Abstract

In the past three decades, several case studies have documented specific industries and instances whereby collusion was welfare-enhancing rather than harmful as is usually assumed. Specifically, two distinct “efficient cartel” hypotheses claim that inter-firm coordination can increase economic efficiency in industries with a large degree of avoidable fixed costs and/or variable output. This paper performs the first systematic empirical test of these hypotheses via an examination of cartel performance under the National Industrial Recovery Act of 1933, a two-year cartel experiment in the United States. While I find a wide variation in welfare changes during cartelization, there is no compelling evidence that differences in fixed costs are the cause. I do, however, find robust empirical support for the hypothesis that industries with highly variable output experience higher welfare gains (or less negative welfare declines) under collusion.