Market definition: An analytical overview

Antitrust Law Journal

Volumn 74, Issue 1, 2007, Pages 129-173

  Baker, Jonathan B ^a  

^a AMERICAN UNIVERSITY   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 33947657050     PISSN: 00036056     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Review

References (151)

1. Eastman Kodak Co. v. Image Technical Servs., Inc., 504 U.S. 451, 469 n.15 (1992) ("Because market power is often inferred from market share, market definition generally determines the result of the case."); Robert Pitofsky, New Definitions of Relevant Market and the Assault on Antitrust, 90 COLUM. L. REV. 1805, 1807 (1990) ("Knowledgeable antitrust practitioners have long known that the most important single issue in most enforcement actions - because so much depends on it - is market definition.").
2. The focus of this essay is on the use of market definition to evaluate the possibility of anticompetitive seller conduct. One of the parallel market definition issues that arises when evaluating the possibility of anticompetitive conduct by buyers (the exercise of monopsony power) is addressed briefly infra note 26. Special issues arising in the definition of innovation markets are not considered.
3. For a black-letter survey of the extensive case law on market definition, see ABA SECTION OF ANTITRUST LAW, ANTITRUST LAW DEVELOPMENTS 525-602 (5th ed. 2002) [hereinafter ANTITRUST LAW DEVELOPMENTS].
4. The use of business documents to define antitrust markets is criticized in Geoffrey A. Manne & E. Marcellus Williamson, Hot Docs vs. Cold Economics: The Use and Misuse of Business Documents, 47 ARIZ. L. REV. 609, 633-46 (2005).
5. Although markets are sometimes described casually by naming the firms that participate in them, they are defined by their products and locations (and targeted buyers for a price discrimination market), not by their participants. See, e.g., U.S. Dep't of Justice & Federal Trade Comm'n, Horizontal Merger Guidelines §§ 1.12, 1.22 (1992, revised 1997) [hereinafter Merger Guidelines].
6. The identification of market participants - a task that generally requires market definition - may be important to the analysis of the competitive effects of firm conduct even when market shares are not important. For example, in evaluating possible coordinated effects of merger, market participants must be identified in order to identify a maverick firm and analyze the effects of merger on its incentives, but market shares need not invariably play a role in that undertaking.
7. ANTITRUST LAW DEVELOPMENTS, supra note 3, at 68.
8. See generally Jonathan B. Baker, Mavericks, Mergers, and Exclusion: Proving Coordinated Competitive Effects Under the Antitrust Laws, 77 N.Y.U. L. REV. 135, 152-54 (2002) (survey of the economic literature relating price to market concentration).
9. See, e.g., U.S. Dep't of Justice & Federal Trade Comm'n, Statements of Antitrust Enforcement Policy in Health Care, Statements 7 & 8 (1996) (antitrust "safety zones" based on market shares); Merger Guidelines, supra note 5, § 1.51 (general standards based on post-merger market concentration and its increase). In the Merger Guidelines, market concentration is evaluated using the Herfindahl-Hirschman Index (HHI), a statistic derived from market shares.
10. The statutory reference in Clayton Act § 7 to a "line of commerce" and "section of the country" is frequently cited as mandating delineation of a product and geographic market, respectively, in litigation under that statute. See, e.g., ANTITRUST LAW DEVELOPMENTS, supra note 3, at 328.
11. For an argument that Clayton Act § 7 should be interpreted to allow courts to rely primarily on direct evidence when market shares are largely uninformative as to competitive effects, see Jonathan B. Baker, Product Differentiation Through Space and Time: Some Antitrust Policy Issues, 42 ANTITRUST BULL. 177, 185-90 (1997).
12. FTC v. Indiana Fed'n of Dentists, 476 U.S. 447, 460-61 (1986) (evaluating a Sherman Act § 1 claim under the rule of reason); Re/Max Int'l v. Realty One, Inc., 173 F.3d 995, 1018-20 (6th Cir. 1999) (monopoly power under Sherman Act § 2 established with direct evidence). Courts have been willing to rely on direct evidence to prove adverse competitive effects in Sherman Act litigation, particularly in cases involving "nearly naked" horizontal restraints, but they appear most inclined to do so when the direct evidence is consistent with what would be inferred about market power from an informal estimate of market concentration based upon a rough market definition. Andrew I. Gavil, A Comment on the Seventh Circuit's Republic Tobacco Decision: On the Utility of "Direct Evidence of Anticompetitive Effects," ANTITRUST, Spring 2005, at 59.
13. Even when market shares, one output of the market definition process, are not strongly probative of market power or anticompetitive effect, however, another output, the identity of the sellers of close substitutes, may remain important. For example, if competitive effects are inferred from estimates of elasticities of demand, it is important to account for substitution to all close rivals but it need not be important to know market shares.
14. For a discussion, see Baker, supra note 10, at 182.
15. Market shares may be relevant here, however, if the distribution of buyer second choices among sellers (which identify localized competition) is strongly correlated with the distribution of buyer first choices (which are the basis for market shares). Merger Guidelines, supra note 5, § 2.211.
16. Direct evidence of anticompetitive effect might include price increases demonstrably unrelated to benign causes, such as higher costs or improved product quality. Direct evidence of market power (from which anticompetitive effect may be inferred) might include evidence as to demand elasticities. Two examples illustrate the connection between demand elasticities and market power. First, if a firm or group of firms raise price when demand grows less elastic, that evidence could be used to show that those sellers exercise market power. Second, if a merger among sellers of differentiated products would make the demand facing either less elastic, that evidence might be used to show that the loss of localized competition arising from merger would make it profitable for the merged firm to exercise market power by raising price for one or both firms' products.
