Dear Statalist,
I'm trying to estimate the following treatment effects model:
Y=bX+aZ+u,
where Z is a treatment variable (0,1) and Y is an ordered variable
(1,2,3).
The "treatreg"-command don't take into account the ordered nature of Y.
So I tried a two stage estimation.
In the first stage I ran a probit regression on Z: Z=c*W+e and
calculated the inverse Mill's.
In the second stage I included these inverse Mill's in the original
regression on Y and ran an ordered probit: Y=bX+aZ+d*inverse Mill's+u1.
I suspect the the standard errors received that way are incorrect. Is
that right and is there any appropriate procedure in Stata?
Thanks a lot,
Georgi Tsertsvadze
University Hannover
Department of Economics
K�nigsworther Platz 1
30167 Hannover
Germany
*
* 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/
