Export Entry and the Gender Wage Gap: When Large Firms Drive Worker-Level Estimates

This paper examines the effect of export entry on the within-firm gender wage gap. Existing studies provide mixed evidence on this relationship. I show that part of this inconsistency reflects how the standard worker-level empirical approach aggregates firm-level treatment effects. When linked employer-employee data are used to estimate effects that operates at the firm level, ordinary least squares (OLS) does not recover a simple average across firms. Instead, firms receive implicit OLS weights that are proportional to the number of worker observations they contribute, giving disproportionate influence to large firms. Applying the standard empirical approach on German data, I estimate that export entry reduces the within-firm gender wage gap by approximately 1.2 percentage points. However, this average effect disappears once the estimation is reweighted by inverse firm size, thereby removing the mechanical firm-size component of the implicit OLS weights. The apparent reduction in the gender wage gap is therefore driven by a small number of large exporters rather than representing a general pattern across firms. The paper highlights an important issue that arises whenever individual-level data are used to estimate treatment effects operating at the group level: in the presence of heterogeneous group-level effects, OLS estimates reflect implicit weighting schemes rather than simple averages across groups. Recognizing these implicit weighting structures is therefore essential for interpreting empirical estimates and drawing policy conclusions.