Speaking of error cost analysis, this paper from a trio of lawyers in the General Counsel’s Policy Studies’ group at the FTC has a section entitled “Error Costs: The False Positive/ Negative Debate.” A frustration for me in discussing the error cost issue with respect to antitrust policy is that many people do not seem to understand that it is error costs and not just errors that we are concerned with. A common refrain is: show me a false positive monopolization case! We bring so few, we don’t have to worry about false positives anymore. QED. Of course that is wrong. The social costs of false positives are not about the single business practice that is condemned by an antitrust judgment. The real costs are the chilling of pro-competitive behavior by other firms in response to the expectation of the same type of judgment against them. You cannot just count cases. Those aren’t where the costs are! Not to mention much of the expectation of liability from pro-competitive behavior is to do with the threat of settlements. Oh, and you want a false positive? Here’s one.
Much of the confusion surrounding this basic point of the error-cost approach to antitrust rules can be read in the hearings on the Section 2 Report with proponents of more interventionist antitrust policy constantly invoking the mantra that we just don’t see that many false positives in the cases as evidence in favor of the lack of error costs. Some go so far as to argue that the social costs of a false negative in the context of monopolization is likely to outweigh the costs of a false positive. While monopolization can have the same economic impact as cartel behavior, of course, economic theory tells us that there are some offsetting forces to correct market failure in the case of false negatives but none for false positives. This was one of Easterbrook’s key points in the Limits of Antitrust. The other key point, which is not as well appreciated, is that errors are more likely when we do not have a good basis for identifying anticompetitive conduct. This is nowhere more true than in the case of monopolization. In this era of no monopolization enforcement, there ought to be bundles of empirical examples abound documenting exclusionary distribution contracts with convincing statistical evidence. The literature isn’t there. So I’m not sure exactly how one would identify the false negative if they saw it. On the other hand, the bulk of the literature on vertical restraints and single firm conduct suggests that the conduct is pro-consumer most of the time. Another point in favor of fearing false positives instead of negatives.
Anyway, back to the paper. Its generally very good in framing the debate and pointing out what Section hearing panelists had to say on the issue. It doesn’t quite, at least to my tastes, separate out the issue of errors versus error costs nor the theoretical underpinnings of the error-cost approach sufficiently. Still, its good and better than most on this issue. And they do cite Easterbrook’s argument with respect to higher social costs for false positives relative to negatives. However, it wraps up the discussion with a little bit of a tone of “some say false positives, some say false negatives, lets ignore both of them because there is no evidence.” For example, on this issue it includes the following paragraphs:
This debate reflects both the potential promise of decision theory as an analytical framework and its current limits as a calibrated tool. While decision theory provides “a way to organize our thinking about legal standards,” the lack of reliable data limits its ability to identify optimal rules.186 As one panelist observed, “[F]alse positives [and] false negatives” should be considered “on the basis of empirical data and not on theoretical assumptions.”187 Yet the hearings suggested no basis for reliably quantifying the likelihood and magnitude of false positives and negatives under potential liability rules.
When evidence is limited, decision theory primarily provides general directions and broad insights, leading courts and enforcers to identify circumstances in which concerns regarding either false positives or false negatives are likely to be especially significant, and where greater tolerance or heightened vigilance may be appropriate.188 The Supreme Court’s application of decision theory in antitrust cases has reflected these limitations, identifying two areas—predatory pricing and predatory buying—in which concerns regarding false positives warrant the use of a specially-designed test.189
The conclusion here on the error cost debate seems to be that the error-cost framework (and the decision-theory that necessarily goes with it) is less useful when we do not have empirical data on the likelihood and magnitude of false positives and negatives. Perhaps no data came out of the hearings on these issues, but there is a substantial economic literature on single firm practices ranging from RPM to exclusive dealing to tying. See, e.g. this recent survey of that literature by colleagues of these FTC authors over in the Bureau of Economics. There are authors surveys as well. E.g., Lafontaine & Slade.
Here’s a quote from the first linked survey piece by Luke Froeb, James Cooper, Dan O’Brien and Michael Vita summing up the state of play:
Empirical analyses of vertical integration and control have failed to find compelling evidence that these practices have harmed competition, and numerous studies find otherwise. Some studies find evidence consistent with both pro- and anticompetitive effects; but virtually no studies can claim to have identified instances where vertical practices were likely to have harmed competition.
The error cost approach allows one to focus on an evidence-based antitrust policy. And frankly, at least in the monopolization area, the empirical basis for a more aggressive policy just isn’t there no matters. Assertions about the rarity of judicial errors in favor of plaintiffs in antitrust cases do not change that given the current state of our empirical knowledge.