st: Hausman test for endogeneity for HT panel estimator

["simone sasso" <sasso@merit.unu.edu>]

Hello,
I am working on a large panel dataset. I've run the Hausman test to compare random vs. fixed effect finding that random effect are not consistent. I would like to try using the Hausman and Taylor model, beliving that in my analysis some of the estimators are endogenous and some are exogenous. In order to run the Hasman test between FE and HT estimators, do I first have to arbitrarly assume which are the variable that are endogeous or how can I see that?
Many thanks in advance for your help,
Simone