st: correct errors for the Heckman two-step method

["Robinson, Adina Barbulescu" <adina.barbulescu@utk.edu>]

Hello, 

I currently control for selection bias with the Heckman two-step method. In the first step, I estimate the selection model (xtprobit) and obtain the inverse mills ratio. In the second step (and with a different file), I estimate the main model (xtreg) to which I add the inverse mills ratio obtained from the first step. For various reasons I cannot combine the two files (selection file and main model file) and use the heckman command. As a result, the standard errors I obtain are not the correct ones. Stata produces the correct standard errors following the heckman command, but I don't think this is the case here. 

Does anyone know how I can estimate the correct errors following the Heckman two-step method when the selection and the main model are in two datasets that cannot be combined? 
 
Many thanks!

Adina





*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/