Abstract:
In this paper we provide semiparametric estimation strategies for a sample selection model with a binary dependent variable. To the best of our knowledge, this has not been done before. We propose a control function approach based on two di erent identifying assumptions. This gives rise to semiparametric estimators which are extensions of the Klein and Spady (1993), maximum score (Manski, 1975) and smoothed maximum score (Horowitz, 1992) estimators. We provide Monte Carlo evidence and an empirical example to study the nite sample properties of our estimators. Finally, we outline an extension of these estimators to the case of endogenous covariates.
