Printer Friendly

Predatory pricing, the theory of the firm, and the recoupment test: an examination of recent developments in Canadian predatory pricing law.


Federal antitrust law in the United States on predatory pricing is clear. The U.S. Supreme Court set out a two-stage test for determining predation in Brooke Group Ltd. v. Brown & Williamson Tobacco Corp. (1) First, it must be shown that the alleged predator priced below an appropriate measure of cost. (2) Second, under the recoupment aspect of the test, it must be demonstrated that there is a reasonable prospect that the alleged predator would be able to recoup its investment in below-cost prices. (3) The plaintiff must show that "there is a likelihood that the predatory scheme alleged would cause a rise in prices above a competitive level that would be sufficient to compensate [the alleged predator] for the amounts expended on predation, including the time value of the money invested in it." (4) The Court offered examples of where predation would be unlikely to lead to recoupment: where barriers to entry are low, or where the market structure is diffuse and competitive. (5) In Brooke Group itself, the Court found that recoupment, and therefore predatory pricing, was implausible given the oligopolistic nature of the market. (6) While the oligopoly as a group may have had an incentive to engage in predatory pricing, no individual oligopolist, in the Court's view, would have profited from doing so.

This article examines the rationale for the recoupment test. (7) The analysis is motivated primarily by recent developments that have left Canadian competition law on predation generally, and recoupment specifically, in flux. In Canada in 1992, the Director of Investigation and Research of the Canadian Bureau of Competition Policy (the Bureau, now the Commissioner of Competition) adopted the Predatory Pricing Enforcement Guidelines (Guidelines) which outlined his approach to enforcement of the criminal predatory pricing provision of the Competition Act, s. 50. An important element of the approach was treating recoupment as a necessary element of a predation allegation. The Guidelines state that, "In the context of a predatory pricing complaint, it is necessary to determine whether or not the alleged predator appears to have the power to recoup its initial losses by raising prices to above-normal levels once its target/rival has been driven from the market." (8) Like the U.S. Supreme Court in Brooke Group, the Guidelines focus on factors such as the alleged predator's market power, as indicated by considerations like market share and entry barriers, to help guide the recoupment analysis.

The necessity of the recoupment test has never been the subject of judicial determination in Canada, and thus has never been the law of Canada. (9) The Bureau took advantage of the freedom the dearth of case law provides and changed its position on the recoupment test. In 2002, the Commissioner of Competition released new draft guidelines, Enforcement Guidelines for Illegal Trade Practices: Unreasonably Low Pricing Policies Under Paragraphs 50(1)(b) and 50(1)(c) of the Competition Act (Draft Guidelines). The Draft Guidelines propose abandoning reliance on the recoupment test, suggesting that the likelihood of recoupment should be treated simply as a factor to be considered rather than a necessary element of predatory pricing. This surprising turn of events comes with very little explanation in the Draft Guidelines themselves. The Commissioner states,
   [Since the time of the original Guidelines], there have been
   changes in the economy as well as developments in economic
   thinking concerning low-pricing behavior. For this reason, the
   original guidelines have been updated to reflect a modern
   perspective on low-pricing issues. These guidelines have adopted
   three principal changes. First, the ability to recoup losses will
   no longer be considered as the primary screening criterion.... (10)

Unfortunately, the Commissioner is silent on what developments in the economy or economic thinking justify this change in direction.

Canadian competition policy has departed from the recoupment test in another significant way. In response to a merger of the two largest Canadian airlines, Air Canada and Canadian Airlines, the federal government amended the Canadian Competition Act to establish that it is an anticompetitive act for a dominant airline to set price below avoidable cost. (11) As a matter of the current statute and regulations, then, there is no recoupment test in determining whether a dominant airline has committed an anticompetitive act of predation. (12) Rather, a price-cost comparison is the sole test. A recent case, Commissioner of Competition v. Air Canada (13) applied this test and found that Air Canada had indeed committed anticompetitive acts by pricing below avoidable cost on some routes in Canada. (14)

The direction of Canadian law on predation is not clear at the moment. The Competition Bureau has not adopted the Draft Guidelines and thus they have no formal status. And in 2004, the Minister of Industry proposed amendments to the Competition Act that would have removed its airline-specific provisions. (15) If the amendment had passed, the statutory rejection of the recoupment test in airlines would no longer exist. However, the government fell before the proposals were passed, and at this moment there is no similar bill before Parliament. In announcing the proposed amendment to remove airline-specific provisions in 2004, the Minister of Industry did not denounce the provisions in substance, but rather concluded that developments in the airline industry, such as the entry and growth of airlines other than Air Canada, implied that the approach was no longer necessary. (16) This is not the same as declaring that the provisions were wrong-headed to begin with because of their failure to include a recoupment test. (17) Nor has the Competition Bureau explicitly rejected the approach in the Draft Guidelines, but rather the Bureau has simply failed to adopt them. The fate of recoupment in Canada remains unsettled.

While not yet settled, these recent developments indicate the declining influence of the recoupment test in Canadian predatory pricing law. This trend, along with the Bureau's claim without explanation that economic thinking no longer requires the recoupment test, invite an examination of the economic justification for the recoupment test. This is the central motivation for this article.

Another motivation arises from some recent work examining the empirical plausibility of predation. In the course of his extensive study of predation, Lott considers the possibility that firms might strategically precommit to engage in predation even where it is not profitable. (18) In my view, a careful examination of this possibility is useful in its own right in understanding the recoupment test and, moreover, casts doubt on some of Lott's empirical results.

A final motivation for this article is to begin to account for the theory of the firm in undertaking antitrust analysis. While it may not always be significant, the fact that firms are not simply "black boxes" that inevitably seek to maximize profits, but rather are contractual and quasi-contractual affiliations of parties whose interests may be in tension with one another, may sometimes be relevant to competition policy. In this article I bring to bear on an analysis of predation recent thinking in corporate finance on the private benefits that managers enjoy from being in charge of the firm. These insights are useful in weighing the pros and cons of the recoupment test.

The analysis of recoupment generally and the possibility of strategic commitment in particular requires a terminological clarification. The recoupment test defines "predatory pricing" by reference to the probability of future profits that such a strategy brings. If pricing below cost imposes losses on competitors as well as on the alleged predator, but the losses to the predator cannot be recouped in the relevant market through higher future prices, then the prices are not predatory under the recoupment test. Put another way, "unprofitable predatory pricing," that is, where a firm engages in below-cost pricing in order to impose losses on rivals but also incurs net losses itself, is not predatory pricing at all under the recoupment test. For the purposes of this article, however, I will use the term "unprofitable predatory pricing" even though such pricing under the recoupment test would not as a legal matter be predatory. (19)

In order to justify my conclusion that it is not good policy to downgrade recoupment as Canadian authorities have done, the article proceeds by setting out the strongest arguments that could be made in support of these changes and against recoupment. The article shows that there are significant deficiencies with the two arguments most commonly advanced in favor of reliance on recoupment. First, it is argued that where predatory pricing is unprofitable in the market in which it takes place, that is, where recoupment is not plausible, it causes no social harm and therefore should not be illegal. Part II shows that this argument is unpersuasive. Second, another approach contends that even if unprofitable predatory pricing could cause social harm, antitrust law need not penalize it because such behavior is self-deterring. Part III challenges this argument. It shows that unprofitable predatory pricing is not necessarily self-deterring. Part III shows that managers at firms may prey out of self-interest, to protect private benefits of control, for example, even if doing so is not profitable for the firm. Further, Part III shows that the firm may have strategic incentives to allow unprofitable predatory pricing to take place. In particular, I examine how the principal-agent structure of many corporations lends itself to strategic manipulation of an agent's incentives to prey in order to deter entry or encourage exit even if such predation would be unprofitable. Firms may adopt this strategy because the possibility of unprofitable predatory pricing deters entry or encourages exit. (20) This in turn suggests that unprofitable predatory pricing is not entirely self-deterring. In reaching this conclusion, part III provides criticism of Lott's analysis of precommitment to prey. A careful examination of reputational models of predation and of strategies designed to create a credible predatory threat based on reputation suggests that some of the factors Lott examines are simply not relevant. (21) Evidence on other factors he examines that are relevant, such as concentrated share ownership, supports the reputational model.

While part III shows that unprofitable predatory pricing is not necessarily self-deterring, which is the strongest argument one could make against the recoupment test, part III nevertheless concludes that analysis does not support downgrading the importance of the test as Canadian guidelines and legislation have done. Instead, my analysis offers a new interpretation of the recoupment test: not only is profitable predation unlikely if recoupment is implausible, but in most such cases unprofitable predatory pricing is also unlikely. Factors relevant to recoupment, like the competitive nature of the market, are also relevant to the probability of unprofitable predatory pricing. Thus, the recoupment test not only explicitly evaluates the prospect of profitable predatory pricing, it also implicitly addresses the prospect of unprofitable predatory pricing.

While the analysis in parts II and III involves consideration of the possibility that downgrading reliance on recoupment creates underdeterrence, part IV reviews issues relating to overdeterrence. The analysis in part HI suggests that relying on recoupment will not create large problems of underdeterrence, while the analysis of part W shows that there would be significant concerns about overdeterrence if recoupment were abandoned or marginalized in importance.

In summary, the article provides a possible economic theory for downgrading recoupment in enforcing predatory pricing laws in that it shows why unprofitable predatory pricing can be plausible. The article nevertheless fails to endorse such a move. Recoupment obviously and explicitly captures profitable predatory pricing, but implicitly captures most instances of unprofitable predatory pricing as well. The inclusion of a recoupment requirement largely deters profitable and unprofitable predatory pricing, while avoiding significant problems of overdeterrence that would arise in its absence. Canadian authorities would be wise to return to their former reliance on the recoupment test.


