Dan Crane and Thom (who has promised more remarks!) have now both posted their prepared remarks for the Section 2 hearings panel on bundled discounts. Both call for bright-line, administrable liability rules for all forms of unilateral exclusionary conduct, and have important things to say about designing antitrust rules for bundled discounts. Both are worth reading in their entirety. Administrable rules that sensibly balance Type I and II errors are certainly an indisputably admirable goal for antitrust analysis and bundled discounts have proven to be a particularly tricky form of conduct for Section 2 analysis. Despite all of the agreement around here between Thom, Dan and I on the design of antitrust rules in a world of costly Type I errors, I think I have found a topic upon which I can at least offer a mild dissent (or at least a different perspective) regarding the usefulness of the analogy of various anticompetitive theories of bundled discounting practices to exclusive dealing.
The overlap between exclusive dealing and bundled/ loyalty discounts is frequently addressed by commentators, and is a topic of newfound interest in what has become the quest for a “holy grail’, one size fits all standard for Section 2 analysis of exclusionary conduct. At times, I detect a tension between the analysis of bundled discounts and exclusive dealing contracts which both purport to exclude exclude by depriving rivals from the opportunity to compete for distribution sufficient to support minimum efficient scale. For example, I discuss what I perceive to be a tension in Professor Hovenkamp’s very sensible analysis of bundled discounts and exclusive dealing in this post:
Hovenkamp concludes that adminstrative costs justify a predatory pricing-type rule in the context of for bundled discounts where the anticompetitive mechanism is de facto â€œforeclosureâ€ or deprivation from distribution resources (i.e. shelf space) that would prevent rivals from achieving minimum efficient scale and extend the duration of monopoly by increasing barriers to entry. One would think that it would follow from Hovenkampâ€™s position that a predatory pricing-type rule would also be sensible for exclusive dealing and tying arrangements where the anticompetitive mechanism is the economic equivalent. To the contrary, Hovenkamp advocates rule of reason analysis (p. 201) for exclusive dealing and tying, noting that â€œforeclosure concerns can be assessed meaningfully only via the rule of reasonâ€ and that â€œthe antitrust law of exclusive dealing,â€ which generally requires proof of substantial foreclosure as a necessary condition of competitive harm, â€œseems to be on the right track.â€
The basic tension here is that the anticompetitive theories underlying both forms of conduct require foreclosure of a rival sufficient to deprive the opportunity to compete for minimum efficient scale. Of course, the pro-competitive side of the ledger differs. One might sensibly believe that the standard for the two forms of exclusion should be different because lower prices are inherently pro-competitive whereas exclusive dealing may not invoke the same immediate consumer benefits. This is certainly a sensible position. But it only suggests that the standard for bundled discounts ought to be more difficult to satisfy than the exclusive dealing standard given equal administrative costs and the same anticompetitive mechanism. This point is not sufficient to render the exclusive dealing analogy fruitless. I offer below some tentative thoughts on the usefulness of the exclusive dealing analogy to bundled discounts.
The basic point is this: we still know that substantial foreclosure is a necessary condition for competitive harm and it seems like in an area where we know so little about the economics of the conduct at issue, it would be wasteful not to design a standard that incorporates this necessary condition.
Both Dan and Thom are certainly aware of this point, and to their credit, address it explicitly on their testimony. Dan does so the most directly in arguing that the analogy to exclusive dealing is insufficient:
[E]xclusive dealing analysis focuses on whether the defendantâ€™s contractual practices â€œforecloseâ€ a substantial share of the relevant market to rivals. â€œForeclosureâ€ is an empty concept unless it means that rivals cannot compete for the business. To put it another way, if a rival would be able profitably to match the defendantâ€™s bundled discount, what we have is ordinary price competition and not foreclosure. The problem with using exclusive dealing analysis to assess bundled discounts is that exclusive dealing analysis begins with the assumption that whatever contracts are covered by the exclusive dealing contracts are â€œforeclosedâ€ to rivals, a fact not at all in evidence in bundled discount cases.
This is a minor portion of Crane’s testimony, but I am not necessarily convinced by this argument that “foreclosure” concepts are irrelevant to cases involving allegations of exclusionary bundled discounts. A few analytical points:
1. Foreclosure is not an empty concept. The notion of foreclosing a rival from access to distribution sufficient to achieve minimum efficient scale is not meaningless. In fact, it is a necessary condition of most models of anticompetitive exclusionary conduct. It may well have been deprived of meaning by court opinions in recent years — and I think that is the point Dan is appropriately making here. That is fair enough. But if the very fact that an exclusive contract “excludes” some rival from access to some thing is sufficient to “foreclose” for Section 2 purposes, then the standard is truly vacuous.
To be sure, the foreclosure inquiry is full of pitfalls and challenges. How do we define the market for distribution? How much foreclosure is enough to shift the burden? What about high foreclosure and short term contracts? But that doesn’t mean the concept is empty. It just means the analysis is difficult. Perhaps it also means, given the obvious procompetitive benefit of lower prices, that we ought to have different standards for bundled discounts and exclusive dealing. This is a point I want to make clear: that the analogy bears fruit does not mean we should have identical standards from an error-cost perspective. However, substantial foreclosure is still a necessary condition for anticompetitive effect caused by exclusion of rivals. We invite trouble if we start devising antitrust standards that ignore the little that we do know about the necessary conditions for anticompetitive effect.
