Dennis Carlton and Michael Waldman have posted an insightfulÂ DOJ working paper on antitrust safe harbors for unilateral conduct involving quantity discounts and bundling. The discussion is very timely in light of the Microsoft CFI decision,Â AMC Report, Section 2 Hearings,Â and various monopolization cases in the United States, EU, and other antitrust jurisdictions. The Carlton & Waldman paper is short, very accessible, and makes several very important points about the benefits of safe harbors to guide antitrust policy in this area generally and some weaknesses in the proposed AMC approach to bundling. Anybody interested in single firm conduct issues in antitrust should read this paper.
The issue they raise — safe harbors for single firm conduct — is one I’ve written about quite a bit. And I want to test out some thoughts on it here that I’ve sketched out partially in some academic writing and blog posts with respect to safe harbors for quantity discounts, loyalty rebates, exclusive dealing and competition for distribution more generally. I am on record defending two very specific safe harbors (one for short-term contracts and another for contracts that foreclose < 40% of the distribution market). I’ll return to the issue of a foreclosure safe harbor in a moment, and that will be the focus of the post, but for now, let me start with Carlton & Waldman’s framing of the antitrust problem of exclusion:
An antitrust claim involving exclusion requires that there be harm to a rival, harm to consumers and a linkage between the harm to the rival and the harm to consumers … This reasoning suggests that all mechanisms of exclusionary pricing conduct that do not alter a rivalâ€™s costs of operating or impair his ability to exist should not trigger an antitrust violation. In particular, this means that if there are no such effects, as for example occurs when the production technology is constant returns to scale, then there can be no anticompetitive harm. This does not mean that the rivalâ€™s business is unaffected nor that consumers are unaffected by a new pricing policy, but simply that the mechanism of harm, if there is one, has nothing to do with excluding a rival.
Carlton & Waldman focus in on the key issue for antitrust policy related to competition for distribution in the form of discounting conduct: the question of whether the defendant’s conduct has deprived a rival of scale to a degree that it is foreclosed from profitable access to the market altogether, or to a sufficient degree that its competitive constraint on the exercise of the defendant’s monopoly power is reduced, and competition is harmed.
So far so good. This economic insight is at the heart of the “foreclosure” requirement that appears in exclusive dealing cases that involve analytically identical claims concerning exclusion. Carlton & Waldman address claims of exclusion involving single product pricing in a predatory pricing framework and make the following statement about the “recoupment” requirement of the standard two pronged Brooke Group analysis:
[The recoupment prong] is a reflection of the principle that with constant returns to scale rivals will always constrain price and there can be no recoupement. The reason is that with no fixed costs, entry is always possible and guarantees that there is a competitive constraint on price. [The recoupment requirement] is phrased more practically to cover deviations from constant returns to scale that are not so large as to allow recoupment. With no possibility of recoupment, there is no reason to incur the initial loses associated with pricing below cost.
This is very interesting, and perhaps optimistic, understanding of courts are doing when the apply the recoupment requirement. My preliminary reaction is that most single product predatory pricing cases involve an analysis of barriers to entry at the recoupment stage as if the court was answering the question: “can the monopolist increase prices for a sustained period of time without attracting entry and therefore, recoup the losses associated with its period 1 prices?” I don’t think the courts explicitly conceptualize the recoupment prong in the way Carlton & Waldman describe here as it relates to scale. Rather, my tentative view is that analysis concerning the potential to deprive rivals of scale, the presence of substantial economies of scale, and even foreclosure are generally missing from these single product predatory pricing cases.
To be clear, thats not to say that courts are not analyzing in the recoupment prong the issue of whether the pricing scheme is likely to exclude rivals in some sense. But I think this sort of scale and foreclosure analysis that is typically present in exclusive dealing cases is generally absent in single product pricing cases. Now, I do believe that the recoupment requirement in these cases should be applied in the manner Carlton & Waldman suggest it already is. In fact, that is basically where I am going with this post. Keep reading and I’ll explain why I think this would be a good idea.
