Michael,
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of
> Bonnal, Michael (CBA)
> Sent: Thursday, February 08, 2007 2:24 PM
> To: [email protected]
> Subject: st: Negative Hausman Test in STATA
>
> Dear all,
>
> When performing a Hausman test between FE and RE models, I
> have a negative chisquare value.
> STATA suggests to use the command suest, but I could not make
> it work for my panel.
> I used the option ROBUST for both FE and RE estimations, but
> the value is still negative.
> Would you have any suggestion?
The latest version of xtoverid will generate robust Hausman-type tests
of FE vs RE that are guaranteed to be nonnegative, and more besides.
