The study of strategic behavior—by which 1 mean efforts by established firms to take up advance positions in relation to actual or potential rivals and/or to respond punitively to new rivalry—is enormously complex. The early entry barrier models emphasized ex ante positioning.15 More recent work on predatory pricing has emphasized ex post responses.
Objections that have been or could be leveled at early entry barrier models and related applications to predatory pricing include: (1) The structural preconditions are not carefully stated; (2) whether it is more attractive to bar rather than accept entry is assumed but not demonstrated; (3) attention is focused on total costs, but the composition of costs and the characteristics of assets matter crucially and have been neglected; (4) the incentives to engage in predation are weak; and (5) cost asymmetries between established firms and potential entrants are asserted but rarely addressed. Recent work has made headway with each objection.
1. Structural Preconditions
As discussed above, the early entry barrier models purported to be oligopoly models. But the question of how oligopolists managed to achieve effective concurrence of market action—with respect to price, output, investment, and so on—was not addressed. The relevance of such models outside of the dominant firm context was thus questionable.
Recent models in the entry barrier tradition have essentially abandoned the oligopoly claim. The issues are posed instead in a duopoly context between a “sitting monopolist” and a potential entrant. Those who would apply those models to oligopoly presumably have the heavy burden of demonstrating their transferability.
Similar care has been taken in assessing claims of predation. The hazard here is that the legal process will be misused to discourage legitimate rivalry. There is growing agreement that the structural preconditions that must be satisfied before claims of predation are seriously entertained are very high concentration coupled with barriers to entry (Williamson, 1977, pp. 292-93). Joskow and Klevorick (1979, pp. 225-31) and Ordover and Willig (1981) concur and propose a “two-tier” test for predatory pricing. The subset of industries for which strategic behavior warrants public policy scrutiny would thus appear to be the following: (1) the sitting monopolist/duopolist situation; (2) regulated monopolies; (3) dominant firm industries; and (4) what William Fellner has referred to as “Case 3 oligopoly” (1949, pp. 47-49), an industry where an outside agency (e.g. a union) enforces collective action.17
2. Rationality of Preentry Deterrence
In principle, entry can be deterred in any of three ways: (1) by expanding output and investment in the preentry period, thereby to discourage the incen-tive to enter; (2) by threatening aggressive postentry responses; and (3) by imposing cost disadvantages on rivals. The latter two are addressed below. The first is in the spirit of Bain and Modigliani and has been dealt with more recently by Avinash Dixit, who models the entry problem in a duopoly context (1979, 1980). That permits him simultaneously to display and assess the profitability and feasibility of having the sitting monopolist adopt any of three postures: (1) behave in an unconstrained monopoly fashion; (2) expand output and investment so as to deter entry; and (3) accept entry by taking up a Stackelburg leadership position vis-à-vis the entrant. Dixit demonstrates that entry deterrence is optimal when fixed costs—actually, durable investments of a firm-specific kind—are of “intermediate” degree, whence the complaint that entry deterrence is an imposed rather than derived result can be dismissed if the requisite conditions are satisfied.
3. Costs, Assets, and Credibility
The standard entry barrier model assumes that potential entrants have access to the same long-run average total cost curve as do established firms. But the composition of costs, as between specific and nonspecific, is ignored. That poses the following anomaly: Extant firms and potential entrants are indistinguishable if all costs are nonspecific. The only “effective” entry deterring policy in circumstances where all costs are nonspecific is to set price equal to total cost, which is to say that entry deterrence is without purpose. The crucial role of sunk costs in entry deterrence is evident from an examination of Dixit’s (1979) formulation of the entry problem.
Even granting that entry deterrence sometimes is optimal, a further ques- tion arises as to how large a monopoly distortion can develop by reason of temporal asymmetry (the sitting monopolist has assets in place at the outset) and fixed cost conditions. Schmalensee has recently addressed the issue and shows that the preentry present value of excess profits that can be realized by established firms “cannot exceed the capital (start-up) cost of a firm of minimum efficient scale” and that scale economies are therefore of little quantitative importance from a welfare standpoint (1980, pp. 3, 8). That result is questionable, however, because it ignores the reputation effect incentives discussed under 3.4 below.
A related issue that has come under scrutiny is the matter of credible tiireats. This goes to the issue of what postentry behavior is appropriately ■mputed to the sitting monopolist. As Curtis Eaton and Richard Lipsey ob- (1980, p. 721), both credible and posturing threats take the same namely, “if y0u take action X, 1 shall take action Y, which will make 8mt X.” But credible and noncredible threats are distinguishable in that the party issuing the threat will rationally take action Y only if credibility conditions are satisfied. If the Nash response to X is indeed to take action Y, the threat is credible. But if, despite the threat, X occurs and the net benefits accruing to the party issuing the threat are greater if he accommodates (by taking action Z rather than Y), then the threat will be perceived as posturing rather than credible. Since such threats will be empty, Eaton and Lipscy have urged that analysis of strategic behavior focus entirely on threats for which credibility is satisfied. The translation of that argument into investment terms discloses that the sitting monopolist must invest in durable, transaction-specific assets if he is to preempt a market and deter entry successfully.
