Estimation and Speci fication Testing of Panel Data Models with Non-Ignorable Persistent Heterogeneity

This paper shows how a simple modi…fication of estimators based on the RandomEffects principle can preserve the consistency and asymptotic efficiency of the methodin panel data despite non-ignorable persistent heterogeneity driven by correlationsbetween the heterogeneity and the regressors. The approach is extremely easy toimplement and allows straightforward tests of the signi…ficance of such correlationsthat lie behind the non-ignorable persistent heterogeneity. The method applies tolinear as well as nonlinear panel data models, static or dynamic.[...]