Consider a situation in which there are no possible strategic advantages from preying. For example, in the situation in which entry has already occurred and there are no other potential entrants, deterring entry is pointless, and engaging in predation is unprofitable for the incumbent. In such circumstances, the firm will lose money if it preys. Setting aside for the moment the incentives to engage in unprofitable predatory pricing, one argument supporting the recoupment test is that since unprofitable predatory pricing of this kind does no social harm, there is no need to condemn it. The U.S. Supreme Court adverts to this reasoning in its Brooke Group opinion, noting that while there may be "some inefficient substitution toward the product being sold at less than its cost, unsuccessful predation is in general a boon to consumers." (22) Other courts have gone further, stating for example that "[p]redatory pricing schemes that fail at the recoupment stage may injure specific competitors ..., but do not injure competition (i.e., they do not injure consumers) and so produce no antitrust injury.... Such futile below-cost pricing effectively bestows a gift on consumers, and the Sherman Act does not condemn such inadvertent charity." (23) Another court contends that "[p]redatory pricing is only harmful when the predator succeeds in recouping the losses it suffered by its earlier below-cost pricing." (24) Hunter and Hutton echo this approach, stating that "without ... supra-competitive prices, there can be no harm to society, as opposed to competitors and the alleged predator itself." (25)

There are, however, at least two types of societal harm that may result from unprofitable predatory pricing. First, below-cost pricing, whether or not it results in recoupment, itself creates a social deadweight loss. (26) As Brooke Group acknowledges, but Hunter and Hutton do not, below-cost pricing engenders an inefficient use of resources. If price is set below cost, some purchases will be made by consumers who realize a positive gain at the below-cost price, but who would not purchase if the unit were sold at marginal cost. Given that the social costs of supplying this group of consumers exceed the social benefits, there is a deadweight loss from such pricing. Some resources consumed in supplying the product at predatory prices would be better deployed in supplying other products; in this sense, other consumers are harmed by below-cost prices in the predatory sector. (27) During the period of predatory prices, there is a social loss whether or not recoupment is feasible.

There is a second potential societal harm from unsuccessful predatory pricing. Brooke Group states that even if predatory pricing were likely to eliminate a competitor, "there is still the further question whether it would likely injure competition in the relevant market." (28) To address this question, the Court states that the plaintiff must demonstrate that recoupment is likely. (29)

The relationship between recoupment and harm to competition, in the sense of harm to social welfare rather than to a particular competitor's welfare, is not as strong as the Court implies. Clearly, if recoupment is likely, social harm would result from supra-competitive prices in the future. But, while it may be sufficient, recoupment is not a necessary condition for these social harms. It may be, for example, that predatory pricing is successful in encouraging exit from the market or discouraging entry into the market, and therefore there will be a period during which prices may be set at supra-competitive levels, yet the profits from predatory pricing do not exceed the costs to the predator. For example, barriers to entry may be low enough to allow entry in the face of supra-competitive prices such that full recoupment is unlikely, but entry may take time, during which time the predator can reap supra-competitive profits. The fact that these profits do not exceed the costs of predatory pricing does not imply that there are no social harms during the period of supra-competitive pricing.

Unprofitable predatory pricing may be socially harmful for these reasons. During the period of low prices, there is a deadweight loss from below-cost pricing, and to the extent that the pricing affects competition, there may be deadweight losses from supra-competitive prices, whether or not full recoupment is likely.


Commentators have generally not ignored the social costs entailed by unprofitable predatory pricing. The reason why antitrust law should not intervene in the case of unprofitable predatory pricing, they argue, is not that there are no social costs, but rather that the firm is punished by the market for engaging in such activity. By definition, unprofitable predatory pricing is costly to the firm and is something the firm will seek to avoid. As Easterbrook puts it, "It would be foolish to devote additional resources to preventing conduct that penalizes itself." (30) I show in this section that unprofitable predatory pricing is not necessarily self-deterring, but that nevertheless, predatory pricing law that relies on a recoupment test will indeed largely deter unprofitable predatory pricing as well as profitable predatory pricing.

A. Predation and private benefits of control

Again, consider a situation in which there are no possible strategic advantages from preying; for example, entry has already occurred, there are no other potential entrants, and engaging in predation is unprofitable for the incumbent. Predation may occur in these circumstances because of agency problems within the firm.

Antitrust policy and commentary almost always treat firms as profit-maximizing "black boxes." If a firm seeks only to maximize profits, it is clearly correct to observe that unprofitable, nonstrategic predatory pricing will be self-deterring. By definition, a profit maximizer does not engage in an unprofitable activity, at least if it can foresee that the activity will be unprofitable. As principal-agent theory suggests, however, the firm is not simply a machine unrelenting in its quest for profits, but rather comprises a variety of actors with potentially differing objectives. (31) Shareholders are likely to seek to maximize profits and the value of the firm, but their agents, who manage the firm on a day to day basis, may have a variety of non-profit-maximizing objectives. Managers may seek leisure, (32) material perquisites like executive jets, (33) prestige from "empire-building" (34) or even harm to rival managers they personally dislike. (35) Managers realize private benefits, either pecuniary or nonpecuniary, from controlling a particular firm. These factors may suggest that, all things equal, a manager may prefer to prey on an entrant than to acquiesce and share the market. If a manager acquiesces and shares the market with the recent entrant, for example, sales will drop at her firm leading to a possible decline in compensation, since company size and pay are correlated. Her prestige as a "captain of industry" may fall as well. Dislike of rivals may also lead to predation if it is likely to harm rival managers, perhaps by causing the rival's shareholders to wind up their firm. Managers may also act in ways that are economically "irrational," but are consistent with observed behavioral regularities. For example, Tor suggests that loss-averse managers may invest in a predation strategy, even if it is likely to fail, in order to avoid losses resulting from increased competition. (36)

More generally, it is reasonable to expect that the private benefits of control will tend to be correlated with the potential profit of the firm as indicated by its market power. Consider a perfectly competitive firm. Any managerial choice that fails to maximize profits risks the failure of the firm, since in perfect competition if marginal costs at one firm are not identical to that at others, the firm will fail. Only with some ability to raise price above marginal cost will a manager be able to make choices based on maximizing the private benefits of control. Cost-ineffective private jets, for example, are easier for a firm to absorb without failing the greater the firm's market power. (37)

There are, of course, sources of discipline on management other than product market competition. Direct shareholder supervision, (38) executive compensation, (39) managerial labor markets, (40) and capital markets--particularly the market for corporate control (41)--may also serve to align managerial and shareholder interests. This does not undermine the correlation between private benefits of control and market power. Some market power is a necessary condition for managerial ability to make choices furthering the private benefits of control, even if it may not be sufficient. If other sources of discipline, such as the market for corporate control, are imperfect, then it will follow that market power will facilitate managerial choices made to realize private benefits of control.

The correlation between market power and private benefits of control suggests that managers realize a greater share of the benefits of market power than do shareholders. This consideration alone may be sufficient to create a private incentive for managers to engage in predation even if it is not profitable for the firm generally. Consider the following example. Suppose a monopoly exists but new entry would create something close to a perfectly competitive market because of post-entry Bertrand (price) competition. This means that the private benefits of control, through perquisites like jets for example, will be limited (though not zero) if the incumbent firm acquiesces (does not prey) and shares the market. On the other hand, if the incumbent preys, suppose that the entrant, which depends on external financing, is driven from the market because its lender, unaware that losses were caused by predation rather than an unprofitable market, refuses to extend further credit. (42) Once exit occurs, the managers of the incumbent will resume enjoying private benefits of control.

Even if the expected losses incurred by the predator exceed expected gains from future market power, predation may be plausible. This is because managers may realize a disproportionate share of future profits from market power through their enjoyment of private benefits of control and are thus willing to incur overall losses. If acquiescence in the successful entry of the entrant eliminated private benefits of control because of product market discipline, a manager may be willing to have the firm take a net loss to preserve these benefits. (43) While full recoupment may not be feasible in a particular case, predation may be plausible because of managers' seeking to protect the private benefits of control.

It is worth noting that corporate law fails to address this type of predatory pricing. Corporate officers are subject to fiduciary duties that require them to act in the best interests of the corporation. (44) Unprofitable predatory pricing violates this duty. However, in managing the company, officers make decisions that are insulated from judicial scrutiny by the "business judgment rule," which establishes a presumption that, in the absence of evidence of pecuniary self-interest or gross negligence, a managerial decision does not violate fiduciary duties. (45) Pricing decisions are unlikely to involve either gross negligence or an obvious pecuniary conflict of interest, (46) and thus are likely to be immune from scrutiny under corporate fiduciary law. (47)

Setting aside strategic considerations that I will address below, faced with the potential that managers will engage in unprofitable predatory pricing, entrepreneurs going public and concerned about maximizing returns have an incentive to address this potential agency problem, perhaps by choosing an ownership structure that lends itself to the threat of takeover; (48) that is, one in which the manager is threatened with ouster by an acquirer. The firm's owners, however, will not have the same incentives to minimize the risk of unprofitable predatory pricing as would a planner seeking to maximize social welfare. The social planner would account for the losses that the firm incurs by pricing below cost, as would the firm, but the social planner would also account for the social losses resulting from any period of supra-competitive prices (even if these prices do not make up for losses from predation), while the firm would not. For example, suppose predatory pricing in a particular market would succeed both in driving out a rival and in giving rise to supra-competitive pricing, but would not be profitable overall; the initial losses exceed the supra-competitive profits. The firm would account only for the overall effect on profit in minimizing agency costs, while the social planner would account also for the social costs of supra-competitive prices. At the margin, the firm and the social planner take different attitudes toward the investment that should be made to minimize agency problems and the possibility of unprofitable predatory pricing.

