This case presents three antitrust claims arising from the defendant’s conduct in the retail gasoline market in Las Vegas, Nevada. The plaintiffs contend that the defendant engaged in predatory pricing between 1985 and 1989, selling self-serve, cash-only gasoline below marginal cost. The plaintiffs claim that the alleged predatory pricing was an attempt by the defendant to monopolize the market, in violation of Sherman Act § 2. The plaintiffs also claim that the predatory pricing scheme involved a conspiracy to restrain trade, in violation of Sherman Act § 1, and primary-line price discrimination, in violation of Clayton Act § 2, as amended by the Robinson-Patman Act, 15 U.S.C. § 13 (a).

The district court granted summary judgment in favor of the defendant on all three antitrust claims, concluding that the defendant did not possess enough power in the market to allow the predatory scheme to *1430 succeed, and therefore that the plaintiffs had not suffered any injury cognizable under the antitrust laws. We have jurisdiction over the plaintiffs’ timely appeal. 28 U.S.C. § 1291 . We affirm in part and reverse and remand in part.

The evidence before the district court on summary judgment reveals the following facts: Gasoline sold in Las Vegas is first produced from crude oil in Los Angeles refineries. Wholesale marketers then pump the gasoline to Las Vegas storage terminals via a common carrier pipeline operated by the Cal-Nev Pipeline Co. (“Cal-Nev”). Ninety-five percent of Las Vegas’ gasoline travels this 250-mile route. Wholesale marketers sell the gasoline to retail marketers, who then sell the gasoline to Las Vegas motorists.

As of 1991, there were more than 275 retail gasoline stations in Las Vegas. Although the grades and types of fuel vary, retailers sell gasoline through two types of service. Some gasoline is sold only on a self-serve, cash-only basis. Motorists purchasing this product must pump their own gasoline and must pay cash. Other gasoline is sold on a full-serve basis. In full serve, a service station attendant pumps the gasoline for the consumer, checks the oil and tires, washes the windows and may perform other minor services. The motorist also has the option of paying either with cash or a credit card. The consumer pays a premium for these services, which means that the price for full-serve gasoline is generally higher than the price for self-serve gasoline. Some retail marketers sell only self-serve, cash-only gasoline; others sell both self-serve, cash-only gasoline and full-serve gasoline. No marketer sells only full-serve gasoline.

The plaintiffs, Rebel Oil Co., Inc., and Auto Flite Oil Co., Inc. (collectively “Rebel”), are retail marketers of gasoline in Las Vegas who sell only self-serve, cash-only gasoline. Rebel operates 16 retail stations under various gasoline brand names. Nine stations operate under the “Rebel” brand name, six stations- operate under the “Unocal” brand name and one operates under the “Texaco” brand name. In addition to its retail sales, Rebel is one of the several wholesale marketers who ship gasoline via the Cal-Nev pipeline and sell to retail marketers.

The defendant, Atlantic Richfield Co. (“ARCO”), is a retail and wholesale marketer of gasoline in Las Vegas, as well as a major driller and refiner of crude oil in Los Ange-les. ARCO supplies gasoline to 53 retail stations in Las Vegas bearing the “ARCO” brand name. These stations sell only self-serve, cash-only gasoline. Of those 53 stations, Prestige Stations, a subsidiary of ARCO, owns and operates 15 stations. The remaining 38 stations are owned and operated by independent dealers who purchase the gasoline wholesale from ARCO and then sell the product at retail for their own account. Thirteen of these dealer stations are leased from ARCO; the remaining 25 dealer stations are operated by “contract dealers” who either own the stations or lease from third parties. The largest “contract dealer” is Terrible Herbst, Inc. (“Terrible Herbst”), which controls 23 stations under the “ARCO” brand name.

Besides Rebel and ARCO, other major retail marketers of gasoline in Las Vegas are Southland Corp. and Texaco Inc. As of 1991, Southland owned and operated 89 “7-11” stations, and Texaco owned and operated five gasoline stations. In addition, numerous independent dealers sell under varying brand names. These dealers either own the stations or are franchised dealers. Although the parties offer conflicting numbers, it is undisputed that at least 67 independent dealers sell under the “Texaco” brand name; 16 independent dealers sell under the “Unocal” brand name; 17 independent dealers sell under the “Chevron” brand name; and 12 other independent marketers sell under various names. The number of stations operated by a marketer does not necessarily determine that marketer’s share in gasoline sales. The “7-11” stations, for example, sell far less volume in gasoline than other marketers because their sales are primarily in the grocery retail market.

The facts of this case developed against a backdrop of change in the gasoline business. *1431 In the era of high fuel consumption motor vehicles, major-brand marketers affiliated with major oil companies were dominant. They enjoyed superior locations and facilities, national advertising, and customer loyalty. Independent marketers, who purchased gasoline from major oil companies for resale under their own brand name, were minor participants — “middle-of-the-block dumps,” in ARCO’s words. Because they sold only self-serve, cash-only gasoline, independent marketers enjoyed low overhead and, hence, could charge less than stations selling equivalent quality gasoline under major brand names. During the 1970s, a growing number of cost-conscious motorists patronized these independent marketers. Independent marketers saw substantial growth in their business. Taking a cue, ARCO in 1982 adopted a nationwide strategy to compete directly with the independent discount marketers. Developed by ARCO’s Special Planning Unit (SPU), the new strategy called for the elimination of full-serve and credit-card sales. Under the SPU’s new strategy, all sales would be self-serve and cash-only. Dealers would be provided with incentives, such as volume discounts, to increase sales volume and match the prices of the discount independents. In an internal memorandum, the SPU predicted that “depending on the degree and rapidity of competitive attrition, a lasting period of quite acceptable profitability could ensue.” ARCO’s new strategy increased its sales and market share nationwide.

In January, 1990, Rebel filed this antitrust suit against ARCO, pursuant to Section 4 of the Clayton Act, which allows private parties to sue antitrust violators for damages. Rebel claims that between 1985 and 1989, ARCO executed the SPU’s new strategy in Las Vegas with “more specific vengeance,” charging predatory prices in an attempt to take away market shares from its competitors and, eventually, monopolize the gasoline market in Las Vegas. Relying on affidavits obtained from former ARCO dealers, Rebel claims that ARCO controlled not only the prices charged at the 15 stations it operated through its subsidiary, Prestige Stations, but also the prices charged at the 38 stations operated by independent dealers. Rebel also obtained affidavits from an expert who compared ARCO’s prices in Los Angeles and Las Vegas markets. ARCO supplies both markets with gasoline from the same Los Ange-les refinery. The expert concluded retail prices in Las Vegas for self-serve, cash-only gasoline, when adjusted for transportation costs, were consistently 6 to 14 cents per gallon below those charged in Los Angeles. The expert concluded that ARCO’s retail prices in Las Vegas were consistently below the wholesale prices of all other wholesale suppliers in Las Vegas, and at times were 10 cents or more per gallon below ARCO’s marginal cost. 1 Rebel contends that Terrible Herbst, ARCO’s major contract dealer, conspired with ARCO in the predatory scheme.

Rebel asserts that ARCO’s pricing scheme forced 37 competitors out of the Las Vegas gasoline market, including both independent discount marketers and major oil companies, such as Exxon, Shell, Conoco, Mobil and Philips. According to Rebel, non-ARCO stations decreased in number, from 258 to 222, during the alleged predation. Rebel contends that the marketwide attrition occurred despite the fact that Las Vegas is one of thé fastest growing retail gasoline markets in the United States. Rebel’s own share of the self-serve, cash-only gasoline market dropped from 30 percent in 1982 to less than 10 percent in 1990. Rebel claimed losses totalling $2 million. Rebel was forced to mostly withdraw from the retail market and instead concentrate on the wholesale market. Rebel claims it has stayed in business only through its non-gasoline revenues.

According to Rebel’s expert affidavits, .when the alleged predation ended in 1989 *1432 AlRCO had captured 54 percent of the market for self-serve, cash-only gasoline. The experts contend that ARCO then engaged in price gouging in order to recoup the losses that resulted from the predatory scheme. To demonstrate that recoupment occurred, Rebel’s experts compared ARCO’s prices in Las Vegas with those charged in Los Ange-les, adjusted for transportation costs. The data showed that from September 1989 through March 1990, and again from September 1990 through mid-April 1991, ARCO’s prices in Las Vegas were between 4 to 19 cents per gallon higher than its prices in Los Angeles. Rebel contends that ARCO was able to charge prices above competitive levels without losing market share, demonstrating that ARCO had the power to harm competition in the gasoline market. Rebel contends that ARCO was able to maintain prices above competitive levels during the “recoupment” phase because Las Vegas marketers had been “disciplined” by ARCO’s previous predation and refused to challenge ARCO’s su-pracompeütive prices. In essence, Rebel contends that the Las Vegas gasoline market today is a “disciplined” oligopoly in which each oligopolist shares in the supracompeti-tive profits.

In the fall of 1990, the district court limited discovery solely to the issue of whether ARCO had sufficient market power to charge prices above competitive levels. Rebel Oil Co. v. Atlantic Richfield Co., 133 F.R.D. 41, 44 (D.Nev.1990). The district court justified the limited discovery on the ground that, absent a showing of market power, Rebel could not demonstrate that it suffered “antitrust injury.” Id. There was no discovery on predatory pricing, intent and collusion.

