Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: RE: Instrumental variables


From   "Padraig Dixon" <[email protected]>
To   <[email protected]>
Subject   Re: st: RE: Instrumental variables
Date   Tue, 13 Jul 2004 16:01:38 +0100

Cordula,

If theory beind your model does not suggest any "external" instruments, one
indirect solution would be to reformulate your model as a dynamic one (i.e.,
with a lagged dependent variable on the right hand side), and then estimate
the model with GMM - GMM is an instrumental variables estimator that uses
the orthogonality between certain lagged values of the dependent variable
and error differences to generate instruments that are used in an optimal
fashion. Although you may be keen to estimate a plain vanilla static fixed
effects panel, many economic relationships are often best modelled using
dynamics, although this of course may be inappropriate in your case.

It would be worth reading the very helpful paper in the link below by Steve
Bond on the GMM approach.

http://cemmap.ifs.org.uk/docs/cwp0902.pdf

Padraig Dixon



----- Original Message ----- 
From: "Cordula Stolberg" <[email protected]>
To: <[email protected]>
Sent: Tuesday, July 13, 2004 3:39 PM
Subject: Re: st: RE: Instrumental variables


> Julie,
>
> Thanks for that. It's good to hear that I gave you the opportunity to
> contribute to the list...:)
>
> You wouldn't by any chance know another solution to my instrumental
> variables problem? The thing is that there is no obvious instrument in my
> case.
>
> Thanks again,
> Cordula
>
> --On 13 July 2004 09:27 -0500 "Trivitt, Julie" <[email protected]>
> wrote:
>
> > I'm so excited, I finally get to contribute to the statalist after being
> > an observer for a few months.
> >
> > Cordula,
> > I don't think your plan will work.  The instrument you suggested (c1x1)
> > will be perfectly collinear with b1X1 you have in the original equation.
> > You will need to find an instrument for X2 that is not in the original
> > equation.
> >
> > Julie Trivitt
> > Economics PhD Student
> > Walton College of Business
> > University of Arkansas
> >
> >
> >
> > -----Original Message-----
> > From: Cordula Stolberg [mailto:[email protected]]
> > Sent: Tuesday, July 13, 2004 9:15 AM
> > To: [email protected]
> > Subject: st: Instrumental variables
> >
> > Dear Statalisters,
> >
> > Following on from my earlier problems with endogeneity and panel data, I
> >
> > now have a question with regard to the choice of the instrumental
> > variable.
> > I have the regression (panel data, fixed effects)
> >
> > Y b1X1 b2X2 b3dummies
> >
> > and I suspect that X2 might be endogenous. Is it possible to obtain an
> > instrument by first estimating
> >
> > X2 c1X1
> >
> > (where X1 is the same variable as the X1 in the first regression) and
> > then
> > use the predicted values of X2 as the instrument for the first
> > regression?
> >
> > Thanks for any suggestions,
> > Cordula
> >
> >
> >
> > *
> > *   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/
>
>
>
>
> *
> *   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/



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