It strikes me as a reasonable safe harbor in all monopolization claims involving conduct alleged to exclude a rival by depriving it from scale require the showing of “substantial foreclosure” that appears in the exclusive dealing case law. I’ve written about this before in response to Thom and Dan Crane’s comments at the AMC on bundled discounts (see also the thoughtful responses from Thom and Dan in the comments to that post), when I raised the argument that bundled discounts and exclusive dealing should both have a safe harbor for practices that do not generate substantial foreclosure sufficient to deprive a rival of scale:
The basic tension here is that the anticompetitive theories underlying both forms of conduct (ed: loyalty rebate and bundled discounts) require foreclosure of a rival sufficient to deprive the opportunity to compete for minimum efficient scale … 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.
I still think this is right. And while Thom, Dan and I come out slightly differently on how we might apply these safe harbors to loyalty discounts and bundled discounts, we all seem to agree (see the comments to the last post) that a plaintiff’s burden ought to include this showing. This makes sense: If economic consensus is that allegations of exclusion require foreclosure sufficient to deprive rivals the opportunity to compete for minimume efficient scale, and we are ready to accept that this is the state of economic consensus, then we ought to explicitly include this showing in the part of the plaintiff’s burden.
Another reason this screen makes sense in not only in the exclusive dealing context where we normally see it, but also in the bundled discount or loyalty rebate context, is that it is a screen that courts can actually apply. They can and do in the exclusive dealing context all the time and have for over a century. There are other very useful screens that one might apply in the bundled discount context. But there might be easy cases to identify with the foreclosure safe harbor that courts might be able to identify at the summary judgment stage or earlier in the post-Twombly world. For example, Carlton & Waldman’s example of constant returns to scale technology or deviations from constant returns that are small enough that profitable exclusion is unlikely. By the way, I read the logic of Carlton & Waldman as supporting a foreclosure screen in the sense that they seem to believe that this is how courts are applying predatory pricing analysis already.
The case of small deviations from constant returns to scale is very interesting. In the exclusive dealing context, these cases are often dismissed at the summary judgment stage on the grounds that the contracts at issue involve only de minimis foreclosure, or foreclosure less than the 40% or so below which liability is rarely found. This is precisely the safe harbor I argued in favor of in Antitrust Law and Competition for Distribution.
Given that our empirical knowledge about the competitive effects of various single firm conduct is not great, but that we have a consensus that any potential exclusion must involve foreclosure, I think it makes a lot of sense to draw some bright lines that allow courts to filter out cases where anticompetitive exclusion is unlikely. Safe harbors should be based on our theoretical and empirical economic knowledge, and they should be administrable by real judges in real cases. Foreclosure analysis can be complicated, and isn’t always easy. There is no magic number above which foreclosure will always or below which it can never create exclusion. The relationship between foreclosure and anticompetitive exclusion depends on product technology and minimum efficient scale. The percentage of foreclosure can also be manipulated by focusing on unduly narrow channels of distribution or overstated when contracts are short term (I’ve also argued for a one year contract safe harbor). On top of that, foreclosure is a necessary but not sufficient condition. No test is perfect. But the foreclosure safe harbor does seem like it satisfies both of these requirements for a good screen.
Speaking of errors in foreclosure analysis, Carlton & Waldman also make an excellent point here about flawed economic arguments that are often raised about the likelihood of exclusion when sunk costs are involved:
We have seen the following incorrect argument: â€œthere is a sunk cost to enter, hence there are scale economies. Depriving a rival of scale will therefore raise its costs.â€ This argument confuses average with marginal costs. As long as the rival is not driven out, the rivalâ€™s marginal cost determines its competitive effect. Once incurred, a sunk cost does not affect the rivalâ€™s behavior. Depriving an existing rival of scale will not necessarily reduce the rivalâ€™s competitive significance as long as marginal costs are non decreasing and the rival remains in the industry.
