Testing for Asymmetric Employer Learning and Statistical Discrimination

We test the implications of a statistical discrimination model with asymmetric learning. Firms receive signals of productivity over time and may use race to infer worker's productivity. Incumbent employers have more information about workers productivity than outside employers. Using data from the NLSY79, we find evidence of asymmetric learning. In addition, employers statistically discriminate against non-college educated black workers at time of hiring. We also find that employers directly observe most of the productivity of college graduates at hiring, and learn very little over time about these workers.