A few quick reactions to the repudiation of the Section 2 Report, and more importantly, what it means for the future of monopolization enforcement:
First, the most disappointing thing about the withdraw of the Report and this announcement is that it is incredibly dismissive about the long hours of work put into this project by both DOJ and FTC staff, the academic community, as well as business representatives. The idea that identifying instances in which monopolization enforcement can generate a positive rate of return for consumers when one accounts for the cost of errors is an easy challenge that can be solved by resolving on “tried and true” cases is a fallacy. The reason that we had Section 2 Hearings is precisely because even if we had acceptable monopolization standards in the cases, we don’t know whether these tests are consistent with an approach that enhances consumer welfare.
Second, declaring that we know anticompetitive conduct when we see it threatens to continue the trend to write serious economic analysis out of antitrust analysis. It is only a slight exaggeration to say that the withdraw of the Report and announcement threaten to bring antitrust back to the 1960s.
Third, to the extent that the new monopolization approach does not take us back to the 1960s, it will be because the approach envisioned will run squarely into Section 2 case law that embraces the very error cost approach that the AAG rejects — leading to quite an odd tension at a law enforcement agency.
Fourth, I’m not sure what the procedural requirements to do this would be at the DOJ, but the Section 2 Report should be published and contrary to what Varney declared (“the Report and its conclusions should not be used as guidance by courts, antitrust practitioners, and the business community”), the Report should be used for eaxctly those purposes. One of the things that the Report does is summarize existing Section 2 law. It combines the insight of 2 years of testimony and thousands of hours from experts from these communities and, to my mind (and no matter whether I agree with all of the conclusions or not — I don’t), is the single most important document on monopolization law as it exists. The FTC dissenting Commissioners nor Varney have offered a competing statement of Section 2 law. While it may no longer be a useful document to predict what the DOJ will do in a given case, it may well be an excellent document for predicting the outcomes of those cases in federal court.
Fifth, as many of our Section 2 Symposium particpants indicated, the rejection of the error-cost framework is a critical step backwards for antitrust enforcement. One can rationally disagree about the relevant empirical estimates of incidence of anticompetitive conduct and magnitude of social costs of errors, but one cannot rationally and simultaneously commit to rejection of the error cost framework (or the non-existence of false positives) and a consumer welfare approach to antitrust.
Sixth, Varney’s statement points Conwood of all cases as one of its handful of “strong examples of successful challenges to exclusionary conduct and the Department will look to them in establishing its Section 2 enforcement priorities.” Yikes. As I wrote just the other day in the symposium with respect to Conwood‘s bona fides as a paradigmatic Section 2 case, the notion that Conwood represents best practices in Section 2 enforcement is a scary thought. To the contrary, I’ve written elsewhere that Conwood is a fairly good example of a false positive (forthcoming in SCER, but link available here). Here’s what I said in the comments the other day:
It is true most commentators focus on the product destruction and allegations of misleading retailers and discuss Conwood as a classic example of a “cheap exclusion case,” much like the textbook example of blowing up the rival’s factory. But there are some problems with this characterization, not the least of which is that Section 2 still requires that plaintiffs demonstrate actual competitive harm and engage in some analysis on that issue. A close look at the evidence in Conwood suggests that the evidence is woefully insufficient with respect to competitive harm.
Also, commentators frequently (following the Sixth Circuit’s example) ignore the fact that Conwood prevailed under a Section 2 theory that included not just the tortious conduct, but also presumptively pro-competitive conduct such as offering loyalty programs and category management services to retailers. The Sixth Circuit never disaggregated lawful from unlawful conduct for the purposes of liability or damages analysis. But the more important point for our purposes is that a case involving some allegations of indisputably “bad” conduct (product destruction), but little evidence of consumer harm, resulted in a Section 2 judgment and expensive settlements that swept in pro-competitive conduct like exclusive dealing and category management contracts.
