Estimating Panel Data Models in the Presence of Endogeneity and Selection

The present paper proposes a strategy for estimating panel data models in the presence of sample selection and endogeneity arising from endogenous explanatory variables and unobserved heterogeneity. We propose three alternative tests for selection bias and estimation strategies to correct for selection in the presence of endogenous regressors under the availability of valid (strictly exogenous) instruments. The approaches are illustrated using a panel of European Labour Force Survey micro data.