ISSN 1045-6333 WHY ABOVE-COST PRICE CUTS TO DRIVE OUT ENTRANTS DO NOT SIGNAL PREDATION OR EVEN MARKET POWER - AND THE IMPLICATIONS FOR DEFINING COSTS Einer Elhauge Discussion Paper No. 383 09/2002 Harvard Law School Cambridge, MA 02138 The Center for Law, Economics, and Business is supported by a grant from the John M. Olin Foundation. This paper can be downloaded without charge from: The Harvard John M. Olin Discussion Paper Series: http://www.law.harvard.edu/programs/olin_center/ JEL Classifications: K00, K21, L00, L10, L12, L40, L41 WHY ABOVE-COST PRICE CUTS TO DRIVE OUT ENTRANTS DO NOT SIGNAL PREDATION OR EVEN MARKET POWER – AND THE IMPLICATIONS FOR DEFINING COSTS (forthcoming YALE L.J.) By Einer Elhauge1 Abstract Recently, European and U.S. officials have made surprising moves toward restricting firms from using above-cost price cuts to drive out entrants. This Article argues that these legal developments likely reflect the fact that scholarly critiques of cost-based tests of predatory pricing have never been satisfactorily addressed, and offers a better explanation about why restrictions on reactive above- cost price cuts are likely to be undesirable. It begins concluding that “costs” should be defined functionally as whichever cost measure assures that prices above costs cannot deter or drive out equally efficient rivals, and shows how applying that functional benchmark resolves numerous apparent anomalies in current predatory pricing law. It then shows that reactive above-cost price cuts do not necessarily indicate an undesirable protection of market power, but rather can be an efficient response to deviations from the output-maximizing price discrimination schedule in competitive markets, particularly in the airline industry that has been the greatest cause of concern. Even when an incumbent does have market power, restrictions on reactive above-cost price cuts have mainly undesirable effects. They fail to encourage entry but do raise post-entry prices in the bulk of cases, where the entrants are or will predictably become as efficient as the incumbent, or would have entered anyway despite relative inefficiency. They can only weakly encourage less efficient entry since the restrictions cannot protect less efficient entrants in the long run, and even in such cases they have mixed effects on post-entry prices since they give incumbents perverse incentives to raise post-entry prices to speed the day when the restriction expires. In all cases, they impose wasteful transition costs and losses in productive efficiency, and they lessen incentives to create more efficient incumbents and entrants. These adverse effects are worsened by implementation difficulties that cannot be avoided no matter how the rules are defined, including that possible definitions of the moment of entry or exit either make the restrictions ineffectual or make their adverse effects last far longer than any benefits from entry, that they will inefficiently increase or decrease innovation rates, and that any price floor or output ceiling will either cause inefficiencies because of either great uncertainty or inflexibility in the fact of changing market conditions. 1 Professor of Law, Harvard Law School. I am grateful for funding by Harvard Law School, the Handler Foundation, and the John M. Olin Center for Law, Economics and Business at Harvard Law School. I have also benefitted from comments on earlier drafts by Jonathan Baker, William Baumol, Joe Brodley, Bruce Hay, Al Klevorick, Michael Levine, Janus Ordover, Mark Ramseyer, Mark Roe, Hal Scott, Bill Stuntz, and Gregory Werden, and from comments by workshop participants at the American Law and Economics Association Annual meetings, and the U.S. Federal Trade Commission. © Einer Elhauge 2001-02. WHY ABOVE-COST PRICE CUTS TO DRIVE OUT ENTRANTS DO NOT SIGNAL PREDATION OR EVEN MARKET POWER – AND THE IMPLICATIONS FOR HOW TO DEFINE COSTS By Einer Elhauge I. THE CURRENT STATE OF LEGAL FLUX AND SCHOLARLY DEBATE A. Legal Developments and Ambiguities B. The Inadequacy of Traditional Responses in Either Direction II. WHAT PRICING ABOVE COST MEANS A. The Murky and Divided Nature of the Current Debate Over Cost Definitions B. Use Whatever Costs Are Variable During the Period of Predatory Pricing C. Use Variable Costs of Replacing the Victim’s Output, Not of Producing the Predator’s Output D. If Short Term Pricing Can Deter Long Term