Book Description

In this context, a glass ceiling is understood to manifest as a large disparity in the top of the distribution, with less disparity in the middle and bottom of the distribution, conditional on the productivity-related characteristics of workers (such as education). [...] The literature on glass ceilings for women finds large disparities at the top of the conditional earnings distribution: for example, Albrecht finds earnings disparity on the order of 20% at the top decile cutoff of conditional earnings. [...] The papers which find differences in the lower quantiles of the conditional distributions tend to interpret the effects at the bottom of the distribution as being due to some process which is different in spirit from a glass ceiling. [...] At the bottom of the conditional earnings distribution, the difference in log- earnings shrinks by more than half, and at the top of the distribution, it shrinks by almost half. [...] However, given that quantile regressions, which do not control for work characteristics, show larger differentials at the bottom of the conditional earnings distribution, this implies that work characteristics soak up more of the earnings differentials at the bottom of the distribution than at the top.