
This note shows how the Shapley-value can be applied to the regression-based methods that are often used to decompose changes in wage distributions. The method remedies the path-dependency exhibited by existing approaches that compute the contributions due to (i) changes in sample observable characteristics, (ii) changes in the return of characteristics, (iii) changes in the distribution of unobservable characteristics. _x000d_ _x000d_ JEL code: J31_x000d_ Keywords: regression-based decomposition methods, wage distribution, Shapley-value.