Hi everybody. It turned out that the Dawn's problem -- failure to get a positive Hausman test
statistic for endogeneity even when using the -sigmamore- option -- resulted from the fact that she
was estimating by IV with the -robust- option.
The -sigmamore- option of the -hausman- command is incompatible with heteroskdastic-robust
estimation because, basically, there is no single "sigma" when errors are heteroskedastic.
-hausman- without -sigmamore- will be asymptotically valid in conjunction with the robust
covariance estimator, but then you can encounter the negative test statistic problem.
The easiest way to guarantee a positive Hausman test statistic for endogeneity is to estimate the
model with ivreg2 (downloadable from ssc-ideas). Estimate the efficient model using ivreg2, and
put the variables to test for endogeneity in the -orthog- option. This will guarantee a positive
test statistic, even with robust covariance estimators. (It's an application of a result described
in Hayashi (2002), Econometrics, p. 220. Or see our (Baum-Schaffer-Stillman) working paper,
http://fmwww.bc.edu/ec-p/WP545.pdf.)
--Mark
> > > > Date sent: Wed, 6 Nov 2002 13:14:58 -0500 (EST)
> > > > From: Dawn M Brancati <[email protected]>
> > > > To: [email protected]
> > > > Copies to: [email protected]
> > > > Subject: Re: st: Re: hausman test
> > > >
> > > > >
> > > > > Thanks for all the advice. I am working on it now. The sigmamore
> > > > > option is under ivreg. I had already used it, however, and still
> > > > > received the error message that I mentioned.
> > > > >
> > > > > Dawn
> > > > >
> > > > >
> > > > > On Wed, 6 Nov 2002, John A Karikari wrote:
> > > > >
> > > > > > Could I have more information on the "sigmamore" option. I couldn't
> > > > > > find it using the search process. Thanks.
> > > > > >
> > > > > > >>> [email protected] 11/06/02 12:06PM >>>
> > > > > > Dawn,
> > > > > >
> > > > > > If you're using -hausman- after ivreg, you won't get the "failure to
> > > > > > meet the asymptotic assumptions" problem if you use the -sigmamore-
> > > > > > option. This guarantees that the matrix difference in the middle of
> > > > > > the standard Hausman test statistic is positive definite.
> > > > > >
> > > > > > Even simpler, just to install -ivendog- from ssc-ideas. This will
> > > > > > give you both the Wu-Hausman artificial regression F stat, and the
> > > > > > (Durbin-Wu-)Hausman chi-sq statistic, after estimation using
> > > > > > -ivreg-.
> > > > > >
> > > > > > Hope this helps.
> > > > > >
> > > > > > --Mark
> > > > > >
> > > > > > Date sent: Wed, 06 Nov 2002 11:53:37 -0500
> > > > > > From: "WILLIAM EVEN" <[email protected]>
> > > > > > To: <[email protected]>
> > > > > > Subject: st: Re: hausman test
> > > > > > Send reply to: [email protected]
> > > > > >
> > > > > > > An alternative approach to the Hausman test is to include the
> > > > > > residual
> > > > > > > from the reduced form equation into the primary equation and
> > > > > > > perform
> > > > > > a
> > > > > > > t-test on its significance. This avoids any problems with the
> > > > > > > differences in the variance covariance matrices being non-positive
> > > > > > > definite that one can encounter in the original formulation of the
> > > > > > > test.
> > > > > > >
> > > > > > >
> > > > > > >
> > > > > > > Bill Even, Professor
> > > > > > > Department of Economics
> > > > > > > Miami University
> > > > > > > Oxford, OH 45056
> > > > > > >
> > > > > > > Phone: 513-529-2865
> > > > > > > Fax: 513-529-6992
> > > > > > > home page: www.sba.muohio.edu/evenwe
> > > > > > > e-mail: [email protected]
> > > > > > >
> > > > > > > >>> [email protected] 11/06/02 10:29AM >>>
> > > > > > > Hi,
> > > > > > >
> > > > > > > I'd like to know if anyone has any suggestions on how to proceed
> > > > > > when
> > > > > > > you cannot perform the Hausman specification test because your
> > > > > > > data does not meet the asymptotic assumptions of the test. The
> > > > > > > augmented regression test is also based on the same assumptions of
> > > > > > > the Hausman and so is of no help in this regard. I am using the
> > > > > > > Hausman test as
> > > > > > a
> > > > > > > pre-test for two-way causation but cannot perform it in some
> > > > > > > instances. Is any one familiar with any alternatives to theses
> > > > > > > tests for @SLS? If not, is the best approach to report results of
> > > > > > > both the OLS models and the 2SLS models?
> > > > > > >
> > > > > > > Thanks.
> > > > > > > DB
>
> *************************************
> Columbia University
> Political Science Department
> 420 W. 118th Street, 7th Flr.
> New York, NY 10027
> E-Mail: [email protected]
>
>
Prof. Mark Schaffer
Director, CERT
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS
tel +44-131-451-3494 / fax +44-131-451-3008
email: [email protected]
web: http://www.sml.hw.ac.uk/ecomes
________________________________________________________________
DISCLAIMER:
This e-mail and any files transmitted with it are confidential
and intended solely for the use of the individual or entity to
whom it is addressed. If you are not the intended recipient
you are prohibited from using any of the information contained
in this e-mail. In such a case, please destroy all copies in
your possession and notify the sender by reply e-mail. Heriot
Watt University does not accept liability or responsibility
for changes made to this e-mail after it was sent, or for
viruses transmitted through this e-mail. Opinions, comments,
conclusions and other information in this e-mail that do not
relate to the official business of Heriot Watt University are
not endorsed by it.
________________________________________________________________
*
* 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/