4. Reputation Effects
Robert Bork’s original assessment of the benefits of predation, the Areeda- Tumer criterion for assessing predation, Schmalensee’s measure of welfare distortion, and the Eaton and Lipsey treatment of credible threats all address the issue of entry and predation in a very narrow context. A large, established firm is confronted with a clearly defined threat of entry, and its response is assessed entirely in that bilateral context. The rationality of killing a rival (Bork, 1978) or of deterring an equally efficient firm (which has not yet made irreversible commitments) becomes the focus of attention (Eaton and Lipsey, 1980; 1981). If, however, punitive behavior carries signals to that and other firms—in future periods, in other geographic areas, and, possibly, in other lines of commerce— such analyses may understate the full set of effects on which the would-be predator is relying in his decision to discipline a rival. Assessing this requires that the issue of predation be addressed in a richer context in which information asymmetries and reputation effects are admitted.
Although their analyses do not make reference to the composition of costs—in particular, the specific versus nonspecific cost distinction is ignored— and for this reason are incomplete, recent articles by David Kreps and Robert Wilson (1982) and Paul Milgrom and John Roberts (1982) make considerable headway with those issues. As Milgrom and Roberts put it
[P]redation emerges as a rational, profit-maximizing strategy… not because it is directly profitable to eliminate the particular rival in question, but rather because it may deter future potential entrants. The mechanism by which this deterrent effect comes about is that by practicing predation the firm establishes a reputation as a predator. [1982, p. 281]
They develop the argument in an intertemporal game theoretic framework in which the usual assumption of complete information is relaxed.
Crucial to their argument is that potential entrants be uncertain as to how to interpret the behavior of the established firm. As they observe:
There are numerous reasons why this element of uncertainty should exist. On the one hand, the entrants could be [unsurc] about the game being played. For example, it might be that the established firm could actually be involved in some bigger game… A second possibility is that in the game actually being played, the established firm may be able to precommit itself to an aggressive course ot action and may have done so. Other scenarios involve the entrants allowing that the firm is not behaving as a fully rational game theorist. [Milgrom and Roberts. 1982. p. 303]
And they conclude with the observation that acts of ‘‘predation will only rarely need to be practiced. The credible threat of predation will deter all but the toughest entrants” (Milgrom and Roberts, 1982, p. 304).
The claim that predation is irrational and can be dismissed is thus evidently mistaken—or at least that would appear to be the judicious view to maintain until such time as those of the nonpredation persuasion can demonstrate wherein the recent treatments to which I refer are defective.
Applications pose the question of whether the circumstances where repu- tation effect incentives are strong can be recognized. An important consideration is whether local entry is being attempted into a small sector of the total market where the established firm enjoys dominance. Exploratory entry into a local geographic market or into one or a few products in a much broader line of related products would presumably enhance the appeal of sending a predatory signal. The likelihood that the observed behavior is strategic is increased in the degree to which (1) the response is intensively focused on the local disturbance (is carefully crafted to apply only to the market where entry is attempted) and (2)goes beyond a simple defensive response (e.g. holding output unchanged in the face of entry) to include a punitive aspect (e.g. . increasing output as the reply to entry).
5. Cost Asymmetries
Areeda and Turner (1975) take the position that the “predatory impact” of a pro* reduction by a dominant firm can be judged by whether such a reduction will exclude an equally efficient rival. As I have argued elsewhere, that is a peculiar criterion for assessing the welfare benefits of contingent increases in output—“now it’s there, now it isn’t, depending on whether an entrant has appeared or perished” (Williamson, 1977, p. 339). I did not, however, comment on the costs incurred by the entrant except in passing (pp. 296, 303- 4). In consideration of the series of strategic cost disadvantages that an entrant experiences or may be made to bear in relation to an established firm, that is a regrettable oversight.
There are two points here, the first of which is that history matters in assessing costs. Temporal cost differences can arise in operating cost, cost of capital, And learning curve respects. The second and more significant point is that the established firm may by its own actions be responsible for added cost differences of all of those kinds.
Many of the issues here have been developed elsewhere (Williamson, 1968; Spence, 1981), and some are discussed in earlier chapters. Suffice it to observe here that the equally efficient rival criterion is primarily suited to static circumstances where historical differences and contrived cost asymmetries may be presumed to be absent.19 To the extent that actual circumstances are not accurately described in that way, allowance for cost differences may be necessary if an informed assessment of predation is to be realized.
Source: Williamson Oliver E. (1998), The Economic Institutions of Capitalism, Free Press; Illustrated edition.