The court also found fault with the four cost-based tests offered by the government as better alternatives to route-wide AVC in this case. The court held that Tests 2 and 3 were tantamount to an average total cost standard and thus were "against prevailing law" of predatory pricing. Op. 1203-04. The court also held that Tests 1 and 4 were "short-run tests for profit-maximization" that were improper as a matter of law. Op. 1201-02. It then ruled, as a matter of fact, that they were "incurably flawed" because "the tests do not attempt the impossible task of measuring how American would have performed on the relevant routes without the capacity increases," and because they do not "identify the actual costs associated with the capacity additions." Op. 1202.

The United States showed that American, without a legitimate business justification, added money-losing capacity in order to maintain its monopoly power on DFW routes against the challenge of LCCs. Moreover, American reasonably expected to recoup the extensive losses incurred on the new capacity by maintaining its supracompetitive pricing in numerous DFW routes. Finally, American's proposed meeting competition defense is unavailable in this case as a matter of law, and untenable as a matter of fact. American violated Section 2 of the Sherman Act. The district court's contrary view and grant of summary judgment rested on serious errors of law.

1. Predatory Capacity Increases

The district court failed to understand that American's capacity increases were predatory because it failed to make the critical inquiry -- whether the increases made business sense unless they excluded rivals. The government's extensive evidence showed that they did not. The additional capacity was money-losing -- as American expected when it adopted the strategy knowing that it would be "very expensive."

The district court, however, mistakenly refused to examine American's incremental costs and revenues from adding the challenged capacity. It further erred by insisting that American's costs and revenues could be evaluated only on a route-wide basis. The court's approach could not shed light on the key issue -- whether American's predatory conduct made business sense.

By contrast, Professor Berry's incremental cost/incremental revenue studies showed that the challenged conduct was predatory because it cost American more than it brought in -- indeed, substantially more. The district court's various objections to Professor Berry's studies are either demonstrably wrong or, at most, raise disputed questions of material fact that may not be decided against the government on summary judgment.

2. Recoupment

In ruling that recoupment was unlikely, the district court made three fundamental errors. First, it incorrectly held that recoupment may be assessed only and separately in each of the four target markets, although American predated on those four routes to maintain monopoly profits in DFW generally. This ruling misreads Supreme Court and court of appeals precedent and allows multimarket monopolists to eliminate nascent competition before it can grow to serve additional markets dominated by the predator. Second, the court illogically gave greater weight to American's claim that the lack of "structural barriers" to entry at DFW made American incapable of charging supracompetitive prices, than to direct evidence that American could -- and did -- charge supracompetitive prices at DFW. Third, the court improperly failed to credit, at the summary judgment stage, Professor Berry's evidence of the likelihood of recoupment, even though that evidence is confirmed by American's own contemporaneous evaluation of the economics of its predatory strategy.

3. Meeting Competition

The district court's importing of the Robinson-Patman Act's statutory meeting competition defense into the Sherman Act as a defense to predatory conduct was error. There is no support in the text of the Sherman Act, or in Supreme Court and court of appeals decisions, for applying the defense in a Section 2 predatory pricing case. Moreover, the gravamen of the government's complaint was not American's prices. Yet, by failing to recognize the competitive significance of American's capacity additions the district court inflated its new defense to suggest that it could immunize a broad range of predatory conduct by a monopolist so long as the monopolist only matches a rival's price. Finally, the evidence shows that American as a matter of economic reality did undercut its rivals' prices by offering what the court itself described as a "superior" product at the same nominal price.

1. STANDARD OF REVIEW

This Court "review[s] the grant of summary judgment de novo, applying the same legal standard used by the district court under Fed. R. Civ. P. 56(c)." Multistate Legal Studies, Inc. v. Harcourt Brace Jovanovich Legal And Professional Publ'ns, Inc., 63 F.3d 1540, 1545 (10th Cir. 1995). The evidence of the non-moving party "is to be believed; all justifiable inferences are to be drawn in its favor; its nonconclusory version of any disputed issue of fact is assumed to be correct." Id. Thus summary judgment is appropriate only when the evidence "is so one-sided that one party must prevail as a matter of law." Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 252 (1986).

2. THE DISTRICT COURT ERRED IN FAILING TO FOCUS ON THE
   CHALLENGED CONDUCT

The offense of monopolization under Section 2 of the Sherman Act has two elements: "(1) the possession of monopoly power in the relevant market and (2) the willful acquisition or maintenance of that power as distinguished from growth or development as a consequence of a superior product, business acumen, or historic accident." Eastman Kodak Co. v. Image Technical Servs., Inc., 504 U.S. 451, 481 (1992), quoting United States v. Grinnell Corp., 384 U.S. 563, 570-71(1966); TV Communications Network, Inc. v. Turner Network Television, Inc., 964 F.2d 1022, 1025 (10th Cir. 1992). Conduct fulfilling the second requirement is "characterized as 'exclusionary' or 'anticompetitive' . . . or 'predatory.'" Aspen Skiing Co. v. Aspen Highlands Skiing Corp., 472 U.S. 585, 602 (1985). American's summary judgment motion focused only on the anticompetitive, or predatory, behavior element.⁽²⁰⁾ Op. 1193.

1. The Critical Inquiry Is Whether The Challenged Conduct Made
   Business Sense Apart From Its Exclusionary Effect

Conduct undertaken for "valid business reasons," Eastman Kodak, 504 U.S. at 483, is not predatory, even if it disadvantages competitors. But conduct by a monopolist that "threatens to defeat or forestall the corrective forces of competition and thereby sustain or extend the defendant's agglomeration of power" violates the Sherman Act. Id. at 488 (Scalia, J., dissenting).

"Anticompetitive or exclusionary conduct under section 2 is 'conduct constituting an abnormal response to market opportunities.'" Multistate, 63 F.3d at 1550, quoting Instructional Systems Dev. Corp. v. Aetna Cas. & Sur. Co., 817 F.2d 639, 649 (10th Cir. 1987). It does not make business sense unless it eliminates or reduces competition, and thus permits the costs of the conduct to be recouped through the exercise of market power. See Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 222 (1993); Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 588-89 (1986). As Judge Bork has explained:

Predation involves the deliberate seeking of monopoly power by means other than superior efficiency, by means that would not be employed in the normal course of competition. Thus, predation involves aggression against business rivals through the use of business practices that would not be considered profit maximizing except for the expectation that (1) actual rivals will be driven from the market, or the entry of potential rivals blocked or delayed, so that the predator will gain or retain a market share sufficient to command monopoly profits, or (2) rivals will be chastened sufficiently to abandon competitive behavior the predator finds threatening to its realization of monopoly profits.

Neumann v. Reinforced Earth Co., 786 F.2d 424, 427 (D.C. Cir. 1986).

