ANTITRUST Brian Facey, Navin Joneja, Paul Cuomo, and Jeffrey Oliver.....64 The Increasing...

S p r i n g 2 0 1 8 • V o l u m e 3 2 • N u m b e r 2

ANTITRUSTI N T H I S I S S U E

ANTITRUST AND ECONOMICS

When Antitrust MetEconomics

Economic Evidence

Roundtable withEconomists: Practice and Theory

Economic Toolbox

Hot Tub Redux

Battle of the Experts: Two Case Studies

Use of Economics Beforethe Agencies and Courts

Economics of Foreclosure

Court of Justice and Intel

Merger Efficiencies in theU.S. and Canada

Innovation in Merger Review

Network Effects and Market Power

Machine Learning

ARTICLES

JVs in International Merger Control

Are Disgorgement’s DaysNumbered?

Merger Remedies

Economics and Antitrust

� FOR OVER 40 YEARS, Antitrust Law Developments and its annual supplements have been recognized as the most authoritative and comprehensive set of research tools for antitrust practitioners. The 2017 Annual Review of Antitrust Law Developments summarizes developments during 2017 in the courts, at the agencies, and in Congress.

Among other topics, the 2017 Annual Review discusses

• Important judicial decisions and government enforcement across the major areas of antitrust law, includingagreements in restraint of trade, monopolization, M&A, joint ventures, intellectual property, and exemptionsand immunities;

• Court rulings on issues central to private antitrust litigation, including developments related to antitrust injury,standing, damages, the availability of arbitration, the requirements to maintain a class action, the limitations on recovery for foreign purchases, the standards for injunctive relief, motions to dismiss, and motions for summary judgment;

• Antitrust law developments in industry-specific sectors, including health care, energy, communications, and transportation; and

• International developments in the European Union, Brazil, Canada, United Kingdom, and other jurisdictions.

Visit our website at www.shopaba.org

NEW from the American Bar Association

2017 Annual Review of Antitrust Law Developments

Product Code: 5030650

Publication Date: 2018

Page Count: 515

Trim Size: 7 x 10

Format: Paperback

Pricing: $229.00 List Price / $209.00 ABA Member / $189.00 AT Section Member

S P R I N G 2 0 1 8 · 1

C o v e r S t o r i e s

Editor’s Note Economic Evidence in Antitrust Cases and Enforcementby Gregory G. Wrobel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Roundtable with Economists: Discussing Practice and Theory with the Experts . . . . . . . . . 11

Unpacking the Economic Toolbox: How to Make Sense of Your Economic Expert’s Analysis by Sheng Li, Christine Meyer, and Gabriela Monahova . . . . . . . . . . . . . 24

LITIGATION PRACTICE: NOTES FROM THE FIELD

Hot Tub Reduxby Lisa C. Wood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

Battle of the Experts in Merger Litigation: Two Case Studies by Debbie Feinstein and Wrede Smith . . . . . . . . . . . . . . . . . . . . . . . . . 37

The First Cut Is the Deepest: Use of Economics Before the Antitrust Agencies and the Courts by Michael J. Perry and Stephen Weissman . . . . . . . . . . . . . . . . . . . . . . 44

The Economics of Foreclosure: A Lawyer’s Guide by James Keyte, David S. Evans, and Eliana Garcés . . . . . . . . . . . . . . . 49

The Court of Justice and Intel: An Overdue Ruling that Was Still Over-hasty?by Kevin Coates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

Mind the Gap: Merger Efficiencies in the United States and Canada by Brian Facey, Navin Joneja, Paul Cuomo, and Jeffrey Oliver . . . . . . . . 64

The Increasing Cross-Border Importance of Innovation in Merger Review by Jennifer Cascone Fauver, Subbu Ramanarayanan, and Nicola Tosini . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

What Have We Learned in the Last Decade? Network Effects and Market Power by Catherine Tucker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

An Antitrust Lawyer’s Guide to Machine Learning by Ai Deng. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

A N T I T R U S T • S P R I N G 2 0 1 8 • V O L . 3 2 • N O . 2

T A B L E O F C O N T E N T S

Economics and Antitrust

Cover Photo: SuperStock

A N T I T R U S T

ISSN 0162-7996

This magazine is published three times a year (Spring, Summer, and Fall) by theSection of Antitrust Law, American Bar Association, 321 North Clark Street, Chicago,IL 60654.

The subscription price for members of the Anti trust Section is included in their dues.Annual sub scriptions for institutions and individuals not eligible for ABA membershipare $75 per year ($85 for Alaska, Hawaii, U.S. Possessions and foreign countries).Single copy price is $30.

Please address all subscription mail to Section of Antitrust Law, American BarAssociation, 321 North Clark Street, Chicago, IL 60654. Nonprofit standard postagepaid at Atlanta, GA.

Unsolicited original manuscripts and letters to the editor are welcome and shouldbe sent to Tina Miller, at [email protected]. For more information on our publishing procedures and policies, visit us at www.americanbar.org/publications/antitrust_magazine_home.html.

The views expressed herein are the authors’ only and are not necessarily those ofthe authors’ firm, the American Bar Association, or the Section of Antitrust Law.

Copyright © 2018 American Bar Association

A r t i c l e s

INTERNATIONAL DEVELOPMENTS

A Coat of Many Colors: Joint Ventures in International Merger Control by Natalie Yeung and Paddy Law . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

Are Disgorgement’s Days Numbered? Kokesh v. SEC May Foreshadow Curtailment of the FTC’s Authority to Obtain Monetary Relief by M. Sean Royall, Richard H. Cunningham, and Ashley Rogers . . . . . . 94

Comment on “Are Merger Enforcement and Remedies Too Permissive? A Look at Two Current Merger Studies” by John Harkrider by John Kwoka . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101

D e p a r t m e n t s

When Antitrust Met Economics— Letter from Section Chair Jonathan M. Jacobson . . . . . . . . . . . 3

A M E R I C A N B A R A S S O C I A T I O N

S E C T I O N O F A N T I T R U S T L A W

C O N T I N U I N G L E G A L

E D U C A T I O N C A L E N D A R

2 0 1 8

A P R I L 1 1 – 1 3 , 2 0 1 8

The Spring MeetingWASHINGTON, DC

� � �

M A Y 8 , 2 0 1 8

Global Seminar Series: DüsseldorfDÜSSELDORF, GERMANY

� � �

M A Y 1 7 – 1 8 , 2 0 1 8

Antitrust in Health Care ConferenceARLINGTON, VA

� � �

M A Y 3 1 – J U N E 1 , 2 0 1 8

Antitrust in Asia ConferenceSEOUL, KOREA

� � �

A U G U S T 2 – 7 , 2 0 1 8

ABA Annual MeetingCHICAGO, IL

� � �

O C T O B E R 1 8 – 2 0 , 2 0 1 8

Antitrust Masters CourseCAMBRIDGE, MD

� � �

N O V E M B E R 1 5 , 2 0 1 8

Fall ForumWASHINGTON, DC

Please visit http://www.ambar.org/antitrust

for detailed conference information.

2 · A N T I T R U S T

A N T I T R U S TM A G A Z I N E

Damian G. DiddenWachtell Lipton Rosen & Katz

51 West 52nd St.New York, NY 10019-6119

Stacy E. FrazierGeneral Electric Company

1299 Pennsylvania Ave., NWWashington, DC 20004

John D. HarkriderAxinn, Veltrop & Harkrider LLP

114 West 47th St. New York, NY 10036-1510

Jeffrey A. JaeckelMorrison & Foerster LLP

2000 Pennsylvania Ave., NWWashington, DC 20006-1831

Janis Kestenbaum Perkins Coie LLP 700 13th St., NW

Washington, DC 20005

James A. Keyte The Brattle Group120 West 45th St.New York, NY 10036

William J. KolaskyHughes Hubbard & Reed LLP

1775 I St., NWWashington, DC 20006

James G. KressBaker Botts LLP

1299 Pensylvania Ave., NWWashington, DC 20004-2408

Associate Editors

Kellie Lerner Robins Kaplan LLP 601 Lexington Ave. New York, NY 10022

Michael A. Lindsay Dorsey & Whitney LLP50 South 6th St.

Minneapolis, MN 55402-1498

Anna Lyle-SmytheSlaughter and MaySquare de Meeûs 40Brussels 1000 Belgium

Christine S. MeyerNERA Economic Consulting

360 Hamilton Ave.White Plains, NY 10601

Robin L. MooreFederal Trade Commission600 Pennsylvania Ave., NWWashington, DC 20580-0002

Andrea MurinoGoodwin

901 New York Ave., NWWashington, DC 20001

Jason O’ConnorFederal Trade Commission600 Pennsylvania Ave., NWWashington, DC 20580

Jack E. Pace IIIWhite & Case LLP

1155 Ave. of the AmericasNew York, NY 10036-2787

Sonia K. Pfaffenroth Arnold & Porter Kaye Scholer601 Massachusetts Ave., NW


Jeffrey Prisbrey Charles River Associates

1201 F St., NWWashington, DC 20004-1229

Ian Simmons O’Melveny & Myers LLP

1625 Eye St., NW Washington, DC 20006-4061

Robert A. SkitolWashington, DC

Joshua SovenWilson Sonsini Goodrich

& Rosati1700 K St., NW


Vanessa A. TurnerAllen & Overy LLP

Avenue de Tervueren 268A Brussels 1150 Belgium

David P. WalesSkadden Arps Slate Meagher

& Flom LLP 1440 New York Ave., NW Washington, DC 20005

Nicholas A. WidnellBoies Schiller Flexner LLP1401 New York Ave., NW Washington, DC 20005

Willard K. TomMorgan, Lewis & Bockius LLP 1111 Pensylvania, Ave., NW Washington, DC 20004-2541

ABA Section of Antitrust Law Publications Officer

Jonathan I. GleklenArnold & Porter Kaye Scholer601 Massachusetts Ave., NW


Design & Production

Gary ArchambaultArchambault Design LLC

PO Box 969/115 Maple St. • Litchfield, CT 06759860/361-9430

[email protected]

Executive EditorTina Miller

81 Seventy Acre Rd.Redding, CT 06896tel 203/938-8507 fax 203/[email protected]

Editorial Board ChairGregory G. Wrobel

Vedder Price PC222 North LaSalle St.Chicago, IL [email protected]

Articles EditorLisa A. Jose Fales

Venable LLP575 7th St., NW


Articles EditorEllen Meriwether

Cafferty Clobes Meriwether & Sprengel LLP 1101 Market St.

Philadelphia, PA 19107

Litigation Practice EditorLisa C. WoodFoley Hoag LLP

155 Seaport BoulevardBoston, MA 02110

S P R I N G 2 0 1 8

from the Section ChairWhen Antitrust Met

Economics

Dear Colleagues,

THIS EXCITING EDITIONof ANTITRUST features piecesfocusing on an array of issues atthe intersection of economicsand antitrust.

Economics today is central to everyantitrust case. But that was not always so. From 1890 throughthe 1940s, economic analysis was largely an afterthought.The work of the Temporary National Economic Committee(1938–1941), which was highly critical of large firms andtheir impact on the economy, began to change all that. It ledto governmental exposure to what economists had dubbedthe structure-conduct-performance paradigm—the idea thatconcentrated markets lead to lessened competition which, inturn leads to poor economic performance.1 That concept,championed in the United States by Joe Bain,2 prompted leg-islative interest in strengthening merger laws. A catalyst for that interest was the Supreme Court’s 1948

decision in Columbia Steel,3 which refused to hold illegal ahighly unpopular acquisition by then-dominant U.S. Steel,allowing Congress to conclude that anti-merger enforcementunder the Sherman Act and the original 1914 version of the Clayton Act was inadequate. The upshot was the 1950Celler-Kefauver amendment to Section 7 of the Clayton Act,designed to halt mergers that increased concentration due to“a fear of what was considered to be a rising tide of economicconcentration in the American economy.”4

Celler-Kefauver led to Brown Shoe 5 in 1962 and, a yearlater, Philadelphia National Bank,6 the Supreme Court’s firstserious foray into economics as a guidepost for antitrust deci-sion-making.7 That decision, widely reported to have beenwritten by Justice Brennan’s then-law clerk, Richard Posner,expressly endorsed the use of economics to create a pre-sumption of illegality for mergers resulting in “undue” con-centration—found there because the combined market sharesexceeded 30 percent. A few years later, active structuralism reached its zenith in

the Neal Report.8 It recommended blocking most mergers inmarkets with four-firm concentration levels above 50 percentand breaking up oligopolies (defined as markets with four-firm concentration of 70 percent or more) such that no firmwould have a market share exceeding 12 percent. Even Areedaand Turner, in the first edition of their treatise, recommend-

ed “no fault monopoly” proceedings—to break up monop-olies through divestitures in equitable proceedings by thegovernment.9

A change in this approach began gradually, but then accel-erated rapidly. The SCP paradigm came under attack from anumber of lawyers and economists associated with theChicago School,10 and their work gave rise to the highly influ-ential Airlie House Conference in 1974.11 At the same time,the Department of Justice suffered its first loss in an anti-merger case under Celler-Kefauver in General Dynamics.12

That development was followed in 1977 by the SupremeCourt’s full embrace of economic analysis in Sylvania,13 andlater by the highly influential 1982 Merger Guidelines.14 Overthe next several years, many of the Supreme Court’s pre-eco-nomic decisions were formally (and informally) overruled.15

As a result, today, vertical intrabrand restraints are virtuallylawful per se, monopolization cases are confined to seriouslyexclusionary conduct, and mergers are allowed absent a highprobability of increased prices or other types of tangible con-sumer harm.These developments fostered concerns in the 1980s among

the pro-enforcement community that reliance on economicswas hampering antitrust enforcement. Fredrick Rowe ex -pressed these views most clearly in his superb article about the“Faustian pact” between antitrust law and economics.16 Theconcerns were magnified by the steep drop in merger enforce-ment in the 1986 to 1988 period. But what we have observedover the last few decades is quite different. Economics has pro-vided a grounding for different antitrust theories of harm byensuring they are theoretically sound. And it has provided newtools for antitrust enforcers, many of which are quite expan-sive. These theories and tools have allowed antitrust enforcersto reexamine competitive behaviors and to better understandtheir competitive implications. The upshot is that economicshas not limited antitrust enforcement, but has provided aclear, principled framework under which the antitrust agen-cies are able to pursue investigations and cases—and to honein on those cases that are most likely to be harmful to anti -trust’s true constituency: U.S. consumers. Consider, for instance, how far we have come with regard

to market definition. From the limitations of the quasi-eco-nomic but difficult to maneuver “practical indicia” approachto identifying antitrust-relevant markets in Brown Shoe Co.v. United States, we have at our disposal now sophisticatedeconomic tools to help us stake out the appropriate marketboundaries. As the agencies’ 2010 Horizontal Merger Guide -lines acknowledge, tests like the hypothetical monopolist/small but significant and non-transitory increase in price(SSNIP) test provide deeper insights regarding available sub-stitutes and demand elasticities than qualitative observa-tions alone. Moreover, economic tools have helped agencies to better

understand and identify markets where anticompetitive pricediscrimination may result.17 Despite initial fears that the1982 Guidelines’ SSNIP test would lead to very large mar-

S P R I N G 2 0 1 8 · 3

D E P A R T M E N T S


D E P A R T M E N T S

kets, permitting very large mergers and limiting the ability tobring monopolization cases,18 almost the opposite has beentrue. The SSNIP test, and its price discrimination (or “tar-geted customers”) variant, have in fact identified compara-tively small markets and allowed aggressive merger enforce-ment when the facts allow.19 Statistics from the Departmentof Justice and Federal Trade Commission’s annual reportsdemonstrate the agencies are actively pursuing cases today—on average, bringing about 15 to 25 enforcement actionseach per year.20

Economics has also allowed us to focus more directly onactual effects, recognizing that market definition is “not anend in itself, but is useful to the extent it illuminates themerger’s likely competitive effects.”21 Scholarly work devel-oping and testing economic tools have allowed us to assessprobable competitive effects more effectively than was previ-ously possible.Economics has also permitted us to reevaluate more accu-

rately other competitive actions previously presumed to beharmful. Take resale price maintenance (RPM), for example.Early cases, like Dr. Miles Medical Co. v. John D. Parks & Sons Co.,22 concluded RPM was necessarily unlawful becauseit removed merchants’ (or other distributors’) discretion to setprices. Because this eliminated some amount of (intrabrand)price competition, courts perceived RPM as going to theheart of what antitrust laws were intended to prevent. Considering the underlying economics, however, led the

courts to more enlightened conclusions. In Leegin CreativeLeather Products, Inc. v. PSKS, Inc.,23 the Court relied heavi-ly upon economic theory and insights. It explained that whileRPM may increase some prices, the “economics literature isreplete with procompetitive justifications for a manufacturer’suse of resale price maintenance,” including—importantly—aligning incentives between manufacturers and retailers andavoiding free riding.24 The Court further acknowledged thatthese procompetitive justifications often spur competitionbetween brands in ways that benefit consumers, by encour-aging brands to improve showrooms, invest in productdemonstrations, train employees, and preserve brands’ repu-tations. Because fostering these many benefits is a goal ofantitrust, the court refused to find such arrangements per seunlawful. That is, the Court adopted an economicallygrounded approach that utilized theory and empirical evi-dence to establish a framework that better preserves the con-sumer welfare benefits at the heart of antitrust law.These and numerous other decisions and agency actions

reflects the many benefits of closely integrating economicsinto antitrust analysis and case law. The close kinship betweenantitrust and economics that has enhanced our understand-ing and enforcement of competition laws continues to thisday. The articles in this issue reflect this ongoing relationship.They tackle some of the most interesting and challengingissues for antitrust law today—including how mergers impactinnovation, the treatment of network effects, and the bene-fits of new and existing economic tools—providing both

interesting and practical insights into the state of modernantitrust law.�

All the very best,

Jonathan M. JacobsonChair, ABA Section of Antitrust Law2017–2018

1 EDWARD CHAMBERLIN, THEORY OF MONOPOLISTIC COMPETITION (1933); EDWARDS. MASON, ECONOMIC CONCENTRATION & THE MONOPOLY PROBLEM (1964).

2 JOSEPH S. BAIN, BARRIERS TO NEW COMPETITION (1956). 3 United States v. Columbia Steel Co., 334 U.S. 495 (1948). See also FED.TRADE COMM’N , THE PRESENT TREND OF CORPORATE MERGERS AND

ACQUISITIONS, 80th Cong., 1st Sess. (1947).4 Brown Shoe Co. v. United States, 370 U.S. 294, 315–16 (1962).5 Id.6 United States v. Phila. Nat’l Bank, 374 U.S. 321 (1963).7 United States v. E.I. duPont de Nemours & Co., 351 U.S. 377 (1956), wassomething of an exception, relying on cross-elasticity of demand (in wayslater criticized) as the guidepost for market definition.

8 PHIL C. NEAL ET AL., REPORT OF THE WHITE HOUSE TASK FORCE ON ANTITRUSTPOLICY 2 reprinted in 2 ANTITRUST L. & ECON. REV. 11 (1968–1969) [here-inafter Neal Report]. Robert Bork dissented. Interestingly, William Baxter didnot.

9 PHILLIP AREEDA & DONALD TURNER, ANTITRUST LAW ¶¶ 630–638 (1st ed.1978).

10 The seminal piece was Robert H. Bork & Ward S. Bowman, Jr., The Crisis inAntitrust, 1965 COLUM. L. REV. 363.

11 The papers are collected in INDUSTRIAL CONCENTRATION: THE NEW LEARNING(Harvey Goldschmid et al. eds., 1974).

12 United States v. Gen. Dynamics Corp., 415 U.S. 486 (1974).13 Cont’l T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36 (1977).14 U.S. Dep’t of Justice, Merger Guidelines (1982).15 These are collected in Jonathan Jacobson, Scholarship & Antitrust (2016),

https://www.wsgr.com/publications/PDFSearch/jacobson-0116.pdf.16 See, e.g., Frederick M. Rowe, The Decline of Antitrust and the Delusion of

Models: The Faustian Pact of Law and Economics, 72 GEO. L.J. 1511 (1984).17 U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines

§§ 3–4 (2010).18 Gordon B. Spivack, New Merger Guidelines Are Substantially Different, LEGAL

TIMES, Aug. 2, 1982, at 38.19 E.g., FTC v. Sysco Corp., 113 F. Supp. 3d 1 (D.D.C. 2015) (broadline food-

service distribution to national customers); United States v. H&R Block, Inc.,833 F. Supp. 2d 36, 58–60 (D.D.C. 2011) (do-it-yourself tax preparationproducts).

20 See Fed. Trade Comm’n, Competition Enforcement Database, https://www.ftc.gov/competition-enforcement-database (last visited Jan. 18, 2017);U.S. Dep’t of Justice & Fed. Trade Comm’n, Hart-Scott-Rodino AnnualReport, Fiscal Year 2016, https://www.ftc.gov/system/files/documents/reports/federal-trade-commission-bureau-competition-department-justice-antitrust-division-hart-scott-rodino/p110014_fy_2016_hsr_report_final_october_2017.pdf.

21 Horizontal Merger Guidelines, supra note 17, § 4. 22 220 U.S. 373 (1911). 23 551 U.S. 877, 895–96 (2007). 24 Id. at 889.

S P R I N G 2 0 1 8 · 5

Editor’s Note:Economic Evidence inAntitrust Cases andEnforcementB Y G R E G O R Y G . W R O B E L

THE COVER THEME FOR TH I Sissue focuses on economic evidence in antitrustcases and enforcement. This Note poses threequestions for consideration that relate to thetheme:

(1) How has the core discipline of industrial organizationeconomics which supplies much of this evidence performedin meeting the evolving practical needs of parties, enforce-ment agencies, and courts that seek to apply relevant eco-nomic principles to the antitrust disputes they face.(2) How have courts and enforcement agencies managed

pretrial proceedings and investigations to capture the impor-tant insights that economic evidence affords, while balancingthe cost, time, and effort that parties must undertake to sup-ply this evidence.(3) Do the theories, analytical tools, and insights of indus-

trial organization economics differ among jurisdictions andtheir various antitrust/competition laws.The brief anecdotes below focus on some issues of current

interest and concern in antitrust and competition lawenforcement, and will serve as context for these comments.We will look for opportunities in the future to discuss inmore detail how economic evidence is performing across abroader range of current antitrust cases and agency investi-gations in the United States and other jurisdictions.

Two-Sided MarketsThe U.S. Supreme Court has granted certiorari in State ofOhio v. American Express Company (Amex).1 The case presentsimportant issues of how economic theory and evidence applyto two-sided markets, and is a fitting and timely focus forconsideration of the first question posed above.

Tirole Nobel Award. Jean Tirole received the 2014Nobel Prize in Economics, for which the Economic Sciences

Prize Committee compiled scientific background to supportthe award, describing Tirole’s important contributions, interalia, on network competition and two-sided markets:2

In a two-sided market, the two sides (say, buyers and sellers)interact via a platform. Examples include operating systems,payment cards (credit, debit or charge cards), shopping malls,and TV-channels.

A concrete example of a two-sided market is given by cred-it-card networks (such as Visa, Mastercard, or AmericanExpress). The two sides of the market are the consumers andthe retailers. If a certain credit-card company charges retail-ers a high transactions fee, a retailer might decide to notaccept this card. This might, however, lead consumers whoprefer this card to shop elsewhere. On the other hand, thereis a positive feedback loop between merchant acceptanceand consumer usage. In a pioneering article, Rochet andTirole (2003) analyzed the equilibrium of this kind of two-sided market, and studied its welfare properties. The modelwas generalized in Rochet and Tirole (2006). Key questionsaddressed in these articles include the equilibrium pricingstructure, and the extent to which consumers and retailers usemore than one network (“multi-homing”).

In platform markets, demands from the two sides can be verydifferent. For example, advertisers might desire that there bemany viewers or readers, whereas viewers and readers oftenprefer that there be few advertisers. As a result, prices thatwould be clearly anti-competitive in a one-sided market canbe highly competitive in a two-sided market. For example,offering newspapers for free would be a sign of predatorypricing if the newspaper’s only source of revenue came fromreaders, but may be entirely consistent with competitivepricing if advertising revenues are important. Because con-ventional tests for anti-competitive behavior are not appli-cable in platform markets, the work by Rochet and Tirole(2003) has had an immediate impact on competition policy(see Evans, 2009).

Without delving into technical details of theoretical analy-sis by Tirole and others, the Committee’s summary reflectsimportant points about economic theory for two-sided mar-kets:� A two-sided market is a single market, not two separatebut related economic markets.

� This is due at least in part to positive feedback between thetwo sides of the market, which is an important character-istic for analyzing pricing and other conduct by the plat-form owner.

� Participants on each side may exhibit much differentdemands/preferences, so an equilibrium pricing structurethat might be deemed anticompetitive in a one-sided mar-ket (or for one side of a two-sided market), can be highlycompetitive in a two-sided market.

� Examples of two-sided markets offered by the Commit -tee—credit-card networks, newspapers—have been oper-ating for many years, but the Committee noted thatTirole’s 2003 article has had an immediate impact oncompetition policy.Although over-simplified, this summary will serve as a

template (Tirole Template), to view arguments related to

Gregory G. Wrobel, Editorial Board Chair of ANTITRUST, is a shareholder and

head of the Antitrust Practice Group of Vedder Price P.C. All opinion

expressed herein are his alone and do not necessarily reflect those of his

firm or any of its clients.

E D I T O R ’ S N O T E


among card companies satisfies plaintiffs’ initial burdenunder the rule of reason.

� Two sets of users is not sufficient for a platform to be twosided; increased usage on each side must also benefit theother side (referred to in economics as two-sided exter-nalities).

� The court of appeals departed sharply from prior antitrustanalyses and rulings involving two-sided markets.

� Formal economic analysis of two-sided platforms is rela-tively new, but antitrust analysis of industries involvingtwo-sided platforms is not. For over 50 years, courts haveanalyzed the competitive impact of restraints on one sideof a two-sided platform by focusing on how competitionamong competing suppliers on that side of the market isaffected.

� Amex, as the proponent and enforcer of its rules, is in thebest position to understand and quantify any cognizablebenefits to cardholders resulting from any relevant two-sidedness of its business.

� Under standard rule of reason analysis, the first step is todetermine if a restraint injures competition between andamong platforms on the “side” where the restraint isimposed. If a plaintiff satisfies this showing, the defendantcan then show procompetitive benefits that may or maynot offset the anticompetitive impacts.The economists argued that the Amex credit card system

may not be a true two-sided platform unless increased usageon each side benefitted the other side, but did not argue (inany detail at least), that evidence from the case shows thatsuch benefits were not present. The economists also relied onhistorical legal precedent that long predates Tirole’s work,arguing that the plaintiffs need not prove an impact on theoverall equilibrium combined prices to merchants and card-holders in the platform market. As noted above, analysis ofcombined equilibrium prices seems to be a central tenet ofTirole’s theoretical work.

Petitioners’ Merits BriefThe states that petitioned for Supreme Court review sub-mitted arguments on the merits that largely diverge from theTirole Template:� Merchant and cardholder services are two separate anti -trust markets because the services are complements, notsubstitutes.

� Multi-sided platforms are not new. Newspapers bringtogether readers and advertisers and predate the ShermanAct. The Supreme Court treated the readership market asseparate from the advertising market in its 1953 decisionin Times-Picayune.

� The court of appeals departed sharply from this priorantitrust analysis of two-sided markets.These arguments eschew the Tirole Template and rely in -

stead on historical legal precedent to argue that the SupremeCourt should treat credit card platforms as two separate anti -trust (if not economic) markets.

economic theory about two-sided markets in selectedSupreme Court briefs in Amex.3

Economists’ Amicus Brief Supporting Petition for CertiorariA group of economists submitted an amicus brief in supportof the petition for certiorari filed by several states. Theirarguments reflect the Tirole Template only in part (e.g.,describing a single market), but diverge with arguments aboutburden shifting and historical legal precedent:� The economic literature on two-sided platforms is new,complex, and evolving, and should not be applied bycourts until its implications are fully understood.

� The court of appeals erred in ruling that proper antitrustanalysis must consider the two-sided net pricing account-ing for the effects of Amex rules on both merchants andcardholders, rather than whether the rules injured com-petition among credit card platforms.

� The court of appeals erred in ruling that plaintiffs had toprove an adverse net effect on combined prices to mer-chants and cardholders.

� In two-sided markets, the relevant competition occurs atthe platform level via price pairs offered by each platformto the two sides.

� A competitive two-sided market, through consumers’choices, will effectively decide the preferred and compet-itive price relationships (the price pair) and, as an inci-dental matter, the overall “price level” (the sum of theprices) in the two sides.

� Whether this sum goes up or not on the Amex platformdoes not relate to whether the Amex rules are or are notanticompetitive. Nor does it relate to how the restraintsmight distort and interfere with competition among plat-forms in two-sided markets.

� The Amex rules on the merchant side of the Amex plat-form directly interfered with this competition, and anti-competitive harm followed because the Amex rules alteredthe price pairs that different card platforms can offer.Perhaps the key point of divergence from the Tirole

Template is the economists’ argument that proof was notrequired of an impact on the overall (equilibrium) combinedprices to merchants and cardholders in the platform market,which seems to be a central tenet of Tirole’s theoretical work.The economists argued that the Amex rules interfered withprice pairs that other (mostly larger) credit card systems couldoffer, and that this evidence was sufficient to show anticom-petitive harm in the market.

Economists’ Amicus Brief Supporting Petitioners on the MeritsThe same group of economists submitted an amicus briefsupporting petitioners on the merits. Their merits argumentsreflect a similar pattern of divergence from the Tirole Tem -plate:� Evidence that the Amex rules impeded price competition

S P R I N G 2 0 1 8 · 7

United States’ Merits BriefThe United States opposed the petition for certiorari, butsubmitted a merits brief in support of petitioners with argu-ment that also largely diverges from the Tirole Template:� Amex had the burden of proving the impact of its rules inthe separate market for cardholder services.

� Merchant and cardholder services are two separate anti -trust markets because the services are complements, notsubstitutes.

� Competition on different sides of a two-sided platform isproperly analyzed as separate, albeit interdependent, anti -trust markets, as illustrated by the Supreme Court’s 1953decision in Times-Picayune.

� Plaintiffs do not have the burden of proving net harm tocardholders and merchants.

� The court of appeals erred in focusing exclusively on Amexpricing, because the Amex rules enabled rival card net-works to charge higher fees, and have blocked low-costrivals.The United States adopted a position similar to that of the

petitioner states, diverging from the Tirole Template in favorof historical legal precedent to argue that credit card systemscompete in two separate antitrust markets.

Economists’ Amicus Brief Supporting Amex on the MeritsTwo groups of economists filed amicus briefs on the meritssupporting Amex. Their briefs reflect much closer adherenceto the Tirole Template, not surprisingly given that the SecondCircuit’s decision in favor of Amex reflects key aspects ofTirole’s theoretical analysis.

Evans/Schmalensee.� The risk of error from ignoring customers on one side ofa platform is heightened for platforms that provide serv-ices that are consumed jointly and unseverably by twodifferent types of customers.

� This is not a matter of burden shifting. There is simply noway to know, especially in the case of such platforms,whether a practice is anticompetitive without at least con-sidering both types of customers and the overall competi-tion among platforms.

� In such cases there is a single service that is subject tocompetition, and it is that service that is interchangeableamong the customers that use it.

� The relevant literature, which started in 2000 with the cir-culation of a working paper version of Rochet and Tirole’sseminal contribution, is now 17 years old, encompasseshundreds of published papers, several major books, and isa standard and noncontroversial part of the modern indus-trial organization literature.

� Access and transaction prices affect the overall use of aplatform. How they do so depends on the structure ofdemand for participants to join and use the platform afterhaving joined. It is common, though certainly not univer-

sal, for two-sided platforms to lose money on one side ofthe platform.

� Considering the impact of challenged conduct on bothsides of the interaction is very different than the usualevaluation of procompetitive benefits in the second stageof a rule of reason inquiry. First, it is possible that the con-duct harms parties on which a restraint has not beenimposed, and failure to consider both sides at the firststage could lead to a false negative. Second, it is possiblethat the conduct benefits parties on one side of the plat-form. That benefit is part of the economic surplus gener-ated by the interaction between the parties and should beaccounted for in determining whether the practice reducesconsumer welfare.

� Platforms that provide similar jointly consumed services aresubstitutes for each other, and their products are inter-changeable as a matter of business reality. Market definitionfor platforms that provide services that are jointly con-sumed and unseverable should therefore focus on identi-fying suppliers that provide services that are interchange-able in this sense.

� The relevant antitrust market should thus consist of thosecompeting suppliers whose services are interchangeable. Toassess whether the conduct at issue in the case was anti-competitive at the first stage of the rule of reason analysisit is necessary to consider both parties to those transac-tions. The Second Circuit found that the Plaintiffs had notmet their burden because there was no evidence that theprice of those transactions increased and, most impor-tantly, there was no evidence that the conduct had reducedthe market output of transaction services.

Sidak/Willig.� Nondiscrimination rules enable networks to compete vig-orously for cardholders with rewards and benefits that canbe funded only with merchant fees. These rules might sti-fle one form of competitive pressure on the networks dueto their repression of steering, but they stimulate com-petitive pressure on networks to persuade merchants toaccept their cards with the balance of their fees, services,and the customers they attract.

� The rules allow two-sided market participants to competefor both merchant acceptance and cardholders with two-sided product differentiation—higher or lower merchantfees, corresponding higher or lower cardholder benefits,and corresponding bigger-spending or smaller-spendingtotal customers for the merchants. Without the rules,individual merchants would be motivated to steer cus-tomers to the network with the lowest merchant fees,thereby disabling the two-sided product differentiation.

� In the two-sided market, a credit-card network should bepermitted under the antitrust laws to compete with astrategy that needs to restrict a merchant that accepts thecard from discriminating against its use at the point of sale.To conclude otherwise, as Petitioners urge, would suppress

product differentiation in this two-sided market.� A two-sided market involves two distinct groups of con-sumers. Network externalities exist between those twogroups: the value that consumers on one side of the mar-ket derive from the consumption of the good or serviceincreases with the number of consumers on the other sideof the market. To prosper in a two-sided market, a firmneeds to compete against alternative platforms by appeal-ing to both groups of consumers with optimally balancedprices and benefits for each side of the market, such thatthe firm achieves an optimal aggregate price posture.

� Competition among credit-card networks for the businessof cardholders and merchants plays out on several dimen-sions: appeal of the cards and its services and rewards topotential cardholders; fees and services and the commerceof the cardholders that acceptance brings to the merchants;and the effectiveness of the network externalities that linkmerchants and cardholders that is driven by finding theright mixture of the terms offered to cardholders and mer-chants. A unilateral decision by a credit-card network toadopt non-discrimination rules does not suppress thesedimensions of competition in their total effect.

� Economist now widely embrace the definition of Tiroleand Jean-Charles Rochet that a multi-sided market is amarket “in which one or several platforms enable interac-tions between end-users and try to get the two (or multi-ple) sides ‘on board’ by appropriately charging each side.”

� Because a change in demand or cost on one side of a two-sided market affects the level and relationship of prices onall sides of the market, one must account for both sideswhen defining a relevant market. The hypotheticalmonopolist test may be analytically superfluous to thecharacterization of two-sided markets: the assessment ofconduct asks whether the challenged practice harms com-petition in its total effects on both sides of the market, tak-ing both sides as well as their interactions into account.

� Two-sided platforms have highly visible and readily iden-tifiable attributes, which this case illustrates. For instance,here two discrete groups of customers—merchants andconsumers—interact to create a single unit of output: acredit-card transaction. The volume of the output thatthe two sides create depends directly on the allocation aswell as the level of the aggregate price that Amex chargesthe two sides.

� From an economic perspective, increased cardholder ben-efits are equivalent to decreased prices on the cardholderside of this two-sided market. Holding a platform’s aggre-gate price constant, a decrease in the price for cardholderswill by definition require an increase in price for mer-chants. Thus, evidence of increased merchant fees in itselfyields inconclusive information about competition in thetwo-sided market for credit-card networks.The arguments of these two groups of economists mirror

the Tirole Template in all important respects: credit cardplatforms compete in a single market for credit card transac-



tions; positive network effects between the two sides of themarket motivate rival platforms to devise what each plat-form views as an optimal service/cost mix for both sides of theplatform; this mix may reflect higher pricing and conductrestraints on one side of the platform in order to attract cus-tomers on the other side; credit card platforms are well estab-lished, but the insights of Tirole’s theoretical work is now wellestablished as well, and provides the analytical frameworkappropriate to analyzing the competitive impact of restraintssuch as those used by Amex.

Amex’s Merits BriefAmex also hewed much closer to the Tirole Template with itsmerits brief:� Credit card services are simultaneous two-sided platformswith network effects and multi-homing (i.e., merchantsand consumers use multiple competing card networks/platforms).

� Network effects require the card network to balance pricesand other service terms.

� Proof of impact of the Amex rules on both merchantsand consumers is necessary due to feedback effects andsimultaneous consumption of services.

� Consumers and merchants jointly consume credit cardtransactions and their demands/preferences are often intension.Amex presented more succinct economic arguments than

the two groups of economists supporting their position, butthe brief still reflects each of the essential elements of theTirole Template.Economic theory will not answer all of the difficult and

disputed factual issues that Amex and other antitrust casespresent, but should at least establish a scientific framework foranalysis of case facts. The briefs in this single (albeit impor-tant) case do not capture the full impact of economic theo-ry applied to two-sided markets, but the arguments summa-rized above show significant divergence on (1) whethercompetitive impact should be analyzed in a single market orin two separate markets (filtered through legal argumentsabout burden-shifting standards), (2) whether positive net-work effects exist for the Amex platform, and how theseeffects impact economic analysis restraints on platform par-ticipants, (3) how equilibrium pricing (and output) for bothsides of the platform should be analyzed for anticompetitiveeffects, and (4) whether recent insights from economic the-ory for two-sided markets by Tirole and others is well enoughunderstood to be applied in actual cases.We may each draw different conclusions on the implica-

tions of this divergence in considering our first question:how well are industrial organization economists meeting thepractical needs of parties, courts, and enforcement agencieswho seek to apply relevant economic principles in antitrustlitigation and enforcement. The briefs discussed above at least suggest that, in the

aftermath of Tirole’s 2003 and 2006 articles, economists

S P R I N G 2 0 1 8 · 9

either have not yet settled on analytical frameworks for defin-ing economic markets for two-sided platform services, andanalyzing pricing, output, and ancillary restraints of plat-form owners, or that economists who take sides in importantantitrust cases will always be able to frame arguments thatchallenge or disregard settled analytical frameworks.

Patent TrollsPractical issues related to economic evidence are illustrated byantitrust litigation and enforcement in matters involving theemergence and growth of non-practicing entities that holdlarge portfolios of patents, referred to by some as patenttrolls.4

The district court in Intellectual Ventures I LLC v. CapitalOne Financial Corp.,5 entered summary judgement in favorof the plaintiff (IV), on antitrust counterclaims asserted byCapital One. The procedural path and grounds for disposi-tion of the case in the district court make it a fitting and time-ly focus for consideration of our second question, of howcourts and enforcement agencies are managing antitrust casesand investigations to capture important insights from eco-nomic evidence while balancing the burden associated withproviding this evidence.In earlier rulings, the district court entered summary judg-

ment against IV on its infringement claims for four patents,based in part on a ruling of invalidity as to two of the patentsin a separate case between the parties. The Federal Circuitaffirmed these rulings. The district court also denied IV’smotion to dismiss Capital One’s antitrust counterclaims basedon alleged sham patent litigation and unlawful acquisitions ofpatents, which led to “another round of extensive (and expen-sive) discovery regarding liability on the antitrust claims.”6

The district court judge appointed a technical advisor toassist the court in understanding the economic experts’ work,and held a tutorial with the testifying economists. Summaryjudgment materials on the antitrust claims included a jointrecord of 286 exhibits over 13,000 pages in length.The court’s summary judgment ruling on the antitrust

counterclaims (excerpted and paraphrased below), illustratessome of the key disputes about economic evidence that anti -trust claims against patent assertion entities may present, onissues or market definition, monopoly power, and willfulconduct:7

� The exercise of monopoly power with regard to a smallnumber of patents usually does not offend antitrust law.But it is another matter to acquire a vast portfolio ofpatents that are essential to technology employed by anentire industry and then to compel its licensing at take-it-or-leave-it prices, because it is not economically feasible todetermine if alternative technologies are available that arenot covered by the portfolio.

� Capital One’s economic expert viewed IV’s conduct afterhaving acquired monopoly power as critical to antitrustscrutiny (i.e., patent aggregation, concealment, and liti-gation, which gave IV insurmountable bargaining power

enabling it to demand take-it-or-leave-it supracompetitiveprices to license its patent portfolio). The expert analo-gized the portfolio to a “cluster market” that IV promotesas a single product for which there are no close substitutes.

� IV’s economic expert argued that patents in the portfoliotouch on a large number of distinct technology markets,each of which must be analyzed using SSNIP analysis.The court observed that it stretched plausibility to the nearbreaking point to assert that it would be economicallyfeasible for Capital One to discern the particulars of eachof IV’s thousands of patents. The court noted that IVdesignated nine Ph.D.s to support the good faith basis forinfringement claims on only four patents, suggesting avastly broader scope of analysis and cost for Capital Oneto determine whether technologies it acquired before IVmade its licensing demands infringed the thousands ofpatents in IV’s portfolio.

� The court observed that, although the Antitrust Guide -lines for the Licensing of Intellectual Property apply theSSNIP analysis favored by IV’s expert, the Guidelines donot address the near impossibility of doing so with a col-lection of intellectual property as massive as IV’s. Thecourt also observed that, even if cluster market analysis isnot the appropriate framework for analyzing relevantantitrust markets in cases involving large patent pools heldby non-practicing entities, there is something concerningfrom an antitrust perspective about the way in which IVengages in its licensing business.The court stated that it would deny summary judgment

if the only issues were relevant market definition, IV’s pos-session of monopoly power, and the willful acquisition ormaintenance of that power, but then granted summary judg-ment based on Noerr immunity for IV’s IP litigation conduct,and collateral estoppel arising from rulings in a related casebetween the parties.IP cases with antitrust counterclaims present difficult

issues for economic evidence on key issues of relevant mar-ket definition, whether the non-practicing entity possessesmonopoly power, and whether its patent acquisition/enforce-ment conduct reflects willful exercise of that power. Tied upin these issues are potentially overwhelming costs related tothe sheer volume of expert evidence that may be warrantedto litigate these claims.These challenges underscore the important role of dis-

trict courts in managing the pretrial path of litigation. Closerreview of the case docket may reveal explanations for thepath that was chosen in this case, but the court’s eventualgrant of summary judgment based on Noerr immunity andcollateral estoppel suggests that a great deal of effort andexpense may have been avoided if those issues were resolvedbefore merits discovery, and before the economic expertsperformed work on the substantive antitrust claims.Of course, sequencing discovery and expert work in this

way may extend the time span of a case, a result that may alsobe undesirable. Economic analysis will not provide ready


1 0 · A N T I T R U S T

nobel_prizes/economic-sciences/laureates/2014/advanced-economicsciences2014.pdf.

3 Briefs discussed in the text are available on the Supreme Court website athttps://www.supremecourt.gov/search.aspx?filename=/docket/docketfiles/html/public/16-1454.html. The short summaries here do not distill allof the arguments in the briefs, but rather seek to focus on elements of argu-ment that—to this reader at least—show whether and how economic the-ory on two-sided markets is serving as an analytical framework for analysisof the question presented for Supreme Court review: “Under the ‘rule of rea-son,’ did the Government’s showing that Amex’s anti-steering provisions sti-fled price competition on the merchant side of the credit-card platform suf-fice to prove anticompetitive effects and thereby shift to Amex the burdenof establishing any procompetitive benefits from the provisions?”

4 See, e.g., Fed. Trade Comm’n, Patent Assertion Entity Activity (Oct. 2016),https://www.ftc.gov/reports/patent-assertion-entity-activity-ftc-study.

5 No. PWG-14-111, 2017 U.S. Dist. LEXIS 197714 (D. Md. Nov. 30, 2017).6 Id. at *24. 7 Id. at *30–46. 8 Paul Lugard & Lee Roach, The Era of “Big Data” and EU/U.S. Divergence for

Refusals to Deal, ANTITRUST, Spring 2017, at 58. 9 Id. at 62–63.

answers to these tensions, but careful planning of pretrial pro-ceedings may help assure that the cost and burden of eco-nomic analysis is used sparingly, and when necessary in con-junction with work on other potentially dispositive issues ina case.

Big DataEU competition law enforcers have raised concerns and arepursing active investigations and enforcement actions focusedon use of very large data sets and data analytics by Internetsearch services and social media platforms used by consumers.U.S. antitrust enforcers for the most part have not pursuedsimilar conduct investigations apart from issues that mayarise in merger reviews. This apparent divergence is a fittingfocus for our third question, of whether the theories, analyt-ical tools, and insights of industrial organization economicsdiffer among jurisdictions and their various antitrust/com-petition laws.In a recent article in this publication,8 Paul Lugard and

Lee Roach noted apparent divergence in US and EU agencyenforcement involving “big data” and offered three explana-tions:� U.S. case law leaves little room to accommodate forcedsharing by competitors.

� The EU’s increased use of the commitment procedure inArticle 9 of Regulation 1/2003 provides the EuropeanCommission with a powerful enforcement tool byenabling it to accept commitments in exchange for ter-mination of a conduct investigation.

� Individual EU member states are not bound by case lawunder Article 102 TFEU and thus may seek to requiredominant firms to share customer data with their com-petitors.Importantly, none of these explanations derive from an

underlying divergence in economic theory or evidence, andthe authors’ discussion about whether big data will becomesubject to an antitrust duty to deal with competitors suggeststhat economists may play an important role in forging greaterconvergence through focused economic analysis of businesspractices and markets where big data is widely used.9

Economic analysis and evidence is deeply embedded inantitrust and competition law doctrines, and is an essentialtool in agency enforcement and private litigation. The arti-cles in the cover theme—and hopefully the questions andcomments posed here—will serve as aids in addressing themany practical and technical issues that this important evi-dence presents.�

1 838 F.3d 179, cert. granted, 86 U.S.L.W. 3184 (U.S. Oct. 16, 2017) (No.16-1454).

2 Scientific Background on the Sveriges Riksbank Prize in Economic Sciencesin Memory of Alfred Nobel 2014, Jean Tirole: Market Power and Regulation,compiled by the Economic Sciences Prize Committee of the Royal SwedishAcademy of Sciences 29–30 (Oct. 13, 2014), http://www.nobelprize.org/

Antitrust Masters Course

October 18–20, 2 018

Grand Hyat t Chesapeake Bay Golf Resort, Spa, and Marina

Cambridge, MD

www.ambar.org/atmasters • #atmasters

Conference Chair: Howard Feller

� JOIN US FOR ONE OF THE MOST highly-regarded training opportunities for seasoned antitrust practitioners with 4 to 10 years of experience. The Masters Course—offered only every other year—

provides attendees with a unique chance to enhance theirknowledge, refine their practice skills, and gain valuable perspectives on the cutting-edge developments in antitrustlaw, all while networking with government enforcers and distinguished antitrust practitioners. The three-day event will feature an immersion in the trends and tensions in currentantitrust doctrine. Interactive sessions will be led by a terrificline-up of renowned antitrust law professors. Lunches anddinners will showcase key government enforcers offeringvaluable insight into current agency views and importantenforcement trends. Limited attendance gives the Masters Course a personal

feel and allows attendees to become acquainted and socializewith colleagues. Whether you are an experienced law firmassociate or partner, government attorney, or in-house counsel, you will not want to miss this exceptional event.

C O V E R S T O R I E S

S P R I N G 2 0 1 8 · 1 1

Editors’ Note: On January 22, 2018, a group of leading economists sat down with Jeff Jaeckel and James Keyte, Associate Editors

of ANTITRUST magazine, for a roundtable conversation. The wide-ranging discussion included practice pointers for working with expert

economists and exploration of issues that cut across many of the hot topics in antitrust today—coordination across multiple juris-

dictions, big data, novel analytical tools in merger review, issues that can arise around class certification, and a role for populism

in antitrust.

Obviously, things will come up and you may have to adjustthe plan.But where I most often see bills get out of hand is when

an expert just comes up with a million different ideas andtries to pursue all of them without first evaluating which ofthose ideas are actually likely to advance the ball. In additionto not perhaps being very effective, that approach certainly isnot very efficient.I think this basic concept is especially applicable in situa-

tions where there is a lot of data. If you are asking your clientfor everything under the sun you get huge amounts of data.

James Keyte isthe Director ofGlobal Developmentat the Brattle Group,the Director of the FordhamCompetition LawInstitute, a formerantitrust partner atSkadden Arps, andan Associate Editor of and frequent contributor to ANTITRUST magazine.

Phillip Johnson iscurrently ManagingDirector at Econ OneResearch. He hasworked as an expertin antitrust (andother economicissues) since 2000,including antitrustcases, such as TFT-LCD, CRT, and theHigh Tech Employees and Duke/UNCno-poaching cases.

Dennis W.Carlton is theDavid McDanielKeller Professor ofEco nomics at theBooth School ofBusiness at theUniversity of Chicagoand is associatedwith the economicconsulting firmCompass Lexecon. He served as theDeputy Assistant Attorney General in the Antitrust Division during2006–2008, and on the AntitrustModern ization Commission. He hasworked as an expert in numerousdomestic and foreign cases involvingissues in antitrust, regulation, and intellectual property.

Jeff Jaeckel is a partner atMorrison & FoersterLLP and co-chair ofthe firm’s GlobalAntitrust &Competition LawPractice Group. He is an AssociateEditor of ANTITRUST

magazine.

Roundtable with Economists: Discussing Practice and Theory

with the Experts

Gregory K.Leonard is a part-ner at EdgeworthEconomics LLC specializing inapplied micro-economics andeconometrics. He has written widely in the areasof antitrust, industrial organization, econometrics, intellectual property, and class certification, and is a SeniorEditor of the Antitrust Law Journal.

Maria Maher is aprincipal at Corner -stone Research. A competition economist with over 20 years ofinternational experience in a wideranging of industries,including financialservices, telecommunications, chemicals, energy, and commodities, Dr. Maher consults on complex mattersrelated to abuse of dominance, cartels,mergers, market investigations, con-tracts, and damages.

Carl Shapiro is a Professor of theGraduate School at the Univer sity of California atBerkeley. He servedas a Member of the Pres ident’sCouncil of EconomicAdvisers during2011–2012 and forthe two years immediately prior to that,he was the Deputy Assis tant AttorneyGeneral for Economics at the AntitrustDivision, a position he also held during1995–1996.

M O D E R A T O R S P A N E L I S T S

P A N E L I S T S

JAMES KEYTE:We’d like to start with some practical issues.Greg, law firms are facing ever increasing pressure to reducetheir own fees, but from the client’s perspective, it doesn’tmatter whether the fees go to the law firm or its expert econ-omist. In your experience, how can economic experts andeconomic consulting firms help manage and control theirshare of the cost?

GREG LEONARD: The approach that I have found to bemost effective is to think very clearly in advance about whatit is that you want to do and to then execute that plan.


1 2 · A N T I T R U S T

It then takes a tremendous amount of time to understand andprocess that data and it generates a bunch of additional ques-tions. In this context, attempting to implement a large num-ber of unfocused ideas can quickly run up large bills.Whereas, if you think in advance, here is what I really

want to do and I’m going to find out what data the client hasthat is actually directly responsive to that set of ideas, thenyou can focus things, not waste their time, and spend yourown time much more efficiently and effectively.

JEFF JAECKEL: Greg, you anticipated our next question,about cost control in this era of abundant data. Dennis, weunderstand—because expert economists always tell us—thatthe more data, the better. But the acquisition and analysis oflarge data sets can be very expensive and not every matter cansupport that expense.How do you think about the cost-benefit analysis, and

how do you work with counsel to deliver the most reliable,persuasive analysis while still being sensitive to the associat-ed cost?

DENNIS CARLTON: I think it’s not just the availability of bigdata. It’s also the availability of enormous electronic recordsof communications. And I would second what Greg said, theproper role for the economist is to advise the clients what arethe most relevant economic questions to answer.In my experience an economist can do that pretty well in

conjunction with lawyers, but I often find that manylawyers—especially on big cases—just want you to do any-thing you can think of that could possibly be relevant inorder to be prepared for whatever the other side might do.That might be the right thing to do for the client if it is a bigcase.On the other hand I’m still of the view that the most

effective economic presentation isn’t presenting a millionstudies. It’s focusing on the ones that are most relevant tomake your points and to rebut those of your opponent. AndI think that streamlines the process. My own experience is that neither judges nor lawyers seem

terribly interested in a presentation that includes all mannerof potentially relevant analyses, even if a client might be. Itsometimes seems the lawyers convince clients that it’s worthpursuing all these potentially relevant, but in the scheme ofthings not crucially important, investigations. That’s been myexperience.

JAMES KEYTE: Phil, econometric models and economicanalysis, of course, can be highly technical and complicated,and they are growing even more sophisticated all the timewith big data sources and different types of simulation andmodels.How can you make your work and analyses accessible in

that environment to council, to judges, to lay juries?

PHIL JOHNSON: It’s a big challenge. The volume of data

and documents has been increasing as Dennis and Greg indi-cated, but also the complexity and sophistication of the analy-sis that is put forward. It can get very technical, resulting inthe appearance that the experts sometimes seem to be writ-ing at each other rather than writing for the court and thelawyers and others in the courtroom who may need to under-stand the disputes.I usually think about presenting information in terms that

I heard some time back—of the hierarchy of clear commu-nication by pictures, words, and then numbers. The bestway to tell a story is with pictures. Words are less clear, butthey are still better than drawing out a lot of numbers. So when we have an analysis that we want to be under-

stood, we look for the picture we can provide. And that canbe an interesting challenge with a lot of analyses these days,but that is very important. Another thing that we have to bear in mind is to avoid

some of the highly technical back and forth that is really socomplicated that it will not be well understood and at the endof the day if it’s not key to the competing stories, it is not like-ly to be decisive. We want to focus on the central narrativeand try to make that story come across clearly and persua-sively, hopefully with images that do a much of the work oftelling the story.

MARIA MAHER: I just wanted to introduce an additionalconsideration—deadlines. Huge data sets and sophisticatedtools mean the analysis is more complex. That can pose chal-lenges in contexts like mergers where deadlines can some-times be tight.

JAMES KEYTE: Carl, obviously you have delivered manypresentations before experienced antitrust counsel and econ-omists at the DOJ and the FTC and elsewhere, and have alsobeen before judges and juries. How in your experience do youadjust the substance or the style of presentation for differentaudiences?

CARL SHAPIRO: Well, there is inevitably a big differencebetween presenting economic analysis to a judge and pre-senting it before an expert agency. When you are presentingto the DOJ or the FTC, a significant part of the audience areeconomists who are working in detail with the data and arefamiliar with economic models and methods. That reallychanges the exercise.I think an on-going question is how much of the really

detailed work that economists do has an impact in court. Ifthe judge thinks certain results are just a black box with duel-ing experts, then the judge has no good way of telling whichnumbers coming out of the black box are more reliable. Inthat circumstance, the analysis may not have much impact.But when you combine the quantitative analysis with otherqualitative evidence from documents and testimony to givea coherent picture of the business context and tell a story,then the detailed economic analysis can have a big impact.

S P R I N G 2 0 1 8 · 1 3

JEFF JAECKEL: Maria, there can often be a different style ofpractice in Europe, and obviously an absence of lay juries andlitigation. Across all of these subjects we have been talkingabout—cost containment, manner of presentation and per-suasion—how do you find the different environment inEurope affecting the manner of preparing and presentingyour analysis?

MARIA MAHER: I’m not sure that there is much of a differ-ence between the EU and the U.S. in preparing and pre-senting our work. As experts we need to be able to present ourwork in a way that the lay person can understand. Picking up on Carl’s point about the difference between

presenting the work to a court versus the expert economist atthe agencies, it’s also that often the lawyers are our clients,too, and we need to present the work so that the lawyers canbest understand it. In that regard, we do need to tell a clearstory. The story needs to fit the facts of the case and then bepresented in a coherent way. You can’t just have economet-ric results or abstract models that are not grounded in thefacts of the case. One of the differences between the U.S. and UK is the

introduction of what we call “hot tubbing” in the UK courts,where you may have opposing experts responding to questionsin tandem or, indeed, cross-examining each other. While theexperts may speak the same language, you still need to ensurethat the court follows what are complex arguments. At the endof the day, you need to present the analysis in layman’s lan-guage and avoid jargon as much as possible. That’s the bestapproach in Europe and other regions.

JEFF JAECKEL: And for others who have practiced in theU.S. and in other jurisdictions, do you have other thoughtson differences in your analytical substance or presentationstyle between the different jurisdictions?

CARL SHAPIRO: Let me add this. I testified about a year agoin front of the CAT, the Competition Appeals Tribunal in theUK. There you have an interesting hybrid with a three-judgepanel and one of the judges is an economist. They were reviewing a decision by the Competition and

Markets Authority. I was very impressed by the level ofsophistication that the panel brought to bear on the antitrustanalysis, including the economics. They also used a “hottub” format, with concurrent expert testimony, so the pres-entation of the economic evidence was very different than ina U.S. courtroom.

DENNIS CARLTON: Yes, I would second that. I have testifiedin New Zealand and in Canada with the same situation thatCarl just described, where there is someone appointed as amember of the tribunal who has training in economics. Andobviously that can make a big difference.I’ve also presented before the EU, and what’s different

about “testifying at the EU” is that it’s not the adversarial pro-

cedure that we would have in a court in the United States. Soit really does differ a lot depending on the country and thestyle, it seems to me, as well as the economic sophisticationof the audience. I have not found any audience in interna-tional proceedings as economically sophisticated as, for exam-ple, the economists at the Department of Justice or theFederal Trade Commission.

JAMES KEYTE: Let’s expand on that a little bit with Phil andfocus on writing. Obviously, practitioners have to learn howto write clearly, persuasively, and accurately. Sometimes itmay be difficult for economists coming out of PhD pro-grams to learn how to write for an audience that consists ofpractitioners and judges, as opposed to what they did intheir PhD programs. Have you found that to be a challenge when working

with young economists? And how do you deal with thatwithin your own shops?

PHIL JOHNSON: This is a challenge for younger economistsand it can also be a challenge for academic economists whoaren’t quite as experienced at communicating in the litigationsetting. My role is often to be between other economists andpractitioners—our clients. It sometimes requires some backand forth on language to get complicated ideas into layman’slanguage or at least a language that the layman will better findaccessible.One practice that’s helpful is to separate the more techni-

cal aspects into an appendix, preferably with an explanationthat is as near to plain English as possible. I think that a nice consequence of working to explain a

complicated idea to the lawyers we’re working with is that inworking to make it clear we come to understand the idea andits importance between ourselves. Only if our clients under-stand our analysis and can communicate it to the court, arewe succeeding at our job. Usually as long as we have time to do that back and forth

and we have counsel who is willing to invest the time tocommunicate with us about what they do and do not under-stand, we will get to a point where things are clear enoughthat hopefully the court will also understand it.

CARL SHAPIRO: I think it’s important for all of the senioreconomists to train the younger economists. Done well, theserelationships can be very synergistic. The fresh PhDs comewith a lot of technical skills and they want to use them, butthey also want to learn about how industries work and howto apply their skills in a real-world setting. We typically form teams of people with differing levels of

seniority. In my experience, it’s important for the senior peo-ple to moderate the tendency of the younger economists tofocus too much on the technical material, where they canshine. The senior economists need to guide the whole proj-ect in a direction that will be most informative and reflectbusiness reality. The senior people need to apply their expe-

1 4 · A N T I T R U S T

CARL SHAPIRO: Well, if by red flag, you mean, “That is it,we have a high number here, we’re done,” then no, I havenever seen that. However informative it may be, UPP is onlyone part of the overall merger analysis, just as HHIs are onlyone part of the overall merger analysis.

JAMES KEYTE: Carl, what I meant by “red flag” is, “Oh,there is something here. We need to look more carefully,including other quantitative and qualitative evidence.” Asyou highlighted, the UK was using this as a rebuttable pre-sumption. And so, in the U.S. for example, has it ever gotten to that

point where it is being used more as a presumption than asa warning that should put people on the alert to do a morerobust analysis on other parameters?

CARL SHAPIRO:Well, the FTC has used UPP as a screeningtool for settlement purposes. So, I think that is getting a lit-tle bit closer to what you are asking.

DENNIS CARLTON: My view is a little different than Carl’s.I think there is no question that UPP—and the literature thatJoe and Carl authored—made very valuable contributions,especially for how you should interpret the opportunity costand pricing impact of having a substitute product in yourportfolio in addition to your original product. I think thereis no question that they have provided a very useful inter-pretation.My concern with UPP has always been that it is a place to

start and end the analysis. Really, it is a shortcut for mergersimulation. As Greg said—and I think my experience is sim-ilar—when people use UPP, they will eventually want toturn it into a price prediction. And that means you are doinga simple merger simulation.So in terms of UPP being a step on the path to a merger

simulation, I think it is sensible. However, it is no substitutefor the merger simulation. UPP can be misleading in manycircumstances. People sometimes have the notion that UPPis somehow easier—that it doesn’t require as much fancyeconometrics as a merger simulation. This impression iswrong as I have explained in my articles. You would needmany of these same underlying, sophisticated notions regard-ing substitution and margins for UPP as you would for amerger simulation.My own view is that UPP can be useful in terms of

enlightening us how to interpret some of the incentives forpricing, but that it really is a step along the way to a fulleranalysis.

JEFF JAECKEL: Dennis, you had expressed some skepticismof the UPP methodology back in 2009 or 2010. You had alsosuggested, in response to the changes in the Merger Guide -lines, that there was a lack of empirical evidence showing theusefulness of merger simulation as a tool to predict the com-petitive effects of a merger.


rience to establish priorities so the team does not go off onwild goose chases or build complex but unreliable models.With the right team, it is very productive and there is goodlearning going on as well. I certainly keep learning fromyounger economists and I hope the reverse is true.

JEFF JAECKEL: Carl, let’s talk about one of our favoritetopics, UPP. It has been several years since you, Joe Farrell,and others brought the framework or concept of “upwardpricing pressure” to the DOJ and the FTC, and later foundthose concepts incorporated into the 2010 Horizontal Merg -er Guidelines. Did the agencies use that tool as you expect-ed? And do you foresee changes on the horizon in the use ofUPP?

CARL SHAPIRO:Well, I think the usage of UPP is evolving.I’d be curious to know what Maria thinks because when wewere revising the Merger Guidelines in 2009 and 2010, theOFT (now the CMA) was really ahead of the U.S. agenciesin terms of using this tool quite actively. I think FTC uses theUPP tool a bit more often than DOJ, partly because theFTC has a different industry focus. The state of economic knowledge regarding UPP is also

advancing. There have been numerous articles published onUPP since it was explicitly included in the Merger Guidelinesin 2010, just as there were numerous articles published aboutunilateral effects following the issuance of the 1992 mergerguidelines. Overall, I think that UPP is well-established andwell-supported and has proven workable in practice. Usage ofUPP will continue to evolve.

