Bias Correction for Within-Group Estimation of Panel Data Models with Fixed Effects and Sample Selection

For linear panel data models with fixed effects and sample selection, we correct for selectivity bias in the within-group estimator. The proposed procedure is equivalent to a pooled weighted least squares based on pairwise differences augmented with correction terms. A computationally affordable method of estimating nuisance temporal correlation parameters in the selection equation errors is also proposed. Analytic standard errors are derived for the multi-step estimator. Our method is easier to implement and performs well in comparison to the minimum distance approach according to simulations