st: Alternative to Hausman test for FE vs RE models

[Udaya Wagle <udaya.wagle@wmich.edu>]

Hi,

 

I am trying to figure out whether FE or RE is more appropriate for my model that I am estimating with panel data. Obviously, the regular ‘hausman test’ reports, “V_b-V_B is not positive definite.” Even the ‘sigmaless’ and ‘sigmamore’ options that are supposed to minimize the likelihood of getting non-positive differences have failed to resolve the problem. I also figured that the ‘suest’ technique is not available when ‘xtreg’ is used. I saw postings made sometimes ago (http://www.stata.com/statalist/archive/2005-08/msg00807.html) that suggested using an augmented regression test that avoids problems such as not having positive definite difference of covariances. Due to my unfamiliarity with working with macro, however, I could not make good use of it. I would appreciate help with operational suggestions particularly pertaining to Stata 9.

 

Thanks in advance.

Udaya Wagle