Why (ever) define markets

Harvard Law Review

Volumn 124, Issue 2, 2010, Pages 437-517

  Kaplow, Louis ^a  

^a HARVARD LAW SCHOOL   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

EID: 79851487318     PISSN: 0017811X     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Review

References (360)

1. See, e.g., 1 ABA SECTION OF ANTITRUST LAW, ANTITRUST LAW DEVELOPMENTS 550-54 (Jonathan M. Jacobson et al. eds., 6th ed. 2007) [hereinafter ANTITRUST LAW DEVELOPMENTS]
2. ABA SECTION OF ANTITRUST LAW, MARKET POWER HANDBOOK 5-9, ch. 2 (2005) [hereinafter MARKET POWER HANDBOOK]
3. DG Competition Discussion Paper on the Application of Article 82 of the Treaty to Exclusionary Abuses ¶ 23 (Dec. 2005) [hereinafter DG Competition Article 82], available at http://ec.europa.eu/competition/antitrust/art82/ discpaper2005.pdf.
4. See, e.g., 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.")
5. ANTITRUST LAW DEVELOPMENTS, supra note 1, at 549 ("Defining a relevant market is often a critical issue, and sometimes the critical issue, in an antitrust case.")
6. 2B PHILLIP E. AREEDA, HERBERT HOVENKAMP & JOHN L. SOLOW, ANTITRUST LAW 135 (3d ed. 2007) ("In resolving market or 'monopoly' power issues, the courts have typically relied heavily on market definition and on the defendant firm's share of the market thus defined.")
7. Jonathan B. Baker, Market Definition: An Analytical Overview, 74 ANTITRUST L.J. 129, 129 (2007) ("Throughout the history of U.S. antitrust litigation, the outcome of more cases has surely turned on market definition than on any other substantive issue.")
8. 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."). For the European Union
9. see, for example, Commission Notice on the Definition of Relevant Market for Purposes of Community Competition Law, 1997 O.J. (C 372) ¶ 4 [hereinafter Commission Notice on Market Definition], which notes that "[t]he definition of the relevant market in both its product and its geographic dimensions often has a decisive influence on the assessment of a competition case."
10. See also Thomas E. Kauper, The Problem of Market Definition Under EC Competition Law, 20 FORDHAM INT'L L.J. 1682 (1997) (comparing the approaches in the European Union to those in the United States, more often emphasizing similarities than differences, particularly with regard to monopolization and mergers).
11. For example, they do not depend on whether a particular econometric technique is sound or can readily be applied in light of data limitations, or on what sorts of internal documents are found, or on how savvy purchasers are in predicting their responses to hypothetical price increases. This Article takes no position on how often adjudicators or agencies should employ more sophisticated econometric techniques, rely on informal methods of assessing the quality of substitutes, and so forth. It addresses only the use to which all such information should be put.
12. Leading prior critiques of market definition are discussed in notes 78, 79, and 80 (at which point they can more meaningfully be contrasted with the present analysis, especially that in Part IV), and past treatments of cross-elasticities (which suggest but do not develop some of the points in Part V) are presented in note 83. Some previous mentions by U.S. courts of market definition's limitations are noted in section VI.E.
13. See Horizontal Merger Guidelines Review Project, FED. TRADE COMM'N, http://www.ftc.gov/bc/workshops/hmg/index.shtml (last visited Oct. 30, 2010) (links for the initial questions and comments)
14. Public Comments on Revision to the Horizontal Merger Guidelines, FED. TRADE COMM'N, http://www.ftc.gov/os/comments/hmgrevisedguides/index.shtm (last visited Oct. 30, 2010)
15. U.S. DEP'T OF JUSTICE & FED. TRADE COMM'N, HORIZONTAL MERGER GUIDELINES (2010) [hereinafter U.S. MERGER GUIDELINES].
16. As Part IV makes clear, the argument does not depend at all on the quality of this estimate of market power or the manner in which it was derived (for example, whether through econometric analysis or the conjectures of customers).
17. See, e.g., AREEDA, HOVENKAMP & SOLOW, supra note 2, at 232 ("Finding the relevant market and its structure is typically not a goal in itself but a mechanism for considering the plausibility of antitrust claims that the defendants' business conduct will create, enlarge, or prolong market power.")
18. Phillip Areeda, Market Definition and Horizontal Restraints, 52 ANTITRUST L.J. 553, 565 (1983)
19. Baker, supra note 2, at 130; Commission Notice on Market Definition, supra note 2, ¶ 2 ("The main purpose of market definition is to identify in a systematic way the competitive constraints that the undertakings involved face⋯ . It is from this perspective that the market definition makes it possible inter alia to calculate market shares that would convey meaningful information regarding market power for the purposes of assessing dominance or for the purposes of applying Article [101 (formerly 81)].")
20. see also Gregory J. Werden, The History of Antitrust Market Delineation, 76 MARQ. L. REV. 123, 189 (1992) (suggesting that the idea that "market delineation must be based on the underlying concern with market power was spreading rapidly in the early 1980s but certainly was not universal").
21. See, e.g., U.S. MERGER GUIDELINES, supra note 5
22. Guidelines on the Assessment of Horizontal Mergers Under the Council Regulation on the Control of Concentrations Between Undertakings, 2004 O.J. (C 31) 5.
23. As discussed in section V.C, some particular methods do focus on changes in market power without being directly concerned with levels, such as in the use of diversion ratios in the analysis of the unilateral effects of horizontal mergers in industries with product differentiation.
24. See, e.g., Joseph Farrell & Carl Shapiro, Antitrust Evaluation of Horizontal Mergers: An Economic Alternative to Market Definition, B.E. J. THEORETICAL ECON., Jan. 2010, art. 9, 1. Even then, however, levels are implicitly employed because premerger markups must be ascertained to determine how much firms' profits suffer as sales fall due to postmerger price increases.
25. A priori, it seems plausible that the justification for market power inquiries would be relevant to whether market definition is a useful tool. However, since there is substantial agreement that the primary (if not sole) purpose of defining markets is to make inferences about market power, itself understood in a particular way, it turns out that further exploration of the underlying purposes of the market power inquiry is largely immaterial to the present investigation. In part, this simplification is possible because the criticisms developed here are conceptual and general rather than pragmatic and particular.
26. More precisely, the inquiry is ordinarily about the extent to which the profit-maximizing price exceeds the competitive level. (In the examination of horizontal mergers, particularly with the use of critical loss analysis
27. see infra section V.C, it is often asked instead whether a given price increase, usually five percent, following the U.S. Merger Guidelines, would be profitable, or at least break-even, rather than profit-reducing.
28. See U.S. MERGER GUIDELINES, supra note 5, § 4.1.2. Typically, if the premerger price was set at a maximizing level and a postulated price increase is break-even, the profit-maximizing price increase would be less than the postulated amount, sometimes approximated as half as much.)
29. Regarding the focus on price, see Louis Kaplow & Carl Shapiro, Antitrust, in 2 HANDBOOK OF LAW AND ECONOMICS 1073, 1080 (A. Mitchell Polinsky & Steven Shavell eds., 2007), which notes that "although anticompetitive harm can come in the form of reduced product quality, retarded innovation, or reduced product variety, our discussion will follow much of the economics literature and most antitrust analysis in focusing on consumer harm that comes in the form of higher prices. This limitation is not as serious as may first appear because higher prices can serve as a loose proxy for other forms of harm to consumers."
30. See, e.g., U.S. MERGER GUIDELINES, supra note 5, § 1 ("A merger enhances market power if it is likely to encourage one or more firms to raise price, reduce output, diminish innovation, or otherwise harm customers as a result of diminished competitive constraints or incentives⋯ . For simplicity of exposition, these Guidelines generally discuss the analysis in terms of ⋯ price effects.")
31. AREEDA, HOVENKAMP & SOLOW, supra note 2, ¶ 501, at 109
32. MARKET POWER HANDBOOK, supra note 1, at 5 ("The economic concept of market power is central to the legal analysis of most antitrust cases.")
33. Dennis W. Carlton, Market Definition: Use and Abuse, 3 COMPETITION POL'Y INT'L 3, 5 (2007)
34. William M. Landes & Richard A. Posner, Market Power in Antitrust Cases, 94 HARV. L. REV. 937, 937, 939 (1981)
35. Gregory J. Werden, Demand Elasticities in Antitrust Analysis, 66 ANTITRUST L.J. 363, 373-74 (1998) (tracing the U.S. Supreme Court's acceptance of this economic definition of market power beginning in the 1960s)
36. id. at 373 ("The courts of appeals have widely used the economic definition of market power.")
37. DG Competition Article 82, supra note 1, ¶ 24 ("Market power is the power to influence market prices, output, innovation, the variety or quality of goods and services, or other parameters of competition on the market for a significant period of time⋯ . An undertaking that is capable of substantially increasing prices above the competitive level for a significant period of time holds substantial market power and possesses the requisite ability to act to an appreciable extent independently of competitors, customers and consumers.")
38. VALENTINE KORAH, AN INTRODUCTORY GUIDE TO EC COMPETITION LAW AND PRACTICE 119 (9th ed. 2007) ("In Continental Can, the ECJ implicitly accepted the Commission's definition of a dominant position based on the economists' concept of power over price ⋯ .").
39. A commonly cited definition is that "[m]onopoly power is the power to control prices or exclude competition." United States v. E.I. du Pont de Nemours & Co. (Cellophane), 351 U.S. 377, 391 (1956). Two differences may be noted briefly. Most obviously, this formulation adds the alternative of being able to exclude competition. Some have rationalized this phrasing as indicating that a monopolist (the focus of inquiry in Cellophane) might use its power either to raise price or instead, perhaps in a predatory manner, to exclude competition. Others have interpreted "or" as "and," emphasizing that the ability to raise prices over a significant period of time requires somehow keeping competitors at bay.
40. See, e.g., Richard Schmalensee, Another Look at Market Power, 95 HARV. L. REV. 1789, 1795 (1982). The other, more subtle difference- echoed more sharply in some other versions as well- is the reference to the ability to "control" price (others have referred to a firm having discretion or, in the words of Case 322/81
41. Michelin v. Comm'n, 1983 E.C.R. 3461 § 30, at 3503, in the European Union, the ability "to behave to an appreciable extent independently of its competitors and customers"). Most of these alternatives are vague and potentially misleading. A firm with little market power can still control its price, although significant elevations would be unprofitable, and a firm with great market power cannot simply ignore competitors and customers and, moreover, if it wishes to maximize profits, will feel compelled to select a single, particular price, just as would its low-power counterpart.
42. See, e.g., PHILLIP AREEDA, LOUIS KAPLOW & AARON EDLIN, ANTITRUST ANALYSIS 485-86 (6th ed. 2004)
43. Kaplow & Shapiro, supra note 12, at 1098
44. see also DG Competition Article 82, supra note 1, ¶ 23 (in elaborating that dominance requires the ability to behave independently of competitors and customers, states: "The notion of independence, which is the special feature of dominance, is related to the level of competitive constraint facing the undertaking(s) in question. For dominance to exist the undertaking(s) concerned must not be subject to effective competitive constraints. In other words, it thus must have substantial market power." (footnote omitted)). These and other differences in definitions of market power are ignored here.
45. See, e.g., AREEDA, HOVENKAMP & SOLOW, supra note 2, ¶ 503b, at 118-19
46. Kaplow & Shapiro, supra note 12, at 1080
47. Landes & Posner, supra note 13, at 939, 941.
48. Specifically, MC is the level of marginal cost at the current quantity of output. If marginal cost is rising with quantity and price is elevated, it follows that, if price were reduced from a supracompetitive level to the competitive level, quantity would be higher and thus marginal cost would be higher, so L would overstate the markup relative to the competitive price.
49. One could instead examine, for example, the fraction P/MC (which equals 1/(1-L)), in which case perfect competition would be associated with an index value of 1, price 10% in excess of marginal cost would result in a value of 1.1, and so forth. Or one could consider (P-MC)/MC (which equals L/(1-L)), which in turn is equivalent to (P/MC) - 1, so perfect competition would (as with the Lerner index) have an index value of 0, price 10% in excess of marginal cost would have a value of 0.1, and so forth. Although these alternatives may seem a bit more straightforward, all are substantially equivalent (since they are all monotonic transformations of L), the use of the Lerner index is conventional in economic analysis of the subject, and as will be seen (notably, in expressions (2) and (4)), L has simpler properties that render its use convenient.