Regarding the characterization of the evidence in that case as “quite strong,” I’m not alone in my belief that it is incorrect. Consider that the damages calculations at trial have been heavily criticized by many antitrust and evidence scholars. See, e.g. D.H. Kaye’s analyses in the Virgina Law Review and Jurimetrics (2003), a scathing amicus brief in support of a writ for certiorari from several leading economists attacking the damages estimates, and a lengthy critical discussion in Herbert Hovenkamp’s Antitrust Enterprise which makes Conwood the poster child of sorts of the case against private litigation.
All in all, these predictions are looking better by the day.
Exactly what I would say, Josh. Yes, there is a common set of problems, but that doesn’t mean that there is a common solution. The whole point of an error cost framework is to take account of of both the likelihood and the consequences of erroneous enforcement. I think at least the likelihood (and perhaps the consequences) are higher with exclusion than collusion. Also, when DG Comp flat out invites companies to report on their exclusionary rivals or suppliers, and when antitrust has been used so successfully as a weapon in the past, I find it hard to be so sanguine about the potential problems of overly enthusiastic rivals or customers. (BTW: In my experience this is a Section 7 problem, as well).
One point to interject here is that it is perfectly possible to both (1) understand that exclusion can lead to the same effects as collusion under some conditions, (2) think such a result would be a bad thing for consumers and the proper subject of antitrust scrutiny, (3) believe that as a practical matter, the error cost differences between the two should lead to very different liability rules and enforcement strategies. Thurman’s comment avoids the error cost point by invoking a similarity of possible effects. But the difference is that we know MUCH more about the effects of horizontal agreements between competitors on price (though we certainly don’t know everything) than we do about single firm conduct. In the latter case, enforcement runs a higher risk of false positives. We have much better technology for understanding the link between horizontal restraints and consumer welfare losses than we do in the monopolization area — hence the need for the Section 2 hearings and all the debate appropriate the appropriate scope.
As for excluded rivals being shy about running to regulators, that does not seem to to be the case in Europe in recent years and I’m also skeptical about the claim in the United States. Nonetheless, perhaps the hesitation on the part of excluded rivals is a function of low levels of monopolization enforcement, a trend not likely to continue in the next few years. So, while theoretical concerns about retaliation on excluded rivals are interesting and plausible, there is plenty of reason to believe this is now or will be a bigger problem in Section 2 than under Section 1.
It is interesting that all but one of the stylized facts Geoff claims (risk aversion, imperfectly specified rules, errors by enforcers and courts, business uncertainty about legal rules, private enforcement and costs of litigation and penalties) — that is, all but “competitor access to enforcers’ ears” — apply on their face to collusive conduct (e.g. horizontal agreements that facilitate coordination) just as they apply to exclusionary conduct like monopolization. And the role of competitors in monopolization enforcement is not as significant as Geoff assumes. Excluded rivals are often reluctant to complain to enforcers on the record and to testify in court against monopolists, particularly when (as is common) they are also customers of the dominant firm or suppliers to it, and must deal with the dominant firm in the future regardless of how the agency investigation comes out. In consequence, the “competitor access” factor doesn’t strongly distinguish collusion from exclusion either, and Geoff’s argument against section 2 enforcement turns out to be equally an argument against section 1 enforcement. So, Geoff, are you questioning the antitrust prohibition against monopolization in particular, and distinguishing that from the rest of antitrust enforcement, or do you see all of antitrust enforcement (including enforcement against cartels and mergers to monopoly) as an institution we are better off without? It seems to me that anyone who thinks cartels are a problem should also be concerned with exclusion — after all, a firm or group of firms could achieve the same anticompetitive end of reducing industry output by colluding with its rivals or by excluding those rivals from inputs or the market.
Right–my reference to Salinger was not to the one example he cites, but to his explanation of why thre might not have been a lot of examples cited at the hearings, even though the problem may be widespread.
I really don’t know why, however, this is so contentious. I am with Josh about hunches, but I must say that it takes very little guesswork to believe that this is a problem. Given risk aversion; given that the law does not perfectly prohibit only inefficent behavior; given that enforcers and courts are not perfect; given that businesses do not know exactly what is efficient; given private enforcement and competitor access to enforcers’ ears; and given the costs of litigation and penalties (including treble damages and injunctions)–how could it be otherwise? Without clear evidence one way or the other to settle the question, how do you arrive at the conclusion that false positives aren’t a problem? What assumptions do you make to reach that conclusion?