MARIA MAHER: I agree, the UK agencies at the time, theOFT and the Competition Commission (CC), broke somenew ground in cases like Somerfield/Morrison—which wasdone at the CC in 2005—to move away from market sharesand to think harder about whether the merger would intro-duce an incentive to raise prices—and did so by analyzingmargins and diversion ratios. And indeed trying to predict theeffect of the merger on prices. The approach has continuedto be an important part of analyzing horizontal mergers.

JAMES KEYTE: Does anybody else have experience in the EUor the UK with the current use of UPP—whether that hasfollowed the UK’s lead or whether that has been drawn backon in any way in the EU or the UK?

GREG LEONARD: I think that UPP has continued to beused. Not maybe in the strict sense of UPP but often takingUPP and then plugging it into a second step that gives youa predicted price change. This, in my experience, is prettycommon.

JAMES KEYTE: And is it, in all of your experiences, beingused as a red flag or as, in some sense, real evidence of likelyeffects?

S P R I N G 2 0 1 8 · 1 5

Obviously, time has passed, the analytical tools haveimproved, and much work has been done over that period.With the benefit of that experience and hindsight, do youhave thoughts now on the usefulness of merger simulationand merger analysis? And are there certain types of transac-tions that you think are particularly well-suited for thatapproach?

DENNIS CARLTON: Well, I think it’s a good question. Myunderstanding of the recent literature and some surveys ofmerger simulation is that there are some times where it workswell and times where it has not worked very well. The main issue when you do a merger simulation is that

you’re assuming for simplicity a certain type of competi-tion—usually Bertrand competition in prices—and it is astatic exercise, ignoring dynamic issues. So, you don’t payattention to changes over time and you don’t pay attentionto product changes or innovation. And in some industries,that would be okay, but not in others. Moreover, my experience is that people rarely test whether

the Bertrand assumption is valid. And for that reason, the lit-erature has shown that sometimes it works, sometimes itdoesn’t, and when it doesn’t, these retrospectives have shownthat they suspect the difficulty is in the assumed static com-petition game. So, I think merger simulation can be useful. I would sum-

marize its main usefulness as a way for an economist to get abetter understanding of demand interactions among a vari-ety of products. As econometric techniques have improved,we have the ability now to estimate pretty sophisticateddemand systems. So if product overlaps arise in a mergerbetween two firms, we can get good estimates of the variouselasticities and cross elasticities of demand. How do you interpret those elasticities and cross elastici-

ties? I think merger simulation gives you a way to interpretthem. I still have skepticism whether the exact predictionsfrom the merger simulation ultimately work out for a par-ticular industry. And I think there is still a lot more work thathas to be done to indicate when it’s likely to work and whenit’s not.

GREG LEONARD: I would agree largely with what Dennissaid—that probably the most important thing is that wehave a better understanding about the extent of substitu-tion, particularly between the products of the merging par-ties. Econometrics and some of the data that we’ve been talk-ing about would certainly be helpful for that. I will note, of course, that there are some industries for

which the merger simulation framework is better suited thanothers—for example, retail products. Once we have a goodidea about the extent of substitution, that, together withmargins, can feed into UPP, to calculate that opportunitycost. But the opportunity cost measure by itself is hard toevaluate. You’re often left saying, “I really would like some-thing a little bit more.” That is why people then take the UPP

opportunity cost and make certain further assumptions, suchas a particular functional form for demand, to get a versionof merger simulation, as Dennis said. So, I think it really is about taking the demand elasticities

and getting a more intuitive interpretation of them that canhelp us assess the merger. Again, as in all these things, we haveto be cognizant of what assumptions we are making, such asthe model of competition, constant marginal cost, and fixedproduct characteristics. As Carl said, when you get a result, whether it’s from

UPP or a merger simulation, you should go back and makeit part of an integrated analysis that looks at other things,which in turn perhaps allows a test of the assumptions usedin the merger simulation. And then, the amount of weightyou end up putting on it depends on the outcome of all thatanalysis. So, I think merger simulation is very useful. I’ve used it a

lot myself and I’ve always been a big proponent of it.However, it does have to be recognized that it’s one piece ofa larger analysis. It has certain assumptions—if you can testthose assumptions you should do so, and if you can’t thenyou at least have to consider the sensitivity of the results tothose assumptions in a given circumstance.

MARIA MAHER: That’s absolutely right. Picking up on thepoint that Dennis made, these simulations to some extent dohave their limitations because they don’t take into accountwhat is increasingly important like innovation or changes inquality, and these are a lot more difficult to measure. At theend of the day, you do need to consider that the UPP ormerger simulation is just one component or element of thewhole analysis. It’s particularly important to keep in mind thelimitations of our merger simulation toolbox when we arelooking at high-tech, dynamic, or innovative markets.

JAMES KEYTE: Moving to vertical mergers, which certainlyare being discussed and thought about a lot lately, I wantedto ask you, Carl, in your experience, has the framework forthe economic analysis of vertical mergers changed or evolvedover the past several years?

CARL SHAPIRO: Yes, I think it has. Steve Salop has playedan especially important role here, through his extensive writ-ings about vertical mergers. Steve’s papers with Serge Moresion the vertical version of UPP, or GUPPIs, provide someimportant tool, the vGUPPIs, that are used in practice. While economists have made real strides in our under-

standing of vertical mergers over the past 30 year, we have hadno clear guidance from the agencies and no recent case lawon vertical mergers. The most recent vertical merger guide-lines date from 1984 and are widely considered to be a deadletter. We have some recent clues. Jonathan Sallet gave a veryinformative speech on vertical mergers in November 2016,“The Interesting Case of the Vertical Merger,” when he wasDeputy Assistant Attorney General, but he is no longer at the


1 6 · A N T I T R U S T

DOJ. And we can learn quite a bit about agency practice overthe years from the competitive impact statements associatedwith consent decrees in vertical mergers. But the absence ofmodern vertical merger guidelines is unfortunate. In short,while the economic analysis of vertical mergers has madegreat strides, it is hard to tell just what approach the DOJ andthe FTC will take in the future regarding vertical mergerenforcement.

JAMES KEYTE: And, when we think of vertical mergers inthe context of highly differentiated products—whether it beGame of Thrones, CNN, etc.—how does an economist real-ly assess market power in a vertical merger in the context ofhighly differentiated products? What are the challenges there? And let me give a little more context, practitioners often

think that you look at the degree of market power upstreamand the effects downstream, at least in the vertical context.But in the context of traditional market definition and mar-ket power analysis, how is that complicated by differentiat-ed products that might be viewed as must-have products?

CARL SHAPIRO: Suppose we have a vertical merger in whichone downstream competitor purchases an upstream inputand we are concerned about how the merger will affect down-stream competition. In this situation, it is important to lookinto the importance of the input that is being acquired. We can ask a series of well-defined questions. If the down-

stream rivals are denied access to that input, what substitutes,if any, would they turn to? How much would they be weak-ened as downstream competitors? What if these downstreamrivals can still obtain the input but are forced to pay higherprices for the input? These are going to be some of the keyquestions in the analysis.

DENNIS CARLTON: The distinction is that by its nature ahorizontal merger eliminates a competitor while a verticalmerger does not. That doesn’t mean that a vertical mergercan’t be harmful. I have written some papers that identify cir-cumstances in which vertical issues can create problems. Butyou don’t want to lose sight of the fact that in general thereare inherent efficiencies in vertical mergers and the surveys ofvertical mergers have shown them to be generally desirable. Now, when you turn to a question like how would you

analyze a situation where as a result of a vertical merger themerged firm raises the price of an input to rivals, I agree withCarl that you must examine the consequences on competi-tion. And if that price increase has only a trivial effect onrivals and there are efficiencies associated with the merger,one should think hard about whether you are impeding effi-cient industrial organizational change by blocking the merg-er in order to protect a rival. There is one other point that I have always found very per-

suasive but I must admit I am not sure any of the regulatoryagencies have yet found persuasive in thinking about thepossibility of vertical mergers creating harms. When there is

a merger that claims to generate efficiencies, the MergerGuidelines explain that you need to show that the efficien-cies are merger-specific and could not otherwise be achievedby contract. Ronald Coase and Oliver Williamson wonNobel Prizes for explaining why you can sometimes achievethrough vertical merger what you cannot achieve throughcontract. But exactly this same logic explains that when someone

claims that there is harm from a vertical merger: you shouldask whether that could have been achieved by contract, pre-merger. If it could have been achieved pre-merger but was not,that would be evidence that the vertical merger also will notresult in harm. That is, a claim that a vertical merger createsharm must explain why that harm is merger-specific. I havenever seen that question examined and it seems to me it’s avery relevant one. That is beyond what you asked but I thinkit is relevant in the context of understanding vertical mergers.

JEFF JAECKEL: Maria, with the continued globalization ofcompetition laws and enforcement and the steadily increas-ing incorporation of economic principles by competitionforces outside the United States, practitioners in these largemulti-jurisdictional matters often find themselves consider-ing whether it is necessary or desirable to retain multipleexperts so that they have an expert in each of the affected orrelevant jurisdictions. We are interested to hear about your experience and what

you would advise practitioners or what you would say is bestpractice when operating across multiple jurisdictions andpresenting economic analysis to multiple regulators.

MARIA MAHER:Well, from a practical point of view, I thinkit’s important to have some element of a local team if there’sgoing to be a lot of discussion with the Commission or NCAabout the effects of the merger. Where local teams are need-ed, there are coordination challenges, but there’s not a fun-damental problem with coordinating across jurisdictions.Indeed, there can be learning across jurisdictions. And ofcourse the time differences can work to your advantage, whenone side of the Atlantic goes to sleep, the other is getting intotheir day. So, I think there are definite advantages to having a local

aspect to teams but also, sometimes, a core team workingacross jurisdictions.

JAMES KEYTE: And this is a follow-up on that with Carl,Dennis, and Greg. I think all of you have participated in presentations and testimony across jurisdictions. Have younoticed a practical difference in the way competition enforcerswork with the economist? How they think about whether itis a process issue or substantive issue? And have you had toadjust how you present in those jurisdictions?

DENNIS CARLTON: I would say the economics are always thesame regardless of the venue. The relevant insight an econo-

S P R I N G 2 0 1 8 · 1 7

mist can bring to the table is his understanding of how var-ious industries work and how they are impacted by variousevents. Now, the antitrust law differs across jurisdictions. So, the

relevance of what you say will obviously differ, and yourattorneys can advise you as to what is the relevant topic ofconversation for an economist to address in one jurisdictionversus another. In terms of the underlying economics, that obviously stays

the same. But which economic analyses are the most relevantto a competition authority can vary from location to location.

GREG LEONARD: I can speak to China specifically as repre-sentative of jurisdictions other than the U.S. and EU. So Iwould say there are two things that are different that you haveto keep in mind. One is that China considers additional factors beyond con-

sumer welfare when evaluating mergers. Some of these addi-tional factors may not be economic considerations per se butnonetheless need to be kept in mind, such as national securi-ty concerns. Some others have economic components, albeitnot ones that we would normally think about here in theU.S. For instance, the impact of the merger on producer sur-plus or the national economy can be quite a concern in China. And then the other thing, I think, is just that the antitrust

law in China is relatively new. This actually goes back to oneof the earlier questions about practical issues and guidancewhen you’re dealing with other jurisdictions. MOFCOM, the agency there that handles mergers, is rel-

atively new compared to the agencies in this country and, ina lot of ways, they are still figuring out how they want to docertain things. So, you do have a different job going in there where you

may give a much fuller explanation regarding why you’redoing what you’re doing, other jurisdictions where it hasbeen used, and how it can help the agency assess effects of themerger. I would say it is a somewhat different approach––they

have very smart people who are very interested to what youhave to say but just with a different level of experience and asomewhat different set of factors that they need to consider.

JAMES KEYTE: Let’s move to some litigation topics. Daubertis very relevant to economic experts. Phil, in light of Comcastand Tyson Foods, some practitioners believe that lodging aDaubert challenge is almost necessary to avoid waiver claimslater. Whether challenging an opponent’s report or defend-ing your own, how do you and your counsel prepare for aDaubert hearing? Or, in your experience, what are the bestpractices there?

PHIL JOHNSON: If you are looking to avoid an adverseDaubert ruling, the key is taking the dispute and putting itinto language where the court will understand why youranalysis is valid, is well-supported economics, and is based

on the facts. Conversely, the court needs to understand thesupposed fundamental flaws in the expert analysis if a partyseeks to have it excluded, i.e., what the problems are with itsfactual and analytical basis—make that really focused andclear because the court won’t want to have a lot of back andforth on confusing issues; they want it as clear as possiblewhat is in dispute.

JAMES KEYTE: Are there other differences in working withexperts outside the United States?

MARIA MAHER: In Europe, we don’t have Daubertmotions.So this is probably one area where there is a big differencebetween the EU and the U.S. But courts will still need todecide how much weight to put on an expert’s evidence. Inthe UK, there is an independent duty to the court and that’staken very seriously by experts. If the court feels that the expert would not be sufficient-

ly independent, then an expert might be dismissed or not beallowed by the court. An example that recently happened isin one case where a party had been advising a firm under amarket investigation by the CMA for a couple of years. Whenthe appeal went to court, they were not allowed to use thatexpert and so they had to bring some other experts on board.So, that can happen. In terms of what tools or what methods we use, it’s up to

the experts to explain them and their purposes as well aspossible to the court. In litigation in the UK, economists areplaying an increasingly important role and a role that is lead-ing to innovation in the management of cases. In a recent high court case, for example, a high court Jus -

tice set out a process for managing expert evidence where youneed to have early mutual engagement of the experts. Theopposing sides’ experts meet and try to discuss areas wherethere is agreement on methodology and approach. Theexperts need to identify where there are areas of dispute andjust how material are those areas of dispute. It is importantto then also articulate the reasons why there might be dis-agreement and to decide how you are going to handle orresolve those disputes. I think that’s a really interesting approach in the UK,

along with the increasing use of “hot tubbing.”

DENNIS CARLTON: I’d like to add that I’ve been involved ina number of both international cases involving arbitration or“hot tubs” and some domestic arbitration. In such situa-tions, you can delineate to a panel of arbitrators or a judgewhat you and the opposing expert agree upon and what youdisagree upon. And that turns out to be a very useful exercisein a complicated litigation for a panel of judges. It is donevery rarely in the United States in a formal court. I havedone it in arbitrations. I have been involved in an arbitration in which you pro-

ceed in steps, in which the arbitrator says, “All right experts,let’s first decide on the database. Okay, now that I have


1 8 · A N T I T R U S T

decided on the common data base to use, tell me what econo-metric techniques you want to use to answer the relevant eco-nomic questions. Proceeding that way narrows the areas ofdisagreement and allows the arbitrator to really hone in onwhat are the critical differences between the experts. Such a procedure can prevent the experts from going too

far afield from each other in areas where there could be agree-ment if someone paid attention and forced them to reachagreement.

CARL SHAPIRO: I agree with what Dennis said. The hardpart I found about giving concurrent expert testimony wasactually producing that joint expert report, which is wherethe experts tell the judges, “Here is what we agree, and thenhere is where we part ways as experts, taking different per-spectives.” Writing the joint expert report can be painful, butreally important and very valuable to the court in narrowingthe issues and reaching issue joinder. I would be delighted if U.S. courts could find a way to

move in that direction, perhaps through more use of court-appointed experts or neutral experts to push the party expertsto identify areas of agreement because that is very useful forthe decision maker.

GREG LEONARD: To add one point about special masters: ithas always seemed to me when there are very difficult ques-tions, having a special master would be at least a step in theright direction. That said, I have been involved in a couple ofcases where the judge has ignored the recommendations ofthe special master, which I found kind of astonishing. In one case it was because, the special master said, “Eco -

nomics says you should do it this way.” And the judge said,“The law does not agree with you” or “it is not allowed—soI am going to disregard you.” In the other case, it was a pretty clear recommendation

from the special master and the judge just did not adopt it.And I thought the recommendation was well-reasoned andso forth. It wasn’t like it was an out-there recommendation,although it would have severely adversely impacted oneparty’s litigating position. Accordingly, it was not clear to mewhy the judge disregarded the recommendation. What is thepoint of seeking help only to ignore it?Judges can only take the help if they are willing to accept

it. That’s one problem with the special master situation, andI hadn’t fully appreciated that it could be a problem.

JAMES KEYTE: Let’s move to another area of litigation whereeconomics plays a role and talk about class certification.Dennis, it seems that econometric analysis has an increasinglyimportant role to play in class certification proceedings, espe-cially in antitrust matters. Has that been consistent withyour experience? And how has the role of economic analysisin class certification evolved over your career?

DENNIS CARLTON: Class certification is still a bit of a mud-

dle. My understanding is the requirements for class certifi-cation differ across the circuits to a large degree, althoughmaybe it is narrowing. But the fundamental problem I findis that there’s a bit of a mismatch between what an economistwould describe as the purpose of the class certification processand some of the legal issues. From an economic perspective, you want to have class

actions in order to economize on the transaction costs forindividual plaintiffs, who collectively might have an incentiveto do something, but individually wouldn’t. The goal of cer-tifying a class should be to deter bad behavior on the part ofa firm. If the firm is guilty, you want to be able to assess dam-ages to the firm that are a reasonable estimate of the totalharm to the economy. And that would achieve the econom-ic purpose of deterrence. (I ignore some complications tokeep the discussion focused on the main point.)From a legal point of view, it seems there is a second issue

that gets inserted. The legal issue is that it is not fair to paydamages to someone who either hasn’t been damaged or forwhom an average measure of damages—however you calcu-late it—is not very accurate. But an accurate measure of harmto each individual is not a requirement in order to have anaccurate measure of total harm and the latter is what is need-ed for optimal deterrence. And that second issue has been useda lot to show that it’s too hard to come up with an econo-metric model that is accurate for everybody in the class andtherefore the class cannot be certified. Sometimes when I readwhat judges think they were accomplishing by certifying ornot certifying the class, it is not obvious whether they havepaid much attention to the economic logic––to the econom-ic underpinnings which I think provide the logic of why thereare class actions. To address your question about econometrics, there is

more use of econometrics, whether it has helped or not Idon’t know. The point is that it would help a lot if judgespaid attention to the fundamental purpose of having a classaction in making the decision as to whether it’s appropriateto certify a class.

JAMES KEYTE: To follow up—part of the purpose of theclass action is to have a trial that can address all the issues asa group that would have been addressed for each individualplaintiff, had their cases proceeded. Phil, what in your vieware the current issues in class certification proceedings andwhere do you think they are headed?

PHIL JOHNSON: One issue is what is needed to satisfy thetypicality and predominance requirements, particularly inindirect purchaser cases where plaintiffs need to establishthat a component with an elevated price upstream results inimpact on consumers downstream. There are many issuesthat get raised and the analyses needed to address them all canbe daunting. Arguments on one side can seem to call for every corner

of the entire chain of distribution from top to bottom and all

S P R I N G 2 0 1 8 · 1 9

possible product varieties to be addressed with their ownempirical analysis. Then there are also the arguments that theanalyses aren’t conclusive because of imperfections in thedata, such as sample representativeness and measurementerror. On the other side, there are good arguments that eco-nomic theory and a wealth of empirical analyses in variousindustries show that impact does generally make its waydownstream. Another issue is ascertainability—the task of showing the

class members can be identified. It sounds simple but whenit involves determining whether or not a person has a specificcomponent in their product, it may not be. Ascertainabilityadds another complication at the class cert phase—and itseems to me has been some confusion about just what needsto be hown. Obtaining class certification is already very challenging

and it seems like the bar to clear class certification might begetting higher. As Dennis mentioned earlier, the height of thebar seems to vary across the circuits. Perhaps we will finallybe getting uniformity, but I’m not counting on it.

JAMES KEYTE: Then how do you work with lawyers in thelitigation context with project design where you might havelaw that varies in different circuits or jurisdictions and youdon’t want to design a project that makes sense economical-ly but may not be the one that is actually consistent with thestandards that happened to be controlling that case, whetherit is in class certification, monopolization, or dominance?

CARL SHAPIRO: Often economists like to say, “Oh, we needto teach the lawyers more economics, so they can under-stand what we’re doing and how great it is.” And I confess tohaving those thoughts sometimes. But I think the reverseproposition is also important, namely, that the economistsneed to understand the legal environment in which they areworking. Economists need to understand the applicable legal stan-

dard. They need to know what the jury instructions willlikely be. Or for class certification, they need to understandwhat will be the standards for certifying a given proposedclass or not. Many economists do not take the time to learnthese elements of the law. They just want to do the econom-ics. I think that can be unwise.

JAMES KEYTE: Just as an example, let’s say you are dealingwith whether it is a tie-in or some other alleged vertical,alleged monopoly conduct. In some circuits you might beable to argue about the “one monopoly rent.” In other cir-cuits, it has already been rejected. Do you find that you aregetting that kind of guidance on any types of cases? Or is thatsomething really that the lawyers need to do better?

DENNIS CARLTON: I think it’s a balance. I think the lawyersobviously have to guide you, especially if different circuitshave very different criteria. You don’t want to spend your

time doing a project that is legally irrelevant. On the other hand, you can’t lose sight of the underlying

economics. In other words, if a lawyer tells you the law isinconsistent with economic reasoning then there is not muchyou can do other than to say, “Well, I can’t help you on thatpart of the law except to explain why (a) it makes no eco-nomic sense and (b) because it makes no economic sense, youmight well be misinterpreting that law.” And, hopefully, youcan convince the judge of an interpretation that is consistentwith common sense and the economics. But I think you dohave to rely on guidance from the lawyers because, as Carlsaid, economists have to be sensitive to what the legal stan-dards are.

MARIA MAHER: In the EU both economists and lawyersproviding them with guidance in class certification cases facethe challenge that all of this is still really new in the EU. Itwas only two years ago that the European Commission pro-posed a recommendation on class certification. The Commis -sion’s proposal recommended the member countries go foran opt-in regime. In 2015, the Consumer Rights Act waspassed, which allows the UK to have both an opt-out and anopt-in regime. Since all of this is so new in Europe, lawyers are also look-

ing to the economists to help provide them with some guid-ance as to how you go about meeting the hurdle for class cer-tification. In the UK that includes the need to identify theclass and for the class to raise common issues.In the two class certification cases in the UK, one was a

MasterCard consumer class action alleging damage frominterchange fees and the other was a case following-on froman OFT decision that a seller of mobility scooters hadengaged in resale price maintenance with a number of retail-ers. Both of those first two cases failed to be certified, but ithas clarified some aspects of the legal hurdle. The short sum-mary is that economists will need to do more in future cases.

JEFF JAECKEL: Maria, we’re going to stay with you to talka little bit more about Europe. The U.S. courts have a longhistory of relying on economic tools to estimate damages incivil antitrust cases, often including follow-on cases to cartelinvestigations. How do European member state courts handle the quan-

tification of damages and private claims? And do you seedifferences in the way the parties use the economists, giventhe differing development of those legal regimes?

MARIA MAHER: There is more coherence of approach devel-oping in the EU with the passage of the directive on damagesactions. Member states had to transpose the directive intotheir own national laws by the end of 2016. The EuropeanCommission has also published a practical guide in quantify-ing damages in relation to breaches of EU competition law. One area where there are some distinctions with the U.S.

is the need to show liability in follow-on damages cases. In


2 0 · A N T I T R U S T

damages cases in the U.S. I believe claimants still need toshow liability, whereas in the EU when it’s a damages casethat follows on from a European Commission finding of aninfringement, the claimant doesn’t need to show the liabili-ty, it is presumed. That means the economics will focusdirectly to the quantification of the damages. Also, the practical guide and the damages directive make

clear that there is scope for a pass-on to impact the quantumof damages actually suffered (albeit the burden is on thedefendant to establish pass-on). In addition, indirect pur-chases may claim damages and so there’s scope for bothdirect and indirect purchaser claims here.

JAMES KEYTE: Let’s talk a little bit about intellectual prop-erty. Standard essential patents (SEPs) continue to be a hottopic, both in the U.S. and outside the U.S. Carl, as an econ-omist, how do you approach the subject of standard essentialpatents and the interaction of intellectual property, antitrust,and bargaining? It is a lot to assess in one basket. And howdo you do that as an economist?

CARL SHAPIRO: Well, I’m glad we have two hours for myanswer. The starting point is that many standard-settingorganizations have rules regarding the licensing of standardessential patents on fair, reasonable, and non-discriminatoryFRAND terms. These FRAND rules represent a private,contractual arrangement setup by the industry participants todeal with the serious, and I would say obvious, risk of patentholdup that would otherwise arise and cripple the wholestandard-setting process.There are a whole set of issues about how these FRAND

rules are established and enforced. Those are not particular-ly antitrust issues. They are contractual and governance issues.However, when a patent holder breaches its FRAND com-mitment and charges above FRAND royalty rates, that caneasily have market-wide impacts that injure final consumers.Furthermore, the companies directly paying the above-FRAND rates may well pass-through those rates in the formof higher prices, injuring final consumers. These breakdownsor failures of the FRAND system can thus raise seriousantitrust issues. The notion that patent holdup is a fictional or minor

issue does not make any sense in terms of the economics. Thereal question is, “How do we deal with it?” I expect we willsee an ongoing debate about the circumstances under whichwe need antitrust enforcement to supplement private solu-tions especially—private actions involving allegations that apatent holder has breached its FRAND commitment. Patent law itself has evolved a lot here in terms of the

treatment of reasonable royalties. It has become much moredifficult for patent holders to obtain injunctions, particular-ly if they have committed to license their patents on reason-able terms. The evolution of patent law has been quite positive over

the past ten years, and that somewhat reduces the need for

antitrust. But these changes in patent law do not eliminatethe need for antitrust remedies as well.

JAMES KEYTE: Any other thoughts from others on SEPs,whether it is in the U.S. or outside the U.S.?

DENNIS CARLTON: I would add just that I’ve been in a num-ber of these cases and written a bit about it. I do think it canbe a serious problem that can harm consumers. Whetheryou want to call it an antitrust problem or contractual prob-lem—that really has to do with legal issues, probably abouttreble damages. But I’ve always wondered why there aren’tmore requirements on an SSO to have mechanisms in place,arbitration mechanisms, to resolve these issues. As Carl indicated, an SSO could say, if you have made a

FRAND commitment that means if there’s a dispute youdon’t have the right to withdraw, you have the right to go toarbitration—let’s say baseball-style arbitration in which anarbitrator is going to determine the rate. But you’re notallowed to ask for an injunction. And the burden should beon the SSO to have that mechanism. I’ve never understoodwhy that wasn’t the approach followed and why an SSOthat had failed to do that wouldn’t be liable for an antitrustviolation brought by someone who claims that the royaltydemanded is too high.

JEFF JAECKEL: Shifting gears again now to issues aroundunilateral conduct potentially outside the SEP context. Greg,it seems that one of the eternally difficult questions in U.S.antitrust law is how to identify unilateral conduct that cross-es the line from aggressive but legitimate competition into theanticompetitive, predatory or exclusionary conduct. Howcan the economist help counsel, the courts, and the enforce-ment agencies draw the line?

GREG LEONARD: That is one of the more difficult areas inantitrust because, except in the most obvious situations, thereare possible procompetitive effects and possible anticompet-itive effects, and to get it right requires figuring out whatthose effects are and balancing them against each other. I’d like to go back to SEPs for just for a minute. I think

we are going to see a lot more issues come up about licens-ing of portfolios of SEPs because what one person thinks ofas an essential patent, somebody else may think is not essen-tial. And so, you have these big portfolios which are calledessential, but some of the patents in the portfolio may not be. Obviously, there are important transaction cost savings by

licensing everything at once. And so, in a case where a patentowner is alleged to have tied non-essential patents to trueSEPs through portfolio licensing, there is a lot to figure out.Is the patent owner just trying to save on transaction costs?Or is the patent owner trying, for example, to evade a RANDobligation by tying non-essential patents to essential patents?Figuring all that out can be very difficult. But that’s true

in almost all exclusionary conduct cases, including cases

S P R I N G 2 0 1 8 · 2 1

where the question is whether loyalty discounts are in effectbeing used as an exclusionary device. A number of tests havebeen developed to assess conditional pricing. Those efforts areall very valuable but we are still in the middle of figuring outwhich one of those tests, if any, is useful in a broad sense andwhether different tests are useful for different contexts. So Ithink it’s an area that is wide open and will certainly be hotlydebated in the coming years.

PHIL JOHNSON: One thing I find notable is that courts out-side of the merger context—in antitrust /IP cases in partic-ular—still seem, to a large degree, married to the idea that itis necessary to have a formal relevant market analysis inalmost all circumstances. In setting where you are not look-ing at the prospective effects of a merger, but instead arelooking retrospectively, e.g., at anticompetitive use of patent,it seems reasonable and much more direct to look at theactual effects of the conduct. I have seen a fair bit of confusion about the necessary role

of relevant market definition in such cases. More clarity isneeded about circumstances where an effects-based approachis sufficient

MARIA MAHER: Again, this might be an area where there issome difference between the EU and the U.S. legal systems.In the EU, you have the concept of abuse of dominance. Sowhen you are looking at exclusionary conduct it is underArticle 102 of the Treaty of the Functioning of the EuropeanUnion which prohibits the abuse of a dominant position. The way you would normally start is to first determine

whether or not the firm is dominant. In order to do that youstart off with market definition––by looking at a relevantmarket and then determining whether the firm that isengaged in the allegedly exclusive behavior is dominant inthat relevant market.But once that step is done—and I personally think it is

absolutely right that you need to look at the effects of theconduct—the courts will consider whether a given conductis abusive in itself or requires an effects based decision. Thatquestion was most recently considered in the recent Inteldecision of the European Court of Justice where the ques-tion was whether certain rebates were abusive in nature.And what was good—at least good from the economistpoint of view—is the European Court of Justice sent theIntel decision back to the General Court to look more atwhether the rebates were actually foreclosing competition onthe market.

DENNIS CARLTON: If I could just add one fundamentalpoint that follows from that, an effects-based approach criesout for what economists call a reduced form analysis.Sometimes the proof is in the pudding, and an effects-basedapproach—which I often analogize to what economists dowhen they perform a reduced form analysis—can be veryinformative. And I think it’s dismissed too often.

A reduced form is looking at the effect of an action, say,exclusive dealing on prices. For example, looking at situationsin which you’ve had historical episodes of, say, exclusive deal-ing—and comparing those situations with situations whenthere is no exclusive dealing to see whether exclusive dealingraises price. (There are some caveats to pay attention to inorder to get the analysis correct.)That type of reduced form analysis is distinct from what

is called structural analysis, which for example, involves adetailed modeling of the competitive interactions amongfirms based on assumptions of how the firms compete andhow consumers respond to various types of incentives. Onecan use a structural analysis to predict what would happen toprices from say exclusive dealing. Although structural analy-sis can be useful, a reduced form analysis can be simpler andmore direct. Looking at historical events in which you have had actu-

al experience with the conduct in question always seems tome to be an important element of an empirical analysis. A lotof the trends have been to move away from that and I thinkthat’s a mistake.

JAMES KEYTE: And, Dennis, if I have it right, the challengein the reduced form is specificity and causation? Whetheryou are looking at a natural experiment or some other regres-sion-type analysis, correct?

DENNIS CARLTON: Yes, that’s correct.

JAMES KEYTE: Maria, I have a follow-up question for you,also highlighting what some practitioners see as the differencebetween the EU and the U.S. when it comes to abuse ofdominance or dealing with firms with high shares or amonopoly position. Many economists in the U.S. embracecreative disruption even if it is competition for the field andresults in a dominant position. Some would say the EU is more interested in a level-

playing field. The idea is that consumers are better off withmany competitors and that is perhaps what has led to dif-ferent treatments in the EU and the U.S. of dominant firmsin assessing or scrutinizing their behavior. Do you see thatas a distinction, or how do you account as an economist for the different treatment of some of the dominant firms,whether e-commerce or a platform, between the EU and theU.S.?

MARIA MAHER: You often do hear those differences in thedebate. And creative destruction is clearly an important forcein European economies just as in the U.S. As an economist,my approach to these cases always boils down to an effects-based approach. As I said, I think that things are changing inEurope. The ECJ’s decision in Intel recently in the last yearhas moved the dial towards a more effects-based approach. Soeven if a firm is dominant, it is not necessarily enough to sayit has engaged in a given conduct.


2 2 · A N T I T R U S T

JAMES KEYTE: Populism and nationalism are having amoment in the national discourse, including in antitrust,for some at least. Carl, antitrust and populism, what are yourthoughts?

CARL SHAPIRO:Well, I’ve written a paper called “Antitrustin a Time of Populism.” This paper is on my website and isforthcoming in the International Journal of Industrial Organi -zation. I will say a couple of things that are in that paper, keythemes. First, a lot of what you see in the press about the supposed

loss of competition in the United States economy is not wellsupported by the evidence, particularly the numbers that arethrown around about market concentration. No doubt there are some markets where we have had

some loss of competition, and I would hope that the antitrustagencies will learn from those cases, but there is no evidenceto indicate that it is a widespread phenomenon. Second, we are now seeing a challenge to the 40 or 50

year old consensus regarding the use of the consumer welfarestandard in the antitrust. So far as I can determine, that chal-lenge is not well thought through. Those calling for theabandonment of the consumer welfare standard have offeredno practical alternative, much less one that would be supe-rior for the purpose of promoting competition. If and whenthe advocates who are pushing for a dramatic rethinking ofantitrust offer a workable alternative, we can compare thatto current practice based on the goal of promoting compe-tition.I addressed these issues in my paper and also in the testi-

mony that I gave in front of the Senate Judiciary Committeein December 2017. But I would advise all antitrust practi-tioners to pay close attention to the questions being raised inthe press and by politicians about whether antitrust enforce-ment has gone astray over the past 50 years, even if youstrongly disagree with that view. These questions reflect adeep and widespread frustration with the perceived power oflarge companies in the United States and with perceivedinequities in the American economy. My own view is that the underlying causes behind these

questions can be found in two areas: (1) the excessive polit-ical power of large corporations, that (2) the big increase ininequality in income and wealth in the United States inrecent decades. I very much share both of those concerns, andantitrust can indeed help somewhat on both fronts. But Istrongly believe that antitrust should continue to be aboutpromoting competition. We need other policies, such as cam-paign finance reform or tax policies that would address theunderlying concerns more directly.

DENNIS CARLTON: I’d distinguish between the political/non-academic approach and an academic approach. As Carlsaid, a lot of the thrust for changing antitrust is coming fromthe political arena. I will leave that aside. I agree with Carl,antitrust isn’t the way to attack many of those problems.

But there are some growing challenges from the academ-ic literature. The Kwoka book raises questions. It’s been crit-icized, but the Kwoka book raised questions about past merg-er policy. One of the criticisms he raised was particularlyfocused on the airline mergers. I have a recent paper show-ing that the airline mergers have been pretty good. But he’sa reputable scholar and I think people should take his workseriously and address it.The most recent challenge has been by DeLoecker and

Eeckhout. They claim to show that there has been an increasein market power since 1990. That’s the best paper I’ve seenon the topic, though one of my students, James Traina, haschallenged the findings. And I agree with Carl that a lot ofthe other studies that have come out, I think are just not con-vincing. So from the point of view of academic research I think

there is some dispute as to whether market power is increas-ing. But from the point of view of antitrust policy, the ques-tion you have to ask is, to the extent that there is an increasein market power, is that increase arising because of laxantitrust policy or is it because of technological change orother factors. And I suspect that a lot has to do with tech-nology to the extent that there’s been any increase in a gapbetween price and marginal cost.

CARL SHAPIRO: Let me add to that. I think it’s a reallyimportant point that that Dennis just made. There is grow-ing evidence that there have been some increases in price-costmargins and profits across many industries. That empiricalfinding is still being tested, but if we accept that finding, weget to the deeper questions: (1) what caused that change, and(2) what, if anything, should we do about it? There is a per-fectly valid explanation for the empirical finding: fixed costshave gone up, perhaps in the form of R&D or investment ininformation technology. With larger fixed costs, economiesof scale are greater, and we expect competition to cause firmsto become larger because larger firms are more efficient thansmaller firms. Over time, these larger firms will need to beable to get sufficient price/cost margins to finance the R&Dand other fixed costs that are associated with efficiency.The really big question is how much growing economies

of scale explain the observed increases in concentration andthe observed increases in price/cost margins. An alternativeexplanation is that firms are earning excess profits (“rents”)based on growing barriers to entry and/or exclusionary con-duct. The empirical literature that has found these increasedmargins has not gotten to that next question yet. That isentirely understandable. They are documenting changes inmargins and profits. But that is where the economic literatureneeds to go. Identifying these underlying causes is very hard,but I hope the field of industrial organization economicswill get there.

JAMES KEYTE: Carl, just so I understand, so they’re not yetdocumenting whether that correlates with changes in

S P R I N G 2 0 1 8 · 2 3

Herfindahls or market structure that would otherwise havebeen indicative of the concern.

CARL SHAPIRO: Looking across industries, you wouldexpect some correlation between market concentration andprice/cost margins, and we see that. However, observing thatcorrelation does not tell us whether the underlying cause isfirms exploiting economies of scale, which is a natural andhealthy part of the competitive process, or exclusionary con-duct, which is not. Put differently, we do not know that thehigher price/cost margins observed in some industries reflectconsumer harm associated with diminished competition.They may reflect the lower variable costs that are associatedwith economies of scale, not higher prices caused by exclu-sionary conduct. Either way, however, these higher price/costmargins tend to point toward a more assertive approach tohorizontal merger enforcement.

PHIL JOHNSON: I agree with what’s been said about how evi-dence of generally increasing market power doesn’t seem allthat persuasive yet. I think one thing besides the technological and other

issues that were mentioned leading to some increase in con-centration in the economy is to look at regulatory frameworksthat support that increased concentration and whether that,rather than antitrust, might be the place to look to reform,to make sure that competition is healthy––IT protections andprofessional licensing and other things that create those bar-riers to entry in a lot of industries.

GREG LEONARD: Yes. And I was going to add that. I thinkthat’s exactly a place where perhaps populism could be har-nessed in a way that we might all agree with. Whether theFTC for instance has the power to do anything about it Idon’t know, but professional licensing regulations and cronycapitalism are things that the populists are upset about, andI think we would agree they can be bad for competition aswell.

JEFF JAECKEL: Thank you for your participation today.�

� IN RECENT YEARS, technology markets have been a major focus of antitrust policy, agency enforcement, and private litigation. Courts, agencies, commentators, and practitioners have grappled with whether technologyindustries are somehow different when it comes toantitrust. Antitrust analysis of technology markets has continued to evolve rapidly—it is arguably one of the least settled analytical areas in antitrust law. The Handbook on Antitrust in Technology Industries is

intended to serve as a comprehensive review and analysisof the application of antitrust law and principles to thetechnology industry. It will serve as an important and useful tool for practitioners who seek an introduction to antitrust, as well as antitrust practitioners who findthemselves coming to grips with the complex field of technology. This book discusses these topics: the fundamental law

and economics of technology industries that make analysisof antitrust issues in these sectors both unique and challenging; the substantive antitrust law applicable totechnology industries, including horizontal and verticalrestraints; the enforcement of antitrust principles in technology markets in the contexts of mergers and acquisitions; and single-firm conduct.



Handbook on Antitrust inTechnology Industries

Product Code: 5030646Publication Date: 2017 Page Count: 418Trim Size: 6 x 9Format: Paperback

Pricing:$249.00 List Price / $209.00 ABA Member / $179.00 AT Section Member


2 4 · A N T I T R U S T

Unpacking the Economic Toolbox: How to Make Sense of Your Economic Expert’s AnalysisB Y S H E N G L I , C H R I S T I N E M E Y E R , A N D G A B R I E L L A M O N A H O V A

tionships between two or more variables. For example,demand regressions can estimate consumers’ propensity tosubstitute between products in response to price changes(referred to as “cross-price elasticity of demand”), and theseelasticities can be used as inputs for determining market def-inition and market power. In cartel and collusion cases,benchmark and before/after-type regressions are often used toestimate alleged overcharges. Given the technical nature and variety of regressions used

in litigation, some attorneys may lack the background todiscuss them in depth with an economist. Below, we providea framework for guiding the conversation between the attor-ney and the economist toward the central issues involved inevaluating a regression analysis. � Establishing the causal link. Basic regressions show cor-relation between variables, but correlation does not implycausation. When working with your economist to estab-lish antitrust damages, it is important for attorneys todetermine the degree to which the evidence shows a causallink between the damages being estimated and the con-duct at issue.

� Defending the model. The government or opposing pri-vate litigant frequently argues that regression models are“incomplete” or “misspecified.” There is no single besteconomic model for all cases. Instead, one must employ theright economic model for each situation, and the modelmust be adapted to take into account relevant market fac-tors. To best prepare for these “battles of the experts,” it isoften instructive to discuss with your economist why theeconomist used the model for the market setting at hand,the key assumptions underlying the model, and the evi-dence and analyses that support the assumptions.

� Working with data limitations. In a perfect world, datawould be error-free and economists would have all thedata they need to estimate the ideal model for each case.In reality, data can be spotty, limited in scope, or prone toerrors. Data limitations impose real restrictions on resultsthat can be derived from econometric analysis. It is impor-tant that the economist explain to the attorneys and theirclients any limitations in the data, the implications of

ATTORNEYS OFTEN THINK THATconversing with an antitrust economist is akinto hearing someone speak a foreign language.While the economist is eagerly describing themodel specification, the attorney is thinking

that it all sounds like Greek to her. But this does not have tobe the case. The tools in the economist’s toolbox are merelystatistical and mathematical embodiments of the antitrustprinciples that are very familiar to the legal practitioner. Moreimportantly, a communication gap between the antitrustattorney and economist is a serious problem that can reducethe level of service that they provide to the client.In this article, we explain four frequently used economic

tools and how economists apply them to common antitrustissues. We first discuss regression analysis as applied to com-mon antitrust issues. We then explain how critical loss analy-sis (CLA), the upward pricing pressure (UPP) model, andmerger simulation are applied in the review of mergers andacquisitions. For each economic tool, we provide a practicallist of the strengths and limitations of these techniques andthe key issues that attorneys and economists need to discussso that they and their clients are fully prepared for how theantitrust agencies or an opposing private litigant might cri-tique the economists’ work.

Economics Tools in Antitrust Litigation:Regression Analysis Regression Analysis in Plain English. Regression analysisprovides a powerful tool for examining correlations betweeninput and output variables. Economists use different types ofregressions to provide support for antitrust damages analysisas well as other inquiries where data can shed light on rela-

Sheng Li is a Senior Consultant at NERA Economic Consulting; Christine

Siegwarth Meyer is a Managing Director at NERA Economic Consulting

and an Associate Editor of ANT ITRUST ; and Gabriella Monahova is a Senior

Consultant at NERA Economic Consulting. The opinions expressed in this

article are those of the authors and do not necessarily reflect the views

of the firm or its clients.

S P R I N G 2 0 1 8 · 2 5

such limitations, and how the economist has addressedthese limitations. Establishing the Causal Link. Isolating the economic

impact of the alleged harmful act is essential to any antitrustdamages analysis. Economic damages reflect how the plain-tiff’s economic position would have differed, if at all, in theabsence of the anticompetitive conduct. As such, economicdamages must be causally linked to the defendant’s conductat issue.As noted above, while regression analysis is often used to

quantify economic damages, the basic regression frameworkdoes not address the issue of causality, which is central to eco-nomic damages. Basic regressions can only measure correla-tion between the outcome (e.g., quantity sold) and the inputvariable (e.g., price). That is, while basic regressions canquantify correlation between designated “input” and “out-put” variables, correlation—however strong and statisticallysignificant—does not imply causation. One way that non-causal correlations can arise is in situ-

ations where both the “input” and “output” variables aredriven by a third factor, leading to what economists call spu-rious correlation. For example, before/after pricing models ofantitrust damages may yield skewed results due to spuriouscorrelation if the models are not properly specified to accountfor market trends. In before/after models, economic damagesare assessed by comparing prices in the impact period toavailable prices before and/or after the alleged period ofantitrust impact (i.e., the “clean” period). However, marketfactors unrelated to the conduct at issue, such as shifts in con-sumer tastes, may change at the same time as the allegedanticompetitive conduct. As an illustration, retail prices for DVD players have

declined steadily since the early 2000s, driven by shifts inconsumer tastes towards Blu-ray and online video streamingtechnologies.1 Therefore, if an economist used a before/aftermodel to estimate antitrust price impacts during this period,then the economist must specify the model so that it can sep-arate the effects of the alleged conduct at issue from those ofunrelated market factors or combine the results of the modelwith additional information from documents or testimony toestablish causation. Otherwise the estimates from the modelwould be skewed by spurious correlation, as they would mis-attribute price changes caused by unrelated shifts in con-sumer tastes to the alleged conduct at issue.Comcast v. Behrend provides an illustration of when sta-

tistical correlation is not sufficient to demonstrate causa-tion.2 In Comcast, plaintiffs alleged four theories of antitrustimpact attributed to defendant’s actions, and the plaintiffs’economic expert estimated an econometric damages modelby assuming that all four theories were valid. However, theSupreme Court accepted only one of the plaintiffs’ theoriesof antitrust impact, and the plaintiffs’ expert’s econometricmodel did not link the estimated damages to any one par-ticular theory of impact.3 This lack of explicit causal linkbetween the theory of harm and the estimated model proved

to be instrumental, as the Court ruled that “[t]here is noquestion that the model failed to measure damages resultingfrom the particular antitrust injury on which [Comcast’s] lia-bility in this action is premised” citing “the model’s inabili-ty to bridge the differences between supra-competitive pricesin general and supra-competitive prices attributable to” theparticular theory of anticompetitive impact at issue.4

The takeaway from the above examples is that regardlessof what statistical estimation methods the economist uses, theexpert must explain how the results of those estimations arecausally linked to the conduct at issue, and are not driven byspurious correlations with unrelated market factors. If theeconomic expert’s explanation of the causal link appearscounterintuitive or opaque, attorneys should ask the econo-mist to explain what evidence supports the link. The causallink must be grounded in evidence in the record and consis-tent with economic theory. Well-specified causal links shouldbe intuitive to a non-technical audience.There are a variety of methods that can establish causa-

tion, including identifying a causal mechanism based oncontemporaneous documents, and using instrumental vari-ables, which is a method that estimates causal effects whenthere is a complicated interdependence between the inputand output variables of interest. The details of instrumentalvariables regression analysis are beyond the scope of this arti-cle, but it is important to note that for a regression model towithstand scrutiny, the attorneys and economists must ensurethat the asserted causal links are grounded in evidence in therecord and consistent with economic theory.5

Defending the Model. The objective of economic dam-ages analysis is to determine how a plaintiff’s economic situ-ations would have differed in the hypothetical but-for worldwhere the defendant did not engage in anticompetitive con-duct. When opposing economics experts spar during litiga-tion, the dispute is often centered on whether the econo-metric model used is appropriate for the problem at hand andwhether the model properly accounts for market conditions.Predicted but-for outcomes should isolate the effects on theplaintiff of the alleged conduct from the effects of all otherevents that are not related to the actions at issue. To best pre-pare for these “battles of the experts,” it is important to dis-cuss with your economist the list of inputs used in the regres-sion and to think about whether this list accounts for allrelevant market factors.The regression input variables that are beyond the key

input of interest are referred to as control variables. Regres sionmodels need a full set of control variables that account formarket factors in order to produce reliable predictions. Forexample, in Freeland v. AT&T Corp, the district court reject-ed the plaintiffs’ expert’s regression analysis, which was intend-ed to show the anticompetitive impact of tying and locking ofcellphone prices, because the analysis failed to account for theintroduction of the smartphone and the industry’s shift fromanalog to digital technology, both major market develop-ments that could have significantly affected prices during the


2 6 · A N T I T R U S T

relevant period.6 Similarly, in Blue Dane Simmental Corp. v.American Simmental Association, the Court of Appeals for theEighth Circuit rejected an economic expert’s analysis becausethe expert had neglected to consider any input variables otherthan the alleged cause of injury.7 As these examples illustrate,failing to account for major market factors results in unreli-able estimates, and may also lead to the exclusion of theexpert’s testimony by the court.There is no fixed list of universal control variables to

account for relevant market factors; the appropriate set ofcontrols depends on the facts in each case. However, a use-ful framework for thinking about the completeness of thecontrol variables is to consider all of the factors that mayaffect demand and supply in the market(s) at issue. For exam-ple, to predict the prices of shipping services, the regressionwould need to include supply-side inputs that affect the costof providing shipping services, including the weight and sizeof each shipment, distance shipped, shipping speed, and fuelprices. Demand-side inputs for predicting shipping pricesmay include indicators for seasonal demand (e.g., holidayshopping season), indices of global trade volumes, interna-tional exchange rates, and changes in trade regulations (e.g.,renegotiations of NAFTA or Brexit). One way to get a sense of the completeness of the regres-

sion model is to ask your economist to list the input factorsused in the model and why each is included. If there are fac-tors that could potentially affect the supply or demand in therelevant market, but are not included in the model, ask whythey are not included. Another way to assess the completenessof the regression model’s list of control variables (and antic-ipate how the model may be challenged) is to check whetherthat list is consistent with what is found in the academic lit-erature on the same topic.For certain input factors, there are often multiple ways to

account for or measure that factor. For example, economistscan account for seasonal shopping trends by using month-of-year or quarter-of-year variables. Generally, reliable regres-sion predictions should not be sensitive to minor changes ininput variables, and should yield similar results with similarvariables (such as, for example, whether the econ omist usesmonth-of-year or quarter-of-year variables). Testing a regres-sion model prediction’s sensitivity to such minor changes in inputs is a good method for assessing the model’s relia-bility.In addition to the proper accounting of market factors,

another common point of contention in “battles of theexperts” is whether the economist is using the right type ofregression model in the first place. Different econometricmodels rely on different assumptions and may be suited fordifferent data structures. For example, in assessing antitrustdamages, two common regression models are before/aftermodels and benchmark models. As discussed above, inbefore/after models, the idea is to compare prices in a “clean”period unaffected by the conduct at issue, against prices in aperiod that are “affected” by the alleged conduct.

Key to the reliability of before/after models is correctlyspecifying the “clean” and “affected” periods to match thetiming of the allegations at issue, and to ensure that the con-trol variables fully account for differences in market condi-tions between the alleged impact period and the benchmarkperiod that are unrelated to conduct. The correct specifica-tion of the “clean” and “affected” periods would also likely bea point of contention between opposing economists in casesusing before/after models. A properly specified “clean” and“affected” period is firmly grounded in the facts in the record.As counsel typically has better knowledge of the record thanthe economics expert, this is an essential topic of conversationwith your economist. Benchmark models compare prices in “clean” benchmark

sales to prices “affected” by the conduct at issue. Similar tothe before/after model, the reliability of benchmark modelsdepends on correctly specifying the “clean” and “affected”sales to match the allegations at issue and ensuring that the control variables fully account for differences betweenaffected sales and benchmark sales that are unrelated to con-duct. Economists often debate the correctness of the chosenbenchmark. The choice of a particular benchmark is anassumption that requires support with additional analysis.When economics experts disagree on the choice of bench-marks, which benchmark prevails would depend on howstrongly each benchmark is supported by market facts andeconomic theory.

Working with Data Limitations. Even a perfect recipewill not taste good if the ingredients are spoiled or of poorquality. Similarly, having a perfectly specified regressionmodel will not produce reliable results if the data inputs arebad. Predictions from regression models are only as reliableas their data inputs. When working with economics experts,it is useful to understand any limitations that data availabil-ity imposes on their analysis, and also discuss the bestapproaches for managing those limitations. Regression analysis is a tool that relies on both the quali-

ty and the quantity of the data. If there are fewer data points,then the regression estimates suffer from reduced precision.If there is additional noise in the underlying data, then it ismore difficult to isolate the effects of interest from the back-ground noise. The negative effects of data noise can general-ly be mitigated using larger sample sizes, but this is not alwaysan option. However, the fact that data limitations may pre-clude analysis using the ideal theoretical economic modeldoes not mean that no economic data analysis should beconducted. To the contrary, econometric models can often beadapted to work with limited data. Below are some approaches that can be used to adapt

models to work within data limitations. � Estimate a simpler model. “Don’t let the perfect be theenemy of the good,” as the saying goes. While complexmodels can provide better insights in theory, complexmodels also come with more extensive data requirements.The complexity of the economic model scales in propor-

S P R I N G 2 0 1 8 · 2 7

tion to the richness of the available data. For example, inmarket definition analysis, the complexity of econometricsemployed can range from intricate structural models ofmarket supply and demand to simpler regressions thatmeasure cross-price elasticities, to price correlation analy-sis between product pairs.

� Use next-best alternatives for data inputs. In some cases,when certain input variables are missing, coarser alterna-tives may be available. For example, household income isoften an important variable for estimating prices anddemand for products. However, if household income fromindividual customers is not available in the data, it may bepossible to generate a “best estimate” of household incomebased on the zip code of the customer’s address or wherethe purchase took place (if those variables are available).

� Augment the model using reasonable assumptions. Incertain situations, reasonable assumptions may be used tofill in gaps in the data or simplify models. For example, ifdetailed product cost information is not available, esti-mating a pricing model assuming constant marginal costmay be a reasonable alternative if documents and testi-mony in the record firmly support that assumption.It is important to note that in cases where the economic

models are adapted to data limitations, the predictions ofthose models are necessarily weakened. This is an unavoidableconsequence of working with restricted data. When talkingto economists about how to manage data limitations, attor-neys should ask about the potential weaknesses of data lim-itations, as well as whether there are ways to address thoseweaknesses using complementary qualitative analyses. Forinstance, if data limitations restrict the usefulness of regres-sion analysis for identifying product substitution patterns,qualitative analysis such as consumer surveys can sometimesprovide additional economic evidence.

Economics Toolbox in Mergers & AcquisitionsEconomists who perform merger analysis also often have tochoose between tools that are relatively straightforward toimplement and describe and those that are more rigorous.Below we discuss two concepts—critical loss analysis (CLA)and upward pricing pressure (UPP) —that belong to the for-mer category of tools, and another method—merger simula-tion—that belongs to the latter.8 CLA is used primarily formarket definition and for analysis of unilateral effects. Ana -logous in nature, UPP is a tool for evaluating the incentive offirms to unilaterally raise prices post-transaction. Merger sim-ulation is similarly applied to predict unilateral competitiveeffects, and it can potentially speak to the magnitude of theproposed transaction’s impact on various economic variables.

Critical Loss Analysis (CLA).CLA in Plain English. CLA is used extensively to inform

the definition of the relevant product or geographic market.9

The main market definition tool used by the Federal TradeCommission and the Department of Justice, and one that hasbeen increasingly adopted by the courts,10 is the Hypothetical

Monopolist Test, which takes a narrowly defined set of prod-ucts and tests whether a hypothetical monopolist would havean incentive to impose “at least a small but significant andnon-transitory increase in price (SSNIP) on at least one prod-uct in the market.”11 The monopolist would find it unprof-itable to impose a SSNIP if there are sufficiently close sub-stitutes outside of the proposed market that consumers wouldturn to in the event of a price increase. Thus, if the analysisshows that it would not be profitable for the monopolist toimpose a SSNIP, the proposed market is expanded to includethose substitutes, and the test is repeated.CLA is used to measure the incentive of the hypothetical

monopolist to impose a SSNIP. If the monopolist increasesits price, it faces a tradeoff: it would make more money oneach retained sale, but it would also lose sales because someconsumers would decide to purchase products outside of themarket or not purchase at all at the new price. Critical loss isthe drop in quantity as a result of the price increase thatmakes the monopolist indifferent between raising its priceand keeping it the same. In the most commonly used version,the critical loss is calculated to equal the value of the SSNIP(often taken to be 5 percent)12 divided by the sum of theprice-cost margin and the value of the SSNIP.13 The criticalloss is then compared to a measure of the “actual loss,” whichis the quantity that the economist believes will in reality belost to products outside the proposed market in the event ofa SSNIP. If the actual loss is smaller than the critical loss, themonopolist would find it profitable to raise its price and theproposed market is determined to be the relevant product orgeographic market.Potential Questions and Weaknesses.While this calcula-

tion seems relatively simple, there are important underlyingassumptions that the economist must verify to ensure themethod is properly applied. For instance, the calculationhinges on a measure of the gross margin, so discuss with youreconomist what data or documents were used to calculate the margin and whether they were interpreted properly.14

Econ omists often assume that the monopolist has constantmarginal cost when conducting the Hypothetical MonopolistTest using CLA, but this assumption may not be appropriatein some market settings and may be subject to challenge byenforcers or opposing economists. If an economist employsthis assumption, the attorneys should discuss with them if itis appropriate for the given market setting, and how theresults might change if the assumption is relaxed. Perhaps the most important component of CLA is the

measurement of the actual loss following a price increase.15

Data are usually not available to directly measure the loss ofsales that would result from a 5 percent price increase. Thus,economists often rely on imperfect measures of substitution,such as win/loss data, customer surveys, or (if the data andtiming permit) regression estimates from natural experiments,such as entry or exit of products. Economists should discusswith the attorneys the methods used to estimate the actualloss, the reliability of the data, and whether the estimated


2 8 · A N T I T R U S T

actual loss is consistent with the operation of the client’sbusiness and the industry in general. For example, in UnitedStates v. H&R Block, the opposing experts relied on IRSswitching data (plaintiff’s expert) and various surveys (defen-dants’ expert) to estimate the actual loss and obtained dif-ferent results that the court had to reconcile.16

Because the actual loss is often difficult to estimate, eco-nomic experts and attorneys may attempt to make inferencesabout the relevant market based only on the critical loss.Doing so could lead to inconsistent conclusions, as describedin detail in articles by Michael Katz and Carl Shapiro andDaniel O’Brien and Abraham Wickelgren.17 These articlesexplain that observing a small critical loss—which happenswhen the gross margin is large—should not automatically beinterpreted in favor of expanding the relevant market. Highgross margins imply that the actual loss is also likely to besmall (because firms generally enjoy high margins when theyface more inelastic demand), so it could still be the case thatthe actual loss is smaller than the critical loss. This issue arose during the preliminary injunction hearing

for the proposed acquisition of Wild Oats by Whole Foodsin 2007.18 The merging parties’ economic expert presented acritical loss analysis to support expanding the relevant prod-uct market put forward by the FTC. However, the mergingparties’ expert did not estimate the actual loss that wouldresult from a SSNIP, but rather “predicted” that it was large(based primarily on his review of market studies), whichimplied that a SSNIP would be unprofitable.19

The FTC’s expert critiqued this analysis based on the argu-ments described by O’Brien and Wickelgren.20 None theless,the district court concluded that “premium, natural, andorganic supermarkets” (PNOS) was not a distinct productmarket and denied the motion for a preliminary injunction,citing the merging parties’ critical loss analysis along withsupporting analyses based on third-party market studies, tes-timony from other supermarkets, and internal documentsfrom the merging parties.21 The district court’s decision waslater overturned by the D.C. Circuit Court of Appeals, citingconcerns about potential harm to a “core of committed cus-tomers” for PNOS that were not captured in the critical lossanalysis performed by the merging parties’ expert.22

Upward Pricing Pressure (UPP).UPP in Plain English.UPP was proposed by Joseph Farrell

and Carl Shapiro as a way to evaluate the merged firm’s incen-tive to increase prices unilaterally following a merger of com-petitors selling differentiated products.23 Rather than focusingon market definition, UPP aims to answer directly the relevantantitrust question: Is the merged firm likely to increase itsprice after the merger is completed?24 Given the tight overlapbetween market definition and anticompetitive effects in uni-lateral effects cases, the two calculations have a lot in common.The key insight here is that in a merger of two competi-

tors, sales that would have been lost to the competitor priorto the merger are internalized by the merged firm, whichincreases the incentive to raise prices post-merger (hence,

upward pricing pressure). Specifically, if a certain volume ofsales is internalized (or “recaptured”) by the merged firm, themagnitude of the upward pricing pressure is equal to thatsales volume multiplied by the merged firm’s price-cost mar-gin at the pre-merger price.25 In simple terms, multiplying therecaptured sales by the margin provides a measure of theprofit that the merged firm stands to gain from internalizingthe sales that would have been lost but for the merger. UPP also recognizes that the merger can lead to efficien-

cies that decrease the marginal cost of the merged firm andprovide an incentive to reduce price—referred to as down-ward pricing pressure. The UPP index is equal to the differ-ence between the upward and downward pricing pressure,and if it is determined to be positive, this is a signal that themerger is likely to result in a price increase.26

Potential Questions and Weaknesses. As with CLA, themain questions that the attorneys and economists shoulddiscuss concern the measurement of the price-cost marginand the volume of sales that is recaptured by the merged firmwhen it increases its price. The latter is often called the diver-sion ratio between the two merging parties, and measures theextent to which the parties are close substitutes for each otherin the eyes of consumers. Obtaining a precise measure of diversion is a complex and

data-heavy exercise (and might not even be possible, especiallyin the early stages of a merger investigation), so economistsoften rely on indirect measures coming from win/loss data,customer surveys, and estimates from natural experiments. Inthe absence of reliable data of this kind, economists sometimesresort to the use of share-based diversions, which only requireinformation on the market shares of firms. However, the con-ditions under which share-based diversions provide a reason-able estimate of the closeness of substitution are often not sat-isfied in practice, so share-based diversions may be limited intheir applicability.27

Because of these issues, measures of diversion are often apoint of contention between opposing economist. FTC v.Sysco provides a good example.28 In its decision to grant a pre-liminary injunction in FTC v. Sysco, the court reviewed diver-sion ratios calculated by the government’s expert, which werebased on request-for-proposal and bidding data, as well asordinary course of business records that tracked sales oppor-tunities. The court acknowledged defendants’ criticism ofthe diversion estimates in that the input data used wereincomplete and unreliable and might have reflected lost salesthat were unrelated to a price increase. The court ultimatelysided with the government, stating that it “hesitates to rely on[the government experts’] precise aggregate diversion per-centages,” but “when evaluated against the record as a whole,[the government experts’] conclusions are more consistentwith the business realities” in the relevant market.29 Thecourt’s decision in FTC v. Sysco illustrates the limitationsthat data issues can impose on diversion ratio analysis, but italso shows how those limitations may be overcome throughother supporting analysis.

