusion. • The group takes significant steps to promote “financial and/or clinical integration.”85 • Although not explicitly included in the Statements, the FTC and DOJ also consider whether individual physicians may “opt out” of the collective negotiation and bargain independently with payers. When physicians can opt out, payers can reject a joint venture’s fee schedule and instead negotiate independently with each member physician. At least in principle, this reduces the group’s leverage. When a venture is outside the safe harbor, the implementation of the messenger model and the degree of integration are critical. When the messenger goes beyond gathering and disseminating rate proposals, or when the financial and clinical integration are minimal, the agencies are more likely to allege that a physician joint venture that

84  However, physician integration is clearly no guarantee of clinical or efficiency improvements. For example, the Physician Group Practice (PGP) Demonstration, a pay-for-performance pilot study launched by Medicare in 2005, involved 10 physician groups that were selected not randomly but on the basis of an expectation that they would have the scale and structure to perform well. Nearly all groups met quality benchmarks, but in no year of the program did more than half of the groups achieve meaningful savings relative to spending under Traditional Medicare (CMS 2011). 85  A group is financially integrated if it accepts the financial risk for the medical services provided to its patients, for example, by accepting capitation or participating in gainsharing. A group is clinically integrated if it uses health information technology to exchange clinical information, develops treatment protocols, and evaluates member physician quality.

92   Cory S. Capps and David Dranove engages in joint negotiations over price and other terms is subject to per se condemnation. The following cases illustrate the application of this framework and the boundaries of acceptable physician joint venture conduct.

4.4.2.  Four Illustrative Cases DOJ and FTC enforcement actions against physician groups tend to focus on two key questions. First, is the group setting prices? Second, is the group sufficiently integrated that collective price negotiations are subject to a rule of reason analysis rather than per se condemnation? The following cases illustrate the breadth of enforcement activity in this area.

4.4.2.1.  Federation of Physicians and Dentists In United States v. Federation of Physicians and Dentists, which was settled in 2002, competing orthopedic surgeons in Delaware agreed to negotiate exclusively through the Federation to resist fee reductions proposed by Blue Cross and Blue Shield of Delaware and other insurers.86 Virtually all of the orthopedic surgeons rejected the Blues’ proposed fee schedule and notified patients that they intended to terminate their Blue Cross contracts. The federation argued that it was acting as a third-party messenger, and the DOJ countered that a legitimate messenger cannot collectively negotiate on behalf of providers or organize a refusal to deal. In the consent decree, the Federation agreed to abide by the requirements of the messenger model. Even so, the Federation allegedly engaged in similar activities in other states, which led to further enforcement actions by the DOJ several years later.87

4.4.2.2.  Advocate Health Partners In 2006, the FTC challenged the conduct of Advocate Health Partners (AHP), which represented more than 2,600 independent physicians, approximately 300 hospital-employed physicians, and the Advocate hospitals.88 The FTC alleged that AHP “collectively negotiated the prices and other contract terms at which their otherwise competing member physicians would provide services to the subscribers of health plans, without any efficiency-enhancing integration of their practices sufficient to justify their conduct.” The FTC also alleged that, in 2001, AHP terminated all its member physician contracts with United Healthcare and then terminated United’s contracts with Advocate

86  United States v. Fed’n of Physicians and Dentists, No. 98-475 (D. Del. 2002), available at http://​www. justice.gov/atr/cases/f200600/200654.htm. 87  United States v. Fed’n of Physicians and Dentists, No. 1:05-cv-431 (S.D. Ohio filed June 24, 2005), available at http://www.justice.gov/atr/cases/indx26_b.htm. 88  FTC, “Advocate Health Partners, et al.; Analysis of Agreement Containing Consent Order to Aid Public Comment,” 72 Federal Register, no. 4 (January 8, 2007) at 784–787, available at http://www.gpo.​ gov/fdsys/pkg/FR-2007-01-08/pdf/E7-27.pdf. A PHO, or Physician-Hospital Organization, represents physicians and hospitals in joint negotiations with payers.

Healthcare Provider and Payer Markets   93

hospitals. Faced with the loss of nearly 3,000 physicians and the Advocate hospitals, United ultimately agreed to pay AHP physicians rates that were 20 to 30 percent higher than what it paid under direct contracts with area physicians. The resulting consent decree between the FTC and AHP precludes AHP from negotiating on behalf of member physicians, coordinating threats not to deal with a health plan, or facilitating information exchanges among member physicians regarding rates or contracting. The decree does allow AHP to pursue “qualified risk-sharing joint arrangement” or a “qualified clinically-integrated” joint arrangement. Advocate Health Partners still exists and has since taken substantial steps towards clinical integration within the group and with the hospitals where they practice.

4.4.2.3.  North Texas Specialty Physicians North Texas Specialty Physicians (NTSP) was a rare case in which the physician association did not settle shortly after being sued. The FTC sued NTSP, a group of approximately 600 physicians in the Fort Worth, Texas, area, in September 2003 and, after a full trial on the merits, an administrative law judge (ALJ) issued a ruling in favor of the FTC:89 The government proved its case . . . physicians participating in NTSP, who are otherwise competitors of each other, communicated to NTSP the minimum prices that they were willing to accept for physician services and . . . NTSP used this information to negotiate higher rates and more favorable terms for non-risk contracts . . . NTSP has engaged in a combination, contract, or conspiracy that has unreasonably restrained trade.

The ALJ further concluded that NTSP had established “no valid procompetitive justifications” for its joint setting of prices for physician services.90 NTSP appealed to the FTC commissioners, who issued a unanimous decision upholding the ALJ’s ruling. NTSP then appealed to the Fifth Circuit, where a three-judge panel also unanimously upheld the ALJ’s ruling.91

4.4.2.4.  TriState Health Partners The FTC does not deem all joint ventures of otherwise independent physicians suspect or subject to per se treatment, even when the venture negotiates pricing terms on behalf of its members. In particular, the FTC will deem a venture to be subject to a rule-ofreason analysis when the venture is structured so that (1) it is likely to generate cost

89 

Initial Decision at 1–2, In re North Texas Specialty Physicians, FTC Docket No. 9312 (ALJ Nov. 15, 2004), available at http://www.ftc.gov/os/adjpro/d9312/041116initialdecision.pdf. 90  Id. at 51–53. 91  US Federal Trade Commission Final Order, In re North Texas Specialty Physicians, No. 9312 (F.T.C. Nov. 29, 2005), available at http://www.ftc.gov/os/adjpro/d9312/051201finalorder.pdf, aff ’d in part and remanded sub nom. North Texas Specialty Physicians v. F.T.C., No. 06-60023 (5th Cir. May 14, 2008), available at http://www.ftc.gov/os/caselist/0660023/080516opinion.pdf, modified sub nom. In re North

94   Cory S. Capps and David Dranove savings and/or quality improvements and (2) joint contracting is “reasonably necessary” to obtain those efficiencies. Under a rule-of-reason analysis, the FTC will weigh the likely efficiencies against the potential or actual harm to competition. A 2009 business review letter issued to TriState Health Partners illustrates how the FTC weighs these factors (Meier 2009). TriState proposed to contract on behalf of a majority of the physicians and the only hospital in the Hagerstown, Maryland, area. After a review of TriState’s proposed structure, the FTC concluded that TriState’s proposal amounted to a “bona fide effort to create a legitimate joint venture among its physician and hospital participants that has the potential to achieve significant efficiencies.” The factors identified as most compelling to the FTC were as follows: • A  “potential to create substantial integration among its participants, with the potential to produce significant efficiencies.” • Joint contracting with payers appears to be “subordinate and reasonably related to TriState’s plan to integrate . . . [and] deliver coordinated care.” • TriState’s structure would facilitate meaningful “opt out” decisions by TriState’s member physicians, which the FTC concluded would limit TriState’s ability to exercise market power. Notably, the FTC appears to have taken it as given that TriState possessed market power: the venture included the only hospital in Hagerstown, Maryland, and two-thirds of the physicians on the staff of that hospital were part of TriState. Thus, the FTC’s decision regarding TriState strongly suggests that physician joint ventures, even ones that combine a substantial fraction of otherwise independent providers in an area, can escape investigation and prosecution if they bear financial risk, have a credible plan that is at least somewhat likely to make care delivery more efficient and effective, and do no more than is necessary to limit health plans’ ability to avoid contracting with TriState providers outside of the joint venture.92

4.5.  Antitrust and Insurers We now turn our attention to antitrust issues associated with purchasers of healthcare services. We begin by providing a brief history of private health insurance in the United States. We then discuss antitrust issues facing insurers as purchasers and as sellers. Texas Speciality Physicians, No. 9312 (F.T.C. Sept. 12, 2008, available at http://www.ftc.gov/os/adjpro/​ d9312/080912orderonremand.pdf). 92 

The FTC reiterated these points in its 2013 review of another proposed multiprovider joint venture, the Norman Physician Hospital Organization (Meier 2013). The Norman PHO includes approximately 280 physicians and the three-hospital Norman Regional Health System, and primarily serves patients from the area in and around Oklahoma City.

Healthcare Provider and Payer Markets   95

4.5.1.  Private Health Insurance in the United States Private health insurance emerged during the Great Depression. Concerned that their patients could not afford care, a handful of hospitals began charging local residents a monthly prepayment (about $2 a month) in exchange for guaranteed free medical care. Baylor University Hospital is credited as the first to offer such a deal. Shortly thereafter, hospitals in St. Paul, Minnesota, offered a similar deal under the name “Blue Cross.”93 By 1940, there were 56 Blue Cross plans spanning nearly every state. Physicians soon offered their own prepayment plans under the Blue Shield trademark. Over the ensuing decades, Blue Cross and Blue Shield plans remained independent organizations only loosely governed by the national Blue Cross and Blue Shield Association, which controls the use of the “Blue” trademarks. Individual Blue plans may only operate under the “Blue” brand within the geographic boundaries licensed by the Association. Thus, for example, Blue Cross and Blue Shield of Illinois does not compete for business with Blue Cross and Blue Shield of Indiana.94 Shortly after the emergence of the Blues, for-profit life insurers and property/casualty insurers such as Aetna and Cigna entered the health insurance market. By 1940, about 12 million Americans had health insurance; the Blues and the commercial insurers each claimed about half of the market. After World War II, demand for health insurance soared, catalyzed by IRS rules that exempted employer-sponsored health insurance from income taxation (Thomasson 2003). By 1960, most employed Americans and their dependents had private health insurance. All states regulate private health insurance. During the 1930s and 1940s, states enacted enabling regulations that granted the Blues nonprofit status. In exchange, Blues plans had to “community rate” certain policies (i.e., they could not adjust premiums based on individuals’ or groups’ past or predicted health expenditures). States no longer require the Blues to community rate. At the same time, many Blues have converted from nonprofit status to mutual or for-profit status. The nation’s largest insurer by gross revenues is Wellpoint, a for-profit company that controls Blue Cross and/or Blue Shield trademarks in 14 states. The states license commercial insurers, so “national” insurers such Aetna and Cigna must operate individual insurance companies in each state where they do business. Today, there are 38 distinct Blue Cross and Blue Shield plans that, collectively, provide coverage to over 90 million enrollees (Donahue 2012). Aetna, United Healthcare, Cigna, and Humana each do business in over 40 states. There are hundreds of other smaller commercial insurers that do business in just one or a handful of states. 93  The origins of Blue Cross and Blue Shield plans are summarized in Cunningham and Cunningham (1997) and Austin and Hungerford (2009). 94  Blues plans can compete in one another’s territories if they do not use the “Blue” brand name— these are referred to as “unbranded” products. WellPoint’s UniCare product is one example. Formally, “Blue Cross” and “Blue Shield” are separately licensed; where the two licenses are separately held, the corresponding entities can compete directly in the same geographic area. For example, in California, the Blue Cross license is held by WellPoint and the Blue Shield license is held by Blue Shield of California (BSC). WellPoint and BSC compete directly in California.

96   Cory S. Capps and David Dranove

4.5.2.  The Geographic Boundaries of Health Insurance Competition Until the growth of managed care, it was reasonable to think of each state as a distinct geographic health insurance market. State boundaries described the extent of each commercial insurer’s business, and, with some exceptions, the Blues plans’ boundaries loosely corresponded to state borders. Thus, in any given state, the same insurers competed for business throughout the state. Historically, the Blues plans held the largest market shares in most states.95 In 1968, for example, Blues plans in the median state had a market share of 34 percent and exceeded 50 percent in six states. Since the late 1960s, the Blues have increased their market shares in many states. According to an analysis by the American Medical Association, Blue shares now reportedly exceed 50 percent in 17 states (AMA 2011).96 Under managed care, it is generally inappropriate to treat the state as the geographic market. Managed care organizations compete by assembling networks of providers, and enrollees strongly prefer networks that offer local access. Thus, MCOs competing in, say, Boston, must contract with Boston-area providers. Because it is costly to manage provider networks, not all insurers licensed to do business in a given state will have networks throughout the state. Thus, it is more appropriate to define the geographic boundaries of insurance competition under managed care as the metropolitan area, because this approximates the area from which employers draw their employees and therefore the area in which each insurer must assemble a network. According to the most recent American Medical Association report, 83 percent of metropolitan areas are highly concentrated; in the large majority of those areas, the largest plan is a Blues plan (AMA 2011). Over time, the percentage of US employers with substantial operations in multiple states has increased. These employers may prefer “one stop shopping” for insurance and may also wish to give their employees access to local networks as they travel. Large, national commercial insurers provide access to all of their local networks. Blues plans also provide national networks through the “BlueCard” program (BCBS 2012). Dranove, Gron, and Mazzeo (2003) show that these national plans constitute a unique market

95 

The Sourcebook of Health Insurance Data reports the combined Blues market shares by state dating back to the 1960s; see, for example, Health Insurance Association of America (1970). 96  Accurate data on health insurer market shares are not publicly available, and researchers have identified a number of apparent flaws in the AMA data and calculations. For example, Dafny et al. (2011) study the AMA data and alternate sources of health insurance market share data and report that “the two datasets most commonly used to estimate market concentration differ considerably from each other (both in levels and in changes over time), and reflect implausibly high volatility in market shares.” Capps (2009) reaches a similar finding. Contrary to the picture painted in the AMA reports, Melnick, Shen, and Wu (2011, 1731) report, using 2004 data, that “64 percent of hospitals operate in [MSAs] where health plans are not very concentrated and only 7 percent are in the most concentrated health plan markets.” The disparity is attributable to two factors. First, the most populous MSAs tend to have both more hospitals and less-concentrated insurer

Healthcare Provider and Payer Markets   97

for employers seeking national coverage. An important implication is that the extent of competition in the market for national employers may differ from the extent of competition in the market for local employers, where the latter is confined to insurers with local networks. The relevant product market may also differ with respect to small employers and large employers. The latter are more likely to self-fund and can purchase administrative services from either a national insurer or a noninsurer third-party administrator.

4.5.3.  Selective Contracting and Insurer Market Structure One of the major features of health insurance markets is the persistence of market structure. Prior to the growth of managed care, the dominance of many Blues plans could be attributed to state regulation, combined with deliberate strategic choices made by the Blues. Recall that the Blues were originally owned by the providers whose services they were covering. The Blues could therefore pay themselves lower reimbursement rates (through the UCR and cost-based reimbursement systems discussed earlier) and assure themselves an advantageous cost position. The Blues dissipated some of this advantage by maintaining low copayments so as to boost demand for medical services (Frech and Ginsberg 1978). The Blues no longer enjoy these same protections, and yet they continue to dominate many markets. In fact, there is considerable correlation (p = .34) between Blues shares in 1968 and today.97 This suggests that there may be other aspects of managed care contracting that create barriers to entry and growth. Handel (2013) finds that enrollees display a remarkable degree of inertia when choosing their health plans, often leavings thousands of dollars on the table by making the wrong choice.98 But this does not explain why employers would consistently offer the same plans (usually Blues plans). One possibility is that, just as the Blues once used UCRs and cost-based reimbursements to obtain lower provider prices, today’s dominant insurers use their sheer size to obtain lower prices from providers. Lower provider prices could enable a dominant insurer to earn economic profits even as it sets its premiums below the level that would induce entry.99 In short, an initial scale advantage could create a powerful source of sustainable advantage for

markets. Second, Melnick, Shen, and Wu use a cutoff HHI of 3,200 to define the “most concentrated” health plan markets, whereas the AMA uses the more standard 2,500 cutoff (an HHIs of 3,200 roughly corresponds to three equally sized firms, while 2,500 corresponds to four equally sized firms). Data limitations and conflicting conclusions aside, it seems clear that the various Blue plans have long maintained leading market shares in many areas. 97 

This is based on the authors’ calculations and data in Health Insurance Association of America (1970) and AMA (2011). 98  Interestingly, Handel finds that such inertia creates a benefit in that it reduces the problem of adverse selection. 99  Other factors may also limit entry. These include state regulation, employer switching costs, and the likelihood that a new health plan in an area would tend to draw a more adversely selected pool of enrollees than an established plan.

98   Cory S. Capps and David Dranove an incumbent health plan.100 These scale advantages have also triggered some of the most contentious antitrust cases involving health insurers.

4.5.4.  Health Insurance and Monopsony Antitrust cases involving health insurers tend to focus on their monopsony power vis-àvis providers. Early monopsony cases met with little success. In Kartell v. Blue Shield of Mass., a group of physicians challenged Blue Shield’s policy of disallowing balance billing.101 Plaintiffs claimed that the ban constituted monopolization or an attempt to monopolize. The US Court of Appeals for the First Circuit accepted the plaintiffs’ contention that Blue Shield had monopsony power; however, the court concluded that Blue Shield was “itself the purchaser of doctors’ services” rather than an intermediary “third force” in the marketplace and, therefore, that Blue Shield had the right to specify the terms of its purchases.102 In particular, the court concluded that by aggregating purchasing power, Blue Shield legitimately obtained discounted rates that benefited its subscribers.103 In Ball Mem’l Hosp. v. Mutual Hosp. Ins., the plaintiff hospital objected to the local Blue Cross Blue Shield plan’s effort to establish a PPO. The US Court of Appeals for the Seventh Circuit rejected the plaintiff ’s claim that the local Blues plan, Mutual Hospital Insurance, had monopsony power.104 While Mutual Hospital Insurance controlled over 50 percent of the market, the court held that entry barriers were minimal and therefore concluded that Mutual Hospital Insurance could not exercise any market power. In Reazin v. Blue Cross & Blue Shield of Kan., the court questioned whether a Blues plan with a large market share would be threatened by entry.105 In this case, the Blues plan obtained larger discounts from two hospitals in exchange for making them exclusive members of its 100  Two recent studies have concluded that insurers with higher market shares in a metropolitan area pay lower prices to providers. Moriya, Vogt, and Gaynor (2010) use a database containing three years of claims for 11 million commercial health plan enrollees and find that “increases in insurance market concentration are significantly associated with decreases in hospital prices” Melnick, Shen, and Wu (2011) reach a similar finding. 101  Kartell v. Blue Shield of Mass, 749 F.2d 922 (1st Cir. 1984), cert. denied, 471 U.S. 1029 (1985). Providers balance bills by billing patients for charges in excess of those allowed by the payer. 102  The court explained its reasoning as follows: “Antitrust law rarely stops the buyer of a service from trying to determine the price or characteristics of the product that will be sold. Thus, the more closely Blue Shield’s activities resemble, in essence, those of a purchaser, the less likely that they are unlawful.” In this respect, the court’s reasoning follows that traditionally applied to firms with monopoly power in output markets: exercising legitimately obtained market power in the form of higher prices does not, in itself, violate the antitrust laws. Id. at 924–28. As we discuss elsewhere, however, contractual restrictions that serve to maintain or extend market power can lessen competition and violate the antitrust laws. 103  Ultimately, then, the court concluded that even purchasers with substantial market power have a unilateral right to demand lower prices—even monopsony prices—and that that right would be meaningless if physicians were free to bill patients for the difference. (“If it is lawful for a monopoly buyer to buy for the account of another, how can it be unlawful for him to insist that no additional charge be made to that other?” Id. at 928.) 104  Ball Mem’l Hosp. v. Mutual Hosp. Ins, 784 F.2d 1325 (7th Cir. 1986). 105  Reazin v. Blue Cross & Blue Shield of Kan., 899 F.2d 951 (10th Cir. 1990), cert. denied, 497 U.S. 1005 (1990).

Healthcare Provider and Payer Markets   99

hospital network. The excluded hospital, Wesley Medical Center, and some of its medical staff complained that this constituted a conspiracy resulting in a vertical restraint. While the court found for the plaintiffs, the facts of the case were somewhat unusual (Wesley offered a competing HMO plan and the other two hospitals agreed not to grant similar discounts to Wesley’s plan). Thus, this case did not materially affect efforts by other health insurance plans to grant discounts in exchange for exclusivity in managed care networks. Blue Cross & Blue Shield of Kansas’s arrangement with the two exclusive hospitals in the Reazin case is similar to a most favored nation (MFN) clause. To date, courts have been reluctant to strike down MFN clauses, and in Ocean State Physicians Health Plan v. Blue Cross and Blue Shield of R.I., the court used logic similar to that in Kartell to uphold an MFN.106 The jury in Ocean State ruled that the Blues had monopsony power to enforce its MFN but awarded no damages. The district court then issued a directed verdict of no monopsony power, which the appellate court upheld.107 This brief review shows that different federal courts have offered different reasons for being suspicious of monopsony cases. The Kartell court argued that monopsonists pass on their savings to consumers. The Ball Memorial court asserted that incumbent insurers enjoyed few barriers to entry. Clearly, economic analysis of health insurer monopsony should consider consumer injury and entry barriers, but it is not clear whether the Kartell and Ball Memorial analyses get these analyses right.

4.5.4.1.  Linking Monopsony to Monopoly The Kartell analysis appears to say that monopsony harm is predicated on monopoly power. Indeed, buy-side harm does not usually arise in the absence of sell-side harm. (We discuss sell-side issues in the next section.) An exception can arise when the scope of the upstream and downstream markets differs substantially. For example, in Cargill, the DOJ alleged that a merger of grain traders would lessen competition in selected geographic markets for the purchase of corn, soybeans, and wheat.108 The DOJ did not, however, allege that the customers of the grain traders would face any price increases or output restrictions, presumably because grain prices are determined on the world market, and Cargill’s postmerger share in that output market would be quite small.109 Without such a divergence, competition in the output market is likely to thwart attempts at monopsonization. 106  Ocean State Physicians Health Plan v. Blue Cross and Blue Shield of R.I., 883 F.2d 1101 (1st Cir. 1989), cert. denied, 494 U.S. 1027 (1990). 107  Indeed, the circuit court essentially stated that, with the potential exception of predation, MFN provisions are per se legal: “We agree with the district court that such a policy of insisting on a supplier’s lowest price—assuming that the price is not ‘predatory’ or below the supplier’s incremental cost—tends to further competition on the merits and, as a matter of law, is not exclusionary.” Id at 1110. 108  Complaint at 17–19, United States v. Cargill, No. 1:99CV01875 (U.S.D.C. July 8, 1999), available at http://www.usdoj.gov/atr/cases/f2500/2552.pdf. Even without a lessening of competition in output markets, monopsony power can reduce total welfare because it distorts incentives. Monopsonists suppress their purchase of monopsonized inputs, and the monopsonized producers will respond by shifting from the best deployment of their efforts to some lesser alternative, or perhaps by reducing quality. 109  The same possibility exists when the relevant sell-side product market is considerably broader than the buy-side market. In the Weyerhaeuser case the jury found—and the Ninth Circuit upheld— that Weyerhaeuser obtained a monopsony over alder logs but that a separate market for the sale of

100   Cory S. Capps and David Dranove By the same token, consider a case where a merged insurer accounts for a relatively high share of a number of physician practices’ revenue but faces strong competition from rival insurers. If the merged insurer were to reduce reimbursements, the rival insurers could respond strategically by offering a reimbursement rate slightly above the merged insurer’s attempted monopsony rate but below the premerger rate. This would reduce those rivals’ costs while also drawing physicians away from the merged insurer. An accompanying slight reduction in health insurance premiums would divert customers from the merged insurer. This strategy by rivals would be likely to render unprofitable the attempt to monopsonize. This analysis seems to confirm the Kartell view that monopsony requires monopoly power, but the converse is not necessary true. For example, because small group insurance accounts for a relatively small share of total commercial enrollment, a merger could result in anticompetitive effects in a sell-side market for the sale of commercial health plans to small employers yet have only a negligible effect on buy-side concentration. Put another way, monopoly harm need not be predicated on monopsony power. The Kartell view can be upset by expanding the analysis to consider quality. When a powerful insurer secures lower prices from medical providers, this may lead providers to reduce quality (Dranove and Satterthwaite 2000). We might expect the insurer to internalize the cost of reduced quality, because it must sell this lower cost / lower quality product to enrollees. But quality in medical care may be a public good. That is, providers tend to set a somewhat uniform quality level for all patients rather than differentiate their level of care according to each patient’s insurance carrier. Thus, if a powerful insurer pays lower fees, providers may reduce quality somewhat for all patients. The powerful insurer’s subscribers may be worse off, but subscribers to other insurance plans may fare even worse, because they experience a quality reduction without the benefit of lower cost. Under these conditions, the monopsonist insurer obtains a competitive advantage over other insurers even as it leaves its subscribers worse off. In this way, monopsony harm need not be predicated on monopoly power.

4.5.4.2.  Entry Barriers The Ball Memorial finding that barriers to entry are low predated the growth of managed care. This finding received considerable criticism for seeming to overlook the importance of brand reputation, medical underwriting, and the tax advantages enjoyed by the Blues. The growth of managed care has likely increased barriers to entry. It is costly for insurers to create and manage provider networks.110 Moreover, if large insurers can use their monopsony power to negotiate lower prices from in-network providers, this will alder lumber had not been established, presumably because of competition from other hardwoods. See Confederated Tribes of Siletz Indians of Or. v. Weyerhaeuser Co., 411 F.3d 1030 (9th Cir. 2005); Weyerhaeuser Co. v. Ross-Simmons Hardwood Lumber Co., 127 S. Ct. 1069 (2007). For a discussion of the interaction between the buy side and sell side in that case, see Werden (2007). 110  In some markets, incumbent, local insurers rent their networks to other insurers. In a rental network agreement, the incumbent negotiates discounted rates from providers. The entrant “rents” the network by paying the same negotiated rates when its enrollees use those providers, along with a network access fee paid to the renting insurer. An incumbent insurer need not rent its network, of course.

Healthcare Provider and Payer Markets   101

give them a cost advantage that can allow them to undercut potential or actual entrants. This can allow monopsonists to sustain their market power over the longer term. Note once again how monopsony and monopoly go hand in hand.

4.5.4.3.  Monopsony, Medicare, and Medicaid: Further Theoretical Issues When measuring market shares in monopsony cases, an insurer’s leverage is driven in large part by its share of providers’ business. Most providers serve Medicare and Medicaid patients, who can account for half or more of their business.111 This raises a critical question: should Medicare and Medicaid be included when measuring an insurer’s market share and assessing monopsony power? Given the size of the Medicare and Medicaid programs, it would seem that suppliers have very good alternatives should an insurer demand discounts. But size is not everything. In most states, Medicaid pays far lower rates than private insurers. When given a choice between replacing all patients covered by a powerful insurer with all Medicaid patients or giving a discount to the powerful insurer, the latter choice usually wins. Medicare has been more generous than Medicaid, and providers might be more willing to substitute from private insurers to Medicare.112 Even so, this may not limit the exercise of monopsony power. In a traditional market, if a powerful purchaser demands excessive price concessions, suppliers take their business elsewhere, selling fewer inputs to the powerful purchaser and more to its rivals. This dynamic limits the ability of a purchaser to exercise its power. Now consider the medical market. If physicians refuse to provide services to a powerful insurer, they could take their business to Medicare. But the amount of business that can be provided to Medicare is more or less fixed; Medicare prices will not adjust to the influx of new suppliers, and demand for Medicare services is unlikely to grow. That is, doing business with Medicare is a zero-sum game: if some providers are able to do more business with Medicare, others will invariably do less. Thus, Medicare does not necessarily discipline private insurers in the input market. This factor argues against including government payers in the input market when analyzing monopsony power. At the same time, prices in healthcare input markets are commonly set in negotiations.113 As we described in section 4.3.1.4, when prices are determined in negotiations, 111  In this respect, there is a fundamental distinction between the input and output markets: providers can turn to government payers as an alternative to commercial payers, but commercial health plan customers (firms and individuals) generally cannot substitute to Medicare or Medicaid. 112  In some markets, physicians receive similar rates from Medicare and commercial insurers (the same is rarely true of hospitals). 113  Essentially all prices paid by commercial insurers to contracted hospitals are negotiated prices. In the case of physicians, pricing can be negotiated (particularly with larger provider groups) or can follow a posted price model. In the posted price model, an insurer circulates a fee schedule to physicians, and physicians choose to accept or reject the offered rates; as the offered rates are higher, more physicians will accept. Even in a posted price model, if the government sector is more attractive in a local area, physicians are likely to demand higher rates to provide any given volume of services to privately insured patients. This follows from a standard model of price discrimination in a mixed economy (Dranove, Ramanarayanan, and Watanabe 2012).

102   Cory S. Capps and David Dranove outcomes are driven by the outside options of the negotiating parties. All else equal, as government payers are either more remunerative or account for a higher proportion of providers’ business, or both, providers’ outside options in negotiations with commercial insurers will be more attractive. This consideration argues for including government payers in the input market when analyzing monopsony power. In short, economic theory does not yield an a priori bright line answer to the question of whether Medicare and Medicaid should be included in the calculation of market shares when analyzing monopsony powers.114 Government payers are relevant, but in ways that differ from commercial payers and to an extent that varies according to market circumstances. Simple share analyses based on fully including or fully excluding Medicare and Medicaid are likely to be misleading. This leaves little choice but to undertake a more intensive and fact-specific exploration of the market or markets at issue in antitrust cases that involve monopsony concerns.

4.5.4.4.  Aetna-Prudential and United-PacifiCare Many of these theoretical and empirical considerations were on display in two insurer mergers that were contested, partly on the basis of monopsony concerns, by the DOJ: the 1999 Aetna-Prudential merger and the 2005 United-PacifiCare merger. The complaints in both cases described similar theories of monopsony harm. Specifically, the DOJ cited two relevant buy-side market shares as indicators of the likelihood of monopsony harm: (1) the postmerger share of individual physician practices’ business accounted for by the merging insurers and (2) the postmerger share of all physician business in the locality. The theory of harm was not simply that it is more difficult for a physician to replace a greater number of lost patients than a smaller number of lost patients, although that is obviously true. Rather, the theory was that a high postmerger share of individual physician practices’ revenue can create monopsony power under scenarios in which it is increasingly costly for a physician practice to replace additional lost patients. That is, monopsony power is more likely to result from higher postmerger buy-side shares when the cost to providers of losing patients increases more than proportionally, or nonlinearly, with the number of lost patients.115 The DOJ’s buy-side theories of harm relied on the premise that the nonlinearity condition will hold when postmerger shares at the physician practice level and locality-wide level are high. As an example in which the increasing costs condition would hold, suppose physician groups can replace one lost patient per day. If a group loses 1 patient, it would lose 1 patient-day of revenue; if it loses 2 patients, it would lose 3 patient days of revenue; if 114  Capps (2010) discusses the analysis of buy-side market power in health plan mergers and recommends basing market share calculations on shares of profits rather than on revenue or on payers’ number of enrollees. Using shares of profits would, for example give no weight to Medicaid if Medicaid patients are unprofitable at the margin, and it would give more weight to private insurers insofar as they pay rates above government rates. 115  Absent this condition, simply doubling the size of an insurer through a merger would not be likely to affect bargaining outcomes: although the provider is twice as dependent on the insurer after the merger, the insurer is also twice as reliant upon the provider.

Healthcare Provider and Payer Markets   103

it loses 5 patients, it would lose 15 patient days of revenue (5 + 4 + 3 + 2 + 1 = 15); and so on. Thus, losing 5 patients is 15 times as costly as losing 1 patient. In this context, a high area-wide insurer share can be viewed as reducing the arrival rate of replacement patients (because replacement patients must come from insurers other than the merged entity).116 With respect to market definition, the DOJ identified metropolitan areas as relevant geographic markets in both cases. Regarding product market definition, in the earlier case, Aetna-Prudential, DOJ made no substantive distinction between commercial and government payers, describing the relevant product market as follows:117 There are no purchasers to whom physicians can sell their services other than individual patients or the commercial and government health insurers who purchase physician services on their behalf. . . . Physicians’ services, therefore, also constitute an appropriate relevant product market within which to assess the likely effects of the proposed transaction [on physicians].

Six years later, in United-PacifiCare, the DOJ still defined the market for physician services based on all payers, but it took explicit note of differences in government and private payment levels:118 As a result of the proposed Transaction, United will account for a large share of total payments to all physicians in the Boulder and Tucson areas. . . . These revenue shares understate the importance to physicians of payments from commercial health insurance plans. The total payments made to physicians include revenue earned by treating patients covered by Medicare and Medicaid, which account for a substantial amount of revenue for many physicians. Physicians typically consider commercial health insurance business more profitable than Medicare and Medicaid business. Many physicians use their commercial health insurance business to compensate for the lower revenue earned from Medicare and Medicaid business.

4.5.4.5.  Empirical Evidence A number of papers published since the advent of managed care have studied the relationship between concentration in the insurance market and prices paid to providers. For the most part, these studies find a negative relationship between concentration (or health plan market shares) and prices paid to providers.119 Although it is consistent with 116  This theory was explained in more detail in Schwartz (1999). See also c ­ hapter 6, section III in US DOJ and FTC (2004). 117  Complaint at ¶ 27, United States v. Aetna Inc., No. 3-99 CV 398-H (N.D. Tex. June 21, 1999), available at http://www.justice.gov/atr/cases/f2500/2501.pdf. 118  Complaint at ¶ 41, No. 1:05CV02436, United States v. UnitedHealth Grp. (D.D.C. Dec. 20, 2005), available at http://www.justice.gov/atr/cases/f213800/213815.htm. 119  E.g., Adamache and Sloan 1983; Feldman and Wholey 2001; Sorensen 2003; Moriya, Vogt, and Gaynor 2010; and Melnick, Shen, and Wu 2011. For a more comprehensive review, see Gaynor and Town (2011).

104   Cory S. Capps and David Dranove an exercise of monopsony power, this stylized fact does not establish that insurers are exercising such power or that prices paid to providers are below competitive levels. There are two primary alternative possibilities. First, causality could run in the other direction, with insurers that are more able to obtain favorable provider pricing in an area (through means other than monopsony power) gaining market share. Second, if providers possess market power in an area, economic theory does not offer a clear prediction as to whether the effect of monopsony power on the part of insurers will be inefficiently low provider rates. In principle, these alternative explanations could be tested empirically. For example, if more favorable provider pricing causes an insurer to gain market share rather than the converse, insurers that negotiate more favorable pricing should be observed to charge lower premiums (the mechanism by which they would gain market share). However, none of the studies that document a negative relationship between insurer concentration and provider prices have linked that result to premiums. This reflects the dearth of suitable data on premiums rather than a lack of scholarly interest. One exception is the recent study by Dafny, Duggan, and Ramanarayanan (2012) of the effects of the 1999 Aetna-Prudential merger. They find that in metropolitan areas most affected by the merger, (1) physician earnings grew more slowly than in unaffected areas, (2) total employment of healthcare providers (of all types) fell, and (3) nurse earnings increased relative to physician earnings. These results are strongly consistent with the hypothesis of a postmerger exercise of monopsony power in areas of higher overlap.

4.5.4.6.  MFNs and Exclusive Territories In Kartell and the other early cases against the Blues, courts seem to adopt two economic theories: (1) in general, the exercise of monopsony power benefits consumers and (2) the specific use of MFN clauses benefits consumers. Economic theory is not as kind to these principles, and there is empirical evidence that MFNs can harm consumers (Scott-Morton 1997a, 1997b). Economists agree that many monopsony claims should be handled through a data-intensive analysis and not dismissed as easily as in Ocean State—as the mirror image of monopoly power, monopsony power does not automatically benefit consumers and can instead result in inefficiency and reduced output. There is a recent movement to revisit MFNs; for example, both government and private plaintiffs have launched recent challenges to the use of MFNs by Blue plans. In 2010, the DOJ sued BCBS of Michigan, alleging that its use of MFN and “MFN-plus” provisions in contracts with roughly half of the hospitals in Michigan had reduced competition by “(1) reducing the ability of other health insurers to compete with Blue Cross, or actually excluding Blue Cross’ competitors in certain markets, and (2) raising prices paid by Blue Cross’ competitors and by self-insured employers.”120 Two main mechanisms underlie DOJ’s theory of harm. First, when a hospital agrees to an MFN with BCBS, cutting price to other insurers becomes more costly for the hospital, 120  Complaint at 2, United States v. Blue Cross Blue Shield of Mich., No. 2:10-cv-14155 (E.D. Mich. Oct. 18, 2010), available at http://www.justice.gov/atr/cases/f263200/263235.pdf.

Healthcare Provider and Payer Markets   105

because the hospital must then also reduce its price to BCBS. Second, the DOJ alleged that MFNs make it more difficult for other insurers to enter or expand in an effort to compete with BCBS.121 Ultimately, in response to the State of Michigan’s enactment of a law prohibiting health insurers from entering into MFNs with healthcare providers, the DOJ dismissed its case on the eve of trial.122 However, BCBS of Michigan has also been sued by a class of customers and, separately, by Aetna.123 Those cases remain active. Several other recent private suits have also challenged the use of MFNs by Blues plans, as well as the long-standing use of exclusive territories by the Blue Cross and Blue Shield Association, the entity that controls the licensing of the Blue Cross and the Blue Shield trademarks.124 The outcomes of these cases are all still uncertain, but even at this early stage it seems clear that the use of vertical restrictions now invites a greater degree of antitrust risk than in years past.

4.5.5.  Health Insurers and Monopoly Many if not most US metropolitan areas have concentrated insurance markets, yet there is relatively little evidence on the effect of market concentration on insurance premiums. This largely reflects two related data issues. First, it is difficult to find useful information about health insurance premiums.125 There are a handful of national surveys of enrollees, but these have limited sample sizes, and respondents enrolled in employer-sponsored plans usually do not know the premiums. Second, several sources provide some information on health plan market shares, but Dafny and colleagues (2011) observe that the shares are not highly correlated across surveys and that there is little to no correlation in changes in plan shares over time—that is, the available data are not highly reliable. An important implication of this fact is that empirical studies that rely on plan fixed effects to examine the impact of changes in market shares on changes in premiums may be unreliable. Two recent studies that use detailed plan data from a benefits consulting firm present the most compelling evidence to date that market structure affects pricing. Dafny 121  It is common for hospitals to grant more favorable pricing to large insurers (essentially, a volume discount), so it is not clear that, under normal circumstances, a hospital would offer a small insurer more favorable pricing than it offers to BCBS of Michigan in any event. 122  US Department of Justice, “Justice Department Files Motion to Dismiss Antitrust Lawsuit Against Blue Cross Blue Shield of Michigan After Michigan Passes Law to Prohibit Health Insurers from Using Most Favored Nation Clauses in Provider Contracts,” news release, March 25, 2013, available at http:// www.justice.gov/opa/pr/2013/March/13-at-345.html. 123  Aetna Inc. v. Blue Cross Blue Shield of Mich., No. 2:11-cv-15346 (E.D. Mich. 2012). The court in the Aetna case rejected BCBS of Michigan’s motion to dismiss (Flaherty 2012). 124  Richards v. Blue Cross & Blue Shield of Ala., No. cv-01133 (N.D. Ala. Apr. 17, 2012), Cerven v. Blue Cross and Blue Shield of N.C., No. 5:12-cv-00017 (W.D.N.C. Feb. 7, 2012), and UPMC. v. Highmark, Inc., No. 2:12-cv-00692 (W.D. Pa. May. 23, 2012). See also Longstreth (2013). 125  Gaynor and Town (2011) find similar shortcomings in data on private health insurers in European nations.

106   Cory S. Capps and David Dranove Duggan, and Ramanarayanan’s (2012) study of the Aetna-Prudential merger finds that premiums increased in metropolitan areas where the merger led to a nontrivial increase in the HHI. Extrapolating from this finding, they estimate that the cumulative effect of all recent insurance mergers on premiums is about 7 percent. The second, Dafny’s (2010) study, uses a novel approach to identify market power. Modeling bargaining between employers and insurers, she predicts that insurers with market power would charge higher premiums to employers that are more profitable. If insurers did not have market power, such price discrimination would not be feasible and premiums should be unrelated to employer profitability. She finds that there is a relationship between employer profits and premiums in markets that have eight or fewer insurers, and that the relationship is strongest in markets with four or fewer insurers. This finding is consistent with research by Dranove, Gron, and Mazzeo (2003), who examine entry patterns in metropolitan areas. These patterns suggest that premiums continue to decrease until there are six or more insurers in a market.

4.5.5.1.  Antitrust Enforcement of Health Plan Mergers The DOJ has been fairly active in its review of health plan mergers. Since 2005, the DOJ has required divestitures in United-PacifiCare, United-Sierra, Humana-Arcadian, and BCBS of Montana–New West Health. Prior to 2005, the last time DOJ required a divestiture in a health insurer merger was the 1999 combination of Aetna and Prudential. In contrast to hospital mergers, relevant geographic markets have not been controversial in health plan mergers. Instead, the DOJ has consistently taken the position that health insurance markets are local, typically using the metropolitan statistical area (MSA) as the geographic unit of analysis. Definition of the relevant product market has, however, been more contentious. There are two areas of contention, the first being more relevant to mergers that involve commercial health insurance plans and the second being more relevant to mergers among Medicare Advantage plans.126 In the commercial arena, in the 1999 Aetna-Prudential merger, DOJ alleged that HMO insurance, which featured tighter utilization restrictions and narrower sets of in-network providers, was in a distinct product market from indemnity and PPO insurance. Based on this distinction, DOJ excluded the majority of enrollment in plans offered by the various Blue entities. Had the relevant product market included PPO and indemnity offerings, the alleged postmerger shares and concentration levels in the overlap markets would have been lower.127 126  Under the Medicare Advantage program, private entities including health insurers and some provider-led organizations submit bids to CMS. Bids are based on the entity’s estimated costs, inclusive of a profit margin, of providing the same set of services covered by traditional Medicare. Insofar as a plan can bid below Traditional Medicare’s costs, a portion of the savings accrue to taxpayers, and a portion must be allocated to additional benefits such as reduced cost sharing or services not covered by traditional Medicare. 127  Complaint, United States v. Aetna Inc., No. 3-99 CV 398-H (N.D. Tex. June 21, 1999), available at http://www.justice.gov/atr/cases/f2500/2501.pdf.

Healthcare Provider and Payer Markets   107

By 2006, when it reviewed the United-PacifiCare merger, however, the DOJ dropped the HMO-PPO distinction. This does not indicate a lack of consistency but instead reflects market shifts that began in the late 1990s and accelerated in the 2000s. First, by the mid-2000s, indemnity insurance had all but disappeared. Second and more significantly, in response to market demands, most HMO plans had significantly reduced their utilization controls (e.g., prior approval for specialist visits) and expanded their networks. As a result, HMOs and PPOs were much more similar in 2006 than in 1999. In United-PacifiCare, DOJ introduced a distinction between small group and large group employers. The primary rationale for this distinction is that most large group employers self-fund, meaning that they directly pay for the covered healthcare expenses of their enrollees and turn to third parties only for the related administrative services, such as benefit design, claims processing, and provider contracting. Because many entities—known as third-party administrators (TPAs)—that are not health insurers offer these services, large group employers generally have a broader set of competitive alternatives. In contrast, self-funding is a less viable option for smaller employers, roughly defined as firms with fewer than 50 employees, because they cannot reliably predict healthcare spending and are usually too small to bear the risk of exceptionally costly claims. The DOJ drew a similar distinction in its 2011 consent decree with BCBS of Montana in relation to its acquisition of New West Health Services.128 Moving forward, DOJ is likely to continue to draw this distinction. An interesting but open question is the effect that health insurance exchanges, which are intended to expand insurance offerings in the individual and small group segments, will have on product definition and competitive effects analyses in future health insurer mergers. The second context in which product market definition has played a critical role is mergers of health insurers offering Medicare Advantage plans.129 To date, the DOJ has alleged that the relevant product market excludes traditional Medicare entirely. In most areas of the country, between 40 percent and 85 percent of the eligible population selects traditional Medicare. As a result, postmerger shares and concentration levels will vary dramatically depending on whether or not traditional Medicare is included. For example, in United-Sierra, DOJ defined a Medicare Advantage only relevant product market and, as a result, concluded that the merger would give United a 94 percent market share in the Las Vegas area; had DOJ included traditional Medicare, the postmerger share would have been about 37 percent.130 DOJ reiterated this relevant product market in its 128 

Complaint, United States v. Blue Cross and Blue Shield of Mont., No. 1:11-cv-00123 (D. Mont. Nov. 8, 2011), available at http://www.justice.gov/atr/cases/f277100/277177.pdf. 129  Because most Medicare Advantage plans already pay provider rates that are close to the administratively set Medicare rates, monopsony harm has not been an issue in these mergers. 130  The DOJ stated: A sufficient number of seniors in the Las Vegas area would not switch away from Medicare Advantage plans to traditional Medicare in the event of a small but significant reduction in benefits under the plans, or a small but significant increase in price, to render the benefit decrease or price increase unprofitable. Complaint at 6, United States v. United Health Grp. Inc., No. 1:08-cv-00322 (D.D.C. Feb. 25, 2008), available at http://www.justice.gov/atr/cases/f230400/230447.pdf.

108   Cory S. Capps and David Dranove 2011 consent decree in Humana-Arcadian, where it imposed divestitures in 45 counties.131 Moving forward, DOJ appears likely to continue to analyze Medicare Advantage mergers with this same product market in mind. Even where it is the case that Medicare Advantage is a relevant market, treating traditional Medicare as a priori fully in or fully out of the relevant product market is unsatisfactory; full inclusion invites underenforcement and full exclusion risks overenforcement.132 Health plan offerings are differentiated, and the degree to which traditional Medicare is a substitute is likely to vary across customers and geographies. Whether as an alternative or a complement to structural analysis, approaches that do not hinge upon market definition and presumptions from market shares, such as merger simulation and upward pricing pressure (UPP) analysis, could avoid these problems (Farrell and Shapiro 2010).

4.6.  Competition Policy in an Era of Healthcare Reform and Accountable Care Organizations The US Supreme Court’s five to four decision in June 2012 to uphold the Affordable Care Act (ACA) ensures that the US healthcare system will continue to rely on market forces in pursuing the twin goals of cost containment and quality improvement.133 Competition requires competitors, of course, so the success of the ACA will depend in no small part on antitrust enforcement. Through the 1990s, advocates of market competition had cause for skepticism, as the courts did little to stand in the way of provider mergers that seemed to create substantial market power. At the same time, health insurers, especially Blues plans, have continued their decades-long dominance in many markets. The tide has seemingly turned in the past decade, with the FTC enjoying a string of

131 

Complaint at 7–8, United States v. Humana Inc., No. 1:12-cv-00464 (D.D.C. March 27, 2012), available at http://www.justice.gov/atr/cases/f281600/281618.pdf. The DOJ argued that traditional Medicare would not preclude a SSNIP by Medicare Advantage plans because “[m]‌ost successful Medicare Advantage plans, including those in the relevant geographic markets, offer substantially richer benefits at lower costs to enrollees than traditional Medicare does.” Id. at 7. 132  This problem arises frequently in unilateral effects cases, where competitors are differentiated and sharp lines are often hard to draw. Indeed, in the 2010 Horizontal Merger Guidelines, the agencies note that “measurement of market shares and market concentration is not an end in itself, but is useful to the extent it illuminates the merger’s likely competitive effects” and that “[s]‌ome of the analytical tools used by the Agencies to assess competitive effects do not rely on market definition” (US DOJ and FTC 2010, 7). The 2010 guidelines also highlight diversion analysis (on component of UPP analysis) as informative of postmerger unilateral incentives to increase price. 133  Natl. Fed’n of Indep. Bus. v. Sebelius, Nos. 11-393, 11-398, and 11-400, 2012 BL 160004 (U.S. June 28, 2012), available at http://www2.bloomberglaw.com/public/document/​ Natl_Federation_of_Independent_Business_v_Sebelius_No_11393_US_Ju.

Healthcare Provider and Payer Markets   109

successful challenges to hospital mergers, and with significant government and private actions against large Blues plans. Nearly all of the antitrust cases involving providers and payers must deal with the tension between potential pricing power and possible cognizable efficiencies. This tension is prominently on display in several important provisions in the ACA that create Accountable Care Organizations (ACOs) to serve Medicare enrollees (CMS and other observers expect that ACOs will also serve the privately insured). ACOs represent a reincarnation of sorts of 1990s Integrated Delivery Systems (IDSs)—vertically integrated organizations that bear risk for the totality of a patient’s health spending. IDSs largely failed to realize integration efficiencies, and many of them fell apart in the 2000s (Burns and Pauly 2002). ACOs, however, need not suffer the same fate. ACO managers may heed the lessons of failed IDSs, which by now are reasonably well understood. And ACOs will surely benefit from information technology in ways that were not available to IDSs. All of this does not mean that the way forward is clear, but there is some hope that ACOs may achieve efficiencies that eluded IDSs. Nobel Prize–winning economist George Stigler (1968) famously noted that competition among different organizational forms “sifts out the more efficient enterprises.”134 Applied to delivery reform, if the system fosters ongoing learning and improvement through experimentation, innovation, and emulation—while sifting out inefficient delivery models—real improvement will be likely. Consistent with this notion, the ACA provides only high-level descriptions of the organizing principles of ACOs: financial accountability, integration, and the use of information technology. The ACA leaves the design and operation of ACOs to ACOs, and a variety of organizational forms, including physician-led, hospital-led, and partnerships with payers, have already emerged. However, if the market is to sift out more efficient delivery models, meaningful competition is essential. With competition, efficient organizations will grow while the less efficient will either improve or fail. This highlights another important similarity between IDSs and ACOs: in addition to their promise, both also have the potential to enhance or entrench market power. Balancing potential efficiencies from integration against potential enhancements of market power will continue to be a challenge for state and federal antitrust agencies. The ACO provisions of the ACA reflect a view widely held by providers and shared by many policymakers that integration is essential for reducing healthcare spending. As we have discussed in this chapter, however, providers have often hid behind the cloak of efficiencies to justify mergers, when in fact mergers led to higher prices with few if any demonstrable cost savings. In part, this is because much of the actual consolidation has involved combinations of horizontal rivals rather than complementary producers. Although vertical integration in the healthcare sector has, to date, more often than not failed to achieve efficiencies, it has not been found to systematically enhance market power and increase prices in the ways that certain horizontal combinations have. In any 134  Stigler made this observation specifically about firms of different sizes. However, his underlying notion of the “Survivor Principle” applies to all aspects of organization.

110   Cory S. Capps and David Dranove case, the ACA, and the ACO movement more generally, could afford some provider systems a veil of procompetitive legitimacy as they seek to gain or use market power to capture rents and exclude rivals while avoiding the difficult job of creating value for patients. Compared with the current norm of fee-for-service reimbursement in which nearly all insured patients are profitable at the margin, accountability is hard work—as another Nobel Prize–winning economist once warned: “The best of all monopoly profits is a quiet life” (Hicks 1935). Effective antitrust enforcement that facilitates efficiency-enhancing integration while also protecting competition will preserve the incentives for payers and providers to undertake the hard work of becoming accountable.

References Adamache, Killard W., and Frank A. Sloan. 1983. “Competition between Non-Profit and For-Profit Health Insurers.” Journal of Health Economics 2(3): 225–43. American Bar Association Section of Antitrust Law. 2010. Antitrust Health Care Handbook. 4th ed. Chicago: ABA. American Medical Association. 2011. Competition in Health Insurance: A Comprehensive Study of U.S. Markets, 2011 Update. Chicago: American Medical Association. Antitrust Modernization Commission, 2007. Report and Recommendations. http://govinfo.​ library.unt.edu/amc/report_recommendation/amc_final_report.pdf. Areeda, Phillip E., and Herbert Hovenkamp. 2012. Antitrust Law 3rd ed. New York: Aspen. Austin, D.  Andrew, and Thomas L.  Hungerford. 2009. The Market Structure of the Health Insurance Industry. Washington, DC: Congressional Research Service. Binmore, Kenneth, Ariel Rubenstein, and Asher Wolinsky. 1986. “The Nash Bargaining Solution in Economic Modeling.” Rand Journal of Economics 17(2): 176–88. Birkmeyer, John, Cathryn Gust, Onur Baser, Justin Dimick, Jason Sutherland, and Jonathan Skinner. 2010. “Medicare Payments for Common Inpatient Procedures:  Implications for Episode-Based Payment Bundling.” Health Services Research 45(6): 1783–95. Blue Cross and Blue Shield Association. 2012. BlueCard Program. http://www.bcbs.com/​ already-a-member/coverage-home-and-away.html. Burns, Lawton R., and Mark V. Pauly. 2002. “Integrated Delivery Networks: A Detour on the Road to Integrated Health Care?” Health Affairs 21(4): 128–43. Capps, Cory. 2009. Comment, Project No. P092900, Submission of America’s Health Insurance Plans to the Department of Justice and the Federal Trade Commission on the Horizontal Merger Guidelines Review Project. http://www.ftc.gov/os/comments/horizontalmerger​ guides/545095-00009.pdf. Capps, Cory. 2010. “Buyer Power in Health Plan Mergers.” Journal of Competition Law and Economics 6(2): 375–91. Capps, Cory. 2014. “From Rockford to Joplin and Back Again:  The Impact of Economics on Hospital Merger Enforcement.” Working paper, Washington, DC. Capps, Cory, Dennis Carlton, and Guy David. 2010. “Antitrust Treatment of Nonprofits: Should Hospitals Receive Special Care?” Stigler Center for the Study of the Economy and the State Working Paper no. 232.

Healthcare Provider and Payer Markets   111

Capps, Cory, and David Dranove. 2004. “Hospital Consolidation and Negotiated PPO Prices.” Health Affairs 23(2): 175–181. Capps, Cory, and David Dranove. 2011. “Market Concentration of Hospitals.” Presentation, America’s Health Insurance Plans. http://www.ahipcoverage.com/wp-content/​uploads/ 2011/06/ACOs-Cory-Capps-Hospital-Market-Consolidation-Final.pdf. Capps, Cory, David Dranove, Shane Greenstein, and Mark Satterthwaite. 2002. “Antitrust Policy and Hospital Mergers:  Recommendations for a New Approach.” Antitrust Bulletin 47: 677–714. Capps, Cory, David Dranove, and Mark Satterthwaite. 2003. “Competition and Market Power in Option Demand Markets.” Rand Journal of Economics 34(4): 737–63. Capps, Cory, Marco Varkevisser, and Frederik T. Schut. 2008. “Defining Hospital Markets for Antitrust Enforcement: New Approaches and Their Applicability to the Netherlands.” Health Economics, Policy and Law 3(1): 7–29. Connor, Robert A., Roger D.  Feldman, and Bryan E.  Dowd. 1998. “The Effects of Market Concentration and Horizontal Mergers on Hospital Costs and Prices.” International Journal of the Economics of Business 5(2): 159–80. Cuellar, Allison E., and Paul J. Gertler. 2003. “Trends in Hospital Consolidation: The Formation of Local Systems.” Health Affairs 22(6): 77–87. Cunningham, Robert III, and Robert M. Cunningham Jr. 1997. The Blues: A History of the Blue Cross and Blue Shield System. DeKalb: Northern Illinois University Press. Czernin, F.  Maximilian. 2012. “Blue Cross Files Petition for Rehearing En Banc of Antitrust Immunity Appeal.” Lexology.com, March 9. http://www.lexology.com/library/detail.aspx?​ g=e633977b-e7c2-4bbd-9cae-5b7e37affd00. Dafny, Leemore. 2010. “Are Health Insurance Markets Competitive?” American Economic Review 100(4): 1399–1431. Dafny, Leemore, David Dranove, Frank Limbrock, and Fiona Scott-Morton. 2011. “Data Impediments to Empirical Work in Health Insurance Markets.” BE Journal of Economic Analysis & Policy 11(2): 1–22. Dafny, Leemore, Mark Duggan, and Subramaniam Ramanarayanan. 2012. “Paying a Premium on Your Premium? Consolidation in the US Health Insurance Industry.” American Economic Review 102(2): 1161–85. Danger, Kenneth L., and Harry E.  Frech. 2001. “Critical Thinking about ‘Critical Loss’ in Antitrust.” Antitrust Bulletin 46(2): 339–55. Donahue, Debra. 2012. “Blue Cross Blue Shield Entities Confronting Strong Competition.” Healthcare Business Strategy. http://www.markfarrah.com/healthcarebs.asp?article=108. Dranove, David, Anne Gron, and Michael Mazzeo. 2003. “Differentiation and Competition in HMO Markets.” Journal of Industrial Economics 51(4): 433–54. Dranove, David, and Richard Ludwick. 1999. “Competition and Pricing by Nonprofit Hospitals: A Reassessment of Lynk’s Analysis.” Journal of Health Economics 18(1): 87–98. Dranove, David, Subramaniam Ramanarayanan, and Yasutora Watanabe. 2012. “Delivering Bad News: Market Responses to Negligence.” Journal of Law and Economics 55(1): 1–25. Dranove, David, and Mark Satterthwaite. 2000. “The Industrial Organization of Health Care Markets.” In A. J.  Culyer and J. P.  Newhouse, Handbook of Health Economics. Amsterdam: Elsevier. Dranove, David, Mark Shanley, and Carol Simon. 1992. “Is Hospital Competition Wasteful?” Rand Journal of Economics 32(2): 247–62.

112   Cory S. Capps and David Dranove Dranove, David, Mark Shanley, and William D.  White. 1993. “Price and Concentration in Hospital Markets: The Switch from Patient-Driven to Payer-Driven Competition.” Journal of Law and Economics 36(1): 179–204. Elzinga, Kenneth G., and Thomas F.  Hogarty. 1973. “The Problem of Geographical Market Delineation in Antimerger Suits.” Antitrust Bulletin 18(45): 45–81. Elzinga, Kenneth G., and Thomas F.  Hogarty. 1978. “The Problem of Geographical Market Delineation Revisited: The Case of Coal.” Antitrust Bulletin 23(1): 1–18. Elzinga, Kenneth G., and Anthony W. Swisher. 2011. “Limits of the Elzinga-Hogarty Test in Hospital Mergers: The Evanston Case.” International Journal of the Economics of Business 18(1): 133–46. Farrell, Joseph, and Carl Shapiro. 2010. “Antitrust Evaluation of Horizontal Mergers: An Economic Alternative to Market Definition.” BE Journal of Theoretical Economics 10(1): Article 9. Feldman, Roger, and Douglas Wholey. 2001. “Do HMOs Have Monopsony Power?” International Journal of Health Care Finance and Economics 1(1): 7–22. Flaherty, Scott. 2012. “Aetna Beats Blue Cross Bid to Nix Mich. Antitrust Suit.” Law360. June 15. Frech, Harry E., III, and Paul B.  Ginsburg.1978. “Competition among Health Insurers.” In Warren Greenberg, Competition in the Health Care Sector, 167–87. Germantown, MD: Aspen Systems Corporation. Frech, Harry E. III, James Langenfeld, and Robert McClure. 2004. “Elzinga-Hogarty Tests and Alternative Approaches for Market Share Calculations in Hospital Markets.” Antitrust Law Journal 71(3): 921–47. Garmon, Chris. 2009. “Hospital Competition and Charity Care.” Forum for Health Economics & Policy 12(1): Article 2. Gaynor, Martin, Samuel A. Kleiner, and William Vogt. 2013. “A Structural Approach to Market Definition with an Application to the Hospital Industry.” Journal of Industrial Economics 61(2): 243–289. Gaynor, Martin, and Robert J. Town. 2011. “Competition in Health Care Markets.” In Handbook of Health Economics, vol. 2: 499–637. Amsterdam, Elsevier. Gaynor, Martin, and Robert J.  Town. 2012. The Impact of Hospital Consolidation—Update. Princeton, NJ: Robert Wood Johnson Foundation (The Synthesis Project). Gaynor, Martin, and William B. Vogt. 2003. “Competition among Hospitals.” Rand Journal of Economics 34(4): 764–85. Gowrisankaran, Gautam, Aviv Nevo, and Robert Town. 2013. “Mergers When Prices Are Negotiated: Evidence from the Hospital Industry.” Working paper, University of Arizona, Tucson, AZ. Haas-Wilson, Deborah, and Christopher Garmon. 2011. “Hospital Mergers and Competitive Effects:  Two Retrospective Analyses.” International Journal of the Economics of Business 18(1): 17–32. Hammer, Peter J., and William M. Sage. 2002. “Antitrust, Health Care Quality, and the Courts.” Columbia Law Review 102(3): 939–55. Handel, Benjamin. 2013. “Adverse Selection and Inertia in Health Insurance Markets:  When Nudging Hurts.” American Economic Review 103(7): 2643–2682. Health Insurance Association of America. 1970. Source Book of Health Insurance Data. New York: Health Ins. Inst. of America. Hicks, J. R. 1935. “The Theory of Monopoly.” Econometrica 3(1): 1–20. Ho, Katherine. 2006. “The Welfare Effects of Restricted Hospital Choice in the US Medical Care Market.” Journal of Applied Econometrics 21(7): 1039–79.

Healthcare Provider and Payer Markets   113

Horn, Henrik, and Asher Wolinsky. 1988. “Bilateral Monopolies and Incentives for Merger.” RAND Journal of Economics 19(3): 408–19. Hough, Douglas, David Gans, and Kai Liu. 2010. “Size Matters: The Impact of Physician Practice Size on Productivity.” Working paper, Johns Hopkins University, Baltimore, MD. Katz, Michael L., and Carl Shapiro. 2003. “Critical Loss: Let’s Tell the Whole Story.” Antitrust 17(2): 49–56. Keeler, Emmet, Glenn A.  Melnick, and Jack Zwanziger. 1999. “The Changing Effects of Competition on Non-profit and For-Profit Hospital Pricing Behavior.” Journal of Health Economics 18(1): 69–86. Kopit, William G., and Robert W. McCann. 1988. “Toward a Definitive Antitrust Standard for Nonprofit Hospital Mergers.” Journal of Health Politics, Policy and Law 20(1): 137–69. Krishnan, Ranjani. 2001. “Market Restructuring and Pricing in the Hospital Industry.” Journal of Health Economics 20(2): 213–37. Langenfeld, James, and Wenqing Li. 2001. “Critical Loss Analysis in Evaluating Mergers.” Antitrust Bulletin 46(2): 299–337. Leibenluft, Robert. 2010. Top Health Care Antitrust Developments for 2010. Unpublished manuscript. Washington, DC. Longstreth, Andrew. 2013. “Blue Cross and Blue Shield Antitrust Lawsuits Pile Up.” Chicago Tribune, February 20. Luft, Harold S., Susan C.  Maerki, James C.  Robinson, Deborah W.  Garnick, and Stephen J. McPhee. 1986. “The Role of Specialized Clinical Services in Competition among Hospitals.” Inquiry 23(1): 83–94. Lynk, William. 1995. “Nonprofit Hospital Mergers and the Exercise of Market Power.” Journal of Law and Economics 38(2): 437–61. Meier, Markus H. 2009. Letter from Markus H.  Meier, Assistant Director F.T.C. Bureau of Competition Health Care Division, to Christi J. Braun. Esquire, April 13. http://www.ftc.gov/​ os/closings/staff/090413tristateaoletter.pdf. Meier, Markus H. 2013. Letter from Markus H.  Meier, Assistant Director F.T.C. Bureau of Competition Health Care Division, to Michael E. Joseph. Esquire, February 13. http://www.​ ftc.gov/os/2013/02/130213normanphoadvltr.pdf. Melnick, Glenn A., Yu-Chu Shen, and Vivian Wu. 2011. “The Increased Concentration of Health Plan Markets Can Benefit Consumers through Lower Hospital Prices.” Health Affairs 30(9): 1728–33. Moriya, Asako S., William B.  Vogt, and Martin Gaynor. 2010. “Hospital Prices and Market Structure in the Hospital and Insurance Industries.” Health Economics, Policy and Law 5(4): 459–79. Newhouse, Joseph. 1970. “Toward a Theory of Nonprofit Institutions: An Economic Model of a Hospital.” American Economic Review 60(1): 64–74. Noether, Monica. 1988. “Competition among Hospitals.” Journal of Health Economics 7(3): 259–79. O’Brien, Daniel P., and Abraham L. Wickelgren. 2004. “A Critical Analysis of Critical Loss.” Antitrust Law Journal 71(1): 161–84. Pauly, Mark, and Michael Redisch. 1973. “The Not-For-Profit Hospital as a Physician’s Cooperative.” American Economic Review 63(1): 87–99. Philipson, Thomas J., and Richard A.  Posner. 2009. “Antitrust in the Not-for-Profit Sector.” Journal of Law and Economics 52(1): 1–18.

114   Cory S. Capps and David Dranove Robinson, James C. 1988. “Market Structure, Employment, and Skill Mix in the Hospital Industry.” Southern Economic Journal 55(2): 315–25. Robinson, James C., Deborah W. Garnick, and Stephen J. McPhee. 1987. “Market and Regulatory Influences on the Availability of Coronary Angioplasty and Bypass Surgery in U.S. Hospitals.” New England Journal of Medicine 317(2): 85–90. Robinson, James C., and Harold S.  Luft. 1985. “The Impact of Hospital Market Structure on Patient Volume, Average Length of Stay, and the Cost of Care.” Journal of Health Economics 4(4): 333–56. Sage, William M., David A. Hyman, and Warren Greenberg. 2003. “Why Competition Law Matters to Health Care Quality.” Health Affairs 22(2): 31–44. Savitt, Todd L. 2011. “Physician Price Fixing in 19th Century Virginia.” Blog. http://blog.hsl.vir​ ginia.edu/feebill/essays/. Schwartz, Marius. 1999. “Buyer Power Concerns and the Aetna-Prudential Merger.” Address, Northwestern University School of Law Fifth Annual Health Care Antitrust Forum, October 20. http://www.usdoj.gov/atr/public/speeches/3924.htm. Scott-Morton, Fiona. 1997a. “The Strategic Response by Pharmaceutical Firms to the Medicaid Most- Favored-Customer Rules.” Rand Journal of Economics 28(2): 269–90. Scott-Morton, Fiona. 1997b. “The Interaction between a Most-Favored-Customer Clause and Price Dispersion: An Empirical Examination of the Medicaid Rebate Rules of 1990.” Journal of Economics & Management Strategy 6(3): 157–74. Scott-Morton, Fiona. 2012. “Contracts that Reference Rivals.” Speech, Georgetown University Law Center, April 5. http://www.justice.gov/atr/public/speeches/281965.pdf. Sorensen, Alan. 2003. “Insurer-Hospital Bargaining: Negotiated Discounts in Post-deregulation Connecticut.” Journal of Industrial Economics 51(4): 469–90. Spang, Heather R., Richard J. Arnould, and Gloria J. Bazzoli. 2009. “The Effect of Non-rural Hospital Mergers and Acquisitions: An Examination of Cost and Price Outcomes.” Quarterly Review of Economics and Finance 49(2): 323–42. Stigler, George. 1968. The Organization of Industry. Chicago: University of Chicago Press. Tenn, Steven. 2011. “The Price Effects of Hospital Mergers: A Case Study of the Sutter-Summit Transaction.” International Journal of the Economics of Business 18(1): 65–82. Thomasson, Melissa A. 2003. “The Importance of Group Coverage: How Tax Policy Shaped U.S. Health Insurance.” American Economic Review 93(4): 1373–84. Thompson, Aileen. 2011. “The Effect of Hospital Mergers on Inpatient Prices: A Case Study of the New Hanover–Cape Fear Transaction.” International Journal of the Economics of Business 18(1): 91–101. Town, Robert J., and Gregory Vistnes. 2001. “Hospital Competition in HMO Networks.” Journal of Health Economics 20(5): 733–53. Town, Robert J., and William B. Vogt. 2006. “How Has Hospital Consolidation Affected the Price and Quality of Hospital Care?” Robert Wood Johnson Foundation Synthesis Project Research Report no. 9. http://www.rwjf.org/files/research/no9researchreport.pdf. US Centers for Medicare and Medicaid Services. 2010. National Health Expenditures:  2010 Highlights. http://www.cms.gov/Research-Statistics-Data-and-Systems/​Statistics-Trendsand-Reports/NationalHealthExpendData/Downloads/highlights.pdf. US Centers for Medicare and Medicaid Services (CMS). 2011. “Medicare Physician Group Practice Demonstration:  Physicians Groups Continue to Improve Quality and Generate Savings Under Medicare Physician Pay-for-Performance Demonstration.” http://www.cms.

Healthcare Provider and Payer Markets   115

gov/Medicare/​Demonstration-Projects/DemoProjectsEvalRpts/downloads/PGP_Fact_ Sheet.pdf. US Department of Justice. 2010a. “Department Alleges Agreements with Hospitals Stifle Competition, Resulting in Higher Health Insurance Prices for Michigan Consumers.” News release, October 18. http://www.justice.gov/atr/public/press_releases/2010/263227.htm. US Department of Justice. 2010b. “Decision to Abandon Deal Follows Justice Department’s Decision to Challenge the Acquisition.” News release, March 8. http://www.justice.gov/opa/​ pr/2010/March/10-at-234.html. US Department of Justice. 2011a. “Department Says United Regional’s Contracts Unlawfully Maintain Monopoly Power.” News release, February 25. http://www.justice.gov/opa/pr/2011/​ February/11-at-249.html. US Department of Justice. 2011b. “Divestiture and Additional Relief Prevent Harm from Anticompetitive Agreement between Blue Cross and Blue Shield of Montana and Five Hospital Owners of New West Health Services.” News release, November 8. http://www.justice.gov/​ atr/public/press_releases/2011/277133.htm. US Department of Justice. 2013. “Justice Department Files Motion to Dismiss Antitrust Lawsuit Against Blue Cross Blue Shield of Michigan after Michigan Passes Law to Prohibit Health Insurers from Using Most Favored Nation Clauses in Provider Contracts.” News release, March 25. http://www.justice.gov/opa/pr/2013/March/13-at-345.html. US Department of Justice and Federal Trade Commission. 1996. Statements of Antitrust Enforcement Policy in Health Care. http://www.justice.gov/atr/public/guidelines/hmg.​ htm. US Department of Justice and Federal Trade Commission. 1997. Horizontal Merger Guidelines. Issued April 2, 1992, revised April 8, 1997. http://www.justice.gov/atr/public/guidelines/​ hmg.htm. US Department of Justice and Federal Trade Commission. 2004. Improving Health Care: A Dose of Competition. Washington, DC: US Department of Justice and Federal Trade Commission. http://www.justice.gov/atr/public/health_care/204694.htm. US Department of Justice and Federal Trade Commission. 2010. Horizontal Merger Guidelines. Washington, DC: US Department of Justice. http://www.ftc.gov/os/2010/08/100819hmg.​pdf. US Federal Trade Commission. 2002. “Federal Trade Commission Announces Formation of Merger Litigation Task Force.” News release, August 28. http://www.ftc.gov/opa/2002/08/​ mergerlitigation.shtm. US Federal Trade Commission. 2007. “Advocate Health Partners, et al.; Analysis of Agreement Containing Consent Order to Aid Public Comment,” 72 Federal Register, no. 4 (January 8, 2007) at 784–787. US Federal Trade Commission. 2009. “Statement of the Bureau of Competition Director Richard Feinstein on the FTC’s Closure of Its Investigation of Consummated Hospital Merger in Temple, Texas.” December 23. http://www.ftc.gov/os/closings/091223scottwhitestmt.pdf. US Federal Trade Commission. 2011. “FTC Bureau of Competition Director Issues Statement on Providence Health & Services’ Abandonment of Its Plan to Acquire Spokane Cardiology and Heart Clinics Northwest.” News release, April 8. http://www.ftc.gov/opa/2011/04/provi​ dence.shtm. US Federal Trade Commission. 2012a. “FTC Order Will Restore Competition for Adult Cardiology Services in Reno, Nevada.” News release, August 6. http://ftc.gov/opa/2012/08/​ renownhealth.shtm.

116   Cory S. Capps and David Dranove Vistnes, Greg. 2000. “Hospitals, Mergers, and Two-Stage Competition.”Antitrust Law Journal 67(3): 671–92. Vita, Michael G., and Seth Sacher. 2001. “The Competitive Effects of Not-for-Profit Hospital Mergers: A Case Study.” Journal of Industrial Economics 49(1): 63–84. Werden, Gregory. 1990. “The Limited Relevance of Patient Migration Data in Market Delineation for Hospital Merger Cases.” Journal of Health Economics 8(4): 363–76. Werden, Gregory. 1996. “A Robust Test for Consumer Welfare Enhancing Mergers among Sellers of Differentiated Products.” Journal of Industrial Economics 44(4): 409–13. Werden, Gregory, 2007. “Monopsony and the Sherman Act: Consumer Welfare in a New Light.” Antitrust Law Journal 74(3): 707–37. Zwanziger, Jack, and Glenn A. Melnick. 1988a. “The Effects of Hospital Competition and the Medicare PPS Program on Hospital Cost Behavior in California.” Journal of Health Economics 7(4): 301–20. Zwanziger, Jack, and Glenn A.  Melnick. 1988b. “Hospital Behavior under Competition and Cost-Containment Policies:  The California Experience, 1980 to 1985.” JAMA 260(18): 2669–75.

Cases Cited Aetna Inc. v. Blue Cross Blue Shield of Mich., No. 2:11-cv-15346 (E.D. Mich. 2012). Arizona v. Maricopa Cnty. Medical Soc’y, 457 U.S. 332 (1982). Ball Mem’l Hosp. v. Mutual Hosp. Ins., 784 F.2d 1325 (7th Cir. 1986). Brooke Grp. Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209 (1993). California v. Sutter Health Sys., 84 F. Supp. 2d 1057 (N.D. Cal.), aff ’d mem., 2000-1 Trade Cas. (CCH) U 87,665 (9th Cir. 2000), revised, 130 F. Supp. 2d 1109 (N.D. Cal. 2001). Cascade Health Solutions v. PeaceHealth, 515 F.3d 883 (9th Cir. 2008). Cerven v. Blue Cross and Blue Shield of N.C., No. 5:12-cv-00017 (W.D.N.C. filed Feb. 7, 2012). Confederated Tribes of Siletz Indians of Or. v. Weyerhaeuser Co., 411 F.3d 1030 (9th Cir. 2005). Copperweld Corp. v. Independence Tube Corp., 467 US 752 (1984). In re Adventist Health Sys./West, FTC Docket No. 9234, 117 FTC 263 (ALJ Dec. 9, 1992), 117 FTC 295 (F.T.C. Apr. 1, 1994). In re Advocate Health Partners, No. 031-0021 (F.T.C. Dec. 29, 2006), available at http://www.ftc.​ gov/os/caselist/0310021/061229do0310021.pdf (decision and order), http://www.ftc.gov/os/​ caselist/0310021/061229agree0310021.pdf (consent agreement). In re Evanston Northwestern Healthcare Corp., No. 9315 (F.T.C. Aug. 6, 2007), available at http://​ www.ftc.gov/os/adjpro/d9315/070806opinion.pdf. In re Inova Health Sys. Found., No. 9326 (F.T.C. June 17, 2008), available at http://www.ftc.gov/​ os/adjpro/d9326/080617orderdismisscmpt.pdf. In re Intel Corp., No. 9341 (F.T.C. Dec. 29, 2010), available at http://www.ftc.gov/os/adjpro/​ d9341/101102inteldo.pdf. In re North Carolina Bd. of Dental Examiners, No. 9343 (F.T.C. Dec. 7, 2011), available at http://​ www.ftc.gov/os/adjpro/d9343/111207ncdentalopinion.pdf, on appeal sub nom. North Carolina Bd. of Dental Examiners v. FTC, No. 5:11-cv-00049 (E.D.N.C. filed June 27, 2011), available at http://www.ftc.gov/os/caselist/0810137/110627ncbdnotice.pdf).

Healthcare Provider and Payer Markets   117

In re North Texas Specialty Physicians, FTC Docket No. 9312 (ALJ Nov. 15, 2004), available at http://www.ftc.gov/os/adjpro/d9312/041116initialdecision.pdf, aff ’d, No. 9312 (F.T.C. Nov. 29, 2005), available at http://www.ftc.gov/os/adjpro/d9312/051201finalorder.pdf, aff ’d in part and remanded sub nom. North Texas Specialty Physicians v. F.T.C., No. 06-60023 (5th Cir. May 14, 2008), available at http://www.ftc.gov/os/caselist/0660023/080516opinion.pdf, modified sub nom. In re North Texas Specialty Physicians, No. 9312 (F.T.C. Sept. 12, 2008), available at http://www.​ ftc.gov/os/adjpro/d9312/080912orderonremand.pdf). In re ProMedica Health Sys., No. 9346 (F.T.C. Mar. 28, 2012), aff ’d per curiam, No. 12-3583 (6th Cir. Apr. 22, 2014). Jefferson Parish Hosp. Dist. v. Hyde, 466 U.S. 2 (1984). Kartell v. Blue Shield of Mass. 749 F.2d 922 (1st Cir. 1984), cert. denied 471 U.S. 1029 (1985). LePage’s Inc. v. 3M, 324 F.3d 141 (3d Cir. 2003) (en banc), cert. denied, 124 S. Ct. 2932 (2004). Messner v. Northshore Univ. Healthsystem, 669 F.3d 802 (7th Cir. 2012). Natl. Fed’n of Indep. Bus. v.  Sebelius, Nos. 11-393, 11-398, and 11-400, 2012 BL 160004 (U.S. June 28, 2012), available at http://www2.bloomberglaw.com/public/document/​ Natl_Federation_of_Independent_Business_v_Sebelius_No_11393_US_Ju. Ocean State Physicians Health Plan v. Blue Cross and Blue Shield of R.I., 883 F.2d 1101 (1st Cir. 1989), cert. denied, 494 U.S. 1027 (1990). Ortho Diagnostic Sys., Inc. v.  Abbott Labs., Inc., 920 F.  Supp.  455, reargument denied, 926 F. Supp. 371 (S.D.N.Y. 1996). Peoria Day Surgery Ctr. v. OSF Healthcare System, No. 06-1236 (C.D. Ill. Dec. 30, 2009). Preferred Health Services, Inc., 139 F.T.C. 266 (2005) (consent order). Reazin v. Blue Cross & Blue Shield of Kan., 899 F.2d 951 (10th Cir.1990), cert denied, 497 U.S. 1005 (1990). Richards v. Blue Cross & Blue Shield of Ala., No. cv-01133 (N.D. Ala. filed Apr. 17. 2012). Ukiah Adventist Hosp. v. FTC, No. 93-70387 (9th Cir. May 18, 1994). United States v. Aetna Inc., No. cv 398 (N.D. Tex. filed June 21, 1999) available at http://www.​ justice.gov/atr/cases/f2500/2501.htm. United States v. Blue Cross Blue Shield of Mich., No. 2:10-cv-14155-DPH-MKM (E.D. Mich. filed Oct. 18, 2010), available at http://www.justice.gov/atr/cases/f263200/263235.pdf. United States v. Blue Cross and Blue Shield of Mont., No. 1:11-cv-00123 (D. Mont. filed Nov. 8, 2011) available at http://www.justice.gov/atr/cases/f277100/277177.pdf. United States v. Cargill, No. 1:99CV01875 (U.S.D.C. filed July 8, 1999), available at http://www.​ usdoj.gov/atr/cases/f2500/2552.pdf. United States v. Carilion Health Sys., 707 F. Supp. 840 (W.D. Va. 1989). United States v. Fed’n of Physicians and Dentists, No. 98-475 (D. Del. 2002), available at http://​ www.justice.gov/atr/cases/f200600/200654.htm. United States v. Fed’n of Physicians and Dentists, No. 1:05-cv-431 (S.D. Ohio filed June 24, 2005), available at http://www.justice.gov/atr/cases/indx26_b.htm. United States v. Long Island Jewish Med. Ctr., 983 F. Supp. 121 (E.D.N.Y. 1997). United States v. Mercy Health Serv., 902 F. Supp. 968 (N.D. Iowa 1995), vacated as moot, 107 F.3d 632 (8th Cir. 1997). United States v. Rockford Mem. Hosp., 717 F. Supp. 1251 (N.D. Ill. 1989), aff ’d, 898 F.2d 1278 (7th Cir.), cert. denied, 498 U.S. 920 (1990).

118   Cory S. Capps and David Dranove United States v. UnitedHealth Grp., No. 1:05CV02436, (D.D.C. May 23, 2006), available at http://​ www.justice.gov/atr/cases/f216400/216423.htm. United States v. UnitedHealth Grp., No. 1:08-cv-00322 (D.D.C. Feb. 25, 2008), available at http://​ www.justice.gov/atr/cases/f230400/230451.htm. United States v. United Reg’l Health Care Sys., No. 7:11-cv-00030 (N.D. Tex. Sept. 29, 2011). UPMC v. Highmark, Inc., No. 2:12-cv-00692-JFC (W.D. Pa. filed May 23, 2012). US Fed. Trade Comm’n v. Butterworth Health Corp., 946 F. Supp. 1285 (W.D. Mich. 1996), aff ’d per curiam, No. 96-2440 (6th Cir. July 8, 1997). US Fed. Trade Comm’n v. Freeman Hosp., 911 F. Supp. 1213 (W.D. Mo. 1995), aff ’d, 69 F.3d 260 (8th Cir. 1995). US Fed. Trade Comm’n v. OSF Healthcare Sys., No. 11-C-50344 (N.D. Ill. Apr. 5, 2012). US Fed. Trade Comm’n v. Phoebe Putney Health Sys., No. 1:11- CV-58 (M.D. Ga. June 27, 2011), rev’d and remanded, No. 11-1160 (U.S. Feb. 19, 2013). US Fed. Trade Comm’n v. St. Luke’s Health System, No. 1:12-cv-00560-BLW-REB (D. Idaho filed Mar. 12, 2013), available at http://www.ftc.gov/os/caselist/1210069/index.shtm. US Fed. Trade Comm’n v. Tenet Healthcare Corp., 17 F. Supp. 2d 937 (E.D. Mo. 1998), rev’d, 186 F.3d 1045 (8th Cir. 1999). US Fed. Trade Comm’n v. University Health, 1991-1 Trade Cases ¶69,400 (S.D. Ga.) and 1991-1 Trade Cases ¶69,444 (S.D. Ga.), rev’d, 938 F.2d 1206 (11th Cir. 1991). West Penn Allegheny Health System, Inc. v. UPMC, 627 F.3d 85 (3d Cir. 2010). Weyerhaeuser Co. v. Ross-Simmons Hardwood Lumber Co., 127 S. Ct. 1069 (2007). White Sands Health Care System, L.L.C., 139 F.T.C. 15 (2005) (consent order).

CHAPTER 5

I N T E R NAT IONA L A N T I T RU ST I N S T I T U T ION S OLIVER BUDZINSKI

5.1. Introduction The fundamental problem addressed in this chapter is simple in its structure: business companies compete increasingly on international or even worldwide markets, whereas the regulation and governance of this business behavior remain bound to territorially defined jurisdictions. In other words, the geographical area of competition often exceeds the geographical scope of competition rules and policy. In even other words, while competition among enterprises is globalizing, competition policy remains national.1 This basic problem entails three more specific research questions: (1) Do we need international antitrust institutions in correspondence to internationalizing competition among enterprises? And if the answer is in the affirmative, then (2) what is the design of effective and efficient international antitrust institutions? Eventually if we know about (2), then (3) how can effective and efficient international antitrust institutions be implemented? This chapter deals with question (1) in its second section and addresses questions (2)  and partly (3)  in its third and fourth sections, along with presenting the state of development of actual international antitrust institutions. In doing so, I focus on

1  Strictly speaking, this is incorrect since the European Competition Policy System represents an effective supranational competition policy regime. However, like other competition policy systems of big territories (e.g., the US one), it remains bound to a territorial jurisdiction that in many cases is geographically smaller than the extent of the markets in which companies compete. Therefore, and even though it is legally and politically incorrect, I will treat European competition policy as a “national” one throughout this chapter.

120   Oliver Budzinski the International Competition Network since it appears to play the most important role in the contemporary world of international antitrust institutions. Restrictions in space imply, however, that this focus comes at the expense of dealing more closely with the Organization for Economic Cooperation and Development (OECD) as a forum for cooperation in international antitrust.2 Furthermore, section 5.4 brings together the two preceding sections by contrasting the current state of international competition policy with the needs of effective and efficient international antitrust institutions that are developed in the second section. In doing so, it derives an agenda of open problems that put up a challenge for the existing international antitrust institutions regarding their future development. Moreover, it discusses a proposal from the economic literature that may serve to solve these open problems. Section 5.5 concludes. Analyzing international antitrust institutions represents a multidisciplinary task, comprising economics, legal sciences, and political sciences. Accordingly, contributors to this research come from all three disciplines, working with very differing methods and theories. Written by an economist, this chapter naturally takes an economic perspective; however, arguments, reasoning, and references from the other two disciplines are included and—hopefully—appropriately valued. Some of the central vocabulary in this chapter may receive different definitions and usage in different disciplinary contexts. Therefore, let me make explicitly clear that antitrust and competition are used interchangeably regarding rules, policies, economics, or regimes. Furthermore, the term institution is used in an economics understanding, implying that institutions are understood to be an interrelated set of rules and can be distinguished from organizations: while organizations belong to the “players of the game,” institutions represent the “rules of the game” (North 1990; Vanberg 1994).

5.2.  The Economics of International Antitrust The economics of competition represent a vital, comprehensive, and also diversified research area encompassing several “schools” of thought and widespread conflicting opinions about various details of pro- and anticompetitive arrangements and business strategies. However, even in the absence of any unifying economic theory of competition,3 the vast majority of competition economics schools, scholars, and researchers 2 

Topics and activities of the OECD competition forum do not differ in drastic ways from the ICN. However, the ICN comprises even more country participants and appears to be a bit more ambitious in seeding international antitrust institutions. See on the impressive OECD activities http://www.oecd.org/​ topic/0,3699,en_2649_37463_1_1_1_1_37463,00.html (accessed July 1, 2012). 3  For the fundamental improbability of any unifying competition theory see Budzinski (2008b), who also presents an updated overview on the most popular schools of economic thought about competition and their main differences.

International Antitrust Institutions   121

agree on the necessity of having competition rules (prohibiting anticompetitive arrangements and business strategies)4 and an effective enforcement of these rules (competition policy combating anticompetitive arrangements and business strategies). Joining this mainstream position, naturally, represents a precondition for engaging in an economic analysis of the pros and cons of international antitrust institutions. However, accepting and advocating the existence of competition rules and policy on a national level of jurisdiction does not automatically imply advocating international antitrust institutions. The question whether national competition regimes can effectively and efficiently govern cross-border anticompetitive arrangements and business strategies is not a trivial one and, consequently, has attracted the attention of economic research. As a result, the problems of (1) cross-border externalities of national competition policies, (2) deficiencies of multiple procedures, (3) loopholes in the protection of competition, and (4) the diversity of competition philosophies and regimes around the world are discussed.

5.2.1.  International Externalities of National Competition Policies If companies compete on relevant markets that are geographically larger than the jurisdiction of competition policy regimes, incongruence between governance and governed actors occurs that is likely to produce cross-border externalities. If the relevant market— delineated according to the state-of-the-art methods in antitrust economics—includes several competition policy regimes, then each regime will deal with cases that (1) may be caused within and/or outside the jurisdiction and (2) may display anticompetitive effects within and/or outside the jurisdiction. While the first issue points at the problem of extraterritorial enforcement power (see section 5.3.1), the effects of the second issue are closely related to the goals of the competition policy regimes in question. The typical and predominant situation is that each regime pursues inbound-focused policy goals, like the protection of competition in domestic markets or the “national” (consumer) welfare. As a consequence, each jurisdiction focuses only on domestic effects, ignoring possible effects abroad. 4 

Much more than the question whether competition rules and policies are necessary at all, the different economic theories of and approaches to competition disagree on the exact borderline between procompetitive and anticompetitive arrangements and conduct. However and notwithstanding, anticompetitive arrangements and conduct usually encompass anticompetitive interbusiness cooperation (cartels), anticompetitive market concentration through mergers and acquisitions as well as business strategies abusing market power with predatory, squeezing, deterrence, handicap or foreclosure effects (unilateral anticompetitive strategies). Accordingly, anticartel policy, merger control, and abuse control are the three typical categories of competition policy. The term anticompetitive arrangements and conduct will be used in this chapter to encompass these three types of anticompetitive business phenomena. Furthermore, and sometimes somewhat neglected in economics, competition rules to combat grossly unfair behavior (espionage, sabotage, false and misleading advertising, consumer protection, etc.) exist in most jurisdictions.

122   Oliver Budzinski Given inbound-focused welfare goals, Barros and Cabral (1994) as well as Head and Ries (1997) demonstrate that an asymmetric allocation of producers and consumers among the jurisdictions of the relevant international market already suffices to cause negative externalities: the uncoordinated, rational decisions of the affected regimes are divergent and incompatible with each other. Due to the inbound focus of the decisions, negative effects on competition and/or welfare abroad are not taken into consideration. Even without any anticompetitive intentions on the side of the involved competition authorities, these negative externalities imply a reduction of welfare compared to a rational decision by a(n international) regime that fully encloses the geographical market in question (see also Sykes 1999; Kaiser and Vosgerau 2000; Tay and Willmann 2005; Haucap, Müller, and Wey 2006; Mehra 2008, 2011). Competition policy goals focusing on domestic welfare goals, furthermore, offer scope for strategic competition policies (Fox 2000; Guzman 2004; Budzinski 2008a, 53–64; Motta and Ruta 2011), which represent a variation to strategic trade policy and, thus, an element of so-called beggar-my-neighbor policies. The underlying idea is to increase domestic welfare by redirecting rents of foreigners into the jurisdiction, so that domestic welfare is increased at the expense of foreign welfare. Competition policy can be used in order to create so-called national champions, that is, domestic companies with market power on international markets. Those companies enjoy the ability to exploit consumers in other countries with the profits ending up “at home.” Instruments of a strategic competition policy may include a more permissive merger control toward domestic mergers than toward mergers between foreign companies, selective nonenforcement of anticartel rules toward home companies, antitrust exceptions for domestic key industries, or the allowance of pure export cartels (creating damages only abroad but securing supracompetitive profits for domestic companies). Strategic competition policies actively distort competition on international markets, thus intentionally producing negative externalities, reducing world welfare at the attempted benefit of single jurisdictions.5 Not only, but particularly, in the case of strategic competition policies (see the economic modeling of Motta and Ruta 2012), jurisdictional conflicts may result from diverging and mutually incompatible decisions of several national competition authorities on the same antitrust case in an international market. Famous examples include the European challenges of the US mergers Boeing–McDonnell Douglas (Fox 1998; Kovacic 2001) and GE-Honeywell (Reynolds and Ordover 2002; Gerber 2003). In both cases, negative externalities on trade relations and diplomatic relations between the two jurisdictions followed or were on the brink of happening. Furthermore, these two famous cases only represent the tip of the iceberg: Klodt (2005, 45–65) and Budzinski (2008a, 40–49)

5 

From an economic point of view, most real-world strategic competition policies are unlikely to achieve their goal (increase of domestic welfare). The rent-shifting reasoning often serves to conceal political-economic motives based upon vested interests and lobbyists’ influence on imperfect political decision processes (Kerber and Budzinski 2003, 2004).

International Antitrust Institutions   123

list an impressive number of jurisdictional conflicts on competition policy issues, involving a multitude of different countries (more recently also China and Russia).

5.2.2.  Deficiencies of Multiple Procedures Having markets with a geographical scope encompassing multiple noncoordinated competition policy regimes inevitably leads to multiple procedures. International mergers as well as companies with global strategies (à la Microsoft) often face competition policy procedures from (sometimes by far) more than twenty jurisdictions. Next to conflicting outcomes, these multiple procedures exert costs on business companies (transaction costs) and taxpayers (administrative costs). Being subject to x competition policy procedures implies the necessity to submit large-scale information in x different languages, acquire knowledge about x different legal systems, pay notification and other fees to x authorities, pay x times for legal assistance, and so on and so forth (overview: ICN 2002). These transaction costs easily amount to significant sums of millions of US dollars, making cross-border mergers and acquisitions, cooperation agreements and alliances, as well as unilateral strategies considerably more expensive—including the eventually procompetitive ones. To some extent, this discourages beneficial strategies, arrangements, and combinations of assets and, thus, harms global welfare (Evenett 2002; De Loecker et al. 2008). Moreover, a system of noncoordinated domestic competition policies puts a considerable burden on taxpayers. In the end, it is vastly the same facts that all these competition authorities are trying to investigate in an international case involving multiple procedures. Most of the doubling, tripling, and multiplying of investigation costs merely increases the “production costs” of competition policy without yielding additional “output,” insights, or benefits.

5.2.3.  Loopholes in the Protection of Competition A system of noncoordinated national competition policies with inbound-focused policy goals cannot provide any global welfare-optimal worldwide protection of competition if not all jurisdictions dispose over an effective competition policy regime. Although the number of countries without any competition regime has significantly decreased during the last two decades, this does not necessarily imply that all the new regimes (as well as some of the older ones) are also effective. As a consequence, loopholes arise, and they appear to be especially relevant where companies from jurisdictions with strong (but inbound-focused) competition regimes (often comparatively wealthy industrialized countries) restrict competition in jurisdictions with weak or without competition regimes (often comparatively poorer developing countries).6 6 

Jenny (2003a, 2003b) as well as Levenstein and Suslow (2004) offer an impressive set of examples.

124   Oliver Budzinski

5.2.4.  Diversity of Competition Regimes The existing competition regimes in the world differ, inter alia, regarding two dimensions. First, the goals of competition policy diverge in focusing on different welfare standards (consumer welfare vs. total welfare; allocative welfare vs. dynamic welfare; etc.), on different intermediate goals (different efficiency concepts, protection of the competitive process, liberty of competition, structure-conduct-performance concepts, etc.), and on the inclusion of “nonwelfare” goals (market integration, economic development, fairness, high employment, international competitiveness, etc.). Apart from the fact that the academic controversy about the “right” goal is not settled (inter alia, Fox 2003b; Foer 2005; Farrell and Katz 2006; Haucap, Müller, and Wey 2006; Heyer 2006; Kerber 2009; Vanberg 2011; Werden 2011), democratic societies may want to substitute some welfare in favor of other values, like fairness. As long as the principal explicitly takes the welfare trade-off into consideration, this choice is legitimate, although economists may remain skeptical toward the underlying value arguments. As a consequence, even in a world without political imperfections and deficiencies, national competition regimes cannot be expected to produce mutually homogenous goals. And if “one size does not fit all” (Evans 2009; Gal 2009; Fox 2012), then a diversity of competition regimes is beneficial in the sense of being conformal with the people’s preferences. Second, the absence of a unifying theory of competition (Budzinski 2008b) implies that different regimes will legitimately base their theories of competitive harm on diverging economic approaches. Consequently, the dividing line between procompetitive and anticompetitive arrangements and conduct will not be the same in each jurisdiction. Similarly, they will use different investigational and analytical methods. The resulting competition among competition regimes has been subject to several economic analyses (Easterbrook 1983; Hauser and Schoene 1994; Smets and Van Cayseele 1995; Sinn 1999; Fox 2000; Guzman 2001; Kerber and Budzinski 2003, 2004; Stephan 2004). While one should keep in mind that institutional competition is a difficult concept that next to similarities also bears important differences to goods market competition (Sinn 1990; Tirole 1994), there is an area where competition among competition regimes yields benefits. If many competition regimes experiment with different theories and methods (or even with differing goals), then this offers scope for dynamic institutional learning (dynamic and evolutionary efficiency). In case of diversity, competition regimes learn from their own successes and failures and, additionally, they learn from other regimes’ successes and failures. This mutual learning speeds up the learning process and offers potential to dynamically improve competition policy (Kerber and Budzinski 2003, 2004). Mutual learning through decentralized experimentation is particularly fruitful if there is no safe academic consensus on the “right” solution—as is the case with competition economics (Budzinski 2008b). However, the benefits of a system with decentralized elements do not imply that any type of coordination is necessarily eroding these learning effects.

International Antitrust Institutions   125

5.2.5. Summary In summary, the economic analysis of international antitrust demonstrates that a system of noncoordinated national competition policies with inbound-focused policy goals will not be optimal in terms of global welfare. Thus, there is scope for beneficial international antitrust institutions. At the same time, the economic analysis highlights that in the absence of perfect competition economic knowledge decentralized elements do possess merits. Altogether, the issues (1) cross-border externalities of national competition policies, (2) deficiencies of multiple procedures, (3) loopholes in the protection of competition, and (4) learning potential from the diversity of competition philosophies and regimes represent criteria against which types of international antitrust institutions can be assessed. While (1)–(3) are motivated by stationary efficiency considerations, (4) focuses on dynamic and evolutionary efficiency considerations.

5.3.  Strategies toward International Antitrust Institutions 5.3.1.  The Unilateral Strategy: Extraterritorial Enforcement through the Effects Doctrine This scenario stays very close to the system of noncoordinated and territorially bound competition policies that was underlying the economic analysis in section 5.2. However, the effects doctrine7 represents an extraterritorial enlargement of domestic competition policy competences by offering a possibility for countries to unilaterally react to anticompetitive arrangements and conduct caused abroad (inward competition restraints). According to this doctrine, a competition regime claims jurisdiction over any business arrangement or conduct that affects its domestic markets, irrespective of the location or the nationality of its participants (Griffin 1999; Fox 2003a). Even though the effects doctrine has been applied successfully in many cases, it cannot heal the shortcomings of noncoordinated national competition policies as outlined in sections 5.2.1–5.2.3. On the contrary, the effects doctrine reinforces the inbound focus of competition policy and, due to the larger (now also extraterritorial) scope of regime jurisdiction, it tends to further increase the problem of multiple procedures. Obviously, the effects doctrine also does not help to close down loopholes originating from nonexistent or ineffective competition regimes. It actually further amplifies power asymmetries even among the “really working” regimes. While regimes with big internal markets like the 7  The Effects Doctrine was first applied by the United States in the so-called Alcoa-Case in 1945 and has subsequently been implemented by most other countries. See also on the following and with comprehensive references Budzinski (2008a, 33–49, 168–73).

126   Oliver Budzinski United States or the EU have sufficient power to enforce their competition rules against foreign and multinational companies (by threatening to tax or block their domestic turnovers), regimes with smaller domestic markets or such from poorer countries (less important domestic markets) may face a powerless situation against foreign and multinational companies. The attempt of extraterritorial competition law enforcement may in such cases be countered by threats from the respective companies to stop supplying the markets of the country in question. The case with the externalities is a bit more complex. If every regime effectively applied the effects doctrine, then the ignorance of purely outbound anticompetitive arrangements and conduct (e.g., export cartels) would not matter anymore. Negative externalities would insofar be internalized as every jurisdiction would protect their domestic markets and, thus, effectively prevent inbound restrictions of competition. However, along with this effect, the scope for jurisdictional conflicts increases since the probability of incompatible decisions on cross-border cases increases. In a perfect enforcement world, conflicting decisions within an effects doctrine regime imply that the most restrictive decision prevails (Geradin 2009). Consequently, a complete internalization of negative externalities cannot be expected even with perfect effects doctrine enforcement. As the preceding paragraph already pointed out, however, the effects doctrine can only be effectively enforced by sufficiently powerful regimes in reality. Therefore, the theoretical potential for internalization is rather unlikely to materialize under real-world conditions. Quite to the contrary, power asymmetries between regimes may actually increase negative externalities when the effects doctrine is used as an instrument of strategic competition policy. In summary, the unilateral strategy via the effects doctrine is not appropriate to improve global welfare.

5.3.2.  The Cooperative Strategy: Bilateral Agreements and Comity Principles Starting to become widespread in the mid-1970s, bilateral cooperation agreements between competition policy regimes became very popular, with a heyday in the late 1990s (overview: Budzinski 2008a, 51). In the absence of more ambitious international antitrust institutions, these bilateral agreements—sometimes initiated to supplement and complement bilateral trade agreements—usually focused merely on cartel prosecution and were predominantly concluded between established competition policy regimes of comparatively industrialized countries (see for more comprehensive analyses, inter alia, Fullerton and Mazard 2001; Zanettin 2002; Jenny 2003b). The contents of bilateral cooperation agreements can be categorized in the following way: • Notification:  competition authorities inform each other about ongoing antitrust procedures and exchange very general information, but their investigations, proceedings, and decisions remain completely autonomous.

International Antitrust Institutions   127

• Consultation: competition authorities exchange more detailed information, particularly regarding technical- issues (market definition, case facts, etc.) on a completely voluntary and discretionary basis. It remains within the discretion of the national agencies whether they pay respect to the interests of other countries in their independent decision. • Mutual Assistance:  the cooperating regimes mutually assist each other regarding information gathering and/or sanctioning in order to overcome the problems of extraterritorial investigation and enforcement. However, mutual assistance only includes cases in which the laws of the cooperating jurisdictions produce mutually compatible assessments, investigational procedures, and sanctions. • Negative Comity: following this traditional comity principle implies abstaining from extraterritorial enforcement of the domestic competition rules if there is explicit resistance by the foreign regime (i.e., no hostile extraterritorial enforcement). Thus, with negative comity each competition regime agrees to respect serious interests and the sovereignty of the other jurisdictions. Almost all the bilateral cooperation agreements focused on these less ambitious types of cooperation; many included merely mutual notification and consultation. For several reasons, these types of bilateral cooperation represent an insufficient supplement to noncoordinated extraterritorial competition policy. Although they do alleviate the shortcomings of noncoordination, significant limitations remain. First, jurisdictional conflicts may be reduced by avoiding misunderstandings. However, genuine conflicts of decisions as well as serious conflicts of interests cannot be solved “just” by mutual notification and consultation. Second, the problems of multiple procedures are not tackled. Third, the asymmetry of power is not resolved. Developing countries were severely underrepresented in existing bilateral cooperation agreements,8 and the bargaining power of big industrialized jurisdictions may allow for discriminating agreements or the discriminating use of it (Fox 2000, 2003a; Jenny 2003a, 2003b). Fourth, and for the same reasons, loopholes are not effectively closed. Notwithstanding, the potential of cooperation should not be overlooked. For instance, the prosecution of hardcore cartels seems to represent a comparatively beneficial field for bilateral cooperation.9 With a view to leniency programs for whistleblowers in hardcore cartels, preliminary modeling hints that, under some special circumstances, limited information exchange between competition agencies may be optimal to destabilize cartels and deter their formation (Choi and Gerlach 2010). In other areas, in merger control, for instance, information exchange faces severe limitations because of 8 

This has been changing more recently. However, it is merely the big industrializing countries like Brazil, China, and so on and not in any way a comprehensive inclusion of the developing world. 9  The prosecution of the international vitamins cartels may represent an outstanding success story (First 2001a). However, the absence of industrial policy interests plays an important role. “Where no industrial policy at the source is concerned, cartel enforcement at the national level with deep cooperation of sister agencies is the success story (in progress) of international antitrust. Where industrial policy intervenes, however, the gap is great, as in the case of OPEC, marketing boards and commodity cartels” (Fox 2003a, 373).

128   Oliver Budzinski confidentiality rules, and the availability of extensive information gathered by one jurisdiction may encourage another to engage in a case just because it can—even if the nexus of this jurisdiction to the merger effects may be rather weak. This then contributes to further multiply the number of procedures. At the end of the day, even if there was an overall kaleidoscope of bilateral cooperation agreements, involving every antitrust jurisdiction in the world, this network would most probably be characterized by significant incoherence. In multijurisdictional cases, each jurisdiction would have to handle a multitude of bilateral duties, probably not consistent with each other since the inbound-focus of competition policy remains unchanged. Positive Comity:  if a domestic market is negatively affected by an anticompetitive arrangement or conduct originating from the partner’s jurisdiction, its antitrust authority can request its partner authority to take enforcement actions on behalf of the affected country. The cooperating authority applies its own antitrust laws, however, with a view to combat the outbound restrictions on competition, which affect the partner jurisdiction, while the affected jurisdiction relinquishes its own procedure and relies on its cooperation partner to protect its interests. This advanced and far-reaching comity principle is the only one within bilateral agreements that actually modifies the inbound-focused policy and adds a responsibility to protect competition abroad and, consequently, also international markets as a whole. In doing so, it provides the potential to internalize most negative externalities. However, it does so by relying on one competition regime to provide a positive externality for the cooperating regime (by targeting to increase its welfare, too), which may cause an incentive problem.10 Indeed, the positive comity principle in the U.S.-EU agreement did not succeed very well in the so-called Amadeus case.11 Furthermore, the fundamental problems of bilateralism still apply. Altogether, a cacophony of bilateral agreements cannot realistically be expected to work under real-world conditions. “It seems over-optimistic to imagine that a world-wide framework for competition policy could be built up piecemeal from a network of bilateral agreements. . . . [I]‌t would be virtually impossible to ensure that all the agreements were compatible with each other” (Meiklejohn 1999, 1247). The popularity of bilateral agreements ceased with the implementation of the International Competition Network in 2001 (see section 5.3.4). 10 

See, for an economic analysis of advanced comity principles, Budzinski (2008a, 165–66, 199–203). In 1997 the US Department of Justice formally requested the European Commission investigate allegedly anticompetitive practices by European airlines, discriminating against US enterprises. Specifically, the EU computer reservation system Amadeus was said to be acting in an exclusionary way, thus deterring the concurrent US system SABRE. However, the procedure seemed to proceed rather sluggishly and was hampered by deviating investigation techniques and evidentiary standards between the United States and the EU. The case was eventually settled in 2000 by private agreements between the airlines and their reservation system companies. See also Zanettin (2002, 188–89). 11 

International Antitrust Institutions   129

5.3.3.  The Multilateral Strategy: International Antitrust Institutions in Trade Agreements The basic idea to fight international anticompetitive arrangements and conduct on an international level has been so straightforward to the political sphere that as far back as in 1927, the League of Nations’ World Economic Conference in Geneva put the problem of international cartels on its agenda, discussing options for a coordinated international anticartel policy effort (Wells 2002, 10–11). This early initiative did not have any chance of success, however, since in the 1920s a consensus that hardcore cartels are detrimental to welfare and should be combated by antitrust policy was just about to form.12 Still, less than two decades later, the next attempt to establish multilateral antitrust institutions appeared on the agenda. This time, it was driven by the desire to create a coherent and comprehensive postwar world economic order, consisting of international institutions and organizations for the governance of (1)  the monetary system and international currency relations (International Monetary Fund; the World Bank Group), (2) public cross-border restrictions to competition, i.e., trade barriers (Havana Charter and International Trade Organization; in advance established in 1947 as the General Agreement on Tariffs and Trade, GATT), and (3) private cross-border restraints of competition (the 1948 Havana Charter; International Trade Organization). While the first two institutions were set into force while the window of opportunity due to the global catastrophe of World War II was open, the international antitrust institution-part missed out and was subsequently abandoned in 1953 due to a lack of ratification by leading members states (Wells 2002, 116–25). However, the idea of international antitrust institutions being a complement to trade liberalization rules remained virulent. The benefits of trade liberalization can only be reaped in a sustainable way if the competition-intensifying effects of opening up national markets for international competition (Budzinski 2008a, 27–32) are not counteracted by the emergence of cross-border anticompetitive arrangements and conduct, re-establishing the preliberalization noncompetitive equilibrium. Therefore, effective means against international cartels and against international market dominance need to accompany trade liberalization. This is in line with theoretical economic thinking (inter alia, Ross 1988; Feinberg 1991; Jacquemin 1995; Cadot, Grether, and de Melo 2000; Hamilton and Stiegert 2000; Gaudet and Kanouni 2004; Mehra 2011; rather contrasting: Hauser and Schoene 1994). Consequently, competition provisions somewhat survived on the agenda of the World Trading System and in some instances found their way into regional trade 12 

In many European countries of the late nineteenth and early twentieth centuries, cartels (including even horizontal price-fixing arrangements) were viewed as (1) a legitimate part of the freedom of business and (2) an effective instrument against the permanent threat of deflation (because of their price“stabilizing” effects), which was inherent to the contemporary gold standard currency system as soon as domestic economic growth met with deficits in the balance of trade. On the origins and development of competition law and policy in Europe see Gerber (1998).

130   Oliver Budzinski agreements, albeit predominantly in rather rudimental shape (Alvarez, Clarke, and Silva 2005; Cernat 2005; Evenett 2005). After the establishment of the World Trade Organization (1995, comprising GATT, the General Agreement on Trade in Services GATS, and the agreement on Trade-Related Intellectual Property Rights, TRIPS), international competition resurfaced on the agenda, leading to the adoption of WTO antitrust institutions in the Doha Declaration (2001). However, in the aftermath of the Cancún conflicts, centering predominantly on agricultural markets issues, the antitrust provisions were provisionally abandoned in 2004—and since then a reappearance has not looked likely. While the recurring attempts to establish multilateral competition rules can easily be motivated both by the shortcomings and limits of unilateral and bilateral approaches (see sections 5.4.1 and 5.4.2) and by the complementary nature of trade liberalization and protection of competition on international markets, the likewise recurring failures to actually establish international antitrust institutions have motivated additional economic research. From a game-theoretic perspective, negotiations on international antitrust institutions among sovereign nations resemble the characteristics of a prisoner’s dilemma game. Even if adopting international antitrust institutions would represent the world welfare optimum, the players may end up in an inferior equilibrium because it is individually rational to choose strategic competition policies (beggar-my-neighbor policies) in the absence of an effective institution. Due to the incentive structure, such an institution is notoriously difficult to establish outside specific “windows of opportunity”—at least in rather simplistic game-theoretical models (à la Budzinski 2003). More advanced models (building upon so-called supergames) allow for much more differentiated analyses that also display self-enforcing cooperation patterns (Cabral 2003, 2005). However, dynamic prisoner’s dilemma games show that cooperation is possible but not necessary and may take long to be successfully established.

5.3.4.  The Network Strategy: The ICN after Ten Years During the years when the hitherto last attempt to establish WTO competition rules failed, a new avenue toward international antitrust institutions surfaced. On its route a multilateral perspective was combined with a focus on voluntary cooperation among competition agencies, and within one decade the resulting network developed to become the most important international antitrust player in the world. There have been attempts to establish voluntary multilateral cooperation before. In 1967, the Organization for Economic Cooperation and Development (OECD) created a forum for its members to debate international competition issues and issue consensus-based recommendations on competition policy—with the latter goal being abandoned in the 1990s (Zanettin 2002, 53–57). Furthermore, in 1980, the United Nations Conference on Trade and Development (UNCTAD) adopted a so-called Restrictive Business Practices Code with the particular aim of protecting developing

International Antitrust Institutions   131

countries against inbound anticompetitive arrangements and conduct by powerful multinational enterprises. It attempted to ban, inter alia, price-fixing arrangements and other hardcore cartels, as well as boycotts. However, the comparatively ambitious code lacked enforceability (First 2003). At the end of the day, both initiatives failed to produce considerable effects regarding a satisfying level of protection of international competition (Wells 2002; First 2003). Based on the concept of a Global Competition Initiative developed by the International Competition Policy Advisory Committee to the U.S. Department of Justice (ICPAC 2000), fifteen national competition agencies (including the European Commission) established the International Competition Network (ICN) in October 2001 (Finckenstein 2003; Janow and Rill 2011). Membership of the ICN has now risen to 121 competition agencies from more than one hundred jurisdictions all around the world.13 Being a network of competition agencies and calling itself a virtual organization, the ICN neither is based on an international contract, nor has its own administrative staff or budget. The ICN is led by a steering group consisting of leading officials from member agencies with the board positions rotating among the members.14 Annual conferences of all member agencies with participation of different stakeholder groups represent the major “decision body.” The actual work is done in so-called working groups (WGs), which typically start out by reviewing and comparatively evaluating the current practices of the member agencies. They constitute themselves project-oriented and expire if the respective agenda has been finished. The general goal of the WGs is to develop best-practice recommendations that are subsequently consensually adopted by the annual conference. In addition to the substantive WGs, administrative working groups address problems of internal governance. Currently, the ICN consists of five substantive and two administrative WGs, which are overviewed in figures 5.1–5.6. The voluntariness of cooperation and the nonbinding character of all best-practice recommendations represent a fundamental principle and an important characteristic of the ICN. Still, the eventual goal of the ICN is about improving international competition governance. By promoting multilateral cooperation among competition agencies and by creating a common competition culture, convergence of national and regional competition policies, starting with procedural issues but aiming at substantive issues as well, is on the long-run agenda (ICN 2011; Mitchell 2011, 5).15

13  This includes several regional-but-international competition agencies like the European Commission or the CARICOM Competition Commission. The number of member agencies is higher than the number of participating countries because some jurisdictions have more than one competition agency. Figures according to http://www.internationalcompetitionnetwork.org/members/​ member-directory.aspx (accessed June 12, 2012). Motta (2012, 1) mentions 123 member agencies. 14  Currently (May 2014), Andreas Mundt (Federal Cartel Office of Germany) represents the chair of the ICN. From the establishment of the ICN on, the Canadian Competition Bureau has taken a leading responsibility for the administration. 15  See, for recent academic analyses of the ICN, inter alia, Fox (2009), Coppola (2011), Hollman and Kovacic (2011), Lugard (2011), and Sokol (2011).

132   Oliver Budzinski

Agency Effectiveness

Mission

Output

Identifying key elements of a wellfunctioning competition agency and good practices for strategy and planning, operations, and enforcement tools and procedures. Sharing of experience among ICN members and NGAs and developing and disseminate good practices for agency effectiveness

• Competition Agency Practice Manual (Strategic Planning and Prioritization & Effective Project Delivery) • Workshops and Teleseminars • Curriculum Project • Series I. Competition Policy Fundamentals (5 Modules) • Series II. Horizontal Restraints (1 Module) • Series III. Dominant Firm Conduct (1 Module) • Series IV. Mergers (1 Module)

Cartel

FIGURE  5.1  Agency

Ongoing Work • Continue to develop the Agency Practice Handbook (Effective Knowledge Management & Human Resources Management & proposals for ex-post evaluation as well as communication and accountability) • New work on Effectiveness of different types of Investigative Processes • Curriculum Project • Agency Effectiveness • Developing Agencies • Competition Advocacy • Unilateral Conduct, Horizontal Agreements

Effectiveness Working Group

Mission

Output

Ongoing Work

Addressing the challenges of anticartel enforcement across the entire range of ICN members and amongst the agencies with different levels of experiences. At the heart of antitrust enforcement is the battle against hardcore cartels directed at price fixing, bid rigging, market sharing and market allocations.

• Compilation of “Good Practices” from the Anti-Cartel Enforcement Manual of the ICN Cartel Working Group • Anti-Cartel Templates • Compilation of the world’s largest collection of cartel awareness and outreach materials. • Cartel Workshops annual conferences.

• Enforcer roundtable telephone discussions on cartel awareness & outreach efforts • New Anti-Cartel Enforcement Manual Chapter on Case Resolution • Coordinating the completion of the Anti-Cartel Enforcement Templates by additional ICN Members, update of old templates

FIGURE  5.2  Cartel

Working Group

During its first decade, the ICN has produced an impressive output of more than 10,000 pages of “virtual” paper. While the dozens of comparative analyses of worldwide existing practices and institutions regarding specific competition policy fields represent a valuable stock of knowledge, inter alia, also for competition economics, law, and policy researchers, the main institutional contribution of the ICN is represented by the consensually adopted best-practice recommendations as well as by enforcement manuals on various topics (ICN 2011). They include, for instance, the Recommended Practices for Merger Notification and Review Procedures, the Anti-Cartel Enforcement Manual, and the Market Studies Good Practices Handbook (see also figures 5.1–5.5). The question whether purely voluntary cooperation, resting on the publication of consensual best-practice recommendations, can actually be successful triggered theoretical and empirical economic research. Budzinski (2004a, 2004b) analyzed the economics of

Merger

Mission

Output

Promoting the adoption of best practices in the design and operation of merger review regimes in order to enchance the effectiveness of each jurisdiction’s merger review mechanisms; to facilitate procedural and substantive convergence; and to reduce the public and private time and cost of multijurisdictional merger reviews.

• Merger Guidelines Workbook • Recommended Practice for Merger Notification Procedures • Merger Remedies Report • Investigative Techniques Handbook • Model Confidentiality Waiver • Practical Guidance Materials • Workshops and Teleseminars

FIGURE  5.3  Merger

Unilateral Conduct

Mission

Working Group

• Recommended Practices on the assessment of dominance/ substantial market power • Recommended Practices on the application of unilateral conduct rules to state-created monopolies • Workshop and Teleseminars

FIGURE  5.4  Unilateral

Advocacy

• Project onn Economic Analysis in Merger Review • Work to promote familiarity, use, and implementation of the MWG’s work product • ICN Framework for merger review cooperation • International Merger Enforcement Cooperation. • Potential ICN Merger Workshop

Output

Examine challenges involved in analyzing unilateral conduct of dominant firms and firms with substantial market power, facilitate greater understanding of the issues involved in analyzing unilateral conduct, and promote greater convergence and sound enforcement of laws governing unilateral conduct

Mission

Ongoing Work

Ongoing Work • Continue to Develop a Unilateral Conduct Workbook • Discuss Recommended Practices or other Guidance for the analysis of unilateral conduct • Consider Further Work Identified in the Group’s Long Term Plan

Conduct Working Group

Output

Develop practical tools and • Best Practice Recommendations guidance, and facilitate experience • Recommendation on sharing among ICN member International Cooperation agencies, in order to improve the • Recommendation on effectiveness of ICN members’ Cooperating in combating activities in advocating the Hard Core Cartels adoption of competition principles in • Recommendation on government and promote the Competition Assessment of development of a competition culture Governmental Market within society. Regulation • Best Practices for the sharing of confidential information in cartel investigations • Best practices to help governments advocate procompetitive reform • Advocacy Toolkit Part 1 (Advocacy process and tools) • Teleseminars FIGURE  5.5  Advocacy

Working Group

Ongoing Work • Raising Awarness of the Benefits of Competition (the Benefits project) • Competition Advocacy Toolkit Part 2 (communications and public relations) • Market Studies Information Store and Market Studies Good Practice Handbook. • Competition Advocacy Experiency Sharing Teleseminars.

134   Oliver Budzinski Merger Cartel Unilacteral Conduct Agency Effectiveness Advocacy 2000

2001

Advocacy

2002

2003 2004

2005

2006

2007

2008

2009

2010

2011

2012

Telecom Antitrust Enforcement in Regulated Sectors Capacity Building/Competition Policy Implementation Current WG FIGURE  5.6  Timeline

Former WG

of Current and Former Working Groups

combing consensual best-practice recommendations with peer pressure. Even though it remains completely voluntary whether individual competition policy regimes bring their practices and institutions in line with the published ICN best-practice recommendations or not, the consensual character of the recommendations and their public availability creates peer pressure. Agencies that have agreed that a certain practice is the best one will face a loss of reputation if they stick to an inferior practice—even according to their own evaluation expressed in the consensually adopted ICN recommendation. Thus, the combination of published best-practice recommendations and peer pressure sets strong incentives to actually comply with the ICN recommendations on the regime level. Furthermore, it is in line with behavioral economic thinking that a systematic and cooperative discussion of competition policy matters among the competition agencies has the potential to harmonize views on competition and antitrust issues, thus, promoting the targeted common competition culture (Budzinski 2004a). Once this “cognitive” harmonization process has taken off, it can develop strong force. However, particularly in the early period considerable obstacles may impede this process altogether. Nonetheless, peer pressure through publication and transparency of superior antitrust practices, which have been consensually acknowledged as superior, should promote a widespread adoption of the ICN best-practice recommendations by the member authorities. This economic theory reasoning is supported by early empirical analyses, suggesting that ICN best-practice recommendations actually influence competition regime reforms and implementation processes in member jurisdictions (Rowley and Campbell 2005; Evenett and Hijzen 2006).

International Antitrust Institutions   135

5.4.  Challenges and Unsolved Problems: The Way Forward 5.4.1.  The Success Story of ICN Without any doubt, the ICN has managed many impressive achievements in its first decade—and more so than many experts were expecting. First of all, the combination of consensual best-practice recommendations and peer pressure through the publication of the recommendations has been effective in the sense that many countries cited the ICN recommendations as motivation and guideline for domestic reforms of antitrust institutions. Moreover, both scientific analysis (Rowley and Campbell 2005; Evenett and Hijzen 2006) and internal assessment (ICN 2011) confirm that many member jurisdictions indeed reformed their competition rules to be more in line with the ICN recommendations. Thus, there is a harmonization effect on national competition policy regimes through the ICN membership that has potential to reduce jurisdictional conflicts over antitrust issues as well as to decrease the volume and severity of negative externalities, albeit not to zero. Second, the ICN has been very successful in promoting the implementation of competition regimes in developing and transitory countries. The impressive rise in membership is partly due to the establishment of new competition policy regimes in previously antitrust-free jurisdictions, and the ICN played a considerable role in this process. Furthermore, the ICN comprehensively engaged in capacity building for agencies in newly established and also in previously defunct or ineffective competition policy regimes. This has contributed to reduce loopholes in the worldwide protection of competition, which were due to a lack of effective competition policy regimes, in particular in many developing and transitory countries (Sokol 2009). And the newly established regimes have to a large extent particularly used the ICN best-practice recommendations as a role model for their antitrust institutions. Third, the ICN has published compilations of current practices in member jurisdictions (inter alia, merger review including substantive assessment and prohibition standards, anticartel enforcement techniques, unilateral conduct, competition advocacy, etc.). In many cases, for instance in the case of the unilateral conduct compilation, the main function of the compilations is to highlight the differences among member jurisdictions. While not directly promoting harmonization, the resulting transparency serves to improve the mutual understanding of differing and potentially incompatible case decisions and, thus, may contribute to reducing conflicts over such decisions (“informed divergence”; Mitchell 2011, 6). Fourth, the ICN has produced handbooks, manuals, and toolkits on many down-toearth competition policy practices. They represent an important practical help for competition agency officials regarding the everyday handling of cases. Together with the curriculum project (see ­figure 5.1), they serve as materials for the training of agency staff

136   Oliver Budzinski and proved particularly useful to young agencies that lack long-standing experiences in how to deal with antitrust cases. Fifth, it is certainly a success story that the ICN managed to actually issue an impressive number of consensually adopted best-practice recommendations (see fi ­ gures 5.1– 5.5). This achievement alone exceeds the output of former multilateral cooperation attempts. It proved to be considerably supportive for the success of cooperation that competition agencies have been driving the process and negotiated the agreements— instead of governments and government officials. Even across jurisdictions, the interests of competition agencies are significantly more homogenous and consensus-suited than the interests of governments. Eventually, a rather informal effect is often cited by participants as representing the main benefit from the ICN: mutual experience-sharing and getting-to-know each other (ICN 2011; Mitchell 2011, 3). The strong working relationship developed through the face-to-face contact in ICN seminars, workshops, and conferences facilitates informal cooperation also outside the direct ICN scope.

5.4.2.  Limits of the ICN Approach? Notwithstanding the achievements, the fifth aspect, however, already hints at some inherent limits of the ICN approach to international antitrust institutions from an economic perspective. A closer look on the best-practice recommendations reveals that there are barely any recommendations on substantive issues. The recommendations that were possible in consensus among all the members predominantly refer to procedural issues like transparency of notification requirements, fees, timetables, and so on. One must not underestimate that this type of best-practice recommendation represents important progress in international antitrust for both interacting agencies and norm addressees (the companies). However, along with the lack of substantial convergence (substantive rules and standards, delineation between pro- and anticompetitive effects, theories of harm, assessment practices and policies, etc.), the potential of the ICN to internalize negative externalities from diverging and incompatible case decisions appears to be rather limited, and this limited scope has effectively been reaped in the first decade. Without consensus on more ambitious best-practice recommendations, diminishing returns on further “low controversial” recommendations must be expected for the second decade. With respect to the problem of negative externalities, the economic analysis identifies the inbound focus of competition policy, that is, the absence of an international welfare goal for national competition policy regimes, as a sufficient condition to create negative cross-border externalities (see section 5.2.1). This problem has not been addressed by the ICN so far. Furthermore, it appears to be rather unlikely that an institutional arrangement like the ICN can be capable of introducing a world welfare goal for national competition policy regimes. Since it is the very nature of the ICN to rely on consensus and voluntary participation and implementation, it cannot provide any

International Antitrust Institutions   137

binding, contractual agreement that in case of defection may be enforced in member jurisdictions. Thus, the only way would be to issue a best-practice recommendation on antitrust goals (world welfare) and hope for (1) a consensual adoption on an annual conference and (2)  voluntary compliance to the recommendation by the member jurisdictions. Since this typically refers to “hard” law, the members of the ICN—competition agencies—would not be able to implement that recommendation without support from the legislative chambers (e.g., parliaments) and executive institutions (e.g., government) in their jurisdictions. This might well represent a limit to the “soft” law approach of the ICN. Another problem of international competition governance—the deficiencies of multiple procedures (see section 5.2.2)—has been alleviated by the ICN only to a negligible extent. Due to the imperfect convergence of procedures through the adopted best-practice recommendations, the costs of multijurisdictional antitrust case handling have been decreased marginally. However, since there has been no reduction of the number of antitrust procedures in conjunction with, for instance, a multijurisdictional merger, the vast majority of transaction and administration cost burdens remain unchanged. In the end, there is still nothing remotely close to a one-stop shop. Ironically, the impressive increase in active competition policy regimes around the world has actually increased the number of jurisdictions that declare themselves competent for international and particularly intercontinental competition cases. This in turn increases the deficiencies of multiple procedures and most probably more than compensates for the cost improvement due to soft and imperfect procedural harmonization. With the ICN as it is now, it is difficult to see how the second decade can bring significant improvements. The ICN does not entail direct case-related cooperation, but exactly this is necessary if considerable efficiency gains from international antitrust institutions are to be realized. Even though the ICN indirectly facilitates case-related cooperation because the member agencies and their staffs know each other and know whom to call for informal exchange and cooperation over a given case (ICN 2011; Mitchell 2011), this grass-roots effect—which without any doubt is highly important and helpful for everyday work—remains rather limited in the absence of an institutionalized case-related cooperation. The loopholes in the worldwide landscape of competition regimes (see section 5.2.3) have been substantially reduced by the ICN’s activities. Next to the impressive increase in newly established competition regimes, the ICN has also been very active in arming previously rather ineffective competition regimes. However, there has been virtually no change in a particularly problematic area, which is the power asymmetry when it comes to enforcing domestic competition rules against multinational companies by means of the effects doctrine (see section 5.3.1). If domestic markets are not sufficiently important for the business of the multinational, then the multinational remains in a position to avoid compliance by boycotting the respective country. The threat of this alone influences the decisions of smaller and less powerful regimes. Again, the regime of the uncoordinated effects doctrine can only be overcome by (1) replacing inbound competition policy goals with international welfare standards

138   Oliver Budzinski and (2) adopting a case-related cooperation approach. As has been argued in the preceding paragraphs, both seem to be difficult to achieve with an ICN of the current nature and structure. The fourth criterion to assess international antitrust institutions from an economic perspective (as derived in section 5.2) is the diversity of regimes reflecting the diversity and the provisional nature of economic thinking on competition. It refers to the dynamic and evolutionary efficiency of international antitrust institutions. The ICN highlights this by systematically reviewing the different practices in the member jurisdictions, and its compilations of the differences create transparency that serves to speed up mutual learning processes. Actually, the ICN best-practice recommendations represent the result of such a learning process. However, this is exactly where problems kick in:  with a best-practice result that leads to all member jurisdiction harmonizing their regimes according to this result, the dynamic learning process comes to an end. This implies no more future learning due to a lack of experiments with new insights and new methods, theories, and so on. Thus, the provisional economic knowledge of the time of the best-practice recommendation becomes a persistent standard, and scientific progress of the future will find it much more difficult to enter the stage.16 If learning from diversity is useful for finding today’s best practices, then learning from diversity will also be useful to detect future’s best practices. Consequently, three hazards are incorporated in the ICN’s harmonization approach. First, the identification of best practices to some extent relies on and promotes academically controversial practices (like the case-by-case effects approach in merger control). Second, the injection of new scientific knowledge is deterred. Both hazards together may lead to a deficient harmonization. Third, the ICN best-practice approach implicitly assumes that there actually are one-size-fits-all benchmarks. However, best practices for long-industrialized countries’ competition regimes may differ from such for newly industrialized or developing or transitory countries’ ones. Of course, the reasoning in this paragraph must be qualified to the extent that it becomes only relevant when the ICN is unexpectedly successful in achieving substantive harmonization also. In summary, the first decade of the ICN must be hailed for bringing the most significant progress to global competition governance so far. However, from the viewpoint of global economic welfare, there are still a lot of challenges and unsolved problems, covering all the four criteria (international externalities, deficiencies from multiple procedures, loopholes, and regime diversity), that can be formulated from an economic perspective. Moreover, and even more seriously, it appears to be rather doubtful whether in its current form (purely voluntary cooperation, reliance on consensus and peer pressure), the ICN is well suited and well equipped to address the remaining issues.

16  It should be emphasized here that the process of scientific knowledge accumulation and progress is a permanent one and all social science knowledge must be provisional by the nature of the (creative, innovative, and intentional) object of research (Budzinski 2008b, 313–17).

International Antitrust Institutions   139

Ironically, the (unexpected) success of the ICN’s first decade may imply bad news for its second decade since the potentials have already been exploited, so that from now on diminishing returns of the network strategy must be expected.17

5.4.3.  A Way Forward? Toward a Multilevel Lead Jurisdiction Model So, how can international antitrust institutions be designed to embrace all four criteria with their conflicting incentives toward more centralization (internalizing externalities and reducing multiple procedures; stationary efficiency) on the one hand and preservation of regime diversity (dynamic and evolutionary efficiency through decentralization) on the other hand? The economic literature offers two interesting concepts to approach this balancing act. The first concept is the idea of a lead jurisdiction model (Campbell and Trebilcock 1993, 1997; Trebilcock and Iacobucci 2004). It extends the positive comity concept (see section 5.3.2) by allocating competence and responsibility for multijurisdictional competition cases to one of the affected regimes that subsequently handles and decides the case with a view to avoiding anticompetitive effects in the overall geographic market (i.e., in all affected jurisdictions) and by relying on the assistance of the other involved regimes.18 The second concept is the idea of multilevel governance (Kerber 2003) in which regimes on different vertical levels (regional, national, supranational) are interconnected with each other. In such a complex multilevel system of institutions, the design of competence allocation rules, managing the interfaces of the participating regimes, becomes particularly important. Economic

17 

When the ICN envisions enforcement cooperation to be an important topic for its second decade, for instance, it refers to “facilitating and promoting informal cooperation and exchange of nonconfidential materials, which may help to foster better inter-agency relations and indirectly promote future formal cooperation; developing tools to facilitate identification of agencies reviewing or investigating matters and contacts in those agencies; promoting the exchange of experience and identifying and disseminating practical tips relevant to cooperation through the ICN blog17 and webinar programs; developing advocacy materials on the value of cooperation; and creating ICN guidance, such as investigational checklists and/or model cooperation agreements or confidentiality provisions, for use by ICN members” (Mitchell 2011, 6). This ICN-typical list—while without doubt representing valuable topics—demonstrates the type of consensual output that the ICN is capable of and thus entails prospects and limits at the same time. Replacing national welfare goals by international welfare goals or introducing one-stop shops for multinational cases goes far beyond the ambition level expressed here. 18  Two subconcepts can be distinguished: in the voluntary lead jurisdiction model (Campbell and Trebilcock 1993, 1997), the lead jurisdiction proposes a decision and each of the affected regimes decides on its own whether to adopt this paragon decision or not, whereas in the mandatory lead jurisdiction model (Trebilcock and Iacobucci 2004; Budzinski 2008a, 166–68, 203–6) the decision of the lead jurisdiction binds the other affected jurisdictions. Note an important difference between the ICN-style network approach and the lead jurisdiction concept: while the ICN works on the level of guidelines and principles, the lead jurisdiction model works on the case level, assigning lead jurisdiction to one of the involved regimes on a case-by-case nature (according to predefined allocation criteria).

140   Oliver Budzinski analysis reveals that different competence allocation rules (such as the effects doctrine, interjurisdictional commerce clauses, turnover thresholds, nondiscrimination, principle of origin doctrine, and relevant markets rule, or x-plus rule) are more or less appropriate when it comes to specified horizontal or vertical regime interfaces (Budzinski 2008a, 151–217). With a view to the four economic problems of international antitrust (as derived in section 5.2), it represents an interesting step to combine these two concepts toward a multilevel lead jurisdiction model (Budzinski 2009, 2011). The advantage of adding the vertical multilevel dimension to the lead jurisdiction concept lies in the option to introduce a referee authority, monitoring and supervising the impartiality of the assigned lead jurisdictions and providing conflict resolution if necessary. Thus, the antitrust institutions on the global level are not about materially deciding cases. Instead, they allocate lead jurisdiction according to agreed-upon criteria on a case basis,19 monitor and supervise the lead jurisdiction in respect of its impartial treatment of anticompetitive effects in the overall relevant international market (irrespective where—in which jurisdiction—the effects display), and settle conflicts in case of affected jurisdictions allege that their domestic effects were disregarded by the lead jurisdiction. Consequently, “only” procedural competences are assigned to the global level, and all material and substantive decision competences remain on the level of the existing national and regional-supranational regimes. From an economic perspective, the charm of this concept is that it (1) replaces the inbound focus of existing competition policy regimes by a focus embracing all effects in the relevant geographic (international) market, (2) provides a one-stop shop for the norm addressees (thus avoiding deficient transaction and administration costs of multiple procedures), (3) closes many loopholes due to the lead jurisdiction being powerful and also providing protection of competition abroad, and (4) maintains diversity of competition regimes because each assigned lead jurisdiction handles and decides the case according to this regime’s antitrust rules and procedures, but with the explicit inclusion of cross-border effects. On the downside, it requires an international agreement on procedural rules (in particular, criteria for allocating case-specific lead jurisdiction as well as for monitoring and conflict resolution mechanisms) and willingness to accept (1) procedural decisions by the international level and (2) material decisions by the lead jurisdiction as long as all effects are treated impartially irrespective of their jurisdictional location. This certainly represents a higher hurdle for consensus than the ICN-style network cooperation, but certainly a lower hurdle than consensus on binding global competition rules within the WTO framework. And from an economic perspective, such a multilevel lead jurisdiction model appears to be welfare-superior to these alternatives. However, the concept of a multilevel lead jurisdiction model is 19  This only refers to multijurisdictional cases; cases that display effects merely within one of the existing competition policy regimes are solely subject to this regime’s jurisdiction. For possible delineation criteria as well as for possible criteria for selecting lead jurisdictions see Budzinski (2009, 377–81; 2011, 81–87), which also provide more details on the following discussion.

International Antitrust Institutions   141

far from being comprehensively researched. Furthermore, an interesting exploration would be whether such a model could develop from the contemporary ICN when it seriously strives to solve the economic problems of international antitrust in its second or third decade.

5.5. Conclusion The global governance of competition represents an important economic problem. Economic theory clearly shows that noncoordinated competition policies of regimes that are territorially smaller than the international markets on which business companies compete violate cross-border allocative efficiency and are deficient with respect to global welfare. At the same time, some diversity of antitrust institutions and policies promotes dynamic and evolutionary efficiency so that globally binding, worldwide homogenous competition rules do not represent a first-best solution—even when neglecting obvious agreement and implementation difficulties. Since 2001, the world of international antitrust institutions has been significantly influenced by the then-established International Competition Network. This multilateral forum for voluntary cooperation among competition agencies has been a success story in its first decade—by far exceeding most experts’ expectation. The ICN has considerably contributed to alleviate the negative economic effects from the previous, virtually noncoordinated world of international antitrust. However and notwithstanding, from an economic welfare point of view, considerable challenges and problems remain on the agenda. Whether the ICN in its current structure and nature has the potential to solve the remaining problems represents a decisive question for the future of international antitrust institutions. Despite the success story of its first decade, however, economic analysis justifies skepticism whether the contemporary ICN is up to the remaining challenges. In particular, a change from inbound-, national-welfare-focused competition policies to such pursuing supranational and suprajurisdictional welfare goals, as well as cooperation on concrete, specified cases, is necessary from an economic perspective. However, both topics are hardly compatible with the contemporary governance principles of the ICN. A way forward can be expected from the economic concept of a multilevel lead jurisdiction model that possesses the potential to balance allocative and dynamic efficiency. It remains an open question, though, whether such a model could evolve out of the ICN during the next decade(s).

Acknowledgments I thank Nadine Neute and Sharyn Zwerenz for invaluable editorial assistance.

142   Oliver Budzinski

References Alvarez, Anna María, Julian Clarke, and Veronica Silva. 2005. Lessons from the Negotiation and Enforcement of Competition Provisions in South-South and North-South RTAs. In Philippe Brusick, Ana María Alvarez, and Lucian Cernat, Competition Provisions in Regional Trade Agreements:  How to Assure Development Gains. New  York:  United Nations, pp. 125–161. Barros Pedro P., and Luís M. Cabral. 1994. Merger Policy in Open Economies. European Economic Review 38: 1041–1055. Budzinski, Oliver. 2003. Toward an International Governance of Transborder Mergers? Competition Networks and Institutions between Centralism and Decentralism. NYU Journal of International Law and Politics 36: 1–52. Budzinski, Oliver. 2004a. The International Competition Network: Prospects and Limits on the Road towards International Competition Governance. Competition and Change 8: 243–66. Budzinski, Oliver. 2004b. The International Competition Network as an International Merger Control Institution. In John Chen, International Institutions and Multinational Enterprises: Global Players, Global Markets. Cheltenham: Edward Elgar, pp. 64–90. Budzinski, Oliver. 2008a. The Governance of Global Competition: Competence Allocation in an International Multilevel Competition Policy System. Cheltenham: Edward Elgar. Budzinski, Oliver. 2008b. Monoculture versus Diversity in Competition Economics. Cambridge Journal of Economics 32: 295–324. Budzinski, Oliver. 2009. An International Multilevel Competition Policy System. International Economics and Economic Policy 6: 367–89. Budzinski, Oliver. 2011. Mehr-Ebenen Governance, Leitjurisdiktionskonzepte und globaler Wettbewerb. In Renate Ohr. Governance in der Wirtschaftspolitik. Berlin:  Duncker & Humblot, pp. 73–96. Cabral, Luís M. 2003. International Merger Policy Coordination. Japan and the World Economy 15: 21–30. Cabral, Luís M. 2005. An Equilibrium Approach to International Merger Policy. International Journal of Industrial Organization 23: 739–51. Cadot, Olivier, Jean-Marie Grether, and Jaime de Melo. 2000. Trade and Competition Policy: Where Do We Stand? Journal of World Trade 34: 1–20. Campbell, Neil, and Michael J.  Trebilcock. 1993. International Merger Review:  Problems of Multi-jurisdictional Conflict. In Erhard Kantzenbach, Hans-Eckart Scharrer, and Leonard Waverman. Competition Policy in an Interdependent World Economy. Baden-Baden: Nomos, pp. 129–163. Campbell, Neil, and Michael J. Trebilcock. 1997. Interjurisdictional Conflict in Merger Review. In Leonard Waverman, William S. Comanor, and Akira Goto, Competition Policy in the Global Economy. London: Routledge, pp. 89–126. Cernat, Lucian. 2005. Eager to Ink, but Ready to Act? RTA Proliferation and International Cooperation on Competition Policy. In Philippe Brusick, Ana María Alvarez, and Lucian Cernat, Competition Provisions in Regional Trade Agreements: How to Assure Development Gains. New York: United Nations, pp. 3–38. Choi, Jay Pil, and Heiko Gerlach. 2010. Global Cartels, Leniency Programs and International Antitrust Cooperation. CESIFO Working Paper No. 3005.

International Antitrust Institutions   143

Coppola, Maria. 2011. One Network’s Effect: The Rise and Future of the ICN. Concurrences 3: 222–29. De Loecker, Jan, Jozef Konings, and Patrick Van Cayseele. 2008. Merger Review: How Much of Industry Is Affected in an International Perspective? Journal of Industry, Competition and Trade 8: 1–19. Easterbrook, Frank H. 1983. Antitrust and the Economics of Federalism. Journal of Law and Economics 26: 23–50. Evans, David S. 2009. Why Different Jurisdictions Do Not (and Should Not) Adopt the Same Antitrust Rules. Chicago Journal of International Law 10: 161–88. Evenett, Simon J. 2002. How Much Have Merger Review Laws Reduced Cross-Border Mergers and Acquisitions? In J. William Rowley, International Merger Control:  Prescriptions for Convergence. London: International Bar Association, pp. 39–55. Evenett, Simon J. 2005. What Can We Really Learn from the Competition Provisions of RTAs, in Philippe Brusick, Ana María Alvarez, and Lucian Cernat, Competition Provisions in Regional Trade Agreements: How to Assure Development Gains. New York: United Nations, pp. 39–65. Evenett, Simon J., and Alexander Hijzen. 2006. Conformity with International Recommendations on Merger Reviews:  An Economic Perspective on “Soft Law.” Nottingham:  University of Nottingham Research Paper No. 04. Farrell, Joseph, and Michael L. Katz. 2006. The Economics of Welfare Standards in Antitrust. Competition Policy International 2: 3–28. Feinberg, Robert M. 1991. Antitrust Policy and International Trade Liberalization. World Competition 14: 13–19. First, Harry. 2001. The Vitamins Case:  Cartel Prosecutions and the Coming of International Competition Law. Antitrust Law Journal 68: 711–34. First, Harry. 2003. Evolving toward What? The Development of International Antitrust. In Josef Drexl, The Future of Transnational Antitrust: From Comparative to Common Competition Law. Berne: Staempfli/Kluwer, pp. 23–51. Foer, Albert. 2005. The Goals of Antitrust. AAI Working Paper No. 09. Fox, Eleanor M. 1998. Antitrust Regulation across National Borders: The United States of Boeing versus the European Union of Airbus. Brookings Review 16: 30–32. Fox, Eleanor M. 2000. Antitrust and Regulatory Federalism: Races Up, Down, and Sideways. New York University Law Review 75: 1781–807. Fox, Eleanor M. 2003a. Can We Solve the Antitrust Problems by Extraterritoriality and Cooperation? Sufficiency and Legitimacy. Antitrust Bulletin 48: 355–76. Fox, Eleanor M. 2003b. We Protect Competition, You Protect Competitors. World Competition 26: 149–65. Fox, Eleanor M. 2009. Linked-In: Antitrust and the Virtues of a Virtual Network. International Lawyer 43: 151–74. Fox, Eleanor M. 2012. Competition, Development and Regional Integration:  In Search of a Competition Law Fit for Developing Countries. In Josef Drexl, Mor Bakhoum, Eleanor M. Fox, Michal S. Gal, and David J. Gerber, Competition Policy and Regional Integration in Developing Countries. Cheltenham: Edward Elgar, pp. 273–290. Fullerton, Larry, and Camelia C. Mazard. 2001. International Antitrust Co-operation Agreements. World Competition 24: 405–23. Gal, Michal S. 2009. Antitrust in a Globalized Economy: The Unique Enforcement Challenges Faced by Small and Developing Jurisdictions. Fordham International Law Journal 33: 1–56.

144   Oliver Budzinski Gaudet, Gérard, and Rams Kanouni. 2004. Trade Liberalization and the Profitability of Domestic Mergers. Review of International Economics 12: 353–58. Geradin, Damien. 2009. The Perils of Antitrust Proliferation: The Globalization of Antitrust and the Risks of Overregulation of Competitive Behavior. Chicago Journal of International Law 10: 189–212. Gerber, David J. 1998. Law and Competition in Twentieth Century Europe: Protecting Prometheus. Oxford: Oxford University Press. Gerber, David J. 2003. The European Commission’s GE/Honeywell Decision:  US Responses and Their Implications. Journal of Competition Law (Zeitschrift für Wettbewerbsrecht) 1: 87–95. Griffin, Joseph P. 1999. Extraterritoriality in U.S. and EU Antitrust Enforcement. Antitrust Law Journal 67: 159–99. Guzman, Andrew T. 2001. Antitrust and International Regulatory Federalism. New  York University Law Review 76: 1142–63. Guzman, Andrew T. 2004. The Case for International Antitrust. In Richard A.  Epstein and Michael S. Greve, Competition Laws in Conflict: Antitrust Jurisdiction in the Global Economy. Washington, D.C.: AEI Press, pp. 99–125. Hamilton, Stephen F., and Kyle Stiegert. 2000. Vertical Coordination, Antitrust Law, and International Trade. Journal of Law and Economics 41: 143–156. Haucap, Justus, Florian Müller, and Christian Wey. 2006. How to Reduce Conflicts over International Antitrust. Conferences on New Political Economy 23: 307–43. Hauser, Heinz, and Rainer Schoene. 1994. Is There a Need for International Competition Rules? Aussenwirtschaft 49: 205–22. Head, Keith, and John Ries. 1997. International Mergers and Welfare under Decentralized Competition Policy. Canadian Journal of Economics 30: 1104–23. Heyer, Ken. 2006. Welfare Standards and Merger Analysis:  Why Not the Best? Competition Policy International 2: 29–54. Hollman, Hugh M., and William E. Kovacic. 2011. The International Competition Network: Its Past, Current and Future Role. Minnesota Journal of International Law 20: 274–323. International Competition Network (ICN). 2002. Report on the Costs and Burdens of Multijurisdictional Merger Review. Naples. http://www.internationalcompetitionnetwork.​ org/costburd.doc. Accessed January 15, 2004. International Competition Network (ICN). 2011. The ICN’s Vision for Its Second Decade. http://www.internationalcompetitionnetwork.org/uploads/library/doc755.pdf. Accessed September 30, 2011. International Competition Policy Advisory Committee (ICPAC). 2000. International Competition Policy Advisory Committee:  Final Report, Washington, DC:  Department of Justice, Antitrust Division. Jacquemin, Alexis. 1995. Towards an Internationalisation of Competition Policy. World Economy 18: 781–89. Janow, Merit E., and James F. Rill. 2011. The Origins of the ICN. In Paul Lugard, The International Competition Network at Ten. Antwerp: Intersentia, pp. 21–37. Jenny, Frédéric. 2003a. Competition Law and Policy:  Global Governance Issues. World Competition 26: 609–24. Jenny, Frédéric. 2003b. International Cooperation on Competition: Myth, Reality and Perspective. Antitrust Bulletin 48: 973–1003.

International Antitrust Institutions   145

Kaiser, Bruno, and Hans-Juergen Vosgerau. 2000. Global Harmonisation of National Competition Policies. In Hans-Juergen Vosgerau, Institutional Arrangements for Global Economic Integration. Houndmills: Palgrave Macmillan, pp. 35–59. Kerber, Wolfgang. 2003. International Multi-level System of Competition Laws:  Federalism in Antitrust. In Josef Drexl, The Future of Transnational Antitrust:  From Comparative to Common Competition Law. Bern: Staempfli/Kluwer, pp. 269–300. Kerber, Wolfgang. 2009. Should Competition Law Promote Efficiency? Some Reflections of an Economist on the Normative Foundations of Competition Law. In Josef Drexl, Laurence Idot, and Joel Moneger, Economic Theory and Competition Law. Cheltenham:  Edward Elgar, pp. 93–120. Kerber, Wolfgang, and Oliver Budzinski. 2003. Towards a Differentiated Analysis of Competition of Competition Laws. Journal of Competition Law 1: 411–48. Kerber, Wolfgang, and Oliver Budzinski. 2004. Competition of Competition Laws:  Mission Impossible? In Richard. A.  Epstein and Michael S.  Greve, Competition Laws in Conflict: Antitrust Jurisdiction in the Global Economy. Washington, DC: AEI Press, pp. 31–65. Klodt, Henning. 2005. Towards a Global Competition Order. Berlin: Liberal. Kovacic, William E. 2001. Transatlantic Turbulence: The Boeing-McDonnell Douglas Merger and International Competition Policy. Antitrust Law Journal 69: 805–74. Levenstein, Margaret C., and Valerie Y. Suslow. 2004. Contemporary International Cartels and Developing Countries: Economic Effects and Implications for Competition Policy. Antitrust Law Journal 71: 801–52. Lugard, Paul. 2011. The ICN at Ten: Origins, Accomplishments and Aspirations. In Paul Lugard, The International Competition Network at Ten. Antwerp: Intersentia, pp. 5–19. Mehra, Poonam. 2008. Conflict in Cross Border Mergers. Mumbai: Indira Ghandi Institute of Development Research WP-2008-030. Mehra, Poonam. 2011. Supranational Agency: A Solution for Conflict in International Mergers. Homo Oeconomicus 27: 545–74. Meiklejohn, Roderick. 1999. An International Competition Policy: Do We Need it? Is it Feasible? World Economy 22: 1233–49. Mitchell, Sophie. 2011. A Clear Vision for the ICN’s Second Decade. Antitrust Chronicle 7: 1–8. Motta, Eduardo Pérez. 2012. My Roadmap as ICN Chair. http://www.internationalcompetition​ network.org/uploads/library/doc792.pdf. Accessed June 13, 2012. Motta, Massimo, and Michele Ruta. 2011. Mergers and National Champions. In Oliver Falck, Christian Gollier, and Ludger Woessmann, Industrial Policy for National Champions, 91–117. Cambridge, MA: MIT Press. Motta, Massimo, and Michele Ruta. 2012. A Political Economy Model of Merger Policy in International Markets. Economica 79: 115–36. North, Douglass C. 1990. Institutions, Institutional Change and Economic Performance. Cambridge: Cambridge University Press. Reynolds, Robert J., and Janusz A.  Ordover. 2002. Archimedean Leveraging and the GE/ Honeywell Transaction. Antitrust Law Journal 70: 171–98. Ross, Thomas W. 1988. On the Price Effects of Mergers with Freer Trade. International Journal of Industrial Organization 6: 233–46. Rowley, J. William, and Neil A. Campbell. 2005. Implementation of the International Competition Network’s Recommended Practices for Merger Review:  Final Survey Report on Practices IV-VII. World Competition 28: 533–88.

146   Oliver Budzinski Sinn, Hans-Werner. 1990. The Limits to Competition Between Economic Regimes. Empirica 17: 3–14. Sinn, Hans-Werner. 1999. The Competition between Competition Rules. Working Paper No. 192. Munich: CESifo. Smets, Hilde, and Patrick Van Cayseele. 1995. Competing Merger Policies in a Common Agency Framework. International Review of Law and Economics 15: 425–41. Sokol, D. Daniel. 2009. The Future of International Antitrust and Improving Antitrust Agency Capacity. Northwestern University Law Review 103: 1081–96. Sokol, D. Daniel. 2011. The ICN in the Context of International Antitrust Institutions. In Paul Lugard, The International Competition Network at Ten, 149–61. Antwerp: Intersentia. Stephan, Paul B. 2004. Against International Cooperation. In Richard A. Epstein and Michael S.  Greve, Competition Laws in Conflict:  Antitrust Jurisdiction in the Global Economy. Washington, DC: AEI Press, pp. 66–98. Sykes, Alan O. 1999. Externalities in Open Economy Antitrust and Their Implications for International Competition Policy. Harvard Journal of Law & Public Policy 23: 89–96. Tay, Abigail, and Gerald Willmann,. 2005. Why (No) Global Competition Policy Is a Tough Choice. Quarterly Review of Economics and Finance 45: 312–24. Tirole, Jean. 1994. The Internal Organization of Government. Oxford Economic Papers 46: 1–29. Trebilcock, Michael J., and Edward Iacobucci. 2004. National Treatment and Extraterritoriality: Defining the Domains of Trade and Antitrust Policy. In Richard A. Epstein and Michael S.  Greve, Competition Laws in Conflict:  Antitrust Jurisdiction in the Global Economy. Washington, DC: AEI Press, pp. 152–176. Vanberg, Viktor J. 1994. Rules and Choice in Economics. London: Routledge. Vanberg, Viktor J. 2011. Consumer Welfare, Total Welfare and Economic Freedom:  On the Normative Foundations of Competition Policy. In Josef Drexl, Wolfgang Kerber, and Rupprecht Podszun, Competition Policy and the Economic Approach. Cheltenham: Edward Elgar, pp. 44–71. Von Finckenstein, Konrad. 2003. International Antitrust Policy and the International Competition Network. In Barry E. Hawk, Annual Proceedings of the Fordham Corporate Law Institute Conference on International Antitrust Law & Policy. New York: Juris, pp. 37–46. Wells, Wyatt. 2002. Antitrust and the Formation of the Postwar World. New York: Columbia University Press. Werden, Gregory J. 2011. Consumer Welfare and Competition Policy. In Josef Drexl, Wolfgang Kerber, and Rupprecht Podszun, Competition Policy and the Economic Approach. Cheltenham: Edward Elgar, pp. 11–43. Zanettin, Bruno. 2002. Cooperation between Antitrust Agencies at the International Level. Oxford: Hart.

CHAPTER 6

C OM P E T I T ION P OL IC Y I N P U B L IC C HOIC E P E R SP E C T I V E FRED S. MCCHESNEY, MICHAEL REKSULAK*, AND WILLIAM F. SHUGHART II

Competition is an important cornerstone of our nation’s economic foundation. Vigilant antitrust enforcement preserves and protects competition and delivers American consumers lower prices, higher quality goods, and more innovation. The Antitrust Division undertakes this vigilance using a measured approach that relies on sound competition and economic principles. We galvanize the tremendous skills of our lawyers and economists to evaluate each matter carefully, thoroughly, and in light of its particular facts. Sharis A. Pozen, Acting Assistant Attorney General, Antitrust Division, congressional testimony, December 7, 2011

6.1. Introduction More than a quarter-century has passed since Robert Tollison (1985, 906) set out to “review the state of the (small) art with respect to developing a positive public choice theory of antitrust and to illustrate the potential of this approach.” Not only was he successful in developing such a positive theory, but its potential has long since been proven in the only way that ought to count among academic scientists, namely its ability to explain outcomes observed in the real world. As George Stigler (1975, 140) succinctly put it: “The theory tells us to look, as precisely and carefully as we can, at who gains and who loses, and how much, when we seek to explain regulatory policy.” *  Any opinions, findings, and conclusions or recommendations expressed in this chapter are those of the authors and do not necessarily reflect the views of the Qatar National Research Fund.

148   Fred S. McChesney, Michael Reksulak, and William F. Shughart II The evidence accumulated since then has painted a picture in stark contrast to the sentiments expressed in this chapter’s epigraph. Far too often, rather than delivering to “American consumers lower prices, higher quality goods, and more innovation,” the antitrust laws, their enforcement, and the remedies adopted supposedly to restore competitive market conditions have not advanced those stated goals. The benign view of government as the independent arbiter of antitrust policies that serve the public interest has long since been confined to the dustbin of history. Posner (1974, 335) correctly described (and dismissed) the “public interest theory,” which holds that “regulation is supplied in response to the demand of the public for the correction of inefficient or inequitable market practices.” Subsequent analyses by public choice scholars and others added to Posner’s open-eyed view by showing that the demand for regulation and antitrust enforcement is indeed real. However, demand usually does not originate with “the public.” It can be traced to various special interest groups and to the self-interests of those involved in administering the antitrust medicine. The relevant actors include competitors and trade groups as well as bureaucrats, elected officials, judges, and economists themselves. The public choice approach to antitrust, as shown below, aims to explain outcomes of this struggle by various interests in the real world. That means not spending inordinate amounts of time and resources on the question of why antitrust policies so often fail to achieve their supposed objectives,1 a gap between theory and practice that Asch (1970) called the “Antitrust Dilemma.” Too frequently, conventional wisdom explains away the dilemma by ascribing it to failures on the part of the individuals involved to understand basic economic principles.2 Stigler (1975, 140) was right again when he labeled such “an explanation of a policy in terms of error or confusion [as] no explanation at all,” highlighting that “anything and everything is compatible with that ‘explanation.’ ” Rather, public choice inquires into the predictive power of models of self-interested individuals making optimal choices given the constraints (budgetary, informational, institutional) they face and the objectives they can be shown to pursue. In doing so, public choice theory has closed the behavioral system—to quote James Buchanan (1972, 11–23)—in the profession’s discussion of antitrust, too. That means utilizing the power of the economic model—including its ongoing refinements incorporating insights from the decision sciences and behavioral psychology showing the consequences for human action of uncertainty, loss aversion, and bounded rationality—to frame the discussion in terms of how decision-makers act in all types of markets, including political ones. Consequently, the fairy tale of the “public servant,” who maximizes her own utility while shopping in Apple’s online store, but is interested only in 1 

Shughart and Tollison (1991), for example, present evidence that both the Sherman Act and the Clayton Act increased the level of unemployment in the economy and, thereby, reduced social welfare. 2  DeLorme, Frame and Kamerschen (1997, 331) asked “whether the Sherman Act, once passed into law, had an effect on the trusts’ behavior.... These results suggest that the Sherman Act, once enacted, was ineffective or was unenforced in dealing with the trusts.” Shughart and Tollison (1991, 38) write: “On those occasions when antitrust policy fails to move the economy toward the competitive ideal (as it does quite often by many accounts), such failures are attributed to correctable errors.”

Competition Policy in Public Choice Perspective   149

enhancing the general welfare when engaged in bureaucratic bean-counting, has been discarded as a serious analytic model.3 Viewed through the lens of individual utility-maximization, the landscape of antitrust suddenly becomes much less black-or-white. It leads to an understanding, as expressed by Reksulak (2010, 425), “that neither the courts nor government antitrust agencies are equipped unambiguously to act in a manner that reduces the ‘deadweight loss’ of certain business practices and, thus, increases overall welfare in society.” Moreover, that one mountain, on which the antitrust Sisyphus labors, is suddenly replaced by a continuum of peaks and valleys in which numerous actors pursue their individual and/or group goals. In such a setting, the analysis does not become easier, but more nuanced. It permits the researcher to ask many more questions. In the general analytical framework developed by public choice scholars, antitrust is just another form of economic regulation, which countless students of the regulatory process have shown to be vulnerable to “capture” by the very industries regulators are appointed or elected to oversee, or by other groups having important stakes in regulatory decisions. In that sense, public choice analyses of antitrust began the search for a social science equivalent to a “grand unified theory” (GUT) for the physical sciences, supplying a model that could explain the obvious gaps between the stated purposes and actual effects of public policies across a broad spectrum of policy issues.4 While physicists are still doing a GUT check, exactly that is what has been done by countless public choice researchers over the last few decades, thus closing the behavioral model in more than one way. This chapter presents a snapshot of the results with respect to the origin of antitrust legislation, its enforcement, the parochial impetuses for bringing antitrust complaints, as well as the interaction of the various interests at play in this policy arena. Obviously, one chapter cannot aspire to survey the existing public-choice literature on antitrust and regulation comprehensively. Previous, book-length treatments of some of the topics discussed here can be found in Shughart (1990) and McChesney and Shughart (1995). Readers of this chapter might bear three thoughts in mind. First, much of what a public choice perspective has to say about antitrust specifically is an application of the general public-choice model of nonmarket decision-making. There are many ways in which public choice offers modes of analysis that, as a first approximation, would seem to illuminate the areas of antitrust discussed here, but that have not yet undergone close examination under the antitrust microscope. At the same time, 3 

All three authors of this chapter have, at various times, been federal employees. While we have certainly encountered many hard-working, loyal, patriotic, and upright fellow civil servants, it was undoubtedly observable that homo economicus’s tendency to maximize personal utility had a seat at the table in many a decision-making process, as is only human. Utility, in that sense, also includes the objective of increasing the power and prestige of the public office one holds (see, for example, Niskanen 1968 and 2001). 4  Although the public choice perspective on antitrust policy overlaps the Chicago School’s many contributions to the analysis of economic regulation, the two approaches are distinct (McChesney and Larkin-Wong 2013).

150   Fred S. McChesney, Michael Reksulak, and William F. Shughart II antitrust has furnished many specific applications of the more general model, as described below. Second, like much study by economists of government action, it is sometimes easier in principle than in practice to discuss theories and methods separated logically but intertwined realistically. There is an obvious conceptual distinction, for example, between the origins of a particular law at a given point in time and the way it then is enforced over the years. But in fact, understanding the initial reasons why politicians passed legislation furnishes at least clues, if not answers, as to how and why government agencies and private individuals later enforce it. Efforts to divide the discussion into different compartments in this chapter do not imply a lack of overlap among them. Finally, it is noteworthy that American antitrust has generated so much positive (nonnormative) analysis while so little is known of antitrust elsewhere. Antitrust—or competition policy—has its own set of laws and somewhat distinct institutional features in different countries. In the United States, the antitrust statutes are enforced by two separate federal agencies, the Antitrust Division of the US Department of Justice (DOJ) and the Federal Trade Commission (FTC), by 50 state attorneys general, and by private parties who have standing to sue. The Competition Commission, operating under that name since 1999, enforces antitrust law in the United Kingdom. In the rest of Europe, the European Commission enforces transnational law established by the EC Treaty, but individual nations also have their own competition authorities that sometimes work in tandem with the Commission but at other times function less cooperatively with it. Private rights of action are not part of European competition law, although there is considerable pressure to follow the American example so as to permit them. To varying degrees, countries outside North America and Europe have promulgated their own competition (antitrust) laws and established their own enforcement apparatuses. Some countries have no antitrust law at all. With such a rich menu of institutional possibilities, one might predict that public choice analysts would be eager to venture abroad to explore them. But the opposite is true: almost no public choice inquiry into antitrust (competition) policy outside the United States has been attempted, and certainly none of any prominence. The paucity of studies means that the benefits of applying public choice to foreign competition regimes would be relatively high. However, the costs of inquiring into foreign competition law using a public choice approach also are relatively high, starting with unfamiliar institutions and possibly different languages. That said, it is noteworthy that much of what constitutes foreign competition law, at least formally, would be familiar, as much of it is based on American law. In Europe, for example, Sections 81 and 82 of the EC Treaty, concerning collusion and monopolization, derive from Sections 1 and 2 of the Sherman Act.5 In appraising mergers, European officials are supposed to follow a set of guidelines virtually identical in main part to the Department of 5 

The European law uses the language of “abuse of dominant market position” rather than “restraint of trade,” however.

Competition Policy in Public Choice Perspective   151

Justice / Federal Trade Commission Merger Guidelines enforced in the United States. The costs of comparing the American and European experiences are admittedly made higher by the fact that the FTC does not follow its own guidelines (Coate and McChesney 1992).6 In sum, the studies of American antitrust summarized herein furnish a blueprint of possible inquiries that could be undertaken elsewhere in the world. Those inquiries would facilitate additional testing of the public choice approach discussed below.