Investments, Then Use Magnitude of Predator Costs for the Sorts of Costs Variable to the Victim, But Look to the Future to Measure Cost Magnitudes E. If (as Likely) Short Term Pricing Cannot Deter Long Term Investments, Then Use Sorts of Costs Variable to Predator During the Period of Predatory Pricing F. Conclusion on Proper Cost Measure III. REACTIVE PRICE CUTS TO DRIVE OUT ENTRANTS NEED NOT INDICATE INCUMBENT MARKET POWER – AND THE IMPLICATIONS FOR DEFINING COSTS AND MARKETS WHERE COMMON COSTS EXIST A. Individual Routes in Hub-and-Spoke Systems Cannot Be Assumed to be Separate Markets B. Why Competitive Markets Induce Price Discrimination that Maximizes Output C. Why Competitive Price Discrimination Will Often Require Reactive Above-Cost Price Cuts IV. RESTRICTING ABOVE-COST PRICE CUTS HAS ADVERSE EFFECTS EVEN WHEN THE INCUMBENT DOES HAVE MARKET POWER AND IMPLEMENTATION DIFFICULTIES ARE IGNORED A. Effects on Likelihood and Consequences of Less Efficient Entry 1. Consequences for Less Efficient Entrants Who Would Have Entered Without Any Restriction 2. Effects for Less Efficient Entrants Whom the Restrictions Encourage to Enter B. Effects on Likelihood and Consequences of Efficient Entry 1. Post-Entry Effects 2. Ex Ante Effects on Creation of More Efficient Entrants 3. The Restrictions Cannot Reasonably Be Construed or Modified to Eliminate These Effects on Efficient Entrants C. Effects For Entrants Who Can Overcome Their Initial Efficiency Disadvantage 1. When Overcoming Incumbent Efficiency Advantage Necessitates Some Deterioration in Incumbent Efficiency 2. When Increased Entrant Efficiency Suffices to Overcome Incumbent Efficiency Advantage 3. Entrants That Share the Incumbent’s Declining Cost Curve. D. Ex Ante Effects on Incumbent Incentives 1. The Likelihood and Legality of Encouraging Limit Pricing 2. Reduced Incentives to Create Efficient Incumbents E. Summary of Effects and Assessment of Possible Tradeoffs F. The Restrictions Cannot Reasonably Be Construed or Modified to Eliminate the Market Power Requirement V. UNAVOIDABLE IMPLEMENTATION DIFFICULTIES WORSEN THE ABOVE EFFECTS A. When Is the Moment of Entry? B. Difficulties in Defining the Incumbent Price Floor or Output Ceiling C. Post-Entry Quality Changes D. Conclusion on Implementation Difficulties VI. THE BAUMOL BAN ON IMPERMANENT REACTIVE PRICE CUTS A. Post-Entry Effects B. Implementation and Incentive Problems C. Ex Ante Effects D. Conclusion on Banning Impermanent Price Cuts WHY ABOVE-COST PRICE CUTS TO DRIVE OUT ENTRANTS DO NOT SIGNAL PREDATION OR EVEN MARKET POWER – AND THE IMPLICATIONS FOR HOW TO DEFINE COSTS By Einer Elhauge ©2002 Einer Elhauge. All Rights Reserved In the early 1990s, antitrust law on both sides of the Atlantic appeared to have reached a consensus that predatory pricing required proof of below-cost prices.1 But the last few years have witnessed a surprising movement toward prohibiting firms from responding to entry with above-cost price cuts. The European courts got things rolling with a 1996 decision holding it illegal for monopolists to adopt selective above-cost price cuts that sacrificed revenue in order to eliminate entrants.2 Then in 1998, the U.S. Department of Transportation proposed a regulation banning major incumbent airlines from reacting to entry with above-cost price cuts or capacity increases that resulted in “substantially” lower short-term profits than alternative pricing would have.3 In May 1999, the U.S. Department of Justice brought the American Airlines litigation based on the similar theory that it was predatory to respond to entry with business practices that (even if above-cost) “clearly” sacrificed profits.4 This government theory was supported by several expert economists, including the Nobel Prizewinning Professor Joseph Stiglitz.5 And now an important new article by Professor Aaron Edlin proposes the even broader rule that, where an entrant charges at least 20% below the prevailing price, a monopolist cannot respond with any price cut at all for 12-18 months or until its loses its monopoly.6 All these positions restrict reactive above-cost price cuts (or output increases) even if they result in prices that meet (rather than undercut) the entrant’s price, on the notion that buyers would likely stick with the incumbent unless the entrant can offer a lower price. The basic concept underlying these new legal developments and proposals is hardly new. Some courts and scholars have long thought reactive above-cost price cuts designed to drive out entrants were predatory, and the idea was a standard staple of Socratic dialogue in antitrust classes.7 The Edlin proposal is the same as Professor Williamson’s famous 1977 proposal except that it substitutes a ban on incumbents lowering their price for Williamson’s ban on incumbents increasing 1 See infra Part I.A. 2 Id. 3 Id. 4 Id. 5 U.S. v. AMR, 140 F. Supp. 2d 1141, 1180, 1191 (D. Kansas 2001). 