This analysis suggests that if antitrust law were to penalize predatory pricing even where unprofitable, the firm would have an added reason to ensure that the incentives of management are aligned with the profit-maximizing motives of shareholders. (49) An analogy can be drawn to reckless driving. Reckless driving is to some extent self-deterring, since the reckless driver endangers herself. However, the driver may not account for harm to others, and this is why we criminalize such behavior.

While ostensibly addressed only to the question of whether predatory pricing could be profitable, there is an important respect in which the recoupment test nevertheless addresses the agency cost type of predatory pricing. A significant constraint on suboptimal agency behavior is competition in the product market. Where markets are competitive, managerial misbehavior may imperil the firm's existence. Consequently, executives concerned about their jobs and reputations in the managerial market will have less latitude in competitive markets to make suboptimal, self-interested choices. Moreover, managers in competitive markets will have smaller private benefits to protect by preying on rivals.

In setting out situations in which recoupment would be unlikely, the Supreme Court in Brooke Group aptly included a situation where "the market is highly diffuse and competitive...." (50) The original Canadian Guidelines also emphasized the importance of market power in establishing recoupment. Where market power is absent, not only is predatory pricing very unlikely to be profitable, it is also unlikely to take place if it is unprofitable. Management in a competitive market will not engage in predatory pricing, even if it would advance certain personal interests, such as animus towards a competitor, because of the risk of the failure of the would-be predator. While I have shown that agency problems could motivate unprofitable predatory pricing, this conclusion does not undermine the relevance of the recoupment test, but rather offers an additional rationale for it: the recoupment test implicitly addresses the prospect of unprofitable predatory pricing by self-interested corporate managers. (51)

B. Unprofitable predatory pricing and strategic considerations

1. REPUTATION There is another important situation in which unprofitable predatory pricing may cause social harm and may not be self-deterring: unprofitable predatory pricing in one market may have socially harmful effects on other markets. Specifically, if predatory pricing creates a reputation for the firm, this reputation may result in supra-competitive profits in other markets in which the firm participates. (52) Hence, unprofitable predatory pricing in a particular market may cause social harm by creating or preserving monopoly profits in other markets, with the social deadweight losses these entail.

The reputational theories of predation can be thought of in at least two ways. One involves a single market in which a predator seeks to encourage exit and to deter potential entrants in future periods. A broad view of the recoupment test accounts for this possibility, given that the predator would recoup its investment in predatory pricing in the particular market in question by establishing a reputation that preserves market power in the future. Indeed, the Canadian Guidelines noted that predation may create barriers to entry by creating a reputation for predation. (53) A second version is that a predator may seek to encourage exit and deter entry in a variety of markets in which it participates at a particular point in time. The Draft Guidelines advert to the effect of multimarket competition and reputation on incentives to prey, (54) though there is some controversy in the United States about whether the Brooke Group test accounts for this motivation for predation, with some courts expressing skepticism about the role of reputation in supporting a predatory pricing allegation. (55)

While legal commentators have noted that reputation may be important in making predatory pricing that is unprofitable in one market a profitable strategy overall, they have tended not to examine closely existing economic theories of reputation and predation. In my view, a careful review of economic theory in this area provides additional insights into the robust response of the recoupment test to both profitable and unprofitable predatory pricing. In the remainder of this section I review the game-theoretic foundation for the potential significance of reputation in analyzing predatory pricing. This foundation will prove useful in discussing the firm's incentives to deter future unprofitable predatory pricing.

Predation for reputation requires a dynamic, or multiperiod, game. In a one-shot game, unprofitable predatory pricing cannot create a profitable reputation because there is no future in which reputation will be valuable. In a multiperiod game, there are two general types of game in which reputation may be profitable for the firm. First, the game may be infinitely repeated. Second, the game may be repeated, but only a finite number of times. The theoretical foundation for reputation as a profitable strategy is different in each case.

With respect to infinitely repeated games, it is not difficult to show that a profit-maximizing firm may engage in predation. Suppose there is an incumbent monopolist who faces an infinite number of potential entrants who would enter sequentially the markets in which it participates. (56) If the entrant enters, the incumbent, if it chooses to prey, loses money for a period but then earns monopoly profits each following period. If it chooses to acquiesce, it never loses money, but makes duopoly profits (which are lower than monopoly profits) each period into the future. Assume also that it is more profitable for the potential entrant to stay out than to enter and be preyed upon, although if the incumbent acquiesces it is more profitable to enter than to stay out.

In this situation, the following is a possible equilibrium. The prospective entrant adopts the following strategy: if the incumbent has preyed in every previous market in which entry has taken place, I will not enter; if the incumbent has ever acquiesced, I will enter. The incumbent adopts the following strategy: if I have never acquiesced and entry occurs, I will prey; if I have ever acquiesced and entry occurs in market n, I will acquiesce in that market. These strategies form a Nash equilibrium: they are the best strategy each player could make given the other's strategy. Moreover, the strategies remain optimal at every point in time; in technical terms, they form a subgame perfect equilibrium.

In the case of an infinitely repeated entry game, concerns about reputation may lead a rational incumbent to threaten credibly to prey in a given period, or even in a given series of periods, even though it is unprofitable to do so in that given period. Reputation is relevant in the following sense: if ever the incumbent acquiesces, entry will occur thereafter; thus its concern about deterring entry will lead it always to prey rather than acquiesce.

There are two problems with such a model as a predictor of what will occur in practice. First, it may not be plausible to imagine an infinite number of potential entrants, or even sufficient uncertainty about the number of potential entrants. (57) Second, in the infinitely repeated game described above, there are an unlimited number of equilibria. For example, the following is also an equilibrium. The potential entrant could adopt the strategy: always enter regardless of history. The incumbent could adopt the strategy: always acquiesce. Again, this is a Nash equilibrium in that each strategy is a best response to the other strategy. Moreover, it is rational at any point in time to adhere to this strategy. While an infinitely repeated game may give rise to predation, it may give rise to any number of equilibria.

Now consider the alternative situation in which the potential predator participates in finite repetitions of a game. Assume that the payoffs for the incumbent are the same as those indicated above in the situation where an incumbent faces entry: the incumbent is best off in a particular round if the entrant does not enter, is worst off if it preys, and achieves a middle ground if it acquiesces. The entrant makes a positive profit if it enters and the incumbent acquiesces, but loses money if the incumbent preys.

Suppose now that rather than playing the game an infinite number of times, this game is played a large, finite number of times; think of the incumbent as a firm which is a monopolist in many different geographic markets and faces potential entry in each one. A strong intuition is that reputation may still be important. The incumbent may wish to prey facing the first couple of entrants in different markets in order to develop a reputation useful for deterring entry. However, as Selten points out, this intuition is incorrect, at least where there is full information. (58)

Selten's approach is to follow a process of backwards induction. In the last period, the incumbent will not prey. It is more profitable to accommodate than to prey in any given period. In the last period there is no future round for which a reputation is valuable, hence the incumbent should simply accommodate. Entry will therefore occur in the last round. But now consider the second-to-last round. No matter what the incumbent does in this penultimate round, it will acquiesce in the last round. Therefore, there is no reputational effect from preying in the penultimate round. Given the absence of reputational effects, the potential entrant will enter in the penultimate round, and so on. The only solution to this finitely repeated game is one in which the entrant enters in every market and the incumbent accommodates in every market.

The irrelevance of reputation in what could be a game of very many rounds seems counter-intuitive--indeed, Selten refers to the result as the "chain-store paradox." It turns out, however, that the result depends crucially on the assumption of full information. Kreps and Wilson (59) and Milgrom and Roberts (60) show that a small probability that the payoffs are not as they are described above may lead to rational predation even in a finitely repeated game.

Suppose there is a small, exogenous probability that, rather than preferring to accommodate than to prey, the incumbent prefers to prey than to accommodate (assume that there is a small probability the incumbent is "irrational"). Referring to the above discussions of agency costs, I will discuss interpretations of this probability below. Kreps and Wilson and Milgrom and Roberts show that this small probability may lead to predation even by a rational incumbent. The intuition is as follows. Potential entrants are aware that there is an exogenous, small probability that the incumbent will prey whether predation is a money-loser or not. The incumbent knows that if it does not prey in a given period, all future entrants will know that it is not an "irrational" predator. Therefore, it has an interest in preying in early periods in order to preserve, and indeed enhance, the perceived probability that it is irrational. Predation becomes a rational strategy even for rational incumbents as a result of the existence of a small initial possibility that the incumbent is irrational.

In both the infinitely repeated game, and the finitely repeated game with imperfect information, predation may be a rational strategy because of reputation. Even if predation is unprofitable in a particular market, it may be profitable overall for the predator.


(a) Strategic (mis)alignment of managerial and shareholder incentives When predatory pricing results from agency problems and has no strategic effects, firm ownership may have suboptimal incentives to address it, as discussed. But if predatory pricing has strategic effects, the firm has an incentive to encourage it to some extent, rather than deter it. First, I consider infinitely repeated games, then finitely repeated games.

With respect to infinitely repeated games, a profit-maximizing firm may nevertheless prey in a particular market, even if it is unprofitable to do so in that market, because of the value of reputation. As shown above, an infinitely repeated game may result in a firm's deterring entry with credible threats of predation, depending on the incumbent's and potential entrants' strategies. In such circumstances, predatory pricing may be profitable for the firm overall, and thus the firm will not avoid embracing such strategies. In particular, the firm will seek to adopt measures, such as incentive pay, that motivate managers to seek to maximize profits. In infinitely repeated games, these incentives will encourage predation.

