Predatory Pricing: Factors Affecting Error Costs and Definition, Study notes of Law

Download Predatory Pricing: Factors Affecting Error Costs and Definition and more Study notes Law in PDF only on Docsity! The Yale Law Journal Volume 89, Number 2, December 1979 A Framework for Analyzing Predatory Pricing Policy* Paul L. Joskowt and Alvin K. Klevorickt Recent literature on the law and economics of antitrust has devoted increasing attention to the issue of "predatory pricing"-a dominant firm's use of price to restrict competition by driving out existing rivals or excluding potential ones. A number of scholars-including Areeda and Turner, Baumol, Bork, Posner, Scherer, and William- son-have contributed to this discussion,' and each has taken a different * A large number of people have stimulated our thinking on this subject. We wouldlike to acknowledge, in particular, the helpful comments Bruce Ackerman, Richard Levin, Richard Schmalensee, Jeff Strnad, and Oliver Williamson made on earlier versions of this paper and the helpful research assistance of Jeff Strnad. These people, of course, bear no responsibility for any faults that may remain. t Professor of Economics, Massachusetts Institute of Technology and Visiting Professor,John F. Kennedy School of Government, Harvard University. ++ Professor of Law and Economics, Yale University. 1. The current discussion began with proposals made by Professors Phillip Areeda and Donald Turner in the mid-1970s. See Areeda & Turner, Predatory Pricing and Related Practices under Section 2 of the Sherman Act, 88 HARv. L. REv. 697 (1975) [hereinafter cited as Areeda-Turner Proposal]. A year later, Professor Frederick Scherer commented on the Areeda-Turner contribution and made proposals of his own. See Scherer, PredatoryPricing and the Sherman Act: A Comment, 89 HARV. L. REv. 869 (1976) [hereinafter cited as Scherer Proposal]. A debate between Areeda and Turner and Scherer ensued. See Areeda & Turner, Scherer on Predatory Pricing: A Reply, 89 HARv. L. REv. 891 (1976) [hereinafter cited as Areeda ir Turner on Scherer]; Scherer, Some Last Words on Predatory Pricing, 89 HARv. L. REv. 901 (1976) [hereinafter cited as Scherer Response]. A year later, Professor Oliver Williamson introduced an analysis that led him to a new substantive proposal. See Williamson, Predatory Pricing: A Strategic and Welfare Analysis, 87 YALE L.J. 284 (1977) [hereinafter cited as Williamson Proposal]. Williamson also engaged in adebate with Areeda and Turner. See, e.g., Areeda & Turner, Williamson on Predatory Pricing, 87 YALE L.J. 1337 (1978) [hereinafter cited as Areeda & Turner on Williamson]; Williamson, Williamson on Predatory Pricing II, 88 YALE L.J. 1183 (1979) [hereinaftercited as Williamson Response]. Most recently, Professor William Baumol has proposed another approach to predatory pricing. See Baumol, Quasi-Permanence of Price Reduc- tions: A Policy For Prevention of Predatory Pricing, 89 YALE L.J. 1 (1979). Aside from these four proposals, several other positions are evident in the literature. 213 The Yale Law Journal Vol. 89: 213, 1979 approach to the predatory pricing problem. A variety of different "rules" have been suggested, and these suggestions have played key roles in the decision of recent Sherman Act and Federal Trade Com- mission Act cases.2 Although each author discusses the approaches of writers who have preceded him and although the opinions in the cases have compared the various approaches, no unified structure has been provided for evaluating the alternative approaches and choosing among them. It is not always clear from the discussions in the respective contri- butions exactly why the several authors have been led to different conclusions. In fact a variety of factors seem to account for the di- vergent positions at which they arrive. First, the various commenta- tors rely on alternative theoretical models. 3 Some of them use a static framework 4 to examine firm behavior, while others concentrate on See, e.g., R. BORK, THE ANTITRUST PARADOX 149-55 (1978) (suggesting predatory price cutting is unlikely to exist and any rule prohibiting it is likely to harm consumers more than the absence of any legal sanction); R. POSNER, ANTITRUST LAW: AN ECONOMIC PERSPEC- TIVE 188, 191 (1976) (suggesting that predatory pricing is most usefully defined as "pricing at a level calculated to exclude from the market an equally or more efficient competitor" and suggesting as a safeguard that plaintiff be required to show relevant market has characteristics-many purchasers, or defendant operating in more markets than competi- tors or potential entrants-predisposing it to effective predatory pricing) [hereinafter cited as POSNER BOOK]. Finally, attention has been given to predatory pricing in general works on antitrust law and policy. See, e.g., 3 P. AREEDA & D. TURNER, ANTITRUST LAW 711722 (1978) [hereinafter cited as AREEDA-TURNER TREATISE]; Posner, The Chicago School of Antitrust Analysis, 127 U. PA. L. REv. 925, 939-44 (1979) [hereinafter cited as Posner Article). 2. See, e.g., Hanson v. Shell Oil Co., 541 F.2d 1352, 1358-59 (9th Cir. 1976), cert. denied, 429 U.S. 1074 (1977) (applying Areeda-Turner rules in Sherman Act § 2 case); Interna- tional Air Indus., Inc. v. American Excelsior Co., 517 F.2d 714, 723-26 (5th Cir. 1975), cert. denied, 424 U.S. 943 (1976) (applying Areeda-Turner rules in predatory pricing case brought under Robinson-Patman Act); Borden, Inc., [1978] TRADE REG. REP. (CCH) 21,490 (FTC cease and desist order with multiple opinions discussing and applying various rules from predatory pricing literature). 3. Professor Richard Schmalensee has pointed out that often a multitude of economic models can provide potential frameworks for analysis of any particular issue in an anti- trust case. See Schmalensee, On the Use of Economic Models in Antitrust: The ReaLemon Case, 127 U. PA. L. REv. 994, 995-97 (1979). He notes that choosing among models may be difficult: Because the literature contains a host of internally coherent models with different assumptions and implications, and because the methods of economic analysis can be used to construct new models, analysis of any particular industry or behavior pattern may pose difficult problems of model selection. Often such problems can only be dealt with satisfactorily by creative theoretical analysis, along with careful organiza. tion and evaluation of available evidence. Id. at 996. Professor Schmalensee illustrates the importance of model selection by examin- ing how some economic issues were treated in the Federal Trade Commission's Initial Decision of Borden, Inc., [1976] 3 TRADE REG. REP. (CCH) 21,194, modified, [1978] TRADE REG. REP. (CCH) 21,490. See Schmalensee, supra, at 998-1043. 4. See R. BORK, supra note 1, at 149-52; Areeda-Turner Proposal, supra note 1, at 703-20. Static analysis examines the behavior of economic agents at a single moment. 214 Predatory Pricing conduct for economic efficiency and other goals of antitrust law.14 A related factor that may lead to differing conclusions, and the last one we will mention here, is the degree of concern about the liti- giousness of market participants. Some commentators are very con- cerned about the use of private treble damage suits against alleged predators when such suits are of dubious merit. Others, while worried about such suits, are also concerned about how dominant firms will re- spond to particular predatory pricing rules in an attempt to avoid litigation.' 5 This list of possible reasons for different policy recommendations with regard to predatory pricing is not exhaustive. It is clear that a large number of diverse factors play a part in the presentation of the several policy conclusions, and thus it seems important to de- velop a comprehensive framework for comparing and evaluating al- ternative approaches to the predatory pricing problem. Each of the factors we have just noted as relevant is incorporated in the decision- theoretic' 6 framework developed here. The analytical structure we suggest will make clearer exactly how these various factors and judg- ments enter into the formulation of a policy toward predatory pricing. Furthermore, it will be apparent as we proceed that this structure also provides a framework for evaluating alternative rules in other areas of antitrust. In designing a policy toward predatory pricing, one is confronted with the difficult task of inferring long-run market outcomes from observable short-run behavior and short-run market conditions. Any such inference entails uncertainty and hence the possibility of error; an assessment of long-run considerations is necessarily "speculative and indeterminate."' 17 Nevertheless, such an assessment is required be- cause the essence of predatory pricing is the alleged predator's sacrifice of short-run gains for greater long-run gains. Consider a particular market, and suppose that we are trying to 14. See Borden, Inc., [1978] TRADE REG. REP. (CCH) 21,490, at 21,518, 21,523 (Pitof- sky, Comm'r, concurring). 15. See Scherer Proposal, supra note 1, at 883 (Areeda-Turner rules will encourage dominant firms to hold costly excess capacity); Williamson Proposal, supra note 1, at 295-302 (describing responses of dominant firm to various rules). 16. Decision theory is concerned with making optimal decisions when various aspects of the world are uncertain. See, e.g., M. DEGROOT, OPTIMAL STATISTICAL DECISIONS (1970) (advanced treatment); H. RAIFFA, DECISION ANALYSIS (1968) (elementary treatment). The specific application of decision theory in this article is presented at pp. 218-19 infra. 17. Areeda &Y Turner on Scherer, supra note 1, at 897. At least one court has agreed with Areeda & Turner and has chosen to ignore long-run considerations in light of the "limitations of the judicial process." Janich Bros. v. American Distilling Co., 570 F.2d 848, 857 n.9 (9th Cir. 1977), cert. denied, 439 U.S. 829 (1978). 217 The Yale Law Journal select a rule that takes long-run considerations into account and that can be applied if and when a charge of predatory pricing is leveled against a firm in this market. The evaluative structure we develop indicates that for each proposed predatory pricing rule one should consider, for this market, the probability of each kind of error"' the rule can lead to, the cost of each kind of error, and the cost of implementing the policy. Our decision-theoretic framework directs that we choose the policy that would minimize the sum of the expected costs of error and the costs of implementation that would result if the policy were ap- plied to the market we are considering. If all markets were identical in their structural and behavioral aspects, then having found the op- timal predatory pricing rule for one market, we could apply it with confidence to all others. But, as one might expect, different markets are not identical with respect to the features that determine the sum of the expected error costs and the costs of implementation for alter- native rules. Hence, our decision-theoretic evaluative mechanism re- veals that no single rule will be best for all market situations; if a predatory pricing rule is formulated with one particular market in mind, we cannot be sure that it should be applied to other market situations. What is needed is an approach that can accommodate important market differences: the characteristics of firms and markets that af- fect the probabilities of error, the error costs, and the implementation costs of alternative policy approaches. We will demonstrate how con- sideration of the links between certain firm and market characteris- tics, on the one hand, and the probabilities of error, error costs, and rule-implementation costs, on the other, can be used to develop a two-tiered "structuralist" rule-of-reason approach to be applied in cases of alleged predatory pricing behavior. To avoid misunderstanding, it should be emphasized that we are using a decision-theoretic framework as an analytical device for eval- uating alternative standards or approaches, under the assumption that each standard or approach evaluated would be applied uniform- ly to all cases. We are not suggesting that a judge or a jury should go through our evaluative calculus in deciding a particular case. Rather, our examination of alternative policies, and our use of the decision- theoretic framework to evaluate their respective expected performance, leads us to suggest a particular approach to predatory pricing that 18. See pp. 223-24 infra (distinguishing the types of errors a rule can generate). 218 Vol. 89: 213, 1979 Predatory Pricing most appropriately accounts for uncertainty and the costs of making incorrect decisions, ex ante. The two-stage rule-of-reason approach we suggest would then be applied uniformly on a case-by-case basis.' 9 I. Identification of the Predatory Pricing Problem Although there is a substantial and growing literature and case law on predatory pricing, no definition of predatory pricing seems to have commanded the assent of all concerned. 