A root-N consistent estimator for some fixed-effects panel data sample selection models

The aim of this note is to present a root-N consistent estimator for the regression coefficients in a fixed-effects panel data sample selection model that has one selection equation and two regression equations. Other than the condition that the error terms satisfy a control function type of restriction and the biases are nonzero in both equations. The proposed procedure does not require parameterizing the selection equation nor does it require estimating the selection equation. Moreover, the proposed procedure only requires a standard IV estimation and hence is easy to implement.