In the US, black workers spend more time in unemployment, lose their jobs more rapidly, and earn lower wages than white workers. This paper quantifies the contributions of statistical discrimination, as portrayed by negative stereotyping and screening discrimination, to such employment and wage disparities. We develop an equilibrium search model of statistical discrimination with learning based on Moscarini (2005) and estimate it by indirect inference. We show that statistical discrimination alone cannot simultaneously explain the observed differences in residual wages and monthly job loss probabilities between black and white workers. However, a model with negative stereotyping, larger unemployment valuation and faster learning about the quality of matches for black workers can account for these facts. One implication of our findings is that black workers have larger returns to tenure.
Keywords: Learning; Screening discrimination; Job search; Indirect inference.
JEL Codes: J31; J64; J71.
The views expressed are those of the author(s) and do not necessarily represent those of the Bank of Lithuania.