In the fall of 1992, the parties filed cross motions for summary judgment. The district court granted summary judgment in favor of ARCO on all three antitrust claims. Rebel Oil Co. v. Atlantic Richfield Co., 808 F.Supp. 1464 (D.Nev.1992). The court concluded that ARCO’s market share in the retail gasoline market in Las Vegas was insufficient as a matter of law to establish market power. In addition, the court concluded that Rebel failed to demonstrate that barriers to entry prevented other retailers from entering the retail gasoline market. The combined lack of market share and entry barriers, the court said, indicated that ARCO lacked the power to charge prices above competitive levels as a means of recouping predatory losses. In essence, the district court held that Rebel failed to put forth sufficient evidence of market power to support a jury verdict. See Celotex Corp. v. Catrett, 477 U.S. 317, 325 , 106 S.Ct. 2548, 2553-54 , 91 L.Ed.2d 265 (1986) (summary judgment appropriate if nonmoving party fails to put forth sufficient evidence for an element essential to case).

Summary judgment is appropriate when the pleadings, affidavits and other material present no genuine issue of material fact and the moving party is entitled to judgment as a matter of law. We review a grant of summary judgment de novo and evaluate the evidence most favorably to the nonmoving party to determine whether any genuine issues of material fact remain and whether the district court correctly applied the relevant substantive law. Thurman Industries, Inc. v. Pay ’N Pak Stores, Inc., 875 F.2d 1369, 1373 (9th Cir.1989).

Rebel contends that it produced sufficient evidence in the record regarding market power to create a genuine issue of material fact for trial for each of its antitrust claims. Rebel also argues that the court erred as a matter of law by applying a “monopolization” standard rather than an “attempted monopolization” standard to the Sherman Act § 2 claim. Rebel also argues that while a showing of market power is necessary for the “attempt to monopolize” claim under Sherman Act § 2, no such showing is required for the price fixing claim under Sherman Act § 1 or the price discrimination claim under Clayton Act § 2. We analyze the issue of “market power” separately as to each of Rebel’s three antitrust claims.

Rebel’s attempted monopolization claim is based on the theory that ARCO conspired with its dealers to set predatory prices in an attempt to gain monopoly power. To establish a Sherman Act § 2 violation for *1433 attempted monopolization, 2 a private plaintiff seeking damages must demonstrate four elements: (1) specific intent to control prices or destroy competition; (2) predatory or anti-competitive conduct directed at accomplishing that purpose; (3) a dangerous probability of achieving “monopoly power”; and (4) causal antitrust injury. McGlinchy v. Shell Chem. Co., 845 F.2d 802 , 811 (9th Cir.1988).

The fourth element, causal antitrust injury, is an element of all antitrust suits brought by private parties seeking damages under Section 4 of the Clayton Act. See Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 489 , 97 S.Ct. 690, 697-98 , 50 L.Ed.2d 701 (1977). Under Section 4, private plaintiffs can be compensated only for injuries that the antitrust laws were intended to prevent. Id. To show antitrust injury, a plaintiff must prove that his loss flows from an anticompetitive aspect or effect of the defendant’s behavior, since it is inimical to the antitrust laws to award damages for losses stemming from acts that do not hurt competition. Atlantic Richfield Co. v. USA Petroleum, Inc., 495 U.S. 328, 334 , 110 S.Ct. 1884, 1889 , 109 L.Ed.2d 333 (1990). If the injury flows from aspects.,of the defendant’s conduct that are beneficial, or neutral to .competition, there is no antitrust injury, , eyatuif the defendant’s conduct is illegal .per se. See id. ' " “ '

In deciding ^whether the plaintiff was injured by an anticompetitive'aspect or effect of the defendant’s behavior, care .must be taken in defining “competition.” Competition consists of rivalry among competitors. Hasbrouck v. Texaco, Inc., 842 F.2d 1034, 1040 (9th Cir.1987), aff'd, 496 U.S. 543 , 110 S.Ct. 2535 , 110 L.Ed.2d 492 (1990). Of course, conduct that_eliminates rivals reduces competition. But reduction of competition does not invoke the Sherman Act until it harms consumer welfare. Products Liab. Ins. Agency, Inc. v. Crum & Forster Ins. Cos., 682 F.2d 660, 663 (7th cir.1982); see Reiter v. Sonotone Corp., 442 U.S. 330, 343 , 99 S.Ct. 2326, 2333 , 60 L.Ed.2d 931 (1979) (Congress designed the Sherman Act as a “consumer welfare prescription”) (quoting Robert H. Bork, The Antitrust Paradox 66 (1978)). Consumer welfare is maximized when economic resources are allocated to their best use. National Gerimedical Hosp. and Gerontology Ctr. v. Blue Cross of Kansas City, 452 U.S. 378 , 387-88 & n. 13, 101 S.Ct. 2415 , 2421 & n. 13, 69 L.Ed.2d 89 (1981), and when consumers are assured competitive price and quality. Products Liab. Ins., 682 F.2d at 663-64 . Accordingly, an act is" deemed" anticompetitive under*The Sherman Act only when ifr harms' both alloca-tive efficnericy’an3" raises ffie prí'cés 'Of "goods above' competitive ’ levels or 'diminishes "their quality. Cf. Brook Group Ltd. v. Brown & Williamson Tobacco Corp., — U.S. -, -, 113 S.Ct. 2578, 2588 , 125 L.Ed.2d 168 (1993) (below-cost pricing is not anticompeti-tive in itself because, although’ it causes allo-cative inefficiency, it brings lower aggregate prices in the market).

The — defendant’s.-.alleged_ conduct here involves predatory pricing. Predatory pricing occurs in two stages. In the first stage, or “price war” period, the defendant sets prices below its marginal cost hoping to eliminate rivals and increase its share of the market. During this phase, the predator, and any rival compelled to challenge the predatory price, will suffer losses. 3 Though rivals may suffer financial losses or be eliminated as a result of the below-cost pricing, injury to rivals at this stage of the predatory scheme is of no concern to the antitrust laws. Id. Only by adopting a long-run strategy is a predator able to injure consumer welfare. See Cargill, Inc. v. Monfort of Colorado, Inc., 479 U.S. 104, 117 , 107 S.Ct. 484, 493 , 93 L.Ed.2d 427 (1986). A long-run strategy requires the predator to drive rivals from the market, or discipline them sufficiently so that they do not act as competitors normally *1434 should. Id. If the predator reaches this long-run goal, it enters the second stage, the “recoupment” period. It then can collect the fruits of the predatory scheme by charging supraeompetitive prices — prices above competitive levels. The predator’s hope is that the excess profits will allow it to recoup the losses suffered during the price war. Brook Group, — U.S. at-, 113 S.Ct. at 2589 .

In order unilaterally to raise prices above competitive levels, the predator must obtain sufficient market power. A predator has sufficient market power when, by restricting its own output, it can restrict marketwide output and, hence, increase marketwide prices. Phillip Areeda & Donald F. Turner, Antitrust Law ¶ 501, at 322 (1978) (hereinafter Areeda & Turner). Prices increase mar-ketwide in response to the reduced output because consumers bid more in competing against one another to obtain the smaller quantity available. Ball Memorial Hosp., Inc. v. Mutual Hosp. Ins., Inc., 784 F.2d 1325 , 1335 (7th Cir.1986). Without market power to increase prices above competitive levels, and sustain them for an extended period, a predator’s actions do not threaten consumer welfare. 4

Market power may be demonstrated through either of two types of proof. One type of proof is direct evidence of the injurious exercise of market power. If the plaintiff puts forth evidence of restricted output and supraeompetitive prices, that is direct proof of the injury to competition which a competitor with market power may inflict, and thus, of the actual exercise of market power. See FTC v. Indiana Fed’n of Dentists, 476 U.S. 447, 460-61 , 106 S.Ct. 2009, 2018-19 , 90 L.Ed.2d 445 (1986). The more common type of proof is circumstantial evidence pertaining to the structure of the market. . To demonstrate market power circumstantially, a plaintiff must: (1) define the relevant market, (2) show that the defendant owns a dominant share of that market, and (3) show that there are significant barriers to entry and show that existing competitors lack the capacity to increase their output in the short run. See Ryko Mfg. Co. v. Eden Serv., 823 F.2d 1215 , 1232 (8th Cir.1987), cert. denied, 484 U.S. 1026 , 108 S.Ct. 751 , 98 L.Ed.2d 763 (1988); Ball Memorial Hosp., 784 F.2d at 1335.

In opposing ARCO’s motion for summary judgment, and in supporting its own cross motion, Rebel submitted circumstantial evidence to the district court purporting to show that ARCO possessed market power. We must determine whether this circumstantial evidence was sufficient to create a genuine triable issue with regard to market power in the Sherman Act § 2 claim.

We begin with the issue of market definition. As noted above, circumstantial evidence of market power requires that the plaintiff, at the threshold, define the relevant market. A “market” is any grouping of sales whose sellers, if unified by a monopolist or a hypothetical cartel, would have market power in dealing with any group of buyers. See Phillip Areeda & Herbert Hovenkamp, Antitrust Law, ¶ 518.1b, at 534 (Supp.1993) (hereinafter Areeda & Hovenkamp). If the sales of other producers substantially constrain the price-increasing ability of the monopolist or hypothetical cartel, these other producers must be included in the market. Stated differently, a “market” is the group of sellers or producers who have the “actual or potential ability to deprive each other of significant levels of business.” Thurman Indus., 875 F.2d at 1374 . Market definition is crucial. Without a definition of the relevant market, it is impossible to determine market share.

