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

Re: st: Hausman test


From   "Mark Schaffer" <m.e.schaffer@hw.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Hausman test
Date   Sun, 08 Aug 2004 22:49:55 +0100

Andrea,

From:           	"Andrea Molinari" <A.Molinari@sussex.ac.uk>
To:             	<statalist@hsphsun2.harvard.edu>
Subject:        	Re: st: Hausman test
Date sent:      	Thu, 5 Aug 2004 13:55:16 +0100
Send reply to:  	statalist@hsphsun2.harvard.edu

> Thanks! That did the trick...

Careful!  This is the right thing to do only if you were doing it 
backwards (=wrongly) to begin with.  Stata needs to know which is the 
efficient model and which is the inefficient-but-consistent model.  
Reversing the order is equivalent to telling it to switch them 
around, but if the order is backwards then you have a useless 
(=wrong) test statistic.

The root of the problem is that to guarantee a positive test 
statistic, you need to use the same estimate of the error variance 
throughout the calculation - either the random effects estimate or 
the fixed effects estimate.  Stata's -hausman- command lets you do 
this with the -sigmamore- or -sigmaless- options, but the last time I 
checked, xtreg doesn't leave the right information in the estimation 
results that would enable these options to work.  Without these 
options, you get a statistic that is still asymptotically correct, 
but prone to occasional misbehaviour in finite samples.

I don't know about getting suest to work with panels.  Your other 
option is to run the test using the artificial regression approach.  
This is guaranteed to generate a positive test statistic - I think 
Davidson and MacKinnon (1993) describe how to do it.

Hope this helps.

--Mark

> Andrea
> 
> ----- Original Message ----- 
> From: "David Greenberg" <dg4@nyu.edu>
> To: <statalist@hsphsun2.harvard.edu>
> Sent: Wednesday, August 04, 2004 11:59 PM
> Subject: Re: st: Hausman test
> 
> 
> > Try doing the estimation of the two models in the opposite order. David
> Greenberg, Sociology Dept., New York University
> >
> > ----- Original Message -----
> > From: Andrea Molinari <A.Molinari@sussex.ac.uk>
> > Date: Wednesday, August 4, 2004 6:22 am
> > Subject: st: Hausman test
> >
> > > Dear all,
> > > When estimating a Hausman test between fixed and random effects
> > > models for a
> > > subsample, I got a negative chi2 value (!) together with the following
> > > message:
> > >
> > > model fitted on these data fails to meet the asymptotic
> > > assumptions of the
> > > Hausman test; see suest for a generalized test
> > >
> > >
> > > but I could not implement suest for panel. Does anyone have a hint
> > > on how to
> > > do this?
> > >
> > >
> > >
> > > Thanks!
> > >
> > >
> > > Andrea
> > >
> > >
> > >
> > >
> > > *
> > > *   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/

Prof. Mark E. Schaffer
Director
Centre for Economic Reform and Transformation
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS  UK
44-131-451-3494 direct
44-131-451-3008 fax
44-131-451-3485 CERT administrator
http://www.som.hw.ac.uk/cert

*
*   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 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index