In short, distinguishing legitimate competition from unlawful predation requires a common-sense business inquiry: whether the conduct would be profitable, apart from any exclusionary effects. Conduct that causes a firm "to forgo profits," but is nonetheless "rational" only because the conduct's ability to eliminate competition gives the firm "a reasonable expectation of recovering, in the form of later monopoly profits, more than the losses suffered," is predatory. Matsushita, 475 U.S. at 588-89.⁽²¹⁾

For example, in Brooke Group, the Supreme Court went to some length to set out the analytic framework for deciding whether a monopolist's low pricing would be profitable but for exclusionary effects. A plaintiff must establish "two prerequisites:" "[f]irst, . . . that the prices complained of are below an appropriate measure of its rival's costs"⁽²²⁾ (509 U.S. at 222); and "second . . . that the competitor had a reasonable prospect, or, under § 2 of the Sherman Act, a dangerous probability, of recouping its investment in below-cost prices." Id. at 224. In this case, the government showed both that (1) the capacity additions at issue were in fact money-losing endeavors but (2) American fully expected to recoup its losses by maintaining its hub dominance and monopoly profits.⁽²³⁾

2. The District Court Erred In Holding That It Was Impermissible
   To Consider Separately The Costs And Revenues Attributable To
   The Challenged Conduct In Determining Whether That Conduct
   Was Predatory

The government offered extensive evidence that American's capacity additions made no business sense except by preventing the development of an LCC hub at DFW. That evidence demonstrated that American developed a "very expensive" strategy (GA 1192) to prevent LCCs from operating profitably at DFW -- a strategy explained succinctly by CEO Crandall: "'If you're not going to get them out then [there is] no point to diminish profit.'" Op. 1153. Professor Berry's extensive quantitative analysis confirmed that the incremental costs American incurred as a result of the challenged capacity additions substantially exceeded the incremental revenues attributable to them. Thus, American's conduct made sense only as what Judge Bork called "an investment in future monopoly profits." Robert H. Bork, The Antitrust Paradox 145 (1978).

The district court held this evidence irrelevant, as a matter of law, precisely because it "address[ed] only the incremental costs and revenues and those costs and revenues only from the added capacity," i.e., because it focused specifically on the allegedly predatory conduct. Op. 1202 (emphasis original). Instead, the court considered only the aggregate costs and revenues to American of serving each entire route. Op. 1196. This approach wrongly aggregated predatory conduct with lawful, profitable conduct, and thus camouflaged the anticompetitive nature of the capacity additions.

1. The Challenged Additions To Capacity Are Properly Viewed As
   The Defining Element Of A Distinct Course Of Conduct

The district court refused to view American's additions to capacity as a distinct course of predatory conduct because, in its view, "American's fare prices and its production level -- whether characterized as 'output' or 'productive capacity' are two sides of the same coin." Op. 1194. The court reasoned that American's decision to lower its fares would stimulate increased passenger demand, and it would "place the defendant in a competitive straight jacket" if it were forbidden to increase output and forced to turn away new customers. Id. It therefore ruled that if American's pricing on the route as a whole was profitable, the capacity additions were justified.

The court's reasoning is fundamentally flawed. It is a truism that an increase in market output normally goes hand in hand with a cut in the market price. But the money-losing capacity additions the government challenged were not triggered by American's fare cuts. Indeed, they usually occurred months after the fare cuts. And the capacity American added was far in excess of what might be considered reasonably necessary to carry the local passengers that American actually did carry. Thus, the capacity additions the government challenged were the defining characteristic of a distinct, and predatory, strategy.

American's price cuts -- which the government did not challenge -- often occurred long before the challenged capacity additions. In Wichita, for example, American published fares matching Vanguard's fares 17 months before its predatory capacity additions. During those 17 months, American did not add flights or otherwise expand capacity, Vanguard continued to compete, and the government did not challenge American's conduct. It was only when Vanguard announced that it was beginning non-stop service on additional routes, including DFW to Phoenix and Cincinnati, creating four non-stop Vanguard "spokes" from DFW when there had previously been one, that American increased capacity on DFW-Wichita by 35 percent. (GA 711, 1092). It was American's conduct after this capacity increase that was challenged as predatory.

American's pricing typically was not substantially different in a period of alleged predation from that in the immediately preceding period. For example, in DFW-Wichita, in the non-predatory period after Vanguard's entry and American's price cuts, American had monthly average fares of $52 to $75. During the predatory period, after American added money-losing capacity in October 1996, American's fares stayed within that range, averaging $58 to $61 monthly. Op. 1169. Examining the price data in these instances reveals a substantial price drop when American initially responded with nonpredatory price cuts to the new LCC competition, but reveals little or no price change when American's response turned predatory with money-losing capacity additions months later. (GA 508, 514, 520). The American conduct challenged by the government as predatory was the money-losing capacity additions.

Moreover, the court could not, and did not, find that the specific capacity additions at issue served primarily to satisfy pent-up demand of local passengers wanting to travel at low fares. American added at least [Redacted] seats for every additional local passenger it carried.⁽²⁴⁾ American has not even tried to argue that it simply miscalculated demand when it undertook the challenged capacity additions, and the evidence would in any event belie such a claim. As one American document explains following Vanguard's cancellation of its non-stop Kansas City service: " [Redacted] they added schedules to drive NJ [Vanguard] from the market [Redacted] ." (GA 1041). Indeed, American's own computer models for capacity use on routes recommended against American's use of capacity on the subject routes -- but American overrode that advice.⁽²⁵⁾

In short, American added the challenged capacity not because it was a "normal business act[] undertaken in pursuit of profit,"⁽²⁶⁾ but as a predatory strategy to maintain its monopoly. (GA 615). It was error for the district court to refuse to focus its analysis on that conduct. See generally United States v. Microsoft Corp., 253 F.3d 34, 59-78 (D.C. Cir. 2001) (en banc), cert. denied, 70 U.S.L.W. 3264 (2001) (examining the business justifications for alleged predatory actions individually).

2. Brooke Group And Subsequent Cases Do Not Support The
   District Court's Refusal To Focus On The Costs And Revenues
   Attributable To The Challenged Conduct

The district court read Brooke Group and subsequent cases as mandating a comparison of revenue and cost for each route as a whole, and therefore rejected Professor Berry's Tests 1 and 4 because they "address only the incremental costs and revenues and those costs and revenues only from the added capacity." Op. 1199; 1202-03. However, both Brooke Group and the rationale behind comparing revenue and cost demonstrate that it is proper to analyze just the costs and revenues associated with the capacity additions at issue. The district court therefore erred in refusing to focus on whether the specifically challenged conduct was profitable except as a predatory strategy.

a.   The plaintiff in Brooke Group claimed that the defendant, B&W, sold generic cigarettes -- a small "segment" of the stipulated "national cigarette market" -- below cost to force plaintiff to raise its generic price to a level closer to that of branded cigarettes. 509 U.S. at 212-15, 230-31. It was disputed whether B&W sold generic cigarettes below an appropriate measure of cost but undisputed that it "made money on its overall cigarette sales." Liggett Group, Inc. v. Brown & Williamson Tobacco Corp., 748 F. Supp. 344, 362 & n.42 (M.D.N.C. 1990). After the jury returned a verdict for the plaintiff, the district court granted judgment NOV without deciding whether B&W sold generics below cost. Rather, in reasoning similar to that of the court below, it held that the "predatory pricing evidence must show that B&W lost money in the . . . market for all cigarettes in the United States. [Plaintiff] has not and cannot do this."⁽²⁷⁾ 748 F. Supp. at 362. The court reasoned that selling below cost in just generic cigarettes could not injure competition because it did not make B&W's entire cigarette business unprofitable and thus did not threaten competition in cigarettes generally. Id. at 363. The Fourth Circuit did not reach the below-cost pricing issue and affirmed on the basis of B&W's inability to recoup any losses from pricing below cost. Liggett Group, Inc. v. Brown & Williamson Tobacco Corp., 964 F.2d 335 (4th Cir. 1992).

The Supreme Court, however, went beyond the Fourth Circuit's ruling and took the "not customary" step of reviewing the sufficiency of the evidence on both below-cost pricing and recoupment. 509 U.S. at 230. That review did not follow the district court's lead in comparing revenues and costs for the whole cigarette market. Rather, the Court proceeded to the factual dispute avoided by both lower courts on the comparison of revenues and costs for the generic segment alone. The Court expressly found "sufficient evidence in the record from which a reasonable jury could conclude that . . . [B&W's] prices on its generic cigarettes were below its costs." Id. at 231 (emphasis added). It concluded that this evidence satisfied the first prerequisite of its predation test despite the fact that it and the parties recognized that generic cigarettes were part of a larger market, and despite the uncontroverted evidence that B&W did not sell below cost over the relevant market as a whole.⁽²⁸⁾ Like the Supreme Court's analysis in Brooke Group, the government's analysis in this case involved a comparison of revenues and costs for the particular conduct alleged to be predatory, rather than for the entire market/route.