There are two possible definitions of the market in the present case. Rebel contends the relevant market includes all retail sales of gasoline in Las Vegas, except for sales of full-serve gasoline. This is the mar *1435 ket in which ARCO exclusively operates. ARCO disputes Rebel’s narrow market definition. ARCO contends the market is broader, consisting of all sales of retail gasoline in Las Vegas, including full-serve gasoline.

The dispute between Rebel and ARCO focused on cross-elasticity of demand: whether consumers view the products as substitutes for each other. See E. Thomas Sullivan & Jeffrey L. Harrison, Understanding Antitrust and Its Economic Implications § 6.04[2] (1988) (hereinafter Sullivan & Harrison). If consumers view the products as substitutes, the products are part of the' same market. Rebel’s expert concluded that self-serve, cash-only gasoline and full-serve gasoline are not substitutes. He stated that consumers of full-serve gasoline base their purchase strictly on the availability of services, for which they pay a premium. Likewise, self-serve, cash-only gasoline consumers do not consider full-serve gasoline as a substitute, he said, because they will always buy the lower cost gasoline, even if the premium for full-service is less than the cost of the service.

The district court accepted ARCO’s position, concluding that both self-serve, cash-only gasoline and full-serve gasoline should be included in the relevant market. ARCO introduced affidavits from an expert who said the two products were correlated in price, indicating that the products are substitutes for each other. 5 The court said ARCO’s affidavits were “more persuasive” than those submitted by Rebel. Rebel contends that the district court’s use of the word “persuasive” indicates that the court improperly weighed the evidence, a role inappropriate at summary judgment. See Lippi v. City Bank, 955 F.2d 599, 613 (9th Cir.1992).

Rebel correctly points out that the definition of the relevant market is a factual inquiry for the jury, and the court may not weigh evidence or judge witness credibility. High Technology Careers v. San Jose Mercury News, 996 F.2d 987, 990 (9th Cir.1993). However, that an issue is factual does not necessarily preclude summary judgment. If the moving party shows that there is an absence of evidence to support the plaintiffs case, the nonmoving party bears the burden of producing evidence sufficient to sustain a jury verdict on those issues for which it bears the burden at trial. Celotex, 477 U.S. at 324 , 106 S.Ct. at 2553 . If Rebel’s evidence cannot sustain a jury verdict on the issue of market definition, summary judgment is appropriate. See id. at 325 , 106 S.Ct. at 2553-54. This rule does not direct courts to resolve questions of credibility or conflicting inferences. What it requires courts to do is assess whether the jury, drawing all inferences in favor of the nonmoving party, could reasonably render a verdict in favor of the nonmoving party in light of the substantive law. Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 249-52 , 106 S.Ct. 2505, 2510-13 , 91 L.Ed.2d 202 (1986). The determination requires application of the standard that courts apply in motions for a directed verdict or a judgment notwithstanding the verdict. See id. at 251 , 106 S.Ct. at 2511-12.

As a preliminary matter, we note that expert opinion is admissible and may defeat summary judgment if it appears that the affiant is competent to give an expert opinion and that the factual basis for the opinion is stated in the affidavit, even though the underlying factual details and reasoning upon which the opinion is based are not. Bulthuis v. Rexall Corp., 789 F.2d 1315, 1317 (9th Cir.1985). While the affidavit of Rebel’s expert meets this basic standard, the inference to be drawn from expert affidavits must, as Anderson requires, be sufficient to support a favorable jury verdict. In the context of antitrust law, if there are undisputed facts about the structure of the market that render the inference economically unreasonable, the expert opinion is insufficient *1436 to support a jury verdict. Eastman Kodak Co. v. Image Technical Serv., Inc., 504 U.S. 451, 468-69 , 112 S.Ct. 2072, 2083 , 119 L.Ed.2d 265 (1992). As the Supreme Court explained

Brook Group, — U.S. at-, 113 S.Ct. at 2598 (citations omitted) (quoting Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574 , 594 n. 19, 106 S.Ct. 1348 , 1360 n. 19, 89 L.Ed.2d 538 (1986)). The inquiry is whether the inference to be drawn from the expert’s opinion is “reasonable given the substantive law which is the foundation for the claim or defense.” See Richards v. Neilsen Freight Lines, 810 F.2d 898, 902 (9th Cir.1987).

Our independent review of Rebel’s expert affidavits compels the conclusion that it would be unreasonable for a juror to infer from those affidavits that full-serve sales of gasoline should be excluded from the relevant market. Rebel’s expert relied on “demand elasticity” — that is, whether a price rise in self-serve, cash-only gasoline would cause self-serve consumers to shift their demand to full-serve gasoline. A price differential between two products may reflect a low cross-elasticity of demand, if the higher priced product offers additional service for which consumers are willing to pay a premium. Thurman Indus., 875 F.2d at 1376 (citing Photovest Corp. v. Fotomat Corp., 606 F.2d 704, 715 (7th Cir.1979) (drive-thru photo processing, for which consumers paid a premium, was relevant market apart from conventional photo processing)). But defining a market on the basis of demand considerations alone is erroneous. Virtual Maintenance, Inc. v. Prime Computer, Inc., 11 F.3d 660, 664 (6th Cir.1993) (citing Areeda & Hovenkamp, ¶ 518.1, at 543), cert. dismissed, — U.S. -, 114 S.Ct. 2700 , 129 L.Ed.2d 829 (1994). A reasonable market definition must also be based on “supply elasticity.” Id.; Twin City Sportservice, Inc. v. Charles O. Finley & Co., 512 F.2d 1264, 1274 (9th Cir.1975). Supply elasticity measures the responsiveness of producers to price increases. Sullivan & Harrison, § 6.02[3]. If producers of product X can readily shift their production facilities to produce product Y, then the sales of both should be included in the relevant market. Areeda & Turner ¶ 521a, at 354. The affidavit of Rebel’s expert fails to account for the fact that sellers of full-serve gasoline can easily convert their full-serve pumps, at virtually no cost, into self-serve, cash-only pumps, expanding output and thus constraining any attempt by ARCO to charge supracompeti-tive prices for self-serve gasoline. The ease by which marketers can convert their full-serve facilities to increase their output of self-serve gasoline requires that full-serve sales be part of the relevant market; it is immaterial that consumers do not regard the products as substitutes, that a price differential exists, or that the prices are not closely correlated. Areeda & Turner ¶ 521, at 354.

While sellers in a normally functioning market would convert their full-serve pumps to self-serve to restrain supracompetitive pricing by ARCO, Rebel’s expert contended that Las Vegas is not a normally functioning market. Rebel states: “As the market actually functions, the full-service islands simply do not represent any competition to supracompetitive pricing by ARCO.” Rebel contends that ARCO’s predatory below-cost pricing has so punished existing marketers of gasoline in Las Vegas that these marketers are now a “disciplined oligopoly” and have no incentive to convert their pumps to challenge ARCO’s pricing. This contention is irrelevant to market definition in the present context. Rebel’s theory has no bearing on the threshold question of whether those marketers are potential competitors who should be included in the relevant market. See Thurman Indus., 875 F.2d at 1374 (a relevant market includes those sellers who have the “actual or potential ability” to compete and deprive the defen *1437 dant of significant amounts of business). The fact that some marketers are oligopolists does not mean they cannot deter an attempted monopolization. Even oligopolists may compete (and the oligopoly will vanish) if one of their number strays from the pack with the intent to violate the Sherman Act. See Bork, The Antitrust Paradox 101-04 (“Nothing in the theory [of oligopoly] compels the conclusion that oligopolists do not behave as competitors.”). 6

Although Rebel failed to demonstrate its proposed definition of the relevant market, this is not fatal to Rebel’s attempt-to-monopolize claim. In the alternative, Rebel claimed that even if full-serve gasoline is included in the relevant market, ARCO would have significant market power. The district court found that the relevant product market included full-serve gasoline. We affirm that determination and are free to address the remaining issues regarding market power. See R.C. Dick Geothermal Corp. v. Thermogenics, Inc., 890 F.2d 139, 144 (9th Cir.1989) (en banc).

We next address the second factor in the market power analysis: market share. Measurement of market share is necessary to determine whether the defendant possesses sufficient leverage to influence marketwide output. With a dominant share of the market’s productive assets, a firm may have the market power to restrict marketwide output and, hence, increase prices, as its rivals may not have the capacity to increase their sales quickly to make up for the reduction by the dominant firm. Ball Memorial Hosp., 784 F.2d at 1335.

Rebel’s experts contended that ARCO possessed a 44 percent share of the retail market consisting of both self-serve, cash-only gasoline and full-serve gasoline. As a preliminary matter, ARCO disputes Rebel’s claim that it possesses 44 percent of the relevant market. The dispute essentially hinges on how to define ARCO as a “firm.” ARCO directly operates 15 “ARCO”-branded stations in Las Vegas; another 38 “ARCO”branded are operated by independent dealers who purchase the gasoline at wholesale from ARCO and sell at retail on their own account. ARCO points out that if only the 15 company-owned stations are aggregated, ARCO’s market share is only 11.5 percent. Rebel contends that the sales capacity of the 38 dealer stations must be included in ARCO’s market share because these independent dealers are not “independent” at all. Rebel claims that ARCO conspired with Terrible Herbst, the largest dealer and owner of 23 stations, in the predatory scheme. 7 As for the remaining ARCO dealers, Rebel presented affidavits from past and present dealers who stated that ARCO controls their retail prices. Rebel contends that because ARCO supplies gasoline at the wholesale level to these dealers, ARCO is in a position to influence the ultimate retail price. Rebel also cited pricing data showing that “ARCO”branded stations are tightly clustered in product price and do not appear to act as independent entities.