With respect to finitely repeated games, it was shown above that where there is perfect information and all parties are profit-maximizers, predation will not be profitable since backwards induction leaves reputation irrelevant. However, where there is some exogenous probability that the firm is "irrational," Kreps and Wilson and Milgrom and Roberts show that it may be profitable for the firm to engage in predation that is unprofitable in a particular market in order to develop a reputation that deters entry. But for this model to have any relevance, there must be some reason why the potential entrants believe there to be a possibility that the firm will behave "irrationally." There is an ad hoc flavor to such an assertion. (61) In this section, I will build on Milgrom and Roberts" suggestion about one of the reasons for the possibility of irrationality: "the fact that firms involve many individuals, each with his or her own preferences and information, suggests that the appropriate model of the firm would be one of group decision making, and there is no compelling reason for choices in such situations to correspond to the maximization of a single utility function." (62) Agency relationships may affect the "rationality" of the firm's behavior.

The higher the perceived probability of "irrationality," the greater the probability that even a rational incumbent will prey in order to preserve and enhance the perception that it is irrational. The following factors will contribute to the entrant's assessment of the probability that the incumbent would prey regardless of profit. First, there are the preferences of the managers and the private benefits of control. Obviously, it is difficult for outsiders to know exactly what a manager's preferences are, but there may be a sense of them. For example, a particular manager may be known to enjoy being the "only game in town", or being a "captain of industry"; in either case, the manager may prefer predation to acquiescing and sharing the market. Alternatively, the potential entrant may know of animus between her and the incumbent's managers. Empire building is thought to be a common desire of managers, and this generality alone may give rise to the possibility that the manager would prefer to prey rather than to acquiesce. A positive correlation of private benefits of control and market power will also be important to outsiders, who will anticipate that managers may have an inclination to prey to preserve private benefits of control even if it is not profitable to do so because of the initial cost of predation. Finally, executive pay and firm size are positively correlated in practice, so managers may be reluctant to share a market with an entrant.

Managers" intrinsic preferences and wishes alone do not necessarily dictate the firm's choices. A second set of relevant factors goes to the alignment of managerial interests and the shareholders' interests in profit maximization. There may be a number of factors that serve to align managerial and shareholder objectives. Fear of ouster as the result of shareholder activism, the firm's failure, or takeover, a desire to earn significant pay because of incentive-laden executive compensation, a desire to be highly regarded in the managerial labor market, and other factors may all compel the manager to act as a profit maximizer. Even if a manager was a known empire builder, if the firm were to fail as the result of predation, such a manager would be less likely to engage in it.

The factors that serve to align shareholder and managerial incentives, such as executive compensation, the pattern of equity ownership and the likelihood of a hostile takeover, may be observable to outsiders; thus potential entrants could have some sense of whether or not managerial objectives are likely to focus on profit maximizing, and thus whether "irrational" behavior is probable or not. Of course, as I will discuss below in more detail, to the extent that these factors can be altered easily, the relationship between observable characteristics of the firm and managerial behavior will be undermined.

Subject to this caveat, surmising managerial preferences and incentive structures may give entrants some picture of the likely "irrationality" of the firm. Potential entrants will assess the initial likelihood of unprofitable predation in light of these factors. This analysis of the rationality of the firm does not get away from ad hoc treatment of irrationality entirely, but it does add some empirical plausibility to the notion by relying on well-known agency problems within a firm.

Thus, agency problems may give rise to predatory pricing whether profitable or not. As set out above, in nonstrategic settings, it may be too costly to eliminate the possibility of unprofitable predatory pricing. When strategic motivations are taken into account, shareholders may intentionally create room for the possibility of agency problems.

It has been long recognized that there may be strategic reasons for a party to wish to destroy its own rationality in order to do better in a strategic interaction with another party. (63) For example, in a game of "chicken," where the winner is the last driver to veer from a collision course with another car, a known insane person is more likely to be expected to stay the course longer, and therefore win, than a sane person. As Schelling points out, one way of destroying this rationality, or at least of signaling a different motivation from that which comes naturally to the game-player, is to delegate to an agent with a different set of preferences. (64) For example, in a bargaining game an agent may be given inflexible instructions that are difficult to change. As a consequence, the agent may strike a better bargain than the principal, who is confined to bargain in light of her own preferences. (65) Various commentators have shown in various economic contexts that a principal may gain strategically by delegating to an agent. (66)

In the present context, it may be profitable for the incumbent to prey for reputation if there is at least some initial prospect of irrational play by the incumbent. Assuming that preying in the current period is profitable relative to acquiescing if it deters entry in the following periods, if the probability of the incumbent's irrationality is large enough, both a rational and irrational incumbent will prey and entry will be deterred. The strategic motivation for delegation suggests that the initial probability of irrationality can be treated as endogenous to some extent. Now assume that before the game is played the firm hires managers and chooses along a variety of dimensions, such as capital structure, the alignment of managerial and shareholder interests. The greater the alignment, the less likely it is that the manager would pursue unprofitable predatory pricing. Of course, the choice of alignment will affect not only the potential for managerial self-indulgence which results in predation, but it will also affect the profits earned when not preying. Specifically, profits will be higher as the alignment of interests between managers and owners increases.

Setting aside predatory pricing law, the firm will face the following trade-off in choosing alignment. On the one hand, the greater the alignment, the higher profits will be in periods of no predation; on the other, the lower the alignment, the greater the probability of irrational predation, and thus the greater the probability of deterred entry. It is possible that the latter motivation is important and that creating a reasonably high probability of unprofitable predatory pricing is actually profitable for the firm because of strategic entry deterrence. Indeed, it is conceivable that predation even in a one-shot game could be sufficiently probable so as to deter entry.

Vickers provides a helpful discussion of the distinction between seeking to maximize profits, and maximizing profits in the result. (67) In the strategic context described here, if potential entrants know that a manager seeks to maximize profits, they will enter. However, if there is a possibility that managers will not seek to maximize profits, potential entrants may anticipate predation by "irrational" managers or "rational" managers mimicking "irrational" managers for reputational purposes. Thus, by creating a possibility that the manager does not seek to maximize profits, the firm may maximize profits in the result by deterring entry.

As noted above, in considering the likely incentives of a manager facing a given array of disciplinary sources, it is important to consider whether the discipline, or lack of it, will change once a period of predation begins. Pre-predation, an outsider may canvass a number of factors and conclude that a manager does not currently face sufficient discipline to deter self-interested, but non-profit-maximizing predation. However, before an outsider could conclude that predation is therefore plausible, she must also consider whether existing sources of discipline would be altered in light of predation. For example, it may be that existing executive compensation fails to provide significant profit-based compensation. If so, the manager may be insulated from the losses of a corporation during a period of predation and therefore predation by managers is more likely. However, if it were the case that compensation schemes could be altered relatively easily to enhance the alignment of managerial and shareholder incentives, then observing an existing compensation scheme may not increase or decrease one's expectation of the likelihood of managerially optimal unprofitable predation. If a period of unprofitable predation began, the compensation scheme could change to discourage it.

Thus, in many situations in which strategic gains accrue from delegating to an agent, there is an important qualification given the prospect of renegotiation. (68) While the shareholders may initially delegate to managers in a way that gives rise to the prospect of unprofitable predatory pricing, if such delegation were easily changed, the threat of predation may not be credible to the entrant in a finitely repeated game. Consider backwards-induction analysis. In the last period, if there were entry notwithstanding a reputation for predation, the firm's owners would increase the alignment of management's incentives with shareholders so as to avoid a costly price war. If this were possible and likely, the entrant would enter in the last round regardless of reputation. Consequently, the second-to-last round entrant will enter, knowing that preying for a reputation is not worth the costs because entry will occur in the last period regardless. The incumbent will have an incentive to reset alignment in the second-to-last round because entry is inevitable in the last round regardless of its actions in the second-to-last round. This incentive to renegotiate will also exist in the third-to-last round, and so on. If alignment factors were easily reset, such that the probability of "irrationality" was reduced to zero, and if it were the only source of potential "irrationality," predation for reputation in a finitely repeated game would not occur.

It is therefore important to focus on alignment factors that cannot be easily changed. If a particular source of discipline can be changed easily when a period of predation begins, potential entrants will not be influenced by it when evaluating the likelihood of irrational predatory pricing. With respect to such flexible sources of discipline, the firm ought to adopt the optimal approach in the absence of any strategic motivation, since anticipated renegotiation will undermine any strategic benefits of a different choice. (69)

What factors, then, are likely to be important in practice? Executive compensation, for example, may be relatively easy to change quickly. A compensation contract that insulates the manager's pay from firm performance, or rewards high market share rather than profits or share price, could encourage unprofitable predation by self-interested managers. However, such a contract's existence, even if observable, may not alter outsiders' perceptions about the likelihood of unprofitable predatory pricing given the ease with which compensation contracts could be varied to avoid a price war. If compensation were easy to change, existing compensation contracts would have little effect on perceived incentives.

A more promising source of strategically useful managerial incentives is the structure of corporate ownership. If a firm is widely held, the manager personally will not suffer much through her shareholding from a period of unprofitable predatory pricing. If she realizes private benefits of control that are jeopardized by entry, she may prefer predatory pricing over acquiescing even if predatory pricing is not profitable overall for the firm. However, an existing widely held equity ownership structure does not determine the incentives for predation because of the possibility that that structure will change in the event of predation. That is, a takeover could occur. Thus, whether a widely held structure increases the probability of unprofitable predation depends on the ease with which a takeover could occur. Takeovers are intrinsically costly, but are particularly so if the firm has adopted, or can adopt, defenses to hostile bids.