Also notable is the fact that the Court had no difficulty with the plaintiff's theory of liability -- that the defendant used below-cost pricing in part of the market to reduce competition generally in the market. Id. at 230. Below-cost pricing in generic cigarettes was aimed at maintaining high prices for branded cigarettes. In this case, predation was similarly used to achieve an anticompetitive objective going beyond the scope of the predatory conduct. Predatory capacity increases on four particular routes were aimed at maintaining American's supracompetitive profits at DFW generally.

b.   Moreover, while Brooke Group does not prescribe any particular comparison of revenues and costs,⁽²⁹⁾ it nevertheless provides significant guidance as to what sorts of comparisons are appropriate in predation cases generally. It quoted a prior decision referring to the issue of "whether recovery should ever be available . . . when the pricing in question is above some measure of incremental cost." Id. at 223, quoting Cargill, Inc. v. Monfort of Colorado, Inc., 479 U.S. 104, 117-18 n.12 (1986), quoting Matsushita, 475 U.S. at 585 n.9. Significantly, all three of these decisions refer to the concept of "incremental cost." And as the district court observed, the leading antitrust treatise concludes that these decisions suggest "essentially a marginal cost test." Op. 1198, quoting A&H ¶ 723d2, at 231-32. Marginal cost is the cost of the last unit of output produced. See, e.g., Pacific Engineering & Production Co. of Nevada v. Kerr-McGee Corp., 551 F.2d 790, 796 n.7 (10th Cir. 1977).

The often-stated rationale for comparing price and marginal cost is to determine whether the defendant made sales that were unprofitable, and thus irrational, apart from any exclusionary effect. This Court has explained that when the alleged conduct is "monopolization through predatory pricing," the relationship between the marginal cost of an item and its sales price is "extremely valuable because: 'There is no reason consistent with an interest in efficiency for selling a unit at a price lower than the cost that the seller incurs by the sale.'" Pacific Engineering, 551 F.2d at 797, quoting Richard Posner, Exclusionary Practices and the Antitrust Laws, 41 U. Chi. L. Rev. 506, 519 (1974). The same is true for a comparison of price with incremental cost. As explained by now-Justice Breyer, the "intuitive idea" behind comparing price to incremental cost in a predatory pricing case is that a firm would not charge prices "that fail to cover these 'avoidable' or 'incremental' costs" "unless it later expected to raise its prices and recoup its losses." Barry Wright, 724 F.2d at 232.

Comparing price with a firm's market-wide average variable cost ("AVC") does not serve the same analytic purpose. Contrary to the district court's view, it is not true that "[a]ny sales above average variable costs help to cover a firm's fixed costs." Op. 1198. Marginal cost easily can exceed AVC, and when it does, no contributions to fixed costs are made by sales at prices above AVC but below marginal cost.⁽³⁰⁾ Moreover, if capacity costs were treated as fixed, adding capacity could enable a firm to make additional sales at prices above AVC, but those additional sales easily could fail to cover the costs of adding the capacity. In this situation, the relationship between price and AVC tells us nothing about whether adding capacity could be profitable except as a predatory strategy.

Nonetheless, courts have often used AVC as a "surrogate" for marginal cost in predatory pricing cases "because marginal cost cannot be determined from conventional accounting methods" and is, therefore, often difficult to ascertain. Northeastern Telephone Co. v. AT&T, 651 F.2d 76, 88 (2d Cir. 1981); accord Pacific Engineering, 551 F.2d at 797 (suggesting use of AVC "in the alternative" for marginal cost).⁽³¹⁾ The United States does not object to courts' use of AVC in many predatory pricing cases. Comparing price to AVC can be an appropriate test in many circumstances and may provide a useful "safe harbor" when incremental or marginal cost cannot confidently be measured. But the use of a market-wide AVC test, comparing price to average cost over a defendant's entire sales in a market, see Op. 1198-99, plainly is inappropriate in a case that involves a challenge to well-defined incremental conduct. In the government's challenge to American's capacity additions here, there is a well-defined and measurable increment of conduct the costs and benefits of which can be measured. The evidence shows that American lost money on that increment, and thus that its conduct makes sense only as an effort to harm new competition and maintain its monopoly.⁽³²⁾ Here, a market-wide price-cost comparison risks masking the reality of predation, just as it did in the Brooke Group district court's analysis. The market-wide approach focuses on the profitability of the defendant's overall operations in the market, and not on the profitability of the challenged exclusionary conduct.

Professors Areeda and Hovenkamp show this clearly:

Suppose, for example, that a firm refined 1000 gallons per day of its own gasoline, which has variable costs of $1 and which normally sells at a price of $1.10. Because this volume exhausts the firm's refining capacity, it then supports predation by purchasing additional gasoline at $1.06, and selling all its gasoline at a price of $1.03.

A&H ¶740 at 378-79. It is irrelevant, they explain, that the original 1,000 self-produced gallons are each sold at a profit, or that the sale of all the gasoline is profitable (i.e., that the gasoline is priced above AVC in the aggregate); "what matters is that . . . incremental costs are greater than incremental revenue." Id. at 379 & nn.5 & 6 (emphasis added). Under the district court's approach, however, the refiner selling at a price of $1.03 could have sold up to 1,000 gallons a day of gasoline purchased at a cost of $1.06 without having engaged in predation actionable under the antitrust laws.

As this hypothetical illustrates, even in fairly conventional predatory pricing cases, a market-wide price-cost comparison is inappropriate when, as in this case, the challenged conduct is a well-defined increment to a clearly established, and non-predatory, course of conduct, and the costs and benefits of that "incremental" conduct can be confidently quantified. It is even more clearly true in this case, where the challenged conduct is defined by an identifiable set of capacity additions. In such circumstances, reliance on a market-wide AVC test is likely to disguise the unprofitability of challenged conduct, while an incremental cost test is likely to reveal it. (GA 651-52). As in the refinery hypothetical and in Brooke Group, it was both feasible and appropriate in this case to examine the costs and benefits attributable to American's incremental conduct -- additions of capacity. It was legal error for the district court to ignore that analysis in favor of its route-wide analysis. And it led to the wrong result.⁽³³⁾

The district court did not recognize the force of this argument because it erroneously believed that an airline's cost of serving additional passengers is always close to its route-wide AVC. It relied on Continental Airlines, Inc. v. American Airlines, Inc., 824 F. Supp. 689 (S.D. Tex. 1993), for the proposition that "in the airline industry, 'variable costs change very little, AVC is close to MC (marginal cost) at all levels of output.'" Op. 1200 n.15. But Continental's authority for this proposition is Morales v. TWA, Inc., 504 U.S. 374 (1992), in which the Supreme Court noted that once an individual airline flight has been scheduled, costs "increase very little with each additional passenger."⁽³⁴⁾ 504 U.S. at 389. This is true but, as Professor Berry observed, it is also irrelevant when, as here, a carrier is not attempting to increase output from existing capacity -- filling empty seats on scheduled flights -- but instead adds capacity (new seats) through additional flights or larger aircraft. (GA 339). The district court's reasoning would lead to the absurd conclusion that American's cost of adding 10 passengers on an existing flight is the same as its cost of adding a new flight to carry those same 10 passengers. This error in the district court's reasoning was fundamental to the incorrect result it reached.

c.  Finally, there is no support in precedent for the district court's conclusion that, "[g]iven the government's relevant market definition, its claim requires proof that American underpriced its services on the entire market/route, not just one particular fraction of those services." Id. at 1202-03. Critically, Brooke Group rejected this logic. As explained above, although the alleged relevant market was all cigarettes, the Court compared revenues and costs in just the generic cigarette segment of the market. See supra pp. 37-39.

