Section 2 Symposium: Josh Wright on An Evidence Based Approach to Exclusive Dealing and Loyalty Discounts
Josh Wright is a Professor of Law at George Mason Law School, a former FTC Scholar in Residence and a regular contributor to Truth on the Market.
The primary anticompetitive concern with exclusive dealing contracts is that a monopolist might be able to utilize exclusivity to fortify its market position, raise rivals’ costs of distribution, and ultimately harm consumers. The unifying economic logic of these anticompetitive models of exclusivity is that the potential entrant (or current rival) must attract a sufficient mass of retailers to cover its fixed costs of entry, but that the monopolist’s exclusive contracts with retailers prevent the potential entrant from doing so. However, the exclusionary equilibrium in these models are relatively fragile, and the models also often generate multiple equilibria in which buyers reject exclusivity. At the exclusive dealing hearings where I testified, a sensible consensus view emerged that a necessary condition for exclusive dealing or de facto exclusive contracts such as market-share discounts or loyalty discounts to cause competitive harm is that they deprive rivals of the opportunity to compete for access to distribution sufficient to achieve minimum efficient scale. The Report (p. 137) reflects this consensus:
In particular, exclusive dealing may be harmful when it deprives rivals “of the necessary scale to achieve efficiencies, even though, absent the exclusivity,” more than one firm “would . . . be large enough to achieve efficiency.”68 In other words, exclusive dealing can be a way that a firm acquires or maintains monopoly power by impairing the ability of rivals to grow into effective competitors that erode the firm’s position. As one panelist put it, “the exclusive dealing case that you ought to worry about” is where exclusivity deprives rivals of the ability to obtain economies of scale.
The Report also goes on to note the competitive justifications for exclusive dealing, ranging from the variety of ways in which exclusive dealing can prevent free-riding, facilitate relationship-specific investments, and intensify manufacturer competition for scarce retailer shelf space or access to distribution with the benefits of that intensified competition passed on to consumers in the form of lower prices or higher quality.
The situation antitrust enforcers find themselves in with respect to exclusive dealing is not unfamiliar. On the one hand, there are a set of possibility theorems which indicate that exclusive dealing and de facto exclusives can lead to anticompetitive outcomes under some specified conditions, including substantial economies of scale or scope. On the other, there are a set of sensible and economically rigorous pro-competitive justifications for the practice. On top of that is the casual empiricism that we observe exclusive dealing contracts in competitive markets and adopted by firms without significant market power. As David Evans noted on the first day of our symposium, quite a bit can be learned about the relative probabilities of anticompetitive and pro-competitive uses of certain types of business behavior by understanding the incidence of use by competitive firms. Exclusive dealing is no different.
Still, we find ourselves between battling theories. The standard error cost approach to this problem, an approach discussed by many in this symposium as a powerful tool to ensure that our liability rules do not do not needlessly harm consumers by overdeterring pro-competitive conduct or under-deterring anticompetitive conduct, is to turn to the evidence. What do we know about the incidences of anticompetitive exclusive dealing and de facto exclusive dealing contracts? The question is not one of the logical validity of any of the competing theories. It is one of their empirical (and therefore policy) relevance. A sensible approach to designing antitrust liability rules for exclusive dealing would be to design a conduct-specific standard sensitive to the particular relative risks of Type I and Type II errors informed by the best available existing evidence. Of course, it should be noted that more evidence is always better and there is certainly a need for more empirical research about single firm conduct. But the limited nature of the evidence does not mean we have zero information to update our priors on the critical policy question.
So what does the evidence say? What approach would it lead to? And how does that approach compare with that endorsed in the Section 2 Report? I’ll focus on those issues in the remainder of the post.
Let’s start with the existing evidence. While empirical evidence on exclusive dealing contracts’ competitive consequences are scarce, even relative to other types of vertical restraints such as RPM, the evidence is not insignificant and supports the view that exclusive dealing is much more likely to be output-enhancing than anticompetitive. For instance, Heide et al. conducted a survey of managers responsible for distribution decisions and found that the incidence of exclusive dealing was correlated with the presence of “free-ridable” investments. Both Asker and Sass separately examine the welfare consequences of exclusive dealing in the beer market by observing the effect of exclusive dealing on total market output, as well as the output and prices of rival distributors, concluding that exclusive dealing is output increasing and does not generate foreclosure. Wright (2007) provides evidence that restrictive shelf space contracts are not associated with anticompetitive effects. Lafontaine and Slade’s survey of exclusive dealing contracts and vertical integration concludes that the practices are generally efficient and not associated with anticompetitive outcomes (see also Cooper et al, 2005; I also summarize the empirical evidence on exclusive dealing here). After their own detailed and comprehensive review of the existing empirical evidence, Lafontaine and Slade conclude that:
In general then, the empirical evidence leads one to conclude that consumer well being tends to be congruent with manufacturer profits, at least with respect to the voluntary adoption of vertical restraints. When the government intervenes and forces firms to adopt (or discontinue the use of) vertical restraints, in contrast, it tends to make consumers worse off.