6.2.  Origins of Antitrust Once a government cure is introduced, it stays for years or decades. Antitrust law to prevent or end monopoly is continued long after the monopoly has been ended by technical invention or other cures. Tullock, Seldon, and Brady (2005, ix)

The history of international competition law shows a clear trend, however, insofar as states embrace—sooner or later—some sort of antitrust law and an enforcement regime. The spirit animating the international growth of competition law derives from American history and experience. With the passage of the Sherman Act in 1890, the United States was the first to promulgate a national system of antitrust law. Antitrust was an American gift to the rest of the world, which many other nations over time have accepted. Antitrust law in the United States relies in large part on seven sections of three statutes: the Sherman Act (1890), the Clayton Act (1914), and the Federal Trade Commission Act (1914). The websites of both the Federal Trade Commission and the DOJ’s Antitrust Division, to this day, identify these laws as the legal basis for their activities.7 Major amendments to the 1914 laws were later introduced by the Robinson-Patman Act of 1936, the Celler-Kefauver Act (1950), the Hart-Scott-Rodino Act (1976), the Magnuson-Moss Warranty / Federal Trade Commission Improvement Act (1976) and the Antitrust Amendments Act of 1990.8 6  Two recent studies (Shan et al. 2012; Lin and Zhao 2012) attempt to extend the analysis of merger control to the practices in China. However, the short time-period of merger enforcement that has passed since the enactment of China’s Anti-Monopoly Law (AML) in 2008 does not yet seem to permit the drawing of firm conclusions. 7  The Antitrust Division states that “the federal government enforces three major federal antitrust laws” (http://www.justice.gov/atr/about/antitrust-laws.html; last accessed May 18, 2014). The Federal Trade Commission (FTC) specifically mentions that “the Federal Trade Commission Act and the Clayton Act, both passed by Congress in 1914, give the Commission authority to enforce the antitrust laws” (http://www.ftc.gov/bc/about.shtm; last accessed May 18, 2014). 8  In 2011, the DOJ’s Antitrust Division and the FTC announced modifications to the Hart-Scott-Rodino (HSR) premerger notification rules. Advertised as “changes to reduce the filing burden and streamline the form parties must file when seeking antitrust clearance of proposed mergers and acquisitions” (http://www.justice.gov/opa/pr/2011/July/11-at-888.html; last accessed May 18, 2014), the modifications were judged by practitioners to potentially increase the burden on filers through expanded information and documentation requests (see, for example, Fried Frank Antitrust & Competition Law Alert® No. 11-06-28 (http://bit.ly/HxSrIE; last accessed May 18, 2014) or Berg and Casas 2011). The DOJ also

152   Fred S. McChesney, Michael Reksulak, and William F. Shughart II For many decades, passage of the Sherman, Clayton, and Federal Trade Commission Acts were seen as clearly compatible with public policies intended to protect the economy from unfettered monopoly power and to prevent the exploitation of consumers by the charging of higher-than-competitive prices. This was true despite the fact that most economists of the time objected to the Sherman Act (Gordon 1963; DiLorenzo and High 1988). In his presidential address to the American Economic Association, Stigler (1982, 3) observed that “a careful student of the history of economics would have searched long and hard, on July 2 of 1890, the day the Sherman Act was signed by President Harrison, for any economist who had ever recommended the policy of actively combatting collusion or monopolization in the economy at large.” The lack of evidence in favor of antitrust as motivated by concerns for competition may reflect what many commentators at the time, and throughout its history, have considered as an alternative rationale for antitrust. Antitrust has been seen as a way to favor small business by adopting a “big is bad” model of antitrust enforcement, a persistent theme in cases brought under Section 2 of the Sherman Act that forbids “monopolization.” Use of the law to pursue large firms so as to favor smaller ones obviously has public choice implications, as discussed further below. Moreover, as recounted in Gordon (1963, 156), and pointed out by Lawrence Chamberlain (1946, 30), in introducing his bill, “Senator Sherman’s action was not in response to persistent public clamor. It did not reflect any generally recognized demand for legislative protection from monopoly.” Although not quoted by Gordon, Chamberlain continues on to emphasize that, “Indeed, Congress had received but a single petition on the subject—a memorial from the Kansas legislature. Trust regulation was not a burning issue.” The public interest theory of the antitrust laws, which is not actually a theory in any meaningful sense, was put under even closer scrutiny when more and more evidence regarding the puzzling effects of antitrust activity was uncovered. Kaysen and Turner (1959, 154), while championing it, found antitrust activity often to “do no positive good.” Stigler (1966, 225) studied the “quantitative notion of the effects of antitrust laws . . . in the past 75 years.” He appeared surprised by their apparent lack of significant impact on industrial concentration. Bork (1966, 7) reminded the courts that they were supposed to “distinguish between agreements or activities that increase wealth through efficiency and those that decrease it through restriction of output” with the goal being the “maximization of wealth or consumer want satisfaction.” Based on his analysis, Bork concluded that the outcomes of antitrust practice generally did not seem to conform to that ideal (see, also, Bork 1978). Posner (1976) combined a legal analysis of antitrust laws with economic sophistication and suggested that changes were needed to bring antitrust practice in line with what generally were assumed to be its objectives.9 A number of explanations have been put recently withdrew a report prepared under the administration of George W. Bush announcing procedural changes with respect to the enforcement of Section 2 of the Sherman Act; new merger guidelines were issued jointly by the DOJ and the FTC on August 19, 2010 (see Shughart and Thomas 2013). 9  As a result of his service on the American Bar Association’s Commission to Study the Federal Trade Commission, which issued a report in 1969 highly critical of the direction the FTC had taken since its

Competition Policy in Public Choice Perspective   153

forward for the passage of the Sherman Act. They are discussed in Shughart (1997, especially ­chapter 9) and include the above-mentioned consumer welfare hypothesis, the small-business hypothesis, and the bureaucratic incentives hypothesis. The consumer welfare hypothesis is difficult to square not only with the results of antitrust enforcement following the passage of the Sherman Act but also with the fact that many of the most concentrated industries had consistently produced larger outputs and lower prices for consumers during the 1880s and 1890s. DiLorenzo (1985, 80; emphasis in original) concludes that “it is usually assumed, without evidence, that the trusts were restricting output” and demonstrates that “prices in these industries were generally falling, not rising, even when compared to the declining general price level.” DiLorenzo (1984, 195) also suggests that antitrust efforts had “perverse effects,” treating “as wasteful or unproductive costs that must be incurred to have a well-functioning market process.” Only a few years ago, the question still had to be asked whether, evaluated at the macro level, antitrust policy improves consumer welfare; it does not (Crandall and Winston 2003). Grandy (1993, 373) is convinced that “the legislative history of the Sherman Act fails to support the consumer-welfare hypothesis and suggests that Congress focused its concern on producer behavior.” The small-business hypothesis proposed that Congress was animated by Jeffersonian sentiments, wanting to protect smaller, less efficient firms from competition from large business agglomerations (“trusts”) that were likely able to exploit economies of scale. Consequently, agricultural interests were often hypothesized to have pushed for the passage of the Sherman Act. It is noteworthy that the populist Granger and Alliance movements, which identified monopolies as the source of price pressures (lower prices for outputs, higher prices for inputs), were politically strong when America’s foray into national antitrust began. However, Stigler dismissed empirically any political explanation linking antitrust to other movements, concluding (1977, 2) that “to a believer in the rational behavior of political participants, the ‘agrarian distress’ explanation for the appearance of antitrust policy is a real puzzle.” Boudreaux and DiLorenzo (1993) suggest that a more powerful explanation for antitrust is found in trends in antitrust legislation in the several states that had enacted their own antitrust laws over roughly the decade before the Sherman Act was passed. They report strong evidence that protectionist motives rather than efficiency goals led to the passage of antitrust statutes by states located mainly in the Mississippi River Valley. “Rural cattlemen and butchers,” Boudreaux and DiLorenzo (1993, 93) conclude, “were especially eager to have statutes enacted that would thwart competition from the newly centralized meat processing facilities in Chicago.”10 establishment in 1914, and, from mid-1963 to mid-1965, as an assistant to Commissioner Philip Elman, who wrote a series of dissents arguing in effect that “much of what the Commission did was not worth doing” (Posner 2005, 761), Judge Posner once favored abolishing that agency. He has since recanted (Posner 2005). 10  Related studies include Boudreaux, DiLorenzo, and Parker (1995)—investigating precursors to the Sherman Act—and Libecap (1992)—tracing the impact of the rise of the Chicago meat packers, which took advantage of scale economies and the invention of refrigerated railroad cars to ship “dressed” beef and pork to markets on the East Coast at much lower prices than had prevailed beforehand.

154   Fred S. McChesney, Michael Reksulak, and William F. Shughart II The third hypothesis, that influence by the private antitrust bar and public servants tasked with enforcing antitrust statutes led to the passing of antitrust legislation, has not gained much traction. There is little evidence of pressure by these groups to have major antitrust laws introduced or enacted. Nevertheless, as shown below, considerable scholarship indicates that the personal interests of public servants are important inputs to subsequent antitrust enforcement activities of the government, once the statutes had been passed. In summary, corroborated by analyses of antitrust’s origins, the evidence points away from a public interest theory and towards an interest-group theory. Lande (1982, 68) argued “that Congress passed the antitrust laws to further economic objectives, but primarily objectives of a distributive rather than of an efficiency nature.”11 Bradley (1990, 742) writes that, “as applications of the law would demonstrate, the Sherman Act discouraged scale economies that promoted lower costs and prices, penalized successful market entrepreneurship, and rewarded the political entrepreneurship of less-efficient business rivals.” The Clayton Act, justified as an answer to a presumed ineffectiveness of the Sherman Act,12 has been found instead to illustrate the influence of interest groups at work. Benson, Greenhut, and Holcombe (1987, 810)  suggest that the FTC and Clayton Acts “can clearly be seen as a transfer of rights to big business in response to the larger-than-intended transfer away from big business that had been the result of the Sherman Act.” Shughart (1997, 326)  discusses the Clayton Act as a compromise between competing interests being championed by the Senate and by the House of Representatives and remarks that Section 7 of the act contained a “glaring loophole.”13 However, that loophole was not an inadvertent mistake; Ekelund, McDonald, and Tollison (1995, 274–75) conclude that it was “intended to transfer wealth among firms.” Ramírez and Eigen-Zucchi (2001, 158–59) “show that senators responded to interest groups when casting their votes. Senators from states with a high proportion of smaller firms were two times more likely to vote in favor of the Act than senators from states with a high proportion of larger firms.” A small number of studies have supplied evidence of interest-group influence on the enforcement of the law on mergers in Europe, but without adopting the public choice framework explicitly. Weir (1992) was to our knowledge the first scholar to do so. He analyzed—using a probit approach—the extent to which the then-named Monopolies and Mergers Commission (UK) was guided by the “public interest.” Weir (1992, 11  Kirkwood and Lande (2008, 192) do, however, seem to take this argument a step too far when they suggest that “the fundamental goal of antitrust law is to protect consumers” without, seemingly, considering the evidence that suggests otherwise. 12  In 1911, Chief Justice Edward D. White had limited the scope of the law’s “restraint of trade” language in his opinion announcing the breakup of Standard Oil (NJ) by defining an enforcement theory called the “rule of reason.” Only contracts that “unduly” or “unreasonably” restrained trade run afoul of the Sherman Act under this interpretation. For more discussions of this early application of the Sherman Act, see Reksulak et al. (2004), Reksulak and Shughart (2011), and Reksulak and Shughart (2012). 13  Section 7 of the Clayton Act originally applied to mergers consummated through the purchase of share capital (equities), allowing asset acquisitions (a close substitute) to proceed without opposition. The loophole was closed legislatively in 1950.

Competition Policy in Public Choice Perspective   155

33) concluded “that very few of the issues which make up the public interest, as outlined in the 1973 Fair Trading Act and expanded in the Guidelines to Merger Policy (1978), appear to influence the Commission” during the period of investigation, 1974–1990. Fifteen years later, Aktas, de Bodt, and Roll (2007, 1096) established the existence of a pattern in which “European regulatory intervention is more likely when European firms are harmed by increased competition.”14 Similarly, studying a sample of mergers in the European Union between 1990 and 2002, Duso, Neven, and Roller (2007, 459) conclude that “the commission’s decisions during the period 1990–2002 cannot be explained solely in terms of protecting consumer surplus. In other words, the institutional and political environment does matter.” Dinc and Erel (2011, 30), relying on a dataset including large corporate merger proposals in Europe between 1997 and 2006, report a continuing trend in which “governments where the target firms are located tend to oppose a foreign merger attempt while supporting domestic ones that create so-called ‘national champions,’ or companies that are deemed to be ‘too-big-to-be acquired.’ ” Overall, while it does not explain everything about antitrust’s beginnings, the interest-group explanation of the origins of antitrust laws possesses more explanatory and predictive power than any of the alternatives, especially with respect to hypothesized “public interest” or “consumer welfare” goals. Neither “the public” nor economic efficiency, in the view of contemporary economists, were likely to benefit from the new law. Furthermore, as the results of cases brought under the statute demonstrated early on, the stated objectives of this law were often far removed from the actual effects, even or especially when the prosecution was successful. The question, then, remains why that was so. That query focuses the observer’s attention on the matter of enforcement.

6.3. Enforcement How can economists fashion remedies for antitrust problems to increase their confidence that the remedies do not cause more harm than good? Carlton (2007, 155)

Reviewing the “attitudes of economists toward monopoly as a problem in public policy” during his Richard T. Ely Lecture, Stigler (1982, 1) mentioned “the coolness with which American economists greeted the Sherman Act” and suggested that it “rested more on disbelief in the efficacy of the Sherman Act than on hostility to its purpose.” If that was, indeed, the case, then much responsibility for the “failures” would have to be laid at the doorstep of bureaucratic and judicial enforcement of this and other major antitrust laws. 14 

On the other hand, Aktas, de Bodt, Delanghe, and Roll (2011) indicate that a 2002 decision by the Court of First Instance, criticizing the economic analysis advanced by the European Commission to block several acquisitions, along with subsequent regulatory reforms, seems to have reduced this kind of protectionism.

156   Fred S. McChesney, Michael Reksulak, and William F. Shughart II Shughart (1996, 217) identifies yet another culprit, concluding that “economists have in fact not had a positive influence on antitrust policy, but have instead actively contributed to its use as a way of subverting competitive market forces.” He presents evidence that many an “ ‘exotic’ antitrust theory” (1996, 219) can be explained by the desire of publicly employed economists eventually to pursue private sector employment prospects as antitrust experts. McChesney (1996, xv), additionally points out that the Supreme Court’s announcement of an “antitrust injury doctrine” has been “a boon to economists who hire on as experts in antitrust litigation. If there has ever been a case for which no economist could be found to testify whether the practice challenged was efficient, or whether the harm suffered was the sort of thing that the antitrust laws were designed to guard against, I have not heard of it.” Writing about institutional incentives faced by bureaucrats tasked with enforcing the nation’s antitrust regulations, Tullock (1987, 337) observes that “they have no motive . . . to seek simple solutions or solutions that reduce the scope” of antitrust activities.15 More recently, as analyzed in Shughart and Thomas (2013), enforcement activities in the United States under the Obama administration seem to be geared towards behavioral (or “conduct”) remedies, as opposed to structural remedies for antitrust law violations. The former are “difficult to enforce and also are vulnerable to incentive and information problems, which are the principal causes of ‘government failure’ ” (Shughart and Thomas 2013, 3). One example of this approach may be the recent case in which the Justice Department brought suit against Apple and various book publishers for violations of Section 1 of the Sherman Act (a price-fixing conspiracy). It also seems to fit the pattern of antitrust authorities doing another company’s bidding. Amazon, which had achieved a dominant position in the e-book market with an ingenious pricing strategy ($9.99), taking losses in order to subsidize the sales of its Kindle e-book reader, had seen its market share decline after Apple and a group of e-book publishers introduced a new sales model. The switch from the traditional “wholesale model” to an “agency model,” denying retailers like Amazon the freedom to set their own prices, resulted in an almost immediate increase in average end-user prices for e-books. Although the successful antitrust action in this case, which Apple still is contesting, plausibly could bring lower prices in the short run, it may also be true that it will serve only to facilitate a new spurt in the growth of Amazon’s market share, which may or may not be beneficial for consumers in the longer run (Shughart and Thomas 2013). Such general observations, applying the private-interest paradigm and interest-group analysis to actors in the antitrust arena (including lawyers, economists, judges, and legislators, among others) have by now sketched a consistent pattern. Whenever a so-called public interest argument has to hold its own against a “private interest” explanation for antitrust enforcement, the latter usually carries the day for the unbiased observer.16 15  Shughart and Tollison (1987, 278) find that “the antitrust laws have been quite ineffective in deterring subsequent violations, since almost one-fourth of FTC activity is devoted to repeat offenders.” 16  Baker and Shapiro (2007, 2008) express concern that the public choice approach to antitrust has resulted in merger enforcement being “too lax” (2008, 29) and they suggest “reinvigorat[ing] merger enforcement” (2008, 35). For a rejoinder, see Muris (2008).

Competition Policy in Public Choice Perspective   157

6.3.1.  Antitrust Bureaucracy and Congressional Oversight [L]‌obbyists with greater incomes and who work for larger departments exhibit a preference for agency lobbying . . . [M]ore seasoned lobbyists working at larger firms may tend to specialize in agency lobbying, while newer lobbyists (especially those who have recently come off Capitol Hill) and smaller firms concentrate on Congress. McKay (2011, 135)

Two contributions by Tollison (1983) and by McCormick and Tollison (1981) started empirical research attempting to address the much bemoaned “failure” of government actions intended to further a “public interest.” In their 1981 book, the authors laid out a research program to analyze politicians as suppliers of special-interest legislation. They examine how important wealth transfers often are in passing laws, what incentives face various interest groups that actively lobby legislators, and how such lobbying affects the gains that policymakers may make from trading legislative access for votes, campaign contributions, or both. They also asked how the institutional structures of parliaments and constitutional checks and balances between the legislative and the executive branches influence the behavior of self-interested decision-makers in such circumstances. Those questions boil down to an analysis of “politicians . . . as brokers of wealth changes in the polity” (McCormick and Tollison 1981, 123). Naturally, these insights can be applied to many subfields of economics. Writing while serving as director of the Bureau of Economics at the Federal Trade Commission, Tollison (1983, 220) recommended that “industrial organization scholars need to spend less time on normative discussion of antitrust policy and more time on the positive issues associated with the behavior of antitrust enforcers.” That challenge was taken up, and a flurry of studies soon appeared investigating, among other things, how congressional oversight affects FTC decision-making, what effect the dual enforcement of antitrust law in the United States has had, and whether FTC action impacted recidivism in accordance with public interest theories. For the most part, they concluded that the public interest was little served (e.g., Faith, Leavens, and Tollison 1982, Shughart and Tollison 1985).17 Politics has a statistically demonstrable role in FTC enforcement actions (Coate, Higgins, and McChesney 1990).18 17  Much of the early research that followed Tollison’s call for more attention to the positive economics of antitrust enforcement is collected in Mackay, Miller, and Yandle (1987). 18  Weingast and Moran (1983, 766) extended the Stigler-Peltzman insight by developing “a model of agency decision making based on the premise that agencies are controlled by the legislature.” By applying their model of identifying membership on critical congressional committees overseeing the FTC they concluded that the agency was “remarkably sensitive to changes in the sub-committee composition” (793). Muris (1986) criticized some of the strong conclusions reported in Weingast and Moran (1983) regarding congressional influences and suggested that the authors had overlooked “the importance of the agency’s staff, both in its ideological character and its career goals” (Muris 1986, 888). In a rejoinder, Weingast and Moran (1986) relied on data gathered by Muris (1986) to reiterate that political actors play important roles in determining agency actions. Of course, evidence for either channel of influence

158   Fred S. McChesney, Michael Reksulak, and William F. Shughart II Faith, Leavens, and Tollison (1982) established that members of powerful oversight committees in Congress were able to influence the outcomes of bureaucratic decision-making processes related to antitrust. The authors (342) identified a pattern in which “representation on certain committees is apparently valuable in antitrust proceedings,” especially with regard to actual action brought against alleged antitrust violators that were headquartered in locations with influential members on relevant congressional committees, finding that congressional representation yielded favorable treatment for constituents. Numerous follow-up investigations likewise demonstrated political influence over bureaucratic activity in many other regulatory realms (e.g., Weingast and Moran 1983; Shughart, Tollison, and Goff 1986; Higgins and McChesney 1986; Young, Reksulak, and Shughart 2001; Garrett, Marsh, and Marshall 2006; Young and Sobel 2013). The principal-agent relationship between bureaucratic agencies and congressional overseers would not necessarily be cause for concern, if indeed public interests also were served. However, geographically based political representation motivates action centered on local interests. In antitrust, as in other policy areas, all politics is local: the overall results at a national level often turn out to be very different from what any public interest pursuits may have demanded. Of course, whenever oversight may not be as tight, studies of bureaus have shown many incentives for public employees to reinterpret their mandates in light of personal interests. Tullock (1965) and Downs (1967) pointed out how the economic analysis of bureaucratic entities could be fruitful in understanding policy outcomes. Tullock (1965, 177) supplied an explanation for why an “over-expanded bureaucracy,” driven by the goal of commanding an ever-larger organizational structure, may lead to a vicious cycle of inefficiency, breeding further expansion and more inefficiency. Niskanen (e.g., 1968, 1971, 1975, 1991, 1994, 2001) asked what bureaucrats maximize. Answers to that question, as it turns out (see Kiewiet 1991, for an overview of empirical studies), include agency budgets, future employment opportunities, personal perquisites, and the relative size of one’s own agency. Once the suspect ideal of a robotic bureaucrat fulfilling mandates given by public-interest-driven legislators is seen to be unrealistic, the gates are thrown open to various ways in which bureaucrats may interpret their mandates while maximizing their own utility (see, also, Wilson 1989 and Horn 1995). Makris (2003, 3) provides an overview of how Congress’s limited oversight resources generally can lead to more and more delegation to the agency and to “slippage.” He writes (2003, 1) that the “agency is likely to have superior information over its production technology relative to the legislature.” Often, bureaucratic activity can best be explained by a utility-maximization model that takes into account the incentives and constraints faced by an agency’s individual employees.19 merely reinforces the argument in this chapter, namely that the nature of institutional constraints as well as the ways in which utility-maximization by members of congressional oversight committees and by staff determine the outcomes of regulatory processes; no matter how loud is the “public welfare” oratory emanating from either Congress or the antitrust bureaucracy. 19  Dewatripont, Jewitt, and Tirole (1999) argue that professionalization and the specialization of officials may make them less inclined to deviate from their official mandates.

Competition Policy in Public Choice Perspective   159

This point applies to antitrust enforcement agencies specifically. Shughart (2000, 284) notes that “the number of economists and support staffers employed by a private consulting firm that specializes in antitrust” rose dramatically after the Reagan administration. Thus, “big antitrust” enforcement activity was back on the agenda, soon after a concerted effort had been undertaken to limit it. There is also evidence (Shughart 1990, 98) that government antitrust lawyers are more likely to press for cases to litigate than are the staff economists at the DOJ and the FTC. Lawyers’ future job prospects are directly related to their actual trial experiences, so settling a case is counterproductive from the individual lawyer’s point of view (Weaver 1977; Katzmann 1980). So, too, is spending effort in seeing that firms found guilty of violating the antitrust laws actually comply with court-ordered remedies or settlement agreements.

6.3.2.  Judicial Review I share with the authors a rejection of the notion that the judiciary’s role is one of representing the underrepresented groups in the political process. I join in their criticism of the romantic view that the members of the judiciary are the unique guardians of some mystical “public interest,” something that is wholly ignored by other branches of government. Buchanan (1975, 903)

Once a case finds its way to the courts, the method of analysis does not have to change. Public choice studies of judicial selection, decision-making on the bench, the impact of judicial appointments, and other institutional features of the judiciary have confirmed that a homo economicus can be found under black robes, as well.20 Tullock (1975, 761) proposed “an economic model of the Anglo-Saxon adversary court system” in which judges were analyzed as utility-maximizing economic agents subject to institutional constraints.21 While studying efficiency differences between two distinctive institutional settings (the adversary court system and the inquisitorial court system), Tullock argued that the incentives faced by utility-maximizing judges might affect their supply of effort. Although “in theory, one could be held in contempt of court” (Tullock, Seldon, and Brady 2005, 11) for adopting such an approach, many scholars took up the challenge to test empirically whether judges’ behavior could be so described. 20  Baye and Wright (2011, 1) test the hypothesis that “antitrust analysis has become too complex for generalist judges” and find that “decisions involving the evaluation of complex economic evidence are significantly more likely to be appealed and decisions of judges trained in basic economics are significantly less likely to be appealed than are decisions by the untrained counterparts.” 21  Ordover and Weitzman (1977) criticized Tullock’s conclusions regarding the (in)efficiencies of an adversarial judicial system but implicitly acknowledged the value of an economic model for studying judicial behavior and/or the judicial system as a whole. Their comment is couched in the language of industrial organization, discussing, for example, “that amongst frequent litigants, the reaction functions

160   Fred S. McChesney, Michael Reksulak, and William F. Shughart II Landes and Posner (1975, 875)  too had encouraged economists to examine more “critically the idea of a nonpolitical, or ‘independent,’ judiciary,” an effort also applauded by Buchanan (1975). Samuels (1975, 907) pointed out as well that “both legislature and judiciary are political phenomena, although their form thereof differs.” The Landes-Posner model (1975, 894) deviated from earlier studies by “not venerating the courts as repositories of some special wisdom, integrity, morality, or commitment to principle. . . . [C]‌ourts do not enforce the moral law or ideals of neutrality, justice, or fairness; they enforce the ‘deals’ made by effective interest groups with earlier legislatures.” Benson (2001, 584) notes that “Court time in the United States is increasingly being taken up by political issues or by disputes involving opposing interest groups.” Posner (2011) shed light more recently on the motivations of judges, describing their utility functions not in monetary terms but as including elements that permit judges to indulge their policy preferences. However, Anderson, Shughart, and Tollison (1989, 217) reported evidence that “the legislature is in a position to reward the judiciary for behavior that enhances the value of legislative output by means of the power of the purse. Simply stated, judges’ salaries (and appropriations for the judicial branch generally) may be a positive function of the judiciary’s effective degree of independence.” Baum (1994, 752) discusses other potential influences on judges: maintaining one’s current judicial position, promotion to a higher court, and attaining attractive nonjudicial jobs. Anderson (2001, 307)  reviews the literature and concludes that judges “tend to respond in predictable ways to incentives in the form of salaries, budgets, workload reductions and various other perquisites.” A recent paper by Landes and Posner (2009, 823) examines “rational judicial behavior” statistically and finds that “ideology matters more in the Supreme Court than in the court of appeals.” They (2009, 824) suggest that additional insights into judicial motivation and behavior could be gained by identifying “from media accounts court of appeals judges who have had good prospects for promotion to the Supreme Court, based on media speculation, and see whether they dissent more than their peers, or otherwise behave differently, in order to attract attention or otherwise enhance their promotion prospects.”22 In surveying the literature, Hanssen and Fleck (2013) examine how “discussion of the benefits accruing from independent courts tends to mask the reality that even the most insulated judges operate in politically defined settings.” They cite numerous studies presenting empirical evidence that goals of office retention, career concerns, and may be of a non-COURNOT type. The reaction function that may emerge from such repeated trial games will have associated NASH-equilibria that strictly dominate the NASH equilibrium associated with COURNOT reaction functions.” Tullock’s results—but not his method of analysis—were also called into question by McChesney (1977, 510), who concluded that “casual empiricism indicates further that the externalities are probably greater in the civil law courts than in the common law courts.” See, also, Tullock’s (1977) response. 22 

Federal judges who perceive opportunities for promotion to a higher court have been found to be more likely to uphold the constitutionality of the U.S. Sentencing Commission (Cohen 1991) and to impose harsher sentences on guilty defendants in antitrust cases initiated by the U.S. Department of Justice (Cohen 1992).

Competition Policy in Public Choice Perspective   161

strategic considerations regarding interactions with legislators and voters, among others, influence judicial decision-making. Given those results, Hanssen (2004a, 2004b) questions what level of judicial independence may be optimal, assuming that incentives matter here as well. Hanssen (1999a, 1999b) also analyzes litigation rates when judges are appointed versus elected. “Independence” in terms of life-tenure actually seems to provide some insulation and “net positive effect on decision uncertainty” by, ceteris paribus, leading to “more litigation when judges are appointed” (Hanssen 1999a, 205).23 Most recently, Iaryczower and Shum (2012a, 230) designed and tested a “voting model for the US Supreme Court in which votes reflect both justices’ personal ideologies as well as their endeavors to ‘get it right’ ”. Although ideology has great predictive powers, the authors see evidence that case-specific information can counteract those leanings to some extent. In a later paper (Iaryczower and Shum 2012b, 22), the same authors, however, find a somewhat more worrisome result for supreme courts in US states where justices are elected to office. Here, they conclude that “a judge’s voting strategy leans more heavily towards an interest group the larger are its contributions to the judge, and the smaller are its contributions to other members of the court.” The results just summarized concern the judiciary generally. It is not surprising, then, that public choice analysis suggests interest group influences and the personal interests of judges to be significant in antitrust proceedings. That is especially the case since antitrust legislation has few “fixed, bright-line rules of conduct” where business practices (e.g., price fixing) are declared to be per se illegal, but instead operates in an environment of more flexible “rule-of-reason” standards (Greenfield and Matheson 2009, 87).24 Accordingly, “in recent years, debate has often centered on whether antitrust should supply bright-line rules or safe harbors of per se (or near per se) legality to minimize the risks of chilling procompetitive behavior and reduce administrative costs, or whether more searching analysis is needed to capture conduct that is anticompetitive in the particular circumstances” (Greenfield and Matheson 2009, 87). The public choice perspective suggests that more judicial discretion has the potential for more interest-group influence on the process. As noted by Rajabiun (2012, 187), “judicial discretion under the rule-of-reason approach to substantive interpretation limits the predictability and credibility of legal constraints against anticompetitive practices.” Observation that courts often defer to regulatory agencies when deciding whether or 23 

Shepherd (2011, 1753) “contend[s]‌that the public choice theory of judicial behavior has gone from an ‘emerging’ model to a model with broad acceptance across a range of disciplines.” However, she also cautions that one should not overstate the predictive power of such analysis and concludes (1765) that “even in instances when judges’ personal preferences might take precedence over their pleasure in judging, much of our judicial system has been created to minimize the opportunity for judges to act out of self-interest.” 24  The Sherman Act proscribes conduct “in restraint of trade” (Section 1) or tantamount to “monopolization” (Section 2) with no definitions of those terms. Likewise, the Clayton Act leaves undefined the practices (tying and exclusive dealing in Section 3, mergers in Section 7) that are proscribed when they “substantially lessen competition or tend to create a monopoly.” Section 5 of the Federal Trade Commission Act prohibits “unfair methods of competition,” again eschewing any definition of what that phrase means.

162   Fred S. McChesney, Michael Reksulak, and William F. Shughart II not to take on a case may just reflect another way judges express their preferences. Also, it has been suggested that courts may be more likely to overturn regulatory determinations of antitrust misconduct when doing otherwise may “break bargains struck in the legislative marketplace” (Shughart 1990, 133). Taken together, this theory and evidence paints a picture of a system of antitrust enforcement comprised of actors in the legislature, in the executive, and in the judiciary, where in each instance the decision-makers’ determinations can themselves be analyzed by starting with a behavioral model incorporating the incentives and constraints faced by self-interested individuals.

6.4.  Antitrust and Interest Groups Ethical theories may have good guys and bad guys, but economic theory possesses no such distinctions. Economic theory is about the children of Adam Smith, who obey the dictates of self-interest, given the constraints on their behavior that prevail. McCormick and Tollison (1981, 126)

Buchanan and Tullock (1962, 3) emphasized that any useful discussion of collective decisions starts with the recognition that collective outcomes are “composed of individual action.” Public choice analysis models individuals as homo economicus and emphasizes that homo economicus and homo politicus are one and the same. Any behavioral difference between the two lies in the various institutional constraints that individuals face. However, constraints exist in all marketplaces, whether for goods and services, for legislative action, for bureaucratic decisions, or for judicial verdicts. One major institutional dissimilarity nevertheless does exist: decisions taken in public office will often impact a multitude of people, while decisions in private settings are less likely to do so. Thus arises the impetus for trying to understand why public decisions are made and in what manner. There is no analytical place for a bifurcated view of human incentives, which assumes that individuals pursue utility-maximizing goals in their private lives while aspiring to maximize some putative social welfare function when in public office. In both cases, it is the constraints hoisted upon individual self-interest that guide the outcome. In public office, however, wealth transfers can be imposed by the powers possessed by elected or appointed officials. The ability to transfer and redistribute wealth invites rent-seeking activities (see, e.g., Tullock 1967; Krueger 1974; Tollison 1982; Rowley, Tollison, and Tullock 1988; Congleton, Hillman, and Konrad 2008a, 2008b). The smaller the interest group seeking the rent and the more dispersed those who will pay it, the greater is the likelihood that costs of organizing, monitoring, and maintaining interest groups (Olson 1965) can be overcome. In antitrust specifically, the stakes often are enormous for the companies in question. Opposing parties, such as consumers in general or the customers of the defendants, will be large in number and typically not well organized, increasing the probability of interest-group influence and lobbying. Antitrust

Competition Policy in Public Choice Perspective   163

actions are often localized in their impact, if successful, and so provide stronger incentives to congressional delegations in the affected jurisdiction to act. Simultaneously, the cost of thwarted antitrust activity is likely to be so much dispersed among the constituents in the remaining areas of the country that other lawmakers will seldom feel compelled to intervene. Consequently, the playing field is set for rigorous contests about what is permissible competition. Shughart (1990) discusses in depth how private interests are at work throughout the antitrust process. As we have outlined above, such private interests are not confined to business enterprises. They comprise, too, the antitrust bureaucracy, the Congress, the private antitrust bar, and the judiciary.

6.4.1.  Antitrust as a Continuation of Competition by Other Means [R]‌ent-seeking behavior by competitors is widespread (I see it frequently from my vantage point at the Federal Trade Commission) and is costly to the economy. It is costly not only when it succeeds in specific cases, but also when the fear of possible antitrust litigation intimidates firms from engaging in competitive conduct. Miller (1985, 267)

Public choice predicts that antitrust law violations will be complained of strategically, especially by firms whose markets are threatened by new entry, in order to block (or retard) the emergence of competition. Baumol and Ordover (1985) express concern that “far from serving as the bulwark of competition, [antitrust] institutions will become the most powerful instruments in the hands of those who wish to subvert it. More than that, it threatens to draw great quantities of resources into the struggle to prevent effective competition, thereby more than offsetting the contributions to economic efficiency promised by antirust activities” (1985, 247). Such protectionism through government action against domestic and/or international competitors is just a form of rent seeking. One advantage of using antitrust against competitors is the transfer of litigation costs to the taxpaying general public, if the government brings a case. Moreover, if the result of the complaint is either a settlement or a verdict for the government, private firms will bear the costs of ongoing government monitoring. Antitrust agencies have become more adept at identifying this strategy (not the least because economists point out its success). Rent-seeking firms will weigh costs and benefits of eliciting government action against competitors. Firms that feel threatened by entry have incentives to exaggerate the benefits of action and will incur costs themselves to prod government into action. Some antitrust lawsuits (e.g., the few victories nowadays based on predatory pricing claims) may often—given litigation costs and managerial time spent on trial strategy (and on preparing for depositions of key decision-makers)—be pyrrhic, even

164   Fred S. McChesney, Michael Reksulak, and William F. Shughart II if successful. At the margin, however, if the threat of antitrust litigation deters entry, then the rent-seekers have achieved their objective.25 One example of ill-advised antitrust policies was the so-called war on mergers during the 1990s. During a time of particularly large numbers of merger applications—driven by a booming stock market—the government decided to crack down on proposed business consolidations and joint ventures, frequently cheered on by competitors claiming the potential for antitrust injury if the transactions were consummated. Shughart (1998) discusses a number of such cases and dismisses the economic arguments brought forth by the antitrust bureaucracy in each one of them. He concludes (1998, 1) that “Because they respond to the demands of competitors, labor unions, and other well-organized groups having a stake in stopping mergers that promise to increase economic efficiency, the antitrust authorities all too often succeed, not in keeping prices from rising, but in keeping them from falling.” That scenario recalls the findings by DiLorenzo (1985) discussed earlier, which showed similar outcomes a century earlier.

6.5. Conclusion The problem is designing a set of public institutions so that, assuming the employees will take the public-choice approach, the outcome is truly to the public’s advantage. The big problem, of course is that for most of us individually, rent seeking is apt to pay off more highly than attempting to set up an organization that will serve the public interest. Tullock (1987, 342)

Public choice economics, as a social science tasked with explaining the behavior of individuals in every kind of market, treats antitrust seriously. To again paraphrase James Buchanan, it is law enforcement without romance. While incorporating progress made in understanding how individuals in different organizations and agencies would rationally maximize in the face of budgetary constraints, public choice analysis insists on accounting for the self-interests of people as well as for the influence of interest-group politics. This way of analyzing policy outcomes and applications does not deny that individuals may care about the unfortunate in society, their friends or their community. It does, however, embed these objectives into a larger picture of rational actors pursuing objectives in their own best interest, as much as they can determine them given informational and institutional constraints.26 Moreover, from this perspective it follows that 25 

See Shughart (1990, 57–76) and Shughart and Thomas (2013) for summaries of cases initiated over the past two decades that exemplify the use of antitrust to subvert competition. 26  Sokol (2011, 1040) makes the argument that sometimes “Public choice relies not just on the assumption of selfishness, but on selection effects... [M]‌aybe a congressman or bureaucrat does not care about survival in office and just wants to advance the public good. But such people will not stay in power for long unless they act selfish[ly] even if they are not. Selfish leaders or those that merely act selfishly will push out the altruistic ones over time, just as profit-seeking firms in the market will tend to out-compete those that are indifferent to profit.”

Competition Policy in Public Choice Perspective   165

simply substituting incumbent bureaucrats with “better” (more knowledgeable, more “publicly spirited”) people is not the solution. If feasible at all, the way out is paved with better institutional designs and incentives. Seen in this light, antitrust becomes a much more interesting subject than when put into the illusory straitjacket of “consumer welfare.” The intersection of many private interests, with numerous actors in the process, provides room for scientifically sound explanations of the “failure” of antitrust enforcement. In many cases, such failures turn out to be the product of a “goldilocks fallacy,” in which somebody imagines that everything is “just right” about why antitrust originated, how antitrust legislation was written, how the antitrust bureaucracy is designed, and what antitrust cases are selected and adjudicated. As described above, what may be “just right” for one actor in this multifaceted repeated game of competing interests may be very wrong for another. Such dispersion of incentives leads, unavoidably, to interest-group activity and rent seeking, especially so when legislators, bureaucrats, and judges are granted decision-making discretion. In no way does this indicate that public choice necessarily declares antitrust to be a failure conceptually. To the contrary, over the last 50 years, public choice has provided many insights into how to improve the application of antitrust enforcement and remedies by supplying a better understanding of how the process truly works. Only after realizing what is wrong—and why—can one usefully begin to fix it. This chapter has focused almost exclusively on competition policy in the United States for two reasons: First, the idea of a national antitrust law enforcement regime originated there in the late nineteenth century and, so, public choice scholars interested in explaining its origins, purposes, and effects have access to 100 years or more of information collected from the case law and observations on the behavior of the public bureaus and private parties who have standing to sue for relief from supposed or real anticompetitive business practices. Second, and what is perhaps more important, the public choice perspective on competition policy is virtually uncharted territory in Europe, Asia, and the rest of the world. Given the substantial and repeatedly reproduced evidence adduced by public choice scholars showing the influences of special interests on the enforcement of the US antitrust laws, we end with a clarion call for research by others who are not afraid to ask and answer tough questions about competition policy in other nations.

References Aktas, Nihat, Eric de Bodt, Marieke Delanghe, and Richard Roll. 2011. Market reactions to European Merger Regulation:  A  Reexamination of the Protectionism Hypothesis. SSRN Working Paper. Available at http://ssrn.com/abstract=1961188. Last accessed May 18, 2014. Aktas, Nihat, Eric de Bodt, and Richard Roll. 2007. Is European M&A Regulation Protectionist? Economic Journal 117(522): 1096–1121. Anderson, Gary M. 2001. The Judiciary. In William F. Shughart II and Laura Razzolini, eds., The Elgar Companion to Public Choice. Cheltenham, UK, and Northampton, MA, USA: Edward Elgar, pp. 293–309.

166   Fred S. McChesney, Michael Reksulak, and William F. Shughart II Anderson, Gary M., William F. Shughart II and Robert D. Tollison. 1989. On the Incentives of Judges to Enforce Legislative Wealth Transfers. Journal of Law and Economics 32(1): 215–28. Asch, Peter. 1970. Economic Theory and the Antitrust Dilemma. New York: John Wiley and Sons. Baker, Jonathan B., and Carl Shapiro. 2007. Reinvigorating Horizontal Merger Enforcement. SSRN Working Paper. Available at http://ssrn.com/abstract=991588. Last accessed May 18, 2014. Baker, Jonathan B., and Carl Shapiro. 2008. Detecting and Reversing the Decline in Horizontal Merger Enforcement. Antitrust 22(3): 29–36. Baum, Lawrence. 1994. What Judges Want:  Judges’ Goals and Judicial Behavior. Political Research Quarterly 47(3): 749–68. Baumol, William J., and Janusz A.  Ordover. 1985. Use of Antitrust to Subvert Competition. Journal of Law and Economics 28(2): 247–65. Baye, Michael R., and Joshua D. Wright. 2011. Is Antitrust Too Complicated for Generalist Judges? The Impact of Economic Complexity and Judicial Training on Appeals. Journal of Law and Economics 54(1): 1–24. Benson, Bruce L. 2001. Law and Economics. In William F. Shughart II and Laura Razzolini, eds., The Elgar Companion to Public Choice. Cheltenham, UK, and Northampton, MA, USA: Edward Elgar, pp. 547–89. Benson, Bruce L., M. L. Greenhut, and Randall G. Holcombe. 1987. Interest Groups and the Antitrust Paradox. Cato Journal 6(3): 801–17. Berg, Andrew G., and Gregory J.  Casas. 2011. New HSR Rules for Premerger Notification Filings—Effective August 18. The National Law Review. Posted on Saturday, July 30, 2011. Available at http://www.natlawreview.com/article/new-hsr-rules-premerger-notificationfi​lings-effective-august-18. Last accessed May 18, 2014. Bork, Robert H. 1966. Legislative Intent and the Policy of the Sherman Act. Journal of Law and Economics 9: 7–48. Bork, Robert H. 1978. The Antitrust Paradox: A Policy at War with Itself. New York: Basic Books. Boudreaux, Donald J., and Thomas J.  DiLorenzo. 1993. The Protectionist Roots of Antitrust. Review of Austrian Economics 6(2): 81–96. Boudreaux, Donald J., Thomas J.  DiLorenzo, and Steven Parker. 1995. Antitrust before the Sherman Act. In Fred S.  McChesney and William F.  Shughart II, eds., The Causes and Consequences of Antitrust: The Public-Choice Perspective. Chicago: University of Chicago Press, pp. 255–70. Bradley, Robert L., Jr. 1990. On the Origins of the Sherman Antitrust Act. Cato Journal 9(3): 737–42. Buchanan, James M. 1972. Toward Analysis of Closed Behavioral Systems. In James M. Buchanan and Robert D. Tollison, eds., Theory of Public Choice. Ann Arbor: University of Michigan Press, pp. 11–23. Buchanan, James M. 1975. The Independent Judiciary in an Interest-Group Perspective: Comment. Journal of Law and Economics 18(3): 903–5. Buchanan, James M., and Gordon Tullock. 1962. The Calculus of Consent: Logical Foundations of Constitutional Democracy. Ann Arbor:  University of Michigan Press/Ann Arbor Paperbacks. Carlton, Dennis W. 2007. Does Antitrust Need to be Modernized? Journal of Economic Perspectives 21(3): 155–76. Chamberlain, Lawrence H. 1946. The President, Congress and Legislation. New York: Columbia University Press.

Competition Policy in Public Choice Perspective   167

Coate, Malcolm B., and Fred S. McChesney. 1992. Empirical Evidence on FTC Enforcement of the Merger Guidelines. Economic Inquiry 30(2): 277–293. Coate, Malcom B., Richard S. Higgins, and Fred S. McChesney. 1990. Bureaucracy and Politics in FTC Merger Challenges. Journal of Law and Economics 33(2): 463–82. Cohen, Mark A. 1991. Explaining Judicial Behavior, or What’s “Unconstitutional” about the Sentencing Commission? Journal of Law, Economics and Organization 7: 183–99. Cohen, Mark A. 1992. The Motives of Judges:  Empirical Evidence from Antitrust Sentencing. International Review of Law and Economics 12: 13–30. Congleton, Roger D., Arye L. Hillman, and Kai A. Konrad, eds. 2008a. 40 Years of Research on Rent Seeking 1: Theory of Rent Seeking. Heidelberg: Springer. Congleton, Roger D., Arye L. Hillman, and Kai A. Konrad, eds. 2008b. 40 Years of Research on Rent Seeking 2: Applications: Rent Seeking in Practice. Heidelberg: Springer. Crandall, Robert W., and Clifford Winston. 2003. Does Antitrust Policy Improve Consumer Welfare? Assessing the Evidence. Journal of Economic Perspectives 17(4): 3–26. DeLorme, Charles D., W. Scott Frame, and David R. Kamerschen. 1997. Empirical Evidence on a Special-Interest-Group Perspective to Antitrust. Public Choice 92(3–4): 317–35. Dewatripont, Mathias, Ian Jewitt, and Jean Tirole. 1999. The Economics of Career Concerns, Part II: Application to Missions and Accountability of Government Agencies. Review of Economic Studies 66(1): 199–217. DiLorenzo, Thomas J. 1984. The Domain of Rent-Seeking Behavior: Private or Public Choice? International Review of Law and Economics 4(2): 185–97. DiLorenzo, Thomas J. 1985. The Origins of Antitrust: An Interest-Group Perspective. International Review of Law and Economics 5(1): 73–90. DiLorenzo, Thomas J., and Jack C.  High. 1988. Antitrust and Competition, Historically Considered. Economic Inquiry 26(3): 423–35. Dinc, Serdar, and Isil Erel. 2011. Economic Nationalism in Mergers and Acquisitions. Charles A.  Dice Center Working Paper No. 2009-24. Available at SSRN:  http://ssrn.com/​ abstract=1361107. Last accessed May 18, 2014. Downs, Anthony. 1967. Inside Bureaucracy. Boston: Little, Brown. Duso, Tomaso, Damien J. Neven, and Lars-Hendrik Roller. 2007. The Political Economy of European Merger Control: Evidence Using Stock Market Data. Journal of Law and Economics 50(3): 455–89. Ekelund, Robert B. Jr., Michael J. McDonald, and Robert D. Tollison. 1995. Business Restraints and the Clayton Act of 1914: Public- or Private-Interest Legislation? In Fred S. McChesney and William F. Shughart II, eds., The Causes and Consequences of Antitrust: The Public-Choice Perspective. Chicago: University of Chicago Press, pp. 271–86. Faith, Roger L., Donald R. Leavens, and Robert D. Tollison. 1982. Antitrust Pork Barrel. Journal of Law and Economics 25(2): 329–42. Garrett, Thomas A., Thomas L. Marsh, and Maria I. Marshall. 2006. Political Allocation of US Agriculture Disaster Payments in the 1990s. International Review of Law and Economics 26(2): 143–61. Gordon, Sanford D. 1963. Attitudes toward Trusts Prior to the Sherman Act. Southern Economic Journal 30(2): 156–67. Grandy, Christopher. 1993. Original Intent and the Sherman Antitrust Act: A Re-examination of the Consumer-Welfare Hypothesis. Journal of Economic History 53(2): 359–76. Greenfield, Leon B., and Daniel J. Matheson. 2009. Rules versus Standards and the Antitrust Jurisprudence of Justice Breyer. Antitrust 23(3): 87–91.

168   Fred S. McChesney, Michael Reksulak, and William F. Shughart II Hanssen, F.  Andrew. 1999a. The Effect of Judicial Institutions on Uncertainty and the Rate of Litigation:  The Election versus Appointment of State Judges. Journal of Legal Studies 28(1): 205–32. Hanssen, F.  Andrew. 1999b. Appointed Courts, Elected Courts, and Public Utility Regulation:  Judicial Independence and the Energy Crisis. Business and Politics 1(2): 179–201. Hanssen, F.  Andrew. 2004a. Is There a Politically Optimal Level of Judicial Independence? American Economic Review 94(3): 712–29. Hanssen, F. Andrew. 2004b. Learning about Judicial Independence: Institutional Change in the State Courts. Journal of Legal Studies 33(2): 431–74. Hanssen, F.  Andrew, and Robert K.  Fleck. 2013. Judges:  Why Do They Matter? In Michael Reksulak, Laura Razzolini, and William F. Shughart II, eds., The Elgar Companion to Public Choice, 2nd ed. Cheltenham, UK, and Northampton, MA, USA: Edward Elgar, pp. 233–48. Higgins, Richard S., and Fred S. McChesney. 1986. Truth and Consequences: The Federal Trade Commission’s Ad Substantiation Program. International Review of Law and Economics 6(2): 151–68. Horn, Murray J. 1995. The Political Economy of Public Administration. New York: Cambridge University Press. Iaryczower, Matias, and Matthew Shum. 2012a. The Value of Information in the Court: Get It Right, Keep It Tight. American Economic Review 102(1): 202–37. Iaryczower, Matias, and Matthew Shum. 2012b. Money in Judicial Politics:  Individual Contributions and Collective Decisions. SSRN Working Paper. Available at http://ssrn.com/​ abstract=1998895. Last accessed May 18, 2014. Katzmann, Robert A. 1980. Regulatory Bureaucracy:  The Federal Trade Commission and Antitrust Policy. Cambridge, MA: MIT Press. Kaysen, Carl, and Donald F. Turner. 1959. Antitrust Policy: An Economic and Legal Analysis. Cambridge, MA: Harvard University Press. Kiewiet, D.  Roderick. 1991. Bureaucrats and Budgetary Outcomes:  Quantitative Analysis. In André Blais and Stéphane Dion, eds., The Budget-Maximizing Bureaucrat: Appraisals and Evidence. Pittsburgh: University of Pittsburgh Press, pp. 143–73. Kirkwood, John B., and Robert H. Lande. 2008. The Fundamental Goal of Antitrust: Protecting Consumers, Not Increasing Efficiency. Notre Dame Law Review 84(1): 191–244. Krueger, Anne O. 1974. The Political Economy of the Rent-Seeking Society. American Economic Review 64(3): 291–303. Lande, Robert H. 1982. Wealth Transfers as the Original and Primary Concern of Antitrust: The Efficiency Interpretation Challenged. Hastings Law Journal 34: 65–151. Landes, William M., and Richard A. Posner. 1975. The Independent Judiciary in an Interest-Group Perspective. Journal of Law and Economics 18(3): 875–901. Landes, William M., and Richard A.  Posner. 2009. Rational Judicial Behavior:  A  Statistical Study. Journal of Legal Analysis 1(2): 775–832. Libecap, Gary D. 1992. The Rise of the Chicago Packers and the Origins of Meat Inspection and Antitrust. Economic Inquiry 30(2): 242–62. Lin, Ping, and Jingjing Zhao. 2012. Merger Control Policy under China’s Anti-Monopoly Law. Review of Industrial Organization 41: 109–132. Mackay, Robert J., James. C.  Miller III, and Bruce Yandle, eds. 1987. Public Choice and Regulation:  A  View from Inside the Federal Trade Commission. Stanford, CA:  Hoover Institution Press.

Competition Policy in Public Choice Perspective   169

Makris, Miltos. 2003. Administrative Bureaus with Standard Operating Procedures. CMPO Working Paper Series No. 03/062. Available at http://econpapers.repec.org/paper/​ bricmpowp/03_2f062.htm. Last accessed May 17, 2014. McChesney, Fred S. 1977. On the Procedural Superiority of a Civil Law System: A Comment. Kyklos 30(3): 507–10. McChesney, Fred S. 1996. Introduction: The Demand for and Supply of Economics in Modern Antitrust. In Fred S. McChesney, Economic Inputs, Legal Outputs. New York: John Wiley & Sons, pp. xiii–xxi. McChesney, Fred S. and Katherine M. Larkin-Wong. 2013. The Public-Choice Perspective on Antitrust Law. In Michael Reksulak, Laura Razzolini, and William F. Shughart II, eds., The Elgar Companion to Public Choice, 2nd ed. Cheltenham, UK, and Northampton, MA , USA: Edward Elgar, pp. 291–303. McChesney, Fred S., and William F. Shughart II, eds. 1995. The Causes and Consequences of Antitrust: The Public-Choice Perspective. Chicago: University of Chicago Press. McCormick, Robert E., and Robert D. Tollison. 1981. Politicians, Legislation, and the Economy. Boston: Martinus Nijhoff. McKay, Amy Melissa. 2011. The Decision to Lobby Bureaucrats. Public Choice 147(1–2): 123–38. Miller, James C., III. 1985. Comments on Baumol and Ordover. Journal of Law and Economics 28(2): 267–70. Muris, Timothy J. 1986. Regulatory Policymaking at the Federal Trade Commission: The Extent of Congressional Control. Journal of Political Economy 94(4): 884–89. Muris, Timothy J. 2008. Facts Trump Politics: The Complexities of Comparing Merger Enforcement over Time and between Agencies. Antitrust 22: 37–38. Niskanen, William A. 1968. Nonmarket Decision Making: The Peculiar Economics of Bureaucracy. American Economic Review 58(2): 293–305. Niskanen, William A. 1971. Bureaucracy and Representative Government. Chicago: Aldine-Atherton. Niskanen, William A. 1975. Bureaucrats and Politicians. Journal of Law and Economics 18(3): 617–44. Niskanen, William A. 1991. A Reflection on Bureaucracy and Representative Government. In André Blais and Stéphane Dion, eds., The Budget-Maximizing Bureaucrat: Appraisals and Evidence. Pittsburgh: University of Pittsburgh Press, pp. 13–32. Niskanen, William A. 1994. Bureaucracy and Public Economics. Brookfield, VT: Edward Elgar. Niskanen, William A. 2001. Bureaucracy. In William F. Shughart II and Laura Razzolini, eds., The Elgar Companion to Public Choice. Cheltenham , UK, and Northampton, MA, USA: Edward Elgar, pp. 258–70. Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge, MA: Harvard University Press. Ordover, J. A., and Phillip Weitzman. 1977. On the Efficient Organization of Trials: A Comment. Kyklos 30(3): 511–16. Posner, Richard A. 1974. Theories of Economic Regulation. Bell Journal of Economics and Management Science 5(2): 335–58. Posner, Richard A. 1976. Antitrust Law:  An Economic Perspective. Chicago:  University of Chicago Press. Posner, Richard A. 2005. The Federal Trade Commission: A Retrospective. Antitrust Law Journal 72(3): 761–71. Posner, Richard A. 2011. Economic Analysis of Law. 8th ed. Aspen, CO: Aspen Publishers. Pozen, Sharis A. 2011. Statement given before the Subcommittee on Intellectual Property, Competition, and the Internet, Committee on the Judiciary, United States House of

170   Fred S. McChesney, Michael Reksulak, and William F. Shughart II Representatives, Oversight Hearing on the Federal Trade Commission’s Bureau of Competition and the US Department of Justice’s Antitrust Division. Available at http://​ www.justice.gov/atr/public/testimony/278020.pdf. Accessed March 30, 2012. Rajabiun, Reza. 2012. Private Enforcement and Judicial Discretion in the Evolution of Antitrust in the United States. Journal of Competition Law & Economics 8(1): 187–230. Ramírez, Carlos D., and Christian Eigen-Zucchi. 2001. Understanding the Clayton Act of 1914: An Analysis of the Interest Group Hypothesis. Public Choice 106 (1 –2): 157–81. Reksulak, Michael. 2010. Antitrust Public Choice(s). Public Choice 142(3–4): 423–28. Reksulak, Michael, and William F. Shughart II. 2011. Of Rebates and Drawbacks: The Standard Oil Company (N.J.) and the Railroads. Review of Industrial Organization 38(3): 267–83. Reksulak, Michael, and William F. Shughart II. 2012. Tarring the Trust: The Political Economy of Standard Oil. Southern California Law Review Postscript 85(5): 23–32. Reksulak, Michael, William F. Shughart II, Robert D. Tollison, and Atin Basu Choudhary. 2004. Titan Agonistes: The Wealth Effects of the Standard Oil (N.J.) Case. Research in Law and Economics 21: 63–84. Rowley, Charles K., Robert D. Tollison, and Gordon Tullock, eds. 1988. The Political Economy of Rent-Seeking. Boston: Kluwer Academic Publishers. Samuels, Warren J. 1975. The Independent Judiciary in an Interest-Group Perspective: Comment. Journal of Law and Economics 18(3): 907–11. Shan, Pingping, Guofu Tan, Simon J.  Wilkie, and Michael A.  Williams. 2012. China’s Anti-Monopoly Law:  What Is the Welfare Standard? Review of Industrial Organization 41: 31–52. Shepherd, Joanna M. 2011. Measuring Maximizing Judges: Empirical Legal Studies, Public Choice Theory, and Judicial Behavior. University of Illinois Law Review 2011(5): 1753–66. Shughart, William F., II. 1990. Antitrust Policy and Interest-Group Politics. New York: Quorum Books. Shughart, William F., II. 1996. Monopoly and the Problem of the Economists. Managerial and Decision Economics 17(2): 217–30. Shughart, William F., II. 1997. The Organization of Industry. 2nd ed.. Houston: Dame Publications. Shughart, William F., II. 1998. The Government’s War on Mergers: The Fatal Conceit of Antitrust Policy. Policy Analysis no. 323, Cato Institute. Shughart, William F., II. 2000. The Fleeting Reagan Antitrust Revolution. Antitrust Bulletin 45: 271–89. Shughart, William F., II, and Diana W.  Thomas. 2013. Antitrust Enforcement in the Obama Administration’s First Term: A Regulatory Approach. Policy Analysis no. 739, Cato Institute. Shughart, William F., II, and Robert D.  Tollison. 1985. The Positive Economics of Antitrust Policy: A Survey Article. International Review of Law and Economics 5(1): 39–57. Shughart, William F., II, and Robert D. Tollison. 1987. Antitrust Recidivism in Federal Trade Commission Data: 1914–1982. In Robert J. Mackay, James C. Miller III, and Bruce Yandle, eds., Public Choice and Regulation:  A  View from inside the Federal Trade Commission. Stanford, CA: Hoover Institution Press, pp. 255–280. Shughart, William F., II, and Robert D. Tollison. 1991. The Employment Consequences of the Sherman and Clayton Acts. Journal of Institutional and Theoretical Economics 147(1): 38–52. Shughart, William F., II, Robert D. Tollison, and Brian L. Goff. 1986. Bureaucratic Structure and Congressional Control. Southern Economic Journal 52(4): 962–72. Sokol, D.  Daniel. 2011. Explaining the Importance of Public Choice for Law. Michigan Law Review 109(6): 1029–48.

Competition Policy in Public Choice Perspective   171

Stigler, George J. 1966. The Economic Effects of the Antitrust Laws. Journal of Law and Economics 9: 225–58. Stigler, George J. 1975. Supplementary Note on Economic Theories of Regulation. In George J. Stigler, The Citizen and the State: Essays on Regulation. Chicago: University of Chicago Press, pp. 137–41. Stigler, George J. 1977. The Origin of the Sherman Act. Journal of Legal Studies 14(1): 1–12. Stigler, George J. 1982. The Economists and the Problem of Monopoly. American Economic Review 72(2): 1–11. Tollison, Robert D. 1982. Rent Seeking: A Survey. Kyklos 35(4): 575–602. Tollison, Robert D. 1983. Antitrust in the Reagan Administration: A Report from the Belly of the Beast. International Journal of Industrial Organization 1(2): 211–21. Tollison, Robert D. 1985. Public Choice and Antitrust. Cato Journal 4(3): 905–16. Tullock, Gordon. 1965. The Politics of Bureaucracy. Washington, DC: Public Affairs Press. Tullock, Gordon. 1967. The Welfare Costs of Tariffs, Monopolies, and Theft. Western Economic Journal 5(3): 224–32. Tullock, Gordon. 1975. On the Efficient Organization of Trials. Kyklos 28(4): 745–62. Tullock, Gordon. 1977. Reply to McChesney, and Ordover and Weizman. Kyklos 30(3): 517–19. Tullock, Gordon. 1987. Concluding Thoughts on the Politics of Regulation. In Robert J. Mackay, James C. Miller III, and Bruce Yandle, eds., Public Choice and Regulation: A View from inside the Federal Trade Commission. Stanford, CA: Hoover Institution Press, pp. 333–43. Tullock, Gordon, Arthur Seldon, and Gordon L. Brady. 2005. Government Failure: A Primer in Public Choice. 2nd ed. Washington, DC: Cato Institute. Weaver, Suzanne. 1977. The Decision to Prosecute: Organization and Public Policy in the Antitrust Division. Cambridge, MA: MIT Press. Weingast, Barry R., and Mark J. Moran. 1983. Bureaucratic Discretion or Congressional Control? Regulatory Policymaking by the Federal Trade Commission. Journal of Political Economy 91(5): 765–800. Weingast, Barry R., and Mark J. Moran. 1986. Congress and Regulatory Agency Choice: Reply to Muris. Journal of Political Economy 94(4): 890–94. Weir, Charlie. 1992. Monopolies and Mergers Commission, Merger Reports and the Public Interest: A Probit Analysis. Applied Economics 24(1): 27–34. Wilson, James Q. 1989. Bureaucracy:  What Government Agencies Do and Why They Do It. New York: Basic Books. Young, Andrew T., and Russell S. Sobel. 2013. Recovery and Reinvestment Act Spending at the State Level: Keynesian Stimulus or Distributive Politics? Public Choice 155: 449–68. Young, Marilyn, Michael Reksulak, and William F. Shughart II. 2001. The Political Economy of the IRS. Economics and Politics 13(2): 201–20.

CHAPTER 7

A N T I T RU S T SE T T L E M E N T S DANIEL L. RUBINFELD

7.1. Introduction Antitrust litigation has been going through a growth spurt in the past several decades as the result of expanding public enforcement worldwide, active private enforcement in the United States, and initial forays into private enforcement in other areas of the world. Given the large costs to the parties flowing from antitrust trials, it is not surprising that a vast majority of both private and public enforcement actions are resolved through settlement. In this chapter, I will sketch out the conceptual framework underlying the settlement-trial decision. I will also describe some of the empirical evidence on the settlement of both public and private antitrust cases, and in the process I will offer commentary on a number of important policy issues.

7.2.  Private Litigation 7.2.1.  The Settlement-Trial Decision There is a massive and still growing literature on the decision to settle litigation rather than to go to trial.1 While that literature is not antitrust specific, it is nevertheless highly

1  Settlement decisions have been modeled using both cooperative game theory (with a focus on the Nash bargaining model) and noncooperative game theory. See, for example, Priest and Klein (1984). For an overview review of the literature, see Cooter and Rubinfeld (1989), Daughety (2000), Hay and Spier (1998), Spier (2007), and Daughety and Reinganum (2005).

Antitrust Settlements   173

instructive. The literature tells us that settlement decisions can be affected by a host of factors, including, but not limited to, (1) the available savings in litigation costs surplus flowing from settlement (the greater the savings, the more likely a case will settle); (2) the risk aversion of the parties (the more risk averse the parties, the more likely they will settle; (3)  the perceived likelihood of success (the more pessimistic the plaintiff relative to the defendant, the greater the likelihood of settlement); and (4) reputational effects of external benefits flowing from the case (the greater the external benefits from a trial victory, the less likely the case will settle). Furthermore, with respect to litigation involving two parties, the literature tells us that the sequencing of negotiations can make a difference. For example, which party makes the last offer can affect the likelihood of settlement and the settlement outcome.2 A similar set of incentives applies to antitrust cases that have been filed. The settlement-trial decision is further complicated when the defendant is likely to face a sequence of distinct trials by multiple plaintiffs. A recent paper by Bernhardt and Xu (2012) highlights the additional complications that are raised in this context. In this situation, the parties use the outcome of the initial trial to update their priors concerning future trial prospects. In single trials, pretrial costs are sunk when the settlement-trial decision is made, whereas in sequential trials, pretrial costs of future litigation are likely to affect settlement decisions before those pretrial expenditures are made. It is not unusual for antitrust settlements to include most-favored-nation (“MFN”) clauses; these ensure that certain terms of future settlements will apply retroactively to the initial settlement. Spier (2003) has provided an insightful analysis of MFNs in the context of a single defendant facing multiple plaintiffs. In her framework, MFNs discourage the defendant from raising settlement offers over time; this encourages early settlement and reduces the costs associated with delay. Because MFNs have the effect of increasing the bargaining power of defendants, they generally benefit defendants.3 However, as Spier points out, MFNs can in some circumstances also benefit plaintiffs that have some bargaining power in the initial settlement negotiations. There is a downside, however. The risk with MFNs is that they could discourage future settlements by reducing and sometimes eliminating the bargaining range.4 It is often the case that the terms of antitrust settlements in private litigation are kept confidential, through court orders or through private agreements. On their face, secret settlements are troubling; they keep valuable information from the public—information that could inform the decisions of future litigants. The externalities that flow from 2  The outcomes will differ in a world of asymmetric information, depending on whether the informed or uninformed party makes the settlement offer. See, for example, Reinganum and Wilde (1986), Schweizer (1989), Spier (1992, 1997). For an analysis involving multiple parties, see Daughety and Reinganum (2005). 3  Spier’s analysis could be adapted to consider a single plaintiff facing multiple defendants. There, an MFN would discourage the plaintiff from reducing settlement offers over time, and would increase the bargaining power of the plaintiff. 4  The possibility of applying MFNs to trial outcomes as well as settlements raising intriguing additional issues, as Spier points out in her discussion of New Mexico’s MFN with Microsoft in the state attorneys general cases against Microsoft.

174   Daniel L. Rubinfeld secret settlements have been analyzed by a number of authors. To pick one example, Daughety and Reinganum (2002) point to the fact that secret settlements plaintiffs to learn about the likely success of future actions. They note that current plaintiffs may be able to extra rents from defendants if they are willing to reach such agreements. Furthermore, secret settlements may reduce over antitrust deterrence. If this means a reduction in suits with little or no merit, it could be a plus; otherwise there is a strong case to be made for a commitment by firms to settle future antitrust disputes in the open. While the theoretical literature on litigation settlement has direct implications for the analysis of the antitrust cases generally, the implications are particularly noteworthy with respect to cases that have relatively large stakes (for example, in large damages in private cases or significant injunctive remedies in cases brought by the competition authorities).

7.2.2.  Single versus Treble Damages A constructive example of how the incentives to settle antitrust litigation can be affected by the outstanding legal rules is given in the analysis of Perloff and Rubinfeld (1988). The authors consider the implications of a changing legal regime that moves from treble damages to single damages. This remains an issue of some import today, because companies that receive amnesty from federal criminal price fixing under the DOJ’s leniency program are limited to single damages in any follow-on civil litigation. If we want to understand the implications of the leniency program, one important component of our analysis will be an evaluation of the move from treble to single damages. To begin the analysis, assume that the parties are risk neutral and that both have the same expectations about the likely trial outcome. Then, the case will be likely to settle, unless the parties diverge on the appropriate split of the available bargaining surplus (the savings in trial costs). A move from treble to single damages is likely to reduce the bargaining surplus since the lower stakes will likely reduce each party’s expenditure at trial. In any antitrust settlement context, the bargaining process will be further complicated by whether the party making the last offer is relatively informed or uninformed about the likely outcome. If the informed party makes the last offer, strategy can be important because the informed party can use its offer as a signaling device given that it has some knowledge about the likely behavioral response of the uninformed party. The settlement probability can also be increased to the extent that the plaintiff is relatively pessimistic about its chances of success relative to the defendant’s view. This divergence of perspectives will increase the bargaining surplus (or equivalently, the bargaining range). The move from treble to single damages is also likely to reduce the bargaining surplus here because it will reduce both parties’ subjective views as to the plaintiff ’s expected reward from going to trial. Furthermore, settlement becomes more likely if the defendant stands to lose more than the plaintiff will gain (perhaps in terms of litigation brought by other plaintiffs).

Antitrust Settlements   175

Given that “reputation” is likely to be important in antitrust cases brought against successful businesses, it would not be surprising not only that a high proportion of cases settling, but that defendants will be relatively successful when the remaining cases do go to trial. Bourjade, Rey, and Seabright (2009) develop this concept in a theoretical framework, pointing out that “if firms can settle out of court, and if the Courts are reasonably reliable at establishing the truth of allegations, it should be the violators who settle and the innocent firms that refuse.” This prediction is borne out in the empirical study of Perloff and Rubinfeld, who showed a very high settlement rate (86 percent) and over two-thirds of all tried cases won by defendants. Finally, as mentioned previously, cases are more likely to settle, the greater the risk aversion of the parties. It might be thought at first blush that businesses involved in antitrust litigation are likely to be risk neutral. However, not all antitrust cases involve individual plaintiffs and there is no certainty that managers of businesses will act in a risk-neutral manner.5 If risk aversion increases with the magnitude of the losses involved, a move from treble to single damages will reduce the amount of risk and therefore, other things the same, increase the likelihood that cases will go to trial. Settlement decisions in cases that follow-on government suits add an additional theoretical complexity to the basic theoretical model just described. In one important contribution, Briggs, Huryn, and McBridge (1996) point out that “in equilibrium, a defendant can probabilistically signal a strong case by not offering to settle. A violator’s incentive to signal a strong case to deter a treble damage suit forces the government to pursue more trials than it would otherwise. Private plaintiffs are more likely to settle following a government suit than otherwise, but they win a trial with the same probability regardless of whether there was a previous government suit. Data on private suits support the latter two contentions.” The key to any empirical analysis of antitrust settlements is to understand that there is a substantial random element in any settlement bargaining. As a result, not all cases in which there is a bargaining surplus will settle and not all cases in which there is no surplus will go to trial. What are the implications of the stochastic nature of the bargaining process for the evaluation of a leniency program that replaces treble damages with single damages? The basic model of litigation predicts that the settlement probability will increase if the losses expected by defendants are less than the gains expected by plaintiffs for reputational or other reasons (if the defendants have a reputation at stake, they will take the case to trial), or if a reduction in the damage multiplier results in a corresponding increase in the costs of trial. Using data developed in a project sponsored by Georgetown Law School, Perloff and Rubinfeld (table  4.3) found that when plaintiffs are large firms, cases are much more likely to settle (90.9  percent) than when the plaintiff is another type of organization 5 

Koku and Qureshi (2006) compare news of settlements of cases involving NYSE and NASDQ listings. They find that settlement is beneficial to defendant firms (the settlements stop the loss of reputation that could result from an adverse trial outcome), but not to the plaintiff firms (whose stock prices do not change).

176   Daniel L. Rubinfeld (54.5 percent). Apart from this large-firm effect, however, the authors found little relationship between size and settlement rates. The type of litigation also had an effect on settlement rates, with settlement rates being highest when the plaintiff is a competitor, supplier, or buyer and lowest when the plaintiff is a licensee, employee, or dealer. Surprisingly, price-fixing cases and refusals to deal have relatively low rates of settlement, perhaps due the relative optimistic of the plaintiffs with respect to potential damage recoveries. The authors also found that defendants tend to have more at stake than plaintiffs. As a result, defendants settle cases that might be seen as close. The effect of this is to suggest that a move to single damages means some of the cases will involve lower stakes for the defendant and will be tried rather than settled.6 Ultimately the effect of a reduction in the damage multiplier depends on the distribution of settlement gaps measured across a wide range of cases. For those cases in which the bargaining surplus is initially large (perhaps because the plaintiff is relatively optimistic about the trial outcome), the probability of settlement is likely to fall when the damage multiplier is reduced. However, for those cases in which the surplus is negative, the probability of settlement will increase as the multiplier falls. Of course, changes in expected trial costs will affect this calculus as well. If trial costs are reduced, cases that settled before the multiplier was reduced are more likely to settle, while cases that did not settle become even less likely to settle. In Perloff and Rubinfeld’s simulations, the average probability of settlement was found to fall by 8.5 percent when the damage multiplier was reduced from 3 to 1 when measured over all cases, but by only 2.3 percent for class action cases. Perhaps surprisingly, these results suggest a potentially overlooked cost of the DOJ’s leniency program—an increase in the percentage of cases that go to trial and consequently an increase in the cost of private litigation.

7.2.3.  The Empirical Framework A number of authors have evaluated settlement probabilities using stochastic models of settlement behavior in private litigation. Relying on a long and compelling literature on settlement in tort litigation, Perloff, Rubinfeld, and Ruud (1996) have devised a model that is directed towards antitrust litigation specifically. Their model differs from much of the literature in specifically accounting for risk aversion and for the endogeneity of the settlement-trial process. In particular, expectations about trial outcomes that drive the settlement decision are estimated from a separate, but related, model of trial outcomes. Understanding the stochastic nature of settlement bargaining is essential if one is to make policy evaluations of the settlement implications of changes in various legal rules (fee shifting, possibilities for appeal, etc.). The framework used by Perloff, Rubinfeld, and Ruud (1996) provides a useful overview of the issues.7 6 

Many of these conclusions were supported by a probit analysis of settlement rates. A similar, but somewhat less general framework was used in the empirical analysis of Fournier and Zuehlke (1999). 7 

Antitrust Settlements   177

The expected utilities of the plaintiff and the defendant are assumed to depend only on the mean and the variance of their incomes yp and yd; for simplicity, I have dropped the subscripts j = p or d, which would distinguish between the plaintiff and the defendant: EU ( y ) = µ − δσ2 ,



(7.1)



where δ reflects the degree of risk aversion for the plaintiff and the defendant. Now, let Pp and Pd represent the parties’ expectations as to the likelihood that the plaintiff will win at trial and let Dp and Dd represent the corresponding subjective estimates of damages. Finally, let c p and cd be the parties’ respective subjective estimate of expenditures at trial. Then, the defendant’s expected loss from going to trial is Pd Dd + δ d σ2 Dd 2 + cd , 



(7.2)

which is the defendant’s expected utility loss at trial plus the defendant’s trial expenditures. Similarly, the plaintiff ’s expected benefit from going to trial is given by (7.3)

Pp Dp − δ p σ2 Dp2 − c p . 



The available bargaining surplus, S, to the parties from settling is the difference between the defendant’s expected loss at trial and the plaintiff ’s expected gain.

(

) + (δ D + (c + c ) .

S = Pd Dd − Pp Dp  

= α + γ σ2

p

d

d

2

)

(

+ δ p Dp2 σ2 + c p + cd

) (7.4)

d

We can see that the bargaining surplus will be higher, (i) the more pessimistic the defendant is about the likely success of the plaintiff ’s case relative to the plaintiff (the higher is α); (ii) the more uncertain the trial outcome, the more risk averse the parties (the higher is γ); and (iii) the greater the costs of going to trial. The stochastic model of antitrust settlements assumes that the probability of settlement, Ps, is an increasing function of the magnitude of the bargaining surplus:  the greater the surplus, the greater the mutual gains if a settlement agreement is reached. However, the model also reflects the possibility that settlement may not occur even though there is a positive gain from settling. If the model is to be estimated using a probit model, the empirical model to be specified would be given as

(

(

))

Ps = Φ (S ) = Φ α + γ σ2 + c p + cd , 

(7.5)

178   Daniel L. Rubinfeld where Φ is the cumulative normal distribution function. The first term in the probit equation is endogenous, since it reflects the parties’ expectations about trial outcomes, which in turn will be a function of actual trial outcomes of similar antitrust cases in the past. Perloff, Rubinfeld, and Ruud estimate a two-equation model, in which (i) the likely success at trial is estimated in an initial equation (not spelled out here) and (ii) equation (7.5) estimates parameters that rely on the estimated probabilities of trial success. What is noteworthy here is that the model predicts that settlements are more likely when the defendant has more at stake than the plaintiff (possibly due to an adverse reputational effect), that is, when Dd > Dp . In contrast, if the defendant believes that settlement gives a bad signal, the defendant might go to trial to show its resolve Dd < D p . Similarly, if the coefficient on the variance terms in equation (7.5) is positive, the case is more likely to settle. This would be the case, for example, if the parties are collectively risk averse. Recall that for a binomial model, the variance is greatest when the probability of success is .5. It follows that, other things equal, any change in the rules of evidence that causes the probability that the plaintiff will win at trial to move away from .5 will increase the variance of the trial outcome and correspondingly increase the probability of trial. Conversely, anything that moves the probability towards .5 will decrease the probability of trial. The application of this model to the Georgetown database led to some powerful results. The authors found that a higher perceived probability of plaintiff success increases the likelihood of settlement, other things the same, as does the increased risk that flows from uncertain trial outcomes. Furthermore, plaintiffs fare substantially better off when they demand a jury trial than a judge trial. The probability of winning at trial increases by 40  percent and the settlement probability increases even more. In addition, any policy that increases the probability that the plaintiff will be successful at trial from the mean of about 31 percent will increase the variance by about .37 percent. As a result, a 1 percent increase in the probability of plaintiff ’s success will increase the probability of settlement by .13 percent. The relatively small magnitude of this impact should not be underestimated. Risk does matter, and the costs of legal change can be very substantial. To put this in a somewhat stark form, policy changes (such as those affecting discovery rules) that are harmful to plaintiffs are likely to increase the likelihood that cases will be trial as well as the corresponding trial costs.8 A similar conclusion applies to the earlier detrebling discussion. A reduction in the damage multiplier will increase the fraction of cases that are litigated substantially.

(

(

)

)

8 

In one interesting empirical study, Hersch (2006) found that cases in which a jury trial was demanded were more likely to settle than judge-tried cases. The evidence suggested (but did not prove) that jury-demanded cases were weaker, which presumably reduced the plaintiffs’ settlement demands.

Antitrust Settlements   179

7.2.4.  Attorney Incentives Many of the largest private antitrust cases have been brought as class action cases in which the plaintiffs have claimed overcharge damages. Settlement outcomes, like the trial outcomes that they shadow, have raised a number of important normative policy issues. First, overcharges do not measure the actual harm suffered by price fixing, since they fail to account for deadweight loss—the harm to those who would have purchased the product at competitive prices, but opted not to purchase at monopoly prices.9 It follows that the decisions made by class action attorneys to bring cases and to choose between settlement and trial may not be closely aligned with the public interest.10 Second, attorneys’ fees in class settlements are determined by the court. What is the most appropriate basis for determining those fee arrangements? Awards in nonclass cases have typically been based on the percentage method, with one-third a commonly cited fraction. However, in class-action cases, the lodestar method has been frequently utilized. Under lodestar, the court awards reasonable attorney fees, along with a multiple (typically from one to three) that is meant to reward the risk taken by the successful plaintiff. The best empirical evidence on fee arrangements is the Eisenberg-Miller (2004) study of class action settlements generally, of which antitrust cases represented a modest subset. The authors find that the mean percentage award is 21.9 percent, and that there is a sliding scale, with fees constituting a lower percentage of the client’s recovery as the recovery increases. Relying on the fact that their model explains percentage awards better than the lodestar method and their belief that the time and expense of doing a lodestar calculation may be wasteful, the authors suggest that courts utilize a variant of the percentage method.11 This view is supported by the work of Helland and Klick (2006), who find (using data from the Federal Judicial Center) that lawyers facing a lodestar calculation delay settlement to accumulate more hours compared to cases in which courts use a percentage fee method for compensating attorneys.12 Third, many antitrust cases in the United States are brought by attorneys under contingent fee arrangements. It is been commonly thought that contingent fees give plaintiff attorneys an excessive motive, relative to the interest of the client, to settle the case. As the argument goes, if the case is settled, the attorney obtains his or her share of the settlement without having to invest the time that would be required if the case were to go to trial. In addition, it has been thought that the settlement amount will be less than the

9 

For a basic review of these issues, see Hovenkamp (1999, ch. 1). For a recent view of the determination of optimal antitrust sanctions and a review of the literature, see Ginsburg and Wright (2010). 11  The disadvantages of doing the lodestar method may be overstated, since there are firms that specialize in monitoring and evaluate lawyers’ hours and fees. 12  For a discussion of the fairness issues that are raised by class action settlements generally, see Macey and Miller (2009). 10 

180   Daniel L. Rubinfeld amount that would be in the ultimate interest of the client. This argument was based on the intuition that by making a low settlement demand, the attorney can encourage the defendant to accept the settlement. Polinsky and Rubinfeld (2002) show that these intuitions can be misleading. Specifically, when compared to a benchmark in which the client’s welfare if maximized, a contingent fee arrangement can lead attorneys to settle cases less often and for a higher amount. The reason for the misleading intuition is that it often fails to account for the fact that if the attorney takes the case to trial, the attorney will work fewer hours than is in the client’s interest. Incorporating this possibility, we conclude that attorneys’ settlement demands can be higher than the client would like, which would lead to too few settlements. Because the conventional analysis ignores the lower-trial-effort effect, it leads to the conclusion that attorneys will necessarily settle too often and at too low an amount. Ultimately, both the direction and the magnitude of the effect of contingent fees on settlement choices and amounts are indeterminate. Fourth, a common practice has been to award injured consumers coupons rather than cash. To what extent are these coupon remedies inefficient? Polinsky and Rubinfeld (2007) have suggested one procompetitive justification—in certain cases the offer of a choice between cash and coupons can create a sorting mechanism that distinguishes those who were in fact injured and those that were not. Despite this exceptional instance, there is little disagreement among a range of scholars that coupon remedies are likely to be inefficient. The primary source of inefficiency lies in the fact that attorneys’ fees are often based on the face value of the coupons, when, in fact, the coupon redemption rate is often quite low. This will lead to undercompensating of plaintiffs and underdeterrence. Polinsky and Rubinfeld point to an additional inefficiency. Coupons can distort consumption decisions. To paraphrase their argument, assume that demand for a product is stochastic, and assume further that demand in the remedy period is substantially less than demand in the period in which the antitrust injury occurred. The consumer will have a surplus of coupons and will likely buy an excessive amount of the product. (This assumes that coupons are not readily transferable.) In this case, the coupon remedy has the effect of lowering the price of the good below the competitive price. The authors show that the deadweight loss flowing from this inefficiency can be of the same order of magnitude as the deadweight loss flowing from the supracompetitive pricing that resulted from the antitrust injury itself.

7.3.  Public Enforcement 7.3.1. Fines and Leniency Programs While private enforcement has been predominately a US-centric exercise, public enforcement of antitrust has been a growth industry worldwide. Along with more

Antitrust Settlements   181

aggressive enforcement activities by a variety of enforcement authorities has come the opportunity for settlements that reflect a combination of fines and injunctive remedies. Settlements of cartel cases have been particularly important in the United States and in Europe. In this section, I briefly summarize the development of the system in the United States (the European Union has a system that is largely but not entirely similar). Following this, I briefly review some relevant empirical evidence in both Europe and the United States. In 1993 the United States updated and modified its Corporate Leniency Program as a means of strengthening public enforcement of cartels. Since that time these programs have grown internationally with over 50 countries having introduced such programs.13 Initially the program provided immunity from federal prosecution for the first company that came forward and fully cooperated with the government’s investigation. The company then faced single rather than treble damages in private litigation. By almost any account the program has been highly successful. Relatively few federal investigations result in trials on the merits; most are settled with the payment of corporate fines. (As explained previously, there is substantial follow-on litigation, and many of these cases are settled.) The first large cartel fines were assessed in the lysine and citric acid cases, with Archer Daniels Midland paying a $100 million fine. This was followed by fines of $110 million by UCAR International and by SGL in the graphic electrodes conspiracy in 1998 and 1999, respectively. The fines continued to grow over time. With respect to the conspirators in the worldwide vitamins conspiracy, F. Hoffman–La Roche Ltd paid $500 million and BASF AG paid $225 million. More recent large fines were levied with respect to air cargo ($1.6 billion in total), liquid crystal display ($860 million), and dynamic access memory ($730 million). The European Union’s leniency program has also been highly effective in inducing cooperation and in the imposition of substantial fines. As an example, the European Commission imposed fines on four car glass manufacturers that totaled more than 1.3 billion euros. Why do those cartel members not receiving leniency choose to settle cases rather than fully contest the decisions of the enforcement authorities? In the United States, plea bargaining is a standard means of disposing of cases involving both corporate and individual liability. In Europe, the process is somewhat different. Wils (2008) offers an insightful discussion with respect to the European Union’s leniency program. In general, Wils points out that the benefits of settlement include faster resolution, less adverse publicity, less burdensome remedies, and/or a narrower finding of infringement. In 2008 the Commission regulations were modified to allow settlements to incorporate (in some cases) a narrower than initial characterization of the cartel “infringement” and a 10 percent reduction in the fine. Wils argues that the faster resolution of cases is likely to increase the deterrence value of the enforcement. Moreover, the cost savings that flow from settlements allows the enforcement agency to pursue a larger number of cases, which also increases deterrence. In some instances faster resolution and lower cases will create a sufficient incentive for violators to settle cartel enforcement cases. However, in other cases additional benefits 13  The descriptive materials concerning the US program are described in detail in Hammond (2010). For a broad empirical overview, see Wood (2010); see also Spagnolo (2008).

182   Daniel L. Rubinfeld are likely to be required—the previously mentioned reductions in penalties being the primary device that is utilized. It seems likely that the loss of 10 percent of revenues will be outweighed by the benefits that flow from increased deterrence. However, I am unaware of any empirical postsettlement studies that provide direct evidence on this point. A complete study would require an evaluation of the selection of cases that do settle. Such a study would need to evaluate the incentives of both parties to resolve a case. To illustrate, it is possible, as Wils points out, for the competition authority to settle a relatively weak case and to fully litigate a strong case, whereas the incentives of the cartel members might be the opposite. The settlement resolution will depend on the specifics of the settlement process, any information asymmetries, and differences in the risks and rewards faced by each of the parties to the settlement negotiation.14 One paper that does offer some valuable insights into the settlement process under the European Union regulations is Ascione and Motta (2008). The authors offer an analysis of the optimal fine reduction for settling parties by examining the fine reductions awarded for all the European competition law infringements occurred between 1970 and 2007. In the view of the authors, a company that decides to settle likely loses the possibility of appealing to the Court of First Instance. The reason is that when deciding to enter in a settlement the Commission will compare the fine it receives if it settles with the fine it would expect to receive if it pursues an appeal through the courts. If the latter exceeds the former, the undertaking decides not to settle. The authors’ empirical analysis shows that the expected fine reduction of a firm that appeals to the European Community Courts is about 26 percent. This is substantially higher than the 10  percent fine reduction established by the Commission. The authors suggest as a consequence that the 10 percent fine will not create a sufficient incentive to settle, that the settlement participation rate will be low, and that the impact of the settlement inducements on the length of cartels’ prosecution will be small. A byproduct of the increasing globalization of antitrust enforcement has been the increased interaction among the enforcement authorities. This is to be expected in part because the reach of authorities such as the Department of Justice and the European Commission extends to foreign firms and individuals whose anticompetitive actions harm domestic competition and consumers. Moreover, the United States and the EU (among other agencies) have found it advantageous to provide assistance to the newer antitrust authorities and in some cases to enter into agreements to jointly engage in anticartel enforcement activities. With respect to the extraterritorial reach of the authorities, questions have been raised as to the motivations of the US and EU enforcement agencies. To be specific, EU 14 

De Azevedo and Furquim (2010) offer an interesting study of settlements in Brazil. First, they suggest that settlements can both increase the likelihood of detection and save litigation costs. Second, they point out that inducements towards settlements have been limited in use because they are effectively designed for the defendants that are likely guilty in any case. Nonetheless, the authors suggest that the settlement policy in Brazil has not had an adverse effect on leniency agreements, while reducing litigation costs and granting a final resolution in a number of cases. For a review of EU decisions, see Carree, Gunster, and Schinkel (2010).

Antitrust Settlements   183

settlements with major US companies such as Microsoft, IBM, and Intel have raised questions as to whether the European Commission might be acting in support of domestic industries. A recent study of enforcement actions by the European Union and the United States between 1994 and 2009 by Cremieux and Snyder (2010) provides an answer. The authors find little support for the protectionist interest theory. While their results are somewhat mixed, they do find evidence that US antitrust authorities have imposed disproportionately larger fines on EU firms than on US firms.

7.3.2.  Settlements and Consent Decrees Settlements of cases brought by governmental entities have a different character than settlements in private cases. For one thing, in civil cases almost always remedies are injunctive, some of which are structural in nature and others of which are behavioral. For another thing, government entities face real resource constraints that require strategic choices; settling one case can free up resources that will enable the prosecution of other cases. Furthermore, governmental actors may have different goals than their private counterparts, in terms of the creation of appropriate long-run deterrence incentives and/or the achievement of shorter-run political objectives. Settlements in governmental cases also create different incentives for defendants who are concerned about the public impact of continuing litigation and the incentives that settlement will have on future private litigation. In this subsection, I delve briefly into some of these issues. From the defendant’s perspective settling a case may limit the flow of adverse public information. Further, settling may avoid the use of facts and legal conclusions in follow-up litigation. To illustrate, findings relating to market definition and market power in government litigation can be used offensively in treble damages actions as collateral estoppel in private cases. From the government’s perspective, settlements not only save on litigation costs, when resolved through a formal consent decree, settlements can have significant precedential value. Under the US Tunney Act, settlements that take the form of consent decrees must be approved by the courts as being fair, reasonable, and adequate. Along with a public Competitive Impact Statement, the consent decree will include a clear statement of the antitrust concerns that are being remedied. This serves a valuable deterrence goal, while also allowing the government to claim success in its investigative and enforcement process. To illustrate this latter point, when the Antitrust Division of the Department of Justice reports on its civil enforcement merger activities, it typically emphasizes the number of enforcement actions, announcing not only those cases that were tried successfully or unsuccessfully, but also the number of mergers that were abandoned or settled through consent decrees.15

15 

For a broad discussion of the political economy of antitrust, see Ghosal and Stennek (2007).

184   Daniel L. Rubinfeld

7.4.  Concluding Remarks Settlements have played an important role in the resolution of private antitrust litigation for decades. Moreover, a variety of changes in the rules of civil procedure as well as the common law of antitrust have affected the incentives of parties to settle rather than to proceed to trial. What has been of more recent interest has been the expansive growth of public enforcement in many countries outside the United States. Along with that growth has come aggressive public enforcement coupled with leniency programs that encourage settlement. The next decade of enforcement should prove to be an exciting one not only from a policy perspective, but also from the point of view of empirical scholars of antitrust who are looking for a new set of “natural experiments” that will help us to understand the complexities of antitrust settlement bargaining.

Acknowledgment Douglas Spencer provided helpful research assistance.

References Ascione, Aurora, and Massimo Motta. 2010. Settlements in Cartel Cases. In Claus-Dieter Ehlermann and Mel Marquis, eds., European Competition Law Annual 2008:  Antitrust Settlements under EC Competition Law. Oxford: Hart, 67–83. Bernhardt, Dan, and Frances Zhiyun Xu. 2012. Trial Incentives in Sequential Litigation. Draft. Bourjade, Sylvain, Patrick Rey, and Paul Seabright. 2009. Private Antitrust Enforcement in the Presence of Pre-trial Bargaining. Journal of Industrial Economics 57(3): 372–409. Briggs, Hugh C., Kathleen D. Huryn, and Mark E. McBride. 1996. Treble Damages and the Incentive to Sue and Settle. Rand Journal of Economics 27(4): 770–86. Carree, Martin, Andrea Gunster, and Maarten Pieter Schinkel. 2010. European Antitrust Policy 1957–2004: An Analysis of Committee Decisions. Review of Industrial Organization 36: 97–131. Cooter, Robert D., and Daniel L. Rubinfeld. 1989. Economic Analysis of Legal Disputes and Their Resolution. Journal of Economic Literature 17: 1067–97. Cremieux, Pierre, and Edward A. Snyder. 2010. Global Antitrust Enforcement: An Empirical Assessment of the Influence of Protectionism. University of Chicago. Draft. Daughety, Andrew F. 2000. Settlement. In B. Bouckaert and G. DeGeest, eds., Encyclopedia of Law and Economics, vol. 5, 95–158. Daughety, Andrew F., and Jennifer F. Reinganum. 2002. Information Externalities in Settlement Bargaining:  Confidentiality and Correlated Culpability. Rand Journal of Economics 33: 587–604.

Antitrust Settlements   185

Daughety, Andrew F., and Jennifer F.  Reinganum. 2005. Economic Theories of Settlement Bargaining. Annual Review of Law and Social Science, 35–39. de Azevedo, Paulo Furquim. 2010. Cartel Deterrence and Settlements: The Brazilian Experience. Working Paper No. 265, Escola de Economia de São Paolo. Eisenberg, Theodore, and Geoffrey P. Miller. 2004. Attorneys Fees in Class-Action Settlements: An Empirical Study. Journal of Empirical Legal Studies 1: 27–78. Fournier, Gary M., and Thomas W.  Zuehlke. 1999. Litigation and Settlement:  An Empirical Approach. Review of Economics and Statistics 71: 189–95. Ghosal, Vivek, and Johann Stennek, eds. 2007. Political Economy of Antitrust. Boston: Elsevier. Ginsburg, Douglas H., and Joshua D. Wright. 2010. Antitrust Sanctions. Competition Policy International 6: 3–39. Hammond, Scott. 2010. The Evolution of Criminal Antitrust Enforcement over the Last Two Decades. Washington, DC: US Department of Justice. Hay, Bruce L., and Kathryn E. Spier. 1998. Settlement of Litigation. In New Palgrave Dictionary of Economics and the Law, 442–51. Helland, Eric A., and Klick, Jonathan. 2006. The Effect of Attorney Compensation on the Timing of Settlements. Available at SSRN. Hersch, Joni. 2006. Demand for a Jury Trial and the Selection of Cases for Trial. Journal of Legal Studies 35: 119–42. Hovenkamp, Herbert. 1999. Federal Antitrust Policy: The Law of Competition and Its Practice. 2nd ed. St. Paul, MN: Thomson-West. Koku, Paul Sergius, and Anique A.  Qureshi. 2006. Analysis of the Effects of Settlement on Interfirm Lawsuits. Managerial and Decision Economics 27: 307–18. Macey, Jonathan R., and Geoffrey P. Miller. 2009. Judicial Review of Class Action Settlements. Journal of Legal Analysis 1: 167–205. Miller, Nathan. 2009. Strategic Leniency and Cartel Enforcement. American Economic Review 99: 750–68. Perloff, Jeffrey M., and Daniel L. Rubinfeld. 1988. Settlements in Private Antitrust Litigation. In Lawrence J. White, ed., Private Antitrust Litigation: New Evidence, New Learning. Cambridge, MA: MIT Press, 149–84. Perloff, Jeffrey M., Daniel L. Rubinfeld, and Paul Ruud. 1996. Antitrust Settlements and Trial Outcomes. Review of Economics and Statistics 78: 401–9. Polinsky, A. Mitchell, and Daniel L. Rubinfeld. 2002. A Note on Settlements under the Contingent Fee Method of Compensating Lawyers. International Review of Law and Economics 22: 217–25. Polinsky, A. Mitchell, and Daniel L. Rubinfeld. 2007. A Damage-Revelation Rationale for Coupon Remedies. Journal of Law, Economics, & Organizations 23: 653–61. Polinsky, A. Mitchell, and Daniel L. Rubinfeld. 2008. The Deadweight Loss of Coupon Remedies for Price Overcharges. Journal of Industrial Economics 56: 402–17. Priest, George L., and Benjamin Klein. 1984. The Selection of Disputes for Litigation. Journal of Legal Studies 13: 1–55. Reinganum, Jennifer F., and Louis L. Wilde. 1986. Settlement, Litigation, and the Allocation of Litigation Costs. Rand Journal of Economics 17: 557–66. Schweizer, Urs. 1989. Litigation and Settlement under Two-Sided Incomplete Information. Review of Economic Studies 56: 163–77. Spagnolo, Giancarlo. 2008. Leniency and Whistle-blowers in Antitrust. In P. Buccirossi, ed., Handbook of Antitrust Economics. Cambridge, MA: MIT Press.

186   Daniel L. Rubinfeld Spier, Kathryn E. 1992. The Dynamics of Pretrial Negotiation. Review of Economic Studies 59: 93–108. Spier, Kathryn E. 2003. The Use of “Most-Favored Nation” Clauses in Settlement of Litigation. Rand Journal of Economics 34: 78–95. Spier, Kathryn E. 2007. Litigation. In A. Mitchell Polinsky and Steven Shavell, eds., The Handbook of Law and Economics, 262–342. Boston: Elsevier. Wils, Wouter P. J. 2009. The Use of Settlements in Public Antitrust Enforcement: Objectives and Principles. In C. D. Ehlermann and M. Marquis, eds., European Competition Law Annual 2008: Settlements under EC Competition Law. Oxford: Hart, 27–46. Wood, Diane P. 2010. Antitrust Settlements in the United States. In C. D. Ehlermann and M. Marquis, eds., European Competition Law Annual:  2008:  Antitrust Settlements under EC Competition Law. Oxford: Hart, 277–304.

CHAPTER 8

T H E E C O N OM IC S O F A N T I T RU ST C L A S S AC T ION S RO GER D. BLAIR AND CHRISTINE PIET TE DURRANCE

8.1. Introduction When ostensible competitors agree to refrain from competing, that is, when they agree to restrain trade, they commit a felony as their conduct violates Section 1 of the Sherman Act (15 U.S.C. §1). There are criminal sanctions for violating Section 1 that include potentially substantial fines (up to $1 million for individuals and up to $100 million for corporations) and possibly a stiff prison sentence (up to 10 years). To the extent that the unlawful conduct has harmed others, the conspirators have also committed a tort. Aggrieved parties may sue for damages under Section 4 of the Clayton Act (15 U.S.C. §15). Section 4 provides for the recovery of treble damages plus a reasonable attorney’s fee for the successful plaintiff. This provision for private damage actions serves two functions; first, it compensates the antitrust victims, and second, it deters antitrust violators. If each injury is substantial, an individual antitrust victim can sue because the potential award is sufficient to justify the litigation expense. But there are cases in which the total harm is quite large while the individual injuries may be too small to be worth pursuing on an individual basis. Under some circumstances, these small injuries may be pursued collectively in a class action. In this chapter, we examine the economics of antitrust class actions. In section 8.2, we examine the economic rationale for antitrust class actions. In section 8.3, we turn our attention to economic issues that arise in class certification. In particular, we discuss the need for and the difficulty of common proof of impact and damages. In section 8.4, we address possible abuses in filing, litigating, and settling antitrust class actions. We include here a brief look at the Class Action Fairness Act. Finally, in section 8.5, we close with some concluding remarks.

188   Roger D. Blair and Christine Piette Durrance

8.2.  The Economic Rationale for Antitrust Class Actions Section 4 of the Clayton Act provides a private right of action for victims of antitrust violations:1 [A]‌ny person who shall be injured in his business or property by reason of anything forbidden in the antitrust laws may sure therefore . . . and shall recover threefold the damages by him sustained and the cost of suit including a reasonable attorney’s fee. (15 U.S.C. §15)

Thus, Section 4 appears to provide compensation for all antitrust victims.2 In addition, it appears to provide a powerful deterrent. In some cases, however, appearances are deceiving.3 In some instances, an antitrust violation may have many victims with individually small damages. For example, suppose that a price-fixing conspiracy among gasoline retailers raises the price at the pump by a nickel. If the average motorist buys 20 gallons a week, the overcharge per motorist will be $52 per year. Each individual cannot sensibly sue to recover $52. Even after trebling the damages over the four-year statute-oflimitations period, it would only amount to $624. The total overcharges, however, would be massive since there are over 100 million motorists. In such cases, Section 4 loses its compensatory and deterrent functions. Class actions, however, may come to the rescue. An antitrust class action consolidates the damage claims of the entire class, which may well be large enough to pursue. If so, this will preserve the compensatory and deterrent functions of Section 4 of the Clayton Act. In our gasoline example, compensation would seem to be most important for the class members, but deterrence is economically more important. To the extent that the threat of private damages deters antitrust violations, substantial consumer injury will be avoided. Consequently, we will focus next on deterrence.

1  For a forward-looking perspective on antitrust class actions in the European Union (EU) see, e.g., Buccirossi et al. (2012). 2  This, of course, is not strictly true. Cognizable injuries are limited to antitrust injuries; see Brunswick v. Pueblo Bowl-O-Mat Inc., 429 U.S.477 (1977). In addition, only direct purchasers have standing to sue; see Illinois v. Illinois Brick, 431 U.S. 720 (1977). 3  The frequency of private antitrust suits is so numerous, one might wonder at the actual deterrent effect. Between 2000 and 2011, the highest annual number of private antitrust suits recorded was 1,318 suits in 2008. The lowest number of annual suits in that period occurred in 2011 with 475 suits (United States Courts, Judicial Business, tables C-2 and C-2A, http://www.uscourts.gov/Statistics/​ JudicialBusiness).

The Economics of Antitrust Class Actions   189

8.2.1. Deterrence Antitrust violations are economic crimes that are motivated by the quest for unlawful profits. These are neither unwitting nor compulsive crimes; they are calculated and often well-planned clandestine efforts. The economic approach to deterring such crimes is to make them unprofitable. The associated criminal sanctions, fines, and imprisonment are intended to punish and thereby deter antitrust violations, but Section 4 of the Clayton Act obviously contributes to that public effort to deter. Violating the antitrust laws is a risky activity. There is a chance of avoiding detection and prosecution, but there is also a chance of being caught and punished. The key to deterrence is making the crimes unprofitable in expected value terms. In other words, enforcement efforts have to reduce the expected value of the antitrust violation below the value of the firm’s lawful profit.4 Suppose that a prospective antitrust violator will earn profits of Π1 without committing a violation. An antitrust violation will increase profits to Π 2 if the violation is not successfully challenged. Although the damages could be larger, we assume that the single damages ( D ) are equal to the difference between Π1 and Π 2 :

D = Π1 − Π 2 . Now, the expected value of the violation is5



E[Π] = pΠ 2 + (1 − p)(−3D),

where p is the probability of getting away with the violation and (1 − p ) is the probability of a successful suit. If Π1 > E[Π], that is, if the lawful profit is larger than the expected profit of an antitrust violation, the violation will not be profitable on average and, therefore, it will be deterred. On the other hand, if the expected gain exceeds the lawful profit, then the violation will be profitable. Given the current sanctions, the key elements to deterring antitrust crimes are the value of p and the trebling provision.

4  This prescription assumes that the firm is risk neutral. Risk neutrality means that only expected values are relevant to decisions; the variance in outcomes is not relevant. A risk-neutral person will view betting $10 on the flip of a fair coin as equivalent to betting $100 or $1,000. Risk aversion and risk-loving attitudes can be incorporated in the analysis, but do not add much insight. 5  Here we abstract from the litigation costs, which may be considerable.

190   Roger D. Blair and Christine Piette Durrance

8.3. Class Certification: Procedure and Practice Class action litigation requires several procedural steps before litigation can proceed. At some point after a self-designated class representative has filed an antitrust class action, the court must decide whether to certify the class, so the litigation can be conducted as a class action. This is critical because the case usually cannot proceed on an individual basis. In this section, we set out the rules that the court follows in deciding whether to certify the proposed class. We then turn to practical problems that may surface in a class context.

8.3.1.  Federal Rules of Civil Procedure Class certification is governed by Rule 23 of the Federal Rules of Civil Procedure. Rule 23(a) sets out four prerequisites for class certification that must be satisfied by a designated class representative. These are (1) numerosity, which means that the class is so large that joinder is impractical, (2) commonality, which means that there must be at least one question of law or fact that is common to every class member, (3) typicality, which means that the class representative must be a member of the class so that his or her claim is typical of those of the other class members, and (4) adequacy, which requires that the class representative fairly and adequately protect the interests of all class members.6 For our purposes, we will assume that the requirements of Rule 23(a) have been met. Rule 23(b) also has four criteria, but only one of them must be satisfied for class certification.7 The most common antitrust classes are certified under Rule 23(b)(3), which provides for class certification under a predominance standard.8 Generally, Rule 23(b)(3) requires two things. First “questions of law or fact common to the members of the class predominate over any questions affecting only individual members.” Second, a class action must be “superior to other available methods for the fair and efficient adjudication of the controversy.” Satisfying Rule 23(b)(3) can be problematic. Proof of a violation is not sufficient to establish civil liability under Section 4 of the Clayton Act. Moreover, Rule 23(b)(3) requires that the proposed class representative demonstrate that common questions predominate over individual questions. For the most part, this involves liability issues rather than damage issues. But liability— even in a per se case—extends beyond proof of a violation and includes impact, that is, it is necessary to prove that all class members suffered injury to their business or property using common proof.9 To satisfy the predominance requirement, therefore, the proposed class representative must prove that the antitrust violation and the fact of injury can be established 6  Id. at ¶331. For a compact survey with ample citations, see ABA Section of Antitrust Law, Antitrust Law Developments 928–941 (5th ed. 2002). 7  Id. at 941. 8  Id. at 943. 9  See Alabama v. Blue Bird Body Co., 573 F.2d 309, 320 (5th Cir. 1978).

The Economics of Antitrust Class Actions   191

“on a systematic, class-wide basis.”10 For this to be true, proof for one should be proof for all; otherwise individual issues will arise. If the court is faced with the prospect of individualized questions of liability, it would have to conduct multiple minitrials on liability. This would make a class action inappropriate. There is no specific test for determining whether common issues predominate. As a general matter, the best that can be said is that the need for individualized examination of each class member vanishes if there is generalized evidence that either proves or disproves an essential element of the claim on a class-wide basis.11 In many cases, specific counterexamples may prove useful in defeating class certification as they raise the specter of numerous minitrials.

8.4.  Problems in Practice In an antitrust class action, the class representative must establish impact and damages on the basis of common proof. If proof is truly common, then proof for one is proof for all. If common proof is unavailing, then individualized inquires become necessary and the advantages of class treatment disappear. Several recent court decisions have altered the precedent regarding class certification and common proof.12 The courts have begun taking a harder look at the economic evidence available to determine common proof.13 When determining common proof, an economic expert must be able to show the alleged impact on class members using an economic model that avoids individualized inquiries. In prior cases, classes would typically be certified based on a presumption of common proof or expert reports indicating that price-fixing models are conducive to class treatment. But recent trends have required more extensive evidence to support the claim of common proof, including some resolution of the merits. Specifically, the court modified precedent by stating: First, the decision to certify a class calls for findings by the court, not merely a “threshold showing” by a party, that each requirement of Rule 23 is met. Factual determinations supporting Rule 23 findings must be made by a preponderance of the evidence. Second, the court must resolve all factual or legal disputes relevant to class certification, even if they overlap with the merits—including disputes touching on elements of the cause of action. Third, the court’s obligation to consider all relevant evidence and arguments extends to expert testimony, whether offered by a party seeking class certification or by a party opposing it.14 10 

See In re Agricultural Chemicals Antitrust litigation, 1995-2 Trade Cases. (CCH ¶71, 197 (N.D. Fla. 1995). 11  See Vitamins Antitrust Litigation, 209 F.R.D. 251, 262 (D.D.C. 2002). 12  See, e.g., In re Hydrogen Peroxide Litigation (552 F.3d 305 (3d Cir. 2008). 13  These issues are analyzed in several recent papers including Johnson (2011), Johnson and Leonard (2011), Singer (2011), and Johnson, David, and Torelli (2010). 14  552 F.3d 305 (3d Cir. 2008) at 307.

192   Roger D. Blair and Christine Piette Durrance Price

S Pa

Pbf D MR

0

Qa

Qbf

FIGURE  8.1  Hypothetical

Price Fixing Conspiracy

While this instruction is sensible, its precise meaning will only become clear with future application. There are some remaining ambiguities. To illustrate some of the difficulties that may arise in using common proof to prove impact and damages, we consider a hypothetical price fixing conspiracy. In figure 8.1, D represents demand, MR is the associated marginal revenue, and S is the competitive supply. In a textbook world, there is no time dimension, as all sales take place at a single instant, nor is there a spatial dimension, as all sales take place in one spot. In this stylized setting, the competitive price is the price but for the illegal price fixing and is labeled Pbf . The corresponding “but for” quantity is Qbf . Assuming that the cartel successfully maximizes cartel profits, the actual price and quantity will be Pa and Qa , respectively. In this simple world, impact is obvious since Pa exceeds Pbf and everyone pays the same price. Moreover, cognizable damages ( ∆ ) in the aggregate amount to

(

)

∆ = Pa − Pbf Qa .



These damages will be trebled to 3∆. As a result, each class member can submit a claim for his or her proportional share of the award. More specifically, each class member can make a claim for 3 Pa − Pbf q where q is the quantity actually purchased by that particular class member. In this case, damages can be claimed by individual class members with a simple claim form. Since Pa − Pbf is the same for everyone, each class member need only prove that he or she purchased during the class period and the quantity purchased. This, of course, is the easiest possible case for class certification. In practice, however, things are usually more complicated. In the but-for world, price is determined by the forces of supply and demand. Anything that causes supply and/or demand to vary creates a need to estimate additional but-for prices. Actual transactions take place at different places and at different points in time. If we consider a four-year damage period and a product subject to seasonal demands, there will be 16 but-for

(

)

(

)

The Economics of Antitrust Class Actions   193

prices to estimate. Additionally, if there are state-specific factors that influence price, there would then be 50 × 16 or 800 but-for prices to estimate. Over this time frame, one must account for product improvements and changes in market structure due to entry and exit as well as mergers. All of this suggests that estimating the but-for price may be extremely complicated in practice. These complications can make a court uneasy about certifying a class.15 Interestingly, the actual price paid may be even more problematic and far more individualized than one might suppose. For products involving trade-in allowances, the economic value of the actual price paid is the sum of cash plus the fair market value of the item traded in. If the trade-in actually allowed exceeds the fair market value, then there is a hidden discount. This alone introduces an individualized element as each transaction involving a trade-in must be examined. Some customers get free credit, while others may not. But free credit amounts to a price concession.16 There may be off-invoice discounts and year-end rebates that may depend on buyer characteristics (size and location) and negotiating tactics.17 Another way to discount the price is to provide free goods, free delivery, free services, free technical advice, and free warranties. To the extent that these vary across customers, obvious individualized issues arise. If it cannot be shown that such variation does not exist, it would seem that the defendants would be entitled to discovery on this issue.18 This, of course, could involve deposing all of the class members, which would make the case unsuitable for class treatment. The issues raised here are not confined to damages. In a price enhancement case, proof of impact requires a showing that the actual price (Pa ) exceeds the but-for price (Pbf ). If one cannot establish the values of Pa and Pbf with common proof, one cannot prove impact either.

8.4.1.  Indirect Purchaser Actions To further complicate damages in antitrust class action claims, consider a potential class of indirect purchasers. In a pair of decisions—Hanover Shoe19 and Illinois Brick20— the Supreme Court made it clear that only direct purchasers have standing to sue for 15  In Blades v. Monsanto Co., 400 F. 3d 562 (8th Cir. 2005), the court denied class certification because of the localized nature of the markets for genetically modified corn and soybean seeds that exhibited substantial price variation. 16  See Catalano, Inc. v. Target Sales, Inc., 446 U.S. 643 (1980). 17  When transaction prices are negotiated, the actual price paid will be determined at least in part by the negotiating styles of the customers. As a result, proof of antitrust injury is bound to be individualized. See, e.g., Robinson v. Texas Automobile Dealers Association, 387 F. 3d 416, 423 (5th Cir. 2004). 18  Common proof may not exist when the defendants’ rebuttal evidence raises individualized questions. For example, in Rodney v. Northwest Airlines, Inc. 2005-2 Trade Cas. (CCH) ¶ 74, 940 (6th Cir. 2005), the court noted that each of Northwest’s 74 routes could be a separate market, which would require individualized evidence. 19  Hanover Shoe v. United Shoe Machinery Corp., 392 U.S. 481 (1968). 20  Illinois Brick v. Illinois, 431 U.S. 720 (1977).

194   Roger D. Blair and Christine Piette Durrance overcharge damages under Section 4 of the Clayton Act (except under the cost-plus exception),21 although some state statutes allow indirect purchasers to bring actions.22 Similar problems as described above exist for both direct and indirect purchaser classes. For the indirect purchaser class, however, the issue of pass through exists. In a price-fixing case, the colluding firms raise their price above the competitive level, thereby “overcharging” their customers. These customers are the direct purchasers while their customers are the indirect purchasers. Indirect purchasers are injured only to the extent that some portion of the direct overcharge is passed on in the form of higher prices. The indirect purchasers, therefore, must show the extent of the pass through, which will be influenced by supply and demand elasticities as well as market structure. Indeed, the overcharge can be very complicated. Consider the additional complication when the product subject to price fixing changes form following its sale. For example, the manufacturers of lysine—an animal feed additive—fixed prices.23 The lysine was sold at collusive prices to producers of animal feed, who suffered the initial overcharge. The feed was subsequently sold to farmers at prices that reflected the increased lysine cost. The farmers sold cattle to meat packers at prices that reflected the higher cost of feed due to the higher costs of lysine. It is clear that the meat packers’ prices to retail grocers and restaurants reflected their higher costs. Finally, the ultimate consumer was overcharged at the grocery store or the restaurant due to the lysine cartel. It is obvious that there will be severe problems of proof along this chain from the lysine producer to the ultimate consumer of beef. At each step, one will be faced with estimating but-for prices. It will become increasingly difficult to prove the extent of any net overcharge as one moves away from the site of the conspiracy.24 This is the sort of problem that the Supreme Court sought to avoid with its Hanover Shoe and Illinois Brick decisions.

8.4.2.  Prevalence of Antitrust Class Action The standards for class certification appear to be demanding. The existence of common proof of impact and damages may be problematic. Nonetheless, there are numerous class action suits filed each year. Table 8.1 summarizes the number of antitrust class actions filed each year from 2000 through 2010.

21  There is an exception to Illinois Brick for preexisting, fixed-quantity, cost-plus contracts. When the direct purchaser has a preexisting contract with its customer(s) that the price that it will charge equals the cost of the good being resold plus an agreed-upon markup, any overcharge will be passed on in its entirety. 22 In ARC America, the Supreme Court ruled that Illinois Brick only precluded indirect purchasers from suing under federal law; it did not preclude their suing under state antitrust or consumer protection statutes that permit indirect purchasers to sue. 23  See Amino Acid Lysine Antitrust Litigation, 196 WL 355368(N.D. Ill. 1996). See Cotterill, Egan, and Buckhold (2001) for an examination of this case. 24  It should also be fairly clear that the number of potential plaintiffs grows by leaps and bounds as we move away from the site of the lysine conspiracy. This problem would be reduced through class actions, but not all situations are amendable to class treatment.

The Economics of Antitrust Class Actions   195

Table 8.1  Antitrust Class Actions, 2000–2011 Year

Cases Filed

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

407 300 257 204 303 387 600 669 767 376 202 134

During the 2000–2010 period, the average annual number of antitrust class actions filed was 384.25 The number ranged from a minimum of 134 in 2011 to a maximum of 767 in 2008. In figure 8.2, we have plotted the data displayed in table 8.1.26

8.4.3.  Settlements of Antitrust Class Actions Antitrust class actions seldom go to trial. The economic model of litigation reveals why most legal disputes (including antitrust class actions) are resolved through settlement rather than trial.27 For the most part, settlements are efficient, but there is room for abuse because the members of the class are not active participants in the process. In this section, we examine one possible abuse that involves “cash for the lawyers and coupons for the class.” We begin with the basic settlement model and then introduce coupons.28 We then examine the policy response to the possible abuses involving coupon settlements. We start with the heroic assumption that the class and the defendant agree that overcharge damages are equal to a sum that we denote as ∆.29 In the event that the class 25 

It is, of course, important to note that this count overstates the number of antitrust class actions because many are joined. 26  These data are drawn from Lexis Nexis, CourtLink, from January 1, 2000, to November 29, 2011. It is possible that some of this recent decline is attributable to the effects of the Class Action Fairness Act (see e.g., Sherman 2005–6). 27  This model can be traced to Landes (1971) and Gould (1973). For a summary of these and subsequent results, see Miceli (1997, 156–70). 28  Several articles specifically consider coupons in class action settlements. See Miller and Singer (1997); Leslie (2002); and Note: In-Kind Class Action Settlements, Harvard Law Review 109 (1996): 810–27. 29  In most antitrust cases, there is considerable dispute regarding the damage estimate. Such disagreements can be incorporated into the model. Doing so, however, introduces complications that are not relevant to the issues that we examine here.

196   Roger D. Blair and Christine Piette Durrance Cases Filed 2000–2011

900 800 700 600 500 400 300 200 100 0

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 FIGURE  8.2  Change

in the Number of Class Action Cases Filed Over the 2000–2011 Period

wins at trial, the damage award will be trebled to 3∆ . In addition, the plaintiff class will receive its costs (CP ) .30 Victory, however, is never certain. Let p represent the subjective probability that the class attaches to its winning and (1 − p) is the subjective probability of losing. If the class loses, it receives nothing, but incurs its litigation costs. To the class, the net expected value31 of the suit can be represented as E[award] = p(3∆ + CP ) − CP ,



where E is the expectations operator.32 In a very real sense, the suit is simply an asset with an expected dollar value equal to the expected net award of p(3∆ + CP ) − CP . The class owns this asset and can “sell” it to the defendant through a settlement. The class presumably will not settle (i.e., sell) for less than the net expected value of the suit.33 In economic terms, this sum represents its reservation price. 30 

Section 4 of the Clayton Act, 15 U.S.C. §15, provides for treble damages plus the cost of suit including a reasonable attorney’s fee. The trebling by the court is a ministerial function; there is no discretion. 31  Expected values are probability weighted averages: each possible outcome is multiplied by the probability of that outcome and then summed. 32  If the class wins the suit, it will receive treble damages (3∆) plus its costs, and the probability of that is p. If it loses, with probability (1 − p), it receives nothing. The litigation costs (C p ) are actually incurred in either event. Thus, the expected outcome is E[ A] = p(3∆ + CP ) + (1 − p)(0) − C p or p(3∆ + CP ) − CP , which is what we have in the text. 33  This assumes that the class is risk neutral. If the class members were risk averse, they would be willing to settle for something less than the expected net award. In what follows, we assume that both the class and the defendant are risk neutral. On attitudes toward risk and their implications for decision making, see Blair and Kenny (1987, 385–408).

The Economics of Antitrust Class Actions   197

The defendant’s exposure is equal to the trebled damages (3∆) plus the cost of the suit (CP ) plus its own litigation costs (CP ). But these sums must be adjusted to reflect the fact that a loss is not certain. In determining whether to settle, the defendant will consider the expected loss of going to trial, which will be less than the exposure. Accordingly, the defendant will not be interested in “buying” the suit from the class unless the defendant values the suit more than the class values it. For the defendant, there is no upside: the value of the suit is the expected loss associated with going to trial, which is what the defendant avoids in a settlement. The defendant’s expected loss of going to trial can be represented as

E[loss] = q(3∆ + CP ) + CD ,

where q is the defendant’s subjective assessment of the probability of losing and CD denotes the defendant’s legal costs, which he bears whether he wins or loses. For a settlement to occur, the value of the suit to the defendant must exceed the value to the class; otherwise the defendant will be priced out of the market, that is, it will not “buy” out the class’s ownership interest in the suit. In other words, a settlement will occur if the expected loss to the defendant exceeds the expected award to the class. Mathematically, settlement requires that

E[loss] > E[award]

or

q(3∆ + CP ) + CD > p(3∆ + CP ) − CP .

Collecting these terms on the left-hand side, we see that a settlement will occur when

(q − p)(3∆ + CP ) + CD + CP > 0.

This condition will always hold if discovery and legal precedent cause the convergence of the subjective probabilities, that is, if q = p. When the class’s assessment of its chances of winning equals the defendant’s assessment of the class’s chances of winning, the first term on the left-hand side is zero. Since the sum of the litigation costs must necessarily be positive, the condition for a settlement is satisfied. To illustrate this point, suppose that the damages (∆) are $15 million and the anticipated costs of litigation (C p ) are $5 million.34 If the class believes that its probability of winning is .8, then we can find the expected value of the suit by substitution:

E[award] = (.8)($45 + $5) − $5,

34  These figures as well as the rest of the numbers used in our numerical example are purely hypothetical and are used for illustrative purposes only.

198   Roger D. Blair and Christine Piette Durrance which equals $35 million. Since the class expects to receive $35 million by going to trial, it will not take less than that to avoid a trial. Now assume that the defendant agrees that the damages are $15 million. Also assume that the defendant believes that the probability of its losing is .8 (i.e.,) and that its costs of litigation will also be $5 million. Now, the expected loss associated with a trial will be

E[loss] = (.8)($45 + $5) + $5,

which is $45 million. This sum represents the maximum price that the defendant will pay to buy the class’s right to go to trial. Since the class will take anything above $35 million and the defendant will pay anything below $45 million, we should expect a settlement. In this example, we have assumed that p = q = .8. As a result, we have

(.8 − .8)($45 + $5) + $5 + $5 > 0.

Since $10 million in litigation costs can be avoided by settling, it is economically efficient to settle. This is the economic explanation for why the vast majority of cases settle. Settlement talks can break down if the parties are too optimistic about their chances of winning at trial. In the model, optimism is captured by a high value of p for the class and a low value of q for the defendant. There will be no settlement if the class’s assessment of p exceeds the defendant’s assessment of q by enough to swamp the combined litigation costs. In other words, if p is sufficiently larger than q, the expected loss to the defendant will be less than the expected award to the class. As a result, the value of the asset (i.e., the suit) is greater for the class than for the defendant and, therefore, the defendant will not be willing to buy it. Suppose, as above, that 3∆ + CP equals $50 million and that CP = CD = $5 million, but that p = .85 while q = .6. Thus, the class believes that the relative likelihood of its winning is .85 while the defendant believes that its chances of losing are only.6. In this event, the expected value to the class is

E[award] = (.85)($50) − $5, ,

which is $37.5 million, while the expected loss to the defendant is

E[loss] = (.6)($50) + $5,

which is $35  million. Since the maximum amount that the defendant is willing to pay ($35 million) is less than the minimum amount that the class is willing to accept ($37.5 million), no settlement will occur.35 The beliefs of one of the parties will be vindicated at trial.

35  The reasons for the disparity in the subjective probabilities of success are indicative of a discovery failure or of inherent optimism. We do not explore the reasons for the disparity here.

The Economics of Antitrust Class Actions   199

As seen in the economic model of litigation, a decision to settle depends on the relative probabilities of success as well as beliefs about damages and costs of litigation. A defendant may choose to settle if he can avoid additional costs associated with a trial or because the probability of losing is high. A decision to settle, therefore, does not mean that the defendant is necessarily guilty. Rather, a defendant may find it in his financial interest to avoid a trial because the cost of settling is less than the expected loss of a trial. Additionally, the size of a settlement depends on the same factors. If both p and q are small, the expected value of litigation for both the plaintiff and the defendant will be low relative to the trebled damages undiscounted by the probability of success. If the suit is pursued, the range of settlement values available will also be low. While class action settlements are often criticized for undercompensating class members, a small settlement does not necessarily signal abuse of the class. For example, suppose that the damages are $100 million and litigation costs are $10 million for each side. If the liability case is weak, p and q will be small, say, 10  percent. Then E[award] = (.10)($300 + 10) − 10, which is $21  million. The E[loss] = (.10)($300 + 10) + 10, which is $41  million. Thus, a settlement of, say, $31 million is a good deal for the plaintiffs even though it is a far cry from $310 million.

8.4.4.  Cash for the Lawyers, Coupons for the Class In some instances, antitrust class actions are settled with a combination of cash for the class attorneys and coupons that provide discounts on future purchases for the class members. When settlement talks fail because the parties are too optimistic (i.e., when p is sufficiently greater than q), coupons may come to the rescue. First, we should recognize that nearly all antitrust class members are largely passive participants in the litigation process.36 The active participants are the class attorneys, who really make the decisions on behalf of the class. If the class lawyers are satisfied, then a settlement may be reached. The trick is to figure out how to do that when the condition for a settlement does not appear to be satisfied. One way is to appeal to the economic interest of class counsel, which flows from the fees that the court will award. There are two bases for the fee award: (1) the lodestar, which depends upon the hours expended, the risks undertaken by the attorneys, the tasks performed, and so on, and (2) a share of the recovery. We focus here on the share-of-recovery method, which means that the case is subject to a contingent attorneys’ fee. If the defendant prevails, the class counsel will get nothing. If the class prevails, class counsel will receive a share of the recovery (α) . For

36  In principle, the named class representative is there to participate on behalf of the entire class. In practice, however, the class representative is apt to be an inexperienced litigant and, therefore, rely heavily on the class attorneys for advice.

200   Roger D. Blair and Christine Piette Durrance simplicity, we shall assume that α is applied to the trebled damages (3∆) and awarded separately.37 In addition, we ignore other litigation costs (which can be substantial) to reduce notational clutter. As a result, the expected award will be p(3∆) + αp(3∆)  or

E[ A] = (1 + α) p(3∆).

The sum that is expected to go to the class members is p(3∆) and the sum expected to go to the class counsel is αp(3∆). A proposed settlement will look good to risk neutral class counsel provided that there is enough cash to cover the expected attorneys’ fees:  αp(3∆). Thus, the defendant must put together a settlement package that will generate a fee award of αp(3∆) in order to settle the case. Suppose a creative offer were made by the defendant: cash for the lawyers of αp(3∆) and coupons with a nominal value of p(3∆) for the class members. If the redemption rate (r) on the coupons is less than 100 percent, this offer may lead to a settlement (assuming that the court approves it). The class lawyers are no worse off financially than they would be by going to trial. As long as the defendant’s expected loss is greater than the actual value of the cash-coupon offer, the defendant will want to settle. In other words, if

(1 + α)q(3∆) > α( p(3∆)) + r ( p(3∆)),

then the defendant will want to settle for anything less than that amount. The defendant will then be better off than it would be by going to trial. The class members, of course, do not receive p(3∆) because not all coupons are redeemed, that is, r < 1. In the extreme, the coupons could be essentially worthless, the redemption rate would then be zero, and the class members would get nothing. In that event, the class lawyers would get the cash and the defendant would have settled for much less than the class’s expected net recovery at trial. In effect, the class lawyers would get 100 percent of the actual recovery. In our previous numerical example, we assumed that 3∆ = $45 million, p = .85, and q = .6. Continuing to ignore the other litigation costs, the expected damage award to the class at trial, therefore, is $38.25 million. If the class lawyers anticipate receiving a fee award equal to 20 percent of the recovery (α = 0.20) , then the class lawyers expect a fee award of (0.20)($38.25 million), which equals $7.65 million. If the defendant is prepared to pay $7.65 million in cash plus coupons worth $38.25 million in nominal terms, the class lawyers’ fee will exhaust the cash. The defendant’s maximum settlement offer is (1 + α)(q3D) , which is equal to (1.2)(.6)($45) or $32.4 million because it is less

37  In the event of a small judgment, class counsel may request a lodestar fee from the court. Here, we assume that class counsel will opt for a share of the recovery that the court will determine, i.e., α is dictated by the court taking into account effort and risks.

The Economics of Antitrust Class Actions   201

convinced that the plaintiff will win. The real value of the settlement offer will be no more than $32.4 million if the coupon redemption rate is less than 64.7 percent.38 In that event, the class members will actually receive $24.75 million in benefits rather than the nominal value of $38.25 million. The class counsel then receive a fee equal to 30.9 percent of the real value of the coupons rather than 20 percent. But things could be much worse for the class and much better for the defendant. Suppose that the redemption rate were 20 percent (r = .20). In that event, the actual value of the class compensation is only $7.65 million. The class lawyers still get $7.65 million in fees. Thus, the lawyers get 50 percent of the actual value of the coupons. The defendant is happy because the actual cost to settle is only $15.3 million.

8.5.  The Class Action Fairness Act The Class Action Fairness Act (hereinafter CAFA or the Act) of 2005 was enacted because of the challenges outlined in the previous section.39 The act modified various components of class action litigation, while protecting class members and ensuring fair settlement awards. In short, the authors of the act addressed several apparent problems with the then current system. Of particular concern was the fact that “class members often receive little or no benefit from class actions.”40 In many instances, coupons appeared to be the culprit when “counsel are awarded large fees, while leaving class members with coupons . . . of little or no value.”41 The act, therefore, was intended to “assure fair and prompt recoveries for class members with legitimate claims” as well as “restore the intent of the framers of the United States Constitution by providing Federal Court consideration of interstate cases of national importance under diversity jurisdiction”; and “benefit society by encouraging innovation and lowering consumer prices.”42 With respect to coupon settlements, the act has two main provisions. First, the settlement must be reviewed by the court and found to be “fair, adequate, and reasonable.”43 Second, the method for calculating attorneys’ fees in class action settlements has been modified. The act states that “[i]‌f a proposed settlement in a class action provides recovery of coupons to a class member, the portion of any attorneys’ fee award to class counsel that is attributable to the award of the coupons shall be based on the value to class members of the coupons that are redeemed.”44 38 

Actual redemption rates are often far less than 64.7 percent. This critical value is an artifact of our numerical example. 39  Class Action Fairness Act of 2005, 2005 Bill Tracking S. 5., Congressional Information Service, Inc., LexisNexisTM Congressional, http://web.lexis-nexis.com.lp.hscl.ufl.edu/congcomp. 40  28 U.S.C. §1711 (a)(3). 41  28 U.S.C. §1711 (a)(3)(A). 42  28 U.S.C. §1711 (b). 43  28 U.S.C. §1712 (e). 44  28 U.S.C. §1712 (a).

202   Roger D. Blair and Christine Piette Durrance When a settlement involves coupons, the act requires the court to base its fee award on the actual value of the recovery. Thus, the fee depends on the value of the coupons actually redeemed rather than the nominal value of the coupons issued. For the class lawyers to get all of a fee award of αp(3D) the value of the nominal coupons must be increased enough so that the coupons actually redeemed are worth p(3D) . Otherwise, the court will not award a fee of α( p(3D)) to the class lawyers and they will not agree to settle because their expected fee is higher if they go to trial and get cash for the class. Due to the coupon rule, therefore, some settlements will not be reached that might have been reached before the act was passed. Suppose that the coupon redemption rate turned out to be 50 percent in our numerical example. As a result, the actual value of the coupons would be half of the nominal value or $19.125 million (redeemed coupons). If the Court awarded 20 percent of the true value of the coupons to the class counsel, they would receive only $3.825 million rather than $7.65 million. This, of course, is not inappropriate since the lawyers are being compensated on the basis of the total actual recovery rather than some fanciful figure. When the court uses the redeemed value of the coupons, it heightens the lawyers’ sensitivity to these values. This better aligns the interests of the lawyers and the class members.

8.6.  Attorney Compensation Depending on the redemption rate, the value of the case to the class attorneys may be lower under CAFA. Redemption rates are typically very low for a variety of reasons.45 Assume, for example, that $50 million in coupons are issued to the class as part of a settlement. If the expected redemption rate is 10 percent, then the coupons are worth $5 million. The attorneys’ fees now must be calculated as a portion of the $5 million redeemed, not the $50 million issued to the class. This will have an obvious effect on the willingness of the class lawyers to file suits or agree to settle cases once they are filed. CAFA also directly impacts the timing and structure of attorneys’ fees. As a result of the coupon provision, the collection of attorneys’ fees is delayed until the close of the redemption period, when the actual redemption rate can be ascertained. The delay in receiving the fee can have a substantial effect on the value to the attorneys. For example, suppose that the fee awarded is $10 million. If the appropriate interest rate is, say, 10 percent, the present value of the fee is $9.09 million if payment is delayed one year. If it is delayed for two years, the present value drops even further to $8.26 million.46 At the time

45  Settlements often consist of “coupons or other awards of little or no value” or “confusing notices are published that prevent class members from being able to fully understand and effectively exercise their rights” (28 U.S.C. §1711(a) (3)). 46  The present value of a dollar to be received one year from now at an interest rate of 10 percent is $1.00 / 1.10 = $0.909. The present value at the end of two years is $1.00 / (1.10)2 = $0.826. This arithmetic can be applied to the hypothetical $10 million fee in the text.

The Economics of Antitrust Class Actions   203

the settlement is proposed and until the redemption period ends, the attorneys’ portion is subject to a great deal of uncertainty. Since the fee award is conditional upon the value of the coupons actually redeemed, the attorneys’ fees at the time of settlement are expected values, dependent upon the realized redemption rate. Restricting attorneys’ fees to a portion of the coupons actually redeemed could shift counsel toward using the lodestar method.47 Because of the uncertainty of the redemption rate as well as the delay in payment, counsel could prefer the lodestar calculation. In addition, it is also possible that class counsel may delay settlements in order to increase the amount of time and resources dedicated to the case, in order to increase their potential fee.48

8.7.  Concluding Remarks Class action litigation is motivated by goals of both compensation to victims and deterrence. Common proof of impact and damages is difficult to prove, and some recent changes in precedent have provided additional barriers. The economic model of litigation reveals why most disputes settle, but coupons in class actions are a special case. Coupon settlements have been criticized for handsomely compensating class counsel, but leaving little, if anything, for the class members. The 2005 Class Action Fairness Act was passed in an attempt to prevent some of the abuses inherent in class actions. We conclude that class actions can provide a powerful deterrent to potential antitrust violations, even if some cases will not settle that would have settled previously. Aligning the interests of the attorneys and class members is an important component of the compensation goal of class actions.

Acknowledgments We thank our institutions for financial support. This work relies on some of our previous work, including Blair and Piette (2005), Blair and Piette (2006), Blair and Piette Durrance (2007), and Areeda et al. (2007).

47  This, of course, depends upon the alternatives. Presumably, the class counsel will ask the court for a lodestar award when it expects this to be larger than a share of the recovery. 48  See Casper (2005). Whether class counsel has an incentive to delay depends upon its opportunity costs. Hours spent on a class action that does not appear to be very lucrative are hours that cannot be spent on other matters.

204   Roger D. Blair and Christine Piette Durrance

References ABA Section of Antitrust Law Civil Practice and Procedure Committee. 2005. The Antitrust Practitioner 2 (June). Arreda, Phillip E., Herbert Hovenkamp, Roger D. Blair, and Christine Piette Durrance. 2007. Antitrust Law Par. 392. New York: Aspen Publishers. Blair Roger D., and Lawrence W.  Kenny. 1987. Microeconomics with Business Applications. New York: Wiley. Blair, Roger D., and Christine A.  Piette. 2005. Coupons and Settlements in Antitrust Class Actions. Antitrust 20 (Fall): 32–37. Blair, Roger D., and Christine A.  Piette. 2006. Coupon Settlements:  Compensation and Deterrence. Antitrust Bulletin 51: 661–85. Blair, Roger D., and Christine Piette Durrance. Summer 2007. The Economic Pitfalls in Antitrust Class Certification. Antitrust 21: 69–73. Buccirossi, Paolo, Michele Carpagnano, Lorenzo Ciari, Massimo Tognoni, and Cristiana Vitale. 2012. Collective Redress in Antitrust. http://www.learlab.com/pdf/est74351_1340289300.pdf. Casper, Charles B. 2005. Settlements under the Class Action Fairness Act. Antitrust Practitioner 2: 2. Cotterill, R., L. Egan, and W. Buckhold. 2001. Beyond Illinois Brick: The Law and Economics of Cost Pass-Through in the ADM Price Fixing Case. Review of Industrial Organization 18: 45–52. Gould, John. 1973. The Economics of Legal Conflicts. Journal of Legal Studies 2: 279–300. Johnson, John A., Jesse David, and Paul A. Torelli. 2010. Empirical Evidence and Class Certification in Labor Market Antitrust Cases. Antitrust 25(1): 60–63. Johnson, John H., and Gregory K. Leonard. 2011. Rigorous Analysis of Class Certification Comes of Age. Antitrust Law Journal 77: 569–86. Johnson, Paul A. 2011. The Economics of Common Impact in Antitrust Class  Certification. Antitrust Law Journal 77: 533–67. Landes, William M. 1971. An Economic Analysis of the Courts. Journal of Law and Economics 14: 61–108. Leslie, Christopher R. April 2002. A Market-Based Approach to Coupon Settlements in Antitrust and Consumer Class Action Litigation. UCLA Law Review 49: 991–1098. Leslie, Christopher R. Fall 2005. The Need to Study Coupon Settlements in Class Action Litigation. Georgetown Journal of Legal Ethics 18: 1395–418. Miceli, Thomas J. 1997. Economics of the Law:  Torts, Contracts, Property, Litigation. New York: Oxford University Press. Miller, Geoffrey P., and Lori S. Singer. 1997. Nonpecuniary Class Action Settlements. Law and Contemporary Problems 60: 97–154. Sherman, Edward. 2005–6. Class Actions after the Class Action Fairness Act of 2005. Tulane Law Review 80: 1593–616. Singer, Hal J. 2011. Economic Evidence of Common Impact for Class Certification in Antitrust Cases: A Two-Step Analysis. Antitrust 25(3): 34–39.

CHAPTER 9

B E HAV IO R A L E C O N OM IC S A N D A N T I T RU ST MARK ARMSTRONG AND STEFFEN HUCK

9.1. Introduction IN this survey chapter we discuss nonstandard, or “behavioral,” approaches to firm behavior. We present evidence that firms (or experimental subjects playing the role of firms) sometimes depart from the profit-maximizing paradigm. For instance: firms may be content to achieve “satisfactory” rather than optimal profits; firms might rely on simple rules of thumb—such as imitating the strategies of well-performing rivals, or changing strategies only when profit falls below some acceptable threshold—rather than on explicit calculation of complex optimal strategies; firms may base pricing decisions on sunk costs as well as avoidable costs; CEOs may be overly optimistic about the profitability of mergers or other actions they undertake; managers might face incentives that induce them to care about relative rather than absolute profits; firms might punish rivals who behave “unfairly” toward them; and so on. There are a number of reasons why one might expect firms to be better decision-makers than consumers, and this helps to explain the recent focus of much policy on consumer failings.1 First, there are economies of scale in making good decisions. A consumer may have to decide whether to buy a given product just a few times and it may not be worthwhile to invest much effort in making the right decision, while a firm selling to millions of customers has more at stake in getting it right. Relatedly, since 1  See Armstrong (2008) for a review of some of this literature and its implications for consumer protection policy. As observed by Ellison (2006, 145), fifty years ago the focus was instead very much on nonoptimizing behavior by firms rather than consumers.

206   Mark Armstrong and Steffen Huck firms often do the same things repeatedly, they may quickly learn how to do it right, while a consumer buying a rarely purchased product may not have that opportunity. Second, firms compete with one another, while consumers usually do not, and firms that are better at generating profits may succeed and prosper at the expense of firms that make worse decisions. By contrast, consumers rarely “exit” when they make poor market decisions. However, there are a number of considerations that push the other way: • Firms typically operate within a highly complex and uncertain environment, and may need to resort to decision-making shortcuts and rules of thumb. • Firms usually face the added complexity stemming from strategic interaction with rivals, which consumers rarely do when buying products.2 For instance, in even the simplest situations, theories of tacit collusion or reputation-building require firms to follow highly intricate strategies. • In practice, underperforming firms may take a long time to exit, and there is often a long period of decline before an established firm actually leaves the market. More generally, the complexity of the environment may mean that crucial decisions are taken with significant delay.3 • Group decision-making, which is practiced more often by firms than by consumers, could introduce extra biases. For instance, the separation of ownership from control could leave managers free to pursue their own objectives that may differ from maximizing shareholder value. In fact, this last point can be turned on its head: In imperfectly competitive markets, shareholder returns might be enhanced by (deliberately or not) hiring managers whose objectives differ from maximization of profit. That is to say, although it appears paradoxical, actual profits might be enhanced when a firm’s objective departs from profit-maximization. For instance, hiring an “aggressive” or “overoptimistic” CEO, rewarding a CEO based on her performance relative to peers, or employing a CEO who has been trained to base pricing decisions on sunk costs, might all have strategic benefits.4 Alternatively, rather than affecting rival responses, explicitly nonprofit aims might act to boost consumer demand, as appears often to be the case with “fair trade” or “green” products, for instance.

2  Hayek (1945) argued that a central merit of competitive markets, that is, nonstrategic price-taking behavior by consumers and firms, is that agents’ strategies are then relatively simple. 3  The take-up of technological improvements in agriculture provides useful evidence. Ellison and Fudenberg (1993) quote a historian of the English agricultural revolution as writing that “land tilled in very ancient ways lay next to fields in which crop rotations were followed.” They report that the agricultural practices known as the “new husbandry” diffused through England and France at the rate of just one mile per year. 4  Stackelberg’s (1934) model is nowadays interpreted as a model in which one oligopolist “moves first” and commits to this action, and others follow. However, he also interpreted the leader as a “strong type” who behaved aggressively in the market.

Behavioral Economics and Antitrust   207

• Relatedly, the people who succeed in the tough career competition to manage firms might have these kinds of personality traits more frequently than the general population. Sometimes it seems that managers are motivated in part by personal animosity—or occasionally respect—toward a rival. • Different managers have differing skills, and in particular some managers may be better at thinking strategically than others. More generally, an individual manager’s “style” can be important—for good or ill—for a firm’s performance.5 • A manager may on occasion have a personal interest in the firm’s activities (say, a sports team, newspaper, or fine wine)6 beyond the profit generated, and this may cause a divergence from profit-maximization.7 Family-owned firms may also have interests that depart from profit-maximization; for instance, they might put additional weight on the likelihood that the firm survives to the next generation. • Illegal cartels need to find ways to resolve disagreements about market shares, whether cheating has occurred, and so on, without recourse to legally binding agreements. In such an environment, issues such as building trust and esprit de corps among conspirators are important. • Finally, the potential presence of a “behavioral” type of firm in a market could induce a profit-maximizing firm to mimic irrational behavior. For instance, a firm might wish to gain a reputation to fight entry come what may, in order to deter future entry. Milton Friedman (1953, 22) is perhaps the most often quoted exponent of the pervasive view that competition acts to discipline firms to maximize their profits: Let the apparent determinant of business behavior be anything at all—habitual reaction, random chance or whatnot. Whenever this determinant happens to lead to behavior consistent with rational and informed maximization of returns, the business will prosper and acquire resources with which to expand; whenever it does not, the business will tend to lose resources and can be kept in existence only by the addition of resources from outside. The process of “natural selection” helps to validate the hypothesis [of “rational and informed maximization of returns”]—or, rather, given 5  See Bertrand and Schoar (2003) for an empirical analysis of the importance of “manager-fixed effects.” For instance, they find that managers with an MBA tend to follow strategies that are more aggressive. They find that managers differ in their attitude to mergers, dividend policy, and cost-cutting policy. 6  Apparently, a proportion of wine producers in California do not care purely about the profit they generate, and instead enjoy producing high-quality, expensive wine. (Profit-maximizing wineries tend to offer lower quality wine.) See Scott Morton and Podolny (2002). 7  An interesting example of this was seen in the Genzyme-Novazyme merger-to-monopoly, which was approved by the FTC in 2004. This was a merger of two firms both engaged in R & D for treating a rare disease, where the prime danger from the merger was whether the discovery of a successful treatment would be delayed relative to the duopoly outcome. One factor in the decision was that the proposed CEO of the merged entity had two children with the disease, who may therefore not have wished to delay discovery. See the statement by the then FTC chairman, Timothy Muris, available at www.ftc.gov/os/2004/01/murisgenzymestmt.pdf.

208   Mark Armstrong and Steffen Huck natural selection, acceptance of the hypothesis can be based largely on the judgement that it summarizes appropriately the conditions for survival.

As we discuss in our conclusions, this view has much merit when it comes to many kinds of decision errors and behavioral biases, and some decision problems often exhibited by consumers (such as procrastination, for instance) are less plausible in firms.8 Friedman’s view also applies well to highly competitive markets, where one firm’s action is irrelevant for other firms’ choice of action. Nevertheless, at various points in this survey we will see examples of situations where oligopolistic firms prosper when their objective diverges from maximizing profits, when “behavioral” managers are put in place, or when firms are content to achieve merely a satisfactory profit level. In the remainder of this chapter, we discuss several of these points in greater detail. Section 9.2 discusses the related issues of imitative behavior by firms and concerns for relative (not absolute) profit, both of which often make an oligopolistic market more competitive than the orthodox theory suggests. Other forms of social preferences—a desire for vengeance if a rival “cheats” on a collusive agreement, say—are presented in section 9.3, where we argue that such preferences may help sustain collusive agreements. Various kinds of satisficing behavior are discussed in section 9.4, where we show that imperfectly optimizing firms may actually end up with greater profits than their profit-maximizing counterparts. Section 9.5 collects together discussion of the impact of overoptimism by entrepreneurs and managers, the imitation of irrational behavior by profit-maximizing firms, and the possible benefits of including fixed costs in a firm’s pricing decisions. Concluding comments are contained in section 9.6.

9.2.  Imitative Behavior and Concerns for Relative Profit Rather than each firm laboriously calculating its own optimal strategy, even if that were feasible, it is plausible that firms may sometimes choose to imitate the strategies of their successful peers. As Alchian (1950, 218) put it: [W]‌henever successful enterprises are observed, the elements common to these observable successes will also be associated with success and copied by others in their pursuit of profits or success. . . . What would otherwise appear to be customary

8  Feinberg (1995) argues that social welfare may sometimes be higher when firms’ managers are relatively myopic or short-termist, perhaps because of high managerial turnover or the kinds of incentive schemes they are offered, since collusion thereby becomes harder to sustain. For the same reason, though, it seems plausible that shareholders who wish to achieve collusion would not wish to put in place a manager who was myopic or had hyperbolic time preferences.

Behavioral Economics and Antitrust   209

“orthodox,” non-rational rules of behavior turns out to be codified imitations of observed success, e.g., “conventional” markup, . . . “orthodox” accounting and operating ratios, “proper” advertising policy, etc.

Alchian suggests that imitation enables firms to make use of other firms’ private information and optimizing behavior, and to enjoy the benefits of conformity (as imitating firms are likely to do as well as their peers on average). To discuss this issue in concrete terms, suppose that firms are rivals operating in the same market.9 In practice, oligopolists may have little idea of the consumer demand function, or how closely substitutable their rivals’ products are with their own. Nevertheless, they may observe their rivals’ actions and realized profits. When oligopolists imitate the most profitable actions observed in the market, it is possible that the market evolves over time to a highly competitive outcome (more competitive than the one-shot Nash equilibrium). Thus, when this particular behavioral bias is present in a market, the market performs better than the standard profit-maximizing analysis would indicate. To see this, consider a textbook Cournot market where several firms with identical constant marginal costs compete to supply a homogenous product. Suppose, whenever they have the opportunity to change their output, that firms imitate the output choice of the most profitable firm of the previous period. Then, when the market price is above cost, the most profitable firm will be the one with the largest output. Hence, firms with low output will increase their output, imitating the profitable firm, which pushes the price down. (If price is below cost, the most profitable firm is the one with the smallest output, and so imitation will then drive prices up.) Thus, imitation pushes prices toward cost and the market evolves toward the perfectly competitive outcome where price equals marginal cost. In sum, when firms myopically imitate the most profitable strategy, the industry as a whole moves to an unprofitable, highly competitive outcome. See Vega-Redondo (1997) for more details of this model, as well as Schaffer (1989) for a related analysis.10 Markets are usually better modeled as firms choosing prices rather than quantities, and where firms offer differentiated products. In these markets, imitation can also induce firms to compete aggressively, although not to the extent of perfect competition as seen in the (homogeneous product) Cournot case. To illustrate, consider a duopoly with firms labeled 1 and 2, where if the two prices are p1 and p2 then firm 1’s demand is

q1 = 1 + p2 / 2 − p1



(9.1)

9  Ridley (2008) analyzes a model in which a second firm sometimes decides to enter a market only if its rival has first entered, in order to save on the costs of acquiring its own market information. He provides some anecdotes about how competitors of McDonald’s often locate near to a new McDonald’s franchise, and he quotes a manager of a coffee shop chain as saying: “The reason we want to open across the street from every Starbucks is they do a great job at finding good locations.” 10  It is important that firms experiment occasionally; otherwise the process will grind to a halt after one period when all firm imitate the most profitable firm in the first period, and then all actions are unchanged thereafter.

210   Mark Armstrong and Steffen Huck (and similarly for firm 2). For simplicity, suppose costs are normalized to zero. Then profit-maximizing firms are predicted to choose the Bertrand equilibrium prices p1 = p2 = 2 / 3, and each firm sells quantity 2/3. One can check that when both prices are above 1/2 then it is the lower-price firm that makes greater profit. (When both prices are below 1/2 then the higher-price firm has more profit.) Therefore, when the less profitable firm chooses the price of the more profitable firm (and firms experiment every so often as discussed above), prices will settle over time at p1 = p2 = 1 / 2, so that the markup on cost falls by one-third with imitative behavior relative to profit-maximizing behavior.11 An interesting corollary of imitative behavior is that with symmetric firms the familiar distinction between price-setting and quantity-setting behavior vanishes, and the long-run performance of markets with imitative firms does not depend on whether firms choose to compete in prices or in quantities. With profit-maximizing behavior, when firms compete in quantities the outcome differs—typically it is less competitive—from when they compete in prices. (For instance, in the linear demand example of the previous paragraph, when firms compete in quantities one can show that the equilibrium involves each firm setting the output 3/5, which induces each firm to set the price 4/5, which is higher than when the firms compete in prices.) However, when firms choose quantities, the process of imitation converges to the situation where each firm chooses output 3/4, which induces each firm to set the price 1/2, which is exactly the same as when firms chose prices.12 Of course, firms can only imitate the most successful firm when they are provided with the necessary market information (namely, the profits of each firm, together with individual outputs in the Cournot case). But as soon as this information is present, a process as described by Vega-Redondo can make markets very competitive. Vega-Redondo’s paper has inspired several experimental tests that found some support for the basic qualitative prediction of the imitation model (Huck, Normann, and Oechssler 2000; Offerman, Potters, and Sonnemans 2002; and Apesteguia, Huck, and Oechssler 2007). Indeed, even when firms have enough information to calculate the Cournot equilibrium, if they are then given the extra firm-specific information that allows them to imitate the best, this makes their behavior more competitive. There is a close connection between situations in which firms imitate the most profitable action played by their rivals and situations in which firms (or their managers) have as their objective the maximization of relative profits. The reason for this close connection is as follows. Consider the differentiated product duopoly just discussed, and suppose one firm reduces its price to undercut its rival. If the lower-price firm makes greater profit than its rival, this means that the price cut harms it less than its rival, that is, the difference between its profit and its rival’s rises. In the specific linear demand example in

11  12 

See Rhode and Stegeman (2001) for more details. See Tanaka (2000) and Rhode and Stegeman (2001) for a general argument.

Behavioral Economics and Antitrust   211

expression (9.1) above, if firm 1 wishes to maximize its relative profit, it chooses price p1 to maximize the profit difference p1 (1 + p2 / 2 − p1 ) − p2 (1 + p1 / 2 − p2 ) ,



and it is therefore optimal for it to set the price p1 = 1 / 2, regardless of the rival’s price p2. If both firms care purely about their relative profit, they will both set the price 1 / 2, just as eventually occurred when firms imitate each other. Thus, the long-run outcome in markets in which firms imitate coincides with the (one-shot) equilibrium outcome in markets in which firms are fully rational except that their objective is relative rather than absolute profit. The importance of this result is that, while purely imitative behavior on the part of sophisticated firms perhaps seems implausible in many markets, the assumption that firms care about their relative position in a market has strong intuitive appeal.13 First, there is little doubt that many individuals are strongly driven by relative pay. This is evident from behavioral and experimental economics, as well as from the happiness literature (e.g., see Clark, Frijters, and Shields 2008). Indeed, CEOs as a group may have a greater proportion of “rivalrous” people than the population as whole, and such people may put particular weight on their standing relative to their peers. Second, managers often have placed on them (either explicitly or implicitly) incentives that induce them to care at least in part about relative as well as absolute performance. (One reason for this might be to insure managers against common shocks impacting the market as a whole.) For instance, Gibbons and Murphy (1990) document empirically how a CEO’s pay rise and likelihood of retention depend positively on the firm’s performance and negatively on the overall industry performance. Vickers (1985) shows that a firm in a Cournot market can improve its equilibrium profits (both in absolute and in relative terms) when competing against profit-maximizing firms by inducing its manager to care about relative rather than absolute profit.14 For the same reason, if potential managers differ in their intrinsic preferences for absolute or relative profit performance, Miller and Pazgal (2002) argue that a firm may wish to hire a manager with a known behavioral bias toward relative profit so as to gain strategic advantage. By contrast, Fumas (1992) shows that in Bertrand markets rather than Cournot markets, a firm’s profits typically fall if its manager cares about relative instead of absolute profits; a manager who cares about relative profits will set a low price, which in turn will induce low prices from its rivals, and each firm’s profits decline. So far, we have argued that imitation or concern for relative profits makes the static interaction among oligopolists more competitive, but what does this behavior imply for the 13 

Goel and Thakor (2010) propose a model in which merger waves can be caused by concerns for relative compensation by CEOs. If CEO compensation is based in part on firm size, then if one merger occurs that boosts that CEO’s pay, other CEO feel envious and set about finding their own takeover targets. 14  The impact of making the firm wish to maximize relative profits is that the firm then behaves as the Stackelberg leader, even though both firms in fact choose quantities simultaneously. It is important that rivals observe the incentive scheme so that they know the firm’s objective and can react to it accordingly.

212   Mark Armstrong and Steffen Huck sustainability of collusion? Suppose that each oligopolist cares about its profits relative to its rivals. Compared to the setting with absolute profit maximization, there are two effects: • The static Nash equilibrium—which is used to punish firms if they deviate from the collusive agreement—is more competitive, but • The short-run benefits to deviating are larger, since by undercutting its rivals a firm not only boosts its own profits but also harms its rivals. The first effect makes collusion easier, while the second makes it harder, and a priori it is unclear which dominates. However, in a Cournot setting, Lundgren (1996) shows that when firms care purely about relative profit, collusion is impossible to sustain. We can summarize this discussion as follows. An oligopolistic market is made more competitive relative to the textbook situation of firms maximizing profits if firms either (1) imitate the previous most profitable strategies or (2) aim to maximize their relative profits. The (eventual) outcomes under scenarios (1) and (2) coincide, and in both cases the outcome is the same whether firms compete in prices or quantities. In mixed settings, where relative profit maximizers compete against profit-maximizing firms, the former will often perform in absolute terms at least as well as the latter.

9.3.  Vengeful Behavior and Esprit de Corps The previous section discussed situations in which firms and their managers care about relative rather than absolute performance. Another kind of social preferences is present when firms care when their rivals obtain an “unfair” share of industry profits, for instance by cheating on a collusive agreement. It is clear that many people are willing to incur costs in order to harm others who are perceived to have behaved unfairly toward them, a phenomenon that could be called vengeance or spite. Among the most famous and robust experiments in economics are those that study the “ultimatum game.”15 Here, two players try to share a specified prize, and one player (the “proposer”) suggests a way to share the prize between the two players. If the second player (the “responder”) agrees with the proposed shares, the prize is divided accordingly, while if the responder does not agree, then neither player gets anything. If players are interested in obtaining as much of the prize as possible, the predicted outcome (from a one-shot interaction) is that the proposer offers only a tiny share to the responder, who accepts since she prefers a small positive payoff to nothing. However, it is commonly observed that the responder will reject small 15  See Güth, Schmittberger, and Schwarze’s (1982) pathbreaking experiment and the many papers that followed.

Behavioral Economics and Antitrust   213

offers. Because of this, proposers learn or anticipate that it is too risky to make such unfair offers, and outcomes where the proposer offers the responder 40 or 50 percent are often the norm. These experiments have been performed (in poorer countries) when the prize is significant in terms of monthly salary, and the results are similar. The same effect can be observed in markets. Huck, Muller, and Normann (2001) show that a firm with a theoretical commitment advantage (a Stackelberg leader in a Cournot market) finds it hard to exploit that advantage in experimental markets. The reason is that the theoretically disadvantaged firm (the Stackelberg follower) acts more aggressively than predicted by the subgame perfect equilibrium of these market games. In fact, followers appear willing to punish the leader when the leader supplies a quantity above the symmetric Cournot quantity. (They punish the leader by themselves supplying a higher quantity than their most profitable response to the leader’s quantity.) This behavior is in line with the vengeful behavior seen in the ultimatum game when the proposer tries to exploit his first-mover advantage. Similar behavioral effects are observed in experiments that study strategic delegation. As already mentioned, Vickers (1985) showed theoretically that firm owners might want to employ managers simply for strategic reasons. By writing an appropriate incentive contract for the manager, the owner can in effect commit the firm to Stackelberg-like aggressive behavior. However, if all firms do this a highly competitive outcome results: All firms are worse off than when managers care solely about profit. Huck, Muller, and Normann (2004) tested this theory in an experiment, and do not confirm its predictions. If an owner does offer his manager the aggressive contract (and other owners do not), then managers in the weaker position are not content with simply accepting their “equilibrium fate” that would give them a lower salary than their opponent. Rather they behave more aggressively, in line with the observed behavior of the Stackelberg followers mentioned in the previous paragraph. This could mean that attempts to hire “behavioral” types of managers, as in Miller and Pazgal, may also backfire. A natural question is whether the spirit of vengeance might help to sustain collusion. For instance, if a particular collusive arrangement (such as a high price and equal market shares) is considered to represent the “fair” outcome, then if one firm reneges on the agreement and undercuts the price, its rivals may be offended and hence punish the deviator especially aggressively (even at extra cost to themselves). In the previous section we argued that concern for relative profits did not help sustain collusion, since it induced an extra benefit to deviating. But vengeance is subtly different from a concern for relative profit, since it only comes into play when someone is treated unfairly. (There is no extra benefit from harming your rival, unless he has first harmed you.) Thus, because vengeance induces more aggressive punishments for deviating, it could help sustain collusion relative to situations in which firms care only about their own profits.16 16 

See Iris and Santos-Pinto (2013) for a model along these lines.

214   Mark Armstrong and Steffen Huck While vengeful behavior may sometimes be a way to sustain collusion, an alternative method is to foster a sense of loyalty, of esprit de corps, among conspirators. Although information sharing is often an important ingredient for collusive schemes to work, a useful byproduct of regular meetings is that loyalty and friendship may be inculcated, which may make it socially costly to cheat on agreements. Moreover, since illegal cartels cannot enforce agreements with legally binding contracts, trust plays a central role in their operation.17 An important role for antitrust, via leniency programs and the like, is indeed to foster distrust among conspirators. A clear example of an attempt to run a cartel (or cartel-like operation) by means of encouraging a sense of camaraderie is the US steel industry during the years 1907–1911, in which cooperation was fostered through a series of social events and meetings that have become known as the “Gary Dinners” after the chairman of US Steel, Judge Elbert Gary. It is worth quoting one of Gary’s speeches at length: [W]‌e have something better to guide and control us in our business methods than a contract which depends upon written or verbal promises with a penalty attached. We as men, as gentlemen, as friends, as neighbors, having been in close communication and contact during the last few years, have reached a point where we entertain for one another respect and affectionate regard. We have reached a position so high in our lines of activity that we are bound to protect one another; and when a man reaches a position where his honor is at stake, where even more than life itself is concerned, where he cannot act or fail to act except with a distinct and clear understanding that his honor is involved, then he has reached a position that is more binding on him than any written or verbal contract. (See Page 2009, 608)

In essence, if conspirators can find a way to increase the social or psychological cost of cheating, collusion will be more readily achieved. Further historic evidence of the impact of social preferences on anticompetitive behavior comes from the UK shipping cartels in the period 1879 to 1929. Podolny and Scott Morton (1999) document how established cartels behaved more aggressively toward entrants who had low social status or who were foreign, relative to the situation where an entrant had a knighthood, say. (The study controls for the correlation between social status and “deep pockets,” which may make predation less profitable.) The authors argue that the likely reason is not pure snobbishness on the part of the cartel (who typically comprised high-status individuals), but rather that social status was used as a signal of the entrant’s likelihood of cooperating within the cartel, and the expected “transactions costs” of having the entrant as a member of the cartel, if admitted. A related issue concerns the impact of communication between firms, and whether the ability to engage in some form of communication before market interaction aids collusion in the laboratory. (Here, any agreements made in the communication stage are 17  See Leslie (2004) for further discussion of these points and illustrations of the role of trust and distrust in cartel stability.

Behavioral Economics and Antitrust   215

nonbinding and purely “cheap talk.”) This literature is surveyed in Potters (2009), who concludes that, given the opportunity to communicate, firms do use this opportunity to conspire to fix prices, and this ability often has the effect of raising prices in the market. It appears that face-to-face talk has more of a collusive impact than computer-mediated communication, somewhat consistent with the camaraderie among conspirators just discussed. Andersson and Wengstrom (2007) report results from an experiment in which it is costly to send a message from one firm to another. Intriguingly, they find that as the cost of communication rises (which could be interpreted as the outcome of a more vigilant antitrust policy) fewer messages are sent, as is intuitive, but when messages are sent, collusion is more effective. One interpretation is that firms feel more committed to a collusive agreement if costs are needed to reach that agreement. The net impact of making it hard to communicate is that collusion is substantially more prevalent, so that certain forms of antitrust policy might perversely turn out to aid cartel formation and stability.

9.4.  Satisfactory, Not Maximum, Profits Rather than denying the importance of (absolute) profits, one might instead question whether firms really maximize profits. For reasons of complexity, ignorance, or the “easy life,” firms might instead engage in satisficing behavior to secure a target level of profit. In its starkest form, the utility of a firm can then take just two values: good enough, and not good enough. This idea goes back at least to Gordon (1948) and Simon (1955).18 Just as a consumer might not change her bank, say, until the perceived level of service quality she receives falls below some threshold, a firm might not revise its strategy while it continues to obtain satisfactory profits. Only if the realized profit falls below the target level will a firm resort to experimenting with an alternative strategy. Such behavior could give rise to a degree of price rigidity, for instance, even if underlying cost or demand conditions vary. Experimentation could be purely random in extreme cases, or firms may be able to target their attention toward new strategies that will likely boost profits. (For instance, an idea in the early literature on satisficing was that firms often had a degree of “managerial slack,” and when times were hard managers could focus their energies on cutting costs or expanding their markets.) What level of profits a firm considers to be “satisfactory” is likely to depend on its historical returns as well as the performance of its peers and the economy as a whole. 18  For instance, Gordon (1948, 271) wrote: “The fear of bankruptcy and the even more widespread fear of temporary financial embarrassment are probably more powerful drives than the desire for the absolute maximum in profits. . . . Given the fog of uncertainty within which [the businessman] must operate, the limited number of variables his mind can juggle at one time, and his desire to play safe, it would not be at all surprising if he adopted a set of yardsticks that promised reasonably satisfactory profits.”

