Noncompetes in the U.S. Labor Force*

Due to limited empirical evidence and controversial anecdotes, speculation over the ubiquity and importance of covenants not to compete in the U.S. labor market is rampant. We present an equilibrium framework to account for the existence and incidence of noncompetition agreements. We then populate this framework using data from a new large-scale survey of U.S. workers. The data show that noncompetes are a perhaps surprisingly common feature of the labor market. As a lower bound, we estimate that at least one in four employees have ever signed a noncompete, and that at least 12.3% are currently working under one. Of those with college education or above, at least one in ve are currently subject to a noncompete agreement. The occupations in which noncompetes appear most frequently are engineering (30%) and computer and mathematical occupations (28%), though they are prevalent in typically lower-skilled occupations as well, including installation and repair (11%), production occupations (11%), and personal services (12%). The observed heterogeneity in the incidence of noncompetes provides evidence that employers use noncompetes to prevent employees holding key resources from joining competitors. We also examine explicitly whether or not noncompetes are associated with the implications of our framework. We nd that noncompetes are associated with longer tenures, increases in the reservation wage for competitors, and increased training. We also show that employees rarely report negotiating over noncompetes, and that noncompetes are associated with no wage premium at signing but greater wage growth.