Full opinion text
SNEED, Circuit Judge: William Inglis & Sons Baking Co. (Inglis) brought this private antitrust suit to recover treble damages against ITT Continental Baking Co. (Continental), American Bakeries Co. (American), and Campbell-Taggart, Inc., alleging violations of sections 1 and 2 of the Sherman Act, 15 U.S.C. §§ 1, 2, section 2(a) of the Clayton Act, as amended by the Robinson-Patman Act, 15 U.S.C. § 13(a), and the California Unfair Practices Act, Cal.Bus. & Prof.Code §§ 17000-17101. Inglis also charged that Continental had conspired with its parent corporation, International Telephone & Telegraph (ITT), and others in violation of sections 1 and 2 of the Sherman Act. Both Continental and American filed counterclaims against Inglis also alleging antitrust violations, although Continental dropped its counterclaim at trial. Before trial Campbell-Taggart settled with Inglis, and the district court granted summary judgment for Continental with respect to the alleged “vertical” conspiracy between Continental, ITT, and others. Later, Inglis voluntarily dropped its horizontal conspiracy claims under section 1 against the named defendants. Following a one month trial in 1978, the jury returned a verdict against Continental on all remaining claims and awarded damages of $5,048,-000. The jury found that neither American nor Inglis were liable on the claims they filed against each other. Continental then moved for judgment notwithstanding the verdict (JNOV) or, in the alternative, a new trial on all claims. The district court granted the motions for JNOV and, in the alternative, a new trial on the federal claims but refused to grant JNOV for Continental on the state claims. Instead, a new trial was ordered. William Inglis & Sons Baking Co. v. ITT Continental Baking Co., 461 F.Supp. 410 (N.D.Cal. 1978). Inglis now appeals the district court’s entry of JNOV and alternative order for a new trial on the federal claims, and Continental appeals the court’s refusal to enter JNOV in its favor on the state claims, pursuant to 28 U.S.C. § 1292(b). We affirm in part, reverse in part, and remand this case for a new trial. I. STATEMENT OF THE CASE A. The Theory of Plaintiff’s Actions Inglis was a family-owned wholesale bakery with production facilities located in Stockton, California. It manufactured and distributed bread and rolls in northern California. Continental is one of the nation’s largest wholesale bakeries, and was a competitor of Inglis in the northern California market, with production facilities in San Francisco, Oakland, and Sacramento. The primary products involved in this case were the one pound and one and one-half pound loaves of white pan bread. During the period covered by Inglis’ complaint, both Continental and Inglis sold their bread under a “private” label and an “advertised” label. Private label bread is manufactured by the wholesaler on behalf of a particular retail customer and marketed under a label exclusively held by that customer. Advertised label bread generally is a national brand name available to all retail purchasers. Continental’s advertised bread bears the “Wonder” label, while Inglis marketed “Sunbeam” bread. Labeling aside, the principal difference between private label and advertised bread was one of price. Wholesale bakeries typically sold private label bread at a lower price than advertised brands and, because the products were essentially the same, at a lower profit. Both types of bread generally were manufactured at the same production facilities and both could be found on the shelves of most large retailers. Inglis’ complaint, which was filed in 1971 and supplemented in 1977, was founded on charges that Continental sought to eliminate competition in the northern California market for wholesale bread by charging discriminatory and below-cost prices for its private label bread. Inglis claims that it was the principal victim of this predatory scheme, suffering losses since 1967 and eventually going out of business in April 1976, nearly five years after it filed its initial complaint. The theory on which Inglis structured its case was that the growth of private label bread, which began in northern California in 1967 or 1968, began to weaken Continental’s market for Wonder bread. In response to this challenge Continental also began selling private label bread, but the price gap between private label and Wonder bread persisted. Inglis argues that Continental then decided to pursue a strategy of predatory pricing in its sales of private label bread, with the intent of eliminating independent wholesalers like Inglis who were financially less capable of withstanding a price war. The ultimate goal, Inglis asserts, was to acquire a large share of the private label market and then to use the enhanced market power to raise private label prices, which would diminish the competitive disadvantage of Wonder bread. Moreover, Inglis contends, the acquisition of private label accounts would enable Continental to “leverage” more shelf space for Wonder bread from those retailers who also purchased Continental’s private label bread. B. The Evidence Summarized To support its theory, Inglis introduced the following evidence. First, Inglis examined the movement in Continental’s prices during the complaint period, focusing on the one pound loaves of bread. In September 1970 Continental reduced the price of its private label bread from 19 to 18 cents per loaf and maintained that price for nearly two years. In July 1972 Continental further reduced its price to 17.2 cents and maintained that price through the summer of 1973. Thereafter Continental gradually began to raise the price, allegedly because it then knew that Inglis was in its death throes as a competitor. Second, Inglis established that Continental suffered substantial losses from its northern california bakeries from 1971 through 1974, the period during which Continental’s private label prices were at their lowest. Inglis also introduced expert testimony, based on a study of prices during brief periods in 1972 and 1973, tending to show that Continental’s private label prices were below its average variable cost of production. Third, Inglis showed that Continental actively made competing offers to private label accounts held by Inglis. Although Inglis actually lost only one account to Continental, it nevertheless was forced to respond with lower prices of its own and suffer the resulting loss of revenue from sales. Finally, Inglis introduced documentary evidence designed to prove Continental’s intent to drive Inglis from the market. This evidence principally consisted of a report prepared by independent consultants identifying strategies Continental might adopt to combat private label competition. One alternative involved maintaining prices “to hasten wholesaler exit.” Inglis also introduced reports by Continental salesmen targeting Inglis private label accounts for enhanced competitive efforts. Continental’s explanation of events during the complaint period, of course, differed sharply from that of Inglis. First, Continental emphasized the intensely competitive nature of the wholesale bread market in northern California and its own lack of market power. Campbell-Taggart held the largest share of the market and, although some of the evidence is ambiguous, apparently initiated price reductions that other competitors, including Continental, were forced to follow. Second, during