2. What does this mean for exclusive dealing and bundled discount liability rules? Again, this does not mean that I disagree with Professor Crane’s proposed liability rule for bundled discounts — though I would see no problem with keeping the foreclosure requirement along with it. If we are willing to say that foreclosure is a necessary condition for competitive harm caused by depriving rivals from access to minimum efficient scale, then we ought to be able to say that foreclosure should be a part of the liability rule for these types of claims. There are, of course, bundled discount claims that do not alleged this type of exclusion and resemble predatory pricing claims. I leave open the question of whether these claims should have a different standard.
Take, for example, Thom’s proposed standard for bundled discounts which I shall refer to as the “Lambert rule.” The Lambert rule would require plaintiffs to demonstrate that they have made a “good faith” offer to supply the discounter and that the plaintiff cannot “practicably” coordinate with other producers to create a competitive bundle.
But I wonder about two issues with respect to the Lambert rule. First, the case for the Lambert rule over the ultra-administrable Brooke Group standard hinges on the grounds that the latter may generate more Type II errors and the two are at least equally administrable. But assuming the Lambert rule succeeds in reducing Type II error, I’m not convinced that shifting to a “good faith” offer and “practicable” bundle standard represents a move towards easier administration burdens on courts.
More to the theme of this post, the very reason that a above cost per se legality rule might leave room for Type II error (which we readily accept in the single price predatory pricing context as Thom points out in his remarks) is that some above cost discounts might exclude rivals by depriving them of the opportunity to achieve minimum efficient scale. Fair enough. But if such exclusion is the problem, we also know that the anticompetitive theories of exclusionary bundling with above cost discounts and, for that matter exclusive dealing, also require substantial foreclosure of distribution sufficient to deprive the rival of MES for some significant period of time.
So here is my question: why don’t we always require substantial foreclosure to be demonstrated in cases alleging exclusionary conduct which deprives the rival of scale? I’m (again) not suggesting equivalent standards. The foreclosure requirement in the bundled discount context may be in addition to some test which maps other necessary conditions into an administrable legal standard. But why not include it as part of a workable standard? Assessing foreclosure can be elusive, to be sure, but courts have had experience assessing for over century in the exclusive dealing context. Perhaps the marginal increase in admin costs is undesirable, but maybe not. Also, it may further reduce the possibility of Type I error for cases that satisfy some multi-product standard but do not have the capability of harming competition.
3. Bundled discounts, exclusive dealing, and competition for contract. The relationship between exclusive contracts and discount contracts is a natural one. Firms compete for distribution by offering retailers (buyers) a number of different contractual forms. For example, Coke may purchase premium shelf space to generate a promotional increase of its sales by a certain percentage. Coke can compensate the retailer with some sort of de facto RPM scheme which increases the retailer’s margin, a discount, a bundled discount, or a per unit time payment like a slotting allowance. Various forms might be optimal in different settings. I discuss this point at length in my JLE article with Ben Klein with respect to per unit time versus discount contracts for shelf space. The key point, however, is that these contracts are typically an efficient element of the competitive process for distribution.
Highly variable liability rules that are sensitive to the form of contract without significant deference to the principle that antitrust standards should focus on protecting the competitive process rather than the particular form of contract at issue. Again, it does not follow that a single uniform standard for all unilateral conduct is appropriate. It isn’t. The conduct falling under the umbrella of Section 2 involves a diverse set of practices with different pro-competitive explanations, and different necessary conditions for generating consumer welfare harm.
I also admit that I don’t get what exactly what Crane means when he writes that:
“exclusive dealing analysis begins with the assumption that whatever contracts are covered by the exclusive dealing contracts are â€œforeclosedâ€ to rivals, a fact not at all in evidence in bundled discount cases.”
Well, it might not be in evidence because nobody is asking for the evidence. It might also not be in evidence because bundled discount contracts are highly unlikely to satisfy the necessary condition for competitive harm. I suspect the latter — and I think so does Dan. In fact, I take Dan to be making the sensible point that competition for bundled discounts means that rivals can compete for distribution by offering superior discounts at any time, and thus foreclosure in the exclusive dealing sense never occurs in bundled discount cases. But exclusive dealing case are also characterized by vigorous competition for the contract —- it is just that the contract includes an explicit or de facto exclusivity term. Further, the rule of reason analysis applied to exclusive dealing generally acknowledges that short-term contracts make it more likely that competition for exclusives will not foreclose rivals in a manner sufficient to create antitrust injury. In other words, courts are increasingly willing to find that even a high percentage of rival outlets foreclosed through exclusives is not sufficient to find a Section 2 violation when the contracts are short, or there is some other reason to believe that “excluded” suppliers can realign supply contracts, i.e. no barriers to entry.
All of this essentially boils down to one simple premise: allegations of exclusionary distribution contracts require foreclosure of distribution sufficient to deprive rivals the opportunity to compete for minimume efficient scale. If we are ready to admit that this is something we know about the anticompetitive theories, as Hovenkamp, Thom and Crane seem ready to do, shouldn’t we include this necessary condition as part of our liability rule. This can be done by adding it to one of the competing proposals advocated by Thom, Dan, or others based upon an error-cost analytical framework. Both Thom and Dan have certainly moved the ball forward in this gray area between discounting and exclusivity. But it seems to me that in an area where we know so little about what is probable and quite a bit about what is possible, shouldn’t we be designing liability rules in a way that gets the most mileage out of what we do know? I don’t know what the right answer is, and I leave Thom and Dan and commenters to share their thoughts and reactions. But I stand ready to defend the proposition that exclusive dealing analysis and concepts are useful guides to analyzing exclusionary distribution claims of all stripes, including bundled discounts.