Statalist


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

RE: st: RE: Hausman test for FE-RE (xtivreg and xtivreg2)


From   "Schaffer, Mark E" <[email protected]>
To   <[email protected]>
Subject   RE: st: RE: Hausman test for FE-RE (xtivreg and xtivreg2)
Date   Wed, 21 Nov 2007 10:28:40 -0000

Correction - my posting below should read

"the Sargan-Hansen statistic reported by xtoverid after xtivreg or xtreg
is, in fact, an FE vs RE test"

Sorry....

--Mark 

> -----Original Message-----
> From: [email protected] 
> [mailto:[email protected]] On Behalf Of 
> Schaffer, Mark E
> Sent: 21 November 2007 09:44
> To: [email protected]
> Subject: RE: st: RE: Hausman test for FE-RE (xtivreg and xtivreg2)
> 
> Damiano,
> 
> > -----Original Message-----
> > From: [email protected]
> > [mailto:[email protected]] On Behalf Of 
> > [email protected]
> > Sent: 21 November 2007 06:42
> > To: statalist
> > Subject: Re:st: RE: Hausman test for FE-RE (xtivreg and xtivreg2)
> > 
> > Dear Mark, thank you very much for your kind reply but I have to 
> > confess I am still uncertin when it comes to the actual operative 
> > commands. In particular, could you explain ma how a cluster-robust 
> > xtoverid of my xtivreg2 estimation represents an Hausman 
> test for F/R 
> > effect? What's the specific code to be used to make the comparison?
> > 
> > If I consider my previous equation,
> > 
> > xtivreg2 x1 x2 x3 (x4= x5 x6), robust first fe (which is 
> "robust" in 
> > the option...)
> > 
> > and, as you say, I call for a "xtoverid, robust" or a "xtoverid, 
> > cluster(id)", I get only a new estimation of the Sargan-Hansen 
> > statistic. I mean, nothing suggest me a FEvsRE Hausman test.
> 
> If you look at the online help for xtoverid, you'll find a 
> paragraph that explains why xtoverid after xtivreg or xtreg 
> is, in fact, an FE vs RE test.  The help file also has 
> illustrative examples.  All at your fingertips!
> 
> --Mark
> 
> Prof. Mark Schaffer
> Director, CERT
> Department of Economics
> School of Management & Languages
> Heriot-Watt University, Edinburgh EH14 4AS tel 
> +44-131-451-3494 / fax +44-131-451-3296
> email: [email protected]
> web: http://www.sml.hw.ac.uk/ecomes
> 
> 
> > Could you help me (hopefully, by an a short but 
> illustrative example) 
> > with the next steps? Many thanks again.
> > 
> > Damiano
> > 
> > 
> > ---------- Initial Header -----------
> > 
> > From      : [email protected]
> > To          : [email protected]
> > Cc          : 
> > Date      : Tue, 20 Nov 2007 18:57:37 -0000
> > Subject : st: RE: Hausman test for FE-RE (xtivreg and xtivreg2)
> > 
> > > 
> > > > -----Original Message-----
> > > > From: [email protected]
> > > > [mailto:[email protected]] On Behalf Of 
> > > > [email protected]
> > > > Sent: 20 November 2007 18:24
> > > > To: statalist
> > > > Subject: st: Hausman test for FE-RE (xtivreg and xtivreg2)
> > > > 
> > > > I am running (by Stata 9) a panel regression and I have used
> > > > xtivreg2 with "fe" option, in order to estimate an
> > equation, where
> > > > one variable is endogenous. The results seem good, the
> > p-values are
> > > > excellent and the Hansen J statistic is reassuring. 
> > However, I have
> > > > been said that to be sure to correctly use xtivreg2 I need to 
> > > > perform an Hausman test for FE vs RE first, using
> > xtivreg. Is this
> > > > objection correct?
> > > 
> > > Yes and no.  FE is always (under the null, anyway)
> > consistent, whereas
> > > RE may be more efficient but only if the extra orthogonality 
> > > conditions hold.  Since you are happy with your results
> > with FE, you
> > > could just stop there.  Morever, xtivreg2 (in its current
> > incarnation
> > > - I have an upgrade planned) will do only FE, and xtivreg
> > with RE in
> > > Stata 9 won't do, say, cluster-robust SEs (though that
> > might have been
> > > rectified in Stata 10).
> > > 
> > > But if you want to do an FE vs RE test after xtivreg2 (or
> > xtivreg, for
> > > that matter) with FE, it's actually very easy: just use xtoverid, 
> > > downloadable from SSC in the usual way.  It takes cluster
> > etc options
> > > so you can get a cluster-robust FE vs RE test, for example.
> > > 
> > > Cheers,
> > > Mark
> > > 
> > > Prof. Mark Schaffer
> > > Director, CERT
> > > Department of Economics
> > > School of Management & Languages
> > > Heriot-Watt University, Edinburgh EH14 4AS tel
> > +44-131-451-3494 / fax
> > > +44-131-451-3296
> > > email: [email protected]
> > > web: http://www.sml.hw.ac.uk/ecomes
> > > 
> > > > If
> > > > yes, I don't know the correct procedure to construct 
> this kind of 
> > > > test. To be clear, this was the equation I estimated:
> > > > 
> > > > xtivreg2 x1 x2 x3 (x4= x5 x6), robust first fe
> > > > 
> > > > how could I perform an Hausman test for FE-RE? 
> > > > An illustrative example would be of great help. Thank you
> > in advance.
> > > > 
> > > > D.
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > 
> > > > >  
> > > > > 
> > > > > Thanks,
> > > > > 
> > > > >  
> > > > > 
> > > > > Raoul
> > > > > 
> > > > >  
> > > > 
> > > > 
> > > > *
> > > > *   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/
> 

*
*   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