Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: xthtaylor without endogenous time-invariant variables?


From   "Aurora MORDONU" <AMORDONU@cris.unu.edu>
To   <statalist@hsphsun2.harvard.edu>, <JSalvati@imf.org>
Subject   RE: st: xthtaylor without endogenous time-invariant variables?
Date   Fri, 20 Jan 2006 10:49:13 +0100

Dear Jean

I checked again the options of xthtaylor and I did not manage to use it
without specifying endogenous time-invariant variables. 

Do you remember by chance what small modification you made to allow for
that?


I was also wondering whether applying a hausman test to decide upon a FE
or a HT estimator is the best way to do?

Thanks a lot,

Aurora
-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Salvati, Jean
Sent: 18 January 2006 22:12
To: statalist@hsphsun2.harvard.edu
Subject: RE: st: xthtaylor without endogenous time-invariant variables?

In my view, the HT estimator is useful when you have time-invariant
regressors and when *any* of the regressor (not necessarily one of the
time-invariant regressorss) is correlated with the individual effect.
Any correlation between any of the regressors and the individual effect
will make the RE estimator inconsistent.

I agree with Aurora that xthtaylor should not force you to specify
time-invariant regressors correlated with the individual effect.

I actually looked at the code for xthtaylor a few weeks ago. Only a
small modification was required to remove the restriction in question,
and it did not seem to have any adverse consequence on the functioning
of the program. But I did not spend much time on that, and I may have
missed something.

Jean Salvati 

> As far as I understand xttaylor helps in estimating a model 
> in which some of the time-invariant variables are correlated 
> with the unobserved time invariant effect, but you don't want 
> to run a FE model because this would get rid of such observed 
> fixed variables.
> 
> If you don't have any time-invariant variable that is 
> correlated with the unobserved error term then why not just 
> using standard Random Effects?
> 
> robert
> 
> On 1/18/06, Aurora MORDONU <AMORDONU@cris.unu.edu> wrote:
> > Hello,
> >
> >
> >
> > There were already two questions unanswered on the topic. I hope 
> > somebody managed to find the answer meanwhile...
> >
> >
> >
> > How can we use xthtaylor without endogenous time-invariant 
> variables? 
> > If no endogenous constant variable is specified, one gets 
> an error message.
> >
> >
> >
> > Thank you.
> >
> > Best regards,
> >
> >
> >
> > Aurora Mordonu
> >
> >
> > United Nations University-Comparative Regional Integration 
> Studies C/o 
> > Groetseminarie Poterrierei 72 8000 Brugge Belgium
> >
> > *
> > *   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 1996–2021 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index