Hausman test

[Joachim Vanderheyden]

I want to investigate if I have to use a random effects or fixed effects model. Therefore I run the Hausman test.
But the results are very strange.
What could be the reasons for this outcome?

---- Coefficients ----
| (b) (B) (b-B) sqrt(diag(V_b-V_B))
| fixed random Difference S.E.
-------------+----------------------------------------------------------------
schuldgraad | -2.33e-12 -5.91e-06 5.91e-06 .
logomzet | -3.35e-08 .1690572 -.1690573 .
logar~siteit | -5.74e-08 -.13566 .1356599 .
logaantalw~s | -1 -.9164953 -.0835046 .
logschulden | -3.50e-08 -.0063579 .0063579 .
logloonkost | 7.12e-09 .7348673 -.7348673 .
------------------------------------------------------------------------------
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(6) = (b-B)'[(V_b-V_B)^(-1)](b-B)
=-40821.70 chi2<0 ==> model fitted on these
data fails to meet the asymptotic
assumptions of the Hausman test;
see suest for a generalized test

[Carlo Lazzaro]

Joachim:
you may want to consider two community-contributed commands, that you can spot and install via:
- search xtoverid-;
-search RHAUSMAN-.

[Joachim Vanderheyden]

I have installed both, but there is nothing changed.

[Carlo Lazzaro]

Joachim:
what if you run -hausman- with the -sigmamore- option?

[Joachim Vanderheyden]

How do I have to run that option?

I am using the following codes:
xtreg logarbeidsproductiviteit schuldgraad logomzet logarbeidsintensiteit logaantalwerknemers logschulden lognettotoegevoegdewaarde logloonkost, re
estimates store random
xtreg logarbeidsproductiviteit schuldgraad logomzet logarbeidsintensiteit logaantalwerknemers logschulden lognettotoegevoegdewaarde logloonkost, fe
estimates store fixed
hausman fixed random

[Carlo Lazzaro]

Joachim:

hausman fixed random, sigmamore

Provided that I'm always pleased to reply if the subject is in line with my skills/knowledge and I'm not under time constraints, please note that you can save yourself tons of time by taking a look at the the help files that come with both official and community-contributed Stata command:

*Just an excerpt from -help hausman-*
Test the appropriateness of the random-effects estimator (xtreg, re)
. hausman fixed ., sigmamore

[Joachim Vanderheyden]

Thank you very much.