20 It is fair to say, however, that "predatory pricing" is generally used to describe the adoption of a pricing policy that somehow restricts competition by driving out existing rivals or by excluding potential rivals from the market.21 The differences in usage arise when this general idea is given specific content. A. Definition of Predatory Pricing We shall use the following rather simple definition: Predatory pricing behavior involves a reduction of price in the short run so as to drive competing firms out of the market or to discourage entry of new firms in an effort to gain larger profits via higher prices in 19. We are taking a "meta-rule-of-reason" approach to the evaluation of alternative per se rules for predatory pricing cases. Our examination of the expected performance of these per se rules leads us to argue that no one of them should be adopted as the rule for a judge or jury to apply in any particular case. Instead, we will argue that the best way to decide a specific predatory pricing case is by applying a rule of reason, as that is traditionally understood in antitrust law. We will suggest a two-stage structuralist ap- proach as a particular way of organizing a rule-of-reason inquiry in predatory pricing cases. 20. Both courts and commentators have noted that there is no accepted definition of predatory pricing or of attempted monopolization under § 2 of the Sherman Act. See, e.g., Pacific Eng'r & Prod. Co. v. Kerr-McGee Corp., 551 F.2d 790, 795-96 (10th Cir.), cert. denied, 434 U.S. 879 (1977) (no definitive standard defining predatory pricing exists in federal law); NATIONAL COMM'N FOR THE REvIE OF ANTITRUST LAWS AND PROCEDURES, REPORT TO THE PRESIDENT AND THE ATTORNEY GENERAL 144-45 (1979) (scope of Sherman Act § 2 attempted monopolization offense not well-defined) [hereinafter cited as ANTI- TRUST COaMMssIoN REPORT]. Indeed, one economist who has contributed to the predatory pricing discussion has stated that: An attempt to provide a universally acceptable definition for a vague term such as "predatory pricing" probably can contribute little. However, the term does relate to a problem that is real and significant-the design of means to permit full and fair competitive measures by the established firm, without foreclosure of entry. Baumol, supra note 1, at 26 (footnote omitted). 21. See, e.g., Areeda-Turner Proposal, supra note 1, at 697 (predatory pricing exists when there are sales at "unremunerative prices"). At least one court has adopted the Areeda-Turner characterization. See Janich Bros. v. American Distilling Co., 570 F.2d 848, 855 (9th Cir. 1977), cert. denied, 439 U.S. 829 (1978). After utilizing the "unremunera- tive pricing" characterization, the Janich court went on to say that "[p]ricing 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." Id. at 856. The Yale Law Journal responsive to our basic concern for preserving and fostering the vi- tality of competition is increased by the fact that the alleged offense involves a charge of improper behavior by the defendant firm. Policy can draw on a body of received economic theory and accumulated empirical studies when addressing itself to structural remedies that alleviate the characteristics that make monopolistic behavior possible; it is possible to evaluate, with some confidence, structural character- istics of a market that may suggest that more firms are necessary to ensure the health of competition. It is much more difficult to try to define precisely what pattern of behavior, whether pricing or oth- erwise, a firm should be allowed to follow and what it should be enjoined from undertaking when the structure of the relevant mar- ket is not specified.2 7 However, it is precisely such behavioral re- strictions that a policy toward predatory pricing seeks: it must dis- tinguish a "predatory" price cut from a "competitive" one and de- velop behavioral constraints that deter the former, but not the latter. II. The Framework for Analysis Given our working definition of predatory pricing, the central problem in formulating a policy to cope with such behavior is the difficult task of inferring unobservable long-run market outcomes from observable short-run market conditions. Such an enterprise, no matter how carefully it is done, is inherently uncertain and involves the possibility of error both because the actual effects of any kind of observable short-run behavior on long-run outcomes are them- selves uncertain and because our methods of predicting those effects are imperfect. This task, however, is unavoidable: to dismiss entirely an assessment of long-run effects, as for example Areeda and Turner seem to do,28 is to dismiss the essence of the predatory pricing problem. Since a profit-maximizing 29 dominant firm will depart from short- 27. The divergence of views one finds in the recent predatory pricing literature con- cerning the "optimal" rule to apply in evaluating dominant firm behavior, see note 1 supra, is testimony to the difficulty of trying to formulate such behavioral restrictions. 28. See notes 12 &. 17 supra. 29. Our analysis assumes, as have most of the contributions to the recent discussion of predatory pricing, that firms are profit-maximizers. This assumption is common to the earlier contributions despite the fact that, as we have noted, see notes 4 & 5 supra, some commentators have focused primarily on short-run profit maximization while others have assumed the firm seeks to maximize long-run profits. For a discussion of the implications for our analysis of relaxing the assumption of profit-maximizing behavior and recognizing other firm objectives, see note 50 infra. 222 Vol. 89: 213, 1979 Predatory Pricing run maximizing behavior only if it expects that such a move will lead to larger long-run profits,30 the best way to assess whether cur- rent behavior is predatory is to evaluate its expected effects on long-run market outcomes. Recognizing the perhaps substantial de- gree of uncertainty involved, we must use information on the ex- isting market structure and short-run firm behavior to try to infer what the long-run outcomes will be. The potential errors resulting from such an inferential process can be classified into two categories. Adopting the terminology of statistical hypothesis testing, we will characterize the two kinds of errors as "false positives" (or Type I errors) and "false negatives" (or Type II errors). These terms, which we now define, are central to our discussion and the reader is urged to keep them firmly fixed in mind. A false positive or Type I error results when the standard being applied to a particular case labels as predatory behavior that is not, in fact, predatory. On the other hand, a false negative or Type II error occurs when the standard does not label as predatory behavior that is, in fact, predatory. Either error is costly; each causes a loss in economic efficiency. The nature of the costs associated with each type of error is developed below. A. Analysis of Error Costs Two kinds of costs are associated with false positive errors-that is, with errors that involve labeling truly competitive price cuts as predatory. First, welfare losses result from the fact that prices may be kept too high for too long as compared to the levels that would have resulted if the dominant firm had more leeway to adjust prices. Second, inefficient firms may be encouraged to remain in the market or to enter the market. This increases the cost of production of the product above the efficient level and results in a waste of scarce resources and hence in a loss of social welfare. It should be noted that any standard that encourages entry by forcing price to be kept above long-run marginal cost for a period of time necessarily runs the risk of preserving inefficient firms, whether existing ones or en- trants, for some period. The question is whether such a standard leads a sufficient number of firms to enter or to mature so that monopoly pricing is eliminated more quickly than it would have 20. In comparing present and future profits, future profits must be discounted to present value to reflect the fact that they are not available today. A dollar today is, ceteris paribus, worth more than a dollar a year from now. For a simple exposition of discounting to present value, see V. BRUDNEY & Nf. CHIREISTEIN, CASES AND MATERIALS ON CORPORATE FINANCE 35-44 (1979). The Yale Law Journal been otherwise. If it does, then the short-run social cost of such a standard may be worth incurring. The costs associated with false negative errors, with errors that re- sult because the predation standard is too permissive, are of three kinds. First, there is a short-run allocative inefficiency 3' if the stan- dard is so lax that it permits the dominant firm to price below short-run marginal cost for any period of time. Second, there is the deadweight loss32 commonly identified with monopolistic pricing. Under a lax standard, this cost is incurred during those periods after the dominant firm has succeeded in either driving others from the market or disciplining the remaining firms to follow its monopolistic- pricing lead. The third type of social loss caused by a false negative error consists of the cost inefficiencies that may result from the in- sulation of the monopoly or dominant firm after it has successfully restricted competition. If removing the pressure generated by com- peting firms diminishes the dominant firm's incentive to find ways to reduce costs-for example, by innovation-the social costs of pro- ducing that firm's product will be higher than they need be. Many structural characteristics affect these error costs; no single structural characteristic is determinative. Instead, one must look at the interaction of a variety of market characteristics to make reason- ably sound empirical judgments about the relative magnitudes of the two sets of error costs. The relevant structural characteristics can be grouped into three basic categories: (1) factors indicative of short- run monopoly power; (2) conditions of entry into the market; and (3) the dynamic effects of competitors or entrants on the costs of production and the quality of products offered to consumers. 31. Short-run allocative inefficiency occurs in such a situation because short-run mar- ginal cost represents the short-run opportunity cost-in terms of resources used-of producing an additional unit of output. If price is less than short-run marginal cost, and if demand is at all elastic, the price will not give consumers the proper signal about the scarcity value of the good. Thus, consumers will purchase too much of the good and as a result, too many resources will be devoted to producing the particular good in the short run. 32. The deadweight loss of monopolistic pricing can be intuitively understood as follows. The monopoly price, P, will be higher than the price, C, that would prevail in a competitive market, and the latter is equal to the short-run marginal cost of producing the good. Hence, if demand is at all elastic, some consumers will reduce their purchases if the price is P instead of C. For those consumers, the value of such purchases lies some- where between C and P. Thus, the increase in price and concomitant decrease in output brought about by monopolistic pricing causes these consumers to lose the difference be- tween their valuations and C. The sum of all such losses is the deadweight loss. Note that no one recoups such losses. In contrast, for all units that would be purchased whether the price were P or C, consumers must pay P-C more per unit but the firm gains exactly that amount. For a more complete explanation, see F. SCHERER, supra note 5, at 13-19; for a diagrammatic illustration of the deadweight loss concept, see note 51 infra. 224 Vol. 89: 213, 1979 Predatory Pricing simply, the opportunity to extract monopoly profits is inversely re- lated to the elasticity of demand. Hence the less elastic the demand for the product, the greater is the cost of not identifying a practice that is predatory as such-that is, the higher is the cost of a false negative error. On the other hand, the deadweight loss that results from giving the monopolist less leeway to adjust prices downward in response to entry is smaller when demand is inelastic than when it is elastic. The less elastic is demand, then, the lower the cost of a false positive error. 