The aggregation of market shares of several rivals is justified if the rivals are alleged to have conspired to monopolize. 8 Cf. *1438 United States v. American Airlines, Inc., 743 F.2d 1114, 1122 (5th Cir.1984), cert. dismissed, 474 U.S. 1001 , 106 S.Ct. 420 , 88 L.Ed.2d 370 (1985) (airline’s unsuccessful attempt to fix prices with rival required aggregation of both firms’ shares in attempt-to-monopolize claim). For the purpose of the summary judgment motion, the district court properly aggregated all “ARCO”-branded stations together, accepting Rebel’s premise that ARCO’s share was 44 percent. 9

In granting summary judgment to ARCO, the district court held that Rebel’s evidence could not sustain a jury verdict in its favor, because a 44 percent market share was insufficient as a matter of law to give ARCO market power. The court, citing Twin City Sportservice, 512 F.2d at 1274 , stated that “numerous cases” hold that any market share below 50 percent is insufficient.

Rebel argues that evidence of a 44 percent market share is enough to establish a genuine issue of material fact regarding market share. It contends that the district court erred because the court relied on a “monopolization” rather than an “attempt-to-monopolize” standard. Rebel argues that when a claim involves an attempt to monopolize, the quantum of market share necessary to create a genuine issue is smaller than is required in a claim of actual monopolization.

We agree with Rebel that the minimum showing of market share required in an attempt case is a lower quantum than the minimum showing required in an actual monopolization case. It is true, as the district court stated, that numerous cases hold that a market share of less than 50 percent is presumptively insufficient to establish market power. See, e.g., Dimmitt Agri Indus., Inc. v. CPC Int'l Inc., 679 F.2d 516, 528 (5th Cir.1982), cert. denied, 460 U.S. 1082 , 103 S.Ct. 1770 , 76 L.Ed.2d 344 (1983); Twin City Sportservice, 512 F.2d at 1274 . However, these cases and others cited by the district court involve claims of actual monopolization. When the claim involves attempted monopolization, most cases hold that a market share of 30 percent is presumptively insufficient to establish the power to control price. See Jefferson Parish Hosp. Dist. No. 2 v. Hyde, 466 U.S. 2 , 26 n. 43, 104 S.Ct. 1551 , 1566 n. 43, 80 L.Ed.2d 2 (1984); A.A. Poultry Farms, Inc. v. Rose Acre Farms, Inc., 881 F.2d 1396, 1403 (7th Cir.1989), cert. denied, 494 U.S. 1019 , 110 S.Ct. 1326 , 108 L.Ed.2d 501 (1990); Nifty Foods Corp. v. Great Atl. & Pac. Tea Co., 614 F.2d 832, 841 (2d Cir.1980); M & M Medical Supplies & Serv., Inc. v. Pleasant Valley Hosp., Inc., 981 F.2d 160, 168 (4th Cir.1992), cert. denied, — U.S. -, 113 S.Ct. 2962 , 125 L.Ed.2d 662 (1993); Areeda & Turner ¶8350, at 349. ARCO’s market share of 44 percent is sufficient as a matter of law to support a finding of market power, if entry barriers are high and competitors are unable to expand their output in response to supracompetitive pricing. 10

*1439 3

We next address the final factors in market power analysis: barriers to entry and barriers to expansion. A mere showing of substantial or even dominant market share alone cannot establish market power sufficient to carry out a predatory scheme. The plaintiff must show that new rivals are barred from entering the market and show that existing competitors lack the capacity to expand their output to challenge the predator’s high price. See Ryko Mfg., 823 F.2d at 1232.

Entry barriers are “additional long-run costs that were not incurred by incumbent firms but must be incurred by new entrants,” or “factors in the market that deter entry while permitting incumbent firms to earn monopoly returns.” Los Angeles Land Co. v. Brunswick Corp., 6 F.3d 1422, 1427-28 (9th Cir.1993), cert. denied, - U.S. -, 114 S.Ct. 1307 , 127 L.Ed.2d 658 (1994) (quoting Areeda & Hovenkamp, Antitrust Law § 409 at 509-10 (Supp.1992)). The main sources of entry barriers are: (1) legal license requirements; 11 (2) control of an essential or superior resource; (3) entrenched buyer preferences for established brands; (4) capital market evaluations imposing higher capital costs on new entrants; and, in some situations, (5) economies of scale. Id. at 1428 n. 4. In evaluating entry barriers, we focus on their ability to constrain not “those already in the market, but ... those who would enter but are prevented from doing so.” United States v. Syufy Enter., 903 F.2d 659 , 672 n. 21 (9th Cir.1990).

To justify a finding that a defendant has the power to control prices, entry barriers must be significant-they must be capable of constraining the normal operation of the market to the extent that the problem is unlikely to be self-correcting. See Syufy Enter., 903 F.2d at 663 . Barriers to entry are insignificant when natural market forces will likely cure the problem. In such cases, judicial intervention into the market is unwarranted. For example, with easy entry, a predator charging supracompetitive .prices will quickly lose market share (as well as any chance of reaping monopoly profits) as new rivals enter the market and undercut its high price. See id. at 664 .

Rebel introduced affidavits from experts stating that potential new competitors in the Las Vegas market face high barriers to entry, the most significant being a legal license. Since July 1, 1987, Nevada law has barred major oil refiners from entering the market and directly -operating gasoline stations. Nevada Divorcement Law, 15 Nev. Rev.Stat. § 598.677. 12 The law contains a “grandfather” provision that permitted existing company-operated stations 'to remain. At the time, ARCO owned 15 stations in Las Vegas, far more than any other major oil company. Because major oil refiners, which includes all Los Angeles refiners, are barred from entering, Rebel contended that only independent “chain” marketers or individual entrepreneurs are candidates to enter the Las Vegas retail gasoline market. Rebel claims that neither is capable of effectively competing with ARCO.

Rebel’s experts contend that the unique nature and structure of the gasoline market in Las Vegas is a barrier to independent “chain” marketers who would seek to enter the market. Rebel argues there is no year-round independent source of wholesale gasoline in Las Vegas, demonstrated by the fact that 95 percent of the gasoline sold in Las Vegas arrives via the Cal-Nev pipeline. The Cal-Nev pipeline requires a minimum shipment of 420,000 gallons. Rebel’s experts contend that because costs and quality maintenance limit storage of any shipment to one month, a new entrant must have the capacity to sell 420,000 gallons a month, which would require the acquisition of 10 to 15 retail outlets.

*1440 Apart from the large capital investment associated with opening and operating 10 to 15 stations, Rebel contends that both independent “chain” marketers and individual entrepreneurs would have trouble ensuring a reliable wholesale supply of gasoline at a competitive price, for two reasons. First, wholesale sellers will simply note market conditions and raise their wholesale prices, seizing the excess profits that are made available if ARCO raises its prices. Rebel’s expert contends that historical data indicates that wholesale sellers have behaved in this manner. Second, since 1989, regulations in Clark County, Nevada, have required the use of oxygenated gasoline during six months each year. ARCO uses ethanol as an oxygenate, which is 3 cents per gallon cheaper than Methyl Tertiary Butyl Ether (MTBE), the oxygenate used by every other wholesale supplier. Rebel claims this is a significant cost disadvantage to new entrants. Rebel also introduced affidavits indicating that high capital costs of $7.5 to $15 million, the unavailability of loan capital, and the “chilling” effect of ARCO’s prior predatory behavior discouraged potential new competitors in the market.

Assertions in expert affidavits do not .automatically create a genuine issue of material fact. As noted above, we are obligated to look at the record to determine whether, in light of any undisputed facts, the inferences to be drawn from the expert’s affidavits are reasonable. Eastman Kodak, 504 U.S. at 468-69 , 112 S.Ct. at 2083 . “When the expert opinion is not supported by sufficient facts to validate it in the eyes of the law or when indisputable record facts contradict or otherwise render the opinion unreasonable,” summary judgment is appropriate. Brook Group, - U.S. at-, 113 S.Ct. at 2598 .

It is undisputed that seven independent marketers, operating a total of 17 stations, entered the Las Vegas market between 1983 and 1990, not counting attempted entrants who failed in their first year of operation. 13 Most of the new rivals entered before 1988. Only two rivals entered after 1988, operating one station apiece.

Undisputed evidence of entry between 1983 and 1990, the alleged predation period, might contradict an inference that ARCO engaged in predatory pricing during that period, as a firm is unlikely to enter a market when a rival is selling products below cost. But it does not necessarily contradict the threat of market power, which is the issue in’ this appeal. The crux of Rebel’s argument is that ARCO obtained or has come close to obtaining market power since the enactment of the Divorcement Law and the oxygenate regulations-in 1987 and 1988, respectively. We know of no authority that would require, as proof of market power, evidence of entry barriers throughout the period of predation. Indeed, predatory claims, including Rebel’s, suggest that market power is not gained until after years of below-cost pricing, during which competitors exit the market and the defendant’s market share increases. To determine whether Rebel’s claim of entry barriers is. reasonable, we must consider those entries occurring after 1988, when the alleged barriers were in place and when ARCO allegedly obtained the power to charge su-pracompetitive prices. See Cargill, 479 U.S. at 120 n. 15, 107 S.Ct. at 494 n. 15.

