This case involves the nature of permissible competitive practices in the airline industry under the antitrust laws of this country, centered around the hub-and-spoke system of American Airlines. The United States brought this suit against AMR Corporation, American Airlines, Inc., and American Eagle Holding Corporation ("American"), alleging monopolization and attempted monopolization through predatory pricing in violation of § 2 of the Sherman Act, 15 U.S.C. § 2. In essence, the government alleges that American engaged in multiple episodes of price predation in four city-pair airline markets, all connected to American's hub at Dallas/Fort Worth International Airport ("DFW"), with the ultimate purpose of using the reputation for predatory pricing it earned in those four markets to defend a monopoly at its DFW hub.1 At its root, the government's complaint alleges that American: (1) priced its product on the routes in question below cost; and (2) intended to recoup these losses by charging supracompetitive prices either on the four core routes themselves, or on those routes where it stands to exclude competition by means of its "reputation for predation." Finding that the government failed to demonstrate the existence of a genuine issue of material fact as to either of these allegations, the district court granted summary judgment in favor of American, from which the government now appeals. Because we agree that the record is void of evidence that rises to the level of a material conflict, we exercise jurisdiction pursuant to 15 U.S.C. § 29(a) and 28 U.S.C. § 1291, and affirm.

* Airlines are predominantly organized in a hub-and-spoke system, with traffic routed such that passengers leave their origin city for an intermediate hub airport. Passengers traveling to a concentrated hub tend to pay higher average fares than those traveling on comparable routes that do not include a concentrated hub as an endpoint. This is known as the "hub premium" and a major airline's hub is often an important profit center. Entry of low cost carriers ("LCCs") into a hub market tends to drive down the fares charged by major carriers. Consequently, major carriers generally enjoy higher margins on routes where they do not face LCC competition.

Both American and Delta Airlines ("Delta") maintain hubs at DFW, though Delta's presence is considerably smaller than American's. As of May 2000, American's share of passengers boarded at DFW was 70.2%, Delta's share was roughly 18%, and LCC share was 2.4%. As of mid-2000, there were seven low-cost airlines serving DFW. In the period between 1997 to 2000, five new low-cost airlines entered DFW: American Trans Air, Frontier, National, Sun Country, and Ozark. DFW has more low-fare airlines than any other hub airport and the number of passengers carried by low-fare airlines increased by over 30% from May 1999 to May 2000. Nevertheless, LCCs have a significantly higher market share in some other major U.S. hubs.2

LCCs generally enjoy the advantage of having lower costs than major carriers, allowing them to offer lower fares than their major-airline competitors.3 During the period between 1995 and 1997, a number of LCCs, including Vanguard, Western Pacific, and Sunjet, began to take advantage of these lower costs by entering certain city-pair routes serving DFW and charging lower fares than American. The instant case primarily involves DFW-Kansas City, DFW-Wichita, DFW-Colorado Springs, and DFW-Long Beach.

American responded to lower LCC fares on these routes with changes in: (1) pricing (matching LCC prices); (2) capacity (adding flights or switching to larger planes); and (3) yield management (making more seats available at the new, lower prices). By increasing capacity, American overrode its own internal capacity-planning models for each route, which had previously indicated that such increases would be unprofitable. In each instance, American's response produced the same result: the competing LCC failed to establish a presence, moved its operations, or ceased its separate existence entirely. Once the LCC ceased or moved its operations, American generally resumed its prior marketing strategy, reducing flights and raising prices to levels roughly comparable to those prior to the period of low-fare competition. Capacity was reduced after LCC exit, but usually remained higher than prior to the alleged episode of predatory activity.4

The government filed suit on May 13, 1999, alleging that American participated in a scheme of predatory pricing in violation of § 2 of the Sherman Act. In the government's view, American's combined response of lowering prices, increasing capacity, and altering yield management in response to LCC competition constituted an unlawful, anticompetitive response. After reviewing a voluminous record and receiving extensive briefs the district court granted American's motion for summary judgment on all antitrust claims, concluding that the government failed to demonstrate the existence of a genuine issue of material fact as to (1) whether American had priced below cost and (2) whether American had a dangerous probability of recouping its alleged investment in below-cost prices.