Nor is the district court's position supported by the only case it cited on this point, Bushnell Corp. v. ITT Corp., 175 F.R.D. 584 (D. Kan. 1997). There, the court opined that "the Tenth Circuit would . . . focus on the effect on competition within the entire market and conclude that a certain pricing practice with respect to a single product does not violate the antitrust laws," and rejected a predatory pricing claim because the plaintiff did not "allege facts to support a finding of a threat to competition generally within the . . . market." 175 F.R.D. at 588. Bushnell, in turn, quotes Morgan v. Ponder, 892 F.2d 1355, 1362 (8th Cir. 1989), stating that the decided cases on the issue "all focus on the basic question of whether the alleged predatory act poses a genuine threat to the overall competition." These cases merely articulate the principle stated in Brooke Group that "below-cost pricing must be capable, as a threshold matter, of producing the intended effects on the firm's rivals, whether driving them from the market, or . . . causing them to raise their prices to supracompetitive levels within a disciplined oligopoly." 509 U.S. at 225. A handful of sales below any measure of cost could not normally threaten to exclude competition, and hence could not violate Section 2. But this case is not about a handful of sales, and the government's evidence demonstrates that the challenged conduct did have an exclusionary effect. The evidence is clear that American's money-losing capacity additions not only were capable of excluding competition, they in fact did so, and the district court did not find to the contrary.⁽³⁵⁾

3. Focusing On The Incremental Costs And Revenues Attributable
   To The Challenged Conduct Does Not Amount To A "Profit
   Maximization" Test

The district court believed that focusing on the profitability of the challenged capacity increases was the equivalent of a profit maximization inquiry that "would effectively condemn as illegal any pricing or capacity decision which would reduce the 'predator's' profits." Op. 1202. The court was incorrect.

Professor Berry's Tests 1 and 4 compare just two alternatives, both firmly grounded in reality: 1) adding all the capacity that American actually added -- with price matching; and 2) maintaining its prior level of capacity -- with price-matching. It is a "before-and-after" comparison. In making this comparison, there is no search for the hypothetical profit-maximizing strategy. If adding all the capacity were profitable, American would pass Professor Berry's tests, even if some other course of action would be yet more profitable.⁽³⁶⁾

Nor would the tests the government advocated condemn a monopolist simply because its profits fell due to increased competition from rivals. We do not contend that American acted illegally when entry by LCCs reduced its profits and it responded by cutting fares. Nor do we contend that American could not legally add any capacity, if such additions constituted a rational response to market conditions. American crossed the line into illegality, however, when it knowingly lost money implementing a plan to drive out LCCs while expecting to recoup its losses.

The court based its rejection of Tests 1 and 4 as profit-maximizing on what it claimed was the concession of government economic expert Professor Stiglitz, that Tests 1 and 4 were "'profit maximizing'" tests. Op. 1202. However, the court took a phrase from Professor Stiglitz' testimony out of context. It is clear that he was articulating a property common to all price-cost-type tests for predation, rather than one unique to Professor Berry's Tests 1 and 4. Thus, Professor Stiglitz correctly explained in the same part of his testimony that "any price cost test," including the "standard Areeda-Turner [AVC] framework" endorsed by the court, "focuses on the narrow question . . . was there a sacrifice, did the firm experience a loss relative to what it could have otherwise done." (GA 1349-50) (emphasis added). If Professor Stiglitz' testimony is understood to be a sufficient basis for rejecting Tests 1 and 4 as profit-maximization tests, it must also be a sufficient basis for rejecting all price-cost tests, even though they are mandated by Brooke Group and insisted upon by the district court.

Thus the issue of "profit-maximization" is a distraction. As defendant's expert Professor Baumol puts it, one must ask whether American's capacity additions "promise[d] to yield a net addition to the firm's profits [except by driving out competition] over the long run." Baumol, supra, at 55. Professor Berry's Tests 1 and 4 do exactly that. (GA 617-18). And in this case -- as in the A&H gasoline refinery example -- an incremental cost test reveals American's predation by exposing the spike in cost per passenger caused by adding capacity and dropping load factors, while the route-wide AVC test disguises it by averaging that spike over both American's lawful and unlawful conduct. Tests 1 and 4, therefore, address the key issue while a route-wide AVC standard does not.

3. THE DISTRICT COURT ERRED BY DECIDING DISPUTED
   ISSUES OF FACT CONCERNING THE PROFITABILITY OF THE
   CHALLENGED CAPACITY ADDITIONS

Although the district court held that it was improper to focus on the profitability of the challenged capacity additions, rather than the profitability of the route as a whole, it also held that the government's proof was insufficient to establish that the capacity additions were unprofitable. In so doing, it exceeded its proper role on summary judgment by inappropriately resolving disputed issues of material fact.

To apply his incremental cost Tests 1 and 4 to American's capacity additions at issue, Professor Berry used the same data that American's managers use in making their everyday business decisions.⁽³⁷⁾ Both tests demonstrated that American's incremental revenues were less than its incremental costs. The court's rejection of these tests as factually flawed was erroneous.

The court was incorrect to suggest that Tests 1 and 4 are too "static" and therefore fail to account for any "increase in fuel or labor costs," Op. 1202, i.e., that the cost increases Professor Berry identified might have been in part the product of system-wide fuel or labor cost increases that fortuitously coincided with American's challenged capacity additions. Professor Berry explained that he performed regression analyses that account for confounding factors [Redacted] (GA 363 n.126), and thus confirm that his cost tests had provided meaningful results. (GA 352, 362-63, 449-53).

Moreover, the court wrongly insisted that the government perform "the impossible task" of determining precisely how American would have performed if it had not added the capacity. Op. 1202. The law does not require the impossible. It is never possible to know with certainty what would have happened if actual events had not transpired. But in this case, Berry's "before-and-after" analysis provided an adequate basis for a finder of fact to determine whether the costs of American's challenged conduct exceeded its benefits. In fact, most of the predatory episodes were preceded by a period when American had already cut prices but had not yet added the money-losing capacity. These periods provide especially probative evidence of how American would have performed without the challenged capacity additions. See supra pp. 21-22.

The court also rejected Tests 1 and 4 for failing "to identify the actual costs associated with the capacity additions." Op. 1202 (emphasis added). While accounting costs never precisely indicate true economic costs, the law plainly does not require more than reliable approximations. More important, Professor Berry used the same accounting data and cost allocations that American's decision makers use in the regular course of business to assess route and flight profitability. And he found that American's cost accounts were reliable for his purposes [Redacted] (GA 1256-58). Indeed, the court and American's economic expert relied on the same accounting costs when computing their proposed route-wide cost measures. Op. 1199-1200. To reject Professor Berry's tests on this record requires a resolution of disputed issues of material fact not permitted on summary judgment. Instructional Systems, supra, 817 F.2d at 648 n.8.

4. THE GOVERNMENT SHOWED THAT AMERICAN WOULD
   LIKELY RECOUP ITS PREDATION LOSSES

1. The District Court Applied An Erroneous Legal Standard In
   Ruling That Recoupment May Be Considered Only In The Four
   Target Markets Rather Than At The DFW Hub

The government's theory of recoupment followed its theory of the case: American predated on four routes to discourage formation of an LCC hub at DFW and so preserve its monopoly power on dozens of DFW routes. As Professor Berry explained, [Redacted] (GA 423).