50. In the limit, L = 100% implies an infinite price or a marginal cost of zero, so for finite prices and strictly positive marginal cost, we have L < 100%.
51. See, e.g., ANTITRUST LAW DEVELOPMENTS, supra note 1, at 229.
52. Economists' favored methods of predicting unilateral price effects of horizontal mergers of differentiated products suppose that premerger margins can be determined. Joseph Farrell and Carl Shapiro (economists who have spent time in antitrust enforcement agencies) argue that, although "gross margins are hard to measure using public data ⋯
53. methods sufficiently standardized for cross-sectional studies[,] ⋯ firms have an incentive to keep track of their cost functions via managerial accounting tools, for instance to know how far they can profitably cut prices. Such information ⋯ normally is available to antitrust agencies and courts." Farrell & Shapiro, supra note 9, at 18. Others are more skeptical.
54. MASSIMO MOTTA, COMPETITION POLICY 116 (2004)
55. Baker, supra note 2, at 142-43
56. Werden, supra note 13, at 394.
57. Since the initial setting involves perfectly homogeneous goods, the question whether a narrower market might be contemplated does not arise, at least as the market definition problem is ordinarily understood. In a world in which all products are differentiated, one might suppose that the market is originally taken to include only a single product, the one produced by the firm under scrutiny. For mergers, of course, at least two firms' products would need to be considered. For further discussion of differentiated products mergers, see section V.C.
58. It is common to view the problems of product and geographic market definition as similar, and in certain respects they are. The major difference is that, in some settings involving geographic market definition issues, consumers' preferences may have a particular structure that may be highly relevant to the analysis. For example, if the products are otherwise identical, shipping costs for different customers are the same, and preferences over shipping time do not differ across consumers, everyone will value the substitute identically
59. in particular, all will prefer it to the local product if and only if its price, adjusted for transportation costs and delay, is lower. For some purposes, it may be useful to treat geographically remote supply having such characteristics similarly to present rivals and to potential entrants, although aspects of the analysis differ (for example, remote suppliers' response to a local price change will depend on the demand elasticity in the remote market as well as on remote suppliers' cost functions; for further related discussion, see notes 38 and 85 on partial versus general equilibrium analysis). Details will not be pursued here. When consumers' preferences for the product from the remote market vary more broadly- for example, the product may itself have different characteristics or a different brand image, or preferences over delivery time may vary- then geographically remote suppliers are more like local suppliers of differentiated (or simply different) products, so the standard analysis of product markets is more closely applicable. This differentiated products perspective on geographic market definition is emphasized in Louis Kaplow, The Accuracy of Traditional Market Power Analysis and a Direct Adjustment Alternative, 95 HARV. L. REV. 1817, 1835-43 (1982). Subsequently, product differentiation has become a central consideration in the theory of international trade, particularly to explain two-way trade flows between countries at similar stages of economic development.
60. See, e.g., Paul Krugman, The Increasing Returns Revolution in Trade and Geography, 99 AM. ECON. REV. 561, 562-65 (2009) (revised version of 2008 Nobel lecture).
61. See, e.g., Kaplow & Shapiro, supra note 12, at 1080. Beginning with the expression for profits (price times output, minus cost, itself as a function of output; that is, PQ-C(Q), where C is the cost function), one takes the derivative with respect to price (keeping in mind that output is implicitly a function of price, as reflected by the firm's demand curve), rearranges terms, and makes a substitution using the definition of the firm's elasticity of demand.
62. See, e.g., id. at 1081
63. Landes & Posner, supra note 13, at 944.
64. If they are not all small and hence do not all behave as price takers, further analysis is required. Some of the pertinent considerations are mentioned in section VI.C's discussion of unilateral and coordinated effects in horizontal mergers.
65. See also Janusz A. Ordover, Alan O. Sykes & Robert D. Willig, Herfindahl Concentration, Rivalry, and Mergers, 95 HARV. L. REV. 1857 (1982) (extending the analysis to the case in which multiple firms in the homogeneous goods market are not small).
66. This approach derives from Karl Forchheimer, Theoretisches zum Unvollständigen Monopole, in 32 JAHRBUCH FÜR GESETZGEBUNG, VERWALTUNG UND VOLKSWIRTSCHAFT IM DEUTSCHEN REICH 1 (Gustav Schmoller ed., 1908).
67. See Gavin C. Reid, Forchheimer on Partial Monopoly, 11 HIST. POL. ECON. 303 (1979).
68. See, e.g., Kaplow & Shapiro, supra note 12, at 1081-82. This derivation begins by expressing the dominant firm's demand as the total industry demand for the product minus that portion supplied by the competitive fringe. When one takes the derivative of that expression with respect to price, rearranges terms, and uses the definitions of the pertinent elasticities and of market share, the expression in the text results. A version of this derivation first appeared
69. George J. Stigler, Notes on the Theory of Duopoly, 48 J. POL. ECON. 521, 523-24 (1940). Stigler was also the first to use this formula to emphasize the point that market share alone tells only part of the story- one that, in isolation, can be misleading.
70. See id. at 524.
71. r is positive to begin with: as price rises, the quantity supplied to the market by rivals increases. On another note, it should be understood that all of these terms are endogenous, in that they depend on the price. As price rises, both elasticities may change (it is sometimes supposed that the elasticity of demand tends to be higher and of supply perhaps lower as prices rise, although neither need be true), and the dominant firm's share will in general change (as price increases, rivals supply more and the dominant firm less, so its share will be lower at higher prices). Accordingly, each of these terms refers to levels at the prevailing price.
72. If the firm's initial quantity is 50 and that of the industry is 100, then the industry quantity falls by 2, from 100 to 98, and the firm's falls by 2, from 50 to 48. A 2-point fall on a base of 50 is 4%, and the elasticities, recall, are defined in percentage terms.
73. If rivals sell 50 of 100 units, a 2% increase on the base of 50 is 1 unit, which is only 1% of the 100 units in the market.
74. The reader may imagine that we have chosen the closest substitute (setting aside just how that is to be defined) and that the chosen substitute is quite close. None of the analysis to follow depends on how this choice is made or whether it is correct by conventional criteria.
75. If it turned out that broadening markets did make sense, the manner in which one analyzed market shares in the broadened market would presumably bear on the criterion for when the market should be broadened. The analysis in Parts IV and V will cast doubt on whether there could be a coherent principle in any event.
76. Some analysts have discussed how, when markets are combined, the firm's lower share will be offset, for example, by there being in some sense a lower demand elasticity in the broadened market, with the suggestion that market definition would not matter as long as the proper adjustments were made.
77. See, e.g., Kaplow & Shapiro, supra note 12, at 1091
78. Landes & Posner, supra note 13, at 962. But the actual manner in which the values of other terms in the formula need to be adjusted in order to make correct inferences has not been examined. Richard Schmalensee offers the closest exception in an appendix; he presents a formula for the market demand elasticity in the combined market that is a function of the elasticities in the two separate markets and the (weighted average of the two) cross-elasticities of demand between the two products.
79. See Schmalensee, supra note 14, at 1815-16. This formula, however, is addressed to the question of how much market power a hypothetical monopolist of both markets would have, in particular, if it increased the price in both markets by the same proportion (an inquiry like that often used to apply the U.S. MERGER GUIDELINES, supra note 5, § 4.1.1
80. see note 99 for further discussion). This question differs importantly from that of how to determine the market power of the dominant firm in the initial market once the additional market has been combined. Moreover, the new formula requires knowledge not only of the market demand elasticity in the initial market- which, combined with the firm's share in that market (and setting aside rivals' supply response, as Schmalensee does), would be sufficient to determine market power- but also of additional pieces of information: the market demand elasticity in the other market and the (weighted average) cross-elasticities between the two. Therefore, it cannot be that combining markets makes it easier to determine market power.
81. Even this simple adjustment can involve complications, notably, whether to measure shares by units sold, revenue, or some other metric.
82. It should likewise be clear that no other general formula for adjustment would work.
83. The only difference is that, to apply formula (5), we need also to compute S. As it turns out, however, this difference also is immaterial. After all, if any error is made, it will wash out once the adjustments in expressions (6) and (7) are completed, assuming that we use the same (possibly erroneous) value of Ŝ in doing so.
84. But see supra note 36.
85. Regarding the latter omission, one might suppose that the magnitude of the overall supply elasticity in the market being combined would matter. Moreover, this dependence may be thought to exist despite the fact that additional supply of the second product does not in itself reduce demand for the dominant firm's product as is the case when rivals producing goods identical to those of the dominant firm expand their supply. Suppose, for example, that the supply elasticity in the second market is very low. Then, a rise in the price for the product in the first market would not induce as much switching to the product in the second market because, as more purchasers switched, the price of the good in the second market would be bid up as a consequence, reducing the relative benefit of substitution. Implicitly, any such effect could be seen as reflected in the market demand curve in the first market and thus in the market demand elasticity in that market. (It is conventional, however, to define this curve and state this elasticity under the assumption that other prices are fixed; this approach is termed partial equilibrium analysis.
86. See, e.g., DAVID M. KREPS, A COURSE IN MICROECONOMIC THEORY 279-80 (1990). Similar assumptions underlie formula (8) in section V.A for the market elasticity of demand as a weighted sum of other products' cross-elasticities. See infra note 85.)
87. i1 Here, Ri is the total revenue in each of the other markets (besides the two on which we are focusing, which have been denoted markets 1 and 2), and ei1 is the cross-elasticity for each of the other goods. It is apparent that these expressions neither illuminate the basic ideas nor provide a simpler means of assessing market power than that provided in formula (4), a subject further discussed in Part V. One could perform yet further convolutions: expressing the market elasticity of de mand in the combined market in terms of the market elasticity in the original market, as in expression (6); further decomposing that elasticity into a weighted sum of cross-elasticities, as in expression (8) in section V.A; and pulling out the particular cross-elasticity for the second market, but then adding it back in, in order to return to the original elasticity. But, again, this task would be pointless. (If one was interested not in our dominant firm's market power, but in the ability of a hypothetical monopolist in the combined market to proportionately raise prices in the two markets, then there does exist a way to combine information about the two market elasticities and the cross-elasticities between the two products to illuminate this question. See supra note 33; see also infra note 99 (discussing the Merger Guidelines' hypothetical monopolist test).)
88. Regarding the lack of prior recognition, most sophisticated observers are aware of the need to interpret market shares carefully in redefined markets- and in original markets- but have not appreciated that if one wishes to undertake such interpretations in a logical fashion, one is essentially forced to return to the original, narrow, homogeneous goods market. Perhaps the sharpest criticism that is in similar spirit to that presented in the text is offered in Schmalensee, supra note 14, at 1797. He does not mention that the method cannot be used even in the case of homogeneous products once one engages in market redefinition, and his more detailed elaboration, see id. at 1797-98, does not feature the core problem with market redefinition identified in the text, although his discussion does indicate some of what can go wrong when one does not fully undo the market redefinition. See also Kaplow, supra note 22, at 1830 n.54 (discussing a numerical example based on the Cellophane case in which making inadequate adjustment in the broader market may substantially understate market power relative to the correct result derived using the formula in the narrower, homogeneous goods market).
89. See, e.g., Franklin M. Fisher, Economic Analysis and "Bright- Line" Tests, 4 J. COMPETITION L. & ECON. 129, 132 (2008) [hereinafter Fisher, Economic Analysis] ("What, then, does economic analysis have to say about market definition? In one sense, the answer is 'Nothing at all.' The question of what is 'the' relevant market never arises in economics outside of antitrust. Moreover, ⋯ it is not a question that has a precise, well-defined answer.")
90. Franklin M. Fisher, Horizontal Mergers: Triage and Treatment, J. ECON. PERSP., Fall 1987, at 23, 27 ("Market definition is an artificial construction created by antitrust litigation. For any other purpose of economic analysis, the binary question of whether particular firms or products are 'in' or 'out' of a given market is a meaningless one. Even in antitrust cases, that question is not a useful one if substantive results turn on the answer.")
91. Kaplow & Shapiro, supra note 12, at 1091 ("Although we have discussed the economic analysis of market power at some length, the concept of market definition has not yet appeared directly.")
