# Labor market sorting and the gender pay gap revisited

## TL;DR

This paper shows that gender segregation in jobs affects how we measure pay gaps, and using better methods can significantly change the estimated size of these gaps.

## Contribution

The paper introduces a new method for estimating gender pay gaps that accounts for labor market segregation, leading to more accurate results.

## Key findings

- Gender segregation in the labor market leads to non-identifiable pay gaps in certain segments.
- Flexible semi-parametric estimators reduce unexplained gender pay gaps by up to 44% compared to standard methods.
- Enforcing comparability and choosing the right estimator significantly impact pay gap estimates.

## Abstract

This paper shows that gender segregation in the labor market has important implications for the estimation of gender pay gaps. Using Switzerland as an example, we provide evidence that there are sizable segments in the labor market with perfect sorting such that there are no comparable men and women. In these segments, covariate-adjusted gender pay gaps are not identified non-parametrically. Reliability of estimated pay gaps then requires correct functional forms for extrapolation or excluding segments of the labor market with perfect sorting from the analysis. We discuss different estimation choices within this trade-off and show how they affect estimates of unexplained gender pay gaps. We find that enforcing comparability ex ante, estimator choice and functional form restrictions matter greatly. Using a flexible semi-parametric estimator with moderate restrictions on ex ante comparability explains up to 38% more of the raw gender pay gap and results in estimated unexplained gender pay gaps that are up to 44% smaller than standard Blinder-Oaxaca estimates that account for the same wage determinants but ignore lack of overlap.

The online version contains supplementary material available at 10.1007/s00148-025-01115-1.