Page 1 of 1
GMM for Linear System
Posted: Tue May 20, 2014 10:48 am
by BobR
I have an estimation problem that involves two linear equations. They are not identical equations. I can see how to do GMM on individual equations but, having read the manual repeatedly, I can't seem to figure out how to do GMM on the system as a whole. The manual seems to omit this case at least as far as I can see. Any help on how to do this is appreciated.
Re: GMM for Linear System
Posted: Tue May 20, 2014 11:01 am
by TomDoan
Even with linear equations, it's often simpler to just use NLSYSTEM as described in Section 4.10 of the v8 User's Guide. The only two examples that I have for GMM done with SUR are GRNP333.RPF (from Greene's 7th) which doesn't even work well since there isn't enough data to freely estimate the weight matrix, and the Holtz-Eakin, Newey, and Rosen panel VAR example.
Re: GMM for Linear System
Posted: Sat May 24, 2014 11:40 am
by BobR
Thanks for your prompt response Tom. I checked out the file that you mentioned and saw that the GMM set up was in there. So I reran my problem with the instruments and zudep options and I get system IV results. But, just to be clear, the output does not mention GMM I think so is GMM on the system what I am getting? Thanks for your help.
Re: GMM for Linear System
Posted: Sat May 24, 2014 1:57 pm
by TomDoan
Sorry, yes. You're getting GMM. We'll have to change the labeling.