While widely held structures are vulnerable to "renegotiation" through takeover bids, thus undermining the threat of "irrational" predation that is necessary for reputation building, other ownership structures may be relatively difficult to change. For example, consider a firm with a controlling shareholder and widely held minority shares. This structure can lead to significant private benefits of control because of the invulnerability of the firm to a hostile takeover and the control the shareholder exercises over the board of directors and managers. (70) Moreover, the private benefits of control can create inertia for the concentrated ownership structure. (71) Suppose a controlling shareholder structure suboptimally controls agency costs because of managerial entrenchment. Minority shareholders would benefit if the control bloc were diluted. This means that any buyer of part of the control bloc will confer benefits on the minority; the buyer of part of the bloc fails to internalize the benefits of its acquisition. This in turn means that the buyer would have a difficult time profitably purchasing shares at a price that compensates the incumbent controlling shareholder for lost private benefits of control. Adopting a controlling shareholder structure that generates significant private benefits of control, particularly if these benefits are jeopardized by successful entry, could bring strategic benefits for the firm in deterring entry. The structure creates a credible possibility of unprofitable, "irrational" predation that is difficult once established to eliminate.

Another choice about ownership structure that could bring strategic benefits concerns the amount of debt that a firm issues. Managers of solvent corporations owe fiduciary duties to shareholders. Moreover, shareholders elect the board of directors which in turn appoints managers and structures their compensation. If there is a constituency other than themselves that managers seek to serve, it is likely to be equity owners. This implies that there will be an agency cost of issuing debt. (72) Managers making decisions that benefit shareholders may be making decisions that harm debtholders. Indeed, significant debt could imply "irrational" predation. For example, suppose that if a firm acquiesces and does not prey on an entrant, future cash flows will suffice only to pay down the debt that the firm has issued, while if the firm preys and successfully drives the entrant from the market, there would be profits on top of the debt payments that would accrue to equity. Suppose further that preying is expected to be unprofitable, but there is a small chance (say, one in a hundred) that it will be profitable. Managers beholden to shareholders would in this case prey even though in expected terms predation is unprofitable. This is because shareholders realize the upside in the rare cases when predation succeeds, but pass on the downside to debtors if predation fails. (73)

This structure also faces inertia. Any purchase of the firm's equity by a debtholder seeking to protect debtholders' interests would benefit all debtholders. The failure to internalize the benefits of such an acquisition will deter it. Acquiring all the debt prior to an acquisition of equity would face free-rider problems: it would pay individual debtholders to hold out and simply realize the benefits of improved governance rather than sell.

Renegotiation is thus not an insurmountable problem for a firm strategically delegating decisions to managers in a way that creates the possibility of irrational predation. The probability of renegotiation, however, must be considered in any analysis of delegation and predation.

(b) Empirical application of ex ante incentives to delegate: a review of Lott's study Examining Lott's theoretical and empirical study of predation further illustrates the analysis. (74) Lott draws on Kreps, Wilson, Milgrom and Roberts' analysis of reputation to conclude that the agency relationship between managers and shareholders will be important in determining the credibility of a predatory threat. He suggests that whether or not predatory firms have made efforts to align managers' interests with shareholders' will affect the probability of predation. He predicts that firms that are predatory are more likely to have taken steps to misalign managerial and shareholder interests through executive compensation, perhaps by rewarding sales and failing to penalize short-run losses, and through managerial entrenchment like antitakeover charter amendments.

Other commentators have pointed out that Lott takes a narrow view of reputational models by focusing only on uncertainty about managerial incentives. (75) Other sources of uncertainty, like uncertainty over the incumbent's cost function, could also motivate a reputational model of predation. (76) But I show here that even on its own terms, Lott's analysis of reputation and managerial incentives is flawed.

An important consideration in assessing differences between predatory firms and nonpredatory firms, Loft suggests, is the time horizon facing the potentially predatory firm because of its effect on the incentives for reputation investment:
   Given that it pays for firms to make investments in reputation
   during the initial periods, there is no reason to expect these
   firms to differ initially from nonpredatory firms in terms of how
   costly it is to remove their managers.... It is only later than
   the firm might have an incentive to remove the aggressive manager
   rather than bear the short-run financial losses from predation.
   However, during the middle and end periods in the game, predatory
   firms will have a greater return to signaling to entrants that
   potentially "tough" or "irrational" managers are not removed. (77)

Lott's analysis of the relevance of the time horizon is suspect. As I do, he treats "irrationality" in the Milgrom, Roberts, Kreps and Wilson model of reputation as depending on the incentives facing managers. In particular, there may be some probability that the manager will engage in unprofitable predatory pricing out of self-interest. If this probability exists, then even managers who would otherwise not find it privately optimal to prey may prey in order to mimic a manager who does find it optimal to prey. For it to be rational for a firm's managers to engage in unprofitable predatory pricing in order to build a reputation, the probability that a self-interested manager will prey must be positive during all periods, not just those at the end. Conversely, if predatory pricing in early rounds creates a reputation for predation, there is no need to make changes in later periods to ensure that managers have incentives to prey. In later periods, outsiders having observed predation in early periods (78) will worry that the firm already has a sufficiently skewed incentive structure such that managers may prefer predation to profits; this is the whole point of earning a reputation in early rounds. (79)

A more fundamental problem with Lott's conclusion that firms will alter incentives depending on the period is that if incentives were so easily manipulated, renegotiation would undermine any attempt to be strategic in creating a potential wedge between managers' and shareholders' interests. As noted above, if in the last round incentives were easily altered to provide profit-maximizing incentives to managers, the model of predation for reputation unravels. This is because in the second-to-last period, potential entrants recognize that renegotiation will occur in the last round, and thus conclude that predation for reputation will not occur. They therefore anticipate that if they enter, renegotiation will occur in the second-to-last period. And so on.

It must be, therefore, that any strategic effort to increase the probability of misaligned incentives relies on characteristics of the firm that are not easily undone. This analysis undermines Lott's predictions about firms changing incentives depending on the period. It also undermines his empirical analysis. Lott bases his analysis on several factors that are likely easy for the firm to alter on short notice. For example, in supporting his conclusion that predatory pricing is rare, he tests whether incentive pay based on short-run profits was greater at public firms not convicted of predatory pricing than at those that have been so convicted. While acknowledging that his data were imprecise because many aspects of compensation, such as that based on stock options, are not reported in detail, and moreover that the power of his tests was weak given limited data, Lott concludes that the compensation structure at firms convicted of predation was not significantly different from that of firms not convicted of predation. (80)

Given that compensation contracts are easily amended, it would be unlikely that they would have a significant strategic effect. Even if, for example, compensation rewarded market share such that in the last period predation were plausible even though unprofitable, it would nevertheless not do much to deter entry. Outsiders would anticipate that when entry actually occurred in the last period, compensation could be altered better to encourage managers to maximize profits rather than prey. Renegotiation and backwards induction implies that compensation is unlikely to be useful in strategically manipulating managers' incentives to prey. Looking for differences in compensation contracts to determine the credibility of predatory threats is misguided.

Lott also examines measures of managerial entrenchment to determine whether predation is likely, (81) He studies the relationship between predation and various variables that suggest that managers are less likely to be removed. Two of these variables are expenditures on research and development and Tobin's q. (82) While these two statistics may be correlated with low managerial turnover, they may or may not be relevant to a firm's efforts to entrench management. (83) High Tobin's q and R&D may be correlated with low managerial turnover simply because they reflect the fact that management is doing a good job. High R&D expenditure may reflect a managerial team with long-term value in mind, thus making a takeover or termination less likely. Put another way, Tobin's q and R&D may not cause the managers to be entrenched; rather good management may cause high Tobin's q and high R&D and a consequent lower probability of a turnover.

Antitakeover charter amendments, on the other hand, will operate unambiguously to entrench managers. This in turn would make them, all things equal, more likely to engage in unprofitable predation. Lott's analysis of the relationship between antitakeover charter amendments and convictions for predatory pricing is flawed, however, because he again neglects renegotiation. This time, however, renegotiation works in favor of strategically signaling the probability of an irrationally tough manager. Whether or not an antitakeover measure is in place when a bid is contemplated may not be particularly important to the raider since managers can implement antitakeover devices such as poison pills relatively quickly. Lott himself states, "Instituting some types of entrenchment devices can be done at low cost: for example, instituting an antitakeover charter amendment can be done fairly quickly and involves routine legal costs which are fairly low." (84) Or as Coates states,
   [I]magine you are a bidder, considering a high premium bid for a
   target. Imagine the target lacks a pill. Do you conclude that the
   target is easy to acquire, as a matter of law? Clearly not. Unless
   there is something unusual about your target, the target will have
   the ability to--and almost certainly will--adopt a pill as soon as
   you start your bid. (85)

Thus, the fact that Lott failed to discover that firms that have been convicted of predation are more likely to have antitakeover charter amendments (86) may not be at all significant, given the possibility of instituting takeover defenses in the face of an actual bid.

Concentrated share ownership that entrenches managers, and therefore allows significant agency costs and private benefits of control, as discussed above, may be resistant to renegotiation. The inability of minority shareholders to overcome a free-rider problem in compensating the controlling shareholders for lost private control benefits implies that controlling shareholders may remain entrenched even if joint gains were available from altering the share structure. Lott examined the level of share ownership concentration to determine its effects on the probability of a conviction for predation. Specifically, he ran a logit regression predicting predation years for firms that were convicted of predation based on, among other factors, the fraction of shares held by the top four owners. (87) Higher share ownership was a statistically significant predictor of a conviction for predatory pricing. This is consistent with the conclusion that entrenchment through shareholdings may encourage managerial self-interest and may be difficult to alter. (88)

As a final observation on Lott's study, note that where the game involves many rounds, the initial level of misaligned incentives that is required to deter entry need not be very significant. (89) Indeed, it may be that it would be prohibitively costly to align managerial interests such that the probability of self-interested predation by the manager is so small that "rational" firms would not find it worthwhile to prey in order to preserve the perception that they may be irrational. For example, an optimal contract with a risk-averse manager is unlikely to pay the manager only on the basis of profits, thus leaving some room for self-indulgence; making the manager bear more risk in order to eliminate this room may be too costly. (90) Thus, even in the absence of strategic manipulation of incentives, the act of delegation itself may bring strategic benefits by implying some positive prior probability of irrationality. (91) Consequently, public ownership and consequent delegation may be more attractive as a strategic matter than, for example, keeping ownership all in the hands of a single entrepreneur. Any empirical examination of public firms may fail to reveal differences in predatory and nonpredatory firms, but this does not necessarily undermine the usefulness of strategic delegation in deterring entry; it may be that being a public corporation alone creates a wedge between managers and shareholders that suffices to make irrational predation and therefore predation for reputation possible.

