Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Re: Suppressing the Constant in the First Stage with ivreg2


From   Kit Baum <[email protected]>
To   [email protected]
Subject   st: Re: Suppressing the Constant in the First Stage with ivreg2
Date   Wed, 17 Jun 2009 07:00:40 -0400

<>
I don't think so. The FSRs are projecting each one of the endogenous variables on all of the instruments, included and excluded. If you include a variable that takes on a constant value in the regression of interest (by doing, e.g., ivreg2 y iota (y2=z1 z2), noco, where iota=1), ivreg2 will properly include iota in the instrument list, and the FSR will contain iota as an explanatory variable. That is as it should be; if the vector of ones appears in the main regression, it must be an instrument in the projection of the endogenous vars on all of the instruments. So the issue is not whether ivreg2 will do it, but whether there is any theoretical justification for doing so. I think not.

Kit

Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
An Introduction to Stata Programming | http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html



On Jun 17, 2009, at 02:33 , Anson wrote:

I have a particular model where I need to suppress the constant term
from the first stage estimation of a two stage model using Fuller's
technique.  As many of you know ivreg2 is a great plug in for
implementing this estimation, and I would like to stick with it
(rather than coding my LIML problem by hand).  Is it possible to
suppress the constant in the first stage, and include it in the
second?  Thank you all.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index