The fact that entry has occurred does not necessarily preclude the existence of “significant” entry barriers. If the output or capacity of the new entrant is insufficient to take significant business away from the predator, they are unlikely to represent a challenge to the predator’s market power. See Oahu Gas Serv., Inc. v. Pacific Resources, Inc., 838 F.2d 360, 367 (9th Cir.1988), cert. denied, 488 U.S. 870 , 109 S.Ct. 180 , 102 L.Ed.2d 149 (1988); Department of Justice Merger Guidelines ¶3.0 (1992) (entry must be “timely, likely, and sufficient in magnitude, character and scope to deter or counteract the [anticompetitive effects of concern”). Barriers may still be “significant” if the market is unable to correct itself despite the entry of small rivals. This is the position taken by Rebel’s expert. He claims the new *1441 entry after 1988 was de minimis because the new rivals were “all small independent dealers with insignificant volumes, whose operations do not contradict the existence of barriers to entry in the Las Vegas market.”

We addressed a similar question in Oahu, which involved an appeal from the denial of a motion for judgment notwithstanding the verdict. In that case, we held that the entry of two rivals did not preclude the jury’s finding that the defendant had actually monopolized the market. Oahu, 838 F.2d at 367 . We explained that one rival’s founder was a former executive for the monopolist, which supported the inference that his success in the market was not competition on the merits. As for the other rival, we noted that “evidence that the firm remained very small could reasonably preclude a decision that [the rival’s] entry reflected a breakdown of barriers to entry.” Id. at 367 .

The same reasoning applies here. The conclusion of Rebel’s expert that “significant” entry barriers were erected by the Nevada Divorcement Law and the oxygenate regulations is not contradicted or rendered unreasonable by the fact that two small rivals entered the market after the enactment of these laws. ARCO operates or supplies 53 “ARCO”-branded gas stations. A juror could reasonably conclude that two gasoline stations would have insufficient capacity to offset supracompetitive pricing by a would-be monopolist. Rebel’s evidence on entry barriers is sufficient to create a genuine issue of material fact, but this does not end our inquiry.

Market power cannot be inferred solely from the existence of entry barriers and a dominant market share. The ability to control output and prices — the essence of market power — depends largely on the ability of existing firms to quickly increase their own output in response to a contraction by the defendant. Ball Memorial Hosp., 784 F.2d at 1336. Competitors may not be able to increase output if there are barriers to expansion. One such barrier is lack of excess capacity. Excess capacity is the capacity of the rivals in a market to produce more than the market demands at a competitive price. Hovenkamp, Economics and Federal Antitrust Law § 6.10, at 182. If existing competitors are producing at full capacity, they may lack the ability to quickly expand supply and counteract a predator’s supra-competitive pricing. Id. at 1335-36. On the other hand, if rivals have idle plants and can quickly respond to any predator’s attempt to raise prices above competitive levels, the predator will suffer an immediate loss of market share to competitors. In that instance, the predator does not have market power. See Herbert Hovenkamp and Avarelle Silver-Westrick, Predatory Pricing and the Ninth Circuit, 1983 Ariz.St.L.J. 4433, 466 (1983).

Excess capacity is a technical concept that is difficult to measure without an analysis of a firm’s' costs. Instead, evidence of past output expansion may be used as a surrogate. If there is undisputed evidence indicating that competitors have expanded output in the recent past, or have the ability to expand output in the future, summary disposition may be appropriate. See International Distrib. Centers, Inc. v. Walsh Trucking Co., 812 F.2d 786 , 792 (2d Cir.), cert. denied, 482 U.S. 915 , 107 S.Ct. 3188 , 96 L.Ed.2d 676 (1987); cf. Oahu, 838 F.2d at 366 (no market power exists in “a market with low entry barriers or other evidence of a defendant’s inability to control prices or exclude competitors.”) (emphasis added). Pri- or expansion by competitors would suggest that the defendant during that expansion lacked the market power to control market-wide output in the first place. See id. at 792-93. If a firm has not obtained that power and is not reasonably close to obtaining it, “it matters little that high barriers to entry exist to help that firm maintain a monopoly power it could never achieve.” Indiana Grocery, Inc. v. Super Valu Stores, Inc., 864 F.2d 1409, 1415 (7th Cir.1989).

On this point, Rebel’s claim of attempted monopolization falters. It is undisputed that in 1988 and 1989 two firms, Texaco and Southland, expanded their operations in Las Vegas by acquiring the assets of exiting or bankrupt rivals. Texaco acquired 14 gasoline stations from the Exxon Corporation and turned the stations into dealerships. Another 21 former. “Circle K” outlets became Texaco dealerships while “Circle K” was in bank *1442 ruptcy, and another 5 “Jet” stations became Texaco dealerships. In addition, the South-land Corporation added 32 more stores to its chain of “7-11” stores by acquiring the assets of “Stop N Go” markets in Las Vegas.

Rebel argues that these acquisitions do not prove “expansion” occurred, because the acquisitions involved gasoline stations already in the market. The upshot of Rebel’s argument apparently is that because there is no increase in stations, Texaco and Southland dealers have no ability to increase market-wide output than they did before the acquisition. Rebel’s argument ignores an important fact: gasoline stations do not produce gasoline. Stations are merely'retail distribution outlets. Gasoline is produced in Los Angeles refineries, then shipped to Las Vegas via the Cal-Nev pipeline. Competitors do not have to build more gas stations to satisfy customers’ wants. They can simply purchase and transport more gasoline via the pipeline. Cf id. at 1414 (to increase output to offset, grocery stores need not build new productive assets but can deliver more groceries from suppliers). This means that, where sufficient competing retail outlets exist, a would-be monopolist must control marketwide output at the wholesale supply level in order to pose any threat of monopolizing the retail market. In essence, Rebel must show that ARCO had monopoly power, or was dangerously close to achieving it, at the wholesale supply level. See Los Angeles Land Co., 6 F.3d at 1429 n. 6 (by failing to prove that defendant controlled the supply of bowling equipment, plaintiff failed to prove that defendant had market power in the retail bowling market).

As the expansion by Texaco and Southland suggests, the wholesale gasoline supply in Las Vegas is highly elastic, which does not support an inference that ARCO controls the supply of wholesale gasoline. Already supplying 22 dealers, Texaco undertook the task of supplying 34 more dealers in a very short time (six stations were not yet up and running). Although the volume of this increased supply is not contained in the record, it undoubtedly was substantial. Texaco’s expansion occurred in 1989, about the time ARCO is supposed to have acquired market power. The expansion suggests that if ARCO attempted to curtail the total volume of gasoline sold to Las Vegas motorists, existing competitors could easily offset that action by delivering more gasoline. The expansion contradicts Rebel’s position that ARCO is reasonably close to controlling the gasoline supply and retail gasoline prices. See Indiana Grocery, 864 F.2d at 1414 .

Rebel’s economic expert contends refiner-suppliers, such as Texaco, in reality will not deliver more gasoline to dealers if ARCO attempts to raise its price above competitive levels. Rebel explains that any retail dealers who attempt to challenge ARCO’s high prices are prevented from doing so because refiner-suppliers are oligopolists who will raise their wholesale prices whenever ARCO raises its retail prices. As Rebel describes it, “retailers bear the brunt of bad times and are denied the fruits of good times.”

To support his contention that refiner-suppliers are oligopolists who restrict the gasoline supply, Rebel’s expert relied on pricing data showing that, in two successive periods, prices in Las Vegas were 4 to 19 cents higher than prices in Los Angeles, when adjusted for transportation costs. This is direct proof, Rebel says, that ARCO actually exercised market power in Las Vegas in an attempt to recoup predatory losses, and that the oligo-polist refiner-suppliers joined the supracom-petitive pricing. Rebel’s experts concluded that ARCO recouped more than $14 million, with no loss of market share.

Rebel’s evidence cannot, as a matter of law, be the basis for inferring market power in its attempted monopolization claim. Although oligopoly pricing cannot be ruled out as a plausible means to recoup predatory losses, Brook Group, — U.S. at-, 113 S.Ct. at 2590 , oligopoly pricing standing alone does not prove that ARCO has market power*, at least not the degree of market power to raise the concerns of the Sherman Act. “The fact that competitors may see proper, in the exercise of their own judgment, to follow the prices of another manufacturer, does not establish any suppression of competition or any sinister domination,” and does not violate the Sherman Act. In re Coordinated Pretrial Proceedings in Petroleum Prod. Antitrust Litigation, 906 F.2d 432, *1443 444 (9th Cir.1990) (quoting United States v. International Harvester Co., 274 U.S. 693, 708-09 , 47 S.Ct. 748, 753-54 , 71 L.Ed. 1302 (1927)), cert. denied, 500 U.S. 959 , 111 S.Ct. 2274 , 114 L.Ed.2d 725 (1991). The reason for this rule is clear. To pose a threat of monopolization, one firm alone must have the power to control market output and exclude competition. Indiana Grocery, 864 F.2d at 1415 . An oligopolist lacks this unilateral power. By definition, oligopolists are interdependent. An oligopolist can increase market price, but only if the others go along. Areeda & Turner, ¶ 1410b, at 64.

