|
|  |
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: Negative Hausman value?
Pavlos,
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of
> Pavlos C. Symeou
> Sent: 16 October 2006 11:58
> To: [email protected]
> Subject: st: Negative Hausman value?
>
> Dear Statalisters,
>
> I attempt to compare between FE and RE transformations in
> Stata using the hausman command, yet the chi^2 for the test
> is negative (how can a square be negative?).
This can happen in finite samples, unless the same estimate of the error
variance is used throughout the H statistic - see, e.g., Baum et al.
(2003) in the Stata Journal or the Boston College working paper version.
This can generally be handled in Stata with the -sigmamore- or
-sigmaless- options to the -hausman- command. What I am not sure of is
whether -xtreg- stores the right sigma internally in the right place
where -hausman- can make use of it. I have a vague memory that in
earlier version of Stata, -xtreg- did not store sigma in a way that was
compatible with this option, but maybe that's been fixed. You need to
check.
> I provide below
> the commands that I use to estimate the models, store the
> model's estimates and compare them with hausman. Stata
> produces a note with the output which states that the
> differenced variance matrix does not equal the number of
> coefficients being tested.
This can happen when the RE estimator is able to estimate the
coefficient for a time-invariant variable that the FE sweeps out with
the within transformation. In that case, the right number of degrees of
freedom is the smaller number (= number of coeffs of FE estimator not
including the constant). You can force this with the -df- option of
-hausman-.
Hope this helps.
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
> I have noticed that in the output
> provided below and used in the calculations of the hausman
> test, the yearly dummy "_Iyear_2004" is not included. Please
> bear in mind that this yearly dummy is dropped in the initial
> FE and RE model estimations because of collinearity.
>
> I hope this helps previous threads to understand why we
> encounter negative values in the hausman test.
>
> Your feedback would be appreciated.
>
>
>
> xi: xtreg f m lagm m2 lib_fixed lib_mobile t d1-d24 i.year,
> fe est store fe
> xi: xtreg f m lagm m2 lib_fixed lib_mobile t d1-d24 i.year,
> re hausman fe .
>
>
> Note: the rank of the differenced variance matrix (52) does
> not equal the number of coefficients being tested (53); be
> sure this is what you expect, or there may be problems
> computing the test. Examine the output of your estimators
> for anything unexpected and possibly consider scaling your
> variables so that the coefficients are on a similar scale.
>
> ---- Coefficients ----
> (b) (B) (b-B)
> sqrt(diag(V_b-V_B))
> fe . Difference
> S.E.
>
> m 4.126995 4.210606 -.0836103
> .
> lagm -3.659119 -3.719283 .060164
> .
> m2 -.0035987 -.0037695
> .0001707 .
> lib_fixed -.7893976 -.6248966 -.164501
> .
> lib_mobile .8347247 . 8552306
> -.020506 .
> t .1057887 .0839587 .02183
> .
> d1 1.591341 1.425241
> .1661004 .
> d2 1.73248 1.566447 .1660332
> .
> d3 1.33496 1.175148 .1598117
> .
> d4 1.163786 1.009243
> .1545424 .
> d5 1.130861 .9795363
> .1513244 .
> d6 1.53821 1.372196 .1660132
> .
> d7 1.787398 1.618786
> .1686119 .
> d8 2.030746 1.871618
> .1591281 .
> d9 2.329689 2.175233
> .1544567 .
> d10 2.474566 2.379384 .0951821
> .
> d11 2.530941 2.391151 .1397903
> .
> d12 2.702182 2.548153 .154029
> .
> d13 3.072654 2.923146 .1495087
> .
> d14 3.200262 3.048934 .1513281
> .
> d15 3.434791 3.299563 .1352286
> .
> d16 4.161894 4.040091 .121803
> .
> d17 4.693086 4.604569 .0885171
> .
> d18 4.813464 4.765931 .0475332
> .
> d19 4.874817 4.829945 .0448717
> .
> d20 4.763858 4.696174 .0676834
> .
> d21 4.007245 3.914138 .0931068
> .
> d22 3.38889 3.261472 .1274171
> .
> d23 2.073858 1.891516 .1823422
> .
> d24 2.133198 1.90412 .2290783
> .
> _Iyear_1981 .1057826 .1715391 -.0657565
> .
> _Iyear_1982 .429569 .5114474 -.0818784
> .
> _Iyear_1983 .779513 .8763522 -.0968392
> .
> _Iyear_1984 1.270724 1.382196 -.1114715
> .
> _Iyear_1985 1.598364 1.731791 -.1334276
> .
> _Iyear_1986 1.724753 1.88993 -.1651769
> .
> _Iyear_1987 2.117019 2.295973 -.1789541
> .
> _Iyear_1988 2.463715 2.660981 -.1972662
> .
> _Iyear_1989 2.808268 3.025018 -.21675 .
> _Iyear_1990 3.237237 3.452751 -.2155147
> .
> _Iyear_1991 3.653473 3.891927 -.2384541
> .
> _Iyear_1992 4.006215 4.272213 -.2659985
> .
> _Iyear_1993 4.307211 4.585079 -.2778678
> .
> _Iyear_1994 4.561733 4.852067 -.2903343
> .
> _Iyear_1995 4.596097 4.878367 -.2822704
> .
> _Iyear_1996 4.680343 4.956875 -.2765316
> .
> _Iyear_1997 4.754712 4.997591 -.242879
> .
> _Iyear_1998 4.427386 4.6229 -.195514 .
> _Iyear_1999 3.808497 3.95841 -.1499126
> .
> _Iyear_2000 3.243333 3.354241 -.110908
> .
> _Iyear_2001 2.610181 2.692871 -.0826899
> .
> _Iyear_2002 1.963031 2.033931 -.0708998
> .
> _Iyear_2003 1.108533 1.15696 -.0484275
> .
>
> b = consistent under Ho and Ha; obtained from xtreg B =
> inconsistent under Ha, efficient under Ho; obtained from xtreg
>
> Test: Ho: difference in coefficients not systematic
>
> chi2(52) = (b-B)'[(V_b-V_B)^(-1)](b-B) = -30.11
> chi2<0 ==> model fitted on these data fails to meet the
> asymptotic assumptions of the Hausman test; see suest for a
> generalized test
>
>
> *
> * 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/
