RE: st: fixed effects estimation with time invariant variables

["ALICE DOBSON" <alice_dobson@hotmail.com>]

I have two obseervations. First, xthtaylor is more suitable for endogenous 
time-invariant variables. Hence, it relaxes the assumption of orthogonality. 
It is a bigger gun that Jona may proably require. Hence, conducting the 
Hausman test first is a better option.
Second, I often face a similar problem like Jona. For instance, I reporduce 
a Hausman test (fixed vs. random effect) result. How do I interpret this 
result?


hausman fixed random

                ---- Coefficients ----
            |      (b)          (B)            (b-B)     
sqrt(diag(V_b-V_B))
            |     fixed        random       Difference          S.E.
-------------+----------------------------------------------------------------
       beta |    .0227671     .0407022       -.0179352         .023357
------------------------------------------------------------------------------
                          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(1) = (b-B)'[(V_b-V_B)^(-1)](b-B)
                         =        0.59
               Prob>chi2 =      0.4426





> From: Christopher Baum <baum@bc.edu>
> Reply-To: statalist@hsphsun2.harvard.edu
> To: statalist@hsphsun2.harvard.edu
> Subject: st: fixed effects estimation with time invariant variables
> Date: Tue, 26 Jul 2005 13:49:55 -0400
>
> Martha said .............................
>
>
>
> Chris replied: This is quite incorrect. The standard random effects model, 
> xtreg, re, does NOT involve anything beyond a random intercept. She is 
> correct is noting that to use RE there is a maintained assumption of 
> orthogonality between regressors and the random intercept for each unit. As 
> Mark S. said, xthtaylor is a way around the oft-violated orthogonality 
> assumption. Mark is right on in suggesting that  you should do the Hausman 
> test.




> You do not need Some Alternative Software nor GLLAMM to estimate a mixed 
> model in Stata. These models (which indeed combine random and fixed 
> effects) are available with command xtmixed.
>
> Kit
> *
> *   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/
_________________________________________________________________
Is your PC infected? Get a FREE online computer virus scan from McAfee� 
Security. http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963

*
*   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/