the complaint period the market was affected by the growth of so-called “captive” bakeries. Retail stores such as Safeway established their own bakeries, thereby reducing the demand for wholesale bread products. One result was that all of the wholesale bakeries experienced excess capacity during the relevant period. As striking evidence of this, Continental proved that during an eleven-week strike in December 1972 and January 1973, which closed the bakeries operated by Continental and Campbell-Taggart, American and Inglis were able to supply the entire market with their existing capacity. In addition to creating excess capacity, the captive bakeries also exerted pressure on other retailers to provide price-competitive private label products, pressure to which Continental and other wholesalers responded. Finally, Continental emphasized that all of the bakeries were subject to federal price controls from the summer of 1971 until April 1974. Within this period the federal government also imposed a temporary price freeze during the summer months of 1973. According to Continental, the price controls contributed substantially to its inability to raise prices despite increasing costs during the period in which private label prices were at their lowest levels. Continental argues that the price increases which occurred in late 1973 and 1974 resulted not from Continental’s anticipation of Inglis’ demise, but from the expiration of government price constraints. From the evidence presented at trial, this much is plain: The price competition among wholesalers in northern California, all selling substantially similar products, was intense. As a result, at least in part, many bakeries failed to earn a profit and Inglis was forced to discontinue operations. The central question for the jury thus was whether Inglis was a casualty of vigorous, but honest, competition, or the victim of unfair and predatory tactics adopted by a company intent on monopolizing the market. Because there was no “smoking gun,” the jury was asked to choose between the conflicting inferences drawn by Inglis and Continental and it found Inglis’ explanation the more reasonable. C. The Findings and Holdings of the District Court As already indicated, the district court found a lack of competent evidence to support the jury’s conclusion. First it held that Inglis’ expert testimony concerning below-cost pricing by Continental was either insufficient as a matter of law to establish predatory pricing for purposes of section 2 of the Sherman Act, or completely unreliable. The court determined that on the facts of this case, Inglis was required to prove that Continental’s prices were below its marginal cost of producing bread. Proof of pricing below average variable cost, which the court in any event found not to be controlling, was insufficient. With respect to Inglis’ Robinson-Patman Act claim, the court held that Inglis’ failure to show pricing below marginal cost also prevented it from establishing the injury to competition required by Robinson-Patman. The court ordered a new trial on the state claim because it found that the weight of the evidence supported Continental’s defense of meeting competition. Finally, the court held that a new trial was warranted because the jury’s damage award was excessive and unsupported by the weight of the evidence. D. Standards of Review In passing on the district court’s decision, we are mindful of the deference due the verdict of a jury. To determine whether an entry of JNOV is proper, we must apply the same standard applied by the district court. Alioto v. Cowles Communications, Inc., 519 F.2d 777, 780 (9th Cir.), cert. denied, 423 U.S. 930, 96 S.Ct. 280, 46 L.Ed.2d 259 (1975). That is, we must affirm the district court if, without accounting for the credibility of the witnesses, we find that the evidence and its inferences, considered as a whole and viewed in the light most favorable to the nonmoving party, can support only one reasonable conclusion — that the moving party is entitled to judgment notwithstanding the adverse verdict. Davison v. Pacific Inland Navigation Co., 569 F.2d 507, 509 (9th Cir. 1978); Maheu v. Hughes Tool Co., 569 F.2d 459, 464 (9th Cir. 1977); Fount-Wip, Inc. v. Reddi-Wip, Inc., 568 F.2d 1296, 1300 (9th Cir. 1978). Neither the district court nor this court is free to weigh the evidence or reach a result that it finds more reasonable as long as the jury’s verdict is supported by substantial evidence. Marquis v. Chrysler Corp., 577 F.2d 624, 631 (9th Cir. 1978); Cockrum v. Whitney, 479 F.2d 84, 86 (9th Cir. 1973). In contrast, a new trial may be ordered by the district court if, in its opinion, the jury’s verdict was clearly contrary to the weight of the evidence. We may reverse such an order only if we find that the district court abused its discretion as to each ground upon which its decision was based. Traver v. Meshriy, 627 F.2d 934, 940-41 (9th Cir. 1980); Peacock v. Board of Regents, 597 F.2d 163, 165 (9th Cir. 1979); Fount-Wip, supra, 568 F.2d at 1302; Hanson v. Shell Oil Co., 541 F.2d 1352, 1359 (9th Cir. 1976), cert. denied, 429 U.S. 1074, 97 S.Ct. 813, 50 L.Ed.2d 792 (1977). II. THE SECTION 2 ATTEMPT TO MONOPOLIZE CLAIM — PREDATORY PRICING Bearing in mind these standards, we shall consider, first, the district court’s entry of JNOY for Continental with respect to the jury’s finding that Continental attempted to monopolize the wholesale bread market in northern California by predatorily pricing its products and, next, its alternative order requiring a new trial on the ground that Inglis failed to prove predatory conduct. A. The Elements of an Attempt Claim Although the law of this circuit on attempted monopolization has not been static, its current state recognizes three elements of an attempt claim under section 2 of the Sherman Act: (1) specific intent to control prices or destroy competition in some part of commerce; (2) predatory or anticompetitive conduct directed to accomplishing the unlawful purpose; and (3) a dangerous probability of success. E. g., California Computer Products, Inc. v. IBM Corp., 613 F.2d 727, 736 (9th Cir. 1979) (Cal-Comp). To state these elements, however, is merely to begin the process of understanding the legal standards of conduct under an attempt claim. Each element interacts with the others in significant and unexpected ways. Because the parties dispute the nature of their interdependence, we must discuss each in some detail. For reasons that will appear later, the third element will be discussed second rather than last. 1. Specific Intent The element of specific intent appears to have had its genesis in the distinctions — and similarities — between monopolization and attempted monopolization, both of which are proscribed in separate terms by section 2. See Cooper, Attempts and Monopolization: A Mildly Expansionary Answer to the Prophylactic Riddle of Section Two, 72 Mich.L.Rev. 375 (1974). Thus, section 2 embraces not only an uncertain collection of evils termed “monopolization,” but also conduct falling short of that result. By analogy to the law of criminal attempt, the requirement of specific intent is used to confine the reach of an attempt claim to conduct threatening monopolization. See Times-Picayune Publishing Co. v. United States, 345 U.S. 594, 626, 73 S.Ct. 872, 889, 97 L.Ed. 1277 (1953); United States v. Griffith, 334 U.S. 100, 105, 68 S.Ct. 941, 944, 92 L.Ed. 1236 (1948); Swift & Co. v. United States, 196 U.S. 375, 396, 25 S.Ct. 276, 279, 49 L.Ed. 518 (1905). Whatever its origins, the existence of specific intent may be established not only by direct evidence of unlawful design, but by circumstantial evidence, principally of illegal conduct. E. g., CalComp, supra, 613 F.2d at 736; Sherman v. British Leyland Motors, Ltd., 601 F.2d 429, 453 n.47 (9th Cir. 1979); Gough v. Rossmoor Corp., 585 F.2d 381, 390 (9th Cir. 1978), cert. denied, 440 U.S. 936, 99 S.Ct. 1280, 59 L.Ed.2d 494 (1979); Janich Bros., Inc. v. American Distilling Co., 570 F.2d 848, 853-54 (9th Cir. 1977), cert. denied, 439 U.S. 829, 99 S.Ct. 103, 58 L.Ed.2d 122 (1978). Too heavy a reliance on circumstantial evidence incurs the risk of reducing almost to the point of extinction the existence of the requirement. The type of conduct that will support the inference, therefore, must be carefully defined. This court has made it clear that the nature of such conduct varies with the conditions of the market and the characteristics of the defendant. Thus, we consistently have held that the inference may be drawn from conduct that serves as the basis for a substantial claim of restraint of trade. Several cases, for example, have explicitly equated such conduct with an unreasonable restraint of trade in violation of section 1 of the Sherman Act. CalComp, supra, 613 F.2d at 736; Sherman, supra, 601 F.2d at 453 n.47; Gough, supra, 585 F.2d at 390. Actions taken by a firm without market power may support the inference of intent if those actions are “of a kind clearly threatening to competition or clearly exclusionary.” Janich Bros., supra, 570 F.2d at 854 n.4. Some opinions have taken this language to refer to per se violations of section 1. E. g., CalComp, supra, 613 F.2d at 737 & n.10; Gough, supra, 585 F.2d at 390. On the other hand, direct evidence of intent alone, without corroborating evidence of conduct, cannot sustain a claim of attempted monopolization. See Hunt-Wesson Foods, Inc. v. Ragu Foods, Inc., 627 F.2d 919, 926 (9th Cir. 1980); Blair Foods, Inc. v. Ranchers Cotton Oil, 610 F.2d 665, 669 (9th Cir. 1980); Knutson v. Daily Review, Inc., 548 F.2d 795, 814 (9th Cir. 1976), cert. denied, 433 U.S. 910, 97 S.Ct. 2977, 53 L.Ed.2d 1094 (1977); Chisholm Brothers Farm Equipment Co. v. International Harvester Co., 498 F.2d 1137, 1144-45 (9th Cir.), cert. denied, 419 U.S. 1023, 95 S.Ct. 500, 42 L.Ed.2d 298 (1974); Hallmark Industry v. Reynolds Metals Co., 489 F.2d 8, 12 (9th Cir. 1973), cert. denied, 417 U.S. 932, 94 S.Ct. 2643, 41 L.Ed.2d 235 (1974). The necessity of corroborative conduct rests on the fact that direct evidence of intent alone can be ambiguous and misleading. The law of attempted monopolization must tread a narrow pathway between rules that would inhibit honest competition and those that would allow pernicious but subtle conduct to escape antitrust scrutiny. Direct evidence of intent to vanquish a rival in an honest competitive struggle cannot help to establish an antitrust violation. It also must be shown that the defendant sought victory through unfair or predatory means. Evidence of conduct is thus indispensable. The language and purpose of section 2 reinforces this necessity. While the prohibition of attempts to monopolize clearly encompasses actions that fall short of their intended result, it is equally clear that actual steps toward interference with the competitive process, and not boardroom ruminations, is the evil against which section 2 is directed. As Justice Holmes taught us, there is a difference, even in antitrust law, between preparation and attempt. Swift & Co., supra, 196 U.S. at 402, 25 S.Ct. at 281. 2. Dangerous Probability of Success The third element, dangerous probability of success, like the first, also is rooted in the relationship between the separate offenses of monopolization and attempt to monopolize. See, e. g., Swift & Co., supra, 196 U.S. at 396, 25 S.Ct. at 279. Although this element is generally treated as separate and independent, it can be inferred from evidence indicating the existence of the other two. E.g., CalComp, supra, 613 F.2d at 737. However, the proper significance of this third element “has been controversial, even within this circuit.” Hunt-Wesson Foods, supra, 627 F.2d at 925. Part of the uncertainty results, as already indicated, from the tendency to treat the element of dangerous probability of success and proof of market power as equivalent. Although related, they are not equivalent. Another source of uncertainty, also previously suggested, is that a dangerous probability of success has been treated as evidence of specific intent and vice versa. See Lessig v. Tidewater Oil Co., 327 F.2d 459, 474 (9th Cir.), cert. denied, 377 U.S. 993, 84 S.Ct. 1920, 12 L.Ed.2d 1046 (1964). However, our more recent decisions make plain that the permissibility of inferring dangerous probability from proof of specific intent is conditional. That is, a dangerous probability of success may be inferred either (1) from direct evidence of specific intent plus proof of conduct directed to accomplishing the unlawful design, or (2) from evidence of conduct alone, provided the conduct is also the sort from which specific intent can be inferred. These more recent decisions also establish that the dangerous probability of success requirement is not designed as a means of screening out cases of minimal concern to antitrust policy but is instead a way of gauging more accurately the purpose of a defendant’s actions. Accordingly, the level of the probability of success appropriately may be raised by the defendant, as did Continental in this case, even if the plaintiff has made his case without direct proof of dangerous probability. Thus, if market conditions are such that a course of conduct described by the plaintiff would be unlikely to succeed in monopolizing the market, it is less likely that the defendant actually attempted to monopolize the market. Conversely, a firm with substantial market power may find it more rational to engage in a monopolistic course of conduct than would a smaller firm in a less concentrated market. In sum, the dangerous probability of success element is always relevant in analyzing an attempt claim. The nature of its relevance, however, is a function of the state of the evidence offered in support of the other two elements necessary to proof of the claim. 