6 Aaron S. Edlin, Stopping Above-Cost Predatory Pricing, 111 Yale L.J. 941, 945-46 (2002). 7 See Transamerica Computer v. IBM, 698 F.2d 1377, 1386-88 (9th Cir. 1983); International Air v. American Excelsior, 517 F.2d 714, 724 (5th Cir. 1975); Scherer, Predatory Pricing and the Sherman Act: A Comment, 89 HARV. L. REV. 869 (1976); Williamson, Predatory Pricing, 87 YALE L.J. 284, 290-92 (1977) [hereinafter “Williamson, Predatory Pricing”]; Baumol, Quasi-Permanence of Price Reductions, 89 YALE L.J. 1, 2-3 (1979) [hereinafter Baumol, Quasi-Permanence]; see generally III AREEDA & TURNER, ANTITRUST LAW 161-64 (1978) (discussing but rejecting the general theory); III P. AREEDA & H. HOVENKAMP, ANTITRUST LAW 337 (Rev. ed.1996) (specifically considering and rejecting an Edlin-like ban on any price reduction);, ANTITRUST ANALYSIS 178 (2d ed. 1974) (offering typical set of Socratic questions to present this (and other) theories of predatory pricing). -1- their output for 12-18 months after entry.8 Edlin’s proposal also has much in common, as he acknowledges, with Professor Baumol’s ingenious 1978 idea of permitting reactive price-cuts only if they are quasi-permanent.9 These are legendary economists. The approach of the E.U. and U.S. Departments in turn has roots in various cases and scholarship that defined a predatory price as one that would not maximize profits unless it could destroy or discipline competitors.10 The scholars supporting this approach in writings from 1977-81 included such heavy hitters as Professors Sullivan, Ordover, Willig, Joskow, and Klevorick.11 By the early 1990s, however, this earlier wave of theories seemed safely buried, in an apparent triumph for the Areeda-Turner position that predatory pricing must be below cost. But now they have resurfaced in these modern legal developments, partly because cases and scholars defending the cost-based rule rested mainly on conclusory definitions or claims about administrability,12 which never provided a satisfactory theoretical response to the critics nor addressed practical objections to actual industry behavior under such a rule. Critics were particularly provoked by an apparently serious problem confronting the airline industry.13 On many routes there is an incumbent airline that dominates business at that route and sells at a price well above its costs for that route. Periodically, another airline enters the market at a lower price. The incumbent firm then lowers its price to beat (or match) the entrant. The incumbent never prices below its own costs. But because the entrant has higher costs (or lower quality) it cannot compete at the new price, and is driven out of the market. Once the less efficient entrant is safely gone, the incumbent re-establishes the old price. The concern is that such reactive temporary price-cuts not only drive out entrants, but deter similar entry in the future, and thus allow the more efficient incumbent to perpetuate monopoly prices that exceed the price the next most efficient firm would charge. Moreover, although airlines present the concern in particularly stark form, this concern can exist in any industry where 8 Williamson, Predatory Pricing, supra note , at 295-96, 333-36. This is, however, a real substantive difference. If demand is constant, the addition of entrant output means the Williamson rule would allow the incumbent to lower its price (to maintain its output), whereas the Edlin rule would require the incumbent to lower its output (to maintain its price). For that very reason, Williamson had in1977 considered and rejected the alternative of banning incumbents from lowering their prices in response to entry, which apparently originated in the 1976 trial testimony of Professor Oxenfeldt. Id. at 296 n.39, 318-20, 328 & nn.109-110, 338 (referring to this 1970s articulation of the Edlin rule as the “price maintenance” or “price umbrella” rule). Id. But other than applying his own conclusory labels that a price maintenance rule would be “protectionist” and protect “competitors rather than competition,” id. at 328, 338, Williamson never really explains why this reduction in post-entry output should be a decisive objection, especially since under his own model the price maintenance rule would also imply higher pre-entry output. Areeda & Turner, Williamson on Predatory Pricing, 87 YALE L.J. 1337, 1340-43 (1978). This is the opening Edlin cleverly pursues. 