I appreciate the clarification Thurman re: misrepresentation.
But when you say “let’s turn to the merits,” however, can’t we do better than what Christine Varney or Geoff or you or I have as hunches about the prevalence of false positives properly defined? There is a body of evidence out there that tells us that some practices are prevalent in competitive markets, the competitive consequences of some of the practices at issue, and the theoretical conditions under which these practices might lead to consumer welfare losses. Seems like that literature is a much better place to focus than hunches — even mine.
And by the way, the inference you’d like to draw about Varney’s observation about refusing to engage in pro-competitive conduct because of fear of antitrust liability just doesn’t line up with her statement about refusal to engage in *legal conduct.* That’s just a different point altogether.
And as Geoff suggests (and I have earlier), I take the mistaken definition of a false positive as (together with the statements above) as pretty good evidence that the new DOJ is not working within the error cost framework. One must stretch far, and in my view too far to be plausible, to turn the above statements into something that fits the error cost framework.
But the more general point is that all of this talk about hunches about false positives aren’t really turning to the merits. The point of the Section 2 Report was to turn to the merits and collect evidence on both the incidence and magnitude of false positives and negatives in order to inform sensible Section 2 liability rules, burdens, safe harbors, and enforcement decisions. The Salinger quote also provides an explanation for why we might expect very little of this sort of testimony to be forthcoming at public hearings.
The Salinger quote says that evidence of false positives in the hearings was “rare” and it gives only one example, which is apparently decades old. These observations are consistent with AAG Varney’s view that she hasn’t seen any chilling effect of the bar on monopolization on the conduct of dominant firms during the past decade.
First, you asked about Intel/Google, et al., and perhaps there is no evidence that they were restrained. (See my selection bias point). The same is clearly not true for other firms. I was in private practice for much less time than Varney, and I saw it happen–I saw the issue discussed, practices changed, etc. I find her claim quite extraordinary, actually. Also, Michael Salinger addressed exactly this issue in our sympoisum here. Here’s what he said:
Geoff asks “how the hell does Josh (or anyone else not on the inside) know what risks Intel and Google didn’t take that they might have and maybe should have?” AAG Varney’s observation that she hasn’t seen false positives was based on her experience counseling in the industry — a background that essentially put her on the inside. If dominant firms (or the 30% to 50% firms Geoff is worried about) have been chilled from engaging in efficient conduct during the past decade by the threat of monopolization suits, contrary to what the AAG has found in her experience, where are the insiders making that case?
That’s not a fair question, and it’s not a random sample. The wildly successful firms are precisely the ones that didn’t shy away from efficient behavior because it might cause trouble. The right question is how many firms with 30-50% market shares refrained from behavior that might have given them 70% market shares because 50% was enough to get them in Section 2 land? Moreover, how the hell does Josh (or anyone else not on the inside) know what risks Intel and Google didn’t take that they might have and maybe should have? Finally, Josh’s point about Varney not even getting the question right (the problem isn’t “legal” behavior that’s deterred–it’s efficient behavior that’s deterred) is hugely important. An enforcer who either doesn’t know that there is a difference between what is legal and what is efficient and/or who implicitly (or explicitly) claims perfection in enforcement is probably not really getting the error cost framework to begin with.
As Josh says, I think the AAG has a different view about the likelihood of false positives than Josh does, while Josh thinks she rejects the error cost framework. I can see why Josh thinks that — I didn’t mean to use the word “misrepresents” to suggest otherwise and in retrospect I should have used a different word (“misinterprets” perhaps). So let’s turn to the merits. What efficient, pro-competitive conduct has, let us say, Google, Cisco or Intel refrained from undertaking during the past decade for fear of antitrust enforcement (assuming for purpose of argument that these are dominant firms in high-tech markets of the sort that AAG Varney would have been familiar with in her private practice)?