3. Conduct The conduct element of the attempt claim also is closely related to the other two elements. Thus, the first element, specific intent to control prices or exclude competition, may be inferred from certain types of conduct. The third element, dangerous probability of success, also is often dependent on proof of conduct. Finally, evidence of conduct is indispensable even when there is direct evidence of unlawful specific intent. This interrelationship extends to the type and strength of proof required to establish each element. In the absence of direct and probative evidence of specific intent to monopolize, for example, a plaintiff must introduce evidence of conduct amounting to a substantial claim of restraint of trade or conduct clearly threatening to competition or clearly exclusionary. Direct evidence of intent, on the other hand, may permit reliance on a broader range of conduct, simply because the purpose of ambiguous conduct may be more clearly understood. But, in general, conduct that will support a claim of attempted monopolization must be such that its anticipated benefits were dependent upon its tendency to discipline or eliminate competition and thereby enhance the firm’s long-term ability to reap the benefits of monopoly power. Such conduct is not true competition; it makes sense only because it eliminates competition. It does not enhance the quality or attractiveness of the product, reduce its cost, or alter the demand function that all competitors confront. Its purpose is to create a monopoly by means other than fair competition. We now turn to the specific facts and arguments of the case at bar. B. Predatory Pricing Inglis alleged in support of its section 2 claim that Continental set predatory prices for its private label bread with the purpose of eliminating weaker bread wholesalers. In support Inglis presented to the jury testimony and documentary evidence relating to Continental’s intent as well as expert testimony demonstrating the relationship between Continental’s prices and various categories of costs. In granting Continental’s motion for JNOV, the district court determined that Inglis had failed to present competent evidence of predatory conduct according to legal standards which Inglis now challenges on appeal. In granting Continental’s alternative motion for a new trial, the court found that Inglis’ expert testimony was unreliable, and apparently discounted Inglis’ direct evidence of intent on the ground that such evidence was insufficient as a matter of law, absent evidence of predatory conduct. 461 F.Supp. at 422 n.11. 1. When Is a Price Predatory! Much of the dispute on appeal concerns the proper relationship between direct evidence of intent and evidence concerning the relationship between the cost and price of Continental’s products. As already indicated, a distinction must be maintained between a “predatory” price and a “competitive” one. Constraints must be developed that will deter the former, but not the latter. See Joskow & Klevorick, A Framework for Analyzing Predatory Pricing Policy, 89 Yale L.J. 213, 220-22 (1979). Price reductions that constitute a legitimate, competitive response to market conditions are entirely proper. “Pricing is predatory only where the firm foregoes short-term profits in order to develop a market position such that the firm can later raise prices and recoup lost profits .... ” Janich Bros., supra, 570 F.2d at 856; Hanson v. Shell Oil Co., 541 F.2d 1352, 1358 (9th Cir. 1976), cert. denied, 429 U.S. 1074, 97 S.Ct. 813, 50 L.Ed.2d 792 (1977). Although this standard captures the distinction between competitive and anticompetitive price reductions, many authorities have offered more specific economic tests to define the difference. 2. The Areeda-Turner Test One such economic test that has found favor in this and other circuits was developed by Professors Areeda and Turner. See Areeda & Turner, Predatory Pricing and Related Practices Under Section 2 of the Sherman Act, 88 Harv.L.Rev. 697 (1975). In their view a price should not be considered predatory if it equals or exceeds the marginal cost of producing the product. When a firm prices at marginal cost, they argue, only less efficient firms will suffer larger losses per unit of output at that price. Moreover, such pricing enables resources to be properly allocated because the price accurately “signals” to the consumer the true social cost of the product. Therefore, “pricing at marginal cost is the competitive and socially optimal result.” Id. at 711. In contrast, pricing below marginal cost should be conclusively presumed illegal. Recognizing that business records rarely reflect marginal costs of production, Areeda and Turner suggest the use of average variable cost as an evidentiary surrogate. This test has had its critics to whom the creators have responded and even revised some portions of their theory. Courts that have adopted the Areeda-Turner test have not done so unqualifiedly. We are no exception. Our first discussion of the test appears in Hanson v. Shell Oil Co., 541 F.2d 1352 (9th Cir. 1976), cert. denied, 429 U.S. 1074, 97 S.Ct. 813, 50 L.Ed.2d 792 (1977). There we affirmed a directed verdict for the defendant on a section 2 attempt claim. The plaintiffs failed to introduce direct evidence of specific intent and also were unable to demonstrate that the defendant’s prices were below its marginal or average variable cost. The opinion, however, suggested two departures from the Areeda-Turner rule. First, it was recognized that a defendant may be allowed to prove “nonpredatory and acceptable business reasons” for prices that are below even marginal or average variable cost. Id. at 1359 n.6. Second, the possibility was raised that a plaintiff could make a case of predatory pricing if defendant’s prices, although above marginal or average variable cost, were below its short-run profit-maximizing price and if barriers to entry were great enough to prevent entry long enough to permit the predator to reap the benefits of its enhanced market position. Id. at 1358 n.5. The Areeda-Turner test also was discussed in Janich Bros., Inc. v. American Distilling Co., 570 F.2d 848 (9th Cir. 1977), cert. denied, 439 U.S. 829, 99 S.Ct. 103, 58 L.Ed.2d 122 (1978). There we again affirmed the district court’s directed verdict in favor of the defendant. We stated that “an across-the-board price set at or above marginal cost should not ordinarily form the basis for an antitrust violation,” id. at 857 (emphasis added), and that, on the facts of that case, the plaintiff’s failure to present evidence of price below average variable cost justified a directed verdict, id. at 857-58. As in Hanson, no direct evidence of intent was admitted at trial. Id. at 859. Finally, we again upheld a directed verdict in favor of the defendant in California Computer Products, Inc. v. IBM Corp., 613 F.2d 727 (9th Cir. 1979). Our view of the evidence was that the defendant’s price reductions were still “substantially profitable” and were made in response to lower-priced competition. The plaintiff had failed to prove pricing below marginal or average variable cost, “which ordinarily is required to show predatory pricing.” Id. at 740 n.19. However, we also recognized “that refinement of the marginal or average variable cost test will be necessary as future cases arise.” Id. at 743. For instance, limit pricing by a monopolist might, on a record which presented the issue, be held an impermissible predatory practice .... And we do not foreclose the possibility that a monopolist who reduces prices to some point above marginal or average variable costs might still be held to have engaged in a predatory act because of other aspects of its conduct. Id. (citations omitted) 3. This Court’s Approach to Establishing Predation Thus, although we have approved the use of marginal or average variable cost statistics in proving predation, “we have not held that mode of proof to be exclusive.” Arizona v. Maricopa County Medical Society, 643 F.2d 553, 559 n.6 (9th Cir. 1980), cert. granted, 450 U.S. 979, 101 S.Ct. 1512, 67 L.Ed.2d 813 (1981). While Hanson, Janich, and CalComp each upheld directed verdicts when the plaintiffs failed to introduce evidence of prices below marginal or average variable cost, those holdings cannot be divorced from the factual contexts in which they arose. Our approach to proof of intent through use of conduct is to focus on what a rational firm would have expected its