2. Conditions of Entry A dominant firm's short-run monopoly power is not of particular concern in and of itself. The critical question is whether or not the dominant firm can use that monopoly power to maintain prices above the competitive level for some significant period of time, and this depends on the conditions of entry into the market. We have chosen to categorize this part of the structural investigation in terms of "con- ditions of entry" rather than "barriers to entry,"'13 because the tra- ditional examples of "entry barriers" are both too restrictive and too easily misinterpreted. In particular, they do not capture the dy- namics of entry processes in different markets that are essential for determining whether potential competition constitutes an effective constraint on the pricing behavior of a dominant firm. Even if there is only one firm in the market, and the elasticity of demand for its product is fairly low, a firm is not in a position to exercise its monopoly power if new firms with additional production capacity can quickly and easily enter a market should prices be raised to a point above the competitive level. In this case, potential com- petition effectively constrains the ability of a dominant firm to en- gage in monopoly pricing for any significant period of time. On the other hand, if entry is costly and time-consuming, the dominant firm will be in a position to exercise its latent monopoly power. Other things being equal, the more costly entry is and the longer it takes for new firms to enter a market in response to monopoly prices, the higher are the costs of false negative errors and the lower the costs of false positive errors. The combination of a highly concentrated market, low demand elasticity, and entry conditions that indicate that potential competi- tion does not respond or responds only slowly to prices set at supra- 38. For a discussion of entry conditions and barriers to entry, see 2 AREEDA-TURNER TREATISE, supra note 1, at 409; F. SCHERER, supra note 5, at 216-34; Williamson Response, supra note 1, at 1195-96. The Yale Law Journal competitive levels should be an area of primary concern for antitrust enforcement; failing to label a predatory price cut as predatory in such situations would be very costly. Similarly, if the conditions of entry are such that competitors enter the market quickly as prices rise above the competitive level, the costs of not labeling a price cut as predatory when it is will not be very large since potential com- petition severely limits the ability of a dominant firm to exercise monopoly power. There are certain structural characteristics of a market that affect entry conditions and thus the ability of potential competition to pro- vide an effective constraint on the pricing behavior of a dominant firm. First, we should look at the amount of capital required by a new firm to enter a market at minimum efficient scale. If capital requirements are very large, it is less likely that entrants will respond quickly, if at all, as prices rise above competitive levels. Capital re- quirements may represent a constraint on entry, even when prices are above competitive levels; because capital markets may not be perfect,39 evidence of substantial sustained supranormal profits may be needed before the required capital commitments can be made. Even if capital markets are perfect, large capital requirements may be indicative of the need to build large-scale production facilities or to develop extensive distribution networks that may take a con- siderable amount of time to complete. This means that entrants can respond to monopoly prices only after a substantial lag time; a firm with monopoly power easily may be able to exploit this entry lag by raising prices in the interim. Second, we should seek to determine whether the existing dominant firm has been successful in establishing a significant "brand prefer- ence" 40 in the eyes of consumers, not by producing an objectively better product, but by having been first in the market or by having made extensive "image" advertising expenditures. Faced with such a market, an entrant not only may have to invest the resources that are technically required to establish production and distribution net- works but also may have to undertake substantial promotional ex- 39. See Posner Article, supra note 1, at 945; Williamson, Book Review, 83 YALE L.J. 647, 656-59 (1974) (reviewing W. BOWMAN, PATENT AND ANTITRUST LAW: A LEGAL AND ECONOMIC APPRAISAL). 40. See FTC v. Procter & Gamble Co., 386 U.S. 568, 572 (1967); FTC v. Borden Co., 383 U.S. 637, 639-40 (1966); J. BAIN, BARRIERS TO NEW COMPETITION 116 (1956); Schmalensee, supra note 3, at 1002, 1033. "First mover" dominant firms may have a "generic name" advantage, see id., or may have engaged in substantial advertising expenditures to dif- ferentiate their products from others that are functionally equivalent, see FTC v. Procter & Gamble Co., 386 U.S. 568, 572 (1967); F. SCHERER, supra note 5, at 341. 228 Vol. 89: 213, 1979 Predatory Pricing penditures, both to get its product known and to overcome brand loyalties that the dominant firm has achieved. In such situations, successful entry will be more costly and more time-consuming, and, as a result, potential competition will be a less effective constraint on the existing dominant firm. 41 As a result, courts should recognize that because generic names and premium brand images induced by advertising change the conditions of entry in such a way that poten- tial competition is a less effective constraint on prices than it other- wise might be, dominant firm behavior in such markets may be worthy of closer scrutiny. A third characteristic of the conditions of entry in a particular market is the ease with which productive resources or assets can be transferred from one firm to another. For any number of reasons, it may be difficult to transfer to new owners the assets of a firm that exits. If there are substantial benefits to "learning by doing" 42 in production or if production in a technologically sophisticated in- dustry requires the coordination of a number of individuals in im- portant design teams and such teams are broken up when the firm exits,43 transferability will be limited. The dominant firm could, in fact, facilitate this disintegration of the team by hiring away key personnel when the firm exits. If assets are not easily transferable, an exit from the market carries with it a reduction in productive capacity and hence the likelihood of social inefficiency. As a result, the cost of a false negative error is higher when there are frictions in the asset-transfer process. The reverse is also true; that is, ceteris paribus, the easier it is for the assets of an exiting firm to be taken over and used by an entrant, the more likely it is that effective competitive pressures can be main- 41. See Schmalensee, subra note 3, at 1032-43 (analysis in context of processed lemon juice industry); Schmalensee, Entry deterrence in the ready-to-eat breakfast cereal in- dustry, 9 BELL J. ECON. 305 (1978) (analysis in context of ready-to-eat breakfast cereal industry). We recognize that the "brand preference" question raises serious problems. For example, how are we to decide whether the benefits of a "generic name" are really attributable to the "accidental" fact that a particular firm was first in the market rather than to the skill and foresight of a first mover? How are we to determine whether ad- vertising expenditures that establish a brand image and a brand preference represent expenditures that "fool" consumers or actually "inform" them of the true differential worth of a particular product? The answer is that such determinations are difficult. But we do know that in such situations entry is more costly and occurs more slowly and thus a dominant firm has more control over price for a longer period of time. 42. See Arrow, The Economic Implications of Learning by Doing, 29 REv. ECON. STUD. 155 (1962); Hirsch, Manufacturing Progress Functions, 34 Rlv. ECON. & STAT. 143 (1952). 43. See P. DOERINGER & M. PIORE, INTERNAL LABOR MARKETS AND MANPOWER ANALYSIS 15-16 (1971); 0. WILLIAMSON, MARKETS AND HIERARCHIES: ANALYSIS AND ANTITRUST IM- PLICATIONs 60-64 (1975). 229 The Yale Law Journal existing products as well as the development of new and better prod- ucts that better satisfy the desires of consumers.48 Although it is gen- erally thought that the structure of industrial markets affects the rate of technological change, the theoretical and empirical literature supports widely divergent views about which structure is preferable. For some commentators, monopoly power is an important spur to in- novation, while for others competition with patent protection is the most effective inducement for rapid technological change. We do not intend to resolve this debate here; indeed, the debate may well be unresolvable as a general matter. Technological change is still so poorly understood and the history of innovative activity differs so much from one market to another that we doubt that a general theory linking market structure variables to the rate and direction of tech- nological change will be convincingly sustained empirically in the near future. However, this does not mean that the effects of compe- tition on innovation can be ignored in assessing the costs of false positive and false negative errors. On the contrary, since innovation has such important effects on social welfare and since a general theory is lacking, the innovative characteristics of the specific firms and markets under investigation should be examined. In order to assess the relevant error costs, an inquiry should be made into the sources of process and product innovations in the mar- ket being scrutinized. Has the dominant firm been a primary source of technological innovations or have smaller firms and entrants been the innovators? In markets in which competition appears not to foster innovation but in which, perhaps because of economies of scale, a relatively insulated dominant firm is a more likely source of tech- nological progress, the cost of false positives is higher and the cost of false negatives is lower. Under these circumstances, forcing a dom- inant firm to keep its price high enough for long enough so that competing firms enter and thrive may engender substantial sacrifices of cost-saving innovation. And, although a social loss will be incurred if we allow a monopolist who temporarily decreases his price to suc- ceed in excluding other firms, the cost of such a false negative error 48. We recognize, as will anyone familiar with the literature on technological change and market structure, that ascertaining the relationship between the degree of competi- tion and the pace of innovation will be very difficult. See, e.g., F. SCHERER, sutra note 5, at 346-78; Markham, Concentration: A Stimulus or Retardant to Innovation? in INDUSTRIAL CONCENTRATION: THE NEW LEARNING 247 (H. Goldschmid, et al., eds. 1974). Nevertheless, even if such considerations can enter only in extreme cases-where there is reason to believe that competition has a very beneficial or a very deleterious effect on technological progress in production for the market being examined-they are very important in any genuinely long-run view of the social-welfare effects of short-run pricing behavior. 232 Vol. 89: 213, 1979 Predatory Pricing will be mitigated by the enhanced technological progress that results from the monopolist's successful self-preservation. In contrast, failing to identify predatory price actions is more costly and mislabeling a truly competitive pricing strategy as predatory is less costly when the rate of technological change is positively affected by an increase in the number of firms in the market, that is, when small competitors or entrants have been the major sources of innovation. Beyond the consideration of the sources of technological change, a general picture of the dynamics of the market under investigation is important for the evaluation of error costs. Is the market growing rapidly? Is the industry a "declining industry"? Do prices tend to rise and fall with cyclical changes in supply and demand? If an in- dustry is growing very rapidly and the products being sold are new products introduced by the dominant firm, we might expect to find both that there is substantial entry and that the dominant firm is earning profits that appear to be above the long-run competitive level. Rather than indicating a long-run monopoly problem, however, the size and profitability of the dominant firm may be reflecting only the short-run disequilibrium characteristics of a very competitive indus- try. In such a market, the costs of false positives would be relatively high and the costs of false negatives relatively low. Similarly, in a declining industry we are unlikely to observe entry; rather we are likely to observe some firms exiting while the remaining firms achieve larger shares of a declining market. In this case, we would not neces- sarily want to interpret increasing concentration as reflecting in- creases in monopoly power. Although other structural factors may affect the costs of the two kinds of errors associated with any predatory pricing standard,49 our discussion has focused on what we think are the principal elements; 50 49. For example, the structure of the consumer side of the market might also merit examination in particular cases. Is the product purchased by a relatively small number of large, sophisticated industrial firms or is the product purchased primarily by a large number of individual consumers? In the first instance, concentration on the buyers' side might constrain the monopoly power the dominant firm can exercise, while in the second instance the demand side will not impose much restraint on the firm's exercise of monopoly power. 