GEOGRAPHY, COMPETITION AND MARKET DEFINITION: THE PROPER USE OF ECONOMIC THEORY IN ANTITRUST DECISIONS

Testimony of
Abbott B. Lipsky, Jr.
Senior Competition Counsel
The Coca-Cola Company
Atlanta, Georgia
before the
Federal Trade Commission
Hearings on
Market Definition, Market Power and Entry
in Light of Global Competition
Washington, D.C.
October 18, 1995

Thank you for the opportunity to appear and present testimony. I applaud the Commission for its willingness to reconsider antitrust enforcement policies in light of technical change, the evolution of the world economy and new learning.

This panel involves the competitive importance of geographic location and the role of geographic market definition in antitrust enforcement. I was closely involved in drafting the 1982 Merger Guidelines, which introduced the hypothetical-monopolist thought-experiment as the principal market-definition methodology. I hope to contribute three points to the discussion today. First, I want to emphasize the importance of economic reasoning in the process of antitrust-law enforcement. Second, I encourage the Commission to recognize more explicitly the probabilistic nature of market definition. Finally, I ask the Commission to study the possibility that there may be an unwarranted tendency to discount the likelihood and magnitude of future supply responses in evaluating market structure and likely future conduct.

I. ECONOMIC REASONING IN ANTITRUST ENFORCEMENT

In the distant past, economic reasoning was often absent from antitrust decisions. The Supreme Court decision in United States v. Arnold, Schwinn & Co.[1] provides a leading example. As the Supreme Court recognized in overruling Schwinn, however, antitrust decisions uninformed by economic reasoning are themselves inconsistent with fundamental antitrust objectives.[2] It is now well-appreciated that every substantive antitrust decision must reflect sound economic reasoning.

Sometimes the economic reasoning in an antitrust decision is explicit but wrong, as, for example, when the Supreme Court asserted that high demand elasticity disproves monopoly power -- the "Cellophane fallacy".[3] In other cases the economic reasoning is never made explicit, rendering it impossible for the economic actors who are subject to the draconian antitrust penalties to discover how to comply with the law. Both types of error can impose enormous costs on consumers, businesses and the economy. It therefore bears emphasis that the economic reasoning underlying every antitrust enforcement decision must be made explicit if antitrust law is to weed out invalid theories and adopt sound reasoning.

Fortunately, in many areas of contemporary U.S. antitrust law, dramatic progress has been made in this regard. The 1982 Merger Guidelines led to wide adoption of the hypothetical-monopolist approach to market definition and aided in the identification and refinement of other fundamental aspects of merger analysis.[4] The 1992 Merger Guidelines continued this tradition by adopting an explicit analysis of the market mechanisms that can produce suboptimal competitive performance. During the early Reagan Administration the Antitrust Division frequently filed amicus briefs in private antitrust cases. Officials of the Antitrust Division and members of this Commission articulated the relationships between economic reasoning and antitrust-law objectives. These actions triggered Supreme Court clarification of the theoretical underpinnings of antitrust rules in a broad variety of antitrust contexts -- inferring agreement,[5] the competitive effect of vertical price[6] and non-price[7] restraints, the competitive effect of horizontal collaborations[8] and principles of antitrust standing.[9]

More recently the perennially murky area of predatory pricing law made a significant advance in Brooke Group Ltd. v. Brown & Williamson Tobacco Co.[10] In that case the Supreme Court defined a category of anticompetitive conduct and fashioned a test for liability based on economic reasoning, then -- apparently for the first time in its history -- rendered judgment based on application of the legal standard to the uncontested facts of the specific case. In so doing the Court rejected the proposition that courts are bound to treat the economic reasoning and theories advanced by the parties as propositions of fact to be evaluated only by juries. This holding had been foreshadowed in Matsushita,[11] where the fundamental point of difference between majority and dissent concerned the freedom of a court to ignore record testimony of economic experts attempting to link basic facts to a required finding of anticompetitive effect. Two dots do not form a picture, but connecting Matsushita to Brooke Group strongly suggests that the Supreme Court views the validity of economic reasoning as a question of law.