9 Baumol, Quasi-Permanence, supra note , at 4-6; Edlin, supra note, at 978. Again, this does not mean the differences are not substantively significant. See infra at VI. Baumol’s rule was actually first proposed by Professors Areeda and Turner but rejected by them. Areeda & Turner, Predatory Practices under Section 2 of the Sherman Act, 88 HARV. L. REV. 697, 708-09 (1975). 10 See infra Part I.A. 11 Id. & note __. 12 See infra Part I.B. 13 This was the direct motivation for the Department of Transportation and Department of Justice efforts. See 63 Fed. Register at 17920-22; 140 F. Supp. 2d at 1145-69 (recounting similar examples). Airline examples also form the main examples motivating the Edlin analysis. See Edlin, supra note, at 942-43, 980-87. This concern with above- cost airline predation even goes back to Professor Baumol. See Baumol, Quasi-Permanence, supra note , at 2. -2- incumbent firms are more efficient than potential entrants and exploit their market power when entrants are not present to charge prices well above incumbent costs. Indeed, if valid, these concerns would overturn a general current skepticism based on the presumption that predatory pricing is rare because it requires the incumbent to sustain losses on a large number of sales.14 If harmful predation involves profitable above-cost pricing, it would be far more plausible and prevalent. This is a serious concern that can no longer be suppressed with conclusory labels or contestable claims about relative administrability. Unless more seriously addressed, these unanswered concerns about above-cost reactive price cuts will likely continue to influence and expand the development of legal doctrines to deal with those concerns in the U.S. and Europe, both for antitrust law and regulatory agencies, as well as bias conclusions about what counts as a cost whenever a cost-based test is still used. It is thus time to take the idea of restricting above-cost reactive price cuts more seriously. But it is not time to adopt that idea. To the contrary, this Article shows that seriously confronting the idea reveals several heretofore unappreciated flaws in such restrictions. First, such restrictions will often penalize efficient pricing behavior when incumbents do not even have market power. This is because, in many markets, incumbent firms can maximize profits and output by charging more to customers that value the product more highly, thus effectively making them pay a price above their proportionate share of common costs. Even if competition or low entry barriers mean that incumbent firms cannot earn profits from all its customers that exceed economic costs, these firms will often be driven to the profit-maximizing price discrimination schedule in order to cover common costs and maximize output. This includes airlines that do not earn positive economic profits, but do charge more for a ticket that offers one nonstop flight than for the same flight when bundled with a connecting flight. An entrant who cherry picks by selling only to the high value customers at a lower price will thus undercut an output-maximizing price schedule. In order to continue to cover common costs, incumbents will have to react to such entry by lowering their prices to those high-value customers. This reactive above-cost price cut will drive a less efficient entrant out of the market. But this does not mean the price cut protected incumbent market power and harmed efficiency and consumer welfare. To the contrary, it means that the initial discriminatory pricing schedule never indicated market power, and that allowing the normal competitive process of price cuts to drive out the entrant restores the market to an efficient and output-maximizing equilibrium. For airlines, it means that individual routes in a hub-and-spoke system with common costs probably should not be treated as separate markets at all, as critics of airline’s