In Indiana Grocery, the Seventh Circuit rejected an argument similar to Rebel’s. The plaintiff, Indiana Grocery, claimed that the defendant’s predatory pricing “disciplined” rivals to the extent that the defendant and his rivals had engaged in oligopoly pricing. The court rejected the plaintiffs argument that oligopoly pricing indicated market power, stating

Indiana Grocery, 864 F.2d at 1416 (quoting Copperweld Corp. v. Independence Tube Corp., 467 U.S. 752, 768 , 104 S.Ct. 2731, 2740 , 81 L.Ed.2d 628 (1984)) (citations omitted); cf. Crimpers Promotions, Inc. v. Home Box Office, Inc., 724 F.2d 290 , 291 n. 1 (2d Cir.1983), cert. denied, 467 U.S. 1252 , 104 S.Ct. 3536 , 82 L.Ed.2d 841 (1984) (claim that two firms expressly colluded in an attempt to monopolize would be one of oligopoly under § 1 rather than monopoly under § 2). We recognize that a gap in the Sherman Act allows oligopolies to slip past its prohibitions, see Sullivan & Harrison, ¶6.07, at 262-63, but filling that gap is the concern of Congress, not the judiciary.

In summary, Rebel failed to establish a genuine issue of material fact on market power to support its attempted monopolization claim. Although there is a genuine issue regarding market share and entry barriers, there appears to be no genuine issue regarding the ability of ARCO’s existing competitors to increase their output. The undisputed record indicates that the gasoline supply in Las Vegas is highly elastic and that competitors could increase their output if ARCO raised prices. Consequently, Rebel failed to produce sufficient evidence to support a finding that ARCO is dangerously close to obtaining the power to monopolize the market. Rebel’s failure to make a sufficient showing of market power demonstrates that ARCO’s alleged predatory pricing has not threatened consumer welfare in a manner cognizable under § 2 of the Sherman Act. For this reason, Rebel cannot obtain damages, because any injury from ARCO’s alleged below-cost pricing is not antitrust injury under Sherman Act § 2.

Rebel’s claim under Sherman Act § 1 alleges that ARCO conspired with Terrible Herbst, its largest dealer, to fix retail gasoline prices in Las Vegas at predatory levels. To establish a Sherman Act § 1 violation for vertical, maximum price fixing, 14 a private plaintiff seeking damages must demonstrate: (1) collusion to fix predatory prices, and (2) causal antitrust injury. See USA Petroleum, 495 U.S. at 335-45 , 110 S.Ct. at 1889-95 .

Rebel argues that market power is not a prerequisite to antitrust injury if the claim is premised on Sherman Act § 1. It reasons that because vertical price fixing is per se illegal, Arizona v. Maricopa County Medical Soc’y, 457 U.S. 332, 347 , 102 S.Ct.2466, 2474-75, 73 L.Ed.2d 48 (1982), market analysis is not necessary. Rebel confuses proof of liability with proof of antitrust injury. Per se rules relieve plaintiffs of the burden of proving anticompetitive effects, which are assumed, but they do not excuse *1444 plaintiffs from showing that their injury was caused by the anticompetitive aspects of the illegal act. Newman v. Universal Pictures, 813 F.2d 1519, 1522-23 (9th Cir.1987), cert. denied, 486 U.S. 1059 , 108 S.Ct. 2831 , 100 L.Ed.2d 931 (1988). The “mere presence” of a per se violation under Sherman Act § 1 “does not by itself bestow on any plaintiff a private right of action for damages.” Indiana Grocery, 864 F.2d at 1419 . A plaintiff must prove that his injury flows from the anticompetitive aspect of the defendant’s conduct. For example, in USA Petroleum, the plaintiff did not suffer antitrust injury from the defendant’s conspiracy to fix low prices. Although per se illegal, maximum price fixing cannot cause antitrust injury because low prices are the “very essence of competition.” Id. 457 U.S. at 338 , 102 S.Ct. at 2469 -70 (quoting Matsushita, 475 U.S. at 594 , 106 S.Ct. at 1359-60 ).

Unlike USA Petroleum, Rebel accuses ARCO of conspiring to fix predatory, below-cost prices. Some commentators have argued that a competitor harmed by a predator’s below-cost pricing need not show market power in order to prove antitrust injury, because below-cost pricing harms consumer welfare by causing allocative inefficiency-that is, society’s resources are allocated to the predated goods in amounts exceeding-society’s valuation of those goods. 15 See Roger D. Blair & Jeffrey L. Harrison, Rethinking Antitrust Injury, 42 Vand.L.Rev. 1539, 1564 (1989). The Supreme Court in Brook Group foreclosed this theory: “Although unsuccessful predatory pricing may encourage some inefficient substitution toward the product being sold at less than its cost, unsuccessful predation is in general a boon to consumers.” Brook Group, - U.S. at-, 113 S.Ct. at 2588 . Because below-cost pricing is a “boon to consumers,” the losses inflicted on Rebel by the pricing are not the stuff of antitrust injury. It would be incongruous to award damages to plaintiffs for actions that in general benefit consumer welfare. See Indiana Grocery, 864 F.2d at 1419 . To show antitrust injury under Sherman Act § 1, a plaintiff must show that the predator has market power.

Rebel does not allege that ARCO conspired with its competitors in Las Vegas to monopolize the market. Thus, for the same reasons that we stated in our analysis of Rebel’s claim under Sherman Act § 2, Rebel’s evidence is insufficient for a jury reasonably to conclude that ARCO possesses market power, or is dangerously close to obtaining it, under § 1. In light of this conclusion, any injury-in-fact suffered by Rebel as a result of ARCO’s alleged predatory maximum price fixing does not constitute antitrust injury.

Rebel challenges the district court’s summary judgment in favor of ARCO on its price discrimination claim. As with Rebel’s other claims, the court concluded that Rebel could not prove antitrust injury, because Rebel had produced insufficient evidence that ARCO possessed market power. The issues we address are twofold: whether a showing of market power is required to demonstrate antitrust injury in a primary-line discrimination claim, and, if so, whether Rebel presented sufficient evidence of market power for a reasonable jury to decide in its favor.

Section 2(a) of the Clayton Act, as amended by the Robinson-Patman Act, prohibits price discrimination “where the effect of such discrimination may be substantially to lessen competition or tend- to create a monopoly in any line of commerce, or to injure, destroy, or prevent competition with any person who either grants or knowingly receives the benefit of such discrimination, or with customers *1445 of either of them.” 15 U.S.C. § 13 (a) (emphasis added).

Price discrimination occurs in two forms: as a price differential between geographical markets, and as a price differential between purchasers. Price discrimination in the former sense, called primary-line injury, occurs when the seller charges predatory, below-cost prices in one market in an attempt to eliminate competitors there, and charges supracompetitive prices in another market. A competing seller in the predated market may claim injury. Price discrimination in the latter sense, called secondary-line injury, occurs when a seller grants a favorable price to one purchaser but not another. In that case, the disfavored purchaser may sue.

Rebel’s claim alleged primary-line discrimination. As proof, Rebel’s expert relied on pricing data showing that for four years, between 1985 and 1989, ARCO charged predatory prices in Las Vegas that, when adjusted for transportation costs, were 6 to 14 cents per gallon below those charged in Los Angeles. Rebel’s expert also produced data showing that, from 1989 to 1991, ARCO intermittently charged prices in Las Vegas that were 4 to 19 cents per gallon higher than in Los Angeles. Rebel argues that this price discrimination shows that ARCO attempted to eliminate rivals and has charged supracompetitive prices to recoup its losses.

Because primary-line injury is of the “same general character” as predatory pricing schemes actionable under Sherman Act § 2, Brook Group, - U.S. at-, 113 S.Ct. at 2587 , a primary-line injury plaintiff bears essentially the same substantive burden as a plaintiff under the Sherman Act. That is, to hold a defendant liable for charging below-cost prices, a primary-line plaintiff must show that the predator stands some chance of recouping his losses. Id. at-, 113 S.Ct. at 2588 . 16 Under the Clayton Act, as amended by the Robinson-Patman Act, the standard is a “reasonable prospect” of recoupment. Id. This requires the plaintiff to show that the predator has market power, or that he has some reasonable prospect of obtaining it. Absent any chance of such showing, the below-cost pricing poses no threat to competition. Id.

The parties disagree over whether Rebel must show that ARCO has market power in order to satisfy the antitrust injury requirement. Brook Group is not necessarily controlling. Brook Group addressed only the standards for liability in a primary-line claim. The standards for proving a violation differ from the standards for proving antitrust injury. Indeed, the antitrust injury requirement imposes a more stringent standard than the rules of liability. See J. Truett Payne Co. v. Chrysler Motors Corp., 451 U.S. 557, 562 , 101 S.Ct. 1923, 1927 , 68 L.Ed.2d 442 (1981). Substantive liability looks to whether the defendant’s conduct is forbidden; antitrust injury looks to whether the claimed injury “fall[s] squarely within the area of congressional concern.” Blue Shield of Virginia v. McCready, 457 U.S. 465, 484 , 102 S.Ct. 2540, 2551 , 73 L.Ed.2d 149 (1982). “Thus, to define the scope of antitrust injury for a violation of the Clayton Act, it is the purpose of that legislation, and the interests it seeks to protect, which is controlling.” Greater Rockford Energy & Technology Corp. v. Shell Oil Co., 998 F.2d 391, 399 (7th Cir.1993), cert. denied, - U.S. -, 114 S.Ct. 1054 , 127 L.Ed.2d 375 (1994). For example, because the Sherman Act’s concern is consumer welfare, antitrust injury occurs only when the claimed injury flows from acts harmful to consumers.

