I have been trying to estimate a bivariate tobit using Stata's ml subset of
commands, to no avail. I first tried to estimate my model with another
statistical package and ran into convergence problems caused by one of my
variables of interest. To make sure that the problem was not caused by that
other statistical package, I decided to use Stata to verify if I truly had
a data problem.
My problem is that I get an r(491) every single time I run my code, i.e.
Stata cannot find feasible initial values, even when I regress only on
constants to see if it works! I did check and recheck the maximum
likelihood program I wrote, but I cannot find mistakes at this point.
I thus wanted to ask if anyone had ever estimated a bivariate (or
multivariate) tobit using Stata, and if anyone has done it before, would
they mind either looking at my code or sharing theirs?
Thanks in advance,
Marc
__________
Marc F. Bellemare
Ph.D. Candidate
Department of Applied Economics and Management
Cornell University
Ithaca, NY 14853-7801 http://www.people.cornell.edu/pages/mfb24
