Why the Gender Pay Gap Is Wider Than It Looks for Low-Earners
Batbayar et. al: "Quantile Selection in the Gender Pay Gap" CRC Discussion Paper No. 560
The persistent disparity in earnings between men and women remains a central concern for economists and policymakers, yet accurately measuring this gap requires accounting for non-random selection into the labor force. A new study by Egshiglen Batbayar (University of Bonn) and colleagues introduces a semiparametric approach to estimate selection-corrected quantiles of the gender wage gap, providing identification without the restrictive parametric assumptions common in traditional models. Identification relies on exogenous variation that affects latent wages but, conditional on the latent wages and other observed variables, provides no additional information on the selection mechanism. Using German administrative data, the authors demonstrate how employment selection patterns vary significantly across the wage distribution and between genders.
The researchers utilize the individual’s initial wage as an instrument to capture persistent individual heterogeneity, such as ability and motivation, which shapes potential earnings over the life cycle. In the context of a Roy model, this strategy relies on a rank invariance condition, assuming the instrument provides no additional information about an individual’s position in the reservation wage distribution once potential wages and observed characteristics are fixed.
The empirical results reveal that women are positively selected into full-time work across the wage distribution, with the strongest effects occurring at the lower and median levels. Among the less educated, full-time working women tend to possess unobserved attributes linked to higher wages, meaning the observed data overstates their representative earnings. When these selection biases are corrected, the gender wage gap for the low-educated at the median increases from 3.6% to 10.7%.
For men, the selection patterns are distinct, characterized by strong positive selection among the highly educated at the top of the wage distribution. Correcting for this selection actually narrows the gender wage gap at upper quantiles, as top male earners are more positively selected than their female counterparts. Ultimately, the study underscores that ignoring these heterogeneous selection patterns can mask substantial disparities, particularly in the lower part of the wage distribution. Despite these complex selection effects, the authors find that men consistently out-earn women across all quantiles and education groups, confirming that the gender wage gap remains a persistent structural feature of the labor market.
Link: Quantile Selection in the Gender Pay Gap
