Friday, July 23, 2021
The idea that consideration of error costs should inform judgments about actions with uncertain consequences is well established. When we act on imperfect information, we consider not only the probability of an event, but also the expected costs of making an error. In 1984 Frank Easterbrook used this idea to rationalize an anti-enforcement bias in antitrust, reasoning that markets are likely to correct monopoly in a relatively short time while judicial errors are likely to persist. As a result, false positives (recognizing a problem when there is none) are more costly than false negatives. While the problem of error cost bias is not explicitly mentioned all that frequently in antitrust cases, its influence is broad and deep, guiding the formation of presumptions and burdens of proof.
The anti-enforcement bias in antitrust originated long before Easterbrook wrote, and was reflected in the work of George J. Stigler and Milton Friedman, mainly in the 1940s and 1950s. They had attempted to dismantle theories of imperfect competition in favor of models in which competition nearly always prevailed unless restrained by government action.
While the ideas underlying the error cost framework were relatively new ones in law schools, by the 1980s they were already in decline in industrial organization economics. Easterbrook was writing defensively. An empirical revolution in economics was already well engaged in a process that found imperfect competition models to be more testable, more dominant, and more useful for policy judgments. In the process, classical Marshall/Stigler models of perfect competition were all but abandoned as irrelevant.
This empirical work provides increasing evidence that United States antitrust policy, but particularly merger policy, took a significant wrong turn in the mid-eighties. Today the error cost framework exists mainly as a result of rent seeking by firms who stand to profit from the low output and high margin consequences of this antitrust anti-enforcement bias.