S P R I N G 2 0 1 8 · 2 9

Merger Simulation.Merger Simulation in Plain English. Merger simulation

refers to any exercise that simulates a market to predict thepotential effects of a merger. In contrast to CLA and UPP,which consist of straightforward calculations and involve rel-atively modest data requirements, merger simulation is acomplex modeling and estimation procedure that requiresextensive data on the prices and quantities of all products inthe relevant market. The tradeoff is that while CLA and UPPcan provide answers to basic questions (e.g., is the proposedmarket a relevant antitrust market; is the merged firm likelyto raise its price), merger simulation can potentially predictpost-merger prices, quantities, profits, and changes in welfarefor all market participants, and can be used to simulate manycounterfactual scenarios, including proposed divestitures.30

Merger simulations vary by complexity depending on the data and time available to execute them. Typically, theeconomist begins by assuming a model of firm behavior (e.g., price-setting Bertrand competition or quantity-settingCournot competition) and a functional form for the costfunctions of firms.31 Consumer preferences are modeled byassuming a functional form of demand.32 These assumptionslead to a set of parameters that are estimated using pre-merg-er data on prices and quantities—often referred to as cali-brating the model. Then, using these estimated parameters,the economist simulates the merger to predict the equilibri-um prices and quantities that the merged firm would realizein the market post-transaction.33

Potential Questions and Weaknesses. If the merger sim-ulation is done properly, it can provide measures of the direc-tion and magnitude of the unilateral competitive effects.How ever, the simulation requires careful modeling and, ifdone incorrectly, it can lead to wrong answers. A standard cri-tique of a merger simulation is that the assumed competitionmodel does not capture the realities of the industry. Youreconomist should explain how the underlying model of com-petition fits the industry facts at hand. Key issues includewhether the industry is characterized as one with homoge-neous or differentiated products or if the buyer-seller rela-tionship is more accurately represented by an auction model. You should discuss the values of the estimated parameters

in the calibration stage to ensure that they conform to themarket realities. For instance, are the estimated pre-mergerprice-cost margins similar in magnitude to the ones report-ed by the client? Data on pre-merger market shares are fre-quently used to calibrate the merger simulation model andshares are often a point of dispute between opposing econo-mists. For example, in United States v. Oracle, the govern-ment’s expert used market shares that the court found to bean inaccurate representation of the parties’ positions in themarket and therefore the court dismissed the results from thesimulation as unreliable.34

The results from a merger simulation are sensitive to thefunctional form of the demand function and the correspon-ding demand elasticities, so this is an area that might receive

scrutiny from enforcers and opposing economists. Your econ-omist should discuss the limitations of her particular choiceand how alternative choices might affect the estimated post-merger prices. Another question to discuss with the economist is the

treatment of efficiencies, the likelihood and magnitude ofwhich are often disputed by the two sides in a merger inves-tigation. For example, are the merging firms assumed to havelower costs post-merger and, if so, how much? Where are theefficiencies coming from? Most merger simulations assumethat the merger only affects pricing and output decisionsand possibly production costs. However, in some settings, thetransaction might affect entry or exit, changes in productofferings, or changes in other strategic variables such as adver-tising. Discuss with your economist whether the enforcers oropposing economists are likely to raise these issues and, if so,what would be the best way to incorporate them in the merg-er simulation.

ConclusionWhile antitrust attorneys and economists often use differentterminology, providing the highest level of service to clientsrequires that attorneys and economists work together toaddress the same questions about competition in markets.Combining insights from attorneys’ and economists’ differ-ent skill sets and backgrounds is the best way to answer thesequestions. Build your conversations with economists on thiscommon ground and you may find that the language barri-er is not so high after all.�

1 See, e.g., Ryan Faughnder, Home Video Sales Shrank Again in 2016 asAmericans Switched to Streaming, L.A. TIMES, Jan. 6, 2007, http://www.latimes.com/business/hollywood/la-fi-ct-home-video-decline-20170106-story.html.

2 Comcast Corp. v. Behrend, 133 S. Ct. 1426 (2013). 3 Id. at 1430–31. 4 Id. at 1434–35. 5 For more details on instrumental variables, see JOSHUA D. ANGRIST & JöRN-STEFFEN PISCHKE, MOSTLY HARMLESS ECONOMETRICS: AN EMPIRICIST’SCOMPANION ch. 4, at 113 (2009).

6 Freeland v. AT&T Corp., 238 F.R.D. 130 (S.D.N.Y. 2006).7 Blue Dane Simmental v. American Simmental Ass’n, 178 F.3d 1035,1039–41 (8th Cir. 1999).

8 Properly implementing CLA or UPP analysis is by no means an easy exercise.As we discuss below, the use of diversion ratios in UPP analysis is partic-ularly tricky and should be done with care.

9 For cases in which CLA featured prominently, see, e.g., FTC v. TenetHealthcare Corp., 186 F.3d 1045 (8th Cir. 1999); United States v. SungardData Sys., 172 F. Supp. 2d 172 (D.D.C. 2001); United States v. H&R Block,Inc., 833 F. Supp. 2d 36 (D.D.C. 2011).

10 For cases where the Hypothetical Monopolist Test was accepted by the courtas an appropriate tool for market definition, see, e.g., United States v. AetnaInc., No. 16-cv-1494, 2017 WL 325189 (D.D.C. Jan. 23, 2017); FTC v.Staples Inc., 190 F. Supp. 3d 100 (D.D.C. 2016); FTC v. Sysco Corp., 113F. Supp. 3d 1 (D.D.C. 2015).

11 U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines4.1.1. (2010).


3 0 · A N T I T R U S T

12 “The Agencies most often use a SSNIP of five percent of the price paid bycustomers for the products or services to which the merging firms contributevalue.” Id. § 4.1.2.

13 Critical loss was first introduced by Barry Harris and Joseph Simons. SeeBarry C. Harris & Joseph J. Simons, Focusing Market Definition: How MuchSubstitution Is Necessary?, 12 RESEARCH IN L. & ECON. 207 (providing a for-mal overview of the method and the arithmetic).

14 The proper measurement of margins was a topic of disagreement betweenthe economic experts in the recent preliminary injunction hearings in, e.g.,FTC v. Advocate Health Care Network, 1:15-cv-11473, 2017 U.S. Dist.LEXIS 37707 (N.D. Ill. Mar. 16, 2017), and Sysco, 113 F. Supp. 3d 1.

15 Note that the actual loss is a completely separate entity from the criticalloss and has to be evaluated through other means.

16 H&R Block, 833 F. Supp. 2d at 62–63.17 Michael Katz & Carl Shapiro, Critical Loss: Let’s Tell the Whole Story,

ANTITRUST, Spring 2003, at 49; Daniel P. O’Brien & Abraham L. Wickelgren,A Critical Analysis of Critical Loss Analysis, 71 ANTITRUST L.J. 161 (2003).

18 FTC v. Whole Foods Mkt., Inc., 502 F. Supp. 2d 1 (D.D.C. 2007). 19 Id. at 14; see also FTC v. Whole Foods Mkt., Inc., 548 F.3d 1028 (D.C. Cir.

2008). 20 See Rebuttal Expert Report of Kevin Murphy, supra note 18, ¶ 25 (“O’Brien

and Wickelgren, in their critique of CL analysis, use this very methodologyto calculate whether it would be profitable for a hypothetical monopolist toincrease price by a given amount.”).

21 Whole Foods, 502 F. Supp. 2d at 34–36. 22 Whole Foods, 548 F.3d at 1041. 23 Joseph Farrell & Carl Shapiro, Antitrust Evaluation of Horizontal Mergers:

An Economic Alternative to Market Definition, B.E. J. THEORETICAL ECON., Vol. 10, Iss. 1 (Policies & Perspectives), Art. (2010), http://faculty.haas.berkeley.edu/shapiro/alternative.pdf.

24 In its original version, UPP analysis speaks only to the likelihood of a priceincrease, but not to the magnitude of it. See id. (detailed discussion).

25 This general description of the analysis effectively assumes that the merg-ing firms have the same costs and the same price prior to the merger. Torelax this assumption, it may be possible to expand the analysis by calcu-lating two upward pricing pressure indices—one for each of the mergingfirms.

26 Another term often used in the context of UPP analysis is GUPPI, or “GrossUpward Pricing Pressure Index.” See Steven Salop & Serge Moresi, Updatingthe Merger Guidelines: Comments (Georgetown Law Faculty Publicationsand Other Works 1662, Nov. 9, 2009, http://scholarship.law.georgetown.edu/facpub/1662. UPP is viewed as a net index since it subtracts theimpact of efficiencies from the upward pricing pressure. GUPPI ignores effi-ciencies and looks strictly at the upward pricing pressure. See id.

27 In particular, in calculations for share-based diversion ratios, the underlyingassumptions are that all sales lost by the firm/product in question arerecaptured by other firms/products in the market and that all firms/prod-ucts are equally close substitutes in the eyes of consumers.

28 Sysco, 113 F. Supp. 3d 1. 29 Id. at 37. 30 For instance, the economist can answer such question as how big the

merger efficiencies have to be to defeat the incentive for a price increaseor if the entry of a new competitor would be sufficient to discipline pricespost-merger.

31 Of particular interest here is the marginal cost function. Most often, econ-omists assume that marginal costs do not vary with output.

32 In cases where the data are insufficient to perform demand estimation,economists might use other available information to estimate the values ofdemand elasticities, which are then used in the merger simulation.

33 For an in-depth description of merger simulation methods, see e.g., Farrell& Shapiro, supra note 23; Gregory J. Werden & Luke M. Froeb, Correlation,Causality, and All That Jazz: The Inherent Shortcomings of Price Tests for

Antitrust Market Delineation, 8 REV. INDUS. ORG. 329 (1993). A brief non-technical example is provided by Roy Epstein, Merger Simulation andUnilateral Effects: A Primer for Antitrust Lawyers, ECON. COMM. NEWSL. (ABASection of Antitrust Law), Vol. 2, No. 2 (2002). See also Joseph Farrell & CarlShapiro, Horizontal Mergers: An Equilibrium Analysis, 80 AM. ECON. REV. 107(1990).

34 United States v. Oracle Corp., 331 F. Supp. 2d 1098 (N.D. Cal. 2004).

� NINETY PERCENT OF U.S. antitrust enforcementis generated through private actions, and the class actionis perhaps the primary mechanism through which privateactors seek to enforce antitrust laws. Class action law andprocedure are becoming an unavoidable fact of life forantitrust practitioners.This second edition of the Antitrust Class Actions

Handbook is a vital reference for both new and experienced antitrust practitioners navigating the classaction thicket. The book is intended as a comprehensivediscussion of the issues that arise before, during, and after the filing of an antitrust class action. It considers the issues and challenges faced by both plaintiffs anddefendants and focuses on the evolving legal standards as well as the strategic decisions that guide these highlycomplex cases.



Antitrust Class ActionsHandbook Second Edition



S P R I N G 2 0 1 8 · 3 1

solicitors, Sophie Lawrance and Pat Treacy, partners in BristowLLP. Sophie and Pat explained that use of the hot tub method hasincreased in the UK generally, and in particular, in the Competi -tion Appeals Tribunal (CAT), and that court rules have been adopt-ed endorsing the method. The relevant court rules refer to the hottub method as “concurrent expert evidence” or CEE. An initialrule was adopted in 2013, and then expanded upon in a revisedrule just published in December 2017. The rule just adopted inDecember 2017 provides:

Concurrent expert evidence [CEE]

11.1 At any stage in the proceedings the court may direct thatsome or all of the evidence of experts from like disciplinesshall be given concurrently. The procedure set out in paragraph11.4 shall apply in respect of any part of the evidence whichis to be given concurrently.

11.2 To the extent that the expert evidence is not to be givenconcurrently, the court may direct the evidence to be given inany appropriate manner. This may include a direction for theexperts from like disciplines to give their evidence and becross-examined on an issue-by-issue basis, so that each partycalls its expert or experts to give evidence in relation to a par-ticular issue, followed by the other parties calling their expertor experts to give evidence in relation to that issue (and so onfor each of the expert issues which are to be addressed in thismanner).

11.3 The court may set an agenda for the taking of expert evi-dence concurrently or on an issue-by-issue basis, or may directthat the parties agree such an agenda subject to the approvalof the court. In either case, the agenda should be based uponthe areas of disagreement identified in the experts’ joint state-ments made pursuant to rule 35.12.

11.4 Where expert evidence is to be given concurrently, then(after the relevant experts have each taken the oath oraffirmed) in relation to each issue on the agenda, and subjectto the judge’s discretion to modify the procedure—

(1) the judge will initiate the discussion by asking theexperts, in turn, for their views in relation to the issues onthe agenda. Once an expert has expressed a view the judgemay ask questions about it. At one or more appropriatestages when questioning a particular expert, the judge mayinvite the other expert to comment or to ask that expert’sown questions of the first expert;

(2) after the process set out in (1) has been completed forany issue (or all issues), the judge will invite the parties’ rep-resentatives to ask questions of the experts. Such ques-tioning should be directed towards:

Hot Tub ReduxBY L I S A C . W O O D

NINE YEARS AGO, I WROTE A COLUMN IN THISmagazine about a trend in the use of expert witnessesin court.1 The trend, originating in Australia, was tohave experts present their testimony concurrently to the

court, without lawyers acting as intermediaries. The concurrentpresentations allowed the court to directly question the expertsin real time about areas of agreement and disagreement. Rela -tive to typical cycles of lawyer intermediated reports, replies,depositions, direct, and cross, the concurrent presentations wereless structured and more free-wheeling. With a wink, the processwas known as “hot tubbing.”

At the time, I was fortunate to be able to interview a federaldistrict court judge from my home District of Massachusetts,Judge Doug Woodlock, who had used the hot tub techniqueenthusiastically in a few jury-waived proceedings. For a subse-quent column, I interviewed several expert witnesses who hadexperience in Australia with both the hot tub and the similarjoint conference process.2 (The joint conference process hasopposing experts meet, without counsel, before trial in anattempt both to narrow the issues in dispute and to prepare adocument summarizing the issues on which the experts agreeand disagree.)

Since writing those two columns back in 2007, the use ofthese techniques has spread beyond Australia to the UK andCanada, both of which have adopted rules encouraging hot tubsand joint conferences. And several U.S. federal judges now reg-ularly employ hot tubs. I remain a booster and hope that by shar-ing these recent developments, the use of expert hot tubs willcontinue to increase in the United States.

The Use of Hot Tubs and Joint Conferences in the UK The Current Rules. To get a handle on recent hot tub devel-opments in the UK, I interviewed two experienced UK competition

L I T I G A T I O N P R A C T I C E

Notes from the Field

Lisa C. Wood is the Litigation Practice Editor for ANT ITRUST , and a partner

at Foley Hoag LLP, where she chairs the Litigation Department and han-

dles complex litigation and government enforcement matters involving

anti trust, securities, and accounting issues.

(a) testing the correctness of an expert’s view;

(b) seeking clarification of an expert’s view; or

(c) eliciting evidence on any issue (or on any aspect of anissue) which has been omitted from consideration duringthe process set out in (1); and

(3) after the process set out in (2) has been completed in rela-tion to any issue (or all issues), the judge may summarize theexperts’ different positions on the issue and ask them to con-firm or correct that summary.3

Sophie and Pat also provided some helpful context for under-standing how hot tubs work in competition cases in the UK.Juries are not used for civil cases, and most civil business casesare heard by specialist rather than generalist judges. Competitioncases are heard either in the Competition Appeal Tribunal, inwhich hearings are presided over by a three-person panel (onlyone of which need be a judge, and the others may be experts oracademics from the CAT panel), or in the Chancery Division of theHigh Court (in which a single judge presides). Judges are not ran-domly assigned to cases in the Chancery Division, and attemptsare made to assign a judge with competition experience to anycompetition case being heard there.

Sophie and Pat explained that the judiciary who preside overcompetition disputes have considerable discretion in how tomanage cases, including how best to handle expert evidence.Even before rules were adopted encouraging use of the CEE,some UK judges started experimenting with the technique. Inthose matters in which the expert evidence is not presentedconcurrently, direct expert witness testimony is typically pre-sented by written report, and only the cross examination and re-direct occurs before the court.

One way in which judges first started trying to better managethe expert evidence was to request that opposing experts meetto determine the issues actually in dispute and prepare a docu-ment identifying the issues on which they agree and disagree.This meeting, which appears to be much like the joint conferencein Australia, occurs after the experts have submitted their reportsbut before the trial at which the experts will be cross-examined.Counsel does not participate in this meeting among the experts.In those cases in which a hot tub is utilized, the court uses thereport prepared at this expert meeting as an agenda, and in thejudgments and transcripts quoted from below, the court thankedthe experts for working to reduce the issues in dispute during thispre-trial meeting among the experts.

Efficiency was identified by Pat and Sophie as the primary ben-efit of the hot tub method; they explained that the courts in theUK are acutely aware of the importance of efficiency because ofits impact on the attractiveness of the London courts for the res-olution of business disputes. Approximately 80 percent of thecases handled in the Chancery Division of the High Court involveparties from a foreign jurisdiction, and in many cases, both par-ties are from foreign jurisdictions. The London courts are there-fore in competition with other jurisdictions to attract foreign liti-gants who have the option of selecting a particular jurisdiction.

The primary drawback, Pat and Sophie explained, is the extraor-dinary effort by the judiciary required to run a hot tub effectively.

The court has to understand the issues addressed by the com-peting experts well enough to ask good questions, and cover allthe salient points. Sophie and Pat commented that one of the rea-sons that the CAT may have utilized the hot tub method more fre-quently is that in addition to having three people to share the load,they have access to court personnel called référendaires, who arelegally qualified assistants trained to assist the judges. Thejudges in the Chancery Division of the High Court do not have thesame access to such personnel, and of course those proceedingsare presided over by one judge rather than three.

According to Pat and Sophie, some barristers have criticizedCEE because it can limit cross examination. For example, somecounsel fear the court will not be as familiar with the issues asis counsel and thus may miss certain issues during the hot tuband yet limit counsel in her subsequent cross examination, think-ing the issues have already been adequately covered in the hottub. The latest practice advisory, just issued in December 2017,appears to try to address this concern by suggesting that thecourt break up the hot tub into discrete issues and allow coun-sel to conduct cross examination in stages after the hot tub oneach issue. This recommended tweak in the process should alsohelp deal with another criticism, which is that the hot tub methodtakes away the benefit of surprise. If an expert is not crossexamined until after a hot tub on all the key issues, then it is like-ly she will have had the chance to think through some of the chiefcriticisms of her work and come up with a better answer thanmight have been the case on a cross examination conductedwithout a prior hot tub exercise. This is particularly the case if theexpert is permitted to confer with the counsel who retained her,after the hot tub and before cross examination.

The Judiciary’s Views. Two recent speeches by promi-nent UK judges indicate that the UK judiciary had three reasonsfor recommending use of CEE—encouraging more objectivity byexperts, better educating the court on the complex issuesaddressed by the experts, and reducing litigation costs.

In a lecture at the London Conference of the Commercial BarAssociation of Victoria on June 29, 2016, Lord Justice Jacksonexplained that he researched the hot tub method in use inAustralia as part of his efforts at civil justice reform and civil lit-igation costs reduction, and concluded it was a cost saving pro-cedure that merited a pilot study in England and Wales. In thesurveys undertaken after the pilot studies, Justice Jacksonexplained that he confirmed that judges and practitioners alikeconcluded that use of CEE saved considerable time, particularlyat trial. He quoted one judge who explained that he had con-ducted two hot tubs in a case with two sets of experts, each ofwhich lasted 2 days, and estimated this halved the time spent onexpert testimony at trial.

Although views differ on the question of costs saving, mostEnglish judges and practitioners agree that CEE leads to a sav-ing of time at trial. Since trial time is the most expensivecomponent of litigation, this is a valuable saving. Solicitorswho are striving to bring the actual costs of litigation closer tothe approved budget should give serious consideration to pro-posing the use of CEE.


3 2 · A N T I T R U S T

S P R I N G 2 0 1 8 · 3 3

course the hot-tubbing process can be deployed flexibly: thejudge can combine inquisitorial questioning from the benchwith cross examination by counsel. There are thus many dif-ferent ways in which the process can be managed so as tomaximize efficiency.

CEE in Practice. Pat Treacy and Sophie Lawrance provid-ed me with copies of transcripts from four UK competition casesin which a hot tub was employed. Those cases all were tried in2016 or 2017, and were in the following litigations:

� Streetmap v Google [2016] EWHC 253: hot-tub on Day 5 ofthe trial.

� Socrates v Law Society [2017] CAT 10: hot-tub on Day 3 ofthe hearing.

� Agent’s Mutual v Gascoigne [2017] CAT 15: hot-tub at Days7–8 of the hearing.

� Generics, GSK & others v CMA [Paroxetine, combined cases1251–1255/1/12/16]: hot-tub on Days 7–9; judgmentpending.

The judgments entered in these cases and several othersmake enlightening comments about the hot tub method. Forexample, in the Streetmap v. Google matter, which was tried to theChancery Division of the High Court of Justice, the court com-mented that the use of the hot tub created efficiencies andhelped narrow the issues of disagreement among the accom-plished expert witnesses who had both become too much of anadvocate in the court’s view. The court also commented, as hadJudge Green in his speech discussed above, that to be effective,the court must spend considerable time in advance of the hot tubbecoming familiar with the issues, and must have a transcriptprepared of the exercise:

Each side called one economic expert and the court used a so-called “hot-tub” for the joint presentation and scrutiny of thoseexperts’ oral evidence. I believe that is the first time this hasbeen done in a competition case in the UK, and it led to a con-structive exchange which considerably shortened the timetaken by the economic evidence at trial. However, I shouldmention that this process involves considerable preparation bythe court and effectively requires (as in the present case) atranscript since the judge is unable to keep a proper notewhile leading the questioning. Each [expert economist] is apartner in a leading economic consultancy and has frequentlybeen involved in giving economic evidence in competitioncases. Both have undoubted expertise in this field, but I foundthat each displayed a tendency to become an advocate for theparty by which he was instructed. Much of their respectivereports was concerned with presenting various different meas-urements of searches for online maps or online mapping web-sites, and analyzing the results. The fundamental economicissues in the present case are not particularly complex, and onthose the hot-tub process led to a significant measure of agree-ment that was helpful, although the two experts remained verydivided on their interpretation of some of the data they pre-sented.

In the judgment rendered in the British Telecommunications v.Ofcom matter, another case tried before the CAT, in which a hottub was employed, the Tribunal also commended the hot tub forits efficiency in narrowing the issues. After a disagreementbetween the parties on the list of relevant matters for the con-

Justice Jackson also commented that while the hot tub exer-cise requires more advance preparation by the judge, this was“not a bad thing” as the judge would have to master the expertevidence sooner or later in any event. Justice Jackson conclud-ed his lecture by predicting that as more people became famil-iar with the concurrent expert evidence technique, its use wouldincrease:

It is striking that the majority of those who are hostile to theprocedure are judges or practitioners who have never used it.Most (but certainly not all) judges and practitioners who haveused the procedure are supportive. . . . It is hoped that the useof concurrent evidence will increase as the benefits becomemore widely appreciated.

Echoing similar themes, Judge Nicholas Green, a judge fromthe UK High Court who previously practiced as an antitrustlawyer, gave a speech in late 2016 provocatively entitled “Peoplein this Country Have Had Enough of Experts.” Judge Green notedthe increasing complexity of “services and goods that drive mod-ern economies” that can “be so overwhelmingly specialized thatit is beyond all but a few.” Compounding the problem, explainedJudge Green, is that “this same complexity also encompassesthe analytical techniques, those “black boxes” of tricks, whichare now routinely used by experts to unlock the most seeming-ly intractable of technical problems.” If courts are going to con-tinue to treat as admissible up-to-date scientific evidence, some-thing must be done to address the knowledge asymmetrybetween the court and the witnesses before it.

In addressing what courts could do to address this knowledgeasymmetry problem, Judge Green recommended increased useof the hot tub method of expert presentation, including the jointconference among experts in advance of trial. He also noted the other benefits of CEE, including better control over expertadvocacy:

The experience of Australia indicates that the practice is effec-tive in saving both time and costs and repatriates to expertstheir proper role of assisting the court to resolve disputes. Itdoes away with the gladiatorial combat between cross-exam-ining counsel and expert that was hitherto characteristic of lit-igation. The practice has, albeit slowly, begun to take root inthis jurisdiction. A number of judges in the Technology Con -struc tion Court (“TCC”) regularly use “hot-tubbing.” For somejudges it is now the rule rather than the exception. The expe-rience is that experts in the field of construction engage wellwith each other and in court may debate a point constructive-ly to the exclusion of the lawyers who then participate (proba-bly to their chagrin) as no more than interested onlookers.

The judge commented on the importance of judicial prepara-tion to the effectiveness of a hot tub exercise, but also made theimportant point about the role of counsel and experts in prepar-ing the court for that exercise:

For “hot-tubbing” to be effective it presupposes that the judgeis on top of the expert issues prior to trial otherwise theprocess of a judge-led structured dialogue will not be effective.And this presupposes also that the material the judge isrequired to pre-read is comprehensible and manageable. Bothpremises might be optimistic, albeit that experience fromjudges familiar with the process is in fact very promising. Of

current session, the Tribunal provided the parties with a list oftopics that would serve as an agenda:

We found the concurrent expert session very useful. Whilst thesubmissions by the two parties prior to the Hearing did notidentify material areas of agreement, the session itselfenabled the Tribunal to hear Dr. Padilla’s explanation of hiswork, Dr. Caffarra’s criticisms of it and Dr. Padilla’s responsein an open and accessible form. The Tribunal’s ability to leadthe examination with the expertise of Professor Colin Mayerenabled it to evaluate the issues effectively in a single morn-ing instead of the one or two full days that might otherwisehave been required.

In the judgment rendered in the Socrates Training matter, theTribunal made similar remarks in the judgment:

Their written reports [of the two economic experts] were com-mendably clear and their joint statement of issues on whichthey agreed and on which they disagreed was very helpful.Their oral evidence was heard concurrently, in a so-called ‘hottub’, and their constructive and very sensible approach inresponse to the Tribunal’s questions made this a valuable andefficient exercise.

In another judgment entered in a patent matter, in whichFRAND issues arose, the court held a hot tub on issues that hadnot been addressed in the expert reports and that he thoughtshould be addressed at the same time by both experts. Thecourt explained its thinking in the judgment as follows:

During the trial I decided that it would be useful if the evidenceof [the two experts] started with a short period of concurrentevidence (CPR 35PD section 11). This was in order to addresscertain general questions which I would otherwise have askedthe experts during their oral evidence. They were not pointsaddressed to either witness in particular and it seemed to methat the fairest way of dealing with it was to ask them both,hence a concurrent evidence session. . . [T]he concurrentsession took place just before [the first expert’s] cross-exam-ination. It helped me clarify my understanding of some of theeconomic issues. . . .

In the Generics (UK) Limited matter before the CAT, the hot tubexercise lasted for 3 full days. The experts had with them theirreports, the reports of the other experts, and the joint state-ments, but no other materials. The President of the Tribunalbegan the exercise by thanking the experts for the joint state-ment and reminding them of their obligation to be independentof the parties who retained them. The President then summa-rized the process that would be followed for the hot tub in thismatter:

[W]e will identify the questions that we would like you torespond to or discuss. We will address a question initially toone of you, and then invite others to comment. If you wish tospeak at any stage, please indicate appropriately, raise a pen-cil or your finger, or whatever. But please do not speak acrosseach other or to each other. Not only does that disrupt anyorderly discussion, but, as you know, we are having a transcriptand it becomes a nightmare for the transcribers if that startsto happen. You will all have a chance to have your say, so donot worry.

At various stages after we have completed a topic we willallow counsel to ask any supplementary questions of the

expert retained by the other side. . . . We will use your jointstatement as a basis.

At several points during the long hot tub, the judges inter-rupted one of the experts to remind him that the hot tub exercisewas focused on theory and general principles rather than thefacts of the case. Also, the President tried to summarize eachexpert’s position after concluding the discussion of each topic,and then asked the expert whether he agreed or disagreed withthe stated summary. This must have been a challenging exercisefor the experts, who had to listen very carefully, and then try diplo-matically to disagree if necessary with the summary as stated bythe President. On the third day of the hot tub, counsel agreed toshorten the cross examination of the experts to allow for addi-tional time for the hot tub, which may be another act of diplomacy,or an acknowledgment of the efficacy of the hot tub.

The transcript of the hot tub exercise in the Agents Mutualmatter before the CAT was also instructive. The Tribunal began byexplaining how the process would work:

Essentially, it is a collaborative process where we, the Tribunal,engage in a form of conversation with you and between your-selves with a view to educating the Tribunal as to the issuesin play. So what will happen is that I, in conjunction with my col-leagues, will lead the discussion. We’ll try and get you to talkboth to us and amongst yourselves.

During the hot tub, the Chairman of the Tribunal regularlypaused to summarize what he thought a particular expert wassaying, and then asked the expert whether he agreed or dis-agreed with his summary. At one point the Chairman admittedthat he had misunderstood one aspect of the expert’s report andthat the exchanges in the hot tub exercise had cleared up histhinking.

Use of Expert Hot Tubs in CanadaTo get a handle on the use of hot tubs in competition cases inCanada, I spoke with Kate McNeece, who is an antitrust associ-ate at the law firm, Blakes. Kate explained that while Canada hasadopted rules permitting a court to require the use of a hot tub,and a number of administrative tribunals employ hot tubs, thetechnique has not yet been used in any competition case. Kateexplained that The Federal Courts Rules (SOR/98-106) paras.282.1 and 282.2 allow the Court to require expert witnesses totestify as a panel:

Expert witness panel

282.1 The Court may require that some or all of the expert wit-nesses testify as a panel after the completion of the testimonyof the non-expert witnesses of each party or at any other timethat the Court may determine.

Testimony of panel members

282.2 (1) Expert witnesses shall give their views and may bedirected to comment on the views of other panel members andto make concluding statements. With leave of the Court, theymay pose questions to other panel members.

Examination of panel members

(2) On completion of the testimony of the panel, the panel


3 4 · A N T I T R U S T

S P R I N G 2 0 1 8 · 3 5

members may be cross-examined and re-examined in thesequence directed by Court.

These rules were the result of 2010 amendments to the Fed -eral Court Rules.4 The same amendments allowed the court toorder expert witnesses to confer with one another in advance ofthe proceedings in order to narrow the issues (para. 52.6). Katefurther explained that similarly, the Competition Tribunal Rulesalso allow the court to appoint “one or more” independentexpert, which is generally understood to contemplate hot tub-bing.5

In his 2010 article, The Changing Role of the Expert Witness,the Honorable Ian Binnie of the Supreme Court of Canada, makesplain that expert advocacy and lack of independence from coun-sel and litigants is one of the primary reasons for recommend-ing use of a hot tub.6 Justice Binnie recommended a number ofchanges to the way in which expert testimony is presented inCanada, including the requirement that “experts exchangereports and meet face to face for an unmediated discussionbefore trial” much like the joint conference used in Australia andthe joint statement now prepared by experts in the UK. JusticeBinnie also recommended that “a court should be able to requireopposing experts to testify on the same panel and to be subjectto questioning in the presence of each other, with the right toquestion each other in the presence of the trier of fact.” After not-ing that the procedure was used in administrative tribunals inCanada and in courts in Australia, Justice Binnie explained hisunderstanding of the merits of the “hot pot” as he called it:

The theory is that experts testifying in the presence of oneanother are likely to be more measured and complete in theirpronouncements, knowing that exaggeration or errors will bepounced upon instantly by a learned colleague, as opposed tobeing argued about days later, perhaps by unlearned opposingcounsel.

Increased Use of Expert Hot Tubs in U.S. CourtsSince the “unprecedented” use of a hot tub in the 2003 census-challenge trial presided over by Judge Woodlock and two otherfederal judges, the use of hot tubs in federal litigation hasincreased significantly. Adam E. Butt, an Australian litigator whoalso practices in the U.S., in writing about the use of hot tubs inthe U.S., reported that hot tubs have been utilized by severalother U.S. judges (in addition to Judge Woodlock) in a range ofproceedings, including in Daubert hearings, “a claims construc-tion hearing, a class certification hearing and other civil mat-ters.”7 Butt further reported that while “the method has notbeen seen as problematic in non-jury contexts,” its use in jury tri-als meets with different reactions by the judiciary, some “wouldavoid using hot tubbing in jury trials, believing it to be inappro-priate for judges to inquire into or comment on expert evidencein front of jurors.” Butt reported that other judges

do not consider that the jury is off limits but they have theircertain qualifications. For example, Judge Woodlock wouldneed to be comfortable with who the experts were in order touse hot tubbing before a jury. Judge Zouhary would supportusing hot tubbing in jury cases where the expert evidence

was complicated (it helps to comprehend such evidence), butwould avoid using it in simpler matters. Judge Weinstein hasactually now used hot tubbing in one jury trial, in a birthingcase. Nevertheless, he states that he would intervene less insuch settings, because his intervention may be demeaning toattorneys, the jury may give greater reliance to questions/posi-tions put forward by the judge, and the concurrent presentationof evidence (cf. sequential presentation) may create compli-cations in relation to burdens of proof and allowing attorneysto present their case.

The Ohio Bar Association published a fascinating interview8

with Judge Zouhary9 about his first use of the hot tub method inthe context of a class certification hearing. Judge Zouhary beganthe interview by extolling the virtues of the hot tub method:“Throwing everybody in the ‘hot tub’ at the same time allows thecourt, counsel and experts to confront or, ‘splash,’ each otherdirectly, resulting in a better chance of reaching a correct con-clusion.” He further explained that he had adopted the procedureon his own, without knowing of its use in Australia or elsewhere.

Judge Zouhary was prompted to adopt a hot tub to assist himin ruling on a class certification motion. He wanted to test theexpert assertions made in the affidavits and deposition testimo-ny, but wanted to do so short of a full-blown evidentiary hearing.Moreover, he wanted to provide an opportunity for the experts,who disagreed with one another, to have direct contact with himand with one another. This would not be a “traditional hearing forcounsel to wax on.”

In the interview, Judge Zouhary quoted directly from the orderhe issued in advance of the hot tub, which is also available onPacer:

At the beginning of each session, all experts for that sessionwill be sworn. This Court, the experts, and counsel for eachside will then engage in a discussion, structured around thisCourt’s questions. That conversation may include back-and-forth directly between the experts, in a point/counterpointfashion, with this Court moderating. For instance, this Courtmay ask [plaintiff’s expert] to comment on [defendant’sexpert’s] critiques with respect to an aspect of his impactmodel, then ask [defense experts] to respond, and so on. ThisCourt may invite counsel to join in the legal aspects of that dis-cussion, or comment on the legal consequences of the expertback-and-forth (e.g., what would follow, as a legal matter, fromaccepting or rejecting a particular expert’s criticisms). Counselin each session may also make “opening statements” (not toexceed 10 minutes each, delivered before discussion with theexperts) that show why plaintiffs have or have not met Rule23’s requirements.

Judge Zouhary further explained that after setting aside a fullday for this hot tub exercise, he sent “counsel ahead of time aset of questions that [he] wanted to be the focus of our discus-sion,” a technique he also uses in advance of oral argument andother pre-trial hearings. He explained why he liked the hot tub somuch:

I found the experience rewarding and will not hesitate to useit again in the right case. What is “the right case?” One thatinvolves multiple experts and a lengthy record, or perhaps acomplex Markman hearing. The procedure requires the duel-ing experts to focus on the same point at the same time. And

interesting tool. I look forward to hearing of your hot tub experi-ences at [email protected].�

1 Lisa C. Wood, Experts in the Tub, ANTITRUST, Summer 2007, at 96.2 Lisa C. Wood, Experts Only: Out of the Tub and into the Joint Conference,ANTITRUST, Fall 2007, at 89.

3 Paragraph 11 of Practice Direction 35, entitled Experts and Assessors(adopted Dec. 21, 2017). Practice Directions apply to Civil Procedure Rulesin effect in civil cases in the Queen’s Bench Division and the ChanceryDivision of the High Court, and to litigation in the county courts other thanfamily proceedings. They are issued by the Lord Chief Justice.

4 The Federal Courts Rules can be found at http://laws-lois.justice.gc.ca/eng/regulations/SOR-98-106/FullText.html. (SOR/2008-141), para. 80(1).

5 The Competition Tribunal Rules can be found at http://laws-lois.justice.gc.ca/eng/regulations/SOR-2008-141/FullText.html.

6 49 S.C.L.R. 2d 179 (2010).7 Adam E. Butt, Concurrent Expert Evidence in the United States—Is There a

Role for Hot Tubbing?, The Civil Jury Project at NYU School of Law (2018),http://civiljuryproject.law.nyu.edu/concurrent-expertevidence-in-the-united-states-is-there-a-role-for-hottubbing/.

8 Jack Zouhary, Splash: Hot Tubbing in a Federal Courtroom, 29 OHIO LAW. 10(2015), https://www.ohiobar.org/NewsAndPublications/OhioLawyer/Pages/Splash-Hot-tubbing-in-a-federalcourtroom.aspx.

9 Judge Jack Zouhary is a federal district court judge sitting in Toledo, Ohio.He has served as a visiting district court judge in several states, sits byassignment on the Sixth and Ninth Circuit Courts of Appeals, and is anactive Fellow of the American College of Trial Lawyers.

10 Melissa Lipman, Judge Illston’s Tips for Economic Experts in Antitrust Cases,LAW360 (Apr. 16, 2015).

11 See Rovakat v. Comm’r, 255 TCM 29, XIV C (20100), Crimi v. Comm’r, 51TCM (2013); Green Gas Delaware Statutory Trust v. Comm’r, 147 T.C. No.1 (July 14, 2016).

the “point/counterpoint” dialogue—as opposed to the tradi-tional appellate-type monologue—is a better way of evaluatingthe accuracy of an expert’s opinion. There is no hiding.

Judge Susan Illston, of the Northern District of California hasutilized procedures very similar to hot tubs in antitrust cases. Forexample, Judge Illston has had the plaintiff’s and defendant’sexperts testify “back to back on particular, difficult issues.”10

Judge Illston argues that this and other procedures create anorder to the evidence which, in turn, allows the jury to “retain abetter understanding of what those [disputed] issues are.”

Another context in which hot tubs have been used in U.S.courts is in the U.S. Tax Court. In several instances, Tax Courtjudges have, with the consent of the parties, received concurrentevidence from expert witnesses.11

ConclusionWhy, you may ask, do I remain a hot tub enthusiast? In additionto the benefits now recognized by judges and advocates inAustralia, the UK, Canada and the U.S., I believe strongly that theUK hot tub is a more effective way to present complex expert tes-timony. In two of the hot tubs for which I read a transcript, the hottub took place over two or three days, with vigorous questioningby the tribunal and the opposing experts. While some advocatesmay feel as though the hot tub robs them of the opportunity forcross examination, my reaction upon reading the transcripts ofthese longer hot tubs was that it was more effective than crossexamination at ferreting out the issues, and displaying the weak-er or less credible opinions. This was both because of the inter-active nature of the exercise and the fact that the opposingviews on each issue were addressed concurrently.

Another benefit of the hot tub exercises I reviewed is that thecourt was obviously more engaged in the discussion because itwas asking the questions. Watching someone else conduct across examination can be much less exciting and engaging thanundertaking the exercise oneself. It reminded me of the benefitsof an active bench when presenting an oral argument. I alwaysview arguments before a “dead bench” as a wasted opportunity.

There are a few drawbacks to this method, but these largelycan be addressed with careful planning by the advocate. Not allexpert witnesses will be up to the challenge of engaging in twodays of vigorous debate with the court and the opposing expert.Moreover, those experts who suffer from arrogance will be at adisadvantage. A willingness to concede weak points or to creditthe opposing expert will be essential to obtaining the trust andrespect of the court. Thus, as always, one must choose yourexpert witness wisely. The other potential drawback is that not alljudges would have the resources to run an effective hot tub.However, there are many different ways to structure the hot tub—they need not all be a two-day grilling of an expert panel.

In my view, any opportunity for the court to engage directly withan expert should be encouraged, but the advocate must know herexpert and judge when recommending a particular type of hot tubexercise.

The increased use of hot tubs both here and abroad shouldencourage additional judges and litigants to experiment with this


3 6 · A N T I T R U S T

ANTITRUST IN ASIAMay 31–June 1, 2018 � Seoul, Korea

www.ambar.org/atasia

See page 88 for more details

S P R I N G 2 0 1 8 · 3 7

Battle of the Experts in Merger Litigation:Two Case StudiesB Y D E B B I E F E I N S T E I N A N D W R E D E S M I T H

FOR MERGER PRACTITIONERS, ONEof the highlights of a merger trial or preliminaryinjunction hearing is the economists’ testimony. Itis a chance to hear the experts provide an overviewof the entire case and is familiar ground for anti -

trust practitioners, who are well-versed in the economic ter-minology—HHIs (Herfindahl-Hirschman index), SSNIPs(small but significant and non-transitory increase in price),diversions and regressions, to name a few. In contrast, judges in merger trials or preliminary injunc-

tion hearings have a short time period to get up to speed onthe industry, antitrust law, and antitrust economics. They areimmediately faced with conflicting testimony on complicat-ed analyses that, in all likelihood, are unfamiliar to them.How they deal with the “battle of the experts” can often bedispositive to the outcome of a case. In this article, we exam-ine the role expert testimony and analysis played in thecourts’ opinions in the Sysco and Advocate matters.

SyscoIn February 2015, the FTC, 10 states, and the District ofColumbia filed a complaint to enjoin the Sysco/US Foodstransaction.1 The FTC alleged that post-merger, Sysco andUS Foods would have a dominant market share for broadlinefoodservice distribution in the United States and in 32 localmarkets. The court agreed with the FTC, enjoining the merg-er. The economic analysis and testimony of the FTC’s expert,Dr. Mark Israel, was key to the FTC’s arguments in everyfacet of the case. Defendants’ experts attempted to refutevirtually every point Dr. Israel made, and the court oftencredited their arguments. Ultimately, however, the districtcourt largely adopted Dr. Israel’s analyses.

The FTC’s Allegations. As is typical in merger litiga-tions, the FTC in Sysco began with market definition, defin-ing two relevant product markets: (1) broadline foodservicedistribution services and (2) broadline foodservice distribu-

tion services sold to national customers. The FTC definedbroadline foodservice distribution services as a distinct formof foodservice distribution not reasonably interchangeablewith other forms of foodservice distribution.2 Specifically,broadline distributors provide a breadth of products andservices that other foodservice distributors do not provide,including broad geographic coverage through a network ofdistribution centers, large product portfolios in a variety ofcategories, private-label products at lower cost than brandeditems, frequent and flexible delivery schedules, and othervalue-added services such as nutritional information. The FTC alleged that broadline distributors were distinct

from other types of foodservice distributors, which include:(1) systems distributors, whose customers tend to be casualchain restaurants (e.g., Burger King, Wendy’s, and Apple -bee’s) that have limited or fixed menus and therefore demandfewer product items; (2) specialty distributors, which focuson distributing one or a small number of niche product categories, such as fresh produce or Italian food products, and serve customers such as independent restaurants; and (3) cash-and-carry stores like Restaurant Depot and clubstores like Costco and Sam’s Club, which do not offer deliv-ery services or sales representatives dedicated to individualcustomers and therefore do not appeal to larger customers.3

The FTC pointed to Brown Shoe’s “practical indicia,” or fac-tors, used to define the boundaries of a product market.4

One of the Brown Shoe factors is “distinct customers,” whichcourts have applied to include the existence of special class-es of customers.5

The second alleged relevant product market was broadlinefoodservice distribution services sold to National Customers.The FTC asserted that national customers, such as nationalhospitality chains and healthcare group purchasing organiza-tions, are distinguished by the desire for a broadline distribu-tor that can provide consistent products and services at all oftheir locations as well as centralized ordering and billing andvolume discounts.6 Further, the FTC argued that both partiescater to national customers through “national account” teamsdedicated to serving national customers and nationwide pric-ing and terms, among other benefits. The FTC asserted that the relevant geographic market

was “intertwined, and overlapping” with the relevant prod-

Debbie Feinstein is a Partner and Wrede Smith is an Associate in the Anti -

trust Group at Arnold & Porter Kaye Scholer LLP. Ms. Feinstein was the

Director of the Bureau of Competition at the U.S. Federal Trade Commis -

sion from June 2013 to March 2017. In that position, she worked on the

matters discussed in this article.

was the customer’s local area. For National Customers, theassessment relied on qualitative evidence that NationalCustomers negotiate contracts at the national level, broadlinedistributors have dedicated national sales teams, and region-al broadline distributors band together to provide offerings tonational accounts through groups like DMA (DistributionMarket Advantage), a cooperative of nine independentregional distributors established to compete for customerswith multi-regional distribution needs.The FTC’s expert also used quantitative evidence to deter-

mine geographic markets for local customers. In each localmarket, the evidence showed a “draw distance,” measured asthe radius from a distribution center that captured 75 percentof the sales of that distribution center.11 The draw distancewas used to identify “overlap areas” in which Sysco and USFoods both had distribution facilities. An analysis of eachoverlap area and the other broadline distributors that couldcompete in that overlap area showed that the merger wouldsubstantially lessen competition in 32 different local markets.This testimony also was essential to the FTC’s arguments

on competitive effects. Broadline distribution sales data fromdefendants and from third parties showed that the partieswould have a 71 percent post-merger share of sales to Nation -al Customers.12 Moreover, various iterations of market sharecalculations using sales data confirmed that (1) calculating theshares in different ways still resulted in high market sharesand (2) even accounting for the defendants’ proposed divesti-ture of 11 US Foods distribution centers to PerformanceFood Group, another broadline distributor, the market sharesremained problematic. Additionally, economic testimony indicated the defen-

dants were the two largest broadline distributors in theUnited States as measured by (1) broadline distribution rev-enue; (2) number of distribution centers; (3) size of deliveryfleet; and (4) size of salesforce. The FTC’s expert testified thateconomics teaches that in a bid market like broadline distri-bution to National Customers, “the terms offered by thewinning bidder are determined (or at least heavily influ-enced) by the capabilities of the second-best option for agiven buyer. . . .”13 Where, as here, a merger occurs betweenthe top two options for a given buyer, one must look at theprice and terms offered by the third-best option for thatbuyer. In this instance, the third-best option was far lessattractive, demonstrating that the merger would cause harmto National Customers. Similarly, data from US Foods’s ordinary course sales rep-

resentative reporting tool demonstrated that US Foods mostoften viewed Sysco as its largest competitor across localareas.14 Finally, an econometric event study based on Sysco’sentry into the market in Long Island critically showed thatthe entry resulted in a 1.4 percent decline in US Foods’ pricefor customers in that overlap area.15

Defendants’ experts, Dr. Jerry Hausman and Dr. TimothyBresnahan, critiqued every aspect of the FTC expert’s work.Regarding product market, they argued that the aggregate

3 8 · A N T I T R U S T


uct market of broadline foodservice distribution to NationalCustomers.7 The geographic market for National Customerswas the United States, and the parties had a combined 75 per-cent share of sales to National Customers. Furthermore, theFTC argued that Sysco and US Foods were the only two trulynational “broadliners,” with 72 and 61 distribution centers,respectively, compared to 24 distribution centers for the nextlargest broadliner.8 Additionally, the FTC asserted a separaterelevant geographic market for local broadline customers.The FTC argued that the merging parties had a combinedshare of greater than 50 percent in 32 local markets.9

Evidence at Trial and in Proposed Findings of Fact.At trial, the FTC relied heavily on Dr. Israel’s expert reportand testimony to support its case. His testimony outlined anumber of quantitative analyses, including the implementa-tion of an aggregate diversion analysis to conduct the hypo-thetical monopolist test. An aggregate diversion analysis usesgross margin to determine the percentage of customers thatwould need to stay in the market in the face of a price increaseto make the price increase profitable. The FTC’s expert useda gross margin of 10 percent, which resulted in a calculatedaggregate diversion ratio of 50 percent. He compared that fig-ure to the actual aggregate diversion percentage based ondata from defendants’ national and regional requests for pro-posals (RFP), bidding summary information and documents,and US Foods’ Linc database used by local sales representa-tives to track sales opportunities. Those data showed thatwhen one of the parties lost a bid or sale, the bid or sale wentto another broadline distributor over 70 percent of the time.10

Thus, because the actual aggregate diversion percentage wasgreater than the calculated aggregate diversion ratio, Dr.Israel concluded that a hypothetical monopolist in broadlinedistribution would find it profitable to impose a SSNIP, andbroadline distribution is a relevant product market. In addition to using economic tools, the FTC’s expert

provided testimony related to the relevant product marketbased on documents and testimony. For example, he walkedthrough each of defendants’ alleged alternatives to broadlinedistribution (system distribution, specialty distribution, andcash-and-carry stores) and testified that each was not a rea-sonable substitute for broadline distribution due to differingcharacteristics. The testimony supported analyzing the merg-er differently for different classes of customers. In particular,the evidence showed that National Customers have differentneeds, such as negotiating and working under one contractto cover all of the customers’ locations, than other types ofcustomers. Also, local broadline customers are different thanNational Customers, often making purchases without enter-ing into a contract and relying on regular contact with abroadline distributor’s sales representatives to negotiate pricesand place orders.Regarding geographic market, economic testimony bol-

stered the FTC’s argument that the geographic market forbroadline distribution sales to National Customers was theUnited States and the geographic market for local customers

S P R I N G 2 0 1 8 · 3 9

monopolist would profitably and separately impose at leasta SSNIP.”21 The court explained that markets in which onlya subset of customers could be targeted for price increases aretermed “price discrimination markets.”22 The court againfound the FTC expert’s hypothetical monopolist test to be supported by the weight of the evidence and found infavor of a product market of broadline distribution sales toNational Customers. The court found that the relevant geographic market for

broadline distribution sales to National Customers was theUnited States, pointing to the same evidence used to definethe relevant product market: that broadline distributors enterinto nationwide contracts with National Customers and haveteams devoted to such customers.Regarding broadline distribution in local markets, the

court called Dr. Israel’s use of draw distances to determineareas of competitive overlap a “practical approach and solu-tion to an otherwise thorny problem” in the absence of anindustry standard for defining markets at a local level.23 Thecourt noted that the approach generally takes into accountthat driving long distances to provide services has negativeimplications for distributors. It therefore concluded that rel-evant local geographic markets were areas of overlap result-ing from Dr. Israel’s 75 percent draw method. Turning to competitive effects, the court relied on the

FTC’s market share calculations due to the lack of industry-recognized standards for market shares for broadline sales toNational Customers. Noting that the FTC expert ran manyvariations on his market share calculations, the court foundthat the FTC did not need to present market share figures“with the precision of a NASA scientist.”24 The court wasmost convinced by the use of data collected from third par-ties to estimate the split of broadline distributors’ sales toNational Customers and to local customers. The court adopt-ed the assumption that the 16 largest broadline distributorshad the same national/local sales split, which resulted in amarket share of 59 percent for the defendants and an HHIincrease of 1,500 points. This led the court to conclude thatthe merger would result in a significant increase in marketconcentration for broadline sales to National Customers.For local markets, the court again relied upon the FTC

expert’s draw method to determine market shares. The courtagreed with the defendants that the 75 percent draw methodexcluded some competitor sales and did not reflect thenuances of each particular local market, but noted that theFTC expert again conducted a variety of market share cal-culations. Ultimately, the court found the FTC’s marketshare calculations to be informative but not conclusive evi-dence of competitive harm in local markets. Further more, thecourt found these figures to be corroborated by ordinarycourse documents and testimony and, as a result, found thatthe merger would lessen competition in local markets.After finding competitive harm based on high market

shares, the court analyzed the FTC’s evidence on unilateraleffects, one source of which was an empirical analysis of bid-

diversion analysis the FTC presented was conducted improp-erly and resulted in an overly narrow market definition. Theycontended that the formula used for the aggregate diversionanalysis was incorrect, as was the use of a 10 percent grossmargin. According to defendants’ experts, use of the correctformula and gross margin resulted in an aggregate diversionratio of over 100 percent, which showed that broadline dis-tribution is an overly narrow market regardless of the actualaggregate diversion percentage.16 Further, the defendants’experts testified that the data used to calculate the FTC’sactual aggregate diversion percentage was flawed. They testi-fied that the parties do not retain comprehensive RFP dataand that the RFP and bidding data providing the basis for theaggregate diversion percentage was created at the request ofthe FTC during the merger investigation. Finally, they arguedthat US Foods’ ordinary course sales representative reportingtool tracked opportunities rather than actual wins and loss-es and that neither data source describes whether the defen-dants lost business for price reasons or due to another factor.17

Regarding the geographic market, they argued that theFTC expert’s local geographic markets under-reported com-petitor sales. Regarding the unilateral effects, Dr. Bresnahanconducted a switching study showing that customers of oneof the merging parties that switched distributors moved to theother merging party much less frequently than would beexpected based on market shares.18

The Court’s Opinion. The court relied heavily on theFTC’s expert’s testimony in its ruling for the FTC. In itsdetermination that broadline foodservice distribution is arelevant product market, the court noted that “Dr. Israel’s tes-timony served two primary functions. First, he acted as a defacto summary witness, synthesizing the mass of testimonialand documentary evidence gathered by the FTC. . . . Second,Dr. Israel conducted a SSNIP test, using what is known as anaggregate diversion analysis.”19 Explaining that the summa-ry testimony mirrored the court’s discussion of testimonialand documentary evidence, the court focused on the FTC’saggregate diversion analysis. The court was persuaded by thearguments from defendants’ experts that the data the FTC’sexpert relied upon was not comprehensive and showedprospective, rather than actual, sales. Therefore, the court didnot rely on the precise percentages presented in the FTC’saggregate diversion analysis. The court noted, however, thatthe FTC expert’s conclusions better reflected the businessrealities of the food distribution market than those of thedefendants’ expert, based on testimony from market partic-ipants that other types of distributors do not constrain thepricing of broadline distributors.20

Continuing with its product market analysis, the courtaccepted that a product market could be defined by groupsof customers, noting that Section 4.1.4 of the HorizontalMerger Guidelines provides that “[i]f a hypothetical monop-olist could profitably target a subset of customers for priceincreases, the Agencies may identify relevant markets definedaround those targeted customers, to whom a hypothetical

ding opportunities to evaluate closeness of competition. Thecourt compared the FTC expert’s analysis to the defendants’expert’s switching study. The court explained that the FTCexpert’s analysis better captured actual competition betweenthe parties, used a more representative data set, and was cor-roborated by qualitative evidence that the defendants wereclose competitors, particularly for national customers. Finally, the court cited the FTC expert’s merger simulation

model, which used an auction model, based on the theorythat the winning bidder will offer price and service terms justgood enough to beat the second place bidder. Using the auc-tion model and factoring in the defendants’ proposed divesti-ture to Performance Food Group, the model demonstratedthat the merger would harm national customers by morethan $900 million annually.25 The court again acknowledgedthe defendants’ concerns about the RFP/bidding data used inthe merger simulation model, but found that the mergersimulation model supported a finding that the merger wouldsubstantially lessen competition in broadline distributionsales to National Customers.

FTC v. Advocate and NorthShoreThe FTC (joined by the State of Illinois) challenged thecombination of Advocate Health Care and NorthShore Uni -versity HealthSystem in late December 2015.26 The FTCalleged that Advocate and NorthShore were the two leadingproviders of general acute care inpatient hospital services inthe northern suburbs of Chicago, Illinois. The FTC allegedthat the combined firm would have 60 percent of the mar-ket, with the third largest competitor, Northwest Commu -nity, having only 14 percent.27 Initially, the district courtdenied the FTC’s motion for a preliminary injunction on thebasis that “destination hospitals” should be included in thegeographic market. The FTC appealed to the Seventh Cir -cuit, which reversed and remanded. On remand, the districtcourt conducted a full analysis of the FTC expert’s argu-ments and granted the FTC’s motion for a preliminaryinjunction.

The FTC’s Allegations. It was clear from the outset thatthe key battleground would be geographic market defini-tion, an issue on which the FTC had floundered in the past.28

More recent challenges in which the government had beensuccessful involved rural areas where there was little disputeabout the geographic market.29

The complaint did not define a precise geographic market.Instead, the FTC claimed the market “is no broader than theNorth Shore Area.”30 It pointed to case law holding that therelevant geographic market “‘need not be identified with‘scientific precision,’” but rather need only identify “in whichpart of the country competition is threatened.”31 In its open-ing brief, the Commission pointed to a number of factorssupporting its alleged geographic market:32

� The North Shore Area was largely co-extensive withNorthShore’s primary service area used in its ordinarycourse strategic analyses;

4 0 · A N T I T R U S T


� Evidence that patients strongly prefer to receive gener-al acute care services locally, including data on wherepatients go and testimony from the executives of themerging parties; and

� Documents and testimony that the hospitals in the areacompeted with each other and not with other hospitals.

Finally, the FTC argued that a hypothetical monopolist ofNorth Shore Area hospitals could profitably impose a SSNIP.It pointed to analysis by its expert, Dr. Steven Tenn, whofound a high level of intra-market diversion between NorthShore Area hospitals. With respect to competitive effects,along with documents and testimony, the FTC again point-ed to Dr. Tenn’s work. The FTC explained that he had con-ducted an analysis that showed that a significant number ofpatients view the merging parties as their first and secondchoices. Based on that analysis, Dr. Tenn estimated that post-merger prices at the defendants’ North Shore Area hospitalswould rise by an average of 8 percent.33

Defendants’ Daubert Motion. The defendants tookissue with Dr. Tenn’s work from their opening briefs. Thedefendants argued that Dr. Tenn used a novel approach forgeographic market definition that had no support in aca-demic literature or case law, arbitrarily excluded major com-petitors based on an unsupported view that NorthwesternMemorial, a hospital south of the North Shore Area, was a“destination” hospital, and also excluded hospitals simplybecause they competed with only one of the merging hospi-tals but not both or had less than 2 percent share in the market.34

The defendants’ economists critiqued Dr. Tenn’s modelshowing a price increase as failing to measure actual substi-tution between the merging firms. The defendants’ econo-mists claimed they had faithfully applied the FTC’s normalmethod for assessing price increases—which Dr. Tenn failedto use—and found the merger would have no statistically sig-nificant effect on price. In addition, one of the defendants’economists, Dr. Thomas McCarthy, contended that Dr. Tennfailed to account for repositioning that would further makehis finding of a price increase implausible.35

In a move somewhat unusual for merger litigation, thedefendants moved to exclude Dr. Tenn’s testimony altogeth-er, by making a Daubert motion and reiterating the criti-cisms of Dr. Tenn’s work they had made in their openingbrief. The court refused to exclude Dr. Tenn’s testimony,finding that Dr. Tenn properly constructed his geographicmarket using the hypothetical monopolist test. The courtnoted that he analyzed admissions to all of the hospitals in theChicago area and calculated diversion ratios for all of thosehospitals. The court thus concluded that “his hypotheticalmonopolist analysis accounts for competition from all areahospitals, not just those that are included in his proposed geo-graphic market.”36

On the question of whether Dr. Tenn improperly exclud-ed “destination hospitals,” the court found that to be a topicfor cross-examination, not a basis to exclude. Finally, the

S P R I N G 2 0 1 8 · 4 1

court rejected the defendants’ argument that Dr. Tenn’smodel had never been used before, holding that neither theMerger Guidelines nor the academic literature suggest thereis only one way to conduct a merger simulation.37

Evidence at Trial. Based on the pre-trial proceedings, itappeared that the stage was set for a close look at the eco-nomic evidence at trial. Indeed, there was substantial testi-mony at trial from Dr. Tenn (as well as that of the defendants’experts). Much of that testimony was highlighted in theFTC’s post-trial brief. The FTC pointed to Dr. Tenn’s testi-mony on both geographic market definition and competitiveeffects, noting:

� Dr. Tenn’s empirical analysis that patients at North -Shore Area hospitals traveled only short distances forhospital services;

� Dr. Tenn’s conclusion that a hypothetical monopolistowning only the six merging hospitals could impose aSSNIP and that the North Shore Area market (whichincluded five non-party hospitals) was therefore con-servative;

� Dr. Tenn’s analysis showing that there is a significantlevel of substitution between Advocate and NorthShorehospitals based on the hospital choice model (alsoDefen dants’ expert’s preferred method to calculatediversions); and

� Dr. Tenn’s willingness to pay analysis quantifying thatthe merger would result in increased prices.38

The defendants’ brief highlighted a few of their many criti-cisms:

� Dr. Tenn’s model made no sense—showing that 52percent of patients who choose hospitals in the TennNorth Shore Area would divert to a competing hospi-tal outside the area in the absence of their first choice;

� Dr. Tenn arbitrarily excluded certain hospitals, as notedabove;

� Dr. Tenn’s price increase model had never been usedbefore, always predicted a price increase, and failed toaccount for insurers’ bargaining leverage; and

� The defendants’ expert’s accepted model showed themerger is not likely to lead to a material price increase.39

The District Court’s Opinion.The district court’s opin-ion denying the FTC’s preliminary injunction motionfocused almost entirely on Dr. Tenn’s geographic marketanalysis. The court rejected the method by which Dr. Tennhad constructed the geographic market. As foreshadowed bythe court’s Daubert opinion, the question of whether desti-nation hospitals could be excluded from the geographic mar-ket turned out to be critical. The court’s main holding wasthat there was no economic basis for excluding “destinationhospitals.”40 The court found that Dr. Tenn’s rationale forexclusion—that they are not substitutes for Advocate andNorthShore—assumed the answer to the very question thegeographic market exercise is designed to elicit. The courtfound that Dr. Tenn’s assumption that patients like to receivecare close to home was not supported by the evidence, point-

ing to various testimony that suggested that patients traveledlonger distances for outpatient care.The court similarly criticized Dr. Tenn’s requirement that

only hospitals competing with both hospitals should beincluded in the geographic market, noting Dr. McCarthy’stestimony that “you can constrain the postmerger system byconstraining any [one] of its hospitals.”41 This was one of therare times the court pointed to specific evidence from thedefendants’ expert in the opinion. The court also found that Dr. Tenn’s exclusion of desti-

nation hospitals “ignores ‘the commercial realities of th[is]industry,’” specifically that: (1) payers negotiate a single contract for both inpatient and outpatient services; (2) outpa-tient services are on the rise and inpatient services are on thedecline; and (3) outpatient services are a key driver of hos-pital admissions.42

The Seventh Circuit Opinion. The Commissionappealed to the Seventh Circuit which reversed and remand-ed. Once again, the focus was on Dr. Tenn’s analysis. TheSeventh Circuit found that the district court mistook Dr.Tenn’s iterative approach to defining the market for circu-larity and that the court offered no explanation for why a nar-row candidate market would ultimately produce narrowresults.43 With respect to Dr. Tenn’s exclusion of destinationhospitals, the Seventh Circuit noted that demand for thosehospitals differs from demand for general acute care hospitalslike the parties’ hospitals. Next, the Seventh Circuit assessedDr. Tenn’s determination that patients choose hospitals closeto home. The Seventh Circuit found that the evidence onwhich the district court relied related to outpatient care, notinpatient acute care.44 Finally, it turned to Dr. Tenn’s diver-sion ratios. It noted that as to that evidence, the district courthad incorrectly focused on patients and not what insurerswould do. It concluded by asking “how many hospitals caninsurers convince most customers to drive past to save a fewpercent on their health insurance premiums. We should notbe surprised if that number is very small. Plaintiffs have madea strong case that it is.”45

The District Court’s Opinion on Remand. In a remandopinion almost three times the length of the original districtcourt opinion, the court found for the FTC. After recitingsome of the procedural history with respect to geographicmarket, the court began by finding that “even taking theirtestimony with a grain of salt, the record as a whole supportsthe view that insurers genuinely believe that a plan thatexcludes Advocate and NorthShore is not viable in the NorthShore Area.”46 The court noted that even if “some—or evenmany—patients are willing to travel outside the market,”the defendants must show that enough would do so thatinsurers are unlikely to agree to pay supracompetitive pricesto be able to offer a plan to patients unwilling to travel.47

The court next turned to Dr. Tenn’s reliance on diversionratios. The parties argued that if the district court erred byrelying on diversion ratios because they focused on patients,not insurers, then Dr. Tenn’s analysis based on diversion

bolsters the presumption of anticompetitive effects and Dr.Tenn’s conclusions.”54 But the court also considered whetherthe economic arguments made logical sense or were coun-terintuitive in cases where there was no ordinary course evi-dence available, as with the question of whether prices wouldlikely increase.

