Website CV Paper
Fields
Econometrics, Labor, Development
Job market paper
The Anatomy of Wage Compression: Locally Robust Inference for Conditional Inequality and Exact Decomposition
An aggregate decline in wage inequality may reflect convergence in group mean wages or reduced dispersion among comparable workers. This paper develops a methodology that allows for an exact decomposition of these two dimensions conditional on a large number of observable factors. In particular, I combine an exact within–between decomposition of the Theil index with conditional-inequality profiles, accommodating high-dimensional worker, job, and employer characteristics through regularized learners. I derive a Neyman-orthogonal signal for conditional inequality and locally robust estimators for the decomposition, using cross-fitting to provide valid inference. Applying the framework to Portuguese matched employer–employee data from 2011 to 2019, a period of substantial minimum-wage increases, I find that the between-component decline slightly exceeds the aggregate decline, with about 95 percent captured by conditioning on worker and job characteristics and hence a limited role for firm characteristics. Despite a nearly unchanged aggregate within component, conditional inequality generally declines among non-college workers, particularly among high-school and some-college women and men with greater initial exposure to the 2014 minimum-wage increase. College-educated men show little comparable decline. These findings reveal wage compression for specific population groups concealed by aggregate measures.