Other recent examples of this trend also warrant mention. In Blue Cross & Blue Shield United of Wisconsin v. Marshfield Clinic, 1995 U.S. App. LEXIS 26399 (7th Cir. 1995), Judge Posner threw out a jury verdict finding monopolization in a (product) market limited to Health Maintenance Organizations ("HMO's"). Without pausing to comment on the distinction between fact and law, Judge Posner threw out this market definition, observing that an HMO "is basically a method of pricing medical services." Judges are becoming increasingly bold at resolving antitrust cases by drawing connections among economic reasoning, record facts and antitrust-law objectives. Judge Posner is in the vanguard of this trend, but is not alone. Other similar cases include Advo, Inc. v. Philadelphia Newspapers, Inc., 51 F.3d 1191 (3d Cir. 1995), and United States v. Syufy Enterprises, 903 F.2d 659 (9th Cir. 1990).

I also strongly commend to the Commission's attention the recent decision of the Canadian Federal Court of Appeal, overruling (in part) Canada's Competition Tribunal in Southam Inc. v. Director of Investigation and Research, FCA File No. A1668-92 (August 8, 1995) -- a merger case involving print advertising in British Columbia. Two points in this decision bear emphasis. First, the opinion treats both the adoption of a market-definition methodology and the manner in which the selected approach is applied as questions of law. Whether the facts in a particular case satisfy the legal definition that has been adopted is a question of mixed law and fact. Second, and related to the first point, the Court declines to defer to the Competition Tribunal on its market- definition approach, even though the Tribunal is specialized to competition cases.[12]

The Tribunal's decision appeared to the Court of Appeal to have failed a critical reality check: the Tribunal had concluded that general-circulation metropolitan daily newspapers did not compete for print display advertising with community newspapers engaged in selling such advertising. Given a lengthy history of see-saw battles for advertising clients and revenues between the two types of publications, the Court could be forgiven for discounting the Tribunal's judgment. The Tribunal warranted little deference because its conclusion made little economic sense.[13]

For U.S. practitioners suspicious of any interference with jury prerogatives, the notion of market definition as a legal issue must seem threatening. But for careful antitrust scholars examining the broader trends of judicial reasoning and enforcement history, the logic of this approach has strong appeal. Experience and study have shown that some propositions connecting marketplace reality to legal conclusions are untenable. Such propositions -- like the Cellophane fallacy -- must be placed out-of-bounds for fact-finders and the retained experts who offer such propositions in attempting to support legal inferences. No criminal court would permit testimony that people with shifty eyes are more likely to commit felonies -- regardless of the qualifications of the witness. Similarly, no antitrust tribunal should permit testimony that, for example, intellectual property confers market power. Some economic reasoning must be placed out-of-bounds by the courts. Thus the trend to address economic reasoning, including the assumptions and methods of market definition, as legal issues can be expected to intensify.

II. ACKNOWLEDGING THE UNCERTAINTIES OF MARKET DEFINITION

For a variety of reasons connected with the strategy and tactics of antitrust litigation (both administrative and judicial), the process of market definition in antitrust investigations and cases tends to drive out objectivity. Each party asserts its own market definition with confident enthusiasm, when in reality market definition must be regarded as a probabilistic matter. The point is illustrated with unique clarity by the proceedings in New York v. Kraft General Foods, Inc., 1995-1 Trade Cas. (CCH) para. 70, 911 (S.D.N.Y. 1995). The presiding judge, the Honorable Kimba Wood, herself an experienced antitrust practitioner, was confronted with a merger case in which market definition was a critical issue. Judge Wood resorted to the novel device of retaining an economic expert on behalf of the court pursuant to Rule 706, Fed.R.Ev. The particular expert, Dr Alfred E. Kahn -- the distinguished industrial organization, antitrust and regulatory economist -- examined conflicting claims and opinions and provided testimony on market definition that is characteristically perceptive and candid. Dr. Kahn's testimony recognizes the uncertainties of measuring substitutability and of attempting to distill a single definition from the complex brew of relevant factors. Remarking on the difference between the objective expert and the retained expert, Dr. Kahn observes:

I can't refrain from pointing out how unfamiliar this role has been to me and how the [testimony] differ[s] so fundamentally from the nature of testimony that I would give if it were testimony or the way I would go about writing professional articles or books.