92. id. at 1170 ("[T]he common approach of defining markets and looking to market shares ⋯ does not play a central role in economists' analysis of market power.")
93. Adriaan ten Kate & Gunnar Niels, The Relevant Market: A Concept Still in Search of a Definition, 5 J. COMPETITION L. & ECON. 297, 298 (2009) ("[I]t is hard to find a satisfactory description or definition [of what a relevant market is] in textbooks on microeconomics or industrial organization, if the concept is mentioned at all in such readings.")
94. see also Gregory J. Werden, Market Delineation and the Justice Department's Merger Guidelines, 1983 DUKE L.J. 514, 516 (noting that "[e]conomic theory does not require the delineation of markets, and most economists would not know how to begin to delineate one," but going on to state that the U.S. Merger "Guidelines employ the concept of an 'antitrust market,' a market delineated for the sole purpose of antitrust analysis").
95. This observation does, however, raise the question of why they have not previously debunked the practice rather than merely offering reservations. Interestingly, before the market definition / market share paradigm became prominent, economists had expressed hostility toward the approach.
96. See Werden, supra note 7, at 126; see also Edward H. Chamberlin, Product Heterogeneity and Public Policy, AM. ECON. REV., May 1950, at 85, 86-87 ("'Industry' or 'commodity' boundaries are a snare and a delusion- in the highest degree arbitrarily drawn, and, wherever drawn, establishing at once wholly false implications both as to competition of substitutes within their limits, which supposedly stops at their borders, and as to the possibility of ruling on the presence or absence of oligopolistic forces by the simple device of counting the number of producers included.").
97. See, e.g., MICHAEL D. WHINSTON, LECTURES ON ANTITRUST ECONOMICS 100-14 (2006) (surveying literature on the estimation of market power in mergers)
98. Jonathan B. Baker & Timothy F. Bresnahan, Empirical Methods of Identifying and Measuring Market Power, 61 ANTITRUST L.J. 3 (1992) (surveying techniques)
99. Jonathan B. Baker & Timothy F. Bresnahan, Estimating the Residual Demand Curve Facing a Single Firm, 6 INT'L J. INDUS. ORG. 283 (1988) [hereinafter Baker & Bresnahan, Residual Demand]
100. Cory S. Capps, David Dranove, Shane Greenstein & Mark Satterthwaite, Antitrust Policy and Hospital Mergers: Recommendations for a New Approach, 47 ANTITRUST BULL. 677 (2002)
101. Kaplow & Shapiro, supra note 12, at 1178-80 (surveying literature on methods for mergers)
102. sources cited infra note 107 (on critical loss and related methods of assessing competitive effects of differentiated products mergers).
103. See, e.g., Kaplow & Shapiro, supra note 12, at 1187-88
104. Pitofsky, supra note 2, at 1810-11
105. Schmalensee, supra note 14, at 1800-01. As section VI.E notes, courts have long recognized the point.
106. Landes & Posner, supra note 13, at 955 tbl.1 (column for a market size of $200 million).
107. Id. at 958 tbl.2
108. see id. at 947-48 ("Thus the degree of market power in this example for a firm with an 8% market share is identical to that of a firm with an 80% share in a different market, because of offsetting differences in demand elasticities.").
109. See infra note 78.
110. E.g., Kaplow, supra note 22, at 1820, 1827 & n.45, 1847
111. Schmalensee, supra note 14, at 1789, 1813-14
112. AREEDA, KAPLOW & EDLIN, supra note 14, at 492, 495
113. Kaplow & Shapiro, supra note 12, at 1188 & n.171 (also providing the most explicit prior reference to this idea). This concept has also played a central role in criticism of Landes and Posner's claim that market power is systematically overstated on account of the frequent omission of direct consideration of some or all of the four factors that they identify as bearing on the market power implications of a given market share.
114. See Kaplow, supra note 22, at 1819-21.
115. To be clear, this practice is not confined to courts or to casual commentators but is virtually ubiquitous, even being employed by some of the most sophisticated analysts. For example, Franklin Fisher states: "A market will thus be well defined if and only if the share measurements it leads to provide some reasonable index of the true power of the alleged firm ⋯ ." Franklin M. Fisher, Diagnosing Monopoly, Q. REV. ECON. & BUS., Summer 1979, at 7, 13 (emphasis added).
116. There are some limitations. For example, suppose that we imagine the standard reference market to be one with a market elasticity of demand of 2.0 and a rivals' supply elasticity of 1.0. In that market (and in any with a demand elasticity above 1.0) there is a finite upper bound to the Lerner index. (Here, if the market share Ŝ is 100%, it is clear from formula (4) that L = 0.5, so that the profit-maximizing price would be double the marginal cost.) For any such upper bound in the benchmark market, it is possible that there would be an actual market exhibiting even greater market power (simply suppose a 100%
117. This gap has been mentioned. See, e.g., AREEDA, KAPLOW & EDLIN, supra note 14, at 492 ("Once we have a firm's market share, we still need to determine what it implies about the existence or magnitude of that firm's ability to charge more than a competitive price⋯ . The starting point is the lay intuition that a firm with 100 percent of a market possesses market power. It almost surely has the power to restrict output and to charge more than the perfectly competitive firm. But how much more? Five or 500 percent? Such answers, never supplied, are essential if one is to infer market power from market share.")
118. id. at 498-99 ("[T]he power implications of any particular market share remain obscure. The courts have not stated how much power they believe to be associated with given market shares.").
119. The logic of the present analysis does not depend on formula (4) or on defining market power using the Lerner index.
120. Note a possible circularity if application of the standard reference market, now taken to be the average market, is necessary to decide how to define markets in the first place. See, for example, statement 4 in section A and also the discussion in Part IV.
121. And, if it were not, individuals would no longer state that "80% is quite high."
122. See U.S. MERGER GUIDELINES, supra note 5, § 4.1.1 ("The hypothetical monopolist test requires that a product market contain enough substitute products so that it could be subject to post-merger exercise of market power significantly exceeding that existing absent the merger. Specifically, the test requires that a hypothetical profit-maximizing firm ⋯ that was the only present and future seller of those products ('hypothetical monopolist') likely would impose at least a small but significant and non-transitory increase in price ('SSNIP') on at least one product in the market ⋯ ."); DG Competition Article 82, supra note 1, ¶ 14 (SSNIP test used by the Commission for market definition, particularly in merger cases); Commission Notice on Market Definition, supra note 2, ¶¶ 15-17.
123. See infra note 99.
124. In brief, the SSNIP test asks how much a hypothetical monopolist of variously defined markets could elevate price, not how much the merging parties will be able to elevate price. Even if one were actually considering a monopolization (abuse of dominance) claim, the question should be the extent to which the firm in question can elevate price (say, as a consequence of the alleged abusive practice), not in what market a hypothetical monopolist thereof could elevate price to some specified degree. Moreover, regarding the actual monopolist, once one has estimated how much it can elevate price in its own (homogeneous goods) market, the analysis should be viewed as complete rather than employing this estimate as the input into a market definition process requiring further steps (and, as stated, answering a different question). See infra section VI.C.
125. One could generalize the discussion in the text along other dimensions without affecting the analysis. For example, if price discrimination is possible, the degree of exercised market power might vary across customer groups, so there would be multiple estimates and one could define different markets for different groups if one wished.
126. As will be discussed in section VI.D, engaging in a market redefinition inquiry may help to hide (including from oneself) just how poor the information inputs are, but there is no magical process by which the technique is able to create something (better information) that does not exist in the first place.
127. Because the only purpose of defining a relevant market is to make an inference about market power, the only truly pertinent criteria for determining the set of admissible markets would be that they be selected in such a way that the subsequent choice among them yields as accurate an inference as possible, which is question-begging in the same way as is choosing the best market from among the finalists.
128. It is possible, depending on how the choice set is determined, that all the elements will be too broad or all too narrow. For example, the narrowest plausible market may be viewed as the homogeneous goods market, the firm's share in that market may not be very high, and yet, due to a combination of fairly inelastic market demand and rivals' supply, the firm could have significant market power, much more than indicated by its share (where the indication might be based on use of the standard reference market). In that case, all markets would be too broad in the sense that the firm's share in them would understate its market power. This possibility constitutes a further defect in the approaches examined in this section, but will be set to the side. (Of course, choosing between markets, both of which, say, are too broad, is easy if one knows this fact: one simply picks the narrower of the two.)
129. United States v. E.I. du Pont de Nemours & Co., 351 U.S. 377 (1956).
130. See Kaplow, supra note 22, at 1828.
131. As section III.A explains, there must be some notion like a standard reference market for typical statements about the implications of market shares to be made sense of in a manner that communicates. Since the standard reference market table, as explained in section III.B, can be constructed in any manner one likes, the method really is nonrestrictive. (One could add further layers. For example, if a higher share in a narrower market is known to overstate market power, one could make some downward adjustment in the power one infers; likewise for the lower share in the broader market. At the end of the day, there are two possibilities: Either the inference from the higher share in the narrower market differs from the inference from the lower share in the broader market, as is commonly supposed, in which case the argument in the text to follow is fully applicable because it does not depend on the magnitude of the difference. Or the inferences are the same, in which case market definition is irrelevant and thus pointless, in line with this Article's central claim.)
132. Following Part II, each of these measures may be taken to be corresponding values of the Lerner index. One could instead substitute some other chosen measure of the extent of profitmaximizing price elevation without affecting the argument.
133. One could refine this statement to incorporate the uncertainty of estimates, asymmetry and nonlinearity of error costs, and so forth without materially altering the argument to follow.
134. As mentioned, the analysis does not depend on whether we use the standard reference market or some other process, as long as we do not essentially ignore the differences in the market definitions and use some other means, such as that examined in section II.B, to obtain the same market power inferences in both markets- in which case market definition is moot, as previously noted.
135. What constitutes a best estimate will vary by context (for example, screening by enforcers, deciding on summary judgment, or full adjudication). For purposes of the present argument, the best estimate simply refers to that which is best given the information set under consideration (that is, the same information one might have used to define the market in any other manner).
136. For further discussion, see section VI.D.
137. The only way our conclusion could contain information not present in our initial best estimate of market power would be if some information irrelevant to market power went into deciding which market definition was better (or if some information relevant to market power was given more or less weight than that appropriate in forming a best estimate of market power). In that case, however, the inference of market power drawn from the market share in the chosen market could only be worse than our initial estimate. (This claim also is a tautology, in that our initial estimate was stated to be the best estimate we could derive from all pertinent information.)
138. Even then, the coincidence would hold only for a particular firm with a given share in that market. If the firm's share were different, or if we wanted to assess some other firm that sells the same product (but has a different market share), or if we were considering possible mergers between firms that sell the same product, then only by chance would the same market definition continue to work. (Suppose, for example, that the initial firm in question has a profit-maximizing price that is 5% above marginal cost, that this corresponds to a 20% market share on the reference table, and that there happens to exist some market definition that results in the firm having a 20% share in the resulting market. Now, consider another firm that sells the same product and has a 50% share in the market that we just selected; it might have the ability to raise price 10%, but the reference table might indicate instead that a 25% elevation is associated with a 50% market share.)
139. As mentioned in note 64, the only way to avoid the problem is to use a process that interprets shares in the different markets in such a way as to reach the same conclusion about market power- and one corresponding to our original best estimate. In that event, market definition is playing no real role in the process of inferring market power.
140. Multiple choices may work (for example, if the correct conclusion is that market power is insufficient, many broad markets may yield this result), in which case any of these may be selected.
141. Actually, not necessarily. See supra note 61 (mentioning the atypical case in which even the narrowest homogeneous goods market is too broad).