ConclusionIn both cases, and particularly in the remand decision inAdvocate, the courts considered the economic evidence indetail. In Sysco, the court went so far as to discuss the appro-priate aggregate diversion equation to use in this market. InSysco, more so than in the Advocate remand, the court cred-ited the defendants’ experts’ arguments. Both cases also makeclear that courts will dig into the complicated economicissues when necessary to evaluate the expert testimony. Thus,parties should not shy away from putting in detailed eco-nomic arguments and analyses when economists can clearlyexplain them. In the end, however, both courts found—repeatedly—

that the FTC’s expert’s views were more consistent with theordinary course evidence. This underscores the importance ofeconomists painting an overall picture of the industry andhow economic facts fit into that competitive picture. Usingempirical analyses is typically more persuasive than anabstract model. And it is also important for the economiststo grapple with contrary facts, documents, and testimony.Expert testimony will only be persuasive if it is consistentwith the ordinary course evidence and is not based onabstract economic theory.�

1 Complaint for TRO and Prelim. Inj. Pursuant to Section 13(b) of the Fed.Trade Comm’n Act at 14, FTC v. Sysco Corp. et al., No. 1:15-cv-00256(D.D.C. Feb. 20, 2015) [hereinafter Sysco Complaint].

2 Id. ¶ 32. 3 Id. ¶¶ 36–38. 4 Brown Shoe Co. v. United States, 370 U.S. 294 (1962). 5 See, e.g., United States v. Grinnell Corp., 384 U.S. 563, 571–75 (1966); see

also U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guide -lines § 3 (2010), http://ftc.gov/os/2010/08/100819hmg.pdf (“Whenexamining possible adverse competitive effects from a merger, the Agenciesconsider whether those effects vary significantly for different customers pur-chasing the same or similar products.”).

6 Sysco Complaint, supra note 1, ¶¶ 32–33. 7 Mem. in Supp. of Pl. Fed. Trade Comm’n’s Mot. for TRO and Prelim. Inj. at14, FTC v. Sysco Corp., No. 1:15-cv-00256 (D.D.C. Mar. 5, 2015).

8 Sysco Complaint, supra note 1, ¶ 78. 9 Id. ¶ 60.

10 See FTC v. Sysco Corp., 113 F. Supp. 3d 1, 34–35 (D.D.C. 2015). 11 Id. at 50. 12 Id. at 53–54. 13 Pl.’s Corrected Proposed Findings of Fact and Conclusions of Law ¶ 496,

FTC v. Sysco Corp. et al., No. 1:15-cv-00256 (D.D.C. May 26, 2015). 14 Id. ¶¶ 520–521.

ratios must be entirely ignored. The court found it appro-priate to consider the diversion ratios because they showedwhether a price increase would drive fewer patients outsidethe merged entity because they would flee to another merg-ing hospital.48 Finally, the court considered whether Dr. Tennshould have included “closer substitutes” and rejected thatargument. The court noted that inclusion of several of thosehospitals would not change the high post-merger concentra-tion and, in particular, that Northwestern Memorial is a des-tination hospital that cannot fulfill payors’ need for a hospi-tal providing local care in the North Shore Area.49

On remand, the district court examined for the first timeDr. Tenn’s opinion on the likely anticompetitive effects of themerger. It did so despite determining that it need not addressdefendants’ criticisms given the presumption of anticompet-itive effects and the voluminous factual evidence “showingthat Advocate and NorthShore are close competitors whodominate the North Shore Area.”50

Dr. Tenn used a model that showed that prices wouldincrease as a result of the merger. The defendants’ economistsargued that the model was flawed because it (1) predicted aprice increase every time; (2) did not take into account vol-ume loss; and (3) was inconsistent with the way hospitalsengaged in bargaining. The court methodically addressedeach of the defendants’ economists’ arguments and foundtheir criticisms of Dr. Tenn’s anticompetitive effects analysisto be “desultory and superficial, and therefore unconvinc-ing.”51 The court summed up that “Dr. Tenn has persuasivelydemonstrated that the merger is likely to cause a significantprice increase.”52

Next the court turned to the defendants’ economists’rebuttal of anticompetitive effects. Again, the court detailedthe various arguments made by both sets of economists. Thedefendants’ economists calculated that the merger wouldlead to a price decrease or at worst a small price increase. Dr.Tenn’s argued that defendants’ results were “implausible”because they suggested that prices decreased as bargainingleverage increased. The court found that analysis convincing.Finally, the court found that it could not accept the defen-dants’ arguments that repositioning of competitors wouldoffset the anticompetitive effects without stronger evidencethan the generalized testimony the defendants offered.53

In this case, the court again relied on ordinary course evi-dence that supported the FTC economist’s analysis. It notedthat the record as a whole showed that patients prefer hospi-tals close to their home, and that insurers believe that a planthat excludes both merging parties’ hospitals is not viable inthe North Shore Area. Both points supported Dr. Tenn’sgeographic market analysis. In assessing defendants’ critiquesof Dr. Tenn’s assessment that prices would increase, the courtfaulted the defendants for “ignor[ing] the voluminous factu-al evidence, including ‘ordinary course’ documents, showingthat Advocate and North Shore are close competitors whodominate the North Shore Area and whose merger wouldnecessarily entail substantial harm to competition, which


4 2 · A N T I T R U S T

S P R I N G 2 0 1 8 · 4 3

52 Id.53 Id. at *12. 54 Id. at *8.

15 Id. ¶¶ 523–524.16 FTC v. Sysco Corp., 113 F. Supp. 3d at 36.17 Id.18 Id. at 63. 19 Id. at 35. 20 Id. at 37. 21 Id. at 38 (quoting U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal

Merger Guidelines § 4.1.4 (2010)). 22 Id. at 39. 23 Id. at 52. 24 Id. at 54. 25 Id. at 66. 26 In the same month, the FTC also challenged the Hershey/Penn State trans-

action and the Cabell/Huntington transaction in West Virginia. That threehospital transactions were filed in such close proximity was a coincidenceof timing. No hospital transactions had been filed between mid-2013 and2015 and there have been no hospital transactions filed since.

27 Mem. in Supp. of Pl. Fed. Trade Comm’n’s Mot. for a Prelim. Inj. at 20, FTCv. Advocate Health Care Network et al., No. 15-cv-11473 (N.D. Ill. Mar. 9,2016).

28 See, e.g., FTC v. Tenet Health Care Corp., 186 F.3d 1045 (8th Cir. 1999); Cal.v. Sutter Health Sys., 84 F. Supp. 2d 1057 (N.D. Cal. 2000), aff’d, 217 F.3d846 (9th Cir. 2000), as amended, 130 F. Supp. 2d 1109 (N.D. Cal. 2001).

29 See, e.g., ProMedica Health Sys., Inc. v. FTC, 749, F.3d 559, 565 (6th Cir.2014) (parties agreed that the relevant geographic market was LucasCounty, Ohio).

30 Complaint for TRO and Prelim. Inj. Pursuant to Section 13(b) of the Fed.Trade Comm’n Act at 12, FTC v. Advocate Health Care Network et al., No.15-cv-11473 (N.D. Ill. Dec. 22, 2015).

31 Mem. in Supp. of Pl. Fed. Trade Comm’n’s Mot. for a Prelim. Inj. at 13, FTCv. Advocate Health Care Network et al. (quoting FTC v. Cardinal Health, Inc.,12 F. Supp. 34, 49 (D.D.C. 1998)).

32 Id. at 14–19. 33 Id. at 28. 34 Def.’s Opp. to Pl.’s Mot. for a Prelim. Inj. at 9–10, FTC v. Advocate Health

Care Network et al., No. 15-cv-11473 (N.D. Ill. Mar. 23, 2016). 35 Id. at 24–26. 36 Order on Def.’s Mot. to Exclude Tenn’s Testimony at 2, FTC v. Advocate

Health Care Network et al., No. 15-cv-11473 (N.D. Ill. Apr. 6, 2016). 37 Id. at 3. 38 Pl.’s Post-Hearing Brief, at 1, 14, FTC v. Advocate Health Care Network et

al., No. 15-cv-11473 (N.D. Ill. May 24, 2016). 39 Def.’s Post-Hearing Mem. in Opp. to Pl.’s Mot. for Prelim. Inj. at 9, 14,

18–19, FTC v. Advocate Health Care Network et al., 2016 WL 3387163, at*4 (N.D. Ill. May 20, 2016) (No. 15-cv-11473).

40 FTC v. Advocate Health Care Network et al., 2016 WL 3387163, at *4 (N.D.Ill. June 20, 2016).

41 Id. at *5. 42 Id. at *4. 43 FTC v. Advocate Health Care Network, 841 F.3d 460, 464 (7th Cir. 2016).44 Id. at 474. 45 Id. at 476. 46 FTC v. Advocate Health Care Network, 2017 WL 1022015, at *5 (N.D. Ill.

Mar. 16, 2017). 47 Id. at *5. 48 Id. at *6. 49 Id.50 Id. at *8. 51 Id. at *10.

� FEW INDUSTRIES IN THE UNITED STATEShave received more attention from the popular press andantitrust authorities than health care. Whether in responseto business or legislative and regulatory pressures, healthcare companies have increasingly looked to transactionsas a means to achieve cost savings, quality improvement,and population health management. At the same time, theantitrust agencies have promised robust enforcement inhealth care markets. The collision of these forces led tomany significant legal developments in recent years, andalso made producing an up-to-date treatment of theseissues challenging.The revised and updated Health Care Mergers and

Acquisitions Handbook, Second Edition provides guidanceon how courts and the U.S. antitrust agencies analyze andreview mergers in the health care industry. This handbookaddresses the issues that antitrust practitioners and their clients face in planning, executing, and litigatingtransactions in the health care field. Specifically, this handbook details antitrust agency policy and caselaw with respect to hospital, physician practice, and health plan transactions.



Health Care Mergers andAcquisitions Handbook Second Edition




4 4 · A N T I T R U S T

sideration is to understand the differences between the audi-ence at the agencies compared with nearly all courts. Bothfederal antitrust agencies employ large staffs of experiencedand dedicated economists, the Bureau of Economics at theFederal Trade Commission and the Economic AnalysisGroup at the Antitrust Division of the Department of Justice.These in-house economists, a subset of whom is assigned toevery significant merger or non-merger investigation, serve acritical role in both agencies’ decision-making. In addition, the agencies often retain outside economic

experts relatively early in the investigative process, particular-ly for cases that may well end up in litigation. This was cer-tainly the case during our tenure. As with the legal staff at eachagency, these economists typically specialize in particularindustries, with their experience from prior cases informingtheir approach to new matters. As a result, agency econo-mists often expect parties to engage in sophisticated quan ti-tative analysis, and the legal staff and front-office managementoften rely heavily on economic analysis, including the con-clusions and recommendations of economic staff. The 2010 Horizontal Merger Guidelines reflect this heavy

focus on economic analysis in agency merger reviews, withsuch increased emphasis on economics one of the drivingforces for the 2010 revisions to the previous Guidelines.2 TheMerger Guidelines make clear that, while market structureand concentration certainly can play a role, the “measurementof market shares and market concentration is not an end initself, but is useful to the extent it illuminates the merger’slikely competitive effects.”3 The Guidelines outline the typesof quantitative analyses that are frequently employed in merg-er investigations, including diversion ratios, measures ofupward pricing pressure, merger simulation models, and crit-ical loss analyses. At least when considering potential unilat-eral effects, and to the extent the necessary data are available,the agencies “rely much more on the value of diverted salesthan on the level of the HHI for diagnosing unilateral priceeffects in markets with differentiated products.”4

Consistent with the 2010 Guidelines, in our experience theFTC and DOJ routinely rely on similarly sophisticatedeconometric models in merger reviews, including at the crit-

IT IS NEWS TO FEW THAT SOPHISTICATEDeconomic analysis has become an important compo-nent of antitrust practice, whether in advocacy beforethe federal and state antitrust agencies or in govern-ment or private antitrust litigation. Regardless of the

forum, credible and persuasive economic analysis can repre-sent the critical ingredient in evaluating the competitive con-sequences of a proposed merger or business conduct or inassessing the appropriate remedy for an antitrust violation.Indeed, economic analysis plays a central role in nearly everysignificant antitrust matter, even when the ultimate decisionmay emphasize other sources of evidence.1

As a practical matter, however, the role and significance ofeconomic analysis can vary dramatically depending on thesetting. An approach tailored to persuading the agencies maybe unsuccessful, or even in some cases counterproductive, ifapplied in the courtroom, and vice versa. In our experience,including our time at the FTC, we have observed many mat-ters where economic analysis was pivotal in securing a suc-cessful resolution, along with plenty of others (including,candidly, some our own cases while in private practice) wherea different approach might have proven more effective. In this article, we offer our perspective on how economic

analysis is most effectively utilized in antitrust advocacybefore the antitrust agencies and the courts, with the upshotbeing that a party’s economic analyses are much more likelyto be persuasive to the agencies than a federal court, especiallyin merger matters.

Economics at the Antitrust AgenciesIn deciding how and when to present economic analysis tosupport antitrust advocacy, the single most important con-

The First Cut Is the Deepest:Use of Economics Before the

Antitrust Agencies and the CourtsB Y M I C H A E L J . P E R R Y A N D S T E P H E N W E I S S M A N

Michael J. Perry and Stephen Weissman are partners at Baker Botts LLP

in Washington, D.C. The authors previously served as Counsel to the

Director (2014–2016) and Deputy Director (2013–2015), respectively, of

the Federal Trade Commis sion’s Bureau of Competition. They would like to

thank Tom Carter, a Baker Botts associate, for his contributions to this

article.

S P R I N G 2 0 1 8 · 4 5

ical early stages of determining whether to issue a SecondRequest or, in conduct cases and non-reportable deals, toopen a full-phase investigation. In many cases, the agenciescan be persuaded that a transaction is unlikely to harm com-petition even when market structure and concentration wouldexceed the thresholds for a presumption of harm under theMerger Guidelines and case law.5

Although the U.S. agencies do not publish their rationalefor clearing a transaction very often, there are several promi-nent examples documented in public statements. In theFTC’s review of Dollar Tree’s acquisition of Family Dollar,for example, the Commission relied on a sophisticatedeconometric model as a primary means to identify the par-ties’ overlapping stores presenting significant competitiveconcerns. As then-Chairwoman Edith Ramirez explained ina speech discussing the case, the Commission

used an econometric model that allowed us to assess thelikely impact of the transaction by taking into account, forany given local geographic area, the varying degrees of mar-ket power-constraining influence exerted by various retailersin that area. The model showed that in some geographicareas, the only retailers exerting a significant influence wereWalmart and other dollar stores such as Dollar General. Inother geographic areas, supermarkets or pharmacies, by virtueof their proximity and number, also exerted a significantinfluence. This approach produced a sophisticated under-standing of likely competitive effects in part because it didnot assume that a particular retailer fell inside or outside ofany relevant market.6

In other words, the Commission was willing to look beyondany strict definition of the relevant market to evaluate thetransaction’s likely competitive effects.Similar kinds of econometric analysis also played mean-

ingful roles in the outcome of other agency investigations. Inone consumer products transaction cleared with divestituresby the Commission during our tenure at the Commission,econometric price modeling presentations by the parties’economists were influential in a close-call decision to acceptthe parties’ proposed remedy. In our view, those economicanalyses showed that, when accounting for dynamic effi-ciencies resulting from the transaction, the transaction wasunlikely to have price effects in any relevant market. In a more recent transaction, before the Department of

Justice, in which we represented the acquirer of a competitor’ssoftware business serving the real estate industry, the in-depth analyses that our economist team presented aboutswitching to alternative products and the lack of head-to-head pricing between the merging parties undoubtedlyhelped our client clear the transaction at the front officedespite significant concerns expressed by staff. Nor is the agencies’ reliance on sound economic analyses

a recent phenomenon. Perhaps the poster child for the agen-cies’ affording substantial weight to economic modeling is theCommission’s 2002 decision to clear the merger betweencompeting cruise line operators. There, in its closing state-ment, a majority of the Commission voted to clear the trans-

action even though it would “significantly increase concen-tration in a market already highly concentrated,” becausethe evidence did not “support any theory of anticompetitiveeffects.”7 As the then-Director of the Bureau of Economicsexplained: “Basic data analysis can make a difference in deci-sion making.”8

The Commission’s heavy reliance on econometrics inDollar Tree/Family Dollar and the other examples mentionedabove illustrate a broader and more fundamental point,which is that, when data is available, the agencies are gener-ally quite receptive—and often eager—to receive sophisti-cated econometric studies and other quantitative analysesfrom the parties in a merger or conduct investigation. Andthis type of evidence can persuade the agency to clear a trans-action or close a conduct investigation even when the struc-tural evidence may suggest cause for concern. Because, as detailed in the following section, courts con-

tinue to rely more heavily on qualitative evidence, such asinternal party documents and testimony of industry partici-pants, than on economic evidence, the best and sometimesonly meaningful opportunity to use economic analysis togain clearance for an otherwise problematic transaction isbefore the agencies. Thus, there can be little question that thefirst cut is the deepest (to borrow a phrase from the CatStevens song popularized by Sheryl Crow) when it comes topresenting sophisticated economic analysis.

Economics in the CourtsBy contrast to the expert audience at the antitrust agencies,most judges and nearly all jurors have relatively little priorexperience with economics or antitrust law. Even in courtsthat have a relatively large number of antitrust cases, antitrustgenerally comprises a tiny fraction of the overall docket. Thisrelative lack of specialized experience does not necessarilymean that economic analysis is not relevant in litigation, orthat judges and juries are incapable of understanding sophis-ticated economic concepts. In fact, our generalist court sys-tem is remarkably adept at learning and applying specializedareas of the law, including antitrust. Furthermore, in ourexperience, most judges assigned complex antitrust casesdevote considerable effort to familiarize themselves with theeconomic analysis. But courts generally evaluate the eco-nomic evidence within the broader evidentiary record, typi-cally relying more heavily on documents, testimony fromfact witnesses, and other qualitative evidence.In the FTC’s challenge to the proposed Sysco/US Foods

merger, for example, the court considered each side’s eco-nomic expert testimony in considerable detail, but the courtultimately resolved the technical and methodological dis-agreements among the economists by turning to othersources of evidence. Faced with “competing expert testi-monies” that came to opposite conclusions on market defi-nition, the court considered each “in light of the evidentiaryrecord as a whole” and concluded that the economic analy-sis of the FTC’s expert was “more consistent with the busi-

4 6 · A N T I T R U S T


ness realities of the food distribution market” than that of thedefendants’ expert.9 The court also focused its analysis ofthe relevant market on the “practical indicia” outlined inUnited States v. Brown Shoe, quipping that the “Brown Shoepractical indicia may indeed be ‘old school’ . . . [but] BrownShoe remains the law, and this court cannot ignore its dic-tates.”10

Similarly, in both of the recent DOJ challenges to pro-posed health insurance mergers, the courts considered eco -nometrics and other quantitative evidence but generallyemphasized qualitative evidence and the Brown Shoe fac-tors. In Anthem/Cigna, for example, the court resolved con-flicting economic expert testimony by focusing on “realworld evidence.”11 In evaluating competing methodologiesfor performing diversion analysis, the court emphasized theperceived tension between the “economic assumptionsunder lying the various methodologies” and the “internalcommunications that shaped and chronicles these events inreal time,” concluding that “Anthems’ ordinary course doc-uments tell a consistent story that contravenes the firm’s lit-igation position.”12 Likewise, although the court in Aetna/Humana gave significant weight to customer switching data,it evaluated this evidence within the Brown Shoe frame-work.13 And the Aetna court rejected what it perceived to bea “technical” and “purely econometric approach to marketdefinition,” concluding that econometric evidence must beconsidered in light of “the Brown Shoe factors and ordinarycourse of business documents.”14 Thus, in each of theserecent litigated merger challenges, the court emphasizeddocuments, testimony, and other forms of qualitative evi-dence, even when evaluating the quantitative evidence. Perhaps most dismissive of the role of economics in a

merger litigation was the Ohio federal court’s decision not tohear any economic evidence at all in FTC v. Steris/Synergy.15

In that case, which was during our tenure, the Commissionsued to preliminarily enjoin a merger on the theory that thetransaction would eliminate likely future competitionbetween an incumbent provider of sterilization services formedical devices and a future rival poised to introduce a bet-ter competing technology. Prior to the hearing, the courtstrongly suggested that the parties focus their presentationsexclusively on factual evidence of whether the alleged poten-tial entrant was likely to enter the relevant market absent thetransaction. The court, which denied the FTC’s request fora preliminary injunction, saw no need to entertain econom-ic evidence about market definition or industry conditionsthat made it attractive or unattractive for new entry. To be sure, there are exceptions to this general pattern of

courts’ seeming reluctance to hinge their decisions in signif-icant part on economics. In Oracle/Peoplesoft, for example, thecourt cautioned that “strong presumptions based on meremarket concentration may be ill-advised in differentiatedproducts unilateral effects cases,” noting that “modern econo-metric methods” such as merger simulation models may bemore appropriate in these circumstances.16

More recently, in the Third Circuit’s decision reversing thedenial of a preliminary injunction in a FTC hospital mergerchallenge, the court engaged in a detailed analysis of theappropriate methodology for applying the hypotheticalmonopolist test to define the relevant market for health careservices.17 These examples illustrates that federal courts canand do grapple with sophisticated economic models andconcepts. But there is no question that the courtroom is afundamentally different forum than the antitrust agencies,and economic analysis needs to be tailored appropriately toaccount for these inherent differences.

The Problem with Keeping Your Powder DryMost antitrust practitioners understand that the agencies areoften more receptive to econometric studies, economic mod-els, and other sophisticated forms of economic analysis. Butin our experience, in practice advocates often fail to take fulladvantage of the opportunity to use these tools effectively.Most notably, we have observed numerous instances inwhich persuasive economic analyses were presented for thefirst time in litigation. Undoubtedly, this pattern may beattributed to practical realities—for example, many clientsmay be hesitant to invest in in-depth economic analysis earlyin an investigation, and often the data necessary to performrobust studies may not be available to the parties before theagencies are forced to produce third-party sources in litiga-tion. Like wise, parties may be hesitant to present an eco-nomic approach that differs significantly from the standardmodels preferred by agency economic staff, or to appear toendorse the agency’s preferred approach. Despite these lim-itations, however, in most cases we believe there is an oppor-tunity to develop and present persuasive economic analysesearlier in the investigative process.

Effective Use of Economics Before the Agenciesand the CourtsWith these general observations in mind, how can antitrustpractitioners more effectively utilize economic tools in advo-cacy before both the agencies and courts? Below, we outlinesome suggestions for both settings.In seeking to persuade the agencies, in our view the most

important consideration is the importance of earnestly engag-ing with agency economic staff. In many types of transac-tions, economic staff will approach their analysis with well-defined approaches and models. Although there may becompelling reasons why alternative models may accuratelypredict competitive effects in a particular transaction, realis-tically there is usually a much greater chance of persuadingagency economic staff using their default framework. To theextent there may be a strong case for tweaking or refining thisapproach to better fit the circumstances of a particular trans-action, in our experience it is generally most useful to focuson a small number of incremental refinements to the stan-dard model. And, even if there is a strong case to apply a dif-ferent framework, agency staff is most likely to be persuad-

S P R I N G 2 0 1 8 · 4 7

ed where the parties are able to demonstrate that anticom-petitive effects are unlikely even using the agency’s standardapproach. We do note one important factor to consider in adopting

the agency’s preferred framework during the investigativestage. If the matter ends up in litigation, litigation staff mayattempt to use the parties’ good-faith efforts to engage withstaff offensively, as evidence that even defendants’ ownexpert(s) admit the transaction may harm competition. For example, a GUPPI analysis or merger simulation

model presented during the investigation may show somepotential for upward pricing pressure or price effects. Indeed,these tools can generate superficially adverse results in near-ly every merger among competitors, and courts may notappreciate that this inherent feature of this type of analysisshould not be interpreted as evidence the merger is likely tosubstantially lessen competition. In our view, however, this risk is manageable, and it should

not deter engagement with agency economic staff on theirterms. But it is important to keep this possibility in mindwhen presenting such economic analysis, including appro-priate caveats explaining that this analysis is presented forillustrative purposes and does not represent an endorsementof the analytical approach. In some cases, there may also bevalue in using a different economic expert in litigation, par-ticularly if the parties intend to advance a fundamentallydifferent approach in that setting. Recognizing the importance of engaging with agency

economic staff, it is equally important to understand the lim-its of economic analysis within the broader range of evidencethat the agencies will evaluate in reviewing a proposed trans-action. Despite the growing role of quantitative analysis,the agencies of course continue to rely heavily on the parties’ordinary-course-of-business documents and the perspectiveof customers and other market participants. In presentingeconomic analysis, it is critical to remain mindful of theseother sources of evidence, which—if they appear to contra-dict the inferences of the economic work—may cast seriousdoubt on the reliability or relevance of the quantitativeanalysis. In one recent transaction we defended, for example, we

presented what we believe to be compelling economic evi-dence that the parties were not especially close competitors.Though staff could not dispute the results of this analysis,internal documents appearing to show intense competitionbetween the parties made staff skeptical of the conclusions wewere attempting to draw from the quantitative analysis.Although we were ultimately successful in persuading theagency to clear the transaction, we were not able to do sountil we compiled a robust record of ordinary-course docu-ments and third-party evidence corroborating our position.This factor—presenting economic analysis that is consis-

tent with the other sources of evidence—is important in anysetting, but of paramount importance once the advocacytransitions to the courtroom. Whereas agency staff are gen-

erally willing to look past a few bad documents or vocal cus-tomer opposition if the balance of evidence indicates thetransaction is unlikely to harm competition, the same type ofevidence can be fatal in litigation if not handled appropri-ately. In our experience, most judges use economic analysisat most as a supplement to, or interpretive guide for, theother, more familiar sources of evidence. If an economicexpert’s testimony does not fit closely with the qualitative evi-dence, the court is unlikely to afford it much, if any weight.As discussed above, courts frequently evaluate the reliabilityof an economic expert’s testimony based on the degree towhich the expert’s analysis is consistent with other sources ofevidence, particularly the defendants’ ordinary course busi-ness documents.18

For this reason, the economic expert’s role in antitrustlitigation should be to help tell the overall story and to pro-vide the court with a roadmap for applying the evidence tothe analytical framework of the governing case law. The mosteffective experts, guided by experienced trial counsel, willpresent economic analysis—even when it involves compli-cated econometrics or economic models—using an intuitive,common-sense approach that fits comfortably within legalprecedent. Thus, economic testimony remains important,but its function differs from the agency setting.

ConclusionThe recent wave of merger litigations stresses that econom-ics, while certainly relevant to courts, rarely if ever can carrythe day when faced with substantial qualitative evidence tothe contrary. So, if a party is prepared to expend the resourcesto litigate, it is probably a wise investment to front load theeconomic analyses at the agency level.�

1 Our own experience is confirmed by public statements from agency lead-ership spanning Republican and Democratic administrations. See, e.g.,Andrew Finch, Acting Assistant Att’y Gen., Remarks at Global AntitrustEnforcement Symposium 5 (Sept. 12, 2017), https://www.justice.gov/opa/speech/acting-assistant-attorney-general-andrew-finch-delivers-remarks-global-antitrust (“Economics has played, and will continue to play, a funda-mental role in antitrust enforcement.”); Bill Baer, Assistant Att’y Gen., U.S. Dep’t of Justice, Connecting the Antitrust Dots: In Praise of HerbHovenkamp 4 (Oct. 23, 2014), https://www.justice.gov/atr/file/517731/download (“Today, an investigation of—or lawsuit challenging—a potentialviolation of the antitrust laws inevitably involves economists (and ofteneconometricians) on both sides.”).

2 Carl Shapiro, The 2010 Horizontal Merger Guidelines: From Hedgehog to Foxin Forty Years, 77 ANTITRUST L.J. 49, 65 (2010).

3 U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines § 4 (2010) [hereinafter Merger Guidelines], http://ftc.gov/os/2010/08/100819hmg.pdf.

4 Id. § 6.1; see also Shapiro, supra note 2, at 68 (“DOJ puts far more weighton diversion ratios and margins than on the HHI level when diagnosing uni-lateral price effects. This has been the case for many years.”).

5 See, e.g., Malcolm B. Coate & Shawn W. Ulrick, Bureau of Economics, Fed.Trade. Comm’n Transparency at the Federal Trade Commission: The Hori -zontal Merger Review Process 1996–2003, at 58 (Feb. 2005), https://www.ftc.gov/sites/default/files/documents/reports/request-mailed-copy-


4 8 · A N T I T R U S T

transparency-federal-trade-commission-horizontal-merger-review-process-1996-2003/0502economicissues-1.pdf.

6 Edith Ramirez, Fed. Trade Comm’n, The Horizontal Merger Guidelines FiveYears Later 5 (Sept. 29, 2015), https://www.ftc.gov/system/files/docu-ments/public_statements/805441/ramirez_-_georgetown_antitrust_enforcement_symposium_9-29-15.pdf; see also Statement of the Com -mission, Fed. Trade Comm’n, In re Dollar Tree, Inc., FTC File No. 141-0207(July 13, 2015), https://www.ftc.gov/public-statements/2015/07/statement-federal-trade-commission-matter-dollar-tree-inc-family-dollar.

7 Statement of the Federal Trade Commission, In re Royal Caribbean Cruises,Ltd., FTC File No. 021-0041 (Oct. 4, 2002), https://www.ftc.gov/sites/default/files/documents/cases/2002/10/cruisestatement.htm; see alsoJoseph J. Simons, Merger Enforcement at the FTC n.44 (Oct. 24, 2002),https://www.ftc.gov/public-statements/2002/10/merger-enforcement-ftc.

8 David Scheffman, The FTC Cruise Line Merger Investigation, ABA Section ofAntitrust Law “Brown Bag Program,” ANTITRUST SOURCE 6 (Jan. 2003),https://www.americanbar.org/content/dam/aba/publishing/antitrust_source/lerner.authcheckdam.pdf.

9 FTC v. Sysco Corp., 113 F. Supp. 3d 1, 36–37 (D.D.C. 2015); see also id.at 55, 64 (agreeing with the FTC’s analysis of the relevant market becauseits position was “consistent with market realities” and “corroborated byother evidence in the record,” such as ordinary course documents and busi-nessperson testimony).

10 Id. at 27 n.2; see also United States v. Aetna, Inc., 240 F. Supp. 3d 1, 21,23–29 (D.D.C. 2017) (relying on Brown Shoe factors to evaluate the relevantproduct market); FTC v. Staples, Inc., 190 F. Supp. 3d 100, 118–21 (D.D.C.2016) (same).

11 United States v. Anthem, Inc., 236 F. Supp. 3d 171, 198 (D.D.C. 2017); seealso id. at 210.

12 Id. at 219. 13 Aetna, 240 F. Supp. 3d at 21, 26–28. 14 Id. at 39.15 FTC v. Steris Corp., 133 F. Supp. 3d 962 (N.D. Ohio 2015).16 United States v. Oracle Corp., 331 F. Supp. 2d 1098, 1121–22 (N.D. Cal.

2004).17 FTC v. Penn State Hershey Med. Ctr., 838 F.3d 327, 339–41 (3d Cir. 2016).18 See, e.g., Anthem, 236 F. Supp. 3d at 210 (noting that “the refutation of the

defense expert’s criticisms can be found in Anthem’s own files”); Sysco,113 F. Supp. 3d at 55, 64 (agreeing with the FTC’s analysis of the relevantmarket because its position was “consistent with market realities” and “cor-roborated by other evidence in the record,” such as ordinary course docu-ments and businessperson testimony).

� LIKE ITS PREDECESSOR, Proving AntitrustDamages: Legal and Economic Issues, Third Edition is anaccessible introduction to the legal and economic conceptsof antitrust damages for use by counsel who may be new to the area. To serve more experienced antitrust practitioners, the third edition has been completely updated to capture the most important developments in this area and represents the most authoritative and comprehensive resource on the subject of antitrust damages. The third edition also features expanded economic content that address the economic principlesunderlying the measurement of damages. Written by economists, this content provides counsel with a deeperunderstanding of the relevant economic issues in a wayaccessible to those without formal economic training.

Proving Antitrust Damages is organized into three parts:Chapters 1–4 guide counsel through the legal requirements that a plaintiff must satisfy in order to establish a right to recover damages in an antitrust private action; Chapters 5 and 6 identify the economic concepts that are used in calculating damages anddescribe the econometric analyses that are used to differentiate the effects of anticompetitive conduct fromother influences; and Chapters 7–9 discuss commonlyarising issues associated with estimating damages relatedto overcharges, lost profits, and price discrimination under the Robinson-Patman Act.



Proving Antitrust DamagesThird Edition



theantitrustsourcewww.antitrustsource.com

�Go to thesource

S P R I N G 2 0 1 8 · 4 9

market as a whole. And, of course, master these issues early,as no client wants to assess risk (or incur costs) midstream ina case.

Threshold Issues: Market Definition, Market Power, and Entry BarriersAs with any antitrust case or matter, economists and practi-tioners must address alleged misconduct within a statutoryframework, whether for certain mergers that may harm con-sumers (Section 7 of the Clayton Act),3 alleged monopo-lization or attempted monopolization (Section 2 of theSherman Act),4 or an “agreement” (horizontal or vertical) thatmay unreasonably restrain trade (Section 1 of the ShermanAct).5 Especially in the context of conduct that may forecloserivals’ opportunities, virtually all of the statutes require proofthat the defendant possesses market power within a well-defined antitrust market.6 Without sufficient market power(referred to as “dominance” in many jurisdictions), it isunderstood that a firm would not be in a position to affectmarket conditions to the extent of hurting the competitiveprocess and, in turn, consumers. After all, if alleged fore-closing activity does not eliminate substantial competition ina market and leaves consumers with substitute products—orrivals with substitute inputs or distribution paths—con-sumers are not at risk, even if a rival may be “foreclosed.”And, of course, it has long been established that lasting mar-ket power cannot exist where entry barriers are low and it canbe shown that either rapid demand or supply responses toalleged misconduct would quickly thwart any potential con-sumer harm. For aggrieved rivals, proving anticompetitive foreclosure is

particularly critical in the United States, where the “essentialfacilities doctrine” is all but dead, and plaintiffs cannot argue,standing alone, that the antitrust laws should provide themaccess to a particular product or distribution opportunity inorder to compete effectively.7

The first necessary step for economists in a typical Section2 or Section 1 foreclosure matter, then, is to tackle these struc-tural requirements. What are the markets affected by the con-duct at issue and who are the (current or potential) suppliersand buyers in these markets; can firms respond in a way thateliminates the expected benefits from the conduct; what is thepotential evidence of the defendant’s market power and are

The Economics of Foreclosure:A Lawyer’s Guide

B Y J A M E S K E Y T E , D A V I D S . E V A N S , A N D E L I A N A G A R C É S

THE TERM “ FOR EC LO SUR E ” I Sbandied about a lot in antitrust. Enforcers, prac-titioners, and economists alike speak of foreclosure in the context of vertical mergers, allegedmonopolization, Section 1 vertical agreements,

and Section 1 horizontal conspiracy cases when, for example,considering parallel vertical arrangements that allegedlyrestrict output.1 Foreclosure arises in analyzing the variouscompetitive consequences of numerous practices, such asexclusivity agreements, MFNs, tying, loyalty discounts, andpredatory pricing. And, of course, in the age of platformcompetition, an assessment of foreclosure–together with pro-competitive justifications–often takes center stage.2

Foreclosure analysis raises many issues that careful practi-tioners must consider. Foreclosure of what, and with whatimpact? What if foreclosure results in lower prices or bettervalue? And how should one evaluate claims that the defen-dant engaged in “raising rivals cost” (RRC) through verticalcontracting practices? How do economists help connect whatmay harm a firm’s rivals to the requisite harm to the com-petitive process and consumers? Finally, as many foreclosureissues arise in the context of non-price vertical restraints—subject to the full rule of reason—what procompetitive ben-efits could economists present to the fact-finder as part of theoverall assessment of whether the restraints should be con-demned? The answers to these and other questions are fundamen-

tal to the assessment of transaction or litigation risk, andtherefore should be at the top of a practitioner’s to-do listwhere alleged foreclosure effects may be at issue. This dis-cussion should send a clear message to practitioners: do notassume that issues of foreclosure may be avoided on theoret-ical bases, but rather be ready, in any context, to analyze (oneither side of a matter) both the incentives to foreclose andthe likely (if prospective) or actual effects on rivals and the

James Keyte is the Director of Global Development at the Brattle Group

and an Associate Editor of ANT ITRUST magazine. Eliana Garcés is a Prin -

cipal of the Brattle Group. David Evans is the Chairman of the Global

Economics Group. The authors would also like to thank Jeremy Koegel, an

associate at Skadden, Arps, Slate, Meagher & Flom, for his invaluable

assistance in the preparation of this article.

there indicators, such as the firm’s market share, that can beused to infer market power? It is also important to investigatethe entry conditions in the affected markets, including specificindustry dynamics that may affect entry. Are technologiesrapidly changing? Are market shares stable, and if so, why? The answers to these questions provide economic evi-

dence that is essential for addressing a fundamental thresh-old question—does the defendant even have market power—for determining whether a foreclosure theory has enoughsubstance to get off the ground (e.g., survive a motion to dis-miss or a motion for summary judgment.)8

No doubt these also are fundamental questions for con-sidering foreclosure strategies in the context of mergers. Aclear illustration is found in the Justice Department’s chal-lenge to AT&T’s acquisition of Time Warner.9 The govern-ment claims that AT&T would use its control over TimeWarner’s content to raise costs to AT&T’s video distributionrivals and foreclose innovative video distribution models;10

AT&T, in response, asserts that it does not have the ability todo that.Finally, while the term foreclosure sounds like an all-or-

nothing proposition, in practice defendants may engage inconduct that results in their rivals shrinking, which in effectcan reduce, though not eliminate, their competitive role inthe market. And courts certainly have found antitrust viola-tions from practices that do not result in full foreclosure—onthe contrary, partial foreclosure is the norm. Accordingly,our use of the term foreclosure here covers the spectrum offoreclosure possibilities.

Prospective Foreclosure: Assessing IncentivesFrom an economic perspective, there is a significant differ-ence in assessing prospective foreclosure versus actual foreclo-sure. For proposed vertical deals or prospective injunctiverelief, the assessment of foreclosure is a predictive exerciserather than an analysis of what foreclosure, and its effects,actually occurred as a result of the challenged conduct. A major threshold subject that economists must address in

evaluating theories of prospective foreclosure is whether thedefendant is likely to have an economic incentive to foreclosea rival.11 Conceptually, this analysis just involves weighing thebenefits and costs of foreclosure by the defendant. For exam-ple, an economist can compare the anticipated economicbenefits of engaging in, or refraining from, foreclosing con-duct, such as charging rivals higher costs: the expected reac-tions of competitors and customers in particular are instru-mental in this analysis. Economists should also consider the market dynamics

that may undermine (or confirm) the premises of the incen-tives calculation. In practice, this analysis can involve com-plex economics, such as dynamic games, and require theexpert to make various assumptions about decision rules andexpected pay-offs. Depending on the particular facts, such ananalysis may reveal a clear picture as to whether a foreclosurestrategy would make sense. A court would be hard-pressed,

for example, to enjoin conduct that, based on sound eco-nomics, makes no sense to pursue.When a firm with market power at one level (e.g., pro-

duction) operates in a more competitive market downstream(e.g., distribution), there may be no incentives to foreclose thedownstream competitors if the competing distributors aredifferentiated in a way that manages to serve many differentusers, which could help boost the sales of that producer. Forexample, this is why supermarkets that sell their own privatelabel products may not necessarily have an incentive to fore-close rival branded products from their shelves. Of course, in any foreclosure context, the defendant’s

documents, testimony, or other facts, may contradict theeconomic theory and calculations, raising the possibility thatthe economist’s theory, assumptions, or estimations are sus-pect. Thus, in evaluating horizontal mergers between multi-channel video distributors (such as Charter and Time WarnerCable) and vertical mergers between video distributors andcontent providers (such as Comcast and NBC-Uni versal),the U.S. Department of Justice and Federal Com muni -cations Com mission have examined whether the mergerwould enhance the incentive and ability of these entities toforeclose rival video distributors (like Netflix) from key in -puts (broadband or content). Their findings that the merg-ers would do so have been based on a combination of economic theory, econ ometric studies, and company docu-ments. And those findings presumably formed part of thebasis for Comcast abandoning its merger with Time WarnerCable and accepting significant conditions in its acquisitionof NBC-Uni versal.12

Proving the existence of the incentive and ability to fore-close rivals is necessary but it is not by itself sufficient. Plain -tiffs also have to show that the foreclosure strategy has had, orwill have, a significant adverse impact on competition andconsumers. Plaintiffs who are relying on prospective foreclo-sure will attempt to show that, because of the incentives, fore-closure is likely to occur and that anticompetitive effects areprobable. Plaintiffs would also have to respond to any evi-dence presented by the defendant that these predicted effectsare outweighed by procompetitive benefits. These are dis-cussed below in the context of actual foreclosure.

Actual Foreclosure: Reality Checks and Analytical FrameworksIf practitioners are dealing with allegations of actual foreclo-sure, the role and analysis of the economists should be muchmore concrete: the conduct has occurred, and it either did ordid not cause consumer harm, which can be determined byexamining whether the conduct resulted in competitors beingforeclosed and, if so, whether that resulted in harm to com-petition and consumers.13

A Reality Check on Consumer Harm. Even beforedetermining the proper framework for assessing the conductin question, economists for both sides would be well advisedto look for evidence of potential consumer harm by com-


5 0 · A N T I T R U S T

S P R I N G 2 0 1 8 · 5 1

paring consumer welfare today versus an estimate of the“but-for world” without the alleged foreclosing behavior,which is the ultimate rule of reason inquiry.14 While the lawhas not necessarily incorporated such a threshold inquiryinto ultimate effects, it can be quite informative for practi-tioners who are assessing risk and potential damages. If, for example, prices and output did not materially change

from before and after the conduct at issue, plaintiffs shouldknow they will be hard pressed to show an anticompetitiveeffect, irrespective of their theory of harm, and run the risk oflosing on summary judgment. By contrast, if decreases inoutput or increases in price correlate with the conduct atissue, then defendants should know that they may have a lotof work to do (even though the initial burden falls on theplaintiff ) to explain why the output decreases or price increas-es were not the result of the alleged conduct and to developprocompetitive justifications for the challenged activity.Such an inquiry makes sense from a legal perspective as

well. A foreclosure theory supports antitrust liability only ifthe theory, and supporting evidence, show a link between theconduct and harm to the competitive process and consumers.A theory that just shows that the conduct weakens one ormore rivals, without affecting the market as a whole, is notenough to allege consumer harm.15 For example, if, after avertical merger the merged entity cancels a supply contractwith a competitor downstream, the conduct will not bedeemed anti-competitive if the customers of the suspendedfirm can easily find alternative sources of supply.

Price-Based Unilateral Conduct. An important thresh-old question for the lawyers and the economists is to deter-mine what framework to use (or to allege, if in the plaintiff’scamp). Specifically, there is a significant distinction betweena foreclosure theory based on pricing behavior (e.g., retroac-tive discounts on all sales) and one based on non-price con-duct (e.g., exclusivity agreements) that makes it more diffi-cult, or costly, for rivals to sell to consumers by raising thecosts of inputs or denying them access to customers, distri-bution, or important complementary products or services.16

The possible mechanisms of harm, as well as the procom-petitive rationales, differ markedly. Where the theory of the alleged misconduct is based on

the plaintiff not being able to match the pricing level result-ing from defendant’s discounts, the economists will need tobe prepared to address both the issue of whether the price isbelow a relevant measure of cost and the potential for recoup-ment—the ability of the defendant to raise prices after eitherdriving the rival out of business or seriously impairing itscompetitive capabilities.17

The former analysis of pricing and costs can get quitecomplex and may even turn on in which court the case isbeing brought. For example, some courts use an “averageprice-cost” test, which compares the firm’s total revenue forall units sold to its total variable cost—i.e., a comparison ofaverage prices to average variable costs.18 Others comparethe firm’s incremental revenue on the contestable volume

secured by the contracting practice with the incremental costof supplying that volume, an “incremental price-cost” test.19

And some courts, applying these tests, take the view thatconsumer harm is at issue only if an “equally efficient com-petitor” could not survive in the marketplace.20

In all cases of price-based predation, however, an economistwill have to address the subject of recoupment, as consumerswill not be harmed from lower prices if the conduct does notlead to a later ability to raise prices or reduce output—and somuch so that it outweighs the benefits they have gotten fromthe lower prices.21 Essentially, this is a structural inquiry thatthe courts and economists are quite familiar with since BrookeGroup, and certainly should be a threshold inquiry whenassessing risk from a claim based on price predation.Some pricing conduct may be deemed exclusionary, even

though it represents no loss to the company implementing it.Loyalty rebates or retroactive rebates can result in the sameaverage price and the same amount of revenue from the cus-tomer and yet may succeed in foreclosing rivals by structur-ing the price so that customers actually pay more if they donot buy a certain amount from the supplier. While this strat-egy does not require below-cost pricing or recoupment, itdoes require that the firm is a somewhat unavoidable supplierfor a significant fraction of the sales to the customer. A cor-rect assessment of such programs can focus on the price actu-ally paid for the incremental sales over the threshold at whichthe rebates kick in, which is similar to the standard appliedin pure predation cases. Alternatively, one can take theamount of the rebate over the total sales of the customer asthe cost that a new entrant would have to pay to the customeras compensation to be able to supply the contestable share. For practitioners, it is important to remember that quan-

tity rebates, exclusionary rebates, loyalty discounts, or bun-dled rebates can come in many shapes and forms and will beassessed under a rule of reason that may or may not carry aprice/cost analysis similar to that of a predatory analysis.But, irrespective of the particular context, economists willargue that understanding the impact of the practice on thedynamics of competition and entry, as well as on the set ofoptions available to customers, will be particularly relevant inthose cases.Most favored nation clauses (MFNs) may also fall under

the price-related foreclosure rubric. MFNs—where a firmdemands the best price (and possibly other commercialterms) that its customer is willing to extend to any other cus-tomer—can be frowned upon by courts and regulators as amethod to soften competition by making aggressive entry orexpansion more difficult or even worse, as a facilitating prac-tice for horizontal collusion. In the case where MFNs are usedin a non-collusive context, the focus of the analysis shouldcenter on the impact of the clause on the ability and incen-tive of rivals to compete with more attractive offerings, as wellas on the impact on prices and quality in the market. Forexample, the Justice Department has argued that dominanthealth insurance providers have imposed MFNs on providers


5 2 · A N T I T R U S T

ability to expand by competing with better products or prices.Because a firm may price a bundle at a discount compared tothe prices of the products independently sold, the analysis ofthe effect of tying and bundling may in some instances bearsome similarities to that of price-based conduct. Firms may have the anticompetitive incentive to “tie” a

product if such conduct deprives all rivals from the ability toaggressively compete in the tied market. Plaintiffs wouldhave to demonstrate the profitability of the practice by show-ing that customers’ preferences and set of options are suchthat a sufficient number of customers would prefer to buy thebundle rather than walk away from both the tying and tiedproduct sold by the firm. After demonstrating the prof-itability of the conduct compared to the but-for world with-out a tie, plaintiffs will have to show both harm to users onthe tied market and, depending on the tying standard beingapplied, overall consumer harm in the tied market.

The Tough Question: Proving “Power over Price”and Harm to ConsumersWhere the “rubber meets the road” in foreclosure analysis,however, is on the subject of connecting exclusionary conductto harm to consumers. Screening for some evidence of possi-ble harm before embarking on a case is a useful check. In theUnited States, at least, every antitrust plaintiff must prove that the challenged conduct caused harm (or will likely causeharm in the injunction context) to the competitive processand consumers, rather than just to one or more competi-tors.24 From an economics perspective, this means that it willnot be enough merely to show that a rival is paying more foran input, or has trouble distributing its product, as a result ofdefendant’s foreclosing conduct. The plaintiff must demon-strate a connection to consumer harm. This, in turn, boilsdown to the question of whether the conduct sufficientlysoftens the but-for competitive constraints on the defendantsuch that prices or quantity will change significantly for theworse in the market as a whole. The plaintiff must also beready to argue against alternative explanations for the priceand quality changes involving events related to the funda-mentals of market demand or supply. How much, and what kind of evidence, is sufficient is one

of the issues the U.S. Supreme Court has before it inAmerican Express.25 The Second Circuit reversed a lower courtfinding of liability, in part because there was no evidence thatthe non-price vertical restraints at issue reduced output. TheJustice Department and State plaintiffs have pointed to priceevidence they claim is more than enough.26

Historically, it was posited—by Judge Richard Posner27—that absent the “elimination” of a significant rival, there couldbe no harm to consumers, as output would remain in themarketplace. This can be seen, economically, as a focus onwhether the conduct has caused (or is likely to cause) rivals’costs to be driven below minimum viable scale.28 And, morerecently, some of the debate in the United States has centeredon whether the conduct has driven (or is likely to drive)

that prevented entry and expansion of smaller rivals, withalleged resulting harm to consumers compared to a worldwithout the MFNs.22

Non-Price Based Unilateral Conduct. There are manyother ways that firms compete besides on price. They try tosecure efficiencies through vertical integration, invest in effi-cient distribution, and engage in various marketing activi-ties. Firms can get an edge over their rivals when they pursuethese strategies. Disadvantaged rivals are harmed. But anti trustdoes not concern itself with this normal to-and-fro of com-petition. It focuses instead on those cases in which a firmwith significant market power engages in exclusionary conductthat results in harm to the competitive process without ben-efiting consumers.Suspect conduct includes cases in which vertically inte-

grated firms deny rivals access to an input, or raise the costof that input to make it harder for rivals to compete, or enterinto distribution agreements that prevent rivals from gettingaccess—or getting as good or as cheap access—to customers.The key issue for economists in these cases, after settling onmarket definition and market power, is whether the conductlimits the competitive vigor of rivals in ways that result inmarket-wide increases in price, reductions in output, orreductions in quality, any of which are not offset by othergains in efficiency. Economists would not find competitiveharm if the affected market remains competitive and there arealternative inputs or distribution paths available to rivals. To show competitive harm, plaintiffs and their econo-

mists must demonstrate the efficacy of the alleged anticom-petitive action. For example, in a case of input foreclosure, aplaintiff would have to show that other input sources werenot available and that the increase in input cost is actuallymaterial. Perhaps it could obtain a similar input from anoth-er supplier or the cost of the input is such a small portion ofcosts that it has no significant impact on the ability to com-pete on price. The courts will require concrete proof on thisand not just theoretical assertions that a problem might bepossible. If there are exclusive contracts for distribution, theplaintiffs and their economists will need to address whetherthe portion of distribution excluded is significant andwhether it is possible for the plaintiff to bid for exclusiveswhen contracts expire or through termination clauses.23 Inthe case of preferential treatment in distribution, plaintiffswill have to show that the most favorable treatment benefit-ing the defendant deprives its rivals of customers or revenuesto the point of hurting the rivals’ viability and ultimatelydepriving final users of supplier choice. Exclusionary conduct, in principle, can also take the form

of tying or bundling the sale of a product to that of anotherover which the firm has either a monopoly or very strong mar-ket power. In such cases, a firm forces or gives incentives tocustomers who want to buy the main or flagship product,which will be the tying product, to also acquire a tied prod-uct. Rivals in the market of the tied product may be foreclosedfrom a substantial portion of the market, which may hurt their

S P R I N G 2 0 1 8 · 5 3

rivals below minimum efficient scale,29 the proposition beingthat, absent such an effect, the same amount of output wouldremain in the market even if it impacts a rival’s profits.30

By contrast, the most aggressive use of an RRC theory toconnect with consumer harm maintains that any materialincrease in a rival’s marginal costs will provide an incentiveto pass those costs on to consumers, which in turn will, intheory, provide the defendant with the flexibility to increaseprices—i.e., a form of “upward pricing pressure” that can bemodeled as such.31 Professor Steve Salop argues that: “RRCconduct does not require the exit of rivals, or even the per-manent reduction in competitors’ production capacity. Ifthe marginal costs of established competitors are raised, thosecompetitors will have the incentive to raise their prices andreduce their output, even if they remain viable.”32

Ultimately, the inquiry centers on this same question: doesthe restraint provide the defendant with additional andunwarranted “power over price,” such that it could reduceoutput, raise price, or reduce product quality without someform of competitive response to defeat it. The possibleresponse of non-excluded (or partially excluded) rivals, newentrants, or the actions of the excluded firms themselves mustbe considered33—all fodder for economists to explore in detail. The subject of causation—i.e., does any effect on prices

and output result from the alleged restraint–will also be a crit-ical question, and numerous identification techniques areused in empirical analysis to try to isolate the effect of par-ticular conduct.34 Suppose, for example, that a dominantdistributor replaces an MFN that covers all conditions of acontract with an MFN that covers price but not the otherforms of consideration that suppliers and distributors mightoffer each other. The selective MFN is sometimes referred toas a cherry-picking MFN. Economists could compare com-petitive outcomes—prices and output—before and after theimposition of the new MFN. Assuming nothing else signif-icant has changed in the marketplace, it may be possible toinfer the impact of the specific conduct at issue to assesswhether it is anticompetitive or not.

Comparing the Actual and “But-For” Worlds: WhatPart of the Market Is “Contestable”? Another key issuein foreclosure cases concerns how much of the market hasactually been foreclosed, or will be foreclosed, by the conductat issue. In some extreme cases an exclusive dealing require-ment excludes the only rival—an entrant challenging amonopolist for example—from the entire market.35 Clearly,however, there are situations where, prior to the alleged fore-closure conduct, the defendant was already garnering largesales based on product differentiation, service, or reputationand the like. Hence, the “but-for world”—i.e., if the fore-closure had not occurred—may involve significant sales orshare that is not “contestable” in the first place because itwould likely have always gone to the defendant—i.e., the“but-for foreclosure” rate.36

Economists have tended to look at effects through this“but-for” lens for many decades, even if that is not always the

case in the courts.37 The difference in estimated “foreclosurerates” can be quite dramatic, depending on the size of thenon-contestable share that would likely go to the defendantirrespective of the contracting conduct at issue. For example,if a firm with a 55 percent market share—based on productquality, differentiation, or other factors—institutes a loyaltyprogram on all customers that takes its share to 60 percent,the difference between finding a foreclosure rate of 60 per-cent versus 5 percent is enormous and can obviously makethe difference in a case. In recent years, courts have followed economists towards

this “but-for” assessment of foreclosure,38 which has now putthe spotlight on the contestability inquiry. The question forthe economists, then, is to try to identify the contestableamount of output available in the market and assess what partof that is consumed by the alleged misconduct. In this way,the effect of the alleged misconduct is predicated on the con-testable amounts that rivals would otherwise have available tothem. That quantification, by itself, still does not show thatthe conduct caused harm to competition through higherprices, lower output, or reduced quality, however. In ideal cir-cumstances, economists can use a quasi-natural experiment—such as the imposition of the practice in different geographiesat different times—to determine whether there is, or is not,a causal relationship.

Identifying and Measuring Procompetitive Justifi -cations and Effects. Many foreclosure allegations haveinvolved vertical, non-price restraints. Such conduct has beensubject to the “full” rule of reason framework since GTESylvania.39 Indeed, the law is sufficiently settled on thispoint,40 and some even advocate for a presumption that ver-tical restraints are procompetitive and that only “clear andconvincing” evidence of anticompetitive effects on consumersshould be able to render these types of restraints unlawful (asubject not addressed here).41

Issues of burden of proof aside, virtually all cases involv-ing foreclosure claims (absent run-of-the-mill coordinationamong horizontal competitors using vertical restraints) willinvolve indepth analysis of potential procompetitive justifi-cations, including, among others: the prevention of free rid-ing; elimination of double marginalization; inducement ofmarketing support or other forms of loyalty; where relevant,the promotion of procompetitive bundling strategies; othercontractual efficiencies; and, depending on the industry andcircumstances, the promotion of innovation. Other procompetitive justifications from “foreclosing”

contracts or vertical integration can include product improve-ments, cost savings, R&D synergies, and so on. Indeed, thereis a long history in both law and economics on the benefitsof partial or full vertical exclusivity starting with the SupremeCourt’s GTE Sylvania decision.42 Parties will have to supporteach of these or similar rationales or justifications with boththeoretical economics and, where possible, empirical andqualitative evidence.

5 4 · A N T I T R U S T


The Complexities of Competition Involving Multi-sided Platforms. The foreclosure issues involving platformcompetition add complexities to the analysis of anticompet-itive conduct. Multi-sided platforms create value by servingas intermediaries between two or more types of customers thatcreate value by interacting with each other.43 For example,consumers desire on-demand car services and Uber driversneed customers: the Uber application platform intermediatesbetween the two sides, and pricing and production decisionsare based on coordinating the demand between these cus-tomer groups. Similar dynamics apply to many other indus-tries, such as smartphone operating systems, credit card plat-forms, shopping malls, ad-supported media, and equityexchanges. Five of the largest companies in the world by mar-ket capitalization—Apple, Amazon, Facebook, Google, andMicrosoft—now earn a significant amount of their revenue,and in some cases almost all, from operating these sorts ofplatforms.The application of standard antitrust principles to markets

involving multi-sided platforms is made complicated by cer-tain inherently unique dynamics. Multi-sided platforms oftenhave positive externalities in that members on one side canexpect to realize more benefits if there are more members onthe other side. These externalities result in the demand foreach group dependent on the demand by the other group. Asa result of interdependent demand, the profit-maximizingprices to customers on each side of the market do not neces-sarily track demand or costs on that side, and business strate-gies must account for the fact that benefits and costs are cre-ated jointly across both sides of the market.44 A keydistinguishing feature of these platforms, in comparison withtraditional firms, is that the platform is mainly in the businessof selling each group of customers access to the other group,so that each group is an input into supplying the other.Firms within multi-sided platforms, however, engage in

many of the same kinds of conduct as traditional firms, whichcan give rise to antitrust concern when they have marketpower. Exclusive contracts, loyalty programs, and tying, forexample, can be attractive strategies to increase demand onboth sides of the platform. In theory, those strategies could beattractive for established firms (with monopoly power) thatmay wish to use exclusive contracting (or other contractingmechanisms) to deter entry by more efficient rivals.45 Theidea would be to use forms of exclusivity or partial exclusivi-ty on one side of the platform to deprive the new entrant ofthe “critical mass” necessary on both sides of the platform togenerate positive feedback effects.46 This inevitably will be afact-specific inquiry that also must take into account the pro-competitive benefits of the restraints—some unique to multi-sided platform competition—that will always be at play inthese cases.47

A classic case involved FTD, an association of florists thathelped people to send flowers to distant locations. It had arule that prevented florists from joining competing associa-tions. Accordingly, a new entrant could not persuade a local

florist to join because that florist would lose access to con-sumers throughout the country, and without enough florists,a prospective entrant could not persuade consumers to useits service. The U.S. Department of Justice challenged therestraint, FTD settled, and entry commenced.48

The procompetitive benefits of vertical restraints in multi-sided platforms are numerous and, at times, complex. Verticalrestraints can help platforms achieve positive network effects,which will benefit consumers if the positive feedback on oneside of the market outweighs the anticipated price increases onthe other side of the market. Vertical restraints can also ensuresufficient participation on one side to generate critical mass orgrow feedback effects, which provides value to all sides of theplatform. Likewise, vertical restraints can encourage sunk-cost investment and promote “single homing” (where a con-sumer uses only one platform in a particular industry), whichalso generates critical mass.49 And, of course, vertical restraints,as in traditional markets, can prevent free riding, which incen-tivizes investment and risk taking.50 Shopping malls, for exam-ple, enter into low-rent deals with anchor stores that can helpdrive traffic to the mall. Those deals would not make sense ifthe anchor store was also at a competing nearby mall or hada standalone store because then traffic would be divertedthere. A solution to this problem is to impose a geographicexclusivity clause on the anchor store, which prevents thestore from establishing nearby locations. Again, many of these justifications will be case- or indus-

try-specific, and may also be informed by the market positionof the firm asserting the justification. There is little doubt, forexample, that these justifications may get more traction withnew entrants or smaller rivals attempting to achieve criticalmass. But, at the same time, all rivals seek to grow and gen-erate or meet consumer demand, and are likely to assert sim-ilar justifications.

Analysis Will Continue to EvolveAs these foreclosure issues are litigated in the courts anddebated in the antitrust agencies, the academic work sur-rounding foreclosure will march on. The pioneering eco-nomic work on foreclosure, which helped spur the courts todrop the per se condemnation of many vertical restraints, canbe traced to the Chicago School’s observation that, becausethere is single-monopoly profit to be had on a vertical chain,it was not possible to get a bigger profit by leveraging amonopoly from one level to another.51 Modern industrialorganization theory has built on this work, but in doing sohas identified situations in which the single-monopoly prof-it theorem does not necessarily hold. This work, which issometimes based on game theory, shows that there are cir-cumstances under which a firm with market power couldhave the incentive to exclude rivals in adjacent markets.52

Yet another facet of economic work involving foreclosureanalysis centers around the empirical assessment of the antic-ipated effects of the conduct. This analysis can consist ofestimating the anticipated effect or measuring actual effects

S P R I N G 2 0 1 8 · 5 5

if the conduct is already established. Economists have begunexploring the effects of vertical integration and contractingthrough more complex “structural models” that attempt toconstruct, at the individual decision-making level, bothdemand and supply conditions in the before or current envi-ronment (which must be able to replicate the real world) andthen explore effects by importing the change in the market(e.g., vertical integration or contracting).53 And perhaps thesetypes of models may be used to address vertical restraints,although incorporating changes in pricing and strategy canbe quite difficult to capture.From a practitioner’s perspective, the most important

thing is to keep track of what type of modeling the other side(whether agency, plaintiff, or defendant) is using or explor-ing; otherwise, one may be caught short in assessing deal orlitigation risk.

