|From||PATGIRI Rajdeep <Rajdeep.PATGIRI@insead.edu>|
|Subject||RE: st: xtivreg|
|Date||Tue, 16 Nov 2004 20:04:41 +0100|
Thanks for the info. This thing seems to be working fine!
From: Mark Schaffer [mailto:M.E.Schaffer@hw.ac.uk]
Sent: Tuesday, November 16, 2004 12:56 PM
Subject: Re: st: xtivreg
You've got three options:
- Keep using xtivreg. This will generate consistent, albeit
inefficient, estimates even if one or more endogenous regressors is
dichotomous. You could also consider Wooldridge's suggestion of
using the probit to generate fitted values and using these fitted
values as instruments in your IV equation. See e.g. the discussion
on Statalist earlier this year:
- Use a canned program such as treatreg or cdsimeq and include dummy
variables for your fixed effects.
- Program it yourself.
Hope this helps.
From: PATGIRI Rajdeep <Rajdeep.PATGIRI@insead.edu>
To: "'firstname.lastname@example.org'" <email@example.com>
Subject: st: xtivreg
Date sent: Tue, 16 Nov 2004 11:00:38 +0100
Send reply to: firstname.lastname@example.org
> I am trying to run "xtivreg" for my 2-SLS regression. I suspect that
> Stata only performs linear OLS in the first stage. However, I need to
> run a Probit model in the first stage. Is there any way out? I tried
> to use "heckman", but it doesn't seem to handle panel data (it only
> performs a pooled regression). Thanks in advance!
Prof. Mark E. Schaffer
Centre for Economic Reform and Transformation
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS UK
44-131-451-3485 CERT administrator
* For searches and help try: