Wednesday, February 22, 2006
st: RE: 2sls with first stage probit
> -----Original Message----- > From: firstname.lastname@example.org > [mailto:email@example.com] On Behalf Of > firstname.lastname@example.org > Sent: 22 February 2006 01:09 > To: email@example.com > Subject: st: 2sls with first stage probit > > Hi, I was looking through past inquiries about ivreg in stata > when the first stage regression is a probit. I found the > following advice, following Wooldridge (2002): > > . probit w x1-xn z > . predict ghat > . ivreg y x1-xn (w = ghat) > > My question is not so much about the econometrics as it is > about how to tell stata to do what I want it to do. First, > how do I tell stata that I want to provide my own fitted > values from the first stage (I have more than one endogenous > variable)?
Not sure what you mean, but probably you don't want to provide your own fitted values; for one thing, the standard errors will be wrong. The extension of the Wooldridge procedure to multiple endogenous variables is to generate multiple ghats as instruments. But...
> Even better, how can I tell stata that I want it > to run a probit in the first stage of the ivreg procedure.
This looks like Stata's built-in -treatreg- estimator. BTW, it is a system estimator, i.e., the probit equation and the outcome (y) equation are estimated jointly. This means the usual efficiency-robustness tradeoff, i.e., treatreg will be more efficient than single-equation IV, but the estimates for neither equation will be consistent if either one is misspecified.
> Ideally, I would like to know how to run a probit in both > stages and also a probit only in the first stage. Thanks in > advance for any help, > > Cory Koedel > > * > * For searches and help try: > * http://www.stata.com/support/faqs/res/findit.html > * http://www.stata.com/support/statalist/faq > * http://www.ats.ucla.edu/stat/stata/ > >
* * For searches and help try: * http://www.stata.com/support/faqs/res/findit.html * http://www.stata.com/support/statalist/faq * http://www.ats.ucla.edu/stat/stata/
Links to this post: