Why wouldn't a standard Hausman test work here?
On Thu, Sep 24, 2009 at 4:46 AM, Miriam Wüst <[email protected]> wrote:
> Hi
>
> I am trying to implement a Hausman test to test my FE model against a model allowing for more unrestricted covariances. I estimate my FE model with xtreg and estimate my alternative (unrestricted) model with the nl feature (I match sample and theoretical covariances by setting up a number of restrictions on the theoretical covariance matrix). The resulting model has more parameters than the FE model, which assumes some of the covariances to be 0 (restricted). Do I simply add those restricted parameters to the FE parameter vector by adding 0's? Am I making any sense to you - please help :)
>
> Miriam
>
>
>
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/