Dear all,
Thanks to Prof. Baum, a new version of xtheckmanfe is now available.
This command estimates the Correlated random effect Heckman selection model proposed by Wooldridge (1995).
The previous version of the program would estimate standard errors using a clustered bootstrap approach. On this version, however, I added the option to obtain the standard errors using what I call a pseudo two-step approach.
This means that the model itself still follows the two step approach as before, but that standard errors are estimated using maximum likelihood.
Best wishes
Fernando