50. Our discussion of how structural factors affect the costs of false positive and false negative errors has assumed that the dominant firm is a "cost minimizer," that is, given existing technology and input prices, the firm produces its chosen level of output at minimum cost. In the microeconomic theory of firm behavior, cost minimization follows from one of two assumptions: (i) the objective of the firm's management is to maximize short-run profits or, alternatively, (ii) there are numerous competitors or there is easy entry so that those firms that survive in the long run will be cost minimizers whether or not specific individual firms observed at any particular time are cost minimizers. See Nelson & Winter, Factor Price changes and factor substitution in an evolutionary model, 6 233 The Yale Law Journal Vol. 89: 213, 1979 examination of these elements provides a way of identifying the kinds of markets in which the costs of false positive errors and false nega- tive errors are likely to be high or low. B. Analysis of the Probabilities of Error Let us now turn to the probability of making each type of error. Consider first the probability that a false positive error is made: a particular price reduction is declared predatory when it is not. The probability of such an error depends on three factors: the probability that an observed price cut is, in fact, not predatory; the standard used to classify that reduction as predatory or not predatory; and the accuracy with which the standard is applied. Similarly, the probabili- BELL. J. ECON. 466 (1975); Nelson SL Winter, Neoclassical vs. Evolutionary Theories of Economic Growth: Critique and Prospectus, 84 EcoN. J. 886 (1974); Winter, Satisficing, Selection, and the Innovating Remnant, 85 Q.J. EcoN. 237 (1971). In reality, neither of these assumptions necessarily holds in those markets that are likely to be of most concern to antitrust monopoly policy. Managerial objectives may include other variables besides profits, such as sales or the growth rate of the firm, or more personal managerial objectives such as salary, security, power, or prestige. See W. B.UssOL, BUSINESS BEHAVIOR, VALUE AND GROWTH 45-52, 86-104 (rev. ed. 1967); Williamson, Man- agerial Discretion and Business Behavior, 53 AM. ECON. REv. 1032, 1033-38 (1963). Or, there may be so few competitors, or entry may be so difficult, that competition does not perform its "natural selection" role. As a result, a dominant firm may depart from cost minimization and waste resources; this phenomenon has been referred to as "X-in- efficiency," see Leibenstein, Allocative Efficiency vs. 'X-Efficiency,' 56 ANI. ECON. REV. 392 (1966), or "organizational slack," see R. CYERT & J. MARCH, A BEHAVIORAL THEORY OF THE FIRm 36-38 (1963). Furthermore, if an existing firm with short-run monopoly power has as its goal the maximization of long-run profits, it may find it advantageous to allocate resources to deterring entry. See R. POSNER, supra note 1, at 184-85. Thus, for example, a dominant firm may have an incentive to carry excess capacity to deter entry. See Spence, Entry, capacity, investment and oligopolistic pricing, 8 BELL. J. ECON. 534 (1977). What- ever the reason, departures from cost minimization, which seem most likely to occur in markets with monopoly characteristics, represent real social costs of monopoly, costs which antitrust policy should seek to eliminate. Given these considerations, it is reasonable to ask why we have not listed "departures from cost minimization" as one of the structural characteristics affecting error costs. Note first that we have in fact focused on dynamic departures from cost minimization by suggesting that the effects of competition on technological change are important. On the other hand, we have not focused directly on static departures from cost minimization. A number of reasons support our decision. First, "organizational slack" and "X-inef- ficiency" are extremely difficult to measure in any systematic or reliable fashion. Second, the sources of inefficiency resulting from long-run profit maximization constitute particu- lar aspects of behavior that are best addressed as part of an inquiry into behavior rather than as part of the structural analysis. Third and perhaps most important, situations in which departures from cost minimization are most likely to occur are readily identified with structural factors that we have discussed in detail-short-run monopoly power, condi- tions of entry, and the dynamic effects of competition. As a general matter, where structural factors are such that the traditional welfare costs of false negatives are high (substantial short-run monopoly power, difficult entry conditions, etc.), considerations of static departures from cost minimization make them even higher. Rather than trying to estimate departures from cost minimization directly, we identify the structural factors that are likely to make such costs large. 234 Predatory Pricing We will not present a detailed catalogue of each of the structural characteristics we have discussed. However, consider the following example of how structural factors affect the probability that an un- constrained dominant firm will make a predatory price reduction. The greater the short-run monopoly power of the dominant firm, ceteris paribus, the higher is the probability of predatory pricing when the firm is free to choose any pricing strategy. This does not say any- thing about the absolute size of the probability; it does not, for ex- ample, say that if we observe a temporary reduction of price by an unconstrained dominant firm with substantial short-run monopoly power, then it is highly likely that the reduction is part of a preda- tory pricing strategy. All it says is that if we observe such a price reduction, then it is more likely that the reduction is predatory if the dominant firm currently has considerable short-run monopoly power than if it does not. The more short-run monopoly power the firm possesses, the greater will be its ability to raise prices, and thereby increase profits, if it succeeds in eliminating the threat of entry. Hence, extinguishing such threats is more valuable to a firm with more short-run monopoly power than to one with less control, and, in the absence of legal restrictions, such entry-deterring strategies are, ceteris paribus, more likely to be pursued by firms with more market power. The preceding discussion has suggested how structural characteris- tics affect the profitability, and hence the likelihood, of predatory pricing when the dominant firm knows that its pricing policy will not be subject to legal challenge. However, the introduction of a stan- dard to be used in judging the legality of the firm's pricing behavior will alter this assessment of profitability. The probability that viola- tions of the standard will be detected and the penalties imposed for such violations determine whether the firm perceives itself as abso- lutely constrained from pursuing certain pricing policies or perceives only a decrease in the probability that it will actually enjoy the long-term benefits of predatory price cuts. Thus, the likelihood that a firm whose pricing policy is being assessed in terms of a particular legal standard will make predatory price cuts will depend on two factors: the behavior that would be most profitable for the uncon- strained firm given the structural characteristics affecting the un- constrained firm's profit calculus, and the nature and application of the legal standard. Thus, the character of the legal standard influences the error prob- abilities indirectly by affecting the firm's incentives to make preda- 237 The Yale Law Journal tory price cuts. The nature of the legal standard, however, also has a direct effect on the probabilities of false positive and false nega- tive errors. The factors that the standard takes into account affect the probability that the standard will lead to particular kinds of errors; a standard that ignores the relevant considerations discussed above will be more error-prone than one that takes them into ac- count. For example, a rule that is based only on static market con- ditions and ignores dynamic considerations is much more likely to generate errors than is one that gives appropriate weight to dynamic factors. Furthermore, the strictness of the standard in evaluating the fac- tors it does consider also affects the error probabilities. The more stringent the standard is, the greater the probability of a false posi- tive and the smaller the probability of a false negative. For example, a per se rule that declares illegal any price reduction by a firm whose share of a particular product market exceeds sixty percent, will have quite a high probability of declaring a firm's price preda- tory when it is not, while it will also have a relatively low probability of making a false negative error. On the other hand, suppose the rule in effect declares nonpredatory any price at which gross revenues ex- ceed variable costs and requires a demonstration that the firm has set price below average variable cost before the firm's action is declared illegal.53 In this case, there will be a low probability of making a false positive error, but the probability of a false negative error will be high. Finally, the ability of judges and juries to understand and apply any particular standard will affect the probabilities that the particu- lar policy will result in each kind of error. This institutional com- petence issue-the putative inability of judges and juries to make certain kinds of complex determinations and to apply those analyses consistently across cases-is sometimes invoked as a conclusive ground for relying on a per se rule rather than on a rule-of-reason approach. The question of how well these decisionmakers can apply the tools of economic theory to monopoly and monopolization cases has often troubled commentators who view economic efficiency as the goal of antitrust policy.5 4 Even those people who argue that the antitrust laws were not intended for the single-minded pursuit of economic 53. See Areeda-Turner Proposal, supra note 1, at 716-18; pp. 250-52 infra (discussing use of average variable cost). 54. See R. BORK, supra note 1, at 80-81, 411-18; Posner, The Rule of Reason and the Economic Approach: Reflections on the Sylvania Decision, 45 U. Cm. L. Rav. 1, 20 (1977). Vol. 89: 213, 1979 Predatory Pricing efficiency 55 agree that if judges and juries are going to use the cri- terion of economic efficiency as a guideline, it is preferable that they do so in a correct and consistent way. Given judges' and jurors' training and the available empirical in- formation, the ability of these decisionmakers to apply economic prin- ciples correctly and consistently is likely to be imperfect. The critical question is how far short they will fall in applying each of the al- ternative rules. But the answer to this question is only one factor to be considered in assessing error probabilities and choosing an ap- proach to predatory pricing. For although a per se rule may be easier to apply than a more complicated rule, the benefits of its simplicity may well be outweighed by the cost of its inaccuracy if the rule fails to encompass important elements of the market situation. Moreover, the simplicity of such a rule may be more apparent than real. Consider, for example, the simple rule mentioned earlier that would declare illegal per se any price reduction by a firm whose market share exceeds sixty percent. This rule would have a low probability of making a false negative error by declaring a firm's pricing behavior nonpredatory when it was predatory, but it would have a high prob- ability of making the opposite kind of error-declaring predatory a truly nonpredatory price cut. The resulting costs of the inaccurate decisions reached could well outweigh the simplicity of the rule. Application of such a mechanical rule can lead to errors that would have been avoided had a broader examination of available economic information been undertaken. Furthermore, the "sixty percent mar- ket share" rule is simple only until one begins to inquire into the appropriate way to measure market share. What, for instance, are the bounds of the relevant market? How will degrees of substitutability be gauged? There is no "right" way to define the relevant market if the intention is to use this simple measurement to distinguish predatory pricing from competitive pricing.56 Hence, the statement of the rule is deceptively simple; articulating a simple-sounding rule does not ensure that it will be correctly applied. Although not all simple per se rules are subject to the same problems as is the "sixty percent market share" rule, simple per se rules can be misused just as more complicated approaches can.57 55. See, e.g., Pitofsky, supra note 24, at 1051-58, 1060-65; Scherer, supra note 24, at 975-81. 56. See Schmalensee, supra note 3, at 1015-16. 