Dear Statalist Members,
I am estimating a model that includes an endogenous variable that I had
previously estimated using xtabond procedure, i.e., my first stage is an
estimation of a dynamic model. I thought of using the predicted variable
from this first stage directly into the second stage, but I believe my
estimated standard errors would be biased. However, by using xtivreg the
first stage would be estimated as a linear regression. I read in the
Statalist archives regarding a Tobit example that the consistency of the IV
estimators does not depend on the functional form of the first equation.
Would that apply in my case?
Another related question is as follows: Another model I need to estimate
also has an endogenous variable as regressor, but here the second stage is a
Tobit and the first one is linear and I use xtreg fe. How can I estimate
this model? Can I just estimate it in two steps: run the xtreg fe and use
the predicted variable directly into the Tobit estimation? Wouldn't I have
the same problem of standard errors mentioned above? Is there a better way?
Thanks in advance for your help,
M�nica L.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