17. See generally ABA ANTITRUST SECTION, MONOGRAPH No. 12, HORIZONTAL MERGERS: LAW AND POLICY 89-130 (1986).
18. Courts commonly treat both Sherman Act and Clayton Act cases as authority on market definition, regardless of whether the case involves a challenge to a merger, monopolization, or unreasonable agreement. ANTITRUST LAW DEVELOPMENTS, supra note 3, at 530 n.26 (citing cases).
19. United States v. E.I. du Pont de Nemours & Co., 351 U.S. 377, 395 (1956) (Cellophane). In a similar formulation, the Court also stated that the "market is composed of products that have reasonable interchangeability for the purposes for which they are produced-price, use and qualities considered." Id. at 404. The so-called Cellophane fallacy is discussed in Part V.B below.
20. Id. at 400.
21. Compare United States v. Aluminum Co. of Am., 377 U.S. 271, 276-77 (1964) (Rome Cable) (majority defines market based on demand substitution), with id. at 285 (dissent favors broader market based on both supply and demand substitution).
22. One change is discussed infra note 66.
23. For a discussion of the history and application of the hypothetical monopolist test for market definition, see generally Gregory J. Werden, Market Delineation and the Justice Department's Merger Guidelines, 1983 DUKE L.J. 514 (1983);
24. Gregory J. Werden, The History of Antitrust Market Delineation, 76 MARQ. L. REV. 123 (1992);
25. Gregory J. Werden, The 1982 Merger Guidelines and the Ascent of the Hypothetical Monopolist Paradigm, 71 ANTITRUST L.J. 253 (2003) [hereinafter The 1982 Merger Guidelines].
26. The Merger Guidelines' reference to a "small but significant and non-transitory" increase in price is often abbreviated as SSNIP (pronounced "snip").
27. The useful term "candidate market" was originated by Greg Werden.
28. For example, product markets are defined through "successive iterations" of the price increase test. Merger Guidelines, supra note 5, § 1.11.
29. This essay addresses settings in which the allegation concerns the exercise of market power by sellers. In monopsony cases, in which the allegation concerns the exercise of market power by buyers, markets would be defined analogously with respect to the economic force of seller substitution. The "hypothetical monopsonist" issue in market definition in such cases is whether the collective exercise of market power by the buyers of a group of products within a region, which depresses the prices paid to sellers through a reduction in purchases, would be made unprofitable by seller decisions to deal instead with other buyers or to cease participation in the market.
30. See generally Jonathan B. Baker, The Antitrust Analysis of Hospital Mergers and the Transformation of the Hospital Industry, 51 LAW & CONTEMP. PROBS. 93, 101-02 & n.42 (1988);
31. ANTITRUST LAW DEVELOPMENTS, supra note 3, at 553-56.
32. Accord RICHARD A. POSNER, ANTITRUST LAW 148 (2d ed. 2001).
33. Federal Trade Comm'n & U.S. Dep't of Justice, Commentary on the Horizontal Merger Guidelines 5 (2006) [hereinafter Guidelines Commentary].
34. For a similar reason, the significance for competitive effects of demand complementarities should not be accounted for in the market definition step of the analytical process. Contra, David S. Evans & Michael D. Noel, Analyzing Market Definition and Power in Multisided Platform Markets (Working Paper Oct. 2005), available at http://ssrn.com/abstract= 835504 (proposing to account for demand complementarities associated with "two-sided" businesses through market definition);
35. Renata B. Hesse & Joshua H. Soven, Defining Relevant Product Markets in Electronic Payment Network Antitrust Cases, 73 ANTITRUST L.J. 709, 728, 732 (2006) (explaining that the Justice Department proposed to account for reactions by the other side of a two-sided market when defining the product markets in a recent case involving electronic payment systems). Rather, the significance of demand complementarities should be accounted for in the later analysis of competitive effects. See Baker, supra note 27, at 123-40 (discussing competitive significance of the possibility that merging firms also sell demand complements).
36. Any reasonable method of analyzing the competitive effects of this hypothetical merger would need to account for the extent to which aluminum conductor capacity would profitably be diverted to copper conductor production in the event that the price of copper conductor were to rise. Under the Merger Guidelines, that analysis would take place in assigning market shares rather than in market definition.
37. The possible absence of a competitive constraint from nonmerging copper conductor producers unable to expand output inexpensively would be addressed in the analysis of competitive effects, regardless of whether markets are defined from a pure demand-side perspective or whether supply substitution is incorporated into that process.
38. Incumbent aluminum conductor producers may switch into copper conductor production, and new entrants may begin to produce aluminum conductor. But this supply response may not be sufficient to prevent an increase in copper conductor prices because of the relatively small increase in copper conductor output.
39. This kind of problem is more than a theoretical concern: a similar analytical step appears to have been skipped improperly by an expert antitrust tribunal evaluating exclusionary conduct allegations in the Seventh Circuit's decision in Menasha Corp. v. News America Marketing In-Store, Inc., 354 F.3d 661 (7th Cir. 2004) (Easterbrook, J.). The principal question in the Menasha case was whether the product market should be limited to atshelf coupon dispensers (attached to grocery store shelves), giving defendants a large market share. The circuit court expanded the market to reflect supply expansion by providers of "signs and placards, end caps (product racks at the end of aisles), sales, coupons included on (or in) the product's package, coupons distributed at the checkout counter, and . . . the traditional coupons distributed by mail or newspaper." Id. at 664. In the resulting broader market, defendants apparendy had a low market share. But the court never asked whether buyers (product manufacturers) would in fact substitute these forms of promotion for at-shelf dispensers in response to higher dispenser prices, and would do so in sufficient numbers to prevent any exercise of market power among sellers of at-shelf dispensers.
40. The influence of the demand-side focus of the Merger Guidelines on the courts is described in Werden, The 1982 Merger Guidelines, supra note 22.
41. ANTITRUST LAW DEVELOPMENTS, supra note 3, at 77.
42. ANDREW I. GAVIL, WILLIAM E. KOVACIC & JONATHAN B. BAKER, ANTITRUST LAW IN PERSPECTIVE: CASES, CONCEPTS AND PROBLEMS IN COMPETITION POLICY 198-99 (2002).