Lawyers: Start Early and Know Both Sides of theEconomic IssuesThe message for lawyers—both in the U.S. and elsewhere—is not to let issues relating to foreclosure analysis take you bysurprise. Investigations of foreclosure conduct involve com-plex economics and require careful gathering and assessmentof evidence. In particular, practitioners are well advised to stayup to date on both the cases that discuss or rule on foreclo-sure issues and the growing body of economic literature dis-cussing and debating applicable frameworks of analyses. Italso make sense—whether representing plaintiffs or defen-dants—for practitioners to hire experienced economists asearly in the representation as possible. Not only will thisinform risk assessment, it also can help shape the develop-ment of plaintiff theories of harm or defenses in the case priorto discovery and motion practice. And, of course, knowingwhat the economic evidence will likely be on foreclosureissues can be very informative in the context of settlementpositions or resolutions.�

1 See generally C. Scott Hemphill & Tim Wu, Parallel Exclusion, 122 YALE L.J.1182 (2013).

2 See generally David S. Evans, Economics of Vertical Restraints for Multi-SidedPlatforms (Coase-Sandor Inst. for Law & Econ., Working Paper No. 626,2013) [hereinafter Evans, Economics of Vertical Restraints]; David S. Evans,The Antitrust Economics of Multi-Sided Platform Markets, 20 YALE J. REG.325 (2003) [hereinafter Evans, Antitrust Economics].

3 15 U.S.C. § 18.4 15 U.S.C. § 2.5 15 U.S.C. § 1.6 See Jonathan M. Jacobson, Exclusive Dealing, “Foreclosure,” and Consumer

Harm, 70 ANTITRUST L.J. 311 (2002) (tracing the evolution of the case law,the emergence of a market power analysis, and the RRC theory).

7 See Verizon Commc’ns Inc. v. Law Offices of Curtis V. Trinko, LLP, 540 U.S.398, 411 (2004).

8 See Carell v. Shubert Org., Inc., 104 F. Supp. 2d 236, 26–-65 (S.D.N.Y.2000) (granting motion to dismiss on foreclosure claim where plaintiffsfailed to plead foreclosure in a relevant market); see also Adidas Am., Inc.v. NCAA, 64 F. Supp. 2d 1097, 1102 (D. Kan. 1999) (dismissing complaint

and finding that foreclosure must be analyzed with regard to a relevantantitrust market, not a single product within a broader market).

9 Complaint, United States v. AT&T Inc., No. 1:17-cv-02511 (D.D.C. filed Nov.20, 2017), ECF No. 1.

10 Id.11 See Herbert Hovenkamp, Robert Bork and Vertical Integration: Leverage,

Foreclosure and Efficiency, 79 ANTITRUST L.J. 983, 995 (2014) (“[M]embersof the Chicago School typically rejected both leverage and foreclosure the-ories, concluding that they were really two variations of the same thing;there is no point in using vertical integration to foreclosure competitors ifyou cannot leverage your position in one market to obtain more monopolyprofits in a vertically related market.”).

12 David S. Evans, Economic Findings Concerning the State of Competition forWired Broadband Provision to U.S. Households and Edge Providers (Aug. 29,2017) (unpublished paper), https://ssrn.com/abstract=3029006; PressRelease, U.S. Dep’t of Justice, Assistant Attorney General Bill Baer DeliversKeynote Address at the Future of Video Competition and Regulation Con -ference Hosted by Duke Law School (Oct. 9, 2015), https://www.justice.gov/opa/speech/assistant-attorney-general-bill-baer-delivers-keynote-address-future-video-competition.

13 See generally Steven C. Salop, The Raising Rivals’ Cost Foreclosure Paradigm,Conditional Pricing Practices and the Flawed Incremental Price-Cost Test, 81 ANTITRUST L.J. 371 (2017); Jonathan B. Baker, Exclusion as the CoreCompetition Concern, 78 ANTITRUST L.J. 527, 551–56 (2013) (reviewing apotential truncated approach to exclusion, including whether rivals are notexcluded, the role of market power, a quick-look like approach for proof ofactual effects and the difficulty of dealing with procompetitive justificationswith truncated rules).

14 See Jacobson, supra note 6, at 366–67 (“Proof of an increase in rivals’costs is a necessary but not sufficient condition in establishing consumerharm. A likelihood of elevated prices (or reduced outputs, quality, choice, orinnovation) must also be shown.”); Dissenting Statement of CommissionerJoshua D. Wright, McWane, Inc., FTC Dkt. No. 9351, slip op. at 5 (Feb. 6,2014) (“Complaint Counsel makes no effort to establish harm to competi-tion directly, such as by demonstrating that McWane’s conduct had a dele-terious effect upon price or output . . . .”) [hereinafter Wright McWaneDissent], https://www.ftc.gov/system/files/documents/public_statements/202211/140206mcwanestatement.pdf.

15 See Daniel L. Rubinfeld, 3M’s Bundled Rebates: An Economic Perspective,72 U. CHI. L. REV. 243, 262–64 (2005) (arguing that because “pure”above cost, low pricing, or innovation can weaken rivals, the weakening ofrivals should not be viewed as monopoly misconduct, and arguing thatsome form of price-cost test that connects to consumer harm must beapplied).

16 See Salop, supra note 13, at 402. 17 See Brooke Grp., Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209

(1993). 18 See generally Gregory K. Leonard, The Competitive Effects of Bundled

Discounts, in ECONOMICS OF ANTITRUST: COMPLEX ISSUES IN A DYNAMIC

ECONOMY (Lawrence Wu ed., 2007) (comparing different approaches to bun-dled discounts after LePages and suggesting the consideration of bothshort and long-term effects); see Rubinfeld, supra note 15, at 251 (notingthat it was undisputed in LePages that the total output of transparent tapegrew over the relevant time period).

19 See, e.g., Yong Chao & Guofu Tan, All-Units Discounts: Leverage and PartialForeclosure in Single-Product Markets, 30 CANADIAN COMPETITION L. REV. 93,99 (2017) (“One key feature of the [all units discount] is that the total pay-ment from the customer to the supplier drops sharply once the customer’spurchase reaches the [agreed upon] threshold, resulting in negative mar-ginal prices for the units near the threshold.”).

20 See Salop, supra note 13, at 393. 21 Recoupment often is not required outside the United States.22 Complaint, United States v. Blue Cross Blue Shield of Mich., No. 2:10-cv-

14155-DPH-MKM (E.D. Mich. filed Oct. 18, 2010), EFC No. 1; Complaint,United States v. Delta Dental of R.I., No. CA 96 113 (D.R.I. filed Feb. 29,1996), ECF No. 1.

5 6 · A N T I T R U S T


23 See Salop, supra note 13, at 408; see also Joshua D. Wright, Moving BeyondNaïve Foreclosure Analysis, 19 GEO. MASON L. REV. 1163, 1169 (2012).

24 See Benjamin Klein, Exclusive Dealing as Competition for Distribution, 12GEO. MASON L. REV. 119, 122–28 (2003) (supporting MES standard).

25 Ohio v. Am. Express Co., 138 S. Ct. 355 (2017). Professor Evans co-authored an amici curiae brief in this case. Brief for Amici Curiae Prof. DavidS. Evans and Prof. Richard Schmalensee in Support of Respondents (Jan.23, 2018).

26 Brief for the Petitioners & Respondents Nebraska, Tennessee, & Texas, Ohiov. Am. Express Co., No. 16-1454 (U.S. filed Dec. 7, 2017); Brief for theUnited States as Respondent Supporting Petitioners, Am. Express, No. 16-1454 (filed Dec. 7, 2017).

27 See Roland Mach. Co. v. Dresser Indus., 749 F.2d 380, 394 (7th Cir. 1984).28 Minimum viable scale is the smallest that an entrant must achieve for prof-

itability.29 Minimum efficient scale is the lowest production point or level of sales at

which long-run total average costs are minimized.30 See Salop, supra note 13, at 385–86; Wright, supra note 23. 31 Steven C. Salop, Sharis A. Pozen & John R. Seward, The Appropriate Legal

Standard and Sufficient Economic Evidence for Exclusive Dealing UnderSection 2: The FTC’s McWane Case, GEO. L. FAC. PUBLICATIONS & OTHER

WORKS 3 (Aug. 7, 2014) (disagreeing with Commissioner Wright’s view onburden and arguing that Wright’s focus on MES does “not separately traceout the other implications of the entrant facing higher distribution costs andhow those higher costs might affect McWane’s pricing incentive”), http://scholarship.law.georgetown.edu/cgi/viewcontent.cgi?article=2376&context=facpub; id. at 6 (“While the rivals may not be totally frozen out ofthe market, and may continue to be viable, their output and ability to com-pete may be significantly constrained.”).

32 Salop, supra note 13, at 377. 33 See Baker, supra note 13, at 565 (“[T]he exclusionary conduct must be suf-

ficient to harm competition. [It] requires in part that the excluded firm mat-ter competitively; its exclusion must relax a competitive restraint on theexcluding firms. In addition, it requires that any remaining competition—whether from rivals not excluded or not fully excluded, from entrants, or fromamong the excluding firms themselves—not undermine what the relaxationof a competitive constraint has achieved for the excluding firms: their abil-ity to raise market prices.”).

34 See Wright McWane Dissent, supra note 14, at 36 n.43 (Feb. 6, 2014) (dis-cussing the issue of causation and whether a price increase resulted fromthe alleged misconduct).

35 Lorain Journal Co. v. United States, 342 U.S. 143 (1951).36 Wright, supra note 23, at 1185. 37 Id. at 1172–74. 38 See, e.g., R.J. Reynolds Tobacco Co. v. Philip Morris, Inc., 199 F. Supp. 2d

362 (M.D.N.C. 2002), aff’d, 67 F. App’x 810 (4th Cir. 2003); J.B.D.L. Corp.v. Wyeth-Ayerst Labs., Inc., No. 1:01-CV-704, 2005 WL 1396940 (S.D.Ohio June 13, 2005).

39 Cont’l T.V. Inc. v. GTE Sylvania Inc., 433 U.S. 36, 51–52 (1977).40 But see Brief of 28 Professors of Antitrust Law as Amici Curiae Supporting

Petitioners, Ohio v. Am. Express Co., No. 16-1454 (U.S. filed Dec. 14,2017).

41 See Wright McWane Dissent, supra note 14, at 48; Salop, supra note 13,at 393.

42 See Continental TV, 433 U.S. at 55.43 Evans, Economics of Vertical Restraints, supra note 2. 44 See generally Evans, Antitrust Economics, supra note 2. 45 Evans, Economics of Vertical Restraints, supra note 2, at 11. 46 Id. at 17. 47 Id. at 18. 48 David A. Balto, Networks and Exclusivity: Antitrust Analysis to Promote

Network Competition, 7 GEO. MASON L. REV. 523 (1999). 49 Evans, Economics of Vertical Restraints, supra note 2, at 6. 50 Id. at 10–11.

� THE ENERGY INDUSTRY HAS BEEN at the center of thedevelopment of many antitrust laws. In particular, the oil industryhas been the source of many seminal antitrust cases. The electricand natural gas industries also have contributed several importantcases, but have been subject to substantial if not completeantitrust immunity in light of pervasive federal and state regulation until the introduction of competition in the late 1980sand 1990s. Although the electric and natural gas industries continue to be subject to less pervasive regulation, new antitrustissues and concerns about “market manipulation” have becomethe focus of scrutiny in the electric, natural gas, and petroleumindustries. This third edition of the Energy Antitrust Handbook presents

a guide to an industry that has great importance to the U.S. economy. It is written to assist energy, regulatory, and antitrustlawyers in understanding the multilayered complexity of this fieldby providing a basic background on antitrust issues in the energyindustry. By reading this book, lawyers already familiar withantitrust law will gain an understanding of related energy issues,the market structure, and the application of the antitrust laws tothese industries. Lawyers and executives already familiar with the energy industry, but not with antitrust law, will find this bookprovides an understanding of the antitrust laws applicable to energy.



Energy Antitrust HandbookThird Edition



51 David S. Evans & A. Jorge Padilla, Designing Antitrust Rules for AssessingUnilateral Practices: A Neo-Chicago Approach, 72 U. CHI. L. REV. 73 (2005).

52 Salop, supra note 13, at 394. 53 See Gregory S. Crawford et al., The Welfare Effects of Vertical Integration in

Multichannel Television Markets (Nat’l Bureau of Econ. Research, WorkingPaper No. 21832, rev. Apr. 2017), http://www.nber.org/papers/w21832.pdf.

S P R I N G 2 0 1 8 · 5 7

The Court of Justice and Intel:An Overdue Ruling that Was Still Over-hasty?

B Y K E V I N C O A T E S

For a case that began with a complaint in 2000, a final res-olution that is unlikely before 2020 is not a model of legalexpedition. The European Commission is often accused ofintervening too readily in “fast-moving high-technology mar-kets,” but it is hard to look at the Intel case and conclude that“being over-hasty” in this case was the most significant short-coming of the European legal system.Though perhaps that is a little harsh. Although the first

complaint was filed by AMD in 2000, it was not until thecomplaint was supplemented substantially in 2003 that theCommission felt it could move forward with an investigation;and although six years between that second complaint andthe decision is hardly fast-moving, a significant reason for thedelay is that the complaint became a totemic case on the bal-ance between legal form and economic analysis in EU abuseof dominance enforcement, and a real world battleground forthe theoretical policy discussions that led to the Article 102Guidance Paper.2 In Intel, the Commission debated the rel-evance of the historic case law on loyalty rebates (largelyfinding a per se prohibition), the proper role of economicanalysis, and indeed the analysis that should apply in specif-ic cases. Progress on the Intel decision was frustrated by aclash of clans—hardline lawyers, hardline economists, andpragmatists from both camps—frustrating perhaps, but nec-essary as the Commission sought to establish the right prece-dent for cases far beyond the Intel case itself.So, taking from 2003 to 2009 to issue a decision is perhaps

understandable. Though for the European Courts to thentake from 2009 to 2017 without final resolution is perhapsless so. A resolution is overdue.At the same time, there are aspects of the Court of Justice’s

ruling that may prove to have been overly hasty, particularlyin relation to the key issue of the treatment of rebates.Although the case is certainly a welcome nudge in the direc-tion of more economic analysis, it is procedurally confusingand substantively unclear.Though perhaps that is a little harsh too. In the long run,

the fact that we do not yet have a satisfactory resolution to allquestions is perhaps less important than the fact that theCommission, starting over a decade ago, tried to grapplewith these questions of economic analysis at all, and that the

MOST OBSERVERS HAD EXPECTEDthe EU’s long-running Intel case1 to beover by the end of 2017, resulting in a rul-ing from the EU’s highest court thatwould illuminate the EU law of rebates

offered by dominant firms. The Court of Justice issued itsjudgment in September 2017, but the illumination of thedecision, however, did not shine as brightly or fully asobservers had hoped. What happened?

BackgroundThe Commission’s proceedings began in 2000, although (asnoted below), the Commission’s investigation began inearnest only several years later, leading to a decision in 2009.In its decision, the Commission condemned a wide-rangingset of Intel’s practices, many of which took place outside of theEuropean Union. Typically, these were arrangements withoriginal equipment manufacturers (OEMs) of computers andlaptops with whom Intel engaged either in loyalty rebatesthat—in the Commission’s view—were unlawful or, moreegregiously, in “naked restrictions,” and both practices reducedor eliminated the OEMs’ ability to purchase inputs from Intelcompetitors. In the latter arrangements, Intel approachedOEMs who were about to launch AMD products and gavethem money not to do so; that these naked restrictions wereunlawful is hardly controversial, and they were not the focusof the appeal. Rather the appeal focused on the rebates.But the Court of Justice’s September 2017 ruling did not

reach a final conclusion on the Commission’s analysis ofIntel’s rebate system. Instead, the Court referred that aspectof the case back to the General Court. When the GeneralCourt eventually rules in that remand proceeding, the losingparty may well take a further appeal back to the EuropeanCourt of Justice. As a result, it now looks possible that thecase will celebrate its 20th anniversary in 2020 without a finalresolution. And there is at least an outside chance that itmight celebrate its silver jubilee in 2025.

Kevin Coates is a partner in the antitrust group of Covington & Burling,

based in Brussels and London, and is a former Head of Unit in the Direc -

torate General for Competition of the European Commission.

Commission’s efforts have resulted in the EU’s highest courtendorsing the Commission’s approach.In full disclosure, I was working in DG Competition at

the time of the Intel investigation and decision, and I wasinvolved in numerous discussions of what became the 102Guidance Paper, and its implications for various cases, includ-ing Intel. I am not neutral in looking at what the Commis -sion did. That said I believe that the Commission deservescredit for its long project to reconsider Article 102.The Commission deserves credit because it was not forced

by the courts to engage in this reconsideration. The Europeancourts had pushed the Commission, in Article 101 cases,towards a greater use of economics in the assessment ofrestrictions on competition,3 but the impetus towards Article102 reform came not from the courts, but from within theCommission itself—albeit requested by companies and theiradvisers. There had been no major judicial annulments of anyCommission Article 102 cases based on a perception that theCommission’s decisions had unjustifiably failed to use eco-nomic analysis. Rather, it was the opposite. Typically, thecourts themselves were perceived as maintaining a doctri-naire and formalistic approach to Article 102 cases (for exam-ple, by insisting that conduct be merely “capable” of pro-ducing anticompetitive consequences, rather than being“likely” to do so, or by referring to a dominant company’s“special responsibility,”—a concept that frequently irks econ-omists because it appears unrelated to the economic conse-quences of conduct by a dominant firm).In the mid-2000s the Commission unilaterally embarked

upon a difficult and controversial road to re-evaluate itsenforcement policy for unilateral conduct, and this has nowled to the Court of Justice’s Intel ruling, “clarifying” theestablished case-law on rebates and giving it an unmistakeablepush in the direction of more economic analysis. The Intelruling thus vindicates the Commission’s largely voluntarydecision to travel down a controversial road But the vindication of more economic analysis is not the

only consequence of the Court of Justice’s Intel ruling. Twoother issues—jurisdiction and the Commission’s internalinformation-gathering procedures—are highly significant,and it is worth spending a little time on these before return-ing to what the Court said on rebates and abuse of domi-nance.

JurisdictionAdvocate General Wahl is a respected competition law spe-cialist, and his opinions are always worth reading. When hisopinion in Intel was released,4 it raised some eyebrows, not somuch for what he said on rebates, but rather for his views onthe jurisdiction of the Commission over competition lawinfringements.In its appeal, Intel had questioned the Commission’s juris-

diction regarding certain contracts that Intel had concludedwith Lenovo concerning CPUs for delivery in China. Intelhad argued that because these contracts were for the sale of

CPUs outside of the EU and for incorporation into productsthat would not be sold into the EU, then any restraints in thecontracts could not possibly infringe Article 101. Article 101,in Intel’s view, required some EU nexus.Advocate General Wahl agreed in part, and his opinion

sought both to broaden and narrow the Commission’s juris-diction. He sought to broaden the jurisdiction by proposingthat the Court expressly affirm Commission jurisdiction overconduct that had “qualified effects” within the EU5—effectswhich the General Court had defined in a merger case,Gencor, to be foreseeable, immediate, and substantial, aninterpretation which the Court of Justice had not previous-ly had occasion to support.6 The Court previously hadaffirmed jurisdiction only over conduct actually implement-ed within the EU7—while excluding conduct whose effectswere “too remote or purely hypothetical.”8

But while proposing to affirm a broadening of the juris-diction, Wahl nevertheless faulted the Commission’s deci-sion, as upheld by the General Court, as being too broad. Heargued that where contracts were concluded outside of theEU, for products that were manufactured and sold outside ofthe EU, and were not intended for further sale within theEU, then there would be no “foreseeable, immediate andsubstantial effects” within the EU that would bring suchcontracts within the jurisdiction of Article 102.In doing so, the Advocate General rejected the Commis -

sion’s reliance on a theory of a single and continuous infringe-ment (SCI). The Commission had held that all of the variousOEM contracts were part of a single anticompetitive courseof conduct, a plan designed to exclude AMD from the mar-ket. The Advocate General faulted this view, arguing that theEU law “concept of single and continuous infringement ismerely a procedural rule aimed at alleviating the evidentiaryburden of competition authorities.”9 It could not, in his view,provide a mechanism to capture conduct that would, in iso-lation, not constitute an infringement of the EU rules.The Court of Justice, however, rightly rejected Wahl’s

argument that the SCI is a purely administrative convenience.It is certainly procedurally efficient for the Commission to beable to take a set of conduct—say, a series of related cartelmeetings—and characterize them as a single infringement.Imagine a cartel that met every week for five years. Any onemeeting is sufficient to constitute an infringement, and afull analysis of each and every meeting would be repetitiveand redundant. But were the Commission of a mind to carryout that analysis, repetitive and redundant though it may be,would that represent merely a loss of procedural conven-ience? Put another way, would the Commission be entitledto characterize each meeting as a separate infringement? Thatseems unlikely. Regulation 1/2003 entitles the Commissionto levy a fine on each “infringement,” each fine being subjectto the overall cap of 10 percent of the firm’s global turnover.It seems unlikely that the Commission has the power to cir-cumvent the 10 percent cap simply by carrying out the pro-cedural inconvenience of analyzing each individual meeting

5 8 · A N T I T R U S T


S P R I N G 2 0 1 8 · 5 9

and then levying a separate fine for each meeting on thebasis that each was a separate infringement.Although it is undoubtedly convenient not to have to

carry out a full legal analysis for every individual element ofconduct that constitutes a whole course of action, to collapsethe whole of the SCI concept into one of simply adminis-trative convenience seems to ignore the real legal conse-quences of whether or not a set of conduct is a single infringe-ment.There are other conceptual problems with Wahl’s ap -

proach. For example, it ignores the possible relationshipbetween the constituent elements of the sets of conduct. Itseems uncontroversial that if Company A engages in a cartelwith Company B in the EU, and a separate cartel togetherwith Company C in Asia, then absent any links between theconduct, the two cartels would not be regarded as a singleinfringement. But if, for example, A agrees with B that Bwould not enter Asia, on the understanding that A wouldagree with C that C would not enter the EU, then that wouldbe at least some evidence of a single infringement rather thantwo separate ones—and would at least raise an argumentthat C has violated EU law.Further, the Intel case did not involve a cartel. The anti-

competitive effect of a cartel agreement among suppliers ispurchasers and ultimately consumers paying more for thecartelized product. So, if the purchasers and ultimately con-sumers were outside of the EU then there is clearly an argu-ment against such conduct being contrary to EU law. But theanticompetitive effects about which the Commission wasconcerned were not related to high prices (as would be thecase in a supplies cartel), or even the prices that were the sub-ject of the agreements with Intel. Rather they were about thepotential foreclosure of Intel’s main competitor at the time,AMD. If AMD were successfully foreclosed, that would havestrengthened Intel’s market power globally, including in theEU, no matter where—geographically—the source of thatforeclosure took place.Perhaps with these or similar arguments in mind, the

Court of Justice chose not to follow the Advocate General onthis point.The Court did follow AG Wahl in part, however, by

affirming the qualified effects test, albeit with rather briefanalysis. The Court noted that in previous cases it had justi-fied the implementation test on the basis that, “if the appli-cability of prohibitions laid down under competition lawwere made to depend on the place where the agreement,decision or concerted practice was formed, the result wouldobviously be to give undertakings an easy means of evadingthose prohibitions.”10 The Court then recognized that thisprevious justification of the implementation test also justifiesthe qualified effects test, which pursues the same objective.The Court accordingly rejected Intel’s ground of appeal onjurisdiction.The Court did not accept the Advocate General’s opinion

as to jurisdiction over the Lenovo contracts. AG Wahl’s opin-

ion asked that the Court should determine whether thesecontracts in themselves could have had sufficient effects onthe EEA.11 The Court of Justice rejected AG Wahl’sapproach, because it “would lead to an artificial fragmenta-tion of comprehensive anticompetitive conduct, capable ofaffecting the market structure within the EEA, into a collec-tion of separate forms of conduct which might escape theEuropean Union’s jurisdiction.”12 The Court considered itappropriate to take into account the “probable effects” of theconduct on competition,13 and the role of Intel’s conduct as“part of an overall strategy.”14

This seems right. In a broad anticompetitive strategy toforeclose a competitor, a dominant company might enterinto a wide range of anticompetitive conduct around theworld, encompassing dozens or hundreds of contracts. Takenin isolation, many of these individual contracts would nothave any appreciable effect in the EEA. But to exclude thosefrom the analysis of the overall anticompetitive conductseems artificial at best.

ProcedureThe Court also addressed a procedural issue. During theadministrative procedure, the Commission, as is typical inmany cases, had a meeting with a third party and discussedthe complainant’s allegations. What is perhaps less usual isthat this meeting was rather long—around five hours. In theaccess to file procedure—the system whereby parties to acase receive copies of exculpatory and inculpatory documentsin the Commission’s file—the Commission initially disclosedno documents to Intel relating to that meeting. When Intelpressed, the Commission disclosed a short aide-memoirewhich—the Commission admitted—was not written toreflect in full the content of the meeting.Intel argued that this violated its rights of defense as excul-

patory information may have been disclosed at that meetingon which it was prevented from relying given the Commis -sion’s failure to properly record the meeting. The General Court had rejected this argument and had (as

the Commission suggested) accepted a distinction betweenformal and informal meetings. The General Court alsoaccepted the Commission’s view that the meeting had beenan informal one and that the Commission had no obligationto take a full record. The Court of Justice took issue with this,calling into question the concept of “informal” meetings:“There is nothing in the wording of [Article 19(1) of Reg -ulation 1] or in the objective that it pursues to suggest thatthe legislature intended to establish a distinction betweentwo categories of interview relating to the subject matter ofan investigation or to exclude certain of those interviewsfrom the scope of that provision.”15

Nevertheless, the Court determined that although thefailure to take a full record of the meeting had infringedIntel’s rights of defense, the failure was not so serious as towarrant an annulment of the decision, because Intel had nottaken advantage of its various procedural rights, such as ask-


6 0 · A N T I T R U S T

ing the General Court to call the third party to give evidencebefore it.So, although Intel ultimately failed in its appeal on this

point, the Commission also failed to convince the Courtthat the Commission’s procedure (preparing a brief and sum-mary note of the meeting’s content, and making only thatnote available to Intel) was sufficient.Commission procedures are—up until now anyway—

fundamentally based on documents, either pre-existing ones,or ones drawn up for the purposes of the proceedings, suchas complaints, comments on complaints, and replies torequests for information. The Commission does not have aninquisitorial system; meetings take place, but typically for ori-entation rather than detailed fact-finding or inquisition. If aparty or third party has a meeting with the Commission, andthe Commission believes that the party has useful informa-tion, the Commission typically asks for a written submissionor follows up with a formal request for information.A formal power to interrogate witnesses was provided to

the Commission for the first time in Regulation 1/2003. Twopoints distinguish this power from the main Commis sionevidence-gathering tool, a request for information, whichmight explain the Commission’s relatively rare use of itspower to interrogate. First, the process is voluntary: the Com -mission has no power to compel testimony. Second, theCommission has no power to fine a witness (or the entity withwhich the witness is affiliated) for the provision of incompleteor misleading information. This latter point in particularlimits its use as a reliable evidence-gathering tool. What legalweight should be given to evidence—whether inculpatory orexculpatory—provided by a witness who was under no threatof penalty for lying or misleading.The Commission has already tightened up its procedures

compared to the position at the time of adoption of its Inteldecision in 2009. In addition to the Best Practice guidelines,attendees at DG Competition meetings are now routinelytold that a note will be made of the meeting and that the notemay be accessible to other parties to the procedure. It may be,however, that this does not go far enough to meet the require-ments of the Court because it may not be a “full recording”of the meeting.The question of whether or not there is a gap between the

Commission’s current (and improved) practice and theCourt’s requirements will be being discussed inside the DG.The discussions will need to cover:

� What constitutes a recording (audio, visual, detailednote-taking)?

� If detailed note-taking suffices, what constitutes “full”?� And in any event what constitutes a meeting where“information” is being received?

This last point is a particular problem. The Commissionhosts a range of meetings over the lifetime of a case, includ-ing:

� Preliminary, typically highly confidential, pre-complaintmeetings;

� Exploratory meetings with complainant, defendant,and sometimes third parties to understand the issues;

� Fact-gathering meetings that in the past would typical-ly be followed up by confirmatory requests for infor-mation. (These are the meetings that perhaps mostunambiguously will now need to be recorded in full,though how to distinguish these meetings from othersis not a trivial question.); and

� State of play meetings to discuss the procedural status of the case, and the Commission’s current thinking—though often these meetings in practice result in thepassing of “information” from the party to the Com -mis sion.

In addition, there are other (and even more specific) meet-ings, such as meetings under the cartel settlement procedure.These are akin to state of play meetings, but serve an evenmore specific purpose: broadly speaking, they provide a meansfor the Commission to set out its view of a cartel case that theybelieve suitable for settlement. But it is not unusual for thereto be at least some discussion with the party to the meetingwhere, it could be argued, “information” is conveyed. It seems unlikely that the Commission will seek to take full

recordings for all of these meetings. And, frankly, companiesand external counsel may not be happy if they did. Informal,practical discussions are helpful to all sides in progressing acase.It may be that the Commission will try to identify cate-

gories of meetings that will be recorded in full, perhaps in anupdate to the Best Practice guidelines, perhaps in a separatecommunication. Whatever the Commission decides to do,time is of the essence. The Court has not introduced a newrule on the recording of meetings; it has simply clarifiedwhat it sees as a legal obligation on the Commission whichbegan when the power to conduct interviews under Article 19was granted through Regulation 1/2003.

RebatesThe most anticipated part of the judgment is perhaps the partabout which the least can be said with any certainty: what isthe appropriate legal analysis in the EU of allegedly exclu-sionary rebates?The Commission decision condemned two broad groups

of conduct. The first, so-called naked restrictions, arose whenIntel approached OEMs that were contemplating launchingproducts with AMD chips, and gave them money not to doso. Such conduct by any company would be ethically dubious,but by a dominant one, plainly illegal. It is hard to think of amore compelling example of not competing on the merits.The second group involved exclusivity or loyalty rebates.

Intel priced its own products to OEMs in such a way that—in the view of the Commission and the General Court—theprices in practice amounted to an exclusive relationshipbetween Intel and each OEM. It was the analysis of this sec-ond group that was the focus of Intel’s appeal to the Courtof Justice.

S P R I N G 2 0 1 8 · 6 1

The 2009 Commission decision took a belt and bracesapproach to these exclusivity rebates. First, it carried out a“traditional,” form-based legal analysis, running from pages274 to 301 of the decision, which concluded that “[t]herebates and payments in question constitute fidelity rebateswhich fulfil the conditions of the relevant case law for qual-ification as abusive.”16

But the decision then continued, engaging in a muchmore detailed economic analysis of whether the rebates wouldexclude an “as efficient competitor.” This “as efficient com-petitor” analysis ran from pages 302 to 472 of the Commis -sion’s decision, and it looked at a range of factors for eachOEM, including the size and nature of the rebates, the vol-ume purchased, costs, and the contestable and non-con-testable parts of the market.The General Court looked at the traditional analysis,

agreed with the Commission, and then determined it need gono further in its own analysis. Specifically, it determined thatit did not need to conduct a detailed review of the Commis -sion’s economic “as efficient competitor” analysis—althoughit did raise some questions as to the difficulties of applyingthe test.Intel appealed, arguing, in pertinent part, that the General

Court should have looked at the Commission’s economicanalysis. The Court of Justice ruled in favor of Intel: it did not annul the Commission decision but instead referred thecase back to the General Court, with instructions for theGeneral Court to review the economic analysis in the Com -mission decision.Unfortunately, the Court of Justice ruling is not a model

of clarity, and it provides little guidance as to how the analy-sis should be conducted.One could view the Court’s ruling as merely identifying

a procedural flaw in the General Court’s review of the Com -mission decision: if the Commission decides, of its own volition, to rely on economic evidence in its assessment ofwhether particular conduct is abusive, then the GeneralCourt is obligated to review that evidence, and the Commis -sion’s analysis, to determine whether the Commission’s con-clusion is valid.In support of this “procedural” interpretation of Intel,

one can point to paragraph 141 of the Court’s ruling:

If, in a decision finding a rebate scheme abusive, the Com -mission carries out such [a foreclosure] analysis, the GeneralCourt must examine all of the applicant’s arguments seekingto call into question the validity of the Commission’s find-ings concerning the foreclosure capability of the rebate con-cerned.

In itself that paragraph might suggest that the foreclosureanalysis is voluntarily undertaken. But that interpretationsits uneasily with some of the earlier paragraphs in the Court’sjudgment. In paragraphs 134 to 137 the Court rehearses theprecedents on Article 102: that competition on the meritsmay foreclose less efficient competitors; that dominantundertakings have a special responsibility; that Article 102

prohibits exclusionary pricing practices (though note thatthe French text of this paragraph refers to “exclusionary prac-tices”); and that exclusive or mostly exclusive purchasingarrangements, including those where the exclusivity resultsfrom the effect of rebates, may be abusive. That last para-graph, 137, explicitly cites Hoffmann-La Roche.17

And it is then paragraphs 138 and 139 that appear to bekey:

However, that case-law must be further clarified in the casewhere the undertaking concerned submits, during theadministrative procedure, on the basis of supporting evi-dence, that its conduct was not capable of restricting com-petition and, in particular, of producing the alleged foreclo-sure effects.

In that case, the Commission is not only required to analyse,first, the extent of the undertaking’s dominant position on therelevant market and, secondly, the share of the market coveredby the challenged practice, as well as the conditions andarrangements for granting the rebates in question, their dura-tion and their amount; it is also required to assess the possi-ble existence of a strategy aiming to exclude competitors thatare at least as efficient as the dominant undertaking from themarket (see, by analogy, judgment of 27 March 2012, PostDanmark, C-209/10, EU:C:2012:172, paragraph 29).

How should we interpret paragraphs 132 to 139 in lightof paragraph 141 discussed above? The Court was sitting asa Grand Chamber of 15 judges, rather than the more usualthree or five. This is done rarely, typically in particularlycomplex or important cases. If a Court cites a long-estab-lished precedent like Hoffmann-La Roche, and then indicatesthat the precedent needs “clarification,” and if the Court isdoing that in the Grand Chamber, then the “procedural”interpretation of paragraph 141 seems unlikely. Rather, theCourt seems to be moving the case law, at least a little, awayfrom per se rules and in the direction of more economicanalysis, and setting out a number of factors that the Com -mis sion must analyze in determining whether an as-efficientcompetitor would be excluded.The Court continued in paragraph 140 with other indi-

cations of what the Commission should look at:

The analysis of the capacity to foreclose is also relevant inassessing whether a system of rebates which, in principle, fallswithin the scope of the prohibition laid down in Article 102TFEU, may be objectively justified. In addition, the exclu-sionary effect arising from such a system, which is disadvan-tageous for competition, may be counterbalanced, or out-weighed, by advantages in terms of efficiency which alsobenefit the consumer (judgment of 15 March 2007, BritishAirways v Commission, C-95/04 P, EU:C:2007:166, para-graph 86). That balancing of the favourable and unfav -ourable effects of the practice in question on competition canbe carried out in the Commission’s decision only after ananalysis of the intrinsic capacity of that practice to foreclosecompetitors which are at least as efficient as the dominantundertaking.

The English version of paragraph 140 set out here is notthe one issued on September 6, 2017, but is the one which


6 2 · A N T I T R U S T

follows two separate rectification orders of the Court—onSeptember 19 and October 24—amending the language andbringing it into line with what presumably (given the work-ings of the Court) was the drafting language, French. One ofthe amendments is significant in that it emphasizes that thebalancing of negative effects and efficiencies is a separateexercise from analyzing the capacity of the conduct to fore-close.18 The Court seems to be emphasizing here that (1) evenwith rebates that under established precedent would be un -lawful, (2) the Commission needs to analyze the capacity of those rebates to foreclose an as-efficient competitor, and(3) only if those rebates would foreclose an as-efficient com-petitor should the Commission look at possible efficiencyarguments to justify the rebates.Taking paragraphs 132 to 140 as a whole, it seems that the

Commission must look at all of the relevant factors, includ-ing the impact of the rebates on competition from an as-effi-cient competitor, and the possible efficiencies that flow fromthe conduct.The ruling still leaves a number of areas unclear, how-

ever.First, even with this “clarification” of the case law, the

Commission does not appear legally obliged always to carryout this economic analysis, even in rebates cases. The legalobligation kicks in only when “the undertaking concernedsubmits, during the administrative procedure, on the basis ofsupporting evidence, that its conduct was not capable ofrestricting competition and, in particular, of producing thealleged foreclosure effects.”19

We can assume that properly advised undertakings willalways provide such evidence. But at what stage will they doso? There presumably is no obligation to submit such evi-dence before a Statement of Objections. An undertakingmight wish to do so early in the case, in order to convince theCommission that the rebates were procompetitive. Alterna -tively, an undertaking may choose to wait until it has seen theStatement of Objections and to put the arguments in itsReply—either to understand better the allegations against itbefore engaging on the substance or, more opportunistical-ly, to delay proceedings. For good reasons or bad, the Court’sformulation that this is an obligation on the Commission torespond to such evidence rather than to make its case pro-actively, seems to make Supplementary Statements of Objec -tions and longer proceedings more likely.It is perhaps also conceivable that the Commission might

look at the “supporting evidence” provided by the under-taking and conclude that it is not adequate to trigger the obli-gation on the Commission to carry out a detailed economicassessment. That would of course be a high-risk strategy forthe Commission, and it is perhaps not one that the Com -mission would be likely to take, but the option seems to beopen.The second area where the ruling is perhaps unclear is in

relation to the as-efficient competitor test. Paragraph 139provides a strong endorsement of the test but leaves several

issues unaddressed: When will the test be relevant? In allrebates cases? In pricing conduct cases but not non-priceconduct case? In abuse cases more generally? The Court issilent on this.And what does the test require? The Commission’s deci-

sion contained nearly 200 pages of economic analysis, as aresult of many months of work by a small army of PhDeconomists. Is that required in every case? One could imag-ine, instead of 200 pages of econometric analysis, a ten-pagequalitative analysis of the economic characteristics of themarket, of the dominant firm, and of the competitor. Wouldthis be sufficient for pricing conduct? The Court of Justice’sruling does not say. There is another concern, and this is where the Court’s

ruling may have been over-hasty. Although Intel’s appealcriticized the essentially per se characterization of loyaltyrebates in the case-law, it also raised several criticisms of theas-efficient competitor test. As the General Court essential-ly skipped over the economic analysis of the Commission,these were never fully considered. There are also numerouscriticisms of the test in academic literature. What is the fatenow of these criticisms, after the Court of Justice’s endorse-ment of the as efficient competitor test?Given the Court of Justice’s ruling, the General Court will

find it hard to address some of these questions. The Court ofJustice has not only endorsed the as-efficient competitor test,but it has explicitly told the General Court that if the Com -mission conducts a foreclosure analysis, the General Courtmust review the entirety of that analysis. So, a review of thenearly 200 pages is inevitable. The General Court could saythat it believes that this degree of economic analysis is unnec-essary, but it would seem difficult for the Court to acceptIntel’s arguments that the as-efficient competitor test isflawed, now that the Court of Justice has given its blessing tothat test.

ConclusionPerhaps inevitably, the Court’s ruling has left issues unre-solved, but one cannot help but feel—reading the judg-ment—that this case leaves more unresolved than is usual.On procedure, the judgment does not grapple with the

tension between the formal, investigative, typically written,procedure of the Commission and the less formal conversa-tions that inevitably take place in any case in any system ofenforcement. Understandable perhaps, in the context of alargely unminuted five-hour meeting, but nonethelessunhelpfully incomplete in its analysis of the implications ofthe case.On substance, different paragraphs of the judgment point

in different directions, and the interpretation provided inthis article is just that—an interpretation. Other readers mayinterpret the ruling differently. Although the Court should be applauded for “clarifying”

the case law and moving it in the direction of more eco-nomic analysis, the way that the Court has done so leaves too

S P R I N G 2 0 1 8 · 6 3

many issues under-analyzed. It might have been better for theCourt of Justice to restrict itself to noting that a form-basedprohibition on rebates was inappropriate, and that someform of analysis of the rebates’ economic impact was neces-sary. Setting out rules on the form of that economic analysiswithout the issue’s first having been debated fully in thelower court, and then appropriately appealed, might prove tohave been unwise.�

1 Case C-3/37.990––Intel, Comm’n Decision (summary at 2009 O.J. (C 227)13), http://ec.europa.eu/competition/antitrust/cases/dec_docs/37990/37990_3581_18.pdf [hereinafter Commission Intel Decision); Case T-286/09, Intel v. Comm’n, ECLI:EU:T:2014:547 (GC June 12, 2014), http://curia.europa.eu/juris/document/document.jsf?text=&docid=194082&pageIndex=0&doclang=EN&mode=lst&dir=&occ=first&part=1&cid=1613407; Case C-413/14 P, Intel v. Comm’n, ECLI:EU:C:2017:632 (CJ Sept. 62017) [hereinafter Court of Justice Intel Decision], http://curia.europa.eu/juris/document/document.jsf?text=&docid=194082&pageIndex=0&doclang=EN&mode=lst&dir=&occ=first&part=1&cid=76193). For an overviewof the facts of the case and the Commission’s thinking, see NicholasBanasevic & Per Hellström, When the Chips Are Down: Some Reflections onthe European Commission’s Intel Decision, 1 J. EUR. COMPETITION L. &PRAC. 301–10 (2010).

2 Guidance on the Commission’s Enforcement Priorities in Applying Article 82of the Treaty [Now Article 102 TFEU] to Abusive Exclusionary Conduct byDominant Undertakings, 24 February 2009, 2009 O.J. (C 45) 7.

3 See, e.g., Case T-374/94, European Night Servs. v. Comm’n, 1998 E.C.R.II-03141, ECLI:EU:T:1998:198.

4 Case C-413/14 P, Intel v. Comm’n, Opinion of Advocate General Wahl (Oct.20, 2016), ECLI:EU:C:2016:788 [hereinafter Wahl Opinion].

5 Id. at 296. 6 Case T-102/96, Gencor Ltd v. Comm’n, 1999 E.C.R. II-00753, ECLI:EU:T:1999:65.

7 Joined cases C-89/85, C-104/85, C-114/85, C-116/85, C-117/85 and C-125/85 to C-129/85, Ahlström Osakeyhtiö v. Comm’n, 1993 E.C.R. I-01307, ECLI:EU:C:1993:120.

8 Wahl Opinion, supra note 4, at 299. 9 Id. at 319.

10 Court of Justice Intel Decision, supra note 1, at 44. 11 Wahl Opinion, supra note 4, at 325 & 326. 12 Court of Justice Intel Decision, supra note 1, at 57. 13 Id. 44. 14 Id. at 52. 15 Id. at 87. 16 Commission Intel Decision, supra note 1, ¶ 1001. 17 Case 85/76, Hoffmann-La Roche v. Comm’n, 1979 E.C.R. 461, EU:C:

1979:36, ¶ 89. 18 Because the original, unamended, English-language ruling is no longer avail-

able on the Court’s website, I set out here the evolution of paragraph 140.From the original ruling:

The analysis of the capacity to foreclose is also relevant in assessingwhether a system of rebates which, in principle, falls within the scope ofthe prohibition laid down in Article 102 TFEU, may be objectively justified.It has to be determined whether the exclusionary effect arising from sucha system, which is disadvantageous for competition, may be counterbal-anced, or outweighed, by advantages in terms of efficiency which alsobenefit the consumer. (Judgment of 15 March 2007, British Airwaysv. Commis sion, C95/04 P, EU:C:2007:166, paragraph 86).

On September 19, the words “In addition” were added at the start of thesecond sentence:

The analysis of the capacity to foreclose is also relevant in assessingwhether a system of rebates which, in principle, falls within the scope ofthe prohibition laid down in Article 102 TFEU, may be objectively justified.In addition, it has to be determined whether the exclusionary effect arisingfrom such a system . . . . (emphasis added).]

And then on October 24, the phrase “it has to be determined whether” wasdeleted from the same sentence, leaving the final (presumably) version ofthat paragraph to read:

The analysis of the capacity to foreclose is also relevant in assessingwhether a system of rebates which, in principle, falls within the scope ofthe prohibition laid down in Article 102 TFEU, may be objectively justified.In addition, it has to be determined whether the exclusionary effect arisingfrom such a system, which is disadvantageous for competition, may becounterbalanced, or outweighed, by advantages in terms of efficiency whichalso benefit the consumer . . . .

19 See Statements of Advocate General Kokott from January 2018, reportedat https://globalcompetitionreview.com/article/1153170/ecj-ag-kokott-no-need-for-”infamous”-test-after-intel: “Kokott noted that the likely result of theIntel judgment is that non-rebuttable presumptions will “cease to exist” inabuse of dominance cases, at least as far as rebates are concerned.”

ABA Section of Antitrust Law

Student WritingCompetition Winner

CLAYTON J. MASTERMAN

is a fifth-year student in Vanderbilt

Law School’s J.D./Ph.D. program in

law and economics. He has served as an Articles

Editor on the Vanderbilt Law Review, a Janet D.

Steiger Fellow in the Office of the Attorney

General of Tennessee, and as a summer legal

intern in the Federal Trade Commission’s Bureau

of Competition. He holds a Bachelor of Arts in

Economics and Mathematics from Vassar

College.

Clayton’s winning article is titled The

Customer Is Not Always Right: Balancing Worker

and Customer Welfare in Antitrust Law, and was

published in 69 VAND. L. REV. 1387 (2016).

In Summer 2018, Clayton will be a summer

associate at Skadden, Arps, Slate, Meagher &

Flom LLP’s Washington, D.C. office. After

graduation, he will be clerking for Judge Gregg

J. Costa of the Fifth Circuit Court of Appeals.


6 4 · A N T I T R U S T

Mind the Gap:Merger Efficiencies in the United States and Canada

B Y B R I A N F A C E Y , N A V I N J O N E J A , P A U L C U O M O , A N D J E F F R E Y O L I V E R

scale and scope in production, staff rationalization, financialsynergies (e.g., lower cost of capital), and other synergies.4

Intergovernmental agencies such as the Organization forEconomic Co-operation and Development (OECD), as wellas most antitrust authorities, have generally acknowledged theimportance of an assessment of efficiencies in merger review.5

The International Competition Network recommends thatcompetition agencies include an assessment of potential effi-ciencies in their overall merger review analytical framework,noting that certain efficiencies “may bring synergies on apotentially continuous basis, thus enhancing the potentialperformance of the merged entity and the potential benefitto competition and consumers.”6

While the U.S. Horizontal Merger Guidelines expresslyrecognize that “a primary benefit of mergers to the economyis their potential to generate significant efficiencies,”7 theU.S. Supreme Court has not expressly recognized an effi-ciencies defense for mergers. However, the Sixth, Eighth,Eleventh, and District of Columbia Circuits, and mostrecently, the Ninth Circuit, have at least suggested that effi-ciencies could save the day for the right merger—althoughthe Ninth Circuit recently cautioned that “we remain skep-tical about the efficiencies defense in general and about itsscope in particular.”8

By contrast, the Supreme Court of Canada has recentlyapproved a merger to monopoly based on the efficienciesdefense, noting that only “marginal efficiency gains arerequired for the defense to apply.”9 Questions remain in bothcountries and in cross-border cases as to how efficiencies are to be treated in strategic mergers. This is particularlyimportant given the 2014 publication of the Best Practices onCo operation in Merger Investigations by the U.S. antitrustagencies and the Bureau, which encourages them to worktogether.10

Regardless of whether merger laws give primacy to eco-nomic efficiency (as in Canada) or incorporate economicefficiency as an element in merger review (as in the UnitedStates), there is significant value in ensuring consistent resultswith conclusions that are supported by sound economic prin-

THE PEACE BRIDGE, LOCATED ATthe end of Lake Erie, connects the United States to Canada. Opened in 1927, it wasnamed to commemorate 100 years of peacebetween the two countries and remains one of

North America’s most important commercial ports.Notwithstanding the close ties between Canada and the

United States, there is an important difference in the way thatmergers are reviewed on either side of the border. In Canada,the efficiencies defense is typically credited and increasinglydetermines the outcome of transactions that would otherwisebe considered anticompetitive under the law. In the UnitedStates, efficiencies are seldom credited and almost never influ-ence the outcome of mergers that are otherwise deemed anti-competitive. This is important because a significant numberof mergers are reviewed by both the Canadian CompetitionBureau (the Bureau) and the U.S. antitrust agencies.1

The differing treatment of efficiencies can lead to differ-ent results on different sides of the border.2 This is despite thefact that U.S. government officials stated in the aftermath ofGE/Honeywell that convergence is desired whenever possi-ble to avoid one authority blocking a transaction that may beprocompetitive and efficiency-enhancing in another.3 As dis-cussed in this article, when it comes to efficiencies, the dif-ferences are greater than the similarities.

BackgroundEfficiencies remain the primary rationale for almost all merg-ers; they allow the merging firms to achieve economies of

Brian Facey and Navin Joneja are Partners in Blake, Cassels & Graydon

LLP. Paul Cuomo is a Partner and Jeffrey Oliver is a Senior Associate at the

Washington, D.C. office of Baker Botts LLP. The authors thank David Dueck

and Robyn Burns of Blake, Cassels & Graydon LLP for their assistance in

preparing this article. Brian Facey and Navin Joneja act as counsel to

Superior Plus Corp. Paul Cuomo and Jeffrey Oliver acted for Superior Plus

Corp. regarding its proposed acquisition of Canexus Corp. in 2015. The

opinions expressed herein are those of the authors and do not necessarily

reflect the views of either law firm or its clients.

S P R I N G 2 0 1 8 · 6 5

In Tervita, the Supreme Court of Canada confirmed theimportance of efficiencies to Canadian competition law andheld that a transaction cannot be blocked if the efficienciesoutweigh the anticompetitive effects, even if only by a smallamount.16 The Supreme Court of Canada also clarified that,in a Section 96 case, the Bureau has the burden of quantify-ing the anticompetitive effects of a merger, while the merg-ing parties have the burden of quantifying the efficiencies.17

Following the Tervita decision, Superior/Canexus was thefirst case to be cleared publicly by the Bureau on efficiencies.18

The Bureau had previously referred to efficiencies as a factorleading it to clear prior mergers,19 but in no prior case had theBureau concluded that claimed efficiencies were sufficient tocompensate for a substantial lessening of competition thatwould otherwise require a remedy.20 Subsequently, the Bureauhas taken merger efficiencies into account in a number ofcases—including Chemtrade/Canexus21 and First Air/CalmAir22—but it is not clear whether the Bureau would havesought to block these mergers if not for the efficiencies.23

More recently, the Bureau utilized a different approach to efficiencies during its review of Superior’s acquisition ofCan west Propane—a case involving retail distribution ofpropane in Western Canada.24 The Bureau’s position state-ment explains that it took a “market-by-market” approach tothe efficiencies trade-off in Section 96, whereby the efficien-cies in each local market were compared to the anticompet-itive effects in each local market, rather than across the trans-action as a whole.25 We expect that the Bureau’s justificationfor this approach is likely that it enables the Bureau to resolveanticompetitive effects in those markets with few efficiencies,while allowing merging parties to still achieve efficiencies inother markets with relatively fewer anticompetitive effects. The legal basis for this market-by-market approach to the

efficiencies trade-off analysis appears questionable. The word“market” does not appear in the wording of the efficienciesdefense set out in Section 96 of the Act. The language ofSection 96 also appears to contemplate a single order beingapplied to the merger as a whole for the efficiencies trade-off(rather than separate “orders” in each relevant market).26 Amarket-by-market approach to Section 96 was not adoptedin either of the relevant leading decided cases. In fact, theTribunal explicitly rejected such an approach in Canada(Commissioner of Competition) v. Superior Propane, Inc., stat-

ciples. Economics provides an objective basis for assessingantitrust issues, which is important when agencies may befeeling pressure to take into account more political factors,such as the impact on employment. As new issues emerge,particularly in terms of the assessment of vertical mergers, ormergers that can result in significant dynamic efficienciesand innovation, tools that are used to evaluate economicefficiency offer the potential for an objective approach tosuch assessments, often through sophisticated and quantifi-able economic analysis.

Prominence of Efficiencies in Canadian Merger Review Canada has a strong track record of crediting efficiencies inmerger reviews, and even clearing transactions that wouldotherwise be challenged if not for the claimed efficiencies. InCanada, primacy is given to economic efficiency as a statu-tory objective in Section 96 of the Canada Competition Act(the Act),11 which provides a defense to mergers that are oth-erwise likely to lessen or prevent competition if efficienciesfrom the merger are likely to be greater than and offset themerger’s anticompetitive effects.12 As the Supreme Court ofCanada recognized in Tervita Corp. v. Canada (Commis sionerof Competition), this efficiencies defense was introduced fol-lowing a report from the Economic Council of Canada that“identified economic efficiency as the overriding policy objec-tive” behind reforming the Act.13

The efficiencies defense in Section 96 of the Act recognizesthe important benefits that mergers can generate for the econ-omy through cost savings and economies of scale, which makethe Canadian economy more competitive and more efficient.At the same time, the provision acknowledges that certainmergers can also have negative impacts on competition, andseeks to balance these factors to determine whether a mergerwill result in a net economic benefit to the Canadian econo-my. There is no requirement that efficiencies be passed on toconsumers; therefore, fixed cost savings are an important con-sideration in addition to variable cost savings. As a result,efficiencies could save a merger to monopoly in Canada, evenif it leads to higher prices or less choice for consumers. In economic terms, Section 96 posits a cost-benefit analy-

sis that aims to maximize total surplus (the sum of producersurplus and consumer surplus) for Canadian society. Thistotal surplus approach set out in Canada’s statute has ledcommentators to call it the “most economically literate”competition law in the world.14 As set out in the MergerEnforcement Guidelines, the Bureau will exclude efficiencygains that would have been achieved through alternativemeans even if an order from the Canadian CompetitionTribunal were issued; efficiency gains that would not beaffected by a Tribunal order; gains that are redistributive innature (e.g., cost reductions from increased bargaining lever-age); efficiency gains that do not accrue to the benefit of theCanadian economy; and savings resulting from a reductionin output, service, quality, or product choice.15

Regardless of whether merger laws give primacy to

economic eff iciency (as in Canada) or incorporate

economic eff iciency as an element in merger review

(as in the United States), there is signif icant value

in ensuring consistent results with conclusions that

are suppor ted by sound economic principles.


6 6 · A N T I T R U S T

ing that the efficiencies do not need to exceed the anticom-petitive effects in every single market for the efficienciesdefense to apply, as long as the efficiencies exceed the effectsas a whole.27 The Supreme Court of Canada also stated inTervita that Section 96 prevents a remedial order from beingissued “if it is found that the merger is likely to bring aboutefficiencies that are greater than and will offset the anticom-petitive effects resulting from the merger.”28 This suggeststhat the efficiencies trade-off is to be carried out across themerger as a whole rather than separately for each market,which is consistent with the plain reading of Section 96.29

Primacy of Consumer Welfare in U.S. Merger Reviews30The U.S. Horizontal Merger Guidelines recognize the sig-nificant benefits that efficiencies can create for consumers,including “lower prices, improved quality, enhanced service,or new products.”31 While efficiencies have been recognizedby U.S. jurisprudence, unlike in recent decisions in Canada,they are rarely endorsed as a defense that will save an other-wise anticompetitive merger.32 In the United States, the effi-ciencies defense generally lands like a dubious alibi—neces-sarily considered but very seldom credited. This is because U.S. “antitrust laws give competition, not

internal operational efficiency, primacy in protecting cus-tomers.”33 Efficiencies are assessed as part of the overall anal -ysis of anticompetitive effects, and the Horizontal MergerGuide lines state that a merger will not be challenged wherecognizable efficiencies exist such that the merger is not likelyto harm customers.34 In economic terms, efficiencies in theUnited States are effectively assessed under a consumer surplusstandard, where the focus is on the direct impact on con-sumers. This creates tension with the Canadian approach,which, as determined by the Supreme Court of Canada, alsotakes producer surplus into account.35

The Horizontal Merger Guidelines note that efficiencies aremost likely to make a difference to the U.S. agencies’ analy-sis of a merger when the likely anticompetitive effects are notsignificant in the first place. Efficiencies will therefore “almostnever justify a merger to monopoly or near monopoly.”36

The roots of U.S. skepticism toward the efficienciesdefense originated in early cases, such as FTC v. Procter &Gamble Co., where the Supreme Court held that “[p]ossibleeconomies cannot be used as a defense to illegality. Congresswas aware that some mergers which lessen competition mayalso result in economies but it struck the balance in favor ofprotecting competition.”37

While subsequent courts have acknowledged the possibil-ity of an efficiencies defense, they have mostly done so grudg-ingly and with little inclination to actually credit relevantclaims.38 In St. Alphonsus Medical Center—Nampa Inc. v. St.Luke’s Health System, Ltd., the U.S. Court of Appeals for theNinth Circuit noted that “a defendant can rebut a prima faciecase with evidence that the proposed merger will create amore efficient combined entity and thus increase competi-

tion.”39 However, the Ninth Circuit nonetheless rejected thedefendants’ efficiencies defense and, looking beyond the caseimmediately before it, also stated that “we remain skepticalabout the efficiencies defense in general and about its scopein particular.”40

More recently, the U.S. Court of Appeals for the D.C.Circuit affirmed the district court’s decision blocking theproposed Anthem/Cigna merger. The focus on appeal wasthe lower court’s treatment of Anthem’s efficiencies defense,and the D.C. Circuit questioned whether efficiencies are aviable legal defense under Section 7 of the Clayton Act inlight of Procter & Gamble.41 The D.C. Circuit ultimatelydeclined to determine whether efficiencies could save an oth-erwise illegal merger and went on to consider but ultimate-ly discounted the efficiencies presented by Anthem.42

The D.C. Circuit found that Anthem’s claimed efficien-cies were not merger-specific because Anthem failed to showthat it could not have achieved the efficiencies on its ownabsent the merger.43 This approach is consistent with theHorizontal Merger Guidelines, which require that efficienciesnot be achievable through other less anticompetitive means.44

In Canada, by contrast, merger-specificity simply requiresthat efficiency gains would not have been achieved absent themerger,45 and does not depend upon whether such gainscould theoretically have been achieved in another, less anti-competitive way.46

While the efficiencies defense faces serious challenges whena transaction ends up in a U.S. courtroom, they are relevantto the prosecutorial discretion of the antitrust agencies. Therevisions made in 1997 to the 1992 Horizontal Merger Guide -lines were intended to bolster the relevance of the efficienciesdefense in merger review,47 and the 2010 Horizontal MergerGuidelines largely echo this.48 However, despite the recogni-tion by the U.S. antitrust agencies that efficiencies are relevantto enforcement decisions, it can be difficult to gauge just howinfluential such efficiencies actually are in practice.49 Lookingforward, it also remains to be seen how the Trump adminis-tration will treat efficiencies, as Republican-led antitrust agen-cies have historically given more credence to such claims.Even though there is some resistance to taking efficiencies

into account in horizontal mergers, there has been a greaterwillingness to accept efficiencies in the context of verticalmergers and innovation markets given the consumer welfarestandard of U.S. antitrust law. For instance, the eliminationof double-marginalization from vertical mergers tends tolower the prices paid by consumers, while innovation leads tonew or higher quality products and services for consumers.50

For example, the Federal Trade Commission cleared theSynopsys/Avant! merger in 2002 partly based on the poten-tial efficiencies resulting from integrating Synopsys’ front-endand Avant!’s back-end electronic design automation tools forthe design of integrated circuits.51

With respect to innovation, the Horizontal Merger Guide -lines state that the U.S. agencies will consider innovation,including the ability of merged firms to more effectively carry

S P R I N G 2 0 1 8 · 6 7

ginalization (but no other efficiencies), the procompetitiveaspects of vertical mergers are well established.58 The exerciseof quantifying anticompetitive effects ensures that these pro-competitive factors are properly taken into account. Thissuggests that the Bureau should be obligated to quantify theanticompetitive effects of vertical mergers before the Tribunal,regardless of whether the elimination of double marginaliza-tion is treated as an increase in efficiency or reduction in anti-competitive effects.Similarly, dynamic efficiencies result in substantial

improvements to long-term social welfare,59 which suggeststhat dynamic efficiencies should readily be taken into accountin merger review. As Gary Roberts and Steven Salop explain,“Efficiency improvements are not static, one-time-onlyevents. Rather, they occur as part of a rich dynamic processin which efficiency improvements are introduced for privategain but then frequently stimulate competition that createssignificant spill-over benefits for consumers.”60

In Canada, the Bureau’s Merger Enforcement Guidelinesnote that the Bureau assesses dynamic efficiencies, such as the introduction of more efficient processes and improve-ments in product quality and service.61 Moreover, the Bur -eau’s 2015–2018 Strategic Vision states that the Bureau willenforce the Act to ensure that “innovative business models areencouraged.”62 However, this does not appear to have beentaken into account by the Bureau in recent merger reviews,and we are not aware of any mergers the Bureau has clearedon dynamic efficiencies that would otherwise have been chal-lenged. Canadian authorities would benefit from the juris -prudence in the United States on vertical mergers and inno-vation in considering such factors. In the United States, the ongoing uncertainty regarding

the role of efficiencies has the potential to result in divergentoutcomes from those of Canada. The U.S. courts andenforcement agencies appear to recognize that efficienciescan lead to lower prices and higher quality goods and servic-es for consumers, but lack a concrete and consistently appliedframework for incorporating and crediting efficiencies. Inthis respect, the United States may learn from the benefitsobserved in Canada from clearly quantifying anticompetitiveeffects and cognizable efficiencies, which provides a moreobjective basis for determining whether the consumer surplusstandard has been satisfied. There may also be analytical rationales in the United States

to giving greater weight to efficiencies in certain contextswhere the consumer welfare impact of a transaction is lessstark. These may include cases involving intermediate prod-ucts (e.g., a merger of companies that produce a small inputinto a larger consumer-facing product), large customers,and/or bargaining between sophisticated parties (large buy-ers and sellers) where a price increase may result in a transferof economic surplus but does not have a significant impacton the consumer surplus of the end-customers. A further topic that both the United States and Canada

will confront in the near future relates to merger reviews that

out research and development activities. Even though theU.S. agencies view such efficiencies as “generally less suscep-tible to verification and may be the result of anticompetitiveoutput reductions,”52 the FTC cleared Genzyme’s acquisi-tion of Novazyme in 2004 partly on the basis that the com-plementary skills and expertise of each firm would acceleratethe development of a drug for the treatment of Pompe dis-ease.53 Similarly, in 2011 the DOJ considered Google’s claimthat its acquisition of ITA would allow Google to innovatemore effective flight search services. The DOJ ultimately lim-ited its required remedy to retain the claimed efficiency.54

Scope for Greater Consideration of Efficiencies inCanada and the United StatesWhile Canada now has a well-established body of cases inwhich efficiencies considerations have played a prominent(and in some cases dispositive) role, each of the cases entaileda predominantly horizontal theory of harm. This has requireda somewhat predictable trade-off analysis comparing the cog-nizable efficiencies under Section 96 and the anticompetitiveeffects (deadweight loss) arising from the horizontal aspectsof the merger. To date, neither the Bureau nor the courts inCanada appear to have addressed whether or how to utilizethe Section 96 framework in the context of vertical integra-tion issues or other theories of harm/efficiency, such as harmor benefit to innovation. The Section 96 framework is wellpositioned to assess these types of antitrust issues, and thiscould represent the next frontier for mergers relying on theefficiencies defense in Canada. While questions remain onhow vertical issues or innovation aspects of a merger wouldultimately be addressed under Section 96, the frameworkfor assessing these topics is foreshadowed in the Bureau’sMerger Enforcement Guidelines and other materials. For example, a key question for future merger reviews in

Canada and the United States is how efficiencies arisingfrom vertical integration should be assessed. The Bureau’sMerger Enforcement Guidelines note that vertical mergers“frequently create significant efficiencies,” and refer to theelimination of double marginalization as an example.55 InCanada, it is not clear whether the elimination of doublemarginalization would be treated as an increase in allocativeefficiency or as a reduction in the anticompetitive effectsbrought about by a merger.56 While this distinction shouldnot matter for purposes of the outcome of the efficienciestrade-off, it may matter from a legal standpoint whetherthe Bureau has the burden of quantifying the anticompeti-tive effects. In Tervita, the Supreme Court of Canada heldthat the Bureau only has the legal obligation to quantify theanticompetitive effects when the efficiencies defense inSection 96 is invoked, but does not need to do so for pur-poses of determining whether a merger will lead to a sub-stantial lessening of competition under Section 92.57

Although it remains an open question whether the Bureauhas an obligation to quantify the anticompetitive effects forvertical mergers resulting in an elimination of double mar-

6 8 · A N T I T R U S T


require consideration of the potential competitive effects aris-ing from the use of data. Here too, efficiencies considerationsmay be able to play a useful analytical role. For instance, amerged firm may be able to more efficiently use the com-bined data set than either merging party could individually.Or it may prove to be more economically efficient for amerged company to utilize a data set (e.g., because of networkeffects) rather than requiring a merged firm to allow access tothe data by competitors. These are important issues that areat the heart of antitrust analysis The extent to which efficiencies will be taken into account

in future merger reviews remains an ongoing question. Agen -cies in Canada and the United States have much that they canlearn from each other’s jurisprudence regarding the consid-eration of efficiencies in merger review.

