Section 2 Symposium: Thom Lambert on Defining and Identifying Exclusionary Conduct

Thom Lambert —  5 May 2009

lambertThom Lambert is an Associate Professor of Law at University of Missouri Law School and a blogger at Truth on the Market.

There’s a fundamental problem with Section 2 of the Sherman Act: nobody really knows what it means. More specifically, we don’t have a very precise definition for “exclusionary conduct,” the second element of a Section 2 claim. The classic definition from the Supreme Court’s Grinnell decision — “the willful acquisition or maintenance of [monopoly] power as distinguished from growth or development as a consequence of a superior product, business acumen, or historic accident” — provides little guidance. The same goes for vacuous statements that exclusionary conduct is something besides “competition on the merits.” Accordingly, a generalized test for exclusionary conduct has become a sort of Holy Grail for antitrust scholars and regulators.

In its controversial Section 2 Report, the Department of Justice considered four proposed general tests for unreasonably exclusionary conduct: the so-called “effects-balancing,” “profit-sacrifice/no-economic-sense,” “equally efficient competitor,” and “disproportionality” tests. While the Department concluded that conduct-specific tests and safe harbors (e.g., the Brooke Group test for predatory pricing) provide the best means of determining when conduct is unreasonably exclusionary, it did endorse the disproportionality test for novel business practices for which “a conduct-specific test is not applicable.” Under the disproportionality test, “conduct that potentially has both procompetitive and anticompetitive effects is anticompetitive under section 2 if its likely anticompetitive harms substantially outweigh its likely procompetitive benefits.”

According to the Department, the disproportionality test satisfies several criteria that should guide selection of a generalized test for exclusionary conduct. It is focused on protecting competition, not competitors. Because it precludes liability based on close balances of pro- and anticompetitive effects, it is easy for courts and regulators to administer and provides clear guidance to business planners. And it properly accounts for decision theory, recognizing that the costs of false positives in this area likely exceed the costs of false negatives.

While it has some laudable properties (most notably, its concern about overdeterrence), the disproportionality test is unsatisfying as a general test for exclusionary conduct because it is somewhat circular. In order to engage in the required balancing of pro- and anticompetitive effects, one needs to know which effects are, in fact, anticompetitive. As the Department correctly noted, the mere fact that a practice disadvantages or even excludes a competitor does not make that practice anticompetitive. For example, lowering one’s prices from supracompetitive levels or enhancing the quality of one’s product will usurp business from one’s rivals. Yet we’d never say such competitor-disadvantaging practices are anticompetitive, and the loss of business to rivals should not be deemed an anticompetitive effect of the practices.

“Anticompetitive” harm presumably means harm to competition. We know that that involves something other than harm to individual competitors. But what exactly does it mean? If Acme Inc. offers a bundled discount that results in a bundle price that is above the aggregate cost of the products in the bundle but cannot be met by a less diversified rival, is that a harm to competition or just a harm to the less diversified competitor? If Acme pays a loyalty rebate that results in an above-cost price for its own product but usurps so much business from rivals that they fall below minimum efficient scale and thus face higher per-unit costs, is that harm to competition or to a competitor? These are precisely the sorts of hard (and somewhat novel) cases in which we need a generalized test for exclusionary conduct. Unfortunately, they are also the sorts of cases in which the Department’s proposed disproportionality test is unhelpful.

A problem with most of the generalized tests that have been proposed is that they never actually define “unreasonably exclusionary.” Instead, they aim merely to identify instances of unreasonably exclusionary conduct. Like the disproportionality test, the effects-balancing test (which resembles the disproportionality test but doesn’t require that anticompetitive harms “substantially” outweigh procompetitive benefits) is somewhat question-begging in that it equates exclusionary conduct with a balance toward anticompetitive effects but never says what anticompetitive means. Similarly, the “no-economic-sense” test, which would condemn a business practice that would make no economic sense but for its ability to exclude rivals from the market, merely states the circumstances under which one may assume a business practice is unreasonably exclusionary; it never says what it is about an instance of exclusion that makes it “unreasonable.”

The one exception to this observation about definitional deficiencies is Judge Posner’s “equally efficient competitor” test, which says that an action is unreasonably exclusionary insofar as it is capable of excluding from the defendant’s market an equally or more efficient rival. Under that test, the essence of “unreasonable exclusionariness” (or “anticompetitiveness”) is doing something that could rig the game — that could result in a winner other than the most deserving firm. Such conduct is, quite literally, anti-competitive.