The court, however, held as a matter of law that Brooke Group limits consideration of recoupment to "the relevant market," and that because each of the four city-pairs is "a separate relevant market," recoupment may be assessed only in each city-pair market standing alone. Op. 1214. The court was wrong as a matter of law and fact.

In Brooke Group, the Supreme Court said that assessment of recoupment "requires an estimate of the cost of the alleged predation and a close analysis of both the scheme alleged by the plaintiff and the structure and conditions of the relevant market." 509 U.S. at 226 (emphasis added). This certainly does not say, as the district court thought, that recoupment can occur only in a single market where the predatory conduct occurred, even if the plaintiff's theory of predation and evidence of recoupment go beyond that market. If anything, the import of this statement is that the court must analyze the entire scheme alleged by the plaintiff. Here, the scheme includes recoupment beyond the four routes where predation occurred. In Brooke Group, the Supreme Court held that the scheme alleged there, in which defendant "sought to preserve supracompetitive profits on branded cigarettes by pressuring Liggett to raise its generic cigarette prices" (id. at 227) (emphasis added) was "within the reach of the statute." Id. at 230. While both generic and branded cigarettes were in the stipulated relevant market, it is noteworthy that the predatory scheme encompassed predatory conduct in one product segment (generic cigarettes) aimed at "preserv[ing] [defendant's] supracompetitive profits" in another product segment (branded cigarettes). Id. at 217. See supra pp. 38-39.

Moreover, in Multistate (decided after Brooke Group), this Court specifically recognized that a firm might engage in predation in one market to prevent the target of the predation -- "the most likely challenger to the Defendants' monopoly in [a second, related] market" -- from expanding to compete in the monopolized related market. 63 F.3d at 1549 n.6. Similarly, in Advo, Inc. v. Philadelphia Newspapers, Inc., 51 F.3d 1191 (3rd Cir. 1995), the court explained that predation makes sense when a monopolist operates in several related markets because "the predator needs to make a relatively small investment (below-cost prices in only a few markets) in order to reap a large reward (supracompetitive prices in many markets)." 51 F.3d at 1196 n.4; accord A&H ¶727g at 289 (a firm that operates in numerous markets may predate in only one to acquire or maintain "higher prices in the others as well"). See also Patrick Bolton, Joseph F. Brodley & Michael H. Riordan, Predatory Pricing: Strategic Theory and Legal Policy, 88 Geo. L.J. 2239, 2267-68 (August 2000) ("Brodley") (recoupment "may occur in either the predatory market or in a strategically related market where the effects of the predation are felt"); GA 579. The government's theory -- that American predated on four routes to prevent the target LCCs from forming a hub that would affect American's numerous other DFW monopoly routes and discourage other LCCs from making a similar attempt -- parallels the strategic theories endorsed by Multistate and Advo. Indeed, at DFW theory became fact: American began its predatory capacity additions in DFW-Wichita immediately after Vanguard announced expansion of its DFW service to include Phoenix and Cincinnati (GA 711, 1092), and in DFW-Long Beach after Sun Jet announced expansion of its DFW service to include Oakland. Op. 1165. And American was so preoccupied with [Redacted] (GA 891), that on its seasonal Colorado Springs route it was adding capacity in the off season and removing it in the peak season. Op. 1162-63. That American's predatory capacity addition decisions were hub-prevention oriented is further demonstrated by the LCC Strategy's conclusion not to take that [Redacted] approach in DFW-Long Beach "unless [SunJet] increases frequency or adds spokes from DFW." (GA 766) (emphasis added).

The court also was wrong that "[t]he government's approach . . . relies on the subjective impressions of American by a number of its competitors." Op. 1214. Rather, the government's approach relies on American documents showing the airline's concern over massive losses that would result from an LCC hub at DFW, and its deliberate actions to prevent such a hub. See supra pp. 7-10. Even though American knew preventing such a hub "would definitely be very expensive in terms of AA's short term profit[s]" (GA 1192), American also knew that diminishing its short-term profits to get LCCs out of DFW would likely be profitable in the longer run because it would prevent American from suffering Delta's fate at Atlanta where ValuJet successfully established an LCC hub.⁽³⁸⁾ Op. 1150. In addition to removing Vanguard, Western Pacific and Sun-Jet, American's DFW strategy would also " [Redacted] ." (GA 877). The strategy worked: American did get the target LCCs out, and no LCC has developed a hub at DFW or seriously threatened American's monopoly profits there.

Professor Berry's analysis confirmed what American itself had calculated: that adopting a strategy to prevent formation of a hubbing LCC at DFW "would definitely be very expensive" in terms of short-term losses (GA 1192), but was nonetheless rational because American's "predatory losses [would be] easily 'recouped' by fairly modest changes in the probability of the creation of an LCC hub."⁽³⁹⁾ (GA 419). The court's criticism of Berry's analysis as "unsupported by the facts" because it contains only "'examples' and 'exercises,'" Op. 1215 & n.24, is unpersuasive. American's internal analysis was also built on "examples and exercises" (financial projections). Moreover, "'recoupment' refers to an ex ante judgment" of whether "a reasonable firm would have anticipated profitable returns to predation," and "[i]n all cases, reasonable estimates are all that can be expected." A&H ¶727c at 278-79; accord id. ¶726d.4 at 274 ("detailed accounting for an inherently uncertain future is . . . impossible in antitrust litigation.") Berry's study merely confirmed American's own internal evaluation of the impact of an LCC DFW hub. American fully expected to recoup its predation investment -- Crandall told the company as much when the LCC Strategy was endorsed -- and Berry's analysis certainly was a "reasonable estimate" given that American's " [Redacted] ."⁽⁴⁰⁾ (GA 423).

But the district court, relying heavily on Advo, rejected the government's strategic entry deterrence argument as lacking a "limiting principle." Op. 1214, quoting 51 F.3d at 1202. This holding badly misconstrues both Advo and the facts of the present case, including the economic logic of American's conduct.

Advo explained that predation "makes economic sense" when the predator operates in multiple related markets because "a relatively small investment (below-cost prices in a few markets)" may allow the predator to "reap a large reward (supracompetitive prices in many markets)." 51 F.3d at 1196 n.4. And it pointedly noted that this scenario in which predation makes sense was not presented in that case. Id. Thus it is the multiple market scenario in which predation makes economic sense that provides the limiting principle the Advo court demanded but the plaintiff there failed to supply. Id. at 1202. And that very scenario well describes both the airline industry, see supra pp. 3-5, and the government's theory of this case.

Moreover, conduct like American's can effectively serve the predator "as an example for competitors it faces in other markets" (Advo, 51 F.3d at 1196 n.4) -- an example that hit home to other LCCs who "observe[d] the exit" of SunJet, Vanguard and Western Pacific from the targeted DFW routes. Brodley, supra, at 2304.⁽⁴¹⁾ In this respect, this case is analogous to Reazin v. Blue Cross and Blue Shield of Kansas, 899 F.2d 951 (10th Cir. 1990), where the Court laid out the "straightforward" antitrust issues (899 F.2d at 953-54) (emphasis added):

Plaintiff's theory was that Blue Cross, alarmed by a perceived competitive threat from Hospital Corporation of America ("HCA") through its acquisitions of a major Wichita hospital . . . "Wesley", . . . determined to "hurt" Wesley and thereby send a message to other hospitals not to do business with entities Blue Cross believed were competitors . . . . The threatened termination of Wesley because of its affiliation with a Blue Cross competitor made other hospitals less willing to affiliate with, or enter into relationships with, Blue Cross competitors. The result was that Kansas health care consumers were restricted in their access to and benefits from health care financing arrangements involving entities other than Blue Cross, and were deprived of the benefits of competition in that arena.

