(Revised October 2026)

Wage growth is a key indicator of labor market conditions, but common measures often conflate individual wage changes with shifts in workforce composition. This paper develops a composition-adjusted measure of wage growth using nonparametric decomposition and program evaluation methods. The adjusted measure tracks unadjusted growth in stable periods but diverges during disruptions: during the Covid-19 pandemic, wage growth falls from 12.1% to 6.1% after adjustment. The method accommodates rich covariates and is robust to data quality issues such as rounding, heaping and top-coding. It also enables distributional and subgroup analysis using micro data, which offers more accurate views of underlying wage dynamics. We consider three adjustments: distributional regression, propensity score matching, and reweighting. Based on limited Monte Carlo evidence, we therefore focus on a bias-adjusted distributional regression as our preferred measure. When applied to the wage Phillips curve, our adjusted measure responds more strongly to labor market slack than the commonly used alternatives, and most strongly in the lower part of the wage distribution.