"For this study, the author inspects the relationship between segregation in the workplace (measured as the proportion female in job cells) and the gender wage gap using linked longitudinal employer-employee data from the German Employment Register. He extends the literature by controlling for nonrandom sorting of workers into job cells, establishments, and occupations. In line with previous studies, the pooled least squares estimates show that the gender wage gap increases as the job-cell-level proportion of females increases. This increase is attributable to the fact that women experience greater wage declines than men do when additional women enter their job cells. Controlling additionally for unobserved heterogeneity at the individual, establishment, occupation, and job-cell levels considerably reduces the size of the proportion female effects on women's wages while rendering the effects on men's wages insignificant or even positive. The same controls also significantly decrease the proportion effects on the wage gap. The related sorting analysis shows that a good deal of the proportion effects can be explained by unobserved individual ability and suggests that especially women working in job cells with small proportions of females show above-average unobserved individual ability. "

Support: Paper