216   Mark Armstrong and Steffen Huck Cyert and March (1956) discuss some implications of satisficing behavior and test these ideas by empirically analyzing data from manufacturers of farm implements. They conclude that a firm is more likely to resort to aggressive strategies (seeking cost reductions or sales expansions) if it is operating close to its break-even point or if its costs are higher than the industry average, behavior that accords with the idea that firms are most likely to change strategy after they realize low profits. Huck and coauthors (2007) document related effects in an experiment concerning mergers in Cournot markets. As shown by Salant, Switzer, and Reynolds (1983), bilateral mergers in Cournot markets with more than two firms and linear demand are predicted to reduce the joint profits of merging firms when firms are profit-maximizers. Since this prediction appears counterintuitive, the result has been dubbed the “merger paradox.” In Cournot markets firms’ outputs exert negative externalities on their rivals. Hence, after a merger the two insiders have an incentive to internalize some of these negative externalities; that is, they have an incentive to reduce their combined output. Further, as Cournot interactions are games with strategic substitutes, the unmerged firms’ optimal response to this contraction in output is to increase their own outputs (in response to which the merged firm will reduce its output even further, and so on). This leads into a new equilibrium where the market price is higher, the outsiders are better off, while the two merging insiders are worse off. Huck and coauthors test this prediction in laboratory markets with initially three or four firms. Firms are randomly selected to merge, and the merged firm is managed by one of the two subjects playing the role of the two merging firms. Profits are shared equally between the manager-owner and the passive owner. After the merger, profit-maximizing firms are predicted to behave as if one of the firms had vanished, and in particular, a merged firm is predicted to choose the same output as the nonmerged firms. However, a different behavior emerges in the laboratory; the merged firm systematically supplies more than the outsiders. (The outsiders are seen to respond approximately optimally to the behavior of the merged firm.) This implies that the losses from the merger are smaller than predicted. In fact, when there are four firms initially present, the merged firm does not lose at all. Rather, it experiences a significant increase in profits for a few periods and then, when the outsiders are fully adjusted, its per-period profits fall slightly compared to their premerger levels. Huck and coauthors discuss a number of reasons for this effect, and by conducting suitable control experiments they argue that something akin to satisficing behavior explains observed behavior.19 After two firms merge, the firms perceive the danger that their joint profits fall, and pursue aggressive strategies to avoid this (as the firms in Cyert and March did). The result is that the merged firm’s output is significantly higher than its rivals’ outputs, even though postmerger a merged firm and an unmerged firm are

19  Levin (1990) presents a related theoretical analysis of mergers in Cournot markets, in which he allows that a merged firm changes its behavior from a Cournot-Nash player to a Stackelberg player (among other possible behavioral changes), in which case a merger becomes profitable.

Behavioral Economics and Antitrust   217

in a symmetric position. (However, the merged firm’s output is lower than the separate firms’ combined outputs in the premerger phase, and so their strategies are not entirely inert.) Dixon (2000) and Oechssler (2002) discuss an interesting application of satisficing behavior. They consider a number of oligopolies (rather than just one), and suppose that firms observe the following rule of thumb: “If my profits are no lower than the average profits observed across all markets, I keep my strategy unchanged; if my profits are lower than average profits across all markets, I experiment and choose a random strategy.” Thus, firms use the average payoff of all firms as their aspiration level. Such a policy requires information about average profitability, but no information about individual firm actions or profits or about consumer demand. It turns out that when firms behave in this manner (with a small amount of noise added) then behavior in each market eventually becomes collusive. Consequently, practices that look innocent from the orthodox point of view—myopic adjustments toward better strategies—may lead to undesirable outcomes. There can then be collusive effect without any collusive intent. To understand this surprising result, consider a pair of identical duopolies (i.e., there are four firms in all), which interact over time in a prisoner’s dilemma manner using the two possible actions of “compete” or “collude.” Suppose that all firms are initially colluding, in which case all firms are satisfied with their profits. Next, imagine one firm “trembles” and changes its action to “compete.” A duopoly in which both firms collude has higher profits than a duopoly where one firm colludes and the other competes, and so firms in the colluding duopoly remain satisfied with their profits and do not experiment. The only firm dissatisfied with its profit is the colluding firm playing against the competing firm, which will revise its strategy and “compete.” We then have two firms in a market playing “compete,” both of whom will be dissatisfied with their profits, and will experiment and end up both playing “collude.” Thus all firms colluding is the stable steady state of this process (unlike all firms competing, where one “tremble” will eventually induce all firms to collude). In contrast to Adam Smith’s invisible hand, which guides profit-maximizing firms toward outcomes that benefit consumers, with this form of satisficing behavior, a second invisible hand guides firms toward outcomes that exploit consumers. A concept related to satisficing behavior is “approximately” optimal behavior. Firms may not find it worthwhile to calculate their optimal strategy precisely, but instead to cease their search for a good strategy when they get to within some tolerance of the optimal strategy. In technical terms, the outcome of approximately optimal behavior is termed an “ε-equilibrium,” where ε > 0 is the (perhaps small) tolerance that firms have for shortfall in optimal profits. An important insight is that even a small tolerance by firms for suboptimal strategies—that is, a small departure from fully optimizing behavior—can result in significant departures from the outcomes corresponding to optimizing behavior. (In the neighborhood of a firm’s optimal strategy its profits are roughly flat, and so a moderate change in its strategy might have little impact on its own profit, and yet could have a significant impact on its rivals’ profits and strategies.)

218   Mark Armstrong and Steffen Huck To illustrate, consider two symmetric firms supplying a homogeneous product engaged in quantity competition. To be concrete, suppose that the consumer demand curve is q ( p ) = 1 − p and that production is costless. Then the symmetric fully collusive outcome in this market involves each firm supplying quantity 1 / 4, inducing the monopoly price p = 1 / 2. How small can a firm’s tolerance for suboptimal behavior be for this monopoly outcome to be an ε-equilibrium? The most profitable response to the rival’s monopoly output of 1/4 is to supply output 3/8 rather than 1/4. However, the firm’s gain in profit from pursuing the optimal strategy rather than the satisficing strategy of supplying output 1/4 is only 1/64. Thus, the collusive outcome (where each firm supplies quantity 1/4) is an ε-equilibrium provided that ε > 1 / 64. Note that 1/64 is about 6 percent of a firm’s share of the monopoly profit, and so if firms are prepared to optimize to within 6 percent of their exact optimal profits, the monopoly outcome can be sustained without any collusive intent (see Ellison 2006, 170–71). Of course, though, a problem with the notion of ε-equilibrium is its lack of predictive power: firms could also accidentally play more competitively than the Cournot duopoly equilibrium. Baye and Morgan (2004) present an interesting account of how almost-optimal behavior can lead to significant departures from the predictions of fully optimal models. They analyze a static homogenous product Bertrand market, so that profit-maximizing firms are predicted to set prices at marginal cost and make zero profits. They derive an ε-equilibrium involving mixed strategies for choosing prices that yields relatively high profits even for small ε. For example, with duopoly, when ε is just 1 percent of monopoly profit, their ε -equilibrium yields expected profit that is more than 25 percent of the monopoly profit. Baye and Morgan also discuss an alternative model of bounded rationality, which supposes that firms play a so-called quantal-response equilibrium (see McKelvey and Palfrey 1995). Here, a firm is more likely to choose a price that yields higher profit, but is not sure to choose the most profitable price. (This model nests the fully rational model and a model with purely random behavior as polar cases.) While the details of the equilibria in the two models of bounded rationality differ, the broad conclusions—that prices are above cost, profits are positive, and profits fall with more rivals—coincide. Baye and Morgan run laboratory experiments to generate data, and estimate which model of firm behavior best fits the data. Especially for duopoly, they find that both the ε-equilibrium and the quantalresponse equilibrium model fit the data better than either of the extreme fully rational or fully random models.

9.5.  Other Topics 9.5.1. Overoptimism As is well documented in the psychology literature, overoptimism (or overconfidence) about one’s own ability or about the probability of favorable outcomes is apparently

Behavioral Economics and Antitrust   219

common in the population.20 It is plausible that the phenomenon is still more common among entrepreneurs than the population as whole. There is a “winner’s curse” aspect to launching a new business: It is likely that others have already thought about launching a similar product (such as a new restaurant in a particular location), and the entrepreneur who actually decides to start the new business is likely to be more optimistic than others. As a result, even if beliefs about the likely return from the investment are unbiased on average, the entrepreneur will typically be overly optimistic. Unless entrepreneurs rationally take full account of the fact that others have decided not to enter this market, we expect to see (1) high failure rates for new businesses, and (2) credit rationing, or loans being offered only with collateral.21 In an abstract context, Heifetz, Shannon, and Spiegel (2007) argue that optimists will systematically outperform realists in competitive environments, and so will predominate in the pool of successful agents. The key intuition is that optimism can serve as a commitment device if observed by rivals (much as making a manager care about relative profits does in Vickers 1985). Translated into a market context, Heifetz, Shannon, and Spiegel’s analysis suggests that particular forms of managerial overoptimism may sometimes act to soften competition. For instance, consider a Bertrand oligopoly with differentiated products where there is some uncertainty about demand. The more optimistic a manager is—say, about the scale of demand, or the extent of product differentiation among suppliers—the higher the price she will charge. If a manager’s optimism is common knowledge or if competitors at least have some informative signal about her degree of optimism, they will rationally anticipate these higher prices and optimally adjust their own prices upwards as well. Hence, one optimist is enough to increase the prices charged by all firms. And crucially the optimist will earn more money for her firm than the realist would have done (although her rivals will typically earn still more).22 In what is perhaps the best-known paper about managerial irrationality, Roll (1986) suggests that excessive merger activity may be generated by managerial overconfidence. In his theory, a CEO is too confident in the accuracy of his information about the potential profitability of a takeover, and so will be too quick to launch a takeover bid. Predictions of this theory are that the combined gain to bidder and target will be close to zero, and that the bidder’s value will fall on announcement of the bid. Roll suggests that the available empirical evidence does not reject these predictions. Overconfidence by CEOs in their ability to choose investment projects and merger opportunities has 20  For example, Svenson (1981) finds that 93 percent of respondents report that they are above the median in terms of driving ability. 21  See De Meza and Southey (1996) for empirical evidence and a model that predicts that entrepreneurs are more optimistic than the general population. Camerer and Lovallo (1999) document in an experimental study that excessive entry can be driven by overconfidence in own ability. Tor (2002) provides an extended discussion of behavioral economics and entry decisions, and the implications for competition policy. 22  See Englmaier and Reisinger (2014) for such a model. In a Cournot market, hiring a manager who is overoptimistic (say, about the scale of market demand) again confers strategic advantage to the firm, but if all firms do this, they will all be worse off relative to the situation with unbiased managers.

220   Mark Armstrong and Steffen Huck been detected by Malmendier and Tate (2005, 2008). CEOs will differ in the extent of their overconfidence, and Malmendier and Tate identify as overconfident those CEOs who hold on to their stock options until they expire. They interpret this as evidence of the CEO’s overestimation of their firm’s future performance. They show that this group of CEOs is 55 percent more likely to undertake a merger.23 Managerial overconfidence could explain why companies that undertake mergers seem to underperform. Another form of potential overconfidence concerns a manager’s faith that he will not be caught if he pursues an illegal activity such as price-fixing.24 If many cartelists downplay the risk of detection, then policy measures to increase the penalties for illegal cartel activity could be less effective than would be predicted by an approach based on rational profit-maximizing behavior.

9.5.2.  Accounting Anomalies It is a fundamental tenet of profit-maximizing behavior that fixed and sunk costs, while they are important for entry and exit decisions, should not play a role in the determination of prices to customers. For instance, competition authorities do not always put much weight on claimed synergies between merging firms that act to reduce fixed costs, since such synergies are not expected to feed through into lower prices. However, this policy seems to be widely flouted by managers. Long ago, Hall and Hitch (1939) interviewed 38 business executives about their methods for setting prices. Instead of equating marginal revenue to marginal cost, the authors concluded that “there is a strong tendency among business men to fix prices directly at a level which they regard as their ‘full cost.’ ” Al-Najjar, Baliga, and Besanko (2008) describe a number of more recent surveys that report similar attitudes among managers, the majority of whom claim to take fixed and sunk costs into account when setting prices. (They also highlight a managerial accounting textbook that argues against basing prices on marginal costs.) Experimental work confirms that supposedly irrelevant sunk costs can have an impact on how prices are actually chosen. Offerman and Potters (2006) conduct an experiment to investigate pricing in a Bertrand-type duopoly market with product differentiation. In one treatment, there are no sunk costs and pricing behavior is observed to converge to the Bertrand equilibrium. In a second treatment, participants must pay a sunk entry fee to join the market. (This is designed to model auctioning a license for the right to enter a market.) In this second treatment, the average markup of prices over marginal cost is substantially higher than in the first treatment. Interestingly, when they perform the same pair of experiments but with monopoly instead of oligopoly, there is no impact of sunk costs on the chosen monopoly price (which is observed to be close to 23 

This correlation between holding onto stock options and corporate behavior could also be due to the CEO’s inside information. However, Malmendier and Tate argue that this is unlikely to be the explanation since the CEOs who hold onto their stock options do not gain money from doing so. 24  For further discuss of this point, see Stucke (2011).

Behavioral Economics and Antitrust   221

the profit-maximizing level). This is somewhat reminiscent of our discussion of communication costs in section 9.3, where we saw that firms were more inclined to stick to a nonbinding collusive agreement if they incurred costs to reach that agreement. This observation is consistent with Al-Najjar, Baliga, and Besanko (2008), who present a theoretical model to show how the use of full-cost pricing policies might persist in the long-run in oligopoly markets. The main result is by now a familiar one in this survey: By introducing a behavioral bias in managerial decision-making, a firm can gain strategic advantage. In more detail, Al-Najjar, Baliga, and Besanko suppose that firms compete in a Bertrand market with product differentiation. In such a market, if a manager somehow commits to set a high price, the firm’s rivals will also set a high price, and all firms will make higher profits.25 (The effect is akin to hiring an overoptimistic manager or, as previously discussed, placing a suitable incentive scheme on the manager.) But a manager who bases prices in part on fixed and sunk costs effectively commits to set a high price, and so hiring a manager who practices this “naive” pricing policy (or instilling a corporate culture where this form of pricing is used) boosts the firm’s profits. The effect is akin to the strategic tax policy analyzed by Eaton and Grossman (1986), where a country has an incentive to tax the output of a home firm in order to relax competition with a foreign rival. If, for whatever reason, firms take fixed costs into account when setting their prices, this has potentially important implications for merger policy when the merger is expected to generate only synergies in fixed costs. Such mergers would in fact lead to lower prices than would be predicted in a profit-maximizing world. Other alleged accounting anomalies might be explained by similar myopic learning or evolutionary pressures, or reasons of strategic delegation. For instance, a firm might be organized into separate profit centers, each of which is given the task of maximizing its own profit (despite the competitive or contracting externalities that might exist between these profit centers). Thus, a manufacturer might choose to supply its products through an independent retailer rather than sell directly to consumers via an integrated retailer. To see one reason why this might be so, despite the apparent danger of double marginalization, suppose that two manufacturers are competing to sell to consumers. If they sell directly to consumers, then in the absence of collusion we expect to see the Bertrand equilibrium emerge. However, if each firm delegates its retailing operation to a separate division (and each retailing division sees the terms at which the rival retailer sources its product), then by setting a wholesale price above 25 

A numerical example may help to fix ideas. Suppose there is a duopoly with differentiated products where the two firms choose prices and where a firm’s demand is given by expression (9.1) as in section 9.2. But instead of profit-maximizing behavior, suppose that firm i = 1, 2 behaves as if its marginal cost is ci rather than the true cost (which is zero). Suppose that farsighted profit-maximizing shareholders of firm i (or some form of evolutionary pressures) choose a corporate culture such that their manager behaves as if cost was ci, and both groups of shareholders do this before price competition takes place. Then one can show that the equilibrium chosen “costs” are c1 = c2 = 1 / 11, and costs are artificially boosted so as to raise equilibrium prices. In this example, prices rise by approximately 10 percent relative to the situation in which managers base their prices on the true marginal costs.

222   Mark Armstrong and Steffen Huck its production cost a firm can induce its retailing division to price high, which softens competition and boosts profits.26

9.5.3.  Uncertainty about the Rationality of Rivals Even if a firm is fully rational, its behavior will be affected when it believes that its rivals may not be rational. Because of this, a rational firm may have an incentive to mimic nonrational behavior so as to induce its rival to believe it is a nonrational type of firm. Consequently, even the potential for behavioral biases can have a substantial effect on market outcomes. One way in which a rational firm might wish to mimic an irrational firm is in the context of predatory pricing, as analyzed in Kreps and Wilson (1982). If a monopoly incumbent faces a sequence of potential entrants to its market, it may wish to establish a reputation for fighting entry whenever it occurs, even when fighting is actually more costly than accommodating entry. The predictions of the Kreps-Wilson model of predatory pricing are broadly (1) there is little entry in the early period, (2) when there is entry in early periods, the incumbent “fights” even if it is a rational firm, and (3) toward the endpoint there is more entry and less willingness to fight on the part of the rational incumbent. This model, which involves highly sophisticated reasoning on the part of firms, is tested in an experiment by Jung, Kagel, and Levin (1994). The experiment found widespread predatory pricing—defined to be either no entry in the early stages, or a rational incumbent fighting entry if it does occur in the early stages—although the more detailed predictions of the Kreps-Wilson model did not fit the data. (For instance, experimental subjects did not enter more frequently if there had been no previous entry as compared to the situation where there had been fought entry, whereas the Kreps-Wilson model predicts there should be more entry in the former case.) Rather than go to the trouble of mimicking the behavior of behavioral type of firm, which in any case may only deter entry for some of the time, a more straightforward method to deter entry could be simply to hire an “aggressive” manager who is known by potential entrants intrinsically to wish to fight entry whenever it occurs. (For instance, an aggressive manager might be someone who is overoptimistic about the ease by which entry can be successfully fought, and so is willing to fight more often than an unbiased manager would.) The impact of this policy, assuming it is credible to keep this aggressive manager in place after entry occurs, is akin to hiring a “conservative” central banker who is less likely to be tempted to cause inflation when unemployment rises (Rogoff 1985). As Schelling (1960, 142) wrote: “the conspicuous delegation of authority to a military commander of known motivation, exemplifies a common means of making credible a response pattern that the original source of decision might have been thought to shrink from or to find profitless.”

26 

See Bonanno and Vickers (1988) for this model.

Behavioral Economics and Antitrust   223

9.6. Discussion This survey has discussed a number of reasons why firms might not pursue maximum profits:





1. In complex and uncertain environments, the optimization problem may simply be too hard, and firms might resort instead to satisficing and the use of rules of thumb. Decision shortcuts included imitating the actions of well-performing peers, satisficing, or being content to achieve profits to within “ε” of the maximum. These rules of thumb appear attractive since in many situations that require strategic sophistication, such as herding, reputation-building, or collusion among many firms, the complex strategies needed to maximize profits are rarely observed in even in the most simplified laboratory settings and even if participants are highly experienced. 2. Alternatively, optimization might still occur, but with alternative aims or under mistaken beliefs. Thus, a manager might maximize her profits relative to those of her peers, or a manager could be overoptimistic about the profitability of some action. The reason why managers have aims different from maximizing profits could be due to selection effects (e.g., only “competitive” or overoptimistic people rise to become CEOs, or because firms that aim to maximize their relative standing actually obtain greater absolute profits than their profit-maximizing rivals), or because profit-maximizing principals choose to give their managers distorted incentives to gain strategic advantage. 3. Social preferences (other than caring about relative profit as above) may play a role, and a firm might punish a rival if that rival obtains an “unfair” share of profits. Alternatively, face-to-face communication between managers, or having a similar social background among managers, may generate solidarity among conspirators that makes it socially awkward to cheat on collusive agreements.

In some situations, markets are more competitive when firms do not aim to maximize their profits. For example, if firms in a market myopically imitate the action of the most profitable rival, then the market may paradoxically move toward a highly competitive outcome. Alternatively, at least in laboratory settings, firms are often unable to achieve tacit collusion, despite this being an equilibrium option for profit-maximizing firms. Some of the more complex strategies that foster collusion in theory are perhaps too subtle to matter empirically. In other situations, when firms cannot maximize profits, their realized profits are actually increased. For these firms, the best is the enemy of the good. This can be clearly seen in the case of static Bertrand competition with homogenous products. Here, the only equilibrium involves firms setting prices equal to marginal costs, leaving them with no profits. But if satisficing firms are content to choose actions that are only approximately optimal, they may all be able to enjoy substantial profits. Likewise, when

224   Mark Armstrong and Steffen Huck satisficing firms alter their actions only when they underperform relative to average performance, the result may be as if firms were successfully colluding. If a firm adopts a nonstandard objective, it may gain strategic advantage in the market since competitors’ behavior will change in a desirable way. In such an environment, it could be rational to behave irrationally. For instance, in Cournot markets a firm that aims to maximize its relative profits may do better in equilibrium than a profit-maximizing rival. Alternatively, a firm that chooses to base its price on “full costs” rather than marginal costs may do better than if it followed textbook profit-maximizing precepts. Or a firm led by an aggressive manager may deter entry more often than a profit-maximizing manager. These nonstandard objectives could be put in place by farsighted profit-maximizing shareholders (as emphasized in the strategic delegation literature) or they could arise myopically due to evolutionary selection of better performing managers and/or firms. Regardless of the reason why firms have nonprofit objectives, the presence of nonprofit objectives is likely to affect competitive interaction. For instance, a manager operating under an incentive scheme that rewards relative performance is likely to behave more competitively than orthodox analysis would indicate, and this is a factor that a competition authority should take into account. (The effect is the reverse of the situation when there is cross-ownership within an oligopoly, which competition authorities already recognize leads to a blunting of competition.) Thus we see there are several situations in which Friedman’s Darwinian critique of nonprofit maximizing behavior appears to fail. Nevertheless, there are a number of situations in which market competition and market experience do seem to diminish those behavioral biases that do not confer evolutionary advantage. For instance, in perfectly competitive markets, there is no distinction between absolute and relative profit. In addition, as shown in the work of List (2003, 2004), market experience can dampen the bias known as the endowment effect, where the valuation of a good increases when it is owned. Such a bias cannot easily improve own performance in markets with many traders (rather it hinders agents from making otherwise beneficial trades). Likewise, we are not aware of many situations in which procrastination, impulsive behavior, and other self-control problems play a major role in firm behavior (e.g., price fixing is not a “crime of passion”), although these behaviors are prominent in discussions of behavioral economics as applied to consumers. We have seen a number of reasons why firms may not maximize their profits, and this potentially has implications for empirical studies of markets, including the use of merger simulation for antitrust policy. Empirical market studies typically assume profit-maximizing behavior on the part of firms to produce their estimates of, say, marginal costs. In such cases, the analysis may lead to biased estimates of the welfare impact of a merger if, in fact, the merging firms were not profit-maximizers. To illustrate, consider the merger situation in Huck and coauthors (2007), which we discussed in section 9.4. If one took the data from this experiment pre- and postmerger and, in line with the structural approach, assumed these data were generated by profit-maximizing firms, one would conclude that the merger must have induced substantial cost synergies. Only with reduced costs would it be possible for the merged firm to have higher outputs than the outsiders. This would affect estimates of the merger’s welfare consequences. While

Behavioral Economics and Antitrust   225

the true consequences are unambiguously negative (after all, in the experiment the merged firm does operate with the same costs), sufficiently big synergies could offset the loss in consumer welfare. Hence, if there are systematic deviations from orthodox profit-maximizing assumptions, structural approaches that assume profit-maximization might detect increases in welfare when, in fact, welfare is reduced. Likewise, a merger simulation exercise may be inaccurate if the assumption is that firms maximize profits. More fundamentally, much of antitrust policy is founded on an assumption of profit-maximizing behavior by firms. Hovenkamp (2005, 134) goes so far as to say that the “entire antitrust enterprise is dedicated to the proposition that business firms behave rationally.” Courts and regulators in some jurisdictions may not consider seriously conduct (such as predatory pricing, for instance) that does not appear to make “business sense” according to their judgment. Leslie (2010, 269) reports that “if a plaintiff ’s complaint describes a conspiracy that the judge concludes is irrational, then the court rules that the conspiracy must not have happened as a matter of law, regardless of the evidence presented by the plaintiff to support its claim.” In the light of the theories and evidence reported in this chapter, we suggest that a dogmatic attitude toward the pervasiveness of business rationality may lead to instances where harmful behavior goes unpunished. Behavioral economics may sometimes pose at least as many questions as it answers, and complicate antitrust debate. But behavioral economics also sheds light on important market phenomena, and if antitrust policy is to reflect market realities, behavioral economics cannot be ignored. Finally, while this chapter has surveyed behavioral economics as it applies to firms, in future it would be interesting to investigate how it applies also to policymakers (see Leslie 2010, section III.F). Antitrust authorities, like firms, operate within a complex and strategic environment, and may need to resort to rules of thumb and satisficing behavior. Instances of this could include an authority’s use of per se rules, or a reliance on relatively rigid market definitions and market share thresholds. It may also be advantageous to induce antitrust authorities to have an objective or institutional focus that differs from social welfare, in order to alter the response from the firms subject to regulation. Imitative strategies may sometimes be employed, and safety in numbers may be enjoyed, as policymakers look around the world for current “best practice.” (Indeed, recent emphasis on the antitrust implications of behavioral economics may be an instance of this.) Friedman’s point about competitive pressures may have less force in terms of constraining good decision-making by public officials, and the result may be that behavioral biases are more prevalent among policymakers than in the firms they oversee.

Acknowledgment This chapter is an updated, edited, and streamlined version of our earlier survey article “Behavioral Economics as Applied to Firms: A Primer,” Competition Policy International 6 (2010): 3–45.

226   Mark Armstrong and Steffen Huck

References Alchian, Armen. 1950. Uncertainty, Evolution, and Economic Theory. Journal of Political Economy 57: 211–21. Al-Najjar, Nabil, Sandeep Baliga, and David Besanko. 2008. Market Forces Meet Behavioral Biases: Cost Misallocation and Irrational Pricing. Rand Journal of Economics 39: 214–37. Andersson, Ola and Erik Wengstrom. 2007. Do Antitrust Laws Facilitate Collusion? Experimental Evidence of Costly Communication in Duopolies. Scandinavian Journal of Economics 109: 321–39. Apesteguia, Jose, Steffen Huck, and Jorg Oechssler. 2007. Imitation: Theory and Experimental Evidence. Journal of Economic Theory 136: 217–35. Armstrong, Mark. 2008. Interactions between Competition and Consumer Policy. Competition Policy International 4: 97–148. Baye, Michael, and John Morgan. 2004. Price Dispersion in the Lab and on the Internet: Theory and Evidence. Rand Journal of Economics 35: 449–66. Bertrand, Marianne, and Antoinette Schoar. 2003. Managing with Style: The Effect of Managers on Firm Policies. Quarterly Journal of Economics 118: 1169–208. Bonanno, Giacomo, and John Vickers. 1988. Vertical Separation. Journal of Industrial Economics 36: 257–65. Camerer, Colin, and Dan Lovallo. 1999. Overconfidence and Excess Entry: An Experimental Approach. American Economic Review 89: 306–18. Clark, Andrew, Paul Frijters, and Michael Shields. 2008. Relative Income, Happiness, and Utility: An Explanation for the Easterlin Paradox and Other Puzzles. Journal of Economic Literature 46: 95–144. Cyert, Richard, and James March. 1956. Organizational Factors in the Theory of Oligopoly. Quarterly Journal of Economics 70: 44–64. De Meza, David, and Clive Southey. 1996. The Borrower’s Curse:  Optimism, Finance and Entrepreneurship. Economic Journal 106: 375–86. Dixon, Huw. 2000. Keeping Up with the Joneses: Competition and the Evolution of Collusion. Journal of Economic Behavior and Organization 43: 223–38. Eaton, Jonathan, and Gene Grossman. 1986. Optimal Trade and Industrial Policy under Oligopoly. Quarterly Journal of Economics 101: 383–406. Ellison, Glenn. 2006. Bounded Rationality in Industrial Organization. In R. Blundell, W. Newey, and T. Persson, Advances in Economics and Econometrics: Theory and Applications. Cambridge: Cambridge University Press, 142–74. Ellison, Glenn, and Drew Fudenberg. 1993. Rules of Thumb for Social Learning. Journal of Political Economy 101: 612–43. Englmaier, Florian, and Markus Reisinger. 2014. Biased Managers as Strategic Commitment. Managerial and Decision Economics 35: 350–356. Feinberg, Robert. 1995. In Defense of Corporate Myopia. Managerial and Decision Economics 16: 205–10. Friedman, Milton. 1953. Essays in Positive Economics. Chicago: University of Chicago Press. Fumas, Vicente Salas. 1992. Relative Performance Evaluation of Management. International Journal of Industrial Organization 10: 473–89. Gibbons, Robert, and Kevin Murphy. 1990. Relative Performance Evaluation for Chief Executive Officers. Industrial and Labor Relations Review 43: 30–51.

Behavioral Economics and Antitrust   227

Goel, Anand, and Anjan Thakor. 2010. Do Envious CEOs Cause Merger Waves? Review of Financial Studies 23: 487–517. Gordon, R.  A. 1948. Short-Period Price Determination in Theory and Practice. American Economic Review 3: 265–80. Güth, Werner, Rolf Schmittberger, and Bernd Schwarze. 1982. An Experimental Analysis of Ultimatum Bargaining. Journal of Economic Behavior and Organization 3: 367–88. Hall, R., and C. Hitch. 1939. Price Theory and Business Behavior. Oxford Economic Papers 2: 12–45. Hayek, Frederick. 1945. The Use of Knowledge in Society. American Economic Review 35: 519–30. Heifetz, Aviad, Chris Shannon, and Yossi Spiegel. 2007. The Dynamic Evolution of Preferences. Economic Theory 32: 251–86. Hovenkamp, Herbert. 2005. The Antitrust Enterprise:  Principle and Execution. Cambridge, MA: Harvard University Press. Huck, Steffen, Kai Konrad, Wieland Muller, and Hans-Theo Normann. 2007. The Merger Paradox and Why Aspiration Levels Let It Fail in the Laboratory. Economic Journal 117: 1073–95. Huck, Steffen, Wieland Muller, and Hans-Theo Normann. 2001. Stackelberg Beats Cournot: On Collusion and Efficiency in Experimental Markets. Economic Journal 111: 749–65. Huck, Steffen, Wieland Muller, and Hans-Theo Normann. 2004. Strategic Delegation in Experimental Markets. International Journal of Industrial Organization 22: 561–74. Huck, Steffen, Hans-Theo Normann, and Jorg Oechssler. 2000. Does Information about Competitors’ Actions Increase or Decrease Competition in Experimental Oligopoly Markets? International Journal of Industrial Organization 18: 39–57. Iris, Doruk, and Luis Santos-Pinto. 2013. Tacit Collusion under Fairness and Reciprocity. Games 4: 50–66. Jung, Yun Joo, John Kagel, and Dan Levin. 1994. On the Existence of Predatory Pricing: An Experimental Study of Reputation and Entry Deterrence in the Chain-Store Game. Rand Journal of Economics 25: 72–93. Kreps, David, and Robert Wilson. 1982. Reputation and Imperfect Information. Journal of Economic Theory 27: 253–79. Leslie, Christopher. 2004. Trust, Distrust and Antitrust. Texas Law Review 82: 515–680. Leslie, Christopher. 2010. Rationality Analysis in Antitrust. University of Pennsylvania Law Review 158: 262–353. Levin, Daniel. 1990. Horizontal Mergers:  The 50-Percent Benchmark. American Economic Review 80: 1238–45. List, John. 2003. Does Market Experience Eliminate Market Anomalies? Quarterly Journal of Economics 118: 41–71. List, John. 2004. Neoclassical Theory versus Prospect Theory: Evidence from the Marketplace. Econometrica 72: 615–25. Lundgren, Carl. 1996. Using Relative Profit Incentives to Prevent Collusion. Review of Industrial Organization 11: 533–50. McKelvey, Richard, and Thomas Palfrey. 1995. Quantal Response Equilibria for Normal Form Games. Games and Economic Behavior 10: 6–38. Malmendier, Ulrike, and Geoffrey Tate. 2005. CEO Overconfidence and Corporate Investment. Journal of Finance 60: 2661–700. Malmendier, Ulrike, and Geoffrey Tate. 2008. Who Makes Acquisitions? CEO Overconfidence and the Market’s Reaction. Journal of Financial Economics 89: 20–43.

228   Mark Armstrong and Steffen Huck Miller, Nolan, and Amit Pazgal. 2002. Relative Performance as a Strategic Commitment Mechanism. Managerial and Decision Economics 23: 51–68. Oechssler, Jorg. 2002. Cooperation as a Result of Learning with Aspiration Levels. Journal of Economic Behavior and Organization 49: 405–9. Offerman, Theo, and Jan Potters. 2006. Does Auctioning of Entry Licences Induce Collusion? An Experimental Study. Review of Economic Studies 73: 769–92. Offerman, Theo, Jan Potters, and Joep Sonnemans. 2002. Imitation and Belief Learning in an Oligopoly Experiment. Review of Economic Studies 69: 973–97. Page, William. 2009. The Gary Dinners and the Meaning of Concerted Action. S.M.U. Law Review 62: 597–619. Podolny, Joel, and Fiona Scott Morton. 1999. Social Status, Entry and Predation: The Case of British Shipping Cartels 1879–1929. Journal of Industrial Economics 47: 41–67. Potters, Jan. 2009. Transparency about Past, Present and Future Conduct. In Jeroen Hinloopen and Hans-Theo Normann, Experiments and Competition Policy. Cambridge University Press, 81–104. Rhode, Paul, and Mark Stegeman. 2001. Non-Nash Equilibria of Darwinian Dynamics with Application to Duopoly. International Journal of Industrial Organization 19: 415–53. Ridley, David. 2008. Herding versus Hotelling: Market Entry with Costly Information. Journal of Economics and Management Strategy 17: 607–31. Rogoff, Kenneth. 1985. The Optimal Degree of Commitment to an Intermediate Monetary Target. Quarterly Journal of Economics 100: 1169–89. Roll, Richard. 1986. The Hubris Hypothesis of Corporate Takeovers. Journal of Business 59: 197–216. Salant, Stephen, Sheldon Switzer, and Robert Reynolds. 1983. Losses from Horizontal Merger: The Effects of an Exogenous Change in Market Structure on Cournot-Nash Equilibrium. Quarterly Journal of Economics 98: 185–99. Schaffer, Mark. 1989. Are Profit-Maximizers the Best Survivors? Journal of Economic Behavior and Organization 12: 29–45. Schelling, Thomas. 1960. The Strategy of Conflict. Cambridge, MA: Harvard University Press. Scott Morton, Fiona, and Joel Podolny. 2002. Love or Money? The Effects of Owner Motivation in the California Wine Industry. Journal of Industrial Economics 50: 431–56. Simon, Herbert. 1955. A Behavioral Model of Rational Choice. Quarterly Journal of Economics 69: 99–118. Stackelberg, Heinrich von. 1934. Marktform und Gleichgewicht, Vienna: Springer. Stucke, Maurice. 2011. Am I a Price-Fixer? A Behavioral Economics Analysis of Cartels. In Caron Beaton-Wells and Ariel Ezrachi, Criminalizing Cartels: Critical Studies of an International Regulatory Movement. Oxford: Hart. Svenson, Ola. 1981. Are We All Less Risky and More Skillful Than Our Fellow Drivers? Acta Psychologica 47: 143–48. Tanaka, Yasuhito. 2000. Stochastically Stable States in an Oligopoly with Differentiated Products: Equivalence of Price and Quantity Strategies. Journal of Mathematical Economics 34: 235–53. Tor, Avishalom. 2002. The Fable of Entry: Bounded Rationality, Market Discipline, and Legal Policy. Michigan Law Review 101: 482–568. Vega-Redondo, Fernando. 1997. The Evolution of Walrasian Behavior. Econometrica 65: 375–84. Vickers, John. 1985. Delegation and the Theory of the Firm. Economic Journal 95: 138–47.

CHAPTER 10

E X P E R I M E N TA L E C ON OM IC S I N A N T I T RU ST WIELAND MÜLLER AND HANS-THEO NORMANN

10.1. Introduction IN this chapter, we will survey experimental economics research with a focus on antitrust. Chamberlin’s (1948) paper, which is generally thought to be the very first experiment in economics,1 already had a focus on competition and competitive markets. As we will see, antitrust issues are an important topic in experimental economics. We will try to identify the specific contribution of experimental methods as a complement to theoretical and (field data) empirical analysis. The chief motivation for running experiments was already clearly and distinctly expressed by Chamberlin (1948). Chamberlin argues that “[t]‌he data of real life are the product of many influences other than those which it is desired to isolate . . . Unwanted variables cannot be held constant or eliminated.” An experiment, however, allows to “study in isolation and under known conditions the effects of particular forces.” This approach complements the econometric approaches (where “unwanted” variables are controlled for as regressors). On top of that, the “conditions” (market parameters, cost values, etc.) are known to the experimenter, and, hence, equilibrium predictions can be made. This turns out to be a big advantage when experiments aim at testing game-theoretic models.

1  For a more comprehensive survey of the early market experiments, see Holt (1995), and Roth (1993) for an overview of early economics experiments in general. See also Normann (2008), from which this section partly draws.

230   Wieland Müller and Hans-Theo Normann The early market experiments (Chamberlin, 1948, Smith, 1962, 1964, Fouraker and Siegel, 1963, Friedman, 1963) analyze issues with implications for antitrust policy. They discuss which market institutions are likely to yield competitive outcomes and which may not. They also highlight the role of communication for successful price conspiracies. Whereas some experiments attempt to mimic real markets and institutions (Hong and Plott 1982, 1987, and Grether and Plott, 1984), other experiments were designed according to theoretical requirements. In section 10.2 of this chapter2 we will review experimental papers on collusion (tacit and explicit, conscious parallelism) and in section 10.3 those on exclusionary behavior (predatory pricing, foreclosure, exclusive dealing, and bundling). We do not introduce the methods of experimental economics. For a detailed introduction to methods, see Davis and Holt (1993) and Friedman and Sunder (1994); for a brief primer, see Normann (2008).

10.2. Collusion Both the Sherman Act and the European Treaty of Rome prohibit cartel agreements. Policies have changed over time, but with few exceptions explicit agreements are per se illegal, and practices facilitating collusion are scrutinized by antitrust authorities. Tacit (or implicit) collusion is treated entirely differently by law,3 with tacit collusion essentially being legal. Often, the dividing line between tacit and explicit collusion is whether communication about prices occurred. When it comes to anticartel policy, it is generally thought that “factors facilitating collusion” are useful for policy (Scherer and Ross, 1990, Tirole, 1989). The idea is that some industry structures are more prone to cartelization than others and may hence serve as a guide for cartel detection. Among these factors facilitating collusion, the limited number of firms is typically the first item on the list. This conventional wisdom goes back at least to Chamberlin (1948). The intuition is that fewer firms will find it easier to coordinate on collusive prices and punishment strategies. In this section, we will review experiments on tacit and explicit collusion. We will also deal with “intermediate” forms of collusion like price-signaling and conscious parallelism. We will put a focus on the aforementioned conventional wisdom that fewer firms will find it easier to collude.

2  Previous comprehensive surveys on this topic are by Plott (1982, 1989) and Holt (1995). A book-length treatment edited by Hinloopen and Normann (2009) contains more survey work. The current chapter complements the older surveys by reviewing more recent developments. This chapter also puts a strong emphasis on antitrust policy issues rather than broadly on all industrial organization topics. 3  See Kaplow (2011) on the legal treatment of tacit vs. explicit collusion.

Experimental Economics in Antitrust   231

10.2.1.  Tacit Collusion: “Two Are Few and Four Are Many” Presumably the first experiment analyzing tacit collusion in oligopoly is by Fouraker and Siegel (1963). The research reported in that book is, to a large extent, motivated by standard economic textbook models like bilateral monopoly and Cournot and Bertrand oligopoly. Fouraker and Siegel tested the theory of Cournot oligopoly by making participants choose outputs only and by endogenously determining the market clearing price and the resulting profits. They also cover the issue of how the number of firms affects competition in oligopoly. The advantage of these experiments is that one can “study in isolation and under known conditions the effects of particular forces,” to pick up to the Chamberlin quote again. The data can unambiguously be compared to the predictions, which, in contrast to many cases in the field, are of course known to the experimenter. A number of experimental papers test the conventional wisdom—that fewer firms find it easier to collude—in the lab. They analyze the impact of a mere change in the number of firms in the market on industry outcome. Only the number of players changes, but nothing else. No matter how many firms there are, and no matter whether a change in the number of competitors results from a merger or divestiture, firms are symmetric players. Fouraker and Siegel (1963), Dolbear and coauthors (1968), Binger, Hoffman, and Libecap (1990), Isaac and Reynolds (2002) and Huck, Normann and Oechssler (2004) essentially all confirm that market outcomes get more competitive the larger the number of firms. For some oligopoly models, this is exactly what the static Nash equilibrium suggests (Cournot competition, as studied in Fouraker and Siegel 1963, Binger, Hoffman, and Libecap 1990, and Huck, Normann, and Oechssler, 2004). For homogenous Bertrand competition (Isaac and Reynolds, 2002, Fonseca and Normann, 2012) this is not the case as the Bertrand Paradox suggest two firms are sufficient to generate perfect competition, but empirically the result still holds. At a closer look, the experiments reveal a more specific result: tacit collusion—supracompetitive pricing—is more or less restricted to duopoly. There can occasionally be moderately above-competitive prices with three firms, but with four or more firms, markets are competitive (at least in the sense of a noncooperative Nash equilibrium). That prices get more competitive the larger the number of firms confirms the conventional wisdom that a small number of firms facilitates collusion (tacit collusion—no communication—in this case). Figure 10.1 summarizes these findings. The figure compiles data from Dolbear and coauthors (1968), Huck, Normann, and Oechssler (2004), and Fonseca and Normann (2012). For oligopolies with different numbers of firms, the figure shows the average price, the static Nash equilibrium price (dotted line), and the collusive price (dashed line). These three studies were selected because they study more than two variants (that is, oligopolies of at least three different sizes). The top panel of figure 10.1 summarizes the data from Dolbear and coauthors (1968), who study differentiated Bertrand competition. In their design, the static Nash equilibrium price remains constant when the number of firms is varied. Also the incentives to

232   Wieland Müller and Hans-Theo Normann Differentiated Bertrand 25.0

Collusion

20.0

Nash

15.0 10.0 5.0 0.0

n=2

n=4

n=16

Homogenous Cournot 50.0

Collusion

40.0 30.0 20.0

Nash

10.0 0.0

n=2

n=3

n=4

n=5

Homogenous Bertrand 100.0

Collusion

80.0 60.0 40.0 20.0 0.0

Nash n=2

n=4

n=6

n=8

FIGURE  10.1  Average Treatment Prices and Deviations from the Collusive / Nash Predictions in Dolbear et  al. (1968), Huck, Normann, and Oechssler (2004), and Fonseca and Normann (2012)

collude (tacitly) remain fixed. If there is a “numbers effect,” it can only be due to general difficulties to collude among a larger number of players. The results indicate that the duopolies, with an average price of 19.8, are significantly above the Nash price of 18 but below the collusive price of 23. The four-firm oligopolies have an average price just at the Nash prediction. And the 16-firm treatment (for which there is only one group) is more “rivalistic,” with prices below Nash. The middle panel contains Cournot oligopolies with homogenous goods (Huck, Normann, and Oechssler, 2004). Here, the Nash equilibrium price varies with the number of firms (the price is given by 100 minus industry output in this experiment).

Experimental Economics in Antitrust   233

Again, prices are above the Nash level only in the duopoly treatment, and they fall short of the collusive price of 50.5. With four or five firms, averages prices seem to be even be more competitive than Nash, although this may not be statistically significant. The panel at the bottom summarizes Bertrand oligopoly with homogenous goods. The static Nash equilibrium price (which equals marginal cost) is zero here; the collusive price is 100. Fonseca and Normann (2012) report selling prices (winning bids). With four or more firms, markets are rather competitive. The duopolies manage to maintain (at least on average) a price level that is halfway between marginal cost and the monopoly price. Taken together, the results summarized in figure 10.1 indeed suggest that “two are few and four are many” (Huck, Normann, and Oechssler, 2004). Prices above the static Nash equilibrium are restricted to duopolies. And markets with four or more firms can be even more competitive than predicted by Nash.4 This is robust in that it has been shown for three different standard oligopoly models. Why does the regularity “two are few and four are many” arise? As pointed out by Holt (1995), one can distinguish between two possible effects of increasing the number of firms beyond two. Firstly, it is harder to monitor the specific decisions of rival players; and, second, it is more difficult to punish a specific rival. As a result, players’ actions may become ambiguous with more than two players: is a low price a deviation or is it an attempt to punish a previous defection? Below, we will learn about two ways to overcome coordination problems of this sort: communication and targeted punishments. These designs are part of the literature on explicit collusion. Concluding this section, we note Davis and Holt (1994) point out a certain drawback of the literature on “numbers effects”: the asset structure (capacities, cost functions) of firms and industry does not change in these experiments. A change in the number of firms is equal to entry or exit in the industry, and firms are always symmetric. Davis and Holt (1994) deviate from this. Their baseline treatment has five sellers, three large and two small firms, and no seller has market power (where market power is defined as the ability of a single firm to profitably raise the price above the competitive level). The second treatment reallocates some capacity, resulting in two large firms with market power and three small firms. The third treatment conducts a full merger: there are three firms with market power in this treatment, like a merger of the three small firms in the second treatment. The central result is that introducing market power, but keeping constant the number of firms, has a larger impact on prices compared to the reduction of the number of firms while keeping market power constant. Hence, number effects may not be important if they do not come along with changes of concentration.

4 

See Fonseca and Normann (2008) for an analysis how firm numbers and asymmetries interact. Huck et al. (2007) investigate numbers effects resulting from mergers.

234   Wieland Müller and Hans-Theo Normann

10.2.2.  Explicit Collusion Conspiracies against trade involve direct communication and explicit cartel-like agreements about price and output restrictions. As mentioned, such conspiracies or cartels are usually illegal and are therefore not enforceable in the field. Laboratory experiments take the nonenforceability into account by allowing for unstructured direct communication and price agreements but making the agreements nonbinding.5 The classic studies on explicit collusion are by Friedman (1963, 1967). He turned to a core antitrust topic: price conspiracies. Friedman (1967) was the first experiment to allow for direct communication between sellers and thus providing the opportunity to conspire against trade. The market game he used was a duopoly with price competition and product differentiation. Before prices were chosen, written messages could be sent to the other seller who was allowed to respond to the message. Friedman found that in three out of four cases a price conspiracy was formed, and that in 90% of these cases firms adhered to the agreement. Isaac, Ramey, and Williams (1984) were the first to show a significant price effect of seller communication. In their posted-offer markets, sellers were able to freely talk about markets, outputs, and prices before posting their prices and without the buyers knowing these talks. In this conspiracy treatment, prices were significantly higher than in the baseline posted-offer markets without the opportunity to conspire. Prices were above the competitive equilibrium and about halfway towards the monopoly price.6 The conclusion in Isaac, Ramey, and Williams (1984) that prices are “about halfway towards the monopoly price” is an interesting one but can hardly hold in general. Indeed, how much communication helps firms with attempts to collude is a significant question, and one would like to know exactly when communication helps and when it does not (Whinston 2008, p. 21). The question under which circumstances communication helps and how much is the research agenda of Fonseca and Normann (2012). They investigate the “numbers effect” in Bertrand oligopoly with various numbers of firms. They are the first to analyze more than one market setup with and without communication in order to find out when communication is particularly effective. Their results without communication are summarized in figure 10.1. In the oligopolies with communication, Fonseca and Normann

5  Experimental research that studies binding cartel agreements only recently appeared on the agenda. Kosfeld et al. (2009) study public-good settings with enforceable contracts, and Clemens and Rau (2012) investigate binding cartel contracts in a Cournot setting. 6  Davis and Holt (1998) run similar posted-offer experiments. In the treatments corresponding to Isaac Ramey and Williams (1984), they also obtain similar results. A novel aspect is that sellers were allowed to grant discounts to buyers (upon request) in some treatments. If discounts were given, they could not be observed by other sellers or buyers. The conspiracy treatment where discounts were allowed led to nearly perfectly competitive results. In a final treatment, information on net prices including discounts were provided ex post. This treatment led to higher prices again. The antitrust implication is that activities by firms or trade associations to limit or coordinate discounts should be watched with suspicion.

Experimental Economics in Antitrust   235

(2012) observe a significant increase in prices due to communication. Further, there is a decline in prices as the number of firms increases. Average selling prices range between 94 (n = 2) and 55 (n = 8) and the trend appears to be linear in the number of firms. Importantly and in contrast to the aforementioned experiments without communication, there is no discontinuity between two and four or more firms. This finding is new to the experimental literature. With communication (free, unrestricted chats with a messenger-type tool), collusion in the sense of supra-competitive prices can be overcome even with more than two firms. Fonseca and Normann (2012) identify three main channels through which communication supports collusion. First, it helps firms coordinating on a collusive pricing scheme. Second, communication is used for dispute mediation. Third, they find that communication has a long-lasting effect on cooperation (hysteresis) that remains even after the possibility of talking has been switched off. Roux and Thöni (2012) show that individually targeted punishments are also able to overcome the collapse of collusion with more than three players. They investigate Cournot oligopolies with two, four, and six firms. Their standard treatment (without punishments) nicely confirms the above mentioned Cournot results: some tacit collusion in the duopolies, but outputs of the Nash benchmark and above with four and six firms. Next, they introduce a variant with punishments à la Fehr and Gächter (2000): subjects first decide about their output and, after feedback on individual quantity decisions, players may then assign individually targeted punishment points. These punishment points cause a cost to both the punisher and the punished. It turns out that such opportunities to punish improve cooperation, and significant degrees of collusion emerge even with four and six firms.

10.2.3.  Nonbinding Price Announcements and Conscious Parallelism Nonbinding price announcements in the trade press have often been of concern to antitrust authorities.7 This is particularly the case when the announcements lead to parallel pricing or conscious parallelism in an industry. Price announcements and conscious parallelism can be prohibited both in the United States and in the EU. Leading cases include American Tobacco in the United States (see, e.g., Scherer and Ross, 1990), and the Woodpulp case (see Phlips, 1995)  and Teerfarben (dyestuff) case in Europe (see Markert, 1974). Several interesting experiments deal with these practices. The experiments have in common that they do not allow for explicit conspiracies. There was no direct 7 

The advance notice of binding prices has been studied in the pioneering work of Hong and Plott (1982) and Grether and Plott (1984). In those papers, the advanced notices were such that the prices noted in advance could not be retracted. See Normann (2008) for a discussion. Ex post notification of prices can cause lower rather than higher prices. See Huck, Normann, and Oechssler (1999, 2000).

236   Wieland Müller and Hans-Theo Normann communication of any sorts. Communication was impersonal and structured by the experimenter, and was about prices or price changes only. In this way, the experiments attempt to simulate the style of price announcements commonly found in the press of trade associations. Up front, we mention that no experiment in this area has tested for numbers effects. The first experiment in this line is by Holt and Davis (1990). They ran posted-offer markets with three firms. Prior to posting a price, participants could fill in the blank in the following message “____ is an appropriate price for the market this period.” The other firms could respond with an “A” for agreement, “H” for too high, and “L” for too low. First, there are 15 periods without communication followed by at least 10 periods with announcements. They find that the announcements increase prices initially but this effect is only transitory. Sellers tended to post price just a little below the announced price. As a result, prices declined and towards the end matched the level of prices without announcements. Cason (1995) and Cason and Davis (1995) are motivated by the Airline Tariff Publishing (ATP) system. Eight airlines used the ATP system, which enabled them to continuously announce, modify, and retract nonbinding price announcements of future tickets. Customers could not observe these announcements. The airlines were accused of illegally raising ticket prices with the help of the ATP system. The airlines stopped using the system in 1994. Cason (1995) analyses the effects of the nonbinding announcements. The announcements were less structured than in Holt and Davis (1990) but, as above, there was no explicit communication. Crucially, the last price announcement in a one-minute window became binding. This is in contrast to Holt and Davis (1990), where the binding price decisions were made under strategic uncertainty. The results are, however, qualitatively similar to those in that paper. The announcements only have a transitory effect, as prices converge to the competitive level. Cason and Davis (1995) extend this setup to one with multiple markets—a feature obviously important in the airline industry. Here, prices are significantly higher when price announcements resembling the ATP system are possible. Harstad, Martin, and Normann (1998) investigate conscious parallelism. The underlying market model in that paper is price-oligopoly with differentiated products and four asymmetric firms. There are two types of firms, with high or low marginal production cost, and both static Nash equilibrium and joint-profit maximum have these two types of firms charge different prices. There is a baseline treatment 1 without any communication. Treatment 2 is with communication. It is similar to Holt and Davis (1990), although there was no response possibility. Participants could make simple nonbinding price announcements before setting their actual prices. Novel is treatment 3, where firms could make announcements of price changes. Whenever a participant made a price change announcement, the other firms are informed with a message reading “Firm 3 announces a price change of +1.25.” The other participants were then required either to match the increase or not. After all participants had made their match / not match

Experimental Economics in Antitrust   237

decisions, a screen with the vector of price changes would be shown. Finally, actual prices were set independently. Treatments 1 and 2 do not appear to be different. Indeed, the regressions on price-cost margin reported in Harstad, Martin, and Normann (1998) show that this difference is either insignificantly negative or, if some other factors are controlled for, significantly but only moderately positive. This result is consistent with Holt and Davis (1990). Treatment 3, however, does have a positive impact on prices. The differences are substantial and statistically significant in all three regressions reported in the paper. This is an interesting finding compared to the previous literature.

10.2.4.  Price Ceilings and Price Floors Price ceilings and price floors are a common instrument of antitrust policy. They have been used at least since ancient Greek times and today are known from markets as diverse as food, steel, rents, fees for physicians, interest rates, and electricity spot market prices. We will discuss price ceilings first. The effect of such a maximum price and its appeal to the policymaker are easily explained. Price ceilings prevent prices from getting higher than the ceiling but, the argument goes, they do not harm competition that may push prices below the ceiling. (To be precise, only nonbinding price ceilings, that is, those above the competitive equilibrium price, have this property. Binding price ceilings can lower welfare, as they cause excess demand.) If this argument is correct, this policy cannot go wrong. A price ceiling may effectively prevent collusion and at worst has no impact at all. Industrial economists have challenged the conclusion that price ceilings cannot weaken competition with the theory of focal points (Scherer and Ross, 1990). Their argument is based on the observation that many prices can occur as outcomes of collusive equilibria in infinitely repeated games (this is the so-called Folk Theorem). This implies that firms face a coordination problem when attempting to collude without explicit communication. Here, a price ceiling may serve as a focal point on which firms coordinate. If that is the case, price ceilings would facilitate tacit collusion and lead to higher prices—contradicting the logic that the price-ceiling policy cannot go wrong. Several experimental papers deal with price ceilings. Isaac and Plott (1981) and Smith and Williams (1981) analyze double auction markets with price ceilings. They clearly reject the focal-point hypothesis. By contrast, their data suggest that nonbinding price ceilings actually lower prices, consistent with the traditionally expected effect. Prices often failed to converge to the competitive equilibrium, leaving a “buffer” between the price ceiling and actual prices that prevented full convergence. Coursey and Smith (1983) analyze price ceilings with posted-offer markets. They obtain the same results qualitatively as Issac and Plott (1981) and Smith and Williams (1981). Markets converge to the competitive equilibrium, and price ceilings as focal

238   Wieland Müller and Hans-Theo Normann point do not play much of a role. A price ceiling often lowers average prices, as prices can no longer converge from above, and markets converge to a price below the ceiling. The more recent research by Engelmann and Normann (2009) is motivated by two observations. First, the results by Isaac and Plott (1981) and Smith and Williams (1981) are hardly surprising because of the well-known strong tendency of double auction markets to converge to the competitive equilibrium. Second, Coursey and Smith’s (1983) posted-offer markets results present stronger evidence against the focal point hypothesis. Posted-offer market often have prices above the competitive equilibrium, so a price ceiling may reduce average prices. However, in Coursey and Smith’s (1983) design the incentives to collude are extremely small (less than 10% of the profits at the competitive equilibrium). Hence, the second observation by Engelmann and Normann (2009) is that the price ceiling in Coursey and Smith is possibly binding with respect to Nash-equilibrium prices. As mentioned above, posted-offer markets typically have Nash equilibria with prices above the competitive equilibrium that cannot be qualified as collusive. The Nash equilibrium prediction is not reported and cannot be calculated from the details reported in the paper. If there are such supracompetitive Nash prices, this could explain why a price ceiling that is nonbinding with respect to the competitive equilibrium still leads to lower prices. The design of Engelmann and Normann (2009) establishes the upper end of the competitive price range as unique Nash equilibrium and provides substantial incentives to collude at the price ceilings. Therefore, it may constitute a harder test of the focal-point hypothesis. Their results, however, fully confirm the results of Coursey and Smith (1983). Their experimental data do not show a collusive effect of price ceilings at all. Prices are initially lower with a ceiling than without one, and, later in the experiment, they do not differ significantly. The maximum admissible price is virtually never chosen. Engelmann and Müller (2011) resume the search for a collusive focal-point effect of price ceilings in laboratory markets. They changed the design of Engelmann and Normann (2009) in two main respects: fixed pairs of players interacting repeatedly and asymmetric cost schedules for the sellers. Moreover, they ran treatments with and without information about the demand curve. Despite these changes, results again failed to support the focal-point hypothesis. Collusion was as unlikely in markets with a price ceiling as in markets with unconstrained pricing. Overall, static Nash equilibrium predicts the data fairly accurately. Engelmann and Müller (2011) argue this might warrant reinterpretation of field studies on anticompetitive effects of price ceilings. We now turn to price floors. Price floors have been analyzed in Issac and Plott (1981) and Smith and Williams (1981), and, recently, in Dufwenberg and coauthors (2007). The first two papers show that price floors near, but above, the competitive equilibrium raise prices. This is the expected effect. If double auctions normally converge to the competitive equilibrium, then a price floor above the competitive equilibrium will prevent such convergence and prices will be higher.

Experimental Economics in Antitrust   239

Dufwenberg and coauthors (2007), however, find that price floors may actually lower prices. They analyze simple duopoly experiments with perfect Bertrand competition (or, alternatively, a posted-offer market with a rectangular design). They find that, without the floor, subjects’ choices are somehow bound away from marginal cost pricing, whereas, with the price floor, prices substantially above marginal cost are often chosen. As a result average prices are higher without the floor.

10.3.  Exclusionary Behavior Article 102 of the Treaty on the Functioning of the European Union (TFEU) prohibits any abuse of a dominant position and contains several examples for abusive practices: (a) directly or indirectly imposing unfair purchase or selling prices or other unfair trading conditions; (b) limiting production, markets, or technical development to the prejudice of consumers; (c) applying dissimilar conditions to equivalent transactions with other trading parties, thereby placing them at a competitive disadvantage; (d) making the conclusion of contracts subject to acceptance by the other parties of supplementary obligations that, by their nature or according to commercial usage, have no connection with the subject of such contracts. This is not an exhaustive list, and so other price or nonprice strategies may be considered an abuse as well. Importantly, Article 102 TFEU does not forbid certain types of behavior as such, but prohibits dominant firms from using them. Section 2 of the Sherman Act in the United States aims at preventing “monopolization” of markets. In applying this law, courts have not declared unlawful monopolies per se. Rather, they have stated that it is directed “not against conduct which is competitive, even severely so, but against conduct which unfairly tends to destroy competition itself.”8 The Clayton Antitrust Act lists examples of unlawful practices that may lead to monopolization such as (a) price discrimination between different purchasers, if such discrimination tends to create a monopoly; (b) exclusive dealing agreements; or (c) tying arrangements. In this section we review experimental evidence on exclusionary behavior related to Article 102 TFEU, Section 2 of the Sherman Act, and the Clayton Act in the United States. Arguably in the spirit of these pieces of legislation, we concentrate on situations in which firms with significant market power engage in strategic behavior that might hurt consumer or total welfare. In doing so, in subsections 10.3.1 and 10.3.2 we heavily draw on related material presented in van Damme, Larouche, and Müller (2009). In order to economize on space, when we review the experimental literature on abusive practices, we will usually not provide details on the underlying economic theories; for this we refer to, for example, Motta (2004).

8 

Spectrum Sports, Inc. v. McQuillan, 506 U.S. 447, 458 (1993).

240   Wieland Müller and Hans-Theo Normann

10.3.1.  Predatory Pricing The first paper investigating the possible emergence of predatory pricing9 in the laboratory is Isaac and Smith (1985). They implement a posted-offer duopoly market that is served by a large seller and a small one. The large seller has a cost advantage over the other seller, a higher capacity, and a larger cash endowment. Sellers don’t know (the simulated) demand, or each other’s cost functions. There is a predatory price range where the large seller provides 10 units at a price that is (a) lower than the small seller’s minimum average price; (b) lower than the large seller’s marginal cost of his tenth unit, and (c) demand is exhausted. In this case, the small seller cannot earn positive profits, while the large seller does not suffer losses, although his profits at the predatory price-quantity combination are lower than at the competitive equilibrium. In these markets no predatory pricing is observed. This “negative” result induced Isaac and Smith to make several design changes. In a first design change, sellers are required to purchase entry permits. But again, no predatory pricing is observed. As a second design change, sellers are given full information about each other’s costs. Yet, again, no predatory pricing is observed. Thus, this first “search of predatory pricing” was unsuccessful. Isaac and Smith conclude:  “We are unable to produce predatory pricing in a structural environment that, a priori, we thought was favourable to its emergence” (1985, p. 342). Harrison (1988) continues the search for predatory pricing. He modifies the Isaac and Smith design by implementing five simultaneous posted-offer markets and introducing 11 sellers. Four sellers are told that they would become large sellers (with costs as introduced in the Isaac and Smith experiments) if they would choose to enter “their” market, but that they would become small sellers in case they entered any other market. The seven remaining sellers are small sellers no matter which market they chose to enter. Efficiency demands that each of the four large sellers enter his preferred market, each of which would also see entry of a small seller, and that the remaining three small sellers enter the fifth market, in which no seller would have a cost advantage. The fifth market serves as an “active escape opportunity” for the small sellers. Demand is simulated in each market. Finally, at the beginning of each market period, firms choose a market to enter, a price, and a corresponding maximum quantity. With this setup, Harrison reports cases of predatory pricing in the sense defined above. However, Goeree and Gomez (1998) replicate the Harrison study with the result that only 3 out of 144 price decisions of large buyers could possibly be classified as predatory. Note that in this design entry, price and quantity decisions are made simultaneously, which means that, for example, a large seller never knows whether he would be a monopolist in his market or whether he would share it with a small seller.

9  See Gomez, Goeree, and Holt (2008) as well as Capra and coauthors (2000) for other overviews on predatory pricing experiments.

Experimental Economics in Antitrust   241

Consequently, Capra and coauthors (2000) vary Harrison’s design in two respects: (a) small sellers first choose their markets (which become commonly known) before all sellers choose prices and quantities (large sellers can now react to entry and possibly raise prices after exit); (b) large incumbent sellers have complete information about demand and other’s costs, whereas small sellers only know own costs. Presumably, this makes it more likely that a large seller prices more aggressively. In fact, Capra and coauthors (2000) report that predatory pricing consistently arose in most markets. Jung, Kagel, and Levin (1994) report on tests of Selten’s (1978) chain-store game and Kreps and Wilson’s (1982) reputation model.10 In the model by Selten, there is an incumbent (a chain store) that is a monopolist in a finite number of separate markets where it is operating the same technology and producing the same products. In each market, the incumbent faces potential entry. More precisely, in the first market, first the entrant decides about entry and then, if entry occurs, the incumbent decides whether to fight or to accommodate. The outcome of this interaction becomes publicly known before in the second market the same stage game is played between the incumbent and a new entrant, and so on. In Selten’s model the incumbent is “weak” and prefers to share each market in which entry occurs. Consequently, by backward induction each entrant enters and is always accommodated. Kreps and Wilson’s (1982) model is similar to Selten’s; however, entrants have incomplete information regarding the type of the incumbent: with some probability the incumbent is “strong” (low cost) and would always fight entry; with complementary probability the incumbent is “weak” (high cost) and, as in Selten’s model, prefers to share each market. Kreps and Wilson derive a sequential equilibrium in which the weak incumbent fights entry in the first markets to make the entrants believe it is strong. Only towards the end of the game does it start to accommodate entry. The fact that weak incumbents fight entry in early periods of the game can be interpreted as predatory pricing. (In fact, Kreps and Wilson [1982] define predation exactly in this way.) Jung, Kagel, and Levin (1994) implement markets in which a monopolist plays a sequence of eight periods against different entrants. They implement both a version of the Selten setup, as well as a version of the Kreps and Wilson model. Using the Kreps and Wilson (1982) definition of predatory pricing, they report that, for experienced subjects, predatory pricing occurred in 100% of the cases in the Kreps and Wilson model and in 85% of the games in the Selten model. Furthermore, entry rates during early periods of the game are reported to be near 0% in the Kreps and Wilson model and as low as 30% in the Selten model. In the Selten model, the experimental results thus differ markedly from what theory predicts. However, regarding the Kreps and Wilson model, some deviations are also reported. First, entry rates increase constantly towards the end of a game, while the Kreps and Wilson model predicts fixed entry rates during these periods 10 

Jung, Kagel, and Levin (1994) is not the first and only paper reporting experimental results on reputation building. Camerer and Weigelt (1988), Neral and Ochs (1992), as well as Brandts and Figueras (2003) also report on reputation building in the lab. However, they cast the Kreps and Wilson model in a lender-borrower frame.

242   Wieland Müller and Hans-Theo Normann of mixed-strategy play. Furthermore, entrants do not enter more often after periods in which no entry occurred than after periods in which entry and fighting occurred.

10.3.2. Foreclosure Rey and Tirole (2007) define foreclosure as “a dominant firm’s denial of proper access to an essential good it produces, with the intent of extending monopoly power from that segment of the market (the bottleneck segment) to an adjacent segment (the potentially competitive segment).” These authors, however, also point out that a monopolist on an upstream market, faced with competition on the downstream market, may not be fully able to exploit its monopoly power, as it may not be able to make a credible commitment to downstream firms that it will restrict output. Consequently, downstream firms will not accept contracts that allow the upstream producer to extract full monopoly profits. Vertical integration can resolve this commitment problem and enable the upstream monopolist to fully exploit its market power. Such integration may, however, be blocked by the merger regulation.11 Martin, Normann, and Snyder (2001) study the commitment problem of such an upstream monopolist in an experimental setting. The basic structure these authors consider has a single upstream firm that produces an input at constant average and marginal costs and two downstream firms that convert each unit of input into a unit of a homogeneous final good. The upstream monopolist can simultaneously make take-it-or-leave-it contract offers to each of the downstream firms, specifying a quantity and a fixed payment it demands for the bundle. In a first treatment, these offers become publicly known before downstream firms decide. Public contracts serve as a commitment device enabling the upstream monopolist to earn monopoly profit by offering contracts that consist of offering half the monopoly output at half the monopoly profit to each downstream firm. If, however, the contracts are negotiated privately, so that a downstream firm will not observe the contract the upstream monopolist offers to the other downstream firm, the upstream monopolist may no longer be able to obtain the monopoly profit. Predictions in this treatment depend on out-of-equilibrium beliefs of a downstream firm concerning the contract that is being offered to its rival. More precisely, if a downstream firm entertains passive beliefs, it thinks that its rival receives the equilibrium offer. In this case it can be shown that output in the downstream market is higher, and the upstream monopolists’ profits are lower than in the joint-profit-maximizing outcome.12 A third treatment 11  In the second classic foreclosure framework, developed by Ordover, Saloner, and Salop (1990), there is upstream duopoly. They argue that markets with one vertically integrated firm should be less competitive than those where both upstream firms are separate. The experiment of Normann (2011) confirms that vertical integration is anti-competitive—although the price increases were only moderate and qualitatively quite different from what the Ordover, Saloner, and Salop model would predict. 12  There is, however, another equilibrium (with symmetric beliefs) in which the outcome is the same as with public contracts. However, Rey and Tirole (2007) argue that the assumption of passive beliefs is theoretically more sound.

Experimental Economics in Antitrust   243

implements the case where the upstream monopolist integrates vertically with one of the downstream firms. Here, the integrated firm can commit to sell the monopoly quantity through its downstream subsidiary and not supply the other downstream firm at all; hence, there is foreclosure. Martin, Normann, and Snyder (2001) find partial support for the arguments from Rey and Tirole (2007) and the foreclosure theory. When contracts are secret (i.e., in the presence of a commitment problem), outputs are higher and the upstream monopolists’ profits are lower than when contracts are public or in case of an integrated firm. However, the differences are not as pronounced as theoretically predicted. Moreover, the experimental results differ from the theoretical predictions with regard to the division of profits between upstream and downstream firms. Theory predicts that the upstream firm should have all the bargaining power (by making take-it-leave-it offers) such that it should be able to extract all of the industry profits. This is observed most of the time in the treatment with an integrated firm. However, in the two nonintegrated treatments, with either public or secret contracts, the upstream monopolist only obtains a fraction of industry profits. It seems that the threat of the downstream firms to reject the upstream firm’s offer limits the latter firm’s bargaining power.13 This unpredicted bargaining effect provides another rationale for vertical integration, different from the ones that are usually discussed in the literature. As Martin, Normann, and Snyder conclude: “other vertical restraints may not allow the upstream firm to extract as much industry rent as full vertical integration” (2001, p. 479).

10.3.3.  Exclusive Dealing Since the beginning of the twentieth century courts have treated firms using exclusive contracts harshly for fear such contracts could be used to exclude rivals and, thus, hamper competition. Proponents of the Chicago school argued that such fears are not warranted since using exclusive contracts for the sole purpose of anticompetitively excluding rivals would not be in the interest of rational profit-maximizing firms. Recently, this view on exclusive dealing has been challenged by various theorists who describe circumstances under which anticompetitive exclusion of rivals can be profitably used by dominant firms. One prominent contribution to this literature is the “naked exclusion” model14 put forward by Rasmusen, Ramseyer, and Wiley (1991) and Segal and Whinston (2000) (henceforth RRW-SW). There are now several experimental studies based on the setups considered in these papers. 13  Martin, Normann, and Snyder (2001) discuss (dis)similarities of their treatments with the ultimatum game for which experimental results reject the subgame perfect equilibrium prediction with its extremely asymmetric payoff consequences. 14  The term “naked” refers to the sole purpose of an exclusive deal to exclude a rival without having any efficiency justification.

244   Wieland Müller and Hans-Theo Normann To fix ideas, consider an incumbent seller, a more efficient entrant and two buyers with independent demand. Due to economies of scale caused by, for example, fixed entry costs, the entrant needs both buyers to be “free” (i.e., not bound by exclusive contracts with the incumbent) to enter the market profitably. An exclusive contract in this framework takes the form of a payment from the incumbent to a buyer in exchange for the buyer’s promise to buy exclusively from the incumbent. RRW-SW show that, under mild assumptions, the incumbent needs to “convince” only one buyer in the market to sign an exclusive contract to deter entry and extract monopoly profits from both buyers. Indeed, compensating one buyer for the forgone consumer surplus that results from dealing with the incumbent is sufficient to obtain exclusion when different (i.e., discriminatory) contracts can be proposed to the two buyers. Moreover, if buyers are approached sequentially, exclusion is achieved at negligible costs. The idea is that the first buyer anticipates that, if he rejects a contract, the incumbent can convince the second buyer to accept by making him an offer he cannot refuse. Hence, the first buyer will accept any offer, even a small one, which deters entry. This stands in contrast to the case where it is impossible for the incumbent to discriminate between the buyers. In this case, exclusion is not guaranteed, as the monopoly profit the incumbent would earn under exclusion is not sufficiently high to compensate both buyers for their forgone surplus. The buyers’ subgame is then a symmetric coordination game with multiple equilibria, and exclusion occurs only if the buyers fail to coordinate on the (more efficient) rejection equilibrium. Landeo and Spier (2009) report experimental evidence regarding the case of simultaneous offers and distinguish between treatments in which buyers can communicate with each other and those in which they cannot. Their experimental design features an incumbent and two buyers. The following main findings are reported. First, when the incumbent seller must make the same offer to both buyers (no discrimination), it is found that communication reduces the likelihood of exclusion substantially (from 92% to 43%), as buyers in this case are more successful in coordinating on rejecting their offer. Second, in the case where the incumbent can discriminate among buyers and offers are publicly known, the exclusion rate is found to be independent of buyers’ ability to communicate with each other, with the exclusion rate being about 80% in both cases. Third, if the incumbent can discriminate among buyers, but offers are private information to the buyer receiving it, the exclusion rate is about 10 percentage points lower than in the case of discriminatory and publicly known offers. Hence, Landeo and Spier (2009) imply that inefficient exclusion by means of exclusive deals is a serious issue in lab markets. A second experimental study on the RRW-SW framework is Smith (2011), who focuses on the case where an incumbent cannot discriminate between buyers. Smith finds that changes in the number of buyers in the market does not significantly affect exclusion rates (fact 1). However, inefficient exclusion is harder for the incumbent to achieve when he needs a larger fraction of buyers to accept an exclusive deal (fact 2) and when buyers are allowed to communicate with each other (fact 3). Note that fact 3 is in line with a result reported in Landeo and Spier (2009). A possible interpretation of fact

Experimental Economics in Antitrust   245

2 is that more scrutiny on the part of competition authorities is needed when industries have larger fixed costs, smaller technological advantages of the potential entrant, or larger scale efficiencies. A third paper investigating the effects of exclusive deals in lab markets is Boone, Müller, and Suetens (2014). These authors also consider the case of an incumbent and two buyers and run treatments in which the incumbent can make simultaneous discriminatory and nondiscriminatory offers. However, Boone, Müller, and Suetens (2014) mainly concentrate on the case of (discriminatory) sequential offers that differ only in whether or not the offer made to the first buyer is observable to the second buyer when the latter is approached by the incumbent. Recall that with sequential offers the incumbent is predicted to achieve exclusion at negligible costs (a result that is independent of the information held by the second buyer). The findings in Boone, Müller, and Suetens (2014) are as follows: First, exclusion rates in simultaneous discriminatory and nondiscriminatory regimes are not significantly different from one another. This result is in line with a finding reported in Landeo and Spier (2009). Second, exclusion rates in the sequential regimes are significantly higher than in the nondiscriminatory regime and the simultaneous discriminatory regimes. Hence, whereas theory does not predict exclusion to be sensitive to the type of discriminatory regime, the experiment reveals that sequential offers facilitate exclusion to a higher extent than simultaneous offers. Third, exclusion costs for the incumbent are significantly higher than zero in all treatments and statistically indistinguishable across all treatments. Hence, exclusion costs are substantial also under sequential contracting, even though they are predicted to be (close to) zero. From an efficiency point of view, it seems immaterial whether the contracts are offered either simultaneously or sequentially. Therefore, in the light of the second finding in Boone, Müller, and Suetens (2014), an argument can be made that contracts offered sequentially should be interpreted as being more likely to aim at exclusion only compared to contracts that are offered simultaneously. For practical purposes, this would mean that an antitrust authority should be on high alert if the suspected company staggered its contracting with buyers over a certain period of time to get the required sequencing of offers (see also Whinston, 2008, p. 147).

10.3.4. Bundling Bundling refers to the practice of selling two or more distinct goods as a package at a single price. Although bundling is a ubiquitous selling practice that might enable consumers to purchase more attractive packages (at a price discount), bundling firms have come under attack for fears this practice may exclude competitors.15 15 

Another form of price discrimination are rebate schemes. Despite the central role rebate schemes play in the discussion of academics and practitioners and in the case law of abusive practices, experimental results regarding their effects are very scarce. A notable exception is Maier-Rigaud and Beckenkamp (2007), who report that despite high error rates, rebate schemes exert a significant

246   Wieland Müller and Hans-Theo Normann A common setup analyzed in the theoretical literature features a firm that has monopoly power in one market but faces competition by a second firm in another market. Offering its two products as a bundle may allow the multiproduct firm to leverage its monopoly power to the market where it faces a rival and, under certain conditions, decrease economic welfare. The experiments reported in Caliskan and coauthors (2007) and Muris and Smith (2008) try to carefully implement features that were theoretically shown to be conducive to generating anticompetitive exclusion by means of bundling and to reduce welfare (e.g., Nalebuff 2004, Greenlee, Reitman, and Sibley 2008). The rich design features, among other elements, a posted-offer market environment; a bundling firm that is dominant in one market in which a fringe firm is also present but faces competition in another market; the absence of efficiencies from bundling; perfectly and positively correlated demands for the goods sold in the two markets; and the impossibility of hit-and-run entry by means of introducing market entry and exit costs. Summarizing the findings, Muris and Smith (2008, p. 403) state Despite an experimental setting designed to produce welfare decreasing bundling by a monopolist, the experimental results did not generally find that allowing a monopolist to bundle decreased total or consumer surplus. Rather, when compared to settings prohibiting bundling, total surplus increased. Moreover, consumer surplus typically increased when bundling was allowed and, rarely, fell.

Muris and Smith’s (2008) results suggest that bundling may harm competitors but not competition, and they interpret their results as evidence against recent models of exclusionary bundling—a view emphatically contested in Greenlee, Reitman, and Sibley (2011). Hinloopen, Müller, and Normann (2014) also analyze the impact of product bundling in experimental markets, where a firm has monopoly power in one market but faces (imperfect) competition by a second firm in another market. They compare treatments where the monopolist can bundle its two products to treatments where it cannot bundle. Additionally, they vary the order of moves and have the two firms move either simultaneously (Cournot) or sequentially (Stackelberg), where in the latter case it is the multiproduct firm moving first. In contrast to Caliskan and coauthors (2007), firms face a downward-sloping demand curve in both markets. In the Cournot setting without bundling, Hinloopen, Müller, and Normann (2014) find that firms roughly play as predicted by theory. In particular, the multiproduct firm increases the number of units offered in the duopoly market, once it is allowed to bundle its products. When the multiproduct firm is the Stackelberg leader but cannot bundle, the previously observed nonequilibrium behavior of the followers is observed, in the “attraction” in their lab experiments, which leads them to suggest that potential foreclosure effects based on standard risk-neutral maximization behaviour may be underestimated. See Normann, Ruffle, and Snyder (2007) for an experiment on large-buyer discounts.

Experimental Economics in Antitrust   247

sense that Stackelberg followers substantially produce more than predicted, effectively reducing the Stackelberg leader’s first-mover advantage (see Huck Müller, and Normann 2001). However, the predicted asymmetric equilibrium in the sequential-move case of Stackelberg with product bundling is closely attained in the lab; large payoffs asymmetries, which are larger than under Stackelberg competition absent bundling, are observed. Importantly, in both the Cournot and the Stackelberg case bundling significantly reduces consumer and total welfare. In sum, and in contrast to some of the findings in Caliskan and coauthors (2007), in Hinloopen, Müller, and Normann (2014) bundling is found to work as a commitment device that enables the transfer of market power from one market to another with detrimental effects on consumer welfare. In view of the partly opposing findings in the two experimental papers discussed in this section and the paucity of available data, much more experimental and empirical work seems needed to test theories of exclusionary bundling and to delineate conditions under which bundling might or might not be (consumer) welfare reducing. In this section we focused on experimental evidence on exclusionary behavior aimed at driving rivals from the market. We briefly note, however, that there are also some experimental papers analyzing entry deterrence behavior. These papers fall into two categories: limit pricing16 and strategic investment into excess capacity as a barrier to entry. We briefly summarize experimental papers in these two categories in turn. Cooper, Garvin, and Kagel (1997a and 1997b) test a game based on the Milgrom and Roberts (1982) model of limit pricing and find that the strategic use of limit pricing to deter entry does emerge in the laboratory. Müller, Spiegel, and Yehezkel (2009) test the oligopoly limit-pricing model of Bagwell and Ramey’s (1991) and find lab support for the “no-distortion” separating equilibrium highlighted by Bagwell and Ramey’s (1991). In this equilibrium, incumbents play as if there was complete information or no entry threat at all: costs are revealed by first-period prices set by the incumbents, and entry takes place only if it is profitable (i.e., when costs are low). Regarding investment into capacity, in a first paper Mason and Nowell (1998) test a simple model based on Dixit (1980) and find that attempted entry deterrence by incumbents is relatively common and that it becomes more so as time unfolds. However, a substantial fraction of the incumbents do not try to exploit their first-mover advantage. In another paper, Brandts, Cabrales, and Charness (2007) report on a test of the entry-deterrence model by Bagwell and Ramey (1996), whose main feature is that not only the incumbent but also the entrant can (partially) precommit to investing into capacity. The model has two subgame perfect equilibria in pure strategies: one of the firms produces and becomes a monopolist while the other firm shuts down. However, a forward induction argument selects the entrant’s preferred equilibrium, implying that 16  A firm engages in limit pricing if it sets its price and output so that there is insufficient demand for another firm to profitably enter the market. Consumers benefit from lower prices in the phase of limit pricing, but this advantage might be offset by higher prices due to reduced competition if entry is successfully deterred. To the best of our knowledge, there are no actual court cases involving complaints of limit pricing.

248   Wieland Müller and Hans-Theo Normann there is a second-mover advantage. Contrary to this prediction, Brandts, Cabrales, and Charness (2007) find that the incumbent becomes the monopolist three times as often as the potential entrant (and that costly preinstallation is relatively rare for both the incumbent and the entrant).

10.4. Conclusion In this chapter, we reviewed economic experiments with relevance for antitrust policy, specifically with a focus on collusive exclusionary practices. Arguably, competition authorities are using more and more economic reasoning and there is an increasing tendency in the application of theoretical models of, for example, strategic behavior in the analysis of antitrust cases (see, e.g., Brennan 2000, or Muris and Smith 2008). It would, therefore, be important to have a wealth of empirical evidence showing that the models competition authorities rely on are indeed valid. Such evidence would help policymakers and courts to distinguish “normal” competitive behavior from illegal practices. Surely there is no shortage of theoretical models of strategic behavior. What is missing, however, is empirical evidence, which, unfortunately but unsurprisingly, is often hard to come by, due to, for example, private information on costs that is inaccessible for outside observers (see also, e.g. Scherer and Ross, 1990, and Wilson, 1992). A unique advantage of experimental methods is that all parameters of a market are under control and observable. This is not only true for firms’ private information such as costs, but also for conspiratorial activities as in Davis and Holt (1998). The value of experimental economics for antitrust practitioners is to help separate the theories that are empirically relevant from those that are not or to help assess the usefulness of a theory’s (behavioral) assumptions or proposed mechanisms (see, e.g., Bolton, Brodley, and Riordan 2000). Experiments show that tacit collusion is mainly restricted to duopoly; with four or more firms, usually no supracompetitive prices occur. If explicit communication between sellers is possible, price conspiracies are often successful even with six or eight firms. Antitrust authorities are right in looking at any form of direct communication between competitors with suspicion. By contrast, nonbinding price announcements without direct contact among sellers usually only have a transitory effect. An exception to this last result occurs when the price announcements are particularly suggestive to parallel conduct. The contribution of experimental economics to the testing of theories of exclusionary behavior surveyed in section 10.3 of this chapter show that this is indeed possible and delivers useful insights. However, such experimental tests are still scarce, so that there continues to be a mismatch between issues dealt with in courts and those subjected to experimental tests and a general lack of experimental (and empirical) evidence on models of strategic behavior. In particular what seems missing is (more) experimental evidence on issues such as the effects of rebate schemes, bundling and tying, refusal to deal,

Experimental Economics in Antitrust   249

and exclusive-dealing clauses (see also Davis and Wilson 2002, Wellford 2002, or van Damme, Larouche, and Müller 2009). Much more work needs to be done and should prove useful here.

Acknowledgment We are grateful to Tomaso Duso for helpful comments.

References Bagwell, Kyle, and Garey Ramey. 1991. Oligopoly Limit Pricing. Rand Journal of Economics 22:155–72. Bagwell, Kyle, and Garey Ramey. 1996. Capacity, Entry and Forward Induction. Rand Journal of Economics 27:660–80. Binger, Brian R., Elizabeth Hoffman, and Gary D. Libecap. 1990. An Experimentric Study of the Cournot Theory of Firm Behavior. Working paper, University of Arizona. Bolton, Patrick, Joseph F. Brodley, and Michael H. Riordan. 2000. Predatory Pricing: Strategic Theory and Legal Policy. Georgetown Law Review 88:2239–330. Boone, Jan, Wieland Müller, and Sigrid Suetens. 2014. Naked Exclusion in the Lab: The Case of Sequential Contracting. Journal of Industrial Economics 62:137–166. Brandts, Jordi, Antonio Cabrales, and Gary Charness. 2007. Forward Induction and Entry Deterrence: An Experiment. Economic Theory 33:183–209. Brandts, Jordi, and Neus Figueras. 2003. An Exploration of Reputation Formation in Experimental Games. Journal of Economic Behaviour & Organization 50:89–115. Brennan, Timothy J. 2000. The Economics of Competition Policy:  Recent Developments and Cautionary Notes in Antitrust and Regulation. RFF Discussion Paper No. 00-07. Washington, DC. Caliskan, Anil, David Porter, Stephen Rassenti, Vernon L. Smith, and Bart J. Wilson. 2007. Exclusionary Bundling and the Effects of a Competitive Fringe. Journal of Institutional and Theoretical Economics 163:109–32. Camerer, Colin F., and Keith Weigelt. 1988. Experimental Tests of a Sequential Equilibrium Reputation Model. Econometrica 56:1–36. Capra, C. Monica, Jacob K. Goeree, Rosario Gomez, and Charles A. Holt. 2000. Predation, Asymmetric Information, and Strategic Behaviour in the Classroom:  An Experimental Approach to the Teaching of Industrial Organization. International Journal of Industrial Organization 18:205–25. Cason, Timothy N. 1995. Cheap Talk Price Signaling in Laboratory Markets. Information Economics and Policy 7:183–204. Cason, Timothy N., and Douglas D.  Davis. 1995. Price Communications in a Multimarket Context: An Experimental Investigation. Review of Industrial Organization 10:769–87. Chamberlin, Edward H. 1948. An Experimental Imperfect Market. Journal of Political Economy 56:95–108.

250   Wieland Müller and Hans-Theo Normann Clemens, Georg, and Holger Rau. 2012. Rebels without a Clue? Experimental Evidence on Endogenous Cartels and Fringe Firms. Working paper, University of Düsseldorf. Cooper, David J., Susan Garvin, and John H. Kagel. 1997a. Signalling and Adaptive Learning in an Entry Limit Pricing Game. Rand Journal of Economics 28:662–83. Cooper, David J., Susan Garvin, and John H. Kagel. 1997b. Adaptive Learning vs. Equilibrium Refinements in an Entry Limit Pricing Game. Economic Journal 107:553–75. Coursey, Don, and Vernon L. Smith. 1983. Price Controls in a Posted Offer Market. American Economic Review 73:218–21. Davis, Douglas D., and Charles A. Holt. 1993. Experimental Economics. Princeton, NJ: Princeton University Press. Davis, Douglas D., and Charles A. Holt. 1994. Market Power and Mergers in Laboratory Markets with Posted Prices. Rand Journal of Economics 25:467–87. Davis, Douglas D., and Charles A.  Holt. 1998. Conspiracies and Secret Price Discounts in Laboratory Markets. Economic Journal 108:736–56. Davis, Douglas D., and Bart J. Wilson. 2002. Experimental Methods and Antitrust Policy. In Charles A.  Holt and R. Mark Isaac, eds., Research in Experimental Economics, vol. 9, Experiments Investigating Market Power. Boston: JAI Press, 61–94. Dixit, Avinash. 1980. The Role of Investment in Entry-Deterrence. Economic Journal 90:95–106. Dolbear, F. T., L. B. Lave, G. Bowman, A. Lieberman, E. Prescott, F. Rueter, and R. Sherman. 1968. Collusion in Oligopoly:  An Experiment on the Effect of Numbers and Information. Quarterly Journal of Economics 82:240–59. Dufwenberg, Martin, Jacob K. Goeree, Uri Gneezy, and Rosemarie Nagel. 2007. Price Floors and Competition. Economic Theory 33:211–24. Engelmann, Dirk, and Wieland Müller. 2011. Collusion through Price Ceilings? In Search of a Focal-Point Effect. Journal of Economic Behavior and Organization 79:291–302. Engelmann, Dirk, and Hans-Theo Normann. 2009. Price Ceilings as Focal Points? An Experimental Test. In Jeroen Hinloopen and Hans-Theo Normann, eds., Experiments for Antitrust Policies. Cambridge: Cambridge University Press, 61–80. Fehr, Ernst, and Simon Gächter. 2000. Cooperation and Punishment in Public Goods Experiments. American Economic Review 90:980–94. Fonseca, Miguel A., and Hans-Theo Normann. 2008. Mergers, Asymmetries and Collusion: Experimental Evidence. Economic Journal 118:387–400. Fonseca, Miguel A., and Hans-Theo Normann. 2012. Explicit vs. Tacit Collusion: The Impact of Communication in Oligopoly Experiments. European Economic Review 56:1759–1772. Fouraker, Lawrence E., and Sidney Siegel. 1963. Bargaining Behavior. New York: McGraw–Hill. Friedman, Daniel, and Shyam Sunder. 1994. Experimental Methods: A Primer for Economists. Cambridge: Cambridge University Press. Friedman, James W. 1963. Individual Behavior in Oligopolistic Markets: An Experimental Study. Yale Economic Essays 3:359–417. Friedman, James W. 1967. An Experimental Study of Cooperative Duopoly. Econometrica 35:379–97. Goeree, Jacob K., and Rosario Gomez. 1998. Predatory Pricing in the Laboratory. Working paper, University of Virginia. Gomez, Rosario, Jacob K. Goeree, and Charles A. Holt. 2008. Predatory Pricing: Rare Like a Unicorn? In Charles R. Plott and Vernon L. Smith, eds., Handbook of Experimental Economics Results, vol. 1. Amsterdam: Elsevier, 178–84.

Experimental Economics in Antitrust   251

Greenlee, Patrick, David Reitman, and David S. Sibley. 2008. An Antitrust Analysis of Bundled Loyalty Discounts. International Journal of Industrial Organization 26:1132–52. Greenlee, Patrick, David Reitman, and David S. Sibley. 2011. Comment on Muris and Smith, “Antitrust and Bundled Discounts:  An Experimental Analysis.” Antitrust Law Journal 77:669–81. Grether, David M., and Charles R. Plott. 1984. The Effects of Market Practices in Oligopolistic Markets: An Experimental Examination of the Ethyl Case. Economic Inquiry 22: 479–507. Harrison, Glenn W. 1988. Predatory Pricing in a Multiple Market Experiment. A Note. Journal of Economic Behaviour and Organization 9:405–17. Harstad, Ron, Stephen Martin, and Hans-Theo Normann.1998. Experimental Tests of Consciously Parallel Behavior in Oligopoly. In Louis Phlips, ed., Applied Industrial Economics. Cambridge: Cambridge University Press, 123–51. Hinloopen, Jeroen, Wieland Müller, and Hans-Theo Normann. 2014. Output Commitment through Product Bundling:  Experimental Evidence. European Economic Review 65:164–180. Hinloopen, Jeroen, and Hans-Theo Normann. 2009. Experiments and Competition Policy. Cambridge: Cambridge University Press. Holt, Charles A. 1995. Industrial Organization: A Survey of Laboratory Research. In John Kagel and Alvin Roth, eds., Handbook of Experimental Economics. Princeton, NJ:  Princeton University Press, 349–443. Holt, Charles A., and Douglas D. Davis. 1990. The Effects of Non-binding Price Announcements in Posted-Offer Markets. Economics Letters 34: 307–10. Hong, James T., and Charles R. Plott. 1982. Rate Filing Policies for Inland Water Transportation: An Experimental Approach. Bell Journal of Economics 13:1–19. Huck, Steffen, Kai A. Konrad, Wieland Müller, and Hans-Theo Normann. 2007. The Merger Paradox and Why Aspiration Levels Let It Fail in the Lab. Economic Journal 117:1073–95. Huck, Steffen, Wieland Müller, and Hans-Theo Normann. 2001. Stackelberg Beats Cournot: On Collusion and Efficiency in Experimental Markets. Economic Journal 111:749–65. Huck, Steffen, Hans-Theo Normann, and Jörg Oechssler. 1999. Learning in Cournot Oligopoly: An Experiment. Economic Journal 109:C80–C95. Huck, Steffen, Hans-Theo Normann, and Jörg Oechssler. 2000. Does Information about Competitors’ Actions Increase or Decrease Competition in Experimental Oligopoly Markets? International Journal of Industrial Organization 18:39–57. Huck, Steffen, Hans-Theo Normann, and Jörg Oechssler. 2004. Two Are Few and Four Are Many:  On Number Effects in Cournot Oligopoly. Journal of Economic Behavior and Organization 53:435–46. Isaac, R. Mark, and Charles R. Plott. 1981. Price Controls and Behavior of Auction Markets: An Experimental Examination. American Economic Review 71:449–59. Isaac, R.  Mark, Valerie Ramey, and Arlington W.  Williams. 1984. The Effects of Market Organization on Conspiracies in Restraint of Trade. Journal of Economic Behavior and Organization 5:191–222. Isaac, R. Mark, and Stan Reynolds. 2002. Two or Four Firms: Does It Matter? In Charles A. Holt and R. Mark Isaac, eds., Research in Experimental Economics, vol. 9, Experiments Investigating Market Power. Boston: JAI Press, 95–119. Isaac, R. Mark, and Vernon L. Smith. 1985. In Search of Predatory Pricing. Journal of Political Economy 93:320–45.

252   Wieland Müller and Hans-Theo Normann Jung, Yun Joo, John H. Kagel, and Dan Levin. 1994. On the Existence of Predatory Pricing: An Experimental Study of Reputation and Entry Deterrence in the Chain-Store Game. Rand Journal of Economics 25:72–93. Kaplow, Louis. 2011. Direct versus Communications-Based Prohibitions on Price Fixing. Journal of Legal Analysis 3:449–538. Kosfeld, Michael, Akira Okada, and Arno Riedl. 2009. Institution Formation in Public Goods Games. American Economic Review 99:1335–55. Kreps, David M., and Robert Wilson. 1982. Reputation and Imperfect Information. Journal of Economic Theory 27:253–79. Landeo, Claudia M., and Kathryn E. Spier. 2009. Naked Exclusion: An Experimental Study of Contracts with Externalities. American Economic Review 99:1850–77. Maier-Rigaud, Frank P., and Martin Beckenkamp. 2007. Purchase Decisions with Non-linear Pricing Options under Risk: Experimental Evidence. Working Paper No. 2007/10, Max Planck Institute for Research on Collective Goods. Markert, Kurt E. 1974. The New German Antitrust Reform Law. Antitrust Bulletin 19:135–77. Martin, Stephen, Hans-Theo Normann, and Christopher M. Snyder. 2001. Vertical Foreclosure in Experimental Markets. Rand Journal of Economics 32:466–96. Mason, Charles F., and Cliff Nowell. 1998. An Experimental Analysis of Subgame Perfect Play: The Entry Deterrence Game. Journal of Economic Behaviour & Organization 37:443–62. Milgrom, Paul, and John Roberts. 1982. Limit Pricing and Entry under Incomplete Information: An Equilibrium Analysis. Econometrica 50:443–59. Motta, Massimo. 2004. Competition Policy:  Theory and Practice. Cambridge:  Cambridge University Press. Müller, Wieland, Yossi Spiegel, and Yaron Yehezkel. 2009. Oligopoly Limit-Pricing in the Lab. Games and Economic Behavior 66:373–93. Muris, Timothy J., and Vernon L. Smith. 2008. Antitrust and Bundled Discounts: An Experimental Analysis. Antitrust Law Journal 75:399–432. Nalebuff, Barry. 2004. Bundling as an Entry Barrier. Quarterly Journal of Economics 119:159–87. Neral, John, and Jack Ochs. 1992. The Sequential Equilibrium Theory of Reputation Building: A Further Test. Econometrica 60:1151–69. Normann, Hans-Theo. 2008. Experimental Economics in Antitrust. In Wayne Dale Collins, ed., Issues in Competition Law and Policy. Chicago: American Bar Association Book Series, 773–801. Normann, Hans-Theo. 2011. Do Vertical Mergers Enable to Raise Rivals’ Costs? Experimental Evidence. Journal of Industrial Economics 59:506–27. Normann, Hans-Theo, Bradley J.  Ruffle, and Christopher M.  Snyder. 2007. Do Buyer-Size Discounts Depend on the Curvature of the Surplus Function? Rand Journal of Economics 38:747–67. Ordover, Janusz A., Garth Saloner, and Steven C. Salop. 1990. Equilibrium Vertical Foreclosure. American Economic Review 80:127–42. Phlips, Louis. 1995. Competition Policy: A Game Theoretic Approach. Cambridge: Cambridge University Press. Plott, Charles R. 1982. Industrial Organization Theory and Experimental Economics. Journal of Economic Literature 20:1485–527. Plott, Charles R. 1987. Dimensions of Parallelism:  Some Policy Applications of Experimental Methods. In Alvin E. Roth, ed., Laboratory Experimentation in Economics: Six Points of View. Cambridge: Cambridge University Press, 193–219.

Experimental Economics in Antitrust   253

Plott, Charles R. 1989. An Updated Review of Industrial Organization Applications of Experimental Methods. In Richard Schmalensee and Robert Willig, eds., Handbook of Industrial Organization, vol. 2. New York: North-Holland, 1111–76. Rasmusen, Eric B., J. Mark Ramseyer, and John S. Wiley Jr. 1991. Naked Exclusion. American Economic Review 81:1137–45. Rey, Patrick, and Jean Tirole. 2007. A Primer on Foreclosure. In Mark Armstrong and Robert Porter, eds., Handbook of Industrial Organization, vol. 3.  New  York:  North-Holland, 2145–220. Roth, Alvin E. 1993. On the Early History of Experimental Economics. Journal of the History of Economic Thought 15:184–209. Roux, Catherine, and Christian Thöni. 2012. Collusive Relationships and Targeted Punishment: Number Effects in Cournot Oligopolies. Working paper. Scherer, Frederic M., and David R Ross. 1990. Industrial Market Structure and Economic Performance. 3rd ed. Boston: Houghton Mifflin. Segal, Ilya, and Michael D. Whinston. 2000. Naked Exclusion: Comment. American Economic Review 90:296–309. Selten, Reinhard. 1978. The Chain-Store Paradox. Theory and Decision 9:127–59. Smith, Angela M. 2011. An Experimental Study of Exclusive Contracts. International Journal of Industrial Organization 29:4–13. Smith, Vernon L. 1962. An Experimental Study of Competitive Market Behavior. Journal of Political Economy 70:111–37. Smith, Vernon L. 1964. The Effect of Market Organization on Competitive Equilibrium. Quarterly Journal of Economics 78:181–201. Smith, Vernon L., and Arlington W.  Williams. 1981. On Nonbinding Price Controls in a Competitive Market. American Economic Review 71:467–74. Tirole, Jean. 1989. The Theory of Industrial Organization. Cambridge, MA: MIT Press. Van Damme, Eric, Pierre Larouche, and Wieland Müller. 2009. Abuse of a Dominant Position:  Cases and Experiments. In Jeroen Hinloopen and Hans-Theo Normann, eds., Experiments and Competition Policy. Cambridge: Cambridge University Press, 107–59. Wellford, Charissa P. 2002. Antitrust: Results from the Laboratory. In Charles A. Holt and R. Mark Isaac, eds., Research in Experimental Economics, vol. 9, Experiments Investigating Market Power. Boston: JAI Press, 1–60. Whinston, Michael D. 2008. Lectures on Antitrust Economics. Cambridge: MIT Press. Wilson, Robert. 1992. Strategic Models of Entry Deterrence. In Robert Aumann and Sergiu Hart, eds., Handbook of Game Theory, vol. 1. Amsterdam: North-Holland, 305–29.

CHAPTER 11

OP T I M A L A N T I T RU ST R E M E D I E S A Synthesis WILLIAM H. PAGE

11.1. Introduction THE framers of the Sherman Act defined antitrust offenses in the language of the common law, using terms like “contract . . . in restraint of trade” (Section 1) and “monopolize” (Section 2) to signal the new statute’s continuity with ancient policies favoring competitive markets. In prescribing remedies, however, they made a clean break with the past. Before 1890, remedies for private monopolistic conduct were few and weak. Courts might decline to enforce agreements in restraint of trade, but they would not grant affirmative relief to the people and businesses the restraints injured (Lopatka 1989, 3–4) and some states for a time used quo warranto proceedings to revoke the charters of corporations that participated in trusts, although the firm was free to reincorporate in a state with more permissive corporate laws (Hovenkamp 1991, 64). These meager remedies reflected the conviction of nineteenth-century American courts that unenforceable private restraints could not limit competition for long. The Sherman Act departed from this laissez-faire orthodoxy by making violations of Sections 1 and 2 both crimes and civil wrongs. Public and private plaintiffs could now seek a variety of antitrust remedies by filing lawsuits in federal court.1 The Department of Justice could bring criminal actions seeking fines and imprisonment and civil actions “in equity to prevent or restrain” violations (§ 4); individuals and corporations could sue for treble damages (§ 7). In the twentieth century, Congress strengthened these 1 

Public litigation is itself a kind of remedy, even apart from any subsequent remedies (Kovacic 1999, 1288–92).

Optimal Antitrust Remedies   255

antitrust remedies and added new ones. In 1914, the Clayton Act authorized private parties to seek injunctions against threatened damages (§ 16). The same year, the Federal Trade Commission Act created the FTC, a new agency that could study markets and issue judicially enforceable cease and desist orders (§ 5). Under more recent legislation, the FTC may sue in court for injunctive relief and restitution, including disgorgement of illegal gains (15 U.S.C. § 53(b); United States v. Keyspan; Elhauge (2009)). The Hart-Scott-Rodino Antitrust Improvements Act of 1976 enabled more effective merger remedies by requiring firms to notify the federal antitrust agencies of merger plans and to delay consummation of the merger until the appropriate agency decides whether to seek an injunction against it in federal court (15 U.S.C. § 18a). By five amendments since 1955, Congress has increased exponentially the penalties for criminal violations of the Sherman Act (Reeves and Stucke 2011, 1568). Antitrust remedies, according to the Supreme Court, are supposed to “eliminate[] the effects of the illegal conduct” (National Society of Professional Engineers, at 698) and thereby “restore competition” (du Pont, at 326). As these goals suggest, American antitrust remedies rarely attempt to supervise or to direct market activities to achieve preferred outcomes (Page 1991, 37; Elzinga 2001, 649). In this respect, they show the persistence of a hybrid form of laissez-faire in American antitrust: the law uses fines and orders to deter illegal conduct or to remove its effects, not to achieve ideal competitive conditions. It presumes that, if illegal conduct ends, entry and competitive contracting will best advance the public interest. European antitrust law, by contrast, reflects less confidence in the self-correcting power of markets. For example, European law at least nominally prohibits excessive pricing by dominant firms. Its mandatory remedies similarly tend to be more regulatory than their American counterparts, protecting, for example, “the short-term needs of consumers” rather than “encouraging firms to strive for dominance [or] providing long-term incentives to invest in innovation” (Devlin and Jacobs 2010, 275). Economic analysis now guides both the definition of substantive antitrust offenses and the selection of remedies. Since the 1970s, courts and scholars have generally reached consensus that antitrust law should condemn only practices that reduce economic efficiency (Posner 2001, vii–ix). Although disagreements remain on the precise standard of economic welfare (Blair and Sokol 2012), the convergence on a relatively clear goal has reshaped antitrust analysis of a multitude of practices. At the same time, however, both courts and commentators have recognized that, even if private conduct reduces efficiency in a relevant sense, antitrust law should intervene to provide a remedy only if (and to the extent) doing so would improve efficiency. Antitrust courts, like physicians, should first do no harm (Lopatka and Page 1999, 25). Consequently, a substantial literature now examines the optimal use of antitrust remedies in specific contexts or for specific offenses. In this chapter, I draw on these studies to develop an integrated, albeit schematic, account of how courts and agencies should optimally deploy the various antitrust remedies. To set some limits on this daunting task, my focus throughout will be on the American system of remedies and the literature surrounding it, with only occasional

256   William H. Page comparisons to European and Asian approaches. I divide antitrust remedies into two categories: penalties aimed at deterrence and injunctions aimed at prospective remediation.2 The following two parts address these two categories in turn. Each part begins with a description of the economic framework for shaping the remedies under consideration and then applies that framework to examine the most important uses of the remedies for particular offenses.

11.2. Penalties Antitrust offenders are potentially subject to three kinds of penalties: private damages, public fines, and (in the case of individuals) imprisonment.3 The challenge for a practical system of antitrust enforcement is to achieve an approximately optimal level of deterrence by imposing an appropriate combination of these penalties on an appropriate selection of corporate and individual offenders. Most economic analysis of antitrust penalties has focused on the role of monetary penalties (including damages), which are available against offenders for the full range of antitrust violations that inflict social harm. A smaller body of work has begun to consider the role of imprisonment of individuals for criminal antitrust violations, mainly cartels, when corporate employers also pay fines and damages.

11.2.1.  Monetary Penalties Gary Becker (1968) and George Stigler (1970) laid the groundwork for modern economic analysis of monetary penalties in classic studies published over 40  years ago. William Landes built on that foundation to formulate a widely accepted measure of the optimal fine for antitrust violations: “the fine should equal the net harm to persons other than the offender . . . divided by the probability of apprehension and conviction” (Landes 1983, 656–57). The “net harm to persons other than the offender” is the sum of the external harms that the offense inflicts:  the overcharge and the deadweight welfare loss, including increased costs. The overcharge is the wealth transfer from purchasers to the offender. The deadweight loss has two components—the lost consumer surplus value from the restriction of output and any increased production costs the offense imposes on firms other than the offender. The logic of the Landes measure is straightforward. The law condemns certain practices because they cause a deadweight loss and thus reduce social wealth. Firms engage in condemned practices not to cause this deadweight loss, but to profit by an overcharge 2  Unlike Elzinga and Breit (1976), I do not classify injunctions as penalties. Even though they impose costs on offenders, their function is prospective remediation rather than deterrence. 3  For discussion of the practicalities of this complex enforcement system, see O’Connor et al. (2012).

Optimal Antitrust Remedies   257

and perhaps cost savings; the deadweight loss is a byproduct. An optimal penalty should remove the incentive to engage in practices that impose a net social cost, that is, a deadweight loss greater than any cost savings. The net-harm-to-others standard accomplishes this goal by prescribing a fine equal to the illicit profit plus the deadweight loss. If detection is certain, the firm will abjure the practice when the sum of the illicit profit and the deadweight loss (the net harm to others) is greater than the sum of the illicit profit and any cost savings that the practice enables (the offender’s total profit from the practice). A penalty calculated in this way will not deter offenses that nominally violate the law, but generate more value and costs savings than any costs they impose on others. Offenses with these effects are efficient in the sense that they create more social wealth than they destroy. If detection is not certain, then the net harm to others should be multiplied by the inverse of the probability of detection, so that the penalty the offender rationally expects to pay is equal to the net harm the practice imposes. Thus, if the chance of detection is one-third, the fine should be equal to the net harm times three. The Sherman Act, as it happens, does provide for trebling of actual damages, a provision that might be justified as a discounting factor. But if it serves such a purpose, it does so only crudely, because it does not adjust the multiple according to the relative concealability of the offense, which largely determines the probability of detection. Consequently, some commentators have suggested that trebling may be excessive for overt offenses like exclusionary practices, but inadequate for covert offenses like price fixing (Easterbrook 1985, 454–61; Lande 2003, 300; Connor 2012). To approximate the optimal penalty, the system of enforcement must take account of all of the penalties offenders pay, especially public fines, private damages (including settlements), and imprisonment of individual offenders (Rosenberg and Sullivan 2006; Connor and Lande 2012; First 2009). A rational firm will “aggregate the expected punishments and discount (multiply) them by the probability of their imposition” (Posner 2001, 47). Whether the aggregate penalties are sufficient for optimal deterrence under present law is unclear, but Connor and Lande (2012, 430) estimate that “the combined level of U.S. cartel sanctions has been only 9% to 21% as large as it should be to protect potential victims of cartelization optimally.” The remainder of this section will consider how these penalties should interact in a system that seeks to impose approximately optimal fines.

11.2.1.1.  Public Fines Public fines play a necessary but limited role in the American system of deterrence. Unlike European competition law, which has authorized fines in the billions for civil offenses like abuse of dominance, American federal antitrust law authorizes public fines only for criminal offenses, like cartels. Private treble damages, which we discuss in the next section, are the only significant federal monetary penalties available for civil violations.4 Even for criminal offenses, public fines have until recently been a 4  First (2009) argues that public fines should be available for civil monopolization offenses in addition to private damage actions. Elzinga and Breit (1976) argue, by contrast, that public fines should be the exclusive antitrust penalty, because antitrust enforcement is a pure public good, while the prospect of

258   William H. Page trivial part of the system of antitrust remedies, because of low statutory limits. The original version of the Sherman Act set a cap of $5,000 for each offense for corporate defendants. Congress raised that amount to $50,000 in 1955, but that was still far too low for meaningful deterrence. In the period between 1960 and 1969, for example, the total corporate fines came to just over $13 million (Posner 1970). Clabault and Block (1981, vol. 2, 732–33) estimate that the fines imposed were less than 4 percent of the volume of commerce affected by the offenses. It is no longer as clear, however, that the statutory caps for public fines are too low. Congress has increased the maximum fine repeatedly, most recently to $100,000,000 in 2004. Under an alternative provision applicable to all federal crimes, a fine may reach “the greater of twice the gross gain or twice the gross loss” from the offense (18 U.S.C. § 3571(d)), even if this amount exceeds the Sherman Act’s cap. United States Sentencing Commission Guidelines Manual (2013, 2R1.1(d)) provide that a court should calculate a fine, within the applicable statutory maximum, equal to 20 percent of the volume of affected commerce,5 a figure that can increase or decrease based upon the offenders’ culpability score (Kneuper and Langenfeld 2011; Mutchnik et al. 2008). Under this regime, some defendants have paid fines far in excess of $100 million. During the period between 2001 and 2010, the total fines came to just over $4 billion (US Department of Justice, Antitrust Division Workload Statistics FY 2000–2010, 13), an amount 50 times greater, in real terms, than fines paid during the 1960s.6 In January 2012, a single Japanese auto parts manufacturer agreed to pay a fine of $470 million (Pickler 2012). Despite recent increases, there is no way to know if current fines represent an appropriate contribution to an optimal fine. Certainly, the formula used to calculate fines is only weakly related to the optimal penalty. The volume-of-commerce measure bears only a tenuous relationship to the theoretically optimal penalty, which, as we have just seen, is determined primarily by the actual overcharge—a figure that varies widely across cartels (Kaplow 2011, 424–25). The drafters of the guidelines apparently estimated that the usual overcharge is 10 percent of sales, and that other harms attributable to the output restriction add to that amount (Connor and Lande 2005, 523). Kneuper and Langenfeld (2011) suggest that the size of recent public fines justifies using the more precise measures of the overcharge common in private damage litigation as a basis for calculating fines rather than the crude estimate implicit in the Sentencing Guidelines.7 private damages creates perverse incentives for the plaintiffs, especially competitor plaintiffs. Antitrust doctrine, as we will see, places various constraints on private damage actions to harness them in the service of public goals. If Congress were nevertheless to abolish the private damage remedy, it would certainly also have to authorize public fines available for civil offenses in order to achieve the goal of optimal deterrence. 5 

EU guidelines provide for fines up to 30 percent of the volume of sales (Kaplow 2011, 424 n. 193). The $13 million paid during the 1960s would be around $80 million in today’s dollars. The $4 billion paid during the 2000s is 50 times that. 7  The court in United States v. Andreas made just such a calculation. Connor and Lande (2005, 551 n. 182). 6 

Optimal Antitrust Remedies   259

11.2.1.2.  Private Damages Under Section 4 of the Clayton Act, persons injured by antitrust violations can sue for treble damages. Despite the increase in public fines, treble damages represent the largest component of monetary penalties paid by criminal antitrust offenders. They are, of course, the only monetary penalties paid by civil antitrust offenders.

11.2.1.2.1.  Deterrence or Compensation? In the last section, the discussion focused on the role of fines in achieving optimal deterrence, because fines serve only to deter. The Clayton Act (Section 4) also authorizes “any person injured in his business or property” to sue for “threefold the damages by him sustained.” This remedy, according to the Supreme Court, serves the dual goals of deterrence and compensation (California v. ARC America Corp. at 102; Illinois Brick Co. v. Illinois at 148; American Society of Mechanical Engineers, Inc. v. Hydrolevel Corp. at 575–76). The compensatory aspect of the remedy—basing the measure of damages on private harm—provides the mechanism for imposing a deterrent penalty (Posner 2001, 266; Standen 1995). Damages and fines are complementary mechanisms for imposing the optimal deterrent penalty. As we saw in the introductory section, the optimal penalty for antitrust violations is measured by the net harm to persons other than the offender. In calculating fines, courts should appropriately take account of some estimate of the harm that the offense inflicts on other economic actors. Similarly, the net harm standard corresponds directly to the very harms that form the basis for the calculation of damages. Many features of the system of antitrust penalties demonstrate the primacy of this deterrent function over compensation. The damage remedy undoubtedly compensates some injured firms and individuals as a way of inducing them to sue. But it also allows some plaintiffs to recover awards far in excess of their personal net harm, and denies any recovery to others who have suffered significant harm, when those results are necessary for effective deterrence. Where a remedy poses a choice between better deterrence and more complete compensation, the federal courts have opted for deterrence. There are many reasons for this preference, but two are decisive. First, if deterrence is effective, there is no need for compensation, which is costly to administer (Landes 1983, 673). Second, of those hurt by antitrust violations—and that would include the population of most of the developed world—few can practically receive compensation through a judicial remedy. The law of antitrust standing explicitly recognizes this limitation by denying those with remote or speculative harms the right to sue. The clearest example of antitrust law’s preference for deterrence is the Illinois Brick rule, which denies indirect purchasers the right to sue because direct purchasers are better situated to impose the deterrent penalty. Consumers, for example, who might have borne most of an overcharge that retailers passed on to them, have no right to sue under the Sherman Act. The law recognizes that, given the limitations of courts, consumers as a class would have little chance of receiving a meaningful recovery. Illinois Brick may actually provide more effective compensation, because it gives direct purchasers the

260   William H. Page incentive to reduce their prices to indirect purchasers in anticipation of the recovery of the full overcharge for each unit they purchase for the offenders (Landes and Posner 1979, 605). The experience under state indirect purchaser statutes has shown the wisdom of the Supreme Court’s assessment (Page 2012). Courts often do not certify indirect purchaser classes because of the difficulty of proving harm to class members on a common basis. This result was first evident when the litigation occurred in state courts, and has continued under the Class Action Fairness Act, which relaxes federal jurisdictional standards to allow state indirect purchaser class actions to be filed in or removed to federal courts, where they can be consolidated in a single district for pretrial procedures and settlement negotiations. When courts certify a class, the parties almost always settle. Class members often do not receive a significant share of the settlement fund, because it is impractical to distribute it to them. Thus, recognizing indirect purchaser standing does not systematically accomplish the goal of compensation.

11.2.1.2.2.  Optimal Deterrence and the Measure of Damages Scholarship on the damage remedy has considered how to shape the criteria of standing and proof of damages to impose the appropriate deterrent penalty. A private damage remedy, applied literally, would authorize recovery for many injuries that bear no relationship to the optimal penalty. The Supreme Court, however, has used the doctrines of antitrust injury and standing, including the rule of Illinois Brick, to shape the damage remedy in ways that approximate the optimal penalty (Page 1985, 1450). The antitrust injury doctrine seeks to ensure that, in order to recover damages, a private plaintiff must have suffered a harm that was caused by the anticompetitive aspect of the violation (Brunswick at 489; Page 1980; Davis 2003). Private harm is antitrust injury if it corresponds to an output restriction that the offense causes. This standard ties the right to sue and the measure of damages directly to the optimal penalty. The doctrine of antitrust standing supplements the antitrust injury doctrine by limiting recovery to the most efficient category of plaintiffs to impose the appropriate deterrent penalty. The most straightforward application of these doctrines occurs in cases alleging cartels. The overcharge from the cartel is the largest component of the optimal penalty; it is also the measure of damages to purchasers from the cartel. The overcharge to purchasers is thus the clearest example of antitrust injury (Page 1985, 1465). Equally important, it is the component of the optimal penalty that plaintiffs’ experts can most practically prove, using available econometric methods. There is now an extensive economic and legal literature on the appropriate statistical methods for establishing the overcharge by comparing actual prices with an estimate of prices that would have prevailed in the market but for the illegal conduct (Davis and Garces 2010, 352–68; Baker and Rubinfeld 1999; Friederiszick and Röller 2010). As we saw in the last section, the Illinois Brick doctrine confines the right to recover under federal law to direct purchasers, regardless of whether they passed on any portion of the overcharge. This step is necessary in order to ensure that the class of plaintiffs best situated to sue for the overcharge has the incentive to bring suit (Page 1985, 1488).

Optimal Antitrust Remedies   261

Because direct purchasers have the right to sue for the full overcharge, creating rights of action for shares of the overcharge that the direct purchasers may have passed on would only make the process of imposing the optimal deterrent penalty more costly. An output restriction may impose other harms, beyond the overcharge to direct purchasers from the cartel. First, the output restriction itself harms marginal consumers who choose not to purchase at the monopoly price and instead substitute other products, which are presumably priced at marginal cost. These consumers most directly bear the deadweight loss from the reduction in consumers’ surplus. Second, if some firms in the market do not participate in the cartel, but sell at the cartel’s price, their customers also suffer an overcharge, even though the sellers are not themselves violating the law. In theory, the cartel should be liable for both of these overcharges: they are antitrust injury, because they are proportional to the output restriction and thus directly reflect the anticompetitive effect of the cartel. Nevertheless, because these harms are difficult to prove, courts have differed on whether they are compensable in principle; few courts, if any, have actually awarded damages on this theory (Lande 2010, 488–90; Hjelmfelt and Strother 1991). The impracticality of recovering these kinds of harms is an important consideration in calculating whether the combined penalties for cartel practices are optimal. The optimal penalty, of course, includes a discounting factor to account for the probability of detection. Thus the net harm to others must be multiplied by the inverse of the probability of detection. Statutory trebling may be far too low to account optimally for the low probability of detection of cartels, the most concealable antitrust offense. American antitrust partly addresses this concern by providing criminal penalties for cartels—public fines and imprisonment of individual offenders. Because these extraordinarily severe penalties are only available for criminal offenses, they provide, in practice, a larger multiplier of penalties for the most concealable offenses. Antitrust injury also tailors damages to the optimal penalty in cases involving exclusionary practices. If a practice “excludes” rivals by increasing output, the rival’s lost profits are, of course, not antitrust injury (Brunswick). In these cases, of which there are many, the antitrust injury requirement limits the perverse incentives of competitor plaintiffs to use antitrust litigation to reduce competition (Snyder and Kauper 1991). But a dominant firm can restrict output and increase its profit by excluding or weakening rivals, for example by using exclusive contracts to deny them access to needed supplies or outlets. If the exclusion restricts total output and harms consumers, the consumers suffer antitrust injury. In those circumstances, the lost profits of the excluded rival are linked to the output restriction and thus represent antitrust injury as well (Page 1990).

11.2.1.2.3.  Settlements and Deterrence In antitrust litigation, as in litigation generally, most cases settle, some for significant amounts (Connor and Lande 2012, 448).8 When that happens, no court actually makes

8 

Connor (2012) estimates that settlements in cartel cases average 30 percent of damages.

262   William H. Page a determination on the merits or calculates an appropriate penalty. In these cases, the agreed amount is necessarily reduced to account for the costs of suit and for the possibility that the plaintiff might fail to prove the offense, the harm, or both. Nevertheless, the optimal penalty framework serves an important function in shaping the settlements of antitrust claims. The parties bargain “in the shadow of the law” (Mnookin and Kornhauser 1979), so the settlement amount necessarily reflects the parties’ estimates of the likelihood of winning, given the evidence and the applicable legal standards of liability and damages, including the doctrines of antitrust injury and standing. The possibility of the plaintiff losing is one of the uncertainties that determine the probability of detection, an element of the optimal penalty. Consequently, settlement may be optimal, if the parties make an accurate estimate of the plaintiff ’s probability of success, even though the settlement amount is less than the damages a court would have imposed had the plaintiff won after accepting the risk of going to trial.9 Courts recognize the appropriateness of such a discount in evaluating the “fairness” of class action settlements to members of the plaintiff class, to ensure that class counsel has adequately represented their interest in negotiating the settlement.

11.2.1.2.4.  Agency Costs and Deterrence Recently, Daniel Crane has argued that the entire concept of relying on private antitrust damages to deter antitrust violations is flawed because the prospect of damages will not deter the managers within firms who are most likely to engage in practices that violate the antitrust laws (Crane 2011, 175–82). A manager cares mainly about his unit’s present prosperity, because that determines his or her success within the firm. The prospect of an antitrust damage award against the company several years in the future, discounted to present value and further discounted for the likelihood of escaping detection, must seem insignificant to a manager who is only with the firm for a few years. Thus, “it is difficult to see how the threat of a future damages judgment disciplines managerial decision making” (Crane 2011, 178). The threat of a big settlement is also insufficient, because defendants usually only pay settlements on the eve of trial (Crane 2011, 180). Crane’s argument does not reject the assumption that an actor will be deterred by the prospect of an appropriately calculated penalty—that, after all, is a central premise of law and economics. Instead, he relies on the idea of agency costs. If a single decision-maker controlled the firm’s actions, damages would deter. But the incentives of managers diverge from those of the firm and it is costly to realign them. Thus, Crane argues, the firm will be unable to monitor the manager or establish appropriate incentives. This argument conflicts with Richard Posner’s view that a “corporation has effective methods of preventing its employees from committing acts that impose huge liabilities on 9  Of course, a defendant may even pay a positive amount to settle a claim that has no merit, simply to avoid the greater costs of suit (Rosenberg and Shavell (2006). Friedman and Wickelgren (2010) review the literature and show that allowing settlement may sometimes reduce welfare by reducing the incentive of a plaintiff to invest in proving its case at trial, which in turn reduces the likelihood of an accurate result and the expected deterrent penalty for guilty offenders.

Optimal Antitrust Remedies   263

it” (Posner 2001, 271). Posner reasons that the firm will recognize the agency costs that Crane identifies and design incentives to bring the goals of its employees in line with the goal of maximizing corporate profit. For most practices, Posner is surely correct. Major firms will also have antitrust compliance programs that identify the sorts of practices to avoid (Lipsky 2009). Once managers know what those practices are, the firm can punish them for engaging in the practices (and stop the practices) long before the firm has to pay any settlement or damage award. Equally important, most antitrust violations, especially most exclusionary practices, require the firm to enter into contracts with suppliers, distributors, or other actors. In a firm large enough to have market power, these sorts of contracts will be subject to review by antitrust counsel, who have strong incentives to identify illegal practices and bring them to the attention of top executives. For example, in Dentsply, it was unclear which manager originally thought of Dentsply’s published policy, later found unlawful, of requiring its dealers to carry only Dentsply products, but there was no contention that managers implemented the policy without the endorsement of the top management. Similarly, in Microsoft, the defendant never contended that its exclusive contracts and design choices were anything other than company policy. Crane is correct, however, that agency costs many prevent the firm from controlling its managers in some instances. The violations most likely to fall in this category are criminal price-fixing conspiracies (Lande and Davis 2008). Some of the firms that engage in price fixing are small and may not have antitrust compliance programs. Even if they do, the programs may be less effective in controlling price fixing than exclusionary practices, because price fixing is necessarily clandestine.10 A manager may get the idea that he or she can surreptitiously fix prices and enhance the profitability of his or her division (and compensation) without drawing the notice of supervisors and antitrust counsel (Kaplow 2011, 427; Bekenstein and Gavel 1985, 674). Some firms many not penalize the offender significantly (Connor and Lande 2012, 442). In these circumstances, there is an argument that individual liability of the manager is necessary for deterrence. Fines and damages might be ineffective, because managers would likely lack the resources to pay for the full social costs of their actions. This shortcoming in the system is the primary justification for the reliance on imprisonment in the implementation of a system of antitrust deterrence.

11.2.2. Imprisonment Imprisonment has been part of the system of antitrust penalties since 1890, but provided little deterrence for decades, because it was rarely used—only 24 individuals were sentenced to jail in the first half-century of the Sherman Act, and only 18 between 1966 and 1974 (Breit and Elzinga 1976, 31–32). According to the US Department of Justice, Antitrust 10  Legal advice can help firms that participate in trade association meetings to avoid illegal communications among rivals (Peterson 2011).

264   William H. Page Division Workload Statistics FY 2000–2010, 14, however, 24 antitrust defendants were sentenced to jail in 2001 alone, and more than that number have been sentenced to jail each year since then, with an average sentence of almost 20 months (US Department of Justice, Antitrust Division Criminal Enforcement Fine and Jail Charts FY 2000–2010). These figures are significant. Thus, at least in the present century, the measure of the optimal penalty must take account of both private damages and public fines on corporations and fines and imprisonment imposed on individuals. Recent studies have begun to consider whether the present system of remedies imposes monetary penalties and imprisonment that, in the aggregate, approximate the optimal deterrent penalty. One recent study, for example, assigns a dollar value to each year in prison served by an individual defendant, triples that amount to reflect the added deterrent value of individual penalties, and adds this figure (and similarly adjusted individual fines) to corporate fines to estimate the total deterrent penalty imposed by government enforcement (Lande and Davis 2011, 336). But it is not clear how corporate and individual penalties should interact in a system of optimal deterrence. Individual managers commit antitrust violations that can trigger both corporate and individual sanctions. The balance between individual and corporate sanctions in achieving optimal deterrence depends on a multitude of factors, particularly agency costs, which vary with the offense and the circumstances (Polinsky and Shavell 1984). As the last section suggests, corporations themselves have the incentive and the means to control most of the conduct of employees that can trigger enormous liability (Beckenstein and Gabel 1986). For antitrust offenses that involve public actions— exclusionary contracts, refusals to deal, predatory pricing, tying arrangements, and so forth—there is little reason to doubt that the actions of employees are fully attributable to the corporation; the powers that be in the corporation should be fully aware of their managers’ highly visible activities. In these circumstances, there seems to be no good reason to impose separate sanctions on individuals:  the corporation is aware of the conduct and the expected sanctions and can make any calculations necessary to decide whether to proceed with a policy. Firms that engage in this sort of conduct are likely to be large and capable of paying the optimal deterrent fine, including private damages. Fines and imprisonment of employees make more sense for clandestine offenses, especially price fixing. Even here, Richard Posner has argued that imprisonment should be used only as a “last resort” in antitrust cases because courts cannot easily measure the monetary value of time served in calculating the appropriate sentence, and, unlike fines, imprisonment imposes a deadweight welfare loss by sacrificing the productive value of the manager’s services (Posner 2001, 270). Moreover, he argues, even in this era imprisonment is infrequently imposed and thus may have little effect. Even the enormous costs of a prison term may be inadequate to deter if the manager thinks the chances of actually being discovered, charged, and convicted are sufficiently remote. Louis Kaplow (2011, 426–27) echoes these concerns, but identifies instances in which imprisonment may be necessary. First, the optimal fine, including an appropriate multiple, may be beyond the capability of the offenders to pay, particularly if they consumed their gains from the offense before sentencing. Second, it may be necessary to impose a prison sentence in order to account for agency problems within the firm. The top management of the firm may know the possible sanctions for participating in

Optimal Antitrust Remedies   265

a cartel and still choose to do so, but it may find it hard to persuade their managers to participate at the risk of spending time in jail. It may, for example, not pay managers enough ex ante to take account of the risk, and state laws prohibit reimbursement of employees ex post for conscious criminal activity. Alternatively, the firm may want to avoid liability, but find it hard to impose sufficient controls on its employees to prevent them from pursuing profit for the company (and compensation for themselves) through secret participation in cartels (Werden and Simon 1987, 930–31). Ex post, the firm may not be able to impose a significant sanction beyond simple termination. But, Kaplow notes, the government can do so (Polinsky and Shavell 1993). Connor and Lande (2012, 447) conclude that, while both individual and corporate penalties are necessary, the individual penalties, including imprisonment, provide the most effective deterrence. Other considerations weigh in favor of individual liability. First, in instances in which the corporation is held liable for actions of a rogue manager, fines might bring about the demise of the firm with widespread harm to innocent employees and shareholders; imposing individual criminal liability “ensures that each actor bears the consequences of her misdeeds” (Hamdani and Klement (2008) at 304). Second, individual liability many be necessary to give prosecutors sufficient leverage to obtain evidence necessary to expose the cartel. Participants in cartels are well aware of the adage that if nobody talks, everybody walks. An offer of immunity from prosecution can be essential to creating the dynamic in which somebody will talk. For this system to work, of course, the employer must be precluded from compensating the employee for any sanction imposed (Mullin and Snyder 2010, 37–39).

11.3. Injunctions Injunctions direct market actors to do something or to stop doing something. They have the same goal as penalties, but a different mechanism: they are designed to enhance social welfare, but operate prospectively rather than by deterrence. Consequently, they are only appropriate for certain offenses. As Louis Kaplow has observed, if violent crimes were subject only to injunctive relief after the first offense, the streets would not be safe (Kaplow 2011, 429). Injunctions are traditionally available only where there is no adequate legal remedy, such as damages. This limitation is fully justified in antitrust cases. Penalties are the preferable remedy in the abstract, because, if deterrence is effective, the legal system need not devote further resources to correcting and monitoring competitive behavior. For many offenses, such as price fixing, a penalty is both necessary to deter the offense and often sufficient to prevent its recurrence; adding an injunction that merely repeats the law’s clear prohibition accomplishes nothing. Only in special cartel cases is it appropriate to impose additional conduct remedies, such as monitoring or firewall provisions. There are some circumstances, however, in which imposition of a deterrent penalty is insufficient to end the defendant’s illegal conduct and restore competitive conditions. In those cases, a court may issue a prohibitory injunction against

266   William H. Page specific conduct; a mandatory injunction regulating conduct; or a structural injunction requiring a firm to divest (or not acquire) assets. Which, if any, of these orders is appropriate depends on the nature the defendant’s conduct—how it reduced or threatens to reduce efficiency—and the court’s remedial capabilities. To the extent the court can practically do so, it should design a remedy that mirrors the illegal conduct and is inexpensive to monitor. A prohibitory injunction is appropriate when the defendant has engaged in active illegal conduct—for example, imposing an exclusive provision in a supply contract that prevents a supplier or a customer from dealing with the defendant’s rivals. Less often, a mandatory injunction may be necessary if the defendant has illegally refused to deal. If the defendant has illegally acquired durable monopoly power or market power, the court may issue a structural injunction, requiring the defendant to divest (or not to acquire) certain assets (Elzinga, 1977; Shelanski and Sidak 2001, 52). Injunctions are costly to implement, both in the direct costs of administration and the indirect costs of deterring efficient conduct. As the court of appeals in Microsoft recognized, more costly remedies must be justified by better evidence that they will improve matters. Although the court affirmed many of the district court’s liability rulings without much evidence that Microsoft’s conduct actually reduced competition, it adopted a different standard for relief. It insisted that there should be solid evidence “of a significant causal connection between the conduct and creation or maintenance of the market power” to justify costly structural remedies (Microsoft 2001, 106). Even if the remedy enhances static efficiency (taking account of both allocative and productive efficiency), the court must be careful to avoid reducing dynamic efficiency or incurring undue enforcement costs (Shelanski and Sidak 2001). A court should issue an injunction only if it will stop continuing harmful conduct without also impeding productive conduct and if the benefits of the order exceed the direct costs of monitoring. Special constraints apply to private cases seeking injunctive relief. Private plaintiffs seeking injunctive relief must have suffered (or face the prospect of) antitrust injury (Cargill). This condition is necessary, because the right to invoke the law to impose an injunction against a defendant’s conduct is worth the net value of the conduct to the defendant (Page and Lopatka 1996). Thus, for a plaintiff to exercise such a right, it must show that its harm is related to an output restriction that the defendant’s conduct causes. Much recent work on framing of optimal injunctions has occurred in connection with the Microsoft case. The sections that follow will use examples from that case to illustrate both optimal and suboptimal remedial choices.

11.3.1.  Structural Injunctions Some commentators have argued that structural remedies are preferable to conduct remedies, because they directly restore competitive market conditions and do not require continuing judicial supervision. But actual structural remedies, such as the breakup of the Bell system, have entailed enormous costs of their own (Shelanski and

Optimal Antitrust Remedies   267

Sidak 2001, 54–55; Kovacic 1989). Thus, structural relief should be limited to instances where it is necessary to remove anticompetitive effects of the particular offense and can be accomplished without indirect costs that exceed the competitive benefits of the remedy. Antitrust remedies, as we have seen, do not seek to establish ideal competitive conditions, but to restore market conditions that would have existed but for the unlawful conduct. Structural relief is better suited to this goal when the offense itself involves the structure of the firm. The D.C. Circuit in Microsoft, for example, recognized that structural remedies, particularly divestitures of assets, are better suited to correcting illegal acquisitions than illegal exclusionary or collusive conduct by a firm that achieved monopoly power lawfully.11 It required a clear showing of a causal connection between exclusionary conduct and monopoly power in order to justify a structural remedy (Microsoft 2001, 106–7). Under this standard, it reversed the district court’s order to divide Microsoft into separate operating systems and applications companies and later rejected a state’s proposal to require Microsoft to create and make available an open-source version of its Internet Explorer browser (Microsoft 2004, 1230–31). Faced with this analysis of appropriate antitrust remedies, the Department of Justice dropped its request for dissolution of the firm. Dissolving a firm with lawfully acquired monopoly power is problematic for several reasons. First, a court cannot as easily divide a unitary company as it can undo an acquisition, particularly one that is recent or one in which the acquired assets have been preserved as a subsidiary or division. Divestiture “is simpler to effectuate where the firm to be broken up is itself the product of mergers: the mergers suggest the lines along which the firm can be broken up with minimal disruption” (Posner 2001, 105). Second, a divestiture remedy necessarily bears a closer causal connection to an illegal acquisition than to exclusionary or collusive conduct. Divestiture reverses the increase in monopoly power attributable to an acquisition that it undoes; it does not necessarily reverse an increase in monopoly power attributable to exclusionary conduct. Third, dissolution of a unitary company is likely to reduce productive efficiencies, because a firm that grows by internal expansion likely structures itself in a way that minimizes costs (Elzinga and Breit 1976, 1005–6). An order ending exclusionary conduct and the market’s own remedial forces are ordinarily better suited to eroding market share achieved or maintained by exclusionary practices. Structural relief is preferable in horizontal merger cases.12 The rationale for this preference mirrors the rationale for prohibiting anticompetitive mergers in the first place. The law recognizes that certain practices are noncompetitive, but does not prohibit 11  As the court in Microsoft noted, a “corporation that has expanded by acquiring its competitors often has preexisting internal lines of division along which it may more easily be split than a corporation that has expanded from natural growth. Although time and corporate modifications and developments may eventually fade those lines, at least the identifiable entities preexisted to create a template for such division as the court might later decree” (Microsoft (2001), at 106). 12  In recent years, however, enforcers and courts have also considered various conduct remedies in order to avoid anticompetitive effects while preserving the benefits of integration (Kwoka and Moss 2011). I return to this development in the next section.

268   William H. Page them directly because of courts’ institutional limitations. For example, American antitrust law does not condemn unilateral monopoly pricing or interdependent oligopoly pricing. Nevertheless, the law prohibits mergers that make these forms of noncompetitive pricing more likely. Because the law’s approach is structural—prohibiting conduct that creates market structures that are likely to be less competitive—the remedy is usually structural. The Supreme Court has observed that “divestiture is particularly appropriate where asset or stock acquisitions violate the antitrust laws” because it “is a start toward restoring the pre-acquisition situation” (Ford Motor Co. (1972), at 573). The US Department of Justice’s Policy Guide to Merger Remedies (2011, at 5) similarly states that “the Division will pursue a divestiture remedy in the vast majority of cases involving horizontal mergers,” because “[d]‌ivestiture of overlapping assets, usually an existing business entity, can effectively preserve competition that the merger otherwise would eliminate.” Undoing the structural effects of anticompetitive mergers enhances social wealth by avoiding noncompetitive pricing and output restrictions that would otherwise be lawful and by inhibiting unlawful collusion (Wils 2006, 186). An order blocking a merger before it occurs is also a structural remedy, because it preserves a more competitive market structure than the one that would have otherwise prevailed. It substitutes, in other words, a preconsummation prohibition for a postconsummation divestiture. Usually, this substitution is efficient. The Hart-Scott-Rodino mechanism, requiring premerger notification of the enforcement agencies and imposing a waiting period before consummating the merger, reduces the costs of administering structural relief by allowing courts to block a merger before it occurs rather than disentangle the merging firms after the fact. It also avoids the punitive effect of dissolution on shareholders who purchased their interests in the defendant after it acquired monopoly power, and therefore paid the capitalized value of the expected monopoly profits as part of the price of the stock (Elzinga and Breit 1976, 48–49). Studies of early monopolization cases have raised questions about economic benefits of large-scale divestitures, even when mergers created monopoly power (Hale 1940; Adams 1951). Crandall (2001, 136), for example, concludes that there is no evidence that the dissolution of the Standard Oil trust “had a significant effect on output or prices in the U.S. oil industry.” Richard Posner (2001, 108–9) reaches a similar conclusion about American Tobacco, Corn Products Refining, and International Harvester. In these cases, however, the government challenged the mergers years after they occurred, when the firms had rationalized their production and the industry had significantly changed (Elzinga and Breit 1976, 52). Delayed divestiture is likely to be more costly. The district court in American Can, for example, expressed reluctance to “to destroy so finely adjusted an industrial machine as the record shows the defendant to be” (at 903). Dissolutions might have been less effective in these circumstances, because a court would be hard pressed to recreate anything approximating the competitive conditions of an earlier period. On the other hand, economists who have studied the record in the United States Steel case have concluded that the 1901 merger that created the steel trust (which remained a holding company with clearly defined operating subsidiaries) caused lasting harm to American consumers and should have been dissolved even 20 years later

Optimal Antitrust Remedies   269

(Parsons and Ray 1975).13 Modern merger standards and Hart-Scott-Rodino procedures would certainly and properly have blocked at the outset all of the mergers that created the trusts. All of the cases mentioned in the previous paragraph involved horizontal integration and proposed or actual horizontal dissolution. More problematic are structural remedies that dissolve firms vertically. As Herbert Hovenkamp (2010) has recently observed, economists in all eras have viewed vertical integration—the performance of some economic function within the firm instead of by contracting in the open market—as ubiquitous and generally benign. The most famous and economically significant example of vertical divestiture occurred in the wake of the Paramount decision in 1948. The government had contended that Hollywood studios’ ownership of film distributors and exhibitors reinforced a complex web of jointly adopted anticompetitive practices including block booking and clearances. The district court had found most of the alleged practices unlawful, but declined to order vertical divestiture and instead imposed system of mandatory competitive bidding for films. In a famous opinion by Justice Douglas, the Supreme Court reversed the bidding remedy, concluding that it was too complicated for a court to supervise and might reinforce the position of powerful exhibitors (Paramount, at 163–65). Instead, it remanded to the district court to reconsider whether vertical divestiture was appropriate to restore competition, particularly in the first-run segment of the market (Paramount, at 171–75). On remand, the district court found that the studios’ control of a substantial share of first-run theaters “when coupled with the strategic advantages of vertical integration, [conferred] a power to exclude competition from these markets when desired” (Paramount (1949), at 894). The court ordered all of the studios that had not already signed consent decrees to divest their theater holdings. As one recent study observed, the decrees “fundamentally altered the structure of the relationship between producers/ distributors and exhibitors” by prohibiting “contractual practices such as block booking.. the system of runs, clearance periods, and zoning under which films were distributed” and mandating “the divestiture of producer-owned cinemas” (Hanssen 2010, 519). A substantial scholarly literature has examined the consequences of the decrees. Those focusing on the structural aspects of the decree have suggested that the severing of studio control caused exhibition prices to increase, presumably because vertical integration had provided efficiencies and prevented double marginalization (Gil 2011). Hanssen (2010) suggests that vertical integration allowed studios to make welfare-enhancing adjustments to the length of exhibition runs, a variable that is difficult to estimate ex ante at the point of licensing.

13  Stigler (1968, 112) describes the merger as “a master stroke of monopoly promotion.” Mullin et al. (1995, 326) present stock market evidence that the failure of the Supreme Court to require dissolution of U.S. Steel, which remained a holding company in 1920 almost two decades after the merger that created it, harmed consumers. More recent studies also find that the merger was harmful (Gallet 2001; McCraw and Reinhardt 1989; Yamawaki 1985).

270   William H. Page Interestingly, the most significant structural remedy in antitrust history did not dissolve a merger to monopoly. Nevertheless, many observers conclude that it was justified by its unusual regulatory context. The Modified Final Judgment or “MFJ” in 1984 settled the AT&T case after 10 years of litigation (Coll 1986; Kearney 1999). The Bell System held regulated monopolies of local telephony, which it used to disadvantage rivals in long distance, like MCI, that needed to connect to local networks in order to complete their calls. The MFJ dissolved the system, separating long-distance and equipment providers from regulated regional carriers, the so-called Baby Bells, which the decree limited to providing local service. The dissolution eliminated the local carriers’ incentive to favor AT&T in long-distance service and equipment and enhanced efficiency in local service and competition in long distance and equipment (Krouse et al. 1999) until the Telecommunications Act of 1996 supplanted the MFJ with a different system designed to encourage entry in both long-distance and local telephony. Richard Posner characterized the MFJ as “the most successful antitrust structural remedy in history” (Posner 2001, 111). As we noted earlier in this section, the court in the Microsoft case—the most important public monopolization case since AT&T—rejected proposals to dissolve the defendant vertically, by separating the operating systems and applications lines of business,14 or horizontally, by requiring licensing of Microsoft’s intellectual property to new, rival entities (Litan et al. 2000; Salop 2000; Lenard 2001; Levinson, et al. 2001). Both forms of dissolution would have done more harm than good. Horizontal divestiture of Microsoft’s intellectual property to newly created rival operating system companies (so-called Baby Bills) would not have matched up with the liability holdings, which focused on Microsoft’s actions aimed at so-called middleware, particularly the browser and the Java programming language. Equally important, it would have created incompatible Windows rivals and thus sacrificed the benefits of indirect network effects. Vertical divesture would also have failed to correct any proven illegal conduct. More important, it would have hindered efficiency in multiple ways, as it did in the Paramount case. It would, for example, have hindered innovation by the operating systems company by impeding the addition of new functionality to the operating system. Equally important, by creating separate producers of complementary products, each with monopoly power, it would have raised prices through double marginalization.

11.3.2.  Conduct Injunctions When a court determines that a specific offense has occurred (or the parties agree that one has occurred), deterrence evidently failed ex ante to prevent the conduct. A court might determine that identification of the offense and the imposition of a penalty are enough to ensure that the offender will abandon the conduct. But if there is any doubt 14  The district court (Microsoft (1999)) initially ordered this kind of divestiture, but the court of appeals reversed (Microsoft (2001), at 98–107).

Optimal Antitrust Remedies   271

about the conduct recurring and imposing additional social harm, the court will consider enjoining it. For example, in Lorain Journal, a newspaper with a near-monopoly of local advertising refused to sell space to merchants who also advertised on a local radio station. The Supreme Court held that the newspaper was attempting to monopolize the advertising market by denying revenues to its only major rival (Lorain Journal at 150–54). The remedy ordered the newspaper to eliminate its exclusive advertising policy and to announce that it was doing so (Lorain Journal at 157–58). Lorain Journal offers a simple instance of a mandatory contracting remedy—one that requires an offender to deal with rivals or customers. Antitrust law generally permits firms to refuse to deal. The assumption that free contracting is value-enhancing rests on a premise that firms can choose not to contract unless the terms are satisfactory. Moreover, courts hesitate to supervise the process of contracting, because doing so may place the court in the role of a regulator. In a simple case like Lorain Journal, the defendant was only required to sell advertising on equal terms that it was free to prescribe, so the task of supervision was minimal. In AT&T, however, the process of supervising the conduct provisions of the decree required the district judge to act very much as a regulator for years. Conduct remedies are generally preferable to structural remedies when the defendant achieved its monopoly power lawfully, but engaged in specific anticompetitive conduct. In Microsoft, for example, the courts found Microsoft had a lawful monopoly of operating systems sheltered by an equally lawful “applications barrier to entry,” or network effects. The courts also held, however, that Microsoft violated Section 2 by certain actions that hindered the development of middleware like Netscape’s web browser and Sun’s Java. These applications had platform capabilities that, the court found, might eventually have posed a threat to Windows itself by eroding the applications’ barrier to entry. Developers might write applications to the programming interfaces provided by a middleware platform, because their applications could then run on any underlying operating system, not only Windows. Microsoft, however, formed contracts and designed Windows in ways that impeded this path of development (Microsoft (2001), at 60–74). As I noted in the last section, the courts rejected either horizontal or vertical divestiture. Instead, it imposed conduct remedies that responded, with few exceptions, to the specific actions that the courts had condemned. The courts held, for example, that Microsoft had monopolized by forming contracts with developers, Internet access providers, and computer manufacturers that disadvantaged Netscape by imposing exclusivity either directly or indirectly (Microsoft (2001), at 72–74). The final judgment responded to these holdings by barring Microsoft from imposing similar restrictions. Microsoft could not pay its partners not to support rival middleware or to retaliate against them if they did (Microsoft Consent Decree (2002) III.A, III.F. §III.G); nor could it prevent partners from modifying Windows in limited ways that promoted or gave access to rival middleware (§III.C.1). The judgments also required Microsoft to provide a mechanism for deleting access to the Internet Explorer browser in most instances and to allow computer manufacturers to install rival middleware to launch in place of Microsoft’s products (§III.H.1). Microsoft implemented

272   William H. Page the final judgments by adopting a uniform license and royalty schedule for licensees (Microsoft Joint Status Report (2003), at 19–20), and by instituting antitrust review of new contracts and Windows features (Microsoft Supplemental Joint Status Report (2003), at 10–20). It designed Windows to allow each user to define a preferred combination of programs for various functions. Computer manufacturers responded to the new freedom by installing various forms of rival software on Windows machines, including Sun’s Java, but did not delete any of Windows’ functionality (Evans and Nichols 2007). Nevertheless, Internet Explorer’s share of browser usage has fallen drastically. Of course, conduct remedies are more likely to be successful if they respond to real market needs. The European Commission held that Microsoft had violated antitrust law by bundling Windows Media Player with Windows. Although users could install other players, the Commission reasoned that the default installation gave Microsoft an unlawful edge. The remedy was to require Microsoft to offer a version of Windows without Windows Media Player—not merely the means of access to it, but much of the code that actually provides functionality. Microsoft was not required to offer the crippled version at a discount (Microsoft European Commission (2004) ¶¶ 978–84). Microsoft complied by developing “N” versions of Windows, which it offered at the same price as the standard version. Not surprisingly, there was no demand for the less functional version (Page 2009a, 798–800). Conduct remedies are also more likely to be successful if they require minimal supervision. Some remedial injunctions, particularly those focused on policing interoperability, have required supervision that approaches the level of industry-specific regulation. The MFJ that resolved the AT&T case, as we have seen, required a federal district judge to regulate important segments of American telecommunications for over a decade (Kearney 1999). The remedies in Microsoft aimed at ensuring interoperability also required a kind of regulation. These provisions were questionable, first, because the courts did not hold that Microsoft had wrongfully impeded interoperability, so there was reason to think that the remedies would not meet a real market need. These concerns were magnified when the specific remedies required extensive expert supervision and testing. One provision required Microsoft to disclose the application programming interfaces (APIs) that its own middleware applications use to run on Windows (Microsoft Consent Decree (2002) §III.D). Because Microsoft routinely gives developers access to its APIs, the specification of this set of APIs for early disclosure provided few difficulties of implementation. Far more problematic, however, was the “forward-looking” requirement that Microsoft “make available” to developers (by disclosures and licensing) the communications protocols that Microsoft’s server operating systems use to interoperate “natively” (that is, with functionality that is part of the operating system) in networks with Windows clients (Microsoft Consent Decree (2002) §III.E). The idea behind this remedy was to enable developers to write applications for non-Microsoft server operating systems that could interoperate with Windows client computers as well as applications written for Microsoft servers. In the long run, this might permit middleware running on servers to become a “platform threat” to Windows comparable to Netscape’s browser and Java. The

Optimal Antitrust Remedies   273

district court thought this extension of the rationale of the government’s case was necessary because the market was evolving in the direction of more network computing with server-based applications (Microsoft (2002), at 192; Microsoft (2004), at 1222–25). This part of the Microsoft remedy was by far the most costly implement and generated little competitive benefit (Page 2009b; Page  2007). From the outset, the governmentplaintiffs’technicalexperts—aTechnicalCommitteeandatechnicaladvisor— questioned the quality of Microsoft’s disclosures, based on the experts’ own extensive testing programs. Microsoft responded to their many reports of technical difficulties, or bugs, by trying to identify the sources of the problems and fix them. When few developers made use of the protocols the district court prodded the parties to try to improve them. Microsoft undertook what amounts to a program of software development to document the protocols, suspending royalties during the testing phase; it provided free technical support in many forms to assist developers in using the protocols in their products. All of this resembled price and entry regulation. Despite the costs of the program, few developers took licenses to the protocols. There was no evidence that the lack of licensees stemmed from the quality of the documentation Microsoft produced (Page 2007, 127–29). The more likely reason was that the licenses were not wanted: there were free technical means other than Microsoft’s native protocols to allow interoperation between servers and clients. More important, the primary competition for Microsoft’s applications has come from applications in the cloud—web-based applications that users can access with any standards-compliant browser. In an effort to allow the judgments to be more forward-looking, the court misread the direction of technical development. As with the EC’s order that Microsoft create a new version of Windows without a media player, the court required the creation of new products for which there was no demand. The EC also issued an interoperability order that was problematic, but in different ways. Sun Microsystems had complained that Microsoft was not providing information necessary for Sun’s workgroup servers to interoperate with Microsoft products in local networks (Page and Childers 2008). But the EC required disclosures that permitted a level of interoperability that, Microsoft contended, might allow rivals to clone its products. One beneficiary of the disclosures has been Samba, a project that produces a widely used open-source server operating system based on Linux. Using the disclosures from Microsoft required by the EC, Samba has sought (thus far unsuccessfully) to develop a version of its operating system that can act as domain controller in a Windows network using Active Directory, the network file administration and security service distinctive to Windows server operating system. Microsoft argued that to achieve this level of interoperation would involve cloning: disclosure of proprietary algorithms that would allow a rival producer to duplicate a Windows server’s internal logic. The EC concluded that this level of interoperability could be achieved by “functional equivalence” rather than simple copying. Microsoft has responded to the order by publishing “interoperability principles” and instituting a project aimed at ensuring interoperability with open-source products. At this writing, the competitive consequences of the EC’s order and Microsoft’s responses remain uncertain.

274   William H. Page So far, the discussion of conduct remedies has focused on orders that respond, more or less directly, to exclusionary actions. In some instances, however, conduct or behavioral remedies can be used in merger cases as a substitute for divestiture. As I noted in the last section, enforcers generally favor divestiture as a remedy for large-scale horizontal mergers, because it directly undoes or prevents the illegal acquisition of market power and requires little supervision. But divestiture also may prevent firms from achieving productive efficiencies or network effects. In some instances, including some vertical mergers, conduct remedies may adequately address competitive concerns and allow a cost-reducing merger to go forward. The Paramount case, discussed in the last section, may have been one in which a conduct remedy was preferable to divestiture. As Kwoka and Moss (2011) have observed, the most recent Antitrust Division Policy Guide to Merger Remedies (2011), more than earlier official statements, is receptive to conduct remedies such as firewalls and mandatory licensing, transparency, antiretaliation, and nonexclusive contracting restrictions. Recent consent decrees in merger cases confirm the Antitrust Division’s more positive assessment of the benefits of behavioral remedies, particularly in nonhorizontal mergers. For example, the Division settled its challenge to Google’s acquisition of a flight-search software producer in return for a variety of conduct restrictions, including requirements that Google license the software on “fair, reasonable, and nondiscriminatory” terms to travel sites and refrain from certain kinds of exclusive contracting (Kwoka and Moss 2011, 20). While laudably aimed at preserving efficiencies, these sorts of remedies, like the interoperability remedies in Microsoft, require antitrust enforcers to act as economic regulators (Kwoka and Moss 2011, 21–34), a task they should undertake only in well-defined circumstances.

11.4. Conclusion This chapter classified antitrust remedies into two broad categories—deterrent penalties and remedial injunctions—both aimed at the same goal of economic efficiency. The first category includes private treble damage actions and, for criminal violations like cartels, public fines and imprisonment. These remedies are justified to the extent that they advance the goal of optimal deterrence by imposing a sanction equal to the net harm to persons other than the offender, discounted for the probability of detection. Doctrines like antitrust injury and standing shape private damages to approximate the net harm standard; the special penalties for criminal offenses provide a necessary multiplier, because these offenses are by the nature more concealable than others, and address concerns about agency costs by imposing penalties on managers that violate the law surreptitiously. Injunctions are appropriate in instances in which deterrent penalties are likely to be insufficient to prevent anticompetitive conduct prospectively. Structural injunctions are costly, but may be appropriate to dissolve entities that illegally achieve durable

Optimal Antitrust Remedies   275

monopoly power, particularly by merger. Conduct remedies may also be costly, but can be well suited to end specific monopolistic conduct by firms with lawful monopoly power. Injunctive relief is most questionable when it seeks to regulate conduct beyond the practices specifically shown to reduce competition.

Acknowledgments I would like to thank Herbert Hovenkamp and John Lopatka for their comments.

References Adams, Walter. 1951. Dissolution, Divorcement, and Divestiture:  The Pyrrhic Victories of Antitrust. Indiana Law Journal 27:1–37. Baker, Jonathan B., and Daniel Law Rubinfeld. 1999. Empirical Methods in Antitrust Litigation: Review and Critique. American Law and Economics Review 1:386–435. Beckenstein, Alan R., and H. Landis Gabel. 1985. The Economics of Antitrust Compliance. Southern Economics Journal 52:673–92. Becker, Gary S. 1968. Crime and Punishment:  An Economic Approach. Journal of Political Economy 76:169–217. Blair, Roger D., and D. Daniel Sokol. 2012. The Rule of Reason and the Goals of Antitrust: An Economic Approach. Antitrust Law Journal 78:471–504. Clabault, James M., and Michael K.  Block. 1981. Sherman Act Indictments 1955–1980. New York: Federal Legal Publications. Coll, Steve. 1986. The Deal of the Century: The Break Up of AT&T. New York: Simon & Shuster. Connor, John M. 2012. Private Recoveries in International Cartel Cases Worldwide: What Do the Data Show? CPI Antitrust Chronicle, December. Connor, John M., and Robert H. Lande. 2005. How High Do Cartels Raise Prices? Implications for Optimal Cartel Fines. Tulane Law Review 80:513–65. Connor, John M., and Robert H. Lande. 2012. Cartels as Rational Business Strategy: Crime Pays. Cardozo Law Review 34:428–91. Crandall, Robert W., 2001, The Failure of Structural Remedies in Sherman Act Monopolization Cases. Oregon Law Review 80: 109–98. Crane, Daniel A. 2011. The Institutional Structure of Antitrust Enforcement. Oxford: Oxford University Press. Davis, Peter, and Eliana Garces. 2010. Quantitative Techniques for Competition and Antitrust Analysis. Princeton, NJ: Princeton University Press. Davis, Ronald W. 2003. Standing on Shaky Ground: The Strangely Elusive Doctrine of Antitrust Injury. Antitrust Law Journal 30:697–776. Devlin, Alan, and Michael Jacobs. 2010. Antitrust Divergence and the Limits of Economics. Northwestern University Law Review 104:253–91. Easterbrook, Frank H. 1985. Detrebling Antitrust Damages. Journal of Law and Economics 28:445–67.

276   William H. Page Elhauge, Einer. 2009. Disgorgement as an Antitrust Remedy. Antitrust Law Journal 76:79–95. Elzinga, Kenneth G. 1977. The Goals of Antitrust: Other Than Competition and Efficiency, What Else Counts? University of Pennsylvania Law Review 125:1191–213. Elzinga, Kenneth G., and William Breit. 1976. The Antitrust Penalties:  A  Study in Law and Economics: New Haven: Yale University Press. Elzinga, Kenneth G., et al. 2001. United States v. Microsoft: Remedy or Malady? George Mason Law Review 9:633–90. Evans, David S., and Albert L Nichols. 2007. The Behavior of PC OEMs and Growth of Server/ Network Computing Five Years after the Final Judgments in U.S. v. Microsoft, Report on Behalf of Microsoft, August 30. First, Harry. 2009. The Case for Antitrust Civil Penalties. Antitrust Law Journal 76:127–66. Friederiszick, Hans W. and Lars-Hendrick Röller. 2010. Quantification of Harm in Damages Actions for Antitrust Infringements: Insights from Germa Cartel Cases. Journal of Competition Law and Economics 6:595–618. Friedman, Ezra, and Abraham L. Wickelgren. 2010. Chilling, Settlement, and the Accuracy of the Legal Process. Journal of Law Economics and Organization 26:144–57. Gallet, Craig A. 2001. The Gradual Response of Market Power to Mergers in the U.S. Steel Industry. Review of Industrial Organization 18:327–36. Gil, Ricard. 2011. Does Vertical Integration Decrease Prices? Evidence from the Paramount Antitrust Case of 1948. Working paper, Johns Hopkins Carey Business School. Hale, George E. 1940. Trust Dissolution: Atomizing Business Units of Monopolistic Size. Columbia Law Review 40:615–32. Hamdani, Assaf, and Alon Klement. 2008. Corporate Crime and Deterrence. Stanford Law Review 61:271–310. Hanssen, F. Andrew. 2010. Vertical Integration during the Hollywood Studio Era. Journal of Law and Economics 53:519–43. Hjelmfelt, David C., and Channing D. Strother. 1991. Antitrust Damages for Consumer Welfare Loss. Cleveland State Law Review 39:505–20. Hovenkamp, Herbert. 1991. Enterprise and American Law, 1936–1937. Cambridge, MA: Harvard University Press. Hovenkamp, Herbert. 2010. The Law of Vertical Integration and the Business Firm: 1880–1960, Iowa Law Review 95:863–918. Kaplow, Louis D. 2011. An Economic Approach to Price Fixing. Antitrust Law Journal 77:343–449. Kearney, Joseph D. 1999. From the Fall of the Bell System to the Telecommunications Act of 1996:  Regulation of Telecommunications under Judge Greene. Hastings Law Journal 50:1395–472. Kneuper, Robert, and James Langenfeld. 2011. The Potential Role of Civil Antitrust Damage Analysis in Determining Financial Penalties in Antitrust Cases. George Mason Law Review 18:953–86. Kovacic, William E. 1999. Designing Remedies for Dominant Firm Misconduct. Connecticut Law Review 31:1285–319. Kovacic, William E. 1989) Failed Expectations: The Troubled Past and Uncertain Future of the Sherman Act as a Tool for Deconcentration. Iowa Law Review 74:1105–50. Krouse, Clement G., et al. 1999. The Bell System Divestiture/Deregulation and Efficiency of the Operating Companies. Journal of Law and Economics 42:61–83.

Optimal Antitrust Remedies   277

Kwoka, John E., and Diana L.  Moss. 2011. Behavioral Merger Remedies:  Evaluation and Implications for Antitrust Enforcement. Working paper, Department of Economics, Northeastern University. Lande, Robert H. 2003. Why Antitrust Damage Levels Should Be Raised. Loyola Consumer Law Review 16:329–45. Lande, Robert H. 2010. New Options for State Indirect Purchaser Legislation: Protecting the Real Victims of Antitrust Violations. Alabama Law Review 61:447–500. Lande, Robert H. and Joshua P. Davis. 2008. Benefits from Private Antitrust Enforcement: An Analysis of Forty Cases. University of San Francisco Law Review 42:879–918. Lande, Robert H., and Joshua P. Davis. 2011. Comparative Deterrence from Private Enforcement and Criminal Enforcement of the U.S. Antitrust Laws. Brigham Young University Law Review 2011: 315. Landes, William M. 1983. Optimal Sanctions for Antitrust Violations. University of Chicago Law Review 50:652–78. Landes, William M., and Richard A. Posner. 1979. Should Indirect Purchasers Have Standing to Sue under the Antitrust Laws? An Economic Analysis of the Rule of Illinois Brick. University Chicago Law Review 46: 602–35. Lenard, Thomas M. 2001. Creating Competition in the Market for Operating Systems: Alternative Structural Remedies in the Microsoft Case. George Mason Law Review 8:803–41. Levinson, Robert J., R. Craig Romaine, and Steven C. Salop. 2001. The Flawed Fragmentation Critique of Structural Remedies in the Microsoft Case. Antitrust Bulletin, 46:135–62. Lipsky, Abbott B. 2009. Managing Antitrust Compliance through the Continuing Surge in Global Enforcement. Antitrust Law Journal 75:965–95. Litan, Robert E., et al. 2000. Remedies Brief of Amici Curiae, United States v. Microsoft Corp., Civ. No. 98-1232, April 27. Lopatka, John E. 1989. The Case for Legal Enforcement of Price Fixing Agreements. Emory Law Journal 38:1–66. Lopatka, John E., and Page, William H. 1999. A (Cautionary) Note on Remedies in the Microsoft Case. Antitrust 13:25–28. McCraw, Thomas K., and Forest Reinhardt. 1989. Losing to Win: U.S. Steel’s Pricing, Investment, and Market Share, 1901–1938, Journal of Economic History 49:593–619. Mnookin, Robert H., and Lewis Kornhauser. 1979. Bargaining in the Shadow of the Law: The Case of Divorce. Yale Law Journal 88:950–97. Mullin, George Law, et al. 1995. The Competitive Effects of Mergers: Stock Market Evidence from the U.S. Steel Dissolution Suit. Rand Journal of Economics 26:314–30. Mullin, Wallace P., and Christopher M. Snyder. 2010. Should Firms Be Allowed to Indemnify Their Employees for Sanctions? Journal of Law Economics and Organization 26:30–53. Mutchnik, James H., et  al. 2008. The Volume of Commerce Enigma, Antitrust Source 7 (June): 1–10. O’Connor, Kevin J., et al. 2012. Interaction of Public and Private Enforcement. In Albert A. Foer and Randy M. Stutz, Private Enforcement of Antitrust Law in the United States: A Handbook. Cheltenham: Edward Elgar. 280–304. Page, William H. 1980. Antitrust Damages and Economic Efficiency: An Approach to Antitrust Injury. University of Chicago Law Review 47:467–504. Page, William H. 1985. The Scope of Liability for Antitrust Violations. Stanford Law Review 37:1445–512.