We review the grant of summary judgment de novo, applying the same legal standard used by the district court under Fed.R.Civ.P. 56(c), viewing the evidence in the light most favorable to the nonmoving party. Applied Genetics Int'l, Inc. v. First Affiliated Secs., Inc., 912 F.2d 1238, 1241 (10th Cir.1990). Summary judgment is appropriate only if American can show the absence of genuine issues of material fact and its entitlement to judgment as a matter of law. Celotex Corp. v. Catrett, 477 U.S. 317, 323, 106 S. Ct. 2548, 91 L. Ed. 2d 265 (1986). Although no special burden is imposed on a plaintiff opposing summary judgment in an antitrust case, see Eastman Kodak Co. v. Image Technical Servs., Inc., 504 U.S. 451, 468, 112 S. Ct. 2072, 119 L. Ed. 2d 265 (1992), in order to establish a "genuine issue" that entitles it to reach trial on its attempted monopolization claim premised on predatory pricing, the government "must present more than a scintilla of evidence that the alleged predatory conduct makes economic sense," Advo, Inc. v. Phila. Newspapers, Inc., 51 F.3d 1191, 1197 (3d Cir.1995). American need not disprove the government's claim; it need only establish that the proffered facts have no legal significance. Dayton Hudson Corp. v. Macerich Real Estate Co., 812 F.2d 1319, 1323 (10th Cir.1987).

Monopolization claims under § 2 of the Sherman Act require proof: (1) that a firm has monopoly power in a properly defined relevant market; and (2) that it willfully acquired or maintained this power by means of anticompetitive conduct. TV Communications Network, Inc. v. Turner Network Television, Inc., 964 F.2d 1022, 1025 (10th Cir.1992). This is to be distinguished from a business that acquired monopoly power by greater skill, efficiency, or by "building a better mousetrap." Claims of attempted monopolization under § 2 of the Sherman Act require four elements of proof: (1) a relevant geographic and product market; (2) specific intent to monopolize the market; (3) anticompetitive conduct in furtherance of the attempt; and (4) a dangerous probability that the firm will succeed in the attempt. Multistate Legal Studies, Inc. v. Harcourt Brace Jovanovich Legal and Prof'l Publ'ns, Inc., 63 F.3d 1540, 1550 (10th Cir.1995).

In the instant case, the anticompetitive conduct at issue is predatory pricing. The crux of the government's argument is that the "incremental" revenues and costs specifically associated with American's capacity additions show a loss. Because American spent more to add capacity than the revenues generated by the capacity additions, such capacity additions made no economic sense unless American intended to drive LCCs out of the market. Under the government's theory, American attempted to monopolize the four city-pair routes in question in order to develop a reputation as an exceedingly aggressive competitor and set an example to all potential competitors. Fearing American's predatory response, the theory goes, future potential competitors will decline to enter other DFW market routes and compete. If American succeeds in preventing or at least forestalling the formation of an LCC hub at DFW, it will then be able to charge higher prices on other DFW routes and thereby recoup the losses it incurred from its "capacity dumping" on the four core routes.

Scholars from the Chicago School of economic thought have long labeled predatory pricing as implausible and irrational. Frank Easterbrook, a leader of the Chicago School, once concluded that "there is no sufficient reason for antitrust law or the courts to take predation seriously." Frank H. Easterbrook, Predatory Strategies & Counterstrategies, 48 U. Chi. L. Rev. 263, 264 (1981). Chicago scholars argued that lowering prices could only be pro-competitive and any prohibition on such conduct could ultimately deter firms from engaging in conduct that is socially beneficial. Richard J. Pierce, Jr., Is Post-Chicago Ready for the Courtroom? A Response to Professor Brennan, 69 Geo. Wash. L. Rev. 1103, 1106 (2001). Commentators viewed below-cost pricing as irrational largely because of the uncertainty of recouping losses through later price increases. In order for a predatory pricing scheme to be successful, two future events had to take place: first, the victim of the alleged predation would have to exit and second, the predator would have to generate profits in excess of its initial losses. Jonathan B. Baker, Predatory Pricing after Brooke Group: An Economic Perspective, 62 Antitrust L.J. 585, 586 (1994).

