Testing for Statistical Discrimination: The Case of Immigrant Physicians in Israel

Abstract

Two models have been proposed to explain why two groups of workers who have the same level of productivity can receive different wages: (i) the statistical discrimination model, according to which the discrimination is the result of a rational response to uncertainty about productivity that consists in adopting certain characteristics as proxies for productivity and (ii) the taste-based discrimination model, according to which the discrimination is based on prejudice against a particular group. It is usually difficult to determine empirically whether the gap in earnings is due to statistical or taste-based discrimination. However, the conditions for a natural experiment were created in 1989 when Israel introduced a licensing examination for immigrant physicians arriving from that point onward, thus making it possible to determine which type of discrimination is responsible for the wage gap between immigrant and native physicians. Controlling for other factors, the paper finds that: (1) the earnings of physicians who immigrated to Israel after 1989 are significantly higher than those of physicians who immigrated previously and (2) the earnings of immigrant physicians who have taken the licensing examination converge to those of native physicians after an average of 5.5 years. These results constitute evidence that the difference in earnings between native and immigrant physicians is due to statistical, rather than taste-based, discrimination.