3. THE RECOUPMENT TEST AND STRATEGIC PREDATION Let me turn now to the implications of the analysis of unprofitable predatory pricing and self-deterrence for the recoupment test. It is helpful to consider two contexts: one, a context in which predation is likely a profitable strategy, perhaps because the predator earns a profitable reputation for predation; and two, a context in which predation is not a profitable strategy for a number of reasons, including the fact that there are only a relatively small, finite number of rounds in which a reputation could be useful.

The first context is obviously caught by the recoupment test. Antitrust authorities in such a setting will view below-cost pricing with suspicion. The second context, on the other hand, appears to fall outside the explicit ambit of the recoupment test. Given that, as we have seen, firms can take a strategic ex ante approach to incentives in order to establish a credible threat of unprofitable predatory pricing, which is profitable in the result by deterring entry, there appears to be an underdeterrence problem associated with the recoupment test.

But there is an important sense in which the recoupment test implicitly deters unprofitable predatory pricing. The most common reason for rejecting the hypothesis that predation is profitable, and thus that recoupment is plausible, is that markets are competitive. This is also a setting in which the principal-agent problem in firms is much less acute than in other settings--if a competitive firm sustains nontrivial losses, it cannot survive. Moreover, a manager in a competitive firm is likely to enjoy lower levels of private benefits of control to protect by preying than managers of firms with market power. As a consequence, although it is true that unprofitable predatory pricing is not captured by the recoupment test explicitly, the presence of competitive markets suggests both that recoupment is not likely and that unprofitable predatory pricing is unlikely; the recoupment test implicitly accounts for unprofitable predatory pricing.


As noted above, the Canadian Draft Guidelines fail to explain what economic developments led to the proposal to abandon the recoupment test as a necessary element of predation. They do not refer to any theory of why unprofitable predatory pricing may take place. This article provides such a theory by examining reputational theories and the principal-agent problem within firms. However, this article fails to provide a justification for rejecting the recoupment test. The central contribution of this article is to offer an additional rationale for the recoupment test: the test deters profitable predation, but also implicitly deters unprofitable predatory pricing. Competitive markets are conducive neither to recoupment nor to agency costs.

Further support for inclusion of the recoupment test is found in conventional examinations of overdeterrence. To this point I have largely focused on the possibility of underdeterrence from inclusion of a recoupment test, but there are also important considerations relating to overdeterrence. While unprofitable predatory pricing is socially harmful and could be economically rational because of ex ante strategic incentives, it must be true that the more probable is recoupment, the more probable is it that a firm would adopt a predatory strategy. The costs of false positives in predatory pricing law are large and run counter to the objectives of competition policy since they create an incentive to keep prices high. Such concerns about overdeterrence were fundamental in the analysis of the Supreme Court in Brooke Group itself.

Concerns about overdeterrence from deciding not to rely on recoupment are made more acute because of the inevitable pressure on the price-cost test that such a move would bring. (92) Areeda and Turner first proposed a predatory pricing test that examined the relationship between price and cost. (93) Predatory pricing would be made out under their test if prices were shown to be below average variable cost. But, as Trebilcock, Winter, Collins and Iacobucci point out, price-cost tests will be extremely complicated to implement in practice for a variety of reasons. (94) Costs are very difficult to ascertain. For example, attributing costs to different products in a multiproduct firm will be fraught with arbitrariness and error. (95) But more importantly, the appropriate comparison is not between price and accounting costs, but rather between price and opportunity cost. (96) Opportunity cost may be very difficult to measure. For example, goods may be perishable, which implies that the opportunity cost of a sale is lower than the good's production cost. Or the opportunity cost of a particular sale may be low because it results in future sales. For example, in the recent Canadian predation case under the new airline regulations that abandoned a recoupment requirement, Air Canada claimed that when pricing some routes it accounted for profits expected from customers who would travel on other subsequent flights. (97) The Competition Tribunal rejected this argument on the ground that there was insufficient evidence to substantiate it. This may have been a justifiable reaction on the record, but such a holding simply demonstrates the inadequacy of a price-cost test--consider the evidence required to show that subsequent paid flights by passengers generate sufficient profits to offset losses on the sale of tickets on one particular route. Suppose some passengers on a flight from Moncton to Toronto fly on to Vancouver. To determine the profit from the passengers on the Toronto-Vancouver flight, that is, to determine the "beyond contribution," it would be necessary to determine the exact costs and revenues on the Toronto-Vancouver flight. Moreover, it would be necessary to determine whether the Toronto-Vancouver flight gives rise to "beyond contribution" from passengers taking other flights associated with their travel between Toronto and Vancouver. This exercise would need to be repeated for customers traveling from Moncton to Toronto and then to Montreal, New York, Paris, etc. At the limit, the prices and costs of every flight that Air Canada flies would need to be known to determine the opportunity cost of the sale of tickets on one flight. Price-cost tests are difficult indeed to implement in a meaningful way.

Abandoning the recoupment test as a necessary element of predation, as the Canadian Draft Guidelines recommend, will make it easier to substantiate a predatory pricing allegation largely on the basis of a price-cost comparison. Because of the very high risk of error in conducting such a comparison (let alone the costs of such analyses), this is not sensible policy. There will be a significant risk of chilling aggressive but nonpredatory pricing, which runs counter to the objectives of the antitrust policy. On the other hand, looking to the recoupment test acts as an important check on the price-cost comparisons. If the price-cost test suggests predation but recoupment is not plausible because of competitive markets, then predation is neither profitable nor likely explicable owing to agency costs; rather, it is highly likely that something was missing from the price-cost tests.

The results in the Air Canada case illustrate the dangers posed by not relying on the recoupment test. As noted above, under current Canadian law, it is an anticompetitive act for a dominant airline to price below avoidable cost. The Canadian Competition Tribunal undertook to compare costs and revenues on various routes and concluded that prices in Air Canada were below cost. Following the plain meaning of the Competition Act and its regulations, no consideration was given to the motivation for the prices in determining that Air Canada had committed anticompetitive acts. (98) Yet it at least challenges the plausibility of predatory pricing to note that Air Canada went bankrupt shortly after the pricing practices in question. It is difficult to imagine that Air Canada's prices, which often simply responded to competitors' prices, were predatory and were protecting Air Canada's supra-competitive profits. Further support for this conclusion comes from the Competition Bureau's announcement on October 29, 2004 that it had settled its predation claim with Air Canada, announcing that changes in market conditions, particularly the "entry and growth of low cost carriers," implied that it would not be in the public interest to pursue the predation claim further. (99) Shortly thereafter, and also because of changed conditions in Canadian airlines, the Minister of Industry proposed deleting airline-specific provisions, including the predatory pricing provisions, in the Competition Act. (100) The first phase of the Air Canada case, which concluded that Air Canada had committed anticompetitive acts, was decided in July 2003. If market conditions could change sufficiently that the Competition Bureau would decide not to pursue its claim of abuse against Air Canada, and that the Minister of Industry would propose changing the Competition Act, only one year after a finding of a dominant airline committing an anticompetitive act, there is a powerful suggestion that the case never should have been brought in the first place. Amending the Competition Act to delete the airline-specific provisions on predatory pricing would make sense not simply because of the change in market conditions, but rather because the rapid change in market conditions in airlines confirms that reliance solely on price-cost tests in determining predatory pricing allegations will often mislead. Analyzing airline predation allegations pursuant to the recoupment test, on the other hand, would have appropriately accounted for factors such as entry conditions in concluding that predation was not plausible.

Under the original Canadian Guidelines, price-cost comparisons and the recoupment test were necessary to substantiate a predatory pricing allegation. Price-cost tests are flawed for the reasons just given. Additionally, while unprofitable predatory pricing is not necessarily self-deterring, and while the recoupment test is not directed at the question of whether unprofitable predatory pricing is plausible, the recoupment test nevertheless implicitly addresses the concerns that unprofitable predatory pricing presents: competitive markets are conducive neither to profitable nor to unprofitable predatory pricing. Combining the two tests significantly reduces the risk of improperly characterizing lawful pricing as predatory. It is dangerous to diminish the importance of either element of a predatory pricing test, as recent developments in Canadian law have threatened to do.

(1) 509 U.S. 209 (1993). The Court had foreshadowed this result in two earlier cases, Matsushita Elec. Indus. Corp. v. Zenith Radio Corp., 475 U.S. 574 (1986) and Cargill v. Monfort of Colo., Inc., 479 U.S. 104 (1986).

(2) The Court declined in Brooke Group to indicate what the appropriate measure of cost is in all cases, given that the parties in the case at hand agreed that the appropriate measure was average variable cost. Brooke Group, 509 U.S. at 209.

(3) Id. at 224.

(4) Id. at 225.

(5) Id. at 226.

(6) The Court did not hold, however, that an oligopolist under the Robinson-Patman Act could never be found to have engaged in predatory pricing. Id. at 230.

(7) Some recent scholarship has focused on the role of price-cost considerations in analyses of predation. Edlin, for example, suggests that allowing incumbent firms to set price above their own costs but below their rivals' costs may deter entry in a manner harmful to consumers and society as a whole. Aaron S.. Edlin, Stopping Above Cost Predatory Pricing, 111 YALE L.J. 94 (2002). Trebilcock et al. point out the problems in attempting to identify relevant measures of cost given that it is the seller's opportunity cost of the sale that is relevant; the opportunity cost of a sale need not relate to production costs and could even be negative, as where there is promotional value to a sale. MICHAEL J. TREBILCOCK, RALPH WINTER, PAUL COLLINS & EDWARD IACOBUCCI, THE LAW AND ECONOMICS OF CANADIAN COMPETITION POLICY (2002).