43. The demand elasticity at issue in market definition is an own-price elasticity, as discussed below in notes 49 and 134. Cross-price elasticities of demand are relevant because they can be used to identify the next-best substitutes that should be included if the candidate market is expanded and because of the mathematical relationship between own- and cross-elasticities set forth below in note 134.
44. Methods of inferring the magnitude of buyer substitution from natural experiments involving a change in market structure have been used to analyze the unilateral competitive effects of merger among sellers of differentiated products. Jonathan B. Baker, Econometric Analysis in FTC v. Staples, 18 J. PUB. POL'Y & MKTG. 11 (1999). These methods are equally appropriate for the analysis of market definition, as buyer substitution is the central economic force at issue in both inquiries.
45. This is an example of what is termed in econometrics the problem of identification. See generally Jonathan B. Baker & Timothy F. Bresnahan, Economic Evidence in Antitrust: Defining Markets and Measuring Market Power (Stanford Law and Economics Olin Working Paper No. 328, Sept. 2006), available at http://ssrn.com/abstract=931225.
46. See, e.g., Jonathan B. Baker & Daniel L. Rubinfeld, Empirical Methods in Antitrust Litigation: Review and Critique, 1 AM. L. & ECON. REV. 386, 424-27 (1999) (discussing conjoint survey methods).
47. When investigating a merger involving producers of an intermediate good sold to a limited number of industrial buyers, the enforcement agencies frequently seek to interview all the buyers about their willingness to substitute away from the product in the event of a price increase.
48. The reliability of customer statements unsupported by analysis as to buyer inability to substitute away from the products in a candidate market in response to a price increase was questioned in United States v. Oracle Corp., 331 F. Supp. 2d 1098, 1131 (N.D. Cal. 2004). The court refused to consider customer views when defining markets and evaluating the unilateral effects of merger on the ground that those views were not based on an expensive new analysis that would not routinely be undertaken in the ordinary course of business. This conclusion takes an overly narrow view of the possible evidentiary bases for crediting customer views as informed and reliable.
49. Similarly, informal analysis of switching costs may bear on product market definition. See, e.g., Bayer AG, C-No. 4049, 2002 WL 1151015 (FTC May 30, 2002) Complaint ¶ 11 (defining a market for "new generation" chemical insecticides based in part on an informal switching costs analysis of the advantages to buyers of new-generation products relative to older insecticides).
50. The probative value of inferences about the extent of buyer substitution from another aspect of seller behavior, the way sellers set prices relative to costs, is discussed in connection with critical loss analysis in Part IV.C below.
51. But seller documents can raise issues of interpretation. For example, if competitors are numerous, sellers may find it cost-effective to sample representative rivals rather than monitoring all of the rivals that matter to them. Similarly, sellers may devote more attention to monitoring actual rivals than potential rivals, even if the latter also help constrain seller pricing. Also references in seller documents to the "market" may not be useful for antitrust market definition when executives or consultants define markets for non-antitrust purposes using criteria that differ from those that apply to antitrust market definition.
52. For additional discussion, see Baker & Bresnahan, supra note 40.
53. This paragraph simply makes a conceptual point and should not be regarded as endorsing the controversial "critical loss" approach to market definition, which is discussed in Part IV.C below.
54. The first-order condition for profit maximization by a single-product monopolist that does not discriminate among buyers can be written as equating the absolute value of the inverse of the elasticity of the demand function facing the monopolist with the seller's Lerner Index of price-cost margin (price less marginal cost, as a fraction of price), or 1/ε = L. In consequence, a price increase is profitable for a hypothetical monopolist if and only if the inverse elasticity of demand exceeds the Lerner Index (1/ε > L). This equation (or related formulae) can, in theory, be used to define markets by simulating whether a hypothetical monopolist would raise price, by estimating and comparing the two sides of the inequality. See generally Gregory J. Werden, Demand Elasticities in Antitrust Analysis, 66 ANTITRUST L.J. 363 (1998). It is worth noting that changes in the Lerner Index over time (for example, an increase in price-cost margins after the alleged violation occurred) are not informative as to whether current margins correspond to the elasticity of firm demand without information about the way the distribution of buyer willingness to pay, and thus the demand elasticity, changed over the same time period.
55. Complications in defining markets such as these are considered in one industry setting in Randy Brenkers & Frank Verboven, Market Definition with Differentiated Products - Lessons from the Car Market (Centre for Economic Policy Research Discussion Paper No. 5249, Sept. 2005), available at http://www.cepr.org/pubs/dps/DP5249.asp.
56. For example, it may be difficult to determine how much a hypothetical monopolist would raise price, and, thus, whether price would rise by more than the SSNIP, if there is reason to believe that the elasticity of demand varies as price increases. This possibility is highlighted in Luke M. Froeb & Gregory J. Werden, Residual Demand Estimation for Market Delineation: Complications and Limitations, 6 REV. INDUS. ORG. 33 (1991). This issue is discussed below in connection with the Cellophane fallacy in Part V.B, and in connection with critical loss analysis in Part IV.C.
57. When using the relationship 1/ε = L to evaluate a proposed market definition, the demand elasticity parameter summarizes all the available evidence as to buyer substitution. There is no presumption that it is estimated econometrically, though it could be.
58. Marginal cost is not necessarily the price that would obtain absent anticompetitive conduct, which for this purpose may be understood as the competitive price. The competitive price could exceed marginal cost, for example, if products or locations are differentiated, firms engage in competitive price discrimination, or the sellers engage in oligopoly conduct without reaching an agreement or otherwise violating the antitrust laws. If the competitive price were known, that would permit direct proof of the existence or absence of market power, obviating the need for market definition.
