Gender wage gap false positives: a monte carlo simulation

Abstract

This study examines whether and how often the standard regression-based model would falsely detect gender wage discrimination in a “discrimination-free” world. In the current context of the EU Pay Transparency Directive (EU) 2023/970, we build a Monte Carlo simulation that generates individual wages under a zero-discrimination benchmark, driven by observable skills and realistic wage-setting frictions. The results present non-trivial false-positive rates that rise with multiple gender-related tests, while goodness-of-fit (e.g., adjusted R2) does not reliably indicate detection risk. The findings highlight the limitations of identification and the sensitivity of specifications in gender wage gap calculation, indicating that enforcement and compliance assessments must consider multiple testing and robustness checks to prevent erroneous conclusions.