prices to accomplish. As Hanson and Janich suggest, a price should be considered predatory if its anticipated benefits depended on its tendency to eliminate competition. If the justification for a price reduction did not depend upon this anticipated effect, then it does not support a claim of attempted monopolization, even if it had the actual effect of taking sales from competitors. We emphasize a defendant’s rational expectations to avoid penalizing innocent miscalculations that result in anticipated profits being turned into losses, with damaging effects on competitors. Our focus does not require that plaintiffs in all cases come forward with evidence of the defendant’s subjective state of mind. Predatory pricing may be proved by examining the relationship between the defendant’s prices and costs. But such proof must tend to show that the anticipated benefits of the prices, at the time they were set, depended on their anticipated destructive effect upon competition and the consequent enhanced market position of the defendant. In this case Continental has conceded that some of the prices challenged by Inglis were below average total cost. Taken alone this does not brand Continental’s prices as predatory. Pricing below average total cost may be a legitimate means of minimizing losses, particularly when the firm is “temporarily” experiencing “excess capacity” in its productive facilities. When this is the case, the firm’s average variable cost — the sum of those costs that vary with output divided by the total units of output — generally will be less than the firm’s marginal cost — the variable cost associated with producing the last unit of output. Prices below the average total cost of production, but above the average variable cost, may represent a legitimate means of minimizing losses during the period of inadequate demand. Such a price will be sufficient to recover the variable costs of production and at least some portion of the firm’s fixed costs — those costs that would remain even if the firm ceased production. To discontinue production under these circumstances would increase losses because even that portion of its total fixed costs would be lost. Pricing at this level, however, will not be rational over the long term because it will not justify renewal of investment at the previous level. If demand does not increase and thus justify a greater price at the point at which investment must be renewed, discontinuance of production, in whole or in part, is required. Although pricing below average total cost in the short term may be legitimate, it is less likely that pricing below average variable cost will be. Such pricing, if sustained, will not permit the recovery of any portion of the firm’s fixed costs. In addition, the firm, because it cannot recover all its variable costs, has out-of-pocket losses on each unit it sells. The economic case for discontinuance of production is strong. Although pricing below average total cost and above average variable cost is not inherently predatory, it does not follow, however, that such prices are never predatory. Predation exists when the justification of these prices is based, not on their effectiveness in minimizing losses, but on their tendency to eliminate rivals and create a market structure enabling the seller to recoup his losses. This is the ultimate standard, and not rigid adherence to a particular cost-based rule, that must govern our analysis of alleged predatory pricing. Guided by these principles, we hold that to establish predatory pricing a plaintiff must prove that the anticipated benefits of defendant’s price depended on its tendency to discipline or eliminate competition and thereby enhance the firm’s long-term ability to reap the benefits of monopoly power. If the defendant’s prices were below average total cost but above average variable cost, the plaintiff bears the burden of showing defendant’s pricing was predatory. If, however, the plaintiff proves that the defendant’s prices .were below average variable cost, the plaintiff has established a prima facie case of predatory pricing and the burden shifts to the defendant to prove that the prices were justified without regard to any anticipated destructive effect they might have on competitors. These holdings are not inconsistent with the results reached in Hanson, Janich and GalComp. Hanson and Janich, as already noted, upheld directed verdicts in favor of defendants because the plaintiffs had failed to introduce evidence of prices below marginal or average variable cost. However, in neither case was the plaintiff able to prove that the defendant had sacrificed greater profits or incurred greater losses than necessary, in order to eliminate the plaintiff. In CalComp the plaintiff introduced neither direct evidence of predatory intent nor evidence of prices below average variable cost or average total cost. In this case the district court granted Continental’s motion for JNOY because it concluded that Inglis’ evidence concerning Continental’s cost-price relationship was legally insufficient to establish predatory pricing. Although Inglis’ expert testified that during the period examined in his study Continental’s prices were below average variable cost, the district court concluded that because excess capacity existed in the relevant market, only proof of prices below marginal cost could establish predatory pricing. The district court erred. The jury reasonably could have concluded that prices below average variable cost were predatory. Even when excess capacity exists, pricing below average variable cost, to repeat, is sufficiently questionable to support the inference that the prices were designed to eliminate competition. 4. How to Determine Which Costs Are Fixed and Which Are Variable We agree with the district court, however, that a new trial is in order. Inglis’ principal evidence concerning the relationship between Continental’s cost and prices consisted of average variable cost computations by its expert. These computations were based entirely on directions by Inglis’ counsel as to which costs were to be considered variable and which fixed, directions which were derived from categories of cost mentioned in Janich Bros., Inc. v. American Distilling Co., supra, 570 F.2d at 858 & n.11. Inglis maintains that Janich allocated categories of cost as a matter of law and refers us to the recommendation of Professors Areeda and Turner that such categories be fixed to avoid case-by-case dispute. We disagree with Inglis’ interpretation of Janich and we reject the proposal of Professors Areeda and Turner. Janich did not purport to describe categories of fixed and variable costs that were to apply in all predatory pricing cases. Rather, the cost items were listed as examples of costs that typically are variable and fixed. They should be read as illustrative and not prescriptive. To do otherwise would be to forget that the use of costprice comparisons is not an end in itself but a means of interpreting the likely justification for particular pricing decisions. It is true, of course, that the fixed production costs of a firm are those costs that do not vary with output and that would remain even if the firm discontinued production. Likewise, variable costs, as the term suggests, are those costs that do vary with output and that the firm is likely to be most concerned with when contemplating a change in price and consequent changes in