59. See discussion infra Part IV.C. Used in this way, this framework has similar strengths and weaknesses as simulation modeling generally. For example, it can help develop intuition and judgment as to market definition by providing a metric for understanding the significance of demand elasticities, and it can help identify areas where collection of additional information would be useful in pinning down whether a proposed market should be defined. But it may also incorporate a number of assumptions that are difficult to test or modify, making the results unhelpful or misleading. Particular difficulties associated with making inferences as to the magnitude of buyer substitution from price-cost margins are addressed below in Part IV.C.
60. Pitofsky, supra note 1, at 1807.
61. "There is little question that relevant market definition was a more coherent exercise during the 1980s than in previous decades, and that can be attributed in part to the orderly, intellectual approach of the Guidelines." Id. at 1808. See also Gregory J. Werden, Market Delineation Algorithms Based on the Hypothetical Monopolist Paradigm (Working Paper July 27, 2002), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=327282 (proposing a formal mathematical statement of the algorithm).
62. The Merger Guidelines suggest considering in most contexts a price increase of 5% lasting for the foreseeable future. Merger Guidelines, supra note 5, § 1.11. The enforcement agencies frequently instead employ a 10% figure in this role. Robert Pitofsky has criticized the use of the higher figure as encouraging overinclusive market definitions. Pitofsky, supra note 1, at 1823-24.
63. Cf. Malcolm B. Coate & Jeffrey H. Fischer, A Practical Guide to the Hypothetical Monopolist Test for Market Definition 6 (Potomac Papers in Law and Economics Working Paper 06-01, Oct. 2006), available at http://ssrn.com/abstract=940667 ("If the [hypothetical] price increase is profitable, then the gap [in the chain of substitutes] is material and a relevant market is defined. If not, simply find another gap and test again.").
64. Other aspects of the approach to market definition may differ across these settings, particularly depending on whether the alleged harmful conduct was prospective or retrospective, as discussed infra in Parts V and VI.
65. The Merger Guidelines' approach of starting with finely defined candidate markets and expanding them is often difficult to apply in practice, in part, because it is often impossible to understand buyer substitution patterns (including the elasticity of candidate product demand and the identification of next-best substitutes) for finely defined products with sufficient precision to make the necessary judgments.
66. In principle, the starting point could be disaggregated more finely than, for example, regular Coke sold in a 12 oz. can. One could view regular Coke as the combination of flavored syrup manufactured by the Coca-Cola Company, along with complementary products purchased by Coca-Cola, including bottling services, distribution services, and marketing. Similarly, one might ask, why begin market definition in a case involving automobiles with a specific brand and model of car? After all, the car purchased by a retail customer can be disaggregated further: it combines a chassis, tires, an engine, and a radio, among other things, along with the distribution and marketing services provided by a dealer. Shouldn't market definition begin with each of these? Viewed this way, virtually all products can be thought of as combinations of demand complements (which have presumably been bundled or assembled because sellers can do so more cheaply than can buyers). Hence, this perspective would require that market definition commence with numerous candidate markets (e.g., the chassis, tires, engine, etc.) in any individual case. This perspective is not appropriate to antitrust market definition, however, because it is not tied to the buyers who would be harmed by the alleged antitrust violation and so conflicts with the demand-substitution focus of market definition.
67. Products should generally not be disaggregated more finely than what is purchased by the particular buyers alleged to have been harmed by the conduct under review. Cf. Werden, supra note 56, at 25 (observing that market delineation algorithms can be applied to bundles of products that arise as a result of "technological integration, combined packaging, or contractual tying"). An exception might arise if a plaintiff alleged that the conduct under review harmed competition by leading defendants to alter the attributes of the products they sell (for example, in order to raise entry barriers or discriminate in price in a manner that harms consumers). Other examples of the importance of relating market definition to the allegations in the case are considered infra Parts V and VI.
68. Some of the five types of evidence as to buyer substitution discussed in Part III.A. above can be used to calibrate the magnitude of the likely buyer response to a specific percentage price increase more closely than can others. For example, evidence as to which rival products are monitored and responded to by sellers may not correspond readily to a particular percentage price increase or suggest a precise percentage for the buyer response, yet this evidence could nevertheless be strongly probative as to market definition. The quantitative aspects of the market definition approach of the Merger Guidelines provide conceptual clarity, but they do not mandate a systematic preference for quantitative evidence.
69. The scope of the market defined using this algorthim could vary with the size of the SSNIP chosen. Other reasons multiple markets could be defined from a single starting point are discussed below in the next note and in Part III.C.3. Moreover, the actual price increase thought likely to occur as a result of challenged conduct could be greater than or less than the benchmark SSNIP used for market definition. In theory, harm to competition could be found even if only a small price increase is thought likely. In practice, the difficulty of distinguishing a small number from zero with the available evidence, the Merger Guidelines' safe harbors (which discourage government enforcement when market concentration is low even if some harm to competition could be demonstrated), and the resource constraints facing government enforcers and other plaintiffs, combine to make unlikely a challenge to a merger that would increase price by a minute amount.
70. In the context of implementing the conceptual experiment for market definition, a price increase is simply a measurable proxy for any small change (whether on a price or nonprice dimension) that makes what is sold in the candidate market less attractive to buyers. Cf. Michael L. Katz & Howard A. Shelanski, Mergers and Innovation, 74 ANTITRUST L.J. 1, 36 (2007) ("What is needed is a careful analysis of what constitutes a real price increase in the face of cost and product changes."). If market definition is based on a small reduction on a nonprice quality dimension, the resulting market could in theory differ from what would result if the conceptual experiment involves a small increase in price. This could happen if buyers respond differently to small changes in product attributes than to small changes in price or if the product or location constituting the next-best substitute (the basis for expanding a candidate market) differ as the conceptual experiment changes. This observation provides another reason why the process of market definition could result in multiple markets within which the possibility of anticompetitive effects could be analyzed.