In two seminal antitrust opinions, the Supreme Court adopted the skepticism of Chicago scholars, observing that "there is a consensus among commentators that predatory pricing schemes are rarely tried, and even more rarely successful." Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 589, 106 S. Ct. 1348, 89 L. Ed. 2d 538 (1986); Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 226, 113 S. Ct. 2578, 125 L. Ed. 2d 168 (1993). Implausibility of predatory pricing schemes was said to flow from the fact that their success is inherently uncertain. Matsushita, 475 U.S. at 598, 106 S. Ct. 1348. While "the short-run loss is definite ... the long-run gain depends on successfully neutralizing the competition." Id. Moreover, "[t]he success of any predatory scheme depends on maintaining monopoly power for long enough both to recoup the predator's losses and to harvest some additional gain." Id.

Furthermore, caution in predatory pricing cases is the watchword as "the costs of an erroneous finding are high." Brooke Group, 509 U.S. at 227, 113 S. Ct. 2578. Because "the mechanism by which a firm engages in predatory pricing-lowering prices-is the same mechanism by which a firm stimulates competition," mistaken inferences may deter the very conduct the antitrust laws were created to protect. Cargill Inc. v. Monfort of Colo., 479 U.S. 104, 122, 107 S. Ct. 484, 93 L. Ed. 2d 427 (1986) (quotation omitted).

Recent scholarship has challenged the notion that predatory pricing schemes are implausible and irrational. See, e.g., Patrick Bolton et al., Predatory Pricing: Strategic Theory and Legal Policy, 88 Geo. L.J. 2239, 2241 (2000) ("Modern economic analysis has developed coherent theories of predation that contravene earlier economic writing claiming that predatory pricing conduct is irrational."). Post-Chicago economists have theorized that price predation is not only plausible, but profitable, especially in a multi-market context where predation can occur in one market and recoupment can occur rapidly in other markets. See Baker, supra, at 590.

Although this court approaches the matter with caution, we do not do so with the incredulity that once prevailed.

The Supreme Court has formulated two prerequisites to recovery on a predatory pricing claim, conditions that "are not easy to establish." Brooke Group, 509 U.S. at 227, 113 S. Ct. 2578.5 First, the government must prove that "the prices complained of are below an appropriate measure of [American's] costs." Id. at 223, 113 S. Ct. 2578. While the first element is crucial, "[t]hat below-cost pricing may impose painful losses on its target is of no moment to the antitrust laws if competition is not injured." Id. at 225, 113 S. Ct. 2578. Thus, the second prerequisite to recovery on a predatory pricing claim, a demonstration that American had "a dangerous probability of recouping its investment in below-cost prices," must also be met. Id. at 224, 113 S. Ct. 2578. Without a dangerous probability of recoupment, competition remains unharmed even if individual competitors suffer. As frequently noted, "the antitrust laws were passed for the protection of competition, not competitors." Id. (citing Brown Shoe Co. v. United States, 370 U.S. 294, 320, 82 S. Ct. 1502, 8 L. Ed. 2d 510 (1962)).