142. *, that we contemplate at the beginning of step 4 is just the share we read off the standard reference market table in step 2. Unless we make a careless error, we entirely undo in the latter portion of the technique precisely what we did in the first part. Note, however, that this process, like that in subsection A.2, does have the virtue that it obtains a correct answer - that is, as good an answer as is feasible given the available information. In essence, it substitutes, for the familiar all-or-nothing choice whether to combine markets (see, e.g., AREEDA, KAPLOW & EDLIN, supra note 14, at 492
143. Carlton, supra note 13, at 16
144. Fisher, supra note 49, at 16
145. Kaplow, supra note 22, at 1828
146. Kaplow & Shapiro, supra note 12, at 1092
147. Pitofsky, supra note 2, at 1811), a jerry-rigged market definition that is entirely arbitrary except for the fact that, on the standard reference market table, the firm's share yields the correct market power inference. Thus, the fact that this process produces the right answer is made possible by its lack of resemblance to ordinary methods of drawing market boundaries. (Indeed, it does not even matter under this criterion which of many other possible products are combined if, for example, there is a need to generate a lower market share.) One might imagine that, instead of using all available information on extant market power, which all these procedures require, we might jiggle and adjust the market definition, comparing alternatives, until we obtain a market that "looks" like the standard reference market. Yet it should be apparent that such is not possible. Market definitions are abstractions to begin with, and either (A) the standard reference market is a table reflecting empirically generated averages across all markets (averaging over not only apples and oranges, which are quite similar and might even be in the same market, but also cell phones, steel, candy, and houses) - and it is hard to give meaning to the determination of which of two or more market definitions more resembles, under some Rorschach test, such an average - or (B) the standard reference market is a table that is derived in an entirely arbitrary manner, by writing down a series of numbers generated in some fashion to have the desired monotone relationship (higher market share, more market power). Furthermore, if one employed any such resemblance test, it has the obvious deficiency relative to the more explicit techniques considered here of being unlikely to reach as good an answer to the original market power question. See also infra section VI.D (further elaborating the point that markets cannot be defined by how they may appear to our senses).
148. See, e.g., Kaplow, supra note 22, at 1827-28
149. Schmalensee, supra note 14, at 1804.
150. See supra note 33.
151. One of their claims that goes furthest appears in their introduction, where they state: "Among other things, we show how the inference of market power can be adjusted so that defining a market broadly or narrowly will not affect the inference [of market power] that is drawn ⋯ ." Landes & Posner, supra note 13, at 938
152. see also id. at 978. However, although they do discuss at length factors that bear on the magnitude of market power (a major focus of their article), they nevertheless advocate particular market definitions as leading to better- and thus presumably different- inferences than others produce, and they do not in fact develop a way to achieve the asserted neutrality. (Throughout, as Schmalensee emphasizes in his commentary, Landes and Posner seek to employ formula (4), which is valid only for the homogeneous goods market, to allow inferences in all manner of markets, an approach explained in subsection II.B.2 to be invalid
153. Schmalensee, supra note 14, at 1797.) Despite the force of many of their arguments, their conclusion is rather tame: When those elasticities are known or knowable, our analysis provides a method of estimating market power in quantitative terms. The analysis thus should be helpful to enforcement agencies in setting priorities and allocating their resources, and to courts in those cases, which may be few, where estimates of the elasticity of demand and supply are obtainable in a form usable in the litigation process
154. Landes & Posner, supra note 13, at 983. The latest edition of Posner's book on antitrust law also sends mixed signals
155. Compare RICHARD A. POSNER, ANTITRUST LAW 147 (2d ed. 2001) ("The importance that antitrust law attaches to defining a market is another consequence of the law's failure to have developed an approach at once genuinely economic and operational to the problem of monopoly. If we knew what would happen if a group of sellers raised their prices- if we knew how rapidly the price increase would be undone by the response of other sellers- it would be redundant to ask whether the group constituted an economically meaningful market. In other words, if we knew the elasticity of demand facing those sellers, we would know almost all we needed to know.")
156. id. at 148 ("It is lack of confidence in the ability to measure elasticities reliably by the methods of litigation that necessitates definition of the market.").
157. Among the sharpest critics are Farrell & Shapiro, supra note 9, at 1 (arguing that the market definition "approach can be clumsy and inaccurate in industries with differentiated products where the theory of harm is related to unilateral ⋯ effects");
158. Fisher, supra note 49, at 16 ("If one always remembers this, there is no positive harm in engaging in the market definition exercise. Indeed, viewed correctly, arguments concerning whether products are in or out of the market which are made in terms of demand and supply substitutability and hence in terms of constraints are exactly the arguments which one would have to decide in looking at the constraints directly. The trouble is that it is too easy to forget what the analysis is all about. By focusing on whether products are in or out of the market, one converts a necessarily continuous question into a question of yes or no. The temptation is to regard products which are in as all counting equally and products which are out as not counting at all.");
159. Fisher, Economic Analysis, supra note 41, at 132- 33 (discussing Cellophane: "Nothing is to be gained, and much information may be lost by an attempt to force these facts into the Procrustean bed of market definition, insisting on an answer that other flexible wrapping papers are either in or out of the market. If one insists on expressing the facts in such language, then the correct answer is that such papers are 'in' at high prices for cellophane and 'out' at low ones. Evidently, this is not a very useful way to proceed.");
160. Schmalensee, supra note 14, at 1798-99 ("This approach focuses attention on market definition, not on the fundamental question of market power. It has a number of intrinsic weaknesses, some of which are noted by Landes and Posner, and thus is not always reliable or even usable." (footnote omitted))
161. id. at 1808 ("Other approaches to proving the existence of substantial market power are no less valid than the market share approach."). I have also been critical in my own prior writing, beginning with Kaplow, supra note 22, at 1827 n.45 ("In principle, of course, the most natural approach to determining extant market power on the basis of a thorough economic analysis would be to cast aside a priori impressions of market power associated with particular market shares and to derive the measure of market power from scratch, using all the relevant factors." (emphasis omitted))
162. id. at 1828-29 ("But if sufficient information were available to make that comparison [between errors in market power inferences from the two markets] possible, it could be put to better use: a more sensible and straightforward approach would be to discount directly the perceived market power by the degree to which failure to combine markets overstates market power.")
163. id. at 1830 ("Combining markets and then measuring elasticity is no less difficult than directly determining the proper adjustment to one's inference of market power in light of the existence of close substitutes. In fact, the two inquiries are theoretically equivalent." (footnote omitted))
164. See also AREEDA, KAPLOW & EDLIN, supra note 14, at 492
165. Kaplow & Shapiro, supra note 12, at 1171 ("[O]ne often needs to know the right answer - that is, how much market power exists - in order to know which market definition is best."). On geographic market definition in particular
166. see Kaplow, supra note 22, at 1839 ("A plaintiff cannot refute the hypothesis that foreign production costs are lower than domestic production costs [and thus overcome Landes and Posner's geographic market definition presumption] without reliable information concerning both sets of costs. If information on domestic production costs were available, one could directly compute the measure of market power suggested by Landes and Posner ⋯ .")
167. AREEDA, KAPLOW & EDLIN, supra note 14, at 497 ("Imports may have occurred only because [the defendant's] price exceeds the competitive level, which would make clear that actual or future imports do not prevent supracompetitive pricing. Of course, if we know that, we don't need any market definition."). For additional strong remarks in works not focused on criticism of market definition
168. Timothy F. Bresnahan, Empirical Studies of Industries with Market Power, in 2 HANDBOOK OF INDUSTRIAL ORGANIZATION 1011, 1049 n.44 (Richard Schmalensee & Robert D. Willig eds., 1989) (stating in passing that "[t]hese [market power] questions are well-posed, even where the usual method of answering them, defining a 'relevant market' and calculating market shares in it, is senseless"),
169. Frank H. Easterbrook, The Limits of Antitrust, 63 TEX. L. REV. 1, 22 (1984) ("An inquiry into power does not entail the definition of a 'market,' a subject that has bedeviled the law of mergers. Usually the search for the 'right' market is a fool's errand. The seller of 100% of a particular good may have no power if consumers have substitutes or if rivals can make the good as cheaply. On the other hand, there may be tens of possible markets, each offering a little insight into conditions of competition. Market definition is just a tool in the investigation of market power; it is an output of antitrust inquiry rather than an input into decisions, and it should be avoided whenever possible.").
170. See also Janusz A. Ordover & Robert D. Willig, The 1982 Department of Justice Merger Guidelines: An Economic Assessment, 71 CALIF. L. REV. 535, 536 (1983) (in an article largely praising the 1982 Guidelines, including its approach toward market definition, observing: "Arguments for and against a merger that turned upon distinctions between broad and narrow market definitions have always seemed theologic rather than economic. The focus on market definition has seemed to be an inadequate substitute for, and a diversion from, sound direct assessment of a merger's effects.").
171. A particularly harsh attack on market definition appears in Richard S. Markovits, Predicting the Competitive Impact of Horizontal Mergers in a Monopolistically Competitive World: A Non-Market-Oriented Proposal and Critique of the Market Definition-Market Share-Market Concentration Approach, 56 TEX. L. REV. 587, 599-603, 698-728 (1978). His extensive critique, which focuses on horizontal mergers in differentiated products markets, is difficult to interpret. Although some of his bottom lines overlap with some of those reached here, the internal analysis largely differs, in part because many of his arguments draw on his idiosyncratic theories of imperfect competition that he develops in the article (which does not cite the then-leading, modern treatments of the subject by Michael Spence and by Avinash Dixit and Joseph Stiglitz). Some indications of the disconnect with standard economic analysis are that, in 145 pages, he does not use the terms elasticity or cross-elasticity (even though his analysis focuses on these ideas), that a large portion of his argument is predicated on firms' pricing individually to consumers (that is, highly refined price discrimination, without regard for problems of identification or arbitrage), and that, when he writes about coordinated oligopolistic pricing at length, he omits the central point that mergers reduce the parties' incentive to cheat. And some of his other reasons are fairly familiar imperfections in the market definition process that induce most to be careful and seek alternatives when available rather than to reject the method entirely. That said, Markovits's emphasis in differentiated products mergers - on how many buyers whose first choice is one merging party's product have the other party's product as their second choice - anticipates the more formal developments on the subject in subsequent decades. See sources cited infra note 107. An internal summary of his main points on issues pertaining to this Article is as follows: The failure of the traditional approach is partially attributable to the inability of market share data to reveal much information about a large number of highly relevant factors to which it does not even purport to relate. Thus, market shares have little bearing on (1) the amount of potentially pro-competitive efficiencies the merger will generate, (2) many of the factors that influence the effectiveness of potential competition (e.g., the rate of growth of market demand), and (3) many reputational, informational, and other factors ⋯ . However, the argument [here] justifies a far more radical departure from the traditional approach. The market share-market concentration approach ultimately must be rejected because even when a clear break in the chain of competitors exists (1) some firms outside the associated market will be in a position to compete for some of its buyers, (2) various firms within the market will not be universally or equally competitive with each other, and (3) different firms in the same market, and a fortiori in different markets, will have very different distributions of competitive ranks, advantages, and disadvantages.
172. Markovits, supra, at 727
173. see also id. at 590, 603, 699.
174. See, e.g., AREEDA, HOVENKAMP & SOLOW, supra note 2, at 108 ("Chapter 5B then explores the 'alternative' or 'direct' indicators of market power, which can be independent of market definition and are sometimes superior to it⋯ . While the data in actual cases are often too incomplete to permit the more technical economic procedures, better understanding of the economics illuminates the less technical- but also typically less accurate- market definition methodology that antitrust litigants and tribunals customarily employ⋯ . We will see that market definition may not be necessary to prove market power; may not itself suffice to establish the degree of market power; and may be redundant or misleading once the facts necessary for an accurate definition have been obtained.")
175. id. at 136 ("Still, there may be no alternative to the marketdefinition approach when no other observable facts establish the existence and degree of market power more directly and with tolerable accuracy.")
176. Areeda, supra note 7, at 565 ("We have no choice but to do the best we can in defining the market and to moderate the inferences we draw from prospective market shares ⋯ .")
177. WHINSTON, supra note 43, at 78-79, 99-100
178. Baker, supra note 2, at 131 ("Market definition is least useful when market shares would not be strongly probative of market power or anticompetitive effect, while direct evidence as to market power or anticompetitive effect is available and convincing.")
179. id. at 162 ("But if direct evidence is unavailable or not strongly probative, making market definition necessary or useful, the court will likely need to define the market through undertaking a hypothetical prospective market definition analysis (albeit relying importantly on retrospective evidence).")
180. Carlton, supra note 13, at 3 ("Market definition is a crude though sometimes useful tool for identifying market power.")
181. id. at 17 ("Although the clear theoretical construct of market definition can guide one, the absence of estimates of the demand (or cost) system subject this exercise to possible error and arbitrary judgments.")