71. Hence, market definition turns on the own-price elasticity of structural demand. Before 1992, market definition was generally thought to turn on the own-price elasticity of residual demand. The residual demand elasticity, which incorporates supply responses as well as demand responses, remains relevant to the assessment of competitive effects of the conduct under review (possibly making market definition and evidence involving the structural demand elasticity unnecessary). For a recent effort to define markets using an approach that accounts for both demand and supply responses, see Mark Ivaldi & Szabolcs Lörincz, A Full Equilibrium Relevant Market Test: Application to Computer Servers (Working Paper Aug. 2005), available at http://ssrn.com/abstract=743144).
72. This example presumes that the smaller market is expanded through the addition of demand substitutes. At a minimum, the hypothetical monopolist of all soft drinks can profitably raise price for the cola-flavored drinks alone.
73. Moreover, markets could be nested. Suppose product markets must be defined to analyze Coke's acquisition of Pepsi. It is conceivable that starting from Coke, one would define a product market including Coke and Dr Pepper only, while starting with Pepsi, one would need to add first Coke and then Dr Pepper before the market definition algorithm generates a product market. Under such circumstances, competitive effects should be analyzed in both a Coke/Dr Pepper product market and a Pepsi/Coke/Dr Pepper product market. Similarly, to provide another example, in analyzing an allegation of exclusionary conduct involving computer monitors, it is possible that competitive effects should be analyzed in both a monitors market and a broader market for personal computers.
74. Cf. Pitofsky, supra note 1, at 1812-13 ("The tendency to see relevant market definition as an all-or-nothing proposition rather than as an array of estimates with no market description being exactly right has led to the most serious errors in antitrust enforcement").
75. Cf. Katz & Shelanski, supra note 65, at 35 (recommending that plaintiffs "not be held to a standard of establishing a unique, bright-line boundary with a high degree of certainty").
76. This problem is only imperfectly addressed by the discretion the Merger Guidelines provide to vary the SSNIP. Cf. Guidelines Commentary, supra note 29, at 8 ("At times, the Agencies may act conservatively and focus on a market definition that might not be the smallest possible relevant market. . . . If the answer for [a] broader market is likely to be the same as for any plausible smaller relevant market, there is no need to pinpoint the smallest market as the precise line drawn does not affect the determination of whether a merger is anticompetitive."). For further discussion of the shortcomings of the "smallest market principle" and related issues, see Jonathan B. Baker & Steven C. Salop, Should Concentration Be Dropped from the Merger Guidelines?, in ABA SECTION OF ANTITRUST LAW TASK FORCE, PERSPECTIVES ON FUNDAMENTAL ANTITRUST THEORY 339 (2001).
77. For a similar reason, it is often reasonable in practice to expand a candidate market to include a good substitute, regardless of whether the substitute is literally the "next-best" substitute, the criterion suggested by the Horizontal Merger Guidelines. Merger Guidelines, supra note 5, § 1.11.
78. Guidelines Commentary, supra note 29, at 6-7 (citing as examples markets defined as "superpremium ice cream," paper labelstock used for "variable information printing," paper labelstock used for "prime" labels, and services provided by full-service hospitals in Slidell, La.).
79. Brown Shoe Co. v. United States, 370 U.S. 294, 325 (1962).
80. See generally Jonathan B. Baker, Stepping Out in an Old Brown Shoe: In Qualified Praise of Submarkets, 68 ANTITRUST L.J. 203, 206 (2000); Pitofsky, supra note 1, at 1817 (decrying the "free and easy tendency in the cases to carve out scores and even hundreds of submarkets" during the 1960s and early 1970s).
81. Narrow markets that cannot be described absent multiple adjectives are often ridiculed as gerrymandered - carefully crafted in order to make concentration appear high, rather than defined on a principled basis. See, e.g., United States v. Grinnell Corp., 384 U.S. 563, 591 (1966) (Fortas, J. dissenting) (highlighting "unreality" of the "central station protective services which are 'accredited' by insurance companies" market found the by majority by calling it a "strange red-haired, bearded, one-eyed man-with-a-limp classification"). But the number of adjectives is beside the point: the issue is whether the market definition is consistent with the evidence as to demand substitution.
82. See infra Part III.D.
83. The economic force of buyer substitution underlies the analysis of unilateral competitive effects among mergers among sellers of differentiated products, much as it underlies the analysis of market definition. The advantages and disadvantages of using submarkets to evaluate unilateral effects are explored in Baker, supra note 75, at 209-17.
84. Compare United States v. Pabst Brewing Co., 384 U.S. 546, 552 (1966) (three-state market including Wisconsin, Illinois, and Michigan), with Christian Schmidt Brewing Co. v. G. Heileman Brewing Co., 753 F. 2d 1354, 1358 (6th Cir. 1985) (12-state upper Midwest market not clearly erroneous; defense would be entitled to prove a broader national market in a later trial on the merits). Moreover, the scope of a market may change over time as buyer demand alters. Suppose, for example, that widgets were originally sold primarily to hobbyists, whose purchases are largely insensitive to price. Widgets would then likely constitute an antitrust product market for analyzing a merger between widget producers. Over time, assume that widget sellers improve the product and lower its cost and price, attracting a large group of new buyers
85. In United States v. Aluminum Co. of America, 148 F.2d 416 (2d Cir. 1945), a famous monopolization case, Judge Learned Hand defined a primary aluminum product market, excluding secondary (recycled) aluminum. Hand's analysis spawned an economics literature pointing out that a substantial market in used products could, in principle, constrain the price of a new durable good but that a new-product monopolist could offset this competitive force to a significant degree by reducing new production in order to limit the future supply of the used product. See generally GAVIL ET AL., supra note 37, at 578-79 (reviewing economic literature). Whether or not this supply-side force is important (and later empirical economic research concluded that it was not in the Alcoa case because secondary aluminum was only an imperfect substitute to buyers of virgin aluminum), this dynamic is more appropriately evaluated in competitive effects analysis than in market definition, which focuses on buyer substitution not seller behavior.