278   William H. Page Page, William H. 1990. Optimal Antitrust Penalties and Competitors’ Injury. Michigan Law Review 88:2151–66. Page, William H. 1991. Ideological Conflict and the Origins of Antitrust Policy. Tulane Law Review 71:1–67. Page, William H. 2009a. Mandatory Contracting Remedies in the American and European Microsoft Cases. Antitrust Law Journal 75:787–809. Page, William H. 2009b. Measuring Compliance with Compulsory Licensing Remedies in the American Microsoft Case. Antitrust Law Journal 76:240–69. Page, William H. 2012. Indirect Purchaser Suits after the Class Action Fairness Act: Reconciling Multilayer Interests. In Stefan Wrbka et  al., Collective Actions:  Enhancing Access to Justice and Reconciling Multilayer Interests? Cambridge: Cambridge University Press. 275–98. Page, William H. and Seldon J.  Childers. 2007. Software Development as an Antitrust Remedy:  Lessons from the Enforcement of the Microsoft Communications Protocol Licensing Requirement. Michigan Telecommunications and Technology Law Review 14:77–136. Page, William H., and Seldon J.  Childers. 2008. Bargaining in the Shadow of the European Microsoft Decision: The Microsoft-Samba Protocol License. Northwestern University School of Law Review Colloquy 102:332–54. Parsons, Donald O., and Edward John Ray. 1975. The United States Steel Consolidation: The Creation of Market Control. Journal of Law and Economics 18:181–219. Pickler, Nedra. 2012. Japanese Auto Suppliers to Pay Price-Fixing Fine. San Francisco Chronicle, January 30. Polinsky, A. Mitchell, and Steven Shavell. 1984. The Optimal Use of Fines and Imprisonment. Journal of Public Economics 24:89–99. Polinsky, A.  Mitchell, and Steven Shavell. 1993. Should Employees Be Subject to Fines and Imprisonment Given the Existence of Corporate Liability? International Review of Law and Economics 13:239–57. Posner, Richard A. 1970. A Statistical Study of Antitrust Enforcement. Journal of Law and Economics 13:365–419. Posner, Richard A. 2001. Antitrust Law. 2nd ed. Chicago: University of Chicago Press. Reeves, Amanda P., and Maurice E. Stucke. 2011. Behavioral Antitrust. Antitrust Law Journal 86:1527–86. Rosenberg, David, and James P.  Sullivan. 2006. Coordinating Private Class Action and Public Agency Enforcement of Antitrust Law. Journal of Competition Law and Economics 2:159–87. Rosenberg, David, and Steven Shavell. 2006. A Solution to the Problem of Nuisance Suits: The Option to Have the Court Bar Settlement. International Review of Law and Economics 26:42–51. Shelanski, Howard A., and Journal Gregory Sidak. 2001. Antitrust Divestiture in Network Industries. University of Chicago Law Review 68:1–99. Snyder, Edward A., and Thomas E. Kauper. 1991. Misuse of the Antitrust Laws: The Competitor Plaintiff. Michigan Law Review 90:551–98. Standen, Jeffrey. 1995. The Fallacy of Full Compensation. Washington University Law Quarterly 73:145–226.

Optimal Antitrust Remedies   279

Stigler, George J. 1968. The Dominant Firm and the Inverted Umbrella. In The Organization of Industry. Homewood, IL: Richard D. Irwin. Stigler, George J. 1970. The Optimum Enforcement of Laws. Journal of Political Economy 78:526–36. United States Department of Justice. Antitrust Division Workload Statistics FY 2000. United States Department of Justice. Antitrust Division Criminal Enforcement Fine and Jail Charts, 2000–2010. United States Department of Justice. 2011. Antitrust Division Policy Guide to Merger Remedies. June. United States Sentencing Commission. 2010. Guidelines Manual. Werden, Gregory J., and Marilyn Journal Simon. 1987. Why Price Fixers Should Go to Prison. Antitrust Bulletin 32:917–37. Wils, Wouter P.  J. 2006. Optimal Antitrust Fines:  Theory and Practice. World Competition 29:183–206. Yamawaki, Hideki. 1985. Dominant Firm Pricing and Fringe Expansion: The Case of the U.S. Iron and Steel Industry, 1907–1930. Review of Economics and Statistics 67:429–37.

Cases Brunswick Corp. v. Pueblo Bowl-O-Mat, 429 U.S. 477 (1977). California v. ARC America Corp., 490 U.S. 93, 102 (1989). Cargill v. Monfort of Colorado, Inc., 479 U.S. 104, 109 (1986). Case COMP/C-3/37.792—Microsoft Corp., Comm’n Decision, 2007 O.J. (L 32) 23, ¶¶ 978–84 (Mar. 24, 2004), affirmed, Case T-201/04, Microsoft Corp. v. Comm’n, 2007 E.C.R. II-3601 (Ct. First Instance). Corn Prods. Ref. Co. v. United States, 249 U.S. 621 (1919). Ford Motor Co. v. United States, 405 U.S. 562, 573 (1972). Illinois Brick Co. v. Illinois, 431 U.S. 720, 748 (1977). International Harvester Co. v. United States, 248 U.S. 587 (1918)). Nat’l Soc’y Prof ’l Eng’rs v. United States, 435 U.S. 679, 698 (1978). United States v. Aluminum Co. of American, 91 F. Supp. 333, 416 (S.D.N.Y. 1950). United States v. American Can Co., 230 F. 859, 903 (D. Md. 1916), appeal dismissed, 256 U.S. 706 (1921). United States v. American Tobacco Co., 221 U.S. 106 (1911). United States v. Andreas, No. 96 CR 762, 1999 WL 515484, at *2 (N.D. Ill. July 15, 1999). United States v. AT&T, 552 F. Supp. 131, 226–34 (D.D.C. 1982), affirmed, Maryland v. United States, 460 U.S. 1001 (1983). United States v. Dentsply International, Inc., 277 F. Supp. 2d 387 (D. Del. 2003), reversed, 399 F. 3d. 181 (3d Cir. 2005). United States v. E. I. du Pont de Nemours and Co., 366 U.S. 316, 326 (1961). United States v. Keyspan Corp., 763 F. Supp. 2d 633 (S.D.N.Y. 2011). United States v. Microsoft Corp., 231 F. Supp. 2d 144, 192 (D.D.C. 2002)  (Consent Decree), affirmed, Massachusetts v. Microsoft Corp., 373 F.3d. 1199 (D.C. Cir. 2004). United States v. Microsoft Corp., 253 F.3d 34, 106 (D.C. Cir. 2001).

280   William H. Page United States v. Microsoft Corp., 84 F. Supp. 2d 9, 19-24 (D.D.C. 1999). United States v. Microsoft Corp., 97 F. Supp.2d 59, 64 (D.D.C. 2000), reversed, 253 F.3d 34 (D.C. Cir. 2001). United States v. Paramount Pictures, 334 U.S. 131, 140 (1948). United States v. Paramount Pictures, Inc., 85 F. Supp. 881, 894 (S.D.N.Y. 1949), affirmed, 339 U.S. 974 (1950). United States v. United Shoe Machine Corp., 110 F. Supp. 295, 348 (D. Mass. 1953).

CHAPTER 12

P R I VAT E A N T I T RU S T E N F OR C E M E N T I N T H E U N I T E D S TAT E S A N D T H E E U R O P E A N U N ION Standing and Antitrust Injury JEFFREY L. HARRISON

12.1. Introduction Aside from interpretations of the substantive law, the level of private antitrust enforcement is a function of a number of factors, the most important of which are (1) the availability of punitive or exemplary damages; (2) the possibility of recovering attorney’s fees; (3) the difficulty of creating a class of plaintiffs or of obtaining collective redress; (4) the ease of discovery; and (5) definitions of what qualifies one to bring a private action. This discussion is about the last of these—the eligibility issue. To date, and despite pressures toward convergence, the United States and the European Union (EU) have taken different paths with respect to the enforcement of antitrust laws by private parties. U. S. law is more encouraging to private enforcement than EU law. As one author has aptly put it, the EU permits private action while the United States makes it attractive (Sittenreich 2010, Rizzuto 2010). In the EU, the substantive antitrust provisions are those found in the Treaty on the Functioning of the European Union. Procedural requirements as they pertain to private actions are determined by national courts. Consequently, any general statements about the eligibility issue may not hold true in particular cases. Nevertheless, as a general matter, there seems to be little question that Europe as a whole is struggling with the eligibility question, and current policies suggest substantial ambivalence. Official statements

282   Jeffrey L. Harrison favor greater private enforcement, but the actual incentives for bringing privates actions are not compelling. The next section sets out the basic legal analysis of the “eligibility” question in the United States and the EU. The goal is to provide the legal regime within which the economic analysis can be assessed. Four relevant economic models are then examined. The discussion of each model includes an analysis of the importance of standing and injury requirements to the level of private enforcement efforts. There are a number of qualifying notes. Much of the following concerns the deterrent impact of private enforcement in the EU as a whole. Obviously, the overall impact of enforcement efforts will be determined by both public and private actions. The discussion also does not consider the varying impacts of individual member state enforcement efforts. The impact of any member state will also be affected by the availability of collective redress in that state. The availability of public redress is significantly different from state to state.

12.2.  The Eligibility Question As a matter of theory, the possibility of private enforcement exists along a continuum ranging from very likely to nonexistent. For example, a set of antitrust laws could be enacted that involve no public enforcement at all but which allow anyone—whether directly affected or not—to file an action and collect punitive or statutory damages. This might be called the pure “bounty hunter” model. Between the bounty hunter model and one in which there is no private enforcement there are a number—perhaps an infinite number—of positions. The position chosen is ultimately a matter of policy and can be compared to turning on or off a private enforcement faucet. As noted at the outset, this faucet can be adjusted in a number of ways, but for purposes of this chapter the faucet is controlled by how broadly eligibility is defined. A critical policy choice is whether private actions are designed to further the goal of compensating victims of antitrust violations or deterring anticompetitive action in the first place. A regime of antitrust law designed to compensate those harmed may look quite different from one designed to punish and deter anticompetitive actions. The question in this respect is closely related to the difference between a civil action based on a tort theory and the enforcement of criminal law. In one case the objective is simply to restore the party damaged to a “but for” position, while, in the criminal case, the objective is to prevent the damage from occurring. The second objective is achieved by going further than requiring the offending party to pay damages or to disgorge any ill-gotten gains. In fact, those measures may still mean anticompetitive action has a net positive outcome for those involved. Put differently, from the firm’s point of view, there would be an efficient level of anticompetitive behavior. Punitive sanctions, on the other hand, are typically designed to make the efficient level of anticompetitive behavior close to zero. Although the compensation and deterrence goals are obviously quite different in some respects, it is also important to recognize how closely related they are. When an

Private Antitrust Enforcement in the United States and the European Union   283

antitrust violation occurs, there are external effects—generally higher prices, lower output, and reduced consumer surplus. Compensation offsets that harm, and, if all victims were assured of full compensation, a tort-like regime would be attractive to many. On the other hand, the deterrence model achieves the same end—protecting potential victims from a reduction in wealth and consumer welfare—by discouraging the harm in the first place. In short, antitrust violations are penalized not because of a widespread belief that they are morally reprehensible but because deterrence is a more effective way to address the compensation problem.

12.2.1.  The United States In the United States, two doctrines determine eligibility—antitrust injury and antitrust standing. An antitrust injury is the type of injury the antitrust laws are designed to prevent. This limitation results from the statutory requirement that damages only be awarded to those injured “by reason” of activities forbidden by the antitrust laws. A good example illustrating the issue is found in the US Supreme Court case Brunswick Corp. v. Pueblo Bowl-O-Mat. In that case the plaintiff, an operator of bowling alleys, challenged the acquisition of a competing but failing bowling alley by an upstream supplier, arguing that it was an unlawful merger. The Supreme Court, in ruling that the plaintiff had not suffered antitrust injury, did not address the merits of the claim. Instead, the plaintiff was ruled, in effect, “ineligible.” The plaintiff ’s goal in invoking the antitrust laws was to advance its desire to gain market power. The harm from the merger—the inability of the plaintiff to become a monopolist itself—was not “antitrust injury.” This is not to say the merger would not ultimately be anticompetitive. That question was not addressed. Instead, the plaintiff ’s injury was not linked to the possibility of a less competitive market. The antitrust injury requirement is one way in which US law deviates significantly from the pure “bounty hunter” model. The second requirement—standing—is that the plaintiff must not be too remotely affected by the anticompetitive act. While the injury requirement can be viewed as a straightforward interpretation of the “by reason of ” language of the relevant legislation, the standing requirement is a bit more nebulous and probably more clearly reflects judicial philosophy with respect the proper level of private antitrust enforcement and its purpose. The standard has been interpreted to mean that an eligible plaintiff must be an efficient enforcer of the antitrust laws. What this means precisely cannot be determined. Important elements include how directly the impact is felt and the likelihood that the plaintiff or plaintiffs are in the best position to advance the procompetitive goals of the antitrust laws. For example, a buyer or seller may have suffered antitrust injury but only as a result of an anticompetitive action, the impact of which first rippled through several other buyers or sellers. That party may have suffered relatively small damages and it would be quite challenging to establish what the actual measure of damages is. This plaintiff would not be viewed as the best private party to advance the goals of the antitrust laws.

284   Jeffrey L. Harrison The doctrines of antitrust injury and standing developed independently and in a common-law-like fashion. They have both been revisited and refined over the last four decades. Today, it is accurate to say that in order to have antitrust standing one must have suffered an antitrust injury. Conversely, not all those suffering antitrust injury have antitrust standing. It is important to note a third standard that influences the eligibility issue in certain cases. It fits most logically under the concept of antitrust standing, but, because of the sequence in which the US Supreme Court decided the relevant cases, it was not officially viewed as a standing decision. The issue involves the eligibility of those who purchase from those who have paid supracompetitive prices due to collusion of their suppliers. Specifically, if a firm buys from another firm that itself buys from upstream firms that are price fixing, can that indirect purchaser collect damages based on the amount of the overcharge1 that is passed through by the firm from which it purchased? The US Supreme Court, in Illinois Brick Co. v. Illinois, answered that indirect purchasers may not recover. As already noted, the decision was made before modern injury and standing doctrines became focused. A more modern interpretation of the policy is that indirect purchasers from price-fixing firms have suffered antitrust injury but ultimately do not have standing. The direct/indirect purchaser standards, like standing more generally, do not necessarily flow from the US antitrust statutes but reflect judicial efforts to answer questions left open by that legislation. Although these doctrines are relatively well developed in US antitrust law, there are important nuances that should not be missed. First, the fact that a party may lack standing or may not have suffered an antitrust injury does not involve the question of whether there has been a substantive violation of the antitrust laws. For example, as already noted, in the leading antirust injury case, Brunswick Corp. v. Pueblo Bowl-O-Mat, the Court did not reach the issue of whether anticompetitive actions had occurred. Instead, in an injury inquiry, a court examines the plaintiff ’s theory of how it was injured individually and if that injury is the type the antitrust laws are designed to avoid. If not, the inquiry stops. A ruling that a party does not have standing creates the risk of an underenforcement error. A ruling that a party does have standing does not, however, create the risk of overenforcement since that party must still demonstrate a substantive violation. Second, to the extent the United States encourages private actions, it ranks the deterrence goal ahead of the compensation goal. A focus on compensation, as noted earlier, is one that attempts to return those harmed to the ex ante status quo. It its fullest form it requires tracing the impact of the anticompetitive harm through layers of buyers and sellers. For example, a horizontal agreement on the price of a raw material may have an impact on manufacturers of a product, as well as its wholesalers, retailers, and final customers. More importantly it would require that defendants be allowed to employ a pass-on defense. If this were not the case, there would be overcompensation. On the other hand, a focus on deterrence means that it is important to have at least one highly 1  The overcharge is the difference between the price charged by the price-fixing firms and that charged in the absence of price fixing.

Private Antitrust Enforcement in the United States and the European Union   285

motivated plaintiff who has much to gain from a successful action whether or not that gain is equal to the actual loss suffered. This means disallowing a pass-on defense and a punitive element. The perfect example of this deterrence over compensation policy is Illinois Brick. A  compensation function would have likely meant allowing indirect purchasers to recover. This too could potentially reduce the damages paid since the lower the impact as one goes the through the chain of distribution, the lower the probability of an action or a successful showing of actual harm. A policy of no indirect action, when coupled with the disallowance of a pass-on defense means one set of directly affected plaintiffs will have the maximum incentive to bring an action.

12.2.2.  The European Union A comparison of the US and EU standards with respect to injury and standing cannot be a straightforward one in part because private enforcement in the EU is still relatively infrequent and the doctrines are considerably less stable than in the United States. Plus, focusing on these concepts alone could be misleading. Standing and injury, as noted at the outset, are only two of the tools that regulate the level of private enforcement. For example, in terms of private enforcement, lower punitive damages could offset a more liberal antitrust standing standard. This means the regimes could be very much alike with respect to standing and injury but have an overall very different posture towards private enforcement generally. Finally, private antitrust actions must be brought in national courts. The issues of standing and injury are generally regarded as procedural, and each national court’s procedural rules govern. These rules often differ from nation to nation (Cook 2008). Nevertheless, whether or not the terms “standing and injury” are employed, the general issue of eligibility to bring a private action in the EU is a critical one, and the ways potential plaintiffs may be pared down are likely to have counterparts in US law. EU law and national courts have, albeit slowly, begun to deal with the “right plaintiff ” issue. A European Commission green paper in 2005 and a subsequent white paper in 2008 encourage increased private enforcement. Recent discussions, however, suggest this urging has not been successful as an overall matter (Rizzuto 2010; Romain and Gubbay 2011). As one would expect when the progress is slow, uniform answers have yet to emerge from national courts, which will have to entertain private actions. This means in the current time period and context, the best that can be done is to predict where EU policy on these matters will lead. That prediction is that without significant changes, private enforcement in the EU is destined to remain underdeveloped. The core of the difference is explained by the compensation/deterrence distinction. Most indications are that, while the United States has chosen deterrence, the EU has a far greater interest in compensation. In fact, this tendency is consistent with the generalized opposition of European countries to awarding punitive damages (Reyer 2009).

286   Jeffrey L. Harrison A good place to start in attempting to understand the lack of clarity in the EU with respect to private actions and, at the same time, consider the primacy of compensation is Courage v. Crehan, the case most responsible for initiating greater interest in private actions. The court in that case wrote that enforcement “must be open to any individual to claim damages for loss caused to him by a contract or by conduct liable to restrict or distort competition” (Crehan, 1-6323, 2001) The “any individual” wording strongly suggests a broad notion of standing and “caused to him” is consistent with a compensation goal. Perhaps the best general guidance is found in the 2008 white paper, which included a number of specific proposals designed to facilitate an increase in private actions. The white paper clearly focuses on “compensation” as the primary goal of private actions. Thus, plaintiffs should “receive full compensation of the real value of the loss suffered” (emphasis added). In addition, “Full compensation is the . . . first and foremost guiding principle” (European Commission 2008, 3). The drafters of the white paper do observe that greater compensation will mean a greater level of deterrence. As noted later, whether this is a realistic expectation is another matter. Nevertheless, when compared to the United States the preference for compensation at the expense of deterrence is clear. It is possible that this preference reflects a view that European public enforcement fines are high enough to offer sufficient deterrence. One of the stickier issues that falls within the discussion of compensation and deterrence is the availability of a pass-on defense. A  fully developed compensation approach would mean that defendants who were able to show that plaintiffs had passed overcharges onto downstream purchasers would be able to deduct those amounts from an award. Otherwise, there would be overcompensation. The 2006 green paper suggests a variety of options for dealing with the direct/indirect purchaser problem, including both allowing and not allowing the pass-on defense. The more recent white paper is more specific and proposes that defendants (infringers) be permitted to apply a “passing on” defense. It is important to remember, though, that these statements do not ensure compliance by national courts. Indeed, for the most part, the issue has not been addressed but, at this point, the EU appears poised to take a very different approach than the United States. Indeed, one of the arguments for compensation over deterrence is the express desire not to become like the United States (Milutinovic 2010). The possibility that antitrust victims may be indirect and still have “standing” in European courts does not provide a complete answer to the issues of standing and injury. The antitrust injury question, it will be recalled, is directed at the type of injury. On this matter, there is also a lack of clarity. A literal interpretation of the wording in Courage allows an action by someone injured as the result of a market “distortion.” Whether “distortion” is the same as movement from more competitive to less competitive conditions is not clear. Veljko Milutinovic offers the example of a situation in which small competitors collude to compete more effectively against a dominant firm (Milutinovic 2010, 232–33). Suppose the dominant firm then brings a private action against the colluding firms. In the United States there would be antitrust injury questions, if not a clear

Private Antitrust Enforcement in the United States and the European Union   287

resolution, because the goal of the dominant firm would be to use the antitrust laws to decrease competition. In the EU, however, the dominant firm may be viewed as having suffered antitrust injury as a result of a market distortion. At least at this point, there is no legislative authority like that in the United States that would further define the kind of injury that must be suffered before the antitrust laws are involved. As in the case of the United States, the importance of injury and standing standards in the EU can be overstated. Regardless of how liberal EU policies may become with respect to standing and injury, a plaintiff, who may not have suffered antitrust injury as defined under the US system, still has to establish that the substantive provisions of Articles 101 or 102 have been violated.

12.2.3. A Comparison Although EU and US policies with respect to standing, injury, and indirect purchases can be viewed as simply at different stages of development, there are probably more important policy considerations at work. These policy considerations result in almost an opposite approach to private actions. In the United States at one level, private enforcement has been designed to be deterrent-oriented and to encourage private actions. Yet, as the same time, the impact of court decisions has been to make the system far less open to plaintiffs. It does not appear to be by accident that at the same time the US Supreme Court was decreasing the scope of the substantive reach of the US antitrust laws in order to focus on economic efficiency and consumer welfare, it was narrowing the range of possible plaintiffs. Considered together, the effect of these two changes was to decrease the likelihood of so-called false negatives—practices incorrectly labeled as anticompetitive. In the EU a far different dynamic is in play. There seems to be a general desire to increase private actions, but the underlying structure reduces the incentives to bring private actions. In effect, expressions of a desire for greater private actions are offset by the general objection to punitive damages and consolidated actions. In fact, it is hard to imagine a system of private enforcement emerging in the EU as long as single damages remain the norm (Rouhette 2007). A more practical issue also separates the EU from the United States as far as private actions. It is the challenge of achieving some level of uniformity among national courts. Economics alone cannot achieve this goal because a nation’s approach to the issues discussed here is a function of history, culture, and legal system. The implications of the deterrence/compensation distinction are discussed in the following section on the economic theory of damages. The central question is: How do the interests of compensation and deterrence affect questions of standing and injury? This follows from difference between the EU and the United States with respect to these goals. In this analysis the indirect/direct purchaser issue will be viewed as part of a standing analysis. In order to understand the issues more clearly it is useful to examine four specific economic models.

288   Jeffrey L. Harrison

12.3.  Economic Models Much of the deterrence/compensation analysis can be viewed in the context of the question of how one is damaged by anticompetitive behavior.2 There are a number of possibilities depending upon whether victims are consumers or sellers and on the type of offense involved. In the case of consumers, the damage is in the form of lost consumer surplus. For firms that resell, the harm is in terms of lost profits. That is not, however, how damages are uniformly calculated. When the offense has a direct impact on price, the recovery is measured by the amount of the overcharge or the difference between the price paid and price in the market without the price-increasing behavior. In other instances in which the harm is not the result of a direct impact on price, the measure is lost profits. The most interesting questions involve overcharge measures of damages because it is in this context that issues of pass-on, compensation, and deterrence are raised.

12.3.1.  Model 1: Price Fixing and Consumers US law allows a recovery for “three fold” the damages suffered. In the typical price-fixing case this means three times the overcharge3 per unit multiplied by the number of units purchased during the period of price fixing. In figure 12.1 the prices charged in a market in which purchases are made by consumers without price-fixing collusion is compared to one in which price fixing exists. Before collusion, the price is P1 and the quantity purchased is Q1. The consumer surplus enjoyed by buyers is the area of triangle P1AR. Using the standard model, if price fixing results in an elevation of price to P2, quantity purchased will decline to Q2. Now consumer surplus is P2BR. It has been decreased by the area of triangle ABC. In the typical antitrust case, the amount recovered by the affected consumers would be the area P1P2BC. This would be trebled in the United States but not, at this point, in the EU. It is worthwhile to note that the base amount is not consistent with compensation. In particular, consumers who are no longer purchasers but who are nonetheless harmed by virtue of a decrease in consumer surplus are not compensated. In fact, the overcharge measure is more like a disgorgement measure in that it takes away the illicit gains of the conspirators. The overcharge measure itself is, in effect, a de facto standing or injury determination because only those who pay the overcharge after the price increase are eligible to file an action.

2 

It is important to note that the calculation of damages is a different matter than the theory of how one is damaged. The following pages focus on the latter. 3  The evolution of the “overcharge” measure of damages has been examined elsewhere (Harrison 1980).

Private Antitrust Enforcement in the United States and the European Union   289

P R

S B P2

C

A

P1

D Q2 FIGURE  12.1  Price

Q1

Q

Fixing and Consumers

This is not of great significance in the United States, where private enforcement efforts remain robust. On the other hand, in the EU, with its desire to increase private actions and to fully compensate those injured, the overcharge measure seems inappropriate. It is not clear how much difference it would make ultimately, but a decision that those who no longer purchase after a price increase have both standing and have suffered injury, at least at the margin, could offset the impact of some of the policies that discourage private actions in the EU. In order to understand the importance of extending standing to those no longer purchasing, it is useful to recall that the overcharge measure is a disgorgement measure. The maximum impact in a single-damages context is returning to victims the amount taken. If the probability of losing a lawsuit for the overcharge is less than one, there is little incentive not to engage in the anticompetitive action. The inclusion of the harm to those who no longer purchase means the amount collected could exceed the gains to those fixing prices. Increasing damages over the amount taken, thus, decreases the expected value of price fixing, making it less likely. In short, in the EU, a decision that those no longer purchasing are nonetheless harmed is inconsistent with both deterrence and compensation.

12.3.2.  Model 2: Price Fixing to Those Reselling under Imperfectly Competitive Conditions The analysis is slightly more complex when the purchaser from a price-fixing cartel is itself a producer and selling under imperfectly competitive conditions. In this case, the direct victim is the purchaser of an intermediate good that may be sold as a final good or travel further through the chain of production and distribution. In figure 12.2,

290   Jeffrey L. Harrison P

E E

A

P2

B

P1 C2

C2

D F

E

C1

C1 D MR Q2

Q1

FIGURE  12.2  Price

Q

Fixing to Reselling Buyers

D represents the demand that a firm faces for its output.4 For purposes of simplicity it is assumed that marginal cost and variable cost are the same and equal to C1. It is also assumed that one unit of input translates to one unit of output. Following standard microeconomic theory, the firms sells where its marginal cost and marginal revenue intersect. In the case of C1, this is at Q1. The price charged is the highest consistent with that level of output. Here that price is P1. Now suppose, due to the collusion of suppliers, the average cost and marginal cost curve increase to C2. Again, applying the standard analysis, the new profit-maximizing quantity is Q2 and the profit-maximizing price is P2. At this point it is useful to note two possible measures of damages. The overcharge, as in the consumer model, is the difference between C1 and C2, and the total recognized damage is the area of rectangle C1C2DE. On the other hand, the actual harm suffered is lost profits. Prior to the price fixing, the profit, including contributions to overhead, was C1P1BF. After price fixing, profit has declined to C2P2AD. There seems to be little question that an actual compensatory award for the direct purchaser would be C1P1BF – C2P2AD. The questions are how either measure matches up with a deterrence-oriented policy or a compensation-oriented policy and what the implications are for standing and injury. As noted earlier, the overcharge measure is really one that disgorges the gains of the price fixers. If one accepts deterrence as the primary function of private action, these details may not be significant. Any combination of damages and standing standards that results in an expected return to anticompetitive activity that is less than zero to potential

4 

A complete version of this model is available (Harrison 1980).

Private Antitrust Enforcement in the United States and the European Union   291

price fixers achieves the desired end. This is hardly consistent with compensation, but, as noted above, successful deterrence makes the compensation issue moot. Consequently a trebled overcharge measure, particularly without standing for indirect purchasers, it can be argued, achieves both ends. From the perspective of the EU, the implications of the overcharge measure are significant. Of principal importance are the questions of standing for indirect purchasers and the use of a pass-on defense for those engaged in price fixing. As noted above, the green paper suggests four options for dealing with the question; the more recent white paper favors allowing recovery by indirect purchasers and a pass-on defense by defendants. In the graph, defendants could then subtract the area P1P2AE from the overcharge of C1C2DE. Standing for indirect purchasers would mean that this portion of the overcharge would be collected by indirect purchasers. Under the white paper proposal, standing would evidently extend well beyond the most directly affected indirect purchasers. In order to facilitate the ability of these more remote purchasers to recover damages, the white paper recommends that those remote plaintiffs be armed with a rebuttable presumption that the “illegal overcharge was passed onto them in its entirety.” Ultimately, however, without punitive damages the highest level of liability will be equal to the ill-gotten gains. This and the small amounts of potential recovery by remote purchasers lessen the possibility of a deterrent effect. More important from the point of view of the EU is the inconsistency of the overcharge measure with compensation. A lost-profits measure of damages is more consistent with compensation. The recovery could not be limited to the direct purchasers only but to those both above and below that buyer. In particular, if some of the price increase is passed on, those downstream manufacturers are also likely to suffer lost profits. Moreover, the firms engaged in price fixing must reduce their output in a way that is consistent with the higher price. This means fewer inputs are purchased from other suppliers. In addition, those firms charged the higher prices will similarly reduce their output, and suppliers to those firms will also find a lower demand for the inputs sold to those firms. The impact of the price fixing permeates the production and distribution system even further. Firms operating at the same level as those fixing prices will be able to take advantage of the price umbrella provided by the price fixers. They too will reduce output and purchases from upstream firms. Their customers, now also producing less, will raise prices to downstream customers and reduce their upstream purchases. All of these affected parties can be said to have suffered antitrust injury—their injuries are the sort the antitrust laws are intended to prevent. In the United States, unless compensation occurs in an ex ante fashion by virtue of a deterrent effect, compensation is most likely not achieved. At this point it does not appear upstream sellers have been recognized as having standing. Downstream purchasers, even if asking for lost profits, will be regarded as indirect. With the expansive notion of standing evidently found in the EU, there is at least the potential for compensation. But this is only in theory. The lost profit for some in the chain of distribution is likely to be low, and there is a risk of no recovery. In effect, even an expansive notion of standing is unlikely to offset the absence of punitive damages. In sum, the overcharge

292   Jeffrey L. Harrison measure falls short as deterrent, it is not compensatory, and it is not a measure of the competitive harm. In fact, there is probably no measure of damages that would as a practical matter achieve all of these goals. For example, tracing the lost profit and loss in consumer surplus lost through the supply and distribution chain is likely to be impossible. Indeed, it may be impossible to even determine all the affected parties. The United States, at least in price-fixing cases, uses the tools of antitrust standing and injury along with treble damages to create incentives for private actions that act as deterrents. The principal drawback of this consolidation of power in direct purchasers is that if they depend on close ties with suppliers and are able to pass on much of the overcharge, the incentive to take action can be reduced.

12.3.3.  Model 3: Monopsony Power and Collusion of the Buying Side of the Market A very similar analysis to the one described above can be applied to collusion on the buying side of the market. Typically, when one thinks of buying power it involves the purchase of labor. The condition can and does exist at every level of the production and distribution chain and affect a great variety of inputs. What makes the analysis a bit confusing are the questions of why antitrust policy should oppose the use of buying power and, if so, who passes the tests for eligibility. The substantive question flows from the fact that those with buying power seek to lower prices. Lower prices, one might reason, leads to lower costs of production throughout the chain of production and distribution, eventually benefiting consumers. The problem with this analysis has been demonstrated (Blair and Harrison 2010) and can be understood by examining figure 12.3. Demand for the input is determined by how profitable it is for a firm to purchase and use an additional unit of the input. This is marginal revenue product (MRP). Demand and supply intersect at Q1. The price for the input in this competitive market is P1. If buyers collude, they no longer accept the market-determined price but set the price. This means comparing the profitability of each additional unit of input purchased with its marginal cost or marginal factor cost. For all units out to Q2, the marginal revenue product exceeds the marginal factor cost. This net gain ends at Q2, which is now the profit-maximizing level of input purchases. The price is set at the lowest price consistent with obtaining Q2 units of input or P2. From the graph the harm from the use of monopsony power is obvious. As a generally matter, resources are misallocated from what would be allocatively efficient. In other words, they do not find their way to their most valued uses. Second, the purchase of inputs declines. This may not seem to be a problem initially, but a decline in the purchase of inputs means less output is produced. When output decreases, prices increase. As with the selling-side analysis, the economic harm extends both upward and downward through the chain of production and distribution.

Private Antitrust Enforcement in the United States and the European Union   293

MFC

$

S

B C

P1 P2

A D, MRP

0 Q2 FIGURE  12.3  Monopsony

Q1

Q

and Price Fixing by Buyers

In a manner analogous to that with the selling-side models, the graph illustrates the shortcoming of what, in this context, would be an undercharge measure of damages. The difference in the price paid per unit after collusion is P2 − P1. The quantity sold is Q2. Thus the measure of damages is P2 − P1(Q2), or the area P1P2AB. Depending on the status in direct sellers, this amount may be allocated among plaintiffs at two or more levels. The actual harm resulting from monopsony power is the area P1P2CA. This measure accounts for the impact of sellers who decrease sales as a result of the lower price. The inconsistency of the undercharge measure with compensation is obvious. The inconsistency of the same measure if the goal is deterrence, in the absence of punitive damages in some form, is similar to that found in the seller-side models examined above. These observations lead to the questions of standing and injury and of deterrence versus compensation. The easy matter is that of injury. Under US standards firms at all levels suffer the type of damage the antitrust laws were designed to prevent. Under a possible broader EU standard, the same is true. The more critical question is that of antitrust standing. In the United States it is clear that direct sellers to the collusive monopsony will have standing. Indirect sellers almost certainly do not. Whether customers of the direct purchaser—those faced with lower output and higher prices—have standing has not been determined. This combination of factors will invite substantial private action depending on the importance of the relationship of direct sellers to the colluding buyers and, to some extent, the degree to which the undercharge may be passed on to suppliers of the potential plaintiff. The critical factor offsetting a possible reluctance to pursue private actions is the availability of treble damages. The analysis demonstrates the quandary faced by the EU in its effort to encourage private antitrust actions while focusing on compensation and in the absence of punitive damages. As with an overcharge measure of damages with a pass-on defense, plaintiffs are not as likely to bring private actions. This is true even in the presence of even the broadest notion of antitrust standing. The outcome is the potential for dividing up the

294   Jeffrey L. Harrison undercharge into many small pieces, some of which will not be seen as worth the risks of litigation. Again, the approach amounts to one in which unlawfully obtained gains are disgorged and returned to those from whom they were taken. Switching to a lost-profits measure of damages, while still maintaining a very broad notion of standing, addresses the compensation matter but does not seem likely to bring a powerful deterrent element to the EU system. The calculation of lost profits would remain a risky matter, with or without punitive damages. Plus, at some levels the amount of lost profit may be very small. Together these factors are not consistent with enough private action to generate a robust deterrent effect.

12.3.4.  Model 4: Lost Profit Cases So far the analysis has focused on price fixing by sellers and buyers and the resulting over- or undercharge measure of damages. This is not the norm in antitrust. Competitors may not agree on price but on other terms of the exchange. They may also agree to divide a market. When a single firm engages in monopolizing conduct, it may simply make it more costly for a competitor to remain in the market. Similarly, a practice of predatory pricing means competing firms experience lower short-term and possibly long-term profits. In these instances, damage determination is not as simple as calculating the difference between a pre- and postcollusion price. The varied consequences can be appreciated by examining a model of exclusionary conduct. For example, competitors at one level of competition may conspire to deny competitors at the same level access to either suppliers or customers. Figure 12.4 illustrates the variety of competitive harms that may occur (Blair and Page 1996). Prior to exclusionary conduct, the supply in the market is Sf, which is the sum of Sc (the supply of colluding firms) and St (the supply of targeted firms). Demand for the product is D. The equilibrium price is P1 and the equilibrium quantity is Q1. A firm or firms represented by St could be excluded if the dominant firms agree to boycott any suppliers selling to St or boycott customers who would otherwise purchase from the targeted firms. Exclusion reduces industry supply to Sc. The new equilibrium price and quantity are P2 and Q2 respectively. Obviously the firms composing St are injured. The damages would be the lost profits. One path for determining damages would be to determine a “but for” market share and the profit resulting from the associated sales. There are also upstream and downstream harms. Downstream purchasers, who continue to purchase, pay a higher price. Those now priced out of the market are also harmed. Since the industry is now smaller, the demand for upstream inputs has declined and suppliers of those inputs are damaged, as, perhaps, are suppliers to those suppliers. Consequently, under both EU and US policy the issue of eligibility is unlikely to turn on the question of antitrust injury. When standing is addressed under the EU’s broad commitment to compensation, all firms suffering lost profits qualify. This would almost certainly include both downstream firms and upstream firms that are able to demonstrate actual losses. The overall

Private Antitrust Enforcement in the United States and the European Union   295

St

P

Sc

Sf P2 P1

Q2

Q

Q1

FIGURE  12.4  Exclusionary

Conduct

impact of a liberal standing policy is not clear though. Lost profits are difficult to prove under most circumstances and are even more of a problem when a firm is separated from direct victims by one or more layers. In a single-damage environment in which it is difficult to aggregate claims, the probability of robust private enforcement efforts when the measure of damages is lost profits is similarly decreased. Under US law, the antitrust standing of all downstream parties is not as certain. For example, even if those purchasing from the colluding firms now sell less of their output at a higher price, they may be regarded as indirectly affected and not the best suited to represent the interests of those harmed. Plus, the demonstration of actual loss becomes more difficult. Given the higher standing threshold in the United States it is unlikely that eligibility will extend beyond those firms excluded and perhaps to suppliers to those firms who are, in effect, used as the leverage by which the excluded firms are harmed. The question, as addressed above, is whether the possibly of an expansive approach to antitrust standing in the EU is enough to offset the difficulty of collecting punitive damages. The question itself illustrates the importance of treble damages in the United States. Whatever “errors,” limitations, and exclusions exist under US law with respect to standing, they are largely irrelevant if treble damage actions provide the hoped-for deterrent effect.

12.5. Conclusions A comparison of US and EU standards with respect to antitrust injury and antitrust standing is only meaningful in the context of the goals of the two systems of antitrust. Even in that broader context the process is difficult and broad conclusions risky because the relevance of private antitrust actions in the EU has yet to be determined. Although there are expressions in favor of a substantial expansion, there are significant barriers.

296   Jeffrey L. Harrison Although each system of private enforcement is concerned with both deterrence and compensation, when faced with choosing, the United States favors a deterrence goal while the EU favors a compensation goal. There is probably little difference in the antitrust injury requirement between the jurisdictions. On the other hand, the EU appears to have a more liberal standard with respect to standing. This is, in part, a result of its emphasis on compensation. Ironically, the United States may, at least in the context of private actions, do a better job with respect to both deterrence and compensation. The key here is not standing and injury but the ability to aggregate claims and the availability of punitive damages. To understand the compensation point it is important to note that compensation is not necessary if deterrence is successful. Thus, in the United States, with a more powerful system of private enforcement, there is what might be called the ex ante effect of ensuring the harm does not occur in the first place. In effect, treble damages have two results. They create an important incentive for private parties to initiate private actions. Second, that threat and the deterrence it provides can mean there are fewer instances in which compensation would be called for. The EU appears to occupy what might be called a no man’s land with respect to private enforcement. As it currently stands, private actions are unlikely to have an important compensatory or deterrent effect. The reasons are clear. Many of the remedies in the EU, as in the United States, have a disgorgement quality. First, the worst-case outcome for colluding firms may be simply to return the unlawful gains. If the probability of detection and liability is less than one, it is in the interest of firms to collude. Second, even when damages are calculated as lost profits, those deserving of compensation may react to small recoveries. This tendency may be offset by increased criminalization. Two final points should be noted. First, the likely failure of the EU to achieve either the compensation or deterrent goal is not a function of its injury and standing policies. In fact, if anything, EU policy with respect to standing and injury is more inviting to private actions than in the United States. Second, all of this analysis is independent of public enforcement. Aggressive public enforcement can achieve both the ends of deterrence which then renders compensation less pressing. Nevertheless, it is doubtful that even the most inviting injury and standing requirements can offset a policy that excludes punitive damages. In effect, as things stand now in the EU, the next few years can be viewed as an experiment with a number of variables interacting to determine which are the most important in encouraging private enforcement.

References Blair, Roger D., and Jeffrey L.  Harrison. 2010. Monopsony in Law and Economics. New York: Cambridge University Press. Blair, Roger D., and William H. Page. 1996. The Economics of Defining Antitrust Injury and Standing. Managerial and Decision Economics 17:127–42.

Private Antitrust Enforcement in the United States and the European Union   297

Cook, Christopher J. 2008. Private Enforcement of E.U. Competition Law in Member State Courts: Experience to Date and Path Ahead. Competition Policy International 4:3–64. European Commission. 2005. Green Paper on Damages for Breach of the EC Antitrust Rules. COM 672 final. European Commission. 2008. White Paper on Damage Claims for Breach of EC Antitrust Rules. COM 165 final. Harrison, Jeffrey L. 1980. Lost Profits as the Measure of Damages in Price Enhancement Cases. Minnesota Law Review 64: 751–88. Milutinovic, Veljko. 2010. The “Right to Damages” under E.U. Competition Law. Amsterdam: Wolters Kluwer. Rizzuto, Francesco. 2010. The Private Enforcement of European Competition Law: What Next? Global Competition Litigation Review 2010:57–68. Romain, David, and Ingrid Gubbay. 2011. Competition Claims: Uneven Progress, Still. European Antitrust Review 2011:47–52. Rouhette, Thomas. 2007. The Availability of Punitive Damages in Europe: Growing Trend or Nonexistent Concept? Defense Counsel Journal 74:320–40. Sittenreich, Marc A. 2010. The Rocky Path for Public Directors General: Procedure, Politics, and the Uncertain Future of EU Damages Actions. Fordham Law Review 78:2703–50.

Cases Cited Brunswick Corp. v. Pueblo Bowl-O-Mat, 429 U.S. 488 (1977). Illinois Brick Co. v. Illinois, 431 U. S. 720 (1977). Case C-453-/99, Courage v. Crehan [2001] ECR I-6297.

CHAPTER 13

F R E E D OM TO T R A DE A N D T H E C OM P E T I T I V E P R O C E S S AARON EDLIN AND JOSEPH FARRELL

[B]‌y the omission of any direct prohibition against monopoly in the concrete [the Sherman Act] indicates a consciousness that the freedom of the individual right to contract when not unduly or improperly exercised was the most efficient means for the prevention of monopoly . . . [T]‌he right to freely contract was the means by which monopoly would be inevitably prevented. Justice White in Standard Oil (1911),1 emphases added. The freedom to switch suppliers lies close to the heart of the competitive process that the antitrust laws seek to encourage. Justice Breyer in Discon (1998).2

13.1. Introduction Antitrust analysis considers both process and outcomes. That is, it asks both whether an activity or agreement interferes with the competitive process and whether it leads to outcomes that are inefficient and/or injure consumers. The mix of these two strands of analysis varies. For example, horizontal merger analysis is almost entirely outcome-focused,3 while cartel enforcement scarcely addresses outcomes at all;4 form-based and effects-based analyses continue to contest other areas of antitrust. 1 

Standard Oil v. United States, 221 U.S. 1 (1911). NYNEX v. Discon, 525 U.S. 128, 137 (1998). 3  See, e.g., Horizontal Merger Guidelines, U.S. Department of Justice and Federal Trade Commission, August 19, 2010. 4  See, e.g., United States v. Socony, 310 U.S. 150 (1940). 2 

Freedom to Trade and the Competitive Process   299

Recent decades have seen increasing weight placed on direct analysis of outcomes. Some commentators even suggest that whether an act is anticompetitive turns by definition on whether its outcome will be bad. Courts, on the other hand, still frequently stress the competitive process, even when they may not be particularly clear about what that process is.5 We take a step toward understanding what the competitive process entails, because we side with the courts in thinking that an analysis of impacts on the competitive process should remain central to antitrust. A very intuitive and robust benefit of competition is that if firm A  is greedy or incompetent, consumers can trade with firm B instead, in a way that makes both consumers and firm B better off. Firm A would of course love to thwart the formation of this improving coalition, and might sometimes be able to do so either in concert with B or on its own, as we illustrate below. These observations suggest a perspective that in large part the competitive process is the process of sellers and buyers forming improving coalitions. We explore below the idea that the competitive process centrally involves the freedom to strike better deals. From this perspective much—though by no means all—of antitrust can be seen as prohibiting firms’ attempts to restrain or thwart improving trade between their rivals and customers, echoing the Sherman Act’s ban on agreements in “restraint of trade.” In this way antitrust protects B’s and consumers’ freedom to trade to their mutual betterment.6 We also illustrate how some antitrust controversies arise when firm A  itself is an essential participant in the improving coalition.

13.2.  Outcomes, Process, and the Core An entirely outcome-focused view would define restraints of trade as practices or agreements that lower economic (consumer or total) surplus, perhaps measured by quantity traded, in a market. Some mainstream antitrust commentary might approach this view. Posner (2001, p. 22) views the “completed act” of violating section 1 as “an actual restriction of output,” while Posner (1976, p. 53) argues that “antitrust policy is to be shaped by the monopoly problem” after explaining that monopolies lower output and raise price, relative to the perfectly competitive level or the level that would otherwise obtain. Similarly, Hovenkamp (2006, ch. 1)  frames his “legal and economic structure” using outcome measures. Bork (1978, p. 51) states that “[t]‌he only legitimate goal of American

5  Similarly, the central notion of “competition on the merits” is hard to make precise in general. See for instance OECD (2006) and (2005). 6  Antitrust prohibits certain things (all statutes do), but it is a shallow paradox to observe that laws protecting freedom can take the form of prohibitions. (For instance, laws against kidnapping prohibit certain conduct but clearly protect freedom.)

300   Aaron Edlin and Joseph Farrell antitrust law is the maximization of consumer welfare”;7 Salop (2010) also focuses on welfare outcomes (though he stresses that Bork used an odd definition of consumer welfare). The freedom-to-trade perspective, in contrast, stresses process: the freedom of buyers and sellers to change their trading partners whenever that is mutually beneficial. The aspect of the competitive process that we study here is buyers and sellers exercising this freedom and forming improving coalitions, finding mutual gain in bypassing greedy or incompetent incumbents. In a highly competitive market a seller who does not give its customers good deals will find that rivals offer better deals to attract these customers. The process of firms fighting over customers by offering them better and better deals makes consumers better off and also normally increases economic efficiency. Until the outcome reaches what economists call the core, an improving coalition of buyers and sellers is always possible and the improving/competitive process will continue. In fact the core is defined as the set of outcomes or payoffs relative to which no coalition of people could improve all of their payoffs by trading among the coalition (see, e.g., Telser 1978). Because it is flexible as to institutions and modes of trading, the core is more meaningful than perfectly competitive equilibrium in a broad range of markets of interest to antitrust. As Edgeworth (1881, p. 31) wrote, [The core] is attained when the existing contracts can neither be varied without recontract with the consent of the existing parties, nor by recontract within the field of competition. The advantage of this general method is that it is applicable to the particular cases of imperfect competition; where the conceptions of demand and supply at a price are no longer appropriate.

As with perfect competition, outcomes in the core are efficient, making the core a reasonable ideal to pursue if we judge by equilibrium outcomes.8 In the last half-century, antitrust commentators have come to think of the outcome ideal or benchmark as a perfectly competitive equilibrium in which firms are price takers. This framing naturally inspires them to focus on acquisitions and maintenance of market power (often defined as the power to raise price above competitive levels) as a problematic outcome, but does not provide much process guidance of what acts are anticompetitive. Grinnell famously and unhelpfully identifies “willful acquisition” as its process prong.9 On the other hand, if the outcome ideal is the core, as we consider here, then the process focus is relatively clear: whatever restrains, thwarts, or blocks the formation of improving coalitions constitutes a process violation. 7 

By consumer welfare Bork means (confusingly) what is more often called total welfare; but our point here is that he asserts that antitrust is outcome-focused. 8  However, the (usual) proof that outcomes in the core are (Pareto) efficient relies on the prospect of forming the “grand” improving coalition containing everyone in the economy. If such coalitions are very hard to form, it becomes an appealing hypothesis rather than a theorem that the formation of improving coalitions tends toward efficiency. 9  United States v. Grinnell Corp., 384 U.S. 563 (1966).

Freedom to Trade and the Competitive Process   301

Our suggestion is that antitrust protects the formation of improving coalitions, a process that might converge toward the core. Lester Telser (1978) also considered connections between the core and antitrust. He argued that when the core is empty, as it might be in ocean shipping, collusion is justified and perhaps efficient. We are skeptical of the efficiency claim, and do not think that nonexistence of theoretical equilibrium is a compelling reason to suspend antitrust, but these issues are not our focus here. We take it for granted that antitrust seeks to protect the competitive process and explore the implications of characterizing this process as the freedom to form improving coalitions and trade within them. Some may fear that a focus on the competitive process rather than economic analysis of outcomes is a throwback to the (bad) old days of antitrust’s first half-century. Indeed, we think the spirit of freedom to trade animated much early antitrust discussion. Fear of a throwback is, however, needless. We are not endorsing decisions like Dr. Miles;10 indeed, as we discuss below, retail price maintenance (RPM) does not strictly violate freedom to trade. Second, rightly or wrongly, process concerns have remained important in antitrust, so it is important to work toward a better understanding and a coherent definition of the competitive process.

13.3.  Freedom to Trade and Horizontal Agreements under Section 1 To illustrate this view of the competitive process as freedom to trade, suppose that suppliers A and B collude and agree to charge a high price H. If their contract were legally binding, it would prohibit either from trading with a buyer at a lower price L < H .11 In the case of a buyer who pays H to buy from A, the contract thwarts the {B, buyer} coalition that would otherwise improve on the status quo for both of its members. The contract thus restrains the freedom of B and the buyer to trade.12 Typically, of course, a cartel does not contemplate court enforcement; rather, A threatens that if B attracts buyers by offering a better deal, A will punish B through marketplace responses. This threat discourages the otherwise improving coalition. B is not entirely

10 

Dr. Miles Medical Co. v. John D. Park & Sons Co., 220 U.S. 373 (1911). Even if A and B are bound to their collusion by honor alone, an agreement not to deal with one another’s customers restrains trade and discourages each from forming an improving coalition with the other’s customers. 12  In the economics of the core, the fact that an improving coalition of B and the buyer to trade at a lower price would hurt A (relative to a high-priced status quo) does not call for pause in calling it an improving coalition. As a result, not every improving coalition need improve total economic surplus or efficiency. This fact is reflected in competition policy economics, for instance, by the observation (Mankiw and Whinston 1986) that profitable entry that benefits consumers may easily in itself lower economic welfare. The theorem/fact that the eventual resting place of a process of free formation of improving coalitions, the core, is economically efficient is thus not because every such step in the process necessarily increases efficiency. 11 

302   Aaron Edlin and Joseph Farrell “free” to cut price so as to trade with the buyer, because A will punish him for doing so. Thus if we view the competitive process as centrally involving freedom to trade, cartels are a direct core violation of Section 1, whatever their likely outcome might be. We suspect this may “explain” per se treatment of cartels in a more fundamental way than the courts’ mantra that “experience” shows that cartels almost always lead to bad outcomes. Readers may ask why {A, B} isn’t an improving coalition relative to the competitive outcome:  why doesn’t freedom to trade fatally include freedom to collude? Put differently, why isn’t a cartel itself an improving coalition? In the language of the core, {A, B} is not an improving coalition because a coalition with only sellers yields no surplus: worthwhile coalitions must include the trading partners, because without including customers {A, B} can achieve nothing. And the coalition {A, B, customers} with trade at H does not “improve on” the competitive outcome because customers are worse off. 13 In the economic logic of the core, a cartel is prevented by the need to attract customers to the trading coalition. In antitrust, the fact that customers are not at the bargaining table justifies the legal restriction on cartel agreements. Maricopa14 may illustrate the view that restraining rivals’ trade is illegal even in the face of a natural intuition that the effects might not be bad. The Court decided it was per se illegal for doctors to agree on maximum prices—that is, to agree not to charge more than certain prices. In an outcome-focused approach, the Court’s per se condemnation would be puzzling. After all, maximum prices would seem likely to expand output (if doctors have market power or information rents), suggesting at least a need for more analysis. On the other hand, the doctors were nakedly restraining one another’s freedom to trade, although in ways that one might not have thought would be particularly tempting; for instance, each doctor was agreeing not to compete away business from the others by offering very high-quality service at high prices. A freedom-to-trade approach appears to explain the result in Maricopa better than an outcome approach. However, we do not suggest that outcomes can be ignored. To do so might risk returning antitrust to pre–Chicago Board15 paradoxes such as whether all contracts are illegal restraints or whether all trade restrains other trade. In BMI, the Court declined to apply the per se rule.16 Although the case literally involved fixing a price for the blanket license, each copyright holder remained free to trade on his own with a licensee.17 One could thus analyze the case by saying that the blanket licenses did not limit anyone’s freedom to trade. Alternatively, if one did view collective pricing of the blanket license as limiting freedom to trade, a court might allow the limitation if efficiencies are sufficiently compelling. In Engineers,18 engineers agreed not to quote a price until a customer had engaged a particular engineer. They argued that without this restriction, customers would choose 13 

In principle one should specify the terms on which members of a coalition trade; in some cases we take the liberty of leaving that implicit. 14  Arizona v. Maricopa County Medical Society, 457 U.S. 332 (1982). 15  Chicago Board of Trade v. United States, 246 U.S. 231 (1918) 16  Broadcast Music v. Columbia Broadcasting System, 441 U.S. 1 (1979). 17  This right to bypass ASCAP was the product of a prior antitrust settlement. 18  National Society of Professional Engineers v. United States, 435 U.S. 679 (1978)

Freedom to Trade and the Competitive Process   303

low bidders, engineers would cut corners, and bridges would fall down. That defense goes to efficiencies, not to whether the horizontal agreement limits freedom to trade. A court might concern itself exclusively with the competitive process, and strike down the agreement; or it might evaluate the efficiency defense. In actuality, the Supreme Court took a confusing path somewhere in between. It pointed out that consumers can trade off quality against price, potentially the germ of a real efficiency analysis, but did not engage with an implied efficiency theory involving consumer irrationality. The short shrift given to the engineers’ defense seems more compatible with a focus on restraints of the competitive process than with a focus on outcomes. Illustrating some of these tensions and interweavings of process and outcome, consider an art dealer D who agrees to sell a painting to buyer A in a week for $500, but leaves it hanging in his gallery. Buyer B arrives midweek and offers $750 for the painting. The forward contract between D and A prevents D from selling to B at the higher price. Does it “restrain” the freedom of B and D to trade to their mutual benefit and thwart an improving coalition {D, B} ? Had D and A entered a spot contract instead of a forward one, and had A taken the painting home with him, then when B arrived, D would plainly have no painting, so {D, B} would not be an improving coalition (let alone one that is thwarted). Inspired by this observation, one could analyze the forward sale by saying that {D, B} is not an improving coalition because the painting is already sold even though it still hangs in the gallery. Here, and in many situations, this may be the soundest answer. Another tack (less compelling in this instance but perhaps helpful in others) would say that the forward sale did limit {D, B} ’s freedom to trade, but might yet not violate Section 1. First, the forward sale may well have helped D and A to trade in the first place, so it might on balance increase freedom to trade. This approach would try to balance pro- and anticompetitive effects in terms of the competitive process, not directly balance efficiencies. Second, even if forward sales hinder the competitive process, one might allow them if they greatly improve efficiency. This approach would try more directly to balance inefficient outcomes against anticompetitive restraints.

13.4.  Freedom to Trade and Monopolization As the quotation from Justice White suggests, Section 2 tries to ensure that monopolies do not thwart others’ incentives to trade around a greedy or incompetent monopoly. This might suggest glossing Section 2 as A monopoly may not block improving trade between customers and rivals who would offer customers a better deal. How, if at all, can a monopoly do that? Raising rivals’ costs can do it: consider sabotaging a key rival’s factory, or threatening customers who deal with rivals.19 Less obviously, 19  As this sentence indicates, otherwise-improving trade between a rival and a customer could be discouraged from either end. Blowing up a rival’s factory is a popular hypothetical in antitrust

304   Aaron Edlin and Joseph Farrell we show below how a firm might discourage improving coalitions using pricing—not price levels but pricing patterns. A low price may exclude, but only by its merit: that is, only by offering customers a genuinely good deal. But a pricing pattern, it turns out, can hinder the competitive process by forcing rivals to compete against a price lower than the price that customers pay. To see how this can happen, we first note two simple models in which it fails. First, suppose that n firms each set a price simultaneously. Each firm’s price to consumers is also the price against which its rivals compete. If a rival offers a better deal, it gets the business. Second, in the contestability model of Baumol, Panzar, and Willig (1982), an incumbent sets a price, and potential entrants make entry decisions knowing that (for long enough at least to recoup sunk costs of entry) they can compete against that price. Again, if a rival offers a better deal than what customers pay the incumbent, the rival wins the business. Reflecting the powerful intuition of such models of simple price level, or more generally of simple competitive offers, Posner (2001, p. 34) wrote: The framers of the Sherman Act were concerned with the “trust” problem, but what they conceived that problem to be is obscure, and indeed contradictory. They seem to have been concerned with low prices harmful to small-business competitors of the trusts . . . as well as with high prices harmful to consumers. . . . Protecting competitors from low prices and consumers from high prices are incompatible objectives, with a few rare exceptions, such as when a monopolist prices below cost in an effort to intimidate a potential entrant.

Indeed, if competition involved simply naming one price, that price could not simultaneously be too high and too low. But the contradiction disappears if the price confronting competitors is not the price that consumers pay. For instance, monopoly M charges a high price H, but would cut its price upon entry to L. A potential rival R who knows this and who cannot compete successfully against L will not enter; with no entry, customers pay H. If R could profitably attract buyers away from H but not away from L, then R and the buyers are not “free to trade.” Specifically, relative to the situation in which M charges H, there would be an improving coalition of R and customers, but its formation is thwarted by M’s threat to lower price to L. Is M “competing on the merits,” the touchstone of competitive behavior since United Shoe?20 The “merit” of M’s offer is inversely measured by the consumer price H, but it “competes” using L. In that sense, when H and L differ, M is not competing on the merits. We say that a firm has moved the goalposts (thus restraining freedom to trade) if its rivals must compete against a price lower than the price its customers pay. commentary: see, for example, Roundtable: Recent Developments in Section 2, Antitrust Magazine, Fall 2003, p. 20. Somewhat similarly, Conwood involved a claim that one firm ruined another’s sales displays. Conwood v. United States Tobacco, 290 F. 3d 768 (2002). 20 

United States v. United Shoe Machinery Corporation, 110 F. Supp. 295.

Freedom to Trade and the Competitive Process   305

Moving the goalposts restrains R and customers from trading to their mutual betterment, relative to M’s pre-entry price of H. In more concrete effects terms, allowing monopolies to move the goalposts chills entry relative to the contestability benchmark, and may well have bad consequences (see Edlin 2002 and 2012). That said, setting policy to prevent it raises very real difficulties, both practical and conceptual. At the conceptual level, M’s postentry price of L represents, ex post, an improving coalition of M and customers, relative to R’s offer. Thus it has aspects both of blocking and of improving. Antitrust has long wrestled with the concept of exclusionary pricing, often formulated (stressing the paradox) as whether prices can be “too low.” That formulation suggests a focus on the level of price. But, as suggested by the discussion above, an incumbent’s pricing pattern can restrain or tax potential trade by others. Here the low price is charged only out of equilibrium: consumers do not benefit from it.21 In this view, neither high pricing nor low pricing is itself a problem, but a pattern could be. Cheating is moving the goalposts, not putting them in a “bad” spot. Note that nothing in this core logic turns on the relationship between the incumbent’s price and its costs.

13.5.  Bundling and Loyalty Pricing Suppose that buyers buy good 1 from firm A, and may buy good 2 either from firm A or its rival firm B. If A worsens the terms of trade in good 1 when the buyer buys good 2 from B, the worsening creates a tax on trade between firm B and the buyer. When the monopolist A leaves the buyer significant surplus in good 1, relative to likely quantities and prices of good 2 (and it’s fair to ask why A would do so), the tax may exclude trade between B and the buyer even while A charges a high marginal price for good 2. In this way, much as in our monopolization discussion above, B may have to compete against a low price (net of the tax) even while the buyer has to pay a high price for incremental units of good 2 (including the tax if it buys from B) (see, e.g., Inderst and Shaffer 2010 and Farrell, Pappalardo, and Shelanski 2010). These concerns seem to align with the general notion of moving the goalposts and finding a pricing pattern that hinders trade between the buyer and B by means other than simply charging low prices to the buyer. On the other hand, the tax is avoidable if the buyer ceases to buy good 1 from A, so it is limited by the buyer’s surplus in good 1. Moreover, B and the buyer can trade between themselves if they are willing to say goodbye to A, so A’s offers do not thwart that potential improving coalition, and thus do not violate plain freedom-to-trade. Thus, any concern about moving goalposts is in at least some ways broader than a concern about restraints on freedom to trade. When a practice harmfully moves the goalposts, but does not violate freedom to trade, the ban on “unfair methods of 21  Of course, a court would only observe a price that occurs. A lawsuit erupts only if an entrant mistakenly enters anyway, and then is driven out. But the fact that the case is brought only ex post does not imply that the antitrust analysis should focus on the ex post effects.

306   Aaron Edlin and Joseph Farrell competition” in Section 5 of the FTC Act might be a better fit than the ban on restraints of trade in the Sherman Act.22 Regardless of what act is involved, however, we think that a significant set of antitrust controversies arise in cases where those criteria differ: where a rival and buyer(s) would like to bypass a dominant firm in one line of business but where the buyers are not willing to eschew all trade with the dominant firm. To put it another way, consumers and firm B would like to form coalitions that improve their payoffs but that (at least in the short run) require firm A to continue to participate, and A declines to do so. The next section further illustrates how controversies can arise in such cases.

13.6.  Vertical Agreements The process-based freedom-to-trade criterion need not return us to pre-Chicago antagonism to vertical agreements. Some vertical agreements may restrain freedom to trade, but by no means all. Consider for example retail price maintenance (“RPM”), involving a manufacturer M, retailer R, and consumer C, and a retail price r at which R agrees with M that it will sell to C. The RPM agreement means that R is then not free to charge a higher or lower price than r for M’s product. Does this restrain the freedom of {R , C} to trade? The tone of Dr. Miles,23 overruled by Leegin,24 suggested so. But, in fact, freedom to form improving coalitions does not imply the Dr. Miles result and is quite consistent with Leegin. After all, {R , C} is not an improving coalition: by themselves the retailer and customer cannot do better because without M, the retailer has no goods to sell or consume.25 Consider now an exclusive distributorship. M and R agree that M will supply R but not a would-be competing retailer S. Is freedom to trade restrained? M could shift its business entirely to S, so if {M, S, C} is an improving coalition because S is a better (or less greedy) retailer than R, then that coalition is free to form. This argument suggests that freedom to trade is not restrained in the sense above. As with bundling, the thornier issue is whether there is an improving coalition that includes R, here {M, R , S, C} , and whether R’s (or jointly M’s and R’s) refusal to negotiate towards such an improving coalition “should” count as restraining freedom of trade.

22  It has long been asserted that the ban of unfair methods of competition under FTC Section 5 is broader than the bans of the Sherman Act. See, e.g., Federal Trade Commission v. Motion Picture Advertising Service Co., 344 U.S. 392 (1953). Efforts by the FTC to take advantage of this extra power, however, have thus far been thwarted. See, e.g., Kovacic and Winerman (2010) and du Pont v. Federal Trade Commission, 729 F.2d 128 (2d Cir. 1984). One linguistic interpretation of restraint of trade might limit it to restraints on strictly improving coalitions, while harmfully moving the goalposts might be an unfair method of competition even if nothing strictly restrains improving coalitions. 23  Dr. Miles Medical Co. v. John D. Park & Sons Co., 220 U.W. 373 (1911). 24  Leegin Creative Leather Products v. PSKS, Inc., 551 U.S. 877. 25  If R is also a manufacturer, there is a horizontal problem, which is not our focus here, so we assume that R is purely a retailer.

Freedom to Trade and the Competitive Process   307

13.7.  Freedom to Trade and the Goals of Antitrust Commentators as diverse as Bork (1978), Posner (2001), Salop (2010), and Hovenkamp (2006) have argued that antitrust should seek to protect total or consumer welfare. Clearly it does so, but equally clearly it is not so simple. Antitrust doesn’t just fail, but explicitly doesn’t try, to protect total or consumer welfare against certain obvious threats, notably the exercise of legitimately acquired monopoly power by simply raising price. While trying to address those threats would certainly involve administrative difficulties and could affect ex ante incentives, antitrust’s refusal to try seems proud and categorical rather than regretful and pragmatic. We think that a straightforward account of why simple monopoly pricing is legal is found in the quotation from Standard Oil with which we began. Antitrust (mainly) seeks to protect against monopoly abuse not by barring monopoly outcomes, but by safeguarding the competitive process, which we suggest involves freedom to trade and the formation of improving coalitions. Simply charging monopoly prices harms welfare, but does not limit rivals’ freedom to trade, Justice White’s “surest protection against monopoly.” Antitrust protects the potential beneficial trades between competitors and consumers. Since both consumers and competitors gain from such trade, this view can explain why both consumers and thwarted competitors have antitrust rights, even though antitrust protects “competition and not competitors.” Consumers are not protected from all high prices, but only from those that a competitor would be happy to beat but for some thwarting action; this explains why a pure monopoly does not violate the law simply by charging high prices. Competitors meanwhile are not protected from actual everyday low prices, but only from tactics such as moving the goalposts that block them from giving customers a better deal than a monopoly does. Antitrust does not ban everything that reduces consumer welfare. Instead Congress banned restraint of trade and monopolization, which can be understood as a monopoly’s erection of “unnatural barriers” (restraints) that “restrict free competition,” in the words of Judge Wyzanski in United Shoe.26 We explored the idea that these prohibitions may be unified if the competitive process is understood to include the process of sellers and buyers forming improving coalitions. This contrasts with the ahistorical view that “competition” means anything that improves total or consumer welfare.27

26  27 

U.S. v. United Shoe, 110 F.Supp. 295 (D. Mass. 1953). Aff ’d Per Curiam, 347 U.S. 521 (1954). On such redefinition, see, for instance, Farrell (2006).

308   Aaron Edlin and Joseph Farrell

Acknowledgments This chapter has been long in gestation, but at the time of its release as a working paper Professor Farrell was on leave serving as director of the Bureau of Economics, Federal Trade Commission (he no longer holds that position). The opinions in this chapter are those of the authors as individuals and are not intended to represent those of the Federal Trade Commission or any commissioner. We thank David Gilo, Fred Kahn, Hanno Kaiser, Louis Kaplow, Douglas Melamed, Ariel Porat, Ed Rock, Carl Shapiro, Lester Telser, and seminar participants at Tel Aviv University for helpful comments; they are not responsible for our analysis or views.

References Baumol, William J., John C. Panzar, and Robert D. Willig. 1982. Contestable Markets and the Theory of Industry Structure. New York: Harcourt Brace Jovanovich. Bork, Robert. 1978. The Antitrust Paradox: A Policy at War with Itself. New York: Basic Books. Edgeworth, F. Y. 1881. Mathematical Psychics. London: C. Kegan Paul. Edlin, Aaron. 2002. Stopping Above-Cost Predatory Pricing. Yale Law Journal 111: 941–91. Edlin, Aaron. 2012. Predatory Pricing. In Einer Elhauge, Research Handbook on the Economics of Antitrust Law. Northampton, MA: Edward Elgar. Farrell, Joseph. 2006. Complexity, Diversity, and Antitrust. Antitrust Bulletin 51: 165–73. Farrell, Joseph, Janis Pappalardo, and Howard Shelanski. 2010. Economics at the FTC: Mergers, Dominant-Firm Conduct, and Consumer Behavior. Review of Industrial Organization. Hildenbrand, Werner. 1974. Core and Equilibria of a Large Economy. Princeton, NJ: Princeton University Press. Hovenkamp, Herbert. 2006. The Antitrust Enterprise. Cambridge, MA:  Harvard University Press. Inderst, Roman, and Greg Shaffer. 2010. Market Share Contracts as Facilitating Practices. Rand Journal of Economics. Kovacic, William E., and Marc Winerman. 2011. Competition Policy and the Application of Section 5 of the Federal Trade Commission Act. Antitrust Law Journal 76: 929. Mankiw, Gregory, and Michael Whinston. 1986. Free Entry and Social Efficiency. Rand Journal of Economics. Organisation for Economic Co-operation and Development (OECD). 2005. “Competition on the Merits.” OECD Policy Roundtables, June. Organisation for Economic Co-operation and Development (OECD). 2006. “Policy Brief: What Is Competition on the Merits?” June. Posner, Richard A. 1976. Antitrust Law:  An Economic Perspective. Chicago:  University of Chicago Press. Posner, Richard A. 2001. Antitrust Law. 2nd ed. Chicago: University of Chicago Press.

Freedom to Trade and the Competitive Process   309

Salop, Steven. 2010. Question:  What Is the Real and Proper Antitrust Welfare Standard? Answer: The True Consumer Welfare Standard. Loyola Consumer Law Review 22: 336–53. Sjostrom, William. 1989. Collusion in Ocean Shipping:  A  Test of Monopoly and Empty Core Models. Journal of Political Economy 97: 1160–79. Telser, Lester G. 1978. Economic Theory and the Core. Chicago: University of Chicago Press.

PA R T I I

M ON OP OLY Structural Considerations

CHAPTER 14

M ON O P O LY A N D D OM I NA N T F I R M S Antitrust Economics and Policy Approaches LAWRENCE J. WHITE

14.1. Introduction THE primary purpose of antitrust—at least, modern antitrust—is to limit the exercise of market power.1 By market power we mean the ability of an enterprise/firm to maintain the price at which it sells its product at a level that is significantly above its average (unit) costs, when “costs” are understood to encompass “opportunity costs” and thus to include a competitive return on the investment that has been made in the enterprise.2 In this chapter we will specifically be focusing on a “monopoly” or a “dominant firm” as the enterprise that can exercise its market power.3

1 

This chapter will have an unavoidable US orientation, since US antitrust policy is what the author knows best. Indeed, I will use the word “antitrust,” rather than the phrase “competition policy”: The former is a more common US usage; the latter is more commonly used outside of the United States. Nevertheless, the basic ideas that are developed in this chapter have widespread applications in the international contexts that this handbook encompasses, although the specific instances discussed in the chapter will be US-oriented. 2  By using this definition, I will try to sidestep what—at least to this author—appears to be an uninteresting discussion as to whether there is a useful distinction between “market power” and “monopoly power.” 3  I will thus leave the discussion of the collective exercise of market power by a group of firms— whether they are explicitly colluding or, in the case of a small number of firms (“oligopoly”), implicitly recognizing their interdependence—to other chapters in this handbook. In addition, most of the concepts that are developed with respect to “monopoly” in this chapter would apply in a similar fashion to an enterprise that acts as a single buyer—a monopsonist—that exercises its market power vis-à-vis the sellers that it faces. But, again, we will leave that discussion to other chapters in this handbook.

314   Lawrence J. White There are at least seven ways that government policy can limit the ability of an enterprise unilaterally to exercise market power:









1. Prevent/stop agreements among firms that restrict competition among them and that thereby allow the individual firms to continue to exercise unilateral market power. An example of such an agreement would be an understanding between two firms that Firm A would refrain from trying to sell to customers in Firm B’s geographic territory (or to Firm B’s particular category of customers), and vice versa. 2. Prevent/stop mergers that would otherwise allow the merged firm to exercise market power (where the separate entities previously did not exercise market power) or that would otherwise allow a firm with market power to enhance that power. 3. Prevent/stop predatory behavior by a firm that can enhance its market power by driving rivals out of the market.4 4. Prevent/stop exclusionary behavior by a firm that can enhance its market power by thereby disadvantaging its rivals; such behavior is often described as “raising rivals’ costs” (see Salop and Scheffman 1983, 1987). 5. Dismember the firm, so as to create multiple competitive entities where there was a single seller before, so that the competitive behavior of the created firms reduces or eliminates the original firm’s market power. This is often described as a “structural” approach to addressing issues of market power. 6. Regulate the firm that possesses market power—for example, by specifying a maximum price that approximates the firm’s average costs and/or limiting the extent to which the firm charges different prices to different customers for essentially the same item (i.e., price discrimination). 7. Replacing the firm with a public enterprise that would charge a price that is closer to the firm’s average costs (on the assumption that the government enterprise’s costs would be the same as the private firm’s costs).5

The first five of these alternatives are part of antitrust policy and will be discussed in this chapter. Discussions of regulation and of government enterprise will be left to other handbooks, although there are chapters in this handbook that address the intersection of antitrust and these other two policies.6 4  Although, in principle, a firm that previously lacks market power might engage in predatory or exclusionary behavior and thereby acquire market power, by far the most common allegations are that a firm that already has market power has engaged in predatory and/or exclusionary actions so as to enhance its market power (or to prevent its erosion). 5  A variant on this approach would be to establish a public enterprise that would compete with the firm. 6  In principle, “antitrust” could be broadly defined to encompass judicial decisions that would cause judges to become the regulators of the prices that firms with market power are allowed to charge. But, for the most part, US antitrust policy has not gone in that direction. One exception might be Section 2 of the Clayton Act, which, after its strengthening in the 1930s, has come to be known as the Robinson-Patman

Monopoly and Dominant Firms   315

This chapter will proceed as follows: First, I will present the standard textbook treatments of monopoly and the dominant firm and discuss the nontrivial issues of what constitutes a real-world monopoly (or dominant firm) and how one would go about identifying it. With that as a foundation, I will then address the five categories of antitrust policy that were listed above. A brief conclusion will follow.

14.2.  Monopoly and the Dominant Firm: Conceptual and Real-World Issues A monopoly is the sole seller of a distinctive product (or service). Note that a monopolist need not be the sole seller of everything in an economy; it only needs to be the sole seller of something that is sufficiently distinctive (and that it is sold in large enough volumes that it is worth noticing).7 There may be partial or imperfect substitutes for the item; but it does need to be distinct. For example, if a single enterprise were the sole producer and seller of all beef in the United States, then it is likely that most observers (except, perhaps, the owners of the firm, and their antitrust lawyers) would consider this enterprise to be a monopoly, even though lamb and pork and chicken, as well as nonmeat foodstuffs, would be considered to be partial substitutes for beef.

14.2.1.  A Simple Monopoly Framework Perhaps the easiest way to conceptualize the ability of a monopoly to exercise market power is to start with a distinctive item that has no substitutes, as in the following: Suppose that there are 10 million potential customers for a distinctive item that will be called a “gadget.” Each customer is willing to buy, at most, a single gadget each week; and each customer has a maximum willingness to pay (each week) for a gadget of $10.00. At a price of even $10.01, all 10 million customers would refuse to buy; at a price of $10.00 or less per gadget, all 10 million customers will each buy one gadget (per week). The “demand curve” for gadgets in this example is portrayed in figure 14.1, where the quantity of gadgets per week is measured on the horizontal axis and dollar amounts (prices and costs) are measured on the vertical axis.

Act. The Robinson-Patman Act attempts to restrict the ability of firms to practice price discrimination. However, enforcement of the Robinson-Patman Act by the federal antitrust enforcement agencies has effectively disappeared; and even private suits under the Act today are relatively rare. 7 

This is an issue to which I will return below.

316   Lawrence J. White $ 11 10

Monopoly profits (transfer of consumers’ surplus from buyers) PM

y

9

Demand for gadgets

8 7 6 5

x PC

MC = AC Unit costs of producing gadgets

4 PM : monopoly price

3

PC : competitive price

2 1 0

2

4

FIGURE  14.1  A  Simple

6

8

10

12

14 Quantity of gadgets per week (in millions)

Comparison of Monopoly and Competition

Suppose that the unit costs (including a normal profit8 on the investment in the firm) of producing gadgets—regardless of the volume produced per week—are $6 per gadget. These are also represented on figure 14.1. If there is a single producer/seller of gadgets that wishes to maximize its profits, then it is clear that the price that will do so is $10/gadget; this is represented in figure 14.1 as PM. The monopolist would sell 10 million gadgets per week; and, since its unit costs are $6/gadget, the monopolist’s (above-normal) profits would be $40 million per week; in figure 14.1 these profits are represented by the rectangle PMyxPC. Any higher price would cause all of the customers to refrain from buying; any lower price would “leave money on the table,” since no additional customers would be induced to buy. At $10/gadget, the monopolist would be “charging what the market would bear.”9 By contrast, if there were multiple firms that each produced identical gadgets10 and that competed on the basis of price, the equilibrium price would be only $6 per gadget (which is represented as PC in figure 14.1). At any higher price, one or more of the competitive firms would find it worthwhile to charge a price that was closer to $6, in the hopes of attracting all of the customers while still charging a price that yielded profits

8  By this I mean the profit on the investment in the firm that could otherwise be earned in a competitive industry. 9  In more technical terms, at a price of $10, the demand relationship experiences infinite elasticity with respect to price at any higher price and zero elasticity at any lower price. 10  Note that a gadget is still distinct from all other goods and services; but, since each producer/seller’s gadget is identical to all other producer/sellers’ gadgets, they are all selling “commodity” gadgets.

Monopoly and Dominant Firms   317

that were above those that were already embedded in the unit costs; any price that was lower than $6 per gadget would not be worthwhile, since a seller would not thereby be covering its full costs (again, including full opportunity costs), unless the seller hoped that it could thereby permanently drive out its rivals and emerge as an unchallenged monopolist (which could charge $10 per gadget, etc.); so long as it is easy for firms to enter and exit the gadget industry, this last expectation would be unlikely to be held.11 It is clear, then, that the monopolist’s profits have come at the expense of consumers. If competition prevails, consumers will be able to buy at a price of $6 the gadgets that they value at $10, and they enjoy a “consumer’s surplus” of $4/gadget, which aggregated across all consumers would be represented by the rectangle PMyxPC in figure 14.1; if monopoly prevails, consumers pay the full $10/gadget, and the consumers’ surplus that they enjoyed under competition is transferred to the monopolist and is the source of the latter’s profits. Further, it is clear that if consumers valued gadgets at $12 each, the monopolist’s profits would be yet higher; if consumers’ valuation of gadgets were only $8, the monopolist’s profits would be lower; but at either valuation, the competitive price would still be $6/gadget. Thus, while the monopolist is able to charge “what the market will bear” (i.e., the maximum that consumers are willing to pay), the competitive price will be anchored at the competitive industry’s costs (so long as the supply of gadgets can be expanded at the constant unit costs of $6/gadget).

14.2.2.  The “Lerner Index” One convenient way of measuring and representing the market power of an enterprise is provided by the so-called Lerner Index,12 which is written as L = (P − MC ) / P, (14.1)



where L is the indicator of market power, P is the price at which the firm sells its output, and MC is the marginal cost of the firm for the volume of output that the firm is selling.13 For the monopolist, L will always be a positive number that approaches 1.0 as P gets larger relative to MC . In the example above, when the monopolist charges $10/ gadget, L = 0.67; whereas, for the competitive firms discussed above, L = 0. 11 

I will discuss below the issues that surround “predatory pricing,” where the claim is that a firm has deliberately tried to drive rivals out of the market. 12  Since Lerner (1934) was the first to popularize this formulation of market power, this is often called the Lerner Index; for some historical perspective, see Elzinga and Mills (2011). Lerner also popularized the relationship by showing that L = ( P − MC ) / P = − 1/ ε, where ε is the own-price elasticity of demand for the firm’s product. This last relationship can be derived from the first-order condition for a firm to maximize its profits. It is worth noting that Lerner’s derivation was apparently preceded by a similar derivation of this relationship by Amoroso (1930); see Keppler (1994) and Giocoli (2012). 13  In the numerical example above, the firm’s marginal costs are equal to its average or unit costs.

318   Lawrence J. White

14.2.3.  A Somewhat More Complex Monopoly Example Let us now suppose that the demand structure for gadgets is more complex than the simple example above. Suppose instead that various buyers have different willingnesses to pay for their single gadget per week and/or some buyers may be willing to buy larger numbers of gadgets per week when the price of gadgets is lower. Thus, at higher prices fewer gadgets would be bought, while at lower prices more gadgets would be bought. A linear representation of this “sloped” demand relationship (designated as D) is provided in figure 14.2.14 Again, we will assume that the costs of producing gadgets involve constant unit costs, regardless of volume. The monopolist’s problem is now more complex: There is no single price that represents “what the market will bear.” If the monopolist again seeks to maximize its profits and can charge only a single price to its customers—that is, it cannot practice price discrimination by charging a high price to some of its customers (who have a high willingness to pay for gadgets) while simultaneously charging a lower price to other customers (who have a lower willingness to pay)—then it faces a trade-off: The monopolist could sell gadgets at a very high price (and have high per-unit margins), but would sell comparatively few per week; or the monopolist could sell gadgets at a comparatively low price and sell many more per week, but its margins would be slim. The quantitative solution to the monopolist’s problem—what price to charge (or, equivalently, what quantity to sell), so as to maximize its profits—requires the use of simple calculus: If the monopolist’s profits are represented by π, its price by P, the quantity produced/sold by Q, and its costs of producing the requisite gadgets as C, then

π = PxQ − C; (14.2)

and taking the derivative of this equation with respect to the quantity to be sold (or with respect to the price to sell it at) yields the result that the monopolist should sell a quantity of gadgets each week such that the marginal revenue (MR ) from selling a slightly larger (or slightly smaller) quantity is equal to the marginal cost (MC ) of producing that quantity; or MR = MC. (14.3)



This outcome is shown geometrically in figure 14.2, where Q M represents the quantity at which MR = MC, and PM represents the price at which that quantity can be sold. The monopolist’s (maximized) profits can again be represented by the rectangle PM yxPC .

14 

This is the standard diagrammatic representation of monopoly, which appears in virtually all microeconomics textbooks.

Monopoly and Dominant Firms   319

$

Monopoly profits (transfer of consumers’ surplus from buyers) y Deadweight loss of consumers’ surplus because of monopoly

PM

x

PC

z

MC=AC

MR

0

QM

D

QC

Quantity of gadgets per week

PM : monopoly price QM : monopoly quantity that is produced/sold PC : competitive price QC : competitive quantity that is produced/sold FIGURE  14.2  A  More

Complex Comparison of Monopoly and Competition

Further, it is straightforward to demonstrate that an implication of MR = MC is that the monopolist’s profit-maximizing price bears the following relationship to its marginal costs and to the elasticity of demand (ε) :15

PM = MC / (1 + 1 / ε). (14.4)

Thus, the monopolist’s profit-maximizing price will be higher when its marginal costs are higher and when its demand is less elastic (i.e., the quantity demanded is less responsive to price changes). The monopolist will thereby be interested in reducing its costs and in undertaking actions (e.g., advertising and other forms of marketing and promotion) that expand the demand for gadgets but also that make the demand less elastic (e.g., that instills in buyers a stronger sense of brand loyalty and “Gotta’ have it!”).16 15  Where elasticity of demand is defined as the percentage change in quantity that accompanies a given percentage change in price; this is expected to be a negative number, since a price increase will cause a quantity decrease, and conversely. 16  However, as equation 14.4 makes clear, so long as the monopolist’s marginal costs are positive, the profit-maximizing price-quantity combination will be in the “elastic” portion of the firm’s demand curve; i.e., where | ε | > 1.0.

320   Lawrence J. White Also, as was noted above, from equation 14.4, the Lerner Index can be shown to be equal to

L = (P − MC ) / P = − 1 / ε. (14.5)

The Lerner Index thus indicates that the monopolist is exercising more market power when its demand curve is less elastic. In contrast to the monopolist, a competitive group of sellers that each sold identical gadgets and faced the same aggregate demand for gadgets and the same cost structure as does the monopolist would sell gadgets at an equilibrium price of PC . and would sell (in aggregate) a quantity QC . These are illustrated in figure 14.2 as well. Thus, the competitive industry sells at a lower price (again, anchored to its costs of production) and sells a larger aggregate volume. Further, the loss to consumers from the replacement of competition with monopoly is now more complex: In addition to the transfer of consumer surplus from buyers to the monopolist (rectangle PM yxPC ), there is also the loss of consumer surplus by buyers who would have bought gadgets at the competitive price PC but whose willingness to pay for gadgets is less than PM. These gadget purchases (represented in figure 14.2 by the distance QC − Q M ) simply aren’t made; the buyers instead buy other (lesser valued) things; and the aggregated lost value to these shut-out buyers—the social inefficiency of monopoly—can be approximately represented by the area of the triangle yzx. And, again, the Lerner Index will show a positive value for the monopolist, while showing a value of L = 0 for the competitive group of sellers.

14.2.4.  The Dominant Firm and the Reactive Fringe There appear to be industries where there is not a single-firm monopoly, but there is nevertheless a dominant firm with an apparent market share that exceeds 50% and a number of smaller firms that account for smaller shares.17 Modern examples would appear to be Google in online search; Intel in microprocessors; Apple in tablets; eBay in online auctions; and Microsoft in personal computer operating systems. Older examples would include AT&T in long-distance telephone service; Alcoa in aluminum; IBM in mainframe computers; Xerox in photocopy machines; and Kodak in film and cameras. If the other firms in the industry are few and have noticeable market shares, then strategic behavior among them is likely to prevail; and that puts the structure of the industry into the category of “oligopoly,” which is not the topic for this chapter. However, if the other firms are

17  What constitutes the “market” and hence how to measure “market shares” is an important but subtle point to which I will return below.

Monopoly and Dominant Firms   321

small and reactive (rather than strategic), a model that retains the basic features of monopoly can emerge.18 That dominant firm model can be described as follows: Suppose that there is a enterprise that has a special production advantage—say, exclusive access to a superior technology—that allows it to produce a distinctive product (that we will again call a “gadget”) at lower costs than can other firms. In figure 14.3 we represent the (sloped) market demand for gadgets per week and the enterprise’s costs of producing gadgets (which we again portray, for the purposes of expositional simplicity, as a horizontal line of constant unit costs, regardless of volume). If the firm were the sole seller of gadgets, then the analysis would revert to that of figure 14.2. However, there are a number of other firms that can also produce (identical) gadgets, but at higher costs than is true for the enterprise with the special technology (which, from this point forward, we will describe as the “dominant” firm). We will assume that these fringe firms are reactive, in the following sense: Rather than acting strategically, these firms simply react to a perceived price in the marketplace by supplying a quantity of gadgets that is consistent with simple (nonstrategic) profit-maximizing behavior:  the quantity at which their marginal costs (MC ) are equal to their marginal revenue (MR ) , which in turn is equal to the price; that is, the quantity at which P = MR = MC. The aggregate supply of gadgets that would be available from the reactive fringe firms at various prices is also portrayed in figure 14.3.19 In order to maximize its profits, the dominant firm has to take the fringe’s reactive supply into account in formulating its pricing decision.20 One possibility would be to practice “limit pricing”: to set a price that is just below the level that would begin to induce a supply response by the fringe. In figure 14.3, this would be price PL , and the consequent profits would be represented by rectangle PL wvPC .21 However, as a general matter, a superior strategy for the dominant firm is to treat the supply by the fringe as a “given” feature of the market. In essence, the fringe’s supply of gadgets unavoidably subtracts from the demand that is available to the dominant firm. The dominant firm thus has available to it a “residual demand”: the aggregate market demand at every price, minus the fringe’s supply at that price. It is with respect to this residual demand curve (which is portrayed in figure 14.3) that the dominant firm should calculate its marginal revenue and thus arrive at the quantity (and price) at which 18  This model is generally attributed to Stigler (1940), with revivals by Saving (1970) and Landes and Posner (1981); for a discussion, see Elzinga and Mills (2011). However, it appears that Amoroso (1938) preceded even Stigler; see Giocoli (2012). 19  We will also assume that entry by the fringe firms (or their equivalents) is sufficiently easy, so that they cannot be permanently driven from the marketplace by a sustained period of low pricing by the dominant firm. 20  Of course, if the fringe firms’ costs were so high that they were not a realistic challenge to the dominant firm (i.e., if point PL is sufficiently far up the vertical axis), then the dominant firm can simply ignore them, and the analysis reverts to that of figure 14.2 and the unhindered monopolist. 21  If the price of P would cause the fringe firms to go out of business and no other firms would L replace them, then the dominant firm would subsequently have the entire market to itself and would thus be able to charge the unfettered monopoly price. For the remainder of the discussion of the dominant firm we will assume that the fringe cannot be permanently driven from the marketplace. In section 14.3

322   Lawrence J. White

$

Supply by the fringe firms

Overall market demand

Residual demand available to the dominant firm PD

y w

PL PC

x 0

QF

Marginal revenue derived from the residual demand Marginal cost of the dominant firm

v QD

Qtotal

QL

QC

Quantity of gadgets per week

PD : price of gadgets sold by the dominant firm and by the reactive fringe firms QF : quantity of gadgets sold by the reactive fringe firms QD : quantity of gadgets sold by the dominant firm Qtotal : quantity of gadgets sold by the aggregate of the reactive fringe plus the dominant firm PL : price of gadgets if the dominant firm chooses to price at the level that just excludes the most efficient of the fringe firms QL : quantity of gadgets sold at a price of PL PC : competitive price (if superior technology is available to all) QC : quantity of gadgets sold by the competitive industry FIGURE  14.3  The

Dominant Firm and the Reactive Fringe

MR = MC. In figure 14.3 this is indicated by Q D and PD , and the dominant firm’s profits are indicated by rectangle PD yxPC . At the price PD , the aggregate market demand is larger than Q D , and the remainder is supplied by the fringe. Finally, if the dominant firm’s production technology were somehow readily available to a group of competitive firms, the outcome would again be PC and QC . So, again, the dominant firm (as a quasi-monopolist) causes the price to be higher and the quantity sold lower than if full competition (using the superior technology) could prevail. Thus despite the presence of the fringe (or, equivalently, because of the technological limits faced by the fringe), the dominant firm—though technically not a “monopolist,” I will address the issue of “predatory pricing,” whereby a firm is alleged to try to drive rivals permanently out of the market.

Monopoly and Dominant Firms   323

since it is not the sole seller of gadgets—still maximizes its profits on the basis of an unambiguous MR = MC calculation; perhaps “quasi-monopolist” would be an appropriate descriptor. However, what this portrayal does highlight is that the dominant firm’s profits are limited by the fringe firms. If the fringe were absent (so that the dominant firm had the entire market demand to itself) or the fringe were weaker (in the sense that the fringe firms’ costs were higher, so that the dominant firm had more of the aggregate market for itself), the dominant firm could charge a higher price, its market power (as measured by the Lerner Index) would be greater, and its profits would be larger. Accordingly, the dominant firm would be interested in undertaking activities that would raise the costs of its fringe competitors or would otherwise make their economic life more difficult.22 I will return to this point in the discussion of antitrust policy below. Finally, although the dominant firm model that was described above was based on the dominant firm’s superior cost technology in producing a generic item that is readily replicable (but at higher costs) by other firms, the same intuitive idea should carry over to a model of branded (and thus differentiated) goods or services: There could be a firm with a strong brand that is able to garner a large market share in a particular consumer goods category—say, laundry detergent—and a fringe of smaller firms with weaker brands (and that may or may not have the same production costs as the dominant firm). Based on the relative strength of its brand, the dominant firm would be able to charge higher prices and earn higher profits than could the fringe. And the dominant firm would be interested in measures that could weaken the fringe and thereby strengthen the market power of the dominant firm.

14.2.5.  How Does Monopoly Arise? We have already indicated that a monopoly must be selling an item that is distinctive. It would be a contradiction in terms to describe a “monopoly” as one among a number of firms that are all selling the same product. Indeed, for the dominant firm model that was just discussed, we acknowledged that “monopoly” was not quite right and suggested “quasi-monopoly.” But more than just distinctiveness is needed. That “more”—as Bain (1956) recognized—are barriers to entry:  Without barriers to entry, the above-normal profits of the monopolist could not persist. Potential sellers of the same or a closely similar product that are otherwise equally situated would be attracted by those above-normal profits, and their additional supply would cause the equilibrium price to fall. So long as the prospects of above-normal profits remained, entry would continue to occur. Only when the price was driven down to the level of PC would entry cease. There are basically three categories of entry barriers: 1. Ownership of a unique resource. The ownership of a unique resource constitutes one important potential barrier to entry. Examples would include (a) a unique mineral

22 

This would also be true, of course, if the dominant firm faces just a few rivals who act strategically.

324   Lawrence J. White deposit or other natural resource; (b) a unique government franchise (e.g., the right to be the sole provider of local taxicab services); (c) an especially distinctive patent;23 or an especially distinctive production technology.24 This last category is the foundation for the dominant firm model that was discussed above. 2. Economies of scale. If the technology of an industry is such that economies of scale—i.e., that the unit costs of production are always lower at higher rates of production per unit of time than at lower rates25—then a single firm (a “natural monopoly”) that supplies the market would be able to do so most efficiently from a production cost standpoint (i.e., with the lowest unit costs). Figure 14.4 provides an example of the cost curves that would follow from a production technology that exhibits economies of scale and thus would yield a “natural monopoly”: In this example there is an unavoidable (and sizable) fixed cost per period and constant marginal costs over all ranges of production. Since the fixed costs (by definition) do not vary with output, the average fixed costs (AFC) are lower when volumes are higher (i.e., the fixed costs are “spread” over more units); and thus unit costs (or “average costs” [AC], which are the sum of the constant marginal costs26 and the falling average fixed costs) are lower when volumes are higher.27 3. The size and “sunkenness” of needed investments. If entry into an activity requires a relatively large investment and that investment has no good alternative uses (i.e., the costs are “sunk”), then potential entrants would consider such entry to be quite risky. Examples of such investments include a large investment in specialized machinery; substantial research and development (R&D) expenditures that may not yield useful results; and large advertising and other promotional expenditures that are lost if unsuccessful. By contrast, if the necessary expenditure is modest in magnitude and, if spent

23  With respect to patents: There is a long and unfortunate legal history of describing all patents (and other forms of “intellectual property,” such as copyrights and trademarks) as “monopolies.” Since the United States alone currently issues well over 200,000 patents each year (and has issued almost nine million patents since the beginning of the US patent system in 1789), “monopoly” cannot be a useful descriptor for all patents. Instead, recognizing them as distinctive pieces of property—of which only some (probably only a relative few) each year are sufficiently distinctive that “monopoly” may be a useful descriptor for those patents—is a superior framework. 24  This distinctive technology might be based on a specific patent; or it could be based on nonpatented but not easily replicated “know-how” or trade secrets. 25  Economies of scale are conceptually distinct from the advantages of being able to exploit a “learning curve” (or “experience curve,” or “learning by doing”). The former involves lower unit costs’ being associated with higher volumes per period of time and is generally reversible; the latter involves lower unit costs that arise because of accumulated production volume over multiple time periods and is generally irreversible (unless the accumulated learning is somehow forgotten). Another way of thinking about the lower costs that occur as a consequence of a learning curve is that they are akin to the lower costs that arise as a consequence of technological innovation with respect to production processes. 26  Constant marginal costs, as portrayed in figure 14.4, imply constant average variable costs of the same magnitude; and it is really the sum of average fixed costs and average variable costs that yield unit costs. 27  Even if at some comparatively high volume the difficulties of managing a large enterprise might introduce diseconomies of scale, so long as the size of the market is smaller than that volume, a monopoly would still be the most efficient framework for that market.

Monopoly and Dominant Firms   325

$

AC MC AFC 0 FIGURE  14.4  The

Q per time period

Cost Curves for a Technology That Has Economies of Scale

on a tangible item, that item has a viable secondary market, the barriers to entry would be low. Monopoly would be more likely to arise in the former circumstance than in the latter. The ability of the monopoly to exploit its market power (as measured by the Lerner Index) would depend, as before, on the monopoly’s costs and the elasticity of market demand but now also on the extent to which other firms are disadvantaged by these barriers (e.g., the extent of the cost disadvantage that a firm faces from the diseconomies of smaller scale if it tries to enter at a small scale) as well as (for small numbers strategic situations) the potential entrants’ conjectures as to how the monopolist would react to their entry (see, e.g., Modigliani 1958 and Sylos-Labini 1962).

14.2.6.  Does a Positive Lerner Index Necessarily Mean “Market Power”? As was discussed above, the standard indicator of market power, at least by economists, has come to be the Lerner Index (see, e.g., Motta 2004, 41; Carlton and Perloff 2005; Perloff, Karp, and Golan 2007; and Tremblay and Tremblay 2012). For the stylized examples that I have employed thus far, the Lerner Index would serve as a good guide for considerations of antitrust policy: The gadget monopolist (and the dominant firm) has a positive Lerner Index; the competitive industry (that is producing and selling commodity gadgets) would have a Lerner Index that is equal to zero. However, a rigid use of the Lerner Index as an indicator of market power—and thus as an indicator of potential (likely?) antitrust concern—immediately becomes problematic when we move away from a simple world where all of the competitive firms in an industry are selling identical (homogeneous) commodity products; at best, this condition of homogeneity among sellers would roughly apply to sellers of primary agricultural and mineral commodities. Outside of these two areas, products (and services) tend to be

326   Lawrence J. White nonhomogeneous. With nonhomogeneity, the product and the seller are distinctive: In deciding from which seller to buy which product, buyers care about more than just which seller has the lowest price. They care about the attributes of the product and of the seller. The direct implication is that the demand curve that faces each distinctive seller has a negative slope (rather than being horizontal from the perspective of the competitive seller in an industry that is selling a homogeneous product, where only price matters to buyers). In turn, this will mean that the profit-maximizing output for the distinctive seller will be a level of output where P > MR = MC; and, thus, for this firm, L > 0 : The Lerner Index indicates that the firm has market power. But if there are a large number of somewhat similar but still distinctive sellers that are competing with each other (with respect to price and with respect to other attributes) and/or entry by such firms (i.e., somewhat similar but still distinctive) is relatively easy, then the typical distinctive firm will earn only normal profits (despite the fact that P > MR = MC and thus L > 0). This outcome was first recognized separately and independently by Chamberlin (1933) and Robinson (1933):  Chamberlin described the phenomenon as “monopolistic competition”; Robinson termed it “imperfect competition.” Chamberlin’s geometry, which showed this result, is reproduced in modified form in figure 14.5: The typical distinctive firm has a downward-sloping demand curve, denoted as D (and a concomitant marginal revenue curve, denoted as MR), and a cost structure that exhibits economies of scale over at least part of its range of volume (in figure 14.5, the cost structure of pervasive economies of scale from figure 14.4 has been used). The figure shows the equilibrium conditions for the typical distinctive firm that faces competition from other distinctive firms:  P = AC (the firm is earning normal profits); MR = MC (the firm is maximizing profits); and P > MR = MC (the Lerner Index is positive). If the typical firm were earning above-normal profits (i.e., P > AC), entry would occur, which would encroach on the firm’s sales (i.e., its demand curve would shift down and to the left) and push it toward normal profits; if the firm were earning below-normal profits, some firms would exit, leaving more potential customers for the remaining firms (the typical firm’s demand curve would move up and to the right) and again push it toward normal profits. The landscapes of most market economies are dominated by myriad distinctive firms that produce and sell distinctive (branded) products and services—whether as manufacturers or as services providers or as retailers. Although there may be some definitional rigor to the attachment of the concept of “market power” to such firms, it makes little intuitive sense to identify the corner delicatessen or the neighborhood kitchen remodeler or a small machine shop with the term “market power.” To do so risks either trivializing the concept or—in the context of public policy—greatly overstating the realistic domain of public policy concerns.28 Consequently, unless otherwise noted, the remainder of this chapter will describe “market power” as applying to instances where the size of the enterprise is large enough 28  See Pepall, Richards, and Norman (2008, 53–54) for similar concerns over the use of the Lerner Index as a mis-representation of “market power.”

Monopoly and Dominant Firms   327

$

P = AC PC-R AC MR 0

QC-R

D

MC

Q per time period

PC-R : price charged by a Chamberlin-Robinson seller QC-R : quantity produced/sold by a Chamberlin-Robinson seller FIGURE  14.5  The

“Tangency” Equilibrium for a Chamberlin-Robinson Competitor

to warrant special attention from public policy: that is, “significant” market power.29 We will thus remain in the traditional domain of “monopoly” and “dominant firm” that is associated with figures 14.2 and 14.3.

14.3.  Antitrust Policy with Respect to Monopoly and the Dominant Firm As was discussed in the introduction, there are at least five ways that antitrust policy can limit the ability of an enterprise unilaterally to exercise market power. This section will discuss those policies. Before I do so, however, there is a threshold issue that must be addressed: market delineation.

14.3.1.  A Threshold Issue: Market Delineation If “market power” is at issue in an antitrust proceeding, then generally there will need to be some specification of the “market” within which the “market power” has been or 29  This recognition that there must be some relative size threshold for realistic considerations of “market power” can also be found in Fisher (2008) and Baker (2008). It is possible, however, that even with a size threshold, a large firm with a positive Lerner Index might be just a large-scale version of the Chamberlin-Robinson firm of figure 14.5. Areeda and Hovenkamp (2002, 133) state that “Market power need not trouble the antitrust authorities unless it is both substantial in magnitude and durable.”

328   Lawrence J. White will be exercised. The exception is “hard core” price fixing and similar collusive arrangements that are prosecuted under Section 1 of the Sherman Act: These are treated as “per se” violations, in which the plaintiff need show only that the actions occurred and need not show effects (which would likely require the specification of a market). In all other antitrust cases, a relevant market will need to be delineated.30 The needed specification of the market can arise in two broad contexts: (a) A proposed action—for example, a proposed merger—might threaten to create market power where it hadn’t previously existed (or it could enhance/exacerbate market power that already exists); or (b) An enterprise is accused of actions that are manifestations of its already existing market power. We will address each, in turn. 1. Market delineation to deal with a proposed action. Proposed mergers are the most common context in which the potential creation or enhancement of market power arises, and merger analysis will provide a convenient basis for the discussion of this approach to market delineation.31 The goal of antitrust merger enforcement is to prevent mergers that would create or enhance market power. The delineation of a relevant market—in product “space” and in geographic space—is essential for the analysis that should underlie merger enforcement decisions. Since 198232 the US Department of Justice’s Antitrust Division (DOJ) and the Federal Trade Commission (FTC) have been using a set of “Merger Guidelines” (which have been updated periodically, the last such revision occurring in 2010) that have at their heart a market delineation paradigm that runs as follows: Suppose that the analyst starts with a “candidate” market that consists of (at least) the two firms that are proposing to merge. The analyst then addresses the following question: If all of the sellers of the candidate market were combined into a single firm (a “hypothetical monopolist”), could that firm achieve a “small but significant and nontransitory increase in price” (SSNIP) above the current price (or above the level at which the price would otherwise be)? If the answer to this question is yes, then this is a relevant market; if the answer is no (because too many buyers would turn to other sellers of the same or similar items, so as to thwart the attempt to raise the price), then the candidate market must be widened (by including more sellers) until the answer is yes. Generally, the smallest relevant market will be the one that serves as the basis for further analysis of the merger. Price increases of 5% or 10% have usually been used as the measure of “significant” in SSNIP.

However, their subsequent discussion indicates that “magnitude” refers to the deviation between price and marginal cost, and not the size of the enterprise. 30 

Often this is described as “market definition.” Since other chapters in this handbook deal more extensively with merger policy, the discussion here will necessarily be brief. 32  An earlier set of Merger Guidelines had been issued by the DOJ in 1968; however, the approach to market delineation in those Guidelines did not prove satisfactory or lasting. 31 

Monopoly and Dominant Firms   329

In essence, the paradigm rests on the idea that a relevant market is one in which market power could be exercised (by the “hypothetical monopolist”).33 Then subsequent stages of the analysis—by examining the market shares of the merging parties and of other major sellers in the relevant market, the change in market shares that would occur as a direct (pro forma) consequence of the merger; the conditions of entry; the buyers’ side of the market; and so on—try to determine whether the merger that has been proposed is likely to result in the creation or enhancement of market power. The market delineation paradigm that was just described was originally developed for the analysis of the possible creation of market power through (what is now described as) “coordinated effects”; earlier generations of economists would have described this as implicit “oligopolistic coordination.” Thus, the merger analysis was focused on the prevention of the creation or enhancement of market power that would be jointly exercised by a group of oligopolists. Unless the relevant market happened to encompass just the two merging firms (and thus the merger was a “2 to 1” merger), it would not directly address the issue of the creation of market power that could be exercised unilaterally. Nevertheless, the hypothetical monopolist paradigm is valuable for the purposes at hand for the following reason: If one thinks of the group of coordinating oligopolists as trying to emulate a monopolist, then the hypothetical monopolist provides an analytical basis for antitrust enforcement that can limit this emulation. Since 1992, the Merger Guidelines have also encompassed an analytical approach to addressing problems of “unilateral effects”: These problems arise when the two merging firms sell differentiated products that compete with each other. To the extent that there are customers of one of the firms that have considered the merger partner’s products to be their best second choice, the merger eliminates the restraint that each firm exercised on the other firm’s ability to raise prices. Thus, the merger will allow the merged firm unilaterally to raise its prices.34 In principle, a market delineation analysis is not needed for this approach. There only needs to be a finding (which would have to be based on solid empirical analysis) that significant unilateral effects would arise as a consequence of the merger. Or, if it were felt that there must be a delineation of a market for there to be a finding that the postmerger firm would be able to exercise greater market power, then (tautologically) the finding that the merged firm has the ability unilaterally to increase its price (if that increase exceeds a SSNIP threshold) must mean that the two prospective merger partners comprise a relevant market for that product. 2. Market delineation to assess whether a firm already posses market power. In a typical antitrust lawsuit in which a firm is being accused of actions that have monopolized or have attempted to monopolize an industry,35 the plaintiff has to demonstrate that the 33  Although this paradigm was first used in the DOJ 1982 Merger Guidelines, the concept appears to have first been developed by Adelman (1959b); see Werden (2003). 34  More detail on this approach will be provided below. 35  Recall that Section 2 of the Sherman Act does not forbid “monopoly”; instead it forbids actions that “monopolize, or attempt to monopolize.”

330   Lawrence J. White defendant possesses market power; without the possession of market power, the defendant’s actions would be unlikely to have significant consequences. Consequently, again, a relevant market must be delineated. Not surprisingly, the plaintiff will want to claim that the relevant market is narrow and that the defendant’s sales constitute a large share of that market (and that entry is difficult, etc.), which would be a strong indicator that the defendant possessed market power; the defendant, of course, will want to claim the opposite: that the market is quite broad and that the defendant has only a small share (and that entry is easy, etc.) and doesn’t/can’t exercise market power. A famous example of this type of issue arose in the antitrust suit by the DOJ against du Pont in the 1950s, alleging monopolization of cellophane.36 The DOJ alleged that the relevant market was narrow: cellophane; du Pont claimed that the market was much broader and encompassed all flexible wrapping materials. More recently, in the DOJ’s monopolization suit against Microsoft in the late 1990s37 the DOJ argued that the relevant market was operating systems for Intel-compatible personal computers; Microsoft argued for a wider delineation that would have encompassed all software on all platforms for computing (including applications running on servers). Yet more recently, in the DOJ’s antitrust suit against Visa and MasterCard in the early 2000s that alleged monopolization of credit card issuance,38 the DOJ argued for a narrow market of credit and charge cards; Visa and MasterCard argued for a broader payments market that also included debit cards, checks, and cash. Which perspectives were valid? Unfortunately, the market delineation paradigm that works well for the Merger Guidelines’ approach to “coordinated effects” merger cases generally doesn’t apply to such monopolization cases. The Merger Guidelines paradigm addresses a prospective merger and the possibility that this prospective merger might create or enhance market power as a consequence of the completion of the merger. By contrast, in the context of a monopolization case, the goal is usually to try to determine whether the defendant currently has market power. The use of the SSNIP test— asking whether the defendant could profitably raise its price from current levels—ought to be useless: If the firm is maximizing its profits, the answer ought always to be no, regardless of whether the firm does or does not have market power. The defendant’s current price should already be its profit-maximizing price; any increase above the current level should be unprofitable, even for a monopolist.39

36 See U.S. v. E.I. Du Pont de Nemours & Co., 351 U.S. 377 (1956); for discussion, see Stocking and Mueller (1955). 37 See U.S. v. Microsoft Corp., 253 F.3d 34 (2001); for discussion, see, e.g., Rubinfeld (2014).

38 See U.S. v. Visa U.S.A., Inc. Visa International Corp., and MasterCard International Inc., 344 F.3d 229

(2003); for discussion, see Pindyck (2014). 39  Unfortunately, as White (2008) documents, the uselessness of the SSNIP test has not stopped judges in monopolization cases—and even some expert economists—from asking a SSNIP-type of question in these cases. In the du Pont cellophane case, the US Supreme Court asked it, and the majority concluded that du Pont did not have market power because the company could not increase its price of cellophane profitably from current levels—that du Pont was too constrained by competition from other flexible wrapping materials. This inappropriate use of a SSNIP type of question has since come to be known in antitrust discussions as the “cellophane fallacy.”

Monopoly and Dominant Firms   331

If profits were considered to be a reliable indicator of the exercise of market power, they might help address the market power issue.40 Recall that the monopolist of figures 14.1, 14.2, and 14.3 is expected to earn above-normal profits, whereas the competitors in those figures, as well as the Chamberlin-Robinson competitor of figure 14.5, is expected to earn only normal profits. But since the early 1980s most economists have been leery of the use of reported profit rates as evidence that can be used to measure the presence of market power;41 and the use of Lerner Indexes alone won’t help, since both the monopolist of figures 14.1, 14.2, and 14.3 and the Chamberlin-Robinson competitor of figure 14.5 have Lerner Indexes that exceed 0. Unfortunately, there have been no generally accepted market delineation paradigms for monopolization cases that would solve this conundrum.42 The development of an appropriate paradigm remains as a serious need for antitrust policy and jurisprudence.

14.3.2.  Antitrust Policies to Limit the Unilateral Exercise of Market Power As was mentioned in the introduction, there are a least five ways that antitrust policy can limit the unilateral exercise of market power. I will now address these five. 1. Prevent/stop agreements among firms that restrict competition among them and that thereby allow the individual firms to continue to exercise unilateral market power. As Werden (2000) has pointed out, a SSNIP test would be appropriate if the issue that was under litigation was a prospective action (e.g., a proposed exclusionary action) by the defendant against which the plaintiff was seeking an anticipatory injunction. In that event, the question—“Will this action create (or add to the defendant’s) market power?”—could be addressed by SSNIP: “After the proposed action occurs, would the defendant (unilaterally or in concert with other firms) be able to achieve a SSNIP?” But few monopolization cases involve prospective actions. 40  In their commentary on the du Pont cellophane case, Stocking and Mueller (1955) demonstrate that du Pont’s profits from selling cellophane were substantially higher than the company’s profits from selling rayon, where du Pont faced considerably more direct competition; they argue that this indicates that du Pont was exercising market power and thus that cellophane, and not flexible wrapping materials, was the relevant market. 41  For critiques of the use of profit data to indicate the presence of market power, see Fisher, McGowan, and Greenwood (1983); Fisher and McGowan (1983); Benston (1982); and Fisher (1987). Carlton and Perloff (2005) and Perloff, Karp, and Golan (2007, ch. 2) offer eight reasons why accounting data on profits are inappropriate and misleading for cross-section studies that would try to reveal the presence of market power. 42  White (2008) offers some suggestions. Also, for some industries, where good price information is available and prices are known to vary for different localities (e.g., the retail prices for an item that is sold in many different metropolitan areas, or airline fares for many different city-pairs), empirical analyses—e.g., regression analyses where price (as the dependent variable) is regressed on a measure of seller concentration (as an indicator of the likelihood of the exercise of market power) and other control variables—may be able to establish that local markets for that item or service are relevant markets. Many airline fare studies have established local city pairs as relevant markets; see, for example, Borenstein (1989, 1992). When monopolization cases arise that involve airlines, city-pairs are usually accepted as

332   Lawrence J. White Agreements among firms that restrict competition among them are often described in terms of “price fixing”: agreements to set the price collusively rather than allowing the competitive process and the market to determine the price. To the extent that the goal of the colluders is to approximate the monopoly outcome of figure 14.2, the analysis that was developed earlier in this chapter is relevant. However, since this chapter is focused on the unilateral exercise of market power, a different set of agreements will be addressed: market allocation agreements. These might also be described as “live and let live” arrangements: Firms A and B, which produce and sell similar products, agree to sell only within their own specified and separate geographic areas (or to their own specified and separate categories of customers) and not to challenge or encroach upon each other’s territories or customer categories. If firms A and B are thereby the sole sellers within their territories (or to their customer categories), then the agreement achieves the monopoly outcome of figure 14.2 for each firm, as compared with a more competitive outcome that would prevail in the absence of the agreement.43 Enforcement of the American antitrust laws has long recognized the antisocial nature of such market allocation arrangements, and judicial opinions have deemed them to be “per se” violations of Section 1 of the Sherman Act, which forbids “every contract, combination . . ., or conspiracy, in restraint of trade or commerce.”44 A recent manifestation of alleged noncompete agreements has arisen in the pharmaceuticals area. A stylized version of such an agreement is as follows:45 Branded pharmaceutical Company I (for “incumbent”) has a patent on an important drug that expires in, say, 2015; after that date generic versions of that drug can be sold. In 2011 generic pharmaceutical Company E (for “entrant”) believes that Company I’s patent is invalid and files an application with the US Food and Drug Administration to sell a generic version of the drug. Company I sues Company E for patent infringement. Instead of going to trial, the two companies settle the suit on the following terms: Company E agrees to

the relevant markets; see, for example, Edlin and Farell (2004) and Elzinga and Mills (2014a). Similarly, the FTC was able to establish (on the basis of price information) that “big box” office supply retailers in local metropolitan areas were a relevant market in its successful effort to stop the merger of Staples and Office Depot in 1997 (FTC v. Staples, Inc., 970 F. Supp 1066 [1997]; for discussion, see, for example, Baker, 1999; Ashenfelter et al. 2006; and Dalkir and Warren-Boulton 2014); this same kind of information could be used to establish the same relevant markets if (hypothetically) a big-box office supply retailer were accused of exclusionary monopolistic behavior (e.g., of trying to buy all of the good retail sites in a metropolitan area, so as to restrict entry by rival big-box office supply retailers). 43  If the two firms produce identical products and would otherwise compete directly on price without any effort to behave strategically (i.e., they are “Bertrand” competitors), then the competitive outcome would be that shown in figure 14.2 as well. 44  Arguably, the first Supreme Court condemnation of such horizontal allocation arrangements (as opposed to specific price-fixing agreements) came in Addyston Pipe & Steel Co. v. U.S., 175 U.S. 211 (1899). In that case, the manufacturers of pipe agreed not to compete with each other in bidding for municipal contracts. The Court’s condemnation of such horizontal noncompete arrangements has been periodically reaffirmed; e.g., in U.S. v. National Lead Co., 332 U.S. 319 (1947); and in U.S. v. Topco Associates, Inc., 405 U.S. 596 (1972). 45  This version will exclude a great deal of institutional detail. For a more comprehensive analysis of these issues, see, for example, Bigelow and Willig (2014).

Monopoly and Dominant Firms   333

delay its sale of the generic version until 2013; Company I agrees that it will not challenge Company E at that time and agrees to make a payment of $X million to Company E. This kind of agreement certainly has the potential to preserve the market power of Company I. If a patent were not involved and this was simply a payment from I to E to delay E’s entry into I’s market, the agreement would almost surely be condemned as a per se violation of the Sherman Act. Because I’s profits as a sole seller (monopolist) will always exceed the joint profits of the two firms selling in the market (and at the limit, the comparison of profits is the monopoly/competition comparison of figure 14.2), Company I can afford to offer a substantial payment to E for delay, and both companies will be better off than if E entered immediately. But consumers will be worse off. However, because a patent is involved and the true validity of the patent is a central but unknown feature of the agreement and because the courts have generally viewed settlements of lawsuits as a beneficial activity that represents a mutually agreed outcome that also economizes on judicial resources, the courts were generally not hostile to these agreements, despite repeated challenges to these agreements by the FTC and by private plaintiffs. Further, there are circumstances—related to the uncertainties that surround the validity of the patent, as well as uncertainties as to how long a trial would take and how much the litigation costs would be—where a “reverse payment”46 settlement can even make consumers better off as compared with the likely outcome in the absence of the settlement.47 However, at the end of 2012 the Supreme Court agreed to hear the FTC’s appeal of one of its cases, and in June 2013 the Court decided that this type of “reverse payment” arrangement could constitute an antitrust violation and that a “rule of reason” analysis was required.48 2. Prevent/stop mergers that would otherwise allow the merged firm to exercise market power or to enhance existing market power. Since 1914 Section 7 of the Clayton Act has empowered the FTC and the DOJ to stop mergers where “in any line of commerce in any section of the country, the effect of such acquisition may be substantially to lessen competition, or tend to create a monopoly.” As originally written, the Act stopped only mergers that were consummated through the acquisition of shares of stock of one company by the other; this left a huge loophole in the form of one company’s direct purchase of the other company’s assets (including its brand name). This loophole was closed in 1950 by the Celler-Kefauver Amendments to the Clayton Act. As was discussed above, the economic analysis of the possible anticompetitive consequences of mergers has focused on two routes:  “coordinated effects,” whereby the merger would allow greater oligopolistic coordination among rivals; and “unilateral effects,” whereby the merging firms sell competing differentiated products and there are significant numbers of customers of either of the merging firms that consider the products of the other firm to be their second choice. In this latter case, the merged firm would 46 

Because plaintiffs do not usually make payments to defendants to settle lawsuits, these types of settlements have come to be known as “reverse payments.” 47  See Bigelow and Willig (2014) for an elaboration of this argument. 48 See FTC v. Actavis, Inc., 133 S.Ct. 2223 (2013).

334   Lawrence J. White have an incentive to raise its prices (as compared with premerger prices), since its partner would no longer be an independent restraint. This latter price-raising effect is clearest if the firm were able to identify which are the “trapped” customers that meet this condition; it could then practice price discrimination toward those customers and just raise its prices toward those customers.49 But the incentive to raise prices holds generally, even when price discrimination is not possible. The strength of the “upward pricing pressure” (UPP)50 depends on the extent to which the merged firm is able to recapture the customers that the premerger firms would have otherwise lost from a price increase and the profit margins on the “companion” product to which those customers are diverted. Any cost-reduction efficiencies that would accompany the merger and that would reduce marginal costs would offset those demand-driven upward pricing pressures, yielding a net UPP.51 Finally, although a unilateral effects analysis that reveals that a significant net UPP would arise from a proposed merger may not track exactly the competition/monopoly comparison of figure 14.2, it is clearly in the same spirit: For the customers of the merged firm, the merger would be accompanied by a significant increase in prices, which would be the manifestation of the merged firm’s increased market power (and that increase in market power would, of course, be registered by a higher Lerner Index for the merged firm). 3. Prevent/stop predatory behavior by a firm that can enhance its market power by driving rivals out of the market. A potential strategy for a firm (the “incumbent”) that wants to achieve a monopoly position would be to set its prices temporarily at such low levels that all of its rivals are permanently driven from the market; the firm could then raise its prices to the monopoly level that is represented in figure 14.2. Such a strategy is typically described as “predatory pricing.”52 The strategy is best described in terms of a necessary “investment” and then a subsequent “return” or “recoupment”: The investment is the reduced profits (or actual losses) that the incumbent incurs while it is charging the low prices that will drive its rivals from the market; the return is the larger profits that the incumbent earns after the rivals have exited the market. Thus, as is true for the evaluation of any investment, the size of the 49  Of course, in addition to identifying those “trapped” customers, the firm would also have to prevent arbitrage between other (low-price) customers and these high-price customers and deal with the customer unhappiness that would arise when the discriminated-against customers learn that they are paying higher prices than are other customers. 50  This is the phrase that Farrell and Shapiro (2010) have popularized. 51  Also, to the extent that entry would occur in response to the net UPP or other competing firms would reposition their products so as to attract some of these “trapped” customers, the net UPP would be smaller. 52  This discussion will focus on predatory pricing; but, in principle, other activities by the incumbent—excessive advertising or promotion, or flooding the market with capacity, or buying crucial resources in excess of reasonable needs—could fit the “predatory” pattern. For a failed effort by the DOJ to prosecute an instance of alleged predatory flooding of capacity by an incumbent airline, see U.S. v. AMR Corp., 335 F.3d 1109 (2003); for an economics discussion of this case, see, for example, Edlin and Farrell (2004). In addition, the exclusionary actions that are discussed in the following section could also be interpreted through a “predatory” lens.

Monopoly and Dominant Firms   335

investment and the size of the return must be calculated; and, because the return occurs at a later time than the investment, a discount rate must be used. Finally, it is crucial that the rivals be permanently driven from the market. If instead entry is easy, and potential entrants are not deterred by the possibility that the firm might again aggressively try to drive them from the market—perhaps because the entrants are myopic, or because they are strategic and believe that the incumbent can be deterred by an equal determination by the entrants to stay in the market—then there will be little or no return or recoupment. On the other hand, if the firm can establish a reputation for being aggressive, then this may deter potential future entrants; or—in the case of a firm that is selling multiple products in multiple markets—deter potential entrants in other markets.53 Public policy and judicial decisions with respect to predatory pricing have varied over the decades. From the 1930s through the 1960s, a populist theme had a major influence on antitrust: Large firms should be discouraged, regardless of efficiencies; and small firms should be preserved—partly for their own sake (the populist theme) and partly because of fears that large firms would drive out small firms and then be able to impose monopoly pricing (the predation scenario). During the 1930s amendments to the antitrust laws strengthened the Clayton Act Section 2 restrictions on price discrimination (this became known as the Robinson-Patman Act of 1936) and amended the Sherman Act to allow the states to authorize resale price maintenance (through the Miller-Tydings Act of 1937). In the former case, the intent was to prevent large retailers (the “chain stores”) from obtaining better wholesale prices from manufacturers than could smaller retailers; and in the latter case, the intent was to prevent the large retailers from selling at lower prices than could smaller retailers.54 Antitrust judicial decisions reflected these policy trends.55 Beginning in the 1970s, however, the populist streak in antitrust receded, and an emphasis on encouraging competition and limiting the exercise of market power became stronger.56 There was still a concern that predatory pricing could undo competition and create a monopoly. But there was also a stronger concern that inhibitions on aggressive pricing would simply inhibit vigorous competition, to the detriment of consumers.57 Reinforcing 53 

See, for example, Bolton, Brodley, and Riordan (2000). Further, even if a reputation for aggressive behavior doesn’t deter entrants, to the extent that a firm faces strategic rivals in one or more markets, its reputation for being willing to be aggressive may serve to discipline those rivals and restrain them from competitive initiatives. 54  Simultaneously, economic regulation was broadened, so as to restrict competition in interstate trucking, airlines, and banking, and restrictions on competition in railroads were strengthened. And for a brief period, from 1933 to 1935, the Sherman Act’s prohibition on price fixing was effectively suspended, as the National Recovery Administration (NRA), which was authorized by the National Industrial Recovery Act of 1933, negotiated and enforced codes of “fair competition” for industries, which effectively encouraged price fixing. A Supreme Court decision that declared the NRA to be unconstitutional (Schechter Poultry Corp. v. U.S., 295 U.S. 495 [1935]) ended this “experiment.” 55  See, for example, Utah Pie Co. v. Continental Baking Co., 386 U.S. 685 (1967); see also Adelman (1959a) and Levinson (2011). 56  And, simultaneously, there was greater emphasis on deregulation and an encouragement of greater competition in airlines, trucking, rail, and banking. 57  This has come to be known as the problem of “Type I errors”: the condemnation (as “predatory”) of pricing that is simply vigorous but competitive.

336   Lawrence J. White this concern was the realization that many judicial (and regulatory) determinations of what was “below cost” pricing included (for multiproduct firms) arbitrary allocations of fixed and overhead costs into the unit costs that were the standard for which lower prices indicated predation. The so-called Areeda-Turner rule (Areeda and Turner 1975), which established prices that are equal to or greater than marginal costs (with average variable costs serving as a proxy for marginal costs) as a “safe harbor” against charges of predation, provided important guidance for policy and for the judiciary.58 Starting in the 1980s, Supreme Court decisions have taken an increasingly skeptical view of the likelihood of successful predation.59 However, what has been missing from those decisions, as well as from lower court decisions, is the recognition that (as was discussed above) the extra benefits of the acquisition of a reputation for being aggressive may make worthwhile an episode of predatory pricing that otherwise appears to lack a positive investment/return profile (and that is thereby likely to be dismissed as merely vigorously competitive, rather than as potentially predatory).60 4. Prevent/stop exclusionary behavior by a firm that can enhance its market power. All enterprises rely on “upstream” suppliers to provide them with some of their inputs; and all but end-of-the-chain retailers sell to “downstream” customer firms.61 Vertical integration by a firm at one stage into an upstream or downstream stage may bring socially beneficial efficiencies in exploiting production efficiencies and/or overcoming externality problems or informational asymmetries; or a firm that lacks the managerial or production capabilities to integrate vertically into an adjoining stage may still be able to overcome the externalities or asymmetries through vertical contracting practices.62 58  Parallel efforts were being made in the regulatory arena to move pricing decisions away from “fully distributed costs” and toward “incremental costs”; see, for example, Baumol (1968) and Baumol and Walton (1973). The Areeda-Turner rule is clearly conservative, as a reference to figure 14.3 demonstrates: By pricing at the level of PL, the dominant firm eliminates profitable production by the fringe, even though PL is above the dominant firm’s marginal costs. If this could make the fringe disappear permanently, then the dominant firm would have the entire market to itself, even though it would not have violated the Areeda-Turner rule. 59 See Matsushita Electric Industrial Corp., Ltd., et al. v. Zenith Radio Corp. et al., 475 U.S. 574 (1986); Brooke Group, Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209 (1993); and Weyerhaeuser Co. v. Ross-Simmons Hardwood Lumber Co., Inc. 549 U.S. 312 (2007). For economics commentary on each, see, for example, Elzinga (1999), Burnett (1999), and Rausser and Foote (2014), respectively. For a case in which the plaintiff convinced an appellate court of the existence of predation, see Spirit Airlines Inc. v. Northwest Airlines Inc., 431 F.3d 917 (2005); for an economics discussion of this case, see, for example, Elzinga and Mills (2014a). 60  Another approach that may well be worth considering is embodied in the concept of “no economic sense”: that a price or nonprice action should be condemned if it made no economic sense for the firm that was undertaking the action unless the target firm (or firms) disappeared from the market or was otherwise disciplined. See Ordover and Willig (1981, 1999) and Werden (2006). 61  In the case of firms that produce and then sell finished goods to or through “distributors” (whether wholesale or retail), the particular legal arrangement with respect to who owns the goods at which stage can affect whether one would consider the distributor to be “downstream” from the producer, or whether the producer is purchasing “distribution services” from the distributors, which would make them (at least conceptually) “upstream” from the producer. 62  Examples of such practices include resale price maintenance (RPM), territorial exclusivity, exclusive dealing, exclusive dealers, full-line forcing, tying, and bundling. Often, these vertical

Monopoly and Dominant Firms   337

However, a firm that enjoys market power at one stage may also be able to use vertical integration or vertical contracting practices to enhance its market power—essentially by restricting and/or raising the cost to rivals of access to upstream or downstream resources.63 By raising costs to rivals, a firm is able to deflect demand to itself and thereby increase its prices and its profitability. Recall that the dominant firm in figure 14.3 benefits from actions that cause the fringe firms’ supply curve to shift upward. The courts—supported by many economists’ suspicions about anticompetitive nature of vertical restraints—initially approached vertical restraints and even vertical integration via merger with substantial hostility.64 At various times, resale price maintenance, territorial exclusivity, exclusive dealing, and tying were branded as “per se” offenses. However, starting in the 1970s, the courts have taken a more nuanced view toward vertical restraints, recognizing their potential efficiencies and essentially judging them under a “rule of reason.”65 However, a rule-of-reason approach does not mean that the defendant always carries the day with efficiency arguments; see, for example, U.S. v. Visa USA Inc., Visa International Corp., and MasterCard International Inc., 344 F.3d 229 (2003);66 and U.S. v. Dentsply International, Inc., 399 F.3d 181 (2005).67 5. Structural dismemberment. If a firm is found to be exercising market power— because it is found to have a positive Lerner Index (and thus is charging P > MC) and/ or it has engaged in predatory/exclusionary acts—then one way to limit that exercise of market power is to dismember the firm: to break the firm into two or more horizontal (and viable) competitors or two or more vertically related entities (if vertical exclusionary acts are seen as the problem). This dismemberment route was the approach that the Supreme Court adopted in 1911 to DOJ monopolization prosecutions of the Standard Oil Company of New Jersey (in Standard Oil Co. of New Jersey v. U.S., 221 U.S. 1 [1911]) and the American Tobacco Company (in U.S. v. American Tobacco Co., 221 U.S. 106 [1911]). In the decade that followed the DOJ succeeded in breaking up other large industrial firms that had large market shares and that appeared to be exercising market power. contracting practices (or vertical integration) are beneficial in dealing with free-riding problems. This is the argument that is often advanced to support resale price maintenance (see, e.g., Telser 1960), territorial exclusivity, and exclusive dealing (see, e.g., Marvel 1982). 63  As Asker and Bar-Isaac (2014) demonstrate, RPM can be seen as an effort by the incumbent to “bribe” its distributors to refrain from distributing a rival’s products. 64  The story of the initial legal and economics hostility to vertical practices, and the eventual softening, is recounted in White (1989, 2010a, 2010b). 65  An important early decision was Continental T.V., Inc., et al. v. GTE-Sylvania Inc., 433 U.S. 36 (1977); for an economics discussion of this case, see, for example, Preston (1994). The most recent major movement in this direction came in the Supreme Court’s decision in 2007 to treat RPM under the rule of reason, in Leegin Creative Leather Products, Inc. v. PSKS, Inc., 581 U.S. 677 (2007). For an economics discussion of this case, see Elzinga and Mills (2014b). Tying remains technically a per se offense; but, by insisting that a defendant possess market power before a tying prosecution can succeed, the Supreme Court has greatly weakened the sting of the apparent per se status of tying. See, for example, Jefferson Parish Hospital District No. 2 v. Hyde, 466 U.S. 2 (1984); for an economics discussion of that case, see, for example, Lynk (1999). 66  For an economics discussion of this case, see, for example, Pindyck (2014). 67  For an economics discussion of this case, see, for example, Katz (2014).

338   Lawrence J. White By 1920, however, the Supreme Court backed away from this approach, refusing to find that the U.S. Steel Corporation was guilty of monopolization (in U.S. v. U.S. Steel Corp., 251 U.S. 417 [1920]), and for the next two decades the dismemberment approach to addressing market power was largely abandoned. In the 1940s through the 1960s, however, prosecutorial and judicial interest in this approach was revived. The DOJ’s successful prosecutions of Alcoa (in U.S. v. Aluminum Co. of America, 148 F.2d 416 [1945]), of movie chains and distributors (in U.S. v. Crescent Amusement Co., 323 U.S. 173 [1944]; U.S.  v.  Griffith, 334 U.S. 100 [1948]; and U.S.  v.  Paramount Pictures, Inc., 334 U.S. 131 [1948]), of the United Shoe Machinery Corporation (in U.S.  v.  United Shoe Machinery Corp., 347 U.S. 521 [1954]),68 and the Grinnel Corporation (in U.S. v. Grinnel Corp., 384 U.S. 563 [1966]) all resulted in some divestitures of either a horizontal or vertical nature; for example, in Alcoa, the DOJ succeeded in requiring that Alcoa divest its Canadian subsidiary, Aluminium Ltd., which became an independent competitor to Alcoa. Simultaneous with this revived prosecutorial and judicial interest in the dismemberment approach to monopolization problems was a rising academic/intellectual interest in this approach, which peaked in 1959 with the publication of a pro-dismemberment monograph by Kaysen and Turner (1959). The Kaysen and Turner intellectual argument for a “no fault” approach to the dismemberment of firms that were exercising market power (which included oligopolies, as well as monopolies and dominant firms) rested on the findings of Bain (1951, 1956), who showed that corporate profits (and thus the exercise of market power) were positively related to seller concentration (Bain 1951)69 and that economies of scale were relatively unimportant in explaining the sizes of large oligopolistic firms (Bain 1956). Also, an important argument for a structural approach was (and remains) that it is largely self-enforcing: Whereas a behavioral remedy requires enforcement by a federal district court judge, a successful structural remedy (i.e., if the additional competitors that are created by the structural remedy are economically viable) creates the improved competitive conditions that limit the exercise of market power. By the 1970s and 1980s, however, many economists’ faith in the linkage between corporate profits as an indicator of the exercise of market power and seller concentration was substantially weakened,70 as was the belief that economies of scale were unimportant for explaining the size and profitability of major industrial firms (see, e.g., Demsetz 1973, 1974 and Mancke 1974). There was still evidence, based on price data (instead of profit data), that seller concentration and the exercise of market power were still positively associated (see, e.g., Weiss 1989 and Audretsch and Siegfried 1992). Still, the 68 

For an economics discussion of this case, see Kaysen (1956). Bain’s findings were replicated many times, with increasingly sophisticated econometrics, over the next two decades. For summaries of these many studies, as of the early 1970s, see Weiss (1971, 1974). 70  As was discussed, above, during this period substantial doubts were raised about the validity of reported corporate profits as an indicator of true “economic profits” and thus as an indicator of the exercise of market power. In addition, critics raised the possibility that the higher profits that were associated with greater seller concentration were the result of economies of scale; see, for example, Demsetz (1973, 1974) and Mancke (1974). 69 

Monopoly and Dominant Firms   339

intellectual force that might have continued to sustain the dismemberment approach was clearly weakened. One final major dismemberment was achieved by the DOJ, in a settlement of a monopolization case against AT&T in 1982.71 In this case, the DOJ’s argument that AT&T’s ownership of (regulated) local wire-line telephone service, long-distance service, and telephone equipment manufacturing facilities made it difficult for rivals to enter and compete in the latter two areas carried the day, and the local wire-line operating companies were separated from the long-distance and equipment facilities (and the local operating companies themselves were separated into seven regional companies, so as to reduce the possibility of monopsony power in their purchases of telephone equipment). In the 1990s, when the DOJ prosecuted Microsoft for monopolization and initially won its case at the federal district court level, it asked for and got the court to agree to the vertical dismemberment of Microsoft into an operating system (OS) company and a browser-plus-software-applications company;72 the theory/hope was that the OS company would vertically integrate into browsers and applications, while the browser company would vertically integrate into OSs, and thus competition in both areas would increase. However, when the case was appealed, the appellate court—though supporting the DOJ’s claim that Microsoft had engaged in monopolization—refused to endorse the dismemberment remedy and remanded the case to the district court for further consideration of a remedy.73 The case was finally settled with only behavioral remedies and no dismemberment.74 It seems unlikely that, in the absence of a major change in antitrust economics and legal sentiment, there will be any further calls for dismemberment in future monopolization cases brought by the DOJ.

14.4. Conclusion Monopoly and the dominant firm continue to be an important and interesting area for antitrust policy and jurisprudence, as well as for antitrust economics. Although the basic concepts of monopoly and the exercise of market power are clear—at least in antitrust economics—the identification of monopoly and the issue of what to do about it when it is identified remain as contentious issues. Especially troubling is the absence of a clearly developed paradigm for the delineation of a market in most monopolization cases; 71 See U.S. v. AT&T, 552 F. Supp. 131 (1982); for an economics discussion of this case, see, for example,

Noll and Owen (1994). 72 See U.S. v. Microsoft Corp., 84 F. Supp. 2d 9 (1999); 87 F. Supp. 2d 30 (2000); and 97 F. Supp. 2d 59 (2000). 73 See U.S. v. Microsoft Corp., 253 F.3d 34 (2001). 74 See U.S. v. Microsoft Corp., 231 F. Supp. 2d 144 (2002). For an economics discussion of this case, see, for example, Rubinfeld (2014).

340   Lawrence J. White without a paradigm, plaintiffs will always argue for arbitrarily narrow markets, defendants will argue for arbitrarily broad markets, and the judiciary will have little basis for choosing among arbitrary alternatives. Further, the separation of hard-nosed, vigorous competition with respect to price and other attributes from predatory behavior, and the trade-offs between the greater efficiencies that can be brought by vertical restraints and the exclusionary effects of those restraints, will remain as contentious issues for antitrust economics and jurisprudence. In sum, monopoly and the dominant firm remain as an interesting area for antitrust.

References Adelman, Morris A. 1959a. A&P: A Study in Price-Cost Behavior and Public Policy. Cambridge, MA: Harvard University Press. Adelman, Morris A. 1959b. Economic Aspects of the Bethlehem Opinion. Virginia Law Review 45: 684–96. Amoroso, Luigi. 1930. La Curva Statica di Offerta. Giornale degli Economisti 45: 1–26. Translated as The Static Supply Curve, in International Economic Papers, ed. Alan T. Peacock, Wolfgang F. Stolper, Ralph Turvey, and Elizabeth Henderson, vol. 4 (London: Macmillan, 1954), 39–65. Amoroso, Luigi. 1938. Principii di Economica Corporativa. Bologna: Zanichelli. Areeda, Phillip E., and Herbert Hovenkamp. 2002. Fundamentals of Antitrust Law. New York: Aspen Law & Business. Areeda, Philip, and Donald F. Turner. 1975. Predatory Pricing and Related Practices under Section 2 of the Sherman Act. Harvard Law Review 88: 697–733. Ashenfelter, Orley, David Ashmore, Jonathan B. Baker, Suzanne Gleason, and Daniel S. Hosken. 2006. Empirical Methods in Merger Analysis:  Econometric Analysis of Pricing in “FTC v. Staples”. International Journal of the Economics of Business 13: 265–79. Asker, John, and Heski Bar-Isaac. 2014. Raising Retailers’ Profits: On Vertical Practices and the Exclusion of Rivals. American Economic Review 104: 672–86. Audretsch, David B., and John J.  Siegfried, eds. 1992. Empirical Studies in Industrial Organization: Essays in Honor of Leonard W. Weiss. Dordrecht: Kluwer. Bain, Joe S. 1951. Relation of Profit Rate to Industry Concentration: American Manufacturing, 1936–1940. Quarterly Journal of Economics 65: 293–324. Bain, Joe S. 1956. Barriers to New Competition. Cambridge, MA: Harvard University Press. Baker, Jonathan B. 1999. Econometric Analysis in “FTC v. Staples”. Journal of Public Policy & Marketing 18: 11–21. Baker, Jonathan B. 2008. Market Definition. In Issues in Competition Law and Policy, ed. Wayne D. Collins, vol. 1, 315–51. Chicago: American Bar Association. Baumol, William J. 1968. Reasonable Rules for Rate Regulation: Plausible Policies for an Imperfect World. In Prices: Issues in Theory, Practice, and Public Policy, ed. Almarin Phillips and Oliver E. Williamson, 101–23. Philadelphia: University of Pennsylvania Press. Baumol, William J., and Alfred G.  Walton. 1973. Full Costing, Competition and Regulatory Practice. Yale Law Journal 82: 639–55. Benston, George J. 1982. Accounting Numbers and Economic Values. Antitrust Bulletin 27: 161–215.

Monopoly and Dominant Firms   341

Bigelow, John P., and Robert D.  Willig. 2014. “Reverse Payments” in Settlements of Patent Litigation: Schering-Plough’s K-Dur (2005 and 2012). In The Antitrust Revolution: Economics, Competition, and Policy, 6th ed, ed. John E.  Kwoka Jr. and Lawrence J.  White, 213–45. New York: Oxford University Press. Borenstein, Severin. 1989. Hubs and High Fares: Airport Dominance and Market Power in the U.S. Airline Industry. Rand Journal of Economics 20: 344–65. Borenstein, Severin. 1992. The Evolution of U.S. Airline Competition. Journal of Economic Perspectives 7: 45–73. Bolton, Patrick, Joseph F. Brodley, and Michael H. Riordon. 2000. Predatory Pricing: Strategic Theory and Legal Policy. Georgetown Law Journal 88: 2239–330. Burnett, William B. 1999. Predation by a Nondominant Firm: The Liggett Case (1993). In The Antitrust Revolution: Economics, Competition, and Policy, 3rd ed, ed. John E. Kwoka Jr. and Lawrence J. White, 239–63. New York: Oxford University Press. Carlton, Dennis W., and Jeffrey M.  Perloff. 2005. Modern Industrial Organization. 4th ed. New York: Pearson. Chamberlin, Edward H. The Theory of Monopolistic Competition. Cambridge, MA: Harvard University Press, 1933. Dalkir, Serdar, and Frederick R. Warren-Boulton. 2014. Prices, Market Definition, and the Effects of Merger: Staples-Office Depot (1997). In The Antitrust Revolution: Economics, Competition, and Policy, 6th ed, ed. John E. Kwoka Jr. and Lawrence J. White, 166–93. New York: Oxford University Press. Demsetz, Harold. 1973. Industry Structure, Market Rivalry, and Public Policy, Journal of Law & Economics 16: 1–9. Demsetz, Harold. 1974. Two Systems of Belief about Monopoly. In Industrial Concentration: The New Learning, ed. Harvey J. Goldschmid, H. Michael Mann, and J. Fred Weston, 164–83. New York: Columbia University Press. Edlin, Aaron S., and Joseph Farrell. 2004. The American Airlines Case: A Chance to Clarify Predation Policy (2001). In The Antitrust Revolution, 4th ed, ed. John E.  Kwoka Jr. and Lawrence J. White, 502–27. New York: Oxford University Press. Elzinga, Kenneth G. 1999. Collusive Predation: Matsushita v. Zenith (1986). In The Antitrust Revolution: Economics, Competition, and Policy, 3rd ed., ed. John E. Kwoka Jr. and Lawrence J. White, 220–38. New York: Oxford University Press. Elzinga, Kenneth G., and David E. Mills. 2011. The Lerner Index of Monopoly Power: Origins and Uses. American Economic Review 101: 558–64. Elzinga, Kenneth G., and David E. Mills. 2014a. Predatory Pricing in the Airline Industry: Spirit Airlines v. Northwest Airlines (2005). In The Antitrust Revolution: Economics, Competition, and Policy, 6th ed., ed. John E. Kwoka Jr. and Lawrence J. White, 307–35. New York: Oxford University Press. Elzinga, Kenneth G., and David E.  Mills. 2014b. Resale Price Maintenance Wins a Reprieve: Leegin v. PSKS (2007). In The Antitrust Revolution: Economics, Competition, and Policy, 6th ed., ed. John E. Kwoka Jr. and Lawrence J. White, 435–57. New York: Oxford University Press. Farrell, Joseph, and Carl Shapiro. 2010. Antitrust Evaluation of Horizontal Mergers:  An Economic Alternative to Market Definition. B.E. Journal of Theoretical Economics: Policies and Perspectives 10 (1): article 9. Fisher, Franklin M., and John J. McGowan. 1983. On the Misuse of Accounting Rates of Return to Infer Monopoly Profits. American Economic Review 73: 82–97.

342   Lawrence J. White Fisher, Franklin M., John J. McGowan, and Joen E. Greenwood. 1983. Folded, Spindled, and Mutilated: Economic Analysis and U.S. v. IBM. Cambridge, MA: MIT Press. Fisher, Franklin M. 1987. On the Mis-use of the Profits-Sales Ratio to Infer Monopoly Power. Rand Journal of Economics 18: 384–96. Fisher, Franklin M. 2008. Detecting Market Power. In Issues in Competition Law and Policy, ed. Wayne D. Collins, vol. 1: 353–71. Chicago: American Bar Association. Giocoli, Nicola. 2012. Who Invented the Lerner Index? Luigi Amoroso, the Dominant Firm Model, and the Measurement of Market Power. Review of Industrial Organization 41: 181–91. Katz, Michael L. 2014. Exclusive Dealing and Antitrust Exclusion:  “U.S.  v.  Dentsply”. In The Antitrust Revolution: Economics, Competition, and Policy, 6th ed., ed. John E. Kwoka Jr. and Lawrence J. White, 488–517. New York: Oxford University Press. Kaysen, Carl. 1956. United States v. United Shoe Machinery Corporation: An Economic Analysis of an Anti-Trust Case. Cambridge, MA: Harvard University Press. Kaysen, Carl, and Donald F. Turner. 1959. Antitrust Policy: An Economic and Legal Analysis. Cambridge, MA: Harvard University Press. Keppler, Jan H. 1994. Luigi Amoroso (1886–1965): Mathematical Economist, Italian Corporatist. History of Political Economy 26: 589–611. Landes, William M., and Richard A. Posner. 1981. Market Power in Antitrust Cases. Harvard Law Review 94: 937–96. Lerner, Abba P. 1934. The Concept of Monopoly and the Measurement of Monopoly. Review of Economic Studies 1: 157–75. Levinson, Marc. 2011. The Great A&P and the Struggle for Small Business in America. New York: Hill and Wang. Lynk, William J. 1999. Tying and Exclusive Dealing: Jefferson Parish Hospital v. Hyde. In The Antitrust Revolution: Economics, Competition, and Policy, 3rd ed., ed. John E. Kwoka Jr. and Lawrence J. White, 342–63. New York: Oxford University Press. Mancke, Richard B. 1974. Causes of Interfirm Profitability Differences: A New Interpretation of the Evidence. Quarterly Journal of Economics 88: 181–93. Marvel, Howard P. 1982. Exclusive Dealing. Journal of Law & Economics 25: 1–25. Modigliani, Franco. 1958. New Developments on the Oligopoly Front. Journal of Political Economy 66: 215–32. Motta, Massimo. 2004. Competition Policy:  Theory and Practice. New  York:  Cambridge University Press. Noll, Roger G., and Bruce M.  Owen. 1994. The Anticompetitive Uses of Regulation:  United States v. AT&T (1982). In The Antitrust Revolution: The Role of Economics, 2nd ed., ed. John E. Kwoka Jr. and Lawrence J. White, 328–75. New York: HarperCollins. Ordover, Janusz A., and Robert D. Willig. 1981. An Economic Definition of Predation: Pricing and Product Innovation. Yale Law Journal 91: 8–53. Ordover, Janusz A., and Robert D.  Willig. 1999. Access and Bundling in High-Technology Markets. In Competition, Innovation, and the Microsoft Monopoly: Antitrust in the Digital Marketplace, ed. Jeffrey A. Eisenach and Thomas M. Lenard, 103–28. Boston: Kluwer. Pepall, Lynne, Dan Richards, and George Norman. 2008. Industrial Organization: Contemporary Theory and Empirical Applications. 4th ed. Malden, MA: Blackwell. Perloff, Jeffrey M., Larry S. Karp, and Amos Golan. 2007. Estimating Market Power and Strategies. New York: Cambridge University Press.

Monopoly and Dominant Firms   343

Pindyck, Robert S. 2014. Governance, Issuance Restrictions, and Competition in Payment Card Networks:  U.S.  v.  Visa and MasterCard (2003). In The Antitrust Revolution:  Economics, Competition, and Policy, 6th ed., ed. John E.  Kwoka Jr. and Lawrence J.  White, 602–26. New York: Oxford University Press. Preston, Lee. 1994. Territorial Restraints: GTE Sylvania (1977). In The Antitrust Revolution: The Role of Economics, 2nd ed., ed. John E.  Kwoka Jr. and Lawrence J.  White, 311–27. New York: HarperCollins. Rausser, Gordon C., and John Foote. 2014. Monopsony and Predatory Bidding in the Lumber Industry: The Weyerhaeuser Cases. In The Antitrust Revolution: Economics, Competition, and Policy, 6th ed., ed. John E. Kwoka Jr. and Lawrence J. White, 277–306. New York: Oxford University Press. Robinson, Joan. The Economics of Imperfect Competition. London: Macmillan, 1933. Rubinfeld, Daniel L. 2014. Maintenance of Monopoly: U.S. v. Microsoft (2001). In The Antitrust Revolution: Economics, Competition, and Policy, 6th ed., ed. John E. Kwoka Jr. and Lawrence J. White, 627–54. New York: Oxford University Press. Salop, Steven C., and David T.  Scheffman. 1983. Raising Rivals’ Costs. American Economic Review 73: 267–71. Salop, Steven C., and David T. Scheffman. 1987. Cost-Raising Strategies. Journal of Industrial Economics 36: 19–34. Saving, Thomas R. 1970. Concentration Ratios and the Degree of Monopoly. International Economic Review 11: 139–46. Stigler, George J. 1940. Notes on the Theory of Duopoly. Journal of Political Economy 48: 521–41. Stocking, George W., and Willard F. Mueller. 1955. The Cellophane Case and the New Competition. American Economic Review 45: 29–63. Sylos-Labini, Paolo. 1962. Oligopoly and Technological Progress. Cambridge, MA:  Harvard University Press. Telser, Lester G. 1960. Why Should Manufacturers Want Fair Trade? Journal of Law & Economics 3: 86–105. Tremblay, Victor J., and Carol H. Tremblay. 2012. New Perspectives on Industrial Organization: Contributions from Behavioral Economics and Game Theory. New York: Springer. Weiss, Leonard W. 1971. Quantitative Studies of Industrial Organization. In Frontiers of Quantitative Economics, ed. Michael D. Intrilligator, 362–403. Amsterdam: North-Holland. Weiss, Leonard J. 1974. The Concentration-Profits Relationship and Antitrust. In Industrial Concentration: The New Learning, ed. Harvey J. Goldschmid, H. Michael Mann, and J. Fred Weston, 184–233. New York: Columbia University Press. Weiss, Leonard W. 1989. Concentration and Price. Cambridge, MA: MIT Press. Werden, Gregory J. 2000. Market Delineation under the Merger Guidelines: Monopoly Cases and Alternative Approaches. Review of Industrial Organization 16: 211–18. Werden, Gregory J. 2003. The 1982 Merger Guidelines and the Ascent of the Hypothetical Monopolist Paradigm. Antitrust Law Journal 71: 253–76. Werden, Gregory J. 2006. Identifying Exclusionary Conduct under Section 2: The “No Economic Sense Test”. Antitrust Law Journal 73: 413–34. White, Lawrence J. 1989. The Revolution in Antitrust Analysis of Vertical Relationships: How Did We Get from There to Here? In Economics and Antitrust Policy, ed. Robert J. Larner and James W. Meehan, 103–21. New York: Quorum.

344   Lawrence J. White White, Lawrence J. 2008. Market Power and Market Definition in Monopolization Cases: A Paradigm Is Missing. In Issues in Competition Law and Policy, ed. Wayne D. Collins, 913–24. Chicago: American Bar Association. White, Lawrence J. 2010a. Economics, Economists, and Antitrust: A Tale of Growing Influence. In Better Living through Economics, ed. John J. Siegfried, 226–48. Cambridge, MA: Harvard University Press. White, Lawrence J. 2010b. The Growing Influence of Economics and Economists on Antitrust: An Extended Discussion. Economics, Management, and Financial Markets 5: 26–63.

CHAPTER 15

M A R K E T D E F I N I T ION LOUIS KAPLOW

The economics of market definition is a paradox—or, more precisely, an oxymoron. On one hand, market power is a central element of most competition rules throughout the world, and market definition is the most widely used method of assessing market power, taken by some to be mandatory in certain settings. On the other hand, the concept of market definition does not exist in industrial organization economics (Fisher 1987, 2008; Kaplow and Shapiro 2007; ten Kate and Niels 2009), and recent work suggests that it cannot be rationalized in a coherent, useful manner (Kaplow 2010, 2011a, 2011b). Accordingly, this chapter’s focus is conceptual.1 Section 15.1 makes explicit what is necessary if market definition is to be rendered a useful tool in illuminating market power and explains how the predicates are lacking or circular. Section 15.2 explores market share threshold tests, revealing how they conflate empirical questions and legal policy judgments. Section 15.3 specifically examines merger guidelines and the hypothetical monopolist test, showing how the method is inappropriate in principle and misleading in practice even under the best of conditions. Section 15.4 addresses problems

1  There exists substantial empirical work on the measurement of market power. However, given that market definition is not a conceptually coherent economic concept, there cannot really exist empirical work on market definition per se (and that which exists, not surprisingly, most often is really about market power itself). Some economists have suggested to me that market definition is perhaps a useful metaphor for guiding data collection, such as in determining which related products to include. This use of the notion, however, is qualitatively different from the use in competition law that is the focus here. Moreover, the analytical connection is not very close; as the analysis in section 15.1 implies, it would be difficult to use an incoherent means to guide empirical investigation, much less one that needs the output of such analysis as an input in applying the method.

346   Louis Kaplow arising from the common focus on specific substitutes and their cross-elasticities, which practice is induced by the felt need to define a market.

15.1.  On the (In)Coherence of the Market Definition Process Under most competition regimes, market power or changes therein play a central role.2 Exclusionary practices—for example, abuse of dominance in the European Union and monopolization in the United States—require a showing of significant market power as well as anticompetitive acts. Horizontal mergers are condemned when they sufficiently augment market power. And many horizontal arrangements—short of naked price fixing and related behavior—are carefully scrutinized and ultimately prohibited only if nontrivial market power is present. Taking as given the critical importance of market power, and the fairly widespread understanding that it refers to the ability to profitably elevate price, the question remains how best to measure it. The most common method in competition law involves defining the so-called relevant market and then making market power inferences from market shares in that market. This section presents two critical flaws that make the market definition/redefinition process useless or worse (Kaplow 2010, 2011a).

15.1.1.  Market Redefinition and Market Share To begin, it is necessary to ask how market definition and market shares relate to market power, which is defined as the degree to which price (P) may profitably be elevated above a competitive level, where price equals marginal cost (MC). In a direct sense, they are not related. However, measuring market power directly requires assessing marginal cost, which often is difficult in light of the need to allocate joint costs and to determine which costs are fixed versus variable in the pertinent time frame. Still, the difficulty of measuring a parameter does not ordinarily lead economists to measure something altogether different. The answer has to be found in models in which one can derive a relationship between market share and market power. The most familiar is one with a dominant firm facing a competitive group of fringe firms in a market for a homogeneous good. (A variant involving quantity competition in a homogeneous goods market is 2 

As the text indicates, sometimes the market power inquiry is addressed to levels and other times to changes (although changes themselves can be expressed in terms of two levels, before and after, say, a merger). For convenience, the language of levels will most often be employed, although section 15.3 on horizontal mergers focuses on changes. For further discussion, see Kaplow and Shapiro (2007).

Market Definition   347

examined in subsection 15.3.1.) The Lerner Index (L) in this case is determined by the formula (Stigler 1940; Landes and Posner 1981; Kaplow and Shapiro 2007): L=



P − MC 1 S = = . ε d + (1 − S)ε r P εf

(

)(

)

In this formula, the firm’s elasticity of demand is ε f = dQ / dP P / Q , where Q denotes the quantity of the firm’s output; the market elasticity of demand is ε d = dX / dP P / X , where X denotes total market demand; and the rivals’ collective elasticity of supply is ε r = dY / dP P / Y , where Y denotes rivals’ total supply. Finally, S is the dominant firm’s market share, and hence 1–S indicates rivals’ aggregate share. The first equality defines the Lerner Index: the fraction of the price that is in excess of marginal cost. The next is derived from the dominant firm’s profit-maximizing calculus: a higher price raises profits on retained sales but sacrifices profits on forgone sales. The optimal elevation is given by the inverse of the firm’s elasticity of demand, which indicates the rate at which the firm loses sales in all forms. The final equality decomposes the firm’s demand elasticity into the sum of the market elasticity and fringe firms’ supply elasticity, the latter weighted by their share. This summation reflects that sales may be lost to other products or to other suppliers of the same product. The intuitive relevance of market share to market power can be seen in this rightmost term: the S in the numerator is due to the fact that a given price increase (caused by a quantity reduction) is profitable to the dominant firm in proportion to its share of industry sales, and the 1–S in the denominator indicates that a higher dominant firm share makes the rivals’ share lower and hence reduces the impact of a given supply response (the elasticity measuring that response in percentage terms). Using the dominant firm’s share in the homogeneous products market, we can therefore determine its market power if we can ascertain the market elasticity of demand and the fringe’s elasticity of supply (or, with some techniques, Baker and Bresnahan 1988, we might simply eschew all this and measure the firm’s demand elasticity directly). It should be emphasized that this is so regardless of how many substitutes exist for the firm’s product or how strong those substitutes are. In other words, this formula gives us the answer we seek without having to undertake any of the standard market definition analysis. (This statement, in terms of product market definition, has analogues for geographic market definition, which might be viewed as pertaining to rivals’ supply response.) Having concluded that market redefinition is unnecessary, suppose next that we do it anyway. That is, suppose that there are some substitutes deemed to be sufficiently close that we should broaden the market to include them. If this step is undertaken, we then need a formula—a new formula—to infer market power in this now-heterogeneous goods market. But what is that formula? The short answer is that there is none. The above formula is valid only for markets with homogeneous goods. It has long been supposed that perhaps one could apply an

(

)(

)

(

)(

)

348   Louis Kaplow analogue to that formula for the broadened market, making suitable interpretations of the elasticities in the rightmost term—the market share being fairly straightforward, and lower, supposing that the dominant firm does not also produce the now-included substitute products. This strategy is pursued in Landes and Posner (1981), with criticisms advanced, for example, in Schmalensee (1982) and Kaplow (1982). The problem is that these common suggestions amount to little more than hand-waving. For the reinterpreted elasticities to be grounded, there needs to be a model that generates a formula that enables us to perform the necessary translation. As shown in Kaplow (2010), there exists a unique way to reinterpret these elasticities, which involves scaling them down to precisely offset the effects of the lower S in the formula. Put more bluntly, the only correct method involves undoing the market redefinition. This can be seen, for example, by taking the case in which the rivals’ supply elasticity is zero. In that event, if the redefinition reduces the share by, say, 72%, one would simply deem the market elasticity of demand in the combined market to be 72% less than that in the original, homogeneous goods market. Obviously, this fudge factor gives us the correct answer, and any other adjustment will not. But if this method is the only valid way to interpret elasticities in the combined market, there obviously is no point to ever redefining markets. (Note that the requisite demand elasticity correction makes no use of any information about demand in the broadened market.)

15.1.2.  Criterion for Market Definition To choose the best market definition presupposes a criterion for ranking alternatives. For concreteness, suppose that in some case the choice has come down to two definitions, which we will call Narrow and Broad. (More generally, whatever method is used to determine which of these two is superior would be used to eliminate the other alternatives.) Now, whichever of these two market definitions is chosen, we will then have to make a market power inference from the share in the chosen market. But we have just learned that such is not really possible unless we have stayed with the homogeneous goods market. For present purposes, set aside that obstacle and suppose that we do have some means of making a market power inference conditional on the market definition selected. Presumably, we infer a higher level of market power from the larger share in Narrow than we infer from the lower share in Broad. (Otherwise, there would be no point in debating which choice to make.) Moreover, to make the choice of interest, we will suppose that the truth lies somewhere between these two inferences. That is, if we choose Narrow, we expect that we will overstate market power at least somewhat, and if we choose Broad, we will understate power. Given that some error is inevitable in this choice, which one is best? As discussed in Kaplow (2010), the literature is essentially silent on the matter of the criterion, a remarkable observation in light of routine practice by economists, agencies,

Market Definition   349

and courts who have purported to define thousands of markets in the last half-century. The most plausible criterion that seems implicit in prior discussion is that the best market definition is that which involves the smallest error—that is, the smallest gap between the market power we will infer conditional on the chosen market and the truth.3 And, since we do not actually know the truth, the benchmark presumably is our best estimate (or, in many settings, guesstimate might better depict the situation). A moment’s reflection on this criterion and the methodology necessary to implement it reveals an overwhelming problem. We cannot determine which market definition involves a smaller error without first measuring the error from each choice. And to measure the error from either market definition involves determining the (absolute value of the) difference between the market power we would infer conditional on choosing that definition and our best estimate of market power.4 Hence, the first step in applying the criterion requires the formulation of our best estimate of market power. But if that is so, why are we not finished at that point? After all, the purpose of the market definition exercise is to enable an inference about market power. Yet, if our method of choosing the market definition presupposes that we already have a best estimate of market power, there is no further work to be done. Now, that best estimate may sometimes be a very good one and other times it may be rather feeble. In either case, our best estimate is still our best estimate. The market definition process, however, is worse than pointless. It was noted just above that some conscious error is inevitable in our choice of markets. But that error is not inevitable—it is entirely avoided—if we refuse to choose one of the two markets and just stick with our initial best estimate. After all, if we determined that Narrow was the best market definition—which is to say that the market power inference in that market overstates market power (relative to our best estimate) by less than the market power inference in Broad understates market power—we are still, knowingly, overstating market power. Likewise, if Broad is the best market definition, we know we are understating market power. Both choices involve acknowledged error, in addition to the extra effort required to complete the market definition task. This unnecessary error in market power inferences will sometimes translate into error in legal outcomes. For example, when we choose Narrow as best and thus overstate market power, we will sometimes prohibit a practice where the result would be exoneration if we had instead employed our best estimate. Likewise, when Broad is superior, we will sometimes allow practices that should have been condemned. There does exist a solution of sorts: one can make market definition the conclusion of the legal inquiry rather than an intermediate step. That is, one can use one’s best estimate of market power to determine whether it makes sense, say, to disallow a merger or 3 

The implicit use of a symmetric loss function is merely for expositional convenience. Throughout this discussion, the basis for the best estimate (guesstimate) is not addressed. Note that, depending on the context (e.g., initial screening versus adjudication) and available data, different techniques may be appropriate, and in all settings it may be relevant to consider a variety of information, including not only formal econometric analysis but also, for example, impressions of pertinent industry participants and documents internal to the firms under investigation. 4 

350   Louis Kaplow penalize the use of some alleged exclusionary practice. Once that is decided, one could, after the fact, choose the market definition that ratifies the decision. For example, if the decision is to condemn, one would choose Narrow. This patently circular method is superfluous, properly understood. But it does have important virtues. First, it does get the best answer, which the existing method may not. Second, for those agencies, economists, lawyers, or judges operating in a jurisdiction that is understood to require market definition,5 this tactic provides an out—a way to avoid error while formally adhering to the requirement. And we suspect that this way of approaching the problem is at least implicit in some actual practice. For example, some view the government’s strategy and the court’s opinion in the Staples merger case in this manner.6 Finally, if pushed to define a market—a theoretically incoherent enterprise— there is not a better way to proceed. In sum, the market redefinition process entails unnecessary work and produces outcomes that are inferior to those arising when it is eschewed entirely. This finding is, in a sense, convenient in light of the earlier point that there does not exist any valid way to make market power inferences from market shares in redefined markets anyhow—that is, unless one always undoes market redefinitions and reverts to the homogeneous goods market.7

15.2.  Market Share Thresholds Some competition law statutes, many government guidelines and policy statements, prominent court decisions, and legal treatises articulate market power requirements in terms of market shares. For example, for abuse of dominance in the European Union or monopolization in the United States, it might be stated that the dominant firm ordinarily must have a market share of at least 50%. And merger guidelines typically offer various market share thresholds—often in terms of HHIs, the sum of the squares of market

5  Kaplow (2010) points out that the basis for viewing market definition as legally mandatory is weaker in the United States than is commonly supposed. For example, courts (including the Supreme Court) have explicitly stated that market definition is a means to an end (which is determining market power), that direct means may be employed, and that when market definition is indeed used, the market shares must be interpreted in light of relevant economic considerations. Nor is it clear that market definition is compelled by the merger statute, Clayton Act Section 7, either on its face or as interpreted by the Supreme Court. In addition, mainstream legal commentators seem increasingly willing to forgo market definition, at least for mergers in markets with differentiated products (e.g., Areeda and Hovenkamp 2009, 84–88). 6  FTC v. Staples, Inc., 970 F. Supp. 1066 (D.D.C. 1997). 7  This logic applies as well to attempts to use market definition as a quick screen. For example, one still needs to define a market, and this act presupposes some market power guesstimate, which might be employed directly. In some cases, it might be suggested that a fairly broad market definition, in which resulting shares are trivial, is obviously correct. But this idea is tantamount to stating that the market demand or fringe supply elasticity in the narrow market is obviously extremely high, which directly implies that market power is low. That is, it is the obviousness of low market power that underlies the market definition conclusion, rendering the latter redundant.