182. Pitofsky, supra note 2, at 1834 ("The remainder of this Article is devoted to an effort to present an approach to relevant market definition that neither understates nor overstates market power."). Finally, some have proposed to me that at least a number of economists hold more critical views of market definition than suggested by the literature, although the particular reasons for such skepticism are not apparent. In assessing merger review in the European Union, Kai-Uwe Kühn suggests that: [E]mpirical research on the problem of market definition and the effects of mergers has made economists realize that the information needed to determine the unilateral effects of mergers requires as much or less information than an economically rigorous estimation of the market boundaries. This means that it is in principle no more difficult to estimate price effects of mergers directly, skipping the market definition step
183. Kai-Uwe Kühn, Reforming European Merger Review: Targeting Problem Areas in Policy Outcomes, 2 J. INDUSTRY, COMPETITION & TRADE 311, 315 (2002). However, he concludes his brief discussion of the subject on a more moderate note: "But market definition should be regarded as part of the general analysis of the competitive effects of the merger.
184. Id. at 316.
185. For details of the historical development, see Werden, supra note 13, at 398-401. For an important example in the European Union, see Commission Regulation 802/2004 of 7 April 2004 Implementing Council Regulation 139/2004 on the Control of Concentrations Between Undertakings, Annex I, 2004 O.J. (L 133) § 6.I, at 15 (EC) ("A relevant product market comprises all those products and/or services which are regarded as interchangeable or substitutable by the consumer, by reason of the products' characteristics, their prices and their intended use⋯ . Factors relevant to the assessment of the relevant product market include the ⋯ cross-price elasticity of demand ⋯ .").
186. "What is called for is an appraisal of the 'cross-elasticity' of demand in the trade." United States v. E.I. du Pont de Nemours & Co., 351 U.S. 377, 394 (1956)
187. see id. at 400 (further use of cross-elasticity)
188. id. at 395 ("reasonably interchangeable")
189. id. at 404 ("reasonable interchangeability")
190. see also Brown Shoe Co. v. United States, 370 U.S. 294, 325 (1962) ("The outer boundaries of a product market are determined by the reasonable interchangeability of use or the cross-elasticity of demand between the product itself and substitutes for it."). "Over the last decade, most courts of appeals have indicated that market delineation should focus on cross elasticity of demand, generally citing Brown Shoe.
191. Werden, supra note 13, at 400
192. see also ANTITRUST LAW DEVELOPMENTS, supra note 1, at 556-66 (discussing cases on reasonable interchangeability of use and cross-elasticity of demand)
193. Malcolm B. Coate & Mark D. Williams, A Critical Commentary on the Critical Comments on Critical Loss, 53 ANTITRUST BULL. 987, 1019 (2008) ("The case law is clear that market definition turns on the closeness of substitutes, including cross-elasticities of supply and demand for nearby products.")
194. Pitofsky, supra note 2, at 1817 ("Virtually all courts started with the Cellophane cross-elasticity of demand test ⋯ ."). Ironically, in Eastman Kodak Co. v. Image Technical Services, Inc., 504 U.S. 451 (1992), the Court in text endorsed Cellophane's focus on cross-elasticity, but then cited the fourth edition of the (then) Areeda and Kaplow text, followed by an elaboration in the footnote: "What constrains the defendant's ability to raise prices in the service market is 'the elasticity of demand faced by the defendant- the degree to which its sales fall ⋯ as its price rises.
195. Areeda & Kaplow ¶ 342(c), p. 576.
196. Id. at 469 n.15 (alteration in original) (emphasis added).
197. The development presented in this Part does not appear elsewhere. One of the first discussions of the issue appears in Landes & Posner, supra note 13, at 961 n.43, in which they indicate a preference for elasticities without expressing hostility to the use of cross-elasticities. Posner's book takes an even more benign view of standard practice.
198. See POSNER, supra note 78, at 149-50.
199. For stronger statements, see AREEDA, KAPLOW & EDLIN, supra note 14, at 495
200. Kaplow, supra note 22, at 1829 n.52
201. Joseph J. Simons & Michael A. Williams, The Renaissance of Market Definition, 38 ANTITRUST BULL. 799, 826-27 (1993)
202. Schmalensee, supra note 14, at 1803
203. Werden, supra note 13, at 401. Some courts and commentators affirmatively state the relationship backwards, suggesting that the reason to ascertain elasticities is to help assess cross-elasticities (which, as the text to follow explains, is not that surprising given the focus on market definition and thus the apparent need to identify which products are important substitutes).
204. See New York v. Kraft Gen. Foods, Inc., 926 F. Supp. 321, 333 (S.D.N.Y. 1995) ("Cross-price elasticity is a more useful tool than own-price elasticity in defining a relevant antitrust market. Cross-price elasticity estimates tell one where the lost sales will go when the price is raised, while own-price elasticity estimates simply tell one that a price increase would cause a decline in volume.")
205. MOTTA, supra note 20, at 107 ("When own-price elasticity for the product considered, say product A, is high enough to lead us to believe that a hypothetical monopolist would not profitably raise prices of A in a small but significant way, it becomes important to identify which products exercise a constraint on A. Cross-price elasticities might help us to rank the closest substitute (which, together with A, will become the object of the next step of the hypothetical monopolist test).")
206. Michael L. Glassman, Market Definition as a Practical Matter, 49 ANTITRUST L.J. 1155, 1163-64 (1980).
207. The market elasticity of demand is sometimes called the price elasticity of demand, and the cross-elasticity is sometimes referred to as the cross-price elasticity. In this Part, the use of the term elasticity by itself refers to the market elasticity of demand, and the simpler term crosselasticity will be employed.
208. For a derivation, see, for example, Kaplow & Shapiro, supra note 12, at 1091-92. (In the limited other writing on competition policy that mentions this decomposition of the elasticity into a weighted sum of cross-elasticities, different formulas sometimes appear.
209. See AREEDA, HOVENKAMP & SOLOW, supra note 2, at 130 n.4
210. Landes & Posner, supra note 13, at 961 n.43.
211. For explanatory comments, see Werden, supra note 13, at 413 n.172.) The method of derivation tracks the preceding explanation in the text: one differentiates a representative individual's budget constraint (indicating that the sum of the quantity purchased of each good, weighted by that good's price, equals total income, which is taken to be fixed) with respect to the price of good 1, holding other prices fixed, and then rearranges terms and makes substitutions using definitions of other terms to produce expression (8). Because the prices of all other goods are held fixed, this formula embodies partial equilibrium analysis.
212. See supra note 38. For goods that are perfectly elastically supplied (have constant marginal cost), this simplification is unproblematic. However, if the supply curve for some other good is steeply rising- and, moreover, if we are considering a substitute for which the cross-elasticity is rather high and the relative market size is low- then a full general equilibrium analysis would reflect a significant rate of price increase for the substitute, which would to that extent dampen the magnitude of the substitution, implying a greater degree to which price could profitably be elevated in the original market. Such effects are ordinarily set to the side and will not be considered further here, although one can conceive of situations in which this factor would be important. (If one focuses on elasticity rather than cross-elasticity, as argued here, the phenomenon arises implicitly. Note that if one is measuring the elasticity by actual market behavior, such feedbacks might be captured, although they would not be if the analysis held constant- controlled for- prices of such substitutes.)
213. d is used to maintain consistency of notation with that in Part II.
214. Some readers may be familiar with the fact that, say, a 1% rise in x and a 1% fall in y does not leave the product xy unchanged (instead it equals (1.01x)(0.99y) = 0.9999xy). Since the formal definition of the elasticity involves the derivative, however, the elasticity is the limiting value of this factor as the magnitude of the change approaches zero, so formally the elasticity is 1.0.
215. i1? At this point, the left side indicates the change in revenue spent on good 1, which must be equal to the change in revenue spent on all the other goods, which in turn is given by the sum of the cross-elasticities, each weighted by the amount of revenue spent on the respective good.
216. By similar reasoning, negative terms (complements) reduce the elasticity.
217. But see infra note 94.
218. One might think that omitting small values would involve little error. This would be true if, for example, there were only two other goods in existence, the cross-elasticity on one was large and the other tiny, and there were not offsetting expenditure shares (further discussed in subsection 2). However, in a developed economy in which products number in the thousands (or vastly more, depending on how one groups them), looking at the few with the largest cross-elasticities and ignoring myriads of others, even if they each individually have small cross-elasticities, could lead to substantial underestimates of the elasticity.
219. AREEDA, KAPLOW & EDLIN, supra note 14, at 488-89.
220. Id. at 489 n.17.
221. If some other goods were complements, for which the cross-elasticity is negative, then it is possible that the sum of omitted terms would be negative, in which case the method in the text would not provide a lower bound. As a practical matter, the technique would likely be problematic only if strong complements were omitted, which omission should therefore be avoided.
222. Note further that we really cannot tell how tight this lower bound is (whether the true elasticity is only slightly above or far higher) unless we know, in aggregate, what we have omitted. The magnitude of the omission could be estimated if we knew the extent to which consumers switched to all other goods, but in that case we would already know the elasticity.
223. See Kaplow, supra note 22, at 1831 n.56
224. AREEDA, KAPLOW & EDLIN, supra note 14, at 495
225. Werden, supra note 13, at 402
226. AREEDA, HOVENKAMP & SOLOW, supra note 2, ¶ 507a, at 131.
227. For an exception, see Kaplow & Shapiro, supra note 12, at 1092. The U.S. Merger Guidelines might also be viewed as an exception, although their approach to market definition addresses a different question from what is conventional and from what is most relevant to assessing the competitive effects of mergers. See infra note 99.
228. Suppose, for example, that we compare two cases: in one, the cross-elasticity is 2.5, moderately over the minimum level of 2.0 we might demand for inclusion, and in another, the crosselasticity is 10.0. If the revenue shares in each case were the same, the actual impact of the substitute on market power would be only a fourth as high in the former case, even though both are treated the same in conventional analyses.
229. See U.S. MERGER GUIDELINES, supra note 5, § 4.1.1
230. DG Competition Article 82, supra note 1, ¶ 14 (SSNIP test used by the Commission for market definition, particularly in merger cases). It might be that, in a narrow market, the hypothetical monopolist would not quite muster a 5% price increase (SSNIP), so the market is broadened to include a substitutes market that, although being closest, has only a modest cross-elasticity but whose size is very large relative to that of the original market. Levels of and increases in the Herfindahl-Hirschman Index (HHI) in the broader market could readily be very low even if the merger significantly enhances market power. "Significantly" is used in a relative sense: after all, we know that the price elevation must be less than 5%; when the hypothetical monopolist can raise prices at least that much, say 7%, what would be viewed as significant HHIs and increases in the HHIs may readily translate into price increases that are much less than 5%. To make this point concrete, consider the following example. Suppose that the elasticity of demand is as high as 10, in which case a firm with a 100% market share (hypothetical monopolist) would have a Lerner index of 0.10 (simplifying throughout by taking rivals' supply elasticity to be zero). (As explained in section II.A, this translates into a price elevation of just over 11%, but it will suffice for present purposes to use rounded approximations, in this case 10%.) If, prior to the merger, we had Cournot competition and the premerger HHI was 2000, then the premerger margin is 0.2/10 or 2% (using the formula appearing, for example, in Ordover
231. Sykes & Willig, supra note 25, at 1865
232. Kaplow & Shapiro, supra note 12, at 1085, under which the industry-wide average, output-weighted margin equals the HHI of market concentration divided by the elasticity of market demand (and keeping in mind that HHIs are typically measured on a scale of 0 to 10,000 rather than 0 to 1, so conventional HHIs, in terms of the relevant formula, indicate tenthousandths)). Therefore, a hypothetical monopolist would raise its price about 8% (the difference between the monopoly elevation of approximately 10% and the premerger elevation of 2%). This passes the Merger Guidelines' SSNIP test. Suppose that the actual merger raises the HHI from 2000 to 2600, which would be regarded as a large increase from an already substantial level
233. See U.S. MERGER GUIDELINES, supra note 5, § 5.3 ("Mergers resulting in highly concentrated markets [with an HHI above 2500] that involve an increase in the HHI of more than 200 points will be presumed to be likely to enhance market power."). But how much will the price increase? Using the formula, the postmerger HHI corresponds to a postmerger margin of 0.26/10, or 2.6%. (This mode of postmerger analysis, although common, is incomplete because it ignores that the share of the merged firms will not in general equal the sum of their premerger shares and other firms' shares will change as well; for a rigorous analysis
234. Joseph Farrell & Carl Shapiro, Horizontal Mergers: An Equilibrium Analysis, 80 AM. ECON. REV. 107 (1990).) Therefore, the merger that is solidly in the danger range raises price by only about 0.6% (just over half a percentage point). (We are not assuming the presence of merger efficiencies or other factors that under the Merger Guidelines might modify the bottom line.) By contrast, if a merger substantially raised the possibility of coordinated behavior (or was a merger to monopoly), producing the hypothetical monopoly result, but the SSNIP test in the homogeneous goods market suggested an elevation of only 4% (nearly seven times as much), the market would be expanded, and if the "next" market had a high level of revenue relative to that in the original, narrow market, the merger may well pass under the radar screen. This juxtaposition shows how, to a significant degree, the Merger Guidelines provide determinacy at the expense of coherence. The problem is that available information is not used directly to ascertain how much the merger will increase price.