Speaking to the first prerequisite to recovery, the Supreme Court stated that "[p]redatory pricing means pricing below some appropriate measure of cost." Matsushita, 475 U.S. at 584 n. 8, 106 S. Ct. 1348 (quotation omitted).6 Despite a great deal of debate on the subject, no consensus has emerged as to what the most "appropriate" measure of cost is in predatory pricing cases. Costs can generally be divided into those that are "fixed" and do not vary with the level of output (management expenses, interest on bonded debt, property taxes, depreciation, and other irreducible overhead) and those that are "variable" and do vary with the level of output (materials, fuel, labor used to produce the product). Marginal cost, the cost that results from producing an additional increment of output, is primarily a function of variable cost because fixed costs, as the name would imply, are largely unaffected by changes in output. See Rebel Oil Co., Inc. v. Atl. Richfield Co., 146 F.3d 1088, 1092 (9th Cir.1998). For predatory pricing cases, especially those involving allegedly predatory production increases, the ideal measure of cost would be marginal cost because "[a]s long as a firm's prices exceed its marginal cost, each additional sale decreases losses or increases profits." Advo, 51 F.3d at 1198. However, marginal cost, an economic abstraction, is notoriously difficult to measure and "cannot be determined from conventional accounting methods." Northeastern Tel. Co. v. AT&T, 651 F.2d 76, 88 (2d Cir.1981); Pac. Eng'g & Prod. Co. of Nev. v. Kerr-McGee Corp., 551 F.2d 790, 797 (10th Cir.1977). Economists, therefore, must resort to proxies for marginal cost. A commonly accepted proxy for marginal cost in predatory pricing cases is Average Variable Cost ("AVC"), the average of those costs that vary with the level of output.7 See, e.g., Stearns Airport Equip. Co. v. FMC Corp., 170 F.3d 518, 532 (5th Cir.1999); Advo, 51 F.3d at 1198; Arthur S. Langenderfer, Inc. v. S.E. Johnson Co., 729 F.2d 1050, 1056 (6th Cir.1984); Northeastern Tel., 651 F.2d at 88.

The Supreme Court has declined to state which of the various cost measures is definitive. In Brooke Group, the Court accepted for the purposes of the case the parties' agreement that the appropriate measure of cost was AVC, but declined to "resolve the conflict among the lower courts over the appropriate measure of cost." 509 U.S. at 223 n. 1, 113 S. Ct. 2578. In this circuit, we have spoken of both AVC and other marginal cost measures as relevant. See, e.g., Multistate Legal Studies, 63 F.3d at 1549 n. 5 (observing that "evidence of marginal cost or average variable cost is extremely beneficial in establishing a case of monopolization through predatory pricing" (emphasis added)); Pac. Eng'g, 551 F.2d at 797. Because there may be times when courts need the flexibility to examine both AVC as well as other proxies for marginal cost in order to evaluate an alleged predatory pricing scheme, we again decline to dictate a definitive cost measure for all cases. Sole reliance on AVC as the appropriate measure of cost may obscure the nature of a particular predatory scheme and, thus, contrary to what is suggested by the district court, we do not favor AVC to the exclusion of other proxies for marginal cost. Whatever the proxy used to measure marginal cost, it must be accurate and reliable in the specific circumstances of the case at bar.

Conceding that AVC is a good proxy for marginal cost in most cases, the government nevertheless argues that there may be times when looking only to a market-wide AVC test will disguise the nature of the predatory conduct at issue. Where there is a challenge to well-defined incremental conduct, and where incremental costs may be directly and confidently measured utilizing alternative proxies to AVC, argues the government, the market-wide AVC test is inappropriate.

Considering this to be the situation in the instant case, the government proffers four tests that purport to measure reliably incremental costs — the precise costs associated with the capacity additions at issue. Rather than creating independent measures of the costs associated with American's capacity additions, the government's experts rely on cost measures used in AAIMSPAN, American's internal decisional-accounting system (accounting measures that are used for internal decision making, not financial reporting). The government notes that a range of tests are necessary to rule out false positives and assure confidence in the results. Thus, the tests operate as cross-checks to each other to avoid misleading indications of predation.8 Due to similarities among the four tests, the district court grouped them as Tests Two and Three, and Tests One and Four for purposes of analysis. We proceed to consider each test to determine whether it is valid as a matter of law.