reactive price cuts in individual routes have assumed. Rather, predatory pricing claims probably should instead consider whether the airline’s overall price schedule produced revenue that failed to cover the combination of its common and separate costs. Second, even if the incumbent does have market power – and we (heroically) assume away the difficulties of implementing the restrictions -- the effects of these restrictions are generally undesirable. This is not because, as commonly supposed, the restrictions exchange a certain short- 14 See, e.g., AREEDA & HOVENKAMP, supra note , at 224-25 (collecting sources and linking them to argument that rareness of predatory pricing means courts are more likely to erroneously condemn desirable pricing than correctly condemn predatory pricing); id. at 226 (assuming predatory must involve the temporary sacrifice of revenue). But see Brodley, Bolton & Riordan, Predatory Pricing: Strategic Theory and Legal Policy, 88 GEO. L.J. 2239, 2241 (2000) (arguing that modern economic literature contravenes earlier claims that below-cost predatory pricing was irrational). -3- term loss for an uncertain long term gain.15 To the contrary, it turns out to be futile to try to encourage long-term entry with restrictions on reactive above-cost price cuts. Firms that are or will become equally efficient will enter and stay in the market even with the prospect of above-cost price cuts and thus will not be encouraged by the restriction. Less efficient firms will be driven out when any restriction expires by passage of time or loss of monopoly power. On the other hand, the restrictions may be an uncertain short-term gain. Namely, the restrictions may weakly encourage additional entry by less efficient firms despite the lack of long term protection when entry costs are sufficiently low that less efficient entrants find it profitable to engage in hit-and-run entry, staying only for the short run period while the restriction lasts. But even in such cases the effects on prices are mixed because a restriction on reactive price cuts can give incumbents perverse incentives to raise post-entry prices to speed the day when the restriction expires. Further, the restrictions will clearly increase prices and harm consumer welfare in the lion’s share of cases, when entrants are (or will become) equally efficient or when less efficient entrants would have entered even without the restrictions. And in all cases the restrictions will lower productive efficiency and impose wasteful transition costs. Worse, the restrictions will less incentives to create more efficient entrants and incumbents, which will mean higher costs and lower quality for society generally. Third, these adverse effects are worsened by implementation difficulties that are not avoidable but are rather an inherent consequence of trying to regulate firm pricing, output and responsiveness to entry. While prior analysis has assumed an unambiguous moment of entry, in fact that moment has many possible definitions. Defining entry at the moment when the entrant actually begins sales would, given the normal lead time for entry, allow the incumbent to make anticipatory price cuts that have the same effect as a reactive one. So would any definition of the moment of entry that does not coincide when the time when entry is first foreseeable. Defining entry at an earlier moment when entry is foreseeable (such as when the entrant first begins to plan for entry) would likely mean any 12-18 month restriction would expire by the time the entry starts. One might try to avoid the latter problem by lengthening the period of the price restriction, but the longer the period of restriction the greater the inefficiencies that will result from uncertainties or inflexibilities in the price floors or output ceiling in the face of changing market conditions. Further, any definition of entry that begins before the entrant actually begins sales means that incumbent prices would be artificially elevated during a period where this is not offset by possibly lower entrant prices, thus worsening the likely mix of effects. Another difficulty is that any price or output floor will provoke inefficient increases in product quality, and any effort to clamp down on that by restricting product enhancements will hamper efficient innovation. Finally, any price floor or output ceiling will create additional inefficiencies