DISCRIMINATION, BAYESIAN UPDATING OF EMPLOYER BELIEFS, AND HUMAN CAPITAL ACCUMULATION

This article proposes a dynamic labor market discrimination model based on Bayesian updating of beliefs by the employer. The employer forms initial beliefs of worker ability conditional on group and updates his beliefs each period based on observed output. Inaccurate initial priors lead to diminished human capital investment among members of the undervalued group and may generate inequities lasting many periods or even permanently. Statistical discrimination models in which differing variances drive the inequity are special cases of the model.