Simple Estimators for Binary Choice Models With Endogenous Regressors

This paper provides two main contributions to binary choice models with endogenous regressors. First, we propose some variants of special regressor based estimators that are numerically trivial to implement. These estimators provide consistent estimates of binary choice model coef cients when regressors (either discretely or continuously distributed) are endogenous, and when the latent errors have heteroskedasticity of unknown form. We also propose an alternative to the average structural function (ASF) measure of tted values for models having a latent index structure that is easier to calculate than ASF. We use this to provide simple estimators for choice probabilities and marginal effects of the regressors. We illustrate these methods with an empirical application to the estimation of migration probabilities within