You have a hypothesis about Varney’s position that she embraces the error cost approach but assigns possibilities of false positives based on her experience. I have a competing hypothesis — that she rejects it. Its tough to know what to make of the statement (I’ll quote in entirety here so readers can see what we’re talking about), but there is other evidence in support of my view.
Let’s start w. her statement:
“My view and, you stole my thunder, I was prepared to say there is no such thing as a false positive, you know, let’s get real. I have counseled numerous incumbents who are dominant as well as numerous new entrants. I can tell you, at least in my own experience, there is not a dominant incumbent who hasn’t done something that is lawful because they were afraid that it might be reviewed by the DOJ or a state attorney general or an FTC. I just don’t see it. Ten years back in the private sector I have never once seen it, so I think that this ruse of, you know, we have to be restrained in our enforcement because false positives will chill innovation, take an economic toll on society and overall result in negative economic consequence, slowing output, increasing cost, I just think is false. I think the more people in the bars start rejecting this idea of false positives the better off we’re going to be.”
First, Varney states that “there is no such thing as a false positive.” There are ways to state this in a way that is consistent with the existence of false positives but assigning low non-zero probabilities. She didn’t do so.
Second, contrary to your description of her comments, notice that Varney says about her counseling experience that she hasn’t seem a dominant firm that “hasn’t done something that is lawful because they were afraid that it might be reviewed by the DOJ or a state attorney general or an FTC. I just don’t see it.” But that’s not the question is it, Thurman? Isn’t the error cost question whether a firm has refrained from doing something efficient or pro-competitive because of fear of antitrust liability? She doesn’t speak to that. But that is the question. You seem to agree because that is the position you assign to her in noting that she embraces the error cost framework (in your second sentence). But that is not what she says. And the whole reason that we had the Section 2 Report to start with is because of the vexing problem under Section 2 of distinguishing pro-competitive from anti-competitive single firm conduct.
Third, do you believe that the statement about relying on Conwood as an example of what we should be doing under Section 2 is consistent with the error cost approach? I don’t. As quoted above, Varney’s statement points Conwood of all cases as one of its handful of “strong examples of successful challenges to exclusionary conduct and the Department will look to them in establishing its Section 2 enforcement priorities.” Yikes. There are plenty of folks out there (not just me) who think that Conwood is a false positive. If Conwood is an application of the error-cost approach as Varney sees it, then perhaps the term does not mean the same thing to both of us.
Fourth, the error cost necessarily includes an empirical component, i.e. we must rely on the existing empirical evidence to inform our priors. Perhaps Varney’s priors on the false positive rate are that it is zero or that the incidence of anticompetitive single firm conduct is particularly high with respect to vertical restraints, exclusive dealing, vertical mergers, etc. But that view is inconsistent with the evidence. An enforcement strategy that does not update one’s priors based on the evidence and instead resorts to the “tried and true” case law that got us into the mess we are in under Section 2 that led us to the hearings in the first instance is not an error-cost approach.
To be absolutely clear, I”m happy to leave open the possibility that Varney embraces the error-cost approach. I’d be quite pleased actually. But my view of the evidence is thus far not supported by that position. I understand that you most likely have a different view of that evidence. But to describe my claim as “misrepresenting” her position is much too strong an objection based on an objective look at the existing body of evidence.
The claim that the Assistant Attorney General rejects the error costs framework misrepresents her position. A fair reading of her comment (e.g. quoted by Geoff Manne in his May 11 post) is that she was simply saying that in her experience doing antitrust counseling over the past decade, she hasn’t seen any chilling effect of the antitrust laws on the willingness of dominant firms to engage in efficient, pro-competitive conduct. Given the Supreme Court’s skepticism about monopolization enforcement (e.g. the dicta in Trinko) and the attitude of the Justice Department during the previous administration (e.g. the Section 2 report), her observation is not surprising. The enforcement approach she inherited may well result in many false negatives with few false positives — in which case a sensible application of the error cost framework would suggest devoting more DOJ resources to monopolization enforcement, as she seems to be signaling.