[Gretl-users] gmm usage

[Sven Schreiber]

Allin Cottrell schrieb:
> On Fri, 21 Sep 2007, Sven Schreiber wrote:
> 
>> now that I have my first gmm estimates produced by gretl, I 
>> would like to know what else is possible -- I already found out 
>> that "restrict" apparently is _im_possible. What about saving 
>> the estimated coefficients or covariance matrix for further 
>> (scripting) use? (I have read the gmm chapter but couldn't find 
>> anything.)
> 
> You already have the coefficients, since you defined them in the 
> first place (and they are modified in the course of estimation).

Right, so that was sort of a stupid question...

> 
> The covariance matrix is obtained in the usual way:
> 
> matrix V = $vcv

ok good to know; actually I just properly realized a minute ago that GMM
is also available in a dialog window, not just via scripts... let's see
what that gives...

> 
> The usual "restrict" apparatus can't apply, because we can't 
> assume that the parameters enter the model linearly.
> 
> You should, however, be able to save the value of the GMM 
> criterion function, so you can cook up your own restriction test 
> -- but if I'm not mistaken, you can't at present.

yes that would be nice.

-sven