Wednesday, May 1, 2013
Posted by Gus HurwitzI must start by thanking both Danny for inviting me to participate in this symposium, and Scott & Tim for what is an interesting, and surely will be an influential, article. At the same time, as a junior scholar, and knowing that this will be my first introduction to many in this community, I approach this with some trepidation. The reason for this trepidation is simple: my response to the article, written by two established scholars whom I respect a great deal, is best politely described as “critical.”
Parallel Exclusion is an important article – one that is grounded in real concerns and that will influence discussion for years to come. Unfortunately, it overemphasizes a coarsely developed game-theoretic argument that parallel exclusionary conduct is more stable than parallel price-escalation without sufficiently making the case that such conduct is actually harmful. And, to the extent that the conduct is harmful, the paper assumes both the insufficiency of existing antitrust law and the viability of regulatory intervention to address these concerns. In so doing fails to make the case that “U.S. antitrust doctrine should be adjusted to address anticompetitive parallel exclusion more effectively.”
The examples sprinkled throughout the article – few of which are considered in any depth – make both of these points well. In many of the cases, the allegedly anticompetitive conduct has subsequently been shown to be procompetitive (e.g., Paramount). Other of the examples resulted in successful litigation under Section 1 or Section 2 (e.g., Visa-MasterCard, Allied Tube, & American Tobacco).
Troublingly, the majority of the cases used as examples actually militate against a pressing need for greater consideration of parallel exclusion concerns. Space precludes (excludes?) a comprehensive analysis of each example – let’s look instead only at Visa-MasterCard & Allied Tube, the examples with which the authors lead off their article and which therefore are presumably their strongest examples.
If these are their best examples of parallel exclusion, the case for parallel exclusion is quite weak. As an initial matter, both cases ultimately did result in findings of liability – it’s not clear what more adding a parallel exclusion theory to these cases would accomplish.
The complex economics of the Visa-MasterCard litigation helped give rise to the study of a new class of markets in the economics literature (two- or multi-sided markets). In these markets, exclusion is often welfare-increasing. Given that our understanding of these markets was in its earliest stages throughout most of this litigation, had the courts proceeded along a parallel exclusion theory, the likelihood of error would have been very substantial indeed.
In the end, the government successfully litigated Section 1 claims against Visa and MasterCard; Visa and Mastercard have since settled other, related, claims brought by both by the DOJ and by merchants. This, despite ongoing debate whether the alleged conduct was in fact harmful to consumers. Of course, such debate follows almost every major antitrust case – and this is, in fact, very healthy, as it continues to provide antitrust laws and economists with better understandings of whether and how certain behavior in fact harms consumers and competition. As discussed in the second half of this response, the authors’ parallel exclusion theory gives insufficient weight to the likelihood that given conduct is harmful, contrary to current approaches which take harm as central to the analysis.
Turning to Allied Tube, there the exclusionary conduct was only possible because it occurred in the context of an industry organization operating with rules that facilitated collusive conduct. The collusive nature of the conduct was so blatant that the defendant didn’t even contest it at trial, instead attempting to rely (unsuccessfully) on a Noerr-Pennington defense. This is an excellent case to demonstrate why organizations need to be wary of facilitating collusion among their members (conduct that is relatively often anticompetitive); it is a poor case to demonstrate why we should be concerned about firms acting in parallel, since here they weren’t merely acting in parallel but were actively colluding.
The authors describe Allied Tube as “a textbook example of parallel exclusion,” suggesting that the steel conduit interests acted in parallel but independently. But this is not the case – there was express agreement among the colluding parties. Prior to the meeting at which the steel conduit interests voted to exclude the new PVC conduit from the National Electric Code, Allied Tube “met to plan strategy with, among others, members of the steel industry, other steel conduit manufacturers, and its independent sales agents. They collectively agreed to exclude respondent's product from the 1981 Code … .” Allied Tube v. Indian Head, Inc., 486 U.S. 492, 496 (1988).