As one American vice-president explained, " [Redacted] ." (GA 1322-23). American specifically knew that its decisions on capacity levels -- on any route -- sent it competitors "a message." It said, for example, that its "ultimate strategy in JFK-SJU [New York-San Juan] particularly with regard to capacity levels, is likely to send a message to our competitors.") (GA 1166) (emphasis added). American also knew that an " [Redacted] ." (GA 713, 717). Moreover, LCCs observed American's capacity addition responses here at issue, and those responses affected their own decisions on whether or when to enter DFW,⁽⁴²⁾ and their ability to obtain financing to enter DFW.⁽⁴³⁾ As Brodley explains, little more is needed to engage in successful "reputation effect predation."⁽⁴⁴⁾ Brodley, supra at 2300-04; accord GA 571-73, 620-21, 654.

2. The Court Erred In Concluding That American Cannot Charge
   Supracompetitive Prices At DFW

An alleged predator does not have a reasonable prospect of recouping its losses if it cannot charge supracompetitive prices after the target leaves the market. Such an inability would negate any inference that aggressive conduct was predatory. Brooke Group, 509 U.S. at 226; Matsushita, 475 U.S. at 591 n.15. But if the predator actually engages in supracompetitive pricing for any period beyond the transient, that by itself shows the market power necessary to recoup earlier predatory losses. Colorado Interstate Gas Co. v. Natural Gas Pipeline Co. of America, 885 F.2d 683, 695-96 & n.1 (10th Cir. 1989). Such pricing also obviates inquiry into entry conditions since such conditions are surrogates for the existence of power over price already shown directly. Cf. FTC v. Indiana Federation of Dentists, 476 U.S. 447, 460-61 (1986); Reazin, 899 F.2d at 968 n.24.

There is abundant evidence that American has a significant and enduring power over price at DFW, and the district court distilled that evidence into findings that "American's prices in Southwest and LCC-competitive markets may be used as proxies for competitive prices that still permit American to earn a profit and maintain service on the route," and that those "competitive" fares "are significantly lower than fares and yields where American does not compete with Southwest or other LCCs." Op. 1149-50 (emphasis added). Thus, the court's own findings demonstrate that American's fares at DFW (except on routes with Southwest or LCC competition) are significantly higher than the court's competitive proxies. Professor Berry agreed. He found [Redacted] (GA 321) that allowed American to charge supracompetitive prices on more than 30 DFW city-pair routes. (GA 317-323, 484-87). Thus, there is ample proof of American's ability to price supracompetitively at DFW.

Not surprisingly, therefore, there has been no meaningful entry by other carriers into DFW. The development that American feared most -- formation of an LCC hub -- has been prevented. It is uncontroverted that Western Pacific's, Vanguard's and SunJet's attempts to expand their DFW operations were halted. And the minimal LCC entry that has occurred in the past three years proves little, contrary to the court's suggestion.⁽⁴⁵⁾ Op. 1209. The five recent LCC entrants all began service from their own, non-DFW hubs after the government's investigation of American's predation began in October, 1996, and three of the five waited until the government had filed its complaint.⁽⁴⁶⁾ As [Redacted] CEO put it: " [Redacted] ." (GA 1277). See United States v. General Dynamics Corp., 415 U.S. 486, 504-05 (1974) (it is dangerous to rely on an antitrust defendant's behavior "when . . . a suit was threatened or pending").

Finally, it is clear that American still has a very large market share at DFW, while the LCCs have tiny shares. Although the court thought that the LCCs have eroded American's market share while their own share "has increased significantly," Op. 1209, it misread the evidence. The data on which it relied showed that the total LCC passenger share in DFW was [Redacted] two percent (GA 708), yet American's share has remained relatively constant at just over 70 percent. Compare GA 1099 [Redacted] with GA 1368 [Redacted] .⁽⁴⁷⁾ Even the testimony of American's expert Professor Baumol is inconsistent with the district court's determination that it is beyond dispute that entry at DFW is so easy that recoupment cannot occur. He testified that at DFW " [Redacted] ."⁽⁴⁸⁾ (GA 1254). As this Court noted in Reazin, "the fact . . . that no other entrant remotely approached [the incumbent's] domination of the market . . . cuts against the argument that entry barriers were insubstantial." 899 F.2d at 971.

5. NO "MEETING COMPETITION DEFENSE" PROPERLY APPLIES

The district court imported into the Sherman Act the Robinson-Patman Act's statutory "meeting competition" defense to price discrimination claims. It then concluded that American cannot be liable for the offense of predation -- even if it charged prices below an appropriate measure of its costs -- so long as American never charged prices lower than those of its target LCC. Op. 1204. This ruling is incorrect.

There is nothing in text of the Sherman Act that speaks of such a defense -- even in a pure price predation context. The Supreme Court has never mentioned the possibility of such a defense under the Sherman Act. Moreover, such a defense would make Brooke Group's below-cost pricing prerequisite superfluous when it is most important: when an entrenched, high-cost monopolist faces new, more efficient competition.

Further, as the district court recognized, this Court, while mentioning the proposed defense, has never applied it. Op. 1205-06, discussing Multistate, 63 F.3d at 1550 (in declining to apply the defense, this Court cites A&H passage listing numerous obstacles to doing so). Contrary to the district court's opinion, Op. 1206, neither the Sixth nor Seventh Circuit has recognized the proposed defense. In Richter Concrete Corp. v. Hilltop Concrete Corp., 691 F.2d 818, 824 (6th Cir. 1982), the defendants' prices were stipulated to be "above its marginal or average variable cost." Applying a predation test under which plaintiff could prevail by showing pricing below average total cost plus evidence that the pricing was "'unreasonable' or intended to destroy competition," the court merely observed that it was not anticompetitive to lower prices below total cost to meet the competition. 691 F.2d at 824. In Chillicothe Sand & Gravel Co. v. Martin Marietta Corp., 615 F.2d 427, 432-33 (7th Cir. 1980), the court only addressed evidence of intent, holding that prices involving no deliberate sacrifice of present revenues for the purpose of driving rivals out of the market and then recouping through higher profits in the absence of competition are not predatory even if lowered to meet competition. Reiteration of Richter's holding, Op. 1205, adds no support to the district court's logic. Indeed, in a case in which predatory pricing was found, the Ninth Circuit, after Richter, observed that there is "no authority allowing the defense of meeting the competition when a company's prices dropped below its average variable costs." D&S Redi-Mix v. Sierra Redi-Mix and Contracting Co., 692 F.2d 1245, 1248 (9th Cir. 1982).

Even if the court's meeting competition defense could apply "in a Section 2 predatory pricing case," Op. 1205, it cannot apply in the present case where the gravamen of the complaint is not limited to American's pricing. Neither reason nor precedent allows a monopolist to escape liability for any and all forms of predatory conduct merely because its prices do not undercut those of a rival.

The court attempted to avoid the limited relevance of its meeting competition defense by refusing to focus on the significance of American's capacity additions as a course of predatory conduct. Op. 1207-08. As we have already shown, the court misconceived the government's claim and the competitive significance of those capacity additions. American's capacity additions were predatory, and a pricing defense cannot be used to immunize that conduct.