57. For an example of the difficulties engendered in applying an apparently simple rule, see our discussion of In re Certain Welded Stainless Steel Pipe and Tube, United States International Trade Commission Investigation No. 337-TA-29 (Feb. 22, 1978), at pp. 263-64 infra. The Yale Law Journal Two other kinds of implementation costs are generated by any policy that does not provide a "bright line" between what is legal behavior and what is not. First, in the absence of a bright-line stan- dard, the costs of harassing litigation will increase as suits are brought by competitors seeking to protect themselves. An ambiguous rule makes it easier for such potential litigants to cloak their pro- tectionist enterprise in a predatory pricing suit. Such complaints are costly to the defendant firm, and, to the extent the cause is without merit, they are costly to society as a whole. Second, in the absence of a bright-line rule, firms' decisionmaking processes are subject to greater uncertainty; a firm cannot be sure whether its pricing policies are legal. The uncertainty generated by the confusing signals that dominant firms would receive from the case law could cause the predatory pricing standard to have no de- terrent effect or to have a perverse efficiency effect. These costs are highest for an "all-factors" unstructured, case-by-case approach, lower for the Areeda-Turner cost-based rule or the William- son output-restriction rule, and even lower for a per se market share rule.64 Per se rules reduce litigiousness and spurious private damage actions and reduce firms' uncertainty about what pricing behavior is legal. But, once again, although this is an attractive property of per se rules, it is only one factor to consider. III. A Proposed Two-Tier Approach It should be clear from the preceding discussion that the probabili- ty that predatory pricing will occur and the costs that such pricing will engender if it does occur vary according to the inherently dif- ferent characteristics of different markets. As a result, given any par- ticular predatory pricing rule, there will be differences across mar- kets in both the probabilities and costs of false positive and false negative errors. The costs of implementing any given policy will also vary across markets. 65 In sum, the "optimal" rule for one market situation will not be "optimal" for all market situations. We can, of course, conceptualize the process of selecting a per se 64. Current concern about the costs of litigation is reflected in the substantial atten- tion the National Commission for the Review of Antitrust Laws and Procedures gave to arguments that rules not focused on conduct (e.g., per se rules based on market structure) will reduce litigation costs. See ANTITRUST COMMIISSION REPORT, sutra note 20, at 151-54. 65. The ease or difficulty of assessing important facts-for example, market share or price-cost margin-will vary across markets as, for example, the degree of product dif- ferentiation varies. Also, the ability of a dominant firm to adapt to one legal rule or another will depend on the market in which it operates. 242 Vol. 89: 213, 1979 Predatory Pricing behavioral rule that would apply to all market situations. But to choose such a rule correctly we must be prepared to make subjective probability judgments about the distribution of market structures and hence the distribution of alternative error cost/error probability/ implementation cost combinations. Using this subjective probability distribution together with a complete calculation of the expected to- tal cost of each possible rule for each kind of market situation, we could in principle select the rule that minimizes the expected total costs across all market situations. The description of this calculation suggests how infeasible it would be as a practical means of choosing the "optimal" per se rule. Not only would it be extremely difficult to arrive at a rule that would meet with any reasonable consensus, but relying on such a rule would also reduce the ability to use all the information that is likely to be available to us. This brings into question the wisdom of sub- scribing to any single per se rule based on observed firm behavior alone. Restricting or permitting some particular behavior as a gen- eral matter may yield economic gains in a few markets, but losses in many others. A rule based solely on behavioral considerations does not provide a means for distinguishing market situations in which the expected costs of predatory pricing are high from those in which they are low. The decision-theoretic framework that we have presented suggests a way of formulating a rule-of-reason approach to predatory pricing that both preserves the desired flexibility to respond to facts particu- lar to a given market and minimizes implementation costs. The pri- mary objective is to design an approach that makes the probability of a false positive error low when the cost of such an error is high and that makes the probability of a false negative error low when the cost of that type of error is high. Since structural factors that raise the cost of false positive errors lower the cost of false negatives and vice versa, an approach to predatory pricing should employ standards that are carefully conditioned on structural factors. Thus, we sug- gest a two-tier approach that is sensitive to differences in market structure and allows us to screen out market situations where a reasonable threat of predatory pricing exists from those where it does not.60 66. One consideration that appears to have been very important in leading Areeda and Turner to the cost-based rule they formulate is their view that predatory pricing is rare in all markets. See Areeda-Turner ProPosal, supra note 1, at 699; Areeda & Turner on Scherer, sukra note 1, at 894; Areeda & Turner on Williamson, supra note 1, at 1339. This leads them to a rule that provides some protection against false negative errors, but 243 The Yale Law Journal In the first stage, we would examine both the structural char- acteristics of the market in question and the market power of the alleged predator firm to find out if they generate a reasonable ex- pectation that predatory pricing could occur and impose significant economic losses on society. It is necessary, at this first stage, to form a judgment about whether the market structure is such that the ex- pected costs of failing to identify as predatory, pricing behavior that actually is predatory, are likely to be high. Such a judgment would be based on consideration of the structural factors we have discussed above. Unless a reasonable case could be made that there was a serious monopoly problem in the industry, no detailed investigation of the alleged predator's intent or behavior, nor speculation about the long- run consequences of its pricing policy, would be undertaken. In short, a claim that predatory pricing had taken place could be pursued only if the plaintiff could show that the market context in which the behavior was taking place was in fact conducive to predatory pricing. We believe that this first-stage "structural" requirement, using generally accepted definitions of monopoly power, would discourage frivolous predatory pricing cases. A plaintiff would have to show that a significant monopoly problem existed in the market and could not try to supplement weak evidence on the likelihood and effects of pred- atory pricing by appealing to vague predatory pricing standards such as "intent" or "below-cost" pricing. Thus, a claim of monopo- lization through predatory pricing could be pursued only in market situations in which the structural characteristics suggest that there is a reasonable probability that monopoly power has been or could be sustained by the use of price reductions. The concern about in- curring substantial false positive error costs, by labeling as predatory, pricing behavior that is not predatory, would be reduced since in- stances of alleged predation in which such costs were expected to be greatest would be eliminated by the initial "structural" analysis. Only those situations in which the costs of false negative errors were that concentrates on reducing the probability of false positive errors. A similar view appears to lead Bork to the conclusion that predatory pricing should not constitute an antitrust violation. See R. BoRx, suPra note 1, at 149-55. Of course, the most effective way to reduce false positive errors is never to declare a firm's pricing behavior predatory. However, if a policymaker agreed with Areeda and Turner's empirical guess about the frequency of predatory pricing in some markets but not in others, he would probably want to adopt an approach that allowed him to differentiate between these two types of markets and to examine more closely behavior in those markets in which predatory pricing seemed most likely. 244 Vol. 89: 213, 1979 Predatory Pricing (3) Dynamic effects of competition on costs and products: (a) The three smaller firms in the market are the primary sources of process and product innovations. CASE #2: (1) Short-run monopoly power: (a) The alleged predator has forty percent of the market and eight smaller firms of varying sizes comprise the rest. (b) The firm's market share has been gradually declining over time. (c) When the firm has tried to "lead" price increases, the demand for its product has declined substantially. (d) The firm's profits have been somewhat greater than the cost of capital over the previous four years, but profitability has been falling. (2) Conditions of entry: (a) Entry at minimum efficient scale requires initial investment of $10 million. (b) Three new firms have entered the market successfully over the past five years and have grown rapidly. Two others have entered and later failed. (c) Products sold by various firms are fairly homogeneous, and ad- vertising costs are not a substantial fraction of total costs. There is no evidence that any product enjoys "premium brand" advantage. (3) Dynamic effects of competition on costs and products: (a) This product market is not characterized by rapid technological change. (b) Technological change that has occurred has focused on new and improved products, and the dominant firm is the primary creator of new products. CASE #3: (1) Short-run monopoly power: (a) The alleged predator has sixty percent of the market and four smaller firms comprise the rest of the market. In some geographical markets one of the other firms has a larger market share. (b) The firm's market share has averaged sixty percent over the past ten years, but the share has fluctuated considerably from year to year. (c) When the firm has tried to "lead" price increases, the other firms have sometimes followed, but when they have not, the firm has rolled back the price increase as the demand for its product declined dramatically. (d) The firm's profits have on average been somewhat greater than the cost of capital but show a slight downward trend. (2) Conditions of entry: (a) Entry at minimum efficient scale requires investment of at least $200 million to enter all major geographical markets simultaneously. Entry The Yale Law Journal into major geographical submarkets requires initial investment of $20 million. (b) Three new firms have entered in particular geographical areas over the past five years, but only one of these still survives. (c) The market is characterized by some product differentiation and advertising outlays are above average, but the firm's product does not gen- erally carry a "premium brand" price. (3) Dynamic effects of competition on costs and products: (a) The market is characterized by fairly rapid cost-saving technological changes. (b) Both the dominant firm and two of its competitors, one of which is a recent entrant, have introduced new low-cost production processes. Recall that at this stage we are not determining whether the firm is liable, but only whether, on the basis of an overview of structural factors, there appears to be enough of a monopoly problem that a further investigation of firm behavior is warranted. This determina- tion is easy for the first two cases, but difficult for the third. Given the structural characteristics of case #1, an inquiry into dominant firm behavior is certainly in order. The market described is exactly the type in which predatory pricing is most likely to occur, and to result in significant losses in economic efficiency if entry is not allowed to erode the monopoly power of the dominant firm. In case #2, no further inquiry into behavior is necessary. Although there is some evidence of historical monopoly power, this does not ap- pear to be a market in which there is a serious long-term monopoly problem. The competitive process seems to be working efficiently to erode any monopoly power that may exist. Thus, a predatory pricing action brought by one of the unsuccessful entrants should be dismissed without any further inquiry into behavior. The third case is not as readily characterized as the other two. The market is less concentrated than in case #1. Although the market share of the dominant firm has remained fairly high, it is also unstable. Furthermore, the alleged predator is not dominant in all geographical markets. Simultaneous entry into all geographical submarkets is very costly, but it is much less difficult to enter individual submarkets and one case of successful entry has occurred in this way. There is some product differentiation, but an entrant does not have to confront a major entrenched "image" problem. Innovations have been contrib- uted by both the dominant firm and some of its competitors. How- ever, there is some evidence that the dominant firm has power over price and has been able to sustain supranormal profits over a con- siderable period of time. 248 Vol. 89: 213, 1979 Predatory Pricing While in case #1 an analysis of behavior is clearly warranted, and in case #2 it is not, the correct way to proceed in case #3 is less obvious. In close cases such as this, we believe that the decision to go forward must ultimately depend upon the court's judgment as to whether a detailed exploration of behavior will help to provide further informa- tion about the monopoly power that the dominant firm actually possesses or might obtain as a result of this behavior. B. The