Market Definition   351

shares—for when challenges have varying degrees of likelihood, including the case of essentially no likelihood (safe harbors). This standard practice is highly problematic. Conventional wisdom emphasizes that, for any such statements to be meaningful, one must look at the market shares in the relevant (best) market. However, the foregoing analysis shows this requirement to be impossible to satisfy in an economically sensible way. Moreover, even in what is regarded to be a properly defined market, Landes and Posner (1981), among others, have shown how a given share can imply a very wide range of market power levels because the elasticities in the above formula are not limited to single values or even a confined set of values. Kaplow (2011b) elaborates on the serious difficulties created for competition law. Everyone recognizes that market shares in a given market setting need to be interpreted, and that the guidance entailed by market share threshold tests should be understood as presumptions or rules of thumb—not rigid, sharp decision rules. The difficulty is that this acknowledgment only tells us what not to do; it fails to provide an affirmative prescription. This gap is much worse than meets the eye, for it is quite unclear what in principle is the direction of the proper adjustments, much less their magnitude. For example, consider a 50% presumptive required share for abuse of dominance. In a given case, the actual necessary share may be notably higher, perhaps 60% or 70%, whereas there might exist other cases where the requisite market power is implied by a share of 45% or 40%. Statements to this effect are ubiquitous, but it is extremely difficult to make them operational, for they all presume a benchmark that remains unspecified and is quite mysterious. To make sense of the preceding example, key hurdles must be overcome. First and foremost, one needs to know how much market power is taken to be conveyed by the 50% market share contained in the presumptive threshold.8 But the market power tests are not stated in terms of market power levels, and instead usually only in terms of market shares. We might impute some translation between market power and market share—what Kaplow (2010, 2011b) refers to as a standard reference market translation table—but the basis for doing so in practice is unclear. Perhaps the promulgators of the threshold had in mind the textbook “market.” But there is no such thing. Perhaps they meant the typical or average situation. But where do we suppose such figures could have come from? They would have had to specify a sample of markets (whatever that means), then engage in market definition since all agree that the shares are only meaningful in so-called relevant markets (but section 15.1 explains that such is infeasible), and finally determine the market power associated with different shares in each relevant market, even though in the actual markets only one situation prevailed. Only if these obstacles are surmounted can one know how to proceed in a given case. After all, if a firm under scrutiny argues that its 55% share is insufficient, because it conveys atypically low market power for such a share, we can only give meaning to its claim 8 

The text oversimplifies, for example, by supposing that the threshold would vary only with the market power associated with given market shares in different industries and not with the practices under scrutiny or other factors. In a sense, the illustration presents the easiest case for market share thresholds to work.

352   Louis Kaplow and assess its validity if we know what is the typical market power conveyed by a 55% share—and by the 50% share in our presumptive threshold. Note further that, as part of this effort, we would need to determine how much power this firm has, at which point (echoing section 15.1), one should naturally wonder why we are not finished—that is, once we know its actual market power, why should we be asking what market power is typically conveyed by a 55% share, a 50% share, or any other share? The present problem can usefully be illustrated, as in Kaplow (2011b), by considering Judge Hand’s famous pronouncement in the Alcoa case that ninety percent “is enough to constitute a monopoly; it is doubtful whether sixty or sixty-four percent would be enough; and certainly thirty-three per cent is not.”9 Consider three of the infinity of possible meanings of this dictum in terms of market power: Market Share and Market Power in Alcoa Market Share 33% 60/64% 90%

Market Power: Low

Market Power: Medium

Market Power: High

1% 3% 10%

5% 15% 30%

25% 50% 75%

Starting with the “Medium” column, Hand might have thought that a market share of 33% would have implied that Alcoa could elevate price only 5% above a competitive level, which is certainly insufficient (in his view) for the offense of monopolization; that a share of 60/64% would have implied a 15% elevation, which probably is not enough; and that a share of 90% implied a 30% elevation, which certainly is enough. In this case, the implicit market power legal threshold that rationalizes Hand’s opinion is, say, in the neighborhood of a 20% minimum elevation. But, looking at the “Low” column, Hand might have envisioned much smaller levels of market power being associated with the pertinent shares, with his implicit legal threshold perhaps closer to a 5% elevation. Or, examining the “High” column, he might have understood far greater market power levels to be involved with the stated shares, with the implicit legal threshold at 60% or so. Consider what this ambiguity means for Alcoa as a precedent. In a subsequent case, we know we are to interpret market shares in context, and that a given market share may mean more or less than usual. Suppose that the share is 55%, as above. Under Alcoa, who wins on this issue? The short answer is that we have no idea. Now, add that further analysis reveals the share to be associated, in the present case, with a 20% price elevation. Then, if Hand had in mind the “Low” column, there is far more than enough market power; if he meant “High,” there is not nearly enough; and if he meant “Medium,” it is a close call. This indeterminacy—and we have only considered three of the possible interpretations—raises insurmountable problems of application, whether for lawyers giving advice to businesses, agencies deciding when they have a plausible case, or adjudicators seeking to 9 

United States v. Aluminum Co. of Am. (Alcoa), 148 F.2d 416, 424 (2d Cir. 1945).

Market Definition   353

follow the precedent. This predicament seems insoluble, for there is no way to determine what Hand meant. One possibility would be to reanalyze the aluminum industry of over a half-century ago to figure out the actual market power implications of these shares—no easy task and, in any case, whatever we might determine using modern methods has no necessary connection to what Hand was thinking, and the latter is relevant to interpreting the precedent. Likewise for other means of reconstruction. The foregoing, unfortunately, is not merely a nitpick at Judge Hand. Just about any market share threshold in any statute, guideline, opinion, or text is subject to the same problems. (A partial exception is the hypothetical monopolist test in various merger guidelines, the subject of section 15.3.) Indeed, this was the problem at the outset in the example of a 50% presumptive market share threshold for abuse of dominance. By reference to the above table, such a threshold might implicitly refer to a minimum price elevation of roughly 2%, 10%, or 40%—or any other figure since those three columns hardly represent the only possibilities. Market share threshold tests are, on reflection, entirely uninformative about the implicit market power requirement. Moreover, they pose the aforementioned difficulty of how to determine whether the inference of market power in a given case involving, say, a 55% share should be adjusted at all in light of the circumstances, in which direction, and by how much. It seems we would have to know both how much actual market power exists in the present case and also how much market power is implicitly required. But once we knew all that, there would be no point in adjusting the 55% share upward or downward to then compare it with the 50% share in the test, for we would already know whether the market power in the case under consideration met the market power legal threshold. Just as market definition and redefinition has been a subject of extensive discussion and implementation for decades, yet we really cannot make economic sense of the notion, so too market shares and market share threshold tests have been at the center of competition law rules, debates, and applications in countless cases, but without any coherent basis. Regarding market shares, the problem can in part be viewed as one involving the possibility of communications. A market share threshold test is supposed to state to the world something about how much market power is required, but it does not. Frequent arguments that market shares in one or another case imply more or less market power than meets the eye assume that there is some level of market power ordinarily associated with a given market share, which is not so. More fundamentally, it is assumed that—however determined, even if by arbitrary convention—there is some degree of power that both speaker and listener have in mind, but there is not. Terms like “high” and “low” are inherently relative, and all market share discourse that is ultimately linked to market power presupposes a common benchmark, a translation table of sorts. Those who pronounce, criticize, and apply the law seem to be speaking the same language, all the while failing to communicate.10 10  To consider further the degree of this failure, consider administering a survey to economists with significant engagement in competition policy or to members of competition agencies or courts. It might ask: “How much market power (measured by the extent to which a dominant firm profitably elevates price above marginal cost) is typically associated with a market share of 50%?” Conversely, they could be asked: “What market share do you think is typically possessed by a dominant firm that can profitably

354   Louis Kaplow Another useful way to view the problem (Kaplow 2011b) is to recognize that a focus on market shares conflates two qualitatively distinct questions:  How much market power exists in a given case? And how much market power is necessary to meet the legal test? The first is an empirical question, the second a matter of policy. Moreover, market shares are not in the proper units to answer either question. The amount of market power in a given case refers to the degree of profitable price elevation. Stating that a firm has a 50% market share does not answer that question. Likewise, a market power legal threshold might indicate, say, for horizontal mergers, how much of a predicted price increase is necessary to trigger a challenge. Again, stating that the postmerger HHI must exceed 2500 and the increase due to the merger must exceed 200 does not answer the question. Obviously, it is impossible to respond to two distinct questions, one empirical and another about optimal policy, with a single answer. Market shares, however, answer neither one. Applied to Hand’s statement in Alcoa, we can ask whether his famous pronouncement meant to say something about how much market power existed in that case— which he had to determine to apply the legal test—or about how much market power is required in a monopolization case—which he needed to state in order to know whether the test was satisfied, and which must be taken to be his meaning if the case is to serve as a precedent. As just explained, and as illustrated above, he in fact answered neither. From his assertion, we have no idea how much market power Hand thought Alcoa possessed, and we have no clue how much market power he deemed to be legally necessary.

15.3.  Merger Guidelines and the Hypothetical Monopolist Test Agency assessment of horizontal mergers is one of the most important competition law domains, and market power inquiries are the key determinant of most decisions whether to block proposed mergers. Many jurisdictions have merger guidelines, often following the blueprint originally established in the United States involving application of the hypothetical monopolist test for market definition purposes:  for example, US Horizontal Merger Guidelines (2010), EU Guidelines on the Assessment of Horizontal Mergers (2004), EU Commission Notice on the Definition of Relevant Market (1997), EU DG Competition Discussion Paper (2005). In brief and rough terms, the method is as follows (with attention to product market definition for concreteness): One begins with the homogeneous goods market and asks whether a hypothetical monopolist in that market could profitably elevate price, say, 5% above current levels. If so, that is the relevant market. If not, one adds the next group of elevate price by 20%?” One can ponder how widely such answers would vary, if those surveyed were willing to answer at all. Or one could ask them instead how they think others in the target groups would respond to these two questions. Again, contemplate the predicted mean and variance of the responses.

Market Definition   355

substitutes and asks whether a hypothetical monopolist in that redefined, broader market could profitably elevate price 5%. If so, that is the relevant market. If not, one continues until the test is satisfied. Then, various guidelines typically examine the postmerger HHI in the resulting market as well as the increment to the HHI due to the merger in order to determine safe harbors, ranges of likely challenge, and so forth. For example, under the US Guidelines, a merger that raises the HHI by more than 200 points and results in an HHI above 2500 is presumed to be likely to enhance market power sufficiently to warrant a challenge. Subject to various qualifications (such as how many substitutes to add at each stage, and in which order), the hypothetical monopolist test provides a determinate market definition. But we already know from subsection 15.1.2 that it must be problematic. If it were the best market definition, it would still be inferior to direct use of a best estimate of market power. And since it is not the best—for example, it does not even attempt to determine which definition is closest to the truth—it is likely to be even worse. Moreover, we should understand that use of the resulting market shares is problematic. First, whenever the test does not stop with the homogeneous goods market, there is a conceptual problem with making inferences in the broader market. Second, all of the HHI thresholds are in the class of market share tests, which do not provide a basis for application in given cases since they are not denominated in market power terms. In addition, we should be wary of the hypothetical monopolist test on another ground: Why are we employing a hypothetical monopolist test to assess a merger? Unless we are considering a merger to monopoly, the test is patently posing the wrong question. We will now see how all these problems arise in concrete settings. To do so, it is useful to consider the three types of anticompetitive effects that have received the most emphasis in merger guidelines and commentary:  unilateral effects in homogeneous goods markets, unilateral effects in differentiated products markets, and coordinated effects (Kaplow 2011a).

15.3.1.  Unilateral Effects: Homogeneous Goods Beginning with unilateral effects in homogeneous goods markets, the immediate question is why one should ever go beyond the first stage of the hypothetical monopolist test. If it fails in round one, we expand the market, at which point it is no longer one with homogeneous goods, and the pertinent model is inapplicable. To elaborate, one of the few contexts—aside from that with a dominant firm and competitive fringe, examined in subsection 15.1.1—in which there is a model and resulting formula that uses market shares is the Cournot quantity-competition model with homogeneous goods. This formula indicates that the industry-wide average, output-weighted margin equals HHI/| εd | (where the HHI is represented in ten-thousandths, so that its range is from 0 to 1). See, for example, Ordover, Sykes, and Willig (1982) and Kaplow and Shapiro (2007)—and we will ignore the important qualifications in Farrell and Shapiro (1990) because they are not central for present purposes.

356   Louis Kaplow Now, if we really have Cournot quantity competition, this formula tells us that we can compute the price elevation as a function of the market shares (the HHI) in the initial, homogeneous goods industry. That is, if we stick with the original, homogeneous goods market, we can compute our price elevation: specifically, we can compare the elevation using the pre- and postmerger HHIs. But if we follow the hypothetical monopolist test when it tells us to redefine (broaden) the market, application of our formula is no longer valid and there is no other formula to use in its place. Hence, the standard merger guidelines method is the wrong one in this setting. It is also the case that the standard method does a poor job even of generating a consistent ordering of possible mergers: that is, whether or not it indicates that a challenge is appropriate is quite poorly correlated with the predicted price elevation. To see this, consider under the US Guidelines a merger in which the 5% hypothetical monopolist test is barely satisfied for the homogeneous goods market. Suppose further that the merger raises the HHI from 2300 to 2501, in which case it is deemed presumptively to pose a threat, as mentioned. The premerger elevation would roughly equal 0.23 divided by the market elasticity of demand, whereas a hypothetical monopolist, which we have supposed can raise price 5% over the prevailing level, has a Lerner Index equal to 1.0 divided by that elasticity. These relationships imply an elasticity of just above 16, a premerger elevation of about 1.42%, and a postmerger elevation of approximately 1.55%. Therefore, the merger-induced elevation is about 0.13%, that is, under two-tenths of one percent. By contrast, if the hypothetical monopolist in the homogeneous goods market can only raise price 4.9%, we are instructed to expand the market. Now, if the first set of substitutes that we now include involves far more revenue than does the original market, then even a merger to monopoly in the original homogeneous goods market would not trigger scrutiny. But we had just stipulated that such a merger would raise price 4.9%. Let us now combine these two results: the Guidelines method would presumptively condemn the first merger, which raises price by 0.13%, and exonerate the second, which raises price by 4.9%, which is to say, over thirty-five times more than does the first merger. This example provides two lessons. First, the hypothetical monopolist test is a poor means of attempting to assess the price effects of mergers. Other considerations, such as entry or efficiencies, may alter the outcomes. But to see if the test fundamentally makes sense, it is appropriate to consider a basic case in which such factors are held constant— even better, for simplicity, are irrelevant—and ask whether the Guidelines’ hypothetical monopolist test works at all well in even the simplest setting, where no complications are present. Unfortunately, it does not. Second, various merger guidelines’ use of HHI criteria illustrate the point in section 15.2 that it is a mistake to denote market power tests using market share thresholds. This mismatch helps to explain how a given set of HHI ranges can generate such inconsistent outcomes, condemning a merger raising price by 0.13% while allowing one that raises price by 4.9%. Interestingly, merger guidelines provide elaborate discussions of agencies’ methodologies in assessing price effects but do not ordinarily indicate what price effects would be sufficient to condemn a merger (Hausman and Sidak 2007). As a thought experiment, suppose that in a challenged merger, an agency’s experts and the merging

Market Definition   357

parties’ experts simultaneously filed reports, and it turned out that both predicted precisely the same price increase—say, 3.1%, or perhaps 1.4%, or 0.2%. Who would win in each instance? Merger guidelines are silent. Indeed, if they are followed conscientiously, one might argue that direct conclusions on price effects are not decisive. After all, these guidelines purport to call for market definition using the hypothetical monopolist test and assessment based on HHIs in the market that is thereby selected. The hypothetical monopolist test and the various market share thresholds indeed perform very poorly in homogeneous goods industries with quantity competition. But this problem is totally avoidable because the measure needed to apply the first round of the hypothetical monopolist test—the market elasticity of demand for the good in that market—is sufficient to directly assess the effect of the merger on price.11 Contemplating market redefinition is entirely unnecessary and only leads the analysis astray.

15.3.2.  Unilateral Effects: Differentiated Products Now consider unilateral effects when products are differentiated. In this particular setting, it has become fairly accepted that market definition is not very useful, as developed in the literature on critical loss analysis and upward pricing pressure (e.g., Baker and Bresnahan 1985; Harris and Simons 1989; O’Brien and Wickelgren 2003; Farrell and Shapiro 2010). To assess a merger of two firms that, let us suppose, each produce a single product, we can predict price elevation by knowing the firms’ margins on the two products and the diversion ratio between them. Simply put, after the merger, there is a stronger incentive to raise the price on each good because part of the lost sales are to the other product now controlled by the same entity. Because the cost of raising price is lessened, the motive to raise price increases.12 Importantly, determination of the relevant factors—the diversion ratios and the margins on the two products—does not require market definition. Accordingly, the hypothetical monopolist test and the guidelines’ various HHI thresholds are beside the point. One could attempt to employ them, but they hardly would provide a reliable guide. There is one way in which to avoid the problem: apply the hypothetical monopolist test to an initial market with only the two firms’ products and no others. And then stop. The first round of the test indicates how much a hypothetical monopolist would raise price. If it is the hypothetical “monopolist” consisting of the two merged firms and no others, that is our answer. Of course, to apply this variation of the test is to inquire directly into how much the merger will elevate price. Note also that it is important to stop, contrary to the dictates of the hypothetical monopolist test. If one instead expands the market, the HHIs there, matched against the guidelines’ thresholds, will not be very useful. 11  As mentioned, there are complications, which should be taken into account in practice but are orthogonal to the foundational problems with the hypothetical monopolist test. 12  The analysis implicitly assumes that the products are substitutes.

358   Louis Kaplow

15.3.3.  Coordinated Effects Finally, consider the possibility that a merger will enhance the likelihood of coordinated price elevation. In this event, we likewise want to stick with the homogeneous goods market. Price coordination usually involves products that are fungible, or nearly so—as suggested by theory (coordination becomes far more complex as the dimensionality of the problem increases) and by evidence (most prosecuted price-fixing cases involve homogeneous intermediate goods: see Connor 2007, 136–53; Harrington 2006, 98–102; Hay and Kelley 1974, 29–38). Therefore, to predict the price effects of the merger, we need to ascertain the degree to which the merger makes coordination more likely and how much successful coordination would increase price. The hypothetical monopolist test is at most applicable to the latter question, but again no redefinition is involved. In round one, we can ask how much a hypothetical monopolist of the homogeneous goods industry would elevate price above current levels. Once we have the answer, we are done, for our question is precisely how much price would be raised by coordination—the many firms acting as a hypothetical monopolist of that market. We still need to weight our result in terms of the merger’s increment to the probability and degree of success, but that is another matter. Again, observe that knowledge of how much a hypothetical monopolist of some broader, heterogeneous goods market could elevate price, and what are the HHIs before and after the merger in that market, is beside the point.13 Although the proper analysis in these three scenarios differs, some conclusions are common. First, redefining markets is inappropriate in all the cases. Second, the iterative hypothetical monopolist test in particular is incorrect in every instance. Third, the guidelines’ HHI thresholds are not useful either. This analysis of the hypothetical monopolist test and its standard use in merger guidelines serves two purposes. Most obviously, it suggests that current methods, if actually followed, are quite problematic and, relatedly, that if guidelines reforms de-emphasize market definition, that move is sensible at least in principle. Second, the discussion reinforces the conclusions in the first two sections. It has long been understood that the market definition process is imperfect, yet most have found it useful in light of the difficulty of the alternatives. Its complete failure to help— and its consistent tendency to lead us astray—in the realm of horizontal mergers, where the method has received the most attention and refinement, signals that the broader conceptual problems examined in the first two sections are important and insuperable.

13 

One might occasionally want to consider coordination across markets, if the added complexity might be surmountable. In such instances, however, we would wish to analyze the predicted price elevation achievable by those groupings of firms for which coordination was determined to be feasible— not by whatever grouping happened to be selected by the hypothetical monopolist test.

Market Definition   359

15.4.  Elasticities versus Cross-Elasticities In addition to the aforementioned defects with the market definition approach to making market power inferences, it is also useful to consider a problem that has long been understood (e.g., Kaplow 1982, Schmalensee 1982, Simons and Williams 1993, Werden 1998) but whose severity is not fully appreciated. When it is thought necessary to define a market, boundaries must be drawn; hence, particular products are either in or out, an all-or-nothing choice.14 Furthermore, it is common to determine whether a particular substitute is sufficiently close so as to be included by examining its cross-elasticity (or, in the language of many legal materials, interchangeability) with the central product of concern. Substitutes’ cross-elasticities do not, however, appear in the market power formula in subsection 15.1.1. Their relevance is due to their bearing on the magnitude of the market elasticity of demand, the first term in the denominator on the right side of that expression. The precise relationship is as follows: N

ε d ≡ ε11 = 1 + ∑



i =2

Ri ε . R1 i1

On the left side we have the market elasticity of demand, which for notational clarity in this context is helpfully restated as the elasticity of demand for the initial product (designated as product 1) with respect to a change in its own price. This own-price elasticity, in turn, equals 1 plus the revenue-weighted sum of the cross-elasticities of demand with all of the other N products.15 The revenue weight for a product i is the ratio of expenditures on that product (Ri) to expenditures on our initial product (R1), and the cross-elasticity (εi1) considered here is the percentage change in the quantity of product i as the price of product 1 is increased. Although substitutes’ cross-elasticities are relevant to the market elasticity of demand in our original market, there are a number of reasons that we should not ordinarily focus on cross-elasticities (Kaplow 2010). First, information on the overall demand elasticity is what ultimately matters; it subsumes the cross-elasticities for all other products, and information about it is often as easy or easier to obtain and as or more reliable than information on particular cross-elasticities. 14 

Again, analogous reasoning is applicable to geographic market definition. The explanation for the 1-plus term is that if, say, the demand for none of the other products changes when the price of product 1 rises, then the elasticity of demand for that product would be 1, not 0: since the same revenue continues to be spent on product 1, a 1% increase in price implies a 1% fall in quantity (literally so for infinitesimal changes). 15 

360   Louis Kaplow Second, focus on specific cross-elasticities is incomplete in that the cross-elasticities of all other products are relevant. Consideration of only the closest substitutes could omit a great deal. To illustrate, one might in one case include a single, close substitute and in another case omit a dozen more distant substitutes even if the latter collectively are more powerful. For example, if revenue weights were equal and the cross-elasticities of the dozen were each one-fourth that of the single, close substitute, then the dozen together would have three times the impact of the single, close substitute. In some instances there may not be a problem if the inclusion of close substitutes is employed merely to form a lower bound on the market elasticity of demand, which if sufficiently large would rule out anticompetitive concerns.16 Third, inclusion implies overstatement of the restraining force of the substitutes because even close substitutes generally are not perfect substitutes. This point raises the need to interpret market shares in the redefined market, which subsection 15.1.1 indicates is possible only by undoing the redefinition and returning to the homogeneous goods market. Fourth, the importance of the revenue weights is often implicitly ignored. Cross-elasticities indicate percentage changes in demand for a substitute, but the importance of a given percentage change depends on the base: hence the revenue weights. For example, a substitute with a cross-elasticity of 5 but a revenue weight of 0.1 has only half the impact of a substitute with a cross-elasticity of 1 but a revenue weight of 1.0. One type of mistake is that we might be led to exclude a product that has a fairly low cross-elasticity when its large revenue weight indicates that it imposes a strong restraint on price increases of the product in question. On the other hand, when we include products with large revenue weights, shares drop dramatically, which can lead to substantial understatement of market power, as illustrated by the second part of the unilateral effects example in subsection 15.3.1. Moreover, the magnitude of the resulting errors can be huge: the revenue weights in the above example differ by a factor of ten, but when one contemplates the range of goods and services in an economy, these weights can differ by many orders of magnitude. All these difficulties are avoided by focusing on the market elasticity of demand in the original market rather than attending to the cross-elasticities of demand for various particular substitutes. Observe that there is a close relationship between the market definition approach and the mistaken focus on cross-elasticities. If one thinks it necessary to define a market, and if that process is understood as a determination of which goods are inside the market and which are outside, then it seems necessary to inquire whether or not each plausible substitute is sufficiently close, which in turn leads to an inquiry into its cross-elasticity. In contrast, if the direct question of market power is kept front and center in the inquiry, then one is led to ask about the market elasticity of demand rather than to focus on particular cross-elasticities. The foregoing does not, however, imply that cross-elasticities are always irrelevant. Initially, there is the point at the outset that the market elasticity is an aggregate of the 16 

If some of the omitted products were strong complements, one would not obtain a lower bound.

Market Definition   361

cross-elasticities, and sometimes estimating some key cross-elasticities—and applying the revenue weights—will enable a useful lower bound estimate of the overall demand elasticity. Additionally, in certain settings particular cross-elasticities are useful. In predicting the price effects of differentiated products mergers, as discussed in subsection 15.3.2, one needs to know the diversion ratios between the merging firms’ products, which in turn is indicated by the (revenue-weighted) cross-elasticities (e.g., Farrell and Shapiro 2010). Relatedly, if an alleged exclusionary practice is directed at particular substitutes, its potential impact is similarly determined (Kaplow and Shapiro 2007). Therefore, cross-elasticities sometimes are useful, but none of these valid applications involve market definition.

15.5. Conclusion The market definition process is incoherent. In a redefined market, there is no valid way to infer market power from the market shares therein—that is, unless one reverses the market redefinition and reverts to the homogeneous goods market. And there is no sensible way to deem one market definition superior to another without saying which definition entails less expected error, and error can only be defined by reference to some best estimate of market power. Hence, one needs a best estimate of market power to define a market, whereas the whole point of the market definition exercise is to facilitate an inference of market power. Moreover, whatever inference one draws from the market shares in the market that is chosen differs from the best estimate that was needed to choose the market and hence can only lead to worse legal outcomes. The market definition approach is also associated with stating market power tests in terms of market share thresholds. This practice as well is conceptually flawed. Since even in a so-called relevant market any given market share can be associated with greatly differing levels of market power, market share thresholds constitute indeterminate legal tests. The hypothetical monopolist test for market definition and the HHI thresholds in merger guidelines reflect these deficiencies. In none of the basic settings—unilateral effects with homogeneous goods, unilateral effects with differentiated products, and coordinated effects—does the hypothetical monopolist test produce the desired result. In the first and third case, one best predicts price effects of mergers by sticking with the homogeneous goods market, regardless of what the hypothetical monopolist test dictates. For differentiated products, one focuses on the diversion between the products in question, which involves no market definition at all (unless one stipulates the market to be that consisting of the products covered by the merger, again without regard to the outcome of the hypothetical monopolist test). This test, and the guidelines’ HHI ranges, fail to indicate which mergers involve the greatest price effects even in the most basic settings.

362   Louis Kaplow The market definition process tends to produce a focus on particular substitutes and their cross-elasticities with the product in question, which emphasis is a byproduct of the deemed need to determine which products are inside and which are outside the relevant market. But it is the overall market elasticity of demand that determines market power; particular products provide only a partial picture; inclusion of inevitably imperfect substitutes can lead to the understatement of market power; and ignoring the widely varying revenue weights can produce highly misleading market power estimates. Cross-elasticities are sometimes important to consider, but not to define markets. This conceptual examination of the economics of market definition reveals its bankruptcy. The implication is that growing research in recent decades on direct means of inferring market power (e.g., Baker and Bresnahan 1988, 1992; Werden and Froeb 2008; Whinston 2006) should receive increasing attention and further development.

Acknowledgments I am grateful to the John M. Olin Center for Law, Economics, and Business at Harvard University for financial support. Subsequent to drafting this chapter, there have appeared some responsive essays to the body of work surveyed here. For references and discussion, see Kaplow (2012, 2013).

References Areeda, Phillip E., and Herbert Hovenkamp. 2009. Antitrust Law:  An Analysis of Antitrust Principles and Their Application. 3rd ed. Vol. 4. Austin, TX: Wolters Kluwer. Baker, Jonathan B., and Timothy F. Bresnahan. 1985. The Gains from Merger or Collusion in Product-Differentiated Industries. Journal of Industrial Economics 33:427–44. Baker, Jonathan B., and Timothy F. Bresnahan. 1988. Estimating the Residual Demand Curve Facing a Single Firm. International Journal of Industrial Organization 6:283–300. Baker, Jonathan B., and Timothy F.  Bresnahan. 1992. Empirical Methods of Identifying and Measuring Market Power. Antitrust Law Journal 61:3–16. Connor, John M. 2007. Price-Fixing Overcharges: Legal and Economic Evidence. Research in Law and Economics 22:59–153. European Union. 1997. Commission Notice on the Definition of Relevant Market for Purposes of Community Competition Law, O.J. (C 372). European Union. 2004. Guidelines on the Assessment of Horizontal Mergers under the Council Regulation on the Control of Concentrations between Undertakings, O.J. (C 31), 5–18. European Union. 2005. DG Competition Discussion Paper on the Application of Article 82 of the Treaty to Exclusionary Abuses. Farrell, Joseph, and Carl Shapiro. 1990. Horizontal Mergers: An Equilibrium Analysis. American Economic Review 80:107–26.

Market Definition   363

Farrell, Joseph, and Carl Shapiro. 2010. Antitrust Evaluation of Horizontal Mergers: An Economic Alternative to Market Definition. B.E. Journal of Theoretical Economics 10(1) (Policies and Perspectives), Article 9. Fisher, Franklin M. 1987. Horizontal Mergers:  Triage and Treatment. Journal of Economic Perspectives 1(2):23–40. Fisher, Franklin M. 2008. Economic Analysis and “Bright-Line” Tests. Journal of Competition Law & Economics 4:129–53. Harrington, Joseph E., Jr. 2006. How Do Cartels Operate? Foundations & Trends in Microeconomics 2(1): 1–105. Harris, Barry C., and Joseph J. Simons. 1989. Focusing Market Definition: How Much Substitution Is Necessary? Research in Law and Economics 12:207–26. Hausman, Jerry A., and J. Gregory Sidak. 2007. Evaluating Market Power Using Competitive Benchmark Prices Instead of the Herfindahl-Hirschman Index. Antitrust Law Journal 74:387–407. Hay, George A., and Daniel Kelley. 1974. An Empirical Survey of Price-Fixing Conspiracies. Journal of Law and Economics 17:13–38. Kaplow, Louis. 1982. The Accuracy of Traditional Market Power Analysis and a Direct Adjustment Alternative. Harvard Law Review 95:1817–48. Kaplow, Louis. 2010. Why (Ever) Define Markets? Harvard Law Review 124:437–517. Kaplow, Louis. 2011a. Market Definition and the Merger Guidelines. Review of Industrial Organization 39:107–25. Kaplow, Louis. 2011b. Market Share Thresholds: On the Conflation of Empirical Assessments and Legal Policy Judgments. Journal of Competition Law & Economics 7:243–76. Kaplow, Louis. 2012. Market Definition Alchemy. Antitrust Bulletin 57:915–52. Kaplow, Louis. 2013. Market Definition:  Impossible and Counterproductive. Antitrust Law Journal 79:361–79. Kaplow, Louis, and Carl Shapiro. 2007. Antitrust. In A. Mitchell Polinsky and Steven Shavell, Handbook of Law and Economics, vol. 2: 1073–225. Amsterdam: Elsevier. Landes, William M., and Richard A. Posner. 1981. Market Power in Antitrust Cases. Harvard Law Review 94:937–96. O’Brien, Daniel P., and Abraham L.  Wickelgren. 2003. A Critical Analysis of Critical Loss Analysis. Antitrust Law Journal 71:161–84. Ordover, Janusz A., Alan O.  Sykes, and Robert D.  Willig. 1982. Herfindahl Concentration, Rivalry, and Mergers. Harvard Law Review 95:1857–74. Schmalensee, Richard. 1982. Another Look at Market Power. Harvard Law Review 95:1789–816. Simons, Joseph J., and Michael A.  Williams. 1993. The Renaissance of Market Definition. Antitrust Bulletin 38:799–857. Stigler, George J. 1940. Notes on the Theory of Duopoly. Journal of Political Economy 48:521–41. ten Kate, Adriaan, and Gunnar Niels. 2009. The Relevant Market: A Concept Still in Search of a Definition. Journal of Competition Law & Economics 5:297–333. US Department of Justice and Federal Trade Commission. 2010. Horizontal Merger Guidelines. Werden, Gregory J. 1998. Demand Elasticities in Antitrust Analysis. Antitrust Law Journal 66:363–414. Werden, Gregory J., and Luke M.  Froeb. 2008. Unilateral Competitive Effects of Horizontal Mergers. In Paolo Buccirossi, Handbook of Antitrust Economics, 43–104. Cambridge, MA: MIT Press. Whinston, Michael D. 2006. Lectures on Antitrust Economics, Cambridge, MA: MIT Press.

CHAPTER 16

B I L AT E R A L M ON OP OLY Economic Analysis and Antitrust Policy RO GER D. BLAIR AND CHRISTINA DEPASQUALE

16.1. Introduction BILATERAL monopoly is a market structure in which there is a single seller (a monopolist) and a single buyer (a monopsonist). From an antitrust perspective, this is a perplexing market structure because it is in the parties’ mutual interest to cooperate in maximizing the available surplus, but then compete over the relative shares of that surplus. Moreover, the price agreed to by the monopolist and the monopsonist serves only as a mechanism for dividing the surplus and is simply irrelevant as a rationing device, as it does not influence the quantity transacted. The antitrust mischief that this may cause will be explored below. In this chapter, we examine the antitrust implications of bilateral monopoly. In section 16.2, we present the economic model of bilateral monopoly. In this section, we compare monopoly, monopsony, and bilateral monopoly. In particular, we focus on price, output, and social welfare. In section 16.3, we examine some complications for antitrust policy. In particular, unlawful monopoly may lead to monopsony and unlawful monopsony may lead to monopoly. This has serious implications for antitrust policy in Section 1 Sherman Act cases and Section 7 Clayton Act cases, which we examine in section 16.4. We close with some concluding remarks in section 16.5.

Bilateral Monopoly   365

16.2.  The Economics of Bilateral Monopoly A bilateral monopoly exists when a single supplier of a well-defined product has only one customer. In this setting, the single customer is a monopsonist, while the single seller is a monopolist. Together, they form a bilateral monopoly. The monopolist’s effort to maximize profit by restricting output and thereby raising the price is confronted by the monopsonist’s effort to maximize profit by restricting purchases and thereby lowering the price. As a consequence, there is something of a problem in identifying the equilibrium price and quantity. First, consider the usual monopoly problem. A  single seller faces a large number of buyers with no market power. In figure 16.1, the negatively sloped demand curve is labeled D. The marginal revenue associated with the demand curve is denoted as MR. For the analysis of the monopoly problem, the firm’s marginal cost is represented by S in figure 16.1. Profit maximization will lead the monopolist to sell Q1 units of output at a price of P1. Next, consider the usual monopsony problem. A single buyer faces a large number of competing sellers with no market power and a positively sloped supply curve, which is labeled S in figure 16.1. The marginal expenditure associated with the supply is shown as ME. Ordinary profit maximizing behavior of the monopsonist will lead to purchases of Q1 units of output and a price of P2. The quantities are the same in both cases because we constructed the demand and supply curves to yield this result in order to reduce clutter in figure 16.1.

Price

a

ME

S

d P1 P3

P2

b

c

D

MR Q1

Q2

FIGURE  16.1  A  Standard

Quantity

Model of Monopsony

366   Roger D. Blair and Christina DePasquale Complications arise when one combines these two models and attempts to infer the bilateral monopoly solution. It is obvious that the usual marginal analysis fails us here. In our somewhat contrived illustration, the optimal output is the same for the monopolist and for the monopsonist, but the corresponding price cannot be P1 and P2 at the same time.

16.2.1.  The Bilateral Monopoly Solution The theory of bilateral monopoly has a rich history that can be traced to the writings of Cournot (1838) and Menger (1871). In 1928, Bowley offered the theoretically correct solution. Bowley (1928) recognized that there is a profit motive for cooperation to emerge between the upstream monopolist and the downstream monopsonist. This profit incentive arises because joint profits are not maximized at either the monopoly solution or the monopsony solution. The two trading partners can do better, but they must cooperate with one another to do so. Such cooperation may take the extreme form of vertical integration, that is, the joint ownership of the upstream and downstream entities. Alternatively, this cooperation may emerge from the bargaining process. For the latter, it is important to realize that the negotiation involves the optimal quantity in the first stage. That is, agreement on the quantity is essential if joint profits are to be maximized.1 After maximizing the joint profits, the parties can then negotiate over the shares.

16.2.2.  A Simple Model of Bilateral Monopoly Suppose that the final good (Q ) is produced according to a production function:

Q = Q ( x, y ) , (16.1)

where x is the intermediate good traded under bilateral monopoly conditions and y is another input that is supplied competitively at a constant price of p y . The total cost of producing the intermediate good x is given by C ( x ) . If the intermediate good monopolist and its sole customer were vertically integrated, the profit function of the integrated firm would be

Π = P (Q ) Q ( x, y ) − C ( x ) − p y ⋅ y, (16.2)

1  Machlup and Taber (1960) made this point and agreed that a failure to recognize this essential difference between bilateral monopoly and all other market structures created a good deal of analytical confusion.

Bilateral Monopoly   367

where P = P (Q ) is the final good demand function. The firm will maximize its profits by selecting quantities of x and y such that the first partial derivatives of (16.2) vanish:

dP  dQ dC ∂π  = 0 (16.3) =  P +Q  dQ  dx dx ∂x

and

dp  dQ ∂π  − p y = 0. (16.4) = P +Q dQ  dy ∂y 

In other words, integrated profits are maximized where the marginal revenue products of the inputs are equal to their respective marginal costs. It should be noted that this is the only level at which the joint profits are maximized. Now suppose that the single buyer and single seller had not vertically integrated. These firms will conduct arm’s-length negotiations on the price of x (q ) and the quantity of x. But these firms are not assumed to be aware of their circumstances. Bowley (1928) argued that under these market conditions, the negotiations will inexorably lead to the very same employment decisions as those that a vertically integrated firm would reach. The same joint profit-maximizing quantities of inputs x and y will be exchanged. This, of course, will lead to the same quantity of the final good. Thus, a determinate quantity of the intermediate good will be transacted. Moreover, the price and quantity of the final good will also be determinate. The only thing that remains indeterminate is the price of x, but this is not allocatively significant because this price does not influence the purchase of x. The price of x is just a means of dividing the jointly maximized profit. To see these results graphically, we turn to figure 16.2 and employ a special case of the model presented above. Suppose that one unit of an intermediate good ( x ) is transformed into one unit of final good (Q ) . The demand for the final good is downward sloping and the demand for the intermediate good is derived from the demand in the final good market. Assuming that the average cost of transforming one unit of input x into one unit of output Q is constant, we represent the average net revenue as a function of the quantity of x employed and label it D in figure 16.2. With monopoly in the sale of Q , the hypothetical derived demand for x is the curve marginal to D − C T , which is labeled Dx in figure 16.2. In other words, Dx is the net marginal revenue product of the input x. Now, MR x is the curve marginal to Dx and represents the marginal revenue associated with selling this intermediate good to a final good producer that has monopoly power, but no monopsony power. Turning to the cost curves, AC x denotes the upstream monopolist’s average cost of producing input x, and MC x is marginal cost. If the supplier of x were to behave as a pure competitor, MC x would correspond to its supply curve. Then, if the downstream

368   Roger D. Blair and Christina DePasquale Price

MEx

MCx ACx

PU1

PD2 DQ- CT MRx X1

MRPx= Dx X2

FIGURE  16.2  A  Special

Quantity

Case of Bilateral Monopoly

monopsonist were hiring this input from such a competitor, ME x would be the marginal expenditure of the input. From equation (16.3), we know that the vertically integrated firm would maximize its profits at a quantity of x2 where the marginal cost of producing x (MC x ) equals the marginal revenue product of x (MRPx ). In essence, the sum of producer and consumer surplus is maximized at that quantity. In the absence of vertical integration, that is, in a case of bilateral monopoly, the quantity x 3 will maximize the joint profits. One would expect the parties to agree on this quantity of the input. The price of the input is indeterminate. There are limits on the indeterminacy of the input price. We can bound the price range by considering the all-or-none demand schedule (DA ) and the all-or-none supply schedule (S A ) shown in figure 16.3.2 At quantity Q1, the height of the all-or-none demand (P2 ) is the maximum price the monopolist can command. At that price (P2 ), all of consumer surplus has been extracted by the monopolist. Analogously, at Q1 , the height of the all-or-none supply (S A ) is the lowest price since that price ( P3 ) allows the monopsonist to extract the entire producer surplus. These maximum and minimum prices determine the bargaining range. The competitive price is in that range, but there is nothing that particularly recommends that price as a solution.3

2  The all-or-none demand schedule represents price-quantity combinations that leave no consumer surplus. Analogously, the all-or-none supply schedule depicts price-quantity pairs that yield no producer surplus 3  In this specific example, the competitive price divides the surplus evenly, which is consistent with the Nash bargaining solution.

Bilateral Monopoly   369

Price

S

a

SA

P2

b

P1

P3 DA D

c Q1 FIGURE  16.3  Limits

Quantity

on the Indeterminacy of the Input Price

Although the actual price is indeterminate, there are some likely price effects. If buyers join forces in the presence of a monopoly, one would expect price to fall below P2 since the purpose of joining forces is to get a better deal. Analogously, if competing sellers merge in response to monopsony, the price is apt to rise above P3 —again because the firms merged in pursuit of a better deal. These price movements, however, have no competitive significance since quantity does not respond to such price changes. The price movement does have distributive consequences, but these have no impact on social welfare.4

16.2.3.  Formula-Price Contracts in Bilateral Monopoly Bilateral monopoly creates a need for repeated, possibly protracted, and costly negotiations over the price and quantity of the intermediate product.5 Under these circumstances, vertical integration is attractive because it eliminates these costs and, therefore, increases the combined profits of the two firms. Although these enhanced profits provide a powerful incentive for vertical integration, ownership integration may not be an attractive alternative for several reasons.6 First, the integrated operation 4  There are, of course, many complicated market structures that have some elements of bilateral monopoly, but results that may deviate from those of pure bilateral monopoly. See, e.g., Cremer and Riordan (1987), Horn and Wolinsky (1988), Chen (2003), and Symeonidis (2010). 5  This section digresses from the primary focus of whether there should be a “bilateral monopoly” defense. The reader may skip to section 16.3 in order to rejoin the primary theme of the chapter. This section depends heavily upon Blair and Kaserman (1987). 6  Coase (1937), examined this in some detail. This classic paper continues to be worth reading for its insights.

370   Roger D. Blair and Christina DePasquale may experience managerial diseconomies as spans of control are extended in the face of bounded rationality on the part of decision-makers.7 Second, capital costs may increase if lenders and investors cannot be convinced that sufficient expertise exists for the efficient operation of both stages simultaneously.8 Finally, even in the absence of the above diseconomies, successful integration must entail considerable negotiation over the price of the acquired firm and the various ancillary terms that accompany the usual merger agreement. Consequently, it is not surprising that firms confronted with the problem of bilateral monopoly might seek contractual alternatives to ownership integration. The economic literature on contractual forms of vertical integration has focused on a model in which an upstream monopolist sells its output to a competitive downstream industry that employs that input in variable proportions.9 Consequently, this literature has not focused on the case in which bilateral monopoly exists. Historically, the only contractual alternatives that were found in the literature were either (1) a long-term contract or (2) a series of short-term contracts that specify both the price and the quantity of the intermediate good to be exchanged in the case of bilateral monopoly. But these particular contractual arrangements are distinctly unappealing alternatives to ownership integration for two reasons. First, a long-term contract on both price and quantity fails to allow the necessary adjustments to occur when changes in cost or final output demand are experienced. Second, a series of short-term contracts requires repeated negotiations in the face of small numbers bargaining and clear opportunistic incentives. As a result of these various problems, economists have tended to conclude that ownership integration is typically the preferred solution in most cases of bilateral monopoly. Blair and Kaserman (1987) have proposed a formula price contract with several desirable properties. First, it leads each firm to pursue independent profit-maximizing policies that will result in maximum joint profits. Second, it allocates a specified share of these maximized joint profits to each party to the contract. Finally, it automatically adjusts to changes in production costs and final output demand. The purpose of the formula price is to share the maximized profit between the monopolist and the monopsonist. Suppose that each unit of the final product (Q ) requires one unit of the intermediate good ( X ), which is subject to the bilateral monopoly. In the absence of vertical integration, the upstream monopolist’s profit function ( Π U )  is Π U = w ( X ) X − C x ( X ) X,



7 

(16.5)

For a useful analysis, consult Williamson (1973). On this point, see Williamson (1974). 9  For an extensive (albeit somewhat dated) survey of this literature, see Blair and Kaserman (1983). In addition, see ­chapters 14–21 in Blair and Kaserman (2009). 8 

Bilateral Monopoly   371

where w(X) is the price of the intermediate good and C x (X) is the average cost of producing X. The downstream monopolist’s profit function will be

Π D = P (Q ) Q − w ( X ) x − C T (Q ) Q, (16.6)

where P (Q ) is the price of the final good and C T (Q ) is the average cost of transforming one unit of X into one unit of Q. It is useful to examine the profit function if the firms were vertically integrated. If these two firms were vertically integrated, the profit function of the combined operation would be

Π I = P (Q ) Q − C x ( X ) X − C T (Q ) Q. (16.7)

Absent vertical integration, the firms would like to sign a long-term agreement that will generate combined profits equal to the maximized value equation (16.7), which we denote as Π *I, and assign shares of these profits equal to α and 1 − α to the upstream firm and downstream firm, respectively, where 0 ≤ α ≤ 1. Obviously, such an assignment of profit shares must occur through the price of the intermediate product that is determined by the contract. Setting Π U = αΠ I and solving for w, we obtain

w = α (P − C T ) + (1 − α ) C x . (16.8)

Thus, if the upstream monopolist can ensure that the price of the intermediate product is determined by final output price and production costs at both stages through the formula given in equation (16.8), it will be assured of receiving α of the profits available to a vertically integrated monopolist. We can rearrange (16.8) as

w = C x + α (P − C T − C x ) .

Thus, the formula price for the intermediate product is equal to the average cost of producing the product plus α times the optimal integrated monopoly markup over cost at the final good stage. By substituting equation (16.8) into equation (16.6), we find that         Π D = PQ − α (P − C T ) + (1 − α ) C x  X − C T Q          = (1 − α ) Π I . That is, the formula price contract described in equation (16.8) also automatically assigns 1 − α of the integrated profits to the downstream firm.

372   Roger D. Blair and Christina DePasquale Therefore, under the terms of the formula price contract of equation (16.8), Π U = αΠ I and Π D = (1 − α ) Π I. As a result, independent profit maximization by the two nonintegrated firms will lead to combined profits of Π *I. The bilateral monopoly contract negotiation problems are greatly reduced because such negotiation can focus entirely on the value of α . All changes in demand and cost are accommodated automatically in the formula. Thus, the formula price contract provides an attractive alternative to ownership integration in the presence of bilateral monopoly.

16.3.  Complications and Antitrust Policy There are a variety of complications for antitrust policy related to bilateral monopoly. First, there is an old adage that “two wrongs don’t make a right.” In most cases, this is sound advice, but it may not be so here. Consider the case of an unlawful monopolist. Buyers will have an incentive to band together to create a bilateral monopoly. This buyer collusion could raise antitrust concerns that are misplaced. To the extent that the formation of countervailing power improves social welfare, the effort to create countervailing power should go unimpeded. If, however, the monopoly is actually an unlawful cartel, then it should be challenged so countervailing power becomes unnecessary. Similarly, competitive sellers may band together in the face of an unlawful monopsony. This collaboration might well be challenged under Section 1 of the Sherman Act, but would be procompetitive. Again, the appropriate antitrust policy would be to attack the unlawful monopsony so that countervailing power is unnecessary. A related complication arises when we have a dominant seller that falls short of full-fledged monopoly or a dominant buyer that approaches, but does not reach, monopsony status.10 In either case, the consolidation of power on the other side of the market offends Section 1 of the Sherman Act (15 U.S.C. §1) that prohibits contracts, combinations, and conspiracies that restrain trade. In this case, however, the agreement does not restrain trade. The buyers do agree not to compete with one another in buying the monopolized product. But the agreement leads to an expansion of quantity transacted. Consequently, the buyer agreement appears to be procompetitive. Irrespective of what happens to the negotiated price (which would probably fall), social welfare increases as the sum of consumer and producer surplus increases. It should be obvious, however, that the appropriate public policy should be to attack the unlawful monopoly. A somewhat more subtle problem arises when a concentrated oligopoly behaves like a monopolist, but there is no evidence of collusion. According to the folk theorem, the

10  The economic analysis of this variant of the bilateral monopoly model extends beyond the scope of this chapter.

Bilateral Monopoly   373

oligopolists may settle on the monopoly solution without overt collusion. In the case of tacit collusion, there is no obvious antitrust remedy. In that event, a buyer cartel will be procompetitive and the appropriate antitrust response may well be one of benign neglect. Unlawful monopsony leads to symmetric problems. As competing sellers are exploited by an unlawful monopsonist, there is an incentive to form a producer cartel to offset the monopsony power. Such a cartel would run afoul of Section 1 of the Sherman Act, but would actually be procompetitive. Again, the colluding sellers would not compete on price, but their conduct would increase output and improve social welfare. Rather than condoning a producer cartel, antitrust enforcement should be aimed at the unlawful monopsony.11 Second, in some contexts, the formation of countervailing power may be seen as an antitrust violation, when it should not be because the results are procompetitive. The presence of lawful monopoly and competitive buyers creates an incentive for the buyers to enter into a horizontal agreement to negotiate with the monopolist. This horizontal agreement may be seen as a Section 1 violation because the buyers agree not to compete with one another by offering to pay a higher price. This agreement, however, is procompetitive since the quantity transacted will increase and therefore the agreement does not restrain trade or commerce. Unlike the unlawful monopoly, the lawful monopoly is immune from antitrust prosecution. Consequently, the emergence of countervailing power through a horizontal agreement among the buyers is socially desirable. Accordingly, antitrust prosecution would seem to be unwarranted. Much the same thing can be said about a lawful monopsony creating an incentive among competing sellers. As an antitrust policy matter, ignoring the formation of countervailing power appears to be socially desirable, but there are several risks associated with this policy, which we examine below. There is a risk associated with the formation of countervailing power in response to a lawful monopoly. Firms who buy the same inputs very often then sell their output in competition with each other. Thus, the cooperation among these firms as buyers may spill over into cooperation as sellers of their output. Consequently, the net social welfare of permitting their collusion as buyers could be negative. It is important to note that the possibility that these buyers may become price-fixing sellers can quickly offset the gains made possible by allowing the monopolist and the collusive monopsony to determine the joint profit-maximizing level of output. Once the firms begin to act as a collusive monopoly, they will raise prices and restrict output and consequently demand less of the input. The increase in consumer welfare, which was the justification for permitting the buyers to cooperate in the first place, could be quickly eliminated. More subtle is the selling price stability that may result if the parties do not agree on selling price but price individually while knowing the price paid by competitors for one or more inputs. Of course, it is always possible that input cost uniformity will lead to uniformity in the pricing of outputs. And, in industries in which production techniques and input mixes are standardized, one would not expect a great variety in price. This 11  The folk theorem is relevant here. Absent proof of illegal collusion among buyers, antitrust enforcement is problematic.

374   Roger D. Blair and Christina DePasquale would be true even in highly competitive markets. On the other hand, if the market is oligopolistic and these buyers cooperate, the knowledge that they all have paid the same amount for a major input can reduce competitive pressure. An extension of the idea of collusion to create bilateral monopoly would be to permit a merger of all the competitors. Campbell (2007) makes the case that competitive sellers should be permitted to merge to monopoly if they confront a pure monopsonist. He argues that such a merger would result in bilateral monopoly, which is more efficient than the prior market structure. The same, of course, can be said for merging to monopsony to confront a monopoly. In either event, the emergence of countervailing power improves efficiency. At least in the case of merging to monopoly, Campbell argues that the merger would be consistent with merger policy as it enhances consumer welfare. In the specific circumstances analyzed, Campbell’s results are theoretically correct. It is, however, a risky policy choice. Consider the case of a merger to monopsony that creates a bilateral monopoly. Whatever gains are produced by a merger to monopsony will be lost should there be new entrants on the selling side of the market so the monopsonist no longer faces a monopolist. In addition, the formation of a monopsony may increase concentration in the markets in which the monopsonist sells its output and, thus, offsets some or all of the gains. Moreover, since these losses would be the result of price increases by a single firm, an antitrust response would be unlikely. Campbell’s proposal provoked a response from Baker, Farrell, and Shapiro (2008). First, they challenged the empirical relevance of Campbell’s analysis since there are precious few examples of pure monopsony. Based on this, they argue that their more complete theoretical analysis yields results that differ from Campbell’s results. In addition, they examine instances of bilateral monopoly and conclude that the results predicted by Campbell are not found. Finally, they challenge Campbell’s interpretation of merger policy under Section 7 of the Clayton Act. Neither Campbell, nor Baker, Farrell, and Shapiro affect our analysis and conclusions. With the possible exception of a few relatively rare situations, promoting competition is always better than accommodating monopoly power or monopsony power by permitting consolidation on the other side of the market. If the lawful monopoly or monopsony falls apart due to entry or technological change, cartels can be dismantled easily, whereas mergers are more permanent. Thus in weighing the relative risks of cartels and mergers, cartels would appear to pose a lower competitive risk.

16.4.  Antitrust Policy Antitrust policy is captured by statutes and their judicial interpretation. In this section, we turn our attention to antitrust policy regarding countervailing power. First, we examine the judicial attitudes towards the creation of countervailing power through collusion among either buyers or sellers. Second, we analyze the antitrust policy surrounding mergers that create countervailing power.

Bilateral Monopoly   375

16.4.1.  Countervailing Power through Collusion An early example of a countervailing market power rationale can be found in the 1933 case Appalachian Coals v. United States (288 U.S. 244), in which 137 producers of bituminous coal combined to create an exclusive joint selling agency. The central objective of the agency was to sell the coal at the “best price obtainable.” One focus of the joint effort was to control the supply of so-called distress coal—a byproduct of the production of the more desirable sizes and shapes of coal. The Supreme Court reversed a lower court holding that the actions of the producers violated Sections 1 and 2 of the Sherman Act. In what was obviously a Depression era opinion, the Court started with the premise that “[t]‌he mere fact that the parties to an agreement eliminate competition between themselves is not enough to condemn it.” The Court noted both the “deplorable” conditions of the industry and the presence of “organized buying agencies and large customers” possessing “concentrated buying power” and creating a “buyers’ market.” Although hardly resting its opinion on any express application of a countervailing market power theory, industry conditions and particularly the imbalance of market power clearly influenced the Court’s decision. Countervailing power has also played a prominent role in the labor antitrust exemption, which has its statutory origins in the Clayton Act and the Norris-Laguardia Act. The exemption for labor unions was designed to permit the formation of countervailing power for labor in negotiating with large employers. The labor union exemption would be meaningless if it did not exempt the collective bargaining agreements between unions and management. Although there is no statutory exemption for such an agreement, a nonstatutory exemption has emerged through judicial interpretation. In some instances, the labor unions had far more power than the individual employer and, consequently, there was an incentive for employers to form multiemployer bargaining units. Ironically, small employers found it necessary to create countervailing power of their own through the formation of a multiemployer bargaining unit. Consequently, bilateral monopoly emerged with monopoly labor unions confronting multiemployer bargaining units that enjoy monopsony power. The resulting collective bargaining agreements are covered by the nonstatutory labor exemption.12 A third illustration of the recognition of countervailing market power can be found in Balmoral Cinema v. Allied Artists Pictures Corp. (885 F. 2d 313), a case dealing with the commonly analyzed matter of motion picture “split” agreements. The split agreement involves motion picture exhibitors who agree not to compete against each other for the rights to exhibit specific films. In fact, they “split” the market, with one exhibitor having the exclusive rights to “bid” on a particular film while its partner in the split has the right to bid on another film. From the standpoint of monopsony, such an understanding confers buying power on the exhibitor with the exclusive bidding rights. In Balmoral, the United States Court of Appeals for the Sixth Circuit held that such an agreement was not per se unlawful. In so holding, the court expressly relied 12  The judicial rationale for this extension is largely based on the fact that efforts to amend the Taft-Hartley Act to outlaw multiemployer bargaining were unsuccessful.

376   Roger D. Blair and Christina DePasquale upon a monopsony-based countervailing market power rationale. It reasoned that “Exhibitors, as purchasers of films, may be justified in combating the market power of film suppliers by group action. Such action may lower prices to moviegoers at the box office and serve rather than undermine consumer welfare.” In cases where splits create a bilateral monopoly, this reasoning is correct since quantity expands and prices will fall as a result. If, however, splits simply create monopsony power, the results will be socially undesirable.

16.4.2.  Countervailing Power through Mergers In general, the antitrust policy towards mergers is dictated by Section 7 of the Clayton Act and the enforcement policies of the Antitrust Division of the Department of Justice and the Federal Trade Commission. The language of Section 7 of the Clayton Act makes it clear that it is a prophylactic effort to prevent a change in market structure that reduces competition. More specifically, Section 7 provides that No person engaged in commerce or in any activity affecting commerce shall acquire, directly or indirectly, the whole or any part of the stock or the whole or any part of the assets of another person engaged also in commerce or in any activity affecting commerce, where in any line of commerce in any section of the country, the effect of such acquisition may be substantially to lessen competition, or to tend to create a monopoly.

The central concern appears to be with concentration on the selling side, but the competitive consequences of buyer concentration has been recognized and accorded similar treatment. In either event, Section 7 is intended to preserve a competitive market structure. The presumption is, of course, that such a structure yields the most desirable outcomes for consumers. Given this presumption, there does not appear to be much room for considerations of countervailing power, but there should be in some circumstances. Merger to monopsony can be procompetitive in the face of a lawful monopoly, and merger to monopoly can be procompetitive in the face of a lawful monopsony. It would be perverse to use Section 7 to prevent these mergers when preventing them reduces social welfare. But the language of Section 7 does not appear to contemplate mergers that respond to market power on the other side of the market. Consequently, it would take a brave interpretation of the statute to permit mergers that create countervailing power. As far as we can tell, the FTC and the DOJ do not have the requisite courage. In Section 8 of the 2010 Horizontal Merger Guidelines, the role of powerful buyers is recognized, but largely dismissed: The Agencies consider the possibility that powerful buyers may constrain the ability of the merging parties to raise prices, but the Agencies do not presume that the presence of powerful buyers alone forestalls adverse competitive effects flowing from

Bilateral Monopoly   377

the merger. Even buyers that can negotiate favorable terms may be harmed by an increase in market power.

Subsequent discussion makes clear that the Guidelines’ focus is on increased market power by merging sellers. But there is no recognition that the merger may be prompted by the monopsony power of the existing buyers. There is some evidence that the agencies have rejected the possibility of bilateral monopoly and its salutary competitive effects. This can be seen in the FTC’s reaction to proposed state legislation regarding collective bargaining by healthcare providers. Several states have been persuaded that the exercise of monopsony power by large health insurers has resulted in undesirable effects. These states have proposed legislation that would provide a limited antitrust exemption for the healthcare providers to join together for the sole purpose of bargaining with health insurers over reimbursement rates. The FTC has uniformly opposed all such legislation. Physicians have argued that health insurers exercise monopsony power by setting reimbursement rates below the competitive level, thereby leading to fewer services for patients and an overall lower quality of care. If the market for healthcare services is monopsonized on the demand side and competitive on the supply side, allowing physicians to bargain as collusive monopolists would lead to a bilateral monopoly market structure. While efforts by physicians have failed at the federal level (Blair and Herndon, 2004), Alaska, New Jersey, and Texas have all passed bills that permit physicians to bargain collectively. In its reaction to the healthcare bills in Alaska, New Jersey, Texas, and other states, the FTC’s attitude is evident (Blair and Coffin, 2005). The FTC warns that such antitrust exemptions that permit price fixing by physicians will limit the availability of healthcare, increase the cost to patients, and decrease the quality of care. As we have seen in section 16.2, however, introducing countervailing power can potentially increase social welfare. Consequently, there is potential for both parties to enjoy increased profits. To the extent that reimbursement rates rise, physicians will spend more time with their patients and provide increased services to their patients. In reviewing the specifics of the Alaska bill, the FTC also argued that allowing physicians to bargain collectively is unnecessary since the antitrust laws already permit physician groups to discuss fee and nonfee issues. This argument, however, is disingenuous since what physicians presumably need is the ability to negotiate as a group. Discussion alone still leaves independent physicians unable to bargain effectively.

16.5.  Concluding Remarks When a lawful monopolist confronts a lawful monopsonist, we have a market structure known as bilateral monopoly. The welfare consequences of this market structure are superior to those of either monopoly or monopsony. As a result, bilateral monopoly raises no antitrust concerns. The creation of countervailing market power in the face

378   Roger D. Blair and Christina DePasquale of monopoly or monopsony, however, may raise antitrust concerns in some circumstances. Horizontal agreements among buyers or among sellers to create either monopoly or monopsony may violate Section 1 of the Sherman Act, which forbids horizontal agreements that restrain trade or commerce. Horizontal mergers that create countervailing power fall under Section 7 of the Clayton Act. The purpose of Section 7 is to prevent the deterioration of competitive conditions in a market. The economic rationale for this goal is obvious: in the absence of market failure, competition maximizes social welfare. Consequently, a merger to monopoly or to monopsony would appear to violate Section 7. In the presence of lawful monopoly or lawful monopsony, horizontal mergers or horizontal agreements may create countervailing power that will improve social welfare under these conditions. Accordingly, an argument can be made that such agreements or mergers should, therefore, be permitted where the bilateral bargaining outcome appears likely. Based on the static models we have examined in this chapter this recommendation is obvious, but there are some qualifications. First, if the monopoly or the monopsony is not lawful, then the unlawful monopoly or monopsony should be challenged. Restoring competition on both sides of the market is a preferable enforcement strategy. Since things can change—entry of new firms, technological change, new products—preserving competition is the safest course of action. Second, even if the monopoly or monopsony is lawful, that does not mean that it is permanent. Presumably, the economic profit enjoyed by the monopolist or monopsonist provides an incentive for entry. If the evaluation of entry barriers suggests that entry is apt to be relatively prompt, then it would be unwise to permit an agreement or merger that could cause future competitive problems. If entry is not foreclosed, but is apt to be delayed, then there may be a tough intertemporal welfare trade-off to consider. Third, in analyzing the welfare effects of horizontal agreements and horizontal mergers, we have assumed that there was, in fact, monopoly or monopsony on the other side of the market. But that may not be true, and, therefore, permitting horizontal agreements or horizontal mergers may be unwarranted. One reason for the apparent attitude of the FTC with respect to cooperative bargaining among healthcare providers is their skepticism regarding the presence of monopsony power. This, of course, is a factual matter that must be examined on a case-by-case basis.

References Baker, Jonathan B., Joseph Farrell, and Carl Shapiro. 2008. Merger to Monopoly to Serve a Single Buyer: Comment. Antitrust Law Journal 75: 641. Blair, Roger D., and Kristine L. Coffin. 2005. Physician Collective Bargaining, State Legislation, and the State Action Doctrine. Cardozo Law Review 26: 1731. Blair, Roger B., and Jeffrey L.  Harrison. 2010. Monopsony in Law and Economics. New York: Cambridge University Press.

Bilateral Monopoly   379

Blair, Roger D., and Jill Boylston Herndon. 2004. Physician Cooperative Bargaining Ventures: An Economic Analysis. Antitrust Law Journal 71: 989. Blair, Roger D., and David L. Kaserman. 1983. Law and Economics of Vertical Integration and Control. New York: Academic Press. Blair, Roger D., and David L. Kaserman. 1987. A Note on Bilateral Monopoly and Formula Price Contracts. American Economic Review 77: 460. Blair, Roger D., and David L. Kaserman. 2009. Antitrust Economics. 2nd ed., New York: Oxford University Press. Bowley, A. L. 1928. Bilateral Monopoly. Economic Journal 25: 651. Campbell, Tom. 2007. Bilateral Monopoly in Mergers. Antitrust Law Journal 74: 521. Chen, Z. 2003. Dominant Retailers and the Countervailing-Power Hypothesis. Rand Journal of Economics 34: 612. Coase, Ronald H. 1937. The Nature of the Firm. Economica 4: 386. Cournot, Augustin. 1838. Recherches sur les Principes Mathematiques de la Théorie des Richesses. Paris: L. Hachette. Cremer, J., and M. H.  Riordan. 1987. On Governing Multilateral Transactions with Bilateral Contracts. Rand Journal of Economics 18: 436. Horn, H., and A. Wolinsky. 1988. Bilateral Monopolies and Incentives for Merger. Rand Journal of Economics 19: 408. Machlup, Fritz, and Martha Taber. 1960. Bilateral Monopoly, Successive Monopoly, and Vertical Integration. Economica 27: 101. Menger, Carl. 1871. Grundsätze der Volkswirtschaftslehre. Vienna: Braumüller. Symeonidis, G. 2010. Downstream Merger and Welfare in a Bilateral Oligopoly. International Journal of Industrial Organization 28: 230. Williamson, Oliver E. 1973. Markets and Hierarchies:  Some Elementary Considerations. American Economic Review 63: 316. Williamson, Oliver E. 1974. The Economics of Antitrust:  Transaction Cost Considerations. University of Pennsylvania Law Review 122: 1439.

CHAPTER 17

A N T I T RU S T, T H E I N T E R N E T, A N D T H E E C O N OM IC S O F N E T WOR K S DANIEL F. SPULBER AND CHRISTOPHER S. YO O

Network industries have been the subject of some of the most important cases brought under the antitrust laws.1 Government suits against AT&T led to the 1914 settlement following the Kingsbury Commitment, the 1956 settlement that barred AT&T from the computer industry, and most importantly the 1982 Modification of Final Judgment that broke up the largest company in the world. Other cases involving telecommunications companies, such as Trinko and linkLine, have reshaped monopolization doctrine and redefined how antitrust law fits with other parts of the legal regime. The government case against Microsoft spawned the first major use of network economic effects in an antitrust case.2 More recently, enforcement authorities have begun to focus attention on the most recent wave of network-based companies, including Google, Apple, and Facebook. This chapter examines the relationship between antitrust and network industries. It begins with an overview of the types of networks before examining the economic considerations generally thought to play a key role in antitrust review of network industries:  natural monopoly, network economic effects, vertical exclusion, and dynamic efficiency. It then analyzes the implications for antitrust policy, focusing on structural separation and the essential facilities doctrine.

17.1.  Types of Networks There are many types of networks (Spulber and Yoo 2009). Although many of the examples in this chapter are drawn from communication networks, for the most part the 1 

For a general study of network industries and the Internet, see Spulber and Yoo (2009). Network effects arguments also appeared in Microsoft v. Commission. See Spulber (2008c) for additional discussion. 2 

Antitrust, the Internet, and the Economics of Networks   381

same conclusions apply to other types of networks as well. We can distinguish between physical networks with physical transmission facilities and virtual networks, which include two-sided platforms with technological compatibility.3 Network industries in the Transportation, Utilities, and Information sectors represent 9.1% of GDP. The components of Transportation are air transportation, railroad transportation, water transportation, truck transportation, transit and ground passenger transportation, pipeline transportation, other transportation and support activities, and warehousing and storage. Information is comprised of publishing industries (including software), motion picture and sound-recording industries, broadcasting and telecommunications, and information and data-processing services. Finally, Utilities includes power generation and supply, natural gas distribution, water, sewage, and sanitary services. Most of these industries are composed in large part of networks. Not all of these elements represent network contributions since communications includes production of content such as programming, and electricity services include the production of electric power. Although the Transportation and Utilities sectors include trucking and warehousing, it might be instructive also to consider Wholesale Trade as operating distribution networks, which comprises another 5.6% of GDP. The Wholesale Trade sector includes intermediation activities and management of the distribution network, even though some trucking and warehousing are counted separately as part of transportation. Within the Retail sector, some large retail chains also engage in their own wholesale distribution activities (US Department of Commerce 2012). Virtual networks include networks of buyers and sellers connected to each other by technological compatibility and two-sided platforms. Such networks include search engines and Internet portals (Google, Bing), social networking platforms (Facebook, Twitter), online marketplaces (Amazon, eBay), computer operating system platforms (Microsoft’s Windows, Apple’s OS, and Google’s Chrome OS), smartphone operating system platforms (Apple’s iOS, Google’s Android, Samsung’s Bada), and video game platforms (Microsoft’s Xbox, Sony’s PlayStation, Nintendo’s Wii).4

3  We do not consider decentralized social and business networks. Such social and business networks are composed of sets of individuals in a society and the relationships between them. These networks are mechanisms that distribute wealth, transmit information, facilitate business transactions, and form personal relationships (Wasserman and Faust 1994). According to Polanyi (1944, p. 46), “man’s economy, as a rule, is submerged in his social relationships” (see also Granovetter 1973, 1985). The application of graph theory and network design to game theory is related to the theory of social networks (Myerson 1977, Dutta and Jackson 2003). For a discussion of graph theory in the context of the law and economics of communications networks, see Spulber and Yoo (2009). 4  For studies of two-sided markets, in which buyers and sellers interact strategically through centralized communication mechanisms, see Caillaud and Jullien (2003), Rochet and Tirole (2003), and Spulber (2006, 2010). The formal study of the structure of business networks also includes the models of market design, referred to as market microstructure, as examined by Spulber (1996, 1998, 1999, 2002, 2003) and Lucking-Reilly and Spulber (2001). On antitrust in two-sided markets see the discussion in Alexandrov, Deltas, and Spulber (2011).

382   Daniel F. Spulber and Christopher S. Yoo

17.2.  Natural Monopoly Perhaps the classic justification for regulating networks is that they have long been presumed to be natural monopolies. A given production technology is said to exhibit natural monopoly characteristics if it has a subadditive cost function, that is, a single firm can supply the entire market demand at lower cost than could two or more firms. A sufficient condition for subadditivity is for the technology to exhibit scale economies, such as occur when a production technology requires the incurrence of joint and common fixed costs that can be spread across units of output or across multiple outputs. Natural monopoly gives rise to two policy concerns that have served to justify regulation of both entry and prices in network industries. The first concern was that entry would result in cost inefficiencies due to duplication of facilities and the loss of economies of scale. The second concern was that entry regulations to protect a monopoly incumbent would result in monopoly pricing in the absence of additional price regulation. Several important developments in network industries generally have mitigated or eliminated these policy concerns:  technological changes in network architecture, increases in demand for network services, product differentiation, and innovation. Changes in network architecture that have occurred in some network industries mitigate or eliminate these concerns. To the extent that the scale economies are the product of joint or common fixed costs, these economies can be dissipated either by a reduction in the fixed costs needed to create and operate a network or by an increase in the total demand for the services provided by the network. Technological and economic forces have pushed from both of these directions to undermine the natural monopoly rationale. A classic example is the reduction in fixed costs needed to provide local telephone service. Although the provision of local service involves a large number of components, the two that have required the greatest up-front investment in fixed costs have historically been (1) the wires needed to connect individual residences and businesses to the central office facility maintained by the local telephone company and (2) the switching equipment needed to route individual calls to their destinations. Empirical scholars have long disagreed over whether local telephone service was in fact subadditive.5 Technological change has in effect made this debate an anachronism. The fixed costs needed to provide both switching and transmission have dropped dramatically. The advent of first electronic and then digital switching has caused the fixed cost of switching technologies to plummet. The emergence of wireless alternatives to wireline transmission technologies has resulted in significant reductions in the fixed costs of transmission. The first significant deployment of wireless transmission technologies was the use of microwave relay and satellite systems as substitutes for wireline long-distance 5 

For studies concluding that local telephone service was subadditive, see Charnes, Cooper, and Sueyoshi (1988), Röller (1990a, 1990b), Wilson and Zhou (2001), and Gasmi, Laffont, and Sharkey (2002). For studies drawing the opposite conclusion, see Evans and Heckman (1983, 1984), Shin and Ying (1992), and Berg and Tschirhart (1995).

Antitrust, the Internet, and the Economics of Networks   383

services. This was followed by the wide-scale deployment of cellular telephony, personal communication services (PCS), and other wireless technologies that could serve as substitutes for local wireline telephone service. The net result is a dramatic reduction in the fixed costs needed to establish a telecommunications network sufficient to undercut the natural monopoly-based justifications for regulation. Technological change also has made telecommunications markets contestable by reducing the sunk costs associated with market entry. This allows competition for the market so that even if there is a monopoly incumbent and the incumbent firm benefits from significant economies of scale, competitive pressures will drive prices toward average costs. So long as fixed costs are not also sunk costs, any attempt by an existing player to charge supracompetitive prices will only invite hit-and-run entry by firms that gather the available profits and depart as soon as competition drives prices down to competitive levels (Baumol, Panzar, and Willig 1988). The effects of sunk costs are even lower in virtual networks, where suppliers have lower entry costs and customers have greater ease of switching in comparison to traditional network markets. Contestability theory underscores a critical difference between wireline and wireless transmission technologies. Because telephone wires have historically been useless for any other purpose, fixed cost investments in telephone wires can properly be regarded as sunk costs and thus a potential source of barriers to entry. The same is not necessarily true for the infrastructure needed to construct a wireless transmission network. Wireless technologies require equipment located on transmission towers as well as the legal right to use particular portions of the electromagnetic spectrum. Since alternative uses exist for both of these assets (either by other wireless telephone providers or by providers of wireless broadband or other spectrum-based services), investments in wireless network technologies are less likely to be regarded as sunk costs and thus less likely to give rise to the market failures said to be associated with natural monopoly. The weakening of the natural monopoly justification resulting from the reduction in fixed costs has been accompanied by an increase in the demand for the services provided by communications networks. The emergence of personal computing and the analog modem made it possible for subscribers to use their connections to telecommunications networks to send data as well as voice communications. The increase in functionality made possible by the deployment of digital technologies has mitigated the tendency of telecommunications markets to collapse into natural monopolies by greatly increasing the demand for the network services. These analog technologies are in the process of being replaced with digital technologies, such as digital subscriber lines, and by fiber optics, which are enhancing the value of the network connection still further. The increasing packetization of communications technologies has also put pressure on the distinctions between transmission technologies, which once made sense when each medium of communications employed distinct analog encoding formats. Voice over Internet protocol (VoIP) made cable modem systems a viable competitor to the voice services provided by local telephone companies, and telephone companies are developing packet-based television services. The deployment of new transmission technologies, such as fourth-generation wireless technology Long Term Evolution (LTE)

384   Daniel F. Spulber and Christopher S. Yoo and WiMax, will increase the competitiveness of last-mile telecommunications services. Once the shift towards packetization is complete, all forms of communications will simply be different applications traveling on the same data network, and the distinctions between transmission technologies will completely collapse. This combination of reductions in fixed costs and increases the demand for network services has caused much of the telecommunications network to lose its natural monopoly. Multiple facilities-based providers now vie to provide telecommunications to large business enterprises. In addition, intermodal competition from different wireline and wireless technologies has the same effect on the residential and small business market as well.6 Product differentiation also weakens or eliminates the natural monopoly justification for regulation (Spulber 1995). The concept of natural monopoly assumes that products are homogeneous so that cost efficiencies imply the need for a single provider. With product differentiation, there are consumer benefits from having multiple providers offering differentiated products. There is a trade-off between the benefits of product variety and reductions in economies of scale. Regulatory entry barriers would favor economies of scale, but would eliminate the benefits of product differentiation. With multiple providers of network services offering differentiated products, consumers benefit from competition among providers that provide services to address different consumer needs. Competition among providers also involves price competition. Product differentiation allows markets to reach equilibrium with multiple producers each producing on the declining portion of the average cost curve, as was shown by Edward Chamberlin’s seminal work on monopolistic competition (1934). So long as products are differentiated, the existence of unexhausted economies of scale need not necessarily force a network to collapse into a natural monopoly (Yoo 2005). Over time, the introduction of new types of products and services by new entrants generates turnover of market leaders, further weakening the cost-based natural monopoly argument for regulation of network industry. Innovation also can invalidate the natural monopoly argument for regulation (Spulber 1995). The traditional natural monopoly justification for entry and price regulation is based on a static industry with a given cost function. Technological change in network industries implies that the cost function of network firms changes over time. This weakens the argument that a protected monopoly incumbent generates cost benefits due to economies of scale. Entry of more efficient firms will generate cost efficiencies even if there are reduced economies of scale. There is a trade-off between the cost efficiencies generated by entry of innovative firms and the potential cost inefficiencies from smaller scale firms. Competition among firms over time addresses the trade-offs between cost reductions through innovative entry and economies of scale from fewer firms.

6  Distribution networks for water, natural gas, and electricity have not benefited to the same extent from improvements in transmission technologies and convergence.

Antitrust, the Internet, and the Economics of Networks   385

17.3.  Network Economic Effects Another economic characteristic often regarded as a source of market power in network industries is network effects. Network effects exist when the value of a network is determined not only by the services it provides, but also by the number of users connected to the network.7 The notion of network effects springs from “Metcalfe’s Law,” which states that a network with n nodes has (1/2)n(n – 1) potential node-to-node connections. In other words, the number of possible connections increases quadratically with the number of users, and presumably so does the number of potential transactions. If each new connection adds value, larger networks will enjoy greater returns to scale than smaller ones (Gilder 1993). The telephone system has long been regarded as a classic example of a network that exhibits such effects, since the value of a telephone network is determined in large part by the number of people with whom one can communicate through that network. The more people that an individual subscriber can reach through the telephone network, the more valuable the network becomes. Internet-based communications networks exhibit similar networks effects because of the benefits users derive from the ability to communicate with a larger number of other users. Internet-based transaction networks also may exhibit networks effects when users derive benefits from the number of other potential trading partners on the network. Because the Internet is a network of networks, it is often said to exhibit network economic effects, although upon closer inspection the constantly increasing returns to scale may be limited to specific networks, such as social networks, rather than the Internet itself. The ability to access the entire universe of customers need not benefit specialized businesses serving market niches, although there are advantages to serving the long tail (Brynjolfsson, Hu, and Smith 2006). Moreover, increases in the number of network subscribers represent a mixed blessing for applications designed to provide connectivity in that broader usage also causes congestion. The existence of private networks based on the same protocols as the Internet that nonetheless do not interconnect with the public Internet underscores that many network participants do not derive significant benefits from a larger number of connections. As with natural monopoly, there is a trade-off between the benefits of a larger network and the benefits of product variety from multiple networks (Yoo 2012b). Of course, the potential transactions that a network offers are very different from actual transactions. The value of such a network would be weighted by the likelihood of a transaction occurring and the potential benefits of a randomly chosen successful transaction. Thus, while the value may be proportional to the squared number of traders, the proportional weight may be very small indeed. The constantly increasing 7  More generally, buyers and sellers derive cross-market benefits, which may be the result of product variety and scale effects, market liquidity, and connectivity of communications networks (cf. Spulber 2010).

386   Daniel F. Spulber and Christopher S. Yoo returns to scale from network effects also implicitly presume that each new connection has equal value. Empirical research suggests that the distribution of value is far from uniform, with network users placing a disproportionately high value on a small, easily identifiable group of other users (Driscoll 2009, Adams 2012). When that is the case, the marginal returns from increases in network size will be small indeed (Yoo 2012a). The problem is further compounded, if the likelihood of a successful match decreases with the number of potential trading partners, for the usual reason that search costs are high and more traders can mean greater diversity and greater costs of finding a good match. If these costs increase rapidly, they can outweigh the benefits of having more members so that there can be diminishing returns to larger networks. The intensity and quality of meetings may be better at a small party than a large gathering, for example. However, with many people connected to a network there are returns to mechanisms that reduce search costs. If such mechanisms exhibit economies of scale, this can restore the benefits of larger networks. For example, with many people connected to a telephone network, there are benefits from telephone directories. With many people on the Internet, there are benefits from establishing search engines. However, these are subtle concepts that are very different from the idea of simply squaring the number of people with access to a network. Some economists contend that network effects can give rise to a kind of externality that can be a source of market failure. This view draws on the fact that an individual subscriber’s decision to join a network creates benefits for those who have already joined the network that the new subscriber cannot internalize. New subscribers’ inability to capture all of the benefits generated by their adoption decisions arguably creates a wedge between private and social benefit that may cause subscribers to forgo joining a network even though the social benefits of doing so would exceed the costs. The concept of network externality thus suggests that network industries may be uniquely susceptible to market failures that may prevent the price mechanism from playing its usual role in generating efficient outcomes. The market failure argument is based on the notion that market actors cannot coordinate their network participation decisions. However, there are many mechanisms through which market actors can coordinate their participation decisions so as to realize the benefits of using the services of a particular network. In addition, market intermediaries including the network companies themselves provide coordination through pricing, marketing, and provision of incentives for participation such as first-party content (Spulber 2006, 2008a, 2010, Hagiu and Spulber 2012). Through direct coordination or intermediation, market participants realize the benefits of network effects and reduce or eliminate potential inefficiencies. This implies that networks effects do not involve externalities, that is, economic effects outside of market transactions. Another policy concern expressed by some antitrust economists is the problem of technology lock-in. The argument is that network effects cause markets to become “locked in” to existing technologies long after the arrival of new, more efficient network technologies. However, market coordination by market participants themselves, by market intermediaries, and by network firms again addresses these network effects and

Antitrust, the Internet, and the Economics of Networks   387

mitigates concerns regarding the ability of markets to facilitate adoption of new technologies (Spulber 2008b). Understanding market coordination of technology adoption decisions helps to explain the apparent rapid technological change and the entry and exit of firms in network industries. Some theorists have also suggested that network effects can turn network access into a competitive weapon. By refusing to interconnect with other networks, network owners can force subscribers to choose one network to the exclusion of others. The fact that each network’s value increases with the number of subscribers connected to it provides a powerful incentive for new subscribers to flock to the largest network. According to this view, network economic effects create demand-side scale economies that can be a source of monopoly power (see, e.g., Katz and Shapiro 1985, Melody 2002). One oft-cited example of this phenomenon is the attempt by the Bell System to use its refusal to interconnect to combat the emergence of competition in local telephone service following the expiration of the initial Bell telephone patents in 1893. Presumably, refusing to interconnect with independent local telephone systems would protect the Bell System’s market share by ensuring that it would remain the largest (and thus the most valuable) local telephone provider. A review of the theoretical literature reveals that arguments that network economic effects inexorably lead to market failure are too simplistic (Yoo 2002). Even if, for the sake of argument, network effects were to create externalities, a consumer’s decision to switch to a new network actually gives rise to two distinct and countervailing effects. On the one hand, the decision to join a network enhances the value of the new network for those already connected to that network and those who will join that network in the future. The inability to capture all of the benefits created by its network adoption decision would give rise to a positive externality that can cause a consumer to refuse to join a new network even when it would be socially beneficial for it to do so, a phenomenon sometimes called lock-in or excess inertia. At the same time, the decision to join a new network also lowers the value of the old network by reducing the number of people using it. In effect, switching to a new network imposes costs on those connected to the old network that the person switching networks does not bear. This may make that individual willing to adopt a new technology even when the costs to society exceed the benefits, a situation variously called excess momentum or insufficient friction. It is thus possible that network effects could make users too reluctant as well as too willing to change networks. Which is the case depends upon which of these two countervailing effects dominates (Farrell and Saloner 1986a, Katz and Shapiro 1992). As already argued, coordination among market participants helps to address either of these potential consequences of network effects. In addition, the alleged market failures identified by some economic models depend on the assumption that the relevant markets are either dominated by a single firm or highly concentrated (Katz and Shapiro 1986, Besen and Farrell 1994, Crémer, Rey, and Tirole 2000). Conversely, in markets composed of a small number of similarly sized networks, individual networks have strong incentives to interconnect (Katz and Shapiro 1985, Faulhaber 2005, Vanberg 2009). An implication of these models is that

388   Daniel F. Spulber and Christopher S. Yoo policymakers should undertake a detailed examination of the structure of the relevant market before relying on network economic effects to impose antitrust liability. The economics literature indicates that private ordering may well prove quite robust in solving any problems created by network effects (Spulber 2008a). One major reason is that with respect to telecommunications networks, potential network effects arise through direct connections to a physical network that is established and owned (Katz and Shapiro 1985, Farrell and Saloner 1985). Thus, even though individual users may not be in a position to internalize all of the benefits created by their network adoption decisions, the network owner will almost certainly be in a position to do so. The existence of a single network owner allows potential problems associated with network effects to be solved by placing property in the hands of a single owner and protecting it with well-defined property rights (Coase 1960, Hardin 1968). Any benefits created by network participation can thus be internalized and allocated through the direct interaction between the network owner and network users (Liebowitz and Margolis 1994, Spulber 2008a). The benefits of property rights in addressing network effects extend to markets with competition among network owners (Liebowitz and Margolis 1994). Even assuming for the sake of argument that circumstances permit network effects to give rise to the problems of monopolistic dominance and technological lock-in, other features of the market and the structure of consumer preferences might mitigate, if not eliminate, these adverse effects. For example, the market may also dislodge an existing network technology so long as the new network provides additional value that exceeds the value derived from the size of the old network (Farrell and Saloner 1986b, Katz and Shapiro 1994). This is particularly true, given that, after networks have captured a sufficient number of subscribers, the marginal benefit from adding another subscriber is likely to be low, which would greatly reduce the magnitude of any potential externality. Heterogeneity of buyer preferences and product differentiation affect the outcome in markets with network effects. As Michael Katz and Carl Shapiro (1994, 106) have noted, “Customer heterogeneity and product differentiation tend to limit tipping and sustain multiple networks. If the rival systems have distinct features sought by certain customers, two or more systems may be able to survive by catering to consumers who care more about product attributes than network size. Here, market equilibrium with multiple incompatible products reflects the social value of variety.” With network effects, expectations of market actors affect participation rates, but coordination among market participants can shape expectations (Hagiu and Spulber 2012). In a growing market, participation will reflect anticipation of the extent of the network in the future rather than the current extent of the network (Katz and Shapiro 1992, Shapiro 1995, Liebowitz and Margolis 1996). The existence of large users may further mitigate any problems caused by network effects. If a single user controls a significant portion of the network, that user would be able to internalize more of the benefits of its adoption decision, which would help minimize any slippage caused by the existence of the network externality. Furthermore, because large users are in a position to capture a disproportionate share of the benefits resulting from the adoption of a new technology, they have significant incentives to make the investments needed to shift towards the new technology (Katz and Shapiro

Antitrust, the Internet, and the Economics of Networks   389

1994). Indeed, formal models of such market structures indicate that “the sponsor of a new technology earns greater profits than its entry contributes to social welfare. In other words, markets with network externalities in which new technologies are proprietary exhibit a bias towards new technologies” (Katz and Shapiro 1992, 73). Determining the optimal number of networks and the optimal timing of technological change requires a careful balance of the relevant costs and benefits. Antitrust policymakers should bear in mind the relevant trade-offs. A single network can involve monopoly rents, but can also offer benefits from standardization, interoperability, complementary products, and lower transaction costs. Conversely, competition lowers firms’ market power while increasing transaction costs and reducing some of the benefits of standardization, interoperability, and complementary products. Accordingly, some delay in the introduction of new products may reflect efficiency, not market failure. The fact that markets seem fully capable of resolving most of the potential market failures identified by the theoretical literature on network economics suggests that any attempt to remedy these supposed problems through regulation and antitrust should be approached with considerable caution. Indeed, it would seem appropriate to insist on empirical proof that such problems actually exist before considering governmental action to redress them. More detailed examinations of the facts surrounding the examples of anticompetitive problems stemming from network economic effects typically cited in the literature raise serious questions about their empirical foundations (Liebowitz and Margolis 2001, Spulber 2008a, 2008b). Lastly, even proof of the existence of the necessary empirical preconditions for network-induced market failure would not necessarily support intervention. Consider, for example, the manner in which antitrust law would attempt to solve the problems of technological lock-in. Such intervention would necessarily require the government to replace winners in the real-world technological marketplace with what it believed represented the superior technology. Moreover, in order to be effective, the government must do so at an early stage in the technology’s development, when making such determinations is the most difficult. Courts would also typically have to make such determinations on extremely thin information that in most cases would be provided by parties with a direct interest in the outcome. In addition, decision-makers would have to insulate themselves from the types of systematic biases traditionally associated with political decision-making processes. It is for these reasons that even supporters of network externality theories caution that governmental intervention might well make the problem worse, not better (Katz and Shapiro 1994, Bresnahan 1999).

17.4.  Vertical Exclusion The possibility of market power in one market raises the possibility that a network provider could use its market power in that market to harm competition in a vertically related market. This could be done either through vertical integration or through a

390   Daniel F. Spulber and Christopher S. Yoo vertical contractual restraint involving exclusive dealing, territorial restrictions, tying, or some similar restriction. As noted above, technological developments have substantially reduced the likelihood that any network provider will wield monopoly power in many network industries. The same forces that are increasing the competitiveness of every portion of the telecommunications industry eliminate the plausibility that any network provider will have a dominant market position to use as leverage over an adjacent market. At the same time, the conventional wisdom with respect to vertical exclusion has undergone a sea change over the past half-century. While economic theorists during the 1950s and 1960s were quite hostile toward vertical integration, vertical integration is now generally recognized to be less problematic than previously believed. (For overviews, see c­ hapter 23 in this volume and Yoo 2002.) The driving force behind this transformation is the emergence of the so-called one monopoly rent theorem, which holds that monopolists have little, if any, incentive to engage in vertical exclusion. Because there is only one monopoly profit available in any vertical chain of production, a monopolist can capture all of that profit without having to resort to vertical integration simply by charging the monopoly price (Director and Levi 1956, Bowman 1957). Moreover, it is impossible to state a coherent theory of vertical exclusion unless two structural preconditions are met. First, the firm possesses monopoly power in one market (typically called the primary market), since without such power the network owner would not have anything to use as leverage over the upstream market for complementary services (Director and Levi 1956). Second, the market into which the firm seeks to exercise vertical exclusion (called the secondary market) must be protected by entry barriers. If no such barriers to entry exist, any attempt to raise price in the secondary market will simply attract new competitors until the price drops back down to competitive levels (Posner 1976). Unless these structural preconditions are met, the most that vertical integration would do is rearrange distribution patterns. Since that time, the “post-Chicago” school of antitrust law and economics has used game theory to identify exceptions to the one monopoly rent theorem under which vertical integration can harm competition. Most of these exceptions are the result of highly stylized assumptions that do not match well with these industries (Yoo 2008). Even more interestingly, these models presuppose the existence of dominant-firm and oligopoly market structures in the primary market, which necessarily presuppose that both the primary and secondary markets are highly concentrated and protected by entry barriers (see, e.g., Salinger 1988, Hart and Tirole 1990, Ordover, Salop, and Saloner 1990, Riordan 1998). In the absence of such structural features, these formal models recognize that vertical integration may be just as likely to lower prices and increase welfare and that the ability of existing players or new entrants to expand their outputs will be sufficient to defeat any attempt to increase prices above competitive levels (Salinger 1991, Riordan and Salop 1995). The post-Chicago literature has thus done little to disturb the basic conclusions that vertical integration is unlikely to harm competition unless the relevant markets are concentrated and protected by entry barriers.

Antitrust, the Internet, and the Economics of Networks   391

Economic theorists have increasingly recognized that vertical integration could yield substantial efficiencies. For example, if two layers of a vertical chain of distribution are monopolistic or oligopolistic, firms in each layer will have the incentive to try to extract the entirety of the available supracompetitive returns, which could lead to a final price that would be even higher than the monopoly price. Vertical integration can eliminate this so-called double marginalization problem, since a company that spans both layers would rationalize the decision-making between the two levels of production and would avoid the uncoordinated action that would make supracompetitive pricing even worse (Spengler 1950). In addition, to the extent that the inputs can be used in variable proportions, any attempt to charge supracompetitive prices for one input creates incentives for firms to substitute alternative inputs whenever possible. The resulting substitution creates an alternative potential source of inefficiency, as production processes deviate from the most efficient input mix. Allowing the provider of the monopolized input to vertically integrate into manufacturing can allow it to prevent inefficient input substitution (Vernon and Graham 1971). The welfare implications of input substitution are ultimately ambiguous, since prohibiting input substitution enhances the monopolist’s ability to exercise market power, which can create welfare losses sufficient to offset the welfare gains from preventing customers from deviating from the most efficient input mix (Schmalensee 1973, Hay 1973, Warren-Boulton 1974). Finally, scholars building on Coase’s seminal work on the theory of the firm (1937) have demonstrated how vertical integration can reduce transaction costs. One example is the elimination of free riding. For example, suppose that a firm manufactures a technically complicated product that requires significant presale services (such as the demonstration of the product). Telser (1960) argues that retailers will have the incentive to shirk in providing such services in the hopes that other retailers will bear the costs of providing such services. If all retailers respond to these incentives in the same way, the total amount of presale services will fall below efficient levels. A manufacturer facing the possibility of such free riding can either rely on a vertical contractual restraint that specifies the level of presale services that each retailer is required to offer or can vertically integrate into distribution. Either solution effectively aligns the retailers’ incentives with the manufacturers’. The Supreme Court embraced precisely this rationale in Sylvania. Determining whether a particular form of vertical integration will enhance or reduce economic welfare is thus an empirical question that turns on the particular market structure and the nature of the available efficiencies. A recent survey of the empirical literature on vertical integration found that the existing studies overwhelmingly support the proposition that vertical integration and vertical restraints tend to promote, rather than harm, competition (LaFontaine and Slade 2007).8 The conventional wisdom has now largely abandoned its hostility toward vertical integration. The manner in which technology is in the process of increasing the competitiveness of network industries and 8 

For surveys of the empirical literature on vertical restraints that draw similar conclusions, see Cooper et al. (2005) and LaFontaine and Slade (2008).

392   Daniel F. Spulber and Christopher S. Yoo the real efficiencies from vertical integration have effectively undercut the threat of vertical exclusion as a justification for imposing antitrust liability.

17.5.  Dynamic Efficiency All of the rationales discussed up to this point focus on the most efficient way to organize the network that already exists. In focusing on allocating the network that already exists today, these rationales overlook the equally (if not more) important question of how to create incentives to invest in new network technologies that will comprise the optimal network of tomorrow. In other words, antitrust may have to tolerate some degree of static inefficiency in order to promote dynamic efficiency. As one noted treatise observes, US antitrust law reflects “a uniquely American, market-affirming response to [market] power” that “assumes that strong incentives promote efficiency” and that in the absence of entry barriers, market power “will erode under the pressure of market developments.” Indeed, high prices can play a key role in promoting competition, because “where supracompetitive pricing accompanies power, erosion of the power is thought to be more likely because high prices signal the need, and promise a reward, for entry” (Sullivan and Grimes 2006, 84–85). In effect, short-run static efficiency losses must be tolerated if necessary to promote long-run dynamic efficiency gains from innovation and successive entry.9 Imposing antitrust liability whenever firms earn supracompetitive returns would eliminate the primary impetus for competitive entry, in which case the supply curve would never shift outwards in order to bring the market back into long-run equilibrium. Moreover, forcing network owners to share the benefits of their investments with their competitors or limiting the prices they can charge their channel partners would dampen the incentives to invest in alternate network capacity. In effect, forcing a monopolist to share its network rescues other firms from having to provide or obtain the relevant input for themselves. As a result, it can have the perverse effect of entrenching a network monopoly by forestalling the emergence of the substitute capacity. This is particularly problematic in technologically dynamic industries, in which the prospects of developing new ways either to circumvent or to compete directly with the alleged bottleneck are the highest. Such a surrender to the monopoly only makes sense if competitive entry is infeasible. Although some scholars have asserted that because the dynamic efficiency gains will be compounded over time, they will necessarily exceed the short-run static efficiency losses (Ordover and Baumol 1988), this approach seems too simplistic. Whether the dynamic efficiency gains will dominate the static efficiency losses depends on the relative magnitude of the gains and losses, the speed of entry, and the appropriate discount rate. That said, a number of institutional considerations militate in favor of the dynamic 9 

For a discussion of antitrust and dynamic efficiency, see Sidak and Teece (2009).

Antitrust, the Internet, and the Economics of Networks   393

efficiency side of the balance. For example, calibrating the prices needed to implement rate regulation and access regulation will necessarily require the government to engage in an exquisite exercise in line drawing that requires a careful and fact-intensive balance of opposing considerations. This is made all the more complicated by rapid changes in network technology and in consumer demand for network services. Antitrust policy must carefully balance these offsetting considerations. The alternative would be to allow the short-run supracompetitive returns to stimulate entry by alternative last-mile providers whenever such entry is possible. It is for this reason that courts and policymakers have been reluctant to compel access to a resource that is available from another source, even if it is only available at significant cost and in the relatively long run (Areeda and Hovenkamp 2006).

17.6.  Implications for Antitrust Policy The economic critique has important implications for the doctrines and remedies that antitrust courts have applied to network industries. In particular, these insights affect the reliance on structural separation as a remedy and on the essential facilities doctrine as a cause of action.

17.6.1.  Structural Remedies: Vertical Separation One of the most common remedies sought during antitrust litigation in network industries is structural separation. For example, the 1956 consent decree settling the second major case against AT&T abandoned the government’s initial request for divestiture of AT&T’s equipment subsidiary, opting instead to restrict AT&T to furnishing common carrier communications services. The 1982 court order that broke up AT&T required that the local telephone services that remained monopolized be structurally separated from the portions of the business in which competition had become possible: long distance, telephone equipment, and “information services” that combined transmission with data processing. More recently, the federal government initially asked the court hearing the case against Microsoft to require the company to spin off its applications businesses into a separate subsidiary. The rationale generally given for imposing structural separation is that rate regulation may encourage firms to vertically integrate into unregulated lines of business that they can bundle with the regulated service and for which they can charge the monopoly prices denied them by regulation (see, e.g., U.S. Department of Justice 1984, Jefferson Parish (O’Connor, J., concurring in the judgment)). In such cases, it is arguably appropriate to impose what is sometimes called the “Bell Doctrine” or “Baxter’s Law,” which prohibits vertical integration in order to isolate and quarantine the monopolist (Joskow and Noll 1999, Farrell and Weiser 2003).

394   Daniel F. Spulber and Christopher S. Yoo A more subtle version of this argument exists when a firm subject to cost-of-service rate regulation provides an unregulated service that shares joint costs with the regulated service. In that case, a monopolist can attempt to allocate a disproportionate amount of the joint costs onto the regulated service and recover those costs through its regulated rates, which is a possibility given the inherent arbitrariness and uncertainty surrounding methodologies for allocating joint costs. Shifting a disproportionate amount of the joint costs onto the regulated market also allows firms to reduce the size of the markup to cover joint costs included in the prices charged in the unregulated market. The firm’s nonvertically integrated competitors, however, charge prices that reflect costs and competition in the unregulated market (Farrell 1996, Sullivan 1996, Huber, Kellogg, and Thorne 1999, Joskow and Noll 1999). The traditional solution to these problems is to prohibit network providers from offering any unregulated services. A less severe restriction is to require network providers to segregate their regulated services into a separate subsidiary and to require that subsidiary to offer carriage to others on the same terms that it offers carriage to itself. Structurally separating the regulated business from the unregulated business limits the network owner’s ability to shift costs from its unregulated service to its regulated service. Forcing the regulated subsidiary to negotiate interconnection agreements through arm’s-length transactions also eliminates the ability to use bundling of regulated and unregulated services to avoid rate regulation and makes nondiscrimination easier to detect and enforce. Structural separation requirements have proven exceedingly difficult to administer. For example, the court overseeing the implementation of the 1982 decree breaking up AT&T was bombarded with hundreds of requests to waive the decree’s line of business restrictions (Huber, Kellogg, and Thorne 1999). These requests often took from three to four years to process, with estimates of the total welfare loss associated with these delays exceeding $1 billion (Rubin and Dezhbakhsh 1995, Hausman 1997). The logic inherent in this approach suggests that regulation and the resulting quarantine should be limited only to those services that are inherently noncompetitive. Doing so would allow competition to determine outcomes to the greatest extent possible. This rationale thus presumes that the scope of regulation should constantly contract over time, as technological change and increases in demand open larger portions of the telecommunications industry to competition. This dynamic quality of the scope of regulation explains much of the history of telecommunications policy. As noted earlier, initially the entire telephone system was regarded as a natural monopoly. As a result, the entire network was subjected to rate regulation. Over time, it became clear that equipment manufacturing was not subject to the high fixed costs traditionally associated with natural monopoly and was instead potentially open to competition. The emergence of microwave and satellite transmission also lowered the fixed costs of providing long-distance service by allowing new long-distance entrants like MCI to compete without having to establish a nationwide network of high-volume telephone lines. As it became clear that each of these markets was now open to competition, the FCC released them from rate regulation and prohibited regulated entities

Antitrust, the Internet, and the Economics of Networks   395

from offering them on an integrated basis. Eventually, the only portion of the telecommunications industry that remained subject to rate regulation was local telephone service. Over time, even that premise has come under fire, as wireless has reduced the fixed costs needed for transmission and as computer processing has reduced the fixed costs of switching. The growing constriction of the scope of rate regulation and monopoly power has effectively curtailed the analytical foundations for mandating structural separation.

17.6.2.  Behavioral Remedies: The Essential Facilities Doctrine Perhaps the most commonly asserted basis under the antitrust laws for challenging network owners’ supposed attempts to engage in anticompetitive conduct is the essential facilities doctrine (see also ­chapter 6 in volume 2, Spulber and Yoo 2009). The doctrine requires owners of bottleneck elements unavailable elsewhere in the marketplace to provide other firms with access to those elements on reasonable terms. Given the economic forces discussed above, it should come as no surprise that early leading cases associated with the doctrine (such as Terminal Railroad and Otter Tail) arose in network industries.10 In addition, both the 1956 settlement of the federal government’s antitrust suit against AT&T and the litigation that led to the 1984 breakup of AT&T were based on the essential facilities doctrine. The central concern of the essential facilities doctrine is thus vertical exclusion (Werden 1987, Areeda and Hovenkamp 2006). Thus, lower courts have incorporated into the doctrine the same economic considerations discussed above. Courts have applied the same structural preconditions to the essential facilities doctrine: concentration in the primary market (see, e.g., MCI Communications Corp. v. AT&T Co.) and concentration and entry barriers in the secondary market (see, e.g., Alaska Airlines). Moreover, courts applying the essential facilities doctrine consider whether exclusion from the monopoly might be justified by efficiencies (see, e.g., City of Vernon v. Southern California Edison Co., Illinois ex rel. Burris v. Panhandle Eastern Pipe Line Co., Byars v. Bluff City News Co.). Finally, courts have recognized the importance of dynamic efficiency. In the words of the Ninth Circuit in Alaska Airlines, “Every time the monopolist asserts its market dominance” by denying rivals access to a bottleneck input, the rival is given “more incentive to find an alternative supplier, which in turn gives alternate suppliers more reason to think that they can compete with the monopolist. Every act exploiting monopoly power to the disadvantage of the monopoly’s customers hastens the monopoly’s end by making the potential competition more attractive.” The Supreme Court offered its most complete discussion of the trade-off underlying vertical exclusion and the essential facilities doctrine claims in Trinko. Although the 10  Although widely regarded as the seminal essential facilities case, Terminal Railroad is more properly regarded as arising from horizontal concentration rather than vertical exclusion (Spulber and Yoo 2009).

396   Daniel F. Spulber and Christopher S. Yoo Court found “no need either to recognize . . . or to repudiate” the doctrine, even assuming arguendo that the doctrine applied, the Court found it inapposite. As an initial matter, the Court acknowledged that the essential facilities doctrine “serves no purpose” when the input to which access is sought is available through other means, in effect embracing the structural precondition requiring concentration in the primary market. The Court also held that any claim of vertical exclusion must show “a ‘dangerous probability of success’ in monopolizing [the] second market.” This language has been widely interpreted as requiring proof of market concentration and entry barriers in the secondary market before a monopolization claim will lie (Areeda and Hovenkamp 2006, Kauper 2005). The Trinko court also explicitly recognized the importance of dynamic efficiency considerations, noting how the prospect of earning short-run supracompetitive returns can stimulate upgrades to the network infrastructure. In the words of the Court, “The mere possession of monopoly power, and the concomitant charging of monopoly prices, is not only not unlawful; it is an important element of the free-market system. The opportunity to charge monopoly prices—at least for a short period—is what attracts ‘business acumen’ in the first place; it induces risk taking that produces innovation and economic growth.” Using antitrust law to require network owners to share that infrastructure “may lessen the incentive for the monopolist, the rival, or both to invest in those economically beneficial facilities.” Imposing antitrust liability on those who invest in such facilities “seem[ed] destined to distort investment.” The Court did not acknowledge that denying others access to the facilities created with such investments could sometimes be unlawful. Its reasoning suggested that such remedies were limited to a narrow range of circumstances with preexisting business arrangements or when a network owner forgoes short-run profits by refusing to deal with competitors even when they are willing to pay full (presumably monopolistic) prices. In addition, the Trinko Court weighed into the long-standing debate on the relative merits of structural and behavioral relief when it emphasized the problematic nature of the essential facilities doctrine in terms of administrability. In the words of the Court, “Enforced sharing also requires antitrust courts to act as central planners, identifying the proper price, quantity, and other terms of dealing.” Furthermore, because mandated access affects network elements “deep within the bowels” of a local telephone network, they can only be made available if “[n]‌ew systems [are] designed and implemented simply to make that access possible.” Additionally, requests for unbundled access “are difficult for antitrust courts to evaluate, not only because they are highly technical, but also because they are likely to be extremely numerous, given the incessant, complex, and constantly changing interaction of competitive and incumbent [local telephone companies] implementing the sharing and interconnection obligations.” The “uncertain virtue of forced sharing and the difficulty of identifying and remedying anticompetitive conduct by a single firm” suggested that the addition of an antitrust remedy would do little to promote consumer welfare while posing a significant risk of deterring welfare-enhancing behavior. In so doing, Trinko echoed the criticism that the culmination of an antitrust case in such a regulatory decree is tantamount to a confession that the case should never have been brought (Posner 1970, Posner and Easterbrook 1981).

Antitrust, the Internet, and the Economics of Networks   397

Many commentators have concluded that little, if anything, remained of the essential facilities doctrine following Trinko (see, e.g., Fox 2005, Noll 2005, Geradin and O’Donoghue 2005, Shelanski 2007, Frischmann and Waller 2008, Hovenkamp 2008). Moreover, Trinko contradicted the assertions of many commentators that the essential facilities doctrine had greater relevance in the context of regulated industries (Werden 1987, Areeda and Hovenkamp 2006). To the extent that application of the doctrine depends on a monopoly in the primary market, the advent of competition in many network industries radically narrows its scope.

17.7. Conclusion Network industries have exhibited significant growth, substantial competition, and rapid innovation, including Internet-based e-commerce. Some have argued that various economic theories, including natural monopoly, network effects, and vertical exclusion, suggest the need for heightened antitrust scrutiny of network industries in comparison with other industries. The present discussion suggests that other aspects of network industries argue for regulatory and antitrust forbearance, or at least suggest that such industries should not be the subject of heightened scrutiny. Technological changes in communications, product differentiation, and the effects of innovation imply that regulatory limits on entry into network industries do not promote consumer benefits. At the same time, natural monopoly need not create first-mover or incumbent advantages, so that natural monopoly should not justify greater antitrust scrutiny of network industries. Coordination among market participants, by the participants themselves, by network intermediaries, and by network owners suggest that network effects are not a source of market failure and should not justify heightened antitrust scrutiny of network industries. The observation that vertical integration and vertical restraints tend to promote, rather than harm, competition applies to network industries. Again, vertical exclusion arguments should not justify heightened antitrust scrutiny of network industries. Our discussion strongly suggests that antitrust policy markets should not presume that network industries are more subject to monopolization than other industries. Rapid innovation and firm turnover in network industries imply that antitrust policymakers should take into account dynamic efficiency considerations in evaluating competition in network industries.

Acknowledgments Professor Spulber would like to thank the Ewing Marion Kauffman Foundation for research support. Professor Yoo would like to thank the Milton and Miriam Handler

398   Daniel F. Spulber and Christopher S. Yoo Foundation, the New  York Bar Foundation, and the University of Pennsylvania Center for Technology, Innovation and Competition for their financial support for this project.

References Adams, Paul. 2012. Grouped. Berkeley, CA: New Riders. Alexandrov, Alexei, George Deltas, and Daniel F. Spulber. 2011. Competition and Antitrust in Two-Sided Markets. Journal of Competition Law and Economics 7: 775–812. Areeda, Philip E., and Herbert Hovenkamp. 2006. Antitrust Law. 3rd ed. Vol. 3A. New York: Aspen Publishers. Baumol, William J., John C. Panzar, and Robert D. Willig. 1988. Contestable Markets and the Theory of Industry Structure. San Diego: Harcourt Brace Jovanovich. Berg, Sanford V., and John Tschirhart. 1995. A Market Test for Natural Monopoly in the Local Exchange. Journal of Regulatory Economics 8: 103–24. Besen, Stanley M., and Joseph Farrell. 1994. Choosing How to Compete: Strategies and Tactics in Standardization. Journal of Economic Perspectives 8(2): 117–31. Bowman, Ward S., Jr. 1957. Tying Arrangements and the Leverage Problem. Yale Law Journal 67: 19–36. Bresnahan, Timothy F. 1999. New Modes of Competition: Implications for the Future Structure of the Computer Industry. In Jeffrey A.  Eisenach and Thomas M.  Lenard, Competition, Innovation and the Microsoft Monopoly:  Antitrust in the Digital Marketplace, 155–208. Boston, MA: Kluwer Academic Publishing. Brynjolfsson, Erik, Yu “Jeffrey” Hu, and Michael D. Smith. 2006. From Niches to Riches: Anatomy of the Long Tail. Sloan Management Review 47(4): 67–71. Caillaud, Bernard, and Bruno Jullien. 2003. Chicken and Egg: Competition among Intermediation Service Providers. Rand Journal of Economics 34: 309–28. Chamberlin, Edward Hastings. 1934. The Theory of Monopolistic Competition. Cambridge, MA: Harvard University Press. Charnes, A., W. W. Cooper, and T. Sueyoshi. 1988. A Goal Programming / Constrained Regression Review of the Bell System Breakup. Management Science 34: 1–26. Coase, Ronald H. 1937. The Theory of the Firm. Economica, n.s. 4: 386–405. Coase, Ronald H. 1960. The Problem of Social Cost. Journal of Law and Economics 3: 1–44. Cooper, James C., Luke M. Froeb, Dan O’Brien, and Michael G. Vita. 2005. Vertical Antitrust Policy as a Problem of Inference. International Journal of Industrial Organization 23: 639–64. Crémer, Jacques, Patrick Rey, and Jean Tirole. 2000. Connectivity in the Commercial Internet. Journal of Industrial Economics 48: 433–72. Director, Aaron, and Edward H. Levi. 1956. Law and the Future: Trade Regulation. Northwestern University Law Review 51: 258–72. Driscoll, Michael E. 2009. SQL Is Dead. Long Live SQL! November 29. http://www.dataspora.​ com/2009/11/sql-is-dead-long-live-sql/. Dutta, Bhaskar, and Matthew O. Jackson. 2003. Networks and Groups. New York: Springer. Evans, David S., and James J.  Heckman. 1983. Multiproduct Cost Function Estimates and Natural Monopoly Tests for the Bell System. In David S. Evans, Breaking Up Bell, 253–82. New York: North-Holland.

Antitrust, the Internet, and the Economics of Networks   399

Evans, David S., and James J. Heckman. 1984. A Test for Subadditivity of the Cost Function with an Application to the Bell System. American Economic Review 74: 856–58. Farrell, Joseph. 1996. Creating Local Competition. Federal Communications Law Journal 49: 201–15. Farrell, Joseph, and Garth Saloner. 1985. Standardization, Compatibility, and Innovation. Rand Journal of Economics 16: 70–83. Farrell, Joseph, and Garth Saloner. 1986a. Installed Base and Compatibility: Innovation, Product Preannouncements, and Predation. American Economic Review 76: 940–955. Farrell, Joseph, and Garth Saloner. 1986b. Standardization and Variety. Economics Letters 20: 71–74. Farrell, Joseph, and Philip J. Weiser. 2003. Modularity, Vertical Integration, and Open Access Policies: Towards a Convergence of Antitrust and Regulation in the Internet Age. Harvard Journal of Law and Technology 17: 85–134. Faulhaber, Gerald R. 2005. Bottlenecks and Bandwagons:  Access Policy in the New Telecommunications. In Samit K. Majumdar, Ingo Vogelsang, and Martin E. Cave, Handbook of Telecommunications Economics: Technology Evolution and the Internet, vol. 2: 488–517. New York: North-Holland. Fox, Eleanor M. 2005. Is There Life in Aspen after Trinko? The Silent Revolution of Section 2 of the Sherman Act. Antitrust Law Journal 73: 153–70. Frischmann, Brett, and Spencer Weber Waller. 2008. Revitalizing Essential Facilities. Antitrust Law Journal 75: 1–66. Gasmi, F., J. J. Laffont, and W. W. Sharkey. 2002. The Natural Monopoly Test Reconsidered: An Engineering Process-Based Approach to Empirical Analysis in Telecommunications. International Journal of Industrial Organization 20: 435–59. Geradin, Damien, and Robert O’Donoghue. 2005. The Concurrent Application of Competition Law and Regulation: The Case of Margin Squeeze Abuses in the Telecommunications Sector. Journal of Competition Law and Economics 1: 355–425. Gilder, George. 1993. Metcalfe’s Law and Legacy. Forbes ASAP, September 13: 158. Granovetter, Mark. 1973. Strength of Weak Ties. American Journal of Sociology 78: 1360–80. Granovetter, Mark. 1985. Economic-Action and Social-Structure: The Problem of Embeddedness. American Journal of Sociology 91: 481–510. Hagiu, Andrei, and Daniel F. Spulber. 2012. First-Party Content and Coordination in Two-Sided Markets. Management Science 59: 933–49. Hardin, Garrett. 1968. The Tragedy of the Commons. Science 162: 1243–48. Hart, Oliver, and Jean Tirole. 1990. Vertical Integration and Market Foreclosure. Brookings Papers on Economic Activity: Microeconomics 1990: 205–76. Hausman, Jerry A. 1997. Valuing the Effect of Regulation on New Services in Telecommunications. Brookings Papers on Economic Activity: Microeconomics 1997: 1–38. Hay, George A. 1973. An Economic Analysis of Vertical Integration. Industrial Organization Review 1: 188–98. Hovenkamp, Herbert J. 2008. Unilateral Refusals to Deal, Vertical Integration, and the Essential Facility Doctrine. University of Iowa Legal Studies Research Paper No. 08-31. http://papers.​ ssrn.com/sol3/papers.cfm?abstract_id=1144675. Huber, Peter W., Michael K. Kellogg, and John Thorne. 1999. Federal Telecommunications Law. 2nd ed. Gaithersburg, NY: Aspen Publishers. Joskow, Paul L., and Roger G. Noll. 1999. The Bell Doctrine: Applications in Telecommunications, Electricity, and Other Network Industries. Stanford Law Review 51: 1249–315.

400   Daniel F. Spulber and Christopher S. Yoo Katz, Michael L., and Carl Shapiro. 1985. Network Externalities, Competition, and Compatibility, American Economic Review 75: 424–40. Katz, Michael L., and Carl Shapiro. 1986. Technology Adoption in the Presence of Network Externalities. Journal of Political Economy 94: 822–41. Katz, Michael L., and Carl Shapiro. 1992. Product Introduction with Network Externalities. Journal of Industrial Economics 40: 55–83. Katz, Michael L., and Carl Shapiro. 1994. Systems Competition and Network Effects. Journal of Economic Perspectives 8(2): 93–115. Kauper, Thomas E. 2005. Section Two of the Sherman Act: The Search for Standards. Georgetown Law Journal 93: 1623–44. LaFontaine, Francine, and Margaret Slade. 2007. Vertical Integration and Firm Boundaries: The Evidence. Journal of Economic Literature 45: 629–85. LaFontaine, Francine, and Margaret Slade. 2008. Exclusive Contracts and Vertical Restraints: Empirical Evidence and Public Policy. In Paolo Buccirossi, Handbook of Antitrust Economics, 391–414. Cambridge, MA: MIT Press. Liebowitz, S. J., and Stephen E. Margolis. 1994. Network Externality: An Uncommon Tragedy. Journal of Economic Perspectives 8(2): 133–50. Liebowitz, S.  J., and Stephen E.  Margolis. 1996. Should Technology Choice Be a Concern of Antitrust Policy? Harvard Journal of Law and Technology 9: 283–318. Liebowitz, S. J., and Stephen E. Margolis. 2001. Winners, Losers and Microsoft. 2nd ed. Oakland, CA: Independent Institute. Lucking-Reilly, David, and Daniel F. Spulber. 2001. Business-to-Business Electronic Commerce. Journal of Economic Perspectives 15(1): 55–68. Melody, W. H. 2002. Building the Regulatory Foundations for Growth in Network Economies. WDR Discussion Paper No. 0201. http://regulateonline.org/2002/dp/dp0201.htm. Myerson, Roger B. 1977. Graphs and Cooperation in Games. Mathematics of Operations Research 2: 225–29. Noll, Roger G. 2005. “Buyer Power” and Economic Policy. Antitrust Law Journal 72: 589–624. Ordover, Janusz A., and William Baumol. 1988. Antitrust Policy and High-Technology Industries. Oxford Review of Economic Policy 4: 13–34. Ordover, Janusz A., Steven Salop, and Garth Saloner. 1990. Equilibrium Vertical Foreclosure. American Economic Review 80: 127–42. Polanyi, Karl. 1944. The Great Transformation: The Political and Economic Origins of Our Time. New York: Farrar and Rinehart. Posner, Richard. 1970. A Statistical Study of Antitrust Enforcement. Journal of Law and Economics 13: 365–419. Posner, Richard. 1976. Antitrust Law: An Economic Perspective. Chicago: University of Chicago Press. Posner, Richard, and Frank H. Easterbrook. 1981. Antitrust: Cases, Economic Notes and Other Materials. 2nd ed. St, Paul, MN: West. Riordan, Michael H. 1998. Anticompetitive Vertical Integration by a Dominant Firm. American Economic Review 88: 1232–48. Riordan, Michael H., and Steven C. Salop. 1995. Evaluating Vertical Mergers: A Post-Chicago Approach. Antitrust Law Journal 63: 513–68. Rochet, Jean-Charles, and Jean Tirole. 2003. Platform Competition in Two-Sided Markets. Journal of the European Economic Association 1: 990–1029.

Antitrust, the Internet, and the Economics of Networks   401

Röller, Lars-Hendrik. 1990a. Modeling Cost Structure:  The Bell System Revisited. Applied Economics 22: 1661–74. Röller, Lars-Hendrik. 1990b. Proper Quadratic Cost Functions with an Application to the Bell System. Review of Economics and Statistics 72: 202–10. Rubin, Paul H., and Hashem Dezhbakhsh. 1995. Costs of Delay and Rent-Seeking under the Modification of Final Judgment. Managerial and Decision Economics 16: 385–99. Salinger, Michael A. 1988. Vertical Mergers and Market Foreclosure. Quarterly Journal of Economics 103: 345–56. Salinger, Michael A. 1991. Vertical Mergers in Multi-product Industries and Edgeworth’s Paradox of Taxation. Journal of Industrial Economics 39: 545–56. Schmalensee, Richard. 1973. A Note on the Theory of Vertical Integration. Journal of Political Economy 81: 442–49. Shapiro, Carl. 1995. Aftermarkets and Consumer Welfare: Making Sense of Kodak. Antitrust Law Journal 63: 483–512. Shelanski, Howard A. 2007. Adjusting Regulation to Competition: Toward a New Model for U.S. Telecommunications Policy. Yale Journal on Regulation 24: 55–105. Shin, Richard T., and John S. Ying. 1992. Unnatural Monopolies in Local Telephone. Rand Journal of Economics 23: 171–83. Sidak, J. Gregory, and David Teece. 2009. Dynamic Competition in Antitrust Law. Journal of Competition Law and Economics 5: 581–631. Spengler, Joseph J. 1950. Vertical Integration and Antitrust Policy. Journal of Political Economy 58: 347–52. Spulber, Daniel F. 1995. Deregulating Telecommunications. Yale Journal on Regulation 12: 25–67. Spulber, Daniel F. 1996. Market Microstructure and Intermediation. Journal of Economic Perspectives 10(3): 135–52. Spulber, Daniel F. 1998. The Market Makers: How Leading Companies Create and Win Markets. New York: McGraw-Hill. Spulber, Daniel F. 1999. Market Microstructure:  Intermediaries and the Theory of the Firm. New York: Cambridge University Press. Spulber, Daniel F. 2002. Market Microstructure and Incentives to Invest. Journal of Political Economy 110: 352–81. Spulber, Daniel F. 2003. Management Strategy. New York: McGraw-Hill. Spulber, Daniel F. 2006. Firms and Networks in Two-Sided Markets. In Terry Hendershott, Handbook of Economics and Information Systems, vol. 1:  137–200. Amsterdam: Elsevier. Spulber, Daniel F. 2008a. Consumer Coordination in the Small and in the Large: Implications for Antitrust in Markets with Network Effects. Journal of Competition Law and Economics 4: 207–62. Spulber, Daniel F. 2008b. Unlocking Technology:  Antitrust and Innovation. Journal of Competition Law and Economics, 4: 915–66. Spulber, Daniel F. 2008c. Competition Policy and the Incentive to Innovate: The Dynamic Effects of Microsoft v. Commission. Yale Journal on Regulation 25: 247–301. Spulber, Daniel F. 2010. Solving the Circular Conundrum: Communication and Coordination in Two-Sided Networks. Northwestern University Law Review 104: 537–91. Spulber Daniel F., and Christopher S.  Yoo. 2009. Access to Networks:  Economics and Law. New York: Cambridge University Press.

402   Daniel F. Spulber and Christopher S. Yoo Sullivan, Lawrence A. 1996. Elusive Goals under the Telecommunications Act: Preserving Long Distance Competition upon Baby Bell Entry and Attaining Local Exchange Competition: We’ll Not Preserve the One Unless We Attain the Other. Southwestern University Law Review 25: 487–534. Sullivan, Lawrence A., and Warren S.  Grimes. 2006. The Law of Antitrust:  An Integrated Handbook. 2nd ed., St. Paul, MN: Thomson/West. Telser, Lester G. 1960. Why Should Manufacturers Want Fair Trade? Journal of Law and Economics 3: 86–108. US Department of Commerce. 2012. Durable-Goods Manufacturing Led Growth in 2011: Advance GDP by Industry Statistics for 2011. http://www.bea.gov/newsreleases/industry/gdpindus​ try/2012/pdf/gdpind11_adv.pdf. US Department of Justice. 1984. 1984 Merger Guidelines. Federal Register 49: 26,823–37. Vanberg, Margit. 2009. Competition and Cooperation in Internet Backbone Services. In Peter Curwen, Justus Haucap, and Brigitte Preissl, Telecommunication Markets:  Drivers and Impediments, 41–58. New York: Physica-Verlag. Vernon, John M., and Daniel A.  Graham. 1971. Profitability of Monopolization by Vertical Integration. Journal of Political Economy 79: 924–25. Warren-Boulton, Frederick R. 1974. Vertical Control with Variable Proportions. Journal of Political Economy 82: 783–802. Wasserman, Stanley, and Katherine Faust. 1994. Social Network Analysis:  Methods and Applications. New York: Cambridge University Press. Werden, Gregory J. 1987. The Law and Economics of the Essential Facility Doctrine. St. Louis University Law Journal 32: 433–80. Wilson, Wesley W., and Yimin Zhou. 2001. Telecommunications Deregulation and Subadditive Costs:  Are Local Telephone Monopolies Natural? International Journal of Industrial Organization 19: 909–30. Yoo, Christopher S. 2002. Vertical Integration and Media Regulation in the New Economy. Yale Journal on Regulation 19: 171–300. Yoo, Christopher S. 2005. Beyond Network Neutrality. Harvard Journal of Law and Technology 19: 1–77. Yoo, Christopher S. 2008. Network Neutrality, Consumers, and Innovation. University of Chicago Legal Forum 2008: 179–262. Yoo, Christopher S. 2012a. When Antitrust Met Facebook. George Mason Law Review 19: 1146–62. Yoo, Christopher S. 2012b. Internet Policy Going Forward: Does One Size Fit All? In Randolph J.  May, Communications Law and Policy in the Digital Age:  The Next Five Years, 51–65. Durham, NC: Carolina Academic Press.

Cases Cited Alaska Airlines, Inc. v. United Airlines, Inc., 948 F.2d 536 (9th Cir. 1991). Byars v. Bluff City News Co., 609 F.2d 843 (6th Cir. 1980). Case T-201/04, Microsoft v. Commission [2007] ECR II-1. City of Vernon v. Southern California Edison Co., 955 F.2d 1361 (9th Cir. 1992). Continental T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36 (1977). Illinois ex rel. Burris v. Panhandle Eastern Pipe Line Co., 935 F.2d 1469 (7th Cir. 1991).

Antitrust, the Internet, and the Economics of Networks   403

Jefferson Parish Hospital District No. 2 v. Hyde, 466 U.S. 2 (1984). MCI Communications Corp. v. AT&T Co., 708 F.2d 1081 (7th Cir. 1983). Otter Tail Power Co. v. United States, 410 U.S. 366 (1973). Pacific Bell Telephone Co. v. linkLine Communications, Inc., 555 U.S. 438 (2009). United States v.  AT&T Co., Decrees and Judgments in Federal Antitrust Cases, July 2, 1890– January 1, 1918, at 483 (D. Or. Mar. 26, 1914)). United States v. AT&T Co., 552 F. Supp. 131 (D.D.C. 1982), aff ’d mem. sub nom. Maryland v. United States, 460 U.S. 1001 (1983). United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001). United States v. Terminal Railroad Association, 224 U.S. 383 (1912). United States v. Western Electric Co., 1956 Trade Cas. (CCH) ¶ 62,246 (D.N.J. Jan. 24, 1956). Verizon Communications Inc. v. Law Offices of Curtis V. Trinko, LLP, 540 U.S. 411 (2004).

CHAPTER 18

T H E A N T I T RU S T A NA LYSI S OF M U LT I SI D E D P L AT F OR M BU SI N E S SE S DAVID S. EVANS AND RICHARD SCHMALENSEE

18.1. Introduction ECONOMISTS have identified an important class of businesses that are now generally referred to as “multisided platforms.”1 Multisided platforms create value by bringing two or more different types of economic agents together and facilitating interactions between them that make all agents better off. These platforms play critical roles in many economically important industries including payments, mobile phones, financial exchanges, advertising-supported media, and various Internet-based industries. In this chapter we will focus on common aspects of platform businesses that have important implications for firm behavior and competition, but we will also note some of the novel and important ways these implications differ across firms and industries.2

1  They are also sometimes referred to a “two-sided markets.” We resist this terminology because multisidedness is an attribute of an individual business that may not always be shared by all of its competitors. 2  Over the years, several strands of economic analysis have touched on aspects of these types of businesses. Economists have understood how intermediaries facilitate value-creating exchange at least since Smith (1776, 213–22). The microfinance literature modeled exchange institutions such as those for equities. See Spulber (1996). The advertising economics literature has examined the role of the media in connecting advertisers and readers. See Bagwell (2007). The literature on indirect network effects has considered the coordination of demand between users and producers of complementary goods. See Katz and Shapiro (1994) for a survey. On the surface, multisided platforms resemble ordinary multiproduct firms, but, as we discuss below, they in fact differ fundamentally.

The Antitrust Analysis of Multisided Platform Businesses   405

Multisided platforms solve a transaction cost problem that makes it difficult or impossible for agents in different groups to get together. In most cases, greater involvement by agents of at least one type increases the value of the platform to agents of other types. Such indirect network effects function something like economies of scale on the demand side and increase the value economic agents can realize from the platform. In a two-sided setting, the chance of finding a value-increasing interaction depends on how many agents of the first kind an agent of the second kind can reach and often vice versa. A multisided platform creates value by coordinating the multiple groups of agents and, in particular, ensuring that there are enough agents of each type to make participation worthwhile for all types. The fundamental insight that there is a broad class of businesses that act as catalysts for creating value for two or more groups of customers and have economic features not well explained by the standard economics of the firm was made by Jean-Charles Rochet and Jean Tirole (2003) in a paper that started circulating around 2000.3 Their main focus in developing an economic model of multisided platforms was on how the relative prices charged to the two sides of the platform coordinated demand. They showed that the optimal prices—both from the standpoint of profit-maximization and social welfare maximization—could entail pricing below the marginal cost of provision to one side and above the marginal cost of provision to the other side. Evans (2003a) showed that there were numerous industries in which firms acting as catalysts set some prices below marginal cost and sometimes at zero. These include software platforms, advertising-supported media, exchanges, and payment systems. They range from the old (the village matchmaker) to the new (mobile-based social networking). Since the birth of the multisided platform literature, three noteworthy developments have taken place that are particularly relevant to this chapter. First, there has emerged a large and rapidly expanding literature on multisided platforms in economics, antitrust, and strategic management. We have identified more than 200 articles on multisided platforms that have appeared in print or in working paper form since we last surveyed this area in 2007.4 It is still early in the development of this corner of industrial organization, and all the important questions have by no means been answered. Second, competition authorities around the world have used the multisided platform framework to evaluate cases and reach decisions. See OECD (2009) for a survey of the how various authorities are approaching cases involving platforms.5 The multisided platform

3  Other key theoretical papers are Caillaud and Jullien (2003), Armstrong (2006), and Rochet and Tirole (2006). Weyl (2010) generalizes and unifies the models in these papers. Most of the subsequent literature builds on these papers. In the context of information goods, Parker and Van Alstyne (2000) introduced a model with interdependent demand between two groups of customers. Their model is a linear version of the model subsequently developed by Armstrong (2006). 4  Evans and Schmalensee (2007a). An appendix to this paper available at http://papers.ssrn.com/​ sol3/papers.cfm?abstract_id=2185373 provides a bibliography of the literature on multisided platforms through 2012. 5  OECD Competition Committee (2009). The first author was an advisor to the OECD and helped draft the OECD’s report.

406   David S. Evans and Richard Schmalensee literature is regularly cited in submissions in legal and investigative proceedings. Third, a number of large global multisided platforms have emerged as a result of ongoing revolutions involving the Internet, mobile devices, and information technology more broadly, as described in Evans (2008). These platforms have garnered considerable attention from antitrust authorities—for example the US Federal Trade Commission’s and the European Commission’s investigations of Google—and private complainants—such as Qihoo 360’s lawsuit against Tencent in China and the Consumer Watchdog complaint against Facebook in the United States (Forden and Bliss 2012; Kendall, Mock, and Catan 2012; Qihoo v. Tencent). Experienced antitrust practitioners know that no two businesses or markets are alike and may wonder why one class of businesses, multisided platforms, deserves a chapter in a volume on antitrust. After all, grocery stores and copper mines also differ in fundamental ways. Indeed, early in the emergence of the economic literature on multisided platforms, some commentators argued that there was nothing new in the economics of multisided platforms and thus no reason for competition analysis to treat them differently from grocery stores or copper mines. Few hold this view today. The economics literature that has developed since 2000 shows robustly that many results derived from models of one-sided businesses generally do not apply to multisided platforms that serve different interdependent customer groups. The clearest example of this involves price. When competition is imperfect, long-run equilibrium prices exceed marginal cost in traditional models but not necessarily in multisided platform models. Similarly, many of the analytical methods that are commonly used in antitrust matters, such as the SSNIP test or models for evaluating exclusionary abuses through tying, do not necessarily apply, without significant adaptation, to industries with multisided platforms. An additional reason for a separate focus on multisided platforms is that there are important similarities in the business models adopted by multisided platforms in different industries, and the literature has already identified some important determinants of both those similarities and of differences of the sort that distinguish grocery stores from copper mines. The selective survey in this chapter is intended to provide insights from across these industries that readers may find useful in specialized cases. In this sense, surveying multisided platforms is similar to considering franchises. The economics and the case law related to one kind of franchises or multisided platform are relevant to other kinds of franchises or multisided platforms, but in both cases it is important to pay attention to sources of differences between individual businesses and markets. As we noted above, some early commentators argued that the economics of multisided platforms was old wine in new bottles: just a new label for industries with indirect network effects that economists had written about since the mid-1980s. Others claimed that these platforms did not raise any issues that required significant modification to traditional antitrust tools.6 We have asserted above and will demonstrate below that the 6  See, e.g., Plaintiffs’ Pretrial Brief at 17-24, United States v. First Data Corporation, No. 03-2169 (D.D.C. December 10, 2003).

The Antitrust Analysis of Multisided Platform Businesses   407

burgeoning economics literature on multisided platforms has shown that there are new wines here—particularly good, complex vintages in fact—and that one-sided tools often do not apply, at least not without substantial changes, to multisided platforms. Much of the work discussed below is making its way into the interactions among authorities, complainants, defendants, and courts. And there is more to come.

18.2.  Economics of Multisided Platforms It is useful to start with an example of a two-sided platform to establish some basic concepts. OpenTable is an American-based company that provides a platform for fine-dining restaurants and consumers in various cities in the United States and other countries.7 It enables consumers to make and restaurants to accept reservations for tables over the Internet. It helped solve a transaction problem for consumers and restaurants. Consumers used to have to call a restaurant and, assuming they reached someone, ask whether a particular time was available for their party. If they were unable to make a reservation, they would repeat the process for another restaurant, perhaps many times. Restaurants used to have to devote resources to taking phone calls, many of which did not result in a reservation, and keeping track of the reservations they did take. As anyone who has used both methods has observed, the web-based reservation systems enable consumers to search very quickly for restaurants with availability and make reservations at them. For OpenTable to provide this service it needed to have significant numbers of both consumers and restaurants using its platform. OpenTable started by providing table management software to restaurants—a one-sided business. Once it had signed up a sufficiently large number of restaurants in some cities, it developed a web-based platform for consumers to make reservations that would be automatically incorporated into its table management software. OpenTable provided this service to consumers for free. In fact, the price to consumers is slightly negative: consumers earn modest rewards that can be used to reduce their bill at participating restaurants. Restaurants have to license OpenTable’s table management software and pay a fee for every patron they seat who has made a reservation through OpenTable. As of 2011 OpenTable had 25,000 restaurants in the United States and sat 9.0 million consumers a month. It earned $139.5 million in revenue (OpenTable Inc. 2011, 2, 7). OpenTable has several features that are common among multisided platforms. First, it has two sorts of indirect network externalities. There is a usage externality:  both consumers and restaurants benefit when the system is used to make

7 

For the company’s geographic scope, see OpenTable Inc. (2011), 3, 7–8.

408   David S. Evans and Richard Schmalensee a reservation.8 And there is a membership externality: the system is more valuable to consumers the more different restaurants it lets them access. It is not clear that restaurants benefit from having more consumers use the system beyond the benefits related to usage; consumers value variety, but restaurants probably do not. On the other hand, it is clear that the value of the table management software is greater the more consumers use it to make reservations. Second, OpenTable clearly facilitates valuable interactions between two distinct groups of agents:  consumers and restaurants. The fact that members of each group value interacting with members of the other group underlies the indirect network externalities just discussed and provided an opportunity for an entrepreneur to create a profit-making platform that could increase value for these economic agents by reducing the transactions costs they must incur in order to interact. Consumers and restaurants were able to find each other and make reservations before OpenTable, of course. One could even argue that the telephone network was the preexisting platform, which generally had no out-of-pocket costs for consumers. But the transactions costs of using the telephone were nonetheless much higher on both sides of the market. Third, the price structure for the two different types of economic agents, which determines the relative incremental profit earned from the two types, is an important tool in solving the coordination problem between the two sides in order to capture value from the externalities that link them. Most likely subsidizing consumers increased the value of the platform to restaurants, which resulted in OpenTable getting more restaurants, which in turn made if more attractive to consumers. Note also that restaurants pay a fixed membership fee to license the software that enables them to be on the platform as well as a usage fee when they take a reservation. It is not uncommon for platforms to charge fees of both sorts. Consumers pay no membership fee and enjoy a small negative usage fee. This sort of dramatic asymmetry, with one group of agents paying prices below marginal cost, is also not uncommon with multisided platforms.

18.2.1.  Definitions of a Multisided Platform Before relying on the multisided platform literature to analyze a business, one obviously needs to determine whether that business is in fact a multisided platform. Economists have proposed several definitions of a multisided platform; they will generally agree in specific cases. Rochet and Tirole (2006) focused on the price structure in defining two-sided platforms (which they refer to as two-sided markets): A market is two-sided if the platform can affect the volume of transactions by charging more to one side of the market and reducing the price paid by the other side by an equal

8  This distinction between usage and membership network externalities was first made by Rochet and Tirole (2006).

The Antitrust Analysis of Multisided Platform Businesses   409

amount; in other words, the price structure matters, and platforms must design it so as to bring both sides on board.

For a business to be two-sided it must reduce transactions costs for the two sides. Because of indirect network externalities there is interdependence between the demands of the two sides, and the price structure is used to balance membership and usage to maximize platform value. Rochet and Tirole emphasize that a business is not two-sided if the economic agents can easily defeat the pricing structure through side payments.9 In practice, it is hard to know as a theoretical matter whether economic agents could defeat a platform’s pricing structure. Restaurants could impose the charges they pay for reservations onto those who make them, thereby negating the OpenTable subsidies. In fact, restaurants do not do this despite complaining about the charges (Pastore 2010; Stross; 2010; Keneally 2011), perhaps because of transactions costs or because consumers would strongly object to such a charge, since making a reservation by phone is free—except for the time involved. Evans and Schmalensee (2007a) proposed a less formal definition that captures the key features of platform businesses (Evans and Schmalensee 2007a; also see Evans 2003b). A multisided platform (which they call an economic catalyst), “has (a) two or more groups of customers; (b) who need each other in some way; (c) but who cannot capture the value from their mutual attraction on their own; and (d) rely on the catalyst to facilitate value-creating interactions between them.” The focus of this definition is on the role of the platform in creating value that would not exist (or would be much smaller) in its absence. This value is created as a result of solving a coordination—and transaction cost—problem between the groups of customers. The generation and allocation of this value between the multiple sides is determined simultaneously. How much value each side gets determines whether they will participate and also how much is left over as profit for the platform. The price structure is critical in determining that allocation. The multisided platform literature assumes the presence of multiple customer groups with demand that is interdependent in various ways. Perhaps the best test of whether a particular situation involves one or more multisided platforms is whether the multisided platform literature provides useful insights that do not emerge from more standard analysis. In the case of OpenTable it seems apparent both that there are significant usage and membership externalities and that it would be difficult to explain OpenTable’s subsidization of reservation makers without taking those externalities into account. In the remainder of this chapter we distill insights from the multisided platform literature that are relevant to competition policy. Whether these insights apply to a particular entity, or to competition policy issues related to that entity, depends on the extent to which the entity is a multisided platform. When in doubt it is useful to consult the Rochet-Tirole or Evans-Schmalensee definition and to make sure that the assumptions of the multisided platform theoretical model(s) being considered apply to that entity. It 9  Therefore a necessary condition for a firm to be a multisided platform is that the Coase (1960) Theorem, which relies on there being zero transactions costs involved in side payments, does not apply.

410   David S. Evans and Richard Schmalensee is also important to note that many of the theoretical models of multisided platforms are heavily dependent on assumptions that may not apply in a particular application or, for that matter, in any application.

18.2.2. Externalities Externalities are a key aspect of multisided platforms. As we noted above, Rochet and Tirole (2006) observed there are two types of indirect externalities: usage externalities and membership externalities. A usage externality exists when two economic agents need to act together, to use the platform, to create value. We see that with OpenTable. There is a person who wants to dine at a restaurant at a particular time and a restaurant that would benefit from serving that individual at that time. They can enter into a value-increasing exchange only if they can get together. In practice that means the person and the restaurant finding each other and entering into a transaction. OpenTable and similar businesses help generate these usage externalities by making it easier for restaurants and diners to enter into this transaction. They also increase the value of usage externalities by increasing the quality of the matches: they make it easier for people to find the best restaurant for the particular occasion involved. It is possible that the usage externalities are positive for one type of economic agent but negative for another type of economic agent. So long as the net value of these externalities is positive there is a benefit to facilitating interaction, some of which the platform may be able to capture. Some advertising-supported media are examples. An advertiser benefits from being able to communicate with a possible customer, but consumers may place a negative value on seeing ads. The platform enables a value-increasing interaction by subsidizing the consumer so that she is willing to see the ad. Most advertising-supported media do this by bundling subsidized content with advertising. Anderson and Gabszewicz (2006), for instance, discuss how free television stations buy consumers’ attention with programming. There is a membership externality when the value received by agents on one side increases with the number of agents—or some related measure of their aggregate value—participating on the other side. As we noted above, OpenTable is more valuable to consumers the more restaurants that participate in the service. Smart mobile phone software platforms provide another example. Developers of applications value a platform more if there are more potential users; users value a platform more if there are more applications. This phenomenon results in the well-known positive feedback loop (Katz and Shapiro 1985; Farrell and Saloner 1985; Arthur 1989). More agents on one side attract more agents on the other side, thereby fueling growth. Yet another example of membership externalities is provided by the entry of Diners Club, the first general-purpose payment card. At its start it recruited 14 restaurants to take its card and several hundred consumers to carry its card in Manhattan. Driven by membership externalities, more restaurants and consumers joined its system in the

The Antitrust Analysis of Multisided Platform Businesses   411

ensuing months (Evans and Schmalensee 2007b, 1–2). Diners Club ignited. More restaurants joined to get access to consumers who wanted to use the card to pay and more consumers joined to pay at more restaurants. By its first anniversary in 1951, Diners Club had 42,000 individuals who carried its card to make payment and 330 merchants that that accepted the card for payment. By 1956, it was accepted at 9,000 merchants, with an annual transaction volume of $54 million (Evans and Schmalensee 2007b; Evans and Schmalensee 2005, 54). The platform plays a key role in creating these indirect network effects. In fact, as described in Evans and Schmalensee (2010), the major challenge for aspiring platforms is to get enough agents on each side to secure enough critical mass to propel indirect network effects.10 Platforms generate indirect network effects, and thus value for the economic agents they aspire to serve, through pricing, product design, marketing, and other efforts to attract agents on each side. Jullien (2011), for instance, has stressed the value of “divide and conquer” strategies for a start-up or an entrant challenging an established platform: subsidize agents in the most price-sensitive group, then use their participation to attract agents in the other group. The presence of these indirect network externalities has an important implication for economic analysis of multisided platforms. For traditional one-sided industries, economists ordinarily assume that demand depends on the price of the product as well as the prices of complements and substitutes. For multisided platforms, the demand by one group of economic agents also depends on the number (or other measures of the size and quality) of each of the other groups of economic agents that the platform serves. Loosely speaking, the sides are complements in demand. Failure to account for these demand interdependencies in an economic model renders the results of that model suspect if not completely unreliable when such interdependencies are important. We will return to this point when we discuss economic models commonly used in antitrust.

18.2.3. Pricing Pricing in two-sided platforms is more complex than in ordinary multiproduct businesses, and it depends on the nature of the platform. To illustrate both points, we briefly consider pricing in the two most basic models of two-sided platforms. In the first of these models, due to Rochet and Tirole (2003),11 a two-sided monopoly platform operates with no membership externalities, only usage externalities, and levies no membership charges, only per-transaction usage charges. The demand for transactions from group i is given by Di (Pi ), for i = 1, 2, where Pi is the per-transaction charge to members of group i. One can think of the two groups as merchants and consumers and the platform as a payment system that levies only per-transaction fees. The 10 

Zhou (2014) provides an empirical study of the critical mass problem in the US video game market. See also Schmalensee (2002), where this same model is applied to the analysis of payment card systems. 11 

412   David S. Evans and Richard Schmalensee number of transactions that actually occurs is proportional to the product of the groups’ demands in this model, so that, as in real payment systems, there is a value to balanced participation. The platform’s profit is given by

Π = (P1 − C1 ) + (P2 − C2 )  D1 ( P1 ) D2 ( P2 ) ,

where C i is the per-transaction cost of serving a member of group i. Let Ei be the (positive) elasticity of Di with respect to Pi. Then Rochet and Tirole (2003) show that the profit-maximizing prices satisfy the following two optimality conditions:

(P1 + P2 ) − (C1 + C2 ) = 1 , E1 + E2 (C1 + C2 )

and

P1 P2 = . E1 E2

The first of these resembles the classic Lerner condition for monopoly equilibrium; the total markup over cost is lower the higher is either demand elasticity. The second condition, however, makes clear that this is not an ordinary multiproduct firm. Such a firm would generally maximize profit by charging prices that are inversely related to demand elasticities, all else equal. Here, however, that condition is turned on its head: the optimal prices are directly proportional to demand elasticities. Intuitively, the reason is that the platform cares about balanced participation of the two groups, while balance has no value to an ordinary multiproduct firm. In the second basic model, due to Armstrong (2006), a two-sided monopoly platform operates with no usage externalities, only membership externalities, and levies no usage charges, only membership charges. One can think of a heterosexual singles bar or an academic journal. In the first example men value the presence of many women and vice versa; in the second example authors value journals with large audiences, and readers value journals in which many good authors want to publish. The demand of each group for membership depends both on the fee it is charged and on the number of members from the other group. The firm’s profit function in this model is given by

Π = ( P1 − C1 ) D1 ( P1 , Q2 ) + (P2 − C2 )D2 ( P2 , Q1 ) ,

(

)

where Qi is the number of members from group i and Qi = Di Pi , Q j , i = 1, 2, i ≠ j. This model is formally related to the classic model of a monopoly selling complements. In the classic example of coffee and cream, lowering the price of coffee increases the demand for cream because some individuals consume coffee and cream together. Here, however, there are two distinct groups. In the singles bar example, lowering the admission charge to women will increase the demand for admission by men as a reaction to the increased number of women in the bar.

The Antitrust Analysis of Multisided Platform Businesses   413

Unlike the Rochet-Tirole (2003) model, the Armstrong (2006) model does not yield simple optimality conditions that hold for all demand functions. Armstrong (2006) shows that in the special case where the Di functions are linear, the profit-maximizing prices satisfy the following conditions:

Pi − (Ci − θij ) Pi

=

1 , εi

i, j = 1, 2, i ≠ j.

Here εi is the elasticity of Di with respect to Pi, holding Qj constant, and θij is a positive term that measures the impact of increases in Qi on demand from group j, i, j = 1, 2 , i ≠ j . As in the case of complements, prices are lower than they would be in the absence of cross-effects. Schmalensee (2011) shows that in both of these models differences in demand functions can lead to highly skewed pricing of the sort that platform businesses like OpenTable often employ. Weyl (2010) explores a general model that has these two models as special cases, and he shows that they have rather different comparative static properties. While the Rochet-Tirole (2003) and Armstrong (2006) models form the foundation of much of the multisided platform literature, later authors have introduced additional factors in attempts to produce more tailored models of particular platform types. Hagiu (2009), for instance, modifies the Armstrong (2006) model to capture features of platforms like video game consoles, OpenTable, Amazon, or eBay, that serve to connect differentiated sellers with consumers. He finds that the stronger consumers’ preferences for variety, the larger the share of a monopoly platform’s profits that is optimally derived from sellers. He argues that this explains why video game platforms derive very little of their profits from consumers of video games, who have a strong preference for variety, while most other software platforms, like Microsoft’s Windows and Apple’s iOS, derive most of their profits from consumers.

18.2.4.  Consumer and Social Welfare For antitrust an important and immediate implication of the multisided platform models is that an accurate analysis of the impact of any platform decision on consumer welfare must take into account all interdependent customer groups the platform serves. Search engines, for example, provide value to three distinct groups of economic agents: (1) websites that are indexed and made available to people through search queries; (2) people making search queries; and (3) advertisers who are seeking to reach the people who are looking at the search-results page from the query. There are usage and membership externalities across all three groups. The search-engine platform has to balance the interests of these three groups to provide value to them and maximize its own profit. Business decisions that affect the welfare of one group of users are likely to affect the other groups of users through indirect network externalities.

414   David S. Evans and Richard Schmalensee Another welfare issue concerns the relationship between the profit-maximizing decisions by a platform and the social welfare maximizing decisions. There are two potential market failures resulting from multisided platforms. The first one is the traditional market power failure. In the absence of perfect competition, which is an implausible market structure for platform-based industries, the platform will set its overall price level higher than is socially desirable. It will earn at least short-term profits that exceed the competitive level. Since most firms have market power, and many must have some market power to compensate for fixed costs and risk taking, the market power failure is not remarkable for multisided platforms. The market power-related market failure is similar to that for a one-sided firm. The fact that platform-based industries deviate from perfect competition should not raise competition policy concerns any more than the fact that almost all real-world single-sided markets also deviate from perfect competition. The second possible market failure can result from the platform choosing a price structure that does not maximize social welfare. In the two basic models considered just above, Weyl (2010) shows that this distortion arises because the platform considers the impact of its pricing on the marginal users in the groups it serves in order to balance participation, while the impact on the average users is what determines the effect on social welfare. This sort of distortion was first pointed out by Spence (1975) in a model of quality choice by a monopoly. It arises, in principle, whenever a firm with any market power has more than one decision variable and faces buyers who are affected differently by the levels of those variables—that is, almost universally. And, unlike the price level distortion, even its direction depends fundamentally on details of the demand structure: Spence (1975) shows that market-determined quality may be either too high or too low under plausible conditions. Perhaps because of this complexity, this sort of distortion has not received much attention from antitrust authorities or others concerned with public policy. Possible distortions created by payment card interchange fees are a notable exception to this generalization. These fees, which primarily affect the price structure, rather than the overall price level, are paid by merchant acquirers (and passed on at least in part to merchants) to bank issuers (and passed on at least in part to consumers) and thereby affect the relative prices paid by merchants and cardholders. There is no general reason why the profit-maximizing interchange fee would also maximize social welfare. Bedre-Defolie and Calvano (2013), for example, show that under certain assumptions payment networks will set the interchange higher than would be socially optimal.12 However, determining the socially optimal interchange fee depends on detailed features of cost and demand structures and is empirically demanding. Moreover, even when there is a theoretical reason to believe that the socially and privately optimal pricing structures are different, there is no a priori reason to believe that they are very different—as observed by Calvano (2011). 12  There is an immense literature on the economics of interchange fees, begun by Baxter (1983). In addition to Bedre-Defolie and Calvano (2013), other recent contributions include Prager et al (2009), Rochet and Wright (2010), and Wang (2010).

The Antitrust Analysis of Multisided Platform Businesses   415

18.3.  Competition among Multisided Platforms The typical US consumer has many different types of payment methods:  cash, checks from at least one depository institution, at least one debit card, and several credit cards (Federal Reserve Board of Governors 2010; Federal Reserve Bank of Boston 2009). Most retail stores accept cash, checks, debit, and credit cards. Each of these payment methods is provided by a two-sided platform serving consumers and merchants. In the United States, debit cards are provided through platforms operated by MasterCard and Visa, credit cards through platforms operated by American Express, Discover, MasterCard, and Visa; cash through a platform operated by the US Department of the Treasury and the Federal Reserve System; and checks through a platform operated by the Federal Reserve System. Some large retail stores also offer store credit or charge cards, which many consumers also use. These store cards are traditional single-sided products. One can debate which of these payment methods compete so closely with each other they should be included in the same relevant market for a merger or antitrust claim—a topic we consider in the next section. But it is clear consumers and retailers use all of these methods, they all compete with each other to some degree, and this competition has been going on for decades. Diners Club was able to break into the mature payments industry in the United States in 1950 through product differentiation. There were two established payment platforms—cash and checks. Some retailers also had store cards. Diners Club introduced a charge-card platform that, like cash and checks, many consumers and merchants could use for payments. Unlike checks, Diners Club guaranteed payment to the merchant. Unlike store cards, consumers could use the card at many merchants. And unlike cash and checks, the consumer did not have to pay for the transaction until her monthly charge card bill came due. Despite the advantage that cash and checks had from indirect network effects, Diners Club and subsequent payment card entrants were able to introduce a new payment platform. More than fifty years later, although shares of payments have shifted dramatically among them, all of these platforms coexist. This competition and its evolution are hard to reconcile with the simplest models of network effects. In these models, a platform is more attractive the more participants it has, and these demand-side scale economies propel markets to monopoly. In the early literature on direct network effects, there was emphasis on the impact of certain events—including anticompetitive behavior—on tipping markets to monopoly.13 One of the major contributions the multisided platform literature has made to industrial organization has been to demonstrate that indirect network effects are important 13  Useful overviews of this literature are provided by Katz and Shapiro (1994) and Farrell and Klemperer (2006).

416   David S. Evans and Richard Schmalensee across a wide range of industries—from low-tech industries such as shopping malls to high-tech ones such as mobile phone operating systems. In simple models, indirect network effects can also produce demand-side economies of scale that lead to monopoly: increased participation on one side of the platform makes it more attractive to the other side, leading to increased participation there, making participation by the first side more attractive, and so on. But many of the industries in which indirect network effects are important do not have a single monopoly provider and do not seem to be tending toward monopoly. For example, in the United States, in addition to several payment systems, there are several competing financial exchanges, numerous magazines even in narrow categories such women’s fashion, multiple shopping malls that people in metropolitan areas can patronize, several mobile phone operating systems used by consumers and developers, numerous dating venues, and three general purpose search engines. Two features missing from simple models help explain this apparent discrepancy. First, competing platforms typically offer differentiated products. Second, in some settings customers on one or more sides of the business can patronize more than one platform—a phenomenon that has come to be called “multihoming.” Nevertheless, indirect network effects, like ordinary supply-side scale economies, do tend to limit the number of viable multisided platforms in any market. Most multisided platforms thus face a relatively small number of competitors as they seek to gain customers on all sides of their businesses. Often, both static and dynamic competition are important in industries with multisided platforms.

18.3.1.  Product Differentiation The traditional concepts of product differentiation (see Tirole 1988, ­chapter  7) for one-sided firms apply to multisided platforms. There is variation across consumers both in the valuation of various product attributes (horizontal differentiation) and in the willingness or ability to pay for quality (vertical differentiation).14 Firms can select product attributes to appeal to particular groups of consumers and to differentiate themselves from competitors. For one-sided firms, horizontal and vertical differentiation locates the firm near a pool of potential customers and helps determine pricing. For multisided platforms, by determining the customers on one side, horizontal and vertical differentiation affect demand on the other sides. Because of these interdependencies, a platform must usually make differentiation decisions jointly for all of the sides it serves. Moreover, the selection of customers on one side is one possible way to differentiate the platform horizontally or vertically. 14  Ambrus and Argenziano (2009) consider two-sided platforms serving customers who differ in their valuation of an important dimension of quality: the number of customers on the other side of the platform. They show that multiple asymmetric platforms can exist in equilibrium.

The Antitrust Analysis of Multisided Platform Businesses   417

A shopping mall developer, for example, must decide on a number of different product attributes such as location, size, parking, and quality of construction. But it also needs to decide what kind of stores and customers it wants to attract. Those are obviously interdependent. It could be an upscale mall and only rent space to merchants with an upscale clientele. If it succeeds in attracting enough such merchants, it will tend to attract an upscale clientele. In order to do this, of course, it is likely to make other decisions—such as locating close to wealthy towns and using better finishes—that help attract wealthy customers and merchants they tend to patronize.15 Product differentiation, as this example suggests, is a tactic that firms can use to create value by making it easier for agents to find counterparties for value-increasing exchange. The upscale mall, for example, makes it easier for shoppers to find stores that serve their tastes and easier for stores to find customers. In some cases, platforms differentiate themselves by using what Evans (2012a), following Strahilevitz (2006), has called the exclusionary vibe.16 The dating site JDate does not prevent Gentiles from using it but engages in various actions that signal that it is primarily for Jewish singles looking for other Jewish singles. Platforms can also create value for agents on one side by limiting how much competition they face for a match, as observed by Halaburda and Piskorski (2013). Product differentiation is a key reason why many industries with multisided platforms have multiple competitors even though indirect network effects and sometimes economies of scale would seem to propel them to monopolies. Job placement provides an interesting example. The online portion of this industry consists of job boards that help match job searchers with employers through online postings and search. In the United States there are two large job boards that cover many different job categories. But then there are hundreds of other job boards that specialize in different job segments such as professionals (LinkedIn.com) and media jobs (mediabistro.com). By specializing, these job boards presumably increase matching efficiency. Beyond the job boards there are recruiting services that work for employers or employees. The result is a highly fragmented industry of two-sided platforms.

18.3.2. Multihoming The competitive dynamics of multisided platforms depend in theory and in practice on the number of platforms that individual economic agents on each side use, on differences between the two sides in the number of platforms used, and on the ability of an agent on one side to dictate the choice of platform for the other side. Rochet and Tirole (2003) observed that one of the key competitive aspects of multisided platforms was the extent to which economic agents engaged in what they called 15  Galeotti and Moraga-Gonzalez (2009) provide a model of a shopping mall that attracts horizontally differentiated retailers as well as consumers. The analysis of Hagiu (2009) is also relevant. 16  For a discussion of the exclusionary vibe in the context of multisided platforms see Evans (2012a).

418   David S. Evans and Richard Schmalensee “single-homing” or “multihoming.” An economic agent single-homes if she uses only one platform in a particular industry and multihomes if she uses several. In the cases of payments, consumers and merchants both generally use several payment platforms and therefore multihome in this sense. Armstrong (2006) showed the importance of “multihoming” for competition. Suppose platforms in some market create value by having agents of Type A and Type B as members. If Type A agents only join one platform, then Type B agents can only gain access to Type A agents by joining that same platform. That makes the Type A side of a platform what Armstrong called a “competitive bottleneck.”17 When there is single homing on one side and multihoming on the other side in his model, Armstrong shows that platforms will compete aggressively for the single-homing customer who will therefore pay low prices. With these customers on board, the platform will then earn its profits from the customers who multihome on the other side. Armstrong and Wright (2007) show that if competing two-sided platforms are viewed as homogeneous by members of one group but are viewed as differentiated by members of the other, the latter will single-home, the former will multihome, and “competitive bottlenecks” will arise endogenously. Sometimes one set of multihoming agents can dictate the choice of platform to agents on the other side of the market. As we have noted, most consumers use multiple payment methods and even use multiple payment cards, and most merchants accept all of the payment alternatives. One can argue, however, that in practice the consumer dictates which payment system is used. The consumer generally offers one particular payment method at checkout. The merchant then has to decide whether to reject that payment method. If it does, it might lose a sale. In addition, there may be other reasons why the merchant cannot steer the consumer away from his chosen method—laws against surcharging payment methods, restrictions imposed by the platform involved, customs relating to payment acceptance, or transaction costs. If the consumer decides then, by the logic of competitive bottlenecks, the payment platforms have an incentive to compete aggressively for the consumer to use their payment method. That is a possible explanation for providing “rewards” for people to use cards. Bedre-Defolie and Calvano (2013) rely on this assumption in showing that payment card systems have an incentive to subsidize card users at the expense of merchants more than is socially optimal. It is not clear how robust the “competitive bottleneck” argument is, however. In software platforms, for instance, the price structure appears to be the opposite of what the competitive bottlenecks theory would predict. Most personal computer users rely on a single software platform, while most developers write for multiple platforms (Evans, Hagiu, and Schmalensee 2006). Yet personal computer software providers generally

17  This term seems to have first been used in connection with mobile telephony, where networks may compete for subscribers but be in a monopoly position for providing access to those subscribers; see Armstrong (2002).

The Antitrust Analysis of Multisided Platform Businesses   419

make their platforms available for free, or at low cost, to applications developers and earn profits from the single-homing user side.

18.3.3.  Asymmetric Competition The theoretical economics literature on multisided platforms has focused on two major cases:  monopoly platforms and competition among (possibly differentiated) multisided platforms that serve the same customer groups. Many platforms, in fact, face more complicated competitive environments as observed by Evans and Noel (2005) and Eisenmann, Parker, and Van Alstyne (2011). An n-sided platform can face competition from the following: • A single-sided firm on any side. We saw that two-sided payment platforms face competition from one-sided store cards. The three-sided personal computer software platforms competed at least initially with one-sided platforms such as the IBM mainframe (the mainframe integrated an operating system and hardware and did not encourage the development of general complementary applications). • A multisided platform that competes on some but not all sides. The Android operating system, for example, serves as a platform for handset manufacturers, users, mobile operators, and software developers. The Apple iPhone operating system serves as a platform for all of these except handset manufacturers, since Apple does not license its operating system to other handset manufacturers. • A multisided platform that has the same sides plus additional ones. That is true for Apple relative to Android. It is also true to a large degree for search platforms versus social networking platforms. Both earn revenue from advertising, but social networking platforms earn revenue from application developers—especially game developers. Online advertising is an example of many platforms serving one side—advertisers—but providing very different products to secure the attention of consumers. In China, for example, the following platforms all sell online advertising to users but provide very different services to them:  Tencent (instant messaging), Baidu (search engine), Sina (integrated portal), Qihoo (antivirus), Netease (integrated portal), Taobao (online shopping), Sohu (integrated portal), Youku (online video), Tudou (Online Video), Google (search engine), Tianya (community), Home of Websites (site map), Ku6 (online video), Thunder Video (online video), and Phoenix (integrated portal). The platforms intersect on the advertising side but deviate, at least in terms of the services offered, on the user side.18 Of course they are all seeking attention from consumers, which is what they are providing to advertisers. 18  Evans (2013a) provides a general discussion of platform businesses that both seek and provide attention.

420   David S. Evans and Richard Schmalensee

18.4.  Market Definition and Market Power The fundamental service provided by multisided platforms is the ability of economic agents on each side to interact in a valuable way with economic agents on other sides. The OpenTable platform, for instance, makes it easier for consumers and restaurants to schedule a time to dine. A platform may provide economic agents with various products and services to persuade them to use the platform and help facilitate the process of searching, matching, and interacting. OpenTable provides restaurants with table management and reservation software. It provides consumers with a website that enables them to search for restaurants, see ratings, determine availability for particular times and numbers of guests, and to make a reservation. But ultimately the platform is generating value from bringing members of these two groups of customers together. The goods and services provided to each side are usually the means to that end. OpenTable provided restaurants with a suite of software services in part to get them to join its network, and it provided consumers with a suite of tools to get them to use its network. (It would be easy for consumers to multihome; less easy for restaurants.) These linkages across the multiple groups of customers and the products and services being offered to each group have to be accounted for in the analysis of the relevant antitrust market and the assessment of market power.19 Market definition and market power analysis are primarily methods for summarizing succinctly the extent to which competitive constraints limit the ability of a firm to engage in various kinds of behavior that may raise antitrust concerns. Market definition identifies the sources of demand-side and supply-side constraints that matter in assessing market power. Market power analysis assesses whether these constraints prevent a firm from charging prices that are noticeably higher than under competition or otherwise engaging in undesirable behavior that a competitive firm could not do profitably. We therefore begin by considering differences between competitive constraints for multisided firms and those for single-sided firms.

18.4.1.  Competitive Constraints Indirect network effects result in three major considerations for the analysis of competitive constraints. First, the positive feedbacks between the sides that indirect network effects produce need to be considered when analyzing the profitability of increasing price. Consider a platform with sides A and B. An increase in price, or a reduction in quality, to A-type customers will reduce the number of A’s that belong to the platform and the extent to which 19  For our general views on the proper analysis of market definition and market power in antitrust matters see Chang, Evans, and Schmalensee (2011a, 2011b) and Evans (2012b).