We conclude that a plaintiff alleging primary-line discrimination must prove antitrust injury in the same manner that a Sherman Act plaintiff does-by showing that his injury flows from those aspects or effects of the conduct that are harmful to consumer welfare. Legislative history supports our conclusion. Congress first prohibited pri *1446 mary-line discrimination with the Clayton Act of 1914. In its original enactment, the Clayton Act forbid discrimination that might “substantially ... lessen competition or tend to create a monopoly in any line of commerce.” Clayton Antitrust Act, 38 Stat. 730 (1914). The statute specifically applied only to primary-line injury. Its purpose was to prevent large corporations from invading markets of small firms and charging predatory prices for the purpose of destroying mar-ketwide competition. See H.R. No. 627, 63rd Cong., 2d Sess. 8 (1914); Sullivan & Harrison, § 8.03, at 321. Like the Sherman Act, the original Clayton Act’s primary aim was to prevent harm to consumers. See Janich Bros., Inc. v. American Distilling Co., 570 F.2d 848, 855 (9th Cir.1977) (both the Sherman Act and Clayton Act are “aimed at the same economic evil”), cert. denied, 439 U.S. 829 , 99 S.Ct. 103 , 58 L.Ed.2d 122 (1978).

In holding that a primary-line plaintiff must demonstrate an injury flowing from an aspect of the' defendant’s conduct injurious to consumer welfare, we intend in no way to affect the standard for antitrust injury in secondary-line cases.' Secondary-line discrimination is forbidden by the RobinsonPatman Act, which amended the Clayton Act. It is clear that Congress intended the Robinson-Patman Act’s provisions to apply only to secondary-line cases, not primary-line eases. See H.R. No. 2287, 74th Cong., 2d Sess. 8 (1936). 17 The Robinson-Patman Act stands on entirely different footing than the Sherman Act and Clayton Act. While the framers of the Sherman and Clayton Acts intended to proscribe only conduct that threatens consumer welfare, the framers of the Robinson-Patman Amendments intended to punish perceived economic evils not necessarily threatening to consumer welfare per se. Fairness and protection of secondary-line purchasers are the concerns of the Robinson-Patman Act, a conclusion that is confirmed by the language of the statute, legislative history and judicial precedent. 18 Accordingly, we have held that a secondary-line plaintiff can demonstrate antitrust injury if the price discrimination caused him to lose sales and profits; he need not demonstrate any injury to consumer welfare. Hasbrouck v. Texaco, Inc., 842 F.2d 1034, 1042 (9th Cir.1987), aff'd on other grounds, 496 U.S. 543, 557-58 , 110 S.Ct. 2535, 2543-44 , 110 L.Ed.2d 492 (1990); accord J.F. Feeser v. Serv-A-Portion, Inc., 909 F.2d 1524 , 1540 (3d Cir.1990), cert. denied, 499 U.S. 921 , 111 S.Ct. 1313 , 113 L.Ed.2d 246 (1991); Alan’s of Atlanta, Inc. v. Minolta Corp., 903 F.2d *1447 1414, 1427 (11th Cir.1990), but see Boise Cascode Co rp. v. FTC, 837 F.2d 1127, 1143 (D.C.Cir.1988) (injury to competitor may be rebutted by lack of consumer injury). As a ease involving secondary-line injury, not primary-line injury, Hasbrouck is not implicated by the present case.

To prove antitrust injury, Rebel cannot rely solely on evidence that ARCO charged lower prices in Las Vegas than in Los Ange-les. Lower prices, even if below cost, are the essence of competition and are a “boon to consumers.” Brook Group, — U.S. at-, 113 S.Ct. at 2588 . To demonstrate antitrust injury in its primary-line claim, Rebel must show that ARCO has market power.

The required showing of market power in a primary-line claim is somewhat less than is required in a Sherman Act claim. In contrast to the Sherman Act, which' speaks only of various forms of express agreement and monopoly, the Clayton Act, as amended by the Robinson-Patman Act, prohibits price discrimination that may lessen competition. Brook Group, at -, 113 S.Ct. at 2591 . Consequently, the Supreme Court has held that “excessive concentration, and the oligopolistic price coordination it portends, may be the injury to competition the Act prohibits.” Brook Group, at-, 113 S.Ct. at 2591 (citing United States v. Philadelphia Nat’l Bank, 374 U.S. 321 , 83 S.Ct. 1715 , 10 L.Ed.2d 915 (1963)). This suggests that Rebel need only prove that ARCO has a degree of market power to threaten oligopoli-zation, not monopolization. Despite this difference, consumer welfare is still the lodestar for determining antitrust injury in a primary-line claim. To show a lessening of competition, it is not enough to show that the number of rivals has been reduced or that the market is concentrated. Rather, it must be demonstrated that a predator has a “reasonable prospect” of enforcing oligopoly pricing in Las Vegas, and thus of recouping its predatory losses. Brook Group, — U.S. at-, 113 S.Ct. at 2588 .

We now consider whether Rebel’s evidence is sufficient to show antitrust injury in its price discrimination claim.

In our discussion of Rebel’s attempted monopolization claim, we concluded that Rebel’s circumstantial evidence of market share and entry barriers in the retail market could not support a finding of market power to support an attempted monopolization claim, in light of undisputed evidence of recent expansion by Texaco and Southland, two of ARCO’s existing competitors. We must evaluate that evidence again to determine whether Rebel has raised a disputed issue of material fact as to whether ARCO has the market power to “lessen competition” by enforcing an oligopoly. The degree of market power need not be as high as in a monopolization claim. In light of this lesser standard, we conclude that Rebel has established a disputed issue of material fact as to its standing to state a cause of action under the Clayton Act.

Rebel produced data showing that, during the alleged recoupment period, ARCO’s retail gasoline prices were higher in Las Vegas than in Los Angeles, when adjusted for transportation costs. Rebel’s experts opined that this is direct evidence of supracompeti-tive pricing by ARCO, as Los Angeles is a highly competitive retail gasoline market. Further, Rebel produced evidence that ARCO’s market share actually increased during the period when it was charging allegedly supracompetitive prices. Rebel contends that this evidence indicates that ARCO’s existing competitors in the retail gasoline market have been “disciplined” by ARCO’s pattern of below-cost predatory pricing, and are unwilling to risk price competition with ARCO. Rebel argues that this is evidence that ARCO has achieved the market power to enforce a classic disciplined oligopoly, one of the anticompetitive effects that the Clayton Act prohibits. See Brook Group, — U.S. at-, 113 S.Ct. at 2591 .

As we noted above, Rebel has produced evidence of significant barriers to entry in the Las Vegas retail gasoline market. This is crucial, because if entry is unimpeded, supracompetitive profit levels in the market would attract new competitors to the market, who could undercut ARCO’s prices and prevent it from recouping its predatory losses. *1448 See Brook Group, at-, 113 S.Ct. at 2589 . We have already determined that Rebel’s evidence indicates that these barriers to entry are significant, and thus that new competitors are unlikely to enter the market at price and volume levels sufficient to challenge ARCO.

Rebel may rely on this evidence of oligopoly pricing to prove market power on its price discrimination claim. Although such evidence cannot support a claim of monopolization under the Sherman Act, no such limitation exists when the claim involves price discrimination under the Clayton Act. Brook Group, — U.S. at-, 113 S.Ct. at 2591 . Further, the lack of evidence of output restriction is not so significant here. The economic forces at work in an oligopoly are very different than in a monopoly. A predator is able to establish and maintain supracompeti-tive prices in an oligopoly by making it too painful for its existing competitors to challenge its prices, and thus, “disciplining” them. This distinction between oligopolistic and monopolistic practices is crucial to the survival of Rebel’s price discrimination claim. Read in the most favorable light, Rebel’s evidence tends to indicate that no new competition can enter the market to challenge ARCO, and that the existing competition, while it may be able to challenge ARCO, lacks the will to do so.

We conclude that Rebel has produced sufficient evidence to create a disputed issue of material fact as to whether ARCO has sufficient market power to maintain oligopoly pricing. Summary judgment on Rebel’s price discrimination claim was therefore improper; Rebel may be able to demonstrate antitrust injury on this claim. 19

Ill

Rebel has failed to provide sufficient evidence to support a jury verdict on the issue of ARCO’s “market power” under the Sherman Act. As a result, Rebel cannot establish a genuine issue of fact regarding antitrust injury on its Sherman Act § 1 and § 2 claims, and the district court properly granted summary judgment for ARCO on those claims. Rebel’s evidence is sufficient, however, to raise a disputed question of material fact as to whether ARCO achieved sufficient market power to enforce supracompetitive oligopoly pricing. 20 This showing is sufficient to allow Rebel to survive summary judgment on the issue of antitrust injury resulting from price discrimination under the Clayton Act.

The judgment of the district court is AFFIRMED in part and REVERSED in part. We REMAND to the district court for further proceedings consistent with this opinion.

. Marginal cost is the cost that a firm incurs in the production of one additional unit of output. Herbert Hovenkamp, Economics and Federal Antitrust Law 10 (1985). Rebel measured ARCO's marginal cost based upon an agreement that ARCO signed with Tosco Corp. to purchase incremental supplies of gasoline. Rebel claims that ARCO bought the additional supplies of gasoline from Tosco because it was expanding output but was unable to obtain the additional output from its own refineries. The cost of this "last barrel” of gasoline represents ARCO’s marginal cost for gasoline.

. Sherman Act § 2 states: “Every person who shall monopolize, or attempt to monopolize, or combine or conspire with any person or persons, to monopolize ... trade shall be guilty” of an antitrust violation. 15 U.S.C. § 2 (1994).

