I’ve done some more thinking about my recent post on the problems associated with claims that infer greater antitrust enforcement quality solely from enforcement activity and come to the conclusion that my post oversimplified matters. I remain rather skeptical about this inference but wanted to highlight some of the nuances in the debate that I skimmed over in the previous post. In the spirit of highlighting some of these issues, and because I’ve always wanted to try one of these posts, here’s a hypothetical dialogue between myself (JW) and an anonymous economist with a more interventionist bent (AE) in the spirit of highlighting some of the complexities of the potential link between enforcement activity and consumer welfare:
JW: The inference that consumers are better off where antitrust activity is higher doesn’t make much sense to me. For starters, there is too much noise in the data. We’re talking about very small changes in aggregate enforcement rates over time and it is difficult to assign meaning to these changes. Also, antitrust enforcers make errors from time to time and those errors can make consumers worse off. I’m especially concerned with false positives and deterring pro-competitive conduct. Fear of antitrust liability, treble damages, and follow-on private litigation can add up to quite a deterrent. So who’s to say that more is better?
AE: I see where you’re going with this. But can you name a Section 2 case that you don’t think is a “false positive”? But more to the point, there’s no avoiding that there is plenty of noise in the data. But there is an even bigger problem with making the move from activity level to quality than that. The mix of conduct and deals the antitrust agencies see over any given time period is endogenous to their enforcement decisions. That is, profit-maximizing firms condition their behavior accordingly to the enforcement standards they observe. This means that we should pretty much always see a constant fraction of cases that require “borderline” judgments with respect to enforcement. If that’s true, deviations from average enforcement levels don’t tell us anything about whether enforcement is high or low in an absolute sense. It only tells us that enforcement is high or low relative to expectations.
JW: So it sounds like you’re agreeing with me that we can’t use activity levels to say that more enforcement is good as a general matter?
AE: Not so fast. You can always use more than the activity level to make that sort of inference. That is exactly what Baker & Shapiro do in the paper you blogged about previously — they use enforcement levels by enforcers with a non-interventionist history and rhetoric to say something about levels as actual enforcement deviates from expectations. So, yes, I agree you cannot use activity level alone, but that doesn’t mean you can’t say anything about whether more enforcement would be a good thing if you can gauge deviations from expectated levels of enforcement.
JW: Ok. That’s fair enough. But I don’t know whether those deviations are good or bad for consumers. Isn’t that what we are really talking about here? For instance, I might agree with you that unexpected deviations toward greater enforcement efforts against hard core cartels would consistently produce consumer welfare benefits. But who says more enforcement means more enforcement against cartels, which we know harm consumers most of the time, rather than more enforcement against single firm conduct, which we know much less about in terms of competitive effects? If more enforcement means a substitution of resources away from cartel enforcement and toward single firm conduct I would think that we might see more false positives and lower consumer welfare as a result. In other words: More enforcement, higher prices.
AE: Yes, it’s certainly possible. But I doubt it. As long as the average enforcement action produces benefits for consumers, more is better even if the mix of cases changes. I think this is probably true for all types of antitrust intervention. You think this is true for cartels, less true for mergers, and probably not true for single firm conduct. That’s the difference. You’re just wrong about the last two.
JW: We’ll at least we both know who is wrong. And I think we’ve identified where we part ways here. But let me raise one more issue. My public choice theorist colleagues at George Mason will be very upset if I don’t also raise the point that there are other factors that are likely to determine where and what types of cases we see if antitrust activity increases. It is not too much of a stretch to imagine that decisions to increase antitrust enforcement might not be based solely on consumer welfare considerations rather than private incentives of those within the agencies, political factors, and other things that may be negatively correlated to consumer outcomes. For examples, the antitrust ambitions of a few presidential candidates who have pinpointed industries of antitrust concern come to mind …
AE: The public choice angle is interesting. But I’m not quite sure it supports your argument that more enforcement means more single firm enforcement and less resources for cartel enforcement. I’ll have to think about that. Two more points that make this a bit more complicated that we haven’t considered yet. You’ve ignored the possibility that there may be increasing returns to scale over the relevant range in antitrust enforcement as agency officials learn better how to detect, prosecute, and remedy violations. Also, current antitrust enforcement appears to be dedicating lots of resources to cartel enforcement and little to single firm enforcement. Doesn’t this mean that the marginal benefit of devoting resources to single firm cases is higher than for cartel cases?
JW: Not if the marginal benefit from cartel cases is greater than zero! I’m kidding. These are both interesting points. But where’s the evidence that there is increasing returns to scale in antitrust enforcement or that single firm enforcement has on average produced consumer welfare gains? Obviously, some good cases exist. We’re talking averages here. But one can easily imagine that any benefits from good single firm cases could be swamped by the social costs imposed by false positives. This seems to be where we part ways on this subject: I worry that the average impact of more enforcement in the single firm category over time might drag the average below zero and make consumers worse off, while you believe that the averages are all positive for all types of intervention and so don’t lose any sleep when the agencies get more active.
AE: That sounds about right. But lets not forget that your assertions about false positives aren’t support by a great deal of empirical evidence either! I know, I know, you it is really hard to measure a reduction in competitive conduct we don’t see. But still, lets agree that we’re both dealing mostly with anecdotal evidence here. So the real question is who has the burden of persuasion here, isn’t it?
JW: Obviously, you do! I guess we’ll have to agree to disagree on that point for now. We’ve got to leave something to talk about next week.