86. Merger Guidelines, supra note 5, § 1.12. The Merger Guidelines also recognize this possibility when they indicate that the hypothetical monopolist may maximize profits by raising price non-uniformly throughout the candidate market. Id. § 1.12.
87. If sellers cannot reliably separate out a group of relatively loyal customers, the profitability of a price increase would depend on the relative number of loyal customers and switchers in the customer pool as a whole.
88. Accordingly, when defining price discrimination markets, the federal enforcement agencies inquire as to the ability to prevent arbitrage as part of the process of market definition. Guidelines Commentary, supra note 29, at 7.
89. In one case, involving customized products sold through bidding mechanisms, the Justice Department viewed each individual procurement as a separate price discrimination market. Id. at 8.
90. See generally Baker, supra note 75, at 207-09.
91. For a recent defense of these methods, see Mario Forni, Using Stationarity Tests in Antitrust Market Definition, 6 AM. L. & ECON. REV. 441 (2004).
92. See also George J. Stigler & Robert A. Sherwin, The Extent of the Market, 28 J.L. & ECON. 555 (1985). The use of price comparisons to define markets appears to have greater acceptance in European competition practice than in the United States.
93. These methods are not derived from an arbitrage model, however. Instead, they test for pricing similarities, which might result from arbitrage. For an example of a study that exploits a model of arbitrage to estimate the transactions costs to arbitrage among a set of homogeneous products at different locations, as well as the frequency that arbitrage conditions bind, see Pablo T. Spiller & Cliff J. Huang, On the Extent of the Market: Wholesale Gasoline in the Northeastern United Stales, 35 J. INDUS. ECON. 131 (1986). Spiller and Huang conclude that the "economic markets" they identify based on arbitrage did not track the markets that would be defined for antitrust analysis.
94. However, if evidence not involving product price points demonstrates that buyers of some particular high-price/high-quality good would not switch to a low-price/low-quality alternative were the high-quality product to increase in price, the high-quality product might constitute a product market, and the bounds of that market might conveniently be described in terms of goods priced above a certain level. See United States v. Gillette Co., 828 F. Supp. 78, 81-83 (D.D.C. 1993) (defining market for premium writing instruments priced between $50 and $400).
95. See generally Daniel Hosken & Christopher T. Taylor, Discussion of "Using Stationarity Tests in Antitrust Market Definition," 6 AM. L. & ECON. REV. 465 (2004);
96. 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);
97. Jonathan B. Baker, Why Price Correlations Do Not Define Antitrust Markets: On Econometric Algorithms for Market Definition (Federal Trade Comm'n Bureau of Econ. Working Paper No. 149, 1987).
98. When demand elasticities are estimated using time series data, it may be necessary to test for stationarity in order to properly specify and estimate the demand model, however. For a recent example, see John Hayes, Carl Shapiro & Robert Town, Market Definition in Crude Oil: Estimating the Effects of the BP/ARCO Merger (Working Paper Aug. 2001), available at http://www.ftc.gov/bc/gasconf/comments2/oilpaperjohnhayesetal.pdf.
99. Kenneth G. Elzinga & Thomas F. Hogarty, The Problem of Geographic Market Definition in Antitmerger Suits, 18 ANTITRUST BULL. 45 (1973);
100. Kenneth G. Elzinga & Thomas F. Hogarty, The Problem of Geographic Market Definition Revisited: The Case of Coal, 23 ANTITRUST BULL. 1 (1978).
101. Elzinga and Hogarty variously proposed that 75% and 90% of shipments satisfy these tests in order to define a geographic market. In principle, an analogous approach could be employed to define product markets - as a group of products for which buyers purchasing in the product category rarely go outside the set - but in application the approach has been limited to geographic market definition.
102. For criticisms, see, e.g., Gregory J. Werden, The Use and Misuse of Shipments Data in Defining Geographic Markets, 26 ANTITRUST BULL. 719 (1981);
103. Janusz A. Ordover & Robert D. Willig, The 1982 Department of Justice Merger Guidelines: An Economic Assessment, 71 CAL. L. REV. 535, 551-52 (1983); Baker, supra note 27, at 143-48.
104. In terms of the classification scheme for types of evidence set forth in Part III.A, this adjustments can be understood as using information about the characteristics of products and locations known to matter to buyers in order to assess likely buyer substitution in response to an increase in price.
105. See generally Barry C. Harris & Joseph J. Simons, Focusing Market Definition: How Much Substitution Is Enough, 12 RES. IN L. & ECON. 207 (Richard O. Zerbe, Jr. ed., 1989);
106. Frederick I. Johnson, Market Definition Under the Merger Guidelines: Critical Demand Elasticities, 12 RES. IN L. & ECON. 235 (Richard O. Zerbe, Jr. ed., 1989).
107. Alternatively, using the same information in a different way, one might ask whether the demand facing a hypothetical monopolist was so responsive to price - whether the demand elasticity exceeds a benchmark "critical elasticity" level - as to make a price increase unprofitable given accounting price-cost margins.
108. Daniel P. O'Brien & Abraham L. Wickelgren, A Critical Analysis of Critical Loss Analysis, 71 ANTITRUST L.J. 161 (2003);
109. Michael L. Katz & Carl Shapiro, Critical Loss: Let's Tell the Whole Story, ANTITRUST, Spring 2003, at 49;
110. James Langenfeld & Wenqing Li, Critical Loss Analysis in Evaluating Mergers, 46 ANTITRUST BULL. 299 (2001);
111. Dennis W. Carlton, Using Economics to Improve Antitrust Policy, 2004 COLUM. BUS. L. REV. 283 (2004).
112. The Lerner Index equals 3/4. Under the assumptions that these are single-product producers and that prices today do not affect demand tomorrow, each firm acts as though its elasticity of current period residual demand is 4/3.