. "Losses during a price war will be proportionally higher for the predator because he faces the necessity of expanding his output at ever higher costs, while the victim not only will not expand output but has the option of reducing it and so decreasing his costs." Robert H. Bork, The Antitrust Paradox 148 (1979).

. Social welfare is maximized when the price of a good equals its marginal cost — the cost of producing the last unit of output. When a firm with market power cuts output to increase prices, price exceeds marginal cost. This causes a loss to society of all that additional output which the firm could produce by lowering its price to marginal cost. Janich Bros., Inc. v. American Distilling Co., 570 F.2d 848 , 857 n. 9 (9th Cir.1977), cert. denied, 439 U.S. 829 , 99 S.Ct. 103 , 58 L.Ed.2d 122 (1978).

. The price-correlation data purportedly showed that the prices of self-serve, cash-only gasoline and full-serve gasoline tend to move together, i.e., when one goes up, the other goes up, and vice versa. The expert noted that under the "Stigler’' method of proving supply elasticity, high price correlation indicates that consumers view the products as substitutes for each other. See George J. Stigler & Robert A. Sherwin, The Extent of the Market, 28 J.L. & Econ. 555 , 572-73 (1985). Self-serve sellers could not increase price and still expect to serve the same number of customers (and, hence, sell the same quantity of gasoline) as they did at the lower price.

. If we accepted Rebel's argument, we would be assuming the existence of ARCO’s market power before defining the relevant market, a method of analysis that defies logic. Moreover, the purpose of defining the relevant market in an attempt-to-monopolize claim is to determine whether ARCO has sufficient market power to pose a threat of monopoly. The fact that the market is an oligopoly is of no concern, as an oligopoly is not per se illegal under Sherman Act § 2. See Indiana Grocery, Inc. v. Super Valu Stores, Inc., 864 F.2d 1409, 1415-16 (7th Cir.1989).

. As proof of conspiracy, Rebel submitted an article in a trade publication in which Ed Herbst, vice president of the family-owned business, was quoted as saying: "ARCO opened stores in Las Vegas and was selling 5 to 7 cents a gallon below what we paid wholesale. My dad said, 'If you can’t beat them, join them.’ So we joined them."

.To prove a conspiracy to monopolize, Rebel must show that the independent dealers had the specific intent to . conspire to monopolize; it is not enough to show that the dealers merely agreed to go along with ARCO’s pricing. See Belfiore v. New York Times Co., 826 F.2d 177, 183 (2d Cir.1987), cert. denied, 484 U.S. 1067 , 108 S.Ct. 1030 , 98 L.Ed.2d 994 (1988).

. While the issue of collusion is a matter disputed by ARCO, the district court limited discovery strictly to the issue of market power. No discovery was allowed on collusion. Given that fact, all inferences on the issue of collusion must be drawn in Rebel’s favor to properly determine the summary judgment motion. See Fludd v. United States Secret Serv., 771 F.2d 549, 554 (D.C.Cir.1985) (per curiam).

. The Fourth Circuit has stated that claims involving shares between 30 and 50 percent "should usually be rejected, except when conduct is very likely to achieve monopoly or when conduct is invidious, but not so much so as to make the defendant per se liable.” M & M Medical Supplies, 981 F.2d at 168 (citing Areeda & Turner ¶ 835c, at 350). This standard, initially proposed by Professors Areeda and Turner, would include predatory conduct in the "invidious” category. In addition, a separate standard would presume market power if the market share is greater than 50 percent, if the other substantive elements of an attempt to monopolize claim are satisfied. Id. at 168 .

We are reluctant to apply such bright-line rules regarding market share in deciding whether a defendant has market power to restrict output or raise prices. Courts should be "wary of the numbers game of market percentage" when considering attempt-to-monopolize claims. Dimmitt Agri Industries, 679 F.2d at 533 . As Professors Areeda and Turner admit, their refined rules may be "illusory” guides to deciding market power. Areeda & Turner ¶ 835c, at 350. The far wiser approach, which this circuit has observed if not explicitly adopted, is illustrated by Ball Memorial Hosp., 784 F.2d at 1335 (7th Cir.) and Ryko Mfg., 823 F.2d at 1232 (8th Cir.), where the issue of market power was decided by carefully analyzing certain telltale factors in the relevant market: market share, entry barriers and the capacity of existing competitors to expand output. We see no reason to depart from this mode of analysis.

. "It is well known that some of the most insuperable barriers in the great race of competition are the result of government regulation.” United States v. Syufy Enterprises, 903 F.2d 659, 673 (9th Cir.1990).

. A refiner is barred if it meets three conditions: (1) produced more than 30 percent of the domestic and imported crude oil supplied to its refinery; (2) refines gasoline from crude oil; and (3) has a total refinery capacity of more than 175,-000. 15 Nev.Rev.Stat. § 597.340.

. According to Rebel's own expert, these marketers included Domino's, Desert Bait, Green Valley, King 8 and Express.

. Section 1 states: "Every contract, combination in the form of trust or otherwise, or conspiracy, in restraint of trade or commerce ... is declared to be illegal.” 15 U.S.C. § 1 (1994).

. To put it another way, pricing below marginal cost is socially wasteful because the seller produces goods at a cost which is greater than their value to consumers. Janich Bros., 570 F.2d at 857-58 ; Hovenkamp, Economics and Federal Antitrust Law, at 46.

As we have noted previously, allocative efficiency is synonymous with consumer welfare, Chesapeake and Ohio R.R. Co. v. United States, 704 F.2d 373, 376 (7th Cir.1983), and is the central goal of the Sherman Act. Northern Pac. Ry. Co. v. United States, 356 U.S. 1 , 78 S.Ct. 514 , 2 L.Ed.2d 545 (1958) (the Sherman Act "rests on the premise that the unrestrained interaction of competitive forces will yield the best allocation of our economic resources.”)

. The Supreme Court implicitly overruled Utah Pie Co. v. Continental Baking Co., 386 U.S. 685 , 87 S.Ct. 1326 , 18 L.Ed.2d 406 (1967), a controversial case which set forth different standards for primary-line injury. The court explained that Utah Pie was merely an “early judicial inquiry.” Brook Group,-U.S. at-, 113 S.Ct. at 2587 .

. The Robinson-Patman Act "attaches to competitive relations between a given seller and his several customers. It concerns discrimination between customers of the same seller. It has nothing to do with ... requiring] the maintenance of any relationship in prices charged by a competing seller." H.R. Report No. 2287, 74th Cong., 2d Sess. 8 (1936).

. Enacted in 1936, the Robinson-Patman Act amendments were an outgrowth of dissatisfaction. with the original Clayton Act’s inability to prevent large retail chains from obtaining "quantity discounts” from big suppliers, at the disadvantage of small retailers who competed with the chains. H.R. Report No. 2287, 74th Cong., 2d Sess. 3-4 (1936); S.Rep. No. 1502, 74th Cong., 2d Sess. 4 (1936); FTC v. Morton Salt Co., 334 U.S. 37, 43 , 68 S.Ct. 822, 826-27 , 92 L.Ed. 1196 (1948).

The new law broadened the scope of the Clayton Act’s provisions. While the Clayton Act only, proscribed conduct that may "substantially lessen competition or tend to create a monopoly!)]” the new law added the following crucial passage: "or to injure, destroy, or prevent competition with any person who either grants or knowingly receives the benefit of such discrimination, or with customers of either of them.” The purpose of this passage was to relieve secondary-line plaintiffs — small retailers who are disfavored by discriminating suppliers — from having to prove harm to competition marketwide, allowing them instead to impose liability simply by proving effects to individual competitors. H.R. Report No. 2287, 74th Cong., 2d Sess. 8 (1936). As the House Report stated:

The Senate Report stated that the amendments aimed to prevent competitor injury as a means of preventing general market injury — "to catch the weed in the,seed will keep it from coming to flower.” S.Rep. No. 1502, 74th Cong., 2d Sess. 4 (1936). As the Supreme Court long ago stated: "Congress intended to protect a merchant from competitive injury attributable to discriminatory prices ...” Morton Salt Co., 334 U.S. at 49 , 68 S.Ct. at 829-30 ; Greater Rockford Energy, 998 F.2d at 399 -400 n. 11.

. ARCO argues that Rebel's continued presence in the Las Vegas Market undercuts Rebel’s claim of antitrust injury. ARCO contends that, while oligopoly pricing hurts consumer welfare, Rebel, as a competitor in the market, stood to benefit from any oligopoly pricing. See Matsushita, 475 U.S. at 574 , 106 S.Ct. at 1348 . It is undisputed that during the peak of the alleged oligopoly pricing, when the cost of a gallon of gasoline was more than 19 cents higher in Las Vegas than in Los Angeles, Rebel's prices were within a penny of ARCO’s prices. Rebel may have received gains during this period offsetting some of the lost profits it suffered during the period of ARCO's alleged below-cost pricing. Even if this is true, however, the argument goes to the measure of damages, rather than the existence of antitrust injury. If Rebel is ultimately able to prove injury resulting from anticompetitive conduct on the part of ARCO, it should not be precluded from recovery simply because it was not driven completely from the market.

. We, of course, express no opinion on the ultimate merits of the claim. At trial. Rebel may fail to establish a "reasonable prospect” of re-coupment by ARCO, or ARCO may be able to prove facts sufficient to establish a defense to Rebel's claim.