Looking again at instrumental variable estimation of wage models in the gender wage gap literature

Abstract

The evaluation of the gender wage gap centers on the consistent estimation of the parameters of interest in a human capital wage regression model. In previous studies, the consistency of the parameter estimates often depends on the restrictive assumptions of strict exogeneity and mean stationarity of corresponding explanatory variables in the model. This applies, particularly, to the work history variables, that is the work experience and the time out of work variable. In this paper, we specify a wage model that allows these variables to be predetermined and discuss consistent estimation by generalised method of moments estimators. For estimation, we use a new longitudinal administrative data set with particularly advantageous features to measure the main variables, i.e. wages and human capital. Our results show that wage models estimated in levels and using first differences of the endogenous regressors as instruments are not valid.

Citation

Kunze, A. (2001), "Looking again at instrumental variable estimation of wage models in the gender wage gap literature", Polachek, S. (Ed.) Worker Wellbeing in a Changing Labor Market (Research in Labor Economics, Vol. 20), Emerald Group Publishing Limited, Leeds, pp. 373-393. https://doi.org/10.1016/S0147-9121(01)20049-1