113. This is the economic point implicit in the claim that when price-cost margins are high, it takes few lost sales to make a price increase unprofitable, so if lost sales exceeded that small critical value, the antitrust market would be broad.
114. See the discussion of "kinks" in demand (by which quantity is more sensitive to a price increase than a price decrease) in Katz & Shapiro, supra note 97, at 49, 52, and O'Brien & Wickelgren, supra note 97, at 178.
115. See generally Gregory J. Werden, Beyond Critical Loss: Tailoring Applications of the Hypothetical Monopolist Paradigm, 4 COMPETITION L.J. 69 (2005);
116. Kenneth L. Danger & H.E. Freeh III, Critical Thinking About "Critical Loss" in Antitrust, 46 ANTITRUST BULL. 339 (2001);
117. James Langenfeld & Wenqing Li, Critical Loss Analysis in Evaluating Mergers, 46 ANTITRUST BULL. 299 (2001).
118. United States v. Philadelphia Nat'l Bank, 374 U.S. 321 (1963).
119. See generally ABA SECTION OF ANTITRUST LAW, MERGERS AND ACQUISITIONS: UNDERSTANDING THE ANTITRUST ISSUES 60-63 (2d ed. 2004) (surveying cases in which cluster markets have been defined and noting cases where courts have declined to do so). Courts have defined cluster markets to analyze hospital mergers even though the individual services that comprise the cluster are commonly recognized as relevant product markets in nonmerger antitrust cases involving medical services provided by hospitals. See Baker, supra note 27, at 124 n.161 (collecting nonmerger hospital services cases in which cluster markets were not defined).
120. Guidelines Commentary, supra note 29, at 9; Baker, supra note 27 at 123-40.
121. See supra Part III.C.1. Consistent with this perspective, a cluster market may be consistent with a focus on demand substitution when sellers offer buyers substantial transactions cost savings from one-stop shopping. See Ian Ayres, Rationalizing Antitrust Cluster Markets, 95 YALE L.J. 109 (1985).
122. Market definition would begin with the products each group of buyers, corporate and mass market, purchases (suites for one group, components for the other) and would expand each of those multiple candidate product markets only if a higher price would lead to substantial buyer substitution.
123. The fact that some multi-product sellers produce demand complements (the other programs in the suite) would then be taken into account in the analysis of competitive effects within the individual component markets.
124. See generally Steven C. Salop, The First Principles Approach to Antitrust, Kodak, and Antitrust at the Millennium, 68 ANTITRUST L.J. 187 (2000). Cf. Guidelines Commentary, supra note 29, at 5 (the market definition process "is not isolated from the other analytic components in the Guidelines").
125. Merger analysis at the enforcement agencies, thus, generally grandfathers-in any prior exercise of market power. However, the Merger Guidelines allow the premerger price to reflect the competitive price if there is evidence of coordination premerger. Merger Guidelines, supra note 5, § 1.11. This provision can be understood as adopting an implicit presumption that the premerger coordination would break down soon, at least with probability, so that the appropriate but-for price is less than the premerger price.
126. Under such circumstances, a mere continuation of the prevailing price by a hypothetical monopolist would reflect the exercise of market power, and a candidate market should not be expanded even if prices would likely not rise further. Salop, supra note 108, at 197 (terming the failure to recognize this situation the "Price-Up Trap"). Market definition in exclusion cases is discussed in Part VI. Market definition in rapidly innovating industries is discussed in Katz & Shelanski, supra note 65, at 36-37 (recommending use of forward-looking quality-adjusted prices for market definition).
127. The agreement might involve standardizing product definitions or posting prices, for example. This example is framed as an agreement on a facilitating practice rather than as an agreement concerning price in order to make it plausible that a court would analyze the conduct under the rule of reason (rather than a per se rule) and, thus, that the court would define a market as part of its analysis.
128. The firms may have experienced increased input prices, or higher costs of operating at the greater scale that was required to meet an increase in industry demand, for example.
129. These are "defenses" in the casual sense of explanations, not in a technical legal sense of justifications for what would otherwise constitute a violation.
130. It is useful to separate out these defenses, although they are not mutually exclusive. There is no reason, moreover, that a court must analyze all claims and defenses raised in a case in the same market, as discussed supra in Part III.C.3. Accordingly, if a broader market is adopted for the purpose of analyzing some issues, that does not preclude the possibility that a more narrow market would be appropriate for analyzing other questions, and vice versa.
131. This inference would be less plausible to the extent that buyer substitution patterns have changed, for example, because of changes in the available mix of products, changes in buyer income, or changes in buyer tastes. Under such circumstances, what is learned about buyer substitution patterns at the price of 8 one year ago would need to be modified to account for those changes in order to make inferences about buyer substitution patterns at the but-for price today.
132. Some types of evidence about buyer substitution then may be available for a court to consider now. For example, it may be possible to obtain information about how buyers responded to even earlier price changes, or seller documents from one year ago may be available identifying which rivals they monitored and responded to at that time. But the longer ago the alleged conspiracy began, the more difficulty it may be to obtain such evidence.
133. This well-known difficulty is named for a conceptual error made by the Supreme Court in defining the product market in United States v. E.I. du Pont de Nemours & Co., 351 U.S. 377 (1956) (Cellophane).
134. In any case, independent knowledge of the competitive price would likely permit proof or disproof of harm to competition by direct evidence, making market definition unnecessary.
135. A similar approach is proposed in Philip B. Nelson & Lawrence J. White, Market Definition and the Identification of Market Power in Monopolization Cases: A Critique and a Proposal (NYU Center for Law & Business Working Paper No. CLB-03-022 Nov. 2003), available at http://w4.stern.nyu.edu/emplibrary/03-022.pdf. This approach is equivalent to starting with the current output and asking whether a substantial increase in supply within the candidate market would not or would lead to only a small reduction in price.