In lauding the Posner test for providing an actual definition of “unreasonably exclusionary” conduct, I do not mean to imply that the test should be the general means of identifying such conduct. As I have elsewhere explained, the test is somewhat under-deterrent (albeit perhaps appropriately so), and it may be difficult to apply to non-pricing practices. The test does have the virtue, though, of providing an actual definition of exclusionary conduct. When a court, regulator, or business planner is confronted with a novel practice and needs to assess liability, it will want to go back to basics and ask, “What exactly does it mean for a practice to be unreasonably exclusionary?” The Posner test provides an answer to that question.

Perhaps for that reason, the Posner test has been invoked quite successfully when novel practices have arisen. For example, in Ortho Diagnostic v. Abbott Labs., 920 F. Supp. 455 (S.D.N.Y. 1996), the district court went back to basics and relied upon the Posner definition of unreasonably exclusionary conduct to craft a workable rule for evaluating the legality of bundled discounts.

At the end of the day, an ideal general standard for exclusionary conduct might incorporate both a relatively administrable “identifying” test, such as the no-economic-sense test, and a definition of exclusionary conduct, such as that offered by Judge Posner. For example, the no-economic-sense test might be used to establish presumptive liability, but the defendant could rebut by showing that its conduct could not have been unreasonably exclusionary because it could not have excluded an equally efficient rival. In any event, given that the means of both legitimate competition and illicit exclusion are myriad, any approach seeking to give guidance on as-yet-untested business practices ought to include some actual definition of “unreasonable exclusionariness.”

Thom Lambert

Posts

I am a law professor at the University of Missouri Law School. I teach antitrust law, business organizations, and contracts. My scholarship focuses on regulatory theory, with a particular emphasis on antitrust.

4 responses to Section 2 Symposium: Thom Lambert on Defining and Identifying Exclusionary Conduct

  1. 
    Thom Lambert 6 May 2009 at 3:22 pm

    Thanks for the comments, both of which emphasize the administrative difficulties the Posner test creates. I agree that the test may be difficult to administer and, for that reason (among others), may not be an ideal generalized liability test. My point, though, was to emphasize a virtue of the test relative to its alternatives: It alone actually says what it means for an action to be unreasonably exclusionary — it’s unreasonably exclusionary if it’s capable of driving a subset of rivals (those who are “equally efficient”) from the market. Now, we can quibble over what it means to be “equally efficient,” but we can at least come up with a non-circular definition of that concept. By contrast, to say that an action is unreasonably exclusionary if its anticompetitive costs far exceed its procompetitive benefits, where “anticompetitive” costs are those resulting from a harm to competition (not simply to a competitor), where “harm to competition” is…well, something like unreasonable exclusion, is a circular exercise. A court or business planner confronting a novel practice that may disadvantage some competitors can’t take any guidance from that approach. The Posner test at least tells courts and planners what the point of Section 2 is: it is to ensure that dominant firms don’t engage in conduct that could rig the game by driving deserving (“equally efficient,” however that is defined) rivals from the market.

    While the Posner test may not be administrable enough to serve as a general standard for exclusionary conduct, it might work well as a sort of “super safe harbor.” To the extent it adequately reflects the essence of “unreasonable exclusionariness,” a defendant who could show that his conduct did not run afoul of the test should be able to avoid liability.

    Prof. Hovenkamp mentions two other problems with the Posner test: it does not condemn practices that prevent rivals from becoming as efficient as the defendant, and it may permit exclusion of the only competition a dominant firm is likely to face if that competition is less efficient than the dominant firm. I address those two arguments in my article, “Weyerhaeuser and the Search for Antitrust’s Holy Grail,” 2006-07 Cato Sup. Ct. Rev. 277, 304-310. As I explain in that article, I agree that the test is somewhat underdeterrent for the reasons mentioned, but I am not troubled by that fact because (1) there’s already a measure of overdeterrence built into Section 2, given that the exclusionary acts it polices are not clandestine and thus don’t warrant treble damages, and (2) most “scale-depriving” exclusionary acts (those that could prevent a rival from attaining equivalent efficiencies) are either (a) already prohibited by Section 1, (b) subject to immunities that would preclude Section 2 liability in any event, or (c) incapable of excluding an aggressive and competent rival. I refer interested readers to pages 304-10 of the Weyerhaeuser piece, available here.