Transcript at 2334-35. Equally interesting in light of this remark is the further observation:

I will, of course, try to express my judgments in the familiar orderly way on each of the major issues, taking them up in the familiar progression; market definition, measures of concentration, and particularly incremental HHI's, the nature of competition in the industry premerger, the likely effect of the merger and of its dissolution.

But the more I have struggled with these individual issues, and particularly the issue of market definition, the more I have become totally convinced that these judgments and recommendations and positions of the respective parties are totally interdependent.

Obviously, if you feel that the merger is highly likely to impair competition, it will impel you to a conception of the market. Indeed, it will itself entail a conception of the definition of the market that will be different from the conception if you have the other assessment of the probabilities.

And I don't suggest that merely as a tactical choice. Obviously one side wants to get the incremental Herfindahl up and the other wants to get it down. But as a matter of genuine intellectual conviction -- and I won't conceal the fact that I almost begin or began with a conception of the likely consequences of the merger to some extent that influenced my conception of the proper definition of the market.

Of course you can't assess the likely effects of the merger except on the basis of your conception of the definition of the market, of the structure, the behavior of the industry, and how they may be expected to change as a result of the merger or would be expected to change if it were undone.

Id. at 2335-36.

If Fred Kahn finds it difficult to make sense of market definition data, very few economists (if any) could be expected to do better. Dr. Kahn does eventually reach an opinion, and the court relies on it in allowing the merger. But the testimony provides a rare illustration of the probabilistic nature of market definition. Because judgments of antitrust liability rest so critically on market definition, it is too seldom acknowledged that such judgments are themselves highly probabilistic.

Yet, one rarely finds any analysis of Type I and Type II error in antitrust decisions. Retrospective research on the accuracy of the economic predictions underlying previous antitrust decisions is also extremely rare. When the Commission or a court strikes down a merger, for example, it seldom attempts to quantify either the efficiencies or the price increases that might result from a decision for or against the transaction. Yet the magnitude of these gains and losses, the probabilities of incorrect judgments either way and a comparison of these quantities are all critical to any rational antitrust enforcement scheme.

It might be objected that to compound the existing uncertainties of merger enforcement with further uncertainties of probability analysis would ask far more than the institutions of antitrust enforcement are capable of giving. In response, I would recall a maxim frequently applied to the field of antitrust enforcement by Bill Baxter. "I would prefer an approximate answer to the right question over an answer to seven significant digits to the wrong question." To have any hope of rationalizing antitrust enforcement and improving its utility, we have to begin the process of refining our enforcement tools -- including our economic theories and our market definition hypotheses.

In sum, market definitions are probabilistic assertions. Predictions of anticompetitive effect are not facts, but conjectures based on reasoning. Antitrust law must control carefully both the process of market definition and the use of economic theory if it is to maintain claims to utility and public benefit in the long run. I believe that the current spate of tough judicial reality checks demonstrates broad implicit acceptance of these propositions. I advise the Commission to expect more searching judicial and perhaps Congressional and public scrutiny. I encourage the Commission to go with this flow: make your economic reasoning explicit, admit that your judgments rest on probabilistic assertions, and do more R&D to find better methodologies for getting closer to more objectively defensible conclusions.

III. GEOGRAPHIC MARKET DEFINITION AND SUPPLY ELASTICITY

My final point is not necessarily restricted to the issue of geographic market definition, but it is appropriately made in this context. Chairman Pitofsky's 1990 Columbia Law Review article succinctly identifies the options for including different suppliers in the relevant geographic market. The choices are to count only the current in-market sales, to add likely future in-market sales, to further add the total production or even the total capacity of distant suppliers who have or could have in-market sales, or to discount some or all of the above based on import restraints, the shape of relevant supply curves, or a variety of other de facto or de jure barriers.

In my view the rule chosen should not be regarded as critical. As Landes & Posner have emphasized, the important point is to ensure that market definition and competitive assessment methodologies are part of a consistent framework -- the point that Fred Kahn made in emphasizing the interdependence of all of the steps in merger analysis. If a narrow market-definition option is chosen, that must be remembered in assessing competitive effect. If antitrust decisions depend critically on the choice of methodology, something is wrong with the methodology. The conclusion should change only if the facts change.