because they will either embody an inflexible rule, which will cause inefficiencies in market pricing or output given changing market conditions, or a flexible standard, which will create similar inefficiencies because of application imprecision and uncertainties. These implementation difficulties cannot be dismissed as mere administrative concerns since their effect is to raise prices, hamper market flexibility, and distort innovation. These additional adverse effects must thus be added to all the other effects noted above. In sum, the restrictions will not have any benefit outside the limited case where less efficient entrants face low enough entry costs that they might be encouraged to engage in hit-and-run short term entry and exit against an incumbent who was really exercising pre-entry market power. And 15 See infra at I.B, IV.E (collecting current sources stating this is the tradeoff). -4- even in that case, the net effects are mixed even if we did not consider implementation difficulties, and become even worse when we do. Further, the restrictions will have clear adverse effects for cases involving any other sort of entrant, and also discourage investment and innovation in creating more efficient firms. These points are all entirely separate from the lively debate about whether even below-cost predatory pricing should be banned. Many scholars think even below-cost pricing should be legal because it inflicts greater losses on the predator than its victims, rarely garners a future recoupment that compensates for losses given time and uncertainty discounts, and can be thwarted by entrant or consumer counter-strategies, all of which make below-cost pricing self-deterring and too irrational to be credible.16 Others have reached a different conclusion about these arguments based mainly on arguments about differential access to capital to cover losses, multi-market reputational effects, imperfect information, or efforts to mislead rivals (or the capital markets than might fund rivals) about predator efficiency or market conditions.17 This Article takes no position on these disputed issues about the desirability of banning below-cost predatory pricing. Rather, I focus on the separate theoretical grounds for rejecting any restriction on above-cost predatory pricing. All of which has implications for how courts should define “costs,” an issue now normally resolved by rather atheoretical judgment calls that result in a murky and unsatisfactory doctrine. Any definition of “costs” for a doctrine that bans below-cost pricing but not above-cost pricing must turn on the rationale for treating the two differently. The rationale for treating above-cost pricing as permissible depends, as the above summary makes clear, on the assumption that above-cost pricing could not deter or drive out an equally efficient entrant. Likewise the rationale for banning below-cost pricing must be that (if implemented) it could deter or drive out an equally efficient entrant. It thus makes sense that, if one is going to have a doctrine against below-cost predatory pricing, “costs” should be defined in whatever way satisfies the condition that an above-cost price could not deter or drive out an equally efficient firm. This test has important implications for which cost measure to use. In particular, it clarify several longstanding problems in defining the relevant costs for predatory pricing, including what to do when industries have zero marginal costs, when equally efficient firms that choose a capital intensive structure have lower variable costs, when all firms in declining industries have marginal costs below their variable costs, and when an alleged predator strategically times low prices after it has made capital investments (and thus has low variable costs) but the rival is deciding whether to do the same. In this way, our inquiry into why above-cost prices are not predatory will reveal something important about the nature of what is predatory. But before addressing those issues, I begin by outlining the recent legal developments and unsatisfactory state of the current debate about above-cost predatory pricing. 16 See Ordover, Predatory Pricing in 3 THE NEW PALGRAVE DICTIONARY OF LAW & ECONOMICS 77, 79 (ed. Newman 1998) (collecting sources); Easterbrook, Predatory Strategies and Counterstrategies, 48 U. Chic. L. Rev. 263, 269-304, 333-37 (1981). 