136. That is, it is still open for the plaintiff to argue that demand would have been relatively inelastic at the lower price that allegedly would have obtained absent anticompetitive conduct, making the alleged exercise of market power profitable, notwithstanding that demand is somewhat more elastic at a price near but slighdy lower than the current price. If so, the price reduction approach may reduce the Cellophane fallacy problem without avoiding it entirely.
137. Again, buyer substitution today may be measured by assuming a hypothetical price reduction, to reduce the Cellophane fallacy problem.
138. Perhaps firm A's conduct or practices would restrict the ability of firm B to obtain key inputs or obtain access to customers, or they would lead to the destruction of firm B's plant.
139. Firm B might claim that firm A had harmed competition through monopolization, exclusive dealing, tying, group boycott, or agreement (possibly vertical) or merger, for example; the specific antitrust violation alleged would depend upon whether the necessary legal elements are met. It is assumed that firm B alleges that it was harmed by the conduct that allegedly harmed competition, leading to higher prices, and, thus, that the injury alleged by firm B would count as antitrust injury.
140. Again, these are "defenses" in the casual sense of explanations, not in a technical legal sense of justifications for what would otherwise constitute a violation. It is useful to separate them, although they are not mutually exclusive. They may also be related, in that the less significant a competitive constraint firm B poses, the easier it may be for firm C to replace the lost competition. As previously noted, there is no reason that a court must analyze all claims and defenses in the same market.
141. If firm A also points in defense to the ability and incentive of firm C to replace any competition lost from the exit of firm B, a court may also analyze whether a candidate product market consisting of products A and C should be expanded to include product B.
142. The implication for market definition is similar if firm A argues that firm B had the ability but not the incentive to compete aggressively with it, as the nature of the oligopoly interaction is also a supply-side issue. This claim about firm B's incentive may be an unattractive argument in the context of the hypothetical example, however, as it is tantamount to suggesting that firm A should be allowed to exclude firm B on the ground that the two were already colluding tacitly, so that there would be no loss of competition if firm B were to exit.
143. This question may also be analyzed within a market consisting of products A, B, and C, if competition from firm C represents an important constraint on firm A's pricing.
144. Similarly, suppose A is a joint venture rather than a single firm, and the plaintiffs are a group of potential members B which claim that the joint venture will harm competition by excluding them. Suppose further that the venture defends by arguing that the alleged exclusionary conduct could not matter because the joint venture was constrained not to raise price by firms selling product C. The market definition question relevant to evaluating the allegation and defense is whether the relevant antitrust market includes the products and locations of both the members of the joint venture and the potential members (to include the products of firms A and B), or whether it must expanded further (also to include the product of firms C). See Dennis W. Carlton & Steven C. Salop, You Keep on Knocking but You Can't Come in: Evaluating Restrictions on Access to Input Joint Ventures, 9 HARV. J.L. & TECH. 319, 346-47 (2006).
145. If firm B defends by claiming that firm C would replace any lost competition, the magnitude of buyer substitution from products A and B to product C is a relevant consideration because it affects the size of the competitive response by firm C that would be necessary. This buyer substitution issue would be treated as part of the competitive effects analysis, along with the supply-side question of the extent to which firm C expanded output, in much the way that the buyer substitution issue of unilateral effects of merger among sellers of differentiated products is treated as a competitive effects issue rather than a market definition question.
146. Again, these defenses are not mutually exclusive, and the fact that a court adopts a market for the purpose of analyzing some issues does not preclude the possibility that a different market, possibly broader or more narrow, would be appropriate for analyzing other questions.
147. Accordingly, direct evidence on whether price changes are fully explained by shifts in cost and demand, without the need to postulate an increase in market power, could, in principle, discriminate between plaintiff's theory of the case and the defense theory. In an appropriate case, therefore, direct evidence of this nature could resolve the case without need for market definition. The discussion in the text presumes that the available direct evidence is not strongly probative in the case at issue and that it is also worth analyzing the evidence concerning market structure.
148. As should be evident, a complete economic analysis of the exclusion allegation would not be limited to market definition because market definition and the market shares that result do not necessarily determine fully whether the alleged conduct harms competition. Regardless of the scope of the market and the market shares of the firms, a range of other economic considerations not involving demand substitution would likely be relevant to identifying the competitive effects of the challenged conduct. These may include whether firm B actually was excluded; whether firm A's conduct was responsible; whether the exclusion of firm B harmed competition by conferring additional market power on firm A; whether benefits to competition from the exclusionary conduct outweigh the harms; and whether more aggressive competition from firm C replaced any competition lost from the exclusion of firm A. This essay does not address the legal question of whether truncated doctrinal rules, possibly based on market shares, should or do limit consideration of some or all of these issues in the courts.
149. In the exclusion context, this methodology may be more challenging to implement than the corresponding question in a prospective harm case. If firms A and B sell differentiated (rather than homogeneous) products and if firm B has exited, it may be necessary to assess the effects of buyer decision making of two simultaneous hypothetical changes in their economic environment, including one (the exit of firm B) that is not a marginal change, by which both prices and products differ from those presently available. If firm A has changed its product offerings or locations in response to the exit of firm B, it may also be necessary to assess the effects of a third simultaneous hypothetical change in the economic environment.
150. c are the shares of buyer expenditures in the product category (the three products as a group). (This equation is derived by differentiating total expenditures with respect to the price of product A, holding expenditures constant. Expenditure shares are based on price times quantity, not simply quantity.)
151. Continuing the example of the previous note, if the own-price elasticity ε measured post-exclusion is used as an estimator of its pre-exclusion value (notwithstanding the Cellophane fallacy problem), then whether product A is defined as a market would turn on a comparison between that elasticity and the Lerner Index measured at the previous (pre-exclusion) price and marginal cost.