Page 1 of 1

GMM

Posted: Fri Nov 22, 2013 6:21 am
by MarkDijkstra
Hi there,

I'm having problems in estimating a system of equations through GMM. However, this works when I tick the boxes and run the system, but not when I write a program and run it.

More specifically, the estimation method I'm using is 'GMM - Cross Section (White Cov.)', ticking the box for 'Identity Weighting Matrix', but not ticking the box for 'Add lagged regressors'. It works fine then.

In my program, I tried to do this in a similar vein by using:

sysbaselinegmm.gmm(instwgt=tsls,instlag=0,cov=white)

But then Eviews gives me an error (near singular matrix)

I suspect I'm specifying wrong in my program, but I'm unsure what the correct specification is. Hopefully, someone can help.

Re: GMM

Posted: Sat Nov 23, 2013 11:44 am
by EViews Gareth
Best thing to do is estimate via the dialogs, then click on View->Representations to view the command line form of the estimation.