Best Practices for Merger Efficiency ClaimsThe role of efficiencies in merger review routinely generatessignificant debate in antitrust circles and, over the last 20years, the role of efficiencies has emerged as one of only a fewsubstantive differentiators between merger review in Canadaand the United States. In Canada, efficiencies can be used asa defense to preclude the blocking of a merger that might oth-erwise result in a substantial lessening or prevention of com-petition. In the United States, efficiencies are assessed as oneof several evaluative factors that help determine whether atransaction will result in a substantial lessening or preventionof competition, and typically need to be passed on to con-sumers in order to impact the merger review, but even thenthe real impact is often unclear. Each jurisdiction has wres-tled with the role that efficiencies should play within itsrespective legal and policy framework. Despite the differences between treatment of efficiencies

in the United States and Canada during merger reviews, effi-ciency claims remain an important strategy for advocates onboth sides of the border in any given cross-border case. Bestpractices include:� U.S. and Canadian counsel should coordinate closely andearly in the preparation of efficiencies arguments in eachjurisdiction as there will be many areas of overlap in thetypes of efficiencies that count, such as variable cost sav-ings.

� The different criteria in each jurisdiction, however, meanthat certain types of efficiencies may be relevant in onejurisdiction but not the other, likely requiring a differentexpert report in each jurisdiction.

� The Bureau, in particular, prefers to see efficiencies reportsearly in the merger review process when parties are relyingon the efficiencies defense, which will require the prepa-ration of efficiencies materials earlier on in the process.While there are differences in how they are considered and

applied in practice, there is a significant and legitimate scope,or at least the possibility of such scope, for efficiencies inmerger reviews in Canada and the United States. In mattersinvolving likely cross-border reviews, merging parties and

their counsel would be well-advised to consider efficiencies’role in the overall regulatory strategy early, particularly incases where the differential treatment in the United States andCanada could be outcome determinative.�

1 More than one-quarter of the Bureau’s merger reviews involve a significantlevel of cooperation from at least one international antitrust counterpart.See John Pecman, Comm’r of Competition, Can. Competition Bureau,Remarks at the International Privacy Enforcement Meeting (June 4, 2015),http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/03957.html.

2 For example, the Superior/Canexus merger was cleared in Canada on thebasis of the efficiencies defense but challenged in the United States.Compare Can. Competition Bureau, Competition Bureau Statement Regard -ing Superior’s Proposed Acquisition of Canexus (June 28, 2016), http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/04111.html, withPress Release, Fed. Trade Comm’n, FTC Challenges Proposed Merger ofCanadian Chemical Companies Superior Plus Corp. and Canexus Corp.(June 27, 2016), https://www.ftc.gov/news-events/press-releases/2016/06/ftc-challenges-proposed-merger-canadian-chemical-companies.

3 Deborah Platt Majoras, Deputy Assistant Att’y Gen., U.S. Dep’t of Justice,Remarks Before the Antitrust Law Section, State Bar of Georgia (Nov. 29,2001), https://www.justice.gov/atr/speech/ge-honeywell-us-decision.

4 OECD: POLICY ROUNDTABLES, THE ROLE OF EFFICIENCY CLAIMS IN ANTITRUSTPROCEEDINGS, DAF/COMP(2012) 23, May 2, 2013, at 16 [hereinafter OECDEFFICIENCY ROUNDTABLE], http://www.oecd.org/competition/EfficiencyClaims2012.pdf.

5 Id. at 16. 6 INT’L COMPETITION NETWORK: RECOMMENDED PRACTICES FOR MERGER ANALYSIS30 (2008), http://www.internationalcompetitionnetwork.org/uploads/library/doc1107.pdf.

7 U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines29 (2010) [hereinafter Horizontal Merger Guidelines], https://www.ftc.gov/sites/default/files/attachments/merger-review/100819hmg.pdf.

8 St. Alphonsus Med. Ctr.–Nampa Inc. v. St. Luke’s Health Sys., Ltd., 778 F.3d775, 790 (9th Cir. 2015; see also ProMedica Health Sys., Inc. v. FTC, 749F.3d 559 (6th Cir. 2014); FTC v. H.J. Heinz Co., 246 F.3d 708 (D.C. Cir.2001); FTC v. Tenet Health Care Corp., 186 F.3d 1045 (8th Cir. 1999); FTCv. Univ. Health, Inc., 938 F.2d 1206 (11th Cir. 1991).

9 Tervita Corp. v. Canada (Comm’r of Competition), 2015 SCC 3, para. 151(Can.).

10 Can.-U.S. Merger Working Group, Best Practices on Cooperation in MergerInvestigations (Mar. 25, 2014), http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/vwapj/Canada-US-Best-Practices-en-2014-03-25.pdf/$file/Canada-US-Best-Practices-en-2014-03-25.pdf.

11 See e.g., Canada (Comm’r of Competition) v. Superior Propane, Inc., 2002Comp. Trib. 16, paras. 80, 215, aff’d, 2003 FCA 53 (CanLII); Tervita, 2015SCC 3, paras 111–113.

12 Competition Act, R.S.C. 1985, c. C-34, § 96 (Can.).13 Tervita, 2015 SCC 3, para. 85.14 MICHAEL J. TREBILCOCK ET AL., THE LAW AND ECONOMICS OF CANADIAN

COMPETITION POLICY 31 (2002). The Canadian approach, as articulated bythe Bureau, also includes somewhat extraneous considerations where amerger results in these “socially adverse” wealth transfers from low-incomeconsumers. See Tervita, 2015 SCC 3, paras. 90–99. The incorporation ofwealth transfers into the efficiencies trade-off analysis has been criticizedby the Canadian Competition Tribunal because of the need to rely on valuejudgments that go beyond the traditional scope of antitrust law. SeeSuperior, 2002 Comp. Trib. 16, para. 372.

15 Can. Competition Bureau, Merger Enforcement Guidelines 12.20 (Oct.2011) [hereinafter Merger Enforcement Guidelines] http://www.competi-tionbureau.gc.ca/eic/site/cb-bc.nsf/eng/03420.html.

S P R I N G 2 0 1 8 · 6 9

16 Tervita, 2015 SCC 3, paras. 151–155. 17 Id. para 122. 18 Can. Competition Bureau, Competition Bureau Statement Regarding

Superior’s Proposed Acquisition of Canexus (June 28, 2016), http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/04111.html.

19 See, e.g., Can. Competition Bureau, Position Statement (Nov. 5, 2015),http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/03353.html;Can. Competition Bureau, Competition Bureau Will Not Challenge Post -media’s Acquisition of Sun Media (Mar. 25, 2015), http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/03898.html.

20 The Bureau had previously challenged several transactions in which themerging parties have successfully invoked the efficiencies defense inSection 96 in court, including Superior Propane Inc., 2002 Comp. Trib. 16,and Tervita, 2015 SCC 3.

21 Can. Competition Bureau, Acquisition of Canexus by Chemtrade Will Not BeChallenged (Mar. 8, 2017), https://www.canada.ca/en/competition-bureau/news/2017/03/acquisition_of_canexusbychemtradewillnotbechallenged.html.

22 Can. Competition Bureau, Competition Bureau Statement Regarding ItsInvestigations into First Air, Canadian North and Calm Air (Aug. 22, 2017),http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/04295.html.

23 The Bureau does not appear to have concluded that either of these trans-actions would have resulted in a “substantial” lessening of competition, andCanadian competition law only allows a remedial order to be issued wherethe lessening of competition would be substantial.

24 Can. Competition Bureau, Competition Bureau Statement RegardingSuperior Plus LP’s Proposed Acquisition of Canwest Propane from GibsonEnergy ULC (Sept. 28, 2017), http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/04307.html.

25 Id.26 Competition Act, R.S.C. 1985, c. C-34, § 96 (Can.). Section 92 of the

Competition Act also suggests that only one “order” is issued from a legalstandpoint even if that order relates to multiple assets or shares.

27 Superior, 2002 Comp. Trib. 16, para. 140. 28 Tervita, 2015 SCC 3, para. 48. 29 Competition Act, R.S.C. 1985, c. C-34, § 96 (Can.). In our view, the plain

reading of Section 96(1) requires the Tribunal to assess whether the gainsin efficiency from the merger exceed the anticompetitive effects from themerger, and then to assess whether an order of the Tribunal would causethe loss of any of those gains in efficiency.

30 By consumer welfare, we mean what is commonly referred to as consumersurplus in economics, as opposed to producer surplus or total surplus.

31 Horizontal Merger Guidelines, supra note 7, at 29. 32 See, e.g., ProMedica Health Sys., Inc. v. FTC, 749 F.3d 559 (6th Cir. 2014);

FTC v. H.J. Heinz Co., 246 F.3d 708 (D.C. Cir. 2001); FTC v. Tenet HealthCare Corp., 186 F.3d 1045 (8th Cir. 1999); FTC v. Univ. Health, Inc., 938F.2d 1206 (11th Cir. 1991).

33 Horizontal Merger Guidelines, supra note 7, at 31. 34 Id. at 30. 35 Tervita, 2015 SCC 3, para. 91 (citing BRIAN A. FACEY & CASSANDRA BROWN,

COMPETITION AND ANTITRUST LAWS IN CANADA: MERGERS, JOINT VENTURES AND

COMPETITOR COLLABORATIONS 256–57 (2013)). 36 Horizontal Merger Guidelines, supra note 7, at 31.37 FTC v. Procter & Gamble Co., 386 U.S. 568, 580 (1967). 38 See, e.g., FTC v. Staples, Inc., 970 F. Supp. 1066 (D.D.C. 1997); FTC v.

Cardinal Health, Inc., 12 F. Supp. 2d 34, 65 (D.D.C. 1998); FTC v. H.J. HeinzCo., 116 F. Supp. 2d 190 (D.D.C. 2000); FTC v. Sysco Corp., 113 F. Supp.3d 1 (D.D.C. 2015).

39 St. Lukes, 778 F.3d at 790.40 Id.41 Procter & Gamble Co., 386 U.S. at 580. 42 United States v. Anthem, Inc., 855 F.3d 345, 369 (D.C. Cir. 2017).

43 Id. at 371, 374–75. 44 Horizontal Merger Guidelines, supra note 7, at 30. 45 Tervita, 2015 SCC 3, para. 107; Canada (Comm’r of Competition) v. CCS

Corp, 2012 Comp. Trib. 14, para. 262; Superior, 2002 Comp. Trib. 16, para.147–149; Canada (Dir. of Investigation and Research, Competition Act) v.Hillsdown Holdings (Canada) Ltd., [1992] C.C.T.D. No. 4, CT - 1991/001, paras. 71–72.

46 Superior, 2002 Comp. Trib. 16, paras. 147–149. 47 U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines

(1992), https://www.ftc.gov/sites/default/files/attachments/merger-review/hmg.pdf. The 1997 revisions to these guidelines set the standardof proof for efficiencies at “reasonable verification,” required only a con-sideration of “practical” alternatives with respect to merger-specificity, anddefined more explicitly what counts as an efficiency.

48 Horizontal Merger Guidelines, supra note 7. The 2010 Horizontal MergerGuidelines added additional specificity as to the likely sources of a suc-cessful claim of merger-specific efficiencies, state that “improved quality,enhanced service, or new products” along with “lower prices” can producecognizable efficiencies, and state that lower fixed costs in the short termcan create long term customer benefits, “e.g., if they make new productintroduction less expensive.” Horizontal Merger Guidelines, supra note 7, at29, 31.

49 See, e.g., U.S. Dep’t of Justice & Fed. Trade Comm’n, Commentary on theHorizontal Merger Guidelines 55 (2006) [hereinafter Commentary], https://www.ftc.gov/sites/default/files/attachments/merger-review/commentaryonthehorizontalmergerguidelinesmarch2006.pdf (noting that the PayPal/eBay merger was not challenged “principally because other means of pay-ment substantially constrained eBay’s ability to increase fees after theacquisition” but noted that efficiencies “were also a factor in the Depart -ment’s analysis.”).

50 Steven C. Salop & Daniel P. Culley, Potential Competitive Effects of VerticalMergers: A How-To Guide for Practitioners (Geo. Univ. Law Center 32–36,2014), http://scholarship.law.georgetown.edu/facpub/1392/.

51 Press Release, Fed. Trade Comm’n, Federal Trade Commission Votes toClose Investigation of Acquisition of Avant! Corporation by Synopsys, Inc.(July 26, 2002), https://www.ftc.gov/news-events/press-releases/2002/07/federal-trade-commission-votes-close-investigation-acquisition.

52 Horizontal Merger Guidelines, supra note 7, at 31. 53 Commentary, supra note 49, at 53. 54 Competitive Impact Statement, United States v. Google Inc., No. 1:11-cv-

00688 (Apr. 8, 2011), https://www.justice.gov/atr/case-document/file/497671/download.

55 Merger Enforcement Guidelines, supra note 15, para 11.2. 56 Id. paras. 12.4, 12.25. 57 Tervita, 2015 SCC 3, paras. 124–125, 166. 58 See, e.g., Merger Enforcement Guidelines, supra note 15, para. 11.2; Can.

Competition Bureau, Competition Bureau Intervention Telecom Notice ofCon sultation CRTC 2016-192 para. 21 (June 29, 2016), http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/04108.html; OECD:POLICY ROUNDTABLES, COMPETITION POLICY FOR VERTICAL RELATIONS IN GAS -OLINE RETAILING, DAF/COMP(2008)35, May 20, 2009, at 18, https://www.oecd.org/competition/abuse/43040511.pdf; OECD: POLICY ROUND -TABLES, VERTICAL MERGERS, DAF/COMP(2007)21, Nov. 12, 2007, at 58n.111, https://www.oecd.org/competition/mergers/39891031.pdf; OECDEFFICIENCY ROUNDTABLE, supra note 4, at 13.

59 ANDREW TEPPERMAN & MARGARET SANDERSON, INNOVATION AND DYNAMIC EFFI -CIENCIES IN MERGER REVIEW, (2007), http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/02376.html.

60 Gary L. Roberts & Steven C. Salop, Efficiencies in Dynamic Merger Analysis:A Summary, 19 WORLD COMPETITION, 5, 7, 9 (1996).

61 Merger Enforcement Guidelines, supra note 15, para. 12.17. 62 Can. Competition Bureau, 2015–2018 Strategic Vision (June 2, 2015),

http://www.competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/03934.html.


7 0 · A N T I T R U S T

ANTITRUST MERGER ANALYS I Sinvolves predicting how a merger will changethe future course of competition in a market,focusing in particular on the merger’s impacton consumers. The substantial lessening of

competition from a merger is thought to lead to potentiallyharmful effects on multiple dimensions: higher prices, lowerquality, and/or lower rates of innovation. In spite of widespread acknowledgment of the potential

for multi-faceted impact, the evaluation of competitive effectsof horizontal mergers has tended to have a greater focus onthe impact of the merger on price competition, as comparedto its dynamic impact on innovation. Historically, the pres-ence of innovation has been considered in connection withthe assessment of price effects as an offsetting procompetitiveefficiency resulting from a merger. However, when innovationis a critical feature of competition between the merging par-ties, a merger may have implications for competition outsideof any observable price effect by impacting the incentive ofthe merging parties to engage in innovation. Moreover, inthese instances, current market shares and competition fromexisting rivals within a relevant market may not fully capturethe impact of the merger on future market performance.1

The historic focus on price effects was driven, in part, bylack of consensus in the academic literature on the effects ofcompetition on innovation,2 leading to greater uncertaintyamong competition authorities on how innovation effectsought to be incorporated into merger review outside of effi-ciencies.3 Without any clear direction on the relationshipbetween competition and innovation, merger enforcementdecisions were based on limited analysis examining howmergers might impact innovation.

In recent years, the analysis of innovation effects on futurecompetition, outside of its implication for price effects, hasexperienced a revival in merger review. In the United States,reflecting current agency practice, the 2010 Horizontal Merg -er Guidelines formalized the role of innovation in mergerreview by moving it from solely being a merger efficiencyconsideration to advocating for its inclusion when assessingcompetitive effects.4 Similarly, innovation effects are refer-enced throughout the European Union horizontal (and non-horizontal) merger guidelines5 in the context of competitiveeffects and efficiencies.6 The result has been a more robusttrend in recent merger enforcement as it relates to innovationparticularly in the EU, but also in the U.S., whereby inno-vation is not only considered more prominently in the regu-latory agencies’ assessment of competitive effects, but is alsoput forward as a reason to challenge mergers in courts. In this article, we discuss the specific economic mecha-

nisms underlying how mergers might affect the merging par-ties’ incentives to innovate ex post.7 We then discuss thesemechanisms in the context of several illustrative mergerreviews and challenges brought by U.S. and EU antitrustauthorities where concerns about innovation played a mean-ingful role in the agency’s decision to challenge the merger.And, finally, to provide specific guidance to antitrust practi-tioners on navigating agency investigations where innovationconcerns may feature prominently, we discuss the evidenceand analyses used by the antitrust agencies to assess innova-tion effects in recent merger reviews.

Incorporating Innovation Effects into MergerAnalysis: Conceptual BackgroundIn analyzing any competitive effect of a merger on innova-tion, the fundamental inquiry is whether the merger signif-icantly changes the incentives of the merging parties to investin innovation. From a firm’s perspective, the incentive toinnovate is driven by the additional profit earned if it were toinvest in innovation, compared to a scenario where it doesnot. The effects are not restricted to situations where thetwo merging parties currently compete against each other ina product market. Rather, they also can be seen in cases ofpotential competition where the merging parties are notproduct market rivals at present or where the product withrespect to which there would be expected competition

The Increasing Cross-Border Importance ofInnovation in Merger Review

B Y J E N N I F E R C A S C O N E F A U V E R , S U B B U R A M A N A R A Y A N A N , A N D N I C O L A T O S I N I

Jennifer Cascone Fauver is an Associate Director in the Washington, DC office

of NERA Economic Consulting. Dr. Subramaniam (Subbu) Ramanarayanan is

an Associate Director in the New York office of NERA Economic Consulting, and

Adjunct Professor of Competitive Strategy at UCLA School of Management.

Dr. Nicola Tosini is an Associate Director in the Berlin and Brussels offices of

NERA Economic Consulting. The authors have consulted on mergers involving

the analysis of competitive effects related to innovation in the U.S. and EU.

The authors thank Yashaswini Singh for research assistance. The opinions

expressed are those of the authors and do not necessarily reflect the views

of the firm and its clients.

S P R I N G 2 0 1 8 · 7 1

a merger between these firms impact Firm A’s incentives tocontinue investing in innovation? Let us consider a scenario where Firm A and Firm B have

different innovation capabilities and/or whereby Firm B hasnot historically launched any innovative or variant productlines. Because the merger would lead to less intense compe-tition between a new product variant launched by Firm A andthe existing substitute products sold by Firm B, the mergedentity could realize greater revenues (and thus, gross profits)on the new product variant by charging a higher price for it,compared to the price at which it would have been sold but-for the merger. This results in an increased incentive for FirmA to innovate in the product line. This effect needs to be bal-anced against the softening effect that the merger has oncompetition between Firm A’s and Firm B’s existing products,which results in a reduced incentive to innovate. This com-plex interaction in competition between new and existingproducts could lead to an overall positive or negative effect oninnovation incentives.This conceptual framework has been employed by both

U.S. and EU regulators in recent merger challenges. For example, in its 2014 challenge of the merger between

Verisk Analytics, Inc. and EagleView Technology Corp., theFederal Trade Commission argued that the merging partiesnot only competed in the relevant product market (rooftopaerial measurement products), but also competed to offerinnovative new products within this market. The FTC point-ed to Verisk, which was a recent entrant into the relevantproduct market and was, at the time, developing a higher res-olution product to compete with EagleView’s existing prod-uct line. The FTC was concerned that post-merger, Veriskwould have had “less incentive to develop new and betterproducts.”13

Similarly, in its 2015 review of the acquisition of theThermal Power, Renewable Power Grid businesses of Alstomby General Electric, the European Commission argued thatthe merger would have reduced direct competition betweenGE and Alstom for certain types of frames used in heavy dutygas turbines (HDGT) and that the parties would also have areduced incentive to innovate to develop upgrades for thoseHDGT frames. Furthermore, the EC also expressed con-cern that any reduction in the incentive to innovate by themerging parties would have a spillover effect, reducing theincentives of other rivals to innovate.14

Likewise, in its 2015 review of the acquisition of Glaxo -SmithKline plc’s portfolio of oncology products by NovartisAG, the EC expressed concern that the acquisition wouldresult in reduced incentives for Novartis to continue to con-duct clinical research and develop products for the treatmentof cancer.15 At the time of the merger, both parties weredeveloping a pair of B-Raf and MEK (protein) inhib itors forthe treatment of a number of advanced cancers. Roche wasthe only other rival that was also developing these proteininhibitors. The EC expressed concern that because GSK hada lead in the development of the pair of inhibitors for the

between the parties does not yet exist. Potential competitionin a product market may still translate into actual competi-tion in so-called innovation markets.A merger can influence the additional profits related to

innovation through several general mechanisms: (1) an “indi-rect effect” driven by the merger-induced change in productmarket competition; (2) a “direct effect” driven by the merg-er-induced change in innovation competition; and (3) effi-ciencies. Ultimately, the net impact of a merger on the incen-tives and the ability of the merged entity to innovate willdepend on a comparison of the competitive effects of themerger on innovation with any innovation-related efficienciesderived from the merger.

Indirect Effect of a Merger on Innovation Incentives.The traditional assessment of horizontal mergers considersthe loss of competition between rivals in a product marketand the corresponding impact on their incentive to raiseprices. The indirect effect of a merger on innovation consid-ers the effect of the loss of competition between the mergingparties in a product market on the incentive of a mergingparty to innovate in that product market. This indirect effectarises when reduced price competition between the partiespost-merger results in higher prices in a relevant productmarket. In particular, if the merging firms are product-mar-ket rivals, then one of the firms may find it optimal to launchits innovative product8 post-merger at a higher price thanwould have been possible absent the merger.9 This indirecteffect is complex because a merger changes pricing incentivesnot only after, but also before the innovation. In other words, the merger changes not only the pricing

incentives for the new product, but also for the existing prod-ucts of the innovating merging party. In this case, one needsto take into account not only the difference in profits on thenew product with and without the merger but also the dif-ference in the cannibalization of the existing products of theinnovating merging party with and without the merger. Inaddition, the merger induces the innovating merging party tointernalize the effect that the introduction of its new prod-uct has on the profits earned on the sale of the merging part-ner’s existing products.10 Both effects (the reduction in pricecompetition and the cannibalization of sales) are two sides ofthe same coin: without the threat of cannibalization of eachother’s sales, there would be no incentive for the merging par-ties to raise the price at which their new product is offered.11

Whether the indirect effect of a merger on innovation ispositive or negative (i.e., if it enhances or restricts innovation)ultimately depends on how the merger impacts the differencein the merging parties’ profits before and after the innova-tion.12

This is best illustrated through a stylized example.Consider two firms, A and B. Both firms sell existing

products that consumers consider to be substitutes. Firm Ais investing in innovation to produce a variant of its existingproduct that would still compete with Firm B’s product butwould also appeal to a different set of consumers. How would


7 2 · A N T I T R U S T

Like the indirect effect, the direct effect conceptual frame-work has been employed by both U.S. and EU regulators inrecent merger challenges. For example, in 2013, the U.S. Department of Justice

investigated the announced merger between Applied Mater -ials and Tokyo Electron, “two of the world’s largest providersof the tools that are used to manufacture semiconductorchips.”20 The DOJ described the competition betweenApplied Materials and Tokyo Electron as “dynamic wherefirms win or lose successive waves of innovation develop-ments.”21 The DOJ pointed to the merging parties’ capabil-ities, historical experience, engineering staff, and establishedR&D facilities as giving them an advantage over smallercompetitors “to develop the new tool as well as the financialresources (R&D budgets) to take on more risky projects and persevere through setbacks.”22 The DOJ argued on thisbasis that the parties were the likeliest, if not the only, toolmanufacturers capable of competing (innovating) to solvefuture problems faced by semiconductor manufacturers. As a result, the DOJ concluded that such competition toinnovate and develop new tools would have been lost post-merger.In Dow/DuPont,23 the EC extended its review to products

that had not been discovered yet and to innovation compe-tition in the medium to long term. The EC’s investigation inDow/DuPont pertained to, among other products, formulat-ed crop-protection products, i.e., pesticides. Pesticides can bedistinguished by the crop that they are meant to protect, thepest from which they are meant to protect it against, and thetiming of their application. In light of the increasing resist-ance of pests to existing products and the increasingly strin-gent regulatory standard on toxicity, pesticides are subject tocontinuous innovation. According to the EC, the market forpesticides was characterized by the presence of five truly glob-al players, each of which was fully integrated from productdiscovery to marketing.The EC’s review of Dow/DuPont focused on both direct

and indirect effects of the proposed merger on innovationincentives.24 In particular, employing an approach reminis-cent of the “innovation markets” framework in the econom-ic literature,25 the EC considered both innovation competi-tion in “innovation spaces,” defined by lines of research/earlypipeline products, and innovation competition “at the indus-try level.” In other words, the EC was concerned that themerger would have reduced not only competition in specif-ic research areas where both parties were active but also com-petition in the crop protection industry overall, irrespectiveof the presence or absence of current and future overlaps inindividual product markets and overlaps in lines of research/early pipeline products. Once the EC had assessed marketstructure in the pesticides industry, and established theimportance of patents in measuring innovation, it focused itsinvestigation on the closeness of the merging parties as com-petitors in innovation, on the basis of its analysis of patent-ing patterns.26

treatment of skin cancer, the merged entity would have aban-doned Novartis’ whole clinical research program, includingfor a cancer (uveal melanoma), where the merging parties’programs did not overlap.16

Direct Effect of a Merger on Innovation Incentives.More fundamentally, a merger may have a direct effect on theincentives of the merging parties to innovate. The directeffect on innovation incentives differs from the indirect effectin that the indirect effect focuses on how the merger impactsincentives to innovate by modifying product market compe-tition between the merging parties, while the direct effectfocuses on how the merger affects competition between themerging parties to innovate.17 Although the source of thechange in the incentives to innovate is different, the harm tocompetition for direct and indirect effects is the same—lossof new products or delay in their introduction.The direct effect on innovation incentives can be best

illustrated by a patent race to establish a new market, inwhich only the pioneering player will thrive (a so-called win-ner-take-all market). In the race for a winner-take-all market,the indirect effect on the incentive to innovate is not appli-cable because the products of the merging parties will neverbe sold next to each other in the market. At the same time,the direct effect is present because the innovation success ofone merging firm means the innovation failure of the othermerging firm. Whereas before the merger this “crowding outeffect” is not taken into account by the merging parties, it willbe taken into account by the merged entity, leading to anoverall reduction in the incentive to innovate.This reduction in the incentive to innovate could lead to

consumer harm at different time horizons, either manifestingitself in the form of lower incentives to continue developingproducts that are already in the pipeline (short to mediumterm) or in the form of lower investments in R&D for newor future products (medium to long term).18 The harm toconsumers is derived from a possible delay in the introduc-tion of new products and/or a possible reduction in the over-all rate of innovation in an industry.19 This effect of a merg-er on innovation is strongest when the merging parties areclose innovation rivals and represent substantial sources ofinnovation in a market that is concentrated and expected toremain so because of high barriers to entry.The direct effect of a merger on innovation can also be

procompetitive, with a merger leading to a stronger incentivefor the merging parties to invest in innovation. This couldhappen in industries where innovation manifests itself insuccessive generations of technology, with each generationhaving the potential to displace the previous one as the stan-dard. In this case, the innovations of the merging partiesmay not be in head-to-head competition with each otherbut may complement each other. If this is the case, a merg-er between innovators may make one merging party appre-ciate how its innovation effort would enhance the futureprofits of its merger counterpart, thus increasing the incen-tive to innovate.

S P R I N G 2 0 1 8 · 7 3

consumers.30 The traditional imbalance in the standard ofproof between procompetitive and anticompetitive mergereffects is exacerbated in mergers involving innovation effectsgiven that innovation is an intrinsically uncertain activity.In particular, given that the development of new products

involves a large degree of uncertainty, and that in many casesproducts in various stages of development might never actu-ally make it to market, the task of determining overlap infuture products and assessing overall competitive effects netof efficiencies on the basis of that overlap is a daunting one.In addition, balancing anticompetitive and procompetitiveeffects of the merger on innovation would require balancingany possible efficiencies generated by the merger with thepotential anticompetitive effect discounted by the probabil-ity of that effect being realized.

When Are Innovation Effects Likely to Be of Concern? The preceding discussion outlines the various mechanisms bywhich a merger might have an impact on competition whenthe level of innovative activity is itself considered to be animportant measure of market performance. A proposedmerger under review by the antitrust agencies will likely raisegreater concerns regarding a possible adverse impact on inno-vation if either the direct or the indirect effects, or both, actto reduce the incentive to innovate post-merger, in conjunc-tion with a lack of (or a weak case for) countervailing effi-ciencies. As a practical matter, this implies that a merger will lead

to significant concerns regarding innovation if the followinghold true.� Importance of innovation in the industry: Innovation is a keymeasure of market performance in the industry, under-scoring the need for going beyond a static analysis of com-petition focused on price effects alone.

� Innovation competition between the parties: The mergingfirms are important sources of innovation and competeclosely in that space to be the first one to establish a newmarket, implying that stronger negative direct effects arelikely. This competition can be at the level of productsbeing developed, products being discovered, or in terms ofbroader innovation capabilities. However, the more distantthe merging parties’ products are from being launched, themore difficult it is—for outside observers as well as for themerging parties themselves—to determine how closelythey will compete in (or for) the product market and thusto what extent the merger will affect the parties’ incentivesto innovate.

� Product market competition between the parties: A mergingparty is developing a product that would enable it to com-pete in a product market in which its merger partner isalready present.

� Alternate sources of innovation: The merging firms haveonly a few rivals (actual or potential) that provide com-petition on innovation.

The EC was concerned that the merged entity would havereduced competition and therefore innovation effort at boththe innovation-space and the industry level, and judged thatthis concern was corroborated by ordinary course of businessdocuments, according to which the merged entity would havecut down on innovation inputs and output targets. Accordingto the EC, innovation efforts undertaken by rivals wouldnot have offset such reductions, either because they were noteffective enough in the overlapping innovation spaces (thethree other global integrated firms) or because they wereregionally focused (firms in Japan).27

Assessing Efficiencies When Innovation Effects ArePresent. In reaching a final assessment on innovation effects,regulators compare the potential competitive harm stem-ming from a direct or an indirect reduction in the incentivesof the merging parties to innovate with potential innova-tion-enhancing efficiencies generated by the merger. In thisway, the analysis very much follows the traditional approachtaken in merger analysis in weighing any procompetitive effi-ciencies against any adverse competitive effects resulting fromthe merger.One way in which a merger might have a beneficial impact

on innovation incentives is by enhancing the extent to whicha merging party can capture the benefits generated by itsinvestments in innovation, as opposed to having these ben-efits captured by its merging partner or other rivals. Thisnotion, termed appropriability, is well-recognized in the eco-nomic literature and is particularly relevant in industries withineffective intellectual property protection.28 Appropriabilityis impacted by the number of rival innovators––that is, thesmaller the number of rival innovators, the higher the prob-ability that a firm will be able to capture a large share of theadditional value from its innovation without concern thatanother firm would imitate its innovation and launch a com-peting product. Such enhanced appropriability resulting froma merger can thus be a procompetitive efficiency related toinnovation.A merger could also have a beneficial impact on innova-

tion through other efficiency-related mechanisms, such asrealization of economies of scale and scope, which motivatesthe merging parties to undertake investments in R&D thatthey would not have undertaken otherwise. If the mergerinvolves firms selling and developing products that are com-plements, better coordinated pricing in the aftermath of themerger could lead to increased profitability, and thereforeincreased incentives to innovate. Finally, if the merging par-ties have different best practices and capabilities that can becombined to increase the likelihood of successful innova-tion, that could be an important merger-related efficiency aswell.29

While a merger could lead to significant innovation-relat-ed efficiencies, such efficiencies are typically subject to ahigh standard of proof by the antitrust agencies, in that, inthe language of the European and U.S. agencies, they mustbe merger-specific, verifiable, cognizable, and passed on to

� Efficiencies: The effect of the merger on appropriability issomewhat limited due to, say, effective intellectual prop-erty rights.

Types of Evidence Used to Assess Innovation EffectsAssessing the effect of a merger on innovation is substantial-ly more demanding than determining the effect of a mergeron the price of existing products. In particular, the assess-ment of innovation effects not only requires examination ofmarket structure and market characteristics (to determinestrength of entry barriers, for example) but also requires deter-mination of the extent to which the merging parties competein innovation efforts, and the degree to which the mergermight have a direct or indirect impact on these efforts.Evaluating the indirect effect of a merger on the incentives

to innovate entails in the first place a traditional analysis ofprofit margins and diversion ratios between the merging par-ties’ products, albeit in a more sparsely populated empiricalenvironment, given there are no data on sales of new prod-ucts. The analysis of the indirect effect, however, does notstop here, since it also involves analyzing the parameters gov-erning the sunk costs of product development to determinethe profitability of the development and sale of the newproducts.Evaluating the direct effect on the incentives to innovate

and innovation efficiencies is even more complex because itentails determining how firms compete in innovation, anundertaking that requires understanding the nature and evo-lution of R&D investments and efforts not only by the merg-ing parties, but industry wide.Antitrust agencies have relied on a variety of sources of evi-

dence in making these determinations. These include:� Ordinary course documents: Documents generated by theparties during the ordinary course of business (includingboard presentations, strategic plans, and market studies)can be used to assess the nature of innovation in the indus-try and the degree to which innovation is targeted towardsspecific product lines and, more importantly, to infer theextent to which the merging parties are close competitorsin innovation and track each other’s activities closely inthat space. The EC relied heavily on ordinary course doc-uments in its review of Dow/Dupont to conclude that theparties competed “head-to-head for a significant numberof innovation spaces” and that the merger would thereforelikely lead to the “discontinuation, delay or reorientationof the parties’ existing overlapping lines of research andpipeline products.”31 This was also the case in the FTC’s2014 challenge in Verisk/EagleView, where ordinary coursebusiness documents discussing the ongoing innovationand competition between the parties underscored theFTC’s concerns regarding the impact of the merger on theparties’ incentives to continue to innovate.32

� Transaction-related documents: Internal documents gener-ated as part of the evaluation of the transaction might

also be used to assess the impact on innovation. For exam-ple, post-integration planning documents discussing pro-posals to integrate R&D activity or revise innovation out-put targets post-merger could be used by competitionauthorities as evidence of direct harm to innovation com-petition, as was the case in the EC’s review of the mergerbetween GE and Alstom.33

� Historical competition between the merging parties: As isthe case with typical merger analysis, historical competi-tion between the merging parties can provide insight intowhether the merger would impact the parties’ incentive toinnovate post-merger. In the DOJ’s 2013 review of AppliedMaterials/Tokyo Electron, the DOJ identified the parties’historical competition to innovate and create productlines—specifically to solve a tooling problem identified bya semiconductor manufacturer—as a key consideration inhow the merger might impact this innovation going for-ward.34 This was compounded by the fact that the merg-ing parties were identified as the only two entities capableof addressing the tooling problems identified by the semi-conductor manufacturers.

� Analysis of patents: Given the inherent difficulty in meas-uring output from innovation-related activity, an analysisof patents can be particularly helpful in assessing innova-tion effects. Such an analysis could be helpful not just indetermining whether the merging parties are key innova-tors, but also in assessing the extent to which the industryis concentrated from an innovation standpoint. For exam-ple, in its review of Dow/Dupont, the EC analyzed themerging parties’ patent portfolios to argue that they were“more important innovation competitors at industry level”than suggested by their market shares and R&D expendi-tures.35 This conclusion was based on the finding that themerging parties had a particularly strong presence in high-quality patents, with quality of a patent being measured bythe number of citations it receives in future patents.36

� Opinion of industry experts:Competition authorities mightconsult industry experts in forming an opinion as towhether a proposed merger might have an adverse impacton innovation, particularly in situations where there islimited information available in the form of precedent orrelevant past patent activity. For example, in its review ofNovartis/GSK Oncology, the EC consulted “Key OpinionLeaders” as part of its investigation to conclude that themerged entity would abandon Novartis’ clinical researchprogram.37

� Opinion of customers: Competition authorities might con-sult customers of the merging parties to assist them informing an opinion about how or whether the proposedmerger would have an adverse impact on innovation. Forexample, the opinion of customers as to whether otherrivals—current or future—would be in a position to com-pete with yet-to-be-developed products was a key com-ponent of the FTC’s analysis in its review of the 2013merger between Arbitron Inc. and Nielsen Holdings N.V.

7 4 · A N T I T R U S T


S P R I N G 2 0 1 8 · 7 5

In its review, the FTC reasoned that customers identifiedthe merging parties as the “best positioned to compete in”national syndicated cross-platform audience measurementservices.38

ConclusionRecent enforcement actions in the U.S., and particularly inthe EU, have highlighted the willingness of enforcementagencies to review and challenge mergers on the basis of theprojected adverse impact of the merger on incentives of themerging parties to innovate. Both direct and indirect effectsof the merger on innovation incentives appear to have playeda role in the agencies’ decision-making process. As such, it isnecessary for merging parties to understand the specificmechanisms behind how a merger might impact innovation,and to develop support for these mechanisms (or the lackthereof ) using the facts of the case at hand.Given this trend, antitrust practitioners will be well-served

to focus early in the merger review process on potential inno-vation concerns. Critically, it is important for practitioners toanticipate the types of evidence the agencies may rely on intheir assessment of innovation effects, particularly pertainingto head-to-head competition in innovation between themerging parties. Relatedly, on the efficiencies front, practi-tioners need to pay early heed to evidence supporting themerging parties’ claims regarding the proposed transaction’srole in enhancing innovation incentives. At the current time, there appears to be somewhat of a

divergence between the U.S. and the EU in their approach toassessing innovation effects and, in particular, in the extent to which each regulatory regime is willing to define strictinnovation markets as in the EC’s recent approach in Dow/DuPont. This implies that a one-size-fits-all approach mightnot be as effective when dealing with mergers involvingmulti-jurisdictional reviews. Looking forward, the trend in enforcement decisions cou-

pled with the recent stream of economic literature studyinginnovation effects points to a regulatory environment whereconcerns regarding the impact of transactions on innovationwill continue to play a major role in merger review.�

1 Michael L. Katz & Howard A. Shelanski, Mergers and Innovation, 74 ANTI -TRUST L.J. 1 (2007). Katz & Shelanski refer to these two distinct perspec-tives as “innovation impact” (where innovation impacts price competition)and “innovation incentives” (where innovation is a key determinant of com-petition). Id. at 12.

2 For a more comprehensive analysis of the economic literature on competi-tion and innovation, see, e.g., Richard Gilbert, Looking for Mr. Schumpeter:Where Are We in the Competition-Innovation Debate?, in 6 INNOVATION POLICYAND TTHE ECONOMY 159 (2006); Katz & Shelanski, supra note 1; CarlShapiro, Competition and Innovation. Did Arrow Hit the Bull’s Eye?, in THERATE AND DIRECTION OF INVENTIVE ACTIVITY REVISITED 361 (Josh Lerner & ScottStern eds., 2012.

3 For example, in its 2004 review of the 2001 consummated merger betweenGenzyme Corporation and Novazyme Pharmaceuticals Inc., the FTC ex -pressed caution with respect to analyzing the effects of a merger on inno-

vation using innovation market analysis because “economic theory andempirical investigations have not established a general causal relationshipbetween innovation and competition.” See Statement of Chairman TimothyJ. Muris, Genzyme Corp./Novazyme Pharma. Inc., FTC File No. 021 0026(2004) [hereinafter Muris Genzyme/Novazyme Statement (2004)], https://www.ftc.gov/system/files/attachments/press-releases/ftc-closes-its-investigation-genzyme-corporations-2001-acquisition-novazyme-pharmaceuticals-inc./murisgenzymestmt.pdf.

4 U.S. Dep’t of Justice & Fed. Trade Comm’n, Horizontal Merger Guidelines(2010). Id. §§ 1, 6 (unilateral effects); id. § 10 (efficiencies), https://www.ftc.gov/sites/default/files/attachments/merger-review/100819hmg.pdf[hereinafter 2010 Horizontal Merger Guidelines]; U.S. Dep’t of Justice &Fed. Trade Comm’n, Horizontal Merger Guidelines (1992); id. §4 (efficien-cies), https://www.justice.gov/atr/horizontal-merger-guidelines-0.

5 Eur. Comm’n, Guidelines on the Assessment of Horizontal Mergers underthe Council Regulation on the Control of Concentrations Between Under -takings, 2004 O.J. (C 31) [hereinafter EC Merger Guidelines].

6 In fact, the EU’s interest in innovation has stretched beyond mergers andincludes restrictive agreements between firms or cases where there are alle-gations of abuse of dominance. See, e.g., Press Release, Eur. Comm’n,Antitrust: Commission Opens Formal Investigation into Mobile TelephoneNetwork Sharing in Czech Republic (Oct. 25, 2016) (IP/16/3539); PressRelease, Eur. Comm’n, Antitrust: Commission Confirms Inspections in theCar Sector in Germany (Oct. 23, 2017) (STATEMENT/17/4103) (announcinginspections at the premises of car manufacturers in Germany as part of anEU antitrust probe into possible cartel behavior).

7 There is, in addition, an ex ante impact of merger policy on firms’ expectedprofits and innovation decisions, which is outside the scope of this article.

8 For ease of exposition, in this article we focus on the impact of a mergeron product innovation. The analysis of the impact on process innovation issimilar, although not identical.

9 This effect is referred to as the “price coordination” effect. Giulio Federico,Gregory Langus & Tommaso Valetti, A Simple Model of Mergers and Inno -vation, 157 ECON. LETTERS 136 (2017).

10 This effect is referred to as the “innovation externality” effect. Id.11 Cannibalization of sales can be so large as to make the launch of the new

product altogether unprofitable, irrespective of the price at which it wouldbe offered.

12 In the case of potential competition, the existing sales of the innovatingmerging party play no role. This fact simplifies the analysis of indirecteffects, because the change in price incentives only affects the new product.

13 Complaint ¶¶ 3, 6, 23, 40, Verisk Analytics Inc., FTC File No. 141-0085(Dec. 16, 2014), https://www.ftc.gov/system/files/documents/cases/141216veriskcmpt.pdf. The parties ultimately abandoned the transaction.See also Terrell McSweeny, Comm’r, Fed. Trade Comm’n, Opening Remarksat the CRA Conference—Brussels, Belgium: Why Regulate Online Platforms:Transparency, Fairness, Competition or Innovation? 2 (Dec. 9, 2015),https://www.ftc.gov/system/files/documents/public_statements/903953/mcsweeny_-_cra_conference_remarks_9-12-15.pdf.

14 Case COMP/M. 7278—GE/Alstom, Comm’n Decision (Sept. 8, 2015) [here-inafter GE/Alstom Commissioni Decision], http://ec.europa.eu/competi-tion/mergers/cases/decisions/m7278_6808_3.pdf. See also EuropeanComm’n, 1 Competition Merger Brief 17 (2016). The EC approved the merg-er conditional upon Alstom divesting most of its existing products and inno-vation assets for selected frames to a smaller rival with more limited exist-ing innovation capabilities.

15 Case COMP/M. 7275—Novartis/GSK Oncology, Comm’n Decision (Jan.28, 2015), http://ec.europa.eu/competition/mergers/cases/decisions/m7275_20150128_20212_4158734_EN.pdf. See also European Comm’n,2 Competition Merger Brief 1–4 (2015). The EC approved this transactionconditional on Novartis divesting rights to MEK inhibitors to an approvedthird party.

16 This would suggest the existence of substantial economies of scope in thedevelopment of the pair of inhibitors for the treatment of different cancers,which is not discussed in the decision.

17 The direct effect of a merger on innovation incentives is grounded in the con-cept of cannibalization of future product sales and represents the innova-


7 6 · A N T I T R U S T

tion analog of the standard unilateral effects theory that is used in exam-ining the price effects of mergers. Extending this analogy, Farrell and Shapiropropose the “innovation diversion ratio” to measure this internalizationeffect. See Joseph Farrell & Carl Shapiro, Antitrust Evaluation of HorizontalMergers: An Economic Alternative to Market Definition, 10 B.E. J. THEO -RETICAL ECON. Vol. 10, Issue 1, Art. 9 (2010), https//faculty.haas.berkeley.edu/shapiro/alternative.pdf.

18 See, e.g., 2010 Horizontal Merger Guidelines, supra note 4, § 6.4 (“That cur-tailment of innovation could take the form of reduced incentive to continuewith an existing product-development effort or reduced initiative to initiatedevelopment of new products.”).

19 Even in cases where the merged entity proceeds with the introduction ofproducts in the pipeline being developed by both merging parties, theremight be harm to consumers stemming from reduced product market com-petition, as outlined in the discussion of “indirect effects” of a merger oninnovation.

20 Nicholas Hill, Nancy L. Rose & Tor Winston, Economics at the Antitrust Divi -sion 2014–2015: Comcast/Time Warner Cable and Applied Materials/TokyoElectron, 47 REV. INDUS. ORG. 431 (2015).

21 Id. at 432. 22 Id. at 433. 23 Case COMP/M. 7932—Dow/Dupont, Comm’n Decision (Mar. 27, 2017)

[hereinafter Dow/DuPont Commission Decision], http://ec.europa.eu/competition/mergers/cases/decisions/m7932_13668_3.pdf. See alsoEur. Comm’n, 2 Competition Merger Brief 1–8 (2017).

24 Dow/DuPont Comm’n Decision, supra note 23, Annex 4. 25 Richard J. Gilbert & Steven C. Sunshine, Incorporating Dynamic Efficiency

Concerns in Merger Analysis: The Use of Innovation Markets, 63 ANTITRUSTL.J. 569 (1995); see also Katz & Shelanski, supra note 1, at 41–43.

26 Dow/DuPont Commission Decision, supra note 23, Annex 1. 27 Id. at 358–76.

28 The relationship between market concentration and appropriability is dis-cussed at length in the economic literature. See, e.g., Shapiro, supra note2; Gilbert, supra note 2. (2006).

29 The FTC took such considerations into account in deciding to close theirinvestigation of Genzyme Corporation’s acquisition of Novazyme Pharma -ceuticals Inc. in 2001. See Muris Genzyme/Novazyme Statement, supranote 3.

30 2010 Horizontal Merger Guidelines, supra note 4, § 10; EC Merger Guide -lines, supra note 5.

31 Eur. Comm’n, 2 Competition Merger Brief, supra note 23, at 1–8. 32 McSweeny, supra note 13, at 2 (“There was strong qualitative evidence that

Verisk was uniquely positioned to compete against EagleView in providingroof reports.”).

33 GE/Alstom Commission Decision, supra note 14. 34 Hill et al., supra note 20, at 431–35 (discussing the evidence of historical

competition by the merging parties in innovating to create product lines).35 Dow/DuPont Commission Decision, supra note 23, at 389. 36 Id.37 Novartis/GSK Oncology, Commission Decision, supra note 15, ¶ 107. 38 Statement of the Federal Trade Comm’n, Nielsen Holdings N.V. and Arbitron

Inc., FTC File No. 131-0058, at 2 (Sept. 20, 2013) (“This forms the basisfor our concern that there would be anticompetitive consequences from thecombination, despite the fact that others are trying to develop cross-plat-form measurement services of their own. Customer views that Nielsen andArbitron would be by far the two strongest competitors are supported byNielsen and Arbitron statements about the products they are each devel-oping and, in some cases, already beta testing with customers.”), https://www.ftc.gov/enforcement/cases-proceedings/131-0058/nielsen-holdings-nv-arbitron-inc-matter. The FTC approved the merger conditioned on thedivestiture of Nielsen’s assets pertaining to cross-platform audience meas-urement to an approved buyer.

� THE REVISED AND UPDATED State Antitrust Enforcement Handbook, Third Edition

complements the Section's multi-volume treatise, State Antitrust Practice and Statutes, which

analyzes substantive antitrust law on a state-by-state basis. This handbook focuses on the antitrust

enforcement practices of state attorneys general and summarizes the substantive law behind state

antitrust enforcement, while also analyzing the roles and priorities of state enforcers.

The third edition, like the previous editions, offers guidance to counsel who interact with state

antitrust enforcers. This latest edition includes an updated focus on merger analysis in the health

care markets, recent decisions on the scope of states' parens patriae authority, and state law

developments regarding vertical price restraints.



State Antitrust Enforcement HandbookThird Edition


Pricing:$249.00 List Price / $229.00 ABA Member /$209.00 AT Section Member

S P R I N G 2 0 1 8 · 7 7

to a positive feedback loop that rendered competing operat-ing systems’ products unattractive.Similarly, it was argued that Microsoft succeeded in mak-

ing MS Office (spanning Word, Excel, and PowerPoint) thedominant suite of office productivity applications, encour-aging users to standardize on MS Office for both business andpersonal use. The direct network effects for these applicationswere based on easy file sharing across users—users had littleincentive to switch to another word processing software appli-cation if none of their colleagues were using it. Microsoft alsobenefited from indirect network effects, because developershad large incentives to develop software for operating systemsthat had the most users. Microsoft used these dual networkeffects (in operating systems and productivity applications) toits advantage, becoming one of the most profitable corpora-tions in the world.However, even the seemingly unassailable advantage

Micro soft appeared to have in the 1990s has begun to erodein recent years as the pace of technological innovation hasquickened and new paradigms like cloud computing havestarted to emerge at increasing rates. The world of digitaltechnology has moved on from the era of Microsoft’s domi-nance, and Microsoft no longer commands an impressivemarket share of computing devices by any definition.3 It is alsono longer a top-5 company in terms of profitability. (Appleand Google, among other digital platforms, have leapfroggedMicrosoft in recent years.) This article discusses how, as the useof digital technologies and their reliance on digital data (ratherthan hardware standards) have expanded and become morefluid, the nature of the network effects that underlie many dig-ital platforms has also changed.Specifically, this article describes three recent advances in

the analysis of network effects and, in particular, how ourunderstanding of network effects has evolved in the digitaleconomy as the usefulness of technology has become lessdependent on the particular hardware it uses and moredependent on digital interactions. First, network effects donot imply entrenchment and can actually lead to quickerdestabilization of a market leader position. Second, networkeffects tend to be quite localized, meaning that they tend to

SINCE THE EARLY YEARS OF PLATFORMand antitrust analysis, network effects have been animportant consideration when analyzing potentialmarket power. This is because the competitiveadvantage bestowed by network effects was thought

to increase with the size of the firm. This article describesthree recent advances in the analysis of network effects and,in particular, how our understanding of network effects hasevolved in the digital economy. These new findings suggestthat network effects are not the guarantor of market domi-nance that antitrust analysts had initially feared.Back in the 1990s, network effects were looked to as a new

source of potential market power, especially in digital mar-kets. Economists use “network effects” to describe contextsin which a good or service offers increasing benefits the moreusers it has. Network effects can be direct—for example, a faxmachine becomes more useful as other people also use faxmachines. Network effects can also be indirect, so that theyflow across different sets of users. For example, Uber wouldnot be a very useful app for a rider if there were no driversusing the platform. Similarly, drivers would not want to usethe Uber app if no riders were using it. The presence of network effects has implications for the

strategies that firms deploy in such markets—for example, itmight be attractive to subsidize initial users to help build crit-ical mass. Network effects also were thought to potentially bea source of market power because larger firms would havestronger network effects as they had more users, which couldhelp reinforce their incumbency and make it costly for smallor new firms to challenge them.The canonical antitrust case in which this argument was

central to the legal pleadings was the antitrust action1 againstMicrosoft.2 The argument was that Microsoft used its largeuser base to encourage software developers and computerhardware markets to focus their efforts on Windows. This led

What Have We Learned In the Last Decade?

Network Effects and Market Power B Y C A T H E R I N E T U C K E R

Catherine Tucker is the Sloan Distinguished Professor of Management

Science at MIT Sloan School of Management, Cambridge, MA, and Research

Associate at the National Bureau of Economic Research (NBER).


7 8 · A N T I T R U S T

switching costs, due to the time that users had invested in set-ting up and personalizing their profiles and establishing theirnetworks. However, that “sunk cost” was rendered negligiblewhen MySpace users’ friends migrated to Facebook. Parallelto these developments, other social network-type sites, such asFriends Reunited and Yik Yak, have sprung up quickly andthen disappeared.5

In a similar vein, we have seen huge failures by products inthe social networking space that, under traditional theories ofnetwork effects, should have been huge successes. The mostobvious of these is Google Plus, which was one of Google’smost salient attempts to compete in the social networkingarena. Indeed, the executive in charge of the venture high-lighted the huge strategic drive at Google behind GooglePlus: “We’re transforming Google itself into a social destina-tion at a level and scale that we’ve never attempted—orders ofmagnitude more investment, in terms of people, than any pre-vious project.”6 Google Plus enjoyed the support of over1,000 employees (including top engineers), as well as CEOsupport. In theory, Google Plus should have had networkeffects and consequent critical mass on its side. This is becauseit was able to “seed” its initial social network with 90 millionusers through the integration of other Google services, suchas YouTube, in its signup process.7 However, Google Plusfailed to gain traction as a social network. This failure is striking simply because early economic

analysis often relates a platform’s competitive advantage to itsuser base. Here, though, Google Plus quickly failed despitea huge installed user base because the features it offered didnot provide any advantages over existing social networks. Another earlier but similar failure was Google Buzz, which

used data about Gmail’s most-mailed contacts to seed con-tact information. In theory, this should have been a power-ful tool, as it reduced the need for users who switched to theplatform to establish their social network. However, ratherthan encouraging users to embrace the platform, this use ofemail data to quickly build critical mass instead garneredcriticism from a privacy perspective.8 In neither instance wassimply creating accounts for a critical mass of users or pop-ulating these accounts with information enough to sustainusage on the platform.

The Instability of Network Effects Also FrequentlyLeads Users to Use Multiple Platforms, Increasing Com -petitive Pressure.This analysis and the examples above sug-gest that simply having a large number of users is not suffi-cient for a modern platform to enjoy sustainable marketpower. Given this, it is natural to also consider another con-straint on the importance of network effects in digital plat-forms, which is the extent to which users “multihome” or“singlehome.”9 Early work on two-sided markets, often stem-ming from litigation in the credit card industry, highlightedthe point that, after establishing a certain market definition,one of the key subsequent questions is whether users multi-home (use multiple platforms) or singlehome (use a singleplatform) within this market. Multihoming increases com-

be less a function of an entire user base than a function of thescope of a user’s interactions in a digital ecosystem. Third, insome instances, the addition of certain users to an ecosystemcan have negative effects for the relative attractiveness of thatplatform. These shifts mean that it can no longer be assumedthat the mere existence of network effects will lead toentrenchment.

Network Effects in Digital Platforms May ImplyRapid Instability Rather Than EntrenchmentThese changes reflect the changing nature of technology plat-forms. First, the evolution of multiple different devices, suchas smartphones, tablets, and digital assistants, means thatnetwork effects no longer are intertwined with a particulardefinition of hardware, as was the case with the desktop com-puter in the 1990s.Instead, platforms that exhibit network effects may be

completely virtual. Nowhere is this shift more striking thanin the world of purely digital platforms (such as social net-works, ride-sharing apps, or digital marketplaces), which donot depend on any one type of hardware and, as a conse-quence, have low learning costs and require few direct invest-ments from users.

The Instability of Platform Dominance Means thatthe Number of Users in a Platform Is Not NecessarilyPredictive of Market Power. In my previous work, I dis-cussed the instability that has been the hallmark of social net-works.4 On the face of it, a social network should have strongnetwork effects. The only purpose of visiting LinkedIn orFacebook is to connect with others, and if no one else is there,they serve little purpose. However, the history of social net-works has suggested anything but entrenchment. Launched in2002, Friendster is often considered the first real social net-work. However, it was quickly replaced by MySpace, and by2006, MySpace surpassed Google as the most visited websitein the United States. The subsequent decline of MySpace, and the speed with

which users switched to Facebook, was also startling, and hasattracted much academic inquiry. In particular, what is strik-ing is that one might have expected MySpace to exhibit

[P]latforms that exhibit network effects may be

completely vir tual. Nowhere is this shift more striking

than in the world of purely digital platforms (such as

social networks, ride-sharing apps, or digital market-

places), which do not depend on any one type of

hardware and, as a consequence, have low learning

costs and require few direct investments from users.

S P R I N G 2 0 1 8 · 7 9

because they use them to reach similar audiences. Such analy-sis is further complicated by the fact that these are two-sidedplatforms, and that advertisers may (or may not) view suchplatforms as complete substitutes on which to show ads.Even if a user singlehomes when reaching a specific audi-

ence, the rise and fall of such platforms may also be more dra-matic and renders such platforms far more vulnerable to amarginally superior competitor. The sudden decline ofMySpace can be explained by the idea that given limited abil-ity to spend time on two similar social networks, usersswitched far more quickly to a competitor.

Network Effects Are Localized and Therefore Less Powerful Than Previously ThoughtTypically, the first models of network effects considered theirstrength to be a function of the number of total users of thetechnology or platform. However, my research has shownthat this is not correct and that network externalities areoften very local in a way that renders the general size of theplatform unimportant.11 This conclusion rests on two obser-vations. First, in a world of scarce attention, users generallytend to focus only on connections and interactions that mat-ter personally to them when deciding what platform or serv-ice to use. Second, the evolution of platforms has led themto be more personalized and individualized in the servicesthey offer. This, in turn, contributes to the localization of net-work effects. Of course, this has parallels in early antitrustcases involving operating systems in which it was recognizedthat, rather than the number of software applications, whattended to matter was a handful of critical software applica-tions. However, as digital technology has evolved and becomemore widely used, personalization has perhaps made theidentity of those critical applications vary more across users.Take, for example, a restaurant review site. A naive view

would be that a user starts using the website when the num-ber of reviews across the website reaches a certain criticalmass. However, that is unlikely to be the case. For example,if I am in Boston, I simply do not care if there are many ver-sus few reviews of Seattle restaurants. Furthermore, evenwithin Boston I may not care about restaurant reviews forcuisines I would never eat, or restaurants I could not afford.In other words, what matters for my decision to start usingthat platform is reviews of restaurants that I am conceivablyinterested in. All other reviews are irrelevant for my decision.The effects of personalization can also be seen in digital

advertising markets. An advertiser may be unconcerned aboutthe total number of users on a particular platform, but insteadvery interested in the presence of a few (otherwise hard to tar-get) individuals. The entire value of digital advertising,indeed, rests on the premise that it can be used to target spe-cific individuals, rather than the same ad appearing to allusers of a platform.The only exception to this rule is that network effects

become less limited in scope at times of instability and uncer-tainty. I have shown that, in uncertain times, people may take

petition within that market, as platforms may use pricereductions or quality improvements to try to entice users tospend more time on their platforms rather than on others. Itis easy to see the importance of this consideration with cred-it card markets, where customers often carry multiple typesof credit cards, and merchants often accept multiple types ofcredit cards.Whether a market exhibits multihoming or singlehoming

is closely related to the degree of lock-in that a user experienceswhen using a platform. In the Microsoft case, the idea thatusers would not multihome across operating systems, butinstead would singlehome and only use a single operatingsystem, was often used to explain the relative power ofMicrosoft’s network effects. Here, technological standardsmade multihoming across platforms hard, as it would be dif-ficult practically, for example, to switch from the word-pro-cessing environment in Windows to that of another operat-ing system, given the learning costs associated with the switch.However, as technology evolved it became apparent that whatenabled multihoming were different types of uses for tech-nology platforms where the application could be more easilyported and where communication or use did not depend onthe application programming interface (API) to the operatingsystem. As a striking example of this, the advent of cloudcomputing has provided users with easier ways to port docu-ments across operating systems, muting such concerns.In contrast to earlier cases surrounding platforms, which

were often focused on particular hardware configurations,many new platforms are entirely digital and operate inde-pendently of hardware. This facilitates multihoming by users.Ride-sharing platforms are a useful example of this. From itsinception, the ride-sharing industry has been characterized byfierce competition, since users can easily have (for example)both Lyft and Uber apps installed on their phones and choosewhichever service has the most available drivers, and conse-quently the best prices at that very moment. Furthermore,many drivers also have both the Lyft and Uber apps installedand choose to operate on whichever platform currently hasthe most customers and consequently is offering the mostprofitable rides.Even platforms characterized by direct network effects can

exhibit multihoming in the new digital environment. Socialnetworks such as LinkedIn and Facebook were founded withthe intention of allowing an individual to reach out to verydifferent audiences. Facebook was originally intended toallow easy connections across college friends and similarpeers. LinkedIn, by contrast, was intended to provide anenvironment that allows users to connect with potentialemployers and professional contacts. As a consequence, someusers multihome across these two platforms in order to reachseparate audiences.10 However, this highlights the difficultyof market definition in such a market. As a matter of tech-nology, both platforms are similar. However, they may not beprecise substitutes for some users who use them to reach dif-ferent audiences, while other users view them as substitutes

people can be found. Though the usual narrative is one ofcool people avoiding those they deem as uncool on plat-forms and therefore pursuing platforms with a limited (butselective) audience, there are other narratives. For example,when I studied the spread of bitcoin at MIT, one thing Iobserved was the need for those who ordinarily adopted tech-nologies early on to be unique. They rejected the technolo-gy if they felt that those people who were not natural earlyadopters were also using it.15 This was despite the fact thatnetwork effects would have predicted that more users wouldhave led to more (or, at least, not fewer) users, regardless ofthe perceived social attractiveness of that user base.Together, these three forces can lead users to choose plat-

forms that exhibit lower levels of usage or weaker networkeffects. This directly contradicts the earlier wisdom that alarger platform size is always competitively desirable andbestows sustainable market power.