Finally, were a meeting competition defense theoretically allowable here, it would fail factually, for American effectively undercut certain LCC fares by matching LCC fares while offering a superior product. Although the court found that "American's product was superior to an LCC's product because American offered higher frequencies, and (in some instances) a frequent flier program and advance seat selection," Op. 1188, the court refused to permit that finding to defeat summary judgment. It reasoned that if consideration of the difference in product quality were allowed, finding the requisites of the defense satisfied would mean "requiring American to charge a premium for its allegedly superior quality [and] would require courts to engage in a series of subjective price comparisons based on intangible values." Id. at 1208. The district court's reasoning has no application here, however, because the court found that American's nominal prices did match the LCCs' and that American's product is superior. In this situation, no "series of subjective price comparisons based on intangible values" is necessary to a conclusion that the monopolist has gone beyond meeting competition. This is particularly true because one of the most important elements of American's superiority is its frequent flier program, the value of which is readily quantifiable. At least where frequent flier miles are involved, saying that an American flight offered at the same price "matched" the LCC fare is simply false as a matter of arithmetic. Accordingly, the record presents at least an issue of fact sufficient to defeat the defense on summary judgment.

CONCLUSION

For the foregoing reasons the order of the district court granting summary judgment should be reversed.

This government antitrust case, which generated an enormous record and an 80-page district court opinion, raises several important issues. The United States believes that oral argument will materially aid the Court in deciding the case.

Respectfully submitted.

DONNA KOOPERSTEIN RENATA B. HESSE ROBERT D. YOUNG J. RICHARD DOIDGE NINA B. HALE REBEKAH J. FRENCH KARL D. KNUTSEN JENNIFER L. CIHON MAX R. HUFFMAN MARK J. NIEFER    Attorneys    U.S Department of Justice    Antitrust Division    325 7th Street, N.W., Suite 500    Washington, D.C. 20004

R. HEWITT PATE    Deputy Assistant Attorney General* CATHERINE G. O'SULLIVAN ROBERT B. NICHOLSON JOHN P. FONTE DAVID SEIDMAN ROGER W. FONES CRAIG W. CONRATH    Attorneys    U.S. Department of Justice    Antitrust Division    601 D Street, N.W.    Washington, D.C. 20530-0001    (202) 514-2435

* Assistant Attorney General James is recused in this case.

I, Robert B. Nicholson, a member of the bar of this Court, hereby certify that this brief contains 15,695 words, and thus is within the 16,500-word limitation set by the Court's order of September 24, 2001.

______________________ ROBERT B. NICHOLSON

I, Robert B. Nicholson, a member of the bar of this Court, hereby certify that today, January 11, 2002, I caused copies of the accompanying BRIEF FOR APPELLANT UNITED STATES OF AMERICA to be served by e-mail and by Federal Express on the following persons:

______________________ ROBERT B. NICHOLSON

1. Department of Transportation, Docket OST-98-3713, Enforcement Policy Regarding Unfair Exclusionary Conduct In The Air Transportation Industry--Findings and Conclusions On The Economic, Policy and Legal Issues (January 17, 2001) ("DOT Report"), available at http://ostpxweb.ost.dot.gov/aviation/.

2. Id. at 31-35; accord GA 578-79, 609.

3. This is not the first time that the government has sued American for a violation of Section 2 relating to anticompetitive conduct at DFW. In United States v. American Airlines, Inc., 743 F.2d 1114, 1116 (5th Cir. 1984), Robert Crandall (the same top executive involved in the current case) was tape-recorded proposing a price fixing agreement to the president of a low-fare DFW rival. The rival refused and turned the phone tapes over to the Department of Justice, which sued American and Crandall for attempted monopolization at DFW. The case was ultimately settled by consent decree. United States v. American Airlines, Inc., 1985-2 Trade Cas. (CCH) ¶ 66,866 (1985).

4. The district court's opinion, United States v. AMR Corp., 140 F. Supp. 2d 1141 (D. Kan. 2001), which is reproduced in the Addendum to this brief, is cited as "Op." followed by the Federal Supplement page number.

5. DOT Report, supra note 1, at 25-26. Those competitive advantages include the availability of many spoke routes that attract connecting passengers, attractive frequent flier programs, leverage over travel agencies through commission overrides, and "greater name recognition" i.e., "an established brand." Id. at 25-26, 34.

6. [Redacted]

7. Notes taken at the meeting show that the primary objective of American's Strategy was getting LCCs out of DFW. For example, those notes include:
" [Redacted] " (GA 790); and
" [Redacted]

Western Pacific . . . get them out of DFW b/f they are encouraged to put the 2nd [frequency] back in COS-DFW." (GA 919).

8. For example, because the Task Force concluded that matching Sun Jet's fares "on a limited basis appears to be effective," it recommended "not pursuing a more aggressive approach unless [Sun Jet] increases frequency or adds spokes from DFW." (GA 766).

9. The "load factor" for a flight is the percentage of seats filled: the lower the load factor the less profitable the flight. Overall route load factor is the percentage of all the seats on the route that are filled; break-even load factor ("BELF") is the average load factor necessary to make a route profitable. The February 1996 meeting highlighted computing the LCCs' break-even load factors: " [Redacted] " (GA 753); " [Redacted] ." (GA 919).

10. [Redacted]

11. American recognized that in DFW it had [Redacted]

12. Thus the district court erred in saying that the government had alleged predation in more than thirty DFW city-pair markets. Op. 1145-46. The numerous markets the court identified are the DFW markets in which American had a monopoly at the time Professory Berry concluded his report, and in which American's monopoly power was protected by its predatory conduct in the four markets at issue. (GA 317, 484-87).

13. In Kansas City I, American added about [Redacted] seats to carry about [Redacted] local passengers; in Kansas City II about [Redacted] seats for about [Redacted] passengers; in Colorado Springs about [Redacted] seats for about [Redacted] passengers; and in Wichita about [Redacted] seats for about [Red.] passengers. (GA 506-07, 512-13, 518-19, 701-06).

14. American uses computerized scheduling models that aim to " [Redacted] ." (GA 1285-86). American [Redacted] ignored projections from its models that the company would benefit with less capacity in DFW-Kansas City in 1995 (GA 1155 ( [Redacted] ), 1303-08), and in DFW-Wichita in the fall of 1996 (GA 1126 ( [Redacted] ), 1283-95).

15. The "incremental cost" associated with any business decision (e.g., replacing a plant or increasing advertising) is the change in cost borne by the firm as a consequence of that decision. For example, when a firm increases its output, the incremental cost of the additional output is the avoidable cost of producing the extra output: "the costs that the firm would save by not producing the additional product." Barry Wright Corp. v. ITT Grinnell Corp., 724 F.2d 227, 232 (1st Cir. 1983) (Breyer, J.).

16. American's profitability measure VAUDNC shows onboard revenues for a flight minus the expenses American classifies as variable by flight over its 18-month planning horizon (e.g., fuel, food, commissions). Op. 1174. VAUDNC-AC adds to VAUDNC the aircraft ownership costs ("AC"), which Professor Berry concluded were variable on a route. Op. 1175. Professors Areeda and Hovenkamp make a similar point. 3 Phillip Areeda & Herbert Hovenkamp, Antitrust Law ¶723d at 227-28 (rev. ed 1996) ("A&H") & ¶740 at 266-67 (Supp. 2000). See infra note 37.

17. Professor Berry did not apply Test 1 to Long Beach using the VAUDNC or VAUDNC-AC measures because American did not serve that route prior to the beginning of the predation period. Rather, he applied Test 1 by examining American's route profitability measure FAUDNC. (GA 382).

18. Professor Berry performed regression analyses to examine the association between American's profitability and its capacity additions in general over a five-year period. He found that American's capacity additions were unprofitable on average only on DFW LCC routes. (GA 449-53).

19. AVC is the "sum of all variable costs of production divided by output." A&H ¶735b.3 at 320.

20. Anticompetitive conduct is also an element of the offense of attempted monopolization, as the district court noted. Op. 1193.

