ari.dothan <[email protected]>:
If you can write out the moment conditions, you can easily program
your GMM problem in Mata without resort to the new -gmm- command; see
also http://www.stata.com/meeting/snasug08/nichols_gmm.pdf
On Tue, Feb 16, 2010 at 12:49 AM, ari.dothan <[email protected]> wrote:
> Thanks very much for your answer, Brian.
> Based on your two messages I conclude that my best shot would probably be to
> use xtdata for equation 2, then use gmm for the system of two equations,
> while the first one would be the differences equation, and the second one
> would be the level equation. This way I could address the fixed effects in
> both (though not rigorous enough by xtdata), and still have the system
> standard errors taken care of.
>
> An alternative would be to run a differences gmm first of equation 2 alone,
> predict the dependent variable, and use that result for the level equation
> in the simultaneous gmm of both equations.
>
> Do you prefer one of the two strategies I suggested, or would you advise
> differently?
>
> Your input is highly appreciated
*
* 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/