Conclusion and Implications for AntitrustEnforcementTo summarize: � Shifts in the nature of technology away from hardwaretowards purely digital platforms reduce the likelihood ofa positive feedback loop that can reinforce incumbency.Network effects no longer imply entrenchment but insteadcan lead to instability.

� The shift to personalization and individualization in theprovision of digital technologies means that networkeffects are often very localized and unlikely to be simply afunction of the number of users or firm size.

� Network effects may not always be positive. In some cases,having a large network may even act as a detriment,enabling differentiated competition from entrants.All three of these developments have potential implica-

tions for antitrust enforcement in platforms. They suggestthat, in general, platform markets may still be competitiveeven if larger firms in these industries exhibit both sizable userbases and competitive dynamics, which are driven by networkeffects. This implies a tempering of antitrust enforcementactions surrounding market dominance of digital platformspredicated simply on their relative size of user base. It alsoimplies some (but not all) alleviation of concerns that anti -trust authorities may have surrounding stances towards merg-ers that involve the combination of digital platforms.�

1 United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001).2 Timothy F. Bresnahan, Network Effects and Microsoft (Stanford Inst. forEcon. Policy, Discussion Paper No. 00-51, 2001).

3 Janet I. Tu, Microsoft Has Gone from 97 Percent Share of Computer Marketto 20 Percent, SEATTLE TIMES (Dec. 7, 2012), http://blogs.seattletimes.com/microsoftpri0/2012/12/07/goldman_sachs_microsoft_os_has_gone_from_more_than/.

4 Alexander E. Marthews & Catherine E. Tucker, Social Networks, Advertising,and Antitrust, 19 GEO. MASON L. REV. 1211 (2011).

8 0 · A N T I T R U S T


account of the “option value” of having a large critical massof users using a product or platform.12 However, if users areable to predict how they will use a platform, this naturallylimits the scope of network effects.

Network Effects Are Not Always Positive, But Can Be NegativeUsually network effects in platforms are modeled as positive.This means that, from an antitrust perspective, a larger num-ber of users is considered to potentially reduce competitivepressure. However, there are three reasons recent research hassuggested that more users may not lead to a sustainable mar-ket dominance (or power).First is for simple reasons of congestion. An individual user

may prefer to avoid a platform that offers too much choice,and instead choose a platform that fulfills a curation function.For example, a platform such as Witchsy.com (a new plat-form that focuses on offbeat art) may sell many products sim-ilar to those one could find on a more general platform forhandcrafted goods such as Etsy.com. However, in thisinstance a user may appreciate the benefits to them, in termsof both saving time and highlighting trends, of having amore limited selection of curated goods to choose from.Indeed, research has emphasized the importance of a well-designed curation system for encouraging users to engagewith a platform due to lower search costs.13 This means thatthere is no reason to think that “buyers” will always be like-ly to engage with a two-sided platform that has a larger num-ber of “sellers” on the platform. My research has also shownthat a large number of “sellers” can be off-putting to other“sellers” who are contemplating joining the platform, hold-ing the number of “buyers” constant.14 This is because ofmore traditional concerns that sellers might face potentialcongestion and pricing pressure if they are competing withtoo many other sellers on the platform.Second, users’ desires for privacy (or at least for control

over their audience) can lead them to value the ability to havea smaller audience. In some sense, the rise of Facebook (andlater Snapchat) can be explained by a very unintuitive set offeatures that allowed users to limit their potential audience.Through its privacy settings, Facebok offered users the abil-ity to restrict the accessibility of their data. Snapchat wenteven further and encouraged private (and ultimately disap-pearing) conversations between only two users. In both ofthese instances, the ability of users to restrict their audienceand control the usage of the platform—and, by consequence,limit the presumed benefits of network effects of the plat-form—enhanced their growth.Third, there are often more socially and culturally ground-

ed reasons at play that may lead a large number of users to beunattractive. One important concept in branding is that ofa dissociative group. This is a group of people—perhaps“Nerds,” perhaps “Soccer Moms”—who, for whatever rea-son, are deemed uncool or “dissociative.” This leads peopleto spurn platforms or avoid digital ecosystems where such

S P R I N G 2 0 1 8 · 8 1

5 Friends Reunited (http://www.friendsreunited.com/) shut down in 2016; YikYak shut down in 2017. Jefferson Graham, Yik Yak, The Once Popular andControversial College Messaging App, Shuts Down, USA TODAY, Apr. 28,2017, https://www.usatoday.com/story/tech/talkingtech/2017/04/28/yik-yak-shut-down/101045670/.

6 Seth Fiegerman, Inside the Failure of Google+, a Very Expensive Attempt toUnseat Facebook, MASHABLE (Aug. 2, 2015), https://mashable.com/2015/08/02/google-plus-history.

7 Jon Brodkin, Google Doubles Plus Membership with Brute-force SignupProcess, ARS TECHNICA (Jan. 22, 2012), https://arstechnica.com/gadgets/2012/01/google-doubles-plus-membership-with-brute-force-signup-process/.

8 Press Release, Fed. Trade Comm’n, FTC Charges Deceptive Privacy Prac -tices in Google’s Rollout of Its Buzz Social Network (Mar. 30, 2011),https://www.ftc.gov/news-events/press-releases/2011/03/ftc-charges-deceptive-privacy-practices-googles-rollout-its-buzz (last visited Jan. 16,2018).

9 Jean-Claude Rochet & Jean Tirole, Platform Competition in Two-sidedMarkets, 1 J. EUR. ECON. ASS’N 990 (2003); Mark Armstrong, Competitionin Two-sided Markets, 37 RAND J. ECON. 668 (2006).

10 However, this line is also blurring as LinkedIn has deliberately becomemore “social” and Facebook is now often used to reach across professionalcolleagues as well as college friends, potentially meaning that now someusers may also singlehome.

11 Catherine E. Tucker, Identifying Formal and Informal Influence in TechnologyAdoption with Network Externalities, 54 MGMT. SCI. 2024 (2008).

12 Catherine E. Tucker, Network Stability, Network Externalities, and TechnologyAdoption, In 37 ADVANCES IN STRATEGIC MANAGEMENT: ENTREPRENEURSHIP,INNOVATION & PLATFORMS 151 (Jeffrey Furman et al. eds., 2017).

13 Anindya Ghose, Panagiotis G. Ipeirotis & Beibi Li, Designing Ranking Systemsfor Hotels on Travel Search Engines by Mining User-Generated and Crowd -sourced Content, 31 MKTG. SCI. 493 (2012).

14 Catherine E. Tucker & Juanjuan Zhang, Growing Two-Sided Networks byAdvertising the User Base: A Field Experiment, 29 MKTG. SCI. 805 (2010).

15 Christian Catalini & Catherine Tucker, When Early Adopters Don’t Adopt, 357SCI., NO. 6347, at 135 (2017).

� CLAIM SUBSTANTIATION REFERS TO the obligation of advertisers to ensure that themessages they communicate about their products or services are truthful and not misleading. In practice, meeting this obligation can be very challenging for marketers and their legal counsel.There are several reasons for this: The law that governs advertising claim substantiation is drawnfrom a variety of different legal practice areas and disciplines. And because advertisers are legallyresponsible for both express and implied claims, this question of communication and perceptioncan be a crucial first step. Finally, claim substantiation is complicated by the very flexibility of the legal standard. As a general matter, advertisers are required to have a “reasonable basis” for their claims.The purpose of this handbook is to provide legal practitioners with an introduction to the law,

principles, and challenges of claim substantiation. The guide provides legal principles, examples,and numerous suggestions for further research. Topics such as privacy and promotions are touchedupon, but the handbook's main focus is on the legal issues that relate to claims—the express andimplied statements made in advertising about a product or service. The Handbook draws upon theexperience of some of the nation's leading experts in advertising and consumer protection law andthe contributions of attorneys within the Consumer Protection and Advertising Disputes andLitigation Committees of the ABA Antitrust Section. This handbook will assist marketers and theirattorneys in navigating the uncertain waters of claim substantiation and help them in evaluating risk and providing sound legal advice on remedies, liabilities, and opportunities.



Advertising Claim Substantiation Handbook


Pricing:$199.00 List Price / $189.00 ABA Member /$179.00 AT Section Member


8 2 · A N T I T R U S T

expert submissions to a 2017 conference on algorithmic col-lusion hosted by the Organisation for Economic Co-opera-tion and Development provide an excellent overview of thestate of thinking.6

The antitrust community is largely playing catch-up ontechnical aspects of AI and ML. As the Acting Chair of theFederal Trade Commission Maureen K. Ohlhausen hasnoted, “The inner workings of these tools are poorly under-stood by virtually everyone outside the narrow circle of tech-nical experts that directly work in the field.”7

While antitrust attorneys do not need to know all thenuts and bolts of AI and ML technology, a basic under-standing is necessary to assess the implications of the AI/MLresearch on antitrust and related issues, such as algorithmiccollusion. This is the purpose of the discussion below.Specifically, through a series of examples, I introduce funda-mental concepts in ML. Along the way, I also discuss a widevariety of ML applications in the law and economics fields tobuild readers’ understanding of AI/ML. I conclude with abrief discussion of algorithmic collusion. My goal is to introduce the minimum necessary but still

provide a systematic discussion without assuming that read-ers have any prior exposure to this topic. This focus distin-guishes the current article from some of the others in the lit-erature. Readers interested in furthering their understandingof any specific application are encouraged to read the articlesreferenced here.

How Do Machines Learn?Learning Through Examples. If you are an antitrust attor-ney who works with experts in economics or statistics, youprobably already know what a linear regression is. So youalready know one of the most widely used ML techniques.No, you won’t find the term “machine learning” in chapterson either regressions or statistics in the Reference Manual ofScientific Evidence. And yes, this is because neither economistsnor statisticians use the term “machine learning.” Computerscientists do.8 The very technique of a regression, however, isin fact an ML method. The concept of regression is actually simple and intuitive.

We do “mental” regressions all the time. For instance, we allknow roughly the average temperature of the summer wherewe have lived for ten years. When we compute that average,

THERE HAS BEEN GROWING INTERESTin the legal community in artificial intelligence(AI), and more specifically in machine learning(ML).1 A Google search for “machine learningand law” returns 80 million hits.2 Blog posts,

news paper pieces, academic papers, and startups aiming tobring the technology to the legal field are now everywhere. Atleast two major factors are behind the recent explosion ofAI/ML in general: the availability of so-called big data, whichcan be intuitively understood as a massive amount of data orinformation, and ever-increasing computing power. The legal community has used AI/ML technologies to

tackle challenging problems. For example, a set of ML meth-ods known as predictive coding (see more below) has beenused to facilitate the costly litigation discovery documentreview process. ML methods have also been used to predictcase outcomes. Such applications can be highly valuable.3

In the antitrust community, the recent interest in AI is alsodriven in part by concerns about algorithmic collusion. Atleast two ways in which computer algorithms could facilitatecollusion have been identified. First, computer algorithmscould be used to implement a price-fixing agreement, e.g., anagreed-upon price or production level or automating thedetection of “cheating” and retaliation. In 2015, the U.S.Depart ment of Justice charged a seller on Amazon market-place with doing exactly that.4 Note that because these algo-rithms are used as tools by human cartel members, they arehardly learning algorithms.5 This is an important observationbecause the antitrust community is also concerned withanother type of much more sophisticated algorithmic collu-sion, i.e., the possibility that algorithms could ultimatelylearn to collude without human facilitation. Much has beenwritten on this topic, and I will reference some of that liter-ature in this article. The background notes for as well as

An Antitrust Lawyer’s Guide to Machine Learning

B Y A I D E N G

Ai Deng, PhD, is a Principal at Bates White Economic Consulting and a lec-

turer in the Advanced Academic Program, Johns Hopkins University. He has

over a decade of consulting experience and has written extensively on the

use of analytical tools in litigation. The views expressed herein are those

of the author and do not necessarily reflect the opinions of Bates White

or its cl ients or Johns Hopkins University or its affi l iates. Heather

Dittbrenner provided excellent editorial assistance. I thank Julian Chan

and Aparna Sengupta for helpful comments. All errors are mine.

S P R I N G 2 0 1 8 · 8 3

what we do is in essence a regression, albeit a very simple one.The key ingredient for such a calculation is a collection ofwhat we could call examples, i.e., data on temperature in thesummer. Let us consider a more elaborate case. Figure 1 shows U.S.

retailer sales at a monthly frequency between 1992 and 1998.Several salient features are notable. First, the sales are grow-ing roughly at the same rate year by year. Second, the sales areseasonal, peaking in December of each year, dipping inJanuary and February, and then “recovering” starting inMarch. The cycle repeats itself every year in our data. Giventhese regularities, most of us could make a reasonably confi-dent prediction of retail sales in the next year.

Figure 1: U.S. Monthly Retailer Sales

Source: Federal Reserve Bank of St. Louis.

It turns out that these regularities are exactly what a regres-sion can be designed to capture. The mathematical formula-tion of a regression simply formalizes these features and allowsus to deal with more complex situations. Just as with the caseof computing average temperature, in this case, the key inputto our learning process is the collection of examples of his-torical relationship between retail sales and the season (orsimply time). Effectively, these examples guide or supervisehow we learn. Perhaps not surprisingly, the related ML tech-niques (regressions included), i.e., those that rely on the avail-ability of examples, are called supervised learning methods.And the data on examples are also known as training data ora training sample in the sense that they allow us to train thelearning process. You can also see why it makes sense to callthem ML methods. They at least mimic in concept how ahuman learns about our world.9

Example:To antitrust attorneys and economists, the mostfamiliar application of regression is probably a model ofprices of the product in question. The model relates theprices to the observed drivers of prices (supply and demandfactors). Of course, the quantity of interest is not limited toprices. In a merger analysis, for example, economists may alsobuild regression models for market shares. In fact, regression

analysis is rather common in today’s antitrust cases. Not sur-prisingly, there are many references on regressions written forthe antitrust audience.10

Example: As more and more electronic documents are pre-served and become available, identifying relevant documentsin the legal discovery process has become a costly endeavor.Against this backdrop, a set of supervised learning techniques,generally called predictive coding, has been employed to facil-itate this process. A prototypical predictive coding approachworks as follows. First, a subset of potentially relevant docu-ments is selected. Then, human experts review a random sub-sample of these selected documents and mark the relevant andresponsive documents (together with associated metadata,such as the author and date). These marked documents pro-vide the examples necessary for the application of supervisedlearning methods.11 Since the goal is to label a document aseither relevant or not, the problem that predictive codingtries to solve is also known as classification.

Example: A buzzword in the recent AI/ML literature aswell as in the emerging antitrust and AI literature is artificialneural network (ANN), as is the closely related concept of deeplearning or deep neural network. ANN has seen a wide varietyof successful applications ranging from image recognition tomachine translation. But as with any technical jargon, theseterms are extremely vague to anyone outside the technicalfield. It turns out that ANN is just a regression (with techni-cal bells and whistles) and hence another supervised learningmethod. Figure 2 shows two possible relationships betweentwo quantities. On the top graph, the two quantities appearto have a linear relationship in that they appear to movealong a straight line, although not perfectly. The graph on thebottom shows a nonlinear relationship.12 This is why a regres-sion model that reflects a linear relationship is called a linearregression and a regression model that reflects a nonlinear rela-tionship is called a nonlinear regression. ANN is a type ofnonlinear regression, flexible in that it can capture differentand complex shapes of nonlinearity. This flexibility comeswith a cost, however. Typically, for ANN to work well, alarge number of examples is required. Antitrust practitionershave argued that, if ANN is used to design business decisionalgorithms, the complexity of this technology could signifi-cantly complicate antitrust enforcement efforts.13 For a dif-ferent take on this topic, see my paper from 2017.14

Example: Price discrimination has also been a focus ofrecent discussion in the antitrust literature. The idea is thatas companies collect more and more personal data on theircustomers, they may be increasingly capable of price dis-crimination among them. In economic terms, companiesmay be able to use personal data to gauge individual will-ingness to pay. The availability of such data as well as cus-tomers’ past purchasing/spending behavior (again, examples)could be used to train supervised learning methods to betterpredict consumer behavior and hence enable companies tooffer personalized product options and associated pricing. Inan interesting paper, FTC Commissioner Terrell McSweeny

0

50

100

150

200

250

300

Jan-92 Jul-92 Jan-93 Jul-93 Jan-94 Jul-94 Jan-95 Jul-95 Jan-96 Jul-96 Jan-97 Jul-97 Jan-98


8 4 · A N T I T R U S T

between—the shapes. We then put the objects in one groupwhen their differences are “small” and in a different groupwhen the differences are “large.” Figure 3 illustrates this.Note that if we had written down the shape on each object,we would simply look at the label and group the objectswithout the mental measuring process. If we had the labelsof each object, this would effectively turn the task into oneof supervised learning. The lack of labels requires us toapproach the problem differently and is the hallmark of whatis known as unsupervised learning. And the grouping exerciseis known as clustering in ML jargon. Again, in these types oflearning, the concept of distance is a critical ingredient andunderlies even the most sophisticated unsupervised learningtechniques.

Figure 3: Classification as Unsupervised Learning

Example: In document review, another objective is togroup documents based on certain criteria even before weknow whether a document is relevant. For example, one maywant to group documents by author or date or, in morecomplex cases, by content through the use of other unsu-pervised ML algorithms.

Example: Novelty or anomaly detection, which identifies the few instances that are different from the majority, is con-ceptually similar to clustering. In many industries (creditcard, telecommunications, etc.), anomaly detection is huge-ly important in detecting fraud. One simple but powerfulidea behind anomaly detection is to start with characterizingthe “norm.” For example, once we use a credit card longenough, the card company would be able to build a person-al profile for our spending behavior. If we have never madea purchase in a foreign country when a transaction just tookplace in that location, that transaction will be flagged as ananomaly and the card company will issue an alert. In theantitrust domain, similar techniques can be used to detectand monitor cartel formation. I elaborated on how ML/AIcould be leveraged to do so in another article.16 As an exam-ple, Professor Joseph Harrington has argued that a sharpincrease in the price-cost margin could signal the onset of acartel.17 Such a price-cost margin “screen,” as it is common-ly known in the cartel detection literature, fits nicely in theunsupervised learning framework. Thus, despite the con-cerns mentioned earlier that ML/AI could facilitate collusion,the very same set of tools might be used to deter and preventcartel formation.

and Brian O’Dea recently discussed the implications of algo-rithm-aided price discrimination on antitrust market defini-tion.15

Last but not least, we should note that because supervisedlearning relies on the training data, the quality of such datais critical. Biased data could lead to biased results. Supposethat a loan company uses a supervised learning algorithm to make loan decisions. If the training data are somehow“biased” against certain subgroups of the population, theloan decision produced by the algorithm is likely to be biasedas well. Interesting legal questions, as one could imagine,arise in such cases. Or suppose that the human experts fail toproperly classify the relevance of documents before thosedocuments are used to train the predictive coding algorithms.Then the quality of additional documents marked as relevantby the learning algorithm could suffer.

Figure 2: Illustration of a Linear and Nonlinear Relationship

Learning Through Differences. Humans engage inother cognitive tasks. Consider the following example.Suppose there is a mix of triangles and squares. The task is toput different shapes into separate groups. While this task isincredibly trivial for us, let’s think about exactly how ourbrains work in such a situation. One plausible hypothesis isthat we have “a mental ruler” that measures pairwise differ-ences of—or, using a slightly more technical term, distances

S P R I N G 2 0 1 8 · 8 5

Learning Through Trial and Error. There are stillother types of learning. Consider the case of a child learningabout different animals. When a child picks up a toy elephantbut calls it a giraffe, we would correct her. When she gets itright, we congratulate and reward her. And we repeat thatprocess until she gets it. This process is probably the mostcommon way of reinforcing proper behavior. Andrew G. Barto and Thomas G. Dietterich give another

example.18 Imagine that you are somewhere the cellphone sig-nal is not very good, talking on the phone but moving aroundto find the right spot. Every time you move, you ask yourpartner whether she can hear you better. You do this until youeither find a good spot or give up. The ML algorithms thatmimic this type of “trial and error” process are known asreinforcement learning (RL). Notice here that the informationwe receive does not directly tell us where we should go toobtain good reception. In other words, we do not have a col-lection of examples of location and reception as in a super-vised learning case, at least not much in a new environment.We make a move and then assess our current situation. AsBarto and Dietterich put it, “We have to move around—explore—in order to decide where we should go.”19 This is amain difference between RL and supervised learning. Some of the most prominent success stories of RL come

from the field of game play. AlphaGo, an RL algorithm, beatworld champions at the ancient game of Go in 2016 and2017. AlphaGo was, however, recently defeated by the nextgeneration of the algorithm, AlphaGoZero, losing all 100games played.20 In fact, AlphaGoZero uses RL to start fromscratch (hence the zero in the name of the algorithm) andtrains itself by playing against itself. In the emerging antitrustand AI literature, Ashwin Ittoo and Nicholas Petit argue that“RL is a suitable framework to study the interaction of prof-it maximizing algorithmic agents because it shares severalsimilarities with the situation of oligopolists in markets.”21

Particularly relevant to recent concerns regarding algo-rithmic collusion in the antitrust community is the multi-agent learning problem. This is where multiple parties areinvolved in the learning process and their behavior directlyaffects each other. For example, in a zero-sum game, if oneplayer wins, another player must lose. In a coordination gamelike basketball, for example, the incentives of the players onthe same team are generally aligned. In contrast, in a positive-sum game such as prisoner’s dilemma, even though the partiesunderstand that they could achieve a higher overall and indi-vidual payoff if they coordinate their behavior, there is atemptation to defect. The last situation resembles the prob-lem cartel members may face and is the most familiar toantitrust attorneys and economists. In fact, much concern hasbeen expressed in the antitrust community regarding thepossibility of colluding ML algorithms. Research on multi-agent learning in the prisoner’s dilemma type of situation isparticularly pertinent. In another article, I summarize current views and concerns

in the antitrust community regarding the possibility of col-

luding algorithms and discuss insights from a recently pub-lished multi-year AI research study by Jacob W. Crandall etal.22 There, I conclude that not all the assumptions underly-ing some concerns expressed in antitrust scholarship haveempirical support. For example, while algorithms can bedesigned to tacitly or explicitly collude in relatively simple sit-uations such as the classic prisoner’s dilemma, as well as inother situations, real-world competitive decisions are muchmore complex, presenting a nontrivial computational chal-lenge. Second, despite some expressed concerns, the goal ofunilateral profit maximizing is unlikely to lead the algo-rithms into successful tacit collusion. In fact, the researchshows technical challenges that make designing algorithmscapable of learning to cooperate rather complex. Adam Lerer and Alexander Peysakhovich, two Facebook

researchers, recently proposed an RL algorithm in a generalmulti-agent game in which each party has an incentive todefect.23 Their goal is to develop an algorithm that cooper-ates with “cooperators” (and with itself ) while avoiding ex -ploitation by “selfish” defectors. They specifically considered the Coin game, in which the

players decide whether to “pick up” the coins of the otherplayer or only their own. The players maximize their payoffif they pick up only their own coins. If we think of the coinsas geographic markets or customers, then this payoff-maxi-mizing strategy somewhat resembles the anticompetitive con-duct of market or customer allocation. Two findings of thisresearch are of particular interest to antitrust practitioners.First, in their experimental study, the authors find that par-ties can increase their individual payoff by 3,150 percent ifboth adopt the proposed algorithmic strategy and learn tocooperate as opposed to acting always independently andselfishly. Second, suppose one player adopts another learningbut selfish algorithm, i.e., an algorithm that attempts to findthe best strategy based on past actions of others and reactsaccordingly. The authors find that their RL algorithm actu-ally “teaches” this other learning algorithm to cooperatethrough repeated interactions. It is worth noting that their RL algorithm is based on and

motivated by the so-called tit-for-tat strategy, known to per-form well in prisoner’s dilemma situations in previous exper-imental studies.24 Since the goal of these researchers is todesign machines that can cooperate and work with people

Par ticular ly relevant to recent concerns regarding

algorithmic col lusion in the antitrust community is

the multi -agent learning problem . This is where

mult iple par t ies are involved in the learning process

and their behavior directly affects each other.

4 See Press Release, U.S. Dep’t of Justice, Former E-Commerce ExecutiveCharged with Price Fixing in the Antitrust Division’s First Online MarketplaceProsecution (Apr. 6, 2015), https://www.justice.gov/opa/pr/former-e-commerce-executive-charged-price-fixing-antitrust-divisions-first-online-marketplace.

5 “Of course, most algorithms today still operate based on instructionsdesigned by human beings and there is no doubt that humans will be inmost cases responsible for the decisions made by algorithms.” OECDSecretariat, Algorithms and Collusion: Competition and Policy in the DigitalAge 101 (OECD Background Paper, June 2017). See Michal S. Gal, Algo -rithmic-Facilitated Coordination: Market and Legal Solutions, 2 CPI ANTITRUSTCHRO. (2017) (highlighting several ways that algorithms could make collu-sion more effective).

6 See OECD, supra note 5. 7 Maureen K. Ohlhausen, Acting Chairman, Fed. Trade Comm’n, Should WeFear the Things That Go Beep in the Night? Some Initial Thoughts on theIntersection of Antitrust Law and Algorithmic Pricing (May 23, 2017),https://www.ftc.gov/system/files/documents/public_statements/1220893/ohlhausen_-_concurrences_5-23-17.pdf.

8 This is changing as ML is gaining more popularity among fields outside ofcomputer science.

9 In fact, the analogy to human learning does not stop there. ML also makesheavy use of “test” data to “validate” what it learns about the problem.

10 See, e.g., ABA SECTION OF ANTITRUST LAW, ECONOMETRICS: LEGAL, PRACTICAL,AND TECHNICAL ISSUES (2d ed. 2014); see also Ai Deng, Book Review:Econometrics—Legal, Practical, and Technical Issues, 61 ANTITRUST BULL.461 (2016).

11 For more details, see, e.g., Predictive Coding, Presentation for PanelDiscussion at ABA Section of Litigation 2012 Section Annual Conference(Apr. 18–20, 2012), https://www.americanbar.org/content/dam/aba/administrative/litigation/materials/sac_2012/14-1_predictive_coding_written_materials.authcheckdam.pdf.

12 For the readers familiar with regression analysis, the second chart is basedon a simple logistic function. Of course, to be technically correct, nonlin-earity pertains to model parameters, not the quantities themselves.

13 “Therefore, by relying on deep learning, firms may be actually able to reacha collusive outcome without being aware of it, raising complex questions onwhether any liability could ever be imposed on them should any infringementof the law be put in place by the deep learning algorithms.” See OECD, supranote 5, at 31.

14 Ai Deng, When Machines Learn to Collude: Lessons from a Recent AIResearch Study on Artificial Intelligence (Sept. 2017), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3029662.

15 Terrell McSweeny & Brian O’Dea, The Implications of Algorithmic Pricing forCoordinated Effects Analysis and Price Discrimination Markets in AntitrustEnforcement, ANTITRUST, Fall 2017, at 75.

16 Ai Deng, Cartel Detection and Monitoring: A Look Forward, 5 J. ANTITRUSTENFORCEMENT 488 (2017). In fact, I believe that AI/ML could be leveragedto significantly enhance antitrust compliance programs. See Ai Deng, 4Reasons Why We May Not See Colluding Robots Anytime Soon, LAW360(Oct. 3, 2017).


8 6 · A N T I T R U S T

and ultimately simplify our lives, it is understandable thatthey started with an approach that works in a similar context.However, when considering adoption of a pricing algorithmwith “capability to collude” as a design feature, firms havegood reason to be extra cautious. Of course, there are still significant technical challenges and

limitations when designing RL algorithms to take over busi-ness decisions. Several were noted in the Facebook researchpaper itself. From a more practical perspective, we need to rec-ognize that real-world decisions and strategies are much morecomplex than those in the games studied. So part of the chal-lenge is scalability. Equally important is the fact that businessdecisions are typically made under a great deal of uncertain-ty as to the external environment. In fact, the economics lit-erature has emphasized that among other challenges, demanduncertainty is a major hurdle for any tacit collusion agree-ment.25 In most, if not all the situations studied, however, therules of the game are fixed, and there is little to no such exter-nal uncertainty. This appears to be an underappreciated obser-vation in the antitrust literature.26

SummaryAI/ML is suddenly almost everywhere. From smartphoneapps to computer-aided medical diagnosis, from face detec-tion on a digital camera to self-driving cars, it is no wonderthat leading AI researcher and Coursera cofounder AndrewNg recently declared that “AI is the new electricity.”27

The recent interest in AI and ML in the antitrust com-munity seems to be taking us into this exciting yet unfamil-iar territory. Fortunately, the intuition behind AI/ML is notdifficult to understand, thanks to the fact that many of thetechnologies are inspired by how we as humans learn.�

1 In the antitrust literature, ML and AI are often used interchangeably. Formost practical purposes, this is harmless. Strictly speaking, however, AI and ML are two different concepts. To put it intuitively, the difference is verysimilar to the difference between intelligence and learning. Learning is notintelligence but rather a way to achieve intelligence. For more detailed andtechnical discussion, see STUART JONATHAN & PETER NORVIG, ARTIFICIALINTELLIGENCE: A MODERN APPROACH (1995).

2 The search was run on November 19, 2017.3 For an excellent summary of the technical application of ML in law, see HarrySurden, Machine Learning and Law, 89 UNIV. WASH L. REV. 87 (2014).

A Quick Reference Guide

Categories of ML methods28 Key idea Example methods/applications

Examples (also known as Linear regression, Artificial neural network, Supervised learning training data, training sample) Classification, etc.

Unsupervised learning Distance Clustering, Novelty/Anomaly/Outlier detection, etc.

Reinforcement learning (RL) Trial and error Single-agent learning, Multi-agent learning, etc.

Note: While RL is conceptually different from supervised learning (and unsupervised learning), implementation of RL typically uses supervised learning methods. For example, deep RL uses an artificial neural network to train its algorithms.

S P R I N G 2 0 1 8 · 8 7

17 Joseph E. Harrington, Jr., Detecting Cartels, in HANDBOOK OF ANTITRUSTECONOMICS (Paolo Buccirossi ed., 2008).

18 A.G. Barto & T.G. Dietterich, Reinforcement Learning and Its Relationship toSupervised Learning, in HANDBOOK OF LEARNING AND APPROXIMATE DYNAMIC

PROGRAMMING 47–64 (J. Si et al. eds., 2004). 19 Id. at 50. 20 See Larry Greenemeier, AI versus AI: Self-Taught AlphaGo Zero Vanquished Its

Predecessor, SCIENTIFIC AM., Oct. 18, 2017. 21 Ashwin Ittoo & Nicolas Petit, Algorithmic Pricing Agents and Tacit Collusion:

A Technological Perspective (Oct. 2017), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3046405.

22 Jacob W. Crandall et al., Cooperating with Machines (Cornell UniversityWorking Paper 2017), https://arxiv.org/abs/1703.06207.

23 Adam Lerer & Alexander Peysakhovich, Maintaining Cooperation in ComplexSocial Dilemmas Using Deep Reinforcement Learning (Cornell UniversityWorking Paper 2017), ttps://arxiv.org/abs/1707.01068. I thank Paul A.Johnson for helpful discussion about this research.

24 ROBERT M. AXELROD, THE EVOLUTION OF COOPERATION (rev. ed. 2006).

25 See, e.g., Edward J. Green, Robert C. Marshall & Leslie M. Marx, TacitCollusion in Oligopoly, in 2 THE OXFORD HANDBOOK OF INTERNATIONAL ANTITRUSTECONOMICS (Roger D. Blair & D. Daniel Sokol eds., 2014). Note that theauthors also carefully distinguish the legal and economic terminologies ofcollusion.

26 For an exception, see Ittoo & Petit, supra note 21. In addition, there are alsonontechnical reasons we may not see colluding robots any time soon. I offered some thoughts in another article, see supra note 16.

27 See Andrew Ng, AI Is the New Electricity, WALL ST. J. video (Jan. 24, 2018),http://www.wsj.com/video/andrew-ng-ai-is-the-new-electricity/56CF4056-4324-4AD2-AD2C-93CD5D32610A.html.

28 Some readers may also have heard of the term “semi-supervised learning.”As the name suggests, this is a category of methods that try to integratethe ideas of supervised and unsupervised learning. Additional assump-tions are typically required, however, for semi-supervised learning methodsto work well. For an introduction and its potential application in cartel detec-tion and monitoring, see Deng, supra note 16.

� SINCE PUBLICATION OF THE SECONDedition of Insurance Antitrust Handbook, theinsurance industry has been in the headlinesregarding antitrust-related issues, includingmerger challenges, state and federal antitrust enforcement actions, and privateparty multidistrict litigation class action proceedings. This third edition has beencompletely updated to incorporate andassess the changes that have occurred ininsurance antitrust law over the last tenyears. Insurance Antitrust Handbook, ThirdEdition is a general guide to the applicationof the federal antitrust laws to common activities in the insurance industry.Following an introductory discussion,

chapter II provides an overview of the federalantitrust laws. The handbook focuses primarily on federal antitrust laws, but theprinciples discussed are generally applicable

to state antitrust laws, which typically arepatterned on the federal statutes. The nextseveral chapters examine whether the insurance industry's main practices—suchas collective ratemaking, standardization of insurance policy forms, and joint underwriting—would be lawful absent an exemption. The final chapter broadlydescribes how insurance companies andtheir counsel can implement an effectivecompliance program to minimize antitrustexposure

Insurance Antitrust Handbook, ThirdEdition is a valuable resource both for theinsurance industry generally and for lawyersadvising clients regarding federal antitrustlaws.


Insurance Antitrust Handbook THIRD EDITION

Product Code: 5030628

Publication Date: 2017

Page Count: 281

Trim Size: 6 x 9

Format: Paperback

Pricing: $249.00 List Price /

$199.00 ABA Member /

$149.00 AT Section Member

8 8 · A N T I T R U S T

Thursday, May 31, 2018

8:30–8:35 CONFERENCE OPENING

8:35–8:45 WELCOME REMARKS

8:45–9:45SESSION I—RECENT DEVELOPMENTS IN KOREAOur panel of senior Korean practitioners will review recent developmentsin Korean competition law and practice.

9:45–10:00 COFFEE/TEA BREAK

10:00–11:30SESSION II—MERGER REVIEW IN NORTH ASIAMerger reviews in China, Japan and Korea are presenting unique procedural and substantive issues both in domestic and internationalmergers. Our panel of senior private practitioners will explore develop-ments in merger notification as well as key substantive issues, includingthe role, if any, of industrial policy concerns in merger reviews.

11:30–12:00 KEYNOTE SPEAKER

12:00–13:00 LUNCH

13:00–14:15SESSION III—IP LICENSING AND STANDARD ESSENTIAL PATENTSThe competition law treatment of IP licensing and SEPs continued to be a significant issue in a number of Asian jurisdictions as well as the EU,clearly reinforcing the importance of competition law to the exercise ofpatent rights. Our panel will review the latest developments in China, the EU, India, and Korea.


14:30–15:30SESSION IV—PRIVATE DAMAGES ACTIONS IN ASIA—A NEW FRONTIER?Private competition law claims continue to grow in prominence in various Asian jurisdictions. In addition, Asian companies continue to find themselves involved in U.S. litigation both as defendants and increasingly,as plaintiffs. Our panel considers developments in private enforcement in Australia, Japan, India, as well as the U.S., as relevant to Asian corporations.

15:30–17:00SESSION V—IN-HOUSE COMPLIANCE STRATEGIES IN A TIME OF COMPLEXITYThe implementation and administration of corporate compliance programshas become increasingly challenging for major Asian corporations asestablished enforcement agencies increase their activity and newer onesemerge. In addition, competition law compliance is often entwined withrelated compliance requirements in other areas, such as anticorruptionand privacy. Our panel of senior in-house counsel discuss their approaches to compliance.

17:00–18:00 RECEPTION

Friday, June 1, 2018

8:30–10:00SESSION VI—BIG DATA & COMPETITION LAW ENFORCEMENT IN ASIABig data continues to draw attention, not only in the U.S. and EuropeanUnion, but also in Asian jurisdictions such as Japan and Singapore law. Our panel will explore recent agency initiatives and the potential application of domestic competition law regimes to the collection and use of data.


10:30–12:00SESSION VII—DEVELOPMENTS IN CARTEL ENFORCEMENT IN ASIA, THE EU, AND THE U.S. The experience of agencies and defense counsel in major recent international and domestic cartel investigations has in many cases led to a rethinking of critical aspects of cartel enforcement, ranging from leniency programs to litigation. Our panel reviews recent developments in major jurisdictions.

12:00–13:00 LUNCH

13:00–13:30A CONVERSATION WITH AGENCY LEADERS

13:30–14:45SESSION VIII—CONFIDENTIALITY, PRIVILEGE, AND PROCEDURAL TRANSPARENCYThe treatment of information during the course of a civil, administrative or criminal agency investigation remains one of the most challenging practical issues for enforcers and respondents alike, particularly in matters involving multiple jurisdictions with different approaches toagency confidentiality, privilege and transparency. Our panel considerssome of these tough issues.


15:00–16:45SESSION IX—ASIAN ENFORCERS’ ROUNDTABLE

16:45–17:00 CLOSING REMARKS

17:00–18:00 RECEPTION

FACULTY: Yong Seok AHN, Kala ANANDARAJAH, Peter ARMITAGE, Elizabeth AVERY, Subrata BHATTACHARJEE, Logan BREED, Peter CAMESASCA,Kyungsun Kyle CHOI, Cecil CHUNG, Fei DENG, Ninette DODOO, Yao FENG, Andrew FINCH, Samir GANDHI, Kaoru HATTORI, Yung Yung HUI, Iskandar ISMAIL, Jonathan JACOBSON, Youngjin JUNG, Dina KALLAY, Winston KIANG, Sang-Jo KIM, Hyun Ah KIM, Genny KIM, Hyung Bae-KIM,Hiromitsu MIYAKAWA, Kirstie NICHOLSON, Kenneth O’ROURKE, KeunHo PARK, Michele PIERGIOVANNI, Harikumar PILLAY, Fiona SCHAEFFER, Hartmut SCHNEIDER, Daren SHIAU, Sanghoon SHIN, Hiromasa SHIOZAKI, Brent SNYDER, Farhad SORABJEE, Marguerite SULLIVAN, Toshiaki TADA,Randolph TRITELL, Stephen WU, Akinori YAMADA, Natalie YEUNG, Hoil YOON, Sai Ree YUN, Gary ZANFAGNA

A N T I T R U S T I Nas iamay 31–june 1, 2018seoul, south koreathe four seasons, seoul

co-chairssubrata bhattacharjee & hyun ah kim

visitambar.org/atasia

A R T I C L E S

S P R I N G 2 0 1 8 · 8 9

I N T E R N A T I O N A L D E V E L O P M E N T S

A Coat of Many Colors: Joint Ventures inInternational Merger Control B Y N A T A L I E Y E U N G A N D P A D D Y L AW

JO I N T V E N T U R E S O F T E N R A I S Ecomplex merger control questions. Divergence in thetreatment of joint ventures under competition lawsaround the world means that the jurisdictional assess-ment of the same transaction may be diametrically

different. A sound knowledge of international merger controlrules is crucial for identifying where a joint venture may trig-ger a merger filing and, where possible, for advising clients onpossible structuring options to minimize the number ofmerger filings that may be required.In Europe,1 it is well established that the creation of a joint

venture performing on a lasting basis all the functions of anautonomous economic entity (full-function joint venture)2 isa notifiable concentration. This feature of EU merger controlhas been exported to some other countries. There was someuncertainty as to whether the creation of a joint venture byway of an acquisition of joint control of a previously wholly-owned, non-full-function target undertaking (for example,purchase of a 50 percent stake in an existing plant that waspreviously wholly owned) falls within the definition of “con-centration” under Article 3 of the EU Merger Regu lation(EUMR). In the recent judgment of Austria Asphalt GmbH &Co OG v Bundes kartellanwalt,3 which is the first reference fora preliminary ruling on the subject of the EUMR regime, theCourt of Justice of the European Union (ECJ) held that a con-centration is deemed to arise only if such joint venture is of“full-function character.” This article discusses the implications of the Austria

Asphalt case for the treatment of non-full-function joint ven-tures under the EUMR regime and national regimes of cer-

tain member states of the European Union. Looking beyondEurope, we then consider more broadly the differences intreatment of joint ventures in international merger control,and highlight a number of practical points of interest whenplanning cross-border joint venture transactions.

Austria Asphalt JudgmentBackground. The transaction concerned the proposedacquisition by Austria Asphalt GmbH & Co OG of a 50 per-cent stake in an existing asphalt mixing plant in Mürzzu -schlag, Austria, which was previously wholly owned by TeeragAsdag AG (belonging to the Porr AG group). The transactioninvolved the establishment of a joint venture vehicle (GmbH& Co KG, a limited partnership with a limited liability com-pany as its general partner) under Austrian law, throughwhich Teerag and Austria Asphalt each held an effective 50percent interest in, and joint control over, the Mürzzuschlagplant. Importantly, such a joint venture would not be of full-function character because its production would be intend-ed for supply almost exclusively to the parents (which, inturn, would market and sell the product independently) andthe joint venture would not have any significant presence onthe market or independent sales to third parties. The Kartellgericht, the Competition Court in Austria,

refused to consider the application by Austria Asphalt forreview on the ground that the transaction constituted a con-centration with EU dimension under the EUMR and there-fore fell within the jurisdiction of the European Commission(EC). Austria Asphalt appealed to the Oberster Gerichtshof,the Supreme Court in Austria, which stayed the proceedingsand referred the following question of interpretation to theECJ:

Must Article 3(1)(b) and (4) of [Regulation No 139/2004]be interpreted as meaning that a move from sole control tojoint control of an existing undertaking, in circumstanceswhere the undertaking previously having sole controlbecomes an undertaking exercising joint control, constitutesa concentration only where the undertaking [the control ofwhich has changed] has on a lasting basis all the functions ofan autonomous economic entity?

To put the issue in context, it is interesting to observe thatthe Directorate-General for Competition of the EC had orig-inally issued a comfort letter to Austria Asphalt saying thatthe transaction did not appear to constitute a concentrationwithin the meaning of Article 3 EUMR, together with a dis-claimer that the view was that of a Commission service andwas not binding on the EC. However, in the proceedingsbefore the ECJ, the EC advocated a different reading from itsCommission service in the comfort letter. Advocate GeneralKokott commented that this was “extremely regrettable that,on such a fundamental and recurrent issue of competence,the Commission did not first commit to a clear and uniformapproach and then apply it consistently.”4

Ruling of the ECJ. The central question boils down tothe interpretation of whether this transaction constitutes a

Natalie Yeung is a partner and the head of the Asian competition practice

of Slaughter and May. Paddy Law is an associate of Slaughter and May.

“concentration” under EUMR. Articles 3(1)(b) and 3(4)EUMR provide that a concentration shall be deemed to arisewhere a change of control on a lasting basis results from:

Article 3(1)(b): the acquisition, by one or more personsalready controlling at least one undertaking, or by one ormore undertakings, whether by purchase of securities orassets, by contract or by any other means, of direct or indi-rect control of the whole or parts of one or more otherundertakings.

Article 3(4): the creation of a joint venture performing on alasting basis all the functions of an autonomous economicentity shall constitute a concentration within the meaning ofparagraph 1(b).

The ECJ acknowledged that this question of interpreta-tion could not be determined based on the wording of Article3 of EUMR alone because Articles 3(1)(b) and 3(4) are, ona literal reading, capable of conflicting interpretations. Article3(1)(b) covers, in the words of Advocate General Kokott, “alloperations involving a lasting change in the control of exist-ing joint ventures”5 (which could include converting an exist-ing undertaking into a joint venture), regardless of whetherthe joint venture is of full-function character. Article 3(4)does not define “the creation of a joint venture” so this pro-vision “might be understood to mean that the full function-ality applies only to the creation of new joint ventures but notto the change of an existing undertaking into a joint venturecontrolled by two companies.”6

In view of this, the ECJ interpreted Article 3 EUMR byreference to its purpose and general structure and came to theconclusion that the provision must be interpreted as mean-ing that “a concentration is deemed to arise upon a changein the form of control of an existing undertaking which, pre-viously exclusive, becomes joint, only if the joint venturecreated by such a transaction performs on a lasting basis allthe functions of an autonomous economic entity.”7 In otherwords, the ECJ held that a change from sole to joint controlof an existing undertaking would not be a concentration forEUMR purposes if it is not of full-function character.The ECJ’s judgment is not surprising as the rationale

behind its decision underpins the overall objectives of theEUMR regime. First, the ECJ noted that the purpose of theEUMR regime was to “cover operations bringing about a last-ing change in the control of the undertakings concerned andtherefore in the structure of the market.”8 Since the EUMRregime is intended to cover structural changes to the market,it is sensible to define concentration to include the creationof a joint venture from an existing undertaking (i.e., a brown-field joint venture) only if such venture is of full-functioncharacter. Second, the ECJ considered a distinction based onwhether the target undertaking existed before the transac-tion or is newly created to be an “unjustified difference intreatment,”9 as evident in the lack of any such distinction inthe recitals to the EUMR. It is difficult to dispute that con-sistent treatment is the right outcome.

As an aside, the Austria Asphalt judgment left one subtlequestion unanswered. The question referred to the ECJ didnot specifically cover the transaction effecting a change ofcontrol in a pre-existing joint venture to a new controllingshareholder. If we change the facts of Austria Asphalt slight-ly for illustration purposes, assume Austria Asphalt subse-quently exits by selling its 50 percent stake in the Mürzzu -schlag plant to a third party. Such a transaction might engageboth Articles 3(1)(b) and 3(4) EUMR and be capable ofconflicting interpretations. On one hand, the replacement ofan existing controlling shareholder by a third party in analready jointly controlled undertaking may lead to a changein the quality of control due to the identity of a new con-trolling shareholder that may, for instance, be a competitor(as opposed to a financial investor).10 This change would, onits face, fall within the definition of concentration underArticle 3(1)(b) EUMR. On the other hand, there appears tobe room for the view that such a change of control wouldinvolve the creation of a joint venture under Article 3(4), inthe sense that this is a joint venture with different character-istics and in which case the transaction would only be a con-centration if the joint venture is of full-function character. After the Austria Asphalt judgment, there are good argu-

ments that the same principles should apply by analogy, suchthat the replacement of a controlling shareholder by a thirdparty should constitute a concentration only if the joint ven-ture is of full-function character. The court’s reasoning, as dis-cussed above, would equally apply where the quality of con-trol changes in the context of a pre-existing joint venture. Forinstance, the ECJ’s remarks that “Article 3 of the regulationconcerns joint ventures only in so far as their creation pro-vokes a lasting effect on the structure of the market” wouldlend support to this interpretation.11 It is nevertheless worthnoting that the ECJ did not articulate the point in its judg-ment in the same terms as Advocate General Kokott did inher opinion, in which she states that “the conversion of anon-full-function undertaking into a joint venture cannot besubjected to EU merger control on the basis of Article3(1)(b)” and that

the concept of an undertaking must be interpreted func-tionally and encompasses every entity engaged in an eco-nomic activity . . . since an economic activity is in turnunderstood to mean any activity consisting in offering goodsand services on a given market, joint ventures without anautonomous market presence—in other words, without fullfunctionality—are by definition not caught by Article3(1)(b).12

Implications of Austria Asphalt Judgment Within Eur ope. In Advocate General Kokott’s words, although thequestion of interpretation in this first reference for a prelim-inary ruling on EU merger control regime may appear “high-ly technical,” it has a “practical significance which cannot beunderestimated”13 because this serves: (1) in a horizontalsense, to draw the dividing line between the control of con-centrations under EUMR and the enforcement of antitrust

A R T I C L E S

9 0 · A N T I T R U S T

S P R I N G 2 0 1 8 · 9 1

law; and (2) in a vertical sense, to distinguish between thecompetences of the EC and the national competition author-ities of member states. The implications of the Austria Asphalt judgment are

twofold. First, a similar transaction would clearly fall outsidethe EUMR regime. Those who might have liked to takeadvantage of the one-stop-shop principle may now have to gothrough separate jurisdictional analysis under the nationalmerger control rules of the 28 EU member states. Dependingon the transaction itself, this may or may not work in favorof the transaction parties, especially bearing in mind the bur-den of having to potentially file in a significant number ofjurisdictions that do not prescribe the requirement of full-functionality.14 The consolation is that the Austria Asphaltjudgment is expected to bring clarity to the interpretation ofequivalent or similar national merger control provisions mod-eled on the EUMR15 to reduce jurisdictional uncertaintyregarding similar transactions in future. Second, a transaction that is not a concentration auto-

matically falls outside the scope of Article 21 EUMR, whichprovides that the EUMR alone shall apply to concentrationsand, therefore, transaction parties cannot seek legal certain-ty under the ex ante clearance regime. Rather, such a trans-action must be reviewed under Articles 101 and 102 of theConsolidated Version of the Treaty on the Functioning of theEuropean Union (TFEU) and the self-assessment regimeunder Regulation (EC) No. 1/2003 applies. In a transactionthat potentially raises antitrust concerns (e.g., where com-petitors propose to form a joint venture), parties will nolonger be able to exploit the previous textual ambiguity toseek the EC’s blessing through ex ante clearance, but mustdeal with ex post enforcement risks by considering appro-priate mitigation strategies. In the Austria Asphalt case,Advocate General Kokott opined that this is the “inevitableconsequence” of Regulation No. 1/2003 and “there is noth-ing to prevent the national competition authorities frommaking it one of their priorities in relation to the enforce-ment of antitrust law (Articles 101 and 102 TFEU) to payspecial attention to occurrences on highly concentrated mar-kets such as that in the present case.”16

International Merger ControlLooking beyond Europe, the question of when a joint ven-ture transaction must be notified under local competitionlaws is far from a straightforward exercise. Joint venturestend to raise more difficult questions than an outright acqui-sition of shares/assets, a merger or an amalgamation becauseof the divergence in their treatment under merger controlrules. In some cases, the precise structure and sequence ofsteps in a joint venture transaction could significantly impactthe jurisdictional analysis. The first part of this section will highlight some nuances

and practical points of interest in relation to how merger control regimes apply to joint ventures in different ways. Astarting point is to look at the different approaches to defin-

ing “notifiable transactions” for merger control purposes.Some regimes expressly cover joint ventures, whereas othersmay refer to concepts such as control (e.g., EU, China, andSingapore) or acquisitions of an equity interest above a spec-ified threshold (e.g., Japan, Korea, and Taiwan). Furthermore,the concept of full-functionality is not widespread outsideEurope. The second part of this section will illustrate thepotentially significant impact of such divergent treatmentfor cross-border joint ventures through a case study in theliner shipping industry.

Definition of Notifiable Transactions. As a startingpoint, joint ventures are clearly caught by merger controlrules if the creation of a joint venture is specifically definedas a type of notifiable transaction, as under the EUMR. Inmany jurisdictions outside Europe, however, the jurisdic-tional tests do not include any specific reference to joint ven-tures.17 This does not mean that they are categorically exclud-ed from a merger control regime; rather, the jurisdictionalanalysis would turn on whether the joint venture transactionmay consist of any of the categories of notifiable transaction. For instance, where the types of notifiable transactions are

defined to include acquisitions of shares or assets, the creationof a joint venture may well be caught because setting up ajoint venture often involves an acquisition of shares and/orassets that may amount to a change of control or effect anacquisition of an interest above the relevant threshold. Con -versely, in Indonesia, an acquisition of assets is not a type ofnotifiable transaction so it may be possible to structure ajoint venture transaction to involve asset transactions only toavoid a merger filing requirement. In India, as opposed to anacquisition of shares or assets, local merger control rulesdefine the type of notifiable transaction with reference to anacquisition of enterprise. Since an enterprise means a goingconcern that has an existing business, it is understood that agreenfield (contrasted with brownfield) joint venture isunlikely to be notifiable in India.

Joint Control. A cause for complexity in jurisdictionalanalysis of joint ventures is the question of joint control. Insome jurisdictions, joint ventures are caught where they effecta change of control that may be on a de jure, or de facto basis.(The latter will be discussed further in the ONE alliancecase below.) Yet, the level of clarity and specificity in locallaws, regulations, and guidance on this question varies sig-nificantly from jurisdiction to jurisdiction. As a result, it isoften difficult to determine the jurisdictional issue with cer-tainty because there may not be sufficient clarity on howcontrol is interpreted in the local jurisdiction. There are alsocertain “quirks” on this highly technical issue; for example,in Japan, if a joint venture is 50-50 owned by two parentcompanies, each with equal board representation and deci-sion-making power over strategic issues, then neither of theparents would be regarded as having control over the jointventure.

Minority Interest. Some jurisdictional tests refer to aspecified percentage threshold of voting rights or equity inter-

est. As such, an acquisition of a minority interest (even in theabsence of strategic veto rights) may be sufficient to triggera merger filing requirement. This issue may catch companiesby surprise, especially because a few Asian jurisdictions pre-scribe a very low threshold. To name a few examples, the rel-evant threshold of voting rights is 20 percent in Japan andSouth Korea,18 33 percent in Taiwan, and 35 percent in thePhilippines.

Separate Step-by-Step Analysis. It is worth bearing inmind that, even though a joint venture may constitute aconcentration, the transaction is not necessarily analyzed asan integral whole. In Japan, for instance, separate analysis ofeach step of the joint venture transaction may be required toascertain whether the jurisdictional thresholds are met, tak-ing each step on its own, even if the steps occur almost simul-taneously in practice. By way of illustration, if a joint venturetransaction involves two steps: (1) the creation of the jointventure; and (2) the injection of assets into the joint venture,Japanese merger control rules require separate analysis (by ref-erence to the relevant turnover/asset tests) for: (1) the acqui-sition by the parents of shares in the joint venture entity; and(2) the acquisition of assets by the joint venture entity.

P3 Network. The disparity in treatment of joint ven-tures may have a significant impact from a practical per-spective. The recent shipping alliances in the liner shippingindustry provide good examples for illustrating the differentoutcomes that may emerge due to differences in the juris-dictional treatment of joint ventures. Three years ago, Denmark’s A.P. Møller-Maersk A/S,

Swiss firm MSC Mediterranean Shipping Company S.A.,and France’s CMA CGM S.A. proposed to establish a glob-al joint venture called P3 Network alliance.19 By way of back-ground, the P3 Network would have enabled members of thealliance to share vessels and coordinate their operating activ-ities on major shipping routes by establishing a network cen-tre for the joint coordination and management of the P3Network. Under the alliance, each member would retain itsseparate identity with independent sales, pricing and marketfunctions.The P3 Network was considered in Europe, the United

States, and China but it received vastly different treatmentpursuant to different legal frameworks in these jurisdictions.In Europe, the alliance members voluntarily reported the P3Network to the EC but not as a merger, as it was not a full-function joint venture.20 The parties did not receive a clearance“decision,” although on June 4, 2014, the EC informed thealliance members that it did not reach any substantive deci-sion on the alliance and simply opted not to investigate,expressly stating that it might decide to investigate in thefuture.21 In the United States, the Federal Maritime Commis -sion (FMC) (rather than the Federal Trade Commission or theDepartment of Justice) reviewed the P3 Network as a vessel-sharing agreement and granted its approval in March 2014.22

In China, the P3 Network was subject to a PRC merger fil-ing requirement because the Anti-Monopoly Law does not

distinguish between full-function and non-full-function jointventures. It was therefore reviewed as a merger by the Anti-Monopoly Bureau of the Ministry of Com merce (MOF-COM).23 In considering its analysis, MOF COM distin-guished the P3 Network from the more loose-knit alliancescommon in the shipping industry, noting that the operationof the P3 Network through a dedicated network centre meantthat the alliance would give rise to a “tight joint operation.”24

MOFCOM blocked the transaction on June 17, 2014, due tosubstantive competition concerns.The different approaches to joint ventures had two impor-

tant consequences. First, MOFCOM’s approach to lookingat the competition issues was fundamentally different to thenon-merger analysis applied by the EC and the FMC. In itssubstantive analysis, MOFCOM considered the combinedmarket shares of the three members of the P3 Network,whereas the FMC considered that the members continued toact as independent operators in the market. In addition,MOFCOM was under an obligation to investigate and issuea formal decision on the transaction, whereas the EC had theoption of taking a “wait and see” approach.One month following the prohibition by MOFCOM,

Maersk and MSC announced a 10-year vessel sharing agree-ment (2M Agreement) on July 10, 2014. The 2M Agreementreceived the FMC’s approval in October 2014.25 As a coop-eration agreement (short of establishment of a joint ven-ture), it was understood that this revised transaction did notrequire a merger filing with MOFCOM.

Ocean Network Express.The P3 Network case might becontrasted with a more recent joint venture between the threeJapanese lines, Kawasaki Kisen Kaisha, Ltd, Mitsui O.S.K.Lines, Ltd, and Nippon Yusen Kaisha, and again illustratesthat jurisdictional analysis of joint ventures continues to besubject to debate. They entered into a tripartite agreement onMay 8, 2017, to create a “fully integrated joint venture” underthe name of Ocean Network Express (ONE) whereby theirpreviously separate ocean carrier containership operationswould be combined into one new common carrier.26

It is interesting to observe the different approaches takenin China, Europe, and the United States. In China, the jointventure was reviewed as a merger and received uncondition-al approval from MOFCOM on June 7, 2017.27 In Europe,the EC accepted that the ONE joint venture was a concen-tration, exceptionally, on the basis that joint control occurredon a de facto basis because there appeared to be “a strongcommonality of interests between the Parties that will preventthem from acting against each other when deciding on strate-gic matters for the management of the joint venture” andgranted unconditional approval under EUMR on June 28,2017.28 The FMC voted to reject the agreement on jurisdic-tional grounds on May 5, 2017, stating that the parties were“ultimately establishing a merged, new business entity andthat action is among the type of agreements excluded fromFMC review [under the Shipping Act of 1984].”29 This mightbe understood to be pointing the parties to notify the trans-

A R T I C L E S

9 2 · A N T I T R U S T

S P R I N G 2 0 1 8 · 9 3

action to the U.S. antitrust agencies responsible for mergerreview instead.

ConclusionJoint ventures raise interesting jurisdictional questions ininternational merger control. In Europe, the concept of full-functionality has recently been affirmed by the ECJ in theAustria Asphalt judgment to be central to the EUMR regime,and the judgment has helpfully provided some legal clarity onthe textual ambiguities in the interpretation of EUMR. Looking outside Europe, the analysis also requires careful

application of national merger control rules to the specificstructure of the joint venture transaction, based on soundknowledge of the different analytical frameworks and nuancesin key concepts, such as the types of notifiable transactions,definition of control, and treatment of minority interests,adopted in different jurisdictions.�

1 Council Regulation No.139/2004 (on the control of concentrations betweenundertakings), 2004 O.J. (L 24) 1.

2 In brief, a full-function joint venture must have sufficient resources to oper-ate on a lasting basis its business activities independently on a market(e.g., with a management dedicated to its day-to-day operations and accessto sufficient resources) and have its own access to or presence on the mar-ket. For a detailed discussion on the concept of “full-function joint venture,”see Commission Consolidated Jurisdictional Notice Under CouncilRegulation (EC) No 139/2004 (on the control of concentrations betweenundertakings) 2008 O.J. (C 95) 1, ¶¶ 91–109 [hereinafter JurisdictionalNotice].

3 Case C-248/16, Austria Asphalt GmbH & Co OG v. BundeskartellanwaltECLI:EU:C:2017:643 (ECJ Sept. 7, 2017), http://curia.europa.eu/juris/document/document.jsf?text=&docid=194102&pageIndex=0&doclang=EN&mode=lst&dir=&occ=first&part=1&cid=214007.

4 Opinion of Advocate General Kokott in Case C-248/16, Austria AsphaltGmbH & Co OG v. Bundeskartellanwalt, ECLI:EU:C:2017:322, ¶ 22 (Apr. 27,2017), http://curia.europa.eu/juris/document/document.jsf?text=&docid=190185&pageIndex=0&doclang=EN&mode=lst&dir=&occ=first&part=1&cid=214007.

5 Id. ¶ 21.6 Id.7 Case C-248/16, Austria-Asphalt, supra note 3, ¶ 36. 8 Id. ¶ 22.9 Id. ¶ 27.

10 Jurisdictional Notice, supra note 2, ¶¶ 85–88. 11 Case C-248/16, Austria-Asphalt, supra note 3, ¶ 25. 12 Opinion of Advocate General Kokott in Case C-248/16, supra note 4, ¶ 33.13 Id. ¶ 4.14 The EU member states that do not require full-functionality include: Austria,

Estonia, Germany, Latvia, Poland, and the United Kingdom.15 The EU member states that are modeled on the EUMR include Belgium,

Cyprus, Denmark, Finland, France, Hungary, Ireland, Lithuania, theNetherlands, Portugal, Slovenia, Spain, and Sweden.

16 Opinion of Advocate General Kokott in Case C-248/16, supra note 4, ¶ 39.17 Examples include Japan, South Korea, Taiwan, India, Indonesia, Australia,

and New Zealand.18 The relevant threshold is 15 percent for companies publicly listed in South

Korea.19 https://www.maersk.com/press/press-release-archive/2013/20130618-

maersk-line-msc-and-cma-cgm-to-establish-an-operational-alliance.

20 https://www.cma-cgm.com/news/473/cma-cgm-pleased-with-european-commission-affirmation.

21 Id.; see also https://www.maersk.com/press/press-release-archive/2014/20140617-the-p3-network-will-not-be-implemented-following-decision-by-the-ministry-of-commerce.

22 News Release, U.S. Fed. Maritime Comm’n, P3 Agreement Clears FMCRegulatory Review (Mar. 20, 2014), https://www.fmc.gov/NR14-06/.

23 商务部公告2014年第46号商务部关于禁止马士基、地中海航运、达飞设立网络中心经营者集中反垄断审查决定的公告 [Announcement No. 46 of2014, Ministry of Commerce of the People’s Republic of China, Decision ofAnti-monopoly Review to Prohibit Concentration of Undertakings by Pro -hibiting Maersk, MSC and CMA CGM from Establishing a Network Center](June 17, 2014), http://fldj.mofcom.gov.cn/article/ztxx/201406/20140600628586.shtml

24 Id.25 News Release, U.S. Fed. Maritime Comm’n, 2M Agreement Clears FMC

Regulatory Review (Oct. 9, 2014), https://www.fmc.gov/NR14-12/?CategoryId=1&Month=10&Year=2014&Archive=y.