The Antitrust Analysis of Multisided Platform Businesses   421

they participate in it. Since B-type customers value the platform because of their ability to access and interact with A-type customers, the demand by B’s will fall even if the prices they face have remained the same. The result is that the demand by As will fall more since the platform is less valuable to them now that it has fewer B’s. As noted by Armstrong (2006), the demand on each side of the platform is more elastic, and the profitability of a price increase is lower, when these positive feedback effects are considered than when they are not considered. In a study of magazines, Song (2013) found that these effects doubled estimated price elasticities. Of course, it is possible that these feedback effects are small and could be ignored in any particular setting. But if the subject of an antitrust inquiry is a multisided platform, one would at least need to inquire into the strength of these feedback effects in assessing the profitability of raising prices on any side. Second, the competitive constraints on raising price to one side, or engaging in any other strategy, can come directly or indirectly from any and all sides of competing platforms. A platform that considers raising its price to one side, for example, has to consider the extent to which customers leave that side; how that affects customer losses on other sides; the extent to which other platforms pick up those customers; and how the addition of customers on each side of a competing platform increases the value of that platform to the other sides through positive feedback effects. More generally, platforms engaging in any competitive move affecting one group it serves would need to consider counterstrategies aimed at that group or any other it serves. For example, if a platform attempts to engage in an exclusionary strategy on one side it is possible that a rival could counter with strategies that involve that, the other, or both sides. It is, of course, an empirical matter as to whether customers can turn to other platforms, as is the extent to which indirect network effects require looking beyond competition between the similar sides of alternative platforms. Third, the existence of indirect network effects can also limit supply-side substitutability and increase entry barriers for multisided platforms. Successful incumbent firms have, by definition, obtained a critical mass of users on their several sides and benefit from the positive feedback effects between these customer groups. Entrants have to obtain critical mass as well, and that often takes time. Moreover, in a mature market the entrant has the challenge of persuading users who may benefit from these positive feedback effects to switch to a platform that has a smaller and therefore possibly less valuable group of customers on the other side.20 In practice product differentiation and the possibilities of multihoming are critical determinants of the degree of difficulty faced by entrants. 20 In U.S. v. Microsoft the government argued that Microsoft had monopoly power in software operating systems because among other things it was protected by an insurmountable “applications barrier to entry.” In the language of multisided platforms, which had not yet emerged when that case was tried, Windows was a multisided platform serving end users, developers who write applications, and hardware makers. All three are linked by indirect network externalities. Windows had a critical mass of customers on all three sides, and each group had made investments in the platform. The market power question is whether an entrant could have overcome these advantages and secured critical mass for the ignition, and generation of positive feedback effects, for an alternative platform. See United States vs. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001); Evans et al. (2000). The answers to questions of this sort are likely to be market-specific: Apple’s iPhone had an enormous stock of applications (or apps) before the first Android smartphone was marketed, for instance, yet as this is written Android phones are substantially outselling iPhones.

422   David S. Evans and Richard Schmalensee

18.4.2.  Market Power Market power is commonly defined as the ability of a firm (or group of firms) to raise prices significantly above the competitive level, although there is no consensus on exactly how much above the competitive level constitutes significant market power. The competitive level is generally taken to be the price that would prevail under perfect competition, so that price is set at marginal cost.21 There are difficulties and nuances in analyzing market power in single-sided markets that we do not address here (Chang, Evans, and Schmalensee 2011b). Instead we focus on how the existence of demand interdependencies affects commonly used approaches to assessing market power. Some models of multisided platforms, notably the Rochet-Tirole (2003) model discussed above, may provide a theoretical rationale for comparing price and marginal cost by computing a multisided version of the familiar Lerner Index. Similar measures can be computed even when they lack a theoretical rationale, of course, but their interpretation is unclear in such cases. In general, multisided price-cost margins face the same difficulties in interpretation as do the single-sided price cost margins. Because competitive firms need to recover fixed costs, unless marginal cost is rising and exceeds average variable coat, prices cannot be equal to marginal cost in competitive equilibrium. Marginal cost is clearly not the right benchmark for software or Internet-based products, for which marginal cost is typically negligible. And, as Chamberlin (1933) noted long ago, competitive equilibrium in the presence of product differentiation necessarily involves prices above marginal cost. Market share is often used to assess market power for single-sided firms. The theoretical justification for inferring market power from market share is weak for single-sided firms. It is no stronger for multisided ones. In addition, it is not always clear how to compute “share” for multisided firms. Consider a software platform. One of the main “products” that software developers get from the platform is access to users; one of the main “products” that users get is the access to software developers. One could compare shares for each of these sides across platforms and then make a judgment about market power based on looking at the shares for both sides, but there is no reason to expect those shares to be equal. In addition, multisided platforms often provide one of their products for free or at a subsidized price. In these cases it is not possible to calculate a value-based market share, which is what is ordinarily recommended, since the price does not in fact reflect the value received by the user. In some cases there is a common metric for the multiple sides that provides a natural way to compute share. For example, buyers and sellers both use

21  Areeda and Hovenkamp (2002, par. 501); Motta (2004, 115); Elhauge and Geradin (2007, 238). In the EU, dominance has also been defined as holding “a position of economic strength enjoyed by an undertaking which enables it to prevent effective competition being maintained on the relevant market by affording it the power to behave to an appreciable extent independently of its competitors, customers and ultimately of its consumers”. See Case 27/76, United Brands Company and United Brands Continental BV v. European Commission, 1978 E.C.R. (1978).

The Antitrust Analysis of Multisided Platform Businesses   423

payment systems to execute transactions, and one could calculate shares based on the number or value of these transactions. Several authors have warned against basing judgments about market power on analysis of only a single side of a multisided platform (Argentesi and Filistrucchi 2007; Evans 2003b; Song 2013; Weyl 2010; Wright 2004). As noted above, it is theoretically possible and empirically common for platforms to have prices that are significantly above marginal cost on one side and at or below marginal cost on the other side. A platform could have a monopoly in which it earns significantly more than the competitive rate of return yet price at or below marginal cost on one side. Therefore examining price on that side would result in a false negative test result for market power. A platform could also earn a competitive rate of return yet price significantly above marginal cost on one side. Therefore examining price on that side would result in a false positive test result for market power. In principle an analyst could adjust these prices by assessing the marginal cost of securing access to a customer on the other side, thereby lowering the effective margin for the high-price side and raising the effective margin for the low-price side. But this would not be a simple exercise and would still face the same difficulties as single-sided price-cost margin analyses. Profitability is also sometimes used to assess market power for single-sided firms. For multisided platforms the economic rate of return is an appealing measure of market power because it assesses the extent to which the platform has been able, through setting prices for its multiple sides, to earn more than a competitive return. Rates of return vary over time, however, and it is well known that measuring the economic rate of return and comparing it to the competitive rate of return is challenging at best (Fisher and McGowan 1983; Fisher 1987; Feenstra and Wang 2000). The manner in which accounting data are collected and reported makes accurate measurements and comparisons difficult, as does the problem of accounting for risk and its compensation. There is no single reliable method for assessing market power for traditional firms and it is usually recommended that analysts consider multiple sources of evidence to reduce the chances of false positives or false negatives. Exclusive reliance on mechanical measures such as market share or price-cost margins in determining market power is not advisable. The same issues apply to multisided platforms but, as we have discussed, there are further problems. A thorough review of competitive constraints is normally needed, and analysts need to make judgments based on multiple sources of evidence on these constraints. In evaluating these constraints, it is important that the analyst consider the role of interdependent demand as discussed above.

18.4.3.  Market Definition The purpose of market definition is to identify the competitive constraints on the supplier of the product under consideration—the market forces that reduce the profitability of raising prices above competitive levels or lowering quality (European Commission 1997, 5; UK Office of Fair Trading 2004, 3; US Department of Justice and Federal

424   David S. Evans and Richard Schmalensee Trade Commission 2010, 9–10). As noted above, these constraints may be imposed by single-sided firms on any side of the platform under consideration, by other platforms that serve the same customer groups, by other platforms that serve some but not all of the same groups, or by other platforms that serve the same groups plus others. Although some of these may be unimportant in particular instances, it is impossible to rule any of them out without investigation (Rysman 2009). The literature on multisided platforms has examined three issues related to market definition.22 The first concerns the consequences of applying analytical tools that were developed for single-sided firms to defining markets for a product offered on one side of a multisided platform. Evans and Noel (2005, 2008) show that the failure to consider positive feedback effects in demand can result in significantly overstating or understating the breadth of the market, depending on the analytical approach. Consider the simple hypothetical monopolist test.23 Suppose a small but significant, nontransitory price increase is profitable on one side under the assumption that nothing changes on the other side of the platforms included in the hypothetical monopoly. Therefore one could conclude that the products considered constitute a relevant antitrust market. However, as we have noted, a price increase on one side results in a reduction of demand by customers for that side and, through positive feedback effects, a reduction in the demand for the other side; the decline in demand on the other side further reduces the demand on the first side. Consequently, one might conclude after considering the positive feedback effects that the price increase is unprofitable. In that case the market is defined too narrowly. One can identify other techniques, such as those involving critical loss analysis, which when applied to a single side of a multisided platform would result in defining markets too broadly. The key point is that it is wrong as a matter of economics to ignore significant demand interdependencies among the multiple platform sides.24 The second issue concerns determining cases in which it is possible to easily adapt existing tools for market definition to multisided platforms. A special case is when the two sides are tied together in a fixed proportion, as in the Rochet-Tirole (2003) model. 22 

See Evans (2012c) for an overview. It starts with a hypothetical monopolist, which often consists of two firms that seek to merge. It then asks whether this hypothetical monopolist could profitably increase price by a small but significant and nontransitory amount above the current level. If the evidence implies that it could not, then there must be additional substitutes that constrain its pricing. The test proceeds in principle by including substitutes in the hypothetical monopolist until one obtains an aggregate that, if acting as a single firm, could profitably increase price, and that aggregate defines the relevant market. See Kaplow in this volume. 24  More generally, Evans and Noel (2008) show that there are two biases. It is possible to predict whether these biases lead to too narrow or too broad a market in the cases where the platforms are symmetric in the sense that they serve the same groups, face the same elasticities of demand, and have the same degree of positive feedback effects. There is an estimation bias that results from estimating a demand system for the potential group of products that ignores positive feedback effects. This bias leads to markets that are too narrow in the symmetric case. There is the Lerner bias that results from using the single-sided price-cost margin to estimate demand elasticity and then using this in a critical loss formula. This would result in defining markets too broadly in the symmetric case. It is not generally possible to predict the directions of the biases in the empirically more common case of asymmetric platforms. 23 

The Antitrust Analysis of Multisided Platform Businesses   425

It is then possible to define a composite price that is equal to the sum of the prices that each customer side pays. The analyst can, in principle, conduct the SSNIP, critical loss, and Lerner-type market power analysis using this composite price. In the case of a critical loss analysis the Lerner-based elasticity of demand would be based on the composite price and the composite marginal cost of providing the service to the two sides, though the same concerns about these approaches in one-sided analyses would generally apply. Emch and Thompson (2006) propose applying this approach to payment cards. The composite price includes the fees charged to merchant acquirers for each transaction (a network fee plus an interchange fee) and the fees charged to issuers for each transaction (a network fee minus the interchange fee which they are paid). The US Department of Justice adopted this approach in a case involving payment cards.25 Many multisided platforms do not, however, provide services that are consumed in fixed proportions by the multiple sides. For example, in the case of mobile software platforms software developers are provided access to users, while users are provided access to applications. The third area involves developing general tools that can be used for assessing market definition for multisided platforms. These generally involve econometric models that explicitly account for interdependencies in demand between the various platform sides. To date, these models have been developed and deployed entirely in the context of mergers and primarily for newspapers. We discuss them in the next section.

18.4.4.  Multisided Platform Markets As the discussion above demonstrates, multisided platforms do not fit neatly into the standard approaches for assessing market definition and market power. Moreover, extending these approaches to deal with multisided platforms rigorously is technically and empirically challenging. There is therefore a temptation on the part of analysts to do the best they can with standard tools, particularly when a case is presented as dealing with only one side of a platform’s business—magazine advertising rates, for instance. While ideal tools for dealing with many situations do not yet exist, two basic limitations of standard tools must be dealt with to avoid serious errors. First, the analysis needs to consider how positive feedback effects between demands on the multiple sides affect competition on the side under examination. It is a fundamental mistake to ignore these positive feedback effects, without at least first evaluating their significance, and therefore conduct a market definition analysis that focuses on one side and in effect assumes the other sides away. These features of the market could be considered in the market power prong of the analysis, but they must be considered somewhere. Second, the analysis needs consider the welfare of all customer groups. When there are material demand interdependencies the welfare of the customers on the multiple 25  Hesse and Soven (2006); U.S. vs. First Data Corp., No. 03-CV-02169 (D.D.C. May 25, 2004), Final Judgment.

426   David S. Evans and Richard Schmalensee sides are inextricably intertwined, and may move in opposite directions as price structures change. In an empirically based simulation of hypothetical magazine mergers, for instance, Song (2013) found that mergers would sometimes benefit subscribers and harm advertisers and sometimes would have the opposite effects. Similarly, Jeziorski (forthcoming) finds that the 1996–2006 merger wave in US radio harmed advertisers but benefited listeners. In these cases and in general, there is no compelling public policy rationale for ignoring the welfare of any group of customers as an artifice of market definition that focuses only on one side. Song’s (2013) analysis indicates that even if advertisers oppose a particular magazine merger that would harm them and present a compelling analysis of a “magazine advertising market,” it does not follow that the merger is socially undesirable.

18.5. Mergers The basic question in a merger inquiry concerns whether the proposed combination is likely to decrease consumer or social welfare significantly after accounting for unilateral and coordinated effects and efficiency (see Willig and Keating, in this volume; Jayaratne and Ordover, in this volume). We first discuss some significant general issues that arise in the unilateral effects analysis of mergers between multisided platforms in addition to those discussed above concerning the analysis of market definition and market power.26 (We treat coordinated effects generally in section 18.7, below.) While the bulk of the economic literature on multisided platforms has been theoretical, actual and potential mergers involving advertiser-supported media have been subjected to a good deal of empirical analysis. We then provide a brief overview of some recent empirical work on mergers of media businesses in section 18.5.2.

18.5.1.  Predicting Unilateral Effects of Multisided Platform Mergers There are four general points to be made here: First, the “off the shelf ” analytical tools for assessing unilateral effects from mergers between one-sided firms may yield incorrect assessments insofar as they fail to account for interdependencies in demand among the multiple sides. Unfortunately, the multisided extensions of the single-sided tools used for back-of-the-envelope calculations of

26  An early and important contribution to the literature on multisided platforms is Rysman (2004), which developed and estimated a structural model of Yellow Pages. Subsequent papers that have used a similar framework to analyze mergers of multisided platforms have included Affeldt, Filistrucchi, and Klein (2013); Chandra and Collard-Wexler (2009); Filistrucchi, Klein, and Michielsen (2012); Jeziorski (forthcoming); and Song (2013).

The Antitrust Analysis of Multisided Platform Businesses   427

unilateral effects result in complex formulae that require estimates that are not likely to be readily available to the analyst. For example, Affeldt, Filistrucchi, and Klein (2013) and White and Weyl (2012) explore how the UPP analysis introduced by Farrell and Shapiro (2010), and subsequently included in the 2011 Horizontal Merger Guidelines in the U.S., must be modified to account for indirect network effects in analyzing mergers of two-sided platforms. The formulas become a good deal more complex because a change in any one price affects all four quantities. To do a complete UPP analysis of a merger between two two-sided platforms, six diversion ratios must be estimated. These necessarily include estimates of indirect network effects. Similarly, Filistrucchi, Klein, and Michielsen (2012) show how to perform a two-sided SSNIP test: they estimate the profitability of a small but significant and nontransitory postmerger increase in either or both of a two-sided platform’s prices. This analysis requires a complete structural model of the firms’ demands, including both cross-price effects and indirect network effects, a good deal more information than is necessary in the case of ordinary single-sided firms. Second, in some cases it is at least theoretically possible for a merger of two-sided firms to result in price reductions to both sides even in the absence of efficiency gains. In the Chandra and Collard-Wexler (2009) model of a two-newspaper market, the main initial effect of a merger is that if a price increase of one paper causes a reader to switch papers, she is not lost to the merged firm. The way the model is set up, the first reader to switch away in response to a paper’s price increase is always the least profitable for that paper. In fact, both newspapers, which are assumed unable to price discriminate, could well be losing money on their marginal readers at their optimal premerger prices. At those prices, by definition, a tiny increase in either paper’s per-copy price would leave its profits unchanged. (That is, the derivative of each paper’s profits with respect the per-copy price is zero.) Suppose at its optimal premerger price paper A is losing money on its marginal reader. Postmerger, a tiny increase in its price would lower the profits of the merged firm, since the unprofitable marginal reader would switch to the firm’s other paper, thus lowering its profits. (That is, the derivative of the merged firm’s profit with respect to the per-copy price of paper A is negative at the optimal premerger prices.) If a small increase in either paper’s price would reduce the merged firm’s profit, it follows that a small decrease in subscription prices at both papers would increase profits. Because consumers’ propensity to subscribe to either paper is assumed to be correlated with their attractiveness to advertisers at that paper, these price cuts bring in subscribers who lower the average attractiveness to advertisers, thus also making a lower per-subscriber advertising rate optimal. As this discussion indicates, the possibility of this particular result depends on a number of special assumptions, and even then whether prices go up or down depends on the values of particular parameters in the model. This model is certainly not generally applicable. But it remains to be seen whether the possibility to which it points—a platform merger lowering profit-maximizing prices to both sides even without efficiency gains— is also present in other models that are descriptive of other market settings. In the

428   David S. Evans and Richard Schmalensee absence of further study, one must conclude that it appears possible that some mergers of multisided firms can lead to price cuts on all sides of the market even in the absence of efficiency gains—something that is simply impossible for mergers of single-sided firms. Although this possibility may turn out to be a very rare occurrence, its existence emphasizes once more that analysts need to consider the multisided aspects of mergers carefully and avoid mechanical analysis of multisided mergers with traditional one-sided tools. Third, all else equal a merger of multisided platforms would ordinarily increase indirect network externalities by increasing the size of all customer groups and thereby provide efficiency benefits.27 There is no similar presumption in the case of mergers of single-sided firms. To evaluate the impact of the merger on consumer (or social) welfare, analysts need to assess the value of these externalities. It is particularly important to do so since prices could increase to consumers on one or more side, while the value consumers are receiving on that side has increased even more as a result of positive externalities. Fourth, to evaluate the impact of a merger of multisided platforms on consumer (or social) welfare it is necessary to consider the impact of all sides. A merger could benefit consumers on one side but harm those on the other side, and the net effect of the merger across all customer groups could therefore be positive or negative. Suppose, for instance, that OpenTable proposed a merger with a competitor and that it is determined that the merged firm would likely increase prices to restaurants. It does not follow that the merger is undesirable, however. Restaurants would likely have access to more consumers, and that might more than make up for the price increase. And if restaurants single-home and the merged firm does not take the radical step of charging consumers to make reservation, consumers would clearly be better off: they would still face a zero price and could access more restaurants on a single platform.

18.5.2.  Advertising-Supported Media Mergers Newspapers, magazines, and radio stations are two-sided platforms. Newspapers and magazines provide news and entertainment to readers and sell space on their pages to advertisers who want to reach those readers. Radio stations also provide news and entertainment, and they sell airtime to advertisers who want to reach their listeners. All these media solve a usage externality by providing a mechanism to bring a possible buyer and seller together. They pay the reader or listener to come by providing content, often at a subscription price below marginal cost for print media and at a price of zero for over-the-air radio, and they give the advertiser access to that reader or listener. They also generate positive externalities for advertisers; advertisers value media with more

27  We want to emphasize that we are not advocating a lighter standard of review for multisided platform mergers or any presumption that such mergers are procompetitive.

The Antitrust Analysis of Multisided Platform Businesses   429

readers or listeners.28 Consumers could like or dislike advertising. The evidence from studies so far indicates that overall advertising does not generate positive or negative externalities for newspaper readers.29 Chandra and Collard-Wexler (2009) conduct a reduced-form examination of the effects of the consolidation of the Canadian newspaper industry in the late 1990s on prices charged to consumers and to advertisers. They find no evidence that mergers led to increases in either price, contrary to expectations derived from one-sided models, and they interpret this as consistent with their theoretical model discussed above. Several authors have evaluated the effect of newspaper and magazine mergers on prices and welfare by developing and estimating structural models of these platforms that account for the possible demand interdependencies between the two sides. Both Affeldt, Filistrucchi, and Klein (2013) and Filistrucchi, Klein, and Michielsen (2012) use detailed econometric models of the Dutch newspaper market, and they study how taking account of two-sidedness affects the evaluation of hypothetical mergers. Affeldt, Filistrucchi, and Klein (2013) find that when two-sidedness is ignored, the estimate of upward pricing pressure on advertising is essentially zero, while they find substantial upward pressure on advertising rates when account is taken of two-sidedness. Filistrucchi, Klein, and Michielsen (2012) find similar results with a SSNIP test and a full simulation of the postmerger equilibrium: taking into account two-sidedness reveals potential unilateral effects on the advertising side that do not show up in analysis that ignores indirect network effects. Fan (2013) develops a structural model of newspapers that considers the value of characteristics to readers and advertisers and the possibility that the owners could change these characteristics following a merger. She estimates the model for newspapers in Minneapolis and finds that a hypothetical merger increases subscription prices, lowers the quality of content from the readers’ perspectives, reduces circulation, and lowers the 28  Economists have recently developed formal models of advertising-supported media that explicitly recognize that they are multisided platforms. Most of the models to date are based on the assumption that consumers rely on a single platform and that the only way for advertisers to reach a consumer is to advertise on the platform that has a monopoly on that person. That assumption departs significantly from reality. Consumers obtain content from multiple advertising platforms. Advertisers consider these alternatives partly as substitutes in deciding how to allocate their advertising budgets; they also consider these alternatives as complements when they want to reinforce messages in different ways through different media. Unfortunately, many key results of most models depend on that assumption. For example, many papers conclude that platforms charge a monopoly price for advertising (since they each have a monopoly over access to some consumers) and that competition among platforms tends to increase advertising prices (because platforms that have less advertising can attract more consumers). See Anderson and Coate (2005); Calvano and Jullien (2012); and Reisinger (2012). Athey, Calvano, and Gans (2013) develop a model that assumes that some people use one platform, while others stochastically search the Web. They show that under this more realistic assumption the results of the earlier models are reversed. 29  See Argentesi and Filistruchhi (2007), Chandra and Collard-Wexler (2009), Fan (2013), and Song (2013). Note that these results average across different types of advertising in newspapers. We suspect that consumers generally value classified listings—a service that consumers often seek out offline and online—while they may not value display advertising mixed with news and entertainment content.

430   David S. Evans and Richard Schmalensee value that advertisers receive. Given that multisided platform businesses typically offer a variety of services to attract economic agents to the platform, analyses such as Fan’s that consider the impact of mergers on nonprice dimensions are likely to prove important in practice. Song (2013) estimates a structural model of German TV magazines that ignores two-sidedness and one that takes it into account. He simulates a merger to monopoly in this market and concludes that this drastic merger would be much less harmful than an analysis that ignores two-sidedness would conclude. For many magazines he estimates that the merged firm would lower the per-copy price, making consumers better off. While advertising rates would generally increase, the greater circulation induced by lower per-copy prices would generally make advertisers better off as well on balance. Finally, Jeziorski (forthcoming) estimates a structural supply-and-demand model using data from the 1996–2006 merger wave in US radio and taking account of indirect network effects. As noted above, he finds that this merger wave harmed advertisers but benefited listeners. Like Fan (2013), Jeziorski (forthcoming) considers an important nonprice dimension of performance. Consistent with other studies, he also finds that these mergers increased market-specific product variety, which contributed importantly to consumer benefits.

18.5.3.  Some Practical Guidance Analysts face a quandary in examining mergers of multisided platforms. The literature discussed above indicates that standard back-of-the envelope calculations may give highly misleading results for the merger of platforms that have significant interdependencies in demand between customer groups. At least at this point the analogous formulas for multisided platforms require more information than an analyst could easily obtain for an initial screening exercise. They essentially require the estimation of a structural econometric model. But if such a model can be estimated, the analyst should just use that model to estimate the unilateral effects of the merger. Of course, in practice, the data necessary for estimating structural models are rarely available, the estimated models may not be robust, and it may take too much time to collect the data and estimate structural models even when this is feasible. Nevertheless, the best has never been the enemy of the good in sound merger analysis, and it need not be just because multisided platforms are involved. The important point is to recognize the economic structure of these platforms, especially the role of competitive constraints and demand-side efficiencies, and factor that into the overall judgment concerning the merger. In some cases, it may be possible to analyze unilateral effects on each side using traditional tools but factoring in biases that have been identified in the literature. What analysts should not do is commit the classic drunk’s mistake—looking under the streetlight for his lost keys just because the light is better there—by conducting a standard one-sided analysis just because it is easier.

The Antitrust Analysis of Multisided Platform Businesses   431

18.6.  Exclusionary Conduct Economists have developed a variety of models that help analyze whether particular business practices are likely to harm consumers as a result of excluding competitors from the market or benefit consumers by reducing prices or increasing quality. These models are typically based on a variety of assumptions that may or may not apply in any specific market. It is well known that many of these models are sensitive to these assumptions. Results change when assumptions change (Evans and Padilla 2005). Vertical restraints, in particular, can be procompetitive under some conditions and anticompetitive under others. Most of the theoretical models on which antitrust analysis relies assume, explicitly or implicitly, that the businesses considered are single-sided. They therefore are not obviously applicable to situations in which firms serve multiple customer groups with interdependent demands. Moreover, given the sensitivity of many of these models to the specific assumptions on which they rest, there is no basis for assuming that the results of models built for competition among single-sided businesses would apply to competition among multisided businesses. The only way to know for sure is to do the math. A relatively small number of authors have extended models originally developed to study business practices by one-sided firms to consider the effects of those practices or similar ones when engaged in by multisided platforms. We provide an overview of some of this work in this section. It is important to note at the outset that the papers we discuss are more likely to be the first words on the issues they address than the last words. Much more work remains to be done. Overall, the results of these multisided models are also sensitive to assumptions, and it remains true as a general matter that vertical restraints may be pro- or anticompetitive.30 The work to date shows, however, that one-sided results generally do not apply to multisided firms. The point is not that multisided firms cannot or do not engage in anticompetitive practices. They could, in fact, engage in the same anticompetitive practices as single-sided firms or in very different anticompetitive practices, not even known or understood yet. Unfortunately, the work economists have done on single-sided firms does not provide much guidance for evaluating the conduct of multisided firms.

18.6.1.  Critical Mass and Platform Viability Multisided platforms may attempt to use exclusionary practices to deter platform entry. To see why, we introduce the concept of critical mass and its role in launching a platform business as developed in Evans (2010) and Evans and Schmalensee (2010). Consider a 30  See Evans (2013b) for a general discussion including the possible procompetitive benefits of vertical restraints for multisided platforms.

432   David S. Evans and Richard Schmalensee new B2B exchange for a particular group of heterogeneous commodities. To provide value to buyers it must have sellers and to provide value to sellers it must have buyers. And it must have enough of both to be viable, since the likelihood of any buyer or seller finding a mutually advantageous trade increases with the number of potential trading partners. This leads to the well-known issue of liquidity in trading environments (Harris 2003; O’Hara 1995; Pagano 1989). A trading venue is only viable if there is a sufficient volume of bids and asks for trading to occur and therefore for both liquidity providers and liquidity takers to incur the expense of coming to the trading platform. If there is too little liquidity, buyers and sellers will not come to the platform. It there is enough liquidity, more buyers and sellers will come and the platform will in fact grow and the platform will be attractive to market specialists and other liquidity providers. But if there are not enough buyers and sellers, the market will not be attractive to liquidity providers. Critical mass refers to the minimal level of demand that platforms must have on their various sides. Platforms that reach that level are viable and positioned to grow more through positive feedback effects. Platforms that fall short of that level are not viable. In some industries, the level of critical mass may be a small portion of the overall market. For example, Discover appears to be a viable US payment card system even though it has only about a 3.2 percent share of transaction volume.31 When platforms are conceived, they, of course, do not have any customers on any side. The challenge for a new platform is to reach critical mass. In some cases, this involves a sequential process of building up demand on one side and then attracting the other side, perhaps à la the “divide and conquer” pricing strategies discussed by Jullien (2011). For example, a media property can invest in building up an audience and then, when it gets enough people, making that audience available to advertisers. In other cases the platform must have both sides on board at the beginning to have any value. That, for example, is clearly the case for matchmaking platforms such as exchanges and dating venues. In this case, start-ups encourage early adopters and other people on both sides that are willing to join the platform in the anticipation that it will achieve critical mass. Strategies that prevent platform entrants from gaining critical mass or that push platforms below critical mass can therefore exclude competitors and preserve the market for the predator. The usual array of exclusionary tactics may be employed, including exclusive dealing, tying and bundling, and predatory pricing. For such practices to be successful, they would need to be effective against the divide and conquer and other entry strategies available to competing platforms. The existence of interdependent demands, however, also provides rationales for why these strategies could increase consumer welfare even though they discourage entry. Putting congestion effects aside, when there are positive feedback effects between the multiple platform sides, increases in the demand on each side increases the value that the platform can deliver overall. Therefore, tactics such as requiring customers on one 31  Nilson Report No. 988, February 2012, 8. Note, however, that the Discover Card is accepted a most merchants that accept cards in the United States. See PaymentsSource (2010).

The Antitrust Analysis of Multisided Platform Businesses   433

side to consolidate their demand on the platform—for example, exclusive dealing—can benefit not only those customers but also the customers on the other side. These tactics can therefore increase consumer surplus.32 Not surprisingly, as the remainder of this section demonstrates, depending on the assumptions made, the theoretical models of exclusionary strategies show that these strategies can increase consumer welfare, decrease consumer welfare, or have an uncertain effect on consumer welfare.

18.6.2.  Exclusive Dealing The Dallas Morning News and the Dallas Times Herald were competing newspapers in Dallas, Texas. They both obtained content such as columns and comic strips from the Universal Press Syndicate. In August 1989 the Morning News signed an exclusive contract with Universal. The Times Herald subsequently lost readership. It filed an antitrust case, and lost.33 In 1991, the parent company of the Morning News bought the Times Herald and shut it down. Chowdhury and Martin (2013) use this example to motivate their analysis of exclusionary contracts that deny platform rivals access to a key complementary input. They show that if consumers do not have strong preferences for one paper over the other and if fixed costs are substantial, social welfare may be higher in the postcontract, monopoly equilibrium, though consumers are always worse off. Similarly, in the Armstrong-Wright (2007) model of competition between platforms viewed as differentiated by one customer group and homogeneous by the other, exclusive dealing can be used to prevent multihoming by the latter group and to exclude competitors, and the monopoly equilibrium may nonetheless be efficient. In the presence of economies of scale, Segal and Whinston (2000) have demonstrated that an incumbent monopoly can profitably deter the entry of a more efficient rival by persuading sufficient customers to sign exclusive dealing contracts before the entrant appears. Doganoglu and Wright (2010) investigate the effectiveness of this strategy when there are no economies of scale but network effects are present. In the two-sided case, they find that it is profitable for the incumbent to exclude a more efficient entrant by offering attractive exclusive dealing contracts to one side of the market before the entrant appears and then charging high prices to those on the other side. As in the single-sided case with scale economies, entry is deterred by making it impossible for the potential entrant to obtain sufficient customers to be viable. With indirect network effects, locking up either side of the market will make it impossible for an entrant to obtain customers on the other side. In an empirical study of the video game industry, Lee (2012) finds that exclusive contracts can facilitate entry rather than deter it. Video game console platforms serve as intermediaries between people who play games and developers that create games. The 32 

Prices could go up as a result of the value being provided by the platform increasing. Consumer welfare would nonetheless increase so long as the value increases by more than the price. 33  Times Herald Printing Co. vs. A.H. Belo Corp, No. A14-90-00856-CV (Texas App.—Houston [14th Dist.] November 27, 1991).

434   David S. Evans and Richard Schmalensee platform consists of hardware and software that developers can use for writing games and that people then use for playing those games. Developers have a greater incentive to write games for consoles that have more users who are likely to license those games and users are more interested in buying consoles that have games they want to play. Video game console makers have a timing issue as discussed by Haigu (2006). It takes time to develop games. Therefore when a new console maker enters (such as Xbox did in 2001) or Sony creates a new version of its console (as it did with the PlayStation 2 and 3 in 2000 and 2006) it has to persuade developers to write games for that console without knowing the demand for it. Video console providers sometimes write their own games, which become exclusive to them, and sometimes enter into contracts with game providers to develop games exclusively for their console. These practices reduce the availability of games to consumers who have purchased other consoles. However, Lee (2012) finds that in the case of sixth-generation video consoles, exclusive dealing facilitated entry. In the absence of exclusivity, game developers would write games first for the incumbent platform because of its larger user base. Only later, if at all, would they port those games to entrants. Thus the entrants would be unable to match the incumbent’s array of games and would find it difficult to differentiate themselves in a positive way. With exclusivity, however, as long as there are uncommitted game developers, entrants can offer games that the incumbent cannot, thus enabling them to differentiate themselves. Lee’s result is not inconsistent with the theoretical analysis of Doganoglu and Wright (2010), since exclusive contracts by the incumbent that prevented entrants from getting access to popular games could exclude entrants and harm consumer welfare. Nor is it inconsistent with the empirical work of Corts and Lederman (2009), who find that the increasing prevalence of nonexclusive video games in recent years has led to a cross-hardware-platform network effect in which game developers react to the installed base of all platforms. This effect, in turn, is consistent with the persistence of multiple competing hardware platforms.34 This literature does not necessarily lead to the conclusion that exclusive dealing is procompetitive for entrants and anticompetitive for incumbents. The entrant could be a significant multisided platform business, as Microsoft was when it used exclusive contracts to enter the browser business in competition with the dominant incumbent Netscape and the video-game console business in competition with the dominant incumbent Sony PlayStation. In evaluating exclusive dealing contracts used by multisided platform entrants and incumbents, analysts would need to consider the context in which those contracts are being entered into—in particular whether they are part of a plausible anticompetitive strategy by either the entrant or the incumbent—and their procompetitive effects in achieving or maintaining critical mass. 34  The analysis of Hagiu and Lee (2011) is also relevant here. They focus on whether content providers (game developers here) sell to one or more platforms or simply “affiliate” (be compatible with) one or more platforms and sell directly to consumers. In their model if content providers are able to extract significant rents from consumers, perhaps because their wares are highly differentiated, they will choose to affiliate and multihome, while if their rent-extraction ability is low, they will sell exclusively to a single platform.

The Antitrust Analysis of Multisided Platform Businesses   435

18.6.3.  Tying and Bundling In a classic paper, Whinston (1990) showed that in the presence of scale economies in the market for good B, a monopoly seller of good A would under some conditions find it profitable to employ tying contracts to become a monopolist in the B market. In his model, tying is not profitable unless it excludes other sellers of B. Whinston showed that whether or not this reduces social welfare depends on the details of the situation. Does this one-sided analysis apply to multisided firms? As usual, adding sides also adds a layer of complexity. For instance, Li (2009) analyzes a version of the Whinston (1990) model in which a two-sided platform facing competition considers whether or not to tie an unrelated good. She finds that if network externalities are strong enough, tying may be profitable (and may also be efficient) even if it does not exclude. Amelio and Julien (2012) consider a two-sided case in which tying is both profitable and welfare-increasing. Suppose the profit-maximizing price on one side of the business is negative, but it is not feasible actually to charge a negative price. (One could argue that OpenTable must have faced this problem, since, as noted above, it effectively charges a negative price to consumers by providing rewards for usage.) By bundling another good or service, however, it is possible to make the effective price on that side negative. They show that this practice increases consumer welfare in the monopoly case but that it may not do so if there is competition. Rochet and Tirole (2008) also model a situation in which tying increases social welfare. They consider two competing platforms:  A, which offers both credit and debit cards, and B, which offers only a debit card. In their simplest model, allowing A to tie its two cards on the merchant side of the business, so they must accept either both cards or neither card (the so-called honor-all-cards rule), increases social welfare. Tying allows A to rebalance its pricing by raising the interchange fee on debit above the competitive level and lowering the interchange fee on credit. While this is always welfare-enhancing in their simplest model, in more complex models the net effect of tying on welfare is ambiguous. Choi (2010) presents a model designed to shed light on the effects of including Microsoft’s Windows Media Player with Windows, which was the subject of an antitrust case in Europe (European Commission 2004; Ahlborn and Evans 2009). This involves tying two platforms together—an operating system (which connects computer users and application providers) and a streaming media software application (which connects computer users and content providers), though Choi does not model the multisided aspects of Windows. In this model, two platforms, A and B, link content providers to consumers, and A also produces a product M, which must be purchased in order to use either A or B. He assumes that content providers multihome. If consumers single-home, tying A to M will exclude B but may increase welfare if network effects are strong (so there is a large efficiency gain from having more customers on both sides of A) and consumers do not consider A and B to be very different (so the reduction in variety from eliminating B is small). If consumers multihome, however, tying A  to M does not exclude B (there are no economies of scale), and social welfare is unambiguously

436   David S. Evans and Richard Schmalensee increased. This analysis makes clear the importance of understanding where multihoming occurs and, if it is not observed, why it does not occur. When products are consumed in fixed proportions, as in the Choi (2010) model, tying is equivalent to pure bundling—selling only bundles of the two products involved. Chao and Derdenger (2013) investigate mixed bundling—selling the products individually and (at a discount) in a bundle. Consider a monopoly video game platform that is considering a mixed bundling strategy: offering a bundle consisting of a console and some games as well as selling the console alone and allowing video game developers to sell games by themselves. Ignoring indirect network effects, one would expect that the optimal mixed bundling strategy would have higher prices for both the console and games when sold separately than would be optimal if the bundle were not offered, since the bundle enables the firm to segment the market according to the number of games buyers prefer to consume. Chao and Derdenger (2013) present a theoretical model in which network effects make it optimal to reduce both console and game prices if a bundle is offered. In that model, mixed bundling acts as a price discrimination device, as in one-sided models, and the presence of the bundle reduces the cost of cutting console and game prices in order to encourage participation by both consumers and developers. Chao and Derdenger (2013) present an empirical analysis of video game pricing that reaches conclusions consistent with their theoretical model.

18.6.4.  Predatory Pricing The general concerns about predatory pricing are as applicable for multisided platforms as they are for single-sided firms. A platform could try to drive rivals out of business by lowering price to one or several of the groups it serves. That could involve increasing subsidies on one side by providing more or better free products and services. After driving a rival out of business the platform, like a single-sided firm, could raise prices and reduce subsidies, thereby recouping its losses. As with the traditional story, one benefit of engaging in this strategy could be developing a reputation for being a ruthless competitor, thereby discouraging entry or competition in multimarket settings (Kreps and Wilson 1982; Milgrom and Roberts 1982). One should recognize, however, that successful predation may be extremely difficult and unlikely, as it is in single-sided markets, given that it would entail certain up-front losses from pricing below cost in pursuit of uncertain future gains from driving and keeping competitors out of the market. The difference is that standard cost-based tests for detecting predatory pricing generally make no economic sense for a multisided business. Those tests are motivated by the standard theoretical result that profit-maximizing prices are never below marginal cost. But for multisided platforms, as a matter of theory, the profit-maximizing price to one or more sides (though not, of course, to all) could be lower than marginal cost—or any other measure of per-unit cost. Indeed, as we have noted, many multisided platforms

The Antitrust Analysis of Multisided Platform Businesses   437

normally charge prices that are less than marginal cost and often zero or even negative on at least one side.35 The flip side of this problem is that a platform could engage in predatory pricing by maintaining a low price on the “subsidy” side (readers for newspapers) and lower the price on the “money” side (advertisers for newspapers) so much that the platform overall loses money. Of course, in theory it would be possible to determine whether the platform has adopted an unprofitable set of prices, but in practice that would be a more complex analysis than comparing prices and some measure of per-unit cost, as is standard for single-sided firms. The important point is that the fact a multisided platform is providing goods or services to one of the groups it serves at prices that do not recover costs provides no meaningful evidence that the platform is engaging in predatory pricing. On the other hand, a change in pricing policy by a multisided platform that results in an overall lowering of prices to loss-inducing levels and that is not explained by other changes in the market could provide an indication of a possible abuse. As for single-sided businesses, one must go beyond analysis of pricing before concluding that predation has actually occurred.36 It is also important to note that there may be procompetitive explanations for a change in pricing policies for multisided platforms that result from interdependent demands and that would not ordinarily be considered for one-sided firms. For example, in examining allegations that The Times of London engaged in predatory pricing Behringer and Filistrucchi (2011) argue that there is evidence that The Times lowered subscription prices sharply because it recognized that advertising prices, and therefore the value of readers, were increasing. It was then profitable to reduce subscription prices to increase readership so as to be more valuable to advertisers. These authors suggest that perhaps The Times recognized this before its rivals and therefore lowered its prices earlier than they did.37 Predatory pricing strategies could also exploit the role of interdependent demands and critical mass for platforms. By charging unprofitably low prices (including providing subsidies in kind) a platform could make it difficult for a rival that cannot match these low prices (or subsidies in kind) to obtain a critical mass of customers. For example, suppose an incumbent platform faces competition from a venture-backed start-up. To prevent the start-up from getting enough critical mass to ignite its platform the incumbent could increase the content offering it is giving customers for free on one side of its platform. Assuming the startup is liquidity constrained, that could prevent the launch of its platform.38 35 

Motta and Vasconcelos (2012) present a model of platform competition in which below-cost pricing never occurs except to deter entry, so that cost-based tests can be employed, but the model seems unlikely to be widely applicable. 36  See the chapter by Elzinga and Mills in volume 2 of this Handbook on predatory pricing. 37  Motta and Vasconcelos (2012) agree that this episode was probably not an instance of predation, but they point to other cases involving platform businesses that they believe did engage in predatory pricing. 38  Of course, it is well known that predatory pricing strategies generally face a number of challenges, and these challenges also apply to multisided firms. Easterbrook (1981); Carlton and Perloff (2005, 352–60).

438   David S. Evans and Richard Schmalensee

18.6.5. Efficiencies When it comes to considering efficiencies, the main novelty for multisided platforms arises from the fact that platforms serve multiple groups of customers with interdependent demands. Business practices can increase the overall value of the platform both to its owners and to society in a variety of ways. By increasing demand on one side, a platform can increase its value to agents on other sides through indirect network externalities. This is a real social benefit, and the platform is unlikely to be able to capture it all. In addition, Evans (2012a) shows how platforms develop governance systems to reduce bad behavior by platform participants that could reduce the platform’s value. The ability to exclude customers is central to the rules platforms use for this purpose. More generally, Boudreau and Hagiu (2009) discuss the use of platform regulations to increase positive externalities and decrease negative ones. In other cases, business process innovations might increase the efficiency of a platform by providing better and cheaper matches between members of the various sides. As a general matter, a platform could increase consumer welfare overall if it increased the value it delivered by more than it increased the prices it charged. It is possible, however, that change in platform prices and products benefit customers on some sides while making customers on other sides worse off. In evaluating such changes, there is no economic reason why one would focus on losses to one group of consumers and ignore gains by another group. Economists are normally interested in overall consumer welfare. A court or competition authority could erroneously conclude, however, that a business practices harms consumers by analyzing competitive effects in a narrow market that excludes one or more sides of a platform.39 That is a fundamental mistake; correct analysis of multisided platforms always considers all groups served by the platform.

18.7.  Coordinated Behavior Two major issues arise in the analysis of coordinated behavior for multisided platforms: First, the welfare analysis of price fixing becomes more complex. When competing single-sided firms collude on prices, economic theory predicts that they will generally

39  Payment card networks have rules that merchants that have agreed to accept a card brand cannot surcharge consumers for using that brand or selectively refuse to accept cards with that brand. Competition authorities and private litigants have challenged these rules under various theories including tying. In these cases, the competition authorities and courts have focused primarily on the merchant side of the payment platforms and ignored the benefits on the consumer side. Our point is not that these practices are necessarily procompetitive but that the correct analysis needs to consider the welfare of customers on both sides of the platforms. See In Re Payment Card Interchange Fee and Merchant-Discount Antitrust Litigation, No. 05-MD-1720-JG-JO (E.D.N.Y. February 20, 2009), First Amended Supplemental Class Action Complaint; U.S. vs. American Express Co., No. 10-CV-04496-NGCLP (E.D.N.Y. October 4, 2010), Complaint.

The Antitrust Analysis of Multisided Platform Businesses   439

raise prices, reduce output, and reduce consumer and total welfare.40 The theory of multisided platforms yields much less clear predictions. It is possible that competing platforms have adopted prices that are too low on one side and too high on another. That is, there is no guarantee that competition will result in platforms adopting the pricing structure that maximizes consumer welfare. It is therefore possible that a coalition of platforms could fix prices—raising them on one side and lowering them on another— that increase consumer welfare. Several papers have reached this conclusion directly or indirectly. The analysis of payment card systems’ “honor all cards” rule by Rochet and Tirole (2008) points in this direction. In their model that rule has the effect of reducing competition between debit cards, permitting a rebalancing of interchange fees that is socially beneficial under some conditions. Chandra and Collard-Wexler (2009), discussed above in the context of mergers, find that a merger of two newspapers or, presumably, an agreement to fix subscriber and advertiser prices could result in an increase of consumer welfare as a matter of theory and find support for this possibility based on Canadian newspaper data. But, as we noted above, the structural model developed by Fan (2013) implies that a merger of two Minneapolis newspapers would make both readers and advertisers worse off. Two recent studies have considered coordinated behavior by newspapers that stops short of fixing all prices. In a model that considers differentiation of newspapers along political lines, Antonielli and Fillistruchi (2012) find that competition on pricing and political positioning creates greater consumer welfare than collusion. But they also find that joint operating agreements, a device used in the United States that allows newspapers to collaborate on pricing and production but compete on the editorial side, result in lower consumer welfare than allowing newspapers to collaborate on all aspects of their business. Gentzkow, Shapiro, and Sinkinson (2012) develop another structural model that allows newspapers to choose their political positions and estimate it using US data from early in the twentieth century, when it was common for newspapers to declare their political affiliation, and many more cities had multiple papers. They find that the main effect of allowing collusion on circulation prices is to reduce consumer and advertiser surplus, while allowing collusion on advertising prices reduces advertiser surplus but increases consumer, newspaper, and total surplus. It also induces substantial new entry and increased ideological diversity. Because advertisers value readers, collusion on advertising rates makes it profitable to lower circulation prices to increase readership, making consumers better off, and then to extract the increased value to advertisers through higher advertising rates. The second issue is that the existence of interlinked prices raises the question of whether collusion is harder or easier for platforms than for single-sided firms. Clearly the need to agree on multiple prices makes platform-wide collusion harder, all else equal. Moreover, even ignoring this issue, Ruhmer (2011) presents a two-platform model with single-homing in which stronger indirect network effects increase the gains 40  The rare case of firms with substantial market power selling complements provides an exception; see Rey and Verge (2008).

440   David S. Evans and Richard Schmalensee from price-cutting and thereby make collusion harder to sustain. This result suggests that in analyzing proposed mergers of multisided firms, enforcement agencies might be less concerned about collusion than in mergers of single-sided firms, but it is not clear that a significant difference can be rationalized unless indirect network effects are very strong indeed. Ruhmer (2011) also addresses the question of whether multisided platforms can increase profits by colluding on only some of the prices they charge. Evans and Schmalensee (2007a) argued that intense competition on one side between two-sided platforms could eliminate the profitability of collusion on the price charged on the other side. In Ruhmer’s (2011) model, where the two platforms are imperfect substitutes for consumers and competition is not particularly intense, it is possible for firms to profitably collude on prices on only one side. The profitability of such incomplete collusion thus depends on the intensity of competition on the noncollusive side.

18.8. Conclusion As economists, we find the rapidly expanding literature on multisided markets interesting and exciting. It is interesting both because of its technical complexity and because of the growing economic importance of multisided platforms. It is exciting because it reveals patterns of behavior and market outcomes that cannot arise with ordinary single-sided firms. As students of antitrust policy, however, we look as this expanding literature with some trepidation. As this chapter has indicated, correct economic analyses of multisided platforms are more complicated than correct analyses of single-sided firms. Moreover, the relevant theory, at least in its current stage of development, yields fewer clear predictions, and there is relatively little empirical work from which one can draw general lessons. Thus it does not now seem possible to come up with many guidelines that can be used to structure rule-of-reason inquiries, let alone sharp lines that would justify per se rules. But the world and the economic literature are what they are, to paraphrase a successful professional (American) football coach. Multisided platforms are more complicated than single-sided firms. Analyses or policy rules that ignore this complexity are prone to commit serious errors. Just because the economic literature on multisided platforms does not have simple extensions of existing single-sided tools does not provide a license to apply the wrong, single-sided tools to multisided platform issues. In the meantime, we can provide some bits of general guidance. Perhaps the most important is to consider all sides served by a platform carefully and to understand the indirect network effects that link them. Understanding that newspapers are two-sided platforms does not make analysis of them simple. But it does serve to structure the analysis in sound and sensible ways. Understanding that OpenTable is a two-sided platform leads the way to understanding how giving away services to consumers could be profit maximizing, not predatory. Recognizing that there are multiple customer groups

The Antitrust Analysis of Multisided Platform Businesses   441

with interlinked demand may help competition policy analysis identify anticompetitive strategies and identify possible efficiencies that would not be apparent from applying a traditional analysis. Finally, recognizing these multiple customer groups is critical for ensuring that antitrust enforcement does not have the unintended consequence of reducing consumer welfare by causing more harm on one or more sides of a platform than it provides benefit on another side.

Acknowledgments We would like to thank Howard Chang and Steven Joyce for valuable assistance and helpful comments. We are indebted to Marshall Van Alstyne, Mark Armstrong, Yong Chao, Glen Weyl, and the editors for useful comments on an earlier draft.

References Affeldt, Pauline, Lapo Filistrucchi, and Tobias J.  Klein. 2013. Upward Pricing Pressure in Two-Sided Markets. Economic Journal 123(572): F505–23. Ahlborn, Christian, and David S. Evans. 2009. The Microsoft Judgment and Its Implications for Competition Policy towards Dominant Firms in Europe. Antitrust Law Journal 75(3): 887–932. Ambrus, Atilla, and Rossella Argenziano. 2009. Asymmetric Networks in Two-Sided Markets. American Economic Journal: Microeconomics 1(1): 17–52. Amelio, Andrea, and Bruno Jullien. 2012. Tying and Freebies in Two-Sided Markets. International Journal of Industrial Organization 30(5): 436–46. Anderson, Simon P., and Stephen Coate. 2005. Market Provision of Broadcasting: A Welfare Analysis. Review of Economic Studies 72(4): 947–72. Anderson, Simon P., and Jean J.  Gabszewicz. 2006. The Media and Advertising:  A  Tale of Two-Sided Markets. In Victor Ginsburgh and David Thosby, Handbook of the Economics of Arts and Culture, Vol. 1: 567–614. Amsterdam: Elsevier. Antonielli, Marco, and Lapo Filistrucchi. 2012. Collusion and the Political Differentiation of Newspapers. TILEC Discussion Paper 2012-014. Areeda, Phillip E., and Herbert Hovenkamp. 2002. Antitrust Law. Vol. 2A. New York: Aspen Law and Business. Argentesi, Elena, and Lapo Filistrucchi. 2007. Estimating Market Power in a Two-Sided Market: The Case of Newspapers. Journal of Applied Econometrics 22(7): 1247–66. Argentesi, Elena, and Marc Ivaldi. 2007. Market Definition in Printed Media Industries: Theory, Practice, and Lessons for Broadcasting. In Paul Seabright and Jurgen von Hagen, The Economic Regulation of Broadcasting Markets, 225–54. Cambridge: Cambridge University Press. Armstrong, Mark. 2002. The Theory of Access Pricing and Interconnection. In Martin E. Cave, Sumit K. Majumdar, and Ingo Vogelsang, Handbook of Telecommunications Economics, vol. 1: 297–386. Amsterdam: North-Holland. Armstrong, Mark. 2006. Competition in Two-Sided Markets. Rand Journal of Economics 37(3): 668–91.

442   David S. Evans and Richard Schmalensee Armstrong, Mark, and Julian Wright. 2007. Two-Sided Markets, Competitive Bottlenecks, and Exclusive Contracts. Journal of Economic Theory 32(2): 353–80. Arthur, W. Brian. 1989. Competing Technologies, Increasing Returns, and Lock-In by Historical Events. Economic Journal 99(394): 116–131. Athey, Susan, Emilio Calvano, and Joshua S. Gans. 2013. The Impact of the Internet on Advertising Markets for News Media. Working paper. Bagwell, Kyle. 2007. The Economic Analysis of Advertising. In Mark Armstrong and Robert Porter, Handbook of Industrial Organization, vol. 3: 1701–844. Amsterdam: Elsevier. Baxter, William. 1983. Bank Interchange of Transactional Paper: Legal and Economic Perspectives. Journal of Law and Economics 26(3): 541–88. Bedre-Defolie, Ozlem, and Emilio Calvano. 2013. Pricing Payment Cards. American Economic Journal: Microeconomics 5(3): 206–31. Behringer, Stefan, and Lapo Filistrucchi. 2011. Predatory Pricing in Two-Sided Markets: Lessons from the UK Quality Newspapers in the ’90s. Working paper, Unversität Bonn and Tilburg University. Boudreau, Kevin J., and Andrei Hagiu. 2009. Platform Rules:  Multi-sided Platforms as Regulators. In Annabelle Gawer, Platforms, Markets, and Innovation, 163–91. Northampton, MA: Edward Elgar. Caillaud, Bernard, and Bruno Jullien. 2003. Chicken and Egg: Competition among Intermediation Service Providers. Rand Journal of Economics 34(2): 309–28. Calvano, Emilio. 2011. Note on the Economic Theory of Interchange. Comment to the Federal Reserve Board on Proposed Regulation II (Debit Card Interchange Fees and Routing). Available at http://www.federalreserve.gov/SECRS/2011/March/20110328/R-140​ 4/R-1404_030811_69122_621890579792_1.pdf. Calvano, Emilio, and Bruno Jullien. 2012. Issues in Online Advertising and Competition Policy: A Two-Sided Market Perspective. In Joseph E. Harrington and Yannis Katsoulacos, Recent Advances in the Analysis of Competition Policy and Regulation, 179–97. Cheltenham: Edward Elgar. Carlton, Dennis W., and Jeffrey M.  Perloff. 2005. Modern Industrial Organization. 4th ed. Boston: Addison Wesley Longman. Chamberlin, Edward. 1933. Theory of Monopolistic Competition. Cambridge, MA:  Harvard University Press. Chandra, Ambarish, and Allan Collard-Wexler. 2009. Mergers in Two-Sided Markets:  An Application to the Canadian Newspaper Industry. Journal of Economics and Management Strategy 18(4): 1045–70. Chang, Howard H., David S.  Evans, and Richard Schmalensee. 2011a. Market Definition: Assessment of the Relevant Market in Competition Matters. Report Prepared for the Federal Competition Commission of Mexico. Chang, Howard H., David S. Evans, and Richard Schmalensee. 2011b. Market Power: Assessment of Market Power in Competition Matters. Report Prepared for the Federal Competition Commission of Mexico. Chao, Yong, and Timothy Derdenger. 2013. Mixed Bundling in Two-Sided Markets in the Presence of Installed Base Effects. Management Science 59(8): 1904–26. Choi, Jay Pil. 2010. Tying in Two-Sided Markets with Multi-homing. Journal of Industrial Economics 58(3): 607–26. Chowdhury, Subhasish M., and Stephen Martin. 2013. Exclusivity and Exclusion on Platform Markets. Working paper.

The Antitrust Analysis of Multisided Platform Businesses   443

Coase, Ronald H. 1960. The Problem of Social Cost. Journal of Law and Economics 3(1): 1–44. Corts, Kenneth S., and Mara Lederman. 2009. Software Exclusivity and the Scope of Indirect Network Effects in the U.S. Home Video Game Market. International Journal of Industrial Organization 27(2): 121–36. Doganoglu, Toker, and Julian Wright. 2010. Exclusive Dealing with Network Effects. International Journal of Industrial Organization 28(2): 145–54. Easterbrook, Frank H. 1981. Predatory Strategies and Counterstrategies. University of Chicago Law Review 48(2): 263–337. Eisenmann, Thomas R., Geoffrey Parker, and Marshall Van Alstyne. 2011. Platform Envelopment. Strategic Management Journal 32(12): 1270–85. Elhauge, Einer, and Damien Geradin. 2007. Global Competition Law and Economics. Portland, OR: Hart Publishing. Emch, Eric, and T. Scott Thompson. 2006. Market Definition and Market Power in Payment Card Networks. Review of Network Economics 5(1): 15–25. European Commission. 1997. Notice on the Definition of the Relevant Market for the Purposes of Community Competition Law. Official Journal, C 372, 09/12/1997 P. 0005-0013. European Commission. 2004. Chicken and Egg:  Competition among Intermediation Service Providers, Decision of 24 May 2004 Relating to a Proceeding Pursuant to Article 82 of the EC Treaty and Article 54 of the EEA Agreement Against Microsoft Corporation (Case COM/C-3/37.792—Microsoft), Official Journal L 032, 06/02.2007 P. 0023-0028. Evans, David S. 2003a. Some Empirical Aspects of Multi-Sided Platforms. Review of Network Economics 2(3): 1–19. Evans, David S. 2003b. The Antitrust Economics of Multi-sided Platform Markets. Yale Journal on Regulation 20(2): 325–81. Evans, David S. 2008. Antitrust Issues Raised by the Emerging Global Internet Economy. Northwestern University Law Review Colloquy 102: 285–306. Evans, David S. 2010. The Web Economy, Two-Sided Markets, and Competition Policy. Working paper, University of Chicago. Evans, David S. 2012a. Governing Bad Behavior by Users of Multi-sided Platforms. Berkeley Technology Law Journal 27(2): 1201–50. Evans, David S. 2012b. Lightening Up on Market Definition. In Einer Elhauge, Research Handbook on the Economics of Antitrust Law, 53–89. Cheltenham: Edward Elgar. Evans, David S. 2012c. Two-Sided Market Definition. ABA Section of Antitrust Law, Market Definition in Antitrust: Theory and Case Studies. Evans, David S. 2013a. Attention Rivalry among Online Platforms and Its Implications for Antitrust Analysis. Journal of Competition Law and Economics 9(2) 313–57. Evans, David S. 2013b. Economics of Vertical Restraints for Multi-sided Platforms. Competition Policy International 9(1): 1–24. Evans, David S., Franklin M. Fisher, Daniel L. Rubinfeld, and Richard L. Schmalensee. 2000. Did Microsoft Harm Consumers? Two Opposing Views. AEI-Brookings Joint Center for Regulatory Studies. Evans, David S., Andrei Hagiu, and Richard Schmalensee. 2006. Invisible Engines: How Software Platforms Drive Innovation and Transform Industries. Cambridge, MA: MIT Press. Evans, David S., and Michael Noel. 2005. Defining Antitrust Markets When Firms Operate Two-Sided Platforms. Columbia Business Law Review 2005(3): 101–34. Evans, David A., and Michael Noel. 2008. The Analysis of Mergers That Involve Multisided Platform Businesses. Journal of Competition Law and Economics 4(3): 663–95.

444   David S. Evans and Richard Schmalensee Evans, David S., and A. Jorge Padilla. 2005. Designing Rules for Assessing Unilateral Practices: A Neo-Chicago Approach. University of Chicago Law Review 72(1): 73–98. Evans, David S., and Richard Schmalensee. 2005. Paying with Plastic: The Digital Revolution in Buying and Borrowing. 2nd ed. Cambridge, MA: MIT Press. Evans, David S., and Richard Schmalensee. 2007a. The Industrial Organization of Markets with Two-Sided Platforms. Competition Policy International 3(1): 151–79. Evans, David S., and Richard Schmalensee. 2007b. Catalyst Code:  The Strategies behind the World’s Most Dynamic Companies. Cambridge, MA: Harvard Business School Press. Evans, David S., and Richard Schmalensee. 2010. Failure to Launch: Critical Mass in Platform Businesses. Review of Network Economics 9(4): 1–26. Fan, Ying. 2013. Ownership Consolidation and Product Characteristics: A Study of the U.S. Daily Newspaper Market. American Economic Review 103(5): 1598–628. Farrell, Joseph, and Paul Klemperer. 2006. Co-ordination and Lock-in:  Competition with Switching Costs and Network Effects. In M. Armstrong and Robert S. Porter, Handbook of Industrial Organization, vol. 3: 1967–2072. Amsterdam: North-Holland. Farrell, Joseph, and Garth Saloner. 1985. Standardization, Compatibility, and Innovation. Rand Journal of Economics 16(1): 70–83. Farrell, Joseph, and Carl Shapiro. 2010. Antitrust Evaluation of Horizontal Mergers:  An Economic Alternative to Market Definition. B.E. Journal of Theoretical Economics, Policies and Perspectives 10(1): Article 9. Federal Reserve Bank of Boston. 2009. Survey of Consumer Payment Choice. Available at http://​ www.bos.frb.org/economic/cprc/scpc/index.htm. Federal Reserve Board of Governors. 2010. Survey of Consumer Finances. Available at http://​ www.federalreserve.gov/econresdata/scf/scfindex.htm. Feenstra, D. W., and H. Wang. 2000. Economic and Accounting Rates of Return. Research Report 00E42, University of Groningen, Research Institute SOM. Filistrucchi, Lapo, Tobias J. Klein, and Thomas Michielsen. 2012. Assessing Unilateral Effects in a Two-Sided Market: An Application to the Dutch Daily Newspaper Market. Journal of Competition Law and Economics 8(2): 297–329. Fisher, Franklin M. 1987. On the Misuse of the Profit-Sales Ratio to Infer Monopoly Power. Rand Journal of Economics 18(3): 384–96. Fisher, Franklin M., and John J. McGowan. 1983. On the Misuse of Accounting Rates of Return to Infer Monopoly Profits. American Economic Review 73(1): 82–97. Forden, Sara, and Jeff Bliss. 2012. FTC Draft Memo Recommends an Antitrust Suit against Google. Washington Post, October 14. Galeotti, Andrea, and Jose Luis Moraga-Gonzalez. 2009. Platform Intermediation in a Market for Differentiated Products. European Economic Review 53(4): 417–28. Gentzkow, Matthew, Jesse M. Shapiro, and Michael Sinkinson. 2012. Competition and Ideological Diversity: Historical Evidence from US Newspapers. Working paper, University of Chicago and Wharton School. Hagiu, Andrei. 2009. Two-Sided Platforms: Product Variety and Pricing Structures. Journal of Economics and Management Strategy 18(4): 1011–43. Hagiu, Andrei, and Robin S.  Lee. 2011. Exclusivity and Control. Journal of Economics and Management Strategy 20(3): 679–708. Halaburda, Hanna, and Mikolaj Jan Piskorski. 2013. Competing by Restricting Choice: The Case of Search Platforms. Harvard Business School Working Paper No. 10-098.

The Antitrust Analysis of Multisided Platform Businesses   445

Harris, Larry. 2003. Trading and Exchanges:  Market Microstructure for Practitioners. Oxford: Oxford University Press. Hesse, Renata B., and Joshua H. Soven. 2006. Defining Relevant Product Markets in Electronic Payment Network Antitrust Cases. Antitrust Law Journal 73(3): 709–38. In Re Payment Card Interchange Fee and Merchant-Discount Antitrust Litigation, No. 05-MD-1720-JG-JO (E.D.N.Y. February 20, 2009. First Amended Supplemental Class Action Complaint. Jeziorski, Przemyslaw. Forthcoming. Effects of Mergers in Two-Sided Markets: Examination of U.S. Radio Industry. American Economic Journal: Microeconomics. Jullien, Bruno. 2011. Competition in Multi-sided Networks:  Divide-and-Conquer. American Economic Journal: Microeconomics 3(4): 1–35. Katz, Michael L., and Carl Shapiro. 1985. Network Externalities, Competition, and Compatibility. American Economic Review 75(3): 424–40. Katz, Michael L., and Carl Shapiro. 1994. Systems Competition and Network Effects. Journal of Economic Perspectives 8(2): 93–115. Kendall, Brent, Vanessa Mock, and Thomas Catan. 2012. U.S., Europe to Huddle on Google Probes. Wall Street Journal Online, November 29. Keneally, Meghan. 2011. Restaurants Question OpenTable Fees. Pittsburgh Post-Gazette, May 1. Kreps, David M., and Robert Wilson. 1982. Reputation and Imperfect Information. Journal of Economic Theory 27(2): 253–79. Lee, Robin S. 2013. Vertical Integration and Exclusivity in Platform and Two-Sided Markets. American Economic Review 103(7): 2960–3000. Li, Ting. 2009. Tying in Two-Sided Markets. Working paper, Toulouse School of Economics. Milgrom, Paul, and John Roberts. 1982. Predation, Reputation, and Entry Deterrence. Journal of Economic Theory 27(2): 280–312. Motta, Massimo. 2004. Competition Policy:  Theory and Practice. Cambridge:  Cambridge University Press. Motta, Massimo, and Helder Vasconcelos. 2012. Exclusionary Pricing in a Two-Sided Market. CEPR Discussion Paper No. 9164. O’Hara, Maureen. 1995. Market Microstructure Theory. Cambridge, MA: Blackwell. OECD Competition Committee. 2009. Two-Sided Markets, DAF/COMP(2009)20. Available at http://www.oecd.org/daf/competition/44445730.pdf. OpenTable Inc. 2011. Form 10-K for the Period Ending 12/31/11. Available at http://files.share​ holder.com/downloads/ABEA-2TKK09/2145589686x0xS1047469-12-1451/1125914/filing.​pdf. Pagano, Marco. 1989. Trading Volume and Asset Liquidity. Quarterly Journal of Economics 104(2): 255–74. Parker, Geoffrey G., and Marshall W. Van Alstyne. 2000. Internetwork Externalities and Free Information Goods. In Proceedings of the 2nd ACM Conference on Electronic Commerce. New York: Association for Computing Machinery, 107–16. Pastore Mark. 2010. Is OpenTable Worth It? Incanto Letters, October 22. Available at http://​ incanto.biz/2010/10/22/is-opentable-worth-it/. PaymentsSource. 2010. Discover Striving to Raise U.S. Merchants’ Awareness of Card-Acceptance Abilities. April 21. http://www.paymentssource.com/news/-3001446-1.html. Prager, Robin, Mark Manuszak, Elizabeth Kiser, and Ron Borzekowski. 2009. Interchange Fees and Payment Card Networks: Economics, Industry Developments, and Policy Issues. Federal Reserve Board of Finance and Economics Discussion Series, 2009-23.

446   David S. Evans and Richard Schmalensee Qihoo v. Tencent, Civil Judgment No. Yuegaofaminsanchuzi 2/2011 (The High People’s Court of Guangdong Province, P.R. China March 28, 2013). Available at http://www.gdcourts.gov.cn/​ gdcourt/front/front!content.action?lmdm=LM43&gjid=20130328040159946185 (Chinese). Reisinger, Markus. 2012. Platform Competition for Advertisers and Users in Media Markets. International Journal of Industrial Organization 30(2): 243–52. Rey, Patrick, and Thibaud Verge. 2008. The Economics of Vertical Restraints. In Paolo Buccirossi, Handbook of Antitrust Economics, 353–90. Cambridge, MA: MIT Press. Rochet, Jean-Charles, and Jean Tirole. 2003. Platform Competition in Two-Sided Markets. Journal of the European Economic Association 1(4): 990–1029. Rochet, Jean-Charles, and Jean Tirole. 2006. Two-Sided Markets:  A  Progress Report. Rand Journal of Economics 37(3): 645–67. Rochet, Jean-Charles, and Jean Tirole. 2008. Tying in Two-Sided Markets and the Honor All Cards Rule. International Journal of Industrial Organization 26(6): 1333–47. Rochet, Jean-Charles, and Julian Wright. 2010. Credit Card Interchange Fees. Journal of Banking and Finance 24(8): 1788–97. Ruhmer, Isasbel. 2011. Platform Collusion in Two-Sided Markets. Working paper, University of Mannheim. Rysman, Marc. 2004. Competition between Networks: A Study of the Market for Yellow Pages. Review of Economic Studies 71(2): 483–512. Rysman, Marc. 2009. The Economics of Two-Sided Markets. Journal of Economic Perspectives 23(3): 125–43. Schmalensee, Richard. 2002. Payment Systems and Interchange Fees. Journal of Industrial Economics 50(2):103–22. Schmalensee, Richard. 2011. Why Is Platform Pricing Generally Highly Skewed? Review of Network Economic, 10(4): 1–11. Segal, Ilya R., and Michael D. Whinston. 2000. Exclusive Contracts and Protection of Investments. Rand Journal of Economics 31(4): 603–33. Smith, Adam. 1776. An Inquiry into the Nature and Causes of the Wealth of Nations. London: T. Nelson and Sons, Project Gutenberg. Song, M. 2013. Estimating Platform Market Power in Two-Sided Markets with an Application to Magazine Advertising. Working paper. Spence, A.  Michael. 1975. Monopoly, Quality, and Regulation. Bell Journal of Economics 6(2): 417–29. Spulber, Daniel F. 1996. Market Microstructure and Intermediation. Journal of Economic Perspectives 10(3): 135–52. Strahilevitz, Lior Jacob. 2006. Information Asymmetries and the Rights to Exclude. Michigan Law Review 104(8):1835–98. Stross, Randall. 2010. The Online Reservations That Restaurants Love to Hate. New York Times, December 12. Times Herald Printing Co. v. A.H. Belo Corp, No. A14-90-00856-CV (Texas App.–Houston [14th Dist.] November 27, 1991). Tirole, Jean. 1988. The Theory of Industrial Organization. Cambridge, MA: MIT Press. United States Government, in OECD Competition Committee. 2009. Two-Sided Markets, DAF/COMP(2009)20. Available at http://www.oecd.org/daf/competition/44445730.pdf. United Brands Company and United Brands Continetaal BV v. European Commission, 1978 E.C.R. (1978).

The Antitrust Analysis of Multisided Platform Businesses   447

United States v. First Data Corporation, No. 03-2169 (D.D.C. December 10, 2003), Plaintiff ’s Pre-Trial Brief. United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001). U.S. v. First Data Corp., No. 03-CV-02169 (D.D.C. May 25, 2004), Final Judgment. U.S.  vs. American Express Co., No. 10-CV-04496-NG-CLP (E.D.N.Y. October 4, 2010), Complaint. US Department of Justice and Federal Trade Commission. 2010. Horizontal Merger Guidelines. UK Office of Fair Trading. 2004. Market Definition: Understanding Competition Law. Wang, Zhu. 2010. Market Structure and Payment Card Pricing: What Drives the Interchange? International Journal of Industrial Organization 28(1): 86–98. Weyl, E. Glenn. 2010. A Price Theory of Multi-sided Platforms. American Economic Review 100(4):1642–72. White, Alexander, and E. Glen Weyl. 2012. Insulated Platform Competition. Working paper, Tsinghua University and University of Chicago. Whinston, Michael D. 1990. Tying, Foreclosure, and Exclusion. American Economic Review 80(4): 837–59. Wright, Julian. 2004. One-Sided Logic in Two-Sided Markets. Review of Network Economics 3(1): 44–64. Zhou, Yiyi. 2014. Failure to Launch in Two-Sided Markets:  A  Study of the U.S. Video Game Market. Working paper, State University of New York at Stony Brook.

PA R T I I I

M E RG E R S

CHAPTER 19

E F F IC I E N C Y C L A I M S A N D A N T I T RU S T E N F OR C E M E N T HOWARD SHELANSKI

Should it matter to a court or antitrust enforcement agency when potentially anticompetitive actions by firms increase economic efficiency? The answer today may seem obvious, if qualified: yes, the right kind of economic efficiency gains pursuant to the right kind of conduct or transaction should factor centrally into antitrust analysis and enforcement. That answer, however, has not always been so obvious, and the qualifications matter even today, especially in merger review. This chapter will examine the theory, evolution, and current state of the law regarding efficiencies in US antitrust enforcement. It will begin with a brief discussion of the economics underlying efficiency claims and how they relate to the welfare standard of US antitrust law. It will then examine the history of efficiency claims before US courts and enforcement agencies and trace the evolution of the law to its current state. This chapter will then turn to how antitrust enforcement agencies in the United States currently incorporate and assess efficiency claims, focusing on merger investigation but with some discussion of conduct cases as well. I conclude with some open questions and suggestions for further research.

19.1.  Why Efficiency Matters for Antitrust The objective of antitrust law is to prevent harmful economic conduct, not to get in the way of actions that reduce costs, increase output, or speed innovation, all of which one might lump together as “efficiencies.” Absent compelling circumstances, it makes little sense for the law to stand in the way of business strategies that increase economic surplus for society. From an economic standpoint, there is general agreement that efficiency

452   Howard Shelanski considerations should matter when courts or agencies decide whether to block a particular merger or deem particular conduct to be illegal. That broad consensus frays when it comes to specific cases, however; and it does so for three reasons. First, antitrust law and policy in the United States is focused on consumer welfare rather than total welfare, which means that antitrust laws allow conduct that expands the economic pie so long as that expansion does not shift net welfare from consumers to producers. In other words, the firm’s actions must leave consumers no worse off even if they make the firm much better off. Second, which transactions or other business activities will likely lead to efficiency gains is the subject of debate among economists, particularly for mergers and in the areas of unilateral, vertical conduct that enforcement agencies might investigate under Section 2 of the Sherman Act. And third, even where a merger or course of conduct may bring efficiency gains, it may at the same time carry costs for consumers or competition, and there is not a clear consensus on how to balance those costs and benefits in assessing the overall effect of the merger or conduct. Therefore, while the decision of whether to count efficiencies is easy at a general policy level, it can be much less so in specific cases. If one takes a short-run view of supply and demand—a “static” analysis of a given market—then efficiency considerations become less important in judging the effects of mergers or conduct. If producers’ costs and consumers’ demand are fixed, then the effect of business decisions will mostly be to shift the comparative shares of producer and consumer surplus, and the total amount of economic welfare, in relation to some fixed benchmark such as the competitive equilibrium. It is relatively easy in such cases to assess conduct based solely on how it allocates resources given fixed demand and supply conditions. Efficiencies complicate the picture because they shift costs downward and/ or shift demand outward, making it possible for conduct simultaneously to shift surplus to producers but also to increase total welfare and make consumers better off. Once antitrust takes into account the possibility of such efficiency effects, the ability to judge conduct in a purely static way diminishes, and indeed doing so could positively harm economic welfare. This insight was developed most forcefully by Oliver Williamson, who argued that efficiencies (which he called “economies”) should be a valid antitrust defense and change the way agencies and courts analyze conduct and transactions (Williamson 1969). Efficiencies can come into play from the effects of the three main kinds of conduct that the US antitrust laws address: mergers, single-firm conduct, and, though to a somewhat lesser extent, cooperation among competitors. As we will see below, merger policy has come increasingly to recognize that some transactions have a genuine prospect of reducing the combined firm’s marginal costs of production compared to the respective costs borne by the separate firms premerger. Unilateral conduct, particularly that which governs vertical contracting between manufacturers and distributors or retailers, has the possibility of expanding output by overcoming various inefficiencies of simple, linear (i.e., price per unit) contracts, thereby potentially supplying a proconsumer business rationale for conduct that might at first glance appear monopolistic. Finally, cooperation among competitors, conduct that can raise the specter of per se liability for price

Efficiency Claims and Antitrust Enforcement   453

fixing or its equivalent, may receive and pass rule-of-reason scrutiny when the cooperation is beneficial for competition in existing product markets (e.g., Northwest Wholesale Stationers)1 or the introduction of new products (e.g. BMI).2 Taking into account efficiencies such as the above makes antitrust analysis more complicated than it would be if the analysis were more static, were more categorical in linking conduct to liability, or were more focused on competitors than the competitive process. But the incorporation of economic efficiencies into antitrust also makes competition policy better by giving courts and agencies a more complete view of how different kinds of conduct affect competition and output, thereby making antitrust enforcement even more likely to improve consumer welfare. The last proposition, widely agreed upon today even if still subject to some debate, has not always been accepted in US antitrust law. The next section will briefly discuss the history of efficiency analysis in merger review and other areas of antitrust enforcement, and show the evolution to the current state of the law.

19.2.  Efficiency: From Inconvenient Fact to Acceptable Defense For a long time, neither courts nor agencies would consider efficiencies when there was either evidence or a legal presumption of harm to competition. Moreover, harm to competition was defined very loosely, often equated with mergers that merely increased concentration (for example Brown Shoe and Vons)3 or conduct that made life tough for competitors (for example Alcoa).4 There were strong per se rules against certain vertical restraints (for example resale price maintenance after Dr. Miles)5 even as some plausible benefits of such conduct were recognized (as with nonprice vertical restraints). In the area of horizontal restraints, the Supreme Court somewhat ironically proved receptive to some procompetitive justifications early on (for example Chicago Board of Trade and, perhaps, Appalachian Coals)6 but proved at times rigidly doctrinal in the face of evident efficiency benefits from cooperation (for example Topco).7 The evolution in antitrust to a more receptive posture toward efficiencies can be seen most clearly in the areas of merger enforcement and vertical restraints. A good example of the early view of efficiencies in merger review is Brown Shoe, in which the Supreme Court stated that the principle underlying passage of Section 7 of the 1  2  3  4  5  6  7 

472 U.S. 284 (1985). 441 U.S. 1 (1979). 370 U.S. 294 (1962); 384 U.S. 270 (1966). 148 F.2d 416 (2d Cir. 1945). 220 U.S. 373 (1911). 246 U.S. 231 (1918); 288 U.S. 344 (1933). 405 U.S. 596 (1972).

454   Howard Shelanski Clayton Act was the prevention of market concentration.8 Under this view, more forcefully articulated in Philadelphia National Bank, a firm’s market share itself could be so likely to lessen competition that no beneficial effects could offset the risk.9 At around the same time, the FTC took an even dimmer view of efficiencies, characterizing them as not only inadequate to offset competitive harms but as the very source of those harms: in in re Foremost Dairies, the Commission found that when a merger gives an organization a “decisive advantage in efficiency over its smaller rivals” it violates the Clayton Act.10 Efficiencies began to enter merger review when the DOJ issued its first Merger Guidelines in 1968.11 Under those initial Guidelines, merger efficiencies could justify a suspect transaction only in exceptional circumstances, because of both concerns that measuring potential efficiencies is quite difficult and concerns that efficiencies might be realized by more procompetitive means.12 In the subsequent 1982 Guidelines, consideration of merger efficiencies still seemed confined to limited circumstances.13 The 1984 revision, however, “clarified” that, despite their restrictive language regarding merger efficiencies, the 1982 Guidelines actually worked a substantial change in their consideration. This was partly because the 1982 Guidelines articulated a dramatically different “unifying theme”: whereas the 1968 Guidelines stated that the primary role of Section 7 enforcement was to preserve and promote market structures conducive to competition, the 1982 Guidelines (§ 1) stated that mergers should not be permitted to create or 8 

See Brown Shoe Co. v. United States, 370 U.S. 294, 315 (1962). See United States v. Philadelphia Nat’l Bank, 374 U.S. 321, 371 (1963) (“Congress determined to preserve our traditionally competitive economy. It therefore proscribed anticompetitive mergers, the benign and malignant alike, fully aware, we must assume, that some price might have to be paid”); see also FTC v. Procter & Gamble Co., 386 U.S. 568, 580 (1967) (“Congress was aware that some mergers which lessen competition may also result in economies but it struck the balance in favor of protecting competition”); see Fisher and Lande (1983, 1580–83) (“[T]‌he courts and merger enforcement agencies have focused almost exclusively on the prevention of increases in market power. Since only a relatively small probability of increased market power . . . has been sufficient for a finding of illegality and the courts have given no weight to possible countervailing benefits, defendant merger victories not surprisingly have been relatively rare”). Some have argued that these precedents only repudiated the use of speculative or uncertain evidence to demonstrate efficiencies (Areeda and Turner 1980). But see Fisher and Lande (1983, 1595) (“This narrow reading of the language ignores the Court’s stated rationale for rejecting an efficiencies defense. The Court based its holding on the balance that Congress struck ‘in favor of protecting competition, believing that Congress valued competition more highly than efficiency gains. . . . To implement the spirit of the Clayton Act more fully, the court chose . . . [to ignore] efficiencies altogether.”). 10  60 FTC 944, 1084. 11  See U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines § 10 (1968) (hereinafter 1968 Guidelines), reprinted in 4 Trade Reg. Rep. (CCH) ¶ 13,101. 12 See id. (“Unless there are exceptional circumstances, the Department will not accept as a justification for an acquisition normally subject to challenge under its horizontal merger standards, the claim that the merger will produce [improvements in efficiency] because . . . there usually are severe difficulties in accurately establishing [their] existence and magnitude”). 13  See U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines § 5A (1982) (hereinafter 1982 Guidelines) (“Except in extraordinary cases, the Department will not consider a claim of specific efficiencies as a mitigating factor for a merger that would otherwise be challenged”), reprinted in 4 Trade Reg. Rep. (CCH) ¶ 13,102. 9 

Efficiency Claims and Antitrust Enforcement   455

enhance market power or facilitate its exercise. This view—that market structure is not synonymous with competitiveness—was not compatible with a severely curtailed efficiencies defense. The 1984 language therefore suggested that the agencies would generally take merger efficiencies into account when assessing a transaction.14 This revision, however, did not relax all of the conditions placed on merger efficiency analysis. As did earlier Guidelines, the 1984 revisions required efficiencies to be cognizable and well supported by evidence.15 Additionally, efficiencies could not be reasonably achievable by nonmerger means such as internal expansion. And finally, adopting a sliding-scale approach, the 1984 Guidelines required a showing of greater efficiencies the more significant the competitive risks involved in the proposed merger. The agencies left the efficiencies section unchanged when they revised the Guidelines in 1992, but they significantly altered the treatment of efficiencies through a 1997 amendment. Among the 1997 amendments to the Guidelines was a change in the burden of verification— a shift from solely internal analysis of alleged savings to a requirement that the merging firms substantiate reasonably verifiable efficiency claims.16 Additionally, the Guidelines moved from a policy where efficiencies would not be considered if they were reasonably achievable by other means to a standard under which efficiencies would only go unconsidered if they were realistically—rather than theoretically—achievable by some alternative means applicable to the business situation faced by the merging parties. A third significant change amplified the definition of cognizable efficiencies to exclude efficiencies that arise from anticompetitive reductions in output or service, as these efficiencies do not enhance competition or benefit consumers.17 Further, and perhaps most significantly, the 1997 revisions explicitly combined efficiency and competitive effects analysis18 and stated that the agencies might consider out-of-market efficiencies.19 Finally the revision confirmed that 14 

See U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines § 3.5 (1984) (hereinafter 1984 Guidelines) (“In practice, the Department never ignores efficiency claims”), reprinted in 4 Trade Reg. Rep. (CCH) ¶ 13,103. 15  Id. (“Cognizable efficiencies include, but are not limited to, achieving economies of scale, better integration of production facilities, plant specialization, lower transportation costs, and similar efficiencies relating to specific manufacturing, servicing, or distribution operations of the merging firms. The agency may also consider claimed efficiencies resulting from reductions in general selling, administrative, and overhead expenses, or that otherwise do not relate to specific manufacturing, servicing, or distribution operations of the merging firms”). 16  See U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines § 4 (1997) (hereinafter 1997 Guidelines), reprinted in 4 Trade Reg. Rep. (CCH) ¶ 13,104. See also Economic Issues: Merger Efficiencies at the Federal Trade Commission 1997–2007, p. 2 (2009). 17  See generally 2010 Guidelines, § 10 (“In adhering to this approach, the Agencies are mindful that the antitrust laws give competition, not internal operational efficiency, primacy in protecting customers”). 18 1997 Guidelines, § 4. See Muris (1999). 19 See id. at n. 36 (“Section 7 of the Clayton Act prohibits mergers that may substantially lessen competition ‘in any line of commerce . . . in any section of the country.’ Accordingly, the [DOJ and FTC] normally assess[] competition in each relevant market affected by a merger independently and normally will challenge the merger if it is likely to be anticompetitive in any relevant market. In some cases, however, the Agency in its prosecutorial discretion will consider efficiencies not strictly in the relevant market, but so inextricably linked with it that a partial divestiture or other remedy could not feasibly eliminate the anticompetitive effect in the relevant market without sacrificing the efficiencies in the other market(s). Inextricably linked efficiencies rarely are a significant factor in the Agency’s determination

456   Howard Shelanski efficiencies, and by extension inefficiencies, are not only to be assessed based on price effects, but also through potential changes in quality, service, and innovation. In substantial part, these 1997 provisions remain unchanged in the 2010 Guidelines, although the 2010 Guidelines add new language governing consideration of innovation,20 and of fixed cost efficiencies.21 Courts, like the antitrust agencies as reflected in the Guidelines, have become increasingly receptive to efficiencies defenses in merger cases. Indeed, the courts to some degree foreshadowed and perhaps motivated some of the changes the antitrust agencies implemented in 1997. A good example comes from the Eleventh Circuit’s decision in FTC v. University Health, Inc., in which the court found that evidence of significant merger efficiencies could rebut a prima facie case of illegality:22 It is clear that whether an acquisition would yield significant efficiencies in the relevant market is an important consideration in predicting whether the acquisition would substantially lessen competition. Market share statistics, which the government uses to make out a prima facie case under section 7, are not an end in not to challenge a merger. They are most likely to make a difference when they are great and the likely anticompetitive effect in the relevant market(s) is small”). 20 2010 Guidelines, § 10 (“When evaluating the effects of a merger on innovation, the Agencies consider the ability of the merged firm to conduct research or development more effectively. Such efficiencies may spur innovation but not affect short-term pricing. The Agencies also consider the ability of the merged firm to appropriate a greater fraction of the benefits resulting from its innovations. Licensing and intellectual property conditions may be important to this inquiry, as they affect the ability of a firm to appropriate the benefits of its innovation. Research and development cost savings may be substantial and yet not be cognizable efficiencies because they are difficult to verify or result from anticompetitive reductions in innovative activities”). For further discussion of the consideration of innovation effects in merger efficiency analysis, particularly concerning the difficulty of measuring innovation effects and identifying them as merger-specific, and concerning how firm size relates to innovation calculations, and how innovation efficiencies would be counted under total-surplus and consumer-welfare standards, see Katz and Shelanski (2007, 135). 21  Id. at n. 15 (“The Agencies also may consider the effects of cognizable efficiencies with no short-term, direct effect on prices in the relevant market. Delayed benefits from efficiencies (due to delay in the achievement of, or the realization of customer benefits from, the efficiencies) will be given less weight because they are less proximate and more difficult to predict. Efficiencies relating to costs that are fixed in the short term are unlikely to benefit customers in the short term, but can benefit customers in the longer run, e.g., if they make new product introduction less expensive”). 22  See 938 F.2d 1206, 1222 (1991); see also FTC v. Butterworth Health Corp., 946 F.Supp. 1285, 1301 (1996) (“In sum, the Court is persuaded that the proposed merger would result in significant efficiencies[,]. . . . savings that would . . . invariably be passed on to consumers”); see generally U.S. v. Mercy Health Services, 902 F.Supp. 968, 987 (1995) (“The court concludes, however, that for purposes of this case, the defendants have failed to meet even the lower burden of showing an efficiencies defense by a preponderance of the evidence”); U.S. v. Third Nat. Bank in Nashville, 390 U.S. 171, 192 (1968) (“The burden of showing that an anticompetitive bank merger would be in the public interest because of the benefits it would bring to the convenience and needs of the community to be served rests on the merging banks. A showing that one bank needed more lively and efficient management, absent a showing that the alternative means for securing such management without a merger would present unusually severe difficulties, cannot be considered to satisfy that burden”); United States v. Archer-Daniels-Midland Co., 781 F. Supp. 1400 (S.D. Iowa 1991).

Efficiency Claims and Antitrust Enforcement   457

themselves; rather, they are used to estimate the effect an intended transaction would have on competition. Thus, evidence that a proposed acquisition would create significant efficiencies benefiting consumers is useful in evaluating the ultimate issue— the acquisition’s overall effect on competition. We think, therefore, that an efficiency defense to the government’s prima facie case in section 7 challenges is appropriate in certain circumstances.23

Despite the increasing acceptance of merger efficiencies by the antitrust agencies and the courts, the issue is not without debate in policy circles. Some commentators argue that the emphasis on efficiencies is contrary to antitrust law’s original legislative purpose of protecting consumers from mergers that increase market power, regardless of the efficiencies that might increase overall social wealth (Kirkwood and Lande 2008, 191–92). No amount of realized efficiency, the argument goes, can offset the unfair transfer of wealth from consumer to price-raising firms that possess too much market power.24 Such a transfer is analogous to stealing—in the sense that it is a practice condemned not because it is inefficient but because it is unfair.25 An opposing view of the purpose of antitrust law, articulated by Robert Bork among others, argues that Congress’s objective in passing antitrust law was to enhance efficiency (Bork 1966, 7). The legislative history of the Sherman Act, for example, is peppered with condemnations of businesses that possessed enough market power to raise prices.26 Therefore, Bork argues, because higher prices are a proxy for inefficiency, Congress’s concern must have been efficiency. It is beyond the scope of this chapter to try to resolve such debates over the legislative purpose behind the Sherman and Clayton Acts. The purpose of raising the above arguments is only to show that debate continues over the role efficiencies should play in merger and other antitrust review even as the recognition of an efficiency defense becomes more widely implemented and accepted. Another area in which efficiencies have become increasingly accepted as a defense is in vertical restraints. This evolution can be seen most clearly in the evolution away from per se rules against such restraints that began in the mid-1960s. That evolution led to an abandonment of per se liability for vertical nonprice restraints in the Supreme Court’s 1977 GTE/Sylvania27 decision and to the extension of rule-of-reason treatment to vertical price restraints in the Court’s 2007 Leegin28 decision. The main benefit of vertical restraints is that they can help to overcome a variety of inefficiencies that can arise from simple linear pricing between upstream and downstream firms. Those inefficiencies include double-marginalization, underprovision of effort, and free-riding by 23 

Id. Id. at 196. 25  Id. at 200–201. 26  See, e.g., 21 Cong. Rec. 2462, 2457 (1890) (statement of Sen. Sherman) (“[Trusts tend to] advance the price to the consumer”); id. at 2558 (statement of Sen. Pugh) (“[T]‌rusts . . . [destroy] competition in production and thereby increas[e] prices to consumers”). 27  433 U.S. 36 (1977). 28  551 U.S. 877 (2007). 24 

458   Howard Shelanski downstream retailers.29 Vertical restraints can help to overcome such problems in a variety of settings, and therefore may improve efficiencies even while they may diminish some forms of (notably intrabrand) competition.30 For this reason, US antitrust law now treats essentially all vertical restraints case by case, under a rule of reason in which the efficiencies of the vertical restraints factor centrally into the balance between procompetitive and anticompetitive effects of the conduct.

19.3.  Analysis of Efficiencies by the US Antitrust Agencies The US antitrust enforcement agencies today consider efficiencies as a matter of course in their investigations into mergers and other business conduct. With respect to mergers, if the combining parties can show the agencies that efficiencies from the transaction are likely to make the firm more competitive and benefit consumers by reducing prices, improving quality, or introducing new products, such efficiencies can offset concerns that the merger might increase market power, especially in close cases.31 In conduct cases, a firm or set of firms can present efficiencies as a procompetitive business justification for the allegedly anticompetitive conduct that is being investigated. In the horizontal context, agreements or coordination among competitors can sometimes allow firms to use assets more efficiently, produce new goods and services more quickly, cut costs, increase quality, or increase investment in new technology.32 Because coordination can have anticompetitive effects, however, the agencies balance any such efficiency justifications against potential anticompetitive effects of the conduct in applying “rule of reason” scrutiny. Of course, US antitrust agencies and courts do not consider alleged efficiencies in the case of agreements to fix prices, divide territories, or limit output, as such agreements are per se illegal under US law.33 Several forms of unilateral conduct, like exclusive dealing, tying, loyalty discounts, and vertical price and nonprice restraints, may have such procompetitive benefits as preventing free-riding, expanding output, achieving economies of scope, and reducing prices. Such conduct can also, in certain circumstances, harm competition. Accordingly, 29  For a helpful discussion of vertical contracting inefficiencies and how vertical restraints can help to overcome those inefficiencies, see Note for the United States, Roundtable on Vertical Restraints for On-line Sales, OECD (Feb. 25, 2013), available at http://www.ftc.gov/bc/international/​ docs/1302verticalrestraints-us.pdf. 30  Id. 31  See U.S. Dep’t of Justice and Federal Trade Comm’n, Horizontal Merger Guidelines (hereinafter Merger Guidelines) § 10 (2010), available at http://www.justice.gov/atr/public/guidelines/hmg-2010.pdf. 32  See U.S. Dep’t of Justice and Federal Trade Comm’n, Antitrust Guidelines for Collaborations Among Competitors (hereinafter Competitor Collaboration Guidelines) § 2.1 (2000), available at http://www.ftc.​ gov/os/2000/04/ftcdojguidelines.pdf. 33  See, e.g., Nat’l Soc’y of Prof ’l Eng’rs v. FTC, 435 U.S. 679, 692 (1978).

Efficiency Claims and Antitrust Enforcement   459

US antitrust law treats such restraints under the rule of reason and balances the harms and efficiencies they might cause. In all of the above kinds of cases, once a plaintiff or agency has demonstrated that a merger or course of conduct is likely to harm competition, then the burden of demonstrating offsetting efficiencies falls to the party or parties under investigation. This makes sense as the evidence for potential efficiencies is usually in the parties’ possession.34 The antitrust agencies require that such evidence go beyond mere speculation and be both specific and verifiable.35

19.3.1.  Mergers and Efficiencies Mergers can have a number of beneficial effects that one might label “efficiencies.” The efficiency that fits most directly with the conventional economic framework of antitrust is that of lower costs. Other efficiencies that enforcement agencies might take into account, and which may be less quantifiable but still provable, include speeding the introduction of new products or enabling innovation.36 Firms can reduce their costs through merger by bringing together complementary technology and “know-how,” eliminating duplicative assets or activities, and gaining achieving scale economies.37 The resulting savings might appear through reductions in the combined firm’s fixed or variable costs of production. Variable costs change with a firm’s output level, while fixed costs do not change with output, at least over the short run. Variable costs, particularly “marginal costs” of the last unit of production, most directly affect short-run market prices. In a purely competitive market, firms set price at marginal cost, so a reduction in marginal cost has an immediate effect on market price. But even firms with market power reduce prices and expand output when the firms’ marginal costs decline. For example, as a monopolist’s marginal cost of production falls, its profit-maximizing price declines and its privately optimal level of output increases. Therefore, efficiencies that reduce variable costs are more likely to be recognized by antitrust agencies than are other kinds of efficiencies that less directly affect market prices. Fixed cost reductions may more often lead to one-time savings that do not translate into sustained price reductions, although some fixed cost savings may in fact result in short-term price reductions (e.g., when selling prices are determined on a cost-plus basis that incorporates fixed costs).38 Mergers may also allow firms better to achieve “dynamic efficiencies” in the form of increased development and introduction of new technologies and products. The 2010 34 

See, e.g., U.S. Dep’t of Justice and Federal Trade Comm’n, Commentary on the Horizontal Merger Guidelines (hereinafter Commentary) at 50 (2006), available at http://www.ftc.gov/os/2006/03/​ CommentaryontheHorizontalMergerGuidelinesMarch2006.pdf. 35 Merger Guidelines § 10. 36  Id. See also Commentary, at 49. 37  Id. 38  Id., at 58.

460   Howard Shelanski US Guidelines expressly recognize innovation related efficiencies in Sections 6.4 and Section 10. In the former, the Guidelines state that the agencies “consider whether the merger is likely to enable innovation that would not otherwise take place, by bringing together complementary capabilities that cannot be otherwise combined or for some other merger-specific reason.” Section 10 makes clear it is still the burden of the merging parties to show that such dynamic efficiencies are sufficiently verifiable to be cognizable. In a substantial number of transactions over the last 20 years, the FTC and DOJ have expressly examined how a merger might harm or benefit innovation, and in some cases the innovation-related efficiency has factored importantly into approval of a transaction.39 As a general matter, the analysis of merger efficiencies takes place within the framework of US antitrust law’s consumer welfare standard. While there is a vigorous debate about whether competition policy should have as its objective total welfare (as in Canada) or consumer welfare (as in the United States and EU), for the most part the US Guidelines only consider efficiencies where consumer prices are not significantly increased.40 No court has endorsed a merger likely to increase prices on the ground that it was likely to enhance economic efficiency. Courts, like the agencies, have required merging parties to demonstrate that efficiencies would lead to cost savings that would be passed on to consumers. In FTC v. Swedish Match, the court held that the defendants’ efficiency evidence was insufficient to overcome the presumption of illegality because the defendants had not shown how much of their cost savings they would pass on.41 And in FTC v. Libbey, Inc., the district court ruled that, despite the defendants’ efficiencies evidence, the defendants had not shown “why Libbey will not use this opportunity to raise its own prices.”42 Furthermore, in FTC v.  Staples, Inc.43 and Dr.  Pepper/Seven-Up Cos. v.  FTC,44 the court found that potential efficiencies did not excuse the transaction because, despite the potential cost savings, consumers would likely pay higher prices after the merger. So, even mergers that create efficiencies may be illegal if they increase the risk of higher consumer prices.45 39 

See Statement of Chairman Timothy J. Muris, Genzyme Corporation/Novazyme Pharmaceuticals, Inc., No. 021-0026 (F.T.C. Jan. 13, 2004), available at http://www.ftc.gov/os/2004/01/murisgenzymestmt.​ pdf. For a more general discussion of how the agencies have considered dynamic efficiencies in merger review, see Katz and Shelanski (2007, 1) (discussing the dramatic increase in policymaker’s concern with fostering innovation within the context of merger enforcement during the 1980s and 1990s). 40  See 2010 Guidelines, § 10 (“[T]‌he Agencies consider whether cognizable efficiencies likely would be sufficient to reverse the merger’s potential to harm customers in the relevant market, e.g., by preventing price increases in that market. . . . In adhering to this approach, the Agencies are mindful that the antitrust laws give competition, not internal operational efficiency, primacy in protecting customers.”). 41  131 F. Supp. 2d 34, 53 (D.D.C. 2002). 42  211 F. Supp.121, 147 (E.D.N.Y. 1997). 43  970 F. Supp. 1066, 1090-1091 (D.D.C. 1997). 44  798 F. Supp. 762, 777 (D.D.C. 1992). 45  See, e.g., United States v. Franklin Elec. Co., 130 F. Supp. 2d 1025, 1035 (W.D. Wis. 2000) (“Defendants have not made the necessary showing that efficiencies would result and that they would lead to benefits for consumers in the relevant market”); FTC v. Tenet Healthcare Corp., 17 F. Supp. 2d 937, 948 (E.D. Mo. 1998) (“[M]‌any of the projected efficiencies would not benefit Poplar Bluff

Efficiency Claims and Antitrust Enforcement   461

19.3.1.1.  Proof of Efficiencies in Merger Cases The 2010 Guidelines state that “[t]‌he Agencies will not challenge a merger if cognizable efficiencies are of a character and magnitude such that the merger is not likely to be anticompetitive.”46 The Guidelines characterize as “cognizable” those efficiencies that (1) are merger-specific, (2) are verified, and (3) “do not arise from anticompetitive reductions in output or service.”47 Although the Merger Guidelines do not explicitly require that merged firms pass cost savings on to consumers, they do note that “[t]‌he greater the potential adverse competitive effect of a merger, the greater must be the cognizable efficiencies, and the more they must be passed through to customers, for the agencies to conclude that the merger will not have an anticompetitive effect in the relevant market.”48 Some courts have required pass-through of efficiencies to consumers.49 Merger-specific efficiencies are those that are “likely to be accomplished with the proposed merger and unlikely to be accomplished in the absence of . . . the proposed merger.” Alternative means of accomplishing the efficiency (e.g., a joint venture or a contractual arrangement) must be “practical in the business situation” and not “merely theoretical.”50 The Commentary to the Merger Guidelines uses the example of a non-merger-specific efficiency in merging firms that propose to consolidate their packaging facilities.51 Where one of the merging firms planned to shift some of its packaging operations from packaging facilities it was closing to its own packaging facilities, and not a packaging facility then owned by the other party to the transaction, this cost savings could occur without the merger, and it therefore did not meet the merger-specificity requirement.52 Verification of efficiencies means the agencies must be able to verify “by reasonable means the likelihood and magnitude of each asserted efficiency, how and when each would be achieved (and any costs of doing so), [and] how each would enhance the merged firm’s ability and incentive to compete.”53 Efficiencies are most likely to be substantiated when they are analogous to past experiences.54 The agencies will not consider efficiency claims that are “vague, speculative, or otherwise cannot be verified by consumers”), rev’d, 186 F.3d 1045 (8th Cir. 1999); FTC v. Butterworth Health Corp., 946 F. Supp. 1285, 1301(W.D. Mich. 1996) (positing that acquisition would yield substantial cost savings “that would, in view of defendants’ nonprofit status and the Community Commitment, invariably be passed on to consumers”), aff ’d, 121 F.3d 708 (6th Cir. 1997); see also Gifford and Kudrle (2005 at 447–62) (“[m]any courts have indicated that some of the merger generated cost savings must be passed on to consumers, thus suggesting a consumer-surplus approach to the evaluation of efficiency”). 46 Merger Guidelines § 10. 47 

Id.

48 Merger Guidelines §10. 49 

See, e.g., United States v. Long Island Jewish Medical Center, 983 F.Supp. 121, 149 (E.D.N.Y. 1997); FTC v. University Health, Inc., 938 F.2d 1206, 1222 (11th Cir. 1991). 50 Merger Guidelines §10. 51  See Commentary, at 52–53 (citing Fine Look—Snazzy, Disguised FTC Matter). 52  Id. 53 Merger Guidelines § 10. 54  Id.

462   Howard Shelanski reasonable means.”55 Since cost savings are both quantifiable and under the control of the merging parties, they are less likely to be vague and speculative. On the other hand, the realization of dynamic efficiencies often depends in part on events out of the control of the merging parties, and likely to be achieved further into the future. Therefore, dynamic efficiencies are likely to be more speculative than cost savings. Generally, the agencies will challenge a merger that has anticompetitive effects in any relevant market.56 However, the agencies reserve the discretion to “consider efficiencies not strictly in the relevant market, but so inextricably linked with it that a partial divestiture or other remedy could not feasibly eliminate the anticompetitive effect in the relevant market without sacrificing the efficiencies in the other market(s).”57 This recognition of “out of market” efficiencies is important because it can prevent tunnel vision that overly limits the range of acceptable efficiencies to the detriment of society. If costs to “in market” consumers are modest but gains to other consumers would be very large, blocking the merger would prevent small harms but also larger benefits, leaving consumers as a whole worse off. Consistent with this concern, the agencies will be more likely to credit such out-of-market efficiencies “when they are great while the anticompetitive effect in the relevant market(s) is small.”58 Despite this increased allowance for out-of-market efficiencies in the 2010 Guidelines, however, the case law has yet to develop on this issue, and the precedent on the books to date still appears to preclude parties from raising out-of-market efficiencies when defending a merger in litigation.59 On a comparative note, the underlying approach to merger enforcement in the European Union is rooted in the early structural approach of the US agencies but expressly recognizes efficiencies under the European Commission’s 2004 guidelines.60 As in the United States, the EC has stated in its Horizontal Guidelines that it will consider substantiated efficiencies claims in its overall assessment of the merger, so long as those efficiencies are of direct benefit to consumers, merger-specific, substantial, timely, and verifiable. Efficiency considerations will not be taken into account where the merger results in the creation of a monopoly or quasi monopoly. In Microsoft/ Yahoo! Search Business (2010), the Commission accepted the parties’ efficiency claims, noting that the economies of scale involved would increase competition. However, in that case the EC identified no anticompetitive concerns for the efficiencies to offset. In Arsenal/DSP (2009), a case where competition concerns were involved, the Commission found that the noted efficiencies were not merger specific and were lower than what the parties might respectively have achieved without the merger. And in Ryanair/Aer Lingus (2007), the fixed cost efficiencies claimed were not sufficient to outweigh the detrimental market effect of high market shares.61 Consequently, while 55 

Id.

56 Merger Guidelines § 10, n. 14. 57 

Id. Id. 59  United States v. Philadelphia National Bank, 374 U.S. 321 (1963). 60  See Council Regulation 139/2004, 2004 O.J. (L 24) 1 (EC). 61  Id. 58 

Efficiency Claims and Antitrust Enforcement   463

the EC is open to considering efficiency claims, there has been no clear signal that the Commission will be particularly hospitable to efficiencies evidence in cases where it has identified a competitive concern.

19.3.1.2.  Merger Efficiencies in Practice The examples and discussions above raise the question of whether efficiencies to date are still recognized more in theory than in practice, given that the agencies appear to consider efficiencies only in cases where anticompetitive effects are “not great.”62 Related to this question is whether the agencies should expand the field of cognizable efficiencies,63 reduce evidentiary thresholds, or more seriously consider out-of-market and fixed cost efficiencies. Some have argued against such expansion of efficiencies analysis on grounds that a comparative weighing of market-power and efficiency effects is simply too hypothetical and complicated an analysis for courts to undertake (Fisher and Lande 1983). Others have argued that efficiencies should be considered more readily (Muris 1980, 381). Actual practice suggests that the agencies have adopted an intermediate path between wariness about proof of efficiencies and recognition that where efficiencies do exist it could be costly to ignore them. Few merger cases are actually litigated, so any analysis of the law of merger efficiencies must look as much at their internal treatment by the enforcement agencies as at their treatment in court. In a study of 186 second-request investigations in proposed mergers from 1997 to 2007, FTC staff found that the agency did not address efficiency claims at all in 39 of the cases (Coate and Heimert 2009). In 147 investigations, the FTC considered efficiencies, sometimes on the agency’s own initiative (Coate and Heimert 2009). Fixed cost efficiencies were no more likely to be rejected than variable cost efficiency arguments. Verification concerns were the most often raised objection, although the FTC rejected few efficiency claims out of hand. In most instances, the FTC raised concerns about the nature and proof of efficiencies but did not conclusively determine whether the claimed efficiency should be accepted or rejected, leaving it somewhat opaque from the publicly available evidence how the agencies are weighing efficiencies in their merger reviews. Further analysis, especially by the antitrust enforcement agencies themselves, would be useful for public understanding of how best to present efficiency justifications. Greater understanding of whether, and how, efficiencies have factored into the analysis of a particular transaction might also enable retrospective studies of premerger efficiency claims and actual postmerger results. Such studies would improve antitrust enforcement going forward. To be sure, how to measure ex post whether a merger has in fact created efficiencies, and if so what kinds, is challenging. 62 

2010 Guidelines, § 10.

63 Muris makes the case that capital-raising efficiencies (i.e. lower borrowing costs for larger firms),

promotional economies (i.e. lower marketing and advertising costs), and managerial economies (i.e. more productive management techniques of acquiring firm) should all be added to the list of cognizable efficiencies. Muris (1999, 734–35).

464   Howard Shelanski Comparisons of premerger and postmerger profitability might appear indicative of cost savings, but can just as easily reflect increased market power (Meeks and Meeks 1981). Some studies have looked not at profitability but directly at production and operating costs, and compared the premerger and postmerger levels.64 More such studies, across a range of industries, would be helpful for competition policy. Sufficiently rich cost data are often difficult to obtain, however, and disentangling merger-specific efficiency gains from those that would have resulted even absent the merger—say due to technological progress or other factors—requires a good control sample that may not exist if an industry as a whole has undergone a wave of mergers. While such challenges exist, the literature related to merger efficiencies appears to be developing economic, financial, and accounting tools that are promising for future research and for further development of policy toward efficiencies in merger enforcement.

19.4. Conclusion Efficiencies have come to have an important theoretical place in merger review and in antitrust enforcement generally. The burden of production and persuasion on merger efficiencies rests with the merging parties, and the evidence suggests that they will rarely meet those burdens in cases where the investigating agency predicts that the anticompetitive effects of the merger will be substantial. In practice, not only must the magnitude of proffered efficiencies grow with the likely anticompetitive effects, but so must their certainty and cognizability. Nonetheless, efficiencies have become a central part of merger evaluation, and the agencies have over the past decades become more expansive in the kinds of efficiencies they are willing to consider, adding dynamic to static considerations, adding fixed-cost to marginal-cost savings, and adding out-of-market to in-market efficiencies in some circumstances. It remains the case that efficiencies are likely to be most salient in close cases. Moreover, important questions remain about the circumstances under which the enforcement agencies will recognize different kinds of efficiencies going forward. Identifying those circumstances is a task both for further research and for the agencies themselves as merger practice evolves.

64  See, e.g., Kwoka and Pollitt (2010) (concluding that mergers were not consistent with increased efficiency in the electric power industry).

Efficiency Claims and Antitrust Enforcement   465

References Areeda, P. and D. Turner. 1980. Antitrust Law. Vol. 4. Boston: Little, Brown. Blair R. D., and J. Hayes. 2011. The Efficiencies Defense in the 2010 Horizontal Merger Guidelines. Review of Industrial Organization 39: 1, 57–68. Bork, R. H. 1966. Legislative Intent and the Policy of the Sherman Act. Journal of Law and Economics 9: 7–48. Coate, M. B., and A. J. Heimert. 2009. Merger Efficiencies at the Federal Trade Commission 1997–2007. Bureau of Economics. Fisher, A.  A., and R. H.  Lande. 1983. Efficiency Considerations in Merger Enforcement. California Law Review 71: 1580–696. Gifford, D. J. &R. T. Kudrle. 2005. Rhetoric and Reality in the Merger Standards of the United States, Canada, and the European Union. Antitrust Law Journal 72: 423–47. Katz, M.  L., and H. A.  Shelanski. 2007. Mergers and Innovation. Antitrust Law Journal 74(1): 1–84. Kirkwood, J.  B., and R. H.  Lande. 2008. The Fundamental Goal of Antitrust:  Protecting Consumers, Not Increasing Efficiency. Notre Dame Law Review 84: 191–92. Kwoka, J., and Michael Pollitt. 2010. Industry Restructuring, Mergers, and Efficiency: Evidence from Electric Power. International Journal Industrial Organization 28: 645–56. Meeks, G., and J. G. Meeks. 1981. Profitability Measures as Indicators of Post-merger Efficiency. J. Industrial Economics 29: 335–44. Muris, T. J. 1980. The Efficiency Defense Under Section 7 of the Clayton Act, Case Western Reserve Law Review 30: 381–432. Muris, T. J. 1999. The Government and Merger Efficiencies: Still Hostile after All These Years. George Mason Law Review 7: 729–32. Williamson, O. E. 1968. Economies as an Antitrust Defense: The Welfare Tradeoffs. American Economic Review 58: 18–36.

CHAPTER 20

U N I L AT E R A L E F F E C T S BRYAN KEATING AND ROBERT D. WILLIG

20.1. Introduction IN this chapter, the term “unilateral effects” refers to impacts of the incentives a horizontal merger might create for the merged firm unilaterally to raise prices, suppress output, or otherwise lessen competition even if nonmerging firms do not change their behavior (Willig 1991). Unilateral effects might arise even if the merger does not substantially increase the likelihood of successful coordinated interaction between firms, whether tacit or explicit.1 Such effects can occur in a variety of settings including differentiated products markets, markets in which products are relatively homogeneous and firms are differentiated primarily by available capacity, markets in which prices are set via auction or bidding, and markets in which innovation and product variety are of particular importance (DOJ and FTC 1992, § 2; 2010, § 6). In the United States, the antitrust enforcement agencies (the Department of Justice [DOJ] Antitrust Division and the Federal Trade Commission [FTC]) first 1 

Although the United States’ Horizontal Merger Guidelines discuss unilateral and coordinated effects as separate and distinct concepts, they are actually closely related. (See DOJ and FTC (2010), §§ 6 and 7.) As we describe below, analysis of unilateral effects typically makes use of static models that ignore dynamic elements of competition that may be particularly germane to analysis of coordinated effects. (See Willig 1991; Farrell and Shapiro 2010; and Carlton 2010.) Willig (1991) notes: Unilateral effects can be represented in a static model, in the repeated stage-game equilibrium of a supergame, or in a supergame where information problems make it impossible for equilibrium strategies to condition current actions on past ones. Coordinated effects can be represented in Nash equilibria of supergames when equilibrium strategies call for actions to depend on the histories of action profiles.

Unilateral Effects   467

explicitly recognized the importance of unilateral effects in the 1992 version of the Horizontal Merger Guidelines (1992 Horizontal Merger Guidelines).2 The 2010 version of the Horizontal Merger Guidelines (2010 Horizontal Merger Guidelines) significantly expands the discussion of unilateral competitive effects and introduces new formal tools for evaluating them.3 Similarly, competition authorities in other jurisdictions recognize the importance of unilateral effects in merger analysis. For example, in 2010, the Competition Commission and Office of Fair Trading in the United Kingdom released the Merger Assessment Guidelines (2010 Merger Assessment Guidelines) adopting a similar framework for assessing unilateral effects resulting from horizontal mergers (Competition Commission and Office of Fair Trading 2010, § 5.4). The European Commission’s Guidelines on the Assessment of Horizontal Mergers contain a discussion of “non-coordinated effects,” which are analogous to unilateral effects.4 In this chapter, we provide an overview of the theory underlying analysis of unilateral effects in different market settings. Our presentation closely follows the structure of the 2010 Horizontal Merger Guidelines.5 We begin with a discussion of the incentives underlying the pricing of differentiated products and then extend this discussion to incorporate homogeneous product markets, markets in which prices are set by bargaining and/ or bidding, and implications for innovation and product variety.

20.2.  Differentiated Products Many mergers occur in markets in which products are differentiated in that they are not perfect substitutes for one another due to variation in their appeal to consumers.6 In such 2 

DOJ and FTC (1992), § 2. The 1984 Horizontal Merger Guidelines focus primarily on the ability of firms to “coordinate their actions in order to approximate the performance of a monopolist” (§ 1). Nonetheless, those guidelines do develop methodologies that rely on the basic framework of unilateral effects. For example, § 3.1 notes, “The smaller the percentage of total supply that a firm controls, the more severely it must restrict its own output in order to produce a given price increase, and the less likely it is that an output restriction will be profitable” (Department of Justice 1984). 3  For a full discussion of the evolution of unilateral effects analysis in DOJ and FTC (2010) and the practices of the antitrust enforcement agencies, see Shapiro (2010a). 4  European Commission (2004), ¶¶ 24–38. Many other jurisdictions also evaluate unilateral effects. See, e.g., New Zealand Commerce Commission (2003), § 8; Competition Commission of Singapore (2007), §§ 6.3–6.6. 5  See DOJ and FTC (2010). § 6.1 addresses “Pricing of Differentiated Products”; § 6.2 addresses “Bargaining and Auctions”; § 6.3 addresses “Capacity and Output for Homogeneous Products”; § 6.4 addresses “Innovation and Product Variety.” 6  It is useful to distinguish between vertical differentiation, in which products are distinguished by dimensions of quality over which all consumers have the same preference ordering, and horizontal differentiation, in which products are distinguished by dimensions of quality over which consumers have heterogeneous preferences. See Bresnahan (1987) for an example of a model that assumes vertical differentiation. See Berry, Levinsohn, and Pakes (1995) for an example of a model of horizontal differentiation.

468   Bryan Keating and Robert D. Willig settings, unilateral effects may occur inasmuch as merging firms internalize the value of sales that divert to the merging partner as a result of a price increase, thereby creating an incentive to raise prices relative to premerger levels (absent offsetting efficiencies about which we say more below). In the United States, the 2010 Horizontal Merger Guidelines feature a tool for analyzing unilateral competitive effects from horizontal mergers of differentiated products that is referred to as “upward pricing pressure” (UPP).7 The United Kingdom’s 2010 Merger Assessment Guidelines discuss similar analysis.8 The UPP tool was first introduced by Farrell and Shapiro (2010), while having origins in earlier literature, including Willig (1991), Shapiro (1996), Werden (1996), O’Brien and Salop (2000), and Froeb, Tschantz, and Werden (2005). The 2010 Horizontal Merger Guidelines also contemplate the use of merger simulation models to quantify more fully the unilateral effects arising from a merger (DOJ and FTC 2010, § 6.1). Merger simulation models specify systems of demand (and possibly cost) functions along with an assumption about the competitive conduct of firms to assess equilibrium outcomes given assumptions about the structure of the industry. Merger simulation models can take into account changes in costs, quality, and product ownership to compare premerger equilibrium prices and output to predicted postmerger equilibrium prices and output. In doing so, these models provide an estimate of the magnitude of anticipated price, output, and ultimately, consumer welfare effects of a merger. As we discuss below, merger simulation techniques have a long history in the economics literature. UPP and merger simulation analysis derive from the same underlying concepts and rely on many of the same assumptions and input data. They differ primarily in the strength of the assumptions required and the breadth of the conclusions that can be drawn from the results. In the following sections, we first introduce the UPP analysis and then extend the discussion to consider merger simulation models.

20.2.1.  Upward Pricing Pressure Given the primacy of UPP in the recent literature and its place in policy discussion and the merger guidelines, it is useful to emphasize the economic theory that underlies this form of analysis. Before turning to the mathematical foundation of the UPP framework,

7 

While the term “upward pricing pressure” appears only once in the 2010 Horizontal Merger Guidelines (page 21), as we discuss below, the concepts underlying the UPP framework form the basis for the Guidelines’ discussion of unilateral effects. Shapiro (2010a) discusses the relationship between UPP and the existing economics literature. 8  Competition Commission and Office of Fair Trading (2010), § 5.4; Australian Competition and Consumer Commission (2008), §§ 5.1–5.7; Brazil Ministry of Finance and Ministry of Justice (2001), ¶¶ 39–62; Competition Bureau of Canada (2011), §§ 6.13–6.18; European Union (2004), §§ 27–38; Japan Fair Trade Commission (2011), pp. 18–19; New Zealand Commerce Commission (2003), § 8; Competition Commission of Singapore (2007), §§ 6.3–6.6. See also Walters (2007).

Unilateral Effects   469

it is worthwhile to quote the central passage from the 2010 Horizontal Guidelines on the subject (DOJ and FTC 2010, § 6.1): Adverse unilateral price effects can arise when the merger gives the merged entity an incentive to raise the price of a product previously sold by one merging firm and thereby divert sales to products previously sold by the other merging firm, boosting the profits on the latter products. Taking as given other prices and product offerings, that boost to profits is equal to the value to the merged firm of the sales diverted to those products. The value of sales diverted to a product is equal to the number of units diverted to that product multiplied by the margin between price and incremental cost on that product. In some cases, where sufficient information is available, the Agencies assess the value of diverted sales, which can serve as an indicator of the upward pricing pressure on the first product resulting from the merger. Diagnosing unilateral price effects based on the value of diverted sales need not rely on market definition or the calculation of market shares and concentration. The Agencies rely much more on the value of diverted sales than on the level of the HHI [Herfindahl-Hirschman Index] for diagnosing unilateral price effects in markets with differentiated products. If the value of diverted sales is proportionately small, significant unilateral price effects are unlikely. [Footnote 11: For this purpose, the value of diverted sales is measured in proportion to the lost revenues attributable to the reduction in unit sales resulting from the price increase. Those lost revenues equal the reduction in the number of units sold of that product multiplied by that product’s price.]

This passage from the 2010 Horizontal Merger Guidelines describes the basic unilateral incentives of firms to set prices in markets characterized by product differentiation. These incentives can be eloquently expressed via a mathematical model. Let Π1 , Π 2 , and Π m denote the profits of firm 1, firm 2, and the combination of firms 1 and 2, respectively. Let P1 and P2 denote the nominal prices for firms 1 and 2, whose demand functions are given by D1 ( P1 , P2 ) and D 2 (P1 , P2 ) . Assume that marginal costs do not vary with the level of output and the premerger marginal cost of firm i is given by ci0 , while the postmerger marginal cost is given by ci . Then the pre- and postmerger profit functions can be written as

(

) (

)

Π i = Pi 0 − ci0 D i P10 , P20 (20.1)

and

Π m = ( P1 − c1 ) D1 ( H1 , H 2 ) + (P2 − c2 ) D 2 (H1 , H 2 ) , (20.2)

respectively, where H i = Pi − vi is the quality-adjusted price of product i postmerger and vi is the postmerger improvement per unit in product quality resulting from the

470   Bryan Keating and Robert D. Willig merger.9 This implies that, premerger, H i0 = Pi 0 and the premerger demand function already reflects any product quality that consumers value. The UPP framework is based on an assumption that firms engage in static Bertrand price-setting behavior (Farrell and Shapiro 2010). In other words, it is assumed that firms unilaterally and simultaneously set prices to maximize static profits.10 The UPP approach assesses changes local to the pre-merger equilibrium.11 At the premerger equilibrium, the following first-order condition (FOC) for each firm i holds: ∂Π i = Pi0 − ci0 Dii P10 , P20 + D i P10 , P20 = 0, ∂Pi

(

) (

)

(

)



(20.3)



(

)

where P 0 denotes premerger prices for firm i, and Dii P10 , P20 ≡ (∂D i ) / (∂Pi ) . At equii librium prices, firm i has no incentive to raise or lower its price because the extra dollar gained (lost) on each unit of its sales D i P10 , P20 would be exactly offset by the lost (gained) margin associated with incremental sales represented by Pi 0 − ci0 Dii P10 , P20 . Conversely, if the FOC were positive, the firm would have an incentive to raise price because lost sales would not fully offset the increased margin earned on retained sales. Postmerger, these incentives change. To see this, consider the partial derivative of Π m with respect to H1 assuming that v2 = 0 and c2 = c20 for the sake of illustration:

(

)

(

) (

)

∂Π m = ( H1 + v1 − c1 ) D11 ( H1 , H 2 ) + D1 ( H1 , H 2 ) + P2 − c20 D12 ( H1 , H 2 ) . (20.4) ∂H 1

(



)

To understand the economic meaning of UPP, consider equation (20.4) evaluated at premerger quality-adjusted prices, H i = Pi0 . If the expression were positive it would indicate that the combined firm would find it profitable to raise the price of product 1 above its premerger level, all else equal:12 9 

For a fuller discussion of “hedonic” or quality-adjusted prices, see Willig (1978). As Carlton (2010) notes, “The Bertrand model is static in the sense that firms are assumed to make simultaneous decisions about price while recognizing the interdependence of their decisions.” This assumption may not always be appropriate. See Carlton (2010) for a discussion and Jaffe and Weyl (2012) for a more general framework. 11  This approach, sometimes called a “first-order approach,” has its origins in Werden (1996), who used it to assess the compensating marginal cost reduction (CMCR) required to offset the anticompetitive effects of a merger. This has important implications for the inferences that can be drawn from the framework. As we discuss further below, because UPP is based on local deviations from the premerger equilibrium, it is not designed to assess the magnitude of the predicted effects. To assess the magnitude requires additional information or assumptions about the nature of demand facing firms. 12  “All else equal” is an important caveat. The UPP formula as described here does not account for other important factors, including entry, repositioning, all efficiencies, all quality improvements, or rival price responses that affect the welfare impact of a merger. 10 

Unilateral Effects   471

∂Π m ∂P1

(

) ( ) ( ) ( ) ( ) = ( p − c ) D ( P , P ) + D ( P , P ) + (c − c + v ) D ( P , P ) + ( P − c ) D ( P , P ) ∂Π = + (c − c + v ) D ( P , P ) + ( P − c ) D ( P , P ) ∂P = (c − c + v ) D ( P , P ) + ( P − c ) D ( P , P ) ,

= p10 + v1 − c1 D11 P10 , P20 + D1 P10 , P20 + P20 − c20 D12 P10 , P20 pre − merger

0 1

0 1

1 1

0 1 1

1

1

1

∂Π1 ∂P1

0 2

1 1

1

0 1

0 1

0 1

0 2

1

0 2

0 2

1 1

1

0 2

2 1

0 1

0 2

0 2

2 1

0 1

0 2

0 2

0 2

pre − merger

0 1

where

0 1

1

1

1 1

0 1

0 2

0 2

0 1

0 2

2 1

0 1

0 2

(20.5)

= 0 from equation (20.3). pre − merger

Equation (20.5) captures the partial derivative of the profits of the combined firm with respect to the quality-adjusted price of product 1, evaluated at premerger prices. The combined firm has an incentive to raise the price of product 1 if this term is greater than zero:

(c

0 1

) (

) (

) (

)

− c1 + v1 D11 P10 , P20 + P20 − c20 D12 P10 , P20 > 0. (20.6)

The second term in equation (20.6) represents the value of diverted sales described in the Guidelines. To see this, note that D12 (P10 , P20 ) represents the change in demand for 0 0 product 2 given an increase in the price of product 1, while (P2 − c2 ) represents the margin earned on those diverted sales.13 Whereas, when deciding whether to raise its price, the premerger firm takes into account the offsetting effects of lost marginal sales and increased margins only on its own infra-marginal sales, the merged firm also accounts for margins earned on sales that divert to the merging partner.14 This internalization of the value of diverted sales alters the balance between increased profits on inframarginal sales (to the combined firm) and lost profits on marginal sales, leading the merged firm to find it profitable to increase prices, all else equal.

13  The relevant demand response for assessing competitive effects relates to the postmerger demand functions. While merging products may be close substitutes for each other premerger, readily available alternatives may develop in the postmerger world. As an extreme example, suppose product B is the closest substitute for product A premerger such that the diversion ratio from A to B is quite high. Further suppose that product C is also a close substitute for product A, but slightly less so than product B. Postmerger, customers may readily substitute to product C even though they were not observed to do so premerger. Said differently, estimating diversions based on premerger demand functions may fail to capture important repositioning that affects the merging firms’ incentives to raise price (DOJ and FTC 2010, §§ 6.1, 9). See also Carlton (2010) and Willig (2011). 14  If products are complements such that an increase in the price of product 1 decreases demand for product 2, this logic is reversed and a merger will lead to downward rather than upward pricing pressure.

472   Bryan Keating and Robert D. Willig Equation (20.6) allows for the possibility that a merger could decrease the marginal cost of producing product 1 and/or increase the quality of product 1. To the extent that c10 − c1 + v1 D11 P10 , P20 is negative, either due to lower marginal costs or increased product quality, this can offset or eliminate the adverse effects on consumer welfare of what might otherwise be upward pricing pressure. Rearranging equation (20.6) shows that upward pricing pressure arises if

(



) (

)

(P

0 2

− c20 0 1

P

) δ > (c

0 1

12

− c1 + v1 0 1

P

) , (20.7)

where δ12 = [D12 (P10 , P20 )] /[− D11 (P10 , P20 )] is the diversion ratio from product 1 to product 2. The diversion ratio captures the fraction of customers’ demand leaving product 1 in response to a price increase that switches to product 2.15 The left-hand side of equation (20.7) is sometimes referred to as the Gross Upward Pricing Pressure Index (GUPPI), and it captures the incentives of the merged firm to raise the price of product 1 absent off-setting cost efficiencies or quality effects.16 The formulation of the GUPPI offers a number of advantages for use in merger control. First, the GUPPI is an index scaled naturally to observed prices. This means that it can be compared to measures of cost efficiencies or quality improvements also expressed as a percentage of observed prices.17 Second, the GUPPI is composed of inputs, including prices, marginal costs, and diversion ratios, which can often be approximated with

15  See, e.g., Katz and Shapiro (2003). Diversion ratios may be computed in many different ways depending on available data. One straightforward way to estimate diversion ratios is to assume that they are equal to the market share of the receiving product divided by one minus the share of the product with the price rise. This approach is implied by a simple “flat” logit model of demand. Other specifications of demand and more sophisticated econometric modeling allow for more flexible and realistic substitution patterns, which may lead to diversion ratios that are not so simply related to market shares. In some cases, “natural experiments,” whereby the prices or availability of relevant products change in the data, allow direct measurement of diversion ratios. 16  Salop and Moresi (2009); Moresi (2010); and Farrell and Shapiro (2010). The 2010 Horizontal Merger Guidelines describe the implications of equation (20.7):

If the value of diverted sales is proportionately small, significant unilateral price effects are unlikely. [Footnote 11: For this purpose, the value of diverted sales is measured in proportion to the lost revenues attributable to the reduction in unit sales resulting from the price increase. Those lost revenues equal the reduction in the number of units sold of that product multiplied by that product’s price.] (DOJ and FTC (2010), § 6.1) 17  Efficiencies can also be expressed relative to marginal costs by rescaling by the ratio of price to marginal cost.

Unilateral Effects   473

readily available data.18 For these reasons, the GUPPI provides a useful tool that allows enforcement agencies to screen potential mergers for unilateral competitive effects. The right-hand side of equation (20.7) shows that mergers are less likely to result in upward pricing pressure the greater are the marginal cost efficiencies arising from the transaction and/or the greater the improvements in quality. This result is consistent with the 2010 Horizontal Merger Guidelines, which note that mergers may “generate significant efficiencies and thus enhance the merged firm’s ability and incentive to compete, which may result in lower prices, improved quality, enhanced service, or new products” and that “[t]‌he Agencies will not challenge a merger if cognizable efficiencies are of a character and magnitude such that the merger is not likely to be anticompetitive in any relevant market” (DOJ and FTC 2010, § 10). For example, a merger that is expected to generate marginal cost saving equivalent to 10 percent of the premerger price for product 1 would not lead to competitive concerns if the GUPPI was less than 10 percent. Similarly, if a merger is expected to generate quality enhancements that consumers value at 10 percent of the premerger price of product 1, then the merger would not raise any anticompetitive concerns even absent cost efficiencies.19 As equation (20.7) indicates, the benefits of cost reductions and quality improvements are additive, such that any combination of these effects can be summed to determine the threshold for the applicable GUPPI below which the merger would not raise competitive concern with respect to the specific product. The 2010 Horizontal Merger Guidelines do not provide numerical thresholds below which a GUPPI would not indicate potential concern, while the Guidelines do provide such thresholds for applicable Herfindahl-Hirschman Indexes (HHIs) (DOJ and FTC 2010, § 5.3). The 2010 Horizontal Merger Guidelines state only that “[i]‌f the value of diverted sales is proportionately small, significant unilateral price effects are unlikely” (DOJ and FTC 2010, § 6.1). Nonetheless, statements by antitrust enforcers and economic theory provide some basis for assessing what levels of GUPPIs are likely to indicate concern (see Farrell and Shapiro 2010; Willig 2011). For example, in a 2010 speech, then deputy assistant attorney general Carl Shapiro stated, “Current Division practice is to treat the value of diverted sales as proportionately small if it is no more than 5% of the lost revenues” (Shapiro 2010b).20 This practice is consistent with some simple assumptions about demand functions and the level of efficiencies likely to be

18  While proxies for margins and diversion ratios are typically available in the data, important measurement issues can arise. Accounting margins are readily available, but it is often more difficult to acquire information on the incremental margins most relevant to economic analysis. See Carlton (2010). 19  Israel et al. (2012) discuss one approach to estimating the value of quality enhancements in the context of airline mergers. Similar techniques potentially could be applied to other industries such as hospitals or telecommunications. Quality efficiencies may be particularly pertinent in industries characterized by relatively low marginal costs and high fixed costs associated with product R&D and design. (See Willig 2011.) 20  Farrell and Shapiro (2010) suggested that the agencies make a (rebuttable) presumption that marginal cost reductions due to a merger would be 10 percent of premerger marginal costs.

474   Bryan Keating and Robert D. Willig achieved by merging firms. For example, under the assumption that demand is linear in price, it can be shown that changes in marginal costs are passed through to prices at a rate of 50 percent.21 The same analytics apply to the impact of a GUPPI on a merged firm’s profit-maximizing choice of price, given linear demand.22 Thus, here, a GUPPI of less than 5 percent will increase prices by less than 2.5 percent, and this would be counteracted by a quality increase or a marginal cost decrease of 5 percent of the premerger price. Willig (2011) shows that the UPP framework can be readily modified to accommodate additional fact patterns that often arise in the evaluation of mergers. For example, it is straightforward to modify the analysis to account for partial equity stakes.23 The UPP framework can also be modified to account for situations in which firms choose quantities rather than prices. We return to this setting in section 1.2 below. A number of theoretical and practical critiques have been aimed at UPP (see, e.g., Coate and Simons 2010; Schmalensee 2009; Hausman, Moresi, and Rainey 2011; and Carlton 2010). Its most important limitation is related to its greatest asset. UPP analysis, by focusing on small price changes from the premerger equilibrium, investigates the incentives of newly merged firms to raise prices without resorting to assumptions about the nature of demand facing the firms away from the premerger equilibrium. Said differently, the value of diverted sales that is central to the computation of UPP is a measure of the opportunity cost to the merged firm.24 It is well known that the pass-through of cost changes depends on the curvature of the demand function, and not just on its slope or elasticity.25 As a result of its focus on local pricing pressures, UPP cannot be used to assess the magnitude of the predicted price changes, and therefore it can be more difficult to balance the conclusions of the UPP analysis against offsetting effects such as entry, repositioning, and the reactions of both customers and rivals to the merger. A second limitation of UPP is that it does not model strategic responses by nonmerging firms. Depending on the situation, such feedback effects may either enhance or offset merger-related competitive effects. For example, in a differentiated product Bertrand model, absent any efficiencies, it is conventional wisdom that rival firms will respond to 21  This result relies on the assumption that only one price is affected by the merger. If there is diversion in both directions and prices of both products change as a result of the merger, then pass-through will not necessarily be one-half even with linear demand curves. See Willig (1991), Schmalensee (2009), and Hausman, Moresi, and Rainey (2011). 22  See Jaffe and Weyl (2010) for a discussion of why linear demand curves may not serve as an appropriate baseline assumption. 23  For example, Verizon Wireless entered into joint marketing agreements with Comcast and other cable providers. Verizon, which competes with the cable providers through its FiOS product, holds a partial equity stake in Verizon Wireless, which is a joint venture with Vodafone. This partial equity stake may have important implications for analyzing the incentives that the marketing agreements generate for Verizon to raise prices. 24  Additional sales made by one unit of the merged firm must account for the lost profits to another unit of the merged firm resulting from cannibalization of sales. (See Shapiro 2010.) 25  For a discussion of pass-through, see Bulow and Pfleiderer (1983) and Weyl and Fabinger (2012).

Unilateral Effects   475

price increases by the merging firm by increasing their own prices, thus magnifying the effects of the merger. However, if the merger allows the merging firms to increase quality, then rival firms may respond by lowering their prices.26 Another limitation is that UPP can lead to indeterminate conclusions in cases where the UPP indicators for different products involved in a merger point in opposite directions.27 This can occur, for example, if marginal cost or quality effects differ substantially across products. It can also occur if the diversion ratios or margins are very different across products. Finally, practitioners may face challenges in correctly measuring the requisite inputs to the analysis and in interpreting the results (Carlton 2010).

20.2.2.  Merger Simulation Many of these limitations of UPP analysis can be addressed by analyzing mergers with more highly structured models that allow for the operation of strategic interactions among firms and other nonlocal effects.28 In particular, if one is willing to make assumptions about the nature of the demand and cost functions facing firms, then merger simulation provides methods to further assess the competitive effects of a merger.29 By making explicit assumptions about the character of the demand function, the analyst can assess the effect of nonlocal changes from the premerger equilibrium. These movements are driven by changes in the ownership of products in the market. As in the analysis of UPP, 26  Kwoka and Shumilkina (2010) find such a pattern in their study of the price effects of the 1987 merger between USAir and Piedmont. 27  See Carlton (2010). Willig (2011) shows that the UPP framework can be applied to multiple products. However, he concludes that “little (if anything) is added by doing UPP analysis of movements in both prices simultaneously [relative to doing single-product UPP].” (32–3) Jaffe and Weyl (2012) propose a more general framework that aggregates multiple price changes into a single metric. 28  Use of indicative price rise (IPR) analysis represents an intermediate step between UPP and full merger simulation. Like UPP, assessing the IPR requires an assumption of Bertrand behavior and some information on gross margins and diversion ratios. Like merger simulation (but unlike UPP), IPR analysis also requires an assumption about the shape of the demand curve. For example, under an assumption of symmetry (i.e., the merging firms have equal margins and diversion ratios premerger) and linear demand, the IPR is equal to [(md )]/[2(1 − d )], m is the premerger gross margin and d is the diversion ratio. (See Shapiro 1996.) This type of analysis has been used by competition authorities in the United Kingdom and South Africa to assess mergers. See, e.g., Competition Commission (UK), “Somerfield plc and Wm Morrison Supermarkets plc: A report on the acquisition by Somerfield plc of 115 stores from Wm Morrison Supermarkets plc,” September 2005, available at http://www.competitioncommission.​org.uk/assets/competitioncommission/docs/pdf/non-inquiry/rep_pub/reports/2005/ fulltext/501.pdf; Competition Tribunal of South Africa, “In the matter between: Masscash Holdings (Pty) Ltd And Finro Enterprises (Pty) Ltd t/a Finro Cash and Carry,” Case No. 04/LM/Jan09, November 2009, available at http://www.saflii.org/za/cases/ZACT/2009/66.pdf. 29  See Carlton (2010). Merger simulation applied to differentiated product industries has a long history in the economic literature. (See, e.g., Baker and Bresnahan 1985; Hausman, Leonard, and Zona 1994; Werden and Froeb 1994 and 1996; Werden 1997; Nevo 2000b; Epstein and Rubinfeld 2001 and 2004.)

476   Bryan Keating and Robert D. Willig a change in ownership changes the incentives of the merging firm to set prices because the merging firm will internalize diversion between the merging parties’ products. Merger simulations model equilibrium prices and outputs under the premerger ownership structure and compare them to those under the postmerger ownership structure (while also potentially taking into account cost and quality changes resulting from the merger). Merger simulation is regularly used by economists at the US, European, and other antitrust enforcement agencies.30 The 2010 Horizontal Merger Guidelines note: Where sufficient data are available, the Agencies may construct economic models designed to quantify the unilateral price effects resulting from the merger. These models often include independent price responses by non-merging firms. (DOJ and FTC 2010, § 6.1)

However, merger simulation is not without cost. In particular, it may require researchers to make unverifiable assumptions about the nature of demand functions and competition. Research has shown that the choice of how to specify the demand function can have significant implications for the resulting conclusions (see, e.g., Shapiro 1996; Crooke et al. 1999; and Froeb, Tschantz, and Werden 2005). For this reason, the 2010 Horizontal Merger Guidelines ascribe particular importance to robust predictions of competitive effects rather than focusing on particular modeling assumptions (DOJ and FTC 2010, § 6.1).

20.2.2.1.  Basic Framework To present the mechanics of merger simulation, we start from the framework developed above to demonstrate the workings of UPP analysis. Assume that several firms produce substitute differentiated products.31 Also assume—as is standard, but not required—that the form of competition among them is characterized by a static Bertrand pricing game whereby firms choose prices to maximize static profits and take the prices of other firms as given.32 Whereas UPP analysis aims to assess the direction of the incentives to change price, merger simulation goes further to estimate the magnitude of price changes for each product. 30 

For example, an economic expert testifying on behalf of the United States relied on merger simulation analysis in United States District Court (2011), 38–39. Similarly, DG COMP performed merger simulation analysis in the Unilever/Sara Lee transaction in 2011. See De Coninck (2011). See also Brazil Ministry of Finance and Ministry of Justice (2001), ¶¶ 45 (“hypothetical firm”); New Zealand Commerce Commission (2003) § 8; Competition Commission and Office of Fair Trading (2010), § 4.3 (“counterfactual”). 31  For simplicity, we assume each firm offers a single product. It is straightforward to extend the analysis to multiproduct firms. (See, e.g., Nevo 2001.) 32  One can assume other forms of competition between firms. One alternative is to assume that firms behave according to the assumptions of the Cournot model whereby firms choose quantities to maximize profits taking the quantities of other firms as given. The Cournot assumption is more commonly used in models of undifferentiated products, which we discuss further below. Jaffe and Weyl (2012) extend the pricing pressure framework to encompass other forms of competitive behavior. Another possibility is to embed the static Bertrand model within a more general model. (See Bresnahan 1989 and Carlton 2011.)

Unilateral Effects   477

Recall that equation (20.3) provides an equality that characterizes the FOC of the pricing problem facing an independent firm in the market and equation (20.4) provides an equality that characterizes the FOC of the pricing problem facing the merged firm. Equation (20.6) provides an inequality that determines whether there is pressure on the merged firm to increase the price of Product 1. The inequality in equation (20.6) arises because we assess the FOC of the merged firm at premerger prices in order to determine whether the merged firm would have an incentive to raise prices. In contrast, merger simulation techniques proceed by evaluating the vector of (quality-adjusted) prices that would solve the system made up of equation (20.3) for the firms not merging and equation (20.4) for the merging firm (which characterizes postmerger incentives), and comparing it to the vector of (quality-adjusted) prices that would solve the system made up of equation (20.3) for all the firms in the market (which characterizes premerger incentives). There are three important differences between equation (20.3) and equation (20.4). First, equation (20.4) captures the internalization by the merged firm of the effect of its pricing decisions on the profitability of the products of its merging partner. This can be seen in the third term in equation (20.4), which is not present in equation (20.3). This same term appears in equation (20.6) as the value of diverted sales and captures the potential upward pressure on prices from the merger. Second, equation (20.4) reflects postmerger costs c1 , which may differ from premerger costs c10 due to merger-related efficiencies. Third, equation (20.4) contains the prices adjusted for merger-induced changes in quality. These second and third elements also appear in equation (20.6) and may partially or fully offset any incentives by the merged firm to raise quality-adjusted prices. To progress in the estimation of the magnitude of a merger’s effects on prices and, ultimately, consumer welfare, it is necessary to further specify the demand function. The FOC in equations (20.3) and (20.4) depend on both demand as a function of (quality-adjusted) prices D i ( H1 , H 2 ) and the partial derivatives of the demand functions with respect to a product’s own (quality-adjusted) price Dii ( H1 , H 2 ) and with respect to the (quality-adjusted) prices of the other products sold by the firm. The own-price sensitivity of demand, and how that changes with movements in the own price (the curvature of a two-dimensional demand curve), influence the rate at which cost changes drive price changes (the pass-through rate), and therefore the estimated price effects of a merger (see Froeb, Tschantz, and Werden 2005; Weyl and Fabinger 2012). Once the demand function is parameterized, the model can be used to find the equilibrium prices that solve the systems of FOCs before and after the merger. Consider a generic demand function:

q = D ( p, θ ) . (20.8)

478   Bryan Keating and Robert D. Willig We first discuss how to parameterize demand functions, and then discuss the applications to merger simulation.

20.2.2.2.  Parameterizing Demand Functions Calibrating demand functions requires two related steps. First, one must make an assumption about the functional form of the demand function D (i ) . Common assumptions include linear, iso-elastic, logit, and the Almost Ideal Demand System (AIDS) (see Crooke et al. 1999). We discuss several of these in greater detail below. Second, having assumed a functional form for the demand function, one must estimate or calibrate the parameters θ characterizing the demand function. For example, a linear demand function takes the form q = a + bp , where the parameters a and b indicate the intercept and slope and fully characterize the demand function. There are two (related) approaches to obtain the parameters of a demand function. First, one can calibrate the parameters of the demand system using aggregate data. Second, one can use more disaggregated data econometrically to estimate the parameters of the demand system.33 Calibrated models often can be implemented using limited and readily available data on prices, quantities, and demand price elasticities (or margins). As such, they can often be implemented at relatively low cost during the early stages of merger investigations. Econometric models typically require greater resources in terms of both time and data requirements and may be more appropriate for in-depth investigations. Below we discuss the calibration and estimation of demand systems typically used by practitioners to simulate mergers.

20.2.2.2.1.  Linear Demand Starting with the calibration approach, a common assumption is that demand functions facing firms are linear:34 q1

 qn

n

= a1 + ∑b1 j p j j =1



. (20.9)

n

= an + ∑bnj p j j =1

Given information on each firm’s margins, one can derive b jj for each firm j from the characterization that profit maximization entails equating the percentage margin between price and marginal cost (the Lerner Index) to the reciprocal of the firm-level

33 

Epstein and Rubinfeld (2001) refer to these as “calibrated-demand simulation models” and “estimated-demand simulation models,” respectively. 34  See, e.g., Davis (2006). However, as discussed above, Jaffe and Weyl (2010) argue that linear demand may not serve as an appropriate baseline assumption.

Unilateral Effects   479

own-price elasticity of demand. To derive cross-price derivatives, it is common to assume diversion between firms is simply related to the share of each firm scaled to account for substitution with the outside good, where the outside good captures the possibility that some potential consumers do not purchase anything.35 The shares within the industry (s ij, where the superscript i indicates a share relative to the inside goods) reflect diversion only between firms but do not account for diversion outside the market. To capture this possibly important element of competition, one can incorporate a scaling factor (θ where s j = θs ij is the share within the overall market including the outside good), calibrated such that the industry elasticity of demand that is implied by the model matches corresponding external estimates commonly found in the literature for many industries.36 This yields the following equation: εij

i≠ j

=

∂qi p j ∂p j qi

=

∂qi ∂p j  ∂q j p j  q j ∂q j  ∂p j q j  qi ∂p j

=

 θsii  q j ε .  i  ji  1 − θs j  qi



(20.10)

Given the elasticities, it is straightforward to calibrate the parameters of the demand system. And given the parameters of the demand system, one can solve the system of

35  See Willig (1991) for a discussion of the conditions under which this assumption holds. See, e.g., Werden and Froeb (1994) and Nevo (2000b) for a discussion of the outside good. 36  The industry elasticity is the percentage change in total industry output due to a proportional increase in the prices of all inside goods in the linear model, and is given by

ε0



= =

γ ∂f ind (γp) f ind ∂γ γ =1

∑ ∑b i

j

ij

pj

f ind

,

Where



f ind

= Σ i fi = Σ i ai + Σ i Σ jbij p j .

480   Bryan Keating and Robert D. Willig equations implied by equations (20.3) and (20.4) to determine the optimal postmerger prices of the market’s individual firms and the newly combined firm.37

20.2.2.2.2.  Logit Demand The logit model belongs to a class known as discrete choice demand models (see McFadden 1980, 1981). The basic premise of discrete choice demand models is that each consumer in a particular time period chooses a single product that maximizes her utility. Each consumer is assumed to have idiosyncratic preferences for each available product. Discrete choice models are particularly suited to settings (e.g., telecommunications systems, automobiles, transportation itineraries, etc.) in which it is reasonable to assume that consumers purchase products in discrete quantities.38 To derive the logit model of demand, let the indirect utility that consumer i would realize from the choice of product j be given by u ij = α j + βp j + ε ij . (20.11)



The parameter α j captures all aspects of product j (e.g., quality) aside from price that have a common evaluation among the consumers. The price coefficient β is common across products and individuals. The idiosyncratic error term, ε ij , is distributed Type I Extreme Value.39 Suppose the firms can be divided naturally into subgroups. Following Cardell (1997), decompose the error term as follows: ε ij = ϕiG + ρ0 µ ig + ρI νij , (20.12)



where ϕ , µ , and ν are each independently drawn from a unique distribution.40 For each individual i, all firms (denoted by G) share the same taste shock ϕ, all firms within each subgroup (denoted by g) share the same taste shock µ , and each firm receives an idiosyncratic taste shock ν .41 The parameter ρ0 lies in the interval 0, 1 and captures the correlation of the error terms among the inside goods. As ρ0 approaches zero, there is no substitution between the inside and outside goods. The parameter ρI lies in the

(

37 

Since this system of equations is typically a nonlinear function of the prices, it is common to use numerical algorithms available in standard statistical software packages to solve the system. 38  The logit model has been applied to a wide variety of settings, including breakfast cereals, air travel, and hospitals. Such applications are accomplished by (implicitly) assuming that the unit of time over which choices are made is sufficiently short such that each consumer selects only one product within the relevant time period. See Werden and Froeb (1994) for an early example of the application of logit demand models to merger simulation. − ( x −µ )/ σ 39  A Type I Extreme Value distribution has a continuous distribution function F ( x ; µ, σ, 0) = e − e . See Cardell (1997) for further details. 40  See Cardell (1997) for further details. 41  Such models are sometimes referred to as “tree extreme value” models. See Cardell (1997).

Unilateral Effects   481

interval (0,ρ0  and captures the correlation of the error terms among the firms in the same group g. As ρI approaches zero, there is no substitution among the subgroups of firms (preferences would be completely determined by ϕiG + ρ0 µ ig ). When ρI equals ρ0, the model corresponds to a nested logit with a single inside-good nest and the outside good in a separate nest. When both ρI and ρ0 equal one, the model corresponds to a simple logit model (i.e., a logit model with no nests). This model generates the following formulation of choice probabilities:42



sj

=

=

 α j + βPj  exp    ρ I

Dg

ρ0

×

D

ρI ρ0 g ρI

Σ g ∈G Dg  ρ0

S j|g × Sg|G × SG

ρI   ρ0   Σ D  g ∈G  g    ×  ρI ρ0 ,   1 +  Σ g ∈G Dg  ρ0   

(20.13)

where D g = ∑ j′ ∈g exp[(α j′ + βp j′ ) / (ρI )]. By design, the choice probabilities in the model will equal the aggregate expected market shares for each product. Choice probabilities relative to the total market (including the outside good) can be computed as43 s0 =



ε0 . (20.14) βp

s j = sij (1 − s0 ) , (20.15)



where ε 0 is the industry elasticity, estimates of which can often be found in the literature. The partial derivatives of the choice probabilities with respect to prices are given by 42  In the logit model, the choice probability is defined relative to the potential size of the market (including the outside good). It is common to normalize α 01, the quality of the outside good, to be zero. Equation (20.13) transforms into the simple logit share equation if one assumes that ρI = ρ0 = 1. In this case, the share equation is given by



sj =

exp(α j + βp j )

1 + ∑ j′∈g exp(α j′ + βp j′ )

.

43  This can be derived by computing the derivative of the industry share function with respect to a proportional increase in the prices of all inside goods. Werden and Froeb (1994) note that it may be preferable to calibrate the demand function using industry elasticities rather than making an explicit assumption about the size of the outside good. This is because the latter approach “links in an undesirable way the aggregate elasticity of demand with the cross elasticities of demand within the nest.” (n. 6) The multilevel nested logit model accounts for this by explicitly allowing for different correlation parameters ρI and ρ0.

482   Bryan Keating and Robert D. Willig ∂s r ( p ) ∂p j

= Γ rj  β   ρI   ρ j  sr  1 −  1 −  s j|g − I (1 − ρ0 ) s j|g s g|G − ρI s j  , if r=j ρ0   ρI   ρ0    ρI  β   ρI  j = − sr   1 −  s j|g − (1 − ρ0 ) s j|g s g|G + ρI s j  if r ≠ j and r , j ∈g ρ0   ρI   ρ0   β ρ   − sr I (1 − ρ0 ) s j|g s g|G + ρI s jj , otherwise    ρI  ρ0

(20.16)

As with the calibrated linear model, the logit model can be calibrated with just a few data elements. If one assumes a simple logit model ( ρ0 = ρI = 1 ), then it can be calibrated using data on the prices and shares of each firm, the margin or own-price elasticity for at least one firm, and an estimate of the industry elasticity (or industry size).44 If estimates of specific diversion ratios between firms are known, these can be used to calibrate the nesting parameters ( ρ0 and ρI ).

20.2.2.2.3.  AIDS Demand The PCAIDS model is based on the Almost Ideal Demand System (AIDS) introduced by Deaton and Muellbauer (1980) (see also Hausman, Leonard, and Zona 1994). The AIDS model is a particularly flexible demand system, which relies on consumer theory (specifically, symmetry and adding-up restrictions) to reduce the number of parameters to be estimated.45 The PCAIDS model places additional restrictive assumptions on the AIDS model in order to further reduce the data requirements (see Epstein and Rubinfeld 2001, 2004). In particular, the PCAIDS model assumes that diversion between firms is related to revenue shares (similar to what we discussed with respect to linear and logit demand models above). Let the demand system be specified as s = a + bln ( p ) , (20.17)



where s is a vector of revenue shares, a is a vector of intercepts, b is a matrix of price coefficients, and ln ( p ) is a vector of log prices. In order to allow for the possibility that diversion is not proportional to share, specify a matrix of nesting parameters (which can be used to solve for the price coefficients as defined in equation (20.20) below): 44 

See Werden and Froeb (1994). Alternatively, if the margin of at least one firm is known, then this can be used in place of the industry elasticity. 45  See Hausman, Leonard, and Zona (1994) for an early example of an AIDS model used in a merger simulation.

Unilateral Effects   483

 1  ωn  Ω =      , (20.18)    ω  1  n



where ω is a nesting parameter that determines the level of substitution. If Ω is specified such that all elements are equal to one, then all products are in the same nest and diversion is proportional to share. Given estimates of the own-price elasticity of one product (which can be derived from the Lerner Index, making use of the assumption that the price is set to maximize the short-run profit earned on that product), and the industry elasticity, the elements of the b matrix can be calibrated as follows:46

(



)

bkk = sk 1 + ε kk − sk ( ε 0 + 1) , (20.19)



 bkk   sk  Ω(k,:) − Ω * (k, k )   bij =   bkk    sk  Ω(k,:) − Ω * (k, k )

(

(

)

  (Ω − diag(Ω)) s if i = j s  

)

  (ss ′).Ω s  

. (20.20)

if i ≠ j

Given parameter estimates b, shares s, and industry elasticity ε 0 , firm-level elasticities are equivalent to



bii   −1 + s + si ( ε 0 + 1)  i εij =  b  ij + s ( ε + 1) 0 j  si

(

(

)

)

if i = j . if i ≠ j

Given shares s and elasticities ε, margins are given by

µ = −diag ( s ) diag ( ε. * Θ ) s, (20.21) −1

−1

where Θ is the ownership matrix. Costs are given by

46  See Epstein and Rubinfeld (2004), equations 5, 12, and 13. Here Ω*(k,k) indicates a vector with Ω(k,k) as the kth element and zeros elsewhere.

484   Bryan Keating and Robert D. Willig c = (1 − µ ) p, (20.22)



where µ is the profit margin. The effects of a merger are simulated by finding the solution vector of price changes that solves the postmerger first-order conditions (see Epstein and Rubinfeld 2001, equation A3):

(

)

(

)

s + diag ε. * Θpost ′diag ( s ) 1 − Γ∆ −1 1 − µ pre  = 0, (20.23)



(

)

(

)

where Γ = diag c post . / c pre and ∆ = diag ppost . / ppre . The estimated price changes can also be used to estimate post-transaction revenue shares using the demand model specified in equation (20.16). Like the calibrated logit model, the PCAIDS model can be calibrated with just a few data elements: revenue shares and two elasticities, the own-price elasticity for one firm and the industry elasticity. Estimates of the values of these data elements are often readily available. Therefore, an additional benefit of calibrated demand models is that, given the relatively light data requirements, they often can be implemented quickly and efficiently. As a result, such models are often used as a preliminary tool in merger analysis to assess potential competitive effects. An additional advantage of calibrated demand models is that it is straightforward to test the sensitivity of the model to alternative assumptions about the input parameters.47 If sufficient data are available, it may be possible econometrically to estimate demand models.48 For example, in certain industries scanner data may be available that track prices and demand across several distinct markets and/or over time (see, e.g., Ackerberg 2001). In other industries, such as airlines, government agencies may publish similar data. In some cases, detailed transaction-level data may be available.49 In other cases, data may be aggregated to the product level. If available, these data can allow one to develop econometric estimates of demand elasticities. These data can also allow one to assess the trade-off between price and nonprice (quality) characteristics.50 Since econometric estimation of demand models has been discussed extensively elsewhere, we do not discuss it in detail here (see, e.g., Ackerberg et al. 2007; Reiss and Wolak 2007). 47  It is also common practice to test the sensitivity of the analysis to alternative assumptions about the nature of the demand system. If the expected effects are likely to be small in magnitude, then a linear demand system provides a good approximation for any demand system. However, in cases where the magnitude of the effects is likely to be bigger, it is important to test how sensitive the model is to the nature of demand. See Jaffe and Weyl (2010), discussing potential shortcomings of linear demand systems. 48  See Ackerberg et al. (2007) and Reiss and Wolak (2007) for comprehensive surveys of econometric modeling in the context of industrial organization. 49  In such cases, special econometric methods have been developed to make use of available micro data. (See, e.g., Berry, Levinsohn, and Pakes 2004.) 50  For an example, see Israel et al. (2012).

Unilateral Effects   485

However, we note that there are several challenges to developing robust econometric demand models that appropriately capture the relevant characteristics of consumer choice.51 First, when customers may choose multiple units of a good or customer decisions involve continuous choice over multiple goods, models of discrete choice (e.g., logit demand models) may not match reality very well (see, e.g., Dubin and McFadden 1984; Hendel 1999). Second, demand for many types of products (e.g., durable, storable, or experience goods) is inherently dynamic. In other words, current demand will depend on past events (e.g., previous purchases) as well as expectations about future events (e.g., the likelihood of future sales). Estimating dynamic demand models can be computationally complex, though models have been developed that make simplifying assumptions (see, e.g., Bajari, Benkard, and Levin 2007). Third, static demand models may be reliant on assumptions made for computational tractability, which nonetheless have important economic implications. One such example is the assumption in logit models that each individual has some i.i.d. idiosyncratic taste for each product in the choice set. This implies that every product will have some positive probability of being chosen no matter how unattractive its price and nonprice characteristics because there is some positive probability that an individual will have a large idiosyncratic value for the product. This also implies that each consumer’s utility will grow as new products are added to the choice set even if these new products are not especially valuable.52 Similarly, certain assumptions about the shape of the demand function may place undue restrictions on the substitution patterns to be estimated.53 Fourth, because observed prices and quantities are assumed to represent equilibrium outcomes that depend on both supply and demand functions, it can be challenging to distinguish between demand and supply effects. A common approach is to find “instrumental variables” that are correlated with one or the other. For example, common instruments in demand models are variables that shift the cost functions and are excluded from the demand equation. Ultimately, regulatory agencies seek to assess the impact of mergers on economic surplus (either consumer surplus or the aggregation of consumer and producer surplus). The logit model is particularly attractive because analytical representations of changes in consumer surplus as a function of pre- and postmerger prices and qualities can be derived (see Small and Rosen 1981). This is most easily illustrated in the context of a simple logit model. McFadden (1981) and Small and Rosen (1981) show that the compensating variation for consumer i is

51 

See Ackerberg et al. (2007) for a more detailed discussion of these issues. Berry and Pakes (2005) consider a model of demand that does not rely on idiosyncratic error terms. 53  For example, the simple logit demand model implies that a consumer’s preferences for products other than the one chosen do not depend on the product chosen. This so-called “independence of irrelevant alternatives” problem implies very restrictive substitution patterns. In particular, in the simple logit model, diversion ratios are proportional to shares (i.e., they do not take into account the relative characteristics of products other than how those characteristics manifest themselves in market shares). Various models have been proposed to relax this assumption. See, e.g., Berry (1994) and Berry, Levinsohn, and Pakes (1995). 52 

486   Bryan Keating and Robert D. Willig



CVi = −

ln

(∑ exp (δ )) − ln(∑ exp (δ )) , (20.24) j

post j

j

pre j

β

where δtj measures the mean utility of product j in period t and β is the estimated marginal utility of income.54

20.2.2.3.  Use of Merger Simulation Merger simulation has been used in several high-profile cases in both the United States and Europe.55 Nonetheless, the economics literature suggests that caution in the use of merger simulation is warranted (see, e.g., Walker 2005; Peters 2006; Budzinski and Ruhmer 2009). In particular, the literature suggests at least three reasons for caution. • Merger simulation models may be sensitive to assumptions that are difficult to verify (e.g., the precise nature of the demand function facing firms). • Merger simulation models may lead to imprecise estimates (if the available data are insufficient to estimate the relevant parameters with precision) or inaccurate predictions (e.g., because primitive modeling assumptions such as the model of firm conduct are incorrect). • Merger simulation may not encompass all of the relevant considerations related to a specific transaction (e.g., merger simulation cannot easily assess the likelihood of entry or repositioning). Although research assessing the accuracy of merger simulation is limited, the papers that have taken this on suggest caution is warranted. For example, Peters (2006) applies merger simulation techniques to several airline mergers in the United States in the 1980s. He then compares the predictions of the model (based only on premerger data) to retrospective assessments of the effect of mergers on price. He concludes that “standard simulation methods, which measure the effect of the change in ownership on unilateral pricing incentives, do not generally provide an accurate forecast.” Moreover, in some cases the merger simulation model overpredicted price increases and in other cases it under-predicted price increases. Similarly, Weinberg and Hosken (2013) compare the predictions of merger simulation models to estimates of actual price effects from two mergers and find that for one merger the simulation overpredicts price effects (in fact 54  Compensating variation measures the amount of money an individual would need to receive in order to offset merger-induced changes in price and quality, and the equivalent variation is the amount of money that would have the same impact on the consumer’s utility as the merger-induced changes in price and quality. These are natural measures of the consumer welfare effects of a merger. Note that in this expression of the model, there are no income effects, so that the absolute values of the compensating and equivalent variations are equal. See Nevo (2000a) and (2000b). 55  See Budzinski and Ruhmer (2009) for a summary of the use of merger simulation in merger evaluation. More recently, an expert employed by the US Department of Justice used merger simulation in United States of America v. H&R Block, Inc., et al.

Unilateral Effects   487

the simulations predicted price effects where none were actually observed) while for the other merger the simulations understate the actual price effects.56 Ashenfelter and coauthors (2011) conclude that “[n]‌otwithstanding the growing popularity of merger simulation, little is known about how well these merger simulation techniques predict actual merger outcomes.” At a minimum, this suggests that it is important to treat with caution the point estimates of any predicted price increases generated by merger simulation modeling. It further suggests that a premium should be placed on exploring the sensitivity of the estimates to alternative assumptions. In addition, to the extent that the relevant data are available, it suggests the importance of pairing merger simulation modeling with empirical analysis of the effects of concentration on price and welfare. We discuss this in the next section.

20.2.3.  Empirical Evidence The economics literature makes a distinction between structural modeling and nonstructural (or descriptive) econometric modeling (see, e.g., Reiss and Wolak 2007). Structural models include equations that are representations of more than one of the forces that underlie the system’s equilibrium, and as such rest on explicit assumptions about institutional and economic features of the market. For example, structural demand models make explicit assumptions about the nature of consumer preferences. Similarly, structural supply models make explicit assumptions about the nature of firm conduct (e.g., firms may be assumed to set prices to maximize static profits on a market-by-market basis). The specifications of the structural equations embody viewpoints on the causal relationships that drive the determination of market outcomes from market characteristics. Once a system of structural equations is fit to the data, its estimated parameters allow the researcher to simulate counterfactual experiments or to assess welfare effects of alterations in the conditions and characteristics that affect market outcomes. Structural modeling also might be used to endogenize economic decisions that are assumed to be exogenous in a nonstructural modeling framework. Merger simulation models are an example of structural modeling. The structure of the model is what allows a researcher to make predictions about postmerger conditions that cannot be observed. Structural modeling tends to be challenging in its requirements that the most important individual forces be identified and separately plausibly specified. While the complexity of the modeling often requires that simplifying assumptions be adopted for computational tractability, in many cases the reasonableness of the assumptions may be assessed using the results of the corresponding nonstructural analysis (i.e., the “reduced form” of the structural model). Nonstructural empirical models 56  See also Weinberg (2011), which finds that merger simulation underpredicts the actual price effects of a consumer products merger.