17Ordover, supra, at 79-80; Brodley, Bolton & Riordan, supra note , at 2247-49, 2285-2330 (synthesizing recent literature). -5- I. THE CURRENT STATE OF LEGAL FLUX AND SCHOLARLY DEBATE Cost-based tests of predatory pricing have been changed or challenged in recent years in ways that suggest legal developments have been, and will continue to be, influenced by the underlying debate in antitrust economic scholarship on above-cost predatory pricing. After detailing these legal developments, I explain why the easy but largely conclusory answers offered by both sides in the current legal and scholarly debate cannot resolve the issue in either direction, which instead requires the more in depth analysis that the rest of this Article takes up. A. Legal Developments and Ambiguities In the early 1990s, the law on predatory pricing appeared relatively settled. The 1991 decision of the European Court of Justice in AKZO held that when a firm with dominant market power prices below average variable costs, those prices are presumed abusive, and that when it prices above average variable costs but below average total costs,18 its prices are abusive if they are intended to eliminate a competitor.19 This seemed to imply that prices above average total costs could not be abusive even if coupled with such an intent. And in 1993, the U.S. Supreme Court in Brooke decided that one necessary element of predatory pricing was proof that the defendant priced below incremental costs.20 Brooke did not resolve which measure of costs should be used,21 and the lower U.S. appellate courts were divided between those that required prices below average variable cost (or short run marginal cost) and those that would also entertain claims of prices below average total costs (or long run marginal costs).22 But while there was plenty of disagreement on such issues,23 at least there appeared to be common ground on the proposition that unilaterally set prices 18 A fixed cost is a cost that does not vary with output levels. A variable cost is a cost that varies with output levels. Total costs are the sum of fixed and variable costs. Average variable costs are the sum of variable costs divided by output. Average total costs are the sum of total costs divided by output. Average total cost thus always exceeds average variable cost since it is the sum of average fixed and variable costs. See CARLTON & PERLOFF, MODERN INDUSTRIAL ORGANIZATION 28-99 (3rd ed. 2000); AREEDA & TURNER, supra note , at 155; AREEDA & HOVENKAMP, supra note , at 320-21. 19 See AKZO Cmemie v. Commission, [1991] European Court Reports I-3359, ¶¶70-73. Just as U.S. Sherman Act §2 makes it illegal to have monopoly power and engage in unilateral exclusionary conduct, E.U. Treaty 86 makes it illegal to have a dominant position and engage in unilateral abusive conduct. Id. ¶¶34-75. But U.S. and E.U. law differ in the precise degree of market power necessary to satisfy the first element, and the type of conduct deemed to anticompetitively violate the second element. 20 In addition to requiring prices below incremental costs, Brooke required proof of two other elements whose precise definition varied with the antitrust statute in question: (1) sufficient market power to have the requisite anticompetitive effect in the market where the predatory pricing is occurring, and (2) a sufficient likelihood of recouping the investment in below-cost prices after rivals were eliminated or disciplined. Brooke Group v. Brown & Williamson, 509 U.S. 209, 222-25 (1993) (adopting somewhat higher standards of market power and recoupment likelihood under Sherman Act §2 than under the Robinson-Patman Act). 21Brooke, 509 U.S. at 222 n.11. 22 AREEDA & HOVENKAMP, supra note , at 229-230, 242-43, 349 , 368, 395. 23 These disagreements extended beyond the right cost measure. The European Court of First Instance has interpreted E.U. law to reject any requirement to prove a likelihood that the defendant could recoup predatory prices. See Tetra Pak v Commission, Case C-333/94 P, [1996] ECR I-5951. E.U. law also rejected the proposition that the dominant position and predatory pricing have to be in the same market, as long as the firm has a dominant position in some market and the leading position in the market where the predatory pricing happened. Id. However, the E.U. Advocate General had opined that E.U. law actually should properly be interpreted to require a recoupment test, see -6-