output. However, to determine whether particular costs are variable, one must evaluate the relationship of the prospective change in output to that level of output which presently exists. For example, some production decisions of great magnitude may entail the substantial retirement or expansion, as the case may be, of productive capacity, in which case costs typically considered fixed become variable. At the other extreme, small expansions of output may entail no change in costs, such as labor or transportation, that are considered typically variable. In predatory pricing cases the relevant changes in output will be those attributable to the price reduction alleged to be predatory. The reduction of price in such a case no doubt will have resulted in an expansion of output to “clear the market,” i. e., to satisfy the increased demand generated by the price reduction. It is those costs that change as a result of the expanded output that appropriately are considered to be variable. Thus, the first step in determining average variable cost in a predatory pricing case will generally be to compare the costs of production before and after the price reduction. The variable costs would then be those expenses that increased as a result of the output expansion attributable to the price reduction. If the new price is below the average of these costs per unit of output there is good reason, as we have said, to infer that the price reduction was predatory. Price maintenance also may be predatory if other alternatives, such as higher prices and reduced output or terminating production altogether, would have been better means of minimizing losses. If these alternatives would have been considered by the firm as commercially reasonable, it would then be appropriate to consider them in identifying variable costs. Under such circumstances the variable cost items would be those that would have been reduced had the firm adopted another combination of output and price more effective in minimizing losses. These costs should be considered variable because they are costs that the firm had within its control at the time it established the price alleged to be predatory. Thus, an available but abjured course of conduct that would have reduced the amount of specifically identifiable unrecovered costs earmarks such cost items as variable. It is the average of these costs incurred per unit of output as a result of the course of action actually pursued by the producer that must be compared with the price of each unit. It follows that the determination of which costs are variable and which fixed will vary with the facts of each case. Moreover, cost categories are solely for the purpose of providing aid in answering the ultimate question: Did the justification for the defendant’s price depend upon its anticipated destructive effect on competition or was the price justified as a reasonably calculated means of maximizing profits, minimizing losses, or achieving some other legitimate end? Accordingly, we hold that the determination of fixed and variable costs is a matter for the jury under appropriate instructions. In this case a new trial is warranted because the evidence upon which Inglis depended to prove predatory conduct was not based upon a calculation of which costs were fixed and which variable that was rooted in the particular facts of this case. We believe this disposition is also mandated by our recent decision in Pierce Packing Co. v. John Morrell & Co., 633 F.2d 1362 (9th Cir. 1980). C. Other Evidence of Predation Inglis argues, however, that there was sufficient evidence of predatory conduct, apart from proof of prices below average variable cost, to support the jury’s verdict. Inglis did introduce evidence of prices below average total cost. We have already pointed out that, unless the plaintiff can prove that the prices were below average variable cost, it bears the burden of demonstrating that the anticipated benefits of defendant’s pricing were dependent upon its tendency to discipline or eliminate competition and thereby enhance the defendant’s ability to reap the benefits of monopoly power in the aftermath. ,We note further that in examining whether this burden has been discharged, the courts must consider the evidence as a whole, giving plaintiffs “the full benefit of their proof without tightly compartmentalizing the various factual components and wiping the slate clean after scrutiny of each.” Continental Ore Co. v. Union Carbide & Carbon Corp., 370 U.S. 690, 699, 82 S.Ct. 1404, 1410, 8 L.Ed.2d 777 (1962). See California Computer Products, Inc. v. IBM Corp., supra, 613 F.2d at 746; Sherman v. British Leyland Motors, Ltd., 601 F.2d 429, 450-51 & n.41 (9th Cir. 1979); Lessig v. Tidewater Oil Co., 327 F.2d 459, 466 (9th Cir. 1964). Following this approach we cannot say that the district judge abused his discretion in ordering a new trial. Inglis emphasizes the direct evidence of allegedly predatory intent that it introduced at trial. It primarily relies on a report prepared by McKinsey & Co., a management consulting firm commissioned by Continental to study the nationwide impact of private label sales on Continental’s business and to recommend strategies of meeting the competition. This report suggested three principal courses of action and then listed a dozen other issues that might be the subject of further study. One of these alternatives was to “maintain price to hasten wholesaler exit pace.” There is no direct evidence in the record that any further action was taken on this proposal or that Continental ever considered or adopted it as a course of action. The only available evidence indicates that the idea was “discussed” at meetings between McKinsey and Continental personnel in advance of the report’s preparation, and nothing specifically suggests that it received any more interest than other alternatives. Finally, the proposal itself is consistent with the interpretation that it involved nothing more than reducing costs and profit margins to put pressure on less efficient bakeries. Nothing in the McKinsey report explicitly suggested that Continental incur losses in an attempt to eliminate competitors. Reasonably interpreted, it amounts to no more than a recommendation of intensified price competition. The other evidence emphasized by Inglis adds little that would suggest unlawful intent. Inglis introduced semiannual sales reports prepared by Continental account executives in northern California which listed as target goals the acquisition of private label accounts currently held by Inglis. These reports add little to the evidence that Continental did in fact make competing offers to retailers served by Inglis. Equally inconclusive is documentary evidence that Continental personnel in northern California sought “market dominance.” Considered as a whole, then, Inglis’s evidence of intent is inconclusive. The district court did not abuse its discretion in ordering a new trial. III. THE ROBINSON-PATMAN ACT CLAIM Appellant’s Robinson-Patman Act claim was based substantially on the same facts underlying the attempt to monopolize claim. Inglis contended that price differentials between Continental’s private label and advertised bread constituted price discrimination in violation of section 2(a) of the Clayton Act, as amended by the Robinson-Patman Act. Inglis also identified (1) discrimination in the price of private label bread between California and