235. Note that the problem has nothing to do with the change in the revenue weight occurring the next day; if ovens had been present from the outset, the dilemma would be the same. Indeed, this dilemma is simply one of the standard problems with using market definition: in general, no market definition captures the strength of substitution, a point that has already been made in various ways in prior Parts of this Article.
236. Indeed, it might seem natural to consider as the closest substitute that with the highest cross-elasticity. However, the most important substitute- the one responsible for most restraining market power- may well be another good with a lower (possibly much lower) cross-elasticity but a sufficiently greater revenue share to result in a product of the two numbers that is higher than that for the substitute with the highest cross-elasticity. See generally Werden, supra note 13, at 402-06 (discussing cross-elasticities and different ways to rank substitutes).
237. See, e.g., AREEDA, KAPLOW & EDLIN, supra note 14, at 495 n.29
238. Simons & Williams, supra note 83, at 827
239. Werden, supra note 13, at 402 n.152
240. see also Jerry A. Hausman & Gregory K. Leonard, Competitive Analysis Using a Flexible Demand Specification, 1 J. COMPETITION L. & ECON. 279 (2005) (surveying methods of estimating elasticities and cross-elasticities, often noting greater obstacles regarding the latter)
241. Kaplow, supra note 22, at 1830 n.55 ("But empirical estimates of elasticity are usually available only for conventionally defined product markets. Because elasticity already incorporates substitution (cross-elasticity) ⋯ , empirical estimates of demand elasticity could be used directly when product markets are not combined. When product markets are combined, however, no immediately usable empirical data are generally available.")
242. Werden, supra note 13, at 363 ("Demand elasticities are actually being estimated, and the estimated demand elasticities are being used to delineate markets, to measure market power directly, and to predict the competitive effects of mergers."). In fact, it sometimes may be best to estimate the firm's elasticity of demand- which, as formula (2) indicates, may enable an even more direct measure of market power than that provided through estimation of market elasticities, much less numerous cross-elasticities.
243. See, e.g., Baker & Bresnahan, Residual Demand, supra note 43.
244. This point is not always true. An advertising campaign aimed at a particular substitute would be based on evidence of perceived cross-elasticity (and the revenue weight).
245. Although not as logically connected to the choice between the two measures, it is notable that many inquiries are conceptually disconnected from the relevant question. Both elasticity and cross-elasticity are concerned with a (percentage) rate of change in purchasing behavior in response to changes in prices. Observing (a static picture of) what consumers are purchasing at given prices does not address this question. It may in our imaginations: we might assume that the reason so many are purchasing y rather than x is that the price of x is elevated. But the quantities purchased of each good do not in themselves answer that question and certainly do not indicate the rate at which consumers' purchases would shift if price changes slightly, which as just stated is central to measuring both elasticities and cross-elasticities. (This point is distinct from the socalled Cellophane fallacy, which involves asking whether a firm, whose status as a monopolist we seek to ascertain, could profitably raise prices even more than it already has- which, if it is a profit-maximizer, it certainly cannot, for if it could have it already would have- rather than whether the firm's prices already entail significant elevation above a competitive level.) As an illustration of how market power analysis can go awry, see Capps, Dranove, Greenstein & Satterthwaite, supra note 43, who criticize the Elzinga-Hogarty geographic market definition test, which is based on shipment flows, and find that, in a hospital merger setting, it can lead to broad market definitions that imply negligible market power whereas more direct approaches demonstrate that significant price increases may well result.
246. That is, the discussion focuses on how much the merging parties would find it in their interest to raise prices, taking as given their competitors' behavior and thus ignoring how the merger may affect the ability to elevate price in a coordinated manner. Likewise, the discussion presupposes a lack of prior coordination (for if, say, there were already complete coordination, then both firms would already be charging essentially the monopoly price, so the merger would not result in a further price increase).
247. As will be mentioned in section VI.C, analogous considerations are applicable if one wishes to determine, for example, the extent to which an alleged act of monopolization directed at producers of a substitute would, if successful, augment the monopolist's market power.
248. See Kaplow & Shapiro, supra note 12, at 1185-86.
249. See, e.g., Jonathan B. Baker & Timothy F. Bresnahan, The Gains from Merger or Collusion in Product-Differentiated Industries, 33 J. INDUS. ECON. 427 (1985)
250. Michael G. Baumann & Paul E. Godek, A New Look at Critical Elasticity, 51 ANTITRUST BULL. 325 (2006)
251. Coate & Williams, supra note 82
252. Farrell & Shapiro, supra note 9
253. Barry C. Harris & Joseph J. Simons, Focusing Market Definition: How Much Substitution Is Necessary?, 12 RES. L. & ECON. 207 (1989)
254. Michael L. Katz & Carl Shapiro, Critical Loss: Let's Tell the Whole Story, ANTITRUST, Spring 2003, at 49
255. Daniel P. O'Brien & Abraham L. Wickelgren, A Critical Analysis of Critical Loss Analysis, 71 ANTITRUST L.J. 161 (2003)
256. see also Kaplow & Shapiro, supra note 12, at 1085- 86, 1173-75 (discussing the diversion ratio and commenting on the critical loss literature).
257. If the margins differ, then the merged firm is not indifferent to sales shifts. Sales moving from the high-margin good to the low-margin good reduce profits (thus the incentive to raise price of the initially high-margin good is lower than the discussion in the text suggests) but sales moving in the opposite direction increase profits (thus raising the incentive). Note also that, as one changes the price of good 1 while holding the price of good 2 constant, the relative margins will change.
258. This point is a central theme of Farrell and Shapiro, supra note 9.
259. See also Werden, supra note 13, at 407 n.164 ("The central role of demand elasticities in merger simulation may be demonstrated most powerfully by the observation that the use of merger simulation eliminates the need for market delineation because the predictions of merger simulations are not sensitive to the make-up of the product group used in the simulations."). Nevertheless, much of the development and use of the critical loss methodology and related inquiries is for purposes of defining markets, often under the U.S. Merger Guidelines.
260. See, e.g., Malcolm B. Coate & Jeffrey H. Fischer, A Practical Guide to the Hypothetical Monopolist Test for Market Definition, 4 J. COMPETITION L. & ECON. 1031, 1040-44 (2008) (section entitled "Critical Loss Analysis for Market Definition").
261. See, e.g., Baker & Bresnahan, Residual Demand, supra note 43.
262. Even when focusing, more appropriately, on the elasticity rather than on particular crosselasticities, problems of determining the proper cutoff for purposes of market definition still arise. For example, in their econometric analysis of margarine and butter, David Scheffman and Pablo Spiller find that the elasticity for margarine is -3.5, which they deem "high ⋯ indicating that margarine is not a relevant market.
263. David T. Scheffman & Pablo T. Spiller, Econometric Market Delineation, 17 MANAGERIAL & DECISION ECON. 165, 175 (1996). Yet they do not explain why this elasticity should be viewed as high rather than low or medium, or why it indicates that the relevant market is broad rather than narrow (just margarine). Indeed, if one applies the hypothetical monopolist test (so the share is 100%) using formula (4), the Lerner index is approximately 0.29 and the implied degree to which price would exceed marginal cost is 40%, which most would take to be quite substantial market power.
264. Landes & Posner, supra note 13, at 948-50
265. U.S. MERGER GUIDELINES, supra note 5, §§ 5.1
266. see also AREEDA, KAPLOW & EDLIN, supra note 14, at 498
267. Fisher, supra note 49, at 15-16
268. Kaplow, supra note 22, at 1847
269. Schmalensee, supra note 14, at 1801. There is some controversy about whether, as a practical matter, it makes more sense to consider supply responses as part of the market definition exercise or in a separate step.
270. See, e.g., Baker, supra note 2, at 133- 38 (advocating that the steps be separate, as under the U.S. Merger Guidelines). Such pragmatic questions are largely distinct from the focus of this Article (although note 116 indicates a possible connection).
271. r is atypically high, and market power actually is overstated when it is unusually low.
272. See supra note 48.
273. Indeed, it can be confusing to make sense of the rivals' supply elasticity in a combined market. As price is increased, more of each of the, say, two goods will be produced by rivals. However, these two sources of supply will, except by coincidence, increase at different rates. The rates could differ widely; for example, there may be capacity constraints in one market and nearly constant marginal cost with no such barrier to expansion in the other. If what we care about is the firm's ability to elevate the price of the initial good, then these two different supply responses have significantly different implications. (Even a hypothetical monopolist of the markets for both goods, as imagined in applying the U.S. Merger Guidelines' methodology, would care about the differential responses and, as a consequence, would elevate the prices in the two markets by different amounts.) Cf. supra notes 38 & 85 (discussing how the use of partial versus general equilibrium analysis is related to making particular assumptions about the supply response in the market for the other good).
274. See supra note 48.
275. For example, it would be a mistake to focus only on certain sources of supply response when there are many, and if one examines, say, the supply elasticity of a particular rival, it is important to weight it by the rival's size. (Thus, if a given rival will expand its supply 3% in response to a 1% increase in price, but that rival supplies only 1% of the market, the aggregate supply response due to this rival would be only 0.03%, which is negligible.)
276. See, e.g., United States v. Grinnell Corp., 384 U.S. 563, 570-71 (1966). EU Article 102 (formerly 82) prohibits the abuse of a dominant position, which clearly requires both a dominant position (taken here roughly to correspond to significant market power) and some form of abuse. Both elements are heavily contested; that is, there is no strong consensus on how much market power is required or on how to define exclusionary practices (or on the possible relationship between the two, notably, whether and how the requisite power should depend on the practice un der scrutiny). These debates are tangential to the analysis of market definition as a means of measuring market power.
277. See, e.g., United States v. Aluminum Co. of Am. (Alcoa), 148 F.2d 416, 430 (2d Cir. 1945) ("The successful competitor, having been urged to compete, must not be turned upon when he wins.").
278. See, e.g., Carlton, supra note 13, at 3 (expressing a concern about inferring market power from market share on the ground that it is difficult to weigh increased market power against possible efficiency gains)
279. id. at 19 ("My experience is that courts ask whether market power exists in the presence of the alleged bad act, a question with the potential to be answered in a misleading way if one ignores the efficiency justification for the alleged bad act ⋯ .")
280. Fisher, supra note 49, at 18 ("If the share is maintained solely because of low prices or better products, then we are looking at what competition is supposed to do and not at a monopoly. This is, of course, closely related to the legal position that a monopoly acquired by 'superior skill, efficiency, or foresight' does not violate the antitrust laws. I would prefer to say that a large share acquired in such ways is not a monopoly at all.")
281. id. at 28 ("A firm which maintains a large share of the market because of behavior forced on it ('economically inevitable') or solely because of being better ('superior skill, efficiency, and foresight') is a firm which does not have monopoly power at all. Monopoly power is the power to maintain a high share and earn supranormal profits without being better.").
282. Long-run marginal cost may be higher or lower than short-run marginal cost. It may be higher because fixed costs are variable in the long run and thus are included, but it may be lower because capacity may be expanded or technology updated.