21. [Redacted]

22. In reaching this holding the Court expressly rejected "the notion that above-cost prices that are below general market levels or the costs of a firm's competitors inflict injury to competition cognizable under the antitrust laws." 509 U.S. at 223. However, "[b]ecause the parties . . . agree[d] that the relevant measure of cost is average variable cost, [the Court] decline[d] to resolve the conflict among the lower courts over the appropriate measure of cost." Id. at 222 n.1.

23. We address the recoupment prerequisite in Section IV, infra.

24. See supra note 13.

25. See supra note 14.

26. William Baumol, Predation and the Logic of the Average Variable Cost Test, 39 J.L. & Econ. 49, 52 n.4 (1996) ("Baumol").

27. Compare Op. 1202-03 ("Given the government's relevant market definition, its claim requires proof that American underpriced its services on the entire market/route, not just one particular fraction of those services"), and Op. 1200 ("American's revenues exceeded its average variable costs at the route level" for each route in question) (emphasis added).

28. That the Court concluded that the first prerequisite had been met is demonstrated by its language contrasting plaintiff's failure to meet the second (recoupment) prerequisite. 509 U.S. at 231 ("failed . . . however"). See supra p. 31.

29. See supra note 22.

30. The standard textbook graphical depiction of cost curves shows marginal cost below AVC for relatively low levels of output and above AVC for relatively high levels of output. See, e.g., Dennis W. Carlton & Jeffrey M. Perloff, Modern Industrial Organization 31 (3d ed. 2000); F.M. Scherer & David Ross, Industrial Market Structure and Economic Performance 20 (3d ed. 1990).

31. This Circuit's most recent pronouncement is that "neither the Supreme Court nor we have taken a position on which of various cost measures is the definitive one, although we have spoken of marginal and average variable costs as being relevant." Multistate, supra, 63 F.3d at 1549 n.5.

32. Although we did not characterize it as such below, Professor Berry's Test 4 in essence calculated AVC for the capacity additions alone [Redacted] and found that American priced below it.

33. Useful parallels can be drawn between this case and Photovest Corp. v. Fotomat Corp., 606 F.2d 704 (7th Cir. 1979). Fotomat was found to have predated in part by "Saturating the Market" with company-owned stores in an effort to acquire franchised kiosks. 606 F.2d at 717. The court did not examine the overall relationship between Fotomat's price and average variable cost or any other relationship over all of Fotomat's operations in the relevant market. Rather, it found that Fotomat was "apparently willing to lose money [on some individual kiosks] in the short run hoping to reduce the profitability of Photovest's store for future buy-back." Id. at 719. Like Fotomat, American was willing to incur losses by increasing capacity unwarranted by demand for the unlawful purpose of eliminating competition.

34. The district court also quoted a 1984 Antitrust Division business review letter that "[i]n general, airlines' average variable costs exceed their marginal costs." Op. 1200 n.15 As the letter explains: "Once a flight is scheduled to depart, the marginal cost for additional passengers may properly be viewed as quite low." (GA 710d) (emphasis added); accord A&H ¶740b.2 at 382. We do not challenge this statement, which is doubtless true. But it has no pertinence in this case, because this is a case of adding capacity, not adding passengers to an already scheduled flight.

35. The district court also erred by rejecting at the summary judgment stage the government's other evidence that even American's routewide pricing was below a surrogate for marginal cost (Berry's Tests 2 and 3) that was superior to American's proposed measures of cost.

36. For example, the court noted that after increasing capacity in DFW-Colorado Springs in the summer of 1995 [Redacted] (GA 891), American withdrew a flight in December and its "profits for the month increased." Op. 1161-62. The government never claimed that the less-profitable pre-December 1995 capacity level was predatory.

37. Test 1 used VAUDNC to demonstrate that the capacity additions -- the predatory actions -- were unprofitable absent benefits from eliminating competition. Professor Berry correctly utilized VAUDNC-AC when applying Test 4 to the capacity additions. [Redacted] (GA 765), and Professor Berry's VAUDNC-AC accounts for those costs. (GA 329, 339, 342). See supra note 16. American disagreed with Professor Berry's explanation of why aircraft costs are variable in the context of this case. The court said that it did not view aircraft costs "as truly variable" -- a strange observation since the number of aircraft on each route actually did vary -- but described its conclusion as "not essential to the court's holding." Op. 1199 n.13. Both the government's expert economists testified that American's aircraft costs are variable. (GA 339, 342, 647). And even two of American's economic experts, in their academic writings, expressed views consistent with inclusion of aircraft costs, specifically addressing the propriety of treating such capital costs as variable (or avoidable). See Janusz A. Ordover & Robert D. Willig, An Economic Definition of Predation: Pricing and Product Innovation, 91 Yale L.J. 8, 17-18 (1981) ("[T]here are circumstances in which capital costs should be included in the calculation of the cost-based price floors. If capital has alternative uses . . . . [t]hese avoidable capital costs should always be included."); Baumol, supra, at 69-71. Thus, since the aircraft did have profitable alternatives on other routes, there is a powerful case that aircraft costs are variable; at the least, the issue is one of disputed facts.

38. As Professor Berry explained: " [Redacted] ." (GA 414).

39. For example, Professor Berry showed that American's predation investment in Wichita could be repaid even if it reduced the probability of a LCC hub by less than [Redacted] percent for five years. (GA 419-20). American's total investment would be recouped [Redacted] if it delayed an LCC hub for only two years. Id. Of course, more than five years have passed without any LCC hub formation at DFW.

40. The court criticized Professor Berry's computation of American's recoupment on the four subject routes. Op. 1211-13. This criticism is nondeterminative, however, because the real issue is whether American was likely to recoup by preserving its many DFW monopoly routes. As Professor Berry made clear, "in market" recoupment was a secondary issue. As a prologue to his analysis he explained: " [Redacted] ." (GA 415). Professor Berry's ultimate assessment of American's recoupment included all " [Redacted] ." (GA 416-17).

41. Brodley sets forth (at 2302-04) a four-part test, which is met in this case: 1) a dominant multi-market firm; 2) the alleged reputation effect is that a predator is likely to repeat the conduct in the future; 3) the predator deliberately pursues a reputation effect strategy; and 4) potential entrants observe the exit or other adverse effect.

42. For example, one LCC executive explained [Redacted] (GA 1346).

43. For example, one LCC executive explained [Redacted] (GA 1315-17, 1319-20).

44. Los Angeles Land Co. v. Brunswick Corp., 6 F.3d 1422 (9th Cir. 1993), and Rockbit Industries USA, Inc. v. Baker Hughes, Inc., 802 F. Supp. 1544 (S.D. Tex. 1991), cited by the court, are inapposite. Op. 1214-15. In fact, Los Angeles Land's central point is that "a plaintiff cannot establish a monopolization offense by a firm without market power solely on the basis of undesirable or even significantly anticompetitive behavior." 6 F.3d at 1427. Here, the record is replete with evidence of American's market power. And Rockbit did not even involve a reputation claim. Rather, Rockbit claimed that the defendant was using supracompetitive profits in the premium bit market "to subsidize losses in the secondary bit market." 802 F. Supp. at 1552.

45. In the past 10 years there has been only half as much entry -- both overall and by LCCs -- at the DFW hub as at other hub airports. (GA 530).

46. Two LCCs, Vanguard and ValuJet/AirTran, were already in DFW when American's predation began.

47. The district court did not even suggest that another major airline (as opposed to an LCC) could easily enter DFW. Indeed, "no airline with a similar cost structure can compete effectively at another airline's hub." Department of Transportation, Dominated Hub Fares 1 (Jan. 2001), available at http:// ostpxweb.ost.dot.gov/aviation. Accord GA 273-74, 281-87, 529; cf. DOT Report, supra note 1, at 26, 38.

48. Government expert Professor Stiglitz concluded that at DFW [Redacted] (GA 649).