Nevada and (2) discrimination in the price of advertised bread between California and Nevada. The district court entered JNOV and, in the alternative, a new trial, in favor of Continental because it found that Inglis had failed to prove (1) the requisite effect on competition, and (2) causation. A. Competitive Injury Because it is undisputed that throughout the complaint period Continental’s advertised label prices were higher than its private label prices, “price discrimination,” within the meaning of the statute, exists. Nothing more than a difference in price between commodities of like grade and quality is necessary to establish such discrimination. FTC v. Anheuser-Busch, Inc., 363 U.S. 536, 549-60, 80 S.Ct. 1267, 1274, 4 L.Ed.2d 1385 (1960). However, section 2(a) does not prohibit mere price discrimination. To be proscribed it must be shown that its effect “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 of either of them.” Inglis must prove that Continental’s price discrimination produced a requisite effect on competition. Janich Bros., Inc. v. American Distilling Co., 570 F.2d 848, 855 & n.6 (9th Cir. 1977), cert. denied, 439 U.S. 829, 99 S.Ct. 103, 58 L.Ed.2d 122 (1978); Texas Gulf Sulphur Co. v. J. R. Simplot Co., 418 F.2d 793, 806 (9th Cir. 1969); Balian Ice Cream Co. v. Arden Farms Co., 231 F.2d 356, 367, 368 (9th Cir. 1955), cert. denied, 350 U.S. 991, 76 S.Ct. 545, 100 L.Ed. 856 (1956). 1. Primary-Line Competition Section 2(a) is concerned with the protection of competition on several levels. Inglis’ claim alleges adverse effects on so-called primary-line competition, i. e., competition with the seller who has charged discriminatory prices. A typical primary-line case involves a national manufacturer and distributor of products who lowers its price in a local market for the purpose of disciplining or eliminating local competitors, while maintaining higher prices in other, less competitive markets. As the Supreme Court has noted, price discrimination of this sort was the impetus for the original enactment of section 2(a) of the Clayton Act in 1914. FTC v. Anheuser-Busch, supra, 363 U.S. at 543, 80 S.Ct. at 1271. The Robinson-Patman Act amendments, although principally concerned with so-called secondary-line competition between customers of the price discriminator, maintained the Clayton Act’s focus on primary-line competition. Id. at 544, 546, 80 S.Ct. at 1271, 1272. Courts consistently have held that the requisite effect on primary-line competition may be inferred from proof of predatory intent on the part of the price discriminator. Utah Pie v. Continental Baking Co., 386 U.S. 685, 696-98, 87 S.Ct. 1326, 1332, 18 L.Ed.2d 406 (1967); International Air Industries, 517 F.2d 714, 722 (5th Cir. 1975), cert. denied, 424 U.S. 943, 96 S.Ct. 1411, 47 L.Ed.2d 349 (1976); Continental Baking Co. v. Old Homestead Bread Co., 476 F.2d 97, 104 (10th Cir.), cert. denied, 414 U.S. 975, 94 S.Ct. 290, 38 L.Ed.2d 218 (1973); Cornwell Quality Tools Co. v. C. T. S. Co., 446 F.2d 825, 831 (9th Cir. 1971), cert. denied, 404 U.S. 1049, 92 S.Ct. 715, 30 L.Ed.2d 740 (1972); Balian Ice Cream, supra, 231 F.2d at 369. And, as is true in suits alleging attempted monopolization under the Sherman Act, predatory intent may be inferred from proof of predatory conduct. Utah Pie, supra, 386 U.S. at 696 n.12, 87 S.Ct. at 1332, n.12; International Air Industries, supra, 517 F.2d at 722-23; Cornwell Quality Tools, supra, 446 F.2d at 831. 2. Proof of Predatory Intent in Primary-Line Cases In this case we confront the issue whether the principles set forth above in connection with proof of predatory intent in an attempt to monopolize claim pursuant to section 2 of the Sherman Act are equally applicable to proof of predatory intent in a primary-line Robinson-Patman Act suit. We hold that they are. We have previously recognized that where “a price differential threatens a primary line injury, .. . section 2 of the Sherman Act . . . and section 2(a) of the Clayton Act .. . are directed at the same economic evil and have the same substantive content.” Janich Bros., supra, 570 F.2d at 855. Accord, Pacific Engineering & Production Co. v. Kerr-McGee Corp., 551 F.2d 790, 798 (10th Cir.), cert. denied, 434 U.S. 879, 98 S.Ct. 234, 54 L.Ed.2d 160 (1977); International Air Industries, supra, 517 F.2d at 720 n.10. There exists no reason to establish principles for primary-line price discrimination cases different from those we recognized in attempt to monopolize cases. It follows, therefore, that we disagree with the district court’s holding that in its primary-line Robinson-Patman Act claim Inglis was required to prove that Continental’s prices were below its marginal cost of production. A plaintiff may establish the required effects on competition in a primary-line case even though the defendant’s prices were shown to be above marginal cost. However, unless the plaintiff proves that the prices were below the defendant’s average variable cost, the plaintiff bears the burden of establishing that the anticipated benefits of the prices depended on their anticipated destructive effect on competition. If the plaintiff does prove pricing below average variable cost, the burden shifts to the defendant to establish a legitimate business justification for its conduct. Accordingly, we must reverse the district court’s entry of JNOV. However, no abuse of its discretion occurred when the trial court ordered a new trial on the Robinson-Patman Act claim. Although Inglis did establish that Continental’s prices were in many instances below average total cost, it failed to establish by sufficient evidence either that Continental’s prices were below average variable cost or that the benefits of Continental’s prices depended on their anticipated tendency to eliminate competition. Finally, as we explained earlier, these deficiencies are not remedied by the direct evidence of intent offered by Inglis at trial. 3. Sherman and Robinson-Patman Compared An injury to competition proscribed by section 2(a) of the Robinson-Pat-man Act perhaps may be established without proof of predatory intent or predatory pricing, however. Therefore, we do not hold that there exists a complete substantive synchronization of the Sherman and Robinson-Patman Acts. However, no basis for establishing the requisite competitive injury, other than by proof of predatory intent and predatory pricing, is evident in the record before us, nor has Inglis argued one. Certainly, the mere fact that Inglis suffered losses and eventually ceased operations is not sufficient to establish a section 2(a) Robinson-Patman Act violation. In primary-line Robinson-Patman Act cases, such as is this one, the distinction between vigorous, but honest, price competition and predatory assaults on the competitive process is just as important as it is to Sherman Act cases brought under its section 2. Under-these circumstances the analytical standards should be no different. Two other differences between the substantive requirements of the Sherman and Robinson-Patman Acts should be mentioned,' although neither alters the result of the case. First, the offense of attempted monopolization requires proof of a dangerous probability of success while section 2(a) of the Robinson-Patman Act requires only a showing that the price discrimination “may” substantially lessen competition. In this case, however, the distinction is of little significance. Without