While the empirical evidence does not conclusively demonstrate that exclusive dealing is always pro-competitive, there is little support for the anticompetitive theories in the literature and certainly greater support for the conventional pro-competitive view of exclusive dealing contracts. While further empirical research on exclusive dealing contracts and their competitive consequences is needed, and I am currently working on a project with using variance in state exclusive dealing laws in the beer industry to identify effects on consumer welfare, application of the evidence-based antitrust approach suggests that the anticompetitive models supporting more interventionist approaches to exclusive contracts should not form the basis of antitrust policy.
So what kind of antitrust analysis would this sort of empirical evidence suggest for antitrust rules using the error-cost approach? Ideal error cost informed standards would reflect our best estimates of the incidence of anticompetitive exclusive dealing, the relative social costs associated with Type I and Type II errors in this setting, and incorporate safe harbors for conduct that we know is unlikely to generate competitive harms. In this case, the evidence suggests that the probable incidence of anticompetitive exclusive dealing is low and the incidence of pro-competitive exclusive dealing (and in competitive markets) is high. We also know that payments for exclusivity at the retail level are likely to be passed on to consumers in competitive retail markets, implying that the social costs of false positives in this case must account for the loss of that welfare-increasing competition. And relying on the theoretical literature, that the only conditions under which exclusive dealing can possibly lead to anticompetitive outcomes are those where there is substantial foreclosure and rivals do not have access to compete for distribution and thus are at risk of being cut off from distribution in a way that deprives the opportunity to achieve minimum efficient scale.
In my view, the error cost approach leads to at least two clear policy recommendations, both of which I was on record supporting before the hearings in my article Antitrust Law and Competition for Distribution, defended during the hearings, and continue to support. The first is a safe harbor for foreclosure levels less than 40 percent. The second is a safe harbor for exclusive dealing contracts that are terminable in one year or less. Howard Marvel has covered the case of the short term safe harbor in his earlier post, expressing some disappointment in the fact that the DOJ (based on Dentsply) backed away from endorsing this safe harbor. I share Howard’s disappointment in this aspect of the exclusive dealing chapter of the Report. But to their credit, the DOJ does provide a foreclosure safe harbor — or at least indicates that exclusive dealing contracts that foreclose less than 30 percent of the market should not be illegal. This is consistent with the necessary (but not sufficient) conditions of the anticompetitive theories and a sensible start to an evidence-based approach. While I’ve argued for something closer to a 40 percent safe harbor based on my reading of the cases, the DOJ’s position is at the low end of the range in the cases and quite defensible. The important point is that such a safe harbor is based on sound economic theory and evidence. While it is true that we’d like to know more about precisely the level of foreclosure at which competitive harm becomes an issue (e.g. greater accuracy in understanding the how much distribution is required to achieve minimum efficient scale) on a case-by-case basis, and that foreclosure tests often beg the question of what “counts” as foreclosure, the DOJ rule is a good starting point and provides some much needed guidance for what types of distribution arrangements and programs will not be challenged by the agencies.
I’d like to conclude with a comment about single product loyalty discounts, which the Report addresses in Chapter 6. It should be noted that the anticompetitive theories of harm for loyalty discounts often are identical to or analytically approximate the claims in exclusive dealing cases. Specifically, the claim is that the distribution contracts result in a form of de facto exclusivity that will deprive the rival of opportunity to compete for distribution sufficient to achieve minimum efficient scale. The key points here from an evidence-based perpsective are both that we have little empirical evidence that loyalty discounts lead to anticompetitive outcomes, but we do know that the discounts are passed on to consumers and increase welfare. Like exclusive dealing, this state of knowledge ought to lead to a liability rule that places a strong burden on the plaintiff to demonstrate actual competitive harm, and safe harbors based on sound theory and evidence where they can be crafted reasonably. In this case, since the anticompetitive theories all require foreclosure of a significant share of distribution and substantial economies of scale, it is quite sensible in the case of loyalty discounts to allow defendant’s a safe harbor that would make per se legal loyalty discount programs that foreclose less than a pre-specified share of the retail/distribution market. I believe the right starting point for such a safe harbor comes from the cases, and could be set at 40 percent. But building on the DOJ’s analysis, the argument can and should be made that the exclusive dealing safe harbor logically can and should apply to loyalty discounts as well.
I’ll leave others to discuss cost-based approaches to loyalty discounts, such as those discussed by Thom and Dan in their earlier posts on bundled discounts, but note here that the foreclosure based safe harbor approach should be viewed as a complement and not a substitute to the other cost-based standards based on the same sort of nexus with existing economic theory and evidence. The important policy point is to economize on the existing evidence to design standards that minimize the sum of error and administrative costs.