(8) Guidelines Part, available at (emphasis added).

(9) While the necessity of recoupment has not been considered, the recoupment test is consistent with case law on predation. See TREBILCOCK ET AL., supra note 7, at 332-33 (guidelines are consistent with predation cases like R. v. Hoffman-La Roche Ltd., [1980] 28 O.R.2d 164 that reject exclusive reliance on price-cost comparisons and rather examine the totality of economic conditions in the market to decide whether predation has taken place).

(10) See Preface to Draft Guidelines, available at

(11) See Competition Act, R.S.C., 1985, ch. C-34, [subsection] 78-79, available at; Regulations Respecting Anti-Competitive Acts of Persons Operating a Domestic Service (SOR/2000-324), available at For a critical review, see Margaret Sanderson & Michael J. Trebilcock, Bad Policy, Bad Law: Bill C-26 Amendments to the Competition Act on Airline Predation, 20 COMPETITION POL'Y REC. 32 (2002). See also Andrew Eckert & Douglas S. West, Predation in the Airline Industry: The Canadian Antitrust Approach, 47 ANTITRUST BULL. 217 (2002).

(12) William H. Jordan, Predatory Pricing After Brooke Group: The Problem of State 'Sales Below Cost' Statutes, 44 EMORY L.J. 267 (1995), describes a similar approach in some U.S. states' antitrust law that prohibits sales below cost as anticompetitive ipso facto.

(13) [2003] Comp. Trib. 13 (Can.).

(14) A resolution of the case, including an assessment of the implications of the finding of anticompetitive acts, will not be forthcoming given the decision of the Bureau to settle the matter with Air Canada. It is clear, however, that the Tribunal found that Air Canada committed anticompetitive acts simply by setting price below cost.

(15) See News Release, Industry Canada, Minister of Industry Tables Amendments to Strengthen Competition Act (November 2, 2004),

(16) Id.

(17) Another proposed amendment in the now-defunct Bill C-19 would have decriminalized predation and treated it as an abuse of dominance under the civil reviewable practices part of the Competition Act, but this would not have affected the substance of the Bureau's approach to identifying predation.


(19) There is another complicating terminological factor resulting from the analysis of strategic precommitment. It may be profitable in the result to precommit not to seek to maximize profits. See John Vickers, Delegation and the Theory of the Firm, 95 ECON. J. 138 (1987). As I will explain below, "unprofitable predatory pricing" may be unprofitable for the firm if it takes place, but creating the possibility of unprofitable predatory pricing may be profitable for the firm if it deters future entry because of reputation.

(20) Edlin, supra note 7, shows that the possibility of low prices that are above the alleged predator's costs creates social harm by deterring entry ex ante; similarly, here I show that the possibility of unprofitable predatory pricing creates social losses ex ante by deterring entry.

(21) In particular, he fails to appreciate the role of renegotiation of strategically designed incentives.

(22) Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 224 (1993).

(23) Advo, Inc. v. Phila. Newspapers, Inc., 51 F.3d 1191 (3d Cir. 1995), cited in C. Scott Hemphill, The Role of Recoupment in Predatory Pricing Analyses, 53 STAN. L. REV. 1581 n.93 (2001).

(24) W. Parcel Express v. UPS, 190 F.3d 974 (9th Cir. 1999), cited in Hemphill, supra note 23, at 1581 n.93.

(25) Lawson A.W. Hunter & Susan M. Hutton, Is the Price Right?: Comments on the Predatory Pricing Enforcement Guidelines and Price Discrimination Enforcement Guidelines of the Bureau of Competition Policy, 38 McGILL L.J. 830, 836 (1993).

(26) See Richard O. Zerbe & Michael T. Mumford, Does Predatory Pricing Exist? Economic Theory and the Courts after Brooke Group, 41 ANTITRUST BULL. 949, 964-67 (1996).

(27) Id.

(28) Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 225 (1993).

(29) Id.

(30) Frank H. Easterbrook, Predatory Strategies and Counterstrategies, 48 U. CHI. L. REV. 263, 268 (1981).

(31) In the corporate law and economics literature, Easterbrook and Fischel's work has been enormously influential in setting out the view of the firm not as a black box, but rather as a nexus of contracts. See, e.g., FRANK H. EASTERBROOK & DANIEL R. FISCHEL, THE ECONOMIC STRUCTURE OF CORPORATE LAW (1991). They drew on insights developed by R.H. Coase, The Nature of the Firm, 4 ECONOMICA 386 (1937), Armen Alchian & Harold Demsetz, Production, Information Costs, and Economic Organization, 62 AM. ECON. REV. 777 (1972), and Michael C. Jensen & William H. Meckling, Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure, 3 J. FIN. ECON. 305 (1975).

(32) Jensen & Meckling, supra note 31.

(33) For an entertaining account of this managerial preference at R.J.R. Nabisco in the 1980s, see BRYAN BURROUGH & JOHN HELYAR, BARBARIANS AT THE GATE: THE FALL OF RJR NABISCO (1990).

(34) See, e.g., Melvin Avon Eisenberg, The Structure of Corporation Law, 89 COLUM. L. REV. 1461, 1472 (1989); Oliver. Hart, An Economist's Perspective on the Theory of the Firm, 89 COLUM. L. REV. 1757, at 1759 (1989).

(35) See, e.g., Harold Demsetz, Barriers to Entry, 72 AM. ECON. REV. 47 (1982).

(36) Avishalom Tor, Illustrating a Behaviorally Informed Approach to Antitrust Law: The Case of Predatory Pricing, 18 ANTITRUST 52 (2003).

(37) See, e.g., Andrei Shleifer & Robert W. Vishny, A Survey of Corporate Governance, 52 J. FIN. 737 (1997). In discussing the role of competition in addressing agency problems, they observe, "product market competition [while imperfect] is probably the most powerful force toward economic efficiency in the world...." Id. at 738.

(38) See, e.g., Bernard S. Black, Shareholder Passivity Reexamined, 89 MICH. L. REV. 520 (1990).


(40) See, e.g., Eugene F. Fama, Agency Problems and the Theory of the Firm, 88 J. POL. ECON. 288 (1980).

(41) See, e.g., Henry G. Manne, Mergers and the Market for Corporate Control, 73 J. POL. ECON. 110 (1965).

(42) See Patrick Bolton & David S. Scharfstein, A Theory of Predation Based on Agency Problems in Financial Contracting, 80 AM. ECON. REV. 94 (1990).

(43) Suppose, for example, that during the period of predation and any period of acquiescence that the manager realizes no private benefits of control, while during the period following predation, the manager realizes private benefits of control. Predation may be rational for the manager in this case even though it is not profit-maximizing.

(44) See, e.g., Ontario Business Corporations Act, R.S.O. 1990, c. B.16 [section] 134, available at In Delaware, Loft Inc. v. Guth, 23 Del. Ch. 255, 260 (1939) held that the rule "requires an undivided and unselfish loyalty to the corporation [and] demands that there shall be no conflict between duty and self-interest."

(45) See generally Edward M. Iacobucci, A Wise Decision? An Analysis of the Relationship Between Corporate Ownership Structure and Directors' and Officers' Duties, 36 CAN. BUS. L.J. 337 (2002). In Delaware, see, e.g., Aronson v. Lewis, 473 A.2d 805, 812 (Del. 1984).

(46) Predatory prices may indirectly preserve private benefits of control, as I have argued, but it is clear that lower prices do not themselves benefit the manager.

(47) See, e.g., Para-Medical Leasing, Inc. v. Hangen, 739 P.2d 717 (Wash. 1987), in which a manager successfully claimed protection under the business judgment rule in the face of a shareholder's suit accusing him of pricing too low; see also Lussier v. Mau-Van Dev. Inc., 667 P.2d 804 (Haw. 1983) (business judgment rule applies to manager's decision to set low prices).

(48) An ownership structure of 51% control by a majority shareholder/manager, for example, will not be susceptible to a hostile takeover. See Rene Stulz, Managerial Control of Voting Rights, Financing Policies, and the Market for Corporate Control, 20 J. FIN. ECON. 25 (1988).

(49) Easterbrook, supra note 30, at 268, analogizes unprofitable predatory pricing and destroying a profitable plant. There is no need to fine the managers for destroying a plant because deterrence is built in. He observes moreover that, "[i]f costs played no role in the decisions of Exxon's managers they would not be deterred. But if they exhibited such irrationality, there would be no point in invoking the antitrust laws. If the managers did not respond to the cost of destroying the refinery, they would not respond to the cost of paying the fine." Id. at 268 n.14. The analysis here does not consider the managers to be "irrational" by engaging in unprofitable activity. Rather, they are pursuing their own interests. Furthermore, it is not just the managers who would respond to antitrust fines in the analysis here, but also the firm's shareholders who would be more concerned about deterring such unwanted agency behavior because of the fines involved.

(50) Brooke Group Ltd. v. Brown & Williamson Tobacco Corp., 509 U.S. 209, 226 (1993).

