|From||PATGIRI Rajdeep <Rajdeep.PATGIRI@insead.edu>|
|Subject||RE: st: reg3|
|Date||Mon, 22 Nov 2004 10:48:15 +0100|
Following up from my previous email about "reg3", is there any way to check for overidentifying restrictions after reg3? The "overid" module that I know of is available only after ivreg and ivreg2 estimations.
And thanks a lot to those people who replied to my previous email.
From: Mark Schaffer [mailto:M.E.Schaffer@hw.ac.uk]
Sent: Sunday, November 21, 2004 5:01 PM
To: firstname.lastname@example.org; PATGIRI Rajdeep
Subject: Re: st: reg3 with Probit
Quoting PATGIRI Rajdeep <Rajdeep.PATGIRI@insead.edu>:
Omar Keshk's -cdsimeq- sounds like it might do what you want. The alternative is to estimate equation-by-equation, e.g., -ivprob- for your endogenous dependent variable and -ivreg- (or -ivreg2-) for the others.
Hope this helps.
> I am trying to run a system of simultaneous equations, but one of the
> regression equation is a binary dependent variable model, i.e. Probit.
> Is it possible in any way to incorporate into the "reg3" module of
> Stata? I am currently generating the predicted probabilities from the
> probit model first, and then replacing this as dependent variable in
> the probit model and
> run a least square. But I think the standard errors in here are not
> exactly reliable.
> Thanks for any input.
> Best regards,