Second Tier: Behavioral Considerations In those cases in which the firm-specific and market-specific struc- tural characteristics suggest that the efficiency losses of failing to iden- tify, as predatory, pricing that is predatory, are likely to be high, it is necessary to proceed to the second tier of analysis. It is presumed that a firm subjected to the behavioral inquiry of the second stage has substantial market power, which can be exploited by employing strategic measures to maintain a trajectory of supra-com- petitive prices for a longer period of time than could be enjoyed with- out engaging in strategic behavior. Hence, the standard that a de- fendant firm subject to second-tier analysis should have to satisfy to be exonerated will be more rigorous than the test one would propose if the search were for a single behavioral rule to apply to all firms in all markets; a more detailed scrutiny of structure, behavior, and expected performance would be conducted in the context of the particular pric- ing behavior at issue. There is a place in the second-tier analysis of firm behavior for each of the rules proposed by other recent contrib- utors to the discussion of predatory pricing.6 9 However, none of these rules alone provides a satisfactory basis for the second-tier analysis. What considerations ought, then, to enter into the second-stage anal- ysis of firm pricing behavior? Merely observing a price cut in the face of entry or potential entry is clearly insufficient as evidence of predatory behavior. The destruction of monopoly elements through new com- 69. At this tier the court would undertake a more detailed scrutiny of structure, be- havior, and expected performance in the context of the particular pricing behavior at issue than it would conduct at the first tier. The kinds of firm and market characteristics that Scherer would examine in every predatory pricing case, see Scherer Proposal, supra note 1, at 890, would be scrutinized in those cases where the reasonable expectation of substantial false negative error costs led to the second tier of analysis. Elements of the Areeda-Turner, Williamson, and Baumol rules are incorporated at the second tier. See pp. 250-58 infra. Note that some commentators have proposed that different rules be applied in different situations. See, e.g., 3 AREEDA-TURNER TREATISE, supra note 1, at 715d (basic average variable cost test with examination of marginal cost also in some situa- tions); Williamson Proposal, supra note 1, at 331-37 (proposing output-restriction rule, average variable cost test, and average total cost test at various points in litigation). The Yale Law Journal Vol. 89: 213, 1979 our reasoning to this conclusion about the status of pricing below average variable cost is different from that of Areeda and Turner. Their analysis and conclusion are based on static microeconomic models and associated concerns with short-run efficiency considera- tions.78 Since the essence of the predatory pricing problem is dynamic, a static perspective is inadequate. We are trying to use short-run static behavior to infer longer-run intent and consequences. Therefore, the adoption of a strategy of pricing below average vari- able cost by a dominant firm confronted with entry is sufficient to demonstrate predation. A price below average variable cost, and for that matter, a price below average total cost, could not possibly be sustained in the long run since, to survive, firms must cover total costs in the long run. A firm with market power-the ability to control price-would only have an incentive to impose losses on itself when faced with an entrant if the promise of future monopoly gains made such a tactic profitable from a long-run perspective. 2. Pricing Between Average Variable Cost and Average Total Cost Although pricing below average variable cost is a sufficient condi- tion for establishing predatory behavior, we do not believe it should be a necessary condition. Pricing between average variable cost and average total cost also may indicate predation. In a competitive market, the equilibrium market price will equal the average total cost of production,7 9 including a normal rate of re- turn on capital invested,80 and this will, in turn, equal long-run mar- 78. See note 4 supra. 79. For a single-product firm, average total cost is easily defined. In the more likely multiproduct context, we are using "average total cost" to signify the average incremental cost of the commodity of concern and not any arbitrary "fully allocated cost measure." See Baumol, supra note 1, at 9 n.26 (high likelihood that most cases will involve multi- product firms; arbitrary and difficult to fully allocate cost among products). Baumol de- fines the "average incremental cost" of product X "as total company cost minus what the total cost of the company would be in the absence of production of X, all divided by the quantity of X being produced." Id. In using an average total cost test, courts will have to make some effort to guard against "creative accounting." This will be especially difficult in the case of multi- product firms where the relevant average total cost figure is the average incremental cost of the commodity with respect to which predatory pricing is alleged. In calculating this average total cost, the court may look first to the firm's allocation of "overhead" costs as a guide, though we would urge that such guidance be sought only with extreme caution. If the firm's allocations are examined, a minimal requirement should be that preentry, not post-entry, cost allocations be used. 80. Some imputation of a normal rate of return on capital must be included in the average total cost figure. This will generally be lower than the dominant firm's his- torical rate of return on capital. Since the computation of a normal rate of return on capital is complex and uncertain, it may be desirable to take the average rate of return 252 Predatory Pricing ginal cost. Of course, prices in a competitive market will fluctuate over time, sometimes rising above and sometimes falling below this com- petitive standard, as short-run variations occur in demand and supply conditions. But it is highly unlikely that the socially optimal trajectory of prices and entry in any market initially characterized by monopoly or oligopoly would involve immediate post-entry prices that impose a loss on the dominant firm. A price below average total cost could drive equally efficient and perhaps even more efficient rivals from the market or deter such firms from entering; in the face of entry lags, the dominant firm could then resume its monopoly pricing behavior. Al- though a dominant firm maximizing only short-run profits would probably lower its price in response to entry or the threat of entry, it would not decrease its price to a level below average total cost. Hence, we would recommend that if the firm-specific and market- specific structural characteristics considered in the first stage of our proposed inquiry lead to an examination of the defendant firm's be- havior, then, at the second stage, a price response that does not cover average total cost should be presumed predatory unless the dominant firm can show that this strategy maximizes short-run profits. The short-run profit-maximization defense is likely to be valid in only one situation-when substantial excess production capacity exists in the industry. This condition might arise for one of three reasons: first, the alleged predator may be operating in a declining industry; second, the entrant may have entered at a scale sufficiently large that, at a price equal to average total cost, total industry capacity would be underutilized; and third, the dominant firm may follow a conscious strategy of carrying excess capacity so as to deter entry.81 The dominant firm would be allowed to defend its pricing decisions by proving that either of the first two circumstances existed. - But the alleged predator would not be allowed an exception to the average total cost rule in the for manufacturing industries in the United States as a starting point in making this imputation and to require an affirmative showing that a different value is more appro- priate for the firm in question. In this regard, it should be clear that profit calculations based on computations of rates of return on sales or similar markup criteria are inappro- priate and are likely to lead to incorrect conclusions. 81. See Spence, supra note 50; Williamson Proposal, supra note 1, at 294. 82. In Pacific Eng'r & Prod. Co. v. Kerr-McGee Corp., 551 F.2d 790 (10th Cir.), cert. denied, 434 U.S. 879 (1977), an unanticipated decline in demand for a chemical product resulted in significant excess capacity in the industry only a few years after bright pros- pects had induced large-scale entry by several firms. 551 F.2d at 791-92. The dominant firm held price below its average total cost but above both its marginal cost and its average variable cost. Id. at 797. The court found no antitrust violation, arguing that the dominant firm had simply engaged in price competition in the face of excess capacity that had resulted from an unanticipated drop in demand. See id. at 796-97. 253 The Yale Law Journal Vol. 89: 213, 1979 third case, in which the industry's immediate post-entry excess capacity was the result of the dominant firm's conscious strategy of carrying preentry excess capacity. In line with Williamson's analysis, the domi- nant firm could not defend its price reduction on the grounds of excess capacity if the record showed that it had increased its own output in the face of entry.83 With regard to the comparison between the alleged predator's price and its costs, one may wonder why we even mention average variable cost. Since average variable cost is always below average total cost, it would appear to be redundant to look at the former as well as the latter: any price that is below average variable cost will be below average total cost and can be held predatory on that ground. However, we view the average variable cost standard as a quick check of the alleged predatory pricing behavior. Since average variable cost is likely to be easier to calculate8 4 than average total cost and since a price cut to so low a level is obviously predatory, it makes sense to look at this threshold first. However, a price cut to below average total cost remains the primary cost standard at the second tier. 3. Pricing Above Average Total Cost At least in theory, a price cut to a point above average total cost in response to entry could be predatory because such a price cut could represent a sacrifice of short-run profits for longer-run monopoly gain.8 However, as is not true for a dominant firm's price cuts to 83. See Williamson Proposal, supra note 1, at 307-10 (superior welfare properties of output-restriction rule that bars dominant firm from strategically carrying excess capacity). 84. Average total cost is likely to be more difficult to calculate than average variable cost for two primary reasons. First, a calculation of average total cost requires calculating the appropriate elements of fixed costs as well as all the elements of average variable costs. Second, calculating fixed cost elements often raises significant difficulties. Accounting data on fixed costs include costs of management or facilities that may be used for producing goods other than those of concern in a predatory pricing case. The correct imputation of fixed costs requires the calculation of average incremental cost. See note 79 supra. This will involve a significant amount of additional calculation because one must ascertain what a company's cost would be if it did not produce a certain product. As Professor Williamson has noted, the calculation of average variable costs also pre- sents difficulties that have been widely recognized by most commentators concerned with cost-based tests. See Williamson Response, supra note 1, at 1196; Williamson Proposal, supra note 1, at 312 & nn. 70-72. In particular, the process of deciding which costs are variable and which are fixed will be a difficult one; the required line-drawing about the nature of costs makes average variable cost harder to calculate than it might, at first glance, seem to be. See 3 AREEDA-TURNER TREATISE, supra note 1, 715c; cf. Areeda & Turner on Williamson, supra note 1, at 1338 (disputes in defining which costs are "vari- able" should be resolved by somewhat arbitrary rules; most disputed items should be assigned to the variable cost category). 