It is a general impression acquired over almost twenty years of antitrust practice -- some as a government enforcer, more as a private practitioner -- that antitrust law consistently underestimates the source, nature and magnitude of likely future supply responses. I had the impression from studying the materials circulated in advance of this hearing -- many of them dealing with the automobile industry -- that there was sympathy for an argument along these lines, based on the fact that American automakers were blindsided in the 1970's by the unexpected strength of Japanese and European competitors.

To take another example, IBM Corp. was challenged by antitrust enforcers of the 1960's and 70's based on fears that the company might lock up the electronic data processing industry, becoming an invincible "computer utility," dispensing services from monolithic mainframes squirreled away in secure redoubts, like nuclear reactors or ammo dumps. More than ten years after dismissal of the government's monopolization suit against IBM, the fear seems exaggerated -- even silly -- in light of the demolition by the personal computer of a competitive structure that had evolved in the era of brontosaurus-like mainframes.

The inherent flexibility of supply is perhaps most dramatically illustrated by the story of American efforts in World War II to reduce Nazi Germany's war production through strategic bombing of production facilities for critical supply components. The strategy was based on so-called "bottleneck" analysis -- the proposition that certain production choke-points controlled the pace of downstream activities. The underlying assumption sounds suspiciously similar to the logic that underlies many essential facilities claims and merger analyses. It is now widely recognized that the bombing strategy was a failure. German war production kept increasing until Germany was physically occupied.[14]

I urge the Commission to examine systematically the perception of likely supply responses. The Commission should consider whether retrospective study of the assumptions and results of previous antitrust enforcement efforts would help to discover whether fundamental but unstated misconceptions about supply response may underlie some enforcement judgments. I know that the Commission's Bureau of Economics performed at least one retrospective study, entitled "The Effects of FTC Antitrust Challenges on Rival Firms 1981-1987: An Analysis of the Use of Stock Returns to Determine the Competitive Effects of Horizontal Mergers," Nov. 1989. The effort ran into a number of difficult methodological questions, but again, I urge the Commission to persist.

The Commission should focus less on the particular elements of market definition and market power analysis, and more on the probabilistic nature of determinations of these issues and the ultimate legal judgments they support. Don't look for the lost keys underneath the streetlight if you dropped them elsewhere. Ask Bill Baxter or Fred Kahn to loan you a flashlight.

Footnotes:

[1] 388 U.S. 365 (1967).

[2] Continental T.V., Inc. v. GTE Sylvania, Inc., 433 U.S. 36 (1977).

[3] The fallacy is explained in now-Chairman Robert Pitofsky'sarticle, "New Definitions of Relevant Market and the Assault on Antitrust," 90 Colum. L.Rev. 1805 (1990).

[4] The spread of the hypothetical-monopolist methodology is briefly traced in Gregory J. Werden, "Market Delineation under the Merger Guidlines: a tenth anniversary retrospective," 38 Antitrust Bulletin 517 (1993).

[5] Monsanto Co. v. Spray-Rite Service Corp., 465 U.S. 752 (1984).

[6] Business Electronics Corp. v. Sharp Electronics Corp., 485 U.S. 717 (1988).

[7] Jefferson Parish Hospital District No. 2 v. Hyde, 466 U.S. 2 (1984).

[8] Board of Regents v. National Collegiate Athletic Association, 468 U.S. 85 (1984).

[9] Associated General Contractors, Inc. v. California State Council of Carpenters, 459 U.S. 519 (1983).

[10] 113 S. Ct. 2578 (1993)

[11] Matsushita Electric Industrial Co. v. Zenith Radio Corp., 475 U.S. 574 (1986).

[12] Although merely tangential to my basic point, Southam also contains a well-reasoned suggestion that among the considerations governing product market definition, functional substitutability should be regarded as primus inter pares. The court regards functional interchangeability as a necessary condition for inclusion within the same market, and, although it is not sufficient to require inclusion, functional substitutability is recognized as a very important factor. Functional interchangeability of two products certainly calls for an explanation why they should not be regarded as reasonable substitutes within the same market.

[13] Although the views expressed here are my own, I wish to thank Paul Crampton of Davies, Ward & Beck, Toronto, Ontario, for help in understanding the opinions in Southam.

[14] Kenneth P. Worrell, "The Strategic Bombing of Germany in World War II: Costs and Accomplishments," 73 J. Am. History 702 (1986); John Kenneth Galbraith, "After the Air Raids," 32 Am. Heritage 65 (1981).