addressing the dispute about whether section 2(a) requires merely a “possibility” or a more substantial “probability” of competitive injury, we agree with the Seventh Circuit that [i]f [the defendant] is using its competitive power fairly in the market place and respecting the rights of its competitors, then no forecast of future adverse effects on competition based on those facts is valid. If, on the other hand, the projection is based on predatoriness or buccaneering, it can reasonably be forecast that an adverse effect on competition may occur. Anheuser-Busch, Inc. v. FTC, 289 F.2d 835, 843 (7th Cir. 1961) (emphasis' in original). Thus, while we recognize that the language of the Robinson-Patman Act may afford somewhat more latitude than that of the Sherman Act, the difference in this case provides no basis for sustaining the jury’s verdict. The second difference lies in the scope of the competitive injury envisioned by the two statutes. While section 2 of the Sherman Act requires attempted monopolization of a “part” of commerce, and thus is concerned with competitive conditions generally in the line of commerce affected, section 2(a) of Robinson-Patman requires only an impermissible effect on competition by others with the seller employing discriminatory prices. Once again, this difference does not alter our result here. Under either statute Inglis has failed to present sufficient evidence to sustain the jury’s verdict. B. Causation The district court based its entry of judgment against Inglis on its RobinsonPatman Act claim on a second and independent ground. It held that Inglis was required to show that Continental subsidized the lower prices of its private label bread with the higher prices of its advertised bread. Such proof was necessary to establish a causal relationship between the price discrimination (and not merely the lower of the two prices) and any injury Inglis might have suffered. Finding no direct proof of subsidization, the district court held that Inglis could not draw the inference of subsidy withotit proof that Continental set prices below marginal cost. We do not decide whether a showing of subsidization is necessary in a primary-line case, for we hold that even if such a requirement existed, the district court erred in refusing to permit the inference of subsidization except when the plaintiff can prove that the lower of the prices was below marginal cost. Our earlier discussion demonstrates that the marginal cost test does not capture the full range of economically questionable pricing activity, and that weakness condemns it in this context as well. The district court, therefore, erred in entering JNOV for Continental. Continental has urged two additional grounds, which the district court either rejected or did not expressly address, for upholding the entry of JNOV in its favor. We turn now to those issues. C. “In Commerce” Requirements Section 2(a) of the Robinson-Patman Act imposes three jurisdictional requirements relating to interstate commerce. First, the alleged price discriminator must be “engaged in commerce.” Second, the unlawful price discrimination must occur “in the course of such commerce.” And third, “either or any of the purchases involved in such discrimination” must be “in commerce.” These requirements delimit “both the universe of ‘persons’ who are subject to the Act and the type of transactions that can constitute a violation thereof.” S & M Materials Co. v. Southern Stone Co., 612 F.2d 198, 200 (5th Cir.), cert. denied, 449 U.S. 832, 101 S.Ct. 101, 66 L.Ed.2d 37 (1980). Only the third of these requirements is disputed in this case. We begin by noting that the jurisdictional reach of the Sherman Act is different from that of the Robinson-Patman Act. In enacting the Sherman Act Congress “wanted to go to the utmost extent of its Constitutional power in restraining trust and monopoly agreements.” United States v. South-Eastern Underwriters Ass’n, 322 U.S. 533, 558, 64 S.Ct. 1162, 1176, 88 L.Ed. 1440 (1944). Accordingly, the Sherman Act consistently has been construed to reach anticompetitive activities that affect commerce. E. g., Mandeville Island Farms v. American Crystal Sugar Co., 334 U.S. 219, 234, 68 S.Ct. 996, 1005, 92 L.Ed. 1328 (1948). In contrast, the Supreme Court has held that merely showing adverse effects on commerce will not satisfy the jurisdictional requirements of the Robinson-Patman Act. Gulf Oil Corp. v. Copp Paving Co., 419 U.S. 186, 95 S.Ct. 392, 42 L.Ed.2d 378 (1974). Instead, the third requirement’s language, “where either or any of the purchases involved in such discrimination are in commerce,” means that section 2(a) applies only where “ ‘ “at least one of the two transactions which, when compared, generate a discrimination . . . cross[es] a state line.” ’ ” Id. at 200, 95 S.Ct. at 401 (quoting Hiram Walker, Inc. v. A&S Tropical, Inc., 407 F.2d 4, 9 (5th Cir.), cert. denied, 396 U.S. 901, 90 S.Ct. 212, 24 L.Ed.2d 177 (1969) (quoting F. Rowe, Price Discrimination Under the RobinsonPatman Act 79 (1962)). It is immaterial which one crossed a state line. As long as one or both did so, the jurisdictional requirement is satisfied. See, e. g., Moore v. Mead’s Fine Bread Co., 348 U.S. 115, 120, 75 S.Ct. 148, 150, 99 L.Ed. 145 (1954). Continental argues that Inglis has failed to satisfy this jurisdictional requirement. Although the district court found that Continental’s interstate transactions constituted a small percentage of its total sales in the region, these transactions were sufficient “to bring the whole panoply of alleged intrastate price discriminations within the subject matter jurisdiction requirements of Robinson-Patman § 2(a).” 461 F.Supp. at 420-21. The great bulk of both advertised and private label bread sales from Continental’s three northern California plants occurred within California; nevertheless, it did sell both private label and advertised bread to accounts in Nevada. Moreover, the prices for those products generally were higher than the prices for their California counterparts, at least until 1973. In addition, the central price disparity identified by Inglis — that between the advertised and private label bread — was reflected in the interstate sales. That is, Continental’s advertised prices in Nevada were nearly always higher than private label prices in California, and its advertised prices in California were higher than its private label prices in Nevada. Denying none of these facts, Continental argues that the interstate sales involved de minimis amounts and thus will not support Robinson-Patman Act jurisdiction. We reject Continental’s position. We cannot ignore that Continental sold its bread in an interstate market area and that some of the sales which demonstrated a price discrimination occurred across state lines. Nothing in the language or legislative history of the Robinson-Patman Act indicates that some prescribed quantum of interstate transactions is necessary for jurisdiction to attach. Since the price disparity between advertised and private label bread, about which Inglis complains, was represented by some interstate sales, we affirm the district court’s holding that it had jurisdiction to examine that same price disparity as it existed with respect to sales of the same products within California. See Continental Baking Co. v. Old Homestead Bread Co., 476 F.2d 97, 109 (10th Cir.), cert. denied, 414 U.S. 975, 94 S.Ct. 290, 38 L.Ed.2d 2