This is a very important and insightful point that is often ignored in discussions of the economics of exclusion. I’ve discussed the Carlton & Waldman piece without really getting to its main point: the AMC test for bundling is flawed because its first prong will result in a high number of false positives. Their position is that single product predatory pricing case law is sufficient to be extended to quantity discounts and loyalty rebates (and I take it, loyalty rebates that amount to exclusive dealing). I agree with the authors with the caveat that courts should make explicit the focus on the potential to excluse a rival by depriving it of scale in a manner that harms the rival and the competitive process. The best way to make this focus explicit it to borrow from the exclusive dealing case law and import the “foreclosure” requirement. After all, exclusive dealing jurisprudence is designed precisely to answer address an analytically identical claim of exclusion.
So here’s what I propose. All monopolization claims involving claims of exclusion involving pricing conduct or exclusive dealing require a showing of substantial foreclosure. I think 40% sounds like a reasonable safe harbor based on the case law (I make a full argument for this in the Competition for Distribution paper linked above). This can be in addition to other tests, like the AMC proposed approach or incorporated into the recoupment analysis in a single product case. No adjustments are necessary in exclusive dealing or partial exclusive cases. Any objections?
Thank you for your quick response, Josh. Now I see your point and agree on your position.
Just a minor comment: Carlton &Waldman’s safe harbor for bundling seems not that safe: in Greenlee, Reitman and Sibley (GRS)’s paper, they claimed that the single-product firm will price more aggressively under bundling and thus if we apply Carlton &Waldman’s test, B-only consumers will be better off then, and so the safe harbor test will be passed for sure. However, GRS’s numerical examples show that the effects on consumer surplus and total surplus are still ambiguous.
Yong: Great comment, thanks. I should clarify a bit. I agree with you completely that there must be a reason for the bundled loyalty discount rather than exclusive dealing contract in terms of the pro-competitive rationale? E.g. the economic question is why the contract takes this form? What does it do that an exclusive doesn’t in terms of efficiencies? However, the *anticompetitive* theories are largely equivalent (the theory is usually that the bundled discount results similar anticompetitive effects to exclusive dealing in terms of foreclosure, deprivation of access to economies of scale, etc.).
I think because of the pervasive nature of price discrimination in the economy, which is generally pro-competitive and a healthy feature of the dynamic competitive process from a consumer welfare perspective, it would be prudent to force plaintiffs to show both that the contracts achieve substantial foreclosure AND that they result in anticompetitive effects. In other words, we have to pick a test that does not invite Type I error with respect to pro-competitive price discrimination such as metering. This is basically Carlton’s concern in the AMC Report in his Separate Statement, e.g. that common pricing mechanisms that do not raise competitive concerns (like bundling razors and razor blades) would pass the first prong of the test but obviously should not be an antitrust problem. I agree with this. My general position is that antitrust regulation of metered pricing and other forms of competitive price discrimination amount to micro-managing pricing decisions with out much upside in terms of consumer welfare gains and so don’t make much sense.
The question of what role price discrimination should play in antitrust analysis in these cases is an interesting one. Klein & Wiley argue that PD should be a defense. Carlton doesn’t go that far in his separate statement but seems to agree at least in principle. I’ve argued in my Cato paper that price discrimination should just be ignored (e.g. not used to imply antitrust relevant market power or anticompetitive effects).
Thanks for pointing to an interesting article, Josh!
I agree that one thing related to bundled loyalty discount is exclusive dealing, but that’s incomplete. If pure exclusive dealing can do the job, why bundle? There must be some other effects besides exclusive dealing. As Carlton &Waldman and you pointed out, bundling can be a price discrimination tool as shown in the literature and thus it could result in higher consumer surplus.
Generally, bundling would have both price discrimination effect and exclusionary effect. Only looking at either side is incomplete. So I think a good empirical test do need our further analysis from the perspective of welfare analysis. The safe harbor for bundling proposed by Carlton &Waldman based on exclusionary aspect is really insightful. However, I do have a concern that: what if price discrimination effect dominates the exclusionary effect of bundling? Or when those B-only customer’s portion is very small, should we focus only on them for the safe harbor test?