Two of the tests grouped together by the district court, Tests Two and Three, purport to measure incremental cost by looking to whether certain of American's internal cost-accounting measures became negative following the allegedly predatory capacity additions. Both tests rely on an internal accounting measure known as FAUDNC, or "Fully Allocated earnings plus Upline/Downline contribution Net of Costs." United States v. AMR Corp., 140 F. Supp. 2d 1141, 1175 (D.Kan.2001). As the name would imply, FAUDNC is a fully allocated earnings measure, meaning that general operating expenses are arbitrarily allocated by American's decision accounting system to the flight or route level, and do not necessarily represent the exact costs associated with a particular flight or route. FAUDNC reflects 97-99% of American's total costs, which include fixed costs not affected by the capacity additions at issue. Thus, while FAUDNC includes some costs directly caused by a particular flight or operations on a particular route (such as fuel and landing fees), it also includes many costs that are not related to the operation of a particular flight or route (dispatch, city ticket offices, certain station expenses, certain maintenance expenses, American's flight academy, flight simulator maintenance, general sales and advertising). In other words, FAUDNC includes costs that are not entirely avoidable even if American were to abandon an entire route.

Because Tests Two and Three rely on fully allocated costs and include many fixed costs, the district court held that utilizing these cost measures would be the equivalent of applying an average total cost test, implicitly ruled out by Brooke Group's mention of incremental costs only.9 The district court therefore concluded that, by relying on FAUDNC, Tests Two and Three were, by definition, not measures of marginal or incremental cost. We agree with this conclusion. While we will accept alternative proxies to marginal cost beyond AVC, Tests Two and Three are simply not proxies for marginal or incremental cost. Moreover, because these tests rely on "arbitrary allocation of costs among different classes of service," they "cannot purport to identify those costs which are caused by a product or service, and this is fundamental to economic cost determination." MCI Communications Corp. v. AT&T, 708 F.2d 1081, 1116 (7th Cir.1982). Thus, given that Tests Two and Three rely on cost measures that are not, in large part, variable or avoidable with respect to capacity increases, we conclude that they are invalid as a matter of law as a measure of allegedly predatory capacity increases.10 See Stearns, 170 F.3d at 532 (noting that "judgment as a matter of law is appropriate when a plaintiff fails to adequately specify how the challenged pricing undercuts the defendant's variable costs").

As to Tests One and Four, the district court grouped them together, labeling them as short-run profit-maximization tests. Test One examines changes in profitability. It employs FAUDNC, discussed above, and an internal measure of American's variable costs know as VAUDNC,11 as well as a version of VAUDNC that has been modified by the government, VAUDNC-AC.12 If these measures declined following a capacity addition, this test allegedly demonstrates that adding capacity forced American to forgo better profit performance elsewhere. Test Four relies on VAUDNC-AC to compare the supposed revenue from incremental passengers with the average avoidable cost of adding capacity. Under Test Four, if incremental revenues are below incremental costs, this is "evidence of sacrifice." AMR Corp., 140 F.Supp.2d at 1180.

In rejecting tests One and Four, the district court concluded that they were, in essence, short-run profit-maximization tests that focus on whether a company has sacrificed some level of profit to compete more effectively. Courts and scholars have observed that such a sacrifice test would necessarily involve a great deal of speculation and often result in injury to the consumer and a chilling of competition. See 3 Phillip E. Areeda & Herbert Hovenkamp, Antitrust Law ¶ 736c2 (2d ed. 2002); Stearns, 170 F.3d at 533 n. 14 (noting that theories of predation based upon the failure to maximize profits in the short run are "no longer tenable in the wake of Brooke Group"). Upon closer examination, it is clear that rather than determining whether the added capacity itself was priced below an appropriate measure of cost, Test One effectively treats forgone or "sacrificed" profits as costs, and condemns activity that may have been profitable as predatory.13 Rather than isolating the costs actually associated with the capacity additions the government purports to measure directly, Test One simply performs a "before-and-after" comparison of the route as a whole (Appellant's Br. at 48), looking to whether profits on the route as a whole decline after capacity was added, not to whether the challenged capacity additions were done below cost. In the end, Test One indicates only that a company has failed to maximize short-run profits on the route as a whole. Such a pricing standard could lead to a strangling of competition, as it would condemn nearly all output expansions, and harm to consumers. We conclude that Test One is invalid as a matter of law. Test Four does not appear to suffer from this flaw, and we do not reject it for being a short-run profit-maximization test.

