Assortative Matching and the Gender Wage Gap

• manuscript coming in August 2024

Abstract

Are we underestimating assortative matching in gender wage gap analyses? Using the two-stage distributional framework of \textcite{bonhomme_distributional_2019} in a massive Brazilian administrative data from 2010 to 2017, I address this question by eliminating the usual limitations of conventional models, particularly data trimming and limited mobility bias, known to severely impact estimates. The analysis reveals that assortative matching accounts for almost 28 percent of wage variance for female workers, approximately one third larger than traditional estimates. Monte Carlo simulations demonstrate that about 30 percent of the gender wage gap is due to labor market allocations. I also find that violations of additive separability principles impose higher friction to women, accounting for at least 9 percent of the gender wage gap.