[Limdep Nlogit List] Hausman test when time-constant variables are important

[Pavlos C. Symeou]

Dear Limdep-listers,

I have a question regarding the Hausman specification test for OLS 
models. I want to include in my OLS model 7 time-constant variables 
which depict the industry classification of a firm along with other 
explanatory variables.  I have balanced panel data on 100 firms for 5 
years. The Hausman test requires the comparison of comparable variables 
between the FE and RE models. However, since my time-constant variables 
are dropped from the FE model, this implies that the Hausman test 
compares two models with not comparable variables. The correct mode of 
use of the Hausman test is to exclude from both models the time-constant 
variables. The resulting test statistic gives support to the FE model. 
Yet, inclusion of the time-constant variables in the OLS model is 
important for its correct specification so I interact these variables 
with year dummies and include the interaction variables (and exclude the 
original time-constant variables) in both FE and RE models. Now the 
Hausman test statistic supports the RE model.

I would appreciate if you could advise me on model selection. Should I 
arbitrarily choose the RE model that can involve the time-constant 
variables (in their original form) in its estimation? Otherwise