More generally, that many of the examples were successfully litigated under Sections 1 or 2 recalls the argument for price squeeze claims – an argument rejected by the Court in Linkline largely because a price squeeze claim is merely a combination of a RRC/RTD claim with a predatory pricing claim. The essence of the parallel exclusion claim is that firms are acting in parallel (a Section 1 claim) to engage in exclusionary conduct (a Section 2 claim). I don’t know that the authors have identified a single example that could not be analyzed under this framework. At a theoretical level, some Section 1 conduct may only be possible contingent upon Section 2 conduct, or vice versa – but where that may be the case the better approach would be to incorporate awareness of this contingency into the analysis of the Section 1 & 2 claims, not to develop a new hybrid claim.
The discussion so far has focused on the examples offered to demonstrate that parallel exclusion is a problem meriting attention and concern they are insufficient to this task. Moving beyond the examples, let’s turn to the article’s substantive discussion.
Modern antitrust incorporates an appreciation for regulatory humility and an understanding of error costs. Wise men are not our only regulatory angels; the market disciplines our excesses too, and often better. The article shows an implicit awareness of this – it must be why they built their paper around a game-theoretic model that argues that parallel exclusion is more durable than other forms of parallel conduct. The underlying message of their approach is clear: parallel exclusion is durable, therefore the market won’t correct it, therefore wise men must.
But an error costs framework isn’t only concerned with the likelihood of errors. It is also concerned with their costs. Exclusionary conduct is very often procompetitive. Regulatory intervention (which includes judges using antitrust law to tell firms how and with whom they must do business) is very often harmful to the competitive process – indeed, government intervention is often the strongest and most durable barrier to entry, and such barriers are perhaps the purest form of exclusionary conduct. Unless the authors can make the case that their interventions would typically only occur in the minority of cases where exclusion is anticompetitive, and the remedies that they would impose are typically less costly than the conduct at issue, the interventions they seek are unsupportable.
Let us consider this question directly, in two parts: whether parallel exclusion is likely harmful, and whether regulation is an effective remedy.
The article offers a brief discussion of the harms of parallel exclusion. This discussion can be summarized by saying that exclusion “slow[s] or block[s] product innovation.” (They also express some concern about higher prices; but their focus is clearly on harms to innovation.) They start with the assuredly correct proposition that dynamic efficiency contributes more to social welfare than static efficiency, and conversely that harms to dynamic efficiency are more problematic than those to static efficiency.
It is at this point that the arguments makes a great & flawed leap. They assert that exclusion poses a greater threat to dynamic efficiency than does price elevation; and, in industries marked by rapid technological change exclusion is therefore the “supreme evil” against which antitrust must protect. This conclusory statement is based on the wrong premise that exclusion necessarily slows innovation. The authors offer some curious defense of this proposition in footnote 137. This footnote recapitulates the Schumpeter-Arrow debate, but in a way that suggests Arrow has won. It starts by conceding that “some commentators have taken the view that more exclusion of competitors by incumbents, rather than less, would promote innovation,” and then dismissively explains that “[t]hese premises have been challenged on multiple grounds.”
The authors then assert, still in the footnote, that “As an empirical matter, dramatic innovations in the twentieth century have tended to come from outsiders, not incumbent firms.” The primary support that the authors offer is Tim’s book, The Master Switch. While perhaps argued, Tim’s book is hardly a definitive word on the matter. See, e.g., Peter Decherney, Nathan Ensmenger, Christopher S. Yoo, Are Those Who Ignore History Doomed to Repeat it?, 78 U. Chi. L. Rev 1627 (2012) (reviewing Tim Wu’s The Master Switch). The history of the twentieth century contains numerous examples that cast doubt on this assertion: Bell Labs, Apple, Google, and the embarrassment of the 1996 Telecom Act all come to mind. Or consider the economics literature on general purpose technologies, which makes clear that vertical integration (that is, exclusion) may increase the social value of a technology. See Bresnahan & Trajtenberg , General Purpose Technologies "Engines of Growth?", 65 J. Econometrics 83, 94–96 (1995). More generally, the modern economic consensus answer to this question can be summarized as “it’s complicated.”