  2. 
    Herb Hovenkamp 5 May 2009 at 7:39 am

    If what we mean by “balancing” is the expression of benefit against harm in a cardinal unit to which weights can be assigned, then balancing in §2 cases will never occur except in the most obvious situations. And it really does not matter what we want to balance – social harm; consumer loss, net welfare loss, aggregate producer gain, or something else. It is difficult enough simply to articulate all the possible positive and negative effects that a practice can have, and to determine how long a time horizon we are able to examine. So if balancing is to be done, it must be reserved for the smallest possible range of cases.

    Hence the importance of a series of “screens” that enable us to cut away the easier cases and deal with them. First, of course, is power. Second is prima facie capacity to exclude – is the practice capable of excluding one or more rivals from a market or limiting their output so as to prolong, enlarge, or create power in the dominant firm? Third is the “no benefit” class: if the conduct harms in the sense of the second point but confers no benefit at all, there is nothing to balance. Fourth is at least a cursory search for less restrictive alternatives: If the conduct causes harm in the point two sense but arguably produces a social benefit as well, could the benefit have been attained in a less restrictive fashion? This factor must be used sparingly, for a monopolist is not a trustee for the public. Only if we get past this point and are left with conduct that arguably causes both harms and benefits with no obvious less restrictive alternatives, then we may have to engage in a crude form of balancing, with close cases to be resolved in favor of the defendant. That is, nonintervention is the norm, and intervention must be justified.

    To this end, we are always better off if we can formulate more precise rules for specific practices, and in the last 25 years we have made substantial progress with respect to various pricing practices – predatory pricing, predatory purchasing, loyalty discounts, package discounts, and the like. We have hardly resolved all the disputes, but at least the counselor has something to take to the client. Even here we must be cautious, however. The problem with pricing rules is not ambiguity so much as technical specification. For example, the average variable cost (AVC) or average avoidable cost (AAC) tests for predatory pricing seem easy until a firm that produces multiple products in the same plant with nontrivial fixed costs has to figure out how to apply them. Nevertheless, these tests are a significant improvement over the more open ended §2 inquiries.
    Thom makes some excellent points. The Posner “capable of excluding an equally efficient rival” test works very well in cases where we can figure out how to apply it, but so far that is mainly pricing. Further, there is an important question about “actual” as opposed to hypothetical equally efficient rivals. Bundled discounts provide a good example: a dominant firm might bundle and discount products A and B at a price that flunks the AMC’s attribution test (of which more tomorrow) but in the aggregate are above cost. The actual rival, who makes only B, gets excluded even though its costs of producing B are the same as those of the dominant firm. But the practice would not exclude a hypothetical equally efficient firm that produces both A and B. These problems also vexed simple predatory pricing law back in the 1970s and 1980s, but we abandoned them in favor of more coherent administrability. For example, suppose a firm’s economies of scale top out in the 30%-40% market share range. The dominant firm uses above cost price cuts to deny its rivals access to sufficient output to attain such economies. Its prices do not exclude a hypothetical equally efficient rival, but they do exclude the only rivals on the scene, and their persistence may ensure that an equally efficient rival will not emerge, at least not for a long time. This is not an argument for condemning above cost predatory pricing; I’m a firm believer in cost-based tests. Rather, it is simply an observation that even the equally efficient rival test has to be applied with some sensitivity to the circumstances as well as to what is realistically available.

  3. 
    Alden Abbott 5 May 2009 at 7:05 am

    Good points, Thom. But even Judge Posner’s test is less than ideal because it begs the question of what is an “equally efficient competitor.” In a bundling scenario, for example, the fact that a firm may be “equally efficient” in producing and distributing one product does not necessarily mean that it is truly “equally efficient” if the marketplace is characterized by “bundle to bundle” competition. Moreover, the cost calculations that are required for an “equal efficiency” determination may be hard to make if not intractable; problems of joint and common costs (especially in a multi-product scenario) and opportunity cost, among others, may introduce high error costs to that process. This is not to deny that Judge Posner’s contribution is extremely important, just to point out that his metric, as is the case with all other Section 2 tests, presents major complications.

Trackbacks and Pingbacks:

  1. TRUTH ON THE MARKET » PeaceHealth and De Facto Exclusive Dealing, Part III - November 13, 2009

    […] as I’ve noted on this blog and elsewhere, we don’t have an agreed upon understanding of what it means for conduct to be […]