283. See, e.g., Fisher, Economic Analysis, supra note 41, at 139
284. Werden, supra note 13, at 381- 82.
285. See, e.g., Fisher, supra note 49, at 10-12, 20-22. This scenario is questionable, for in a competitive market, price would equal short-run marginal cost even after an unexpected demand spike. Each firm would equate its marginal cost to the higher price; with rising marginal cost curves, this involves some expansion even before new capacity is created. (If capacity is literally constrained, and marginal cost does not rise rapidly as the capacity limit is approached, then firms would produce at capacity, at least in the short run. But a proper definition of marginal cost would have its level rising discontinuously from its prior finite level to infinity at that level of output, and one could thus technically say that the resulting equilibrium was nevertheless consis tent with price equal to short-run marginal cost. Realistically, marginal cost may well rise quickly but not at an infinite rate as "full capacity" is approached, and firms do not often produce at literally 100% of feasible capacity in any event.) Under these circumstances, the imagined temporary divergence- and thus the temporary indication of market power via use of the Lerner index- simply does not arise. (It is true that firms that otherwise would have earned no profits will make some profit temporarily.) Another case sometimes mentioned involves rents on a scarce input owned by a firm. However, if one imagines the resource as if it were owned by a separate entity and thereby treats the opportunity cost of consuming the resource as an input cost, which is theoretically proper, then pricing is not really above marginal cost (although it may exceed a historically based accounting measure of incremental cost).
286. See, e.g., Baumann & Godek, supra note 107, at 329
287. Lawrence J. White, Market Power and Market Definition in Monopolization Cases, in 2 ISSUES IN COMPETITION LAW AND POLICY 913, 921-22 (Wayne Dale Collins ed., 2008
288. cf. Fisher, Economic Analysis, supra note 41, at 137 (expressing a concern about inferring significant market power from a high market share because "a large share can simply represent greater efficiency or product quality on the part of the alleged monopolist").
289. These arguments can be obscure in two ways. First, it may not be explained how prices in excess of marginal cost may be sustained in the presence of competition since firms with marginal cost below price have an incentive to increase output. In supposing that they will refrain, it seems that noncompetitive behavior is contemplated. Second, the meaning of competition is sometimes unclear or the term may be used in unconventional ways, as suggested by the first point. A straightforward case involving price in excess of marginal cost in long-run equilibrium involves product differentiation, where firms incur fixed costs (which limits the extent of entry) and cannot expand output and steal rivals' customers simply by pricing infinitesimally below what others charge, due to consumers having at least some nontrivial preference for different producers' products. See, e.g., JEAN TIROLE, THE THEORY OF INDUSTRIAL ORGANIZATION, ch. 7 (1988).
290. This case is often termed monopolistic competition, a form of imperfect competition, precisely because it does entail each firm possessing some market power (defined in terms of the Lerner index). Further refinement and resolution of these issues is unnecessary for present purposes.
291. Interestingly, this argument about the definition and measurement of market power is generally advanced in the monopolization context without comment on its implications elsewhere, notably, for horizontal mergers. Under this approach, two parties could merge to monopoly and concede that prices would rise, say, 20%, and claim that the government has not made out a prima facie case. The reasoning would be that market power is required for liability and, since market power is defined not as price elevation above marginal cost or what is ordinarily understood to be a competitive price, but above whatever level is necessary to recover fixed costs, incentivize innovation, and so forth, market power has not even been shown to exist unless the monopoly price is demonstrated to be unnecessary in this regard. (Implicitly, the firms are claiming that price was below the appropriate level before the merger.) Of commentators who advance these views about the definition or meaning of market power and concerns about fixed costs, innovation, and so forth, I am unaware of any who then consistently deploy them in this fashion to criticize standard analysis of horizontal mergers.
292. Views closest to those here appear in Schmalensee's commentary on qualms expressed by Landes and Posner, supra note 13, at 956-57, regarding the finding of market power in certain settings.
293. See Schmalensee, supra note 14, at 1795-96
294. id. at 1808 n.74 ("If one followed Landes and Posner's recommendation, most 'clean-handed' monopolists could thus rebut the inference that they had any market power. Evidence of the sources of market power is much more appropriately used to judge the legality of the acquisition and exercise of that power. It would cause considerable confusion if 'market power' were to be redefined (as Landes and Posner implicitly suggest) so that 'clean-handed' firms could never possess it.").
295. Consider another example outside the monopolization context. One commentator would limit market power to prices in excess of levels set by any process, not necessarily competitive as conventionally understood, as long as it is legal; for example, he (like others) holds that price elevation through oligopolistic coordination is legal as long as no secret meetings and the like are involved. See Baker, supra note 2, at 143 n.53.
296. That definition creates problems for the application of a rule of reason analysis to, say, trade association behavior that is alleged to facilitate oligopolistic coordination. If such coordination is legal and is occurring, then the elevated price is the baseline, there is no further elevation relative to that (by definition), and there is thus no market power. As a consequence, if the rule of reason inquiry has a market power prerequisite (so we do not have to scrutinize all manner of cooperative activity), then the trade association would be immune. The response would be to say that, in such a case, the practice should be challenged and, if invalid, then the original premise that the oligopolistic price elevation was perfectly legal no longer holds, so we can find market power after all. As in the example in the text, this manner of thinking and communicating seems needlessly convoluted. What is more surprising is that one would insist on expressing ideas this way rather than in the simpler and more straightforward fashion advocated in the text and practiced broadly.
297. Taking the definitions literally, two internally consistent results are possible, and the formal statements do not provide any basis for selection: (1) There is no market power; hence, there is no violation due to abuse (since it is not counted as an abuse in the absence of market power), and since we have just concluded that there is no finding of abuse, our original statement of no market power is correct (since we deem it nonexistent in the absence of abuse). (2) There is market power; hence, there is a violation due to abuse (which we can deem to exist, having met the market power requirement), and since we have just found an abuse, our original statement that there is market power (which required the simultaneous existence of abuse) is valid. This tension can easily be seen in the preceding note, since Jonathan Baker defines the benchmark price in that case as that reached by "sellers [who] engage in oligopoly conduct without reaching an agreement or otherwise violating the antitrust laws," id. (emphasis added), since the question motivating the market power inquiry is whether the antitrust laws have been violated.
298. If one believes that it is existing practice that is muddled and, in particular, that courts often fail to insist on demonstration of abuse or are too generous in finding it, then the seemingly convoluted approach may nevertheless have pragmatic appeal. In that event, it might be helpful to introduce some new, supplemental terminology, perhaps something like "provisional market power"- that is, market power, as defined by the Lerner index, but deemed tentative until there is a finding that an exclusionary practice (or other anticompetitive behavior) has occurred.
299. The analysis in this brief section is not comprehensive even as to the applications it considers. However, it covers many of the most important settings, and others can largely be assimilated to one or another of them. For example, many horizontal restraints examined under the rule of reason, where market power inquiries may be employed, involve fears about facilitating collusion, so the market power analysis used in the case of coordinated effects in horizontal mergers would roughly be applicable.
300. See Kaplow & Shapiro, supra note 12, at 1185-86.
301. For comments specifically on the approach to market definition in the U.S. Merger Guidelines, see notes 99 and 138.
302. See, e.g., U.S. MERGER GUIDELINES, supra note 5, § 6.
303. See, e.g., SIGRID STROUX, US AND EC OLIGOPOLY CONTROL 187-88 (2004).
304. See, e.g., Kaplow & Shapiro, supra note 12, at 1084-85 (presenting the standard analysis indicating that a firm's markup equals the firm's market share divided by the elasticity of market demand and also the formula indicating that the industry-wide average, output-weighted margin equals the HHI divided by the elasticity of market demand; for an illustration and qualification, see note 99).
305. See, e.g., U.S. MERGER GUIDELINES, supra note 5, § 7.
306. See, e.g., Werden, supra note 41, at 532. If the firms we suppose to be coordinating cannot raise price significantly even if their coordination is perfect, there is no point in inquiring into broader markets. This inquiry would be interesting only if we were supposing that cross-market coordination might occur. The success of such coordination, if achieved, can be determined much as in the case in which we measure the unilateral effects of a merger between two firms producing differentiated products.
307. A number of economists have suggested to me that the market definition metaphor may also summarize decisions about how broadly to collect data on possible substitute products when empirically estimating demand elasticities or for purposes of calibrating merger simulations. Such boundary decisions, however, reflect costs of data collection and empirical hunches on how likely a (moderately to substantially distant) substitute is to influence the estimation of other parameters, a qualitatively different calculus from that involved in defining markets for purposes of making market power inferences. Additionally, some find the metaphor useful to help suggest relevant paths of inquiry at the outset of a market power investigation. This usage- again, quite different from that examined in this Article- may be fairly benign, although pitfalls have been mentioned, including that one may be led to focus on particular cross-elasticities rather than on the demand elasticity.
308. For markets in which even successful oligopolists could not significantly elevate price, one might keep broadening the boundaries, as under the Merger Guidelines, in order to see which broad markets would involve significant price elevations conditional on success. (Indeed, the hypothetical monopolist SSNIP test, as a purely logical matter, seems much more tailored to answering this question, see Kaplow & Shapiro, supra note 12, at 1172, than to those it purports to address, a criticism developed via numerical example in note 99.) However, since coordination would be infeasible in most such broad markets, this method of prioritizing investigations does not seem very sensible.
309. More precisely, as outlined in section IV.A, to have decided that the market power inference from the market share in the broader market is more accurate (or more likely to lead to the right answer) than the market power inference from the market share in the narrower market is to assume the conclusion about market power.
310. There is a real danger of overlooking important instances of market power. For example, in mergers in differentiated product industries, it is easy to reject narrow market definitions as obviously incorrect, but more careful analysis of some such mergers indicates that looks can be deceiving. The most familiar example is FTC v. Staples, Inc., 970 F. Supp. 1066 (D.D.C. 1997). (Even if one disagrees with the government's econometric analysis in that case- a question on which no opinion is offered here- the point is that a conclusion many would have seen as obvious is not.) Additional errors can arise in cases with highly inelastic supply, which may exist with oil refining and electricity generation, for example. In such industries, low shares can convey significant market power (assuming that the market elasticity of demand is not very high), a conclusion that is apparent from expression (4).
311. See, e.g., Kaplow, supra note 22, at 1832
312. AREEDA, KAPLOW & EDLIN, supra note 14, at 492.
313. See supra note 75. One might add that they also may be hypothetical abstractions, concerning, for example, behavior that would occur if a merger were allowed to go forward (making it a bit like asking, at the time a seed is planted, whether the height of a never-before-seen type of tree will be closer to that of one versus another existing tree, a question one could answer only by using available science to predict the height of the prospective tree).
314. See AREEDA, KAPLOW & EDLIN, supra note 14, at 492-93, 498-99.
315. This problem is elaborated in Louis Kaplow, Market Share Thresholds: On the Conflation of Empirical Assessments and Legal Policy Judgments, 7 J. COMPETITION L. & ECON. (forthcoming 2011).
316. Consider a further implication of courts' reluctance to offer quantitative statements about anything other than market share when joined with a further queasiness about being too specific about what market shares suffice. In one hypothetical monopolization case, suppose that one market definition involves a dominant firm share of 90% and another definition results in a share of 55%. Since the court may be uncertain about the right ultimate legal answer if it goes with the latter, it may have some inclination to select the definition yielding the 90% share in order not to have to confront the issue. Likewise, in another hypothetical case, in which one definition yields a share of 55% and another a share of 20%, the latter choice would be more comfortable for the same reason. The murkiness of the market definition process also makes it fairly easy for an adjudicator to reach any decision that might be desired on other grounds (for example, if the claim that a practice is abusive seems frivolous, but the analysis that would support that feeling is timeconsuming and complex, one can often select a broad market definition that renders the other inquiry moot).
317. Even if other approaches were nevertheless allowed, it is hard to see how enforcement agencies and private parties would be prohibited from offering and making central to their case high-quality direct evidence of market power, including methods that economists have been developing in recent decades and will continue to formulate.
318. The problem concerning the use of experts, which is widespread in market definition inquiries, is severe. In light of this Article's analysis, what could an expert say when purporting to offer an opinion on market definition (unless taking this results-oriented out)? For example, if asked whether more conventional analysis of market definition was rooted in accepted economic understanding, an honest answer (other than "no, not at all") would be difficult to come by. As mentioned in subsection II.B.2, the construct really does not exist in industrial organization economics. As a consequence, an argument could be made in U.S. courts that conventional expert opinions on market definition should not be admissible. Rule 702 of the Federal Rules of Evidence- as amended in light of Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993), and subsequent cases- requires that an expert's "testimony is the product of reliable principles and methods." Query what those principles and methods could be with regard to market definition in light of the concept's nonexistence in the field of expertise and the foregoing analysis in the text.