(For completeness, I must note that Parallel Exclusion does cite to other sources. They cite to Bohannan & Hovenkamp for the proposition that “dominant firms have frequently slowed innovation” (emphasis added) – but if we’re operating in the context of dominant firms, a parallel exclusion theory is duplicating the work of Section 2, and therefore not needed. They also cite Jon Baker to support the initial proposition that innovation tends to come from outsiders – but Baker’s conclusions are far more nuanced that those suggested in footnote 137. Indeed, contrary to the article’s baseline conclusion that “U.S. antitrust doctrine should be adjusted to address anticompetitive parallel exclusion more effectively,” Baker urges that “current U.S. antitrust rules and enforcement priorities are on the whole well-targeted to foster innovation.” Baker would likely treat the idea of parallel exclusion under his discussion of “the ongoing debate over the appropriate legal standard for identifying exclusionary acts” in winner-take-all or winner-take-most industries. While he does advocate greater antitrust involvement in such industries, he would accomplish this by tweaking the Section 2 standards.)
More problematic than the weak case that parallel exclusion is harmful, the authors do not consider the possible harms on the other side of the balance: the costs and potential harms of the regulatory edifice. I won’t belabor this point: regulation is costly, regulators are subject to capture, and they often don’t understand the market. Thank goodness the FTC ensured that AOL couldn’t exclude its competitors from the instant messaging market! It is alarming and dangerous that these costs do not seem to factor into the article’s analysis. There is no law without remedy; where, as often the case in antitrust, the remedy is worse than the offense it is better to do without law.
That all said, and to conclude on a more positive note, I am sympathetic to the article’s central doctrinal argument: oligopolistic exclusionary conduct surely can be just as problematic as monopolistic exclusionary conduct, and it is problematic if antitrust doctrine systematically and unjustifiably disadvantages one type of claim over the other. Indeed, this argument isn’t unique to the authors: for instance, both the EU and Canada embrace the concept of joint abuse of dominance – and their approaches have been criticized for much the same reasons that I have discussed here.
The article gives the impression that the authors’ task is beyond mere tweaking of existing standards. I don’t think that they have successfully made the case for reform on the scale envisioned – I am not convinced that the law is seriously miscalibrated to the detriment of oligopolistic exclusion claims, and like Baker, I would try to reach the concerning conduct by tweaking the standards within the existing Section 1 and 2 frameworks. In part, this is because I am more cautious about error costs and the viability of regulatory intervention than Hemphill and Wu. Regardless, several of the authors’ doctrinal recommendations, if read modestly, are quite reasonable and should be taken seriously.
Unlike the authors, I find Section 1 enforcement based upon horizontal agreements to exclude unproblematic. It is curious that so many of the cases used as examples throughout the paper successfully take this approach, but that the authors nonetheless demand more. In cases where firms engage in parallel exclusion that is facilitated by their participation in (for instance) an industry organization, the use of MFN clauses, or agency relationships, (collectively, vertical intermediaries,) I would argue for Section 1 liability for the vertical intermediary and its participants. Such an argument may apply, for instance, to Allied Tube and FOGA (and, more recently, industry Standard Setting Organizations such as JEDEC, see, e.g., Justin (Gus) Hurwitz, The Value of Patents in Industry Standards, 36 AIPLA Q.J. 1, 34 (2007) (“… SSOs themselves should arguably be subject to Section 1 liability for creating rules that are insufficient to protect social welfare from the great power that an SSO wields.”)).
But in the case of oligopolies, I am also sympathetic to the author’s general case for joint monopoly claims. A firm that is able to engage in anticompetitive exclusionary conduct due to the interdependence of that conduct with other firms’ conduct should be subject to Section 2 liability – attenuated by an understanding of error costs. But I would base this on an argument that that firm has market power: it can act unilaterally in a way that the market does not discipline. Traditional understanding of market power is myopically, and dangerously, focused on price – a problem endemic to antitrust analysis in non-commodity industries. Adjusting our understanding of market power to incorporate non-price factors would capture most of the residual conduct that the authors are concerned about that would not be captured by Section 1 claims – and it would do so through a much more modest and theoretically sound adjustment to existing standards.