26 U.S. Fed. Maritime Comm’n, Tripartite Agreement, FMC Agreement No.012475 (Mar. 24, 2017) https://www2.fmc.gov/FMC.Agreements.Web/Public/Document/5528.

27 Table of Unconditional Clearance Decisions published by MOFCOM for thesecond quarter of 2017.

28 Case M.8472—Nippon Yusen Kabushiki Kaisha/Mitsui Osk Lines/Kawasaki Kisen Kaisha & JV, Comm’n Decision (June 28, 2017) (Summaryat 2017 O.J. (C 401) 17), http://ec.europa.eu/competition/mergers/cases/decisions/m8472_128_3.pdf.

29 News Release, U.S. Fed. Maritime Comm’n, Update Regarding FMC Rejec -tion of Tripartite Agreement (May 5, 2017), https://www.fmc.gov/NR17-10/?CategoryId=1&pg=5.

Product Code: 5030642Publication Date: 2017Page Count: 13Trim Size: 4x8.5Format: Spiral boundPrice: $15.00 List Price /$12.00 ABA Member /$10.00 AT SectionMember

� THE CARTEL LAW BASICS book is a practicalresource in “flip-book” form for business people to aidthem in avoiding engaging in cartel behavior in viola-tion of the antitrust laws. This book is available in attractive bulk pricing for distribution to clients.


Cartel Law Basics for Executives

A R T I C L E S

9 4 · A N T I T R U S T

Are Disgorgement’sDays Numbered?Kokesh v. SECMay ForeshadowCurtailment of theFTC’s Authority toObtain Monetary ReliefB Y M . S E A N R O Y A L L , R I C H A R D H . C U N N I N G H A M ,

A N D A S H L E Y R O G E R S

THE FEDERAL TRADE COMMISSION’SBureau of Consumer Protection recovers big dol-lars. Some recent examples: Vemma NutritionCompany agreed to pay $238 million to settlecharges that the multi-level marketing company

operated an illegal pyramid scheme; TracFone, the nation’slargest prepaid mobile provider, paid $40 million to settledeceptive advertising charges; the marketers of Sensa settledcharges of misleading weight-loss claims for $26.5 million;and Google agreed to pay at least $19 million to resolvecharges relating to the in-app purchase feature in the GooglePlay Store.1

From the FTC’s standpoint, there is nothing controversialabout the agency seeking monetary relief from individualsand companies alleged to have engaged in deceptive or unfairpractices that violate Section 5 of the FTC Act. Indeed, theFTC in December 2017 boasted that the Bureau of Con -sumer Protection obtained 168 court orders totaling over$12.72 billion between July 1, 2016 and June 30, 2017,resulting in more than $6.4 billion in consumer refunds.2 The

FTC obtains a substantial portion of these monetary reme-dies pursuant to Section 13(b) of the FTC Act—a provisionthat does not mention monetary remedies and speaks only interms of “injunctions.” Federal circuit courts have, never-theless, construed Section 13(b) to allow the FTC to obtainmonetary relief on the theory that Congress’s use of the word“injunctions” permits the FTC to seek a full panoply of equi-table relief, including disgorgement and restitution.In an article published in the Fall 2016 edition of ANTI -

TRUST, David C. Vladeck, a former Director of the FTC’sBureau of Consumer Protection and a current member of thefaculty at Georgetown University Law Center, summarizedthe precedents in this area and concluded that the argumentthat the FTC lacks authority to obtain monetary relief pur-suant to Section 13(b) “has been repeatedly and uniformlyrejected by every court to address it.”3 Professor Vladeckthen predicted that this “is not going to change” because“Section 13(b) of the FTC Act authorizes courts to grantinjunctions, and that grant of authority empowers courts toorder the full range of equitable remedies, including restitu-tion and disgorgement.”4

The Supreme Court, however, has never approved theFTC’s monetary relief authority under Section 13(b) and, lastsummer, signaled in Kokesh v. SEC that significant limits onthis authority may be on the horizon. In Kokesh, the SupremeCourt held that disgorgement obtained by the Securities andExchange Commission is a “penalty” subject to a five-yearstatute of limitations under 28 U.S.C. § 2462 when orderedpursuant to provisions similar to Section 13(b).5 WhileKokesh’s holding is confined to the SEC and a statute of lim-itations question, the Supreme Court’s reasoning and analy-sis suggest potential implications for the FTC.First, with regard to matters in which the FTC seeks dis-

gorgement, the agency likely will be subject to the five-yearstatute of limitations in 28 U.S.C. § 2462. Each of the “hall-marks of a penalty” the Supreme Court identified in Kokeshapplies equally to the FTC’s use of disgorgement. Thischange alone may significantly decrease the size of the FTC’sdisgorgement awards, as the agency has frequently gone backmuch further than five years in its remedy calculations.6

Second, and perhaps more importantly, the SupremeCourt signaled that federal agencies may not be able to obtaindisgorgement or other remedies with similar characteristics asequitable relief ancillary to an injunction at all. In a footnotein the Kokesh opinion, the Court stated that it was not opin-ing on “whether courts possess authority to order disgorge-ment in SEC enforcement proceedings” or whether dis-gorgement has been “properly applied.”7 At oral argument,several Justices posed questions suggesting discomfort withthe SEC’s current disgorgement practices. For example,Justice Kagan asked if the SEC has “ever set down in writingwhat the guidelines are for how the SEC is going to use dis-gorgement and what’s going to happen to the monies col-lected?”8 In addition, the fact that the Supreme Courtdeemed the SEC’s disgorgement remedy to be a “penalty”

M. Sean Royall and Richard H. Cunningham are partners, and Ashley

Rogers is an associate, with Gibson, Dunn & Crutcher LLP. Mr. Royall

served as Deputy Director of the Bureau of Competition of the Federal

Trade Commission from 2001 through 2003 and served as appellate

counsel for the defendant in FTC v. Commerce Planet, Inc., 815 F.3d 593

(9th Cir. 2016). Mr. Cunningham served as Staff Attorney and Senior Trial

Counsel with the Bureau of Competition from 2004 to 2013. The authors

thank Natalie J. Hausknecht for her valuable contributions to this article.

S P R I N G 2 0 1 8 · 9 5

To obtain a monetary remedy that reached back further intime, the SEC invoked its authority to seek disgorgement asrelief ancillary to an injunction issued pursuant to theSecurities Exchange Act, the Investment Advisers Act, and theInvestment Company Act.13 This authority had been widelyaccepted among the lower courts, even after Congress in1990 gave the SEC the power to seek statutory civil finesinstead.14 Citing lower court decisions, the SEC contendedthat its disgorgement authority was not subject to 28 U.S.C.§ 2462’s statute of limitations because disgorgement is anequitable, “remedial” monetary sanction, rather than a“penalty.”15 The district court concurred and ordered Mr.Kokesh to pay $34.9 million in disgorgement and another$18.1 million in prejudgment interest on top of his civilfine. The Tenth Circuit likewise found “the reasons for th[at]view are clear” and affirmed.16

In a unanimous opinion penned by Justice Sotomayor, theSupreme Court reversed, holding that SEC disgorgementconstitutes a penalty subject to the five-year statute of limi-tations found in Section 2482.17 To reach this conclusion, theCourt began by defining a “penalty” as a “punishment,whether corporal or pecuniary, imposed and enforced by theState, for a crime or offen[s]e against its laws.”18 From thisdefinition, the Court extrapolated three “hallmarks” of apenalty: (1) it is imposed to redress the violation of a publiclaw; (2) it is sought for a punitive purpose such as deterrence;and (3) it is not intended solely to compensate a victim forhis loss.19

Applying these factors, the Court determined that SECdisgorgement was a penalty subject to Section 2482. First, theCourt noted that lower courts imposed SEC disgorgement asa consequence for violating public laws—i.e., a violationcommitted against the United States rather than an aggrievedindividual.20 A securities enforcement action may proceed,the Court reasoned, even if the victims do not endorse it.21

Second, the Court concluded that disgorgement in the SECcontext was imposed for the “inherently punitive” purpose ofdeterring of future public law violations.22 Third, the Courtrecognized that “in many cases,” SEC disgorgement was notcompensatory because lower courts did not require themoney actually be paid to victims.23 The Court summarizedits decision by stating that SEC disgorgement bore “all thehallmarks of a penalty: It is imposed as a consequence of vio-lating a public law, and it is intended to deter, not to com-pensate.”24

The Potential Implications of Kokesh for the FTCKokesh establishes a general definition of what constitutes a“penalty” subject to Section 2462’s five-year limitations peri-od. This definition is unlikely to be confined to the SEC andlogically applies to the FTC’s authority to seek monetaryrelief under Section 13(b). The FTC’s use of disgorgement meets each of the Kokesh

criteria for a “penalty.” The FTC—like the SEC—is notrequired to stand in the shoes of a particular victim when it

suggests that disgorgement may not be available as reliefancillary to an injunction in light of the fact that the Courthas previously said that a court in equity may not enforce a“civil penalty.”9 Because the FTC relies on the same legalrationale as the SEC for its authority to obtain disgorge-ment, Kokesh raises the possibility of future Supreme Courtdecisions reaching the issues of whether and to what extentthe FTC may obtain disgorgement or other monetary relief.Targets of FTC enforcement actions are already arguing

that Kokesh limits the FTC’s ability to obtain monetary reliefin Section 13(b) actions. The agency’s response in the fiveinstances in which Kokesh-related arguments have arisen todate has been to deny that Kokesh applies to the FTC.Notwithstanding this position, the agency has also simulta-neously excluded from its remedy demands any funds asso-ciated with conduct outside of Section 2462’s five-yearstatute of limitations. The FTC has also described its primaryremedy demand as restitution—which focuses on harm tovictims caused by the unlawful conduct—rather than dis-gorgement—which focuses on depriving the wrongdoer ofill-gotten gains, and/or has characterized the purpose of itsmonetary remedy request as victim compensation. Whetherrestitution in the context of a Section 13(b) enforcementaction is substantively different than disgorgement in practiceand does not constitute a “penalty” under Kokesh is an openquestion that undoubtedly will be litigated vigorously as thisarea of case law evolves. The full impact of Kokesh on the FTC’s authority under

Section 13(b) remains to be seen. At a minimum, however,Kokesh provides a sound basis to question the view expressedby Professor Vladeck and others that challenges to the scopeof the FTC’s authority to obtain monetary relief under 13(b)are futile and that “courts are increasingly showing impa-tience with these long discredited arguments.”10

Kokesh Limits the SEC’s Disgorgement AuthorityThe facts at issue in Kokesh are straightforward. The defen-dant, Charles Kokesh, owned two investment-adviser firmsthat provided advice to business-development companies. In2009, the SEC brought charges against Mr. Kokesh, claim-ing he had misappropriated $34.9 million from four of theseclients and filed false reports with the Commission to coverthis up for 14 years. The jury agreed. The remedies phase is where Kokesh got interesting. In

2013, the Supreme Court had held in Gabelli v. SEC that thegeneral five-year statute of limitations in 28 U.S.C. § 2462applied when the SEC sought statutory monetary penal-ties.11 The district court in Kokesh recognized that any mon-etary “penalty” against Mr. Kokesh therefore had to be basedon his misconduct during the last five years of his scheme.The problem for the SEC was that Mr. Kokesh had taken thevast majority of the $34.9 million involved—$29.9 millionto be exact—outside this limitations period. As a result, thedistrict court concluded it could impose a civil fine of only$2.35 million for his unlawful behavior.12

seeks disgorgement, nor is it seeking to vindicate the rights ofany particular victims.25 Lower courts have likewise held thatthe FTC, like the SEC, is not required to pay disgorgementto victims, and the Commission has maintained that it neednot do so.26 Additionally, as with SEC disgorgement, courtshave held the primary purpose of FTC disgorgement is deter-rence of future public law violations.27 For these reasons, itappears very likely that FTC disgorgement will be limited bythe five-year limitations period found in Section 2462.Indeed, the government acknowledged as much in its brief-ing in Kokesh. Specifically, the Solicitor General describedthe FTC as among the agencies whose authority to obtainmonetary remedies would be curtailed by construing Section2462 as applicable to the SEC’s disgorgement authority.28

The more interesting question is whether Kokesh portendsadditional limits on the FTC’s authority to obtain monetaryrelief.There are good reasons to believe that it does. As noted above, the Supreme Court in Kokesh stated in a

footnote that its decision should not be read to opine onwhether courts actually possess authority to order disgorge-ment in SEC enforcement proceedings or “whether courtshave properly applied disgorgement principles in this con-text.”29 This footnote suggests the Court may not considereither of these issues to be settled. Justice Gorsuch also reminded the SEC at oral argument

that the Court had never given its approval to the 50 years oflower court precedent holding a court could order disgorge-ment based on its inherent equitable authority ancillary to aninjunction.30 Justice Gorsuch further noted that the difficul-ty in defining reasonable limits for such a remedy mightstem from the fact that “there’s no statute governing it. We’rejust making it up.”31

Justice Gorsuch was not alone among his colleagues instruggling to identify the source (and consequential limits) of SEC disgorgement. Chief Justice Roberts, too, signaleddiscomfort with the fact that Congress had never specifieddisgorgement or another monetary remedy in the text of therelevant provisions.32 And the Chief Justice was joined byboth Justice Kennedy and Justice Sotomayor in pressing theparties to identify a specific statutory authority for disgorge-ment.33 Justice Kennedy, in particular, explained that he“understood in cases where the aggrieved party is before theCourt, there can be equitable remedies under state law,” buthe queried whether “[i]t is clear that the district court hasstatutory authority” to obtain disgorgement otherwise.34 Jus -tice Sotomayor wondered how, if the money was not beingreturned to the harmed individuals as restitution, a statuto-ry grant to seek equitable relief could serve as the basis for dis-gorgement.35

Kokesh’s holding that disgorgement is not “remedial,” butrather is a “penalty,” suggests that the Court may questionwhether disgorgement is an appropriate equitable powerwhen a statute merely authorizes injunctions. InTull v. UnitedStates, the Court explained that “while a court in equity mayaward monetary restitution as an adjunct to injunctive relief,

it may not enforce civil penalties.”36 Hence, if disgorgementis a “penalty,” then presumably it cannot be a valid equitableremedy under Tull. The entire final section of the Kokesh decision focuses on

the differences between disgorgement, as applied in the lowercourts, and the traditional characteristics of an equitableremedy. The decision notes, in particular, that SEC dis-gorgement is not “remedial” in large part because it does notsimply restore the status quo by returning the defendant tothe place he would have occupied had he not broken thelaw.37 Rather, SEC disgorgement “sometimes exceed[s] theprofits gained as a result of the violation” either by forcing awrongdoer to disgorge gains that accrued to third parties orby failing to account for expenses that reduced the amountof illegal profit.38 The Court cited the Restatement (Third) ofRestitution and Unjust Enrichment for the proposition thatsuch sanctions were punitive ones “that the law of restitutionnormally attempts to avoid.”39

Viewed collectively, these are fairly strong signals that the Court might be uncomfortable classifying an agency’srequested monetary remedy as equitable relief ancillary to itsinjunctive authority when the requested remedy has effectsbeyond merely addressing the harm caused by the defen-dant’s conduct.

The FTC’s Initial Response to KokeshUnsurprisingly, within a matter of days, arguments based onKokesh began surfacing in ongoing FTC enforcement actions.To date, defendants in at least five FTC matters have raisedKokesh-related arguments—FTC v. DirecTV LLC, FTC v. J.William Enterprises, LLC, FTC v. Credit Bureau Center, LLC,FTC v. Publishers Business Services, Inc., and FTC v. AMGCapital Management, LLC.40

FTC v. DirecTV. Fifteen days after Kokesh was decided,DirecTV filed a motion for leave to file an amended answerto add affirmative defenses based on Kokesh asserting that (1) the FTC is not entitled to seek restitution; and (2) evenif it is, the request is subject to a five-year statute of limita-tions.41

In response, the FTC argued that DirecTV’s request wasuntimely. The FTC further argued that Kokesh does not applyto an FTC action seeking restitution because the purpose ofrestitution is compensation rather than punishment, andtherefore, restitution is not a “penalty” under Kokesh.42 As afallback, the FTC also maintained that even if Kokesh appliesto an action seeking restitution, it did not apply in this casebecause the agency’s complaint sought monetary relief “nec-essary to redress injury to consumers” that would be used to“provide redress to eligible consumers,” in contrast with theKokesh complaint, which sought “to disgorge an amountequal to the funds and benefits obtained illegally.”43 Notably,the FTC stated that its proposed order “recognizes the fun-damental distinction between restitution and disgorgementby providing primarily for restitution while allowing for dis-gorgement of any residual funds for which distribution to

A R T I C L E S

9 6 · A N T I T R U S T

S P R I N G 2 0 1 8 · 9 7

consumers is not feasible.” And the agency all but concededthat a five-year statute of limitations applies to disgorgementby representing to the court that it intended to eliminate dis-gorgement from its proposed order, at least as to any “puta-tively time-barred claims.”44

In August 2017, the district court denied DirecTV’smotion on the basis that DirecTV was not diligent in assert-ing these affirmative defenses, expressly declining to “applyKokesh to make broad generalizations.”45 The court did, how-ever, state in dicta that “[a] fair reading of the first line of foot-note 3 in Kokesh does not support the argument that the FTCis barred from seeking restitution” because “the Court explic-itly declined to make any finding whatsoever, much less onerelevant to whether the FTC has authority to seek restitu-tion.”46

FTC v. J. William Enterprises. In October 2017, defen-dants filed a partial motion for summary judgment based onKokesh, asserting that (1) the Justices’ comments during oralargument and the “ominous footnote” in the Kokesh decision“cast[ ] considerable doubt” on courts’ authority to orderrestitution, rescission, refunds, or disgorgement in FTCenforcement actions; and (2) the three-year statute of limi-tations in Section 19 of the FTC act applies.47 In response,the FTC maintained its position that Kokesh “involve[d] anentirely different and distinct statutory scheme” and “didnot change the law . . . regarding the scope of remedies underthe FTC Act.” The agency also asserted that Section 19 of theFTC Act “has no bearing” on relief the FTC seeks underSection 13(b). But the agency again also indicated that it waswilling to apply a five-year limit on its claims in order to“effectively render[] a ruling that Kokesh applies moot.”48

The court accepted the FTC’s arguments and held thatKokesh “provides no basis for this Court to disregard decadesof precedent” permitting the equitable remedies sought bythe FTC and contains “nothing . . . that indicates that theCourt should apply section 19(b)’s statute of limitations tothe FTC’s claims under section 13(b).”49

FTC v. Credit Bureau Center, LLC. In November2017, defendants filed a motion to modify a preliminaryinjunction entered against them based on Kokesh. Defen -dants argued that terms in the preliminary injunction con-tinuing an asset freeze and appointing a receiver were improp-er because they were intended to hold assets for disgorgementand restitution. As such, the “principles” set forth in Kokesh,they contended, confirm that disgorgement and restitutionare “penalties” that the FTC is not authorized to seek underSection 13(b).50

In response, the FTC again asserted that Kokesh “did notalter, let alone reverse” the scope of remedies available underthe FTC Act, but instead narrowly held that “a five-yearstatute of limitations applies when the SEC seeks punitivedisgorgement under the securities laws” and “expressly dis-claimed any broader application.”51 The FTC also cited theDirecTV and J. William Enterprises matters discussed above,stating that “the only two courts to have considered the issue

have similarly concluded that Kokesh does not disturb estab-lished precedent under Section 13(b).”52 Nevertheless, theFTC again emphasized that “compensating victims is theprimary goal” in the agency’s case because the agency askedfor relief “necessary to redress injury to consumers” and pro-posed to measure consumer recovery as “the full amount ofconsumer loss.”53 And again, the FTC maintained that evenif Kokesh applies to FTC actions, “its only possible effectwould be to impose a five-year statute of limitations on anaward of disgorgement to the Treasury to the extent that thedisgorgement amount exceeded defendants’ gains,” whichwould not affect the case because the alleged misconductdates back only to 2014.54

In January 2018, the district court denied the defendants’motion, stating that the defendants’ interpretation of Kokeshwas a “considerable overstatement” because the decision saidnothing about “whether disgorgement and restitution were, asdefendants claim, ‘authorized’ under the securities statute.”55

The court held that controlling Seventh Circuit law “specifi-cally authorizes disgorgement and restitution in FTC suits,”that it saw nothing in the “principles of Kokesh underminingthese decisions,” and that it is “rather reckless to contend” thatJustices’ comments during oral argument in Kokesh “renderexisting precedent infirm.”56

FTC v. Publishers Business Services. In this pendingNinth Circuit appeal, Publishers Business Services contests amonetary judgment of nearly $24 million, arguing (amongother things)57 that Kokesh makes clear that FTC disgorge-ment is a penalty that is outside the scope of permissibleequitable relief under Section 13(b).58

In response, the FTC reiterated the position taken by theagency in both DirecTV and J. William Enterprises: Kokeshwas a narrow decision that left undisturbed “decades of con-sistent law on the availability of equitable remedies under theFTC Act.”59 Yet again, however, the FTC was careful to statein its briefing that even if Kokesh did apply, it has no “prac-tical application” to the present case because the entire judg-ment seeks “equitable monetary relief for consumers” thatfalls within Section 2462’s five-year statute of limitations.60

FTC v. AMG Capital Management. In this pendingNinth Circuit appeal, the appellants filed a brief approxi-mately one month after Kokesh was decided, citing to theSupreme Court’s decision as a basis for challenging a $1.3 billion monetary judgment.61 The appellants argued thatKokesh “mandates the application of a statute of limitationsto FTC 13(b) enforcement actions” and that the Truth inLending Act’s one-year statute of limitations applies to thecase or, in the alternative, that the three-year statute of limi-tations in Section 19 of the FTC Act applies. They also urgedthe Ninth Circuit to reconsider “[t]he availability of mone-tary awards under Section 13(b)—regardless whether a courtlabels them ‘damages’ or ‘equitable restitution,’” and to applythe Kokesh factors to “determine whether the FTC improp-erly uses Section 13(b) to pursue penal monetary relief underthe guise of equitable authority.”62

In response, the FTC again characterized Kokesh as a nar-row decision limited to disgorgement judgments “with thepurpose of punishment and deterrence” under the SecuritiesExchange Act.63 At the same time, the agency took pains tonote that the judgment in the present case was intended tobe used to “compensat[e] a victim for his loss” and that theNinth Circuit “need not reach the issue” of whether Kokeshapplies to FTC “requests for monetary relief” because themonetary judgment did not extend beyond five years.64

While these matters are likely just a taste of the Kokesh-related disputes the FTC can expect to encounter in thefuture, several recurrent themes seem to be emerging from theCommission’s advocacy. First, the FTC is not explicitly con-ceding that Kokesh applies to it. Given that the government’sbriefing in Kokesh argued that the ruling would apply to theFTC, it will be interesting to see if the FTC will maintain thisposition in future Kokesh-related disputes. Second, the FTCis largely abandoning efforts to seek disgorgement outside ofSection 2462’s five-year statute of limitations, and may there-by minimize the likelihood of court rulings applying the coreholding of Kokesh to the agency. Finally, the FTC appears tobe emphasizing that the agency is primarily seeking restitu-tion, as opposed to disgorgement, and characterizing thepurpose of its monetary remedies as victim compensation—for example, by stating that its monetary relief demand is“necessary to redress injury to consumers” (DirecTV andCredit Bureau Center), constitutes “equitable monetary relieffor consumers” (Publishers Business Services), and will beused to “compensat[e] a victim for his loss” (AMG CapitalManagement and Credit Bureau Center).65 This presumablyreflects an effort by the agency to distance its monetary reme-dies from the SEC disgorgement remedy directly at issue inKokesh.

Can FTC Restitution Constitute a Penalty?The FTC’s apparent effort to emphasize that its primarymonetary remedy request is in the form of restitution andthat it is seeking funds for victim compensation raises thequestion whether FTC restitution can be classified as a“penalty” under Kokesh. For the reasons outlined above, thiswould call into question whether the FTC can obtain suchrelief beyond the five-year statute of limitations specified by28 U.S.C. § 2462 and could lead to a finding that the FTCmay not obtain such relief ancillary to an injunction underSection 13(b). The FTC has previously described restitution and dis-

gorgement as “related” equitable remedies that serve dis-tinct purposes.66 According to the agency, disgorgement is“designed to deprive a wrongdoer of his unjust enrichmentand to deter others from future violations,” and restitutionis “intended to restore the victims of a violation to the posi-tion they would have been in without the violation, often byrefunding overpayments made as a result of the violation.”67

Although the FTC’s description of restitution emphasizesvictim compensation and makes no mention of deterrence,

the Kokesh decision suggests that the label attached to themonetary remedy is irrelevant—what matters is the remedy’sobjective, how it was calculated, and how it is used. Applying the Kokesh factors to FTC restitution does not

yield a clear answer. As to the first factor, which focuses onwhether the case proceeds for law enforcement purposes,there is little the FTC can do to avoid the conclusion that thecases in which the agency seeks restitution may proceed “evenif victims do not support or are not parties to the prosecu-tion,” and thus redress the violation of a public law underKokesh.68 Indeed, this is true of all FTC actions under Section13(b).Application of the other two Kokesh factors to FTC resti-

tution is less clear-cut. As to the second factor—the degree towhich the remedy is “punitive” and “deterrent” in nature—as noted above, the FTC describes restitution as focused onvictim compensation, a purpose that is, generally speaking,remedial rather than punitive or deterrent.69 In practice,however, FTC restitution has been calculated in a mannerthat has resulted in defendants being liable for amounts thatsignificantly exceed the alleged harm to victims caused by theoffending conduct. This occurs because some courts havepermitted the FTC to base restitution amounts on roughapproximations of consumer harm70 that may fail to accountfor the fact that some purchasers of the product at issue (a) received significant value, and/or (b) were unaffected bythe offending conduct because they did not see the deceptiveadvertisement or experience the unfair conduct.71 If a resti-tution amount exceeds the amount of consumer harm causedby the conduct, the remedy arguably imposes costs on defen-dants that are punitive and serve to deter future violations.Thus, because the restitution amounts the FTC has soughtand obtained have, at least in some cases, not been closelylinked to the consumer harm actually caused by the defen-dant’s conduct, the second Kokesh factor arguably supportsthe conclusion that the remedy is a penalty.The third Kokesh factor turns on whether the recovered

funds are actually paid to victims as compensation. The FTChas stated that, where it is feasible to do so, it endeavors toremit funds obtained as restitution to victimized consumers.72

However, the FTC has not always accomplished this goaland, in some cases, the agency has sought instead to at leasthave the discretion to direct restitutionary funds to the U.S.Treasury, given difficulties in successfully remitting funds toaffected consumers. To the extent such funds are not returnedto consumers, one might think of this component of a mon-etary remedy as equitable disgorgement, not restitution.73

Kokesh can be read to suggest that the relevant question whenanalyzing this factor is whether all of the funds collected goto the victims, as opposed to the government. Hence, becausethe FTC does not consistently remit all of the funds obtainedas restitution to consumers, this may cut in favor of con-cluding that restitution obtained by the FTC does indeedfunction as a penalty.

A R T I C L E S

9 8 · A N T I T R U S T

S P R I N G 2 0 1 8 · 9 9

ConclusionWhat ultimate implications Kokesh may have for the FTCremains unclear. What is clear is that the agency is alreadyconfronting arguments that Kokesh significantly limits itsability to obtain monetary relief in Section 13(b) matters.While no court to date has issued a decision applying the rea-soning of Kokesh to the FTC, there are strong argumentsthat FTC disgorgement is a “penalty” subject to the statuteof limitations in 28 U.S.C. § 2462, and that this remedy maybe entirely unavailable as relief ancillary to the Commission’sstatutory injunctive authority. The agency does appear to be shifting its approach to

equitable monetary remedies to avoid requiring a court tosquarely decide these issues, including by imposing limita-tions on how far back in time its disgorgement calculationsreach, and characterizing its monetary relief as restitutiondesigned to facilitate victim compensation. Whatever courtsconclude after delving more deeply into these issues, Kokeshalready makes clear that the scope of the FTC’s authority toobtain monetary relief under Section 13(b) is less settledthan some previous commentators have suggested.�

1 Press Release, Fed. Trade Comm’n, Vemma Agrees to Ban on PyramidScheme Practices to Settle FTC Charges (Dec. 15, 2016), https://www.ftc.gov/news-events/press-releases/2016/12/vemma-agrees-ban-pyramid-scheme-practices-settle-ftc-charges; Press Release, Fed. Trade Comm’n,Prepaid Mobile Provider TracFone to Pay $40 Million to Settle FTC ChargesIt Deceived Consumers About ‘Unlimited’ Data Plans (Jan. 28, 2015),https://www.ftc.gov/news-events/press-releases/2015/01/prepaid-mobile-provider-tracfone-pay-40-million-settle-ftc; Press Release, Fed. TradeComm’n, Sensa and Three Other Marketers of Fad Weight-Loss ProductsSettle FTC Charges in Crackdown on Deceptive Advertising (Jan. 7, 2014),https://www.ftc.gov/news-events/press-releases/2014/01/sensa-three-other-marketers-fad-weight-loss-products-settle-ftc; Press Release, Fed.Trade Comm’n, FTC Approves Final Order in Case About Google Billing forKids’ In-App Charges Without Parental Consent (Dec. 5, 2014), https://www.ftc.gov/news-events/press-releases/2014/12/ftc-approves-final-order-case-about-google-billing-kids-app.

2 FED. TRADE COMM’N, OFFICE OF CLAIMS AND REFUNDS ANNUAL REPORT (2017)[hereinafter FTC OFFICE OF CLAIMS AND REFUNDS ANNUAL REPORT (2017)],https://www.ftc.gov/system/files/documents/reports/bureau-consumer-protection-office-claims-refunds-annual-report-2017-consumer-refunds-effected-july/redressreportformattedforweb122117.pdf.

3 David C. Vladeck, Time to Stop Digging: Failed Attacks on FTC Authority toObtain Consumer Redress, ANTITRUST, Fall 2016, at 89.

4 Id.5 Kokesh v. SEC, 137 S. Ct. 1635, 1643–44 (2017). 6 See M. Sean Royall & Richard H. Cunningham, Will “Kokesh v. SEC” Put a

Kink in the Federal Trade Commission’s Disgorgement Hose? (July 10, 2017),https://wlflegalpulse.com/2017/07/10/will-kokesh-v-sec-put-a-kink-in-the-federal-trade-commissions-disgorgement-hose/.

7 Kokesh, 137 S. Ct. at 1642 n.3. 8 Transcript of Oral Argument at 29:16–19, Kokesh, 137 S. Ct. at 1635. 9 Tull v. United States, 481 U.S. 412, 424 (1987).

10 Vladeck, supra note 3, at 94. 11 568 U.S. 442, 444 (2013); 28 U.S.C. § 2462 (“Except as otherwise pro-

vided by Act of Congress, an action, suit or proceeding for the enforcement

of any civil fine, penalty, or forfeiture, pecuniary or otherwise, shall not beentertained unless commenced within five years from the date when theclaim first accrued if, within the same period, the offender or the propertyis found within the United States in order that proper service may be madethereon.”).

12 Kokesh, 137 S. Ct. at 1641. 13 Id. Similar provisions found in each act provide the SEC with the authority

to seek injunctions. See 15 U.S.C. 78u(d)(1) (Securities Exchange Act)(providing SEC may bring an action in the district court “to enjoin [certainunlawful] acts or practices, and upon a proper showing a permanent or tem-porary injunction or restraining order shall be granted without bond”); 15U.S.C. 80b-9(d) (Investment Advisors Act) (providing SEC may bring anaction in district court “to enjoin [certain unlawful] acts or practices and toenforce compliance with this subchapter or any rule, regulation, or orderhereunder”); 15 U.S.C. 80a-41(d) (Investment Company Act) (same).

14 Kokesh, 137 S. Ct. at 1640. 15 See Transcript of Motion for Entry of Final Judgment at 22:17–24:13, SEC

v. Kokesh, 2015 WL 12670314 (D.N.M. Mar. 9, 2015). 16 SEC v. Kokesh, 834 F.3d 1158, 1164 (10th Cir. 2016). 17 Kokesh, 137 S. Ct. at 1643. 18 Id. at 1642 (quoting Huntington v. Attrill, 146 U.S. 657, 667 (1892)).19 Id. at 1643–44. 20 Id. at 1643. 21 Id.22 Id. at 1643–44.23 Id. at 1644.24 Id.25 FTC v. Bronson Partners, LLC, 654 F.3d 359, 373 (2d Cir. 2011). 26 Id.27 Id. (recognizing that the “primary purpose of disgorgement orders is to

deter violations of the” FTC laws); FTC v. Gem Merch. Corp., 87 F.3d 466,469 (11th Cir. 1996) (recognizing “the deterrence function of section 13(b)”disgorgement).

28 See Brief for Respondent at 48, Kokesh, 137 S. Ct. at 1635. 29 Kokesh, 137 S. Ct. at 1642 n.3. 30 See Transcript of Oral Argument at 52:18–21, Kokesh, 137 S. Ct. at 1635

(“Ms. Goldenberg: There are almost 50 years of precedents on how thisshould work and I think the way it worked is—; Justice Gorsuch: Not in thisCourt.”).

31 Id. at 52:14-16. For a Supreme Court tracker, Justice Gorsuch’s choice ofwords may be of particular note. The precedents that agencies, like the SEC,rely on to justify disgorgement and restitution remedies as ancillary toinjunctive authority—Porter v. Warner Holding Co., 328 U.S. 395 (1946), andMitchell v. Robert De Mario Jewelry, Inc., 361 U.S. 288 (1960)—have beendescribed as relics from an era in which the Supreme Court considered ita “duty of the courts to be alert to provide such remedies as are necessaryto make effective the congressional purpose,” rather than to hone to thestatute’s text. Alexander v. Sandoval, 532 U.S. 275, 287 (2001) (quotingJ.I. Case Co. v. Borak, 377 U.S. 426, 433 (1964)). Over the last twodecades, the Court has cautioned against the judicial crafting of statutoryremedial schemes in excess of those Congress carefully delineated, warn-ing that important policy (and thus legislative) judgments are implicated bythe choice of remedies available in a given statutory scheme. See, e.g.,Alexander, 543 U.S. at 286–87.

32 Transcript of Oral Argument at 33:12-18, Kokesh, 137 S. Ct. at 1635(“They’re—they’re sort of backing and filling. I mean, this remedy is outthere, and yes, [Congress is] saying this. But it does seem to me that wekind of have a special obligation to be concerned about how far back thegovernment can go when it’s something that Congress did not addressbecause it did not specify the remedy.”).

33 Id. at 7:20–8:2; 9:5–11.34 Id. at 7:20–8:2.

35 Id. at 9:5–11.36 Tull, 481 U.S. at 424. 37 Kokesh, 137 S. Ct. at 1644–45. 38 Id. at 1644. 39 Id. at 1644–45. 40 FTC v. DirecTV Inc., No. 15-cv-01129-HSG (N.D. Cal. 2017); FTC v. J. William

Enters., LLC, No. 6:16-cv-2123 (M.D. Fla. 2017); FTC v. Publishers Bus.Servs., Inc., No. 17-15600 (9th Cir. 2017); FTC v. AMG Capital Mgmt.,LLC., No. 16-17197 (9th Cir. 2017); FTC v. Credit Bureau Ctr., LLC, No. 17-CV-194 (N.D. Ill. 2017).

41 DirecTV’s Motion for Leave to File Third Amended Answer at 1–3, FTC v.DirecTV LLC, No. 4:15-cv-01129 (N.D. Cal. June 20, 2017), ECF No. 323.

42 Federal Trade Commission’s Opposition to DirecTV’s Motion for Leave to FileThird Amended Answer at 6–8, FTC v. DirecTV LLC, No. 4:15-cv-01129(N.D. Cal. June 30, 2017), ECF No. 330.

43 Id. at 5 (citing Complaint at 14, ¶ 55, SEC v. Kokesh, No. 1:09-cv-01021(D.N.M. Oct. 27, 2009), ECF No. 1).

44 Id. at 7 n.6. 45 FTC v. DirecTV Inc., No. 15-cv-01129-HSG, 2017 WL 3453376, at *6 (N.D.

Cal. Aug. 12, 2017). 46 Id. at *5. 47 Defendants’ Joint Motion for Partial Summary Judgment at 3–10, FTC v. J.

William Enterprises, LLC, No. 6:16-cv-2123 (M.D. Fla. Aug. 2, 2017), ECFNo. 146.

48 Plaintiff’s Response in Opposition to Defendants’ Joint Motion for PartialSummary Judgment at 2, 7, FTC v. J. William Enterprises, LLC, No. 6:16-cv-2123 (M.D. Fla. Sept. 8, 2017), ECF No. 159.

49 FTC v. J. William Enterprises, LLC, No. 6:16-cv-2123, 2017 WL 4776669,at *2 (M.D. Fla. Oct. 23, 2017).

50 Defendants’ Motion to Modify Preliminary Injunction at 1–2, FTC v. CreditBureau Center, LLC, et al., No. 17-CV-194 (N.D. Ill. Nov. 9, 2017), ECF No.156.

51 Federal Trade Commission’s Memorandum in Opposition to Defendants’Motion to Modify Preliminary Injunction at 5–6, FTC v. Credit Bureau Center,LLC, et al., No. 17-CV-194 (N.D. Ill. Nov. 28, 2017), ECF No. 164.

52 Id. at 7. 53 Id. at 8. 54 Id. at 1, 9. 55 FTC v. Credit Bureau Ctr., LLC, No. 17-CV-194, 2018 WL 482076, at *1

(N.D. Ill. Jan. 14, 2018). 56 Id. at *2 (internal quotations and citations omitted). 57 Publishers Business Services has also raised on appeal the argument that

Section 13(b) has “common-law foundations,” so actions brought pursuantto Section 13(b) are therefore subject to a common-law proximate causeanalysis. Initial Brief of Appellants at 32–39, FTC v. Publishers Bus. Servs.,Inc., No. 17-15600 (9th Cir. Aug. 30, 2017), ECF No. 13.

58 Initial Brief of Appellants at 18–26, FTC v. Publishers Business Servs., Inc.,No. 17-15600 (9th Cir. Aug. 30, 2017), ECF No. 13.

59 Brief of the Federal Trade Commission at 22, FTC v. Publishers BusinessServs., Inc., No. 17-15600 (9th Cir. Oct. 30, 2017), ECF No. 22.

60 Id. at 38. 61 Opening Brief of Relief Defendants-Appellants at 84–86, FTC v. AMG Capital

Management, LLC, No. 16-17197 (9th Cir. July 21, 2017), ECF No. 14.62 Id. at 89, 90. 63 Brief of the Federal Trade Commission at 90–92, FTC v. AMG Capital Man -

age ment, LLC, No. 16-17197 (9th Cir. Nov. 28, 2017), ECF No. 37-2. 64 Id. at 91–92. 65 Federal Trade Commission’s Opposition to DirecTV’s Motion for Leave to File

Third Amended Answer at 5, FTC v. DirecTV LLC, No. 4:15-cv-01129 (N.D.

A R T I C L E S

1 0 0 · A N T I T R U S T

Cal. June 30, 2017), ECF No. 330; Brief of the Federal Trade Commissionat 38, FTC v. Publishers Business Servs., Inc., No. 17-15600 (9th Cir. Oct.30, 2017), ECF No. 22; Brief of the Federal Trade Commission at 91, FTCv. AMG Capital Management, LLC, No. 16-17197 (9th Cir. Nov. 28, 2017),ECF No. 37-2.

66 Fed. Trade Comm’n, Policy Statement on Monetary Equitable Remedies inCompetition Cases (July 31, 2003) [hereinafter FTC 2003 Policy Statementon Mone tary Equitable Remedies in Competition Cases], https://www.ftc.gov/public-statements/2003/07/policy-statement-monetary-equitable-remedies-including-particular (withdrawn July 31, 2012).

67 Id.68 Kokesh, 137 S. Ct. at 1643. 69 See FTC 2003 Policy Statement on Monetary Equitable Remedies in Com -

petition Cases, supra note 66. 70 See, e.g., FTC v. Febre, 128 F.3d 530, at 535–36 (7th Cir. 1997) (“The

Commission must show that its calculations reasonably approximated theamount of customers’ net losses, and then the burden shifts to the defen-dants to show that those figures were inaccurate.”); FTC v. JK Publ’ns, Inc.,No. 99-0044, 2000 WL 35594143, at *17 (C.D. Cal. Aug. 9, 2000) (requir-ing the FTC’s restitution amount to “reasonably approximate the amount ofconsumer loss”). The Ninth Circuit’s recent decision in FTC v. CommercePlanet, Inc. highlights how approximate the FTC’s estimate can be. InCommerce Planet, the defendant contested the amount of the FTC’s resti-tution award of $18.2 million as being “arbitrarily” determined because theFTC failed to account for the fact that “not all of the consumers who pur-chased [the product at issue] were deceived.” FTC v. Commerce Planet, Inc.,815 F.3d 593, 604–04 (9th Cir. 2016). The district court noted that the FTCmet its initial burden based on expert testimony that “most” people whosaw the advertisement at issue would have been deceived. FTC v.Commerce Planet, 878 F. Supp. 2d 1048, 1092 (C.D. Cal. 2012). The dis-trict court accepted ipso facto that 50 percent is the lower bound of “most,”and then multiplied the defendant’s net revenues of $36.4 million by thatfraction to calculate the appropriate monetary remedy. Id. The Ninth Circuitspecifically affirmed this methodology, which included no specific assess-ment of how many of the defendant’s customers were deceived by theadvertising practice the FTC successfully challenged or received some valuefrom the service. Commerce Planet, 815 F.3d at 605.

71 See FTC v. Nat’l Urological Group, Inc., 645 F. Supp. 2d 1167, 1212 (N.D.Ga. 2008) (stating that the “primary purpose of restitution in the contextof deceptive advertising is to restore victims to their position prior to thedeceptive sales” and that “the court looks to the price paid by consumersand does not deduct any value received” when calculating the remedyamount). Notably, the FTC very recently took the position in motions prac-tice in the FTC v. DirecTV LLC matter described above that it could havedemanded all of DirecTV’s net revenues during the applicable time period,even though DirecTV offered “legitimate products that some customers val-ued.” Federal Trade Commission’s Opposition to DirecTV’s Motion for PartialFindings at 25, FTC v. DirecTV LLC, No. 4:15-cv-01129 (N.D. Cal. Sept. 25,2017), ECF No. 401.

72 See, e.g., FTC OFFICE OF CLAIMS AND REFUNDS ANNUAL REPORT (2017), supranote 2, at 7 (“As part of its mission to protect American consumers, the FTCworks to get money back to people who are harmed by illegal business prac-tices. . . . In all of its activities, the Bureau works to provide consumerrefunds whenever feasible, and then develops customized refund programsthat are designed to get as much money back to as many injured consumersas possible.”).

73 See, e.g., Gem Merchandising, 87 F.3d at 470 (“[B]ecause it is not alwayspossible to distribute the money to the victims of defendant’s wrongdoing,a court may order the funds paid to the United States Treasury.”); Febre,128 F.3d at 537 (stating that the FTC “often requests orders directingequitable disgorgement of the excess money to the United States Treasury”because “they cannot identify all the consumers entitled to restitution andcannot distribute all the equitable relief ordered to be paid”).

S P R I N G 2 0 1 8 · 1 0 1

Comment on “AreMerger EnforcementAnd Remedies TooPermissive? A Look atTwo Current MergerStudies” by JohnHarkriderB Y J O H N K W O K A

OVER THE PAST FEW YEARS THEREhas been increased academic, policy, and publicattention to the issue of mergers in the U.S.economy—to their causes, including possiblypermissive antitrust, and to their effects, both

narrowly economic and wider in scope. The debate over theappropriate degree of stringency for merger control policy isscarcely new, of course: it predated, and prompted, the firstmerger statutes more than a century ago. But recently economists and policymakers have more

aggressively addressed the issues with data, that is, with care-ful consideration of past experience in an effort to betterunderstand what experience tells us and how its lessons canstrengthen the ability of antitrust policy to serve consumersand competition. My own research in Mergers, MergerControl, and Remedies (MMCR)1 has contributed to thatdebate, as has work by the Federal Trade Commission in itsMerger Remedies Study.2 These two sources come to ratherdifferent conclusions, with the FTC concluding that its rem-edy program has been largely successful whereas my evalua-tion of the FTC and the Justice Department remedies con-cludes that those have too often failed.

In his recent article in ANTITRUST, John Harkrider com-pared these two contributions, endorsing the FTC study’s“much better data and empirical methods” and urging cau-tion about my work, which he says is based on a “limited dataset, unrepresentative of the total population of transactionsand based on public data.”3 In this note, I address two ques-tions central to Harkrider’s critique: first, are my findingsundermined by data limitations of the sort that Harkriderclaims? And second, does my research suffer by comparisonwith the FTC Remedies Study, as Harkrider maintains?

MMCR’s Data and Harkrider’s CritiqueThe research reported in my book is based on three datasets. The first is the FTC’s own enforcement data that showthat increasingly in recent years the agency has concentratedits enforcement effort on mergers in the highest concentra-tion category, ultimately ceasing all challenges to mergersresulting in five or more significant competitors. Second, I compile and conduct a meta-analysis of all the

methodologically sound studies of actual merger outcomesin the economics literature, and find that a high proportionof the 50 or so carefully studied mergers result in priceincreases.Third, using additional data on enforcement actions taken

with respect to these mergers, I find that a substantial frac-tion of those that ultimately are found to be anticompetitivewere in fact cleared by the relevant agency. I also find thatmany mergers addressed through the use of remedies resultin price increases anyway, especially in cases where conductremedies are used.The data base that Harkrider focuses on is my compilation

of all studies of horizontal mergers in the United States (so-called merger retrospectives), which I use for a meta-analysisof mergers and remedies. As noted, Harkrider’s criticismscenter around the quality of the data: their public nature, bal-ance, and representativeness.4 My data are indeed all public,and their scope and numbers reflect that constraint. Harkriderargues that the resulting number of industries is too small andcannot be “a representative sample of transactions.” It is easy to say that more would be better, but what

“more” means in this context deserves further thought. Forpurposes of evaluating merger policy, one would not wantcoverage of all 1000 industries that Harkrider alludes to, noreven all industries in HSR filings. It has been estimated thatthere are 10,000 mergers valued at over one million dollarsper year in the United States, only a minuscule fraction ofwhich raise any competitive concerns—and that is apartfrom those valued at less than one million dollars. Even thevast majority of HSR filings involve mergers that do noteven get a second request, much less found to be competi-tively problematic. There is no need for serious studies cov-ering all these mergers which are not relevant for policy pur-poses. For policy purposes I would argue that one should be

interested in “mergers at the margin,” that is, those that

John Kwoka is the Neal F. Finnegan Distinguished Professor of Economics

at Northeastern University and the author of Mergers, Merger Control,

and Remedies: A Retrospective Analysis of U.S. Policy. He has consult-

ed on many mergers and associated remedies, studied enforcement and

remedies practices in general, and offered comments before and reviews

after the recent FTC Remedies Study.

But simply noting the data limitations of the studies inMMCR does not imply their inferiority since every one ofthe studies examines actual price data to assess merger out-comes. All test for merger-induced changes using state-of-the-art methodologies, such as difference-in-differences, for con-trolling for other influences and isolating the effects ofmergers. All were judged by the consistent and explicit cri-terion of whether postmerger, post-remedy price rose or fell.And they are all detailed in the book for any reader to assess.In contrast, and despite its potential for a truly break-

through study of divestiture remedies and policy, the FTCdid not conduct such a study. It did not report and apparentlydid not conduct a single retrospective analysis of divestiturecases. Its judgments about remedy outcomes were not basedon actual pre and post-remedy price data for analyzing out-comes. It did not use difference-in-difference methodologynow common in the literature and relied on in all studies inmy book. It did not put on the record any new statistical evi-dence with respect to its cases, or even report which divesti-tures they believed were successful and which not. Rather, the FTC study reports that for a little over half the

cases, it relied on interviews and some data (not includingprice) that it had secured from the parties to arrive as itsown judgment as to whether its remedy had preserved com-petition. Inadequate as that process and evidence may be, foranother quarter of the remedy cases the FTC study did notuse any data at all, but simply relied on responses to ques-tionnaires to some market participants. And for the remain-ing cases, it failed even to solicit any outside information,much less data, but instead decided whether its own remedieswere successful based entirely on its internal records andviews of its own staff that oversaw certain industry sectors. Compounding this, for almost half the cases the FTC

declared divestiture remedies to be successful even if they didnot necessarily preserve or restore competition—which is, ofcourse, the objective of a remedy. For some cases, the FTCassessed the remedies based only on whether the divestedassets were still in operation in the industry, and for others thecriterion was weaker yet—simply whether the assets thatwere ordered to be divested were in fact divested. Both ofthese latter criteria do not answer the question of whether theremedies preserved competition.

A R T I C L E S

1 0 2 · A N T I T R U S T

might raise competitive concerns. In that respect, the focusof research—as reflected in this body of merger retrospec-tives—may be more or less on target, since it is mergers at theenforcement margin that attract research attention. There islittle reason to lament the lack of studies of most mergersthroughout the rest of the economy—although, of course,more studies of mergers at the margin is an important mat-ter, as MMCR stresses.This leaves Harkrider’s concern5 that imbalance among

industries in the sample may lead to distortion in the report-ed findings. That concern is refuted by the following simpletest. The price effect reported in MMCR—a 7.2 percentprice rise—is the average across all mergers in the data base.As Harkrider observes, this does give more weight to indus-tries where more mergers have been studied. But I have recal-culated this average effect across industries, rather than merg-ers, thereby eliminating the possibility that the measuredaverage is affected by varying numbers of studies in eachindustry. I do this by first averaging the price effect across allmergers in the same industry, and then averaging those indus-try effects.6 The average price effect across industries is 7.1percent, trivially different from that across mergers. In short,there is no empirical basis for concern that the balance of datadistorts the reported average effect of studied mergers.Finally, I would note Harkrider’s particular criticisms of

studies of mergers among journal publishers(their number,their methodology, and their findings. His criticisms echothose I have heard from others who take exception to theinclusion of various particular studies in the data base. Butthose criticisms and recommendations are at odds with mydetermination that the research protocols in the book be asobjective as possible. As explained in detail in MMCR, allstudies meeting the profession’s publication criteria—peerreview or analogous quality screening—are included, regard-less of my or anyone else’s personal views of the merits of anyparticular study. Substituting judgments about the merits ofparticular studies would have left the research open to doubtabout its objectivity. Whether anyone agrees with the findingsof this research or not, I am satisfied that my methodologyhas avoided reflecting any such subjective judgments.

Harkrider and the FTC Remedies StudyHarkrider offers a number of laudatory comments on theFTC Remedies Study and uses those to draw an unfavorablecontrast with MMCR’s overall approach. Since MMCR cov-ers issues not dealt with in the FTC study, some of thosecomparisons are not apropos. Here I limit my comments tothe issue of divestitures, where a comparison of data bases ispossible. As Harkrider rightly observes, the FTC study hadevery advantage—a comprehensive list of cases involvingdivestiture remedies, access to key nonpublic data, and sub-stantial resources to conduct its analysis. In contrast, thebasis for MMCR was a compilation of published studiesundertaken primarily by outside academics, and based onpublic data.

[T]he FTC study appears to be more of a missed

oppor tunity than a sound study that truly advances

our understanding. . . . I do not think the FTC’s

conclusion that divestiture remedies have in fact

succeeded is well founded.

S P R I N G 2 0 1 8 · 1 0 3

Thus, the FTC study appears to be more of a missedopportunity than a sound study that truly advances ourunderstanding.7 The research in MMCR does not purport tobe definitive and surely would benefit from more retrospec-tive studies to rely on. But I do not think that the FTCstudy’s data or technique represent methodological improve-ments over the studies in MMCR, and I do not think theFTC’s conclusion that divestiture remedies have in fact suc-ceeded is well founded.

ConclusionIn Mergers, Merger Control, and Remedies, I conclude:

The evidence compiled and analyzed in this project pro-vides a compelling argument for the value of merger retro-spectives and certainly for doing more of them. . . . [A] larg-er body of retrospectives, covering the full range of agencydeterminations, and—ideally—based on internal agency dataand other relevant information would over time constitute amore comprehensive basis for understanding the effects ofmergers and merger control.8

I continue to believe that retrospective analyses of merg-ers and remedies represent an important and largely unex-ploited source of insights into enforcement practices. But inorder actually to improve our understanding, any such analy-sis needs to adhere to sound methodology, test actual out-come data against the relevant counterfactual, be strictlyobjective in its standards, and careful in its conclusions. Iremain satisfied that my study meets those criteria and myconclusions represent a constructive step in this quest.�

1 JOHN KWOKA, MERGERS, MERGER CONTROL, AND REMEDIES (2015) [hereinafterMMCR].

2 The FTC’s Merger Remedies Study 2006–2012 (Jan. 2017), https://www.ftc.gov/system/files/documents/reports/ftcs-merger-remedies-2006-2012-report-bureaus-competition-economics/p143100_ftc_merger_remedies_2006-2012.pdf.

3 John D. Harkrider, Are Merger Enforcement and Remedies Too Permissive: ALook at Two Current Studies, ANTITRUST, Fall 2017, at 96.

4 I would note that Harkrider’s criticisms do not apply to my other major database, which consists strictly of data published by the FTC itself.

5 Harkrider conflates two data bases, one covering mergers evaluated singly,the other covering groups of mergers evaluated collectively. The formercomprise the basis for most of my conclusions, specifically including thosehe cautions about, although he cites the imbalance of industries among thegrouped mergers data as evidence against those conclusions. SeeHarkrider, supra note 3, at 97.

6 I have reported this in John Kwoka, The Structural Presumption and the SafeHarbor in Merger Review: False Positives or Unwarranted Concerns, 81ANTITRUST L.J. 837 (2017) (previously posted on ssrn.com in draft).

7 The FTC report does not disclose any detail about its case determinations,making it impossible to assess its conclusions. It could, for example, atleast publish the list of remedy cases that it judged to be successful or not,which would preserve confidentiality of its data and other evidence onwhich it relied.

8 MMCR, supra note 1, at 160.

� THIS THIRD EDITION OF THE State Action

Practice Manual continues to be the most comprehensive

published source of information and guidance concerning

the state action doctrine. This book addresses not only

the historical development and theoretical underpinnings

of the state action doctrine, but also many practical

considerations that arise in the application of the doctrine,

including pleading and procedural issues and strategy

and the role of discovery and expert testimony in cases

applying the state action doctrine. The State Action

Practice Manual is an important source of information

on the state action doctrine for a wide range of antitrust

practitioners representing clients that are subject to state

regulation and thus may be protected from federal antitrust

liability under the doctrine. For practitioners who do not

specialize in antitrust, it provides an accessible source of

guidance on the state action doctrine. This manual will

also be of interest to practitioners at the federal and

state antitrust enforcement agencies.



State Action Practice ManualThird Edition


Pricing:$159.95 List Price / $129.95 AT Section Member

AGENDA

Thursday, May 17, 2018

7:30am–5:00pm Registration

8:00–9:30am Plenary Session: Antitrust Primer

9:45–10:00am Welcome & Introductions

10:00–10:45am Keynote

11:00am–Noon Plenary Session: Year in Review

Noon–1:30pm Lunch

1:30–2:30pm Concurrent Sessions:

Horizontal Collaborations: JVs, JOAs, Service Line Consolidations

Pharma: Pay for Delay—How Are the Courts Applying Actavis?


Hospital Mergers: Counseling Clients in Light of the Appellate Cases

Pharma Product-Hopping & REMS: What Is the Duty to Deal?


Physician Practice Mergers

Hot Topics in Non-Merger Litigation

Friday, May 18, 2018

7:30am–5:00pm Registration

8:00–9:00am Plenary Session: View from the Enforcers

9:00–10:15am Plenary Session: Practical Advice for In-House Counsel

10:30–11:45am Plenary Session: Brave New World Vertical Arrangements

Noon–1:15pm Lunch on your own


Cutting Edge Economics for Future Merger Cases

Competition or Regulation? State COPA Laws & Hospital Mergers


Insurance Mergers—Efficiencies and Monopsony Power

State Boards & Professional Licensing Post-NC Dental


Making Legal Ethics Great Again: The “Challenging” Client

Q&A

he a l t h c a r e

1 0 4 · A N T I T R U S T

A N T I T R U S T I Nmay 17–18, 2018ar l ington, vathe r i tz -car l tonpentagon ci ty

CO-CHAIRSchr ist i j . braunjeff rey w. brennandouglas c. ross

VISITambar.org/athealthcare#athealthcare

Co-Sponsors: ABA Sections of Antitrust Law & Health Law and American Health Lawyers Association

PLEASE JOIN CONFIRMED FACULTY: Gerard ANDERSON, Alicia BATTS, Leah BRANNON, Saralisa BRAU, Christi BRAUN, Jeffrey BRENNAN, Holden BROOKS, Daniel BUTRYMOWICZ, John CALENDER, Mary COLEMAN, Ian CONNER, Michael COWIE, Makan DELRAHIM, Victor DOMEN, Lisl DUNLOP, Richard FEINSTEIN, Deborah FEINSTEIN, Ashley FISCHER, Danielle FOLEY, Lona FOWDUR, Alexis GILMAN, Thomas GREANEY, Kevin HAHM, Melissa HILL, Mark HOROSCHAK, William HORTON, Jonathan JACOBS, Michael KNIGHT, Tara KOSLOV, James LANGENFELD, David MAAS, Sean MAY, John MILES, Peter MUCHETTI, Parker NORMANN, Dina OLDER AGUILAR, Leigh OLIVER, Leslie OVERTON, Mitchell RAUP, Douglas ROSS, Dov ROTHMAN, Mark SEIDMAN, Seth SILBER, Bruce SOKLER, Joshua SOVEN, Jeffrey SPIGEL, Christine WHITE, Paul WONG

AHLA

ABA Section of Antitrust Law

Global Seminar Series: DüsseldorfPublic Interest Considerations and Competition Law

Tuesday, May 8, 2018 • 15:00–19:00Steigenberger Park Hotel

Koenigsallee 1a • 40212 DüsseldorfT: +49 211 13810

WelcomeHartmut Schneider, WilmerHale, Washington, DC

Keynote: Agency Perspectives on Public InterestConsideration

Round 1:The Growing Scrutiny of ForeignInvestments Around the WorldForeign investment reviews have grown in recent years, both

in number and in scope. Jurisdictions continue to scrutinize

investments in industries that raise traditional national security

concerns. But the scope of foreign investment reviews is

not limited to those concerns, as recent matters involving

food producers (Smithfield, GrainCorp), lighting (Lumileds),

or wireless manufacturers (Qualcomm) illustrate. Some

jurisdictions expressly allow foreign investment scrutiny

under general national interest standards—sometimes

even requiring “net benefit” to the host country. Others have

broadly interpreted existing national security standards or

are considering new rules to expand the scope of foreign

investment reviews. This panel will consider:

• What sectors raise national security concerns? What other

national interests should authorities consider when

reviewing foreign investments?

• Should the level of scrutiny depend on the origin of the

acquirer (e.g., China) and/or on the nature of the acquirer

(e.g., state-owned enterprise, sovereign wealth fund, etc.)?

Is it advisable to require a “net benefit” to the host country?

• Which agencies should review foreign investments for

national interest concerns?

• Are there procedural best practices for policymakers

considering changes to their current regimes?

• What does this all mean for practitioners seeking to

help their clients navigate parallel antitrust and foreign

investment reviews?

Moderator:Calvin S. Goldman, Goodmans LLP, Toronto

Panelists:Steven P. Croley, Latham & Watkins LLP, Washington, DC

Christopher Hutton, Hogan Lovells, London

Romina Polley, Cleary Gottlieb Steen & Hamilton LLP, Köln

Kirsten Webb, Clayton Utz, Sydney

Round 2:The Increased Focus on Public InterestConsiderations in Merger ReviewsThis panel will explore public interest considerations unrelated

to foreign investment, including concerns about how a

transaction may impact jobs, small businesses, privacy, or

the concentration of political power. Germany’s experience

with the Edeka/Tengelmann grocery store merger offers a

prominent example of how employment considerations

can override traditional competition law analysis. In other

jurisdictions, public interest considerations can factor into

sector-specific reviews, such as in the banking and finance,

media, telecommunications, or transportation industries.

There is also a lively debate in the U.S. and elsewhere

about whether, as a policy matter, traditional competition

law analysis should be more open to public interest

considerations. This panel will consider:

• What public interest factors should authorities consider in

transactions that do not raise foreign investment concerns

(if any)?

• If such public interest factors are considered, how should

they be balanced against traditional competition law

analysis?

• Is international substantive convergence for the review

of public interest considerations unrelated to foreign

investment desirable and feasible?

• How should practitioners prepare for and address non-

competition concerns in a merger review?

Moderator:Richard M. Steuer, Mayer Brown LLP, New York

Panelists:Antonio Capobianco, OECD, Paris

Hartmut Schneider, WilmerHale, Washington, DC

Daniela Seeliger, Linklaters, Düsseldorf

Patrick Smith, RBB Economics, London

Reception

STATEMENT of EDITORIAL POLICY

ANTITRUST is intended to be read and discussed bylawyers who specialize in counseling and litigation concern-ing the Sherman, Clayton, and FTC Acts and their state andforeign law counterparts, consumer protection law, stateunfair competition laws, and the federal antitrust laws relat-ing to international trade. Like our sponsor, the ABA Sectionof Antitrust Law, we also strive to serve practitioners and cor-porate counsel whose work on other legal issues brings theminto contact with antitrust laws and federal and state com-petition policy.

To this end, ANTITRUST will feature forward-looking arti-cles of practical use to attorneys grappling with new courtdecisions, legislation, or other recent developments. Articlesthat convey a particular perspective not only are acceptablebut are encouraged. In our view, if these articles generate con-troversy, the proper response is to publish responsive articles

or letters to the editor. This we intend to do. Of course, theviews expressed in the articles are the authors’ only and arenot to be attributed to ANTITRUST, its Editors, the Sectionof Antitrust Law, or the American Bar Association unlessexpressly stated.

ANTITRUST welcomes original articles and letters to theeditor. Send correspondence and articles to the Exec utiveEditor, Tina Miller, [email protected]; 203/938-8507; fax 203/938-2845. For more information, visit us on theWeb at http://www.americanbar.org/groups/antitrust_law/publications/antitrust_magazine.html.

American Bar Association321 North Clark StreetChicago, IL 60654

NONPROFIT ORG.U.S . POSTAGE

P A I DAMERICAN BARASSOCIATION

Published by the Section of Antitrust Law, American Bar Association

ANTITRUST Brian Facey, Navin Joneja, Paul Cuomo, and Jeffrey Oliver.....64 The Increasing...

Documents

Transcript of ANTITRUST Brian Facey, Navin Joneja, Paul Cuomo, and Jeffrey Oliver.....64 The Increasing...