85. For example, consider the following set of conditions: capital markets are im- perfect, information flows concerning the riskiness of entry are poor, potential entrants have "thin" capital bases, and production involves high start-up costs as well as sub- 254 Predatory Pricing The sustained price reduction test alone would not suffice to deter such an approach; the average total cost test would enjoin it. As a final example, consider a dominant firm that operates in a market characterized by rapid product innovation. In such a market, a product that is successful today may be superseded by a new and better one in a very short period of time. Thus, the dominant firm could drive an entrant from the market by charging prices below average total cost for a single generation of products with the aim of maintaining its monopoly power over future generations of products. Successful entry today might have ensured competition over all future generations of products; preventing entry today may result in future monopoly.s9 The established firm would recoup its short-run losses on today's product by charging monopoly prices on tomorrow's product. However, the "price increases" would occur on different products. Therefore, if the firm's structural characteristics and those of the market it serves have led to a second-stage analysis of its pricing behavior, further examination asking only whether the dominant firm had sustained for the required period of time its price cut on today's product would not prevent its predatory actions. In contrast, cost-based tests would be more effective in detecting such anticompetitive practices. In each of these instances, a behavioral inquiry focused solely on whether a price cut has been maintained may fail to detect, and hence to stop, predatory behavior. It is critical to recognize that actual markets can be far more complicated than static or dynamic theoretical models of single-product, single-market firms imply. Entry strategies un- doubtedly will differ from one market to the next as the supply and demand characteristics of the markets vary. Naturally, as a consequence, the entry deterrence strategies of dominant firms in the different mar- kets will vary in response. Indeed, our empirical guess would be that firms that fail the first-tier structural standard and thereby make a be- havioral inquiry necessary are far more likely to be operating in com- plex markets than in simple ones. Hence, we believe that the sustained price test, although perhaps adequate for assessing alleged predation in 89. Consider the computer example presented above. See note 45 supra. Suppose that the sequential entry strategy is the easiest way for new firms to enter this market, but that the individual segments are themselves characterized by rapid technological change with new generations of products introduced in each segment every few years. By "permanent- ly" reducing prices to levels such that entrants lose money on the current generation of products, the dominant firm may be able to make entry more costly or induce exit in the hope that competition will be reduced over future generations of products. A price re- duction may thereby be sustained over one generation of products but competition may be reduced over future generations if the real or perceived costs of entry are increased as a result of this kind of pricing behavior. The Yale Law Journal relatively static single-product national markets, is not sufficient, by itself, as a general predatory pricing policy; it should be joined with cost-based tests to form a sound second-stage examination of the de- fendant firm's behavior. C. Overview of the Two-Tier Approach Our proposed approach does not force large industrial firms to create and maintain an uneconomic price umbrella in the markets they serve. For most firms in the economy, the first, structural tier of our approach will eliminate any judicial inquiry into pricing behavior. A firm that is subject to a second-stage examination of its pricing will have com- plete freedom to reduce its prices to the level that would prevail in the long run in a competitive market, and even below that if an excess- capacity or declining-industry case can be demonstrated.90 Such price reductions could be challenged only if they were later reversed in an exercise of monopoly power made possible by the diminished com- petition in the market. The message our proposed behavioral inquiry would convey to firms with monopoly power is that they should not use that power to main- tain their dominant positions. They will know that price cuts below average variable cost will be viewed as clear predatory acts, that prices below average total cost will establish a presumption of predation, and that other price cuts will be questioned only if subsequent increases are not justified by changes in cost or demand conditions. We believe these criteria are sufficiently well-defined that they can be utilized effectively to identify predatory behavior in both government cases and private treble damage actions. Furthermore, their implementation does not require an extensive investigation into the intent of the various pricing responses to entry. Our emphasis at the second tier is on the application of a rule of reason to the analysis of allegedly predatory price cuts once the presence of monopoly power has been established. The substantive content of this second-stage examination of pricing behavior is, in a sense, a selective combination of the rules proposed by Areeda and Turner, Baumol, and Williamson. Pricing below average variable cost constitutes, as Areeda and Turner proposed, a sufficient condition for finding predation. However, under our approach, below-cost pric- ing is not necessary for predation. Instead, we apply Baumol's "quasi- permanent price reduction" rule, and supplement it with the cost- 90. See pp. 252-54 supra. 258 Vol. 89: 213, 1979 Predatory Pricing based tests, to identify predatory behavior. In particular, Williamson's "backstop" average total cost test is an intrinsic part of the second-stage behavioral inquiry we propose. Finally, Williamson's output-restriction rule is reflected in our limitation of the excess-capacity defense to a charge of predatory pricing. The approach that we have outlined does not preclude the ex- amination of other evidence that could inform the court about the intent and effects of the pricing behavior, especially if the price and cost analysis turns out to be ambiguous. Although we are sympathetic to the concerns that have been expressed about the use of internal memoranda and statements of executives to show "intent,"91 we do not believe that such information is useless. However, the use of such information should be structured around two basic questions: 1) Does the documentary evidence clearly indicate that the firm with monopoly power plans to increase its prices once the competition has been driven from the market? 2) Does the documentary evidence clearly indicate an effort to use price cuts as a mechanism to increase artificially the difficulty of entering the market? We are not here referring to random documents indicating a desire to "crush" or "kill" competitors. Rather we are looking for carefully constructed long-run plans to maintain monopoly power by reducing the number of firms in the market and making the entry of new firms more difficult. We doubt that such clear tracks will frequently be left by a dominant firm, but if they are, the courts should not ignore them.92 91. See POSNER Boox, supra note 1, at 189-90 ("availability of evidence of improper intent is often a function of luck and of the defendant's legal sophistication, not of the underlying reality"; company with "executives sensitized to antitrust problems will not leave any documentary trail of improper intent"). 92. The inquiry we have described as constituting the second stage of our two-tier analysis is directed solely at the pricing behavior of the defendant firm since price setting has been the focus of our discussion. As we remarked earlier though, pricing policy is only one instrument of strategic behavior that can be used to establish and maintain monopoly power. As both we and Williamson have emphasized, each predatory pricing standard will engender a strategic adaptation by the dominant firm, which must be con- sidered when evaluating the costs of implementing a particular policy. Allegations of predatory pricing have often been accompanied by charges that the defendant firm has also engaged in other predatory activities of a nonprice nature. The latter have included, for example, "targeted" advertising expenditures, "false" product announcements, and product "manipulations." Increased advertising effort, new product announcements, and product modifications can be nonpredatory responses to competitive entry. But there exists no simple test to determine, as a general matter, whether they promote the interests of consumers and yield true social benefits or merely make it difficult for a new entrant to appear or to 259 The Yale Law Journal Vol. 89: 213, 1979 defendant satisfies the burden of production, the plaintiff would bear the burden of proving the price cut was predatory. IV. The Two-Tier Approach in Context A. Advantage of the Two-Tier Inquiry The two-tier nature of the proposed approach to predatory pricing allows resources for antitrust enforcement to be targeted where they will produce the greatest social efficiency gains. The sequential aspect of the examination is at least as important, however, because it avoids the kind of questionable outcomes produced by an inquiry that focuses only on behavior without first considering issues of market structure Trade Commission found eleven Japanese manufacturers to have engaged in the unfair competitive practice of pricing below average variable cost. See pp. 263-64 infra (dis- cussing substance of allegation of predatory pricing). The plurality opinion of Commissioners Minchew, Moore, and Alberger and the con- curring opinion of Commissioner Ablondi emphasized the respondents' lack of coopera- tion and failure to participate in the investigation. In particular, average variable cost data for the Japanese producers were not made available to the hearing officer. As a result, unaudited calculations of average variable costs of domestic producers were averaged and that average was imputed to the respondent manufacturers and used as a proxy for the respondents' costs. Furthermore, comprehensive transaction price data were unavailable for the particular firms accused of anticompetitive behavior. The Commission staff compiled the basic evidence on prices from questionnaires sent to domestic distributors and domestic importers of welded stainless steel pipe and tube; the resulting price data pertained to the lowest price each domestic distributor paid per quarter for each type of product and could not always be reliably attributed to a given foreign manufacturer. See I.T.C. Steel Case, supra, Plurality Opinion at 25 ("many importers received sales from two or more foreign sources, and .. . the importers were only required to report the names of suppliers, not their prices.'). The Commission attempted, on the basis of the secondary price and cost information, "to attribute the lowest sale price in each quarter by each importer to the imputed costs of production of this article." Id. at 25-26. Because the prices so derived were found to be below the imputed average variable cost in a large number of cases, the Commission held that eleven of the respondents had engaged in unfair methods of competition or unfair acts. Since the respondents did not offer any evidence of a plausible defense for their practices, the Commission concluded that the rebuttable presumption of predatory intent had to be applied. Id. at 33. The dissenting commissioners, Parker and Bedell, concluded to the contrary that "the evidence in this investigation does not contain adequate proof that the importation and sale of welded stainless steel pipe and tube were made by any respondents at prices which were below their respective average variable costs of production over a sustained period." I.T.C. Steel Case, supra, Dissenting Opinion at 5. While the commissioners on the plurality opinion defended their use of the secondary evidence, they did emphasize their inference from the respondents' failure to participate "that facts might have been produced that would have been adverse to them on the issue of unfair methods or acts." I.T.C. Steel Case, supra, Plurality Opinion at 26. In his concurring opinion, Com- missioner Ablondi also decried the nature of the secondary evidence. Rather than relying on such infirm data, he argued for the more direct conclusion "that the [respondents'] failure to comply with discovery in this investigation justifies a presumption of violation, and that is a basis for my decision." I.T.C. Steel Case, supra, Concurring Opinion at 6. Predatory Pricing and monopoly power. A recent decision illustrates that such outcomes can emerge even if the behavioral rule applied is as simple as the "pricing below average variable cost" version of the Areeda-Turner standard. In a recent case before the United States International Trade Com- mission, 94 eleven Japanese manufacturers were found to have engaged in the unfair competitive practice of predatory pricing for selling cer- tain welded stainless steel pipe and tube "at prices lower than the average variable cost of production ... without commercial justifica- tion."0 5 The Commission relied heavily on the Areeda-Turner rule, using average variable cost as a proxy for reasonably anticipated marginal cost. Neither the presiding officer nor the Commission itself adopted a per se rule, which would have held that sales below average variable cost are conclusively predatory.96 Rather, they found that although sales below average variable cost raised a strong presumption of preda- tory intent, the presumption should be rebuttable.97 The contention in the case was that some respondents had unilat- erally set their prices below reasonably anticipated marginal costs. No evidence was presented to indicate that there was a monopoly problem in the industry. Indeed, the plurality opinion held that application of the below-average variable cost rule to raise a rebuttable presumption of predatory intent does not depend on a showing of monopoly power.98 In fact, the large number of firms supplying the United States market combined with the fact that the presiding officer dropped all allega- tions that the Japanese firms had engaged in joint action, combination, contract, or conspiracy99 in restraint of trade, suggests that there prob- ably was not a monopoly problem. The Commission acknowledged that imports, when aggregated, con- stituted a relatively small percentage of the domestic market. 