As with Test One, the district court noted that, in proffering Test Four, the government has not "identif[ied] the actual costs associated with the capacity additions." AMR Corp., 140 F.Supp.2d at 1202. We agree with this conclusion as well. Test Four attempts to reveal American's predatory conduct by measuring and comparing the incremental costs incurred by American when it added capacity to the city-pair routes in question to the incremental revenue it received from the additional capacity. The government's expert who developed Test Four, Steven Berry, characterized it as a comparison of the "average revenue from incremental passengers who traveled after the capacity addition with the average avoidable cost of the capacity addition." See also William J. Baumol, Predation and the Logic of the Average Variable Cost Test, 39 J.L. & Econ. 49, 58 (1996) (opining that average avoidable cost is the proper cost measure for predatory pricing tests). Berry further stated that, when considering an increase in capacity, an avoidable cost test compares, "the incremental revenue generated by the increment of capacity to the avoidable cost of the increment of capacity." Therefore, the only appropriate costs included in Test Four are those costs that American could have avoided by not adding the challenged capacity to the city-pair routes.

Test Four utilizes VAUDNC-AC, the cost component of which includes both aircraft ownership costs and costs characterized as variable over an eighteen-month planning period by AAIMSPAN. See AMR Corp., 140 F.Supp.2d at 1174-75. The costs included in VAUDNC-AC include variable costs American incurs with respect to all of its operations at DFW. Because some of those variable costs do not vary proportionately with the level of flight activity, they are allocated arbitrarily to a flight or route by AAIMSPAN. American identifies these variable, non-proportional common costs as: (1) airport ticket agents, (2) arrival agents, (3) ramp workers, and (4) security. Therefore, American argues that because VAUDNC-AC is an allocated variable cost measure, it cannot be used to calculate the avoidable cost of the added capacity.

The government first responds to American's criticism by arguing that cost allocation is a key component of managerial accounting and a relevant and sensible method by which to assign costs for decision-making purposes. While the government may be correct, this court is not presented with the question of whether cost allocation is a reasonable accounting method or a technique which provides businesses with reliable data to evaluate business decisions. Because the government asserts that Test Four measures average avoidable cost, this court must instead determine whether that assertion is correct. Thus, the government's first response is wholly irrelevant.

The government also alleges that there exists a genuine issue of material fact because its expert reworked Test Four so as to omit the contested costs and the results still indicated predation. The government's expert, however, states that when he reworked the numbers in response to criticism from American's expert, he eliminated the following costs from the test: (1) CTO ticketing, (2) direct reservations, (3) reservation communications, (4) cargo reservations, (5) and dispatch. Although the propriety of including these costs in Test Four was also disputed by American, they are not the costs that American disputed on the grounds that they are allocated arbitrarily to a route or flight by AAIMSPAN. Consequently, the expert's revisions to Test Four are not responsive to American's criticism and no genuine issue of material fact exists.

Because the cost component of Test Four includes arbitrarily allocated variable costs, it does not compare incremental revenue to average avoidable cost. Instead, it compares incremental revenue to a measure of both average variable cost and average avoidable cost. Therefore, Test Four does not measure only the avoidable or incremental cost of the capacity additions and cannot be used to satisfy the government's burden in this case.

We conclude that all four proxies are invalid as a matter of law, fatally flawed in their application, and fundamentally unreliable.14 Because it is uncontested that American did not price below AVC for any route as a whole, we agree with the district court's conclusion that the government has not succeeded in establishing the first element of Brooke Group, pricing below an appropriate measure of cost.15 Our conclusion that the government has not succeeded in establishing a genuine issue of material fact as to the first prong of Brooke Group, pricing below an appropriate measure of cost, renders an examination of whether the government has succeeded in creating a genuine issue of material fact as to the second prong of Brooke Group, dangerous probability of recoupment, unnecessary. Given the exceedingly thin line between vigorous price competition and predatory pricing, see Northeastern Telephone Co., 651 F.2d at 88, the balance the Supreme Court has struck in Brooke Group, and the fatally flawed nature of the alternative pricing proxies proffered by the government, we conclude that summary judgment in favor of American was appropriate.

The order of the district court granting summary judgment to American is AFFIRMED.16