Dear Statalisters,
I would like to perform a two stage estimation where I use xttobit in
the first equation to predict the values of the suspected endogenous
variable and use them in the second equation, an xtpoisson. Reading
Woolridge and statalistes threads, I understood that the standard
errors of the second equation can be biased. If my second equation was
an ols I could have used the Woolridge 2002 suggestion of solving the
problem with ivreg (z= yhat). However, since my second equation is an
xtpoisson, what would you reccommend? Thank you very much in advance.
AR
*
* 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/