100 It in- dicated, however, that this fact was "unimportant" because imports 94. I.T.C. Steel Case, supra note 93. 95. Id., Plurality Opinion at 1. 96. Id. at 21. 97. See id. at 21-22. Three of the six commissioners indicated that "there are too many economic variables in the steel industry to adopt a per se rule." Id. at 22. Furthermore, they found that there is a "significantly greater number of plausible justifications for pricing between ATC and AVC than for pricing below AVC," and hence "sales above AVC but below average total costs must be supported by evidence of subjective intent before this Commission can find them to be unfair within the meaning of Section 337." Id. at 22-23. 98. See id. at 23. 99. See id. at 18. 100. See id. at 37. 263 The Yale Law Journal were significant in the maintenance of competition in the American market.1 1 The Commission concluded that selling at prices below average variable cost, as it found certain Japanese firms had done, was "an unfair act which has the tendency to restrain trade and commerce in the United States by substantially reducing the domestic market share of other foreign competitors."'' 0 2 The fact that all Japanese welded stainless steel pipe and tube imports (not just those of the "predators") had risen to eighty-seven percent of total imports in 1976 from seventy percent in 1974 was crucial to the decision, even though, as the Commission acknowledged, this was a very small proportion of the total American market. 0 3 Furthermore, the decision did not cite any evidence that conditions of entry in this industry were con- ducive to monopoly. The plurality opinion dismissed the Justice Department's argument that unilateral below-cost selling by a nondominant firm is not an "unfair method of competition"'1 4 encompassed by the section of the legislation under which the case was brought. The three commissioners also found without merit the Department's contentions that imports had had a restraining effect on price, and that, therefore, excluding Japanese imports would raise consumer prices and not promote price competition. 10 5 They applied the average variable cost rule and de- termined that eleven of the respondent Japanese firms had engaged in predatory pricing, an unfair method of competition. The Commission's decision appears to have protected particular competitors at the expense of competition. The beneficiaries of the cease and desist order issued by the Commission would be not only other foreign producers but also domestic producers. The restrictive impact of the decision on foreign competition, which the Justice De- partment had argued that it would have, only could have enhanced any market power domestic producers already enjoyed. They, not consumers, were the beneficiaries. Thus, even the simple average variable cost version of the Areeda-Turner rule can be misused if an effort is not made initially to determine whether the structural char- acteristics of the firms and markets involved generate a reasonable ex- pectation that monopoly power has been or could be exercised by the alleged predator. Applying our two-tier approach, the case, as presented in the Com- 101. See id. 102. Id. at 39 (emphasis supplied). 103. id. at 36-37. 104. Id. at 17 n.1. 105. Id. at 46. 264 Vol. 89: 213, 1979 Predatory Pricing mizing price, when barriers to entry are high.1 14 At the same time, the court indicated clearly that it viewed the high-entry-barriers alternative as an exception to the general standard of below-cost pricing rather than as an independent test."15 A similar exception to the Areeda-Turner formulation has emerged in the Ninth Circuit. In Hanson v. Shell Oil Co.," 6 the court cited the International Air Industries discussion of entry barriers as support for its own similar dual test of predation.z17 And later, the district court in ILC Peripherals Leasing Corporation v. International Business Machines Corporation ("Memorex")" 8 interpreted Hanson as estab- lishing the same two tests for predation as International Air Indus- tries had. 1 9 But the Memorex court stated, The second test for predatory pricing should be applied only in the limited circumstances described above [extremely high entry barriers], and should probably be considered an exception to the marginal or average variable cost test rather than an independent test itself.120 An example of an approach to a predatory pricing case that con- siders the kinds of structural characteristics we have emphasized is the concurring opinion by Federal Trade Commission Commissioner Pitofsky in Borden, Inc., Final Order to Cease and Desist.'2' Pitofsky argued that in the presence of certain market-specific and firm-specific 114. The International Air court stated its test explicitly in the form of two alternative sufficient conditions: [I]n order to prevail as a matter of law, a plaintiff must at least show that either (1) a competitor is charging a price below his average variable cost in the competitive market or (2) the competitor is charging a price below its short-run, profit-maximizing price and barriers to entry are great enough to enable the [competitor] to reap the benefits of predation before new entry is possible. Id. (footnotes omitted). 115. See id. at 724 n.31. 116. 541 F.2d 1352 (9th Cir. 1976), cert. denied, 429 U.S. 1074 (1977). 117. See id. at 1358 & n.5. 118. 458 F. Supp. 423 (N.D. Cal. 1978). 119. See id. at 431-32. 120. Id. at 432. The Memorex court found support for its view that "the exception for a market with high barriers to entry is still recognized," id., in the Ninth Circuit opinion in Janich Bros. v. American Distilling Co., 570 F.2d 848, 856-58 (9th Cir. 1977), cert. denied, 439 U.S. 829 (1978). In the latter case, although adopting an average variable cost rule, the court stated that, "As implied in Hanson, 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 supplied) (footnote omitted). 121. [1978] TRADE REG. R P. (CCH) 21,490 at 21,517. The Commission's initial deci- sion in the Borden case was analyzed extensively and cogently by Professor Schmalensee. See Schmalensee, supra note 3, at 998-1043. Our discussion focuses on the Commission's final decision. The Yale Law Journal structural features, a rule that draws the line between reasonable and unreasonably exclusionary pricing at average variable cost is defi- cient.1 22 He argued that because of its past expenditures on image advertising and promotion, Borden enjoyed a pronounced consumer preference and brand loyalty advantage that would not be taken into account by an average variable cost rule. 123 Pitofsky proposed instead that Borden should have been allowed to reduce its price only to average total cost because "[u]nder this standard, the monopolist would be able to respond selectively to competitive challenges down to its level of average full cost, but ordinarily would not be able to drive an equally efficient challenger out of the market."' 24 The point we wish to emphasize about Pitofsky's analysis is not the particular rule he applied in this case but his focus on structural fea- tures of the market in formulating that rule. Pitofsky explained that he was departing from the Areeda-Turner approach because "pro- nounced consumer brand loyalty is a barrier to entry,"'125 that may enable dominant firms to engage in predatory pricing without setting price below average variable cost. Although the cases just discussed evidence concern about the struc- tural characteristics on which the first tier of our approach focuses, the approach we propose differs in an important way from the analyses in these cases. The opinions we have cited view the issue of entry barriers as arising only in exceptional cases and, therefore, give it secondary consideration. In sharp contrast, under our approach an evaluation of structural characteristics is essential to a determination of whether monopoly power exists. Every case would begin with an inquiry into the structural characteristics of the defendant firm and the market in which it operates that would be broader in scope than the kind of examination undertaken in the cases mentioned above. Evidence concerning the relationship between the firm's price and some measure of its costs, which is central to the general rule adopted by the courts whose opinions we have reviewed, would be introduced only in cases that reached the second tier under our approach, those in which structural characteristics could sustain an argument that a mo- nopoly problem exists in the market. In contrast, a court applying a cost-based rule with a high-entry-barrier exception would examine the relationship between the defendant firm's price and its cost in every case brought before it. Moreover, since under our approach the behav- 122. See Borden, Inc., [1978] TRADE REG. REP. (CCH) 21,490 at 21,521-22, 21,524. 123. See id. at 21,521-22. 124. See id. at 21,523. 125. See id. at 21,524. Vol. 89: 213, 1979 Predatory Pricing ioral analysis is undertaken only when structural features indicate that monopoly conditions are of genuine concern, neither the average variable cost standard nor the short-run marginal cost standard is ap- propriate for deciding whether pricing behavior has been predatory. Instead, as we have indicated, a much more expansive rule-of-reason approach to pricing behavior is appropriate for structuring the second- tier behavioral inquiry. 126 The decisions in which courts have adopted the average variable cost standard, noted the high-entry-barriers "exception," and concluded that it did not pertain to the specific case, would have looked quite different under our approach. In each such case, either the structural preconditions for a proper showing of monopoly power were lacking or the court erred in its conclusion about the importance of entry barriers. In the absence of difficult entry conditions or similarly con- ducive structural features of the market, long-run monopoly power is unlikely. Hence, predatory pricing would be irrational and thus im- probable. Therefore, if there were not significant entry barriers, and no other structural evidence was presented that pointed to a monopoly problem, the case should have been dismissed on structural grounds alone. If, on the other hand, the market's structural characteristics implied the existence or prospect of long-run monopoly power, then the exception was relevant, and the court should have looked beyond average variable cost in deciding the issue.127 126. While the more expansive rule-of-reason analysis of pricing behavior in the second stage of our approach necessarily entails a more detailed and more resource- consuming investigation than would application of a per se cost-based rule, for the reasons discussed above, see pp. 238-39, 242-43 supra, we believe that such an inquiry is re- quired. Cf. ANTITRUsT COMMISSION REPORT, supra note 20, at 143 ("some degree of com- plexity in antitrust litigation, where it is due to the generality or breadth of the ap- plicable legal standards, in many cases may be necessary and desirable"). It is interesting to note a parallel between our proposed approach to predatory pricing cases and the approach Professor Richard Posner suggests for determining the legality of restrictions on distribution imposed by a producer. Motivated by Continental T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36 (1977), in which the Supreme Court rejected the rule that nonprice restrictions on dealer competition are illegal per se when imposed in sales contracts, Posner proposes a three-stage inquiry for determining the legality of both price and nonprice restrictions on distribution. See Posner, supra note 54, at 20. In the first stage of his inquiry Posner would ask whether a restriction embraces "so large a fraction of the market as to make cartelization a plausible motivation for the restriction." Id. at 19. If not, Posner believes the restriction should be held lawful without proceeding to the other two stages where specific scrutiny of firm behavior would take place. Id. We do not wish to enter into a discussion of the merits of Posner's proposal but want to point out that it, like our proposed approach, places the question of structural characteristics first. The inquiry moves on to a more detailed examination of firm behavior and its consequences only if the structural characteristics generate, in our terms, a reasonable expectation that cartelization is a possible motivation. 127. As a final note, our approach is consistent with the view of the National Com- 269