319. FTC v. Staples, Inc., 970 F. Supp. 1066 (D.D.C. 1997).
320. Id. at 1075-81
321. see also James A. Keyte & Neal R. Stoll, Markets? We Don't Need No Stinking Markets! The FTC and Market Definition, 49 ANTITRUST BULL. 593 (2004) (arguing that the FTC has been moving away from market definition in merger and nonmerger cases)
322. Robert H. Lande & James Langenfeld, From Surrogates to Stories: The Evolution of Federal Merger Policy, ANTITRUST, Spring 1997, at 5, 6 ("[T]here have been many recent attempts by the Federal Trade Commission and the Department of Justice to shift the focus of investigations away from market definition ⋯ [toward] try[ing] directly to predict what will happen to future industry prices.")
323. Kühn, supra note 80, at 316 (suggesting that the European Commission "appears to be open to the general trend of moving toward a practice of joint determination of market definition and market power instead of insisting on a mechanical two-step exercise of market definition and market share assessment").
324. There are exceptions. See ten Kate & Niels, supra note 41, at 297 ("Some more recently installed competition regimes, such as Mexico in 1992, have even incorporated the term 'relevant market' explicitly in the official language of their competition statutes, thereby turning it from an analytical tool in the determination of market power into a statutory prerequisite to decisions in competition cases.").
325. Cont'l T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36, 54, 56 (1977)
326. (overruling United States v. Arnold, Schwinn & Co., 388 U.S. 365 (1967)).
327. Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 587 (1986).
328. Id. at 596.
329. Leegin Creative Leather Prods., Inc. v. PSKS, Inc., 127 S. Ct. 2705 (2007).
330. Dr. Miles Med. Co. v. John D. Park & Sons Co., 220 U.S. 373 (1911).
331. Leegin, 127 S. Ct. at 2714
332. (quoting Sylvania, 433 U.S. at 58-59). The Leegin Court further emphasized that antitrust principles on vertical restraints were to be formulated by reference to "differences in economic effect" so that "it is necessary to examine ⋯ the economic effects of vertical agreements to fix minimum resale prices" to determine what legal rule should apply.
333. Id.
334. see id. at 2722.
335. See sources cited supra note 13.
336. See, e.g., Spectrum Sports, Inc. v. McQuillan, 506 U.S. 447, 459 (1993) ("The concern that § 2 might be applied so as to further anticompetitive ends is plainly not met by inquiring only whether the defendant has engaged in 'unfair' or 'predatory' tactics. Such conduct may be sufficient to prove the necessary intent to monopolize, which is something more than an intent to compete vigorously, but demonstrating the dangerous probability of monopolization in an attempt case also requires inquiry into the relevant product and geographic market and the defendant's economic power in that market."). It is important to keep in mind, however, that the Court in Spectrum was rejecting an interpretation under which intent or conduct alone was asserted to be sufficient to establish a dangerous probability of success, without any separate market power inquiry; no attention was directed to whether means of establishing market power directly might suffice in lieu of the indirect inference drawn using the market definition process.
337. See also DG Competition Article 82, supra note 1, ¶ 11 ("The concept of dominance contained in Article 82 of the Treaty relates to a position of economic strength on a market. In the application of Article 82 it is therefore necessary to define a relevant market."); Developments- and Divergence- in Merger Enforcement, ANTITRUST, Fall 2008, at 9, 13, 19 (in panel discussion on merger enforcement, European lawyers indicate a greater focus on market definition and market share by the European Commission than by U.S. enforcement agencies).
338. The opinion most associated with this view is United States v. General Dynamics Corp., 415 U.S. 486 (1974), which in turn cites Brown Shoe for the proposition that "statistics concerning market share and concentration, while of great significance, [are] not conclusive indicators of anticompetitive effects."
339. Id. at 498
340. see also Landes & Posner, supra note 13, at 979 (citing various Supreme Court cases in support of "the recent trend toward regarding market share statistics in merger cases as providing merely presumptive evidence of market power, which can be rebutted by bringing in other factors"). The view that market shares require interpretation is not confined to the United States. See, e.g., DG Competition Article 82, supra note 1, ¶ 30 ("In any event, the Commission interprets market shares in the light of likely market conditions, for instance, whether the market is highly dynamic in character and whether the market structure is unstable due to innovation or growth."); id. ¶ 32 ("The strength of any indication based on market share depends on the facts of each individual case. Market share is only a proxy for market power, which is the decisive factor. It is therefore necessary to extend the dominance analysis beyond market shares, especially when taking into account the difficulty of defining relevant markets in Article 82 cases ⋯.").
341. See, e.g., FTC v. Ind. Fed'n of Dentists, 476 U.S. 447, 460-61 (1986) ("Since the purpose of the inquiries into market definition and market power is to determine whether an arrangement has the potential for genuine adverse effects on competition, 'proof of actual detrimental effects, such as a reduction of output,' can obviate the need for an inquiry into market power, which is but a 'surrogate for detrimental effects.'"
342. (quoting 7 PHILLIP AREEDA, ANTITRUST LAW ¶ 1511, at 429 (1986)))
343. Toys "R" Us, Inc. v. FTC, 221 F.3d 928, 937 (7th Cir. 2000) ("[Toys 'R' Us] seems to think that anticompetitive effects in a market cannot be shown unless the plaintiff, or here the Commission, first proves that it has a large market share. This, however, has things backwards. As we have explained elsewhere, the share a firm has in a properly defined relevant market is only a way of estimating market power, which is the ultimate consideration⋯. The Supreme Court has made it clear that there are two ways of proving market power. One is through direct evidence of anticompetitive effects."
344. (citing Ind. Fed'n of Dentists, 476 U.S. at 460-61)
345. Werden, supra note 13, at 380-81 ("While market share has long been the staple of market power analysis, three courts of appeals have held: 'Market share is just a way of estimating market power, which is the ultimate consideration. When there are better ways to estimate market power, the court should use them.'"
346. (quoting Ball Mem'l Hosp., Inc. v. Mut. Hosp. Ins., Inc., 784 F.2d 1325, 1336 (7th Cir. 1986), an opinion by Judge Easterbrook, and citing in support an opinion by now-Justice Thomas, joined by now-Justice Ginsburg))
347. ANTITRUST LAW DEVELOPMENTS, supra note 1, at 229, 347
348. Eric L. Cramer & Daniel Berger, The Superiority of Direct Proof of Monopoly Power and Anticompetitive Effects in Antitrust Cases Involving Delayed Entry of Generic Drugs, 39 U.S.F. L. REV. 81, 100 (2004).
349. See, e.g., LAWRENCE A. SULLIVAN & WARREN S. GRIMES, THE LAW OF ANTITRUST 74 (2d ed. 2006) ("2.6d. A Showing of Anticompetitive Effect May Obviate the Need for Market Definition [(section heading)]. Antitrust focuses on abusive exercise of market power. The exercise of defining the market is merely a tool for determining if that has occurred; it should not be an end in itself. In Eastman Kodak Co. v. Image Technical Services, Inc.[, 504 U.S. 451, 466-67, 477 (1992)], the Supreme Court stressed that 'legal presumptions that rest on formalistic distinctions rather than actual market realities are generally disfavored in antitrust law.' Confronted with Kodak's arguments that aftermarkets were not appropriate for measuring market power, the Court wrote: 'It is clearly reasonable to infer that Kodak has market power to raise prices and drive out competition in aftermarkets, since respondents offer direct evidence that Kodak did so.' Disputes about market definition, then, are of little consequence in the face of actual evidence of anticompetitive effects.")
350. see also AREEDA, HOVENKAMP & SOLOW, supra note 2, at 232-33 (although noting other reasons for the use of market definition, stating: "Finding the relevant market and its structure is typically not a goal in itself but a mechanism for considering the plausibility of antitrust claims that the defendants' business conduct will create, enlarge, or prolong market power⋯ . While the inferences to be drawn from market shares are limited, there may be no practical alternative when the data are insufficient to find the firm's elasticities of supply and demand or its residual demand curves. If known, these numbers would measure market power directly, without any need for market definition. Because they so often lack such data, antitrust courts traditionally define a market and examine the firms' market shares." (footnotes omitted)).
351. Recall that, even if a market definition is required, it makes little difference if discussion of market power inferences need not be influenced much or at all by the market definition that is chosen; that is, the requirement may be treated as a mere formality.
352. U.S.C. § 18 (2006). Importantly, mergers may also be attacked under Sherman Act section 1's prohibition on "[e]very contract, combination ⋯ , or conspiracy ⋯ in restraint of trade or commerce,"
353. id. § 1, so even if a doctrinal impediment exists under the Clayton Act, its significance is unclear. Note further that arguments similar to that in the text on Clayton Act section 7 can be applied to Sherman Act section 2's language referring to monopolization of "any part of the trade or commerce among the several States, or with foreign nations."
354. Id. § 2. The responses in the text are applicable, although it should also be noted that the inference from the statutory language is even weaker with the Sherman Act (in part because the clause seems to refer to Congress's authority to regulate interstate and foreign commerce), and the Supreme Court has for over a century failed to interpret the Act literally.
355. See, e.g., Standard Oil Co. v. United States, 221 U.S. 1, 63-65 (1911) (rejecting the view that "every contract ⋯ in restraint of trade" applies to every contract that in some fashion restrains trade; instead the determination of which restraints are prohibited is governed by a rule of reason).
356. See, e.g., AREEDA, HOVENKAMP & SOLOW, supra note 2, at 233 & n.4 (citing Brown Shoe as among the Supreme Court cases endorsing the need to define markets to satisfy Clayton Act section 7's language, relying on the language quoted in the text to follow, without mentioning the opposite available interpretation).
357. Brown Shoe Co. v. United States, 370 U.S. 294, 325 (1962). Moreover, earlier in the opinion, when discussing the legislative history of the statute, the Court stated that "Congress neither adopted nor rejected specifically any particular tests for measuring the relevant markets, either as defined in terms of product or in terms of geographic locus of competition, within which the anticompetitive effects of a merger were to be judged." Id. at 320-21. In discussing horizontal aspects of the merger, the Court elaborated: Section 7 of the Clayton Act, prior to its amendment, focused upon this aspect of horizontal combinations by proscribing acquisitions which might result in a lessening of competition between the acquiring and the acquired companies. The 1950 amendments made plain Congress' intent that the validity of such combinations was to be gauged on a broader scale: their effect on competition generally in an economically significant market. Thus, again, the proper definition of the market is a "necessary predicate" to an examination of the competition that may be affected by the horizontal aspects of the merger. Id. at 335 (footnote omitted) (quoting United States v. E.I. du Pont de Nemours & Co., 353 U.S. 586, 593 (1957)). This language suggests that the motivation for defining markets was to encompass anticompetitive effects that might lie beyond those confined to the two merging parties rather than to limit the law's application to some subset of demonstrable anticompetitive effects.
358. One might add the general points that formalistic interpretations of the antitrust law have long been rejected (in addition to the prior discussion in the text); that earlier decisions, including Brown Shoe in particular, are no longer followed along many dimensions; and that, for those inclined toward legislative fidelity in this realm, the legislative history of the 1950 enactment creating the current version of section 7 is largely at odds with the central thrust of modern merger law more broadly (and also fails to endorse economic markets understood in terms of horizontal competitors). See, e.g., Derek C. Bok, Section 7 of the Clayton Act and the Merging of Law and Economics, 74 HARV. L. REV. 226, 236-37 (1960) ("To anyone used to the preoccupation of professors and administrators with the economic consequences of monopoly power, the curious aspect of the debates is the paucity of remarks having to do with the effects of concentration on prices, innovation, distribution, and efficiency. To be sure, there were allusions to the need for preserving competition. But competition appeared to possess a strong socio-political connotation which centered on the virtues of the small entrepreneur to an extent seldom duplicated in economic literature.").
359. Indeed, much of current market definition practice in U.S. courts is attributable to the Horizontal Merger Guidelines. See Hillary Greene, Guideline Institutionalization: The Role of Merger Guidelines in Antitrust Discourse, 48 WM. & MARY L. REV. 771 (2006). They have been reformed in 2010 in a manner that reduces the emphasis on traditional market definition, see sources cited supra note 5, and could be relaxed further.
360. As explained in Part IV, the argument does not depend on the quality of one's best estimate of market power or the types of evidence from which it is derived.