488   Bryan Keating and Robert D. Willig typically regress measures of market outcomes (e.g., prices or output) on measures of competitive market structure (e.g., concentration measures or competitor counts), along with other control variables. Market structure is taken to be exogenous in these equations (i.e., determined outside of the system that the analysis seeks to represent). In other words, variations across time or across markets in these variables are treated as a series of “experiments,” which allow the researcher to estimate how price (or output) is a function of the intensity of competition in the market. In principle, the estimated parameters or coefficients of that function can then be applied to other situations with varying levels of competition to make predictions about the resulting prices and output. This approach relies little on economic theory other than to guide the specification of the empirical model in terms of the factors for which to control. Nonetheless, it provides a great deal of information on the joint distribution of market outcomes (prices or output) and market characteristics (e.g., number of competitors, market concentration, etc.) and therefore might be useful for the assessment of the effects of mergers. Thus, in addition to or as an alternative to UPP and merger simulation, econometric evidence is commonly used to assess the likelihood of unilateral effects. Estimation of nonstructural econometric models depends upon variation in observed data on market outcomes and market structure. Such variation typically occurs along one of two dimensions: (1) across time (within a market); or (2) across markets (within a time period). We briefly discuss each.

20.2.3.1.  Intertemporal Variation The 2010 Horizontal Merger Guidelines discuss the context in which the agencies use empirical analysis of historical data to assess prospective mergers:57 The Agencies look for historical events, or “natural experiments,” that are informative regarding the competitive effects of the merger. For example, the Agencies may examine the impact of recent mergers, entry, expansion, or exit in the relevant market. Effects of analogous events in similar markets may also be informative.

The effect of historical events on the market outcomes of interest (e.g., prices and quantities) can provide powerful evidence on the expected effects of prospective mergers. The analysis of such events naturally lends itself to econometric analysis, given appropriate data. However, care must be taken to ensure that correct inferences are drawn from the data. The “experiments” available in the historical data may or may not resemble the merger under consideration. As the 2010 Horizontal Merger Guidelines note, the agencies may look at several different types of events in addition to mergers, including entry, 57 

DOJ and FTC (2010), § 2.1.2. Other competition authorities rely on similar evidence. (See, e.g., European Union 2004, § 29; Competition Commission and Office of Fair Trading 2010, § 5.4.9; New Zealand Commerce Commission 2003, § 3.2; Competition Bureau of Canada 2011; and Brazil Ministry of Finance and Ministry of Justice 2001, ¶ 45.)

Unilateral Effects   489

expansion, and exit. While entry, expansion, and exit may bear some resemblance to mergers to the extent that they lead to significant changes in the competitive structure of a market (by, for example, increasing or decreasing the number of independent competitors in a market or leading to changes in market shares and diversion ratios), such events may also be distinguished in important ways from mergers.58 For example, entry and exit may lead to the addition or loss of specific brands. In differentiated product markets, changes in the set of products available to consumers may have very different effects on pricing incentives and consumer welfare than changes in ownership control over a fixed set of products.59 To see this, return to equations (20.3) and (20.4). As equation (20.3) demonstrates, the pricing incentives faced by an independent firm depend on the own-price partial derivative of demand Dii (i) . The entry or exit of a competing product might impact the own-price elasticity depending on the extent to which the competing product is a close substitute. But equation (20.4) shows that the pricing incentives of the merging firm also depend on the cross-price partial derivative of demand Dij (i) . This effect has no analog in the entry, expansion, or exit scenarios. Analogously, in homogeneous goods industries, which we discuss in greater detail below, entry, expansion, or exit of a firm may lead to significant changes in the overall capacity available in a market. Changes in total market capacity can have different effects on welfare than changes in ownership over a fixed set of capacity assets. To the extent that variation in concentration is driven by previous mergers in the same industry, several issues arise. First, it is likely that few events may be available to analyze so that the standard errors associated with the effects of those events may be large.60 Second, observed mergers may suffer from sample selection issues. Mergers that are most likely adversely to affect prices are those that are most likely to be challenged and/or blocked. Conversely, observed mergers may be unlikely to exhibit large effects on prices precisely because those mergers were deemed unlikely to result in significant adverse effects.61 Third, each merger takes place in the context of specific market and regulatory conditions. The conditions affecting observed mergers may differ dramatically

58  Several empirical papers develop estimation strategies that attempt to disentangle entry effects from concentration effects. For example, a series of articles starting with Mazzeo (2002a and 2002b) use the structural entry model developed by Bresnahan and Reiss (1991) to separately identify the price effects associated with product differentiation (and entry) and concentration. Doane et al. (2012) make use of variation in the number of brands and the number of owners competing in rental car markets to disentangle the two effects. See also Werden and Froeb (2008) for a general discussion of the issue. 59  Of course mergers also can have implications for product variety. For analysis of the effects of new products, see Hausman (1997) and Petrin (2002). 60  For example, see Town and Vogt (2006, 2008) for a summary of empirical analysis of hospital mergers. 61  For this reason, retrospective studies have attempted to examine mergers whose regulatory approval was “close.” For example, Tenn (2011) examines the price effects of a merger of the Sutter and Summit hospital systems in California. The FTC did not challenge that merger, but the California attorney general unsuccessfully tried to block it. For this reason, the author deemed this to be a “marginal” case. The author found that prices at one of the merging hospitals increased significantly after the transaction, though the precise causes of the price increase are unclear. (See Tenn 2011.)

490   Bryan Keating and Robert D. Willig from the conditions facing prospective mergers. If those differences are not or cannot be appropriately controlled for, incorrect inferences may be drawn. It is also important to determine and control for factors that lead to changes in market structure over time, whether by merger, entry, expansion, or exit. For example, these events may be caused by demand or cost shocks, which occur simultaneously. It may not be possible to distinguish econometrically the effects of the event from the effects of exogenous shifts in the demand or cost functions. For example, suppose that a downward shift in demand for air travel on a particular route caused a carrier to remove service from a particular route. Independent of the exit, the downward shift in the demand function would be expected to result in lower prices. Therefore, an econometric analysis based on such exit events may incorrectly suggest that consolidation has no (or even negative) effects on price (see also Carlton 2011).

20.2.3.2.  Cross-Market Variation The study of specific events and the effects on prices exploits variation over time within a particular market. Another source of variation that is potentially relevant comes from cross-market comparisons. For example, the 2010 Horizontal Merger Guidelines discusses empirical evidence based on cross-sectional variation (DOJ and FTC 2010, § 2.1.2): The Agencies also look for reliable evidence based on variations among similar markets. For example, if the merging firms compete in some locales but not others, comparisons of prices charged in regions where they do and do not compete may be informative regarding post-merger prices. In some cases, however, prices are set on such a broad geographic basis that such comparisons are not informative. The Agencies also may examine how prices in similar markets vary with the number of significant competitors in those markets.

Cross-market comparisons rely on analyzing variation in prices and concentration across markets. For example, Brueckner, Lee, and Singer (2013) analyze the effect of concentration in airline markets by examining variation in prices and the number of carriers across thousands of routes. But such studies can be subject to similar endogeneity concerns. Two types of issues arise. First, to the extent that concentration is measured based on functions of market shares (e.g., HHIs), these market shares depend, at least in part, on prices and are therefore endogenously determined.62 Alternatively, concentration may be measured by factors more directly related to the structure of the market such as the number of independent competitors. However, even these types of measures can cause problems. For example, if markets are not observed in long-run equilibrium, then issues similar to those discussed above arise.

62  See, e.g., Peters (2006), who uses the number of independent competitors as an instrument for HHI.

Unilateral Effects   491

For example, rapid growth in demand may lead to both higher prices and more firms competing. Even if markets are observed in equilibrium, the factors determining the equilibria in each market may vary substantially. If the econometric specification does not appropriately control for these factors, then omitted variable bias may be a serious concern. Despite these challenges, nonstructural empirical analysis can often provide insight into the relationship between nominal prices and concentration in a variety of industries. Such evidence may complement or substitute for UPP or merger simulation analysis depending on the setting. It is therefore important clearly to identify the sources of variation that are driving the empirical estimates and assess the extent to which those sources of variation are reasonable proxies for the transaction under consideration.

20.3.  Homogeneous Products Mergers sometimes occur in markets, such as those for commodities, in which products are relatively undifferentiated. Indeed, much of the early history of merger enforcement in the United States focused on homogeneous products markets.63 In such markets, market power is more likely to be manifested by strategic decisions of the merging firm to reduce output (or divert capacity to other uses) rather than to increase price. As the 2010 Horizontal Merger Guidelines recognize, such an output-suppression strategy is more likely to be profitable the larger is the merged firm’s output and the lower are the supply response of rivals and the demand response of consumers.64 This is so because the possibly profitable effect of an output suppression strategy on the merged firm is to increase the price on the merged firm’s inframarginal sales of its products, while sacrificing sales of marginal units. The larger the firm, the greater the number of inframarginal units, and the weaker the supply and demand responses, the less elastic is the residual demand function facing the merging firm and therefore the more price elevation and profit boost are caused by a given output suppression. Both Cournot (quantity-setting) and Bertrand (price-setting) models can be used to assess competitive effects in homogeneous goods industries. We consider each in turn. 63 

See Willig (1991); Shapiro (2010). For example, the 1982 Horizontal Merger Guidelines consider product differentiation in the context of coordinated effects “only in relatively extreme cases,” despite its recognition that “[p]‌roduct variation is arguably relevant in all cases” (Department of Justice (1982), § III.C.1.(a)). 64  DOJ and FTC (2010), § 6.3. See also 2010 Horizontal Assessment Guidelines, §§ 5.3–5.4. See also Australian Competition and Consumer Commission (2008), § 5.17; Brazil Ministry of Finance and Ministry of Justice (2001) ¶ 61; Competition Bureau of Canada (2011) § 6.19; European Union (2004), §§ 31, 34; Japan Fair Trade Commission (2011), p. 18; New Zealand Commerce Commission (2003) § 5.4; Competition Commission of Singapore (2007) § 6.4.

492   Bryan Keating and Robert D. Willig

20.3.1.  Cournot Model Markets in which products are relatively undifferentiated lend themselves to modeling firms as Cournot competitors.65 Under the Cournot model, firms are assumed independently to choose output levels in order to maximize profit. The Cournot model with homogeneous products underlies many common viewpoints about the linkage between prices and market power (see Tirole 1988; Shapiro 1989; and Willig 1991). Indeed, the basic intuition that concentration and market power are positively correlated, a key concept in the 2010 Horizontal Merger Guidelines, can be derived from the Cournot model.66 In particular, the Horizontal Merger Guidelines (and related guidelines around the world) have long featured thresholds of levels of the Herfindahl-Hirschman Index (HHI) of concentration to assess the markets in which competitive effects are most likely.67 It can be shown that in a Cournot model the HHI is positively related to the market-share weighted average price-cost margin.68 In the basic Cournot model, it can be shown that firms’ market shares are inversely related to their marginal costs.69 Cournot models can and have been used to simulate mergers. If one assumes that the merging firm will obtain the lower of the two merging partners’ marginal costs, then absent capacity constraints the merger simulation in the Cournot model with constant marginal costs is equivalent to just eliminating the higher cost firm from the market. Various commentators have questioned the applicability of

65  See Willig (1991): “Little doubt should remain that Cournot models can provide a productive setting for derivation of rich results concerning unilateral competitive effects of horizontal mergers. A remaining question, however, is the applicability of these models to the analysis of real mergers. At their foundational level, these models are based on a particular view of business behavior—that firms unilaterally decide on their levels of output to be sold and that the level of price is that which clears the market of the total output produced. This view of business behavior may very well be a sufficiently apt description of unilateral conduct in markets for commodity products where production capacity and production runs play important roles” (298–99). 66  This fact is explicitly recognized in the 2010 Horizontal Assessment Guidelines, which note that “[t]‌hese [HHI] thresholds may be most informative for mergers in a market where the product is undifferentiated and where competition between firms involves firms choosing what volume to supply to the market. In other cases the significance of these thresholds will be less.” (DOJ and FTC 2010, § 5.3.5). See also Australian Competition and Consumer Commission (2008), § 7.8; European Union (2004), § 27; Japan Fair Trade Commission (2011), p. 21; New Zealand Commerce Commission (2003), § 8; Competition Commission of Singapore (2007), § 5.13; Competition Commission and Office of Fair Trading (2010), § 5.3.4. 67  Department of Justice (1982), § III.A; Department of Justice (1984), § 3.1; DOJ and FTC (1992), § 1.5; DOJ and FTC (2010), § 5.3; Competition Commission and Office of Fair Trading (2010), § 5.3.5; Australian Competition and Consumer Commission (2008), §§ 7.14–7.16; Competition Bureau of Canada (2011) § 5.13, n. 31; European Union (2004), §§ 19–21; Japan Fair Trade Commission (2011), pp. 20–21. 68  For a derivation, see Ordover, Sykes, and Willig (1982) or Carlton and Perloff (2004), appendix 8A. 69  The FOC are given by (∂p) / (∂q)q + ( p − c ) = 0, which implies that quantity q and cost c are inversely related as long as the market demand curve is downward-sloping.

Unilateral Effects   493

the Cournot assumption to many industry settings (see, e.g., Willig 1991; Werden and Froeb 2008). Salant, Switzer, and Reynolds (1983) showed that the Cournot model can lead to unintuitive results. In particular, they showed under a certain set of assumptions (including constant marginal costs) that a merger of Cournot competitors can actually reduce the profits of the merging firms (absent efficiencies) because the reaction by other firms in equilibrium offsets part of the output reduction by the merging firms. Subsequent research showed that, under a range of plausible circumstances, anticompetitive mergers in Cournot settings could be profitable for the merging parties. For example, Perry and Porter (1985) showed that if nonmerging parties face upward-sloping marginal cost curves, their incentives to respond will be dampened and profitable mergers are possible (see also Willig 1991). Farrell and Shapiro (1990a and 1990b) derive general conditions under which mergers in Cournot industries can have adverse effects absent efficiencies. While the UPP analysis in section 20.2.1 above is based on a Bertrand model of competition, it can also be related to Cournot models of homogeneous product competition by modifying the diversion ratio (see Farrell and Shapiro 2010; Willig 2011; Jaffe and Weyl 2012). Willig (2011) introduces a Gross Upward Market Power Pressure Index (GUMPPI), which generalizes the UPP to allow for increases in market power due to change in the merging firms’ prices or output or some combination of the two. Willig (2011) parameterizes the strategic decisions of Firm 1 (or 2) by the parameter s, such that we have P1 ( s ) , Q1 ( s ) , and (∂Π1 ) / (∂s) = 0 (when evaluated at premerger levels of price, output, marginal costs and quality). Then,



∂Π m ∂s

∂H 1 ∂Π 2 ∂Q 1 + + Q1 ∂s ∂s ∂s 2 1 ∂Π1 Q ∂ ∂ Π + v1 + c10 − c1 + . ∂s ∂s ∂s

= (H1 + v1 − c1 ) =

(

)

(20.25)

Evaluated at premerger levels, this implies that



∂Π 2 v1 + c10 − c1 ∂Π ∂s  0 iff  . (20.26) 1 ∂s P10 0 ∂Q − P1 ∂s m

(

)

The GUMPPI[(∂Π 2 ) / (∂s)]/[− P10 (∂Q1 ) / (∂s)]—is the change in profits of Firm 2 due to the exercise of market power by Firm 1 relative to the value of sales lost by Firm 1 due to that exercise of market power. Thus it is analogous to the GUPPI discussed above, generalized to allow for changes in price and/or quantity.

494   Bryan Keating and Robert D. Willig

20.3.2.  Bertrand-Edgeworth Model If products are homogeneous, marginal costs are constant, capacity constraints are absent, and firms compete on prices, the standard logic of the Bertrand model suggests that prices will be competed down to marginal cost (or more generally to slightly below the cost of the second lowest cost firm if firms’ costs are asymmetric). This is because a firm can capture the entire market by slightly undercutting its rivals’ prices, and, therefore, so long as at least two firms are competing they will compete prices down to marginal costs. This is sometimes referred to as the “Bertrand Paradox” since, if firms earn no margins they will be unable to cover fixed costs and will have no incentive to enter the market and compete. In such a model, absent efficiencies, consolidation would have no effect as long as at least two firms remain postmerger, with the same second-lowest level of marginal costs as there was premerger.70 Introducing capacity constraints into the model changes the equilibrium outcome. A model of homogeneous product pricing competition combined with capacity constraints is sometimes referred to as a Bertrand-Edgeworth model (see Edgeworth 1925). To see this, suppose that there are two firms and Firm 1’s capacity K is less than market demand when price equals marginal cost. If both firms price at marginal cost, neither makes positive profits. However, if Firm 2 raises price above marginal cost, Firm 1 does not have sufficient capacity to satisfy market demand. This means that Firm 2 can earn positive margins on the residual demand. Unlike the models we have discussed previously, however, the Bertrand-Edgeworth model typically does not have a pure-strategy equilibrium. To see this, return to the previous example and consider what happens if Firm 2 sets the monopoly price over the residual demand (given that Firm 1 produces up to capacity). In this case, Firm 1 has an incentive to price almost as high as Firm 2 because it can do so without losing any sales.71 But if Firm 1 raises its price to just below that of Firm 2, then Firm 2 has an incentive to slightly undercut Firm 1’s prices. Unless the capacity constraints are binding such that firms are already producing at capacity or capacity constraints are never binding even if firms price at marginal cost, then the only equilibria of the model will be mixed strategy equilibria. Despite the lack of pure-strategy equilibria, the Bertrand-Edgeworth model may provide an appropriate way to model consolidation in industries characterized by homogeneous products, price-setting competition, and fixed capacities. In this model, a firm’s market power is inversely related to its rivals’ capacity.72 This is because the residual

70 

This is not thought to be a reasonable model of competition in many markets in which we observe mergers occurring since we typically observe firms earning positive margins. 71  Assume, for the sake of this example, that Firm 2’s capacity is greater than or equal to Firm 1’s capacity. 72  These findings may be reversed in infinitely repeated versions of the Bertrand-Edgeworth model. We do not address this possibility here as it is more closely related to the logic of coordinated effects. For a fuller discussion, see Compte, Jenny, and Rey (2002).

Unilateral Effects   495

demand facing a firm whose rivals have produced up to capacity will be larger when the collective capacity of the firm’s rivals is smaller. A merger that results in the consolidation of capacity therefore reduces the total capacity of the merging firm’s rivals and increases the merged firm’s market power.73 Kreps and Scheinkman (1983) derive an important link between Cournot and Bertrand-Edgeworth models. They develop a two-stage model in which firms first precommit to capacity and then play a pricing game constrained by the capacity chosen in the first stage. In other words, the second-stage subgame is modeled as Bertrand-Edgeworth competition. Kreps and Scheinkman (1983) show that the unique equilibrium outcome of this two-stage game is identical to the Cournot outcome. Subsequent research by Davidson and Deneckere (1986) demonstrates that the Kreps and Scheinkman (1983) result is sensitive to the chosen rationing rule and that “the equilibrium tends to be more competitive than the Cournot model would predict” (Davidson and Deneckere 1986, 404).74

20.3.3.  Dominant Firm Model Another model sometimes employed to evaluate mergers is the dominant firm model in which one firm is a price setter and all other firms are price takers (i.e., they will passively supply the market so long as the price exceeds their marginal costs).75 In this model, the dominant firm faces a residual demand curve that is the horizontal difference between the market demand curve and the fringe supply curve (Carlton 2004, ch. 4). Absent efficiencies, any merger that bolsters the position of the dominant firm (e.g., a merger of the dominant firm and a firm that was formerly a member of the competitive fringe), potentially results in adverse competitive effects. This is because the dominant firm’s residual supply curve shifts outward, resulting in greater incentive for the dominant firm to increase prices.

73  We are aware of at least one recent case in which the European Commission used a static Bertrand-Edgeworth model to evaluate the unilateral effects of a proposed merger. 74 More generally, rationing rules are important to any model of homogeneous price competition with capacity constraints. This is because when firms produce up to capacity, it is necessary to make an assumption about which customers are served. For example, Allen and Hellwig (1986) show that while, under proportional rationing, the equilibrium converges in probability to the competitive outcome as the number of firms increases, the support for the mixed strategy equilibrium does not. This is because the firm residual demand curve facing the nonconstrained firm resembles the full demand curve and therefore one firm will always set the monopoly price with some positive probability in the mixed strategy equilibrium. In contrast, Vives (1986) shows that under efficient rationing, the support of the mixed strategy equilibrium converges to the competitive equilibrium as the number of firms increases. 75  A dominant firm model is discussed in the 1984 Horizontal Merger Guidelines under the “leading firm proviso” (Department of Justice (1984), § III.A.2). This proviso does not appear in subsequent horizontal merger guidelines, which are generally more focused on mergers in differentiated products industries. The language of “dominance” is much more prevalent in the European Commission Horizontal Merger Guidelines (European Commission 2004).

496   Bryan Keating and Robert D. Willig

20.4.  Markets with Auctions or Negotiated Prices In contrast to markets for consumer goods, where prices are typically publicly stated and nonnegotiable, prices in intermediate goods markets are often set via auction or bilateral negotiation between buyers and sellers. While the exact mechanisms through which unilateral effects can manifest themselves and the models used to assess those effects vary depending upon the manner in which firms set prices, the basic logic is analogous to that which we developed above in the context of stated-price markets. In particular, a merging firm will internalize the extent to which its merging partner is a close substitute, and this internalization can create incentives to raise prices, absent efficiencies. The 2010 Horizontal Merger Guidelines discuss competitive effects in negotiated price markets as follows (§ 6.2): A merger between two competing sellers prevents buyers from playing those sellers off against each other in negotiations. This alone can significantly enhance the ability and incentive of the merged entity to obtain a result more favorable to it, and less favorable to the buyer, than the merging firms would have offered separately absent the merger. The Agencies analyze unilateral effects of this type using similar approaches to those described in Section 6.1.

In this section, we first discuss auction models and the similarities and differences with the differentiated product models discussed above. We then discuss bargaining models.

20.4.1.  Auction Models Auctions are interactions between buyers and sellers that follow specific rules for how and to whom goods are sold. Typically, multiple sellers compete on price (and potentially other dimensions) to sell to a single buyer. The seller with the lowest bid wins the right to sell to the buyer. These types of auctions are typically called procurement auctions and take place in a variety of intermediate goods industries. For example, governments and firms often use procurement auctions to obtain goods and services from private contractors. There are many similarities between the economic modeling of auctions and the modeling of competition in price-setting markets with differentiated products.76 To see these similarities, consider the objective function that bidder i in a first-price sealed bid procurement auction with independent private values (IPV) seeks to maximize: 76 

For a fuller description of the relationship between differentiated product models and auction models, see Einav and Nevo (2006) and Einav and Levin (2010). For an overview of competition policy in bidding markets, see Klemperer (2008).

Unilateral Effects   497

(

)

max (bi − ci ) Pr bi ≤ b j ∀j ≠ i , (20.27)



bi

where bi is bidder i ’s bid. In equilibrium, bids bi must solve the following FOC:



(b − c ) i

i

(

∂Pr bi ≤ b j ∀j ≠ i ∂bi

) + Pr b ≤ b ∀j ≠ i = 0. (20.28) ( ) i

j

It is apparent that equations (20.27) and (20.28) are closely related to equations (20.3) and (20.4). In the optimization problem facing price-setting firms (as in equation (20.3)), a firm considering a price increase must trade off increased margins on inframarginal sales against lost marginal sales. The extent to which it will be profitable to increase prices depends on how many marginal sales will be lost for a given price increase, which is parameterized by the demand function D (i ) and will depend among other factors on the availability of substitute products. In the optimization problem facing firms bidding to supply a good or service in a procurement auction (as in equation (20.27)), a firm considering whether to increase its bid must trade off increased margins in the event that it wins the auction against the reduced probability that it will enter the lowest bid and therefore win the auction.77 The extent to which it will be profitable to increase the bid depends on how the probability of winning the auction decreases with an increased bid, which will depend among other factors on the (unknown) distributions of the costs of competing bidders. Indeed, Einav (2003) points out that “one way of thinking about IPV auctions is as an incomplete information version of a Bertrand price competition in homogeneous product markets.” The primary difference between the literature on differentiated products and the literature on auctions is the body of empirical techniques that are used to estimate the models. The differentiated products literature has focused on parametric estimation of demand systems, while allowing for some asymmetry across competing products, whereas the auction literature typically makes strong assumptions about symmetry across firms, but focuses on nonparametric identification of cost distributions for participating bidders.78 This difference in focus is due primarily to the contrasts in availability of data. For demand models, the researcher often observes quantity and pricing data and can use them as described in section 20.2.2. The analogous indicators in auction models, the probabilities of winning, cannot be observed. As a result, to make auction models empirically tractable, additional assumptions on the symmetry of competing firms are typically imposed.79 77  It is typically assumed that costs are private information. In such cases, bidders cannot predict with certainty the bid required to win an auction. As a result, bidders must trade off higher prices and the expected quantity sold. (See Klemperer 2008.) 78  See Einav (2003) for a discussion. See also Klemperer (2008). 79  See Athey and Haile (2006) for a summary of the empirical auction literature.

498   Bryan Keating and Robert D. Willig Nonetheless, the effects of a merger in the auction context are analogous to those that we have established in stated-price markets. In a first-price auction such as discussed here, the merged firm will internalize the probability that in some instances the second-best bid would have been entered by the merging partner, and shade its bid accordingly. In a second-price procurement auction, the winning firm is paid the bid of the second-lowest bidder. In such an auction it is optimal for each firm to bid its cost. Therefore, a merger will not impact the merging firm’s bid. However, it will affect how much money the buyer must pay since the merger eliminates a bidder, which would have been the highest or second-highest bidder with some positive probability.80 The economics literature has developed techniques to simulate mergers in a variety of different auction settings. In contrast to stated price markets, the analyst must calibrate cost or bidding distributions rather than demand and cost functions. Nonetheless, the basic approach is very similar to the methods used to simulate mergers in differentiated product markets. The analyst must first choose a particular auction model and then calibrate the parameters of the model to observed (premerger) data. Given the calibration of the model, the analyst can simulate the merger by taking into account the effect of coordinated bidding between multiple firms and assess the merger impact by comparing pre- and postmerger bids.81 Auction-based merger simulation has been used in at least one court case. In the 2004 litigation attempting to stop the proposed merger between Oracle and People Soft, the expert for the US government presented a merger simulation model based on an oral procurement auction.82 In contrast, for their analysis of the same combination, experts for the European Commission considered a merger simulation model based on a sealed-bid auction, but the model was eventually abandoned by the European Commission.83

20.4.2.  Bargaining Models Prices charged by suppliers of intermediate goods are often set in bilateral negotiations between buyer and seller. It is typically assumed that negotiations between buyers and sellers can be characterized by the Nash bargaining solution in which the parties accept the price that splits the joint surplus generated by the agreement 80  Tschantz, Crooke, and Froeb (2000) find similar competitive effects in first- and second-price auctions. 81  Waehrer (1999) analyzes mergers with asymmetric first- and second-price auctions. Thomas (2004), Dalkir, Logan, and Masson (2000), and Tschantz, Crooke, and Froeb (2000) analyze asymmetric first-price auctions. Froeb, Tschantz, and Crooke (1998), Brannman and Froeb (2000), and Waehrer and Perry (2003) analyze asymmetric second-price auctions. Mares (2000a and 2000b) analyzes mergers with asymmetric common value auctions. 82  The court ultimately rejected the model, though for reasons only tangentially related to the model itself. (See Werden and Froeb 2008; Baker and Reitman 2012.) 83  See Budzinskie and Ruhmer (2009) for a discussion.

Unilateral Effects   499

between them (see Binmore, Rubinstein and Wolinsky 1986). In the context of this bargaining framework, the disagreement points (each party’s profits absent an agreement) determine the joint surplus to be bargained over. Merger analysis focuses on how a potential merger would change the disagreement points and therefore the equilibrium prices.84 As is standard in any unilateral effects analysis of differentiated products, the degree to which the offerings of the merging parties are substitutes for each other plays a key role in determining the magnitude of the competitive effects of the merger. This can be seen most clearly by way of example in the hospital industry, where bargaining models are often deployed to model merger effects.85 Hospitals and managed care organizations (MCOs) negotiate the rates at which MCOs reimburse hospitals for services provided to MCO enrollees.86 It is typically assumed that a hospital’s objective in the negotiation is to maximize profits, while the objective of the MCO is to maximize the surplus of its enrollees, which is a function of the quality of the hospital network offered by the MCO and the payments that enrollees must make. A hospital derives value from inclusion in provider networks because MCO enrollees are more likely to choose in-network hospitals than out-of-network hospitals. An MCO derives value from including a hospital in its network because, all else equal, enrollees prefer to be able to choose from more in-network hospitals. Hospitals and MCOs bargain over this joint surplus.87 A merger changes the incentives of a hospital to reach an agreement. This is because a merging hospital that fails to reach an agreement with an MCO will internalize the fact that some of the patients who otherwise would have chosen that hospital (and provided some profit margin) would instead choose the other merging hospital. On a stand-alone basis, these lost margins would lower the disagreement profits considered by the hospital. With a merger, the internalization of the 84  A merger could also affect the parameter characterizing how the joint surplus is split. However, it is typically assumed that the split parameter remains unchanged. (See, e.g., Capps, Dranove, and Satterthwaite 2003.) 85  A wide range of approaches has been used to estimate merger effects in the hospital industry. These range from empirical estimates of the relationship between price and measures of concentration and/ or bargaining power to estimation of structural simulation models based on bargaining theory. For a comprehensive survey of empirical analyses of relationships between price and concentration of hospital services, see Vogt and Town (2006, 2008) and Gaynor and Town (2012). For a discussion of structural simulation models, see Balan and Brand (2009) and Gowrisankaran, Nevo, and Town (2013). For an overview of the FTC’s approach to regulatory review of hospital mergers, see Farrell et al. (2011). For a critique of the FTC’s merger simulation model, see Keating et al. (2012). Similar intuition would apply to any industry in which an aggregator (e.g., a provider of a bundle of television networks or a collection of music) negotiates with a provider of content or service for the right in include the content or service in a bundled service (e.g., cable television or internet radio), which it then sells to consumers for a fee. 86 MCOs are private insurers that typically construct provider networks and then market health insurance plans to employers and individuals. 87  Ho (2009) discusses the circumstances under which positive-surplus contracts may not be agreed to in equilibrium.

500   Bryan Keating and Robert D. Willig otherwise diverted profits increases the disagreement profits realized by the hospital. Under Nash bargaining theory, this increase in disagreement profits leads to higher prices by shifting the bargaining power towards the postmerger hospital. The extent to which prices would increase will depend, among other factors, on the degree of substitution between the merging parties, with greater substitution between the merging parties suggestive of greater unilateral effects (all else equal).88 This logic is analogous to the logic of unilateral price effects in differentiated product markets discussed in section 20.2. From an implementation point of view, one notable aspect of recent regulatory review of hospital mergers in the United States is that the FTC has relied on a hybrid model that involves aspects of both structural modeling and nonstructural modeling. The model favored by the FTC to evaluate the price effects of prospective mergers includes two stages of empirical analysis.89 The first stage is based on structural models of consumer demand for hospital services and embodies explicit assumptions about the nature of demand. The FTC uses estimates from the structural demand model to develop proxies for the bargaining power of a hospital. The FTC then uses time series and/or cross-sectional variation in observed prices and in the bargaining power proxies to assess the joint distribution of these variables.90 It uses the estimated parameters from this nonstructural analysis to make predictions about the price effects of prospective mergers.

20.5.  Product Variety and Innovation The 2010 Horizontal Merger Guidelines recognize that competition can affect firms’ incentives along a variety of dimensions beyond the pricing and quality of existing products. While prices, and sometimes product quality, can be modified in the short term, in many industries longer term consumer welfare depends on ongoing innovation and the development of new products. As the 2010 Horizontal Merger Guidelines note, the same basic incentives discussed above affect the incentives of firms to engage in research and development with the goal of introducing new products to the market.91 In particular, merging firms’ incentives to introduce new products and/or retain existing products 88 

The prospective price effects will also depend on the ability of the MCO to mitigate the effects of hospital consolidation by rearranging its relationships with rival hospitals. (See Keating et al. 2012.) As Balan and Brand explain, the same basic intuition applies regardless of whether the merging hospitals engage in “all-or-nothing” bargaining postmerger, or continue to negotiate separately while still internalizing the effects of diversion. 89  For an overview of the FTC’s approach, see Farrell et al. (2011). For a critique of the FTC’s merger simulation model, see Keating et al. (2012). 90 More recently, Gowrisankaran, Nevo, and Town (2013) have developed a structural econometric model based on a Nash bargaining model that they use to predict the price effects of hospital mergers. 91  DOJ and FTC (2010), § 6.4 (“The first of these effects is most likely to occur if at least one of the merging firms is engaging in efforts to introduce new products that would capture substantial revenues from the other merging firm. The second, longer-run, effect is most likely to occur if at least one of the

Unilateral Effects   501

will depend, in part, on the extent to which such products cannibalize sales from the products of the merger partner.92 Changes in product offerings can have important welfare implications.93 The economics literature has developed techniques designed to assess the welfare effects of new products that are closely related to the demand models that we discuss in section 20.2.2. For example, Hausman (1997) uses a multilevel model of demand where the lowest level is based on an AIDS model in reaching the conclusion that the introduction of Apple Cinnamon Cheerios resulted in large welfare effects. Similarly, Trajtenberg (1990) and Petrin (2002) use logit-based demand models to assess the welfare implications of introducing CT scanners and minivans, respectively. Petrin (2002) shows that logit models will tend to overestimate the welfare effects of new products. This is because logit models include idiosyncratic taste terms and welfare is measured as the maximum utility across all products. Adding an additional product also introduces additional idiosyncratic error terms and with it the possibility that some consumer has an extremely high idiosyncratic preference for the new product.94 Thus it is important accurately to model the welfare implications of product variety to the extent it is determined that a merger is likely to affect product variety.95

20.6. Conclusion We have attempted to survey a large and fast growing body of economic research on the theory underlying qualitative and econometric analysis of unilateral effects of mergers. Even though the literature and accomplished analyses make use of many different techniques and work in many different market settings, we believe that we have been able

merging firms has capabilities that are likely to lead it to develop new products in the future that would capture substantial revenues from the other merging firm”). See also European Commission (2004), ¶ 38. 92 

Draganska, Mazzeo, and Seim (2009) develop a structural model to assess endogenous product choice. In their model, a merged firm may offer either greater or lesser product variety compared to stand-alone firms depending, among other factors, on the level of fixed costs. More generally, they recognize that product repositioning by both the merged firm and nonmerging rivals may have important implications for the welfare analysis of mergers. 93  As the 2010 Horizontal Assessment Guidelines note, “If the merged firm would withdraw a product that a significant number of customers strongly prefer to those products that would remain available, this can constitute a harm to customers over and above any effects on the price or quality of any given product” (DOJ and FTC 2010, § 6.4). For the applicable basic theory of consumer welfare, see Willig (1980). 94  See Nevo (2003) and Ackerberg and Rysman (2005) for potential solutions to the issue identified by Petrin (2002). 95  Evidence on a merger’s effect on product variety often comes from internal documents made available to regulatory agencies during the course of a merger review. However, the welfare implications require analysis based on consumer preferences, most reliably expressed through market demands.

502   Bryan Keating and Robert D. Willig to identify a few common themes that characterize the fundamental logic of the theoretical approaches and the econometric applications. We have also been attentive to the needs for methods of analysis of mergers that maintain their economic logic while enabling utilization within the limitations on time and data endemic to the timetable of antitrust policy determinations and enforcement. We conclude on the upbeat note that our survey shows real movement towards better use of rigorous economic theory and econometric methodology to formulate analytic frameworks that are attuned to market institutions and to the practical needs of antitrust practice. We also optimistically note that both confidential and public analyses of unilateral competitive effects have been trending to the use of more inclusive and richer data sets whose availability improves the reliability of policy conclusions, while promoting development of increasingly incisive economic tools.

Acknowledgment Significant portions of this chapter derive from work presented in Willig (2011).

References Ackerberg, Daniel A. 2001. Empirically Distinguishing Informative and Prestige Effects of Advertising. Rand Journal of Economics 32: 316–33. Ackerberg, Daniel A., C. Lanier Benkard, Steven Berry, and Ariel Pakes. 2007. Econometric Tools for Analyzing Market Outcomes. In James J. Heckman and Edward E. Leamer, Handbook of Econometrics, 4171–276. Oxford: Elsevier. Ackerberg, Daniel A., and Marc Rysman. 2005. Unobserved Product Differentiation in Discrete-Choice Models: Estimating Price Elasticities and Welfare Effects. Rand Journal of Economics 36: 771–88. Allen, Beth, and Martin Hellwig. 1986. Bertrand-Edgeworth Oligopoly in Large Markets. Review of Economic Studies 53: 175–204. Ashenfelter, Orley, Daniel Hosken, Michael Vita, and Matthew Weinberg. 2011. Retrospective Analysis of Hospital Mergers. International Journal of the Economics of Business 18: 5–16. Athey, Susan, and Philip A. Haile. 2006. Empirical Models of Auctions. Cowles Foundation Discussion Paper No. 1562, Yale University. Australian Competition and Consumer Commission. 2008. Merger Guidelines. Retrieved from http://www.accc.gov.au/content/index.phtml/itemId/809866. Bajari, Patrick, C. Lanier Benkard, and Jonathan Levin. 2007. Estimating Dynamic Models of Imperfect Competition. Econometrica 75: 1331–70. Baker, Jonathan, and David Reitman. 2012. Research Topics in Unilateral Effects Analysis. In Einer Elhauge, Research Handbook on the Economics of Antitrust Law, 25–52. Northampton, MA: Edward Elgar.

Unilateral Effects   503

Balan, David, and Keith J. Brand. 2009. Simulating Hospital Merger Simulations. Unpublished draft. Retrieved from http://works.bepress.com/david_balan/7. Berry, Stephen T. 1994. Estimating Discrete-Choice Models of Product Differentiation. Rand Journal of Economics 25: 242–62. Berry, Steven T., James Levinsohn, and Ariel Pakes. 1995. Automobile Prices in Market Equilibrium. Econometrica 63: 841–90. Berry, Steven T., James Levinsohn, and Ariel Pakes. 2004. Differentiated Products Demand Systems from a Combination of Micro and Macro Data: The New Car Market. Journal of Political Economy 112: 68–105. Binmore, Ken, Ariel Rubinstein, and Asher Wolinsky. 1986. The Nash Bargaining Solution in Economic Modelling. Rand Journal of Economics 17: 176–88. Brannman, Lance, and Luke M.  Froeb. 2000. Mergers, Cartels, Set-Asides, and Bidding Preferences in Asymmetric Oral Auctions. Review of Economics and Statistics 82: 283–90. Brazil Ministry of Finance and Brazil Ministry of Justice. 2001. Horizontal Merger Guidelines. Retrieved from http://www.seae.fazenda.gov.br/document_center/legislation/​annex-7_ guidelines_brazil.pdf. Bresnahan, Timothy F. 1987. Competition and Collusion in the American Automobile Industry: The 1955 Price War. Journal of Industrial Economics 35: 457–82. Bresnahan, Timothy F. 1989. Empirical Studies of Industries with Market Power. In Richard Schmalensee and Robert D. Willig, Handbook of Industrial Organization, vol. 2: 1011–57. Oxford: Elsevier. Bresnahan, Timothy F., and Peter C. Reiss. 1991. Entry and Competition in Concentrated Markets. Journal of Political Economy 99: 977–1009. Brueckner, Jan, Darin Lee, and Ethan S. Singer. 2013. Airline Competition and Domestic US Airfares: A Comprehensive Reappraisal. Economics of Transportation 2: 1–17. Budzinski, Oliver, and Isabel Ruhmer. 2009. Merger Simulation in Competition Policy: A Survey. Journal of Competition Law and Economics 6: 277–319. Bulow, Jeremy I., and Paul Pfleiderer. 1983. A Note on the Effect of Cost Changes on Prices. Journal of Political Economy 91: 182–85. Capps, Cory, David Dranove, and Mark Satterthwaite. 2003. Competition and Market Power in Option Demand Markets. Rand Journal of Economics 34: 737–63. Cardell, N.  Scott. 1997. Variance Components Structures for the Extreme-Value and Logistic Distributions with Application to Models of Heterogeneity. Econometric Theory 13: 185–213. Carlton, Dennis W. 2010. Revising the Horizontal Merger Guidelines. Journal of Competition Law and Economics 6: 619–52. Carlton, Dennis W. 2011. Use and Misuse of Empirical Methods in the Economics of Antitrust. CPI Antitrust Chronicle 3: 1–14. Carlton, Dennis W., and Jeffrey M.  Perloff. 2004. Modern Industrial Organization. 4th ed. Boston: Prentice Hall. Coate, Malcolm B., and Joseph J. Simons. 2010. Upward Pressure on Price Analysis: Issues and Implications for Merger Policy. European Competition Journal 6.2: 377–96. Competition Bureau of Canada. 2011. Merger Enforcement Guidelines. Retrieved from http://​ www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/03420.html. Competition Commission and Office of Fair Trading. 2010. Merger Assessment Guidelines. Retrieved from http://www.oft.gov.uk/shared_oft/mergers/642749/OFT1254.pdf.

504   Bryan Keating and Robert D. Willig Competition Commission of Singapore. 2007. CCS Guidelines on the Substantive Assessment of Mergers. Retrieved from http://www.ccs.gov.sg/content/dam/ccs/PDFs/CCSGuidelines/​ substantiveassessmerger_Jul07FINAL.pdf. Compte, Olivier, Frederic Jenny, and Patrick Rey. 2002. Capacity Constraints, Mergers and Collusion. European Economic Review 46: 1–29. Coninck, Raphaël. 2011. Recent Developments in EU Merger Control. Presented at the Merger Control Conference, Brussels, Belgium. Retrieved from http://ec.europa.eu/dgs/​ competition/economist/deconinck_ibc_en.pdf. Crooke, Philip, Luke Froeb, Steven Tschantz, and Gregory J Werden. 1999. Effects of Assumed Demand Form on Simulated Postmerger Equilibria. Review of Industrial Organization 15: 205–17. Dalkir, Serdar, John W. Logan, and Robert T. Masson. 2000. Mergers in Symmetric and Asymmetric Noncooperative Auction Markets: The Effects on Prices and Efficiency. International Journal of Industrial Organization 18: 383–413. Davidson, Carl, and Raymond Deneckere. 1986. Long-Run Competition in Capacity, Short-Run Competition in Price, and the Cournot Model. Rand Journal of Economics 17: 404–15. Davis, Peter. 2006. Coordinated Effects Merger Simulation with Linear Demand. Competition Commission. Retrieved from http://www.competition-commission.org.uk/assets/compe​titioncommission/docs/pdf/non-inquiry/our_role/analysis/coordinated_effects_merger_​sim.pdf. Deaton, Angus, and John Muellbauer. 1980. An Almost Ideal Demand System. American Economic Review 70: 312–26. Department of Justice. 1982. 1982 Merger Guidelines. Department of Justice. 1984. 1984 Merger Guidelines. Department of Justice and Federal Trade Commission (DOJ and FTC). 1992. 1992 Merger Guidelines. Rev. ed. Department of Justice and Federal Trade Commission (DOJ and FTC). 2010. 2010 Merger Guidelines. Doane, Michael J., Luke Froeb, Gregory J. Werden, and David M. Zimmer. 2012. Pricing and Market Concentration:  A  New Estimation Strategy. Unpublished paper. Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2034464. Draganska, Michaela, Michael Mazzeo, and Katja Seim. 2009. Beyond Plain Vanilla: Modeling Joint Product Assortment and Pricing Decisions. Quantitative Marketing and Economics 7: 105–46. Dubin, Jeffrey A., and Daniel L. McFadden. 1984. An Econometric Analysis of Residential Electric Appliance Holdings and Consumption. Econometrica 52: 345–62. Edgeworth, Francis Ysidro. 1925. The Pure Theory of Monopoly. In Papers Relating to Political Economy, vol. 1: 111–42. London: Macmillan. Einav, Liran. 2003. A Note on the Analogies between Empirical Models of Auctions and of Differentiated Product Markets. Unpublished paper. Retrieved from http://www.stanford.​ edu/~leinav/avd.pdf. Einav, Liran, and Jonathan Levin. 2010. Empirical Industrial Organization: A Progress Report. Journal of Economic Perspectives 24: 145–62. Einav, Liran, and Aviv Nevo. 2006. Empirical Models of Imperfect Competition:  A  Discussion. Unpublished paper. Retrieved from http://www.stanford.edu/~leinav/pubs/WC_​Discussion.pdf. Epstein, Roy J., and Daniel L. Rubinfeld. 2001. Merger Simulation: A Simplified Approach with New Applications. Antitrust Law Journal 69: 883–919.

Unilateral Effects   505

Epstein, Roy J., and Daniel L. Rubinfeld. 2004. Merger Simulation with Brand-Level Margin Data: Extending PCAIDS with Nests. Advances in Economic Analysis and Policy 4. Retrieved from http://www.degruyter.com/view/j/bejeap.2004.4.issue-1/bejeap.2004.4.1.1212/​ bejeap.2004.4.1.1212.xml. European Union. 2004. Guidelines on the Assessment of Horizontal Mergers under the Council Regulation on the Control of Concentrations between Undertakings. Official Journal of the European Union. Retrieved from http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri​ =OJ:C:2004:031:0005:0018:EN:PDF. Farrell, Joseph, and Carl Shapiro. 1990. Horizontal Mergers: An Equilibrium Analysis. American Economic Review 80: 107–26. Farrell, Joseph, and Carl Shapiro. 2010. Antitrust Evaluation of Horizontal Mergers: An Economic Alternative to Market Definition. B.E. Journal of Theoretical Economics 10. Retrieved from http://www.degruyter.com/view/j/bejte.2010.10.1/bejte.2010.10.1.1563/bejte.2010.10.1.1563.​ xml. Froeb, Luke, Steven Tschantz, and Gregory J. Werden. 2005. Pass-through Rates and the Price Effects of Mergers. International Journal of Industrial Organization 23: 703–15. Gaynor, Martin, and Robert Town. 2012. The Impact of Hospital Consolidation: Update. Robert Wood Johnson Foundation. Retrieved from https://www.rwjf.org/files/research/5973.74582.​ synthesisprojectupdate.hospitalconsolidation.pdf. Gowrisankaran, Gautam, Aviv Nevo, and Robert Town. 2013. Mergers When Prices Are Negotiated: Evidence from the Hospital Industry. NBER Working Paper No. 18875. Hausman, Jerry A. 1997. Valuation of New Goods under Perfect and Imperfect Competition. In Timothy F. Bresnahan and Robert J. Gordon, The Economics of New Goods, 209–37. National Bureau of Economic Research Studies in Income and Wealth, 58. Chicago: University of Chicago Press. Hausman, Jerry A., and Gregory K. Leonard. 2005. Using Merger Simulation Models: Testing the Underlying Assumptions. International Journal of Industrial Organization 23: 693–98. Hausman, Jerry A., Gregory K. Leonard, and J. Douglas Zona. 1994. Competitive Analysis with Differentiated Products. Annales d’Economie et de Statistique 34: 159–80. Hausman, Jerry, Serge Moresi, and Mark Rainey. 2011. Unilateral Effects of Mergers with General Linear Demand. Economics Letters 111: 119–21. Hendel, Igal. 1999. Estimating Multiple-Discrete Choice Models:  An Application to Computerization Returns. Review of Economic Studies 66: 423–46. Ho, Katherine. 2009. Insurer-Provider Networks in the Medical Care Market. American Economic Review 99: 393–430. Israel, Mark, Bryan Keating, Daniel L. Rubinfeld, and Robert D. Willig. 2012. Airline Network Effects and Consumer Welfare. Review of Network Economics: 1–36. Jaffe, Sonia, and E. Glen Weyl. 2010. Linear Demand Systems Are Inconsistent with Discrete Choice. B.E. Journal of Theoretical Economics 10. Retrieved from http://www. degruyter.com/dg/​v iewarticle/j$002fbejte.2010.10.1$002fbejte.2010.10.1.1729$002fbe jte.2010.10.1.1729.xml. Jaffe, Sonia, and E. Glen Weyl. 2012. The First-Order Approach to Merger Analysis. Unpublished paper. Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1765024. Japan Fair Trade Commission. 2011. Guidelines to Application of the Antimonopoly Act Concerning Review of Business Combination. Retrieved from http://www.jftc.go.jp/en/​ legislation_guidelines/ama/pdf/110713.2.pdf.

506   Bryan Keating and Robert D. Willig Katz, Michael L., and Carl Shapiro. 2003. Critical Loss: Let’s Tell the Whole Story. Antitrust 17: 49–56. Keating, Bryan, Paolo Ramezzana, Robert Willig, Margaret Guerin-Calvert, and Nauman Ilias. 2012. Comment on Farrell, Balan, Brand and Wendling (2011): “Economics at the FTC: Hospital Mergers, Authorized Generic Drugs, and Consumer Credit Markets.” Unpublished paper. Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2051028. Klemperer, Paul. 2008. Competition Policy in Auctions and “Bidding Markets”. In Paolo Buccirossi, Handbook of Antitrust Economics, 583–624. Cambridge, MA: MIT Press. Kreps, David M., and Jose A.  Scheinkman. 1983. Quantity Precommitment and Bertrand Competition Yield Cournot Outcomes. Bell Journal of Economics 14: 326–37. Kwoka, John, and Evgenia Shumilkina. 2010. The Price Effect of Eliminating Potential Competition: Evidence from an Airline Merger. Journal of Industrial Economics 58: 767–93. Mares, Vlad. 2000a. Asymmetric Mergers in Common Value Auctions. Unpublished paper, Washington University, St. Louis, MO. Retrieved from https://editorialexpress.com/cgi-bin/​ conference/download.cgi?db_name=NASM2001&paper_id=527. Mares, Vlad. 2000b. Market Concentration and Prices in Common-Value Auctions. Unpublished paper, Washington University, St. Louis, MO. Mazzeo, Michael J. 2002a. Competitive Outcomes in Product-Differentiated Oligopoly. Review of Economics and Statistics 84: 716–28. Mazzeo, Michael J. 2002b. Product Choice and Oligopoly Market Structure. Rand Journal of Economics 33: 1–22. McFadden, Daniel. 1980. Econometric Models for Probabilistic Choice among Products. Journal of Business 53: S13–S29. McFadden, Daniel. 1981. Economic Models of Probabilistic Choice. In C. E. Manski and Daniel L. McFadden, Structural Analysis of Discrete Data, 198–272. Cambridge, MA: MIT Press. Moresi, Serge. 2010. The Use of Upward Price Pressure Indices in Merger Analysis. Antitrust Source. Retrieved from http://crai.co.uk/uploadedFiles/Publications/the-use-of-UPPIs-in-​ merger-analysis.pdf. Nevo, Aviv. 2000a. A Practitioner’s Guide to Estimation of Random-Coefficients Logit Models of Demand. Journal of Economics and Management Strategy 9: 513–48. Nevo, Aviv. 2000b. Mergers with Differentiated Products: The Case of the Ready-to-Eat Cereal Industry. Rand Journal of Economics 31: 395–421. Nevo, Aviv. 2001. Measuring Market Power in the Ready-to-Eat Cereal Industry. Econometrica 69: 307–42. Nevo, Aviv. 2003. New Products, Quality Changes, and Welfare Measures Computed from Estimated Demand Systems. Review of Economics and Statistics 85: 266–75. New Zealand Commerce Commission. 2003. Mergers and Acquisitions Guidelines. Retrieved from http://www.comcom.govt.nz/assets/Imported-from-old-site/BusinessCompetition/​ MergersAcquisitions/ClearanceProcessGuidelines/ContentFiles/Documents/​ Mergers-and-AcquisitionsGuidelines-2003.pdf. O’Brien, Daniel P., and Steven C. Salop. 2000. Competitive Effects of Partial Ownership: Financial Interest and Corporate Control. Antitrust Law Journal 67: 559–614. Ordover, Janusz, Alan O. Sykes, and Robert Willig. 1982. Herfindahl Concentration, Rivalry, and Mergers. Harvard Law Review 95. Perry, Martin K., and Robert H. Porter. 1985. Oligopoly and the Incentive for Horizontal Merger. American Economic Review 75: 219–27.

Unilateral Effects   507

Peters, Craig. 2006. Evaluating the Performance of Merger Simulation: Evidence from the U.S. Airline Industry. Journal of Law and Economics 49: 627–49. Petrin, Amil. 2002. Quantifying the Benefits of New Products: The Case of the Minivan. Journal of Political Economy 110: 705–29. Reiss, Peter C., and Frank A. Wolak. 2007. Structural Econometric Modeling: Rationales and Examples from Industrial Organization. In James J.  Heckman and Edward E.  Leamer, Handbook of Econometrics, 4277–4415. Oxford: Elsevier. Salant, Stephen W., Sheldon Switzer, and Robert J.  Reynolds. 1983. Losses from Horizontal Merger:  The Effects of an Exogenous Change in Industry Structure on Cournot-Nash Equilibrium. Quarterly Journal of Economics 98: 185–99. Salop, Steven C., and Serge Moresi. 2009. Updating the Merger Guidelines: Comments. Public Comment Letter, Charles River Associates. Retrieved from http://www.ftc.gov/os/com​ ments/horizontalmergerguides/545095-00032.pdf. Schmalensee, Richard. 2009. Should New Merger Guidelines Give UPP Market Definition? GCP:  The Antitrust Chronicle. Retrieved from https://www.beta.competitionpolicyinter​ national.com/assets/Free/SchmalenseeDEC-091.pdf. Shapiro, Carl. 1989. Theories of Oligopoly Power. In Richard Schmalensee and Robert D. Willig, Handbook of Industrial Organization, vol. 1: 329–414. New York: North-Holland. Shapiro, Carl. 1996. Mergers with Differentiated Products. Antitrust 10: 23–30. Shapiro, Carl. 2010a. The 2010 Horizontal Merger Guidelines: From Hedgehog to Fox in Forty Years. Antitrust Law Journal 77: 49–108. Shapiro, Carl. 2010b. Update from the Antitrust Division. Remarks as Prepared for the American Bar Association Section of Antitrust Law Fall Forum. Retrieved from http://www.justice.​ gov/atr/public/speeches/264295.pdf. Small, Kenneth A., and Harvey S. Rosen. 1981. Applied Welfare Economics with Discrete Choice Models. Econometrica 49: 105–30. Tenn, Steven. 2011. The Price Effects of Hospital Mergers: A Case Study of the Sutter-Summit Transaction. International Journal of the Economics of Business 18: 65–82. Thomas, Charles J. 2004. The Competitive Effects of Mergers between Asymmetric Firms. International Journal of Industrial Organization 22: 679–92. Tirole, Jean. 1988. The Theory of Industrial Organization. Cambridge, MA: MIT Press. Trajtenberg, Manuel. 1990. Economic Analysis of Product Innovation: The Case of CT Scanner., Harvard Economic Studies 160. Cambridge, MA: Harvard University Press. Tschantz, Steven, Philip Crooke, and Luke Froeb. 2000. Mergers in Sealed versus Oral Auctions. International Journal of the Economics of Business 7: 201–12. US District Court, District of Columbia. United States of America v. H&R Block, Inc., et al. Civil Action No. 11-00948. BAH. Memorandum Opinion. Vives, Xavier. 1986. Rationing Rules and Bertrand-Edgeworth Equilibria in Large Markets. Economics Letters 21: 113–16. Vogt, William B., and Robert Town. 2006. How Has Hospital Consolidation Affected the Price and Quality of Hospital Care? Robert Wood Johnson Foundation. Retrieved from http://​ www.rwjf.org/content/rwjf/en/research-publications/find-rwjf-research/2006/02/​ how-has-hospital-consolidation-affected-the-price-and-quality-of.html. Vogt, William B., and Robert Town. 2008. The Impact of Hospital Consolidation: Update. Robert Wood Johnson Foundation. Retrieved from https://www.rwjf.org/files/research/5973.74582.​ synthesisprojectupdate.hospitalconsolidation.pdf.

508   Bryan Keating and Robert D. Willig Waehrer, Keith. 1999. Asymmetric Private Values Auctions with Application to Joint Bidding and Mergers. International Journal of Industrial Organization 17: 437–52. Waehrer, Keith, and Martin K. Perry. 2003. The Effects of Mergers in Open-Auction Markets. Rand Journal of Economics 34: 287–304. Walker, Michael. 2005. The Potential for Significant Inaccuracies in Merger Simulation Models. Journal of Competition Law and Economics 1: 473–96. Walters, Chris. 2007. Diversion Ahead! Approximating Diversion Ratios for Retail Chain Mergers. Competition Commission. Retrieved from http://www.competition-commission.org.uk/​ assets/competitioncommission/docs/pdf/non-inquiry/our_role/analysis/diversion_ratios.​ pdf. Weinberg, Matthew. 2011. More Evidence on the Performance of Merger Simulations. American Economic Review 101: 51–55. Weinberg, Matthew, and Daniel Hosken. 2013. Evidence on the Accuracy of Merger Simulation. Review of Economics and Statistics 95.5: 1584–600. Werden, Gregory J. 1996. A Robust Test for Consumer Welfare Enhancing Mergers among Sellers of Differentiated Products. Journal of Industrial Economics 44: 409–13. Werden, Gregory J. 1997. Simulating Unilateral Competitive Effects from Differentiated Products Mergers. Antitrust 11: 27–31. Werden, Gregory J., and Luke M. Froeb. 1994. The Effects of Mergers in Differentiated Products Industries: Logit Demand and Merger Policy. Journal of Law, Economics, and Organization 10: 407–26. Werden, Gregory J., and Luke M. Froeb. 1996. Simulation as an Alternative to Structural Merger Policy in Differentiated Products Industries. In Malcolm B. Coate and Andrew N. Kleit, The Economics of the Antitrust Process, 65–88. Norwell, MA: Kluwer Academic Publishers. Werden, Gregory J., and Luke M.  Froeb. 2008. Unilateral Competitive Effects of Horizontal Mergers. In Paolo Buccirossi. Handbook of Antitrust Economics, 43–104. Cambridge, MA: MIT Press. Weyl, E. Glen, and Michal Fabinger. 2012. Pass-Through as an Economic Tool. Unpublished paper. Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1324426. Willig, Robert D. 1978. Incremental Consumer’s Surplus and Hedonic Price Adjustment. Journal of Economic Theory 17: 227–53. Willig, Robert D. 1980. Welfare Analysis of Policies Affecting Prices and Products. New York: Garland Press. Willig, Robert D. 1991. Merger Analysis, Industrial Organization Theory, and Merger Guidelines. Brookings Papers on Economic Activity 1990: 281–312. Willig, Robert D. 2011. Unilateral Competitive Effects of Mergers:  Upward Pricing Pressure, Product Quality, and Other Extensions. Review of Industrial Organization 39: 19–38.

CHAPTER 21

C O O R D I NAT E D E F F E C T S Evolution of Practice and Theory JITH JAYARATNE AND JANUSZ ORD OVER

21.1. Introduction COMPETITIVE analysis of coordinated effects has moved from a singular focus on market concentration to a more sophisticated assessment of the likelihood that a transaction would facilitate coordinated conduct. This evolution is reflected in the US Horizontal Merger Guidelines (HMGs) and in regulators’ approach to recent merger matters in the United States and in the EU. The 1984 US HMGs emphasized the structural effects of a transaction. The 1992 HMGs went beyond the structural presumption and asked whether the relevant market is conducive to coordinated conduct and how a particular transaction would change market conditions in a way that would make coordination more likely, stable, or complete. Since then, attention has been paid to the importance of articulating a mechanism by which coordinated conduct is likely to occur as well as the risk of “parallel accommodating conduct.” Despite these welcome analytical and policy developments, the analytical foundations for predicting coordinated effects remain relatively undeveloped compared with unilateral effects. Coordinated effects analysis focuses on the impact of mergers on the potential for tacit collusion, which is not necessarily deterred through antitrust enforcement by the application of laws against explicit collusion. Hence, it is not surprising that the analytics of coordinated merger effects overlaps with, and is informed by, the economics of tacit collusion. This chapter briefly reviews the pertinent portions of the literature on tacit collusion and shows how its learning is applied or can be applied in the assessment of horizontal mergers. It then quickly reviews several merger cases that were scrutinized through the lens of coordinated interactions and comments on the types of evidence

510   Jith Jayaratne and Janusz Ordover that may be probative to the outcome of the review process. In this respect, it is important to recognize that, as compared to unilateral effects analyses, there is much less analytical rigor and much more reliance on a broad range of indicators, both quantitative and qualitative, when gauging the likelihood of coordinated effects. This is perhaps not surprising (as will be explored below), but makes counseling regarding the risks of coordinated effects more difficult and litigation more unpredictable. However, some progress towards the quantification of the risk of coordinated effects is being made linking the increase in incentives for coordination to the potential changes in the benefits from coordination resulting from a merger.

21.2.  Initial Thinking on Coordinated Effects: The Structural Presumption and Merger Policy In the context of coordinated effects, regulators have long viewed the impact of mergers through the lens of the “structural presumption,” that is, that a reduction in the number of firms in a market resulting in a material increase in concentration elevates the potential for harmful coordination among firms in the relevant market. This presumption is an element of the 1984 US Merger Guidelines, for example, which state that “[w]‌here only a few firms account for most of the sales of a product, those firms can in some circumstances either explicitly or implicitly coordinate their actions in order to approximate the performance of a monopolist” (DOJ 1984, § 1.0). The basic economics underpinning the structural presumption was established in George Stigler’s seminal 1964 paper on oligopoly theory (Stigler 1964). The Stiglerian framework, properly interpreted, is a rich one that shows how a broad range of economic factors other than simply the number of competitors affects the probability of successful coordination in a market. It is therefore useful to summarize Stigler’s framework for understanding collusion in oligopoly and then illuminate that approach with a simple mathematical model. From this point it is relatively straightforward to link Stigler’s insights to the analysis of coordinated effects. Suppose there is an industry with N > 2 identical firms, each producing an identical undifferentiated product at the same marginal cost c. If these firms compete, each will earn a profit of π comp , which depends on the number of active firms, marginal costs, the demand for the product, and the nature of (static) competitive interactions. If these firms collude, each firm would (by assumption) earn a level of profit equal to (1 / N ) * πmon > πcomp , where πmon is the monopolist profit level. Given these assumptions, industry participants have an incentive to collude and raise price to monopoly level. However, the firms’ individual incentives may hinder collusion even in the absence of risks of legal sanctions, as explained by Stigler. Put another way, as explained by Stigler, even if market participants as a group may have an incentive to raise price (or

Coordinated Effects   511

prices) above the current (competitive) benchmark, absent side-payments they simply may not be able to do so in a realistic market setting. Although all firms in the market under consideration can earn more from colluding, each firm has an incentive to cheat on the collusive agreement if it is more profitable in the short run to cheat than to stay in the collusive agreement.1 Suppose that a firm cheats by pricing just a little below the (collusive) monopoly price, diverting all other firms’ sales to itself. It would be able to do so if it faces no capacity constraints and if other firms cannot respond instantly to the deviation by the cheating firm. The cheating firm’s one-period payoff from cheating is (approximately) the entire industry-level monopoly profit, π mon . Suppose further that the deviation from collusion is detected after one period and, once detected, it is punished by all firms reverting to competitive behavior and they act in that manner forever. This is, of course, a drastic assumption, but it helps to identify the incentives.2 With this assumption, once collusion fails, each firm earns just π comp each period forever. As a simplification, assume that the competitive benchmark is Bertrand price competition where each firm maximizes its profits, taking as given the prices of its rivals. Since products are undifferentiated, π comp = 0. The profit stream earned over time by a firm that deviates from the collusive path is V (d ) = π mon + δ * 0 = π mon , where δ (a fraction between 0 and 1) is the “discount factor” used to express future earnings in terms of today’s dollars. δ can be viewed as a metric of a firm’s “impatience,” that is, a lower δ implies that a firm is more willing to trade off future profits for current profits. In assessing whether deviation is profitable, a firm has to compare V (d ) against the earnings over time from conforming to the collusive agreement. Let V (m) represent earnings per period under collusion. It can be shown that V (m) = (1 / N ) * π mon + δV (m) . That is, the net present value of sticking to the collusive agreement today is equal to today’s share of monopoly profit plus the net present value of sticking to the agreement starting tomorrow. There is no incentive to deviate from the monopoly agreement if V (m) > V (d ) . That is, collusion is sustainable if it is compatible with each firm’s individual incentive to stick to a collusive agreement. From the inequality above, we see that collusion will not be undermined by cheating as long as δ > 1 − (1/ N ) . In this simple model, this threshold is more likely to be reached and hence collusion is made more likely (i.e., the incentive compatibility requirement 1 / N > (1 − δ ) is more likely to hold) when (1) the number of firms, N, is small, and/or (2) firms do not discount future profit flows heavily (i.e., are “patient”). This simple model already yields several lessons for the analysis of the coordinated effects. First, ceteris paribus, a reduction in the number of firms may make tacit collusion more likely (as reflected in the structural presumption). Second, even if the market is potentially susceptible to 1 Moreover, there is no reason to assume that all market participants would be necessarily inclined to join a cartel, unless a sufficiently rich set of side-payments is feasible. 2  Other more complex punishment strategies have been explored by economists. See, e.g., Abreu (1998). One message that emerges from these more sophisticated models of tacit collusion is that such collusion is likely to be sustained in a broader range of market settings, as measured by the impatience parameter δ.

512   Jith Jayaratne and Janusz Ordover coordinated interactions (tacit collusion), collusion will not be sustained if market participants are “impatient” enough. Third, if the merger affects the rate at which (some) firms discount the future and makes them more “patient,” it can make collusion more likely, complete, and stable. Of course if a merger can make (some) firms more “impatient,” it can have precisely the opposite effect. While the previous discussion assumes that all firms in the industry have the same discount factor of δ, it is more realistic to assume that δ varies from firm to firm. This poses a coordination challenge when firms with low δs exist in the market. Since these firms are more impatient (i.e., they are more willing to forgo future earnings in favor of immediate earnings), they have a greater incentive to cheat on the collusive agreement.3 Such firms are commonly known as “mavericks,” as they tend to have a disruptive and competitive influence in their industry.4 For collusion to be successful in a market with a maverick firm, the firms must agree on a course of action that is consistent with the maverick’s incentives. This may not be possible: if the other firms believe that the maverick firm will cheat on any tacitly collusive agreement, there may be no way to implement such an agreement.5 The presence of a maverick, then, can be an important consideration in coordinated effects analysis. A merger may increase the risk of postmerger coordination if it eliminates a maverick firm or if it changes a former maverick’s incentives such that it is less likely to deviate from a tacitly collusive agreement (or is more likely to join a coordinating/collusive group of firms). Of course, the reverse is also true: a merger may decrease the risk of postmerger coordination if it creates a new maverick firm (or makes a maverick even more competitive). Baker (2002) explores the many possible ways a merger can impact a maverick, and suggests that coordinated effects analysis should focus on the merger’s effects on the competitive role of maverick firms. Baker and Shapiro (2008) additionally comment on the relationship between the structural presumption and the presence of mavericks by noting that “[t]‌he reduction in the number of sellers raises the odds that a merger involves a maverick—the type of merger most likely to enhance seller coordination—and those odds generally grow the most when the number of sellers is few.” It has to be acknowledged, however, that there is no a priori reason why increasing the number of firms in the market should increase the likelihood of a maverick. Whether it is true or not depends on the distribution from which the firms 3  The Stiglerian model defines δ as a function of the interest rate r such that δ = 1/(1 + r). Formally, then, a firm with a low δ is “impatient” in that it heavily discounts future cash flows. More broadly, though, impatience is a proxy for a firm’s many possible incentives that make it less likely to accept possible collusive agreements that would be accepted by other firms. See also Luis Cabral’s textbook (Cabral 2000) that shows how various features of the industry can be intuitively captured by the (generalized) discount factor. 4  The 2010 US HMGs define a maverick firm as one “that plays a disruptive role in the market to the benefit of customers” (DOJ and FTC 2010, § 2.1.5). 5  Note, though, that the other firms may be able to jointly raise prices above the competitive level even in the presence of a maverick firm. This can be the case, for instance, if the maverick firm is supply constrained (such that it cannot supply the entire market), faces increasing costs of supplying the market, or makes products that are imperfect substitutes for the products sold by the colluding firms.

Coordinated Effects   513

are drawn. For example, if firms differ in their marginal costs, the difficulties in reaching terms of agreement will not arise because the firms are differentially “impatient” but because they will differ in the assessment of the optimal industry price. From an empirical perspective, the structural presumption is less airtight than one might expect. Cross-industry studies of the relationship between measures of accounting profitability and market concentration have found a variety of results, but the overall relationship between profitability and concentration is weak, as noted in Schmalensee (1989). (“The relation, if any, between seller concentration and profitability is weak statistically, and the estimated concentration effect is usually small. The estimated relation is unstable over time and space and vanishes in many multivariate studies.” (976)) Interestingly, in one of the few instances of a direct test of coordinated effects, Ganslandt and Norback (2004) found that there was no relationship between merger-induced changes in the Herfindahl Index of concentration and the conduct of firms in the Swedish retail gasoline market. Moreover, correlations between profitability and concentration—even if they did exist—do not necessarily demonstrate cooperation between firms. It may simply be the case that, in markets with only a few successful firms, those firms will earn substantial profits because they are especially efficient. Lastly, if by coordination one means a deviation for a benchmark Cournot or Bertrand oligopolistic market, then a reduction in the number of firms, ceteris paribus, creates the incentives to lower output and raise prices even absent any change in firms’ conduct. It is clear from the above discussion that neither theory nor empirical evidence unambiguously predicts that a reduction in the number of firms in an industry necessarily makes collusion inevitable, or even more likely. Nonetheless, changes in market concentration can still be used as a starting point for a full-blown assessment of the likelihood and the magnitude of coordinated effects, and indeed play a major part in antitrust policy thinking even today. Sometimes this simple (or simplistic) approach is reflected in a policy view that a merger that reduces the number of firms from four to three is unlikely to be cleared, absent remedies or serious attenuating factors.

21.3.  Beyond the Structural Presumption: Checklist of Market Factors, Coordination Mechanism, and Implications of Pricing Complexity The 1984 US HMGs emphasized the structural effects of a transaction (although they recognized that market factors such as the extent of product differentiation are relevant for assessing the impact of a merger) (see DOJ 1984, § 3.411). The 1992 US HMGs went beyond the structural presumption and asked whether the relevant market is conducive to coordinated conduct (i.e., it focused on an examination of a “checklist” of factors that

514   Jith Jayaratne and Janusz Ordover make a market more or less conducive to coordinated conduct) and how the transaction affects the elements of the checklist. Indeed, antitrust regulators in the United States and in Europe have further recognized the need to articulate how a particular transaction would change market conditions (as reflected in the checklist) in a way that would make coordination more likely, stable or complete. More recently, following academic work, regulators have also recognized the importance of complexity in firms’ market strategies, including pricing, in assessing the likelihood of coordinated effects. We describe each of these developments in this section.

21.3.1.  Checklist Factors Both theoretical and empirical literature have identified certain market characteristics as being particularly relevant for assessing whether a market is vulnerable to successful coordination. Both the 1992 and 2010 US HMGs contain checklists of pertinent market features. Price transparency and demand and cost uncertainty are an example. So far, we have assumed that the relevant market is not subject to any shocks (such as fluctuations in demand or in firms’ costs) and that publicly observed market data, when coupled with each firm’s knowledge of its actions, adequately reveal whether colluders are cheating or not. Not surprisingly, rigorous modeling of coordinated conduct in markets that are beset by imperfect and asymmetric information is quite difficult. More importantly, in these types of markets, successful coordinated conduct is less likely, less sustainable, and less complete. Imperfect and asymmetric information may impede reaching an agreement on the collusive level of price or other competitive variables of concern, may necessitate adjustments in firms’ behavior and so confound the detection of cheating, and may reduce incentives to collude if coordination requires the participants to follow rigid behavioral rules. Moreover, in such markets it may be very difficult to arrive at terms of coordination absent direct communications among firms because such terms may have to specify complex responses to changing market conditions, for example. To illustrate, a pathbreaking paper by Green and Porter (1984), which examines tacit collusion in a market subject to random demand shocks, shows that the putative colluders must not only agree on the level of output (which will exceed the monopolist’s output) but also on the “trigger price”—that is, the price such that if observed price falls below that level, putative colluders will engage in a “price war”—and on the duration of the price war following that triggering event. Surely, it is much more difficult for market participants to arrive at such a complex equilibrium triplet as compared to the benchmark case where coordination is “only” required on price (or output). Moreover, it is not necessarily the case that a reduction in the number of firms makes reaching such an agreement materially easier. Empirical studies (e.g., Suslow 2002 and Dick 1996) have demonstrated the importance of transparency and demand uncertainty for cartel stability. Uncertainty in industry-specific demand shocks in particular have been found to be especially relevant

Coordinated Effects   515

for cartel duration (Suslow 2002) since the impact of such shocks can be difficult to separate from deviations from agreed-upon prices. The 1992 and 2010 US Merger Guidelines list several additional illustrative factors that may be conducive to (or be an impediment to) successful coordination (DOJ and FTC 1992, § 2.1; 2010, § 7). We have already noted the relevance of such factors such as market concentration and monitoring, but there are numerous other factors that bear on the likelihood of successful coordination and its sustainability. For example, the 2010 US HMGs note that a market typically is more vulnerable to coordinated conduct if “a firm’s prospective competitive reward from attracting customers away from rivals will be significantly diminished by likely responses of those rivals” (DOJ and FTC 2010, § 7.2). According to the 2010 Guidelines, a firm is more likely to anticipate strong responses in markets with (a) few significant competitors (which is likely when there are substantial entry barriers), (b) relatively homogenous products, (c) easy customer switching, or (d) meeting-competition clauses (DOJ and FTC 2010, § 7.2). Academic work has identified many other factors that affect the likelihood of successful coordinated conduct including the extent of excess capacity and its distribution among the incumbent firms, contracting practices, and multimarket contact.6 Cartel experience has also shed light on the importance of these factors. For example, the importance of entry barriers is illustrated by the international vitamins cartel, elements of which unraveled quickly once Chinese manufacturers aggressively entered and expanded their vitamins offerings (Connor 2008). More generally, Levenstein and Suslow (2006) find that entry is one of the two biggest challenges (along with adjustments to changing market conditions) faced by cartels. Given the destabilizing effects of potential entry in response to a coordinated increase in prices, members of a successful cartel may need to come to terms on raising entry barriers and deter cartel-destabilizing entry. Such entry-deterring mechanisms can include vertical foreclosure (wherein entrants are denied necessary inputs) and reliance on the state to introduce tariffs and antidumping protection. The value of such a checklist in merger assessment cannot be disputed, but it is important to appreciate its limitations. First of all, every market will be characterized by numerous “checklist” factors with potentially different directional impacts on the likelihood of successful coordinated conduct. For example, pricing may be transparent (which facilitates coordinated conduct) but transactions in the market infrequent and large (which creates incentives to cheat). What is worse, the directional impact of some factors can be quite ambiguous. For example, industry excess capacity can stabilize 6 

For an excellent and quite up-to-date review of the various factors, see Motta (2004). Kumar et al. (2013) note that merger waves have sometimes followed cartel enforcement actions by regulators. They take this as evidence that cartels can sometimes produce higher profits than mergers and firms can resort to mergers when cartels are infeasible. If they are correct and if it is possible to identify industries where a cartel is more likely to earn higher profits than a merged firm, then the fact that firms are choosing to merge in such industries is evidence that the risk of coordinated effects is likely to be low. In other words, firms in such industries must be choosing to merge because successful coordination is infeasible (unless, of course, mergers may alter market conditions such that successful coordination is made possible following the mergers).

516   Jith Jayaratne and Janusz Ordover collusion because it enables swift and potent punishment but also creates incentives to cheat. Hence, whether excess capacity is a factor that facilitates coordination or not depends in complex ways on which firms hold it and why. For another example, multimarket contacts might be seen as a facilitating factor but, as Bernheim and Whinston (1990) show, this need not be the case. Moreover, as recent work by Thomas and Willig (2006) shows, when different markets are subject to local shocks, linking competitive responses across such disparate markets may, in fact, undermine coordination. In the same vein, the impact of product homogeneity on the likelihood of coordinated effects is ambiguous. Indeed, when products are differentiated, the rewards from cheating are reduced (because the deviating firm may not be able to gain a large incremental volume of sales), but, on the other hand, the punishment for deviation is likely to be less painful (because the punished firm will not necessarily lose a large share of its sales). As Kuhn (2008) explains, the impact of product heterogeneity on coordinated conduct depends on the shape of the demand curve and on the type of punishment strategies being considered. Hence, from the enforcement standpoint, it makes sense to focus on the most salient features of a given market (as opposed to running through the checklist, ticking off pluses and minuses) and build a case for or against the merger from these key market features. For example, if the main factor impeding coordination is the difference in the range of products offered by various firms or in their vertical structure, then a transaction that “homogenizes” these features can be more problematic than a transaction that does not substantially affect these salient features.

21.3.2.  Coordination Problem and the Collusion Mechanism The simple model of tacit collusion presented in section 3 vastly understates the difficulties faced by any would-be tacit or explicit colluders. In particular, it assumes that cartel members can clearly identify a desired common course of action. Even if one assumes that there exists some common course of action to which industry participants could in principle agree and that no firm would have an incentive to deviate, there is still the “coordination problem” to be solved. The coordination problem arises for several reasons. First, the set of outcomes acceptable to all members that could be obtained through coordinated conduct need not be a singleton, and all participants may not agree on a specific outcome from this set.7 There may be several plausible focal points on which to coordinate, but to collude successfully, market participants must first discover what these possible outcomes are and then select and implement one of them. If firms disagree on which outcome is to be implemented, 7 

For example, firms A and B may agree that market division is preferable to competition but they may have diametrically different opinions on how the market should be divided. Similarly, if firms A and B have different costs, both may agree that.90 per widget is better than.50 but may disagree about whether.90 is better than $1.10.

Coordinated Effects   517

coordination may fall apart or fail to take root in the first place. Second, firms must have a mechanism for flexibly responding to changing market conditions. This may be impossible or difficult in the absence of explicit communications since “market signaling” may simply be too slow and too garbled.8 Third, firms may have to coordinate on “punishment” strategies in the event that such will have to be implemented. While reversion to “competition” seems like a simple prescription, it may mean different things to different market participants. Fourth, because punishment is costly to all market participants, firms may want to coordinate on when to trigger the punishment phase. Jumping the punishment gun may appear to others as cheating and lead to further unraveling of tacit collusion, but too much restraint in punishing can embolden cheating and thus reduce the chances of successful coordinated conduct. While the obstacles to successful coordination in real markets summarized above are substantial, successful coordination nevertheless does occur (Levenstein and Suslow 2006). However, the obstacles facing market participants also complicate the task of establishing the likelihood of coordinated effects from a merger. This is so if only because the economic foundations on which coordinated effects rest are much less developed and much more complex as compared to the Bertrand and Cournot workhorse models of static competition, and the assessment of the risks does not yield itself to the ready quantification associated with unilateral effects. European and US antitrust agencies’ recent approach to merger matters faces up to the challenge by focusing the analysis on describing and empirically testing, when possible, the plausible mechanism(s) of coordination by which a particular merger would make coordination more likely, more successful, or more complete. This approach aims to explain how a merger is likely to materially relax several of the constraints that limited, prior to the merger, the industry’s ability and incentive to maintain prices above competitive levels in a dynamic market environment. This, in turn, requires showing empirically and qualitatively, in the context of a clearly enunciated model of coordinated conduct, that the merger makes it easier to (1) reach and sustain agreement on key dimensions of competition; (2) detect deviations from the agreement (especially if the industry is not stable), so as to dissuade firms from cheating for fear of punishment; and (3) deter such deviations from coordination by means of more effective punishments (e.g., faster and more costly to the cheating firm). The various checklist factors are clearly relevant to this line of inquiry, but other considerations matter as well. Put another way, an analysis of merger effects must describe the main building blocks of the mechanism of coordination, including the description of the means by which coordination was or would be implemented and sustained. This leg of the analysis addresses several subissues, each of which needs separate analytics. First, what is the particular form the putative collusive conduct could likely take? Will it entail output suppression/price

8  In fact, when markets are subject to frequent shocks, tacit collusion may simply be impossible. Suslow (2002) finds that uncertainty in industry-specific demand shocks is especially relevant for cartel duration.

518   Jith Jayaratne and Janusz Ordover elevation, capacity restriction, or customer allocation? And which business decisions will remain independent? In Arch Coal, for example, the FTC claimed that the mechanism of coordination would entail suppression of capacity growth (hence output) ahead of anticipated growth of demand.9 Such suppression, if agreed upon and successfully implemented, would have the predictable effect of raising the time path of prices relative to what it would have been absent the merger. The Arch Coal court famously—and mistakenly— described the FTC’s theory of coordinated effects as “novel.”10 While coordination on capacity is hardly a novel idea in economics, the court was most likely on a firmer footing when it questioned whether such coordination can be reached absent direct negotiations among the putative colluders. Second, as a part of the examination of the mechanism of coordination, the competition authority (and the court) should assess, assuming that hypothetically the parties can agree to coordination on some aspect of their strategies, how the parties would be able to actually agree on the level (and time path) of the coordinated variable. For example, it is one thing to agree that coordinated conduct will entail customer allocation, it is quite another to agree on the actual allocation absent direct communications. Can an acceptable plan of action actually be reached without direct communications or with just “cheap talk”? Plainly, the more complex the terms of coordination, the less likely it is that these can be implemented absent some direct communications or absent a price leadership being assumed by one of the market participants. These considerations put a premium on simplicity of the form of coordination, but such simplicity has its own costs. Third, assuming that there is an agreement on the terms of coordination, how would such terms of coordination be monitored? In particular, is the market sufficiently transparent to make detection possible, and does the merger improve market transparency? When coordination permits a certain amount of flexibility (in response to market and idiosyncratic shocks) in firms’ market behavior, cheating may be hard to distinguish from adhering to the terms of agreement. This creates a temptation to deviate that can unravel the coordination or impede it from getting off the ground. Also, how will the terms of agreement be enforced? A simple reversion to “competition” may mean different things to different firms and, instead of solidifying coordination, may further destabilize it. Finally, how would the terms of the agreement be adjusted in response to market shocks? In addition to identifying a potential mechanism for coordination, it is also important to articulate how a proposed merger would enable the mechanism to be effective, thereby allowing coordination to occur as a result of the merger. In other words, it is important to identify what conditions prevent (or effectively limit) coordination in 9 

See, e.g., Arch Coal, 329 F. Supp. 2d at 131 (in which the FTC opposed the proposed merger “under the theory that the mechanism of tacit coordination that is most strongly supported by the evidence is a form of output restriction”). 10  Id.

Coordinated Effects   519

the absence of the merger and how such conditions would be altered as a result of the merger. The US Department of Justice (DOJ) appears to have adopted precisely such a framework in 2002 in its analysis of the proposed acquisition of Masonite, a manufacturer of molded door skins, by Premdor, a manufacturer of interior molded doors. In its complaint, the DOJ alleged that premerger the risk of price coordination in the downstream molded door market was reduced by the independent presence of Masonite.11 The DOJ argued that Masonite, which was present only in the upstream door skin market, had an incentive to supply smaller, nonintegrated downstream firms if major downstream firms attempted to elevate molded door prices. That is, Masonite had an incentive to supply potential downstream mavericks that could disrupt any attempt at downstream coordination. This incentive would disappear once Masonite was purchased by Premdor, since Masonite would also benefit from the elevation of molded door prices. Conversely, Premdor’s (albeit small) presence in the upstream market allegedly gave it the ability and incentive to disrupt any attempt by Masonite and other major upstream firms to elevate upstream prices, but this incentive would also be eliminated by the merger.12 Thus, the DOJ articulated how potential maverick behavior prevents coordination in the absence of the merger and how the merger would likely eliminate this constraint on coordination.

21.3.3.  Price Complexity There have been several interesting analytical breakthroughs in the economic analyses of coordinated effects that are of special relevance to assessment of mergers from the vantage point of tacit coordination. This theoretical literature explains how the complexity and volatility of profit-maximizing business strategies can impede pre- and postmerger coordination (Sannikov and Skrzypacz 2007; Athey, Bagwell, and Sanchirico 2004; Athey and Bagwell 2001). If the profit-maximizing monopolist acting alone in the marketplace would have to deploy dynamically complex business strategies—including advertising and marketing, investment in capacity and product development, besides pricing strategies—and the marketplace is subject to significant volatility, then such a monopolist must have the ability to adjust its behavior frequently, possibly even on a day-to-day basis. The problem this creates for incentives for coordinated conduct and for its sustainability is obvious:  if profit maximization requires behavioral flexibility, then, when a given firm makes those adjustments to its behavior, its rivals may have a difficult time distinguishing “innocent” strategic adjustments from cheating (that is, deviations from coordinated conduct). If rivals cannot predict or effectively “read” each other’s profit-maximizing strategies, they may be compelled to forgo flexibility and adopt simpler behavioral strategies in order to secure some, perhaps only limited, 11  Complaint ¶ 35, United States v. Premdor, Inc., No. 1:01CV01696, 2002 WL 1816981 (D.D.C. Apr. 5, 2002). 12  Id., ¶ 36.

520   Jith Jayaratne and Janusz Ordover coordination. Yet simple, stable, and predictable strategies are likely to be far from profit maximizing. Hence, if these strategies were to be adopted, gains from cooperation would be limited at stabilized prices relative to the benefits of strategic flexibility.13 At the same time, deviation profits may increase. There is, thus, a complex trade-off for the putative colluders between the costs of simplifying their behavioral rules to achieve coordination that can be readily monitored and the gains from coordination. Indeed, the more the profit-maximizing behavioral rules have to be bent to the objectives of facilitating tacit coordination, the less likely it is that a simple coordination mechanism can be designed that would enable the potential colluders to agree on such strategies and maintain them in the face of changing market conditions.14 Several past merger analyses have put new analytical focus on the importance of pricing heterogeneity for the likelihood of coordinated effects. Two high-profile cases exhibit the perils of complexity and the challenges that it creates for tacit coordination: Carnival Cruise Lines, and MGM–Mandalay Bay. At the end of 2001, two competing proposed transactions emerged in the cruise ship industry: the friendly creation of a “dual-listed company” combining Royal Caribbean Cruises, Ltd. and P&O Princess Cruises plc, and a competing hostile tender offer by Carnival Corporation for Princess. The three firms were the largest in the North American cruise market and either merger would have resulted in a single firm with a share of about 50 percent of the market. The FTC’s investigation involved extremely complex and detailed empirical analyses, drawing on more than 100 gigabytes of price transaction data. The FTC quickly recognized that neither a unilateral increase in price nor a tacitly coordinated price increase to all passengers was possible. Given that a theoretical monopolist would not find it profitable to raise prices across the board, the focus of the analysis shifted to whether the merger would facilitate anticompetitive price discrimination, that is, whether the merger created a reasonable probability of a focused price increase at some subset of customers. Under US merger antitrust law, the merger would raise competitive concerns not only if it harmed all consumers in the relevant market but also if it harmed a well-defined (i.e., identifiable) and sizeable subset of consumers. Consequently, the analytical challenge was to determine whether the industry could find a group of consumers that could be susceptible to a coordinated postmerger increase in prices. An important feature of the cruise line industry in this analysis is that the industry has a fixed and highly perishable product: the berths on any given sailing of a ship. As a result, pricing in the industry is enormously complex. Determining whether price discrimination was possible was, thus, not an easy task. Cruise lines are concerned first and foremost with filling as many berths as possible because most of the costs of the berth are sunk, so there is little extra cost to filling an open berth but significant lost revenue 13  Under the conditions described, firms may tacitly agree to cooperate before all of the pertinent information is revealed, but some may decide to cheat once certain types of information are revealed ex post. 14  Sannikov and Skrzypacz (2007) show that collusion is impossible when (a) information arrives continuously; (b) firms are able to respond quickly; and (c) goods are homogenous and firms compete in quantities.

Coordinated Effects   521

from letting a berth sail empty. The industry is therefore driven by “yield management,” which is implemented using highly complex models of the evolution of demand over time: sell an empty berth too soon and forgo a chance of selling it later to a customer willing to pay more; wait with an empty berth too long and it may never be sold. This yield management model results in large variation in day-to-day pricing of product. A key focus of the merger analysis in that case, therefore, was on the role of yield management as a tool for aiding potential price discrimination. Specifically, the question was whether there were patterns to the pricing for specific subgroups of customers that reduced the complexity of industry pricing in such a way as to allow for coordinated interaction. In particular, if cruise line operators could identify a specific group of potential customers with relatively inelastic demand for cruising, then a tacit increase in price could be confined to that group or, better yet, to the product that this group was (on average) favoring. An analysis of pricing to various groups of customers—those who booked early, those who booked in more attractive cabins, those who were “frequent cruisers,” and many others—revealed no such discernible pattern. The failure to find any such ways of reducing pricing complexity was a key factor in the FTC’s decision not to challenge the proposed mergers.15 Yield management also played an important role in the analysis of the MGM Mandalay Bay merger in Las Vegas. In 2004, MGM Mirage and Mandalay Resort Group announced that MGM Mirage would buy the Mandalay Resort Group for $7.9 billion. Both companies have a major presence in Las Vegas. MGM Mirage owns Bellagio, MGM Grand, The Mirage, Treasure Island, New York–New York and the Boardwalk Hotel, while Mandalay operates Mandalay Bay Resort, The Hotel, Luxor, Excalibur, Circus-Circus, Monte Carlo, and the Four Seasons at Mandalay Bay. The combined company, which would operate more than half of the approximately 72,000 hotel rooms on the Strip, would be the largest gaming company in the world, with combined revenues of nearly $7 billion a year. However, the evidence showed that there is extreme variation in the price of a standard room in Las Vegas over a period of several months prior to the day of arrival. Indeed, there was extreme variation even on any given day for a standard room for a booking for a specific day. As a result, it would likely not be possible for the hotels to coordinate because there was no focal price point off of which hotels could coordinate their pricing behavior. Moreover, in both cases, the need for pricing flexibility would make it difficult to detect legitimate pricing adjustments from cheating even if one could assume that rivals would be able to monitor actual prices paid by hotel guests.16 The FTC closed its investigation of the transaction without action. As these cases demonstrate, pricing complexity can make it difficult for firms to coordinate tacitly on price and thus lessens the risk that a merger will facilitate collusion. 15 

See Statement of the Federal Trade Commission Concerning Royal Caribbean Cruises, Ltd./P&O Princess Cruises plc and Carnival Corporation/P&O Princess Cruises plc, FTC File No. 021 0041 (Oct. 4, 2002), available at http://www.ftc.gov/os/2002/10/cruisestatement.htm. For more on the investigation, see, e.g., Coleman, Meyer, and Scheffman (2003). 16  In the cruise line business and in the hotel business, rack prices bear little resemblance to the actual prices paid by customers.

522   Jith Jayaratne and Janusz Ordover

21.4.  2010 US Horizontal Merger Guidelines and Parallel Accommodating Conduct The 2010 US Merger Guidelines expand the definition of coordinated interaction to include—in addition to explicit and tacit collusion—“parallel accommodating conduct (PAC) not pursuant to a prior understanding” (DOJ and FTC 2010, § 7). The guidelines explain PAC thusly: Parallel accommodating conduct includes situations in which each rival’s response to competitive moves made by others is individually rational, and not motivated by retaliation or deterrence nor intended to sustain an agreed-upon market outcome, but nevertheless emboldens price increases and weakens competitive incentives to reduce prices or offer customers better terms. Coordinated interaction includes conduct not otherwise condemned by the antitrust laws. (DOJ and FTC 2010, § 7)

Unlike tacit or explicit collusion, PAC is said to require no “agreed-upon market outcome” and is “not motivated by retaliation or deterrence.” This language brings coordinated effects closer to unilateral effects, which also are not motivated by retaliation or deterrence. Perhaps the DOJ and FTC hope that once the defining characteristics of coordination and unilateral conduct are brought closer together, the rigor of unilateral effects analyses can be likewise applied in coordinated effects analyses. As an initial matter, though, it is important to clarify what type of behavior constitutes PAC and whether this does indeed blur the distinction between coordinated and unilateral effects. To illustrate a “simple but central example of parallel accommodating conduct,” former DOJ chief economist Carl Shapiro analyzed in a recent speech a four-to-three merger in a highly concentrated market characterized by homogeneous products, little brand loyalty, and low switching costs. In this market, Shapiro (2010) considers a scenario where postmerger, the merged firm would likely become the industry price leader, and the remaining two rivals would likely follow price increases by the leader. Shapiro explains: This pattern of behavior does not involve any agreement by the other two firms to be followers; doing so is in their own best interest. This pattern of behavior does not involve any agreement that the merged firm will punish the other two firms if they fail to follow; but all three firms know that the merged firm will likely rescind its price increase in that event. Thus, the anticipated pricing behavior does not involve reaching and enforcing an agreement, central elements of the explicit and tacit collusion, the two other forms of coordinated interaction.

Coordinated Effects   523

Upon closer examination, this example of PAC seems more closely related to the standard tacit collusion story than to unilateral effects. Indeed, PAC and tacit collusion appear closely related if one takes a fairly plausible view of what “reaching and enforcing an agreement” entails. Consider a scenario, as in the example above, in which the merged Firm A will only raise prices if it believes Firm B and Firm C will follow. Clearly A does not know with certainty whether B and C will follow. B’s and C’s actions depend on several factors, including what they believe A will do if B and C do not follow. One likely answer is that A will rescind the price increase, since the higher price is profitable only if the other firms follow and is not profitable otherwise. In this scenario, A’s return to the initial price can be seen as a “punishment” for failing to follow its lead. On the other hand, if B and C follow the price increase, the outcome of this accommodating action can be viewed as a tacitly collusive agreement. Viewed through the lens of a traditional tacit agreement, “reaching an agreement” only requires that B and C follow A’s price increase; “enforcing” the agreement only requires that A rescind a price increase if B and C do not follow. Harrington (2012) similarly argues that PAC “does involve ‘retaliation or deterrence’ and thus an evaluation of PAC-generated coordinated effects would need to be conducted along standard lines.” However, he notes that the PAC concept is useful insofar as it highlights that coordination can occur even when mutual understanding is “far from the level that is associated with the concept of an agreement among firms” (p. 15). As in the example above, as long as the mutual understanding between firms is such that each participant understands that leader-follower pricing behavior will result in higher prices (and deviations from leader-follower behavior will result in a return to the initial status quo), PAC can result in coordinated anticompetitive outcomes. Of course, all of this begs the question why such understanding could not have developed prior to the merger. This brings the analysis back to the initial challenge of explaining how the merger changes the market dynamics, besides simply reducing the number of firms. Moreover, as an aside, if the PAC behavior is not only consistent with the postmerger environment, then the traditional merger simulation tool does not work either, inasmuch as it is based on the assumption that firms play a Bertrand game premerger (and postmerger). Consequently, this new analytical tool suggests perhaps that more attention should be paid to the premerger state of competition when assessing both the dangers (and profitability) of coordinated as well as unilateral effects. Moresi and coauthors (n.d.) have developed an index for evaluating the effect of a merger on the likelihood of PAC. Called the “Coordinated Price Pressure Index” (CPPI), the index assesses the likelihood of PAC based on each firm’s unilateral pricing incentives before and after a merger. The CPPI gauges the largest price increase that two firms would be willing to initiate and follow through PAC. A merger’s impact on the potential for PAC concerns is indicated by the increase in the CPPI resulting from the merger. A larger increase in the CPPI after the merger is thought to portend a greater likelihood that the merger will enable PAC. One benefit of the CPPI is that it does not require complex econometric simulations or extensive data inputs. Like the GUPPI—which is used frequently in unilateral effects

524   Jith Jayaratne and Janusz Ordover analysis—the CPPI requires data on margins, market shares, and diversion rations, and so is relatively easy to execute. However, the CPPI is limited to firms interacting in the context of differentiated product pricing competition, and it applies to a particular form of coordinated effects (PAC). Whether it is a potentially useful extension of coordinated effects analysis remains to be seen. In particular, a challenge is to extend the tool to incorporate responses from noncoordinating firms.

21.5.  Current Limitations of Coordinated Effects Analysis Coordinated effects policy analysis has grown and deepened over time. However, compared with unilateral effects analyses—which now encompass myriad sophisticated tests developed since the publication of the 1992 HMGs, at least—there is still much less rigor in judging whether a merger will increase the likelihood of coordination. Compte, Jenny, and Rey (2002) is an early effort at modeling the impact of mergers on the incentives for tacit collusion. They show that the introduction of asymmetric capacity makes coordinated conduct more difficult to sustain when aggregate capacity is limited. One implication of this finding is that, to the extent a merger exacerbates capacity asymmetries in a market, it may make that market less conducive to coordinated conduct. The CPPI metric developed in Moresi and coauthors (n.d.) is a recent effort at generating a relatively general analytical tool for assessing coordinated effects. Kovacic and coauthors (2009) and Gayle and coauthors (2011) are additional recent efforts at quantitative assessments of coordinated effects. In their approach, merger simulation analyses normally used to assess unilateral effects are extended to estimate the payoff from coordinated conduct and the impact of a merger on such payoffs. These papers build on the acknowledgment in the 2010 US HMGs that the payoff (i.e., the increase in profits) from coordination is important for assessing the risk of coordinated effects of a merger (DOJ and FTC 2010, p. 26). The payoff from coordination is especially large if the market is competitive and margins are small in the absence of coordination. These papers show that mergers that generate no concerns about unilateral effects (e.g., those that pass the UPP test) could nonetheless increase the payoff from coordination and thus raise concerns about coordinated effects. However, these efforts are at a relatively early stage compared to unilateral effects analyses. For example, Kovacic and coauthors (2009) and Gayle and coauthors (2011) rely on the premise that coordinating firms act like a firm that is formed by the merger of all coordinating entities. However, tacit and explicit collusion may take the form where the coordinating firms do not maximize joint profits and act like a single firm but instead adopt rules and procedures that sacrifice profits in return for stabilizing coordination. It is unclear if unilateral effects models such as merger simulation readily extend to such situations. Also, in Gayle and coauthors (2011), the authors show that one reason why a merger may

Coordinated Effects   525

create coordinated competitive concern is because it “homogenizes” the product portfolio of the merged firm and that of the rival firms. This is not a novel insight, but it is interesting to follow its implications in a fully specified model of competition. One challenge for coordinated effects analysis is that economic theory does not provide convincing conclusions as to how tacitly colluding firms can solve the “coordination problem.” While collusion theory enables a reasonably robust framework for assessing the potential for explicit collusion, it is less compelling on the topic of tacit collusion. It is standard to model collusion as an infinitely repeated game, where each period is simply a play of the static Cournot or Bertrand game. In this multiperiod game, it is possible that firms’ repeated interactions would allow them to achieve a more profitable (less competitive) equilibrium outcome than during a static one-shot game. In this setting, multiple collusive equilibria are possible, so the standard model assumes explicit bargaining between participants to choose a course of action. By definition, tacitly colluding firms cannot solve this coordination problem through explicit communications. A merger further complicates the analysis since it impacts firms’ incentives and ability to implement and enforce a collusive agreement.17 Without the seemingly clear-cut precision of the unilateral effects analyses, coordinated effects analysis is more likely to rely on a variety of indicators such as the increase in concentration and the presence of checklist factors. This can cause uncertainty for a firm considering a merger, as the relevant antitrust agency may cite any number of indicators to raise coordinated effects concerns. While it is unrealistic to expect policymakers to “prove” that a merger would facilitate postmerger coordination, it should be necessary that the agency explain the mechanism of coordination and how the market conditions—both before and after the merger—impact the likelihood of coordinated interaction. The development of analytical tools used to assess unilateral effects was prompted in part by the greater emphasis on unilateral effects placed by the 1992 US HMGs. Although coordinated effects are by no means neglected in the merger review process, the relative decline in the emphasis placed on coordinated effects is interesting given that changes in firms’ conduct due to a merger can have a substantial impact on a market, as illustrated in Ordover, Sykes, and Willig (1982). They analyze mergers in a market with homogeneous products where firms make strategic output decisions, that is, each firm decides on its profit-maximizing output conditional on the elasticity of market demand and each firm’s expectations of other firms’ reactions to a change in its output. The analysis incorporates scenarios where each firm expects that an increase in its output would induce other firms to lower their output (i.e., each firm conjectures a “competitive reaction” from other firms) or where other firms would increase their output (i.e., each firm conjectures a “cooperative reaction” from other firms) as well as the scenario where each expects no reaction from rivals.18 17 

Specifically, a merger impacts firms’ profit functions and changes the noncollusive Cournot-Nash (or Bertrand-Nash) equilibrium. 18  The greater the cooperative reaction, the higher the aggregate industry margin.

526   Jith Jayaratne and Janusz Ordover The authors find that if a merger between firms with even relatively modest market shares changes firms’ conduct, the impact on price can be comparable to a merger between much larger firms. For example, suppose that a maverick firm that was particularly unlikely to engage in tacit collusion is eliminated as a result of a merger between a firm with 11% market share and another with a 20% market share. Suppose also that as a result of the elimination of the maverick, following the merger, each firm j in the industry expects that other firms would react to a 1-unit increase output by firm j by collectively increasing their output by 0.5 units, whereas before the merger each firm expected a zero output response from other firms. In effect, firms’ conduct switches from one that is neither competitive nor cooperative to tacit cooperation following the merger. In this setting, the increase in the aggregate (industry-level) Lerner Index (margin) as a result of this merger is primarily due to the change in firms’ conduct. To achieve the same effect on the aggregate Lerner Index without a change in firms’ conduct, a firm with 11% market share would have to merge with a firm with a market share greater than 64%.19 Thus, if a merger alters firms’ conduct and makes tacit collusion more likely, that could have a substantial impact on prices and consumer welfare. Nonetheless, coordinated effects have been de-emphasized relative to unilateral effects by regulators’ in the merger review process. Perhaps regulators’ unstated view is that sustained tacit collusion is unlikely outside of explicit cartel arrangements of the type that can be addressed by the enforcement of laws such as Section 1 of Sherman Act in the United States. However, there is very little empirical research that directly assesses the impact of mergers on firms’ conduct. Ganslandt and Norback (2004) investigate changes in conduct following mergers in Swedish retail gasoline markets and find no evidence that increased concentration following mergers led to more accommodating output responses by competitors. More research on this topic is necessary. In sum, as this chapter describes, economics is deepening its understanding of coordinated effects and is developing new analytical approaches to gauging the risks of such effects. The traditional “policy shortcuts” used for this purpose—as reflected in the checklists in the United States and in the EU merger guidelines—are still useful but have to be used with care and with attention to the details of the industry or industries involved. The focus is now on the coherent description of the mechanism by which posited coordination can plausibly be implemented and sustained over time as well as on a development of empirical tests that can be used to gauge incentives and ability to deviate from the coordinated outcome.

19  Kovacic et al. (2009, 414) also consider an example where the impact coordinated effects on prices and output are substantially greater than the impact of unilateral effects.

Coordinated Effects   527

References Abreu, Dilip. 1998. On the Theory of Infinitely Repeated Games with Discounting. Econometrica 56: 383–96. Athey, Susan, Kyle Bagwell, and Chris Sanchirico. 2004. Collusion and Price Rigidity. Review of Economic Studies 71: 317–49. Athey, Susan, and Kyle Bagwell. 2001. Collusion with Private Information. Rand Journal of Economics 32: 428–65. Baker, Jonathan. 2002. Mavericks, Mergers, and Exclusion: Proving Coordinated Competitive Effects under the Antitrust Laws. NYU Law Review 77: 135–203. Baker, Jonathan, and Carl Shapiro. 2008. Reinvigorating Horizontal Merger Enforcement. In Robert Pitofsky, How the Chicago School Overshot the Mark: The Effect of Conservative Economic Analysis on U.S. Antitrust. Oxford: Oxford University Press: 235–91. Bernheim, B. Douglas, and Michael D. Whinston. 1990. Multimarket Contact and Collusive Behavior. Rand Journal of Economics 21(1): 1–26. Cabral, Luis. 2000. Introduction to Industrial Organization. Cambridge, Massachusetts: The MIT Press. Coleman, Mary T., David W.  Meyer, and David T.  Scheffman. 2003. Economic Analyses of Mergers at the FTC: The Cruise Ships Mergers Investigation. Review of Industrial Organization 23:121–55. Compte, Olivier, Frédéric Jenny, and Patrick Rey. 2002. Capacity Constraints, Mergers and Collusion. European Economic Review 46: 1–29. Connor, J. 2008. Global Price Fixing: Studies in Industrial Organization. New York: Springer Verlag. Gayle, Wayne-Roy, Robert Marshall, Leslie Marx, and Jean-Francoise Richard. 2011. Coordinated Effects in the 2010 Horizontal Merger Guidelines. Review of Industrial Organization 39: 39–56. Green, E. J., and R. H. Porter. 1984. Noncooperative Collusion under Imperfect Price Information. Econometrica 52(1): 87–100. Harrington, Joseph, Jr. 2012. Evaluating Mergers for Coordinated Effects and the Role of “Parallel Accommodating Conduct.” Johns Hopkins Department of Economics Working Paper. Kovacic, Willam, Robert Marshall, Leslie Marx, and Steven Schulenberg. 2009. Quantitative Analysis of Coordinated Effects. Antitrust Law Journal 76(2): 397–430. Kuhn, Kai-Uwe. 2008. The Coordinated Effects of Mergers. In Paolo Buccirossi, Handbook of Antitrust Economics. Cambridge: MIT Press: 105–44. Kumar, Vikram, Robert Marshall, Leslie Marx, and Lily Samkharadze. 2013. Buyer Resistance for Cartel versus Merger. Working paper. Levenstein, M. and Valerie Y.  Suslow. 2006. What Determines Cartel Success? Journal of Economic Literature 44(1): 43–95. Moresi, Serge, David Reitman, Steven Salop, and Yianis Sarafidis. N.d. Gauging Parallel Accommodating Conduct Concerns with the CPPI. Working paper, Charles River Associates, Inc. Motta, Massimo. 2004. Competition Policy, Theory and Practice. Cambridge:  Cambridge University Press. Ordover, Janusz, Alan Sykes, and Robert Willig. 1982. Herfindahl Concentration, Rivalry and Mergers. Harvard Law Review 95(8): 1857–74.

528   Jith Jayaratne and Janusz Ordover Sannikov, Yuliy, and Andrzej Skrzypacz. 2007. Impossibility of Collusion under Imperfect Monitoring with Flexible Production. American Economic Review 97: 1794–823. Schmalensee, Richard. 1989. Inter-industry Studies of Structure and Performance. In Richard Schmalensee and Robert D. Willig, Handbook of Industrial Organization. New York: North Holland: 951–76. Shapiro, Carl. 2010. “Update From the Antitrust Division.” Remarks as Prepared for the American Bar Association Section of the Antitrust Law Fall Forum, November 18. Stigler, George. 1964. A Theory of Oligopoly. Journal of Political Economy 72: 44–61. Thomas, Charles J., and Robert D.  Willig. 2006. The Risk of Contagion from Multimarket Contact. International Journal of Industrial Organization 24: 1157–84. US Department of Justice (DOJ). 1984. Horizontal Merger Guidelines. US Department of Justice and Federal Trade Commission (DOJ and FTC). 1992. Horizontal Merger Guidelines. US Department of Justice and Federal Trade Commission (DOJ and FTC). 2010. Horizontal Merger Guidelines.

CHAPTER 22

BU Y E R P OW E R I N M E R G E R R EV I EW DENNIS W. CARLTON, MARY COLEMAN, AND MARK ISRAEL

22.1. Introduction Merger analysis has generally developed with a focus on mergers among sellers and whether a merger will create or enhance seller market power. The issue of buyer power, however, can be important in many merger cases. • First, some mergers will combine two significant purchasers of a product or service. In such cases, the competitive analysis of the merger for those products or services assesses (1) whether the merger creates or enhances buyer power sufficiently to raise competitive concerns, and (2) whether competitive concerns are offset due to efficiencies resulting from the combination. • Second, some seller mergers involve markets where customers have some level of buyer power. In such cases, part of the competitive assessment of the merger may be to determine whether the presence of buyer power will constrain the potential exercise of seller market power following the proposed merger. In this chapter, we address both issues, including providing a summary of the relevant literature. In section 22.2, we discuss the competitive analysis of mergers among buyers. We describe different ways of defining buyer power (the standard textbook case of monopsony power or a broader notion of buyer power) and how these definitions affect the competitive analysis of buyer mergers. We then discuss when such mergers may create competitive concerns or alternatively result in procompetitive benefits. In section 22.3, we outline the role of buyer power in the competitive analysis of mergers among sellers. We detail the conditions under which buyer power might offset potential competitive

530   Dennis W. Carlton, Mary Coleman, and Mark Israel concerns from a merger among sellers. In both sections, we stress the limitations on the predictions that can be derived entirely from theoretical analyses and thus the need for empirical analysis.

22.2.  Assessment of Buyer Power in Mergers between Buyers Mergers of buyers have historically received far less attention than mergers among sellers, although that has been changing in recent years.1 A key question in evaluating buyer mergers is whether buy-side merger analysis should properly be a “mirror image” of sell-side merger analysis. The answer to this question in part depends on how one conceptualizes the “buyer power” that may be created by a buy-side merger.2 If one limits the definition of buyer power to the standard economic textbook case of monopsony power, then as we discuss in section 22.2.1, buyer power is precisely the mirror image of seller power—where “seller power” is defined as the ability to set a single price (to all consumers) above competitive levels—and thus buyer mergers should be evaluated in the same way as standard seller mergers. However, in many situations encountered in antitrust analysis, the standard economic textbook concept of monopsony power may be too narrow, and thus a broader definition of buyer power—which takes into account negotiations between buyers and sellers, neither of whom are price takers—is needed. In such cases, the definition of buyer power relates to the ability to extract additional surplus from the negotiated outcomes. In such cases, the creation of (or an increase in) buyer power can create either competitive harm (less efficient contractual terms and lower output) or competitive benefits (more efficient contractual terms and higher output).3

1 

Two proposed acquisitions that raised monopsony concerns were Aetna’s acquisition of Prudential’s health insurance assets and Cargill’s acquisition of Continental’s grain-trading division (both in 1999). Both transactions were approved by the DOJ’s Antitrust Division subject to divestitures that addressed competitive concerns. See Schwartz (1999) for a discussion of those mergers. Monopsony concerns also arose in UnitedHealth Group’s acquisition of Oxford Health Plans in 2004 and of PacifiCare in 2005. See Capps (2009) for a discussion of buyer power in heath plan mergers. For other recent economic analyses of buyer mergers, see, e.g., Raskovich (2003), Adilov and Alexander (2006), Fauli-Oller and Bru (2008), Carstensen (2008, 2010), Inderst and Shaffer (2008), Blair (2010), and Carlton and Israel (2011). 2  Chen (2008) surveys various definitions of buyer power and Chen (2007) discusses how the welfare implications of buyer power depend on how the term is defined. 3  The fact that mergers in which neither side is a price taker raise complex economic issues (related to the mergers; effect on the efficiency of outcomes from negotiation) is not unique to mergers of buyers. The same issue can arise in mergers of sellers. As with mergers of buyers, theoretically, an increase in seller power from a merger of sellers could result in negotiated outcomes that result in increased, decreased, or unchanged output.

Buyer Power in Merger Review   531

22.2.1.  Monopsony Power: Buyer Power with Competitive Supply One way to define buyer power is simply as classic monopsony power, or “the power to profitably reduce the price of an input below the competitive level.”4 A buyer has monopsony power if, when facing an upward-sloping supply curve, it has the incentive and ability to restrict output to pay lower prices. If one were to equate buyer power and monopsony power, the question of how to treat buyer power would be straightforward: In a manner exactly symmetric to that for monopoly power held by suppliers, monopsony power is inherently inefficient and welfare reducing.5 The potential for buyers to possess monopsony power is not just a theoretical concern. Empirical studies claim to have documented monopsony power faced by dairy farmers (Cotterill, Rabinowitz, and Tian 2003), teachers (Ransom and Sims 2010; Sohn 2008), nurses (Staiger, Spetz, and Phibbs 2010), and athletes.6 However, some studies fail to find significant evidence of monopsony power.7 Economic theory helps to explain the limitations on monopsony power that may generate this result, at least over the longer term. In particular, a reasonable supposition is that for many industries, supply elasticities are near infinite, at least in the long term. If powerful buyers attempt to reduce their demand to drive down prices, capital and labor will flow out of the industry over time. This yields the precise analog of situations in which demand elasticities are near infinite, leaving no scope for the anticompetitive exercise of market power. However, a key distinction is that, although there are certainly many cases where demand elasticities are not infinite—substitution away from “necessity” products is surely limited—supply elasticities are much more likely to be very large, at least over the long term. As a result, there are likely to be more situations in which monopoly power is a serious concern than there are for monopsony power. 4  American Antitrust Institute (2008, p. 99). For a more complete treatment of the economic theory of monopsony, see Blair and Harrison (2010, ­chapter 3). 5  See Carlton and Perloff (2005, p. 108), who observe: “The monopsony deadweight loss triangle (Figure 4.5) is analogous to the deadweight loss that results from monopoly (Figure 4.2a).” See also Blair and Harrison (2010, p. 41): “In the same sense that a monopolist has market power in selling its output, the monopsonist has buying power in purchasing some of its input requirements. The economic objections to monopoly and monopsony are similar: The exercise of market power reduces social welfare.” For more discussion of the symmetry between monopoly power and monopsony power, see Noll (2005, p. 591). 6  See Blair and Romano (1997), who examine monopsony in college athletics, and Twomey and Monks (2011), who examine monopsony in Major League Soccer. For more examples of the possible exercise of monopsony power, see Blair and Harrison (2010, c­ hapters 1 and 8–10), who discuss bid rigging in antique auctions, the activities of Salomon Brothers at Treasury auctions, the efforts of restaurants in Boston vs. American Express, collusion among elite colleges in financial aid decisions, collusion by baseball owners in the free agent market, effects of NCAA rules on coaches’ salaries and players’ scholarships, and monopsony power issues in agriculture and healthcare markets. 7  See, e.g., Boal (2009), who rejects the monopsony model of the market for schoolteachers in South Carolina and Texas; Medcalfe and Thornton (2006), who conclude that the effect of monopsony on teachers’ salaries in Georgia is very small; and Hirsch and Schumacher (1995, 2005), who conclude that monopsony power is not an important determinant of the wages of nurses.

532   Dennis W. Carlton, Mary Coleman, and Mark Israel In addition, although a merger of buyers that creates or increases monopsony power for the merged firm may raise competitive concerns, as with mergers among sellers, this effect may be offset, either partially or completely, by the realization of merger-specific efficiencies (Blair 2010). Importantly though, the ability for firms to obtain lower input prices via the exercise of monopsony power is not itself an efficiency. Arguments to the contrary may be premised on a belief that the lower input prices will translate into lower downstream prices. Such a claim is belied by the fact that a firm with monopsony power obtains lower prices by reducing the quantity of inputs that it purchases and thus it produces less output. Hence, given a downward-sloping demand curve in the downstream market, downstream prices must rise (and total welfare fall) due to the exercise of monopsony power.8 The US Merger Guidelines correctly reflect the anticompetitive nature of monopsony power (DOJ and FTC 2010, Section 12). In some cases, the potential creation of monopsony power has led US Department of Justice (DOJ) or the Federal Trade Commission (FTC) to challenge a merger. For example, Schwartz (1999) explains that consideration of the interaction among a variety of factors, including physician’s switching costs, price discrimination, postmerger market shares, and the merging firm’s market shares across particular physicians, led the DOJ to conclude that the acquisition of Prudential by Aetna would depress the price significantly to a substantial number of physicians in the Houston and Dallas areas. Again, however, we stress the important distinction between the short run and the long run. In the short run, supply elasticities may be relatively low and there may be scope for monopsony power; in the long run, it seems likely that some physicians in the Houston and Dallas area would move out and fewer people would choose to become physicians in these areas—the supply elasticities would become quite large—thus offsetting any monopsony power. As another example, the FTC has frequently challenged mergers among owners of natural gas pipeline gathering systems in cases where the merging firms are the only two or two of a small number of options by which owners of wells producing natural gas can get their product to consumers. Typically these natural gas gathering systems connect to larger pipeline systems and the natural gas is sold at a gathering point or “hub” that is relatively elastically supplied. Thus, although such mergers are likely to have little impact on downstream prices of natural gas, the potential for an exercise of monopsony power could result in lower prices received by natural gas producers and, potentially, an inefficiently low supply of natural gas in the affected areas.9

8  In fact, even if the demand curve facing the firm were perfectly elastic so that the output price does not rise, there could still be an economic inefficiency since the firm would produce too little output. Even if other firms were to make up the reduction in output, there could still be an economic inefficiency if the production by a relatively low cost, inframarginal firm were replaced by more production from higher cost, marginal producers. 9  For examples of such cases, see Conoco Inc. and Phillips Petroleum Company (2002), File No. 021 0040, Docket No. C-4058, and Chevron Corporation, and Texaco Inc (2001), File No. 011 0011, Docket No. C-4023. (Some of the authors have worked on these transactions.)

Buyer Power in Merger Review   533

22.2.2.  Buyer Power Where Suppliers Are Not Price Takers 22.2.2.1.  Broader Definition of Buyer Power The monopsony model provides a very narrow description of “buyer power.” It is based on the restrictive assumptions that a buyer pays the same price per unit for all units of the input and that the amount paid per unit increases (according to the supply curve of the sellers who themselves have no market power as sellers) if more units are purchased.10 These assumptions are widely violated, particularly in the more sophisticated contractual negotiations that are often associated with input purchases where both buyers and sellers have some degree of market power. For example, the monopsony model assumptions are violated by the presence of (non-cost-based) volume discounts or, more generally, nonlinear pricing terms. In this more general, nonmonopsony case, the buyer and seller (neither of whom is a price taker) can be thought of as engaging in a bilateral bargaining game, in which they negotiate jointly over price and quantity (and possibly other terms), unlike the monopsony model in which the buyer sets the price and the sellers produce according to their marginal cost curves, taking that price as given. Because quantity can be negotiated jointly with price, there is no need for a buyer to be constrained to choose price and quantity combinations on the supply curve, and therefore it is not necessarily the case that the buyer must restrict its input purchases to obtain a lower price. In this context, we define an increase in buyer power as the ability of a buyer to obtain a greater percentage of the combined surplus created through an agreement with a given supplier via price and nonprice terms.11 We note that it is possible that the changes that lead to increased buyer power also change total surplus—for example, if more powerful buyers are also able to arrive at more efficient contracts when negotiating with sellers—so that buyers could end up with more surplus even if they capture a smaller share of surplus. We would not call this an increase in buyer power. Rather, our definition of

10 

Chen (2007, p. 22) explains that two important assumptions of the ‘textbook’ theory of monopsony are that the supply curve is upward sloping (because with a horizontal supply curve the monopsonist would be unable to affect the market price and cause an efficiency loss) and transactions between the monopsonist and the supplier occur at a single unit price (because if “nonlinear” pricing is permitted, the efficiency loss of the monopsony may be reduced or eliminated). 11  Others have also used a similar definition. The American Antitrust Institute (2008, p. 99) concludes that “buyer power is best defined as the ability of a buyer to depress the price it pays a supplier or to induce a supplier to provide more favorable nonprice terms.” (This is consistent with our definition if “price” means total revenue.) Chen (2008, p. 247) defines buyer power as “the ability of a buyer to reduce price profitably below a supplier’s normal selling price, or more generally the ability to obtain terms of supply more favorable than a supplier’s normal terms,” where the “normal selling price” is defined as “the supplier’s profit-maximizing price in the absence of buyer power.” Doyle and Inderst (2007, note 29) distinguish between “countervailing power” and “substantial buyer power,” with the latter referring to “the ability (though not necessarily the incentive) to engage in potentially anti-competitive behavior,” but the economic literature generally does not follow this suggestion.

534   Dennis W. Carlton, Mary Coleman, and Mark Israel an increase in buyer power requires that the buyer captures a greater share of the total surplus. We consider the appropriate treatment of buyer power under antitrust policy in this situation.12 Under this model, a relevant question is whether an increase in buyer power that is due to the merger of two buyers should be treated as a merger-related efficiency and thus procompetitive or a merger-related harm and thus anticompetitive. Because this is a bargaining game, either outcome is possible. The relevant question for efficiency versus anticompetitive effect is whether contract terms negotiated after the merger lead the buyer and seller to increase the total amount of surplus.13 Inderst and Shaffer (2007a) explain that, when negotiations are limited to linear tariffs, an increase in buyer power manifests itself in lower prices to final consumers and thus in higher output, which thereby reduces deadweight loss (increases total surplus) and increases consumer surplus. In contrast, when negotiations are not restricted to linear tariffs, a buyer may negotiate higher slotting fees, for example, rather than lower per-unit prices, so that output in the downstream market may not be higher, final consumers may not enjoy lower prices, and there may not be increases in either total or consumer surplus.14 In general, if premerger supply was below the competitive level, it is true that a reduction in marginal price at all quantities will lead to greater output and greater surplus,15 but the fact that a buyer captures more of the surplus does not imply that the relevant marginal price falls, that output is closer to the competitive level, or that total surplus has increased. The contract negotiations may also lead to other contract terms that, a priori, could be either procompetitive or anticompetitive. Examples of nonprice contractual terms between buyers and suppliers include exclusivity requirements,16 supplier contributions for promotional expenses, supplier acceptance of the return of unused or unsold supplies from the buyer, long delays before payment is due, slotting allowances, and listing fees.17 Other contractual terms may produce nonlinear pricing schedules, such as volume rebates and most-favored-customer clauses. 12  Some discussions of classic monopsony power do extend analysis to this situation in which there is power on both sides of the market. See Blair and Harrison (2010, c­ hapter 6), for example. However, in these cases, the authors acknowledge that “[u]‌nder these circumstances, the usual marginal analysis fails us. The final outcome depends on bargaining over how to split the surplus that is available in the market” (p. 141). 13  Note that for this discussion, we are assuming that the merger of buyers does actually increase buyer power, in the sense that it increases the share of surplus captured by the combined buyers. There are certainly models where this is not the case. See, for example, Chipty and Snyder (1999, pp. 326–40) and Raskovich (2003, pp. 405–26). 14  See Marx and Shaffer (2007, 2010) and Miklós-Thal, Rey, and Vergé (2011) for theoretical analyses of buyer power and slotting fees. For theoretical models of how retail buyer power changes the price structure in retail markets, see Chen (2003), Erutku (2005), and Wang (2010). 15  We assume that the negotiated p(q) schedule has the property that the second derivative of the schedule is negative. 16  Such exclusivity requirements need not be anticompetitive if they, for example, enable a buyer to overcome hold-up concerns and thus to make efficient investments that would not have been made absent such requirements. 17  An example of this more expansive definition of buyer power is the alleged ability of the State of Texas to influence the content of grade school and high school textbooks. See Carstensen (2010, p. 17, note 70).

Buyer Power in Merger Review   535

22.2.2.2.  Potential Efficiencies from an Increase in Buyer Power A merger among buyers that creates or increases buyer power could result in efficiencies if buyers and sellers were not able to arrive at an efficient contract (in which marginal price equals marginal cost at the competitive level) premerger, but are able to do so (or at least come closer to doing so) postmerger. This might occur for several reasons. First, the merger could cause a reduction in transactions costs. It may simply be easier for a seller to work out, monitor, and enforce a single, efficient contract with one buyer than it was with multiple buyers. Second, if there is more surplus at stake in negotiations with a larger buyer, there may be a sufficient return to incentivize firms to pay the transactions cost to work out an optimal contract. Third, by combining the information possessed by multiple buyers and reducing the number of buyers that a seller must study, a buyer merger can solve the asymmetric information problems that might otherwise prevent the negotiation of optimal contracts.18 Mergers among buyers may also generate efficiencies in other ways. Dobson and Chakraborty (2008, 336) observe that “retail buyer power may bring benefits to consumers when it spurs supplier competition and efficiency that allow for lower retail prices.” Kirkwood (2012) argues that a buyer merger may be procompetitive either because it allows the merging firms to achieve economies of scale, reduce transaction costs, or avoid duplication of facilities, or because it allows the merging firms to exercise negotiating power in a procompetitive manner by reducing supplier market power and thereby lowering their prices. The theoretical models of Inderst and Wey (2003, 2007, 2011) demonstrate that buyer power can spur dynamic efficiency by inducing suppliers to reduce their marginal costs. Empirical evidence consistent with (but not necessarily proving) these procompetitive effects from buyer mergers comes from Fee and Thomas (2004), who find that horizontal buyer mergers during the period 1980–97 were followed by a deterioration in their suppliers’ margins, and from Bhattacharyya and Nain (2011), who find that horizontal mergers from the period 1984–2003 were followed by large price declines for the suppliers of the merging firms (as well as by an increased likelihood of a wave of mergers among those suppliers). More generally, many of the contractual terms discussed above, which might result from negotiations where there is buyer power, can be procompetitive (although as noted below they can also create competitive harm). As the American Antitrust Institute (2008) explains, such contractual terms may generate efficiency benefits by solving free-rider problems, encouraging new investment, facilitating entry into new markets, allowing for different promotional strategies in different markets, achieving economies of scale in distribution or production, alleviating capital market imperfections, and allowing for uniformity and quality standardization.

18 

Note that analogous arguments could be made in the case of a merger of suppliers. So an alternative statement of our conclusion is that, in markets characterized by bargaining between buyers and sellers, mergers of buyers or suppliers can create efficiencies if they enable the formation of contracts that are closer to the socially optimal ones.

536   Dennis W. Carlton, Mary Coleman, and Mark Israel

22.2.2.3.  Potential Competitive Harm from an Increase in Buyer Power Competitive harm from an increase in buyer power other than that associated with monopsony power might occur if the new contractual terms result in less total surplus (albeit more for the buyer) than prior to the increase in buyer power. As the American Antitrust Institute (2008) explains, contractual terms negotiated in such contracts may generate negative welfare effects by restricting competition at either the buyer or supplier level. Dobson and Chakraborty (2008, 336) observe that “retail buyer power may harm consumer welfare when it restricts and distorts competition to the extent that it leads to reduced choice, quality, and innovation and/or higher prices for consumers.” Kirkwood (2012) describes five potential ways in which buyer mergers might cause harm to downstream competition. The five potential harms listed by Kirkwood are (1) raising rivals costs; (2) so-called waterbed effects (i.e., the merged firm extracts price cuts or other concessions from suppliers, who react by increasing prices to other buyers, which in turn allows the merged firm to raise its own prices in downstream markets); (3) higher downstream concentration as lower prices drive smaller buyers from the market; (4) reduced choice as smaller buyers are driven from the market; and (5) a merged firm that is less efficient due to lower prices. The issues raised by Kirkwood are generally vertical concerns where asymmetries in dealing with upstream suppliers disadvantage some (frequently smaller) downstream firms, which may soften competition and thus result in higher prices downstream. As with many vertical theories, it is not clear when these potential harms might arise or how common they are likely to be.19 Moreover, identifying instances where such effects may occur is likely to be very difficult.20

19 

For example, some of the discussion surrounding the failed AT&T/T-Mobile merger expressed concern that the combined firm—together with Verizon Wireless—would exert so much control over upstream wireless handset manufacturers that other wireless carriers, such as Sprint, would be disadvantaged. Of course, it is also possible that upstream handset manufacturers would have an increased incentive to work with smaller carriers such as Sprint in order to gain a more advantageous negotiating position vis-à-vis all wireless carriers and thus to offset any merger-related increase in buyer power. Hence, this example stresses the importance of evaluating such buyer power effects via the lens of bargaining theory and the need for empirical analysis to resolve these competing theories. (The authors worked on this transaction for ATT and T-Mobile.) 20  Kirkwood (2012) also argues that buyer mergers can cause harm to upstream competition in several ways. The potential harms included (1) reduced innovation upstream as suppliers have lower profits; (2) reduced product variety as suppliers reduce variety due to lower profits; (3) creation of upstream monopsony as the buyer concentrates its purchases in one supplier; (4) possible creation of future monopsony power if the supplier’s selling power is reduced; and (5) coordinated behavior among suppliers in response to the exercise of buyer power. The evidence on the first two points is mixed. Peters (2000, p. 13) examines the German automobile industry and finds that “buyers’ pressure on input prices reduces suppliers’ innovation expenditures and their incentive to develop new products.” However, Dobson and Chakraborty (2008) note that there is little evidence that consumers have been harmed due to buyer mergers with respect to innovation in the UK groceries market. Inderst and Wey (2011) analyze a bargaining model in which the presence of a large buyer spurs a supplier to invest in lowering its marginal cost. Inderst and Shaffer (2007b) and Chen (2006) conduct theoretical analyses of the effect of buyer power on product variety, and both find a welfare trade-off between lower consumer prices and reduced product diversity. However, Dobson and Chakraborty (2008) observe that there is little

Buyer Power in Merger Review   537

Although most of the theories described thus far describe harms that may arise if powerful buyers use their power to the detriment of upstream suppliers, an alternative theory may be that powerful buyers may actually facilitate upstream market power, perhaps via upstream coordinated behavior. Consider an example. Imagine that a large buyer agrees not to buy low-priced products from entrants and thereby limits entry into an industry. As a result, the prices of those products are higher than they would otherwise be because the upstream industry participants enjoy some protection from entry. The large buyer is “paid off ” for its efforts by being offered a low price for the products while the large buyer’s rivals must pay the higher price. A closely related example may occur if there are powerful buyers in a standard setting organization (see, e.g., Gilbert 2011). Such buyers may be able to tell specific suppliers that they will support their preferred standard as long as those suppliers agree to provide products using that standard to the buyer at a low price, while withholding them or charging higher prices to other buyers including future entrants into the market.21

22.2.2.4.  Approach Taken in the US Merger Guidelines We have argued that “buyer power” should not be limited to classic monopsony power, but also should include negotiating power over price and nonprice contractual terms with suppliers (in situations where contracts are negotiated between buyers and sellers, neither of whom is a price taker). As noted, buyers may exercise negotiating power in both procompetitive and anticompetitive ways. Therefore, an analysis of a buyer merger that fails to take into account how the merger will affect the merged firm’s negotiating power misses a potentially important competitive effect. However, Kirkwood (2012) claims that, in conducting their analyses of buyer mergers, the DOJ and Federal Trade Commission (FTC) consider only the “buyer power as monopsony power” theory. In particular, the DOJ/FTC’s Horizontal Merger Guidelines (HMG) adopt an approach to mergers between buyers that largely treats them as the evidence that consumers have been harmed due to buyer mergers with respect to product choice in the UK groceries market. Kirkwood gives no evidence to establish the third and fourth points. As to the fifth point, although there is some evidence that powerful buyers can induce coordination among suppliers, we note that such behavior itself would be subject to antitrust review and thus stopping a merger based on such concerns may not be warranted. Raper et al. (2000) examined the US leaf tobacco market, which had cigarette manufacturers with monopsony power on one side of the market and producers, which were organized as a cartel through a marketing quota program administered by the US Department of Agriculture, on the other. The authors note that cigarette manufacturers played a large role in determining national quota levels and argue that the marketing quota program did not protect leaf producers from the exertion of monopsony power by the cigarette manufacturers. More generally, Carstensen (2008) observes that numerous laws attempt to constrain how buyer power can be exercised in agricultural, labor, and retail markets. For a model of bilateral collusion, see Bloch and Ghosal (2000). 21  A similar example occurs when one of the merging buyers is vertically integrated with an upstream supplier, in which case the concern may be that the merged buyer will have increased ability to favor the affiliated upstream supplier to the detriment of other suppliers. For example, if two cable television operators—one of which also owns one or more television networks—merge the concern may be that the combined cable operator will have more ability to favor its affiliated programming to the detriment of competing television networks.

538   Dennis W. Carlton, Mary Coleman, and Mark Israel “mirror image” of mergers between sellers, explaining that the DOJ/FTC will use essentially the same framework for evaluating seller mergers when evaluating buyer mergers. The section of the HMG on buyer mergers (Section 12) does not even mention negotiating power (DOJ and FTC 2010, Section 12).22 The economic literature discussed above suggests that, to the extent the HMG fail to consider negotiating power, the DOJ and FTC may be missing a potentially important competitive effect of buyer mergers. The economic literature suggests that it may be important for the DOJ/FTC to distinguish between buyer mergers that simply create monopsony power from those that create negotiating power in addition to monopsony power. There are many more avenues through which the latter type of mergers may be procompetitive. In practice, our experience is that the FTC and DOJ take the broader definition of market power among buyers into account. Although there is general agreement in the economic literature that negotiating power can be an important competitive effect that should be considered in the evaluation of buyer mergers, there is sharp disagreement as to whether the tools used for evaluating mergers between sellers are inadequate or unsuitable, without some modification, for evaluating mergers between buyers. For example, Carstensen (2004) argues that, compared with sellers, buyers have different incentives to cheat on tacit coordination,23 have different opportunities to exercise market power,24 and face less deterrence to the exercise of market power. These differences, if significant, would have a number of consequences for the evaluation of buyer mergers. However, there is not a general consensus as to this point of view—particularly as to whether buyer power is more likely at lower levels of share or concentration than seller power. For example, Schwartz (2004) claims that no systematic economic evidence shows that buyer power concerns arise at lower concentration levels than for seller power. Similarly, Scheelings and Wright (2006, 231) conclude: “There is little serious economic evidence that lower market shares are necessary for buyers to obtain market power.” Indeed, some commentators reach the opposite conclusion. For example, Areeda and Hovenkamp (2006, pars. 981b, 982) point out that price fixing by buyers is observed much less frequently than price fixing by sellers, which is suggestive that buyers have more difficulty conspiring than sellers, and thus they assert that concentration thresholds should be higher for buyer mergers than for selling mergers. 22  Regardless of how one may interpret the HMG, our experience is that the government agencies do consider the effect of increased bargaining power in their economic analysis of some cases. 23  According to this argument, the reason why buyers engaged in coordinated monopsony pricing may be less inclined to cheat on the tacitly coordinated price is that, by defecting from the tacitly coordinated price, the cheater pays a higher price for the input and thus is put at a competitive disadvantage relative to the buyers who do not cheat. In contrast, a seller who cheats on tacit coordination charges a lower price than its noncheating competitors charge and thus gains an advantage over them. However, an alternative point of view would be that, in either case, the cheater accepts slightly reduced margins in order to sell substantially more output, increasing total profits. 24  According to this argument, a retailer may need to control a smaller percentage of the total number of outlets in order to obtain better terms from suppliers than a seller would need to obtain higher prices from its customers.

Buyer Power in Merger Review   539

The evaluation of buyer mergers may raise additional issues in some special situations. For example, there are numerous laws that attempt to constrain the exercise of buyer power in agricultural, labor, and retail markets. Examples include the Agricultural Marketing Agreements Act, minimum wage laws, and the Robinson-Patman Act. As Carstensen (2008, 275) points out, “Any comprehensive analysis of buyer power and the appropriate type and level of its regulation requires careful attention to the interaction between the kinds of conduct authorized or permitted by such regulations and the kinds of controls that antitrust law can provide” and, “Basically, antitrust is rarely if ever the sole basis on which effective, socially desirable competition policy can rely.” A second special situation concerns buying groups. The DOJ has publicly stated that it will not object to the creation of “buying groups” that account for up to 35% of the total volume of purchases in certain markets. Carstensen (2010a, 8) claims that this market share threshold is too high since buyer power in a buying group setting, just as in a buyer cartel setting, “can exist at lower levels of apparent concentration than is customarily assumed” and, more generally, he contends that “legitimate buying groups create significant competitive risks meriting stricter scrutiny of such organizations than they have traditionally been accorded.” A special type of buying group is the standard-setting organization (see, e.g., Sidak 2009; Gilbert 2011). As Skitol (2005, 728)  observes, “Information technology standard-setting processes are one context in which carefully structured concerted buying power could be procompetitive and where clarification of antitrust implications would be highly desirable.”

22.2.2.5.  Need for Empirical Analysis It is clear that an increase in buyer power as a matter of theory can either have procompetitive or anticompetitive effects. Moreover, the empirical evidence on the effects of buyer concentration is mixed. For example, Bates and Santerre (2008) focus on contracting between health insurers and hospitals and find that higher health insurer concentration is associated with more hospital services, which they interpret as evidence that greater health insurer concentration results in increased “monopoly-busting” power rather than the greater exercise of monopsony power. Kelly and Gosman (2000) find that, as expected, buyer concentration reduces manufacturer profitability, but somewhat surprisingly, they find that this effect primarily occurs when the manufacturers’ industry is competitive, not oligopolistic. Given these uncertainties, how does one determine whether the increase in buyer power is likely to generate efficiencies or competitive harm? That is, how can one determine the likely effect of an increase in buyer power on the marginal price of a product? One can try to look at existing contracts, as well as company documents on how the merger may change those contracts, in order to ask directly whether the increase in buyer power is likely to lead more efficient contracts, including the use of nonlinear contracts that may permit the marginal price to be set at marginal cost. But forming a prediction of contractual changes is likely to be quite difficult in practice. Hence, we recommend strong emphasis on industry-specific empirical evidence (much as we

540   Dennis W. Carlton, Mary Coleman, and Mark Israel recommend the use of such evidence in seller mergers). In the remainder of this section, we propose some of the types of tests that might be used. First and most basically, to the extent that there have been previous mergers between buyers in the industry, we would encourage strong reliance on the outcomes of these mergers. In particular, the effect of previous mergers on industry output (both input purchases and downstream output) is likely to be particularly informative. In particular, evidence of increased industry output that followed previous buyer mergers suggests important efficiencies from such mergers. Absent previous buyer mergers, alternative sources of variation should be sought to test whether efficiencies are likely to exist as a result of a merger of buyers. One possibility arises from noting the analogy between the efficiencies being discussed here (for horizontal buyer mergers) and efficiencies commonly claimed in vertical mergers. In each case, the argument is that, premerger, buyers and sellers cannot arrive at optimal contracts (with marginal price equal to marginal cost at the competitive output) and that the merger can help to overcome these contractual difficulties. Hence, if there have been previous vertical mergers in the industry, evidence of improved outcomes (e.g., greater industry output) following these mergers supports the claim that there are important contractual inefficiencies to be overcome. Alternatively, if there are certain buyers who do not participate in all geographies or product lines (or have lower share in certain geographies or product lines), then one could attempt to use cross-sectional variation to measure the effect of variation in the number (or market share) of buyers on marketplace outcomes.25 In this case, it will be important (and potentially quite difficult) to control for the other differences across regions or products or to use exogenous sources of variation in buyer counts as instruments.

22.3.  Role of Buyer Power in Assessment of Supplier Mergers In this section, we consider the argument that powerful buyers may be able to mitigate price effects from a merger of two upstream sellers.26 As an initial matter, we note that 25  One example of such empirical analysis can be found in the FTC decision to approve the acquisition of Medco Health Solutions by Express Scripts, Inc. One concern raised in the investigation of this transaction was whether “the combined firm could exercise monopsony power, driving drug dispensing fees so low that they would threaten the important services offered by local pharmacies.” The FTC ultimately rejected this concern, at least in part because “the data reveal that there is little correlation between PBM [pharmacy benefit manager] size and the reimbursement rates paid to retail pharmacies.” (See Statement of the Federal Trade Commission Concerning the Proposed Acquisition of Medco Health by Express Scripts, Inc. FTC File No. 111-0210, April 2, 2012.) (Some of the authors have worked on transactions in this industry.) 26  The economic forces in this section are analogous to those in section 22.2. If buyers and sellers bargain with one another, it need not be the case that mergers of suppliers always reduce total surplus, as

Buyer Power in Merger Review   541

it is far from certain that the presence of powerful buyers will prevent (or even reduce) the price effects that arise from a merger of two suppliers. It is certainly natural to expect that powerful buyers can command lower price levels than other buyers. However, what matters for merger analysis is the change in prices due to the merger. Economic theory indicates that powerful buyers are already likely fully utilizing their power premerger. From this starting point, a merger of suppliers (at least one leading to a significant increase in suppliers’ relative bargaining power) leaves all buyers, even powerful ones, with fewer options to use in negotiations and thus may lead to higher prices even in the presence of very powerful buyers. Indeed, it is entirely possible that the presence of powerful buyers could lead to larger price effects from a merger of suppliers than would occur in the absence of powerful buyers. Consider the following simple example: There are only two sellers. Absent powerful buyers, the sellers recognize their mutual interdependence and set monopoly prices. Hence, a merger will have no effect. Now suppose powerful buyers can prevent monopoly pricing by playing the sellers off against each other (as is often claimed when buyer power is used as a merger defense). In this case, in the presence of powerful buyers, premerger pricing will be at duopoly levels, while postmerger pricing will be at monopoly levels. Hence, the presence of powerful buyers may increase the adverse pricing effects that flow from a merger. Though this is a specialized example, it illustrates that there is no theoretical necessity that the presence of powerful buyers must always lessen the price effects from a merger.

22.3.1.  Conditions Where Power Buyers Can Prevent Anticompetitive Effects from a Supplier Merger The economic literature as well as the horizontal merger guidelines of the United States, European Commission, the United Kingdom, and Canada, and the International Competition Network (ICN) Merger Guidelines Workbook discuss several necessary and/or sufficient conditions for powerful buyers to prevent anticompetitive effects from a supplier merger. Three necessary conditions appear to be generally agreed upon. The first condition is that one or more of the buyers must have both the ability and the incentive to constrain an exercise of market power by the merged firm (DOJ and FTC 2010, 27, Section 8; Canada Competition Bureau 2011, 31–32, pars. 8.1–8.2). The “ability” condition is more likely to be satisfied if either (1) buyers can credibly threaten to switch to alternative sources of supply within a reasonable time frame, or 2) buyers can impose costs on the merged firm. In particular, the “ability” condition is more likely to be satisfied if buyers can (a) immediately switch to other suppliers; (b) credibly threaten to self-supply via vertical integration into the upstream market; (c) credibly threaten to induce expansion of one or more of the merged firm’s competitors or to sponsor entry by a potential supplier not currently in the market; (d) refuse to buy (or delay buying) other products produced by the merged firm; (e) refuse to purchase the merged firm’s products in other geographic markets where competitive conditions are different; or (f) impose costs on the merged firm by, for

542   Dennis W. Carlton, Mary Coleman, and Mark Israel example, giving its products less favorable retail placement (Competition Bureau Canada 2011, 31, par. 8.1; European Commission 2004, par. 6.5). Notice that buyer size, by itself, is not on the list. Although size can be important in certain situations—for example, if a large buyer can provide enough sales to one supplier to immunize that supplier from any punishment following deviation from coordination, this can help to undermine the likelihood of coordination—the simple fact that a buyer accounts for a large fraction of the merged firm’s sales of a product does not imply that the buyer has negotiating power.27 Indeed, some of the actions on the list are more likely to be feasible for small buyers (e.g., the ability to immediately switch suppliers), although others are certainly easier for large buyers (e.g., the ability to induce entry by offering the entrant a large volume of business).28 Also note that some of the actions on the list, when taken by one buyer, are likely to benefit all buyers (e.g., the inducement of entry),29 while others benefit only certain buyers (e.g., switching suppliers or raising the costs of the merged firm) (Inderst and Shaffer 2007a, 20; UK Competition Commission and Office of Fair Trading 2010, par. 5.9.5). Buyers must not only have the ability to constrain the merged firm’s pricing, they must have the incentive as well. In other words, restraining supplier pricing must be more profitable than simply using whatever buyer power exists to negotiate a share of the merged firm’s monopoly profits (Hovenkamp 2005, 544). The “incentive” condition is more likely to be satisfied if (1) buyers cannot pass-through any price increases downstream (ABA Section of Antitrust Law 2008, 219), and (2) actions by one buyer to constrain the exercise of market power by the merged firm (e.g., induce entry) do not produce benefits (e.g., lower price) that spill over to its competitors (European Commission 2004, par. 66). Evidence demonstrating that both parts of the “ability and incentive’ condition are satisfied would include prior dealing between buyers and one efficiencies (e.g., reduced transactions costs) resulting from the merger may lead to sufficiently improved contracts such that total surplus rises. However, in this section, we presuppose that, absent powerful buyers, the merger of sellers would lead to inefficiently higher prices and then ask whether powerful buyers can prevent this effect. 27  Indeed, Raskovich (2003) shows that if a buyer becomes large enough to be pivotal—meaning that without the business of that buyer a seller would shut down some or all of its operations—then the buyer may have less power to extract surplus from a supplier. The reason is that a buyer’s ability to affect a supplier’s decision about whether to operate or shut down implies that the supplier’s fixed costs are at least partially marginal with respect to that buyer, whereas they would be sunk for smaller, nonpivotal buyers. Hence, the supplier effectively “charges” the buyer for some or all of its fixed costs and thus may not be willing to give the buyer as much surplus in negotiation. Put differently, large buyers may have to “cover” more of the fixed costs that are common to serving all buyers. 28  See UK Competition Commission and Office of Fair Trading (2010, par. 5.9.5). However, although excess capacity in the upstream industry facilitates switching by sufficiently small buyers, if there is also a fixed cost to switching, then small buyers will not necessarily find it less costly to switch suppliers than large buyers will. See Inderst and Shaffer (2007, p. 9, note 20). 29  Indeed, the combination of powerful buyers and likely entrants may make for a much more powerful line of argument against merger-related price increases than either condition would make on its own. Powerful buyers may be able to facilitate entry (or at least credibly threaten to do so) that otherwise would not occur, and the existence of likely entrants may give powerful buyers a credible threat to use in negotiations.

Buyer Power in Merger Review   543

or more of the merger parties that show the strength of the former’s bargaining power (Competition Bureau Canada 2011, 31, par. 8.2). As one example, for many mergers involving grocery products, retailers could theoretically provide a constraint on pricing. Sales of many grocery products are heavily concentrated in a relatively small number of retailers (with Walmart often comprising a very high percentage of sales). If several large retailers threatened to discontinue a product following an announced price increase, this might make a price increase unprofitable, where it would not be absent this threat. The question then becomes whether such a threat is credible—that is, would the retailer be able to discontinue the brands at issue without risking losing consumer traffic that would make the price increase unprofitable. The answer will in turn depend on what other options are available (including its own private label), how strong is the brand, and whether the product is an important consumer draw. A retailer might have less credibility threatening to discontinue Coca-Cola, for example, than a brand of dried pasta. Another issue is whether the retailer has the incentive to exercise its buyer power. If the retailer just passes along the price increase (and faces a relatively inelastic demand curve), then it has less incentive to try to maintain lower prices. Thus, assessing whether retailers might provide an additional competitive constraint in consumer products mergers depends on the specific nature of the products at issue and the primary retailers selling the products. In many cases, the FTC and DOJ do not consider retailers as being able or willing to constrain the postmerger pricing of consumer products manufacturers. However, in some cases, the constraint can be important. For example, if the merger involves one large brand and a small brand, there may be little merger-induced change in incentives for pricing of the large brand but a large change for the small brand. However, a price increase on the small brand may threaten its distribution, and if there are other brands that could replace the small brand, this might make the price increase unprofitable. The second necessary condition is that buyer negotiating power must not be limited to offsetting the adverse effect of the merger for a strict subset of customers (e.g., only the large buyer), but rather must constrain the merged firm’s pricing either to all buyers or make the potential competitive harm small enough to be offset from the efficiencies from the transaction (DOJ and FTC 2010, 27, Section 8; European Commission 2004, par. 6.7; ICN 2006, 42). This condition is more likely to be satisfied if (1) there are no bilateral negotiations between customers and suppliers, (2) the market price of the input is transparent to all suppliers and customers, or (3) there are enough large buyers that even if only the large buyers survive, competition will remain intense enough to protect consumers (UK Competition Commission and Office of Fair Trading 2010, par. 5.9.7) We note that this condition may conflict with the “no spillover” incentive condition just mentioned. The third necessary condition is that countervailing buyer power must remain effective following the merger (European Commission 2004, par. 6.7; UK Competition Commission and Office of Fair Trading 2010, par. 5.9.8; ICN 2006, 42). This condition may not be satisfied for a merger of suppliers X and Y if, for example, a buyer who

544   Dennis W. Carlton, Mary Coleman, and Mark Israel historically had negotiating power in individual negotiations with X and Y because it could credibly threaten to shift purchases to the other, cannot find an alternative supplier Z to whom it can credibly threaten to shift purchases after the merger (Competition Bureau Canada 2011, 31–32, par. 8.2). Two other necessary conditions are sometimes mentioned in the literature but are not as widely accepted. One is that there must be no evidence that the merging firms’ industry has been historically prone to coordinated behavior (ABA Section of Antitrust Law 2008, 218). The other, advocated by Hovenkamp (1991, 1370), is that the buyer market must be competitive. The latter is especially controversial and, in fact, is inconsistent with the sufficient condition advocated by Campbell (2007) discussed below. Several sufficient conditions for the presence of powerful buyers to prevent the adverse effects of a merger of suppliers have been suggested in the literature, but none is generally accepted. One possible sufficient condition is if each buyer has the ability and the incentive to constrain an exercise of market power by the merged firm: However, such a condition is likely to be extremely difficult to establish in practice. Campbell (2007) argues that a sufficient condition for permitting a merger to monopoly on the sell side is when the buy-side market consists of a single firm (i.e., a monopsonist). His argument is that bilateral monopoly yields the efficient (i.e., competitive) output level and thus, given the existence of a monopsony on the buy side, the optimal market structure for the sell side is a monopoly. Note that Campbell’s sufficient condition is logically inconsistent with Hovenkamp’s necessary condition that the buyer market be competitive. Campbell’s argument has been criticized by Baker, Farrell, and Shapiro (2008), who counter that modern economic theory routinely finds that bilateral negotiations often do not yield efficient outcomes. Another possible sufficient condition is the presence of “disruptive” buyers. Indeed, George Stigler, in his seminal paper “A Theory of Oligopoly,” assigns a special role to large buyers for their ability to disrupt a cartel (Stigler 1964). However, one problem with implementing this ideal is in verifying the existence of such buyers prior to merger consummation (large size is sometimes viewed as a proxy) (Dick 1996). In addition, the presence of “disruptive” buyers will not necessarily benefit other buyers, as a disruptive buyer may not have the incentive to curb coordinated seller effects on other buyers. Even given the necessary and sufficient conditions just identified, some argue that there are problems in incorporating buyer-negotiating power as a mitigating factor in a merger of sellers and thus the role of buyer power in seller mergers should be limited. There is not, however, general agreement on this point and merger guidelines from most antitrust authorities internationally take into account the potential mitigating effect of buyer power. Areeda and Hovenkamp (2006) argue that the measurement of buyer power poses intractable problems except in those circumstances where buyer power is so significant that sellers clearly are forced to behave competitively. Kirkwood (2012) counters that this measurement issue is overstated because it can be addressed by examining

Buyer Power in Merger Review   545

the tactics available to each buyer to exert leverage over its suppliers and investigating whether those tactics would continue to be available and effective after the merger.30 Areeda and Hovenkamp (2006) also argue that powerful buyers may find it more profitable to share in the merged firm’s excess profits than in negotiating a price closer to the competitive level (consistent with the point made above that powerful buyers may actually help to facilitate the upstream exercise of market power).31 This possibility is also covered by the “incentive” necessary condition discussed above and thus is part of the analysis of whether buyer power should be considered a mitigating factor.

22.3.2.  Effect on Small Buyers Even if evidence indicates that powerful buyers can protect themselves against merger-related price effects, there may still be effects on smaller buyers (Carlton and Israel 2011, 135). An important question for antitrust policy to grapple with is whether to be concerned about the price increases on smaller buyers—even if those price increases drive the small buyer out of business—as long as large buyers survive and compete. We propose the following high-level answer to this question: If there exists a reasonably large number of large buyers, such that the downstream market would remain competitive even without price constraints from the smaller buyers, then there is little cause for concern. Although small buyers may be harmed, total (and consumer) surplus will not diminish. However, there is cause for concern if competition among the larger buyers is imperfect, such that an increase in costs for the smaller buyers would create a “pricing umbrella” for larger buyers to raise prices to end consumers. Although this general framework should serve as the guide to antitrust policy, it may be difficult to determine how competitive the market would be if only large buyers survived. More generally, it is likely to be difficult, as a matter of theory, to determine the extent to which differential cost changes on small and large buyers will affect downstream prices. Hence, appropriate empirical tests are called for. One fruitful test may be to examine situations in which differential cost changes (perhaps due to different production locations, differential effects of exchange rates, etc.) have affected different firms and to study the ultimate effect on downstream prices.

22.3.3.  Need for Empirical Analysis The factors listed above, which determine whether powerful buyers will or will not offset price increases following mergers of suppliers, are not primarily about the size of buyers 30 

Similarly, the US Horizontal Merger Guidelines explain that the DOJ and FTC will examine the choices available to powerful buyers and how those choices would likely change due to the merger. See DOJ and FTC (2010, p. 27, Section 8). 31  Kirkwood (2012) concedes that this may occur in some cases, but in most cases it does not.

546   Dennis W. Carlton, Mary Coleman, and Mark Israel (or other easy-to-observe metrics of buyer power) but rather about specific institutional details with regard to the nature of bargaining between buyers and sellers. As such, the extent to which powerful buyers can offset potential merger effects may be very hard to intuit from theoretical analyses or easy-to-observe industry or firm characteristics. Hence, determining whether the existence of powerful buyers mitigates competitive harms from a merger of suppliers calls for empirical analysis. Several types of empirical evidence are likely to be informative about whether mergers of suppliers are or are not likely to have price effects in the presence of powerful buyers.32 Most basically, to the extent that there have been previous mergers between sellers in the industry—at times when the mix of buyers was similar to today—we would encourage that strong reliance be placed on the outcomes of these mergers, both for input prices and for downstream prices, as these outcomes capture the relevant aspects of the underlying game. It may also be informative to study mergers in similar industries, or in the same industry in other geographic areas, if those alternative markets feature buyers who are positioned similarly to those in the market in question. If no relevant prior mergers exist, alternative sources of variation should be sought to test whether the presence of powerful buyers will mitigate or eliminate merger price effects. If there are certain sellers who are not active in certain geographies, product lines, and so on, then the effect of variation in seller count (or market share) on prices may be highly informative in ascertaining whether buyer power matters. To draw inferences from such cross-sectional variation, however, requires that sufficient controls are available for the other differences across regions and products or that there are sources of exogenous variation in seller counts that can be used as instruments. Other less direct tests of efficiencies from merger may also be possible. For example, if most buyers (and especially powerful buyers) negotiate with all available suppliers when making purchase decisions, this indicates that it is helpful to negotiate with the full set of sellers and thus that supplier mergers are more likely to have significant pricing effects. However, if many buyers negotiate with only a subset of suppliers, this may indicate that not all suppliers are required to obtain competitive prices and thus reduce concerns over merger-related harms. Alternatively, note that, according to standard bargaining models, changing the number of sellers matters to the extent that it affects the outside options that are available to a buyer during negotiations with a given seller. Hence, evidence on other how other factors that affect buyers’ outside options (e.g., the ability to use substitute production processes that do not require the input in question or availability of in-house sources of supply) affect prices may allow the calibration of a model of bargaining effects, which could then be applied to the specific supplier merger at issue.

32  In addition to the empirical tests described here, it may also be informative to ask buyers for their views of the merger, although the reasons for the buyers favoring or disfavoring a transaction must be considered. For example, buyers that are relatively advantaged by the merger might favor the transaction even if prices rise because their profits will increase.

Buyer Power in Merger Review   547

22.4. Conclusion Despite the relative lack of attention that is paid to the topic of powerful buyers in the Horizontal Merger Guidelines, the role of powerful buyers raises thorny questions for merger review. Such questions include the extent to which mergers of buyers can lead to more efficient contracts with suppliers versus anticompetitive harm and the extent to which the presence of powerful buyers can offset higher prices from supplier mergers. The difficulties arise from the fact that answers to these questions turn on specific details of underlying bargaining games, which are hard to observe in practice. Hence, answers to these questions will depend on empirical tests applied to the specific industries and firms in question. In this chapter, we have suggested the types of empirical tests that may be most fruitful. We hope that subsequent literature will continue to analyze these complex issues, develop appropriate empirical tests, discuss methodological issues with the implementation of the tests, and present results for specific mergers in a range of industries.

Acknowledgments The authors wish to thank Greg Pelnar and Dave Grothouse for extremely useful contributions to this chapter, as well as Roger Blair and Daniel Sokol for excellent editorial suggestions. Any remaining errors or omissions are, of course, our fault.

References ABA Section of Antitrust Law. 2008. Mergers and Acquisitions. 3rd ed. Adilov, Nodir, and Peter J. Alexander. 2006. Horizontal Merger: Pivotal Buyers and Bargaining Power. Economics Letters 91: 307–11. Areeda, Phillip E., and Herbert Hovenkamp. 2006. Antitrust Law:  An Analysis of Antitrust Principles and Their Applications. 3rd ed. New York: Aspen. Baker, Jonathan B., Joseph Farrell, and Carl Shapiro. 2008. Merger to Monopoly to Serve a Single Buyer: Comment. Antitrust Law Journal 75: 637–46. Bates, Laurie J., and Rexford E. Santerre. 2008. Do Health Insurers Possess Monopsony Power in the Hospital Services Industry? International Journal of Health Care Finance and Economics 8(1): 1–11. Bhattacharyya, Sugato, and Amrita Nain. 2011. Horizontal Acquisitions and Buying Power: A Product Market Analysis. Journal of Financial Economics 99(1): 97–115. Blair, Roger D. 2010. Merger to Monopsony:  An Efficiencies Defense. Antitrust Bulletin 55(3): 689–98. Blair, Roger D., and Jeffrey L.  Harrison. 2010. Monopsony in Law and Economics. New York: Cambridge University Press.

548   Dennis W. Carlton, Mary Coleman, and Mark Israel Blair, Roger D., and Richard E. Romano. 1997. Collusive Monopsony in Theory and Practice: The NCAA. Antitrust Bulletin 42(3): 681–719. Bloch, Francis, and Sayantan Ghosal. 2000. Buyers’ and Sellers’ Cartels on Markets with Indivisible Goods. Review of Economic Design 5(2): 129–47. Boal, William. 2009. The Effect of Minimum Salaries on Employment of Teachers: A Test of the Monopsony Model. Southern Economic Journal 76(3): 611–38. Campbell, Tom. 2007. Bilateral Monopoly in Mergers. Antitrust Law Journal 74(3): 521–36. Capps, Cory S. 2009. Buyer Power in Health Plan Mergers. Journal of Competition Law and Economics 6(2): 375–91. Carlton, Dennis W., and Mark Israel. 2011. Proper Treatment of Buyer Power in Merger Review. Review of Industrial Organization 39(1): 127–36. Carlton, Dennis W., and Jeffrey M. Perloff. 2005. Modern Industrial Organization. 4th Edition. Boston: Pearson / Addison Wesley. Carstensen, Peter C. 2004. Buyer Power and Merger Analysis: The Need for Different Metrics. Prepared for the Federal Trade Commission workshop on merger enforcement. Carstensen, Peter C. 2008. Buyer Power, Competition Policy, and Antitrust: The Competitive Effects of Discrimination among Suppliers. Antitrust Bulletin 53(2): 271–331. Carstensen, Peter C. 2010. Horizontal Merger Guidelines:  The Omitted Dimension of Buyer Power Comments Submitted to the FTC and DOJ. University of Wisconsin Law School Legal Studies Research Paper Series Paper No. 1104. Chen, Zhiqi. 2003. Dominant Retailers and the Countervailing-Power Hypothesis. Rand Journal of Economics 34(4): 612–25. Chen, Zhiqi. 2006. Monopoly and Product Diversity: The Role of Retailer Countervailing Power. Photocopy, Carleton University. Chen, Zhiqi. 2007. Buyer Power:  Economic Theory and Antitrust Policy. Modern Industrial Organization. In Richard O. Zerbe and John B. Kirkwood, Research in Law and Economics, vol. 22: 17–40. Amsterdam: Elsevier; Oxford: JAI. Chen, Zhiqi. 2008. Defining Buyer Power. Antitrust Bulletin 53(2): 241–47. Chipty, Tasneem, and Christopher M.  Snyder. 1999. The Role of Firm Size in Bilateral Bargaining: A Study of the Cable Television Industry. Review of Economics and Statistics 81(2): 326–40. Competition Bureau Canada. 2011. Merger Enforcement Guidelines. October 6. Cotterill, Ronald W., Adam N. Rabinowitz, and Li Tian. 2003. Milk Market Channel Structure: Its Impact on Farmers and Consumers, and the Inadequacies of Antitrust Enforcement as a Foundation for Dairy Policies:  Evidence from the Northeast Dairy Industry, Testimony in Monopsony Issues in Agriculture:  Buying Power of Processors in Our Nation’s Agriculture Markets. Judiciary Committee, US Senate. Dick, Andrew R. 1996. When Are Cartels Stable Contracts. Journal of Law and Economics 39(1): 241–83. Dobson, Paul W., and Ratula Chakraborty. 2008. Buyer Power in the U.K Groceries Market. Antitrust Bulletin 53(2): 333–68. Doyle, Chris, and Roman Inderst. 2007. Some Economics on the Treatment of Buyer Power in Antitrust. European Competition Law Review 28: 210–19. Erutku, Can. 2005. Buying Power and Strategic Interactions. Canadian Journal of Economics 38(4): 1160–72.