(51) There is a qualification to this, however. The recoupment test typically focuses on competition in the market in which the predation takes place. If that market is competitive, but there are other markets in which the alleged predator participates that are not competitive, the competitive discipline of the one market may not be sufficient to compel the manager to seek to maximize profits. That is, agency problems may arise with respect to one market because of slack generated by profits in other markets. This may help explain the intuition that a multimarket firm with ample resources is more likely to engage in predation. Such an intuition is generally treated with skepticism--simply because a multimarket firm can prey does not mean it is profitable to do so. In Ne. Tel. Co. v. AT&T Co., 651 F.2d 76 (2d Cir. 1981), the Court recognizes this intuition but points out reasons to reject it: "That SNET is a multi-product firm seems to enhance the likelihood that it would profit from a policy of predatory pricing. One might assume that while SNET leased its PBXs at unremunerative rates, it would subsidize its losses with profits earned in other areas of its business. But although subsidization may stave off bankruptcy, it does not appreciably reduce the short-run costs of predation. In terms of lost profits, or in economic jargon, 'opportunity costs,' unremunerative prices will be as expensive to a diversified monopolist as to a single-product firm." Id. at 88-89. Yet participation by the firm in many markets is also relevant to recoupment. As I discuss below, multimarket firms may have an incentive to develop a reputation for preying; indeed, the Draft Guidelines discuss this possibility. See Draft Guidelines, supra note 10, at Part 4.3.a. Thus, recoupment also implicitly addresses multi-market slack since the prospect of preying for reputation is greater in such a case.

(52) For an early work suggesting this effect of predatory pricing, see B.S. Yamey, Predatory Price-Cutting: Notes and Comments, (1972) 15 J.L. & ECON. 129 (1972). For a survey of various theories of predation, including reputation, see Janusz A. Ordover & Garth Saloner, Predation, Monopolization and Antitrust, in 1 HANDBOOK OF INDUSTRIAL ORGANIZATION 537 (Richard Schmalensee & Robert D. Willig eds., 1989). For a recent article suggesting how insights from strategic theory, like the reputation theory of predation, can be incorporated in predatory pricing law, see Patrick Bolton et al., Predatory Pricing: Strategic Theory and Legal Policy, 88 GEO. L.J. 2239 (2000).

(53) Guidelines, supra note 8, at Part

(54) Draft Guidelines, supra note 10, at Part 4.3.a. ("If the incumbent firm is successful at persuading the entrant that its continued presence or expansion in the market will be met with a strategy of unreasonably low pricing, then the entrant will discontinue its expansion and possibly exit the market. The incumbent firm thereby creates a reputation for unreasonably low pricing that deters the entry or expansion of other firms in that market or in other markets in which the incumbent competes. In any given market, an unreasonably low pricing policy used to gain a reputation is more likely when the firm in question operates in more than one geographic or product market.")

(55) See, e.g., United States v. AMR Corp., 335 F.3d 1109 (10th Cir. 2003); Advo, Inc. v. Phila. Newspapers, Inc., 51 F.3d 1191 (3rd Cir. 1995). The recoupment test in Brooke Group, 509 U.S. 209, arguably focuses more on the market in which predation takes place; hence reputation-building predation may escape predatory pricing law. See also Zerbe & Mumford, supra note 26; Timothy J. Trujillo, Predatory Pricing Standards Under Recent Supreme Court Decisions and Their Failure to Recognize Strategic Behavior as a Barrier to Entry, 19 J. CORP. L. 809 (1994).

(56) The following example is adapted from JEAN TIROLE, THE THEORY OF INDUSTRIAL ORGANIZATION (1988).

(57) If the probability of one more entrant next period is large enough, this approximates an infinity of potential entrants.

(58) Reinhard Selten, The Chain-Store Paradox, 9 THEORY & DECISION 127 (1978).

(59) David M. Kreps & Robert Wilson, Reputation and Imperfect Information, 27 J. ECON. THEORY 253 (1982).

(60) MILGROM & ROBERTS, supra note 39.

(61) Kreps & Wilson, supra note 59, at 277, acknowledge this, but point out that "analysis of this sort may require ad hoc assumptions."

(62) MILGROM & ROBERTS, supra note 39, at 303. See also LOTT, supra note 18.


(64) Id.

(65) Id. at 29.

(66) See, e.g., Vickers, supra note 19; Chaim Fershtmann & Kenneth L. Judd, Equilibrium Incentives in Oligopoly, 77 AM. ECON. REV. 927 (1987); Bernard Caillaud & Benjamin Hermalin, The Use of an Agent in a Signalling Model, 60 J. ECON. THEORY 83 (1993); Michael L. Katz, Game-Playing Agents: Unobservable Contracts as Precommitments, 22 RAND J. ECON. 307 (1991); Gillian K. Hadfield, Credible Spatial Preemption Through Franchising, 22 RAND J. OF ECON. 531 (1991); and Nahum D. Melumad & Dilip Mookherjee, Delegation as Commitment: The Case of Income Tax Audits, 20 RAND J. ECON. 139 (1989).

(67) Vickers, supra note 19.

(68) See Katz, supra note 66, in particular for an excellent analysis of renegotiation.

(69) Id.

(70) See, e.g., Randall Morck et al., Management Ownership and Market Valuation: An Empirical Analysis, 20 J. FIN. ECON. 293 (1988); Rene M. Stulz, Managerial Control of Voting Rights, Financing Policies, and the Market for Corporate Control, 20 J. FIN. ECON. 25 (1988).

(71) Ronald J. Daniels & Edward M. Iacobucci, "Some of the Causes and Consequences of Ownership Concentration in Canada, in CONCENTRATED CORPORATE OWNERSHIP 81 (Randall K. Morck ed., 2000).

(72) See, e.g., Jensen & Meckling, supra note 31.

(73) This asymmetry of payoffs generates the agency costs of debt generally. Id.

(74) LOTT, supra note 18.

(75) David E.M. Sappington & J. Gregory Sidak, Are Public Enterprises the Only Credible Predators?, 67 U. CHI. L. REV. 271 (2000).

(76 Id. at 276-77.

(77) LOTT, supra note 18, at 23.

(78) If there have been no attempts at entry, then the firm is perpetually in the initial period in terms of deterring future entrants.

(79) To the extent that Lott is simply observing that firms with a long time horizon are more likely to engage in predation, his observations are correct. It is the apparent implication that firms adjust their incentive structure over time that is suspect.

(80) LOTT, supra note 18, at 36-49.

(81) Id. at 49-59.

(82) Tobin's q is the ratio of the market value of equity plus the book value of debt and preferred stock all divided by the book value of a firm's assets adjusted for inflation by the producer price index.

(83) See also Sappington & Sidak, supra note 75.

(84) Loft, supra note 18, at 25.

(85) John C. Coates W, Takeover Defenses in the Shadow of the Pill: A Critique of the Scientific Evidence, 79 TEX. L. REV. 271, 287-88 (2000).

(86) LOTT, supra note 18, at 53.

(87) This is not a perfect test of the effects of entrenchment, since the relationship between share ownership concentration and entrenchment and the corresponding opportunity for managerial self-interest at the expense of profits may not be monotonic. Morck et al., supra note 70, found that at very low levels of managerial share ownership, increasing share ownership served to align better managerial and shareholder interests. At a certain level, increasing ownership further caused entrenchment. Once managers are entrenched, increasing share ownership will again serve better to align interests. A statistical measurement of share ownership concentration across firms will not, therefore, always give an accurate measure of entrenchment and managerial self-interest.

(88) To reiterate, while increasing managerial shareholdings will, all things equal, reduce the benefits to managers of, say, diverting corporate assets to themselves, it will also reduce the costs by making them less vulnerable to board oversight, proxy battles and most importantly, takeovers. This latter effect could dominate the former. See Daniels & Iacobucci, supra note 71.

(89) LOTT, supra note 18, at 21 acknowledges this.

(90) See, e.g., Bengt Holmstrom, Moral Hazard and Observability, 10 BELL J. ECON. 74 (1979).

(91) Katz, supra note 66.

(92) Edlin has argued that even above-cost pricing can be predatory and instead suggests conduct-based rules to govern predation. See Edlin, supra note 7. In particular, he, like Baumol and Williamson before him, would restrict the freedom of incumbents to lower their prices following entry. Without engaging the debate over conduct-based rules in detail, I simply note that they are notoriously difficult to implement and assume that the two tools available to regulators in practice are price-cost tests and the recoupment test. See, e.g., Einer Elhauge, Why Above-Cost Price Cuts to Drive Out Entrants are Not Predatory--and the Implications for Defining Costs and Market Power, 112 YALE L.J. 681 (2003).

(93) Phillip Areeda & Donald F. Turner, Predatory Pricing and Related Practices Under Section 2 of the Sherman Act, 88 HARV. L. REV. 697 (1975).


(95) See the valiant, but ultimately arbitrary, attempt by the Canadian Competition Tribunal to allocate common costs in Comm'r of Competition v. Air Canada, [2003] Comp. Trib. 13.

(96) See TREBILCOCK ET AL., supra note 94.

(97) This was identified as "beyond contribution." Air Canada, [2003] Comp. Trib. 13.

(98) This finding in itself, however, would not justify finding an abuse of dominance. For such a conclusion to be reached, the Tribunal would have needed to satisfy itself in the second phase of the trial that Air Canada had a practice of committing anticompetitive acts that substantially lessened competition. It would be open to the Tribunal under existing Canadian law to find anticompetitive acts that did not lessen competition at all, let alone substantially. The possibility of such result casts further doubt on the wisdom of the current Canadian approach to predatory pricing in airlines.

(99) Press Release, Competition Bureau, Competition Bureau Settles Case with Air Canada (October 29, 2004), available at

(100) News Release, Industry Canada, supra note 15.

AUTHOR'S NOTE: Thanks to Kevin Davis, Michael Trebilcock, and Ralph Winter for comments on earlier drafts.

EDWARD M. IACOBUCCI, Associate Professor, Faculty of Law, University of Toronto.
COPYRIGHT 2006 Sage Publications, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2006 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Iacobucci, Edward M.
Publication:Antitrust Bulletin
Date:Jun 22, 2006
Previous Article:Antitrust policy and the oligopoly problem.
Next Article:A new look at critical elasticity.

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |