Monday, May 20, 2013
Posted by Scott Hemphill and Tim Wu
We thank the commenters for their thoughtful and generous reactions.
In a nutshell, parallel exclusion is conduct by multiple firms that blocks or slows the entry of would-be competitors. Parallel exclusion can be far more harmful than parallel pricing, yet receives much less attention. In our article, we identify the mechanisms and effects of parallel exclusion; assess its surprising resistance to internal collapse, as compared to ordinary cartels; and offer interpretations of existing antitrust doctrine that would address its anticompetitive forms.
The commenters, a set of first-rate lawyers and economists, offer a variety of extensions, clarifications, and challenges to our legal and economic analysis. In this post, we offer reactions to a few of their reactions.
Doctrine of parallel exclusion
Professors Baker and Waller address the legal doctrine that might be used in a parallel exclusion case. Three points bear special emphasis.
Baker says what’s needed is a test case—a “Prince Charming” in Baker’s Cinderella metaphor—to give a full vetting to the parallel exclusion approach. The point is well taken. A major problem is that potentially important parallel exclusion cases never get past the initial stage of antitrust scrutiny, probably based on a fear of wasting limited resources on unproven theories. It’s a reasonable concern: many federal judges seem resistant to any but the most established theories of antitrust (like price-fixing), so why, if you are an enforcer, put precious lawyers to work on such a case?
Harm should be the guiding star for enforcers in this area. In fact, parallel exclusion cases do get brought when the harm is clear and the conduct outlandish, and also when agreement, though of minor or no importance to the underlying economics, is easy to establish. Unfortunately, we suspect other potential cases are ignored, but we hope our article convinces enforcers to look for the harm that parallel exclusion can cause.
Next, Baker worries that the Supreme Court’s decision in Brooke Group might stand in the way of recognizing parallel exclusion as a viable theory. In that case, the Court considered parallel predation conducted by oligopolists. The Court expressed skepticism about oligopoly predation, in part because it thought that the losses and gains would be difficult to allocate. The “anticompetitive minuet,” the Court wrote, would be “difficult to compose and perform.”
On the other hand, the Court has already recognized parallel exclusion where it takes the form of multiple vertical exclusive agreements. Standard Stations established the principle that multiple exclusive contracts can be added up in evaluating their collective effect. The case has been criticized by commentators, but not on this ground, and the Court has not backed away from this analysis. Moreover, as we emphasize in the article, often the coordination is in fact easy to compose and perform, and in such settings this dicta from Brooke Group has little force.
Finally, we’d like to echo one point emphasized by Waller in his kind and generous review. It is very common for the media, courts, and antitrust lawyers to point to industry-wide practice as proof that particular conduct cannot be harmful. But that’s exactly the wrong way to look at things: an industry-wide practice might be procompetitive, or might instead be the anticompetitive conduct of a constructive monopoly. In discussing the FTC’s Google investigation, observers too often said the fact that Bing did what Google was doing meant the practice couldn’t be anticompetitive, which is just illogical. In a recent UK example, Office of Fair Trading issued a statement of objections alleging that booking.com (owned by Priceline) and Expedia had both entered agreements with InterContinental Hotels, which had the effect of preventing online travel agents from competing by offering discounts. Here, too, the widespread nature of the practice is hardly a defense.
Economics of parallel exclusion
We see an important research opportunity for economists to model, in a rigorous way, the conditions under which inefficient parallel exclusion can be expected to arise. Professors Wickelgren, Baker, and Bar-Isaac suggest helpful next steps along this path. Wickelgren identifies coordination among the excluders, in determining which excluder will work with which buyer, as one element of a rigorous model. This concern is more important for exclusion accomplished through buyer contracts, less so for other exclusionary mechanisms. Wickelgren also notes a further modeling element, compared to off-the-shelf models of exclusion by a single dominant firm: profit erosion through competition among the excluders.
Baker identifies a third issue that is ripe for formal modeling, namely the relative stability of parallel exclusion compared to parallel price elevation. Finally, Bar-Isaac notes the value of further theoretical work to specify the structure of upstream and downstream market segments and the relationship between buyers and sellers. As Bar-Isaac emphasizes, upstream exclusion and downstream exclusion can be complementary, as firms in each segment act one another’s behalf. All of these suggestions are helpful in expanding on the single-firm exclusionary models that have predominated in the economics literature.
Hurwitz takes a critical stance, concluding that our analysis is neither new nor correct. The stance is surprising, because he agrees with much of what we say. He accepts that “oligopolistic exclusionary conduct surely can be just as problematic as monopolistic exclusionary conduct….” He agrees that “it is problematic if antitrust doctrine systematically and unjustifiably disadvantages one type of claim over the other.” And he is “sympathetic” to our advocacy of what he calls “joint monopoly claims,” to be addressed as monopolization under section 2, to handle interdependent action by excluders that would not be readily captured as an agreement subject to section 1. (The details of his proposed test are different from ours.)
So where is the disagreement, exactly? Hurwitz argues that parallel exclusion might not be so harmful, and even if it is sometimes harmful, existing doctrine can handle any harm that arises. How does he know? Because there are cases that use existing doctrine to deal with parallel exclusion problems. The problem inherent in this proof should be immediately apparent.
How much harm is actually caused by parallel exclusion? No one knows. Given that fact, we certainly cannot know whether current doctrine is sufficient merely by looking at cases that use it. Stated otherwise, relying on successful uses of section 1 to combat anticompetitive exclusion tells us nothing about whether those cases are the tip of the iceberg, so to speak, or, as Hurwitz seems to assume, the whole iceberg.
At a deeper level, our main interest is trying to understanding when and how parallel exclusion is harmful; how doctrine ought deal with anticompetitive parallel exclusion is a secondary concern. Cases such as Visa and Allied Tube, among many others discussed in our article, show the potential harms of parallel exclusion. While apparently accepting that Visa and Allied Tube featured anticompetitive schemes, and that antitrust intervention was appropriate in both, Hurwitz draws the conclusion from these examples that “the case for parallel exclusion is quite weak.” By this, he means that antitrust doctrine successfully reached these cases, without any need for our analysis. But that misses the point, because we don’t think it’s the end of the world if (say) a public enforcer uses a section 1 agreement “hook” to reach parallel exclusion cases. It’s a little like prosecuting Al Capone for tax evasion, but at least the government recognizes the harm.
Another point that Hurwitz challenges is our assertion that anticompetitive exclusionary conduct is detrimental to innovation. He says we are discounting Schumpeter, and points out that organizations like “Bell Labs, Apple, [and] Google” have been very innovative. But this confuses large organizations with exclusionary behavior. We are not arguing that large organizations cannot be innovative. Rather, we’re claiming that self-entrenching behavior has bad effects on innovation, not the mere fact of size. Exclusionary conduct may lead to size, but it isn’t necessary to it, nor clearly necessary to large-firm innovation of the kind that Schumpeter and others have celebrated.