Can Statistical Discrimination Explain Inequality?

Christian Hunkler, University of Mannheim

Statistical discrimination theory assumes that employers have incomplete information about real productivity of applicants, even after hiring. This paper focuses on the disputed question whether Phelp's measurement model of statistical discrimination can explain inequality in hiring, i.e. group discrimination. The central argument is that employers perceive differences in the reliability of productivity signals between groups. They trust productivity signals less, if these signals come from specific groups. The theoretical analysis finds the model capable of explaining group discrimination, i.e. inequality, when considering allocation decisions - the hiring of workers. Contrary to intuition the direction of discrimination depends on the relation of workers seeking a job to the number of open positions. Using simulations, I show that the group whose signals are trusted less is discriminated in very competitive labor markets, whereas, under the conditions of less competition, the model even predicts discrimination against those workers whose signals are trusted more.