Buyer Power in Merger Review   549

European Commission. 2004. Guidelines on the Assessment of Horizontal Mergers under the Council Regulation on the Control of Concentrations between Undertakings. Official Journal of the European Union, Official Journal C 031, 05/02/2004 P. 0005-0018. Fauli-Oller, Ramon, and Lluis Bru. 2008. Horizontal Mergers for Buyer Power. Economics Bulletin 12(3): 1–7. Federal Trade Commission. 2001. Chevron Corporation, and Texaco Inc, File No. 011 0011, Docket No. C-4023. Federal Trade Commission. 2002. Conoco Inc. and Phillips Petroleum Company, File No. 021 0040, Docket No. C-4058. Federal Trade Commission. 2012. Statement of the Federal Trade Commission Concerning the Proposed Acquisition of Medco Health by Express Scripts, Inc. FTC File No. 111-0210. Fee, Edward C., and Shawn Thomas. 2004. Sources of Gains in Horizontal Mergers: Evidence from Customer, Supplier, and Rival Firms. Journal of Financial Economics 74(3): 423–60. Gilbert, Richard J. 2011. Deal or No Deal? Licensing Negotiations in Standard-Setting Organizations. Antitrust Law Journal 77(3): 855–88. Hirsch, Barry T., and Edward J. Schumacher. 1995. Monopsony Power and Relative Wages in the Labor Market for Nurses. Journal of Health Economics 14(4): 443–76. Hirsch, Barry T., and Edward J Schumacher. 2005. Classic or New Monopsony? Searching for Evidence in Nursing Labor Markets. Journal of Health Economics 24(5): 969–89. Hovenkamp, Herbert. 1991. Mergers and Buyers. Virginia Law Review 77(7): 1369–83. Hovenkamp, Herbert. 2005. Federal Antitrust Policy: The Law of Competition and Its Practice. 3rd ed. St. Paul, MN: West Publishing. ICN Merger Working Group, Investigation and Analysis Subgroup. 2006. ICN Merger Guidelines Workbook. Inderst, Roman, and Greg Shaffer. 2007a. Market-Share Contracts as Facilitating Practices. Rand Journal of Economics 41(4): 708–29. Inderst, Roman, and Greg Shaffer. 2007b. Retail, Buyer Power and Product Variety. Economic Journal 117: 45–67. Inderst, Roman, and Greg Shaffer. 2008. Buyer Power in Merger Control. In W. D. Collins, Issues in Competition Law and Policy. Chicago: American Bar Association Section of Antitrust Law. Inderst, Roman, and Christian Wey. 2003. Bargaining, Mergers, and Technology Choice in Bilaterally Oligopolistic Industries. Rand Journal of Economics 34(1): 1–19. Inderst, Roman, and Christian Wey. 2007. Buyer Power and Supplier Incentives. European Economic Review 51(3): 647–67. Inderst, Roman, and Christian Wey. 2011. Countervailing Power and Dynamic Efficiency. Journal of the European Economic Association 9(4): 702–20. Kelly, Trish, and Martin L. Gosman. 2000. Increased Buyer Concentration and Its Effects on Profitability in the Manufacturing Sector. Review of Industrial Organization 17(1): 41–59. Kirkwood, John B. 2012. Powerful Buyers and Merger Enforcement. Boston University Law Review 92(5): 1485–560. Marx, Leslie M., and Greg Shaffer. 2007. Upfront Payments and Exclusion in Downstream Markets. Rand Journal of Economics 38(3): 823–43. Marx, Leslie M. and Greg Shaffer. 2010. Slotting Allowances and Scarce Shelf Space. Journal of Economics and Management Strategy 19(3): 575–603.

550   Dennis W. Carlton, Mary Coleman, and Mark Israel Medcalfe, Simon, and Robert J. Thornton. 2006. Monopsony and Teachers’ Salaries in Georgia. Journal of Labor Research 27(4): 537–54. Miklós-Thal, Jeanine, Patrick Rey, and Thibaud Vergé. 2011. Buyer Power and Intrabrand Coordination. Journal of the European Economic Association 9(4): 721–41. Noll, Roger G. 2005. Buyer Power and Economic Policy. Antitrust Law Journal 72(2): 589–624. Peters, Jürgen. 2000. Buyer Market Power and Innovative Activities. Review of Industrial Organization 16(1): 13–38. Ransom, Michael R., and David P. Sims. 2010. Estimating the Firm’s Labor Supply Curve in a “New Monopsony” Framework:  Schoolteachers in Missouri. Journal of Labor Economics 28(2): 331–56. Raper, Kellie Curry, H. Alan Love, and C. Richard Shumway. 2000. Determining Market Power Exertion between Buyers and Sellers. Journal of Applied Econometrics 15(3): 225–52. Raskovich, Alexander. 2003. Pivotal Buyers and Bargaining Position. Journal of Industrial Economics 51(4): 405–26. Scheelings, Richard, and Joshua D. Wright. 2006. “Sui Generis”? An Antitrust Analysis of Buyer Power in the United States and European Union. Akron Law Review 39(1): 207–43. Schwartz, Marius. 1999. Buyer-Power Concerns and the Aetna-Prudential Merger: Address to the 5th Annual Health Care Antitrust Forum. Northwestern University School of Law, Chicago, October 20. Schwartz, Marius. 2004. Should Antitrust Assess Buyer Market Power Differently Than Seller Market Power? DOJ/FTC Workshop on Merger Enforcement, Washington, DC. Sidak, Gregory J. 2009. Patent Holdup and Oligopsonistic Collusion in Standard-Setting Organizations. Journal of Competition Law and Economics 5(1): 123–88. Skitol, Robert A. 2005. Concentrated Buying Power:  Its Potential for Addressing the Patent Holdup Problem in Standard Setting. Antitrust Law Journal 72(2): 727–44. Sohn, Kitae. 2008. The Monopsony Power in the US Teacher Labor Market. Empirical Economics Letters 7(6): 553–59. Staiger, Douglas O., Joanne Spetz, and Ciaran S. Phibbs. 2010. Is There Monopsony in the Labor Market? Evidence from a Natural Experiment. Journal of Labor Economics 28(2): 211–36. Stigler, George J. 1964. A Theory of Oligopoly. Journal of Political Economy 72(1): 44–61. Twomey, John, and James Monks. 2011. Monopsony and Salary Suppression: The Case of Major League Soccer in the United States. American Economist 56(1): 20–28. UK Competition Commission and Office of Fair Trading. 2010. Merger Assessment Guidelines. US Department of Justice and the Federal Trade Commission (DOJ and FTC). 2010. Horizontal Merger Guidelines. August 19. Wang, Hao. 2010. Buyer Power, Transport Cost and Welfare. Journal of Industry, Competition, and Trade 10(1): 41–53.

CHAPTER 23

V E RT IC A L M E R G E R S MICHAEL A. SALINGER

23.1. Introduction A merger is “vertical” if the companies involved operate at different stages of the “value chain” for a good or service. In some cases, one of the companies purchased inputs from the other prior to the merger. For example, in 2010, both Coca-Cola and Pepsi merged with their largest United States bottlers, companies that were purchasing soft drink concentrate from them. In other cases, one of the companies is a potential but not actual supplier to the other. A famous example was the 1956 merger of Brown Shoe Co., which was primarily a shoe manufacturer, and Kinney, which was primarily a shoe retailer. Kinney stores did not sell shoes manufactured by Brown Shoe prior to the merger but started to do so afterwards.1 This chapter describes the development and state of economic analysis to guide the antitrust treatment of vertical mergers. In Brown Shoe, the US Supreme Court ruled that the Brown Shoe–Kinney merger violated Section 7 of the Clayton Act in part because of the vertical aspects of the merger. Despite blistering critiques from antitrust scholars,2 the Supreme Court has never reversed Brown Shoe. Yet there is substantial intellectual and policy agreement in the United States and Europe on some essential elements of the so-called Chicago critique. (Section 23.3 of this chapter explains the Chicago critique of the earlier US policy toward vertical mergers discussed in section 23.2.) It is widely accepted that the competitive 1  Another category of vertical mergers is so-called diagonal mergers, in which one firm (“A”) merges with a supplier (“B”) to one of its competitors even though there is no prospect A can supply B. 2  Commenting on Brown Shoe, Robert Bork wrote, “It would be overhasty to say that the Brown Shoe opinion is the worst antitrust essay ever written. . . . Still, all things considered, [it] . . . has considerable claim to the title” (Bork 1978,210).

552   Michael A. Salinger effects of vertical mergers are different from the competitive effects of horizontal mergers. It is widely accepted that the basis for this difference is that, whereas competitors generally have a mutual interest in raising prices (which hurts consumers), vertically situated firms generally have a mutual interest in lowering prices (which helps consumers). As a result, it is widely accepted that, as the European Commission stated in its 2008 nonhorizontal merger guidelines, “Non-horizontal3 mergers are generally less likely to significantly impede effective competition than horizontal mergers” (European Commission 2008, ¶10).4 The mergers of Coke and Pepsi with their bottlers illustrate the point. If Coke and Pepsi were to propose to merge with each other, the antitrust authorities in many jurisdictions, including the United States and Europe, would almost certainly intervene to block the merger. Yet both vertical mergers gained antitrust approval.5 But there likely remains consensus that vertical mergers can sometimes be anticompetitive, and that is where the challenge for economic analysis begins. In the late 1980s and early 1990s, a series of articles qualified the Chicago school critique of the antitrust treatment of vertical mergers (Salinger 1988, 1991; Ordover, Saloner, and Salop; 1990, Hart and Tirole 1990). The Chicago critique arguably made it appear that vertical mergers should be legal per se, or at least nearly so.6 As section 23.4 explains, the so-called post-Chicago literature on vertical mergers presented models in which vertical mergers could result in an increase in prices. These results cast some doubt on the policy prescription that seemed to emerge from the Chicago critique, that is, per se legality of vertical mergers that are not attempts at avoiding rate regulation at one stage. A severe limitation of the post-Chicago literature was, however, that it was based on highly stylized models; and how to adapt the models to real-world settings is not at all obvious. This is not a criticism of that literature. Even if the original models were too stylized to be ready for “policy prime time,” they were sufficient to establish that the Chicago critique of earlier policy did not definitively settle the economically appropriate antitrust treatment of vertical mergers. In addition to explaining the development of the policy and the literature, this chapter diagnoses why progress has been so disappointing and suggests an approach that might help provide a rigorous foundation for deciding when to intervene in vertical mergers. The first problem is that the literature became side-tracked on an unimportant theoretical issue. Once one gets past that detour, the pricing pressures created by vertical 3  Vertical mergers are one class of nonhorizontal mergers. Conglomerate mergers, which are neither horizontal nor vertical, are the other class. 4  For evidence of acceptance of these points in the United States, compare the vertical merger sections of the 1968 and 1984 Department of Justice Merger Guidelines. (The 1984 Merger Guidelines were the last federal guidelines in the United States to address vertical mergers.) 5  Coca-Cola’s acquisition of the North American assets of its largest bottler, Coca-Cola Enterprises, entailed Coke’s divestiture of its stakes in its European bottlers, so that case only illustrates the tolerance of vertical mergers in the United States. 6  Even Chicago school critics of antitrust hostility to vertical mergers generally acknowledged that vertical integration or mergers could be a strategy for circumventing price regulation at one stage of an industry. The role of regulation was an important factor in the suit that led to the 1984 breakup of AT&T.

Vertical Mergers   553

mergers are clear. The elimination of double marginalization places downward pricing pressure on the final-good output of the merging entity, but upward pricing pressure on its sales of intermediate inputs to competitors.7 (This latter pricing pressure is the “foreclosure effect.”) This leads to a second reason why the literature has not provided clear guidance on what vertical mergers the agencies should block. The effects go in opposite directions, and the techniques for measuring them are not well developed. Thus, demonstrating that the foreclosure effect is likely to outweigh the benefits from eliminating double marginalization with a sufficient degree of certainty is likely to be difficult.8 There might, however, be a far more compelling class of circumstances that give rise to legitimate concerns about the competitive effects of vertical mergers. To see this point, consider two questions. First, name a horizontal merger that would almost surely be illegal and explain the principal reason why. While many answers are possible, a merger between Coca-Cola and Pepsi that I hypothesized above is as compelling as any. The rationale would be strong upward pricing pressure holding constant all aspects of market structure other than the merger. In other words, the basis for a challenge to a proposed Coca-Cola–Pepsi merger would be static pricing effects. The formal economic analysis that would support the challenge would likely be econometric studies showing a substantial cross-price effect between the two products, and company documents revealing that the actual and anticipated prices (and other actions) of the other weighed heavily on each company’s decisions would be key. The second question is to name a vertical/complementary product merger that would almost surely be illegal and explain the principal reason why. Other than vertical mergers designed to avoid rate regulation, answers to this question are likely to be far more controversial. My answer is a hypothetical merger between Microsoft and Intel, circa 2000.9 The underlying principle would not be short-term pricing pressure, which would have been negative due to double marginalization. Rather, it would have been potential competition. The structure of this chapter reflects the interplay between policy and the academic literature. Section 23.2 describes aspects of public policy toward vertical mergers prior to the Chicago school critique. It does not seek to be encyclopedic. Rather, it describes a small number of policy issues that revealed what modern antitrust scholars consider fundamental misconceptions. Section 23.3 then discusses the Chicago 7  Farrell and Shapiro introduced the term “upward pricing pressure” or (UPP) (Farrell and Shapiro 2010). As Moresi and Salop have observed and as is discussed below, their analysis can be extended to vertical mergers (Moresi and Salop 2012). 8  As I discuss below, this point requires some qualifications. One concerns whether, to be anticompetitive, a merger must result in the price of the final good or, alternatively, whether an increase in the price of an intermediate input is sufficient. The second concerns the burden of proof. See section 23.6, infra. 9  In specifying the time frame, I am not suggesting that Microsoft and Intel could merge now without a challenge. What I am suggesting is that it is likely anticompetitive for two firms, each of which is a monopolist or near-monopolist in successive or complementary stages of an industry, to merge. Whatever controversy there might be about how dominant Microsoft and Intel are today, there would have been far less in 2000.

554   Michael A. Salinger critique, explaining both its contribution in revealing flaws in prevailing policy and its limitations. Section 23.4 discusses the post-Chicago literature. It explains both how the game-theoretic models in that literature modified Chicago school models and also the modeling issues that they raised. Section 23.5 returns to policy in both Europe and the United States. In 2008, the European Commission published guidelines for how it would handle nonhorizontal mergers. As input to those guidelines, the European Commission solicited an extensive review of the theoretical literature on the competitive effects of nonhorizontal mergers. As I explain in section 23.5.1., EC guidelines reflect an obvious influence of the literature. The US antitrust authorities have not issued nonhorizontal merger guidelines, but they have acted (typically with consent agreements) with respect to a small number of vertical mergers. Section 23.5.2 discusses some of those cases and argues that while they reflect some influence of the post-Chicago literature, they do not reveal a coherent, economically sound policy. Section 23.6 then returns to the post-Chicago literature and the slow progress in developing practical policy rules from it. Section 23.6.1 discusses why the commitment issues that have occupied much of the literature are not important in practice or even in theory. Section 23.6.2 argues that the literature does reveal the opposing pricing pressures created by vertical mergers but that measuring which dominates with sufficient precision to support a merger challenge is likely to be difficult. Section 23.6.3 then argues that the issue addressed by the post-Chicago literature, while analogous to what is viewed as the key issue in the analysis of horizontal mergers, is not the most relevant issue for assessing which vertical mergers the antitrust agencies should challenge. Section 23.7 contains conclusions.

23.2.  1960s Era Policy In 1968, six years after the Supreme Court’s ruling in Brown Shoe, the US Department of Justice issued merger guidelines that included a section on vertical mergers (US Department of Justice 1968). They merit quoting at some length, as they reveal how radically thinking on vertical mergers has changed. The first paragraph in the section on vertical mergers asserts, With respect to vertical mergers, . . . the Department’s enforcement activity . . . is intended to prevent changes in market structure that are likely to lead over the course of time to significant anticompetitive consequences. In general, the Department believes that such consequences can be expected to occur whenever a particular vertical acquisition, or series of acquisitions, by one or more of the firms in a supplying or purchasing market, tends significantly to raise barriers to entry in either market or to disadvantage existing non-integrated or partly integrated firms in either market in ways unrelated to economic efficiency. . . . [V]‌ertical mergers tend to raise barriers to entry in undesirable ways, particularly the following: (i) by foreclosing equal access to potential customers, thus reducing

Vertical Mergers   555

the ability of non-integrated firms to capture competitively the market share needed to achieve an efficient level of production, or imposing the burden of entry on an integrated basis . . . even though entry at a single level would permit efficient operation; (ii) by foreclosing equal access to potential suppliers, thus either increasing the risk of a price or supply squeeze on the new entrant or imposing the additional burden of entry as an integrated firm; or (iii) by facilitating promotional product differentiation, when the merger involves a manufacturing firm’s acquisition of firms at the retail level. (US Department of Justice 1968, ¶ 11)

Several aspects of this broad enforcement philosophy are notable. First, the focus is on protecting competitors and, in particular competitors that operate at only one stage of an industry. There is no explicit discussion of consumer welfare as the proper objective of antitrust. The approach implicitly rests on the assumption that consumers ultimately benefit from the aggressive policy “over the course of time,” but the argument that protecting competitors ultimately protects competition, without more, is indistinguishable from an argument for protecting competitors. Second, the statement seems to imply a presumption that vertical mergers between firms with nontrivial market shares are anticompetitive. It reveals no recognition of a fundamental difference between horizontal and vertical mergers with respect to the underlying economic effects. As they did with respect to horizontal mergers, the 1968 Merger Guidelines stated structural criteria likely to give rise to a challenge. These confirm the impression that the DOJ policy at the time was outright hostility to vertical mergers by large firms. The 1968 Merger Guidelines indicated that the DOJ was likely to challenge any merger or set of mergers of a firm with at least a 10% share of an input market and one or more firms accounting for at least 6% of the downstream market. By way of comparison, the section on horizontal mergers in those same guidelines stated that even in a market with a four-firm concentration ratio below 75%, the DOJ was likely to challenge a merger between one firm with a 10% share and another firm with a 4% share. In combination, the standards for vertical and horizontal mergers imply that the DOJ believed that any market share of at least 10% raised competitive concerns, and that any firm with such a share should expect a challenge to a merger with a related entity of any significant size, with little or any distinction based on whether the relationship was horizontal or vertical.

23.3.  The Chicago Critique of Early Vertical Merger Doctrine When the DOJ issued the 1968 Merger Guidelines, the Economic Analysis Group in the Department of Justice did not exist, and the DOJ did not have a chief economist. The

556   Michael A. Salinger architects of the 1968 Merger Guidelines were plainly familiar with the economics literature of the time. They emphasize the effect of vertical mergers on “barriers to entry,” a term that had been central to Joe Bain’s work in the 1950s (Bain 1956). Still, economists had far less input into the 1968 Merger Guidelines than they did into all federal US merger guidelines since, starting with those the DOJ issued in 1982. An essential feature of the Chicago critique of antitrust policy was to use formal economic analysis to assess the effects of actions (including mergers) on economic outcomes. This analytical approach focused the analysis on prices and, therefore, effects on consumers. What is considered the Chicago school critique of policy toward vertical mergers rested on three principal ideas/models.

23.3.1.  Single-Monopoly Profit In the model underlying the “single-monopoly profit” principle, one stage of production has only a single supplier (i.e., a monopolist), while the adjacent stage is perfectly competitive. If the monopolized stage is upstream in the sense that it sells to the downstream producers who in turn sell to final consumers, then the model requires the additional assumption that downstream production is of the “fixed proportions” or “fixed coefficients” variety. This additional assumption rules out the possibility that downstream producers can substitute away from the monopolized input. Under these assumptions, the monopolist can earn the same profits by operating at just one stage as it could earn by being a vertically integrated monopolist. Thus, it has no theoretical incentive to leverage its market power into the adjacent stage; and, even if it did so (perhaps by refusing to sell its intermediate output), it would not raise the price to consumers. The result in this model that a monopolist at one stage cannot increase profits by integrating into the adjacent stage is the “single-monopoly profit theorem.”10 Even before the so-called post-Chicago literature developed, there was a literature on the incentives for and effects of vertical integration by an upstream monopolist into a competitive downstream stage with variable proportions technology (Vernon and Graham 1971; Malella and Nahata 1980; Schmalensee 1973). On one level, one might doubt whether this literature ever had any policy relevance, as it was not vertical mergers into a perfectly competitive downstream stage industry that created antitrust concern. However, the literature foreshadowed a set of trade-offs brought out by the post-Chicago literature. The opportunity to substitute away from the monopolized input acted as a constraint on the upstream firm’s market power. Integration would eliminate this constraint and thereby extend the monopoly power of the upstream firm. On the other hand, absent vertical integration,

10 More accurately, it is “a” single-monopoly profit theorem, as there is a related result for the tying of goods consumed in fixed proportions.

Vertical Mergers   557

downstream production was inefficient. Vertical integration eliminated the production inefficiencies. Thus a vertical merger extended market power but also created efficiencies. The literature examined the conditions under which each effect was dominant. Whatever the inherent relevance of the variable proportions literature, it has not played a central role in policy debates for several decades.11

23.3.2.  Double Marginalization The second model underlying the Chicago school critique of vertical merger policy was the successive monopoly model. It examines the incentive of two successive monopolists to integrate and the effect if they do so. While, in contrast to the single-monopoly profit model, successive monopolists do have an incentive to merge, the theoretical effect of such a merger is to cause a reduction in price. To take a simple numerical example, let demand be given by

P = 24 − Q,



(23.1)

where P is the final good price and Q is the quantity demanded. Assuming fixed proportions in production, Q is also the quantity of the intermediate good.12 Let upstream per unit costs be 6. Downstream producers must purchase one unit of the upstream (or intermediate) good per unit of the downstream good and incur additional costs of 6 per unit as well. One standard solution to the model is to assume that the downstream monopolist maximizes its profits taking the upstream price as given (Spengler 1950). That is, the downstream firm chooses Q to maximize:

(

)

Π D = PQ − ( p + 6) Q = 24 − Q − p − 6 Q (23.2)

Note that in this equation, p, the upstream price, is a parameter, not a choice variable. As a result, the downstream firm’s profit-maximizing output, Q*, is a function of p. In inverse form, the relationship is

p =18 − 2Q * . (23.3)

11  I am not aware of anyone who suggests that measuring the elasticity of substitution between the monopolized input and other inputs is relevant for the legality of one or a series of vertical mergers. 12  Given the fixed proportions assumption, the assumption that each unit of output requires one unit of the upstream good merely requires a suitable definition of units and does not impose any additional substantive restrictions.

558   Michael A. Salinger Equation (23.3) is the “derived demand” curve facing the upstream firm. Based on that derived demand curve, its profit function is ΠU =



( p − 6)Q* = (12 − 2Q *)Q*, (23.4)

which is maximized at Q* = 3, implying p = 12 and P = 21. If the two firms were to merge to form a vertically integrated monopolist, their profits (ΠVI ) would be

(

)

(

)

(

)

ΠVI = P − 12 Q = 24 − Q − 12 Q = 12 − Q Q, (23.5)



which is maximized at Q = 6 (and P = 18), a higher output and a lower price than when there are successive monopolists. The result arises because of double marginalization prior to the merger. The upstream firm’s price entails a margin or markup over marginal cost that then becomes part of the downstream firm’s marginal cost. The downstream firm equates (true) marginal revenue to its marginal cost, but its marginal cost exceeds the true marginal production cost. This distortion induces it to charge too high a price relative to the price that maximizes joint profits. A vertical merger eliminates this distortion and causes price to drop. While the above model of the outcome when the firms are separate might seem natural, the literature contains another prominent solution originally due to Cournot ([1838] 1971). In it, the two firms are conceived of as producing complementary products rather than vertically related products. This approach might seem to be the natural way to model cases like Microsoft and Intel, which do not sell to each other but instead sell separately to original equipment manufacturers such as Dell or Hewlitt Packard. But even when one of the firms sells its output to the other, which in turn sells to final consumers, we might think of the downstream firm as selecting a mark-up over its price rather than an actual price. Under this formulation, there are two stages, 1 and 2, setting prices p1 and p2, with equation 1 becoming: Q = 24 − p1 − p2 , (23.1ʹ)

Firm i maximizes: Πi =

( p − 6) Q = ( p − 6) (24 − p − p ) ,     i = 1, 2, i

i

i

j

j = 2, 1. (23.6)

Maximizing π1 with respect to p1 holding p2 constant and maximizing π2 with respect to p2 holding p1 constant yields two equations with two unknowns. Solving them simultaneously gives p1 = p2 = 10 and therefore a total price of 20. This combined price

Vertical Mergers   559

in the “complementary monopoly” model is less than the price of 21 in the “successive monopoly” model, but still greater than the integrated solution of 18.

23.3.3.  Comparative Transactions Costs Another idea generally considered supportive of a tolerant treatment of vertical mergers is Coase’s seminal article “The Nature of the Firm” (Coase 1937). According to Coase, activities at different stages of production can be coordinated through market exchange or through internal organization. Both entail costs. Coase hypothesized that vertical integration arises when the cost of coordinating through internal organization is lower than the cost of market-mediated coordination. Early interpretations of the implications of Coase’s theory for vertical merger enforcement (Bork 1978) presumed that the necessary coordination had to entail achieving efficiencies. As a matter of pure economic logic, this need not be the case. To the extent that firms at adjacent stages can mutually benefit by raising the costs of other firms, improving coordination through vertical integration can conceivably facilitate anticompetitive behavior. Notwithstanding this caveat, the conventional interpretation is that the comparative transactions/organization cost literature weighs in favor of antitrust tolerance of vertical mergers.

23.3.4.  Reprise on the Chicago School Critique Prior to the Chicago school critique, antitrust policy toward vertical mergers reflected outright hostility. Far from viewing the horizontal/vertical distinction as having dramatically different implications for appropriate antitrust treatment, the earlier policy reflected the view that vertical mergers posed nearly as great a threat to competition as do horizontal mergers. By using formal economic analysis of the likely effects of vertical mergers on consumers, the Chicago school critique virtually reversed the policy prescription to a presumption that vertical mergers are procompetitive or at least competitively neutral.

23.4.  The Post-Chicago Qualifications The Chicago critique of vertical merger doctrine successfully convinced antitrust policymakers that the economic relationship between vertically related firms differs fundamentally from the economic relationship among competing firms and that the difference has profound implications for merger policy. It did not, however, resolve all competitive issues with vertical mergers. Suppose, for example, a car manufacturer proposed to purchase a steel company that not only supplies the acquirer but also some of

560   Michael A. Salinger the acquirer’s competitors. Even if the merger results in lower-cost steel for the acquiring car manufacturer, does it provide an incentive to raise the market (as opposed to the internal transfer) price of steel to the acquiring firm’s rivals? Of course, under the assumption that the steel industry is perfectly competitive, any attempt to do so would be doomed to failure. Competing car companies would counter any attempt by the newly merged company to raise steel prices by acquiring from a competing company. But what if the steel industry is not perfectly competitive? The successive/complementary monopoly model cannot address the question of raising the market price of the input to rivals because, given the assumption of downstream monopoly, there are no downstream rivals to competitively disadvantage and no external market for the input after the merger in which to raise the price. For economic analysis to address these questions, one must assume oligopoly for at least one (and possibly both) stages. The post-Chicago literature on vertical mergers refers to a set of models that emerged in the latter half of the 1980s that examined vertical mergers in an oligopolistic setting. Here, I briefly describe four of them, focusing on the key assumptions and results.

23.4.1.  Salinger Model of Successive Cournot Oligopolists Salinger analyzed the effects of vertical mergers when both the input and final output are homogeneous and there is a Cournot equilibrium at each stage (Salinger 1988). Compared with the other post-Chicago models of vertical mergers, the Salinger successive Cournot oligopoly (“SSCO”) model is more general with respect to market structure, allowing for any number of upstream and downstream firms and for any number of those firms to be vertically integrated. Within the model, a vertical merger is represented by an additional vertically integrated firm (along with one less unintegrated intermediate good and one less unintegrated final good producer). A key issue in the model is whether vertically integrated firms participate in the market for the intermediate good (either as buyer or seller). I will return to this issue in more detail in section 23.6. For now, suffice it to say that the model assumptions imply that vertically integrated firms do not participate in the intermediate good market. Given the model assumptions, a vertical merger has two initial effects. First, as noted above, the merging entity withdraws from the intermediate good market, leaving unintegrated final good producers with one less supplier. All else equal, this effect results in a higher price for the intermediate good and therefore higher costs for unintegrated final good producers. Second, the merging entity gets the intermediate good at marginal cost rather than the market price, so its cost of producing the final good goes down. These initial effects then give rise to a third equilibrium effect. The merging entity’s lower cost for the final good induces it to expand output. As a result, the residual demand curve facing unintegrated final good producers and, in turn, the derived demand curve facing unintegrated intermediate good producers shift back. All else (including the number of unintegrated intermediate good producers) equal, this shift in the derived demand curve results in a reduction in the price for the intermediate good. Thus, in equilibrium,

Vertical Mergers   561

there are two effects on the price of the intermediate good and they go in opposite directions. Under some circumstances, the price-raising effect dominates and the price of the intermediate good increases. Under others, the price-lowering effect dominates and the price of the intermediate good drops. In a Cournot equilibrium with a given number of final good producers and given demand, the equilibrium price is a function of the average marginal cost of the firms in the market. When a vertical merger causes the price of the intermediate good to increase, the net effect on the final good price depends on which effect dominates, the lower marginal cost of the merging entity that is due to the elimination of double marginalization or the increased marginal cost of the remaining unintegrated final good producers (due to what might be termed “input foreclosure”). Again, the price-increasing effect dominates in some cases, while the price-reducing effect dominates in others. When a vertical merger causes the price of the intermediate good to drop, marginal cost drops for all final good producers and so, therefore, does the price of the final good. The model rests on specific assumptions and one must be careful not to interpret the results too literally.13 Still, the model brings out the two offsetting pricing pressures created by a vertical merger. In the model, a vertical merger can cause the price of the final good to increase because a reduction in competition in the market for the intermediate good causes an increase in the marginal cost of unintegrated final good producers. However, for a vertical merger to have this anticompetitive effect, this foreclosure effect must be larger than the effect that comes from eliminating double marginalization for the merging entity. Moreover, the foreclosure effect may not occur at all, as the price of the intermediate good can fall rather than increase.

23.4.2.  Ordover, Saloner, and Salop In the Ordover, Saloner, and Salop (“OSS”) model, there are two downstream firms (D1 and D2) selling differentiated output and two upstream (U1 and U2) firms selling a homogenous input that both downstream firms use. They assume Bertrand competition at both stages. Because the input is homogeneous, Bertrand competition absent a merger results in the input being priced at marginal cost. In the simplest version of the model, U1 and D1 consider merging and committing to not supplying the input to D2. With the latter commitment, the merger leaves U2 with a monopoly over sales of the input to D2, which it exploits by raising the price of the input. Under quite general conditions, the increased marginal cost to D2 increases the combined profits of D1 and U1 given the Bertrand equilibrium downstream. This stripped-down version is a basis for a potential concern about vertical mergers. By 13  For example, the model predicts that a vertical merger can be anticompetitive when the upstream stage is more concentrated than the downstream stage but not vice versa. The result is sensitive, however, to the specific assumptions about the timing of decisions. See Salinger 1988, n. 8, and, for a more thorough analysis, Salinger 1989.

562   Michael A. Salinger reducing the number of independent suppliers of the intermediate good, the vertical merger results in an increase in the price of the intermediate good and therefore an increase in the marginal cost of the integrated firm’s rival. The profitability of the merger stems from the raising rivals’ cost (RRC) effect. This RRC effect is quite similar to the anticompetitive mechanism in the SSCO model. A key difference is that the upstream Bertrand assumption eliminates both premerger double marginalization and the possibility of a decrease in the price of the intermediate good for independent downstream producers. In the SSCO model, the RRC effect might (but does not necessarily) occur, and the efficiency of eliminating double marginalization for the merging firm always occurs. OSS abstract away from these trade-offs to explore two other issues. One is whether the integrated firm could credibly avoid undercutting the remaining unintegrated upstream firm in the sale of the input to the downstream firm. The other is whether the remaining unintegrated upstream and downstream firms could profitably merge as a counterstrategy. The resolutions of the two issues turn out to be related to each other, as OSS show by making three alternative assumptions about how the vertically integrated firm might participate in the intermediate good market. One assumption is that the vertically integrated firm engages in Bertrand competition in the intermediate good market. In that case, the equilibrium with one vertically integrated firm is the same as with none or two. For any price above marginal cost charged by the unintegrated intermediate good producer, the vertically integrated firm cannot resist slightly undercutting. Given the Bertrand assumption, a vertically integrated firm believes it can capture the rents of the unintegrated upstream firm while having no material effect on the marginal cost of the unintegrated downstream firm. But, as in single-stage Bertrand competition with equal marginal costs and homogenous products, the equilibrium is to drive the intermediate good price down to marginal cost, which then eliminates any competitive effect. OSS also consider allowing the vertically integrated firm to commit to not supplying the unintegrated downstream at all or, alternatively, to be a first mover in the price game, recognizing that the unintegrated upstream firm will then slightly undercut it to be able to supply the unintegrated downstream firm. This turns out to matter for whether the remaining unintegrated firms have an incentive to merge as a counterstrategy. A second vertical merger in the OSS model effectively restores the premerger equilibrium, as both downstream firms get the input at marginal cost. Thus, the profitability of a second merger turns critically on whether having a single vertical merger causes the joint profits of the remaining independent firms to go up or down. If they go up, a subsequent vertical merger is not profitable, as restoring the premerger equilibrium will cause them to go down. Because, the premerger Bertrand equilibrium is an example of a prisoner’s dilemma, small increases in both prices tend to cause the profits of both vertically related pairs of firms to increase. However, offsetting this is the possibility that the unintegrated upstream firm gets “too aggressive” in the price it sets, thus grabbing for itself a bigger share of the pie than the unintegrated sector gets but at the same time shrinking the size of that pie. In OSS, an anticompetitive merger is not susceptible to profitable

Vertical Mergers   563

counter merger if there is some restraint on the price increase of the intermediate good. This can occur either through a backstop provider or the commitment on the part of the vertically integrated firm to supply the unintegrated downstream firm. Its reason for doing so is not to make profits selling the intermediate good but, rather, to make sure that the remaining upstream firms are jointly profitable enough to stay unintegrated.

23.4.3.  Hart and Tirole Remarkably, Hart and Tirole question the single-monopoly profit theorem itself, and present a model in which a vertical merger is a way for an upstream monopolist to have a complete monopoly, which entails higher prices to consumers and higher profits than a monopoly over a single stage (Hart and Tirole 1990). In their model, two downstream firms produce a homogeneous output. Two upstream firms are capable of producing a necessary input, with one having lower costs than the other. Despite the appearance of duopoly upstream, they allow for the possibility that one of the upstream firms has such a cost advantage that in effect gives it a monopoly. Hart and Tirole allow for completely general nonlinear price-quantity contracts between an upstream and a downstream firm,14 but they do not allow for exclusivity provisions or any other form of commitment by the upstream firms. It might initially appear that a variant of the single-monopoly profit theorem would apply. The upstream firm could contract with one of the downstream firms to supply it with the monopoly quantity in return for a payment that extracts all the monopoly profits. Alternatively, it could contract with each downstream firm to supply half the monopoly quantity in return for a payment that extracts all the available rents. In the Hart-Tirole model, however, this strategy is not feasible because, having written those contracts and received payment, the upstream firm could then offer to sell more to the downstream firms. Under such a setting, a vertical merger would eliminate the incentive to offer to sell additional units of the intermediate good to a downstream firm in order to extract additional profits after having already received the simple monopoly profits. The Hart-Tirole argument extends well beyond their precise structural assumptions. It applies as well to the standard setting for the single-monopoly profit theorem. Under the standard argument, a monopolist upstream sets a single (linear) price that applies to all downstream firms in the competitive downstream stage, is designed to induce the downstream firms in aggregate to demand the monopoly quantity, and captures the monopoly profits for the upstream monopolist. Hart and Tirole point out that an implicit assumption behind this strategy is that, having sold the monopoly quantity at the monopoly price, the upstream monopolist cannot sell more. Technically, Hart and Tirole’s observation about the commitment implicit in the single-monopoly profit theorem is correct. But because firms stay in business for more

14 

Contracts can specify a quantity and a total revenue.

564   Michael A. Salinger than one period, the benefits from establishing a reputation for not flooding the market after having made some sales at a high price would seem to justify such an assumption.

23.4.4.  Salinger Differentiated Products Model In another model of vertical mergers by Salinger, two differentiated upstream producers sell inputs to a downstream multiproduct monopolist (Salinger 1991).15 Each upstream firm sets its (upstream) price (above marginal cost), and then the downstream firm maximizes its profits by choosing the downstream prices for the two products (given the input prices). A vertical merger between the downstream monopolist and one of the two upstream brands results in the downstream firm getting the input from the firm it acquires at marginal cost. The model resembles the successive monopoly model, but the presence of the second product complicates the effect of the vertical merger. One of the effects is similar to the single-product successive monopoly model. Let “Good 1” be the brand of the acquired upstream firm and “Good 2” be the brand of the remaining independent upstream firm. Holding constant its (downstream) price of Good 2, the downstream firm has an incentive to lower its price for Good 1. The second effect is, however, a foreclosure or RRC effect.16 Since the downstream firm gets the input for Good 1 at a lower price as a consequence of the merger, then, holding its price of Good 1 constant, it gets a higher margin on Good 1 and therefore has an incentive to sell more units of Good 1. One tool for doing so is to raise the price of Good 2, which induces customers to switch from Good 2 to Good 1. Holding constant its price for Good 1, the downstream firm has an incentive to increase the price of Good 2 so as to get consumers to shift from Good 2 to Good 1. These two effects interact with each other,17 and there are three possibilities for the net effect: the price of both goods can drop, the price of both goods can increase, or the price of Good 1 can drop while the price of Good 2 increases.18 The only qualitative effect that is not possible is a decrease in the price of Good 2 and an increase in the price of Good 1. When 15  I developed this model when I was on academic leave visiting the FTC in 1985 and 1986. During that time, the Commission evaluated (and ultimately approved) proposals by both Coke and Pepsi to buy some of their bottlers, many of which bottled other brands as well. Think of the upstream firms as selling different flavors of soft drink concentrate and the downstream firm as a bottler. 16  The term foreclosure is often used to mean a complete denial of access to certain transactions. But that meaning is imprecise and would not be the economically meaningful concept in any event. An economically meaningful definition is that a company is foreclosed from a market when it experiences an increase in an input “price,” recognizing that a downstream margin can be interpreted as a price for distribution paid by an upstream firm. 17  Formally, the optimal prices for the downstream monopolist are the simultaneous solution to two first-order conditions (one with respect to each price). Each of the two effects is a shift in one of the first-order conditions. Each pure effect is the shift of one first-order condition. The effect of the merger reflects the simultaneous solution of the two. 18  While it may be surprising that eliminating double marginalization for one good can cause an increase in the price of both goods, the result is an application of a well-established economic principle known as “Edgeworth’s Paradox of Taxation.”

Vertical Mergers   565

a vertical merger causes an increase in the price of both goods, it is anticompetitive. Of course, when it causes a reduction in both prices, it is procompetitive. In the intermediate case, further analysis is necessary. The two effects of a vertical merger in the differentiated monopoly model are quite analogous to those in the SSCO model; and, as in that model, the interaction between them can lead to counterintuitive results. The anticompetitive effect can dominate, but it might not arise at all.

23.4.5.  Empirical Work Given that vertical mergers give rise to offsetting effects, it would seem desirable to have an extensive empirical literature to ascertain in general which dominates and, ideally, conditions to determine when the foreclosure is likely to dominate and other conditions when it is not. Some evidence exists, although it has not been extensive. One industry in which there have been several studies of vertical integration is the cable television industry (Waterman and Weiss 1996; Chipty 2001). An industry feature that makes it attractive for studying vertical integration is that, historically at least, there were relatively small distinct geographic markets each served by a single operator.19 Initially, some of the cable operators were quite small. Over time, a relatively small number of large multiple-system operators (MSOs) accounted for an increasing proportion of the homes passed and subscribers nationwide.20 The large MSOs tended to have ownership stakes in some cable networks. Thus, one could try to assess the effects of vertical integration by comparing offerings and prices across markets and/or over time as both cable systems and cable networks changed hands. One result has emerged in every study of the question. MSOs with an ownership stake in a network were more likely to carry that network and less likely to carry competing networks than were MSOs without an ownership stake in the network. For example, Viacom at one point operated a relatively small number of cable systems and also owned the pay cable channels Showtime and The Movie Channel, which competed with HBO and Cinemax. Viacom systems were more likely to offer Showtime and The Movie Channel than were non-Viacom systems, but less likely to offer HBO and Cinemax. What the studies did not find was any compelling evidence that vertical integration affected prices or the total number of offerings. One industry that received rather extensive attention from the FTC when it was actively pursuing vertical mergers was the Portland cement/ready-mix concrete industry. In an early study, Allen reviewed 55 mergers between cement and ready-mix firms 19  The Federal Communications Commission issues annual reports on the state of competition in programming distribution. As the series of reports documents, telephone companies have increasingly installed fiber-optic networks that allow them to provide multichannel video distribution service, and satellite providers such as DirecTV have grown. 20  See the discussion of the Time Warner–Turner Broadcasting merger in section 23.5, infra.

566   Michael A. Salinger between 1956 and 1969 (Allen 1971). Following the FTC’s primary concern when it was aggressively challenging vertical mergers in this industry, he assessed whether the motive for each merger was to guarantee the cement producer an outlet for its output and, if so, whether the merger accomplished the objective. Using (highly) informal analysis, he concluded that some cement firms may have acquired concrete firms to buy outlets, but that the strategy was unsuccessful. Hortaçsu and Syverson did statistical analysis on plant-level census data from 1963 to 1997 to assess the effects of vertical integration (Hortaçsu and Syverson 2007). They found that concrete prices were on average lower in markets with more vertical integration.21 They concluded that the result was not the direct result of vertical integration, however. Instead they attributed it to the development and efficiency of multiplant concrete firms that also tended to be vertically integrated.22 Hastings and Gilbert studied the effect of Tosco’s 1997 acquisition of BP’s West Coast refining and marketing assets on Tosco’s wholesale gasoline prices (Hastings and Gilbert 2005). Tosco sold unbranded wholesale gasoline to stations that competed with BP’s stations. They constructed a measure of the interaction between Tosco’s stations after the transaction and independent stations, and found that this variable was positively and significantly related to the postacquisition increase in Tosco’s wholesale price, an effect that is consistent with an RRC interpretation.

23.5.  Effect of the Post-Chicago Literature on Policy The post-Chicago literature affected policy, most significantly in Europe but arguably in the United States as well. In Europe, the European Commission commissioned Professor Jeffrey Church to conduct a thorough review of the theoretical literature on nonhorizontal mergers (Church 2004, henceforth “the Church White Paper”) as input to the nonhorizontal merger guidelines it issued in November 2007 (European Commission 2008).23

21  They also found that within given markets, vertically integrated concrete firms charge less than those that are not vertically integrated. The result certainly suggests that vertical integration does not raise prices, although the existing literature does not contain a formal prediction about the relative pricing among firms within markets. 22  They also found that while as much as 70% of cement output was by vertically integrated cement producers, the peak in the fraction of concrete output by vertically integrated concrete firms was only 11%. (Presumably, the vertically integrated cement firms were only partially integrated and sold a substantial portion of their output to other concrete producers.) Given that fact, the historical concern that this industry was particularly prone to anticompetitive vertical integration seems misplaced. 23  October 18, 2008, is the date of publication in the Official Journal of the European Union. The date of adoption was November 28, 2007.

Vertical Mergers   567

23.5.1.  The Church White Paper The Church paper is remarkable for being comprehensive and for providing structure to a vast literature. As I have stated previously (Salinger 2005) and will explain below and as Professor Church has acknowledged (Church 2008), I disagree with how he drew policy inferences from the SSCO and OSS literatures. However, his conclusions about the relevance of the Hart-Tirole model and, more importantly, his recognition that the preemption literature that started with Gilbert and Newbery has important implications for vertical merger policy are apt (Gilbert and Newbery 1982). The Church White Paper divides the post-Chicago literature on vertical mergers between those that focus on input foreclosure and those that emphasize customer foreclosure. He in turn divides the input foreclosure group into three main streams, raising rivals’ costs, pricing commitments, and entry deterrence. He labels the SSCO model and OSS as the seminal papers in the first stream and the Hart and Tirole paper as the seminal paper in the second stream. The third stream draws on a literature that had not previously been considered as applying to vertical mergers, although he is right to observe that it does. In discussing the raising rivals’ cost literature, he focuses on issues of commitment. As noted above, the SSCO model rests on assumptions under which vertically integrated firms do not participate in the market for the intermediate good. A series of papers work with the alternative assumption that in deciding how much to “sell” on the intermediate good market, vertically integrated firms assume that other intermediate good producers hold their output fixed. While this might seem like a natural assumption in a Cournot model, the implication is that vertically integrated firms choose to sell a negative amount, meaning that they choose to buy on the intermediate good market. Under this assumption, vertical mergers lower prices. From this, Church infers that in industries for which successive Cournot oligopoly provides a reasonable approximation, the welfare consequences turn critically on whether vertically integrated firms can commit not to engage in strategic buying. If they cannot, he concludes, a vertical merger is welfare-improving. When they can commit, he concludes that the welfare effects are ambiguous but that an increase in the upstream price is necessary for a vertical merger to be harmful. Drawing on work by Avenal and Barlet, Church also concludes that the welfare effects turn critically on whether a vertically integrated firm’s commitment not to engage in strategic purchases deters upstream entry (Avenal and Barlet 2000). He elaborates on this point by arguing, “The one set of circumstances in which it is unambiguous that a vertical merger leads to anticompetitive harm to consumers downstream involves a monopolist upstream in which a vertical merger deters entry upstream and where technology makes foreclosure credible” (Church 2004, p. 54). The other article that Church characterizes as being seminal with respect to a raising-rivals’-costs effect of vertical mergers is OSS. Reiffen criticized it on the grounds that for the remaining unintegrated firm to increase its price postmerger, it had to be credible for the vertically integrated firm not to undercut it (Reiffen 1992). Without such a commitment, the vertically integrated firm would, according to Reiffen, have

568   Michael A. Salinger an incentive to undercut the unintegrated upstream firm by an infinitesimal amount, thus securing the upstream profit for itself without reducing its downstream profits. According to this argument, the postmerger equilibrium would entail an upstream price equal to marginal cost and would therefore be equivalent to the downstream equilibrium. Church discusses extensions to OSS by Choi and Yi and Chen (Choi and Yi 2000; Chen 2001). Both papers seek to address the supposed “commitment” issue in OSS. Similar to the paper by Avenal and Barlet, Choi and Yi do so through a choice of technology. Chen does so by introducing cost heterogeneity among upstream firms and a cost to a downstream firm of switching suppliers. In the Chen model, unintegrated firms choose to buy from the vertically integrated firm at a price above marginal cost to give the vertically integrated firm a stake in the unintegrated firms’ sales. That stake creates what is now termed upward-pricing pressure on the vertically integrated firm’s downstream good and results in higher equilibrium prices. Going in the opposite direction of this effect is the elimination of double marginalization for the merging firm. With respect to the Hart and Tirole model, Church writes, “The implications of this model for vertical merger policy are unlikely to be significant. It is difficult to see this monopoly commitment problem as a justification for a vertical merger” (Church 2004, p. 85). Church asserts that contractual forms of commitment might eliminate the need to commit through vertical merger, but mere reputation might be sufficient. While dismissing the policy relevance of the Hart-Tirole paper, which is generally considered one of the seminal post-Chicago papers on vertical mergers, he quite perceptively asserts the relevance of the Gilbert-Newbery paper that is nominally about the incentive to innovate. Gilbert and Newbery addressed the question of whether an incumbent or an entrant would have the greater incentive to invest in an improved version of the product. In one of the finest examples of what John Sutton termed “robust theory” (Sutton 1991), they showed that a monopolist has more of an incentive to innovate and preempt an entrant than the entrant has to innovate. The only condition needed for the result is that the monopoly profits exceed the sum of the duopoly profits. As Church observes, a vertically integrated monopolist facing potential competition downstream would have more incentive to merge with an upstream entrant than would a downstream firm. Having made this observation, Church asserts that it is most relevant when a vertically integrated monopolist considers purchasing an upstream entrant that would make downstream entry possible. The agencies might view such a merger as horizontal rather than vertical, which would be consistent with the Chicago view that competitive concerns from mergers arise entirely as a result of horizontal overlaps.

23.5.2.  The EU Nonhorizontal Merger Guidelines The EU nonhorizontal merger guidelines in many ways followed the broad outline provided by the Church White Paper. As noted above, arguably the most important

Vertical Mergers   569

statement in the EU nonhorizontal merger guidelines is, “Non-horizontal mergers are generally less likely to impede effective competition than horizontal mergers” (European Commission 2008, ¶11). Given that, what one then wants to understand about the EU guidelines is what insight they yield into the characteristics of vertical mergers that give rise to the presumably rare challenge. The guidelines lay out several possible theories that could underlie a challenge to a vertical merger. They distinguish between unilateral and coordinated effects,24 and they divide unilateral effects into input foreclosure and output foreclosure effects. Input foreclosure refers to raising rivals’ costs. Output foreclosure refers to denying upstream firms access to downstream markets. The EU guidelines refer to EU cases, not the Church White Paper, but some of the specific points seem to pick up on Church’s themes. For example, they observe that a mechanism for input foreclosure is that “the merged firm may opt for a specific choice of technology within the new firm which is not compatible with the technologies chosen by rival firms” (European Commission 2008, ¶33). This comment seems to draw on Church’s discussion of the Avenal and Barlet and Choi and Yi papers. The guidelines also say, When competition in the input market is oligopolistic, a decision by the merged entity to restrict access to its inputs reduces the competitive pressure exercised on remaining input suppliers, which may allow them to raise the input price they charge to non-integrated downstream competitors. In essence, input foreclosure by the merged entity may expose its downstream rivals to non-vertically integrated suppliers with increase market power. This increase in third-party market power will be greater the lower the degree of product differentiation between the merged entity and other upstream suppliers and the higher the degree of upstream competition. However, the attempt to raise the input price may fail when independent input suppliers, faced with a reduction in the demand for their products (from the downstream division of the merged entity or form independent downstream firms), respond by pricing more aggressively. (European Commission 2008, ¶38, references omitted) 24  In this chapter, I have generally not covered coordinated effects. In part, this is because coordinated effects have played an increasingly small role in horizontal merger enforcement, so it seems unlikely that coordinated effects would be central to the presumably small number of vertical mergers to be challenged. This is particularly so because vertical separation facilitates collusion better than vertical integration (Bonanno and Vickers, 1988). To the extent that horizontal collusion is imperfect, successive oligopoly will result in something closer to the monopoly outcome than single-stage oligopoly. This observation might seem to run counter to Nocke and White, who purport to show that vertical mergers facilitate collusion as a quite general matter (Nocke and White 2007). They get this result even though a vertical merger in the Nocke-White model makes the merging entity more likely to cheat on a preexisting collusive agreement. Offsetting this effect is that the vertical merger reduces the opportunity for each of the remaining unintegrated firms to cheat. But both collusion and cheating take quite specific forms in the model, so their conclusion that vertical mergers as a quite general matter facilitate collusion appears to be that vertical mergers facilitate the rare form of collusion in their model.

570   Michael A. Salinger The first part of that paragraph captures the anticompetitive effects in both the SSCO and OSS models. The remainder of the paragraph and, in particular, the last sentence, draws on the result in the former that the price of the intermediate good sold by unintegrated upstream firms to unintegrated downstream firms might drop as a result of a vertical merger despite the withdrawal of one supplier from the market. Despite Professor Church’s conclusion with respect to the Hart-Tirole analysis, the EC guidelines contain a footnote that gives some credence to the commitment issues raised by their model. They observe with respect to the single-monopoly profit theorem, “One situation in which this [i.e., the ability for a monopolist to extract all available rent] may not be the case would be when the monopolist has a so-called commitment problem which it is unable to solve. . . . Vertical integration may restore the upstream supplier’s ability to commit not to expand input sales as this would harm its own downstream division” (European Commission 2008, ¶44, note 5). The concern with output foreclosure articulated in the EC guidelines is that a vertically integrated firm will refuse to buy from independent input suppliers and thereby induce some of them to exit because they cannot achieve adequate scale. In turn, the exit might cause the price of the input to unintegrated downstream producers to increase and cause the price of the final good to increase as well. The effect is what Krattenmaker and Salop referred to as “Frankenstein monster” (Krattenmaker and Salop 1986). The EU guidelines also have a section on how a vertical merger might facilitate collusion. The discussion follows the factors Stigler articulated as being necessary for collusion to succeed and describes ways in which a vertical merger could conceivably make it easier for colluding firms to reach an agreement, to monitor whether the “coconspirators” are abiding by it, and to punish deviations (Stigler 1964).

23.5.3.  US Enforcement The last US vertical merger guidelines were those issued by the Department of Justice in 1984. Thus, an assessment of US enforcement must reflect cases. A thorough review is beyond the scope of this chapter, but a few cases are worth noting. Following twelve years of Reagan-Bush era antitrust that had intentionally scaled back enforcement, the beginning of the Clinton administration in 1993 shortly followed the development of the post-Chicago literature and coincided with vertical mergers in key sectors of the economy undergoing substantial structural change.

23.5.3.1. A&T-McCaw A prime example was AT&T’s acquisition of McCaw. AT&T manufactured telecommunications equipment (including equipment used in cellular telephone networks) and was the largest provider of long-distance service (with a 70% share of the US market). McCaw had been an early entrant into cellular telephone service. At that time, cellular telephone service was a duopoly in each city (as a result of FCC spectrum auction policy), with the incumbent telephone company (referred to as the “BOC” in the case

Vertical Mergers   571

of former Bell system companies) as one of the providers. McCaw’s national share was approximately 17%. The merger resulted in two primary vertical links. As an equipment manufacturer, AT&T had provided important components of the cellular networks of incumbent telephone companies. It had been a more important supplier to the mobile services of the BOCs than to McCaw, and there was at least some lock-in effect associated with maintenance, upgrades, and the building out of networks in a market that was still in its infancy. The second vertical link was that cellular customers needed long-distance providers. Prior to the merger, McCaw required its customers to use AT&T’s long-distance service, which it purchased from AT&T at wholesale and resold to its cellular customers at a higher rate. As a consequence of the Modified Final Judgment (“MFJ”) that settled the epic U.S. v. AT&T, the BOCs were required to let their mobile phone customers select their long-distance provider. The Justice Department did not block the merger, but it did impose conditions on it. It required that the merged entity continue to operate as separate divisions. It required AT&T to continue to supply equipment to its customers that compete with McCaw on the same terms as it had provided prior to the merger and on terms that were as good as it provided McCaw. It placed some restrictions on the sharing of information between AT&T and McCaw. (Interestingly, however, it did not prevent AT&T from sharing with McCaw its information about customers of McCaw’s competitors who used AT&T long-distance service.) Finally, it required McCaw to allow its customers to choose their long-distance service. One might argue that the case is unique because of the backdrop of the MFJ. For example, BOCs would seem to have a legitimate concern about the sharing of competitively sensitive information. (BOC cellular customers who used a large amount of long-distance service were also, on average, large users of local cellular service, so the information AT&T had as a long-distance supplier arguably allowed it to identify a BOC’s most valuable customers.) Absent the MFJ, the BOC’s might have responded to the merger by not making AT&T long-distance service an option for its customers, but the MFJ precluded that response. The Antitrust Division’s decision to allow AT&T to share this information with McCaw may have been an attempt to put a thumb on the competitive scales to counterbalance what it judged to be other advantages for the BOCs. Similarly, the requirement to allow McCaw customers to choose their long-distance provider was an explicit attempt to increase competition rather than a remedy to correct harm to competition. These caveats notwithstanding, the consent is problematic. The requirement that McCaw give its consumers a choice of long-distance providers seems to reflect concerns about customer foreclosure, not raising rivals’ costs. The conditions with respect to equipment do reflect a concern with raising rivals’ costs, a concern that might have been particularly important given that AT&T was more important as a supplier to McCaw’s competitors than it was to McCaw itself. (The relevance is that there may not have been a double marginalization effect to offset the raising rivals’ cost effect.) The requirement that AT&T maintain McCaw as a separate division appears to allow the vertical merger

572   Michael A. Salinger while imposing a condition that denies the merging entity the benefits of integration. The most compelling rationale for this solution is that the benefits of the merger were to give McCaw, a young, highly leveraged firm, access to capital. AT&T, as the largest provider of a complementary service, may have been willing to finance McCaw on terms that it could not otherwise obtain. In general, though, economists place little weight on access to capital as a merger efficiency.

23.5.3.2.  Time Warner–Turner Broadcasting In 1995, Time Warner, at the time the second largest cable television operator in the United States, proposed to purchase the 80% of Turner Broadcasting that it did not already own. Turner Broadcasting owned several cable networks, including CNN, the first cable news network. At the time of the transaction, CNN was the only cable news network that virtually all cable systems carried. Not only was it the most viewed cable news network, but its coverage of the first US invasion of Iraq in 1990 had greatly enhanced its general reputation as a news source. At the time of the proposed merger, however, CNN’s position in cable news was threatened by the entry of two cable news networks, MSNBC and Fox News Channel. The FTC issued a complaint against the proposed acquisition and entered into a consent agreement with the parties to resolve the complaint.25 While the FTC did not allege that cable news networks constituted a relevant antitrust market, it did allege that the merger would reduce Time Warner’s incentives to carry networks that competed directly with the “crown jewels” it was acquiring and that this reduced incentive would hinder competition in the market for the sale of “Cable Television Programming Services” to multichannel video program distributors (MVPDs). To resolve this aspect of the complaint, Time Warner agreed to carry one national news and information channel. To comply with the consent agreement, Time Warner entered into an agreement with MSNBC. As the consent agreement only required Time Warner to carry one national news channel, it did not at first agree to carry Fox News Channel. Reportedly in response to pressure from New York mayor Rudolph Giuliani, who threatened to use a channel allocated to local government use to carry Fox News Channel, Time Warner did agree to carry Fox News Channel (Lueck 1997).26 Fox News has emerged as a far more popular news network than CNN (or MSNBC).27 The FTC consent was not the reason that Fox gained access to the Time Warner systems, and it is impossible to know what would have happened without the intercession 25  Federal Trade Commission, Agreement Containing Consent Order, In the Matter of Time Warner, Inc., Turner Broadcasting System, Inc., Telecommunications, Inc., and Liberty Media Corporation, available at http://www.ftc.gov/os/1996/09/timewar.pdf, last accessed May 7, 2014. 26  An aspect of this saga not easily captured by economic analysis is the well-publicized animosity between Rupert Murdoch, the chairman of Fox News channel owner News Corporation, and then Time Warner vice chairman Ted Turner. 27  “This most recent win marks 11 consecutive years of FNC’s dominance as the #1-rated cable news network” (Kondloojy 2013).

Vertical Mergers   573

of Mayor Giuliani. However, what is clear is that Time Warner did resist carrying Fox News, which turned out to be a formidable competitive threat to CNN.

23.5.3.3. Premdor-Masonite At issue in the Premdor-Masonite merger was competition in the sale of interior molded doors.28 Producing a molded door entails attaching two molded panels (called “doorskins”) to a frame. In contrast to flat doorskins, a molded doorskin creates the appearance of a solid wood door. (Molded doors are much less expensive than solid wood doors.) At the time of the merger, the market for molded interior doors had two large producers—Premdor and Jeld-Wen29—and a competitive fringe. Jeld-Wen was vertically integrated into molded doorskins, producing virtually all the doorskins it needed for its own door production and selling some to small door producers. Premdor primarily relied on Masonite, the other large producer of molded doorskins and the only one that was not vertically integrated, for most of its supply of doorskins. Masonite also sold doorskins to the smaller doorskin manufacturers. In 2000, Premdor and International Paper entered into an agreement for Premdor to buy its Masonite division. Two years prior to the acquisition, Premdor had taken a 48.5% stake in Fibramold, a small molded doorskin manufacturer, and entered into an agreement to get Fibramold doorskins at a low price in return for a commitment to purchase a minimum volume. In response to this agreement, Masonite had offered lower doorskin prices to Premdor as well as to smaller door manufacturers. The Antitrust Division challenged the merger. It then entered into a consent with Premdor whereby Premdor would sell one of Masonite’s two plants where it manufactured molded doorskins.30 In explaining its actions, the Justice Department put forward a coordinated effects theory of anticompetitive harm. Relying on the standard Stiglerian (Stigler 1964) framework for assessing the likelihood of collusion, it argued the merger would eliminate factors that had previously stood in the way of coordination in what was nearly a duopoly. Specifically, it argued that Premdor believed that Jen-Weld had a cost advantage over it by virtue of being vertically integrated. According to the Justice Department’s competitive impact statement, the merger would break down a barrier to competition by making the cost structures of the two companies more similar.31 Whatever the merits of its actions in the case, the Justice Department’s explanation is problematic both because it turns an efficiency into a factor that facilitates collusion 28  United States v. Premdor et al., Complaint, August 3, 2001, available at http://www.justice.gov/atr/​ cases/f8800/8893.htm, last accessed March 14, 2013. 29  The Justice Department did not name Jen-Weld in its complaint or competitive impact statement, instead referring to it as, “the non-party firm.” 30  United States v. Premdor et al., Hold Separate Stipulation and Order, August 3, 2001, available at http://www.justice.gov/atr/cases/f8900/8907.htm, last accessed March 14, 2013. 31  United States v. Premdor et al., Competitive Impact Statement, August 3, 2001, available at http://​ www.justice.gov/atr/cases/f9000/9017.htm, last accessed March 14, 2013.

574   Michael A. Salinger and because it puts forward the doubtful proposition that vertical mergers can harm competition by facilitating collusion.32

23.6.  Towards a Practical Policy The post-Chicago models have laid an intellectual foundation for the policy position that a vertical merger (or a merger of firms producing complementary products) can be anticompetitive. But the result that some presumably small number of vertical mergers might be anticompetitive, without more, does not provide antitrust agencies, courts, or counsel advising clients the guidance they need. If policy toward vertical mergers is to be something other than per se legality, the challenge for antitrust policymakers and for economists seeking to give advice about how to formulate a coherent policy is to try to glean practical insights from what is now a literature consisting of highly stylized largely game-theoretic models.

23.6.1. Commitment As discussed in section 23.5.1 above, the literatures on both the SSCO and OSS models have focused on the credibility of commitments. In both models, the commitment at issue concerns the participation of the vertically integrated firm in the market for the intermediate good (although in the SSCO model, the supposed commitment issue concerns purchases of the intermediate good, whereas committing not to sell the intermediate good is the supposed commitment issue in OSS). This section explains why the focus on commitment has diverted attention from the economics of vertical mergers.

23.6.1.1.  SSCO Model As explained in the original article, the supposedly controversial assumption regarding vertically integrated firms’ participation in the market for the intermediate good is the following: First, if a vertically integrated firm sells an extra unit of the intermediate good, it conjectures that other intermediate good producers maintain their outputs and that a final good producer increases its output by one unit. Second, if a vertically integrated 32 

A fallacy underlying the Justice Department’s argument is a common misperception that cost asymmetries are a barrier to collusion. Under Stigler’s approach to analyzing the likelihood of collusion, successful collusion requires arriving at an agreement, the ability to detect violations if they occur, and the ability to punish violations (which in part rests on a credible threat to follow through with the punishment). Some argue (incorrectly) that cost differences are a barrier to the first step, as firms with different costs will disagree over the price they would ideally want. To be sure, in a duopoly, the low-cost firm generally prefers a lower price than the one that the high-cost firm wants most. But the

Vertical Mergers   575

firm buys an extra unit of the intermediate good, it assumes that an intermediate good producer expands its output by one unit and other final good producers maintain their output. (Salinger 1988, p. 347)

From a contemporary perspective, the argument seems informal or, to state matters more forcefully, lacking in rigor. The word “conjecture” is out of style. Riordan characterized the assumption as “awkward” (Riordan 1998). But the assumption is completely standard, although the reason why is apparently not well understood. As a preliminary, consider successive Cournot oligopoly without vertically integrated firms. Suppose there are NI intermediate good producers and NF final good producers. Let Qi be the output of the final good by final good producer i (1 ≤ i ≤ N F ) and q j be the output of intermediate good producer j (1 ≤ j ≤ N I ) . Final good production entails fixed proportions, so we can define units so that a final good producer needs one unit of the intermediate good to produce one unit of the final good. The inverse demand curve for the final good is given by P (Q ) , where Q = ∑ iN=F1 Qi = ∑ iN=I1 q j . Assume constant returns to scale at both stages, and define c I as the marginal (and average) cost of the intermediate good and cF be the marginal (and average) cost of the final good other than the need to purchase the intermediate input. A natural approach to modeling the equilibrium would be to specify a two-stage game. In stage 1, intermediate good producers simultaneously choose outputs. In stage 2, final good producers simultaneously choose their output (based on the price determined at stage 1).33 This approach is not the only one possible. One might consider formulating the problem as complementary as opposed to successive oligopoly.34 But it is certainly one approach to consider. The solution technique is to work backwards. The equilibrium at stage 2 is the single-stage Cournot-Nash equilibrium among the final good producers, taking the price of the intermediate good as given. Final good producer i maximizes its profits πiF :

    πiF =  P  Qi + ∑Qk  − pI − cF  Qi , (23.7)     k ≠i

taking pI and the output of all other final good producers as constant. Given standard regularity conditions, the quantity of the final good produced and, therefore, the low-cost firm still wants to collude to a price above the competitive price, and the high-cost firm’s cost disadvantage reduces its incentive to undercut the price the low-firm wants. Whether or not the two firms agree on the price they most want is the wrong question to focus on. The right question is whether cost asymmetries prevent the firms from agreeing that they prefer some price above the noncollusive price. They do not. 33  Some would describe the model as having three stages, with stage 1 being the choice of upstream outputs, stage 2 being the determination of a price for the intermediate good, and stage 3 being the choice of downstream outputs. 34  See the comparison of successive and complementary monopoly in section 23.4, supra.

576   Michael A. Salinger quantity of the intermediate good demanded is a function of pI : Q = f 0 ( pI ) , where the subscript indicates that there are no vertically integrated firms. The derived inverse demand curve for the input is pI = f 0 −1 (Q ) . At stage 1, intermediate good producer j maximizes its profits, π Ij :     π Ij = pI − c I q j =  f 0 −1  q j + ∑qm  − c I  q j , (23.8)     m≠ j

(



)

taking the output of all other intermediate good producers as given. While rarely (if ever) stated as an explicit assumption, any Cournot oligopoly model—whether it has one stage or many—implicitly assumes an asymmetry between buyers and sellers with respect to their participation in the market for the intermediate good. When a seller considers increasing the quantity it sells by 1, it holds constant the quantity transacted by other sellers. It does not (and cannot logically) hold constant the quantity transacted by buyers. An implication of a seller’s Cournot assumption about other sellers is that when it expands its sales by one unit, the total quantity purchased increases by one unit. Some buyer has to purchase the additional unit. A similar point applies to buyers. They are price takers. In a single-stage Cournot model, they need not make any explicit assumption about the behavior of other buyers. However, since each buyer assumes it is a price taker, changing the quantity it purchases does not affect the price and therefore the incentives facing other buyers. In a successive Cournot oligopoly, the assumption that the purchasers of the intermediate good hold the purchases of other purchases constant is explicit because the purchasers of the intermediate good are sellers of the final good and their decisions about how much of the intermediate good to purchase are equivalent to their decisions about how much of the final good to produce. If a buyer holds the purchases of other buyers constant, it must assume that sellers adjust the quantity they sell in response to changes in how much it purchases.35 Now suppose upstream firm 1 and downstream firm 1 merge. The profit of the merged entity, π1V , is

(

)( ) (Q ) − c  (q − Q ) , (23.9)

π1V = Q1  P (Q ) − c I − cF  + pI − c I q1 − Q1



     = Q1  P (Q ) − c I − cF  +  f1−1 

1

I

1

1

where Q1 is the final good output of all firms besides firm 1 and therefore their requirements for the intermediate input, Q1 = f1 ( pI ) , is the relationship giving the Cournot output of the non-vertically integrated firms as a function of the price of the intermediate

35 

Note that downstream firms take the supply of the input as being perfectly elastic even though they move second in the game.

Vertical Mergers   577

good when there is one vertically integrated firm. The first term of (23.9) reflects the firm’s downstream profits. The second term allows for the vertically integrated firm’s participation in the market for the intermediate good. If it chooses qi = Qi , then its production of the intermediate good exactly satisfies its needs for the intermediate good in its production of the final good. Its market “sales” of the intermediate good are qi − Qi , where a negative value would imply market purchases of the intermediate good. After the vertical merger, assume there is still a Cournot equilibrium for the final good. Formally, there is a first order condition in which one differentiates (23.9) with respect to Q1, holding constant the output of the other final good producers. Also note that what one holds constant in taking the partial derivative reflects an important assumption. If one holds q1 constant, then an increase in Q1 either uses up a unit of the intermediate good that the firm could otherwise sell at the market price or requires the vertically integrated firm to buy a unit of the intermediate good at the market price. In either event, the intermediate good used in final production has an opportunity cost of pI , which in turn implies that the vertical merger does not eliminate double marginalization. Alternatively, if one holds constant q1 − Q1 (i.e., the firm’s participation on the market for the intermediate good), then the decision to increase output of the final good by one unit includes the decision to expand output of the intermediate good by one unit (at marginal cost). Similarly, one must decide whether to model participation on the intermediate good market by taking the derivative of the profit function with respect to q1 or q1 − Q1, in either case holding Q1 constant. Outside of a game-theoretic context, the choice would simply reflect a change in variable. As it would not affect the solution, the choice would merely be a matter of analytic or expositional convenience. But in a game-theoretic model, the choice matters because it affects what about the behavior of other firms is held constant. The first order condition for the vertically integrated firm with respect to its output of the final good holding constant its sales of the intermediate good is

∂π1V =  P (Q ) − c I − cF  + Q1P ′ (Q ) = 0. (23.10) ∂Q1 |q1 −Q1 

To ascertain the vertically integrated firm’s optimal sales of the intermediate good, take the partial derivative of (23.9) with respect to its intermediate good sales:

∂π1V = Q1P ′ (Q ) + ( pI − c I ) + f1−1′ (Q1 ) (q1 − Q1 ) . (23.11) ∂ (q1 − Q1 ) |Q 1

Note that Q = Q1 + Q1 and, given the assumption of a Cournot equilibrium upstream, dQ1 / d (q1 − Q1 ) = 1. A  key result is that (23.11) implies that the sum of the first two terms of (23.10) is negative provided that the total price-cost margin exceeds just the upstream price-cost margin (which has to be the case given the Cournot equilibrium

578   Michael A. Salinger downstream). Since f1−1′ (Q1 ) < 0, (23.11) is negative for any nonnegative value of (q1 − Q1 ) , which in turn implies that a vertically integrated firm in successive Cournot oligopoly would choose not to sell any units of the intermediate good. Equation (23.11) equals 0 at a negative value of q1 − Q1 . Taken literally, the result implies that vertically integrated firms would choose to buy the intermediate good. However, it is important to be clear on the assumptions underlying that “result.” It rests on the assumption that dQ1 / d (q1 − Q1 ) = 1. In words, it presumes that, by buying one more unit of the intermediate good (thus causing a reduction one unit reduction in q1 − Q1 ), a vertically integrated firm can cause the supply of the intermediate good available to other final good producers and, therefore, the output of other final good producers to drop by 1. This strategy of buying scarce inputs is not available to any other downstream firm in the model, and there is no reason why vertical integration by a downstream firm into producing the intermediate good would confer upon it the ability to buy up the supplies its downstream rivals need to produce. Under the logic of the Cournot model, equation (23.11) applies only to sales of the intermediate good, that is, nonnegative values of q1 − Q1 . To analyze whether a vertically integrated firm would purchase any units of the intermediate good, the natural assumption to make is that it is a price taker (like every other buyer in the model). If so, it has no reason to purchase the intermediate good at a price above marginal cost if it can produce it for itself at marginal cost. There is nothing awkward about the set of assumptions underlying the SSCO model. It does entail an asymmetry between the effects of sales and purchases of the intermediate good by vertically integrated firms. But it is the same asymmetry between buyers and sellers that exists in the standard Cournot model. With respect to the SSCO model, the issue of commitment with respect to vertically integrated firms’ behavior on the intermediate good market is a non-issue. Using completely standard assumptions, it is a result of the model that vertically integrated firms have no incentive either to buy or sell on the intermediate good market.

23.6.1.2.  OSS Model The main effect captured by the OSS model is that a vertical merger with a downstream producer alters the incentives of an upstream producer to supply the intermediate good to a customer that will use the supply to compete with its downstream merger partner. The credibility issue in the model is that on the margin, a price cut creates a large, discontinuous benefit in the market for the intermediate good and only a marginal raising-rivals’cost benefit in the downstream market. The issue is a consequence of assuming Bertrand competition with undifferentiated products. It is, however, a well-known modeling issue, not an economic issue. (Both Cournot and Bertrand were aware of it.)36 36 

Perhaps it is not surprising as Cournot was a mathematician, but concerns about continuity of payoff functions play prominently in this book. Those concerns explain why he made quantity the decision variable in his oligopoly model and price the choice variable in his model of complementary monopoly with fixed proportions. Ironically Bertrand proposed what we now call the Bertrand equilibrium as reductio ad absurdum with respect to Cournot’s logic.

Vertical Mergers   579

OSS acknowledged the point in their original article, but had several answers to it both there and in their reply to Reiffen (Ordover, Saloner, and Salop 1992). Prior to a vertical merger in the OSS model, neither upstream firm has an incentive to commit to a price. (Doing so would simply invite the other firm to capture the entire market by undercutting slightly.) But after the merger, the merging firm would like to commit to a price for the intermediate good (even recognizing that it will be undercut), and the remaining upstream producer benefits from the vertically integrated firm’s commitment (and therefore has no reason to stop it from doing so). The credibility issues raised by Reiffen are very specific to the undifferentiated Bertrand model underlying OSS, but the economic effect of a vertical merger is not. To the extent that an intermediate good producer can have a material effect on the market price of the intermediate good, a vertical merger gives it an incentive to consider the effect on its sales of the intermediate good on the costs of its downstream rivals and therefore on its own downstream profits. If OSS had made an alternative assumption about competition in the upstream market, the model would have become more complicated. Price would not equal marginal cost prior to a vertical merger. A vertical merger would eliminate double marginalization, and it is not as obvious as one might initially suspect that the price of the intermediate good would increase. The OSS model does not capture all the important price effects of a vertical merger, and that point is a valid comment on the limitations. But to suggest that the effect that they did isolate is not present because it is not credible is at best a point about modeling and not about how vertical mergers affect incentives.

23.6.2.  The Robust Lessons of the Post-Chicago Models and the Challenge in Applying Them Once one dismisses the credibility issues that were the basis of the critiques of the SSCO and OSS models and that were the starting point of the Hart-Tirole analysis, a robust set of results emerges. As in the analysis of horizontal mergers, one wants to start with understanding how a vertical merger alters the merging firm’s pricing and output incentives; and one needs to do so at both the intermediate and final good stage. With respect to final goods, a vertical merger eliminates double marginalization and thereby creates an incentive to increase output and lower price, all else equal. With respect to the intermediate good, however, there is a foreclosure effect. If the merging firm has market power in the intermediate good market, increasing output lowers the price of the intermediate good to other downstream firms. In turn, downstream prices drop, which causes the merged entity to have lower downstream profits. That effect creates an additional economic marginal cost in selling the intermediate good that acts as a form of upward pricing pressure that is analogous to the diversion effect that forms the basis of the Farrell-Shapiro upward pricing pressure analysis for horizontal mergers.

580   Michael A. Salinger As in the analysis of horizontal mergers, full merger simulation would entail additional effects that would depend on such factors as the shape of the demand curve(s) and the precise nature of competitive interactions. The possibility that the price of the intermediate good falls in the SSCO model is an example of such a secondary effect. The rationale behind the use of UPP analysis in horizontal mergers, which focuses just on the change in the first-order conditions, is that these secondary effects are difficult to measure. We should expect these effects to be at least as difficult to measure with respect to vertical mergers as they are with respect to horizontal mergers. To the extent that UPP analysis is a useful tool for horizontal merger analysis, one may try a similar approach in vertical merger analysis. How to implement UPP analysis for vertical mergers depends on what the legal standard is for anticompetitive harm. One possibility, which the European Commission guidelines seem to embrace, is that a reduction in competition means an increase in final good prices, not just intermediate good prices. The alternative is that an increase in intermediate good prices is sufficient to establish harm to competition. Many economists would likely argue for a standard based on the market for the final good. But if the competition statutes preserve competition to protect customers, it is not clear why they should not protect firms that rely on competition among input suppliers to keep the price of their inputs down. If the legal standard is that a likely increase in the price of the intermediate good is sufficient to characterize the merger as being anticompetitive, then far more mergers are likely to be subject to challenge than if a challenge requires a showing that an increase in the price of the final good is likely. If the legal standard requires a showing that final good prices are likely to go up, then a challenge in blocking vertical mergers is that the pricing pressures go in opposite directions. There may be cases when the upward pricing pressure is strong and the downward pricing pressure is not. There are two general circumstances when one might expect such a result. One is when premerger pricing of the intermediate good is at or near marginal cost while downstream margins are substantial. If so, the elimination of double marginalization may be negligible, but the incentive to use pricing of the intermediate good to divert downstream sales away from the upstream division’s existing customers toward its own downstream division may be substantial. The other possibility is a circumstance like the AT&T-McCaw merger in which the upstream firm is a bigger supplier of the downstream division’s rivals than it is of the division it is acquiring. What to do when both pricing pressures are strong may be less clear. With horizontal mergers, the pricing pressure generally goes in one direction; and the challenge in merger analysis is to assess whether it is big enough to pose a serious competitive concern. Measurement error is likely to be present, but there is only one measurement error. With vertical mergers, there are two pricing pressures with measurement error likely on both. If one starts with the presumption that most vertical mergers are procompetitive, then one might impose a strict burden of proof on an agency trying to block a merger to demonstrate that the anticompetitive effects are likely to dominate. It may prove

Vertical Mergers   581

difficult to demonstrate with the required degree of certainty that a substantial upward pricing pressure due to foreclosure outweighs eliminating a substantial degree of double marginalization.

23.6.3.  Is Static Pricing Pressure the Right Issue? Despite wide acceptance of the proposition that competition in innovation is far more important to consumer welfare and economic growth than static price competition, horizontal merger enforcement largely reflects static pricing effects. The basis of the Chicago critique of early vertical merger policy was that the static pricing effects of vertical mergers are much different from the static pricing effects of horizontal mergers and that antitrust policy should therefore reflect greater tolerance for vertical mergers than for horizontal mergers. Post-Chicago analysis suggested that the static pricing effects with vertical mergers are more complicated than the Chicago critique had made it appear. However, as discussed above, the post-Chicago literature suggests a complicated set of trade-offs that may be difficult to weigh in practice. While using market structure to predict market outcomes has generally fallen out of favor among economists and it is common to describe advances in antitrust analysis as being based more on effects than on structure, market structure still plays an important role in horizontal merger analysis. The probability of an antitrust challenge remains substantial for mergers to monopoly and even “three to two” mergers. Mergers of the ninth and tenth largest firms in a market generally get approval easily. As a result, it is reasonable to ask what structural criteria are most conducive to challenging a vertical merger. If the theory behind a challenge concerns short-run pricing pressure, successive monopoly is not the answer. Models predict a price reduction from a merger of successive monopolists. Using the OSS model for support, some might argue that the ideal condition for challenging a vertical merger is (roughly symmetric) duopoly at two successive stages. Under such a theory, a vertical merger between one firm at each stage leaves the remaining independent downstream at the mercy of a monopolist input supplier. An alternative perspective is, however, that successive monopoly is precisely the most important time to intervene. Here, the basis for the challenge is not short-run price competition. Rather, it is potential competition. One of the long-running debates regarding competition in innovation concerns which entities have a greater incentive to innovate and, in particular, whether the incentive is greater for entrants or incumbents. Intuitively, one might expect the incentives for an incumbent monopolist to be weaker, as any innovation on might cannibalize its existing business. While such an argument may be persuasive with respect to a monopolist protected from entry (such as AT&T prior to the 1984 divestiture), it may be misplaced with respect to firms with market positions that are dominant at a point in time but subject to the threat of entry.

582   Michael A. Salinger Gilbert and Newbery demonstrated that, as a quite general matter, a monopolist faced by the threat of entry by a firm that makes a disruptive innovation is greater than the incentive of an entrant (Gilbert and Newbery 1982). When there is successive monopoly and double marginalization exists, however, the monopolist at one stage has a bigger incentive to innovate in the adjacent stage than does a new firm seeking to enter the adjacent stage. In addition to getting the direct profits from entry, the adjacent monopolist’s business would also benefit from the increase in competition. Thus, an adjacent monopolist (or dominant firm) may be uniquely situated to challenge the firm at the adjacent stage. Consider the merger between Intel and Microsoft (circa 2000) hypothesized in the introduction. Both firms were likely charging prices above marginal cost. A  merger between them may have provided a short-run pricing incentive to cut prices.37 Even if that were so, however, both arguably had a powerful incentive either to innovate or to promote innovation in the other’s market. Microsoft had a strong incentive to help AMD succeed in its challenge to Intel’s position and Intel had an incentive to subsidize open source projects that competed with Microsoft Windows and Office.38

23.7. Conclusions The post-Chicago literature on vertical mergers has clarified how vertical mergers affect pricing incentives in ways that the Chicago school critique overlooked. But it has revealed a complicated set of effects that are difficult to integrate into practical policy rules. Antitrust authorities should still seek to understand the pricing pressures created by vertical mergers, and cases may well arise when a vertical merger will likely create substantial upward pricing pressure without an offsetting benefit from the elimination of double marginalization. But they should pay even closer attention to mergers between firms that are dominant at successive stages. Even if the successive monopoly model suggests that the merger will create downward pricing pressure in the short run, each of the parties might be uniquely situated to challenge the dominance of other; and keeping them separate can preserve their incentive to do so.

37 

The fact that both were charging above marginal cost is not sufficient to establish that a merger between them would have solved a double marginalization problem. Both were likely aware of the prisoner’s dilemma nature of their pricing decisions. Even without explicit discussions about pricing, they may have been able to coordinate tacitly and restrain their pricing. 38  This rationale arguably justifies the FTC’s objection to the likely effect of Time Warner–Turner Broadcasting merger on the possibility of entry by other cable news channels.

Vertical Mergers   583

References Allen, Bruce T. 1971. Vertical Integration and Market Foreclosure:  The Case of Concrete and Cement. Journal of Law and Economics 14:251–74. Avenal, Eric, and Corinne Barlet. 2000. Vertical Foreclosure, Technological Choice, and Entry in the Intermediate Good Market. Journal of Economics and Management Strategy 9:211–30. Bain, Joe S. 1956. Barriers to New Competition. Cambridge, MA: Harvard University Press. Bonanno, Giacomo, and John Vickers. 1988. Vertical Separation. Journal of Industrial Economics 36:257–65. Bork, Robert H. 1978. The Antitrust Paradox: A Policy at War with Itself. New York: Basic Books. Chen, Yongmin. 2001. On Vertical Mergers and Their Competitive Effect. Rand Journal of Economics 32:667–85. Chipty, Tasneem. 2001. Vertical Integration, Market Foreclosure, and Consumer Welfare in the Cable Television Industry. American Economic Review 91:428–53. Choi, Jay P., and Sang-Seung Yi. 2000. Vertical Foreclosure with the Choice of Input Specifications. Rand Journal of Economics 31:717–43. Church, Jeffrey. 2004. The Impact of Vertical and Conglomerate Mergers on Competition. Available at http://ec.europa.eu/competition/mergers/studies_reports/merger_impact.​ pdf. Last accessed March 21, 2013. Church, Jeffrey. 2008. Vertical Mergers. In Wayne D. Collins, Issues in Competition Law and Policy. Chicago: American Bar Association, Section of Antitrust Law. Coase, Ronald. 1937. The Nature of the Firm. Economica 4:386–405. Cournot, Augustin. [1838] 1971. Researches into the Mathematical Principles of the Theory of Wealth. N. T. Bacon, trans. New York: Augustus M. Kelley. European Commission. 2008. Guidelines on the Assessment of Non-horizontal Mergers under the Council Regulation on the Control of Concentrations between Undertakings. Available at http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:C:2008:265:0006:0025:en:​ PDF. Last accessed January 9, 2013. Farrell, Joseph, and Carl Shapiro. 2010. Antitrust Evaluation of Horizontal Mergers: An Economic Alternative to Market Definition. B.E. Journal of Theoretical Economics 10:9. Gilbert, Richard J., and David M. J. Newbery. 1982. Preemptive Patenting and the Persistence of Monopoly. American Economic Review 72:514–26. Hart, Oliver, and Jean Tirole. 1990. Vertical Mergers and Market Foreclosure. Brookings Papers on Economic Activity: Microeconomics. 1990:205–76. Hastings, Justine S., and Richard J. Gilbert. 2005. Market Power, Vertical Integration and the Wholesale Price of Gasoline. Journal of Industrial Economics 53:469–92. Hortaçsu, Ali, and Chad Syverson. 2007. Cementing Relationships:  Vertical Integration, Foreclosure, Productivity, and Prices. Journal of Political Economy 115:250–301. Kondloojy, Amanda. 2013. Fox News Channel Is Again Top-Rated Cable News Network for the Month of January. TV by the Numbers, January 29. Available at http:// tvbythenu ​ m bers.zap2it.com/2013/01/29/fox-news-channel-is-again-top-ratedcable-news-netw​ork-for-the-month-of-january/167281/. Last accessed March 14, 2013. Krattenmaker, Thomas G., and Steven Salop. 1986. Anticompetitive Exclusion: Raising Rivals’ Costs to Achieve Power over Price. Yale Law Review 96:209–93. Lueck, Thomas J. 1997. Time Warner Is Said to Agree to Carry Murdoch’s Fox News. New York Times, July 23. Available at http://www.nytimes.com/1997/07/23/nyregion/time-warner-is-​said-toagree-to-carry-murdoch-s-fox-news.html?pagewanted=2&src=pm. Last accessed March 14, 2013.

584   Michael A. Salinger Mallela, Parthasuradhi, and Babu Nahata. 1980. Theory of Vertical Control with Variable Proportions. Journal of Political Economy 88:1009–25. Moresi, Serge, and Steven Salop. 2012. vGUPPI:  Scoring Unilateral Pricing Incentives in Vertical Mergers. Georgetown Business, Economics, and Regulatory Law Research Paper No. 12-022. Available at http://scholarship.law.georgetown.edu/cgi/viewcontent.​ cgi?article=1167&context=fwps_papers. Last accessed March 21, 2013. Nocke, Volker, and Lucy White. 2007. Do Vertical Mergers Facilitate Upstream Collusion? American Economic Review 97:1321–39. Ordover, Janusz A., Garth Saloner, and Steven C. Salop. 1990. Equilibrium Vertical Foreclosure. American Economic Review 80:127–42. Ordover, Janusz A., Garth Saloner, and Steven C.  Salop. 1992. Equilibrium Vertical Foreclosure: Reply. American Economic Review 82:698–703. Reiffen, David. 1992. Equilibrium Vertical Foreclosure: Comment. American Economic Review 82:694–97. Riordan, Michael H. 1998. Anticompetitive Vertical Integration by a Dominant Firm. American Economic Review 88:1232–48. Salinger, Michael A. 1988. Vertical Mergers and Market Foreclosure. Quarterly Journal of Economics 103:345–56. Salinger, Michael A. 1989. The Meaning of “Upstream” and “Downstream” and the Implications for Modeling Vertical Mergers. Journal of Industrial Economics 37:373–87. Salinger, Michael A. 1991. Vertical Mergers in Multi-product Industries and Edgeworth’s Paradox of Taxation. Journal of Industrial Economics 39:545–56. Salinger, Michael A. 2005. Is it Live or Is It Memorex? Models of Vertical Mergers and Antitrust Enforcement. Available at http://www.ftc.gov/speeches/salinger/050927isitlive.pdf. Last accessed May 7, 2014. Schmalensee, Richard. 1973. A Note on the Theory of Forward Integration. Journal of Political Economy 81:442–49. Spengler, Joseph J. 1950. Vertical Integration and Antitrust Policy. Journal of Political Economy 58:347–52. Stigler, George J. 1964. A Theory of Oligopoly. Journal of Political Economy 72:44–61. Vernon, John M., and Daniel A.  Graham. 1971. Profitability of Monopolization by Vertical Integration. Journal of Political Economy 79:924–25. US Department of Justice. 1968. 1968 Merger Guidelines. Available at http://www.justice.gov/​ atr/hmerger/11247.pdf. Last accessed July 17, 2012. Waterman, David, and Andrew A. Weiss. 1996. The Effects of Vertical Integration between Cable Systems and Pay Cable Networks. Journal of Econometrics 72:357–95.

Vertical Mergers   585

Cases Cited Brown Shoe Company v. United States 370 US 294 (1962). Federal Trade Commission, In the Matter of Time Warner, Inc., Turner Broadcasting System, Inc., Telecommunications, Inc., and Liberty Media Corporation, File No. 961-0004, Docket No. C-3709. United States v. AT&T (Modified Final Judgment) 552 F. Supp. 131 (D.D.C. 1982). United States v. AT&T Corp. and McCaw Cellular Communications, Inc., Civil Action No. 94-015555 (HHG). United States v. Premdor Inc., Premdor U.S. Holdings, Inc., International Paper Company, and Masonite Corporation, Civil Action No.: 1:01CV01696.

Index

Accountable care organizations,  109–110 Accounting anomalies,  220–222 Actual punishment policy,  21–25 Advertising-supported media mergers, 428–430 Advocate Health Partners,  92–93 Aetna,  95, 102–104, 106, 532 Affeldt, Pauline,  427, 429 Agency costs, deterrence and,  262–263 Agricultural Marketing Agreements Act of  1937, 539 Agricultural revolution,  206n3 AIDS (Almost Ideal Demand System) demand,  478, 482–486 Airline industry component production, achieving efficient scale for,  43 national security, conferring market power to promote,  55 Airline Tariff Publishing system,  236 Aktas, Nihat,  155 Alchian, Armen,  208–209 Alcoa,  320, 338, 352 Allen, Bruce T.,  565–566 Alliance Movement,  153 Almost Ideal Demand System (AIDS) demand,  478, 482–486 Al-Najjar, Nabil,  220–221 Aluminum Ltd.,  338 Amazon, 156 Amelio, Andrea,  435 American Antitrust Institute,  533n11, 535–536 American Bar Association, Antitrust Law Section,  84, 542, 544 American Express,  415 American Medical Association, health insurance and,  96 American Telegraph & Telephone (AT&T)

as dominant firm,  320 essential facilities doctrine and,  395 injunctions and,  270–272 market power and,  339 as national champion,  54–55 networks and,  380 vertical mergers and,  570–572, 580 vertical separation and,  393–394 American Tobacco Company,  337 Anderson, Gary M.,  160 Anderson, Simon P.,  410 Andersson, Ola,  215 Antitrust Amendments Act of 1990,  151 “Antitrust Dilemma,”  148 Antitrust Division. See Justice Department Antonielli, Marco,  439 Apple,  156, 320, 380, 413 Approximately optimal profit,  217–218 Arcadian,  106, 108 Archer Daniels Midland,  181 Areeda, Phillip,  26, 85, 538, 544–545 Areeda-Turner rule,  336 Arms manufacturing, conferring market power to promote national security,  55 Armstrong, Mark,  205, 412–413, 418, 421, 433 Arrow, Kenneth,  10–11 Asch, Peter,  148 Ascione, Aurora,  182 Ashenfelter, Orley,  487 Asymmetric competition,  419 AT&T. See American Telegraph & Telephone Attorney incentives in settlements,  179–180 Attorneys’ fees in class actions,  199–203 contingent fees,  179–180, 199 lodestar method,  199 in settlements,  179–180 share-of-recovery method,  199–201

588   Index Auction models, unilateral effects analysis, 496–498 Avenal, Eric,  567–569 Azcuenaga, Mary L.,  73 Baby Bells,  270 Bagwell, Kyle,  247 Baidu, 419 Bain, Joe S.,  323, 338, 556 Baker, Jonathan,  512 Baker, Jonathan B.,  374 Baliga, Sandeep,  220–221 Bank Holding Company Act of  1956, 50 Banking industry, mitigating regulation, 50–51 Bank Merger Act of  1960, 50 Bargaining. See also Collective bargaining bargaining models, unilateral effects analysis, 498–500 hospital mergers and,  77–79 Barlet, Corinne,  567–569 Barros, Pedro P.,  122 BASFAG (cartel),  181 Bates, Laurie J.,  539 Baum, Lawrence,  160 Baumol, William J.,  163, 304 Baxter's Law,  393 Baye, Michael,  218 Baylor University Hospital,  95 Becker, Gary,  18, 20, 24, 256 Bedre-Defolie, Ozlem,  414, 418 Behavioral economics,  205–225 accounting anomalies and,  220–222 analysis of,  223–225 Bertrand markets and,  210–211, 218–221, 223 Cournot markets and,  209–213, 216, 224 economies of scale and,  205–206 esprit de corps,  212–215 imitative behavior,  208–212 overoptimism and,  218–220 overview, 205–208 price fixing and,  210 rationality of rivals, uncertainty regarding, 222 reasons for not pursuing maximum profit, 223–225

relative profit, concerns for,  208–212 satisfactory profit versus maximum profit, 215–218 in Stackelberg markets,  213 vengeful behavior,  212–215 Behavioral psychology,  148–149 Behringer, Stefan,  437 Bell Doctrine,  393 Bell System,  266, 270, 387 Benson, Bruce L.,  154, 160 Bentham, Jeremy,  24 Bernhardt, Dan,  173 Bernheim, B. Douglas,  516 Bertrand markets accounting anomalies and,  220–221 behavioral economics and,  223 coordinated effects analysis in,  511, 513, 517, 523, 525 explicit collusion in,  234 imitative behavior and,  210–211 overoptimism and,  219 price floors and,  239 satisfactory profit versus maximum profit in, 218 tacit collusion in,  231, 233 unilateral effects analysis in,  470, 476, 494–495 vertical mergers in,  561–562, 579 Besanko, David,  220–221 Bhattacharyya, Sugato,  535 Bilateral agreements consultation in,  127 international antitrust institutions and, 126–128 mutual assistance in,  127 negative comity in,  127 notification in,  126 positive comity in,  128 Bilateral monopoly,  364–378 antitrust policy and,  372–377 collusion, countervailing power through, 375–376 complications of,  372–374 cost curves in,  367–368 countervailing power and,  372–377. See also Countervailing power economics of,  365–372

Index   589

formula-price contracts in,  369–372 input price, limits on indeterminacy of, 368–369 mergers, countervailing power through, 376–377 overview,  364, 377–378 simple model,  366–369 theory of,  366 Binger, Brian R.,  231 Blair, Roger D.,  187, 364, 370 Block, Michael K.,  258 Blodgett Memorial Medical Center,  80 Blue Cross and Blue Shield Association,  63, 95–98, 105 Blue Cross and Blue Shield of Delaware,  92 Blue Cross and Blue Shield of Kansas,  99 Blue Cross and Blue Shield of Michigan,  65, 104–105 Blue Cross and Blue Shield of Montana, 106–107 Boeing, 122 Boone, Jan,  245 Bork, Robert H.,  3, 152, 299–300, 307, 457 Boudreau, Kevin J.,  438 Boudreaux, Donald J.,  153 “Bounty hunter” model,  282–283 Bourjade, Sylvain,  175 Bowley, A.L.,  366–367 BP (British Petroleum),  566 Bradley, Robert L., Jr.,  154 Brady, Gordon L.,  151 Brandeis, Louis (Justice),  13 Brandts, Jordi,  247–248 Brazil, settlements in,  182n14 Breuckner, Jan,  490 Breyer, Stephen (Justice),  298 Briggs, Hugh C.,  175 British Petroleum (BP),  566 Brown Shoe Co.,  551 Buchanan, James M.,  148, 159–160, 162, 164 Budzinkski, Oliver,  119, 122–123, 132, 134 Bundling in Cournot markets,  246–247 in European Union,  435 as exclusionary conduct,  245–248 experimental economics,  245–248 hospitals and,  86–87

by multisided platform businesses,  435–436 restraint of trade and,  305–306 in Stackelberg markets,  246–247 Bureaucracy, public choice theory of antitrust law and,  157–159 Bureaucratic incentives hypothesis,  154 Bush, George H.W.,  570 Business networks,  381n3 Butterworth Hospital,  80 Buyer mergers,  530–540 buyer power defined,  533–534 competitive supply, buyer power with, 531–532 efficiencies from increase in buyer power, 535 empirical analysis, need for,  539–540 in healthcare markets,  532 Merger Guidelines and,  532, 537–539 “mirror image” question,  530 monopsony and,  531–532 overview, 530 potential competitive harm from increase in, 536–537 seller mergers compared,  530n3 upstream competition, effect on, 536–537n20 where suppliers not price takers,  533–540 Buyer power,  529–547 in buyer mergers,  530–540. See also Buyer mergers overview,  529–530, 547 rationale for antitrust law and,  11–12 in supplier mergers,  540–546. See also Supplier mergers Cable television, vertical mergers in,  565, 572–573 Cabral, Luís M.,  122 Cabrales, Antonio,  247–248 CAFA. See Class Action Fairness Act of 2005 Caliskan, Anil,  246–247 Calvano, Emilio,  414, 418 Campbell, Tom,  374, 544 Canada Competition Bureau Canada,  541–544 mergers in,  9 supplier mergers in,  541–544

590   Index Capitalism, Socialism and Democracy (Schumpeter), 10 Capper-Volstead Act of  1922, 49, 57 Capps, Cory S.,  63, 77–78, 81 Capra, C. Monica,  241 Cargill, 99 CARICOM Competition Commission,  131n13 Carlton, Dennis W.,  529 Carnival Corporation,  520 Carstensen, Peter C.,  33, 538–539 Cartels actual punishment policy and,  22–23 coordinated effects analysis and,  515 empty core and,  40–41 in European Union,  129n12 explicit collusion and,  234 imprisonment and,  264–265 national champions, conferring market power to protect,  54 optimal deterrence and,  261 sanctions imposed on,  257 tacit collusion and,  230 in United Kingdom,  214 vengeful behavior and,  214 Cason, Timothy N.,  236 Celler-Kefauver Act of  1950, 13, 151, 333 Cement industry, vertical mergers in,  566n21, 566n22 Certification of class,  190–191 Chakraborty, Ratula,  535–536 Chamberlain, Lawrence H.,  152 Chamberlin, Edward H.,  229–231, 326, 331, 384, 422 Chandra, Ambarish,  427, 429, 439 Chao, Yong,  436 Charitable Gift Annuity Antitrust Relief Act of  1995, 51, 53 Charitable organizations, conferring market power to achieve social goals,  53 Charness, Gary,  247–248 Chen, Yongmin,  568 Chicago School exclusive dealing and,  243 rationales for antitrust law and,  3, 7, 13 vertical merger policy, critique of,  551, 553–559 China Anti-Monopoly Law of  2007, 5

coordinated effects analysis in,  515 development of law in,  29 evidence in,  30 importance of enforcement regime,  17 international externalities of national competition policies and,  122–123 multisided platform businesses in,  406 Choi, Jay Pil,  435–436, 568–569 Chowdhury, Subhasish M.,  433 Church, Jeffrey,  566–570 Church White Paper,  567–568 Cigna, 95 Cinemax, 565 Clabault, James M.,  258 Class Action Fairness Act of 2005 (CAFA) class actions under,  201–203 damages under,  260 overview, 187 Class actions,  187–203 attorneys' fees in,  199–203 CAFA and,  187, 201–203 class certification,  190–191 common proofs,  191–193 coupons, 199–203 deterrence and,  189 economic rationale for,  188–189 Federal Rules of Civil Procedure and, 190–191 frequency of,  188n3 indirect purchaser actions,  193–194 overview,  187, 203 practical problems,  191–201 predominance requirement,  190–191 prevalence of,  194–195 price and,  192–193 risk neutrality and,  189n4 settlements of,  195–199 treble damages in,  192, 197 Class certification,  190–191 Clayton Act of  1914 bilateral monopoly and,  364, 374–375, 378 class actions under,  187–190, 194 efficiencies and,  453–454, 457 evidence under,  30 exclusionary conduct under,  239 historical background,  151–152, 154 mergers under,  333, 335, 376 remedies under,  255

Index   591

treble damages under,  259 vertical mergers and,  551 Clinton, Bill,  570 CNN, 572–573 Coal industry, countervailing power and,  375 Coase, Ronald H.,  391, 559 Coca-Cola, 551–553 Coleman, Mary,  529 Collard-Wexler, Allan,  427, 429, 439 Collective bargaining efficiencies, creating countervailing power to facilitate,  49–50 by physicians,  377 Colleges and universities, conferring market power to achieve social goals,  53–54 Collusion in Bertrand markets,  231, 233, 234, 235 cartels and,  230, 234 conscious parallelism,  235–237 coordinated effects analysis and,  513, 516–519 countervailing power through,  375–376 in Cournot markets,  231–232 in European Union,  293–294 experimental economics,  230–239 explicit collusion,  234–235 Nash equilibrium and,  231–233 nonbinding price announcements,  235–237 oligopoly and,  231, 234–235 overview, 230 posted-offer experiments,  234n6 price ceilings,  237–239 price fixing and,  234 price floors,  237–239 private enforcement actions and,  292–294 tacit collusion,  231–233 technology, effect of,  214–215 in United States,  293 vengeful behavior and,  212–215 Comity, international antitrust institutions and, 126–128 Commerce Department, Export Trade Certificate Program,  44n22 Commitments, vertical mergers and,  574–579 Comparative transaction costs, vertical mergers and,  559 Compensation, deterrence versus,  259–260, 282–287

Competition asymmetric competition,  419 competitors versus,  9–10 diversity of competition regimes, international antitrust institutions and, 124 healthcare markets, evolving nature of competition in,  68–69 international externalities of national competition policies, international antitrust institutions and,  121–123 loopholes in protection of competition, international antitrust institutions and, 123 monopoly compared,  316, 319 multisided platform businesses, among, 415–419 prior failures in market regulation, mitigating, 51 procompetitive versus anticompetitive conduct, 121n4 profit and,  207–208 public choice theory of antitrust law and, 163–164 restraint of trade. See Restraint of trade restriction on competition, preventing, 331–333 wealth transfers and protection from. See Wealth transfers and protection from competition “Competitive bottlenecks,”  418 Competitive supply, buyer power with, 531–532 Competitors competition versus,  9–10 coordination among to achieve efficient market, 44–48 rationality of rivals, uncertainty regarding, 222 Complete deterrence,  20–22 Component production, achieving efficient scale and,  42–44 Compte, Olivier,  524 Computer industry, injunctions and,  271–274 Concentration effects,  489n58 Conditioning, hospitals and,  86–87 Conduct injunctions,  270–274 Confidentiality in settlements,  173–174

592   Index Conglomerate mergers,  552n3 Congress, public choice theory of antitrust law and, 157–159 Connor, Robert A.,  69, 257, 265 Conscious parallelism,  235–237 Consent decrees, settlements and,  183 Consultation in bilateral agreements,  127 Consumers, price fixing and,  288–289 Consumer Watchdog,  406 Consumer welfare efficiencies and,  460 hypothesis, 153 multisided platform businesses and, 413–414 restraint of trade and,  299–300, 307 Contestability theory,  383 Contingent attorneys' fees,  179–180, 199 Cooper, David J.,  247 Cooperation, efficiencies and,  452–453 Cooperative strategy in international antitrust institutions, 126–128 Coordinated effects analysis,  509–526 in Bertrand markets,  511, 513, 517, 523, 525 cartels and,  515 checklist of market factors,  513–516 in China,  515 collusion mechanism,  513, 516–519 in Cournot markets,  513, 517, 525 in European Union,  514, 517 limitations of,  524–526 market definition and,  358 “mavericks” and,  512 mergers and,  333–334 overview, 509–510 parallel accommodating conduct,  522–524 price complexity, implications of,  513, 519–521 structural presumption,  510–513 unilateral effects analysis compared,  466n1 in United States,  514, 517 vertical mergers and,  469n24 Coordinated Price Pressure Index (CPPI), 523–524 Core, restraint of trade and,  299–301 Corts, Kenneth S.,  434 Cost curves in bilateral monopoly,  367–368

in economies of scale,  325 Countervailing power bilateral monopoly and,  372–377 coal industry and,  375 collusion, through,  375–376 efficiencies, to facilitate,  49–50 labor unions and,  375 mergers, through,  376–377 motion picture industry and,  375–376 split agreements and,  375–376 Coupons in class actions,  199–203 Cournot, Augustin,  366, 558 Cournot markets behavioral economics and,  224 bundling in,  246–247 coordinated effects analysis in,  513, 517, 525 explicit collusion in,  235 imitative behavior and,  209–212 satisfactory profit versus maximum profit in, 216 SSCO model,  560–561 tacit collusion in,  231–232 unilateral effects analysis in,  355, 492–493 vengeful behavior and,  213 vertical mergers in,  560–561, 567, 575–578 Coursey, Don,  237–238 CPPI (Coordinated Price Pressure Index), 523–524 Crandall, Robert W.,  268 Crane, Daniel A.,  3, 262–263 Cremieux, Peirre,  183 Cross-border externalities, international antitrust institutions and,  121–123 Cross-elasticities, market definition and, 359–361 Cross-market variation, unilateral effects analysis, 490–491 Current procedural terminology (CPT) code,  66, 90 Cyert, Richard,  216 Dafny, Leemore,  104–106 Dallas Morning News, 433 Dallas Times Herald, 433 Damages agency costs and,  262–263

Index   593

deterrence versus compensation,  259–260 optimal deterrence,  260–261 punitive damages in private enforcement actions, 285 as remedy,  259–263 settlements and,  261–262 Daugherty, Andrew F.,  174 Davidson, Carl,  495 Davis, Douglas D.,  230, 233, 236–237, 248 Deaton, Angus,  482 de Bodt, Eric,  155 Deficiencies of multiple procedures in international antitrust institutions,  123 Dell, 558 Demand curves in monopoly,  316, 322, 326–327 Demand externalities, market failure and, 45–46 Deneckere, Raymond,  495 DePasquale, Christina,  364 Derdenger, Timothy,  436 Destructive competition, market failure and, 38–42 Deterrence agency costs and,  262–263 class actions and,  189 compensation versus,  259–260, 282–287 complete deterrence,  20–22 optimal deterrence,  260–261 settlements and,  261–262 Developing countries, ICN in,  135 Diagnosis related groups (DRGs),  66 Diagonal mergers,  551n1 Differentiated products, market definition and, 357 DiLorenzo, Thomas J.,  153, 164 Dinc, Serdar,  155 Diners Club,  410–411, 415 Direct purchasers, private enforcement actions by,  284, 287 Direct regulation, exemptions facilitating transition to market orientation,  57–58 Discover,  415, 432 Dissolutions, injunctions and,  268–269 Divergence between opposing views of antitrust law,  8–9 Diversion ratios,  472n15

Divestiture injunctions and,  267–270 of networks,  393 Dixit, Avinash,  247 Dixon, Huw,  217 Dobson, Paul W.,  535–536 Doctors. See Physicians Doganoglu, Toker,  433–434 Doha Declaration,  130 DOJ. See Justice Department Dolbear, F.T.,  231 Dominant firms economic and policy approaches,  313–340 exclusionary conduct, preventing,  336–337 Lerner Index and,  317, 320, 325–327, 331, 337 market delineation and,  327–331 market power in,  325–327, 331–339. See also Market power mergers, preventing,  333–334 overview,  313–315, 339–340 predatory pricing, preventing,  334–336 reactive fringe and,  320–323 restriction on competition, preventing, 331–333 structural dismemberment and,  337–339 unilateral effects analysis,  495 unilateral exercise of market power, policies to prevent,  314, 331–339 Double marginalization, vertical mergers and, 557–559 Douglas, William O.,  269 Dowd, Bryan E.,  69 Downs, Anthony,  158 Dranove, David,  63, 68, 77–78, 81, 96, 106 DRGs (Diagnosis related groups),  66 Dufwenberg, Martin,  238–239 Duggan, Mark,  104–106 Duopoly accounting anomalies and,  220 imitative behavior and,  209–210 satisfactory profit versus maximum profit in, 217–218 upstream duopoly,  242n11 du Pont,  330 Duso, Tomaso,  155 Dynamic efficiency mergers and,  459–460

594   Index Dynamic efficiency (cont.) in networks,  392–393 as rationale for antitrust law,  10–11 Eaton, Jonathan,  221 eBay, 320 Economic justifications for exemptions,  34 “Economic Welfare and the Allocation of Resources for Invention” (Arrow),  11 Economies of scale behavioral economics and,  205–206 cost curves in,  325 learning curves in,  324n25 monopoly, as entry barrier for,  324 Edgeworth, F.Y.,  300, 494–495 Edlin, Aaron,  298 Effects doctrine, extraterritorial enforcement through, 125–126 Efficiencies, 451–464 agency analysis of,  458–464 in buyer mergers,  535 buyer power, from increase in,  535 consumer welfare and,  460 cooperation and,  452–453 countervailing power to facilitate, creating, 49–50 defense of mergers,  453–458 dynamic efficiency. See Dynamic efficiency in European Union,  460, 462–463 evolution of doctrine,  453–458 exclusionary conduct and,  458–459 hospital mergers and,  79n50, 83 importance in antitrust law,  451–453 marginal costs and,  459 mergers and,  459–464 in networks,  392–393 overview,  451, 464 in practice,  463–464 price fixing and,  20n5, 452–453 proof of,  461–463 as rationale for antitrust law,  5–8, 10–11 static efficiency. See Static efficiency unilateral conduct and,  452 in United States,  460 variable costs and,  459 vertical restraints and,  457–458 Eigen-Zucchi, Christian,  154

Einav, Liran,  497 Eisenberg, Theodore,  179 Eisenmann, Thomas R.,  419 Ekelund, Robert B.,  154 Elasticities, market definition and,  359–361 Electric utilities, facilitating transition from state ownership or direct regulation to market orientation,  58 Elzinga, Kenneth G.,  75, 77 Elzinga-Hogart method,  72–75 Emch, Eric,  425 Empty core cartels and,  40–41 market failure and,  38–42 transportation and,  41 Enforcement regimes,  17–31 actual punishment policy,  21–25 complete deterrence,  20–22 development of law,  27–29 effects doctrine, extraterritorial enforcement through,  125–126 efficiencies and,  458–464. See also Efficiencies evidence, 29–30 exemptions improving efficiency of,  58–59 harm internalization,  20–22 imprisonment, 23 optimal enforcement policy,  18–21 overview,  17–18, 30–31 predatory pricing,  25–27 private enforcement actions. See Private enforcement actions procedure,  27, 29–30 public choice theory of antitrust law and, 155–162 substantive standards,  25–27 in United Kingdom,  149 Engelmann, Dirk,  238 England. See United Kingdom Entry barriers in health insurance,  100–101 Entry effects,  489n58 Erel, Isil,  155 Esprit de corps,  212–215 Essential facilities doctrine, networks and, 395–397 European Coal and Steel Community of  1952, 13 European Commission

Index   595

enforcement by,  149 horizontal merger guidelines,  354, 462–463, 467 injunctions imposed by,  272–273 nonhorizontal merger guidelines,  554, 568–571, 580 European Community Treaty actual punishment under,  21 development of law under,  27–28 enforcement under,  149 exclusionary conduct under,  239 European Court of First Instance,  28n8, 155n14 European Court of Justice and General Court, competition versus competitors in,  9 European Union actual punishment policy in,  21–25 airline industry in,  128n11 bundling in,  435 cartels in,  129n12 collusion in,  293–294 comity in,  128 Commission Notice on the Definition of Relevant Market,  354 Competition Law,  27 Competition Policy System,  119n1 competition versus competitors in,  9 conscious parallelism in,  235 consumers, price fixing and,  288–289 coordinated effects analysis in,  514, 517 development of law in,  27–29 DG Competition Discussion Paper,  354 effects doctrine, extraterritorial enforcement through,  125–126 efficiencies in,  460, 462–463 evidence in,  29–30 exclusionary conduct in,  346, 350 fines in,  257 importance of enforcement regime,  17 leniency programs in,  181 lost profits in,  294–295 merger simulation in,  486, 498 mergers in,  9, 155 monopsony in,  293–294 multisided platform businesses in,  406 nonbinding price announcements in,  235 predatory pricing in,  25–27

private enforcement actions in,  281–282, 285–287, 295–296 public enforcement actions, settlements in, 180–183 resellers under imperfectly competitive conditions, price fixing and,  291 supplier mergers and,  541–543 tying in,  435 unilateral effects analysis in,  467 vertical mergers in,  551–552, 554, 568–570 Evans, David S. on asymmetric competition,  419 on coordinated behavior,  440 on efficiencies,  438 on exclusionary conduct,  431 on market definition,  424 on multisided platform businesses,  404, 405–406, 409, 411 on product differentiation,  417 Evanston Northwest Healthcare,  71, 76–77 Evidence in enforcement regimes,  29–30 Excess inertia,  386–387 Excess momentum,  387 Exclusionary conduct,  239–248 bundling, 245–248 efficiencies and,  458–459 in European Union,  346, 350 exclusive dealing,  243–245 foreclosure, 242–243 by hospitals,  84–89 limit pricing,  247n16 multisided platform businesses and, 431–438 overview, 239 predatory pricing,  240–242 preventing, 336–337 rebates, 245–246n15 in United States,  346, 350 upstream duopoly,  242n11 Exclusive dealing experimental economics,  243–245 by multisided platform businesses,  433–434 by newspapers,  433 video games and,  433–436 Exclusive territories in health insurance, 104–105 Exemptions, 33–59

596   Index Exemptions (cont.) economic justifications for,  34 efficiency of enforcement, improving, 58–59 market failure and,  37–51. See also Market failure natural monopoly and,  35–36 overview,  33–35, 59 rent seeking and,  34 transition from state ownership or direct regulation to market orientation, facilitating, 57–58 wealth transfers and protection from competition and,  51–57. See also Wealth transfers and protection from competition Experimental economics,  229–249 bundling, 245–248 collusion, 230–239. See also Collusion exclusionary conduct,  239–248 exclusive dealing,  243–245 foreclosure, 242–243 overview,  229–230, 248–249 predatory pricing,  240–242 Explicit collusion,  234–235 Externalities in China,  122–123 demand externalities, market failure and, 45–46 international externalities, international antitrust institutions and,  121–123 membership externalities,  408, 410–411 multisided platform businesses and,  407–408, 410–411 negative externalities, ICN and,  136 network externalities, market failure and, 45–46 in Russia,  122–123 supply externalities, market failure and, 46–47 usage externalities,  407–408, 410 F. Hoffman-La Roche Ltd.,  181 Facebook,  380, 406 Fairness as rationale for antitrust law,  5–8 Faith, Roger L.,  158 Fan, Ying,  429–430, 439

Farrell, Joseph,  298, 374, 427, 468, 493 Favored activities, subsidization of,  56 Favored classes of market participant, wealth transfers to,  56–57 Federal Communications Commission (FCC),  394, 565n19, 570 Federal Reserve System, payment cards and, 415 Federal Rules of Civil Procedure, class actions under, 190–191 Federal Trade Commission (FTC) bargaining models and,  500 Bureau of Economics,  157 buyer mergers and,  532, 537–538 competition law doctrine, mitigating,  51 coordinated effects analysis and,  515, 518, 520–522 efficiencies and,  454, 460, 463 enforcement by,  149–150, 157, 159 evidence and,  30 General Counsel,  30 healthcare markets and,  63, 65, 69 hospital mergers and,  70–74, 76–80, 82–83, 108–109 Merger Guidelines. See Merger Guidelines mergers and,  333, 376–377 multisided platform businesses and,  406 noncompete agreements and,  333 physician price fixing and fee schedules and, 90–94 remedies sought by,  255 Statement of Antitrust Enforcement Policy in Health Care,  91 supplier mergers and,  543 vertical mergers and,  565–566, 572 Federal Trade Commission Act of  1914, 151–152, 154, 255, 305–306, 306n22 Federation of Physicians and Dentists,  92 Fee, Edward C.,  535 Fee schedules, physicians and,  90–92 Fehr, Ernst,  235 Feldman, Roger D.,  69 Fibramold, 573 Filistrucchi, Lapo,  427, 429, 437, 439 Fines in European Union,  257 as remedy,  257–258

Index   597

settlements and,  180–183 in United States,  257 Fishermen's Collective Marketing Act of  1934, 49, 57 Fixed costs, networks and,  382–384 Fleck, Robert K.,  160 Fonseca, Miguel A.,  231, 233–235 Food and Drug Administration,  332–333 Foreclosure defined, 564n16 experimental economics,  242–243 Formula-price contracts in bilateral monopoly, 369–372 Fouraker, Lawrence E.,  231 Fox News Channel,  572–573 Freedom to trade. See Restraint of trade Freiburg School,  13 Friedman, Daniel,  230 Friedman, James W.,  234 Friedman, Milton,  207–208, 224–225 Froeb, Luke,  468 FTC. See Federal Trade Commission Fumas, Vicente Salas,  211 Gabszewicz, Jean J.,  410 Gächter, Simon,  235 Game theory, settlements and,  172n1 Ganslandt, Mattias,  513, 526 Garvin, Susan,  247 Gary, Elbert,  214 “Gary Dinners,”  214 Gayle, Wayne-Roy,  524 Gaynor, Martin,  77, 81 General Electric,  122 Gentzkow, Matthew,  439 Geographic boundaries of competition in health insurance,  96–97 Georgetown Law School,  175, 178 Germany, antitrust law in,  13 Gibbons, Robert,  211 Gilbert, Richard J.,  566, 568, 582 Giuliani, Rudolph,  572–573 Global Competition Initiative (DOJ),  131 Goals of antitrust law, restraint of trade and, 307 Goeree, Jacob K.,  240 Gomez, Rosario,  240

Google,  274, 320, 380, 406, 419 Gordon, R.A.,  215 Gordon, Sanford D.,  152 Gosman, Martin L.,  539 Gowrisankaran, Guatam,  78 Grand unified theory (GUT),  149 Grandy, Christopher,  153 Granger Movement,  153 Great Britain. See United Kingdom Green, Edward J.,  514 Greenhut, M.L.,  154 Greenlee, Patrick,  246 Grinnel Corporation,  338 Gron, Anne,  96, 106 Grossman, Gene,  221 Gross Upward Market Power Pressure Index (GUMPPI), 493 Gross Upward Pricing Pressure Index (GUPPI),  472–474, 493, 523–524 GUT (Grand unified theory),  149 Hagiu, Andrei,  434, 438 Halaburda, Hanna,  417 Hall, R.L.,  220 Hand, Learned,  7, 352–354 Handel, Benjamin,  97 Hanssen, F. Andrew,  160–161, 269 Harm internalization,  20–22 Harrington, Joseph E., Jr.,  523 Harrison, Glenn W.,  240–241 Harrison, Jeffrey L.,  281 Harstad, Ron,  236–237 Hart, Oliver,  563 Hart-Scott-Rodino Act of  1976, 151, 255, 268–269 Hart-Tirole model of vertical mergers, 563–564 Harvard School,  13 Hastings, Justine S.,  566 Hausman, Jerry A.,  501 HBO, 565 Head, Keith,  122 Healthcare markets,  63–110 accountable care organizations and, 109–110 bargaining models and,  499–500 buyer mergers in,  532

598   Index Healthcare markets (cont.) current procedural terminology (CPT) code,  66, 90 diagnosis related groups (DRGs),  66 evolving nature of competition in,  68–69 health insurance,  94–108. See also Health insurance health maintenance organizations (HMOs), 106–107 hospitals, 69–89. See also Hospitals integrated delivery systems (IDSs),  109 market mechanisms,  65–69 “medical arms race,”  68–69 mergers in,  377 in Netherlands,  65–66n6 overview,  63, 65 physicians, 89–94. See also Physicians preferred provider organizations (PPOs),  67, 69, 106–107 pricing in,  66–67 reform, effect of,  108–110 supplier induced demand in,  68n9 usual, customary, and reasonable (UCR) fees,  67, 97 value chain in,  65–66 Health insurance,  94–108 Aetna-Prudential, 102–103 divestitures, 106 empirical evidence regarding monopsony, 103–104 entry barriers,  100–101 exclusive territories,  104–105 geographic boundaries of competition, 96–97 health maintenance organizations (HMOs), 106–107 insurer market structure,  97–98 managed care,  96 market share and,  96–97n96 Medicaid, 101–102 Medicare, 101–102 mergers, 106–108 MFN clauses and,  104–105 monopoly and,  99–100, 105–108 monopsony and,  98–105 overview, 94 preferred provider organizations

(PPOs),  67, 69, 106–107 private health insurance,  95 selective contracting,  97–98 self-insurance, 67n8 United-PacifiCare, 102–103 in United States,  95 Health maintenance organizations (HMOs), 106–107 Heifetz, Aviad,  219 Helland, Eric A.,  179 Herfindahl-Hirschman Index (HHI) in healthcare markets,  69n10, 72, 81, 106 market definition and,  350, 355–358, 361 unilateral effects analysis and,  473, 492 Hewlett-Packard, 558 Hicks, John,  7 Highland Park Hospital,  71, 76–77 Hinloopen, Jeroen,  246–247 Historical background of public choice theory of antitrust law,  151–155 Hitch, C.J.,  220 HMOs (Health maintenance organizations), 106–107 Hoffman, Elizabeth,  231 Holcombe, Randall G.,  154 Holt, Charles A.,  230, 233, 236–237, 248 Home of Websites (software),  419 Homogeneous goods market definition and,  355–357 unilateral effects analysis and,  355–357, 491–495 Honeywell, 122 Horizontal agreements, restraint of trade and, 301–303 Horizontal differentiation,  467n6 Horizontal integration, injunctions and,  270 Hortaҫsu, Ali,  566 Hosken, Daniel,  486 Hospital mergers,  70–83 bargaining and,  77–79 critical loss analysis,  75–76n37 efficiencies and,  79n50, 83 Elzinga-Hogart method,  72–75 empirical evidence,  499n85 Evanston,  71, 76–77 leverage and,  77–79 litigated cases,  64

Index   599

market definition and,  72–74 nonprofit status and,  79–82 overview, 70–72 quality and,  83 state action immunity and,  82–83 willingness-to-pay and,  77–79 Hospitals, 69–89 bargaining models and,  499–500 bundling and,  86–87 conditioning and,  86–87 exclusionary conduct by,  84–89 mergers, 70–83. See also Hospital mergers Ortho test,  88n75 overview,  69–70, 89 PeaceHealth, 87–88 staff privileges,  84–86 tying and,  86–87 United Regional, 88–89 Hovenkamp, Herbert on behavioral economics,  225 on buyer mergers,  538 on buyer power,  544–545 on healthcare markets,  85 on injunctions,  269 on rationales for antitrust law,  7, 13 on restraint of trade,  299, 307 Huck, Steffen,  205, 213, 216, 224, 231 Humana,  95, 106, 108 Huryn, Kathleen D.,  175 Hylton, Keith N.,  17 Hypothetical monopolist test,  354–358, 424 Iaryczower, Matias,  161 IBM,  183, 320 ICC (Interstate Commerce Commission), mitigating regulation,  50 ICN. See International Competition Network Imitative behavior,  208–212 Imprisonment as remedy,  23, 263–265 Inderst, Roman,  534–535 India, Competition Law of  2002, 5 Indicative price rise (IPR) analysis,  475n28 Indirect network effects,  405 Indirect purchasers class actions,  193–194 private enforcement actions by,  284, 287 Industry elasticities,  481n43

Information pooling, market failure and,  47–48 Injunctions computer industry and,  271–274 conduct injunctions,  270–274 dissolutions and,  268–269 divestiture, 267–270 horizontal integration and,  270 mergers and,  267–269 monopoly and,  267–268 newspapers and,  271 overview, 265–266 structural injunctions,  266–270 vertical integration and,  269 Injury requirement for private enforcement actions, 283–287 Innovation networks and,  384 unilateral effects analysis and,  500–501 Institutional failure. See Market failure Insufficient friction,  387 Insurance industry health insurance,  94–108. See also Health insurance market facilitation in,  48 Intel Corporation,  183, 320, 553, 558, 582 Intended beneficiaries of antitrust law,  4–5 Interest groups, public choice theory of antitrust law and,  162–164 Internal Revenue Service, health insurance and, 95 International antitrust institutions,  119–141 bilateral agreements and,  126–128 challenges facing,  135–141 comity and,  126–128 cooperative strategy,  126–128 deficiencies of multiple procedures in,  123, 137 diversity of competition regimes and,  124, 138 economics and,  120–125 effects doctrine, extraterritorial enforcement through,  125–126 ICN. See International Competition Network international externalities of national competition policies and,  121–123 loopholes in protection of competition and,  123, 137–138

600   Index International antitrust institutions (cont.) multilateral strategy,  129–130 multilevel lead jurisdiction model,  139–141 network strategy,  130–134 overview,  119–120, 141 strategies of,  125–134 trade agreements, in,  129–130 unilateral strategy,  125–126 unsolved problems in,  135–141 International Competition Network (ICN) Advocacy Working Group,  133 Agency Effectiveness Working Group,  132 Cartel Working Group,  132 consensual best-practices recommendations, 135–136 deficiencies of multiple procedures in,  137 in developing countries,  135 diversity of competition regimes and,  138 enforcement cooperation and,  139n17 limitations of,  136–139 loopholes in protection of competition and, 137–138 Merger Guidelines Workbook, 541 Merger Working Group,  133 negative externalities and,  136 network strategy and,  130–134 overview,  119–120, 128, 141 success of,  135–136 timeline of working groups,  134 Unilateral Conduct Working Group,  133 International externalities, international antitrust institutions and,  121–123 Internet. See Networks Internet Explorer (browser),  267, 271 Interstate Commerce Commission (ICC), mitigating regulation,  50 Intertemporal variation, unilateral effects analysis, 488–490 “Invisible hand,”  217 iOS (operating system),  413 IPR (Indicative price rise) analysis,  475n28 Isaac, R. Mark,  231, 234, 237–238, 240 Israel, Mark,  529

Jefferson Parish Hospital,  85 Jeld-Wen, 573 Jenny, Frédéric,  524 Jeziorski, Przemslaw,  426, 430 Judicial review, public choice theory of antitrust law and,  159–162 Jullien, Bruno,  432, 435 Jung, Yun Joo,  222, 241 Jury trial, settlements and,  178n8 Justice Department (DOJ) Antitrust Guidelines for Collaborations among Competitors,  37n4 buyer mergers and,  532, 537–539 consent decrees and,  183 coordinated effects analysis and,  515, 519, 522 Corporate Leniency Program,  174, 176, 180–183 countervailing power and,  376 Economic Analysis Group,  555 efficiencies and,  454, 460 enforcement by,  149–150, 156, 159 exclusionary conduct by hospitals and,  86, 89 Global Competition Initiative,  131 healthcare markets and,  63, 65, 69 health insurance and,  99, 102–108 hospital mergers and,  70–74, 83 imprisonment statistics,  263–264 International Competition Policy Advisory Committee, 131 Merger Guidelines. See Merger Guidelines mergers and,  333, 376 physician price fixing and fee schedules and, 91 Policy Guide to Merger Remedies,  268, 274 remedies sought by,  254 Statement of Antitrust Enforcement Policy in Health Care,  91 structural dismemberment and,  337–339 supplier mergers and,  543 vertical mergers and,  554–556, 571, 573–574 Workload Statistics,  263–264

Japan, antitrust law in,  13 Java (software),  271–272 Jayaratne, Jith,  509

Kagel, John H.,  222, 241, 247 Kaplow, Louis,  264–265, 345, 348, 351–352, 355 Kaserman, David L.,  370

Index   601

Katz, Michael,  388 Kaysen, Carl,  152, 338 Keating, Bryan,  466 Keeler, Emmet,  81 Kell West Regional Hospital,  88 Kelly, Trish,  539 Kingsbury Commitment,  380 Kinney, 551 Kirkwood, John B.,  535–537, 544 Klein, Tobias J.,  427, 429 Klick, Jonathan,  179 Klodt, Henning,  122–123 Kneuper, Robert,  258 Kodak, 320 Kopit, William G.,  79–80 Kovacic, William,  524 Krattenmaker, Thomas G.,  570 Kreps, David M.,  222, 241, 495 Krishnan, Ranjani,  81 Ku6 (software),  419 Kuhn, Kai-Uwe,  516 Kwoka, John E.,  274 Labor unions countervailing power and,  375 favored classes of market participants, wealth transfers to,  57 Lande, Robert H.,  154, 257, 265 Landeo, Claudia M.,  244–245 Landes, William M.,  18, 160, 256, 348, 351 Langenfeld, James,  258 Larouche, Pierre,  239 League of Nations World Economic Conference, 129 Learning curves in economies of scale,  324n25 Leavens, Donald R.,  158 Lederman, Mara,  434 Lee, Darin,  490 Lee, Robin S.,  433–434 Leniency programs in European Union,  181 settlements and,  180–183 Lerner Index dominant firms and,  317, 320, 325–327, 331, 337 market definition and,  347 market power and,  325–327

Leslie, Christopher,  225 Levenstein, Margaret C.,  515 Leverage, hospital mergers and,  77–79 Levin, Daniel,  222, 241 Li, Ting,  435 Libecap, Gary D.,  231 Limit pricing,  247n16 Linear demand,  478–480 Linear tariffs,  534 Linux (operating system),  273 List, John,  224 Lodestar method of attorneys' fees,  199 Logit demand,  478, 480–482, 485n53 Lost profits in European Union,  294–295 private enforcement actions and,  294–295 in United States,  294–295 Loyalty pricing, restraint of trade and, 305–306 Ludwick, Richard,  81 “Lumpiness” of costs,  40n7 Lundgren, Carl,  212 Lynk, William,  80–81 Magnuson-Moss Warranty/Federal Trade Commission Improvement Act of  1976, 151 Makris, Miltos,  158 Malmendier, Ulrike,  220 Managed care bargaining models and,  499–500 health insurance,  96 Mandalay Resort Group,  520–521 March, James,  216 Marginal costs, efficiencies and,  459 Market definition,  345–362 coordinated effects analysis,  358 criterion for,  348–350 differentiated products,  357 elasticities versus cross-elasticities,  359–361 homogeneous goods,  355–357 hospital mergers and,  72–74 hypothetical monopolist test,  354–358 incoherence of process,  346–350 Lerner Index and,  347 market power and,  346–348 market redefinition,  346–348 market share and,  346–348

602   Index Market definition (cont.) Merger Guidelines and,  354 multisided platform businesses and, 423–426 overview,  345–346, 361–362 thresholds of market share,  350–354 unilateral effects analysis,  355–357 Market delineation in dominant firms,  327–331 market power, to assess whether firm possesses, 329–331 proposed action, dealing with,  328–329 SSNIP test,  328–330 Market facilitation, market failure and,  48–49 Market failure,  37–51 competition law doctrine, mitigating,  51 coordination among competitors to achieve efficient market and,  44–48 countervailing power to facilitate efficiencies, creating,  49–50 demand externalities and,  45–46 destructive competition and,  38–42 efficient scale for component production, achieving, 42–44 empty core and,  38–42 information pooling and,  47–48 “lumpiness” of costs and,  40n7 market facilitation and,  48–49 network externalities and,  45–46 networks and,  386, 389 overview, 37–38 supply externalities and,  46–47 theories of,  37–38 Market orientation, exemptions facilitating transition to,  57–58 Market power in dominant firms,  325–327, 331–339 empirical evidence,  345n1 exclusionary conduct, preventing,  336–337 Lerner Index and,  325–327 market definition and,  346–348 market share and,  346–348 mergers, preventing,  333–334 multisided platform businesses and,  422– 423, 425–426 noncompete agreements and,  332–333 predatory pricing, preventing,  334–336 restriction on competition,

preventing, 331–333 specific goals, conferring to achieve,  52–55 structural dismemberment and,  337–339 unilateral exercise of, policies to prevent,  314, 331–339 Market redefinition,  346–348 Market share health insurance and,  96–97n96 market definition and,  346–348 market power and,  346–348 thresholds, 350–354 Martin, Stephen,  236–237, 242–243, 433 Mason, Charles F.,  247 Masonite,  519, 573 MasterCard,  330, 415 “Mavericks,” 512 Maximum profit competition and,  207–208 reasons for not pursuing,  223–225 satisfactory profit versus,  215–218 Mazzeo, Michael J.,  96, 106 McBride, Mark E.,  175 McCann, Robert W.,  79–80 McCarran-Ferguson Act of  1945, 47 McCaw,  570–572, 580 McChesney, Fred S.,  147, 149, 156 McCormick, Robert E.,  157, 162 McDonald, Michael J.,  154 McDonnell Douglas,  122 McFadden, Daniel,  485 MCI, 394 McKay, Amy Melissa,  157 McKenzie Hospital,  86–88 Media, advertising-supported mergers, 428–430 Medicaid,  66, 101–102 “Medical arms race,”  68–69 Medicare,  66, 101–102 Advantage plans,  106–108 Melnick, Glenn A.,  68, 81 Membership externalities,  408, 410–411 Menger, Carl,  366 Merger Guidelines buyer mergers and,  532, 537–539 buyer power and,  547 coordinated effects analysis and,  509–510, 513–515, 522, 524–525 countervailing power and,  376–377

Index   603

dominant firms and,  328–330 efficiencies and,  454–462 evolution of,  454–458 hypothetical monopolist test and,  354–356 multisided platform businesses and,  427 public choice theory and,  150–151 supplier mergers and,  541 unilateral effects analysis and,  466–469, 471, 473, 476, 488, 490–492, 496, 500 vertical mergers and,  555–556 Mergers advertising-supported media mergers, 428–430 buyer mergers,  530–540. See also Buyer mergers buyer power and,  529–547. See also Buyer power in Canada,  9 conglomerate mergers,  552n3 coordinated effects analysis,  333–334. See also Coordinated effects analysis countervailing power through,  376–377 diagonal mergers,  551n1 dynamic efficiency and,  459–460 efficiencies of,  459–464. See also Efficiencies in European Union,  9, 155 healthcare markets,  377 health insurance,  106–108 hospital mergers,  70–83. See also Hospital mergers injunctions and,  267–269 merger simulation,  475–487. See also Merger simulation monopoly and,  269n13 of multisided platform businesses,  426–430 opposing views of antitrust law and,  8–9 preventing, 333–334 seller mergers,  530n3 static efficiency and,  452 supplier mergers,  540–546. See also Supplier mergers unilateral effects analysis,  333–334. See also Unilateral effects analysis in United Kingdom,  154–155 upward pricing pressure and,  334 vertical mergers,  551–582. See also Vertical

mergers Merger simulation,  475–487 AIDS demand,  478, 482–486 in European Union,  486, 498 framework, 476–478 linear demand,  478–480 logit demand,  478, 480–482, 485n53 overview, 475–476 parameterizing demand functions, 478–486 in United States,  486, 498 use of,  486–487 Metcalfe's Law,  385 MFN (Most favored nation) clauses health insurance and,  104–105 settlements and,  173 MGM Mirage,  520–521 Michielsen, Thomas,  427, 429 Microsoft as dominant firm,  320, 330 exclusionary conduct by,  434–435 injunctions and,  266–267, 270–273 market power and,  339 as multisided platform business,  413, 421n20 settlements and,  183 vertical mergers and,  553, 558, 582 Microsoft Office (software),  582 Milgrom, Paul,  247 Miller, Geoffrey P.,  179 Miller, James C., III,  163 Miller, Nolan,  211, 213 Miller-Tydings Act of  1937, 335 Milutinovic, Veljko,  284 Minimum efficient scale,  35–36 Monopoly actual punishment policy and,  22–24 bilateral monopoly,  364–378. See also Bilateral monopoly competition compared,  316, 319 complex example of,  318–320 defined, 315 demand curves in,  316, 322, 326–327 dominant firms. See Dominant firms economies of scale as entry barrier,  324 entry barriers,  323–325 framework, 315–317 health insurance and,  99–100, 105–108

604   Index Monopoly (cont.) hypothetical monopolist test,  354–358 injunctions and,  267–268 market definition. See Market definition mergers and,  269n13 monopsony, linking to,  99–100 multisided platform businesses,  404–441. See also Multisided platform businesses natural monopoly. See Natural monopoly networks and,  380–398. See also Networks one monopoly rent theorem,  390 optimal enforcement policy and,  18–20 ownership of unique resource as entry barrier, 323–324 patents as,  324n23 restraint of trade and,  303–305 rise of,  323–325 single-monopoly profit principle,  556–557 size and sunkenness of needed investments as entry barrier,  324–325 Monopsony buyer mergers and,  531–532 buyer power and,  531–532 in European Union,  293–294 health insurance and,  98–105 Medicaid and,  101–102 Medicare and,  101–102 monopoly, linking to,  99–100 private enforcement actions and,  292–294 rationale for antitrust law and,  11–12 standard model,  365 in United States,  293 Moresi, Serge,  524 Morgan, John,  218 Moss, Diana L.,  274 Most favored nation (MFN) clauses health insurance and,  104–105 settlements and,  173 Motion picture industry, countervailing power and,  375–376 Motta, Massimo,  182, 239 The Movie Channel,  565 MSNBC, 572 Muellbauer, John,  482 Müller, Wieland,  213, 229, 238–239, 245–247 Multihoming, 417–419 Multilateral strategy in international antitrust institutions, 129–130

Multilevel lead jurisdiction model of international antitrust institutions, 139–141 Multiple-system operators, vertical mergers and, 565 Multisided platform businesses,  404–441 advertising-supported media mergers, 428–430 asymmetric competition,  419 bundling by,  435–436 in China,  406 competition among,  415–419 “competitive bottlenecks,”  418 competitive constraints,  420–421 consumer welfare and,  413–414 coordinated behavior,  438–440 critical mass,  431–433 defined, 408–410 economic analysis of,  404n2 economics of,  407–414 efficiencies, 438 in European Union,  406 exclusive dealing by,  433–434 externalities and,  407–408, 410–411 hypothetical monopolist test and,  424 indirect network effects,  405 market definition and,  423–426 market power and,  422–423, 425–426 mergers of,  426–430 multihoming, 417–419 overview,  404–407, 440–441 payment cards,  415, 438n39 platform viability,  431–433 predatory pricing by,  436–437 pricing and,  411–413 product differentiation and,  416–417 social welfare and,  413–414 tying by,  435–436 unilateral effects of mergers, predicting, 426–428 upward pricing pressure and,  427 Muris, Timothy J.,  246 Murphy, Kevin M.,  211 Mutual assistance in bilateral agreements,  127 Mutual Hospital Insurance,  98 Nain, Amrita,  535 Nash equilibrium

Index   605

bargaining models and,  498, 500 nonbinding price announcements and,  236 price ceilings and,  238 tacit collusion and,  231–233 vertical mergers and,  575 National champions, conferring market power to protect,  54–55 National Industrial Recovery Act of  1933, 335n54 National Recovery Administration,  335n54 National security, conferring market power to promote, 55 Natural monopoly exemptions and,  35–36 minimum efficient scale and,  35–36 networks as,  382–384 technology and,  35–36 Need Based Financial Aid Act of  2007, 53 Negative comity in bilateral agreements,  127 Negative externalities, ICN and,  136 Netease (software),  419 Netherlands, healthcare markets in,  65–66n6 Netscape (browser),  271, 434 Network effects,  385–389 Network externalities, market failure and, 45–46 Networks, 380–398 antitrust policy implications,  393–397 behavioral remedies,  395–397 business networks,  381n3 contestability theory and,  383 divestiture of,  393 dynamic efficiency in,  392–393 essential facilities doctrine and,  395–397 excess inertia and,  386–387 excess momentum and,  387 fixed costs and,  382–384 indirect network effects,  405 innovation and,  384 insufficient friction and,  387 market failure and,  386, 389 as natural monopoly,  382–384 network effects,  385–389 one monopoly rent theorem and,  390 overview,  380, 397 product differentiation and,  384 social networks,  381n3 static efficiency in,  392–393

structural separation of,  393–395 sunk costs and,  383 technology, impact of,  382–383 technology lock-in and,  386–387 types of,  380–381 vertical exclusion in,  389–392 vertical integration in,  389–392 virtual networks,  381 Network strategy in international antitrust institutions, 130–134 Neven, Damien J.,  155 Nevo, Aviv,  78 Newbery, David M.J.,  568, 582 Newhouse, Joseph,  79 Newspaper Preservation Act of  1970, 56 Newspapers advertising-supported media mergers, 428–430 exclusive dealing by,  433 favored activities, subsidization of,  56 injunctions and,  271 predatory pricing by,  437 New West Health,  106–107 Niskanen, William A.,  158 Noel, Michael,  419, 424 Noether, Monica,  68 Nonbinding price announcements,  235–237 Noncompete agreements,  332–333 Noneconomic objectives of antitrust law, 12–14 Nonprofit organizations, hospital mergers and, 79–82 Nonstandard economics. See Behavioral economics Norbäck, Pehr-Johan,  513, 526 Normann, Hans-Theo,  213, 231, 233–238, 242–243, 246–247 Norris-LaGuardia Act of  1932, 49, 375 North Texas Specialty Physicians,  93 Notification in bilateral agreements,  126 Nowell, Cliff,  247 Null core cartels and,  40–41 market failure and,  38–42 transportation and,  41 Obama, Barack,  156 O'Brien, Daniel P.,  468

606   Index Ocean shipping, conferring market power to promote national security,  55 OECD. See Organization for Economic Cooperation and Development Oechssler, Jorg,  217, 231 Offerman, Theo,  220 Oligopoly accounting anomalies and,  221 explicit collusion and,  234–235 imitative behavior and,  208–212 SSCO model,  560–561 tacit collusion in,  231 One monopoly rent theorem,  390 OpenTable,  407–410, 413, 420, 428, 440 Optimal deterrence,  260–261 Optimal enforcement policy,  18–21 Oracle, 498 Ordover, Janusz A.,  163, 355, 509, 525, 561 Organization for Economic Cooperation and Development (OECD),  120, 130, 405 OSS model of vertical mergers,  561–563, 578–579 Outcome-focused view of restraint of trade, 299–301 Overoptimism, 218–220 PacifiCare,  102–103, 106–107 Page, William H.,  254 Palmyra Park Hospital,  82 P&O Princess Cruises plc,  520 Panzar, John C.,  304 Parker, Geoffrey,  419 Pass-on defense, private enforcement actions and, 284 Patents as monopoly,  324n23 Patient Protection and Affordable Care Act of  2010, 63, 108–110 Pauly, Mark,  79 Payment cards as multisided platform businesses,  415, 438n39 Pazgal, Amit,  211, 213 PCAIDS model,  482–486 PeaceHealth, 86–88 People Soft,  498 PepsiCo, 551–553 Perloff, Jeffrey M.,  174–176, 178 Perry, Martin K.,  493

Peters, Craig,  486 Petrin, Amil,  501 Pharmaceutical industry, noncompete agreements in,  332–333 Pheobe Putney Health System,  82 Philipson, Thomas J.,  80 Phoenix (software),  419 Physicians, 89–94 Advocate Health Partners, 92–93 caselaw, 92–94 collective bargaining by,  377 fee schedules and,  90–92 Foundation of Physicians and Dentists, 92 North Texas Specialty Physicians, 93 overview, 89–90 price fixing and,  90–92 TriState Health Partners, 93–94 Piette Durrance, Christine,  187 Piskorski, Mikolaj Jan,  417 Plott, Charles R.,  237–238 Podolny, Joel,  214 Polinsky, A. Mitchell,  180 Porter, Robert H.,  493, 514 Positive analysis,  150 Positive comity in bilateral agreements,  128 Posner, Richard A. on healthcare markets,  80 on imprisonment,  264 on injunctions,  268, 270 on market definition,  348, 351 on monetary penalties,  262–263 on public choice theory,  148, 152, 160 on rationales for antitrust law,  7 on restraint of trade,  299, 304, 307 Posted-offer experiments,  234n6 Potters, Jan,  215, 220 Pozen, Shari A.,  147 PPOs. See Preferred provider organizations Predatory pricing in European Union,  25–27 as exclusionary conduct,  240–242 experimental economics,  240–242 by multisided platform businesses,  436–437 by newspapers,  437 preventing, 334–336 substantive standards,  25–27 in United States,  25–27

Index   607

Predominance requirement in class actions, 190–191 Preferred Health Services,  90 Preferred provider organizations (PPOs),  67, 69, 106–107 Premdor,  519, 573 Price ceilings,  237–239 Price complexity, coordinated effects analysis and,  513, 519–521 Price coordination,  358 Price discrimination, opposing views of antitrust law and,  8 Price fixing actual punishment policy and,  22–23 consumers and,  288–289 efficiencies and,  20n5, 452–453 explicit collusion and,  234 imitative behavior and,  210 optimal enforcement policy and,  20 physicians and,  90–92 resellers under imperfectly competitive conditions and,  289–292 Price floors,  237–239 Pricing in healthcare markets,  66–67 multisided platform businesses and, 411–413 predatory pricing. See Predatory pricing Private enforcement actions,  281–296 attorneys' fees in,  179–180 “bounty hunter” model,  282–283 collusion and,  292–294 confidentiality of settlements,  173–174 consumers, price fixing and,  288–289 by direct purchasers,  284, 287 economic models regarding,  288–295 in European Union,  281–282, 285–287, 295–296 by indirect purchasers,  284, 287 injury requirement,  283–287 lost profits and,  294–295 monopsony and,  292–294 overview,  281–282, 295–296 pass-on defense and,  284 punitive damages in,  285 resellers under imperfectly competitive conditions, price fixing and,  289–292

settlements in,  172–180 single damages versus treble damages, 174–176 standing requirement,  283–287 in United States,  281–285, 287, 295–296 Private health insurance,  95 Procedure in enforcement regimes,  27, 29–30 Process-focused view of restraint of trade, 299–301 Product differentiation multisided platform businesses and, 416–417 networks and,  384 Salinger differentiated products model, 564–565 unilateral effects analysis and,  467–491 Product variety, unilateral effects analysis and, 500–501 Professional sports favored activities, subsidization of,  56 supply externalities in,  46–47 Profit approximately optimal profit,  217–218 competition and maximization of profit, 207–208 lost profits, private enforcement actions and, 294–295 reasons for not pursuing maximum profit, 223–225 relative profit, concerns for,  208–212 satisfactory profit versus maximum profit, 215–218 Prudential,  102–104, 106, 532 Public choice theory,  147–165 behavioral psychology and,  148–149 bureaucracy and,  157–159 Congressional oversight and,  157–159 continuation of competition, antitrust law as, 163–164 enforcement and,  155–162 historical background of antitrust law and, 151–155 interest groups and,  162–164 judicial review and,  159–162 overview,  147–151, 164–165 positive analysis and,  150

608   Index Public enforcement actions, settlements in, 180–183 Public interest theory,  152 Punitive damages in private enforcement actions, 285 Qihoo 360 (software),  406, 419 Quality, hospital mergers and,  83 Railroad industry component production, achieving efficient scale for,  43 regulation, mitigating,  50 Rajabiun, Reza,  161 Ramanarayanan, Subramaniam,  104–106 Ramey, Garey,  247 Ramey, Valerie,  234 Ramírez, Carlos D.,  154 Ramseyer, J. Mark,  243–244 Rasmusen, Eric B.,  243–244 Rationale for antitrust law,  3–14 buyer power and,  11–12 competition versus competitors,  9–10 divergence between opposing views,  8–9 dynamic efficiency,  10–11 efficiencies, 5–8 fairness, 5–8 intended beneficiaries,  4–5 monopsony and,  11–12 noneconomic objectives,  12–14 overview, 3–4 static efficiency,  10–11 Rationing rules,  495n74 Reactive fringe, dominant firms and,  320–323 Reagan, Ronald,  570 Rebates, 245–246n15 Redisch, Michael,  79 Regulation direct regulation, exemptions facilitating transition to market orientation,  57–58 market regulation, mitigating prior failures in, 50–51 Reiffen, David,  567–568, 579 Reinganum, Jennifer F.,  174 Reitman, David,  246 Reksulak, Michael,  147, 149 Relative profit,  208–212

Remedies, 254–275 conduct injunctions,  270–274 damages, 259–263. See also Damages fines, 257–258 imprisonment,  23, 263–265 injunctions, 265–274. See also Injunctions monetary penalties,  256–263 overview,  254–256, 274–275 structural injunctions,  266–270 Rent seeking exemptions and,  34 wealth transfers and protection from competition and,  52 Research and development, achieving efficient scale for component production,  43 Resellers under imperfectly competitive conditions, price fixing and,  289–292 Restraint of trade,  298–308 bundling and,  305–306 consumer welfare and,  299–300, 307 core and,  299–301 goals of antitrust law and,  307 horizontal agreements and,  301–303 loyalty pricing and,  305–306 monopoly and,  303–305 outcome-focused view,  299–301 overview, 298–299 process-focused view,  299–301 retail price maintenance and,  301, 306 vertical agreements and,  306 Restrictive Business Practices Code,  130 Retail price maintenance (RPM),  301, 306 Rey, Patrick,  175, 242–243, 524 Reynolds, Robert,  216, 493 Reynolds, Stan,  231 Ries, John,  122 Risk neutrality, class actions and,  189n4 Roberts, John,  247 Robert Wood Johnson Foundation,  71 Robinson, James,  68 Robinson, Joan,  326, 331 Robinson-Patman Act of  1936, 52, 151, 335, 539 Rochet, Jean-Charles on coordinated behavior,  439 on exclusionary conduct,  435 on market definition,  424 on market power,  422

Index   609

on multihoming,  417 on multisided platform businesses,  405, 408 on pricing,  411–413 Roll, Richard,  155, 219 Roller, Lars-Hendrik,  155 Rosen, Harvey S.,  485 Roux, Catherine,  235 Royal Caribbean Cruises, Ltd.,  520 RPM (Retail price maintenance),  301, 306 Rubinfeld, Daniel L.,  172, 174–176, 178, 180 Ruhmer, Isabel,  439–440 Ruinous competition, market failure and, 38–42 Russia, international externalities of national competition policies and,  122–123 Ruud, Paul,  176, 178 Sacher, Seth,  81 Salant, Stephen,  216, 493 Salinger, Michael A.,  551, 560, 564 Salinger differentiated products model of vertical mergers,  564–565 Saloner, Garth,  561 Salop, Steven C.,  307, 468, 561, 570 Samba (software),  273 Samuels, Warren J.,  160 Santerre, Rexford E.,  539 Satisfactory profit,  215–218 Satterthwaite, Mark,  77–78, 81 Schaffer, Mark,  209 Scheelings, Richard,  538 Scheinkman, Jose A.,  495 Schelling, Thomas,  222 Schmalensee, Richard,  348, 404, 409, 411, 413, 431, 440, 513 Schumpeter, Joseph,  10–11 Schwartz, Marius,  532, 538 Scott Morton, Fiona,  214 Seabright, Paul,  175 Segal, Ilya R.,  243–244, 433 Seldon, Arthur,  151 Selective contracting in health insurance, 97–98 Seller mergers,  530n3 Selten, Reinhard,  241 Sentencing Guidelines,  258

Settlements, 172–184 attorney incentives,  179–180 attorneys' fees in,  179–180 in Brazil,  182n14 of class actions,  195–199 confidentiality in,  173–174 consent decrees and,  183 decision between settlement and litigation, 172–174 deterrence and,  261–262 empirical framework,  176–178 fines and,  180–183 game theory and,  172n1 jury trial and,  178n8 leniency programs and,  180–183 MFN clauses and,  173 overview,  172, 184 private enforcement actions,  172–180 public enforcement actions,  180–183 single damages versus treble damages, 174–176 SGL (cartel),  181 Shaffer, Greg,  534 Shanley, Mark,  68 Shannon, Chris,  219 Shapiro, Carl,  355, 374, 388, 427, 468, 473, 493, 512, 522 Shapiro, Jesse M.,  439 Share-of-recovery method of attorneys' fees, 199–201 Shelanski, Howard,  451 Sherman Act of  1890 actual punishment under,  21, 23 bilateral monopoly and,  364, 372, 375, 378 bundling under,  305–306 cartels under,  230 class actions under,  187 collusion under,  328 coordinated effects analysis and,  526 damages under,  259 development of law under,  27 efficiencies and,  452, 457 enforcement under,  149, 155–156 exclusionary conduct under,  239 fines under,  258 framers' intentions,  3 historical background,  151–154

610   Index Sherman Act of (cont.) horizontal agreements under,  301–303 hospital staff privileges and,  84–85 imprisonment under,  263 intended beneficiaries of,  5 loyalty pricing under,  305–306 mergers under,  335 monetary penalties under,  257 monopoly under,  161n24, 303 per se violations,  328, 332 predatory pricing under,  25 price fixing under,  328 remedies under,  254–255 restraint of trade under,  161n24, 299 Showtime, 565 Shughart, William F., II,  147, 149, 153–154, 156, 159–160, 163–164 Shum, Matthew,  161 Sibley, David S.,  246 Siegel, Sidney,  231 Sierra, 106 Simon, Carol,  68 Simon, Herbert,  215 Sina (software),  419 Singer, Ethan S.,  490 Single damages,  174–176 Single-monopoly profit principle, vertical mergers and,  556–557 Sinkinson, Michael,  439 Skitol, Robert A.,  539 Small, Kenneth A.,  485 Small-business hypothesis,  153 “Small but significant and nontransitory increase in price” (SSNIP) test,  328–330, 406, 425, 427, 429 Small buyers, supplier mergers and,  545 Smith, Adam,  217 Smith, Angela M.,  244 Smith, Vernon L.,  237–238, 240, 246 Snyder, Christopher M.,  242–243 Snyder, Edward A.,  183 Social goals, conferring market power to achieve, 52–54 Social networks,  381n3 Social welfare, multisided platform businesses and, 413–414

Soft Drink Interbrand Competition Act of  1980, 51 Sohu (software),  419 Song, M.,  421, 426, 430 Sony PlayStation,  434 South Africa, antitrust law in,  14 Spence, A. Michael,  414 Spiegel, Yossi,  219, 247 Spier, Kathryn E.,  173, 244–245 Split agreements,  375–376 Sports favored activities, subsidization of,  56 supply externalities in,  46–47 Spulber, Daniel F.,  380 SSCO model of vertical mergers,  574–578 SSNIP test. See “Small but significant and nontransitory increase in price” (SSNIP) test Stackelberg markets bundling in,  246–247 vengeful behavior in,  213 Staff privileges in hospitals,  84–86 Staggers Rail Act of  1980, 50 Standard Oil Company of New Jersey,  337 Standard setting, market facilitation and,  49 Standards Setting Organization Development Act of  2004, 49 Standing requirement for private enforcement actions, 283–287 State action immunity, hospital mergers and, 82–83 State ownership, exemptions facilitating transition to market orientation,  57–58 Static efficiency mergers and,  452 in networks,  392–393 as rationale for antitrust law,  10–11 Static pricing pressure, vertical mergers and, 581–582 Stigler, George J.,  3, 109, 147–148, 152–153, 155, 256, 510, 544 Strahilevitz, Lior Jacob,  417 Structural dismemberment,  337–339 Structural injunctions,  266–270 Structural presumption of coordinated effects analysis, 510–513 Structural separation of networks,  393–395 Suetens, Sigrid,  245

Index   611

Sunder, Shyam,  230 Sunk costs, networks and,  383 Sun Microsystems,  271–273 Supplier mergers,  540–546 ability to prevent anticompetitive effects, 541–545 in Canada,  541–544 empirical analysis, need for,  545–546 european Union and,  541–543 overview, 540–541 small buyers and,  545 in United Kingdom,  541–543 in United States,  541 Supply externalities, market failure and, 46–47 Suslow, Valerie Y.,  515 Sutton, John,  568 Switzer, Sheldon,  216, 493 Sykes, Alan O.,  355, 525 Syverson, Chad,  566 Tacit collusion,  231–233 Tangency equilibrium,  327 Taobao (software),  419 Tate, Geoffrey,  220 Technology collusion, effect on,  214–215 natural monopoly and,  35–36 networks, impact on,  382–383 Technology lock-in,  386–387 Telecommunications Act of  1996, 270 Telecommunications industry national champions, conferring market power to protect,  54–55 network externalities in,  45 networks in. See Networks packetization in,  383–384 Telser, Lester,  301, 391 Tencent (software),  406, 419 Thoeni, Christian,  235 Thomas, Charles J.,  516 Thomas, Diana W.,  156 Thomas, Shawn,  535 Thompson, T. Scott,  425 Thunder Video (software),  419 Tianya (software),  419 The Times (London),  437

Time Warner,  572–573 Tirole, Jean on coordinated behavior,  439 on exclusionary conduct,  435 on foreclosure,  242–243 on market definition,  424 on market power,  422 on multihoming,  417 on multisided platform businesses,  405, 408 on pricing,  411–413 on vertical mergers,  563 Tobacco, antitrust law and,  14 Tollison, Robert D.,  147, 154, 157–158, 160, 162 Tosco, 566 Town, Robert J.,  71–72, 77–78, 81 Trade agreements in international antitrust institutions, 129–130 Trajtenberg, Manuel,  501 Transportation industry empty core and,  41 networks in,  381 Treasury Department,  415 Treaty of Rome (1957) generally,  3, 13 cartels under,  230 Treaty on the Functioning of the European Union (2009) generally, 3 development of law under,  27 intended beneficiaries of,  5 private enforcement actions under,  281 Treble damages in class actions,  192, 197 under Clayton Act,  259 single damages versus,  174–176 TriState Health Partners,  94 Tschantz, Steven,  468 Tudou (software),  419 Tullock, Gordon,  151, 156, 158–159, 162, 164 Tunney Act of  1974, 183 Turner, Donald F.,  26, 152, 338 Turner Broadcasting,  572–573 Tying in European Union,  435 hospitals and,  86–87 by multisided platform businesses,  435–436

612   Index UCAR International,  181 “Ultimatum game,”  212–213 UNCTAD (United Nations Conference on Trade and Development),  130 Unilateral conduct, efficiencies and,  452 Unilateral effects analysis,  466–502 auction models,  496–498 bargaining models,  498–500 in Bertrand markets,  470, 476, 494–495 coordinated effects analysis compared, 466n1 in Cournot markets,  355, 492–493 cross-market variation,  490–491 differentiated products,  357 dominant firm model,  495 empirical evidence,  487–491 in European Union,  467 homogeneous goods and,  355–357, 491–495 innovation and,  500–501 intertemporal variation,  488–490 market definition and,  355–357 mergers and,  333–334 merger simulation and,  475–487. See also Merger simulation overview,  466–467, 501–502 product differentiation and,  467–491 product variety and,  500–501 in United Kingdom,  467–468 upward pricing pressure and,  468–476, 488, 491 Unilateral exercise of market power exclusionary conduct, preventing,  336–337 mergers, preventing,  333–334 policies to prevent,  314, 331–339 predatory pricing, preventing,  334–336 restriction on competition, preventing, 331–333 structural dismemberment and,  337–339 Unilateral strategy in international antitrust institutions, 125–126 Unions countervailing power and,  375 Favored classes of market participants, wealth transfers to,  57 United Healthcare,  92–93, 95, 102–103, 106–107 United Kingdom

cartels in,  214 Competition Commission,  150 Competition Commission and Office of Fair Trading,  467, 542–543 enforcement in,  149 Fair Trading Act  1973, 155 Guidelines to Merger Policy,  155 Merger Assessment Guidelines,  467–468 mergers in,  154–155 Monopolies and Mergers Commission,  154 supplier mergers in,  541–543 unilateral effects analysis in,  467–468 United Nations Conference on Trade and Development (UNCTAD),  130 United Regional Health System,  65, 88–89 United Shoe Machinery Corporation,  338 United States. See also specific Act or entity actual punishment policy in,  21–25 collusion in,  293 comity in,  128 conscious parallelism in,  235 consumers, price fixing and,  288–289 coordinated effects analysis in,  514, 517 development of law in,  27–29 effects doctrine, extraterritorial enforcement through,  125–126 efficiencies in,  460 evidence in,  29–30 exclusionary conduct in,  346, 350 fines in,  257 importance of enforcement regime,  17 lost profits in,  294–295 merger simulation in,  486, 498 monopsony in,  293 nonbinding price announcements in,  235 predatory pricing in,  25–27 private enforcement actions in,  281–285, 287, 295–296 private health insurance in,  95 public enforcement actions, settlements in, 180–183 resellers under imperfectly competitive conditions, price fixing and,  292 Sentencing Guidelines,  258 supplier mergers in,  541 vertical mergers in,  570–574 Upstream competition

Index   613

buyer mergers, effect of,  536–537n20 vertical mergers, effect of,  561n13 Upstream duopoly,  242n11 Upward pricing pressure (UPP) mergers and,  334 multisided platform businesses and,  427 unilateral effects analysis and,  468–476, 488, 491 in vertical mergers,  580 Usage externalities,  407–408, 410 U.S Steel Corporation,  338 Usual, customary, and reasonable (UCR) fees,  67, 97 Utilities, networks in,  381 Value chain in healthcare markets,  65–66 Van Alstyne, Marshall,  419 van Damme, Eric,  239 Variable costs, efficiencies and,  459 Vega-Redondo, Fernando,  209–210 Vengeful behavior,  212–215 Vertical agreements restraint of trade and,  306 retail price maintenance and,  306 Vertical differentiation,  467n6 Vertical exclusion in networks,  389–392 Vertical integration injunctions and,  269 in networks,  389–392 Vertical mergers,  551–582 application of post-Chicago School models, 579–581 in Bertrand markets,  561–562, 579 in cable television,  565, 572–573 in cement industry,  566n21, 566n22 Chicago School critique of policy,  551, 553–554, 555–559 Church White Paper,  567–568 commitments and,  574–579 comparative transaction costs and,  559 coordinated effects analysis and,  469n24 in Cournot markets,  560–561, 567, 575–578 double marginalization and,  557–559 empirical evidence,  565–566 in European Union,  551–552, 554, 568–570 Hart-Tirole model,  563–564 multiple-system operators and,  565

Nash equilibrium and,  575 1960s policy,  554–555 OSS model,  561–563, 578–579 overview,  551–554, 582 policy effects of post-Chicago School literature, 566–574 post-Chicago School qualifications,  552– 554, 559–566 recommendations for practical policy, 574–582 Salinger differentiated products model, 564–565 single-monopoly profit principle and, 556–557 SSCO model,  560–561, 574–578 static pricing pressure and,  581–582 in United States,  570–574 upstream competition, effect on,  561n13 upward pricing pressure in,  580 Vertical restraints, efficiencies and,  457–458 Viacom, 565 Vickers, John,  211, 213 Video games, exclusive dealing and,  433–436 Virtual networks,  381 Visa,  330, 415 Vistnes, Greg,  77–78, 81 Vita, Michael G.,  81 Vogt, William B.,  71–72, 77, 81 Wealth transfers and protection from competition, 51–57 favored activities, subsidization of,  56 favored classes of market participant, wealth transfers to,  56–57 market power, conferring to achieve specific goals,  52–55 national champions, conferring market power to protect,  54–55 national security, conferring market power to promote,  55 rent seeking and,  52 social goals, conferring market power to achieve, 52–54 Webb-Pomerene Export Act of  1918, 54 Weinberg, Matthew,  486 Weir, Charlie,  154–155 Wengstrom, Erik,  215

614   Index Werden, Gregory L.,  468 Wesley Medical Center,  99 Wey, Christian,  535 Weyl, E. Glenn,  414, 427 Whinston, Michael D.,  243–244, 433, 435, 516 White, Alexander,  427 White, Edward D. (Justice),  298, 303, 307 White, Lawrence J.,  313 White, William D.,  68 White Sands Health Care System,  90 Wiley, John S., Jr.,  243–244 Williams, Arlington W.,  234, 237–238 Williamson, Oliver,  8, 452 Willig, Robert D.,  304, 355, 466, 468, 516, 525 Willingness-to-pay, hospital mergers and, 77–79 Wils, Wouter P.J.,  181 Wilson, Robert,  222, 241

Windows (operating system),  271–272, 413, 435, 582 Windows Media Player,  272, 435 World Economic Conference,  129 World Trade Organization (WTO),  130, 140 Wright, Joshua D.,  538 Wright, Julian,  418, 433–434 Wyzanski, Charles E., Jr. (Judge),  307 Xerox, 320 Xu, Frances Zhiyun,  173 Yehezkel, Yaron,  247 Yi, Sang-Seung,  568–569 Yoo, Christopher S.,  380 Youku (software),  419 Zwanziger, Jack,  68, 81