Labor market underrepresentation results in minority discrimination: A dynamic hiring model with employer learning

This paper develops a dynamic model of minority labor market discrimination. Employers repeatedly decide to hire either minority or majority job candidates whose productivities are unobservable beforehand. Hiring decisions are based on productivity expectations derived from the observable productivity of employers’ previously hired workers. If employers have fewer minority workers initially—a plausible assumption for (numerical) minorities—they discriminate against minority workers over time. Discrimination results from more dispersed minority expectations across the employer population and stronger effects of additional productivity observations on minority expectations. Both effects are a direct consequence of the minority’s initial underrepresentation in firms. I demonstrate the emergence of minority discrimination formally in a two-period hiring model and show simulation results for longer time frames.