st: reg3 - p-value for difference in coefficients across 2 models

[Sascha Heller <Sascha.Heller@wiso.uni-hamburg.de>]

Dear all,

I am trying to estimate a p-value for the difference in coefficients  
on one variable in a 3SLS regression framework for two subgroups.

Normally, I would use "suest" which does, however, not work for the  
"reg3"-command. One possibility is to estimate a joint model with a  
"subgroup-dummy and variable" interaction term. Unfortunately, the  
variable for which the p-value should be estimated is the dependent  
variable in the second equation for the endogenous explanatory variable.

reg3 (y variable x1 x2 x3) (variable x4 x5 x6)

I would regress the interaction term on the instrumental variables in  
the second equation which seems -let's say- "unorthodox" to me. The  
endogenous explanatory variable must be estimated

Is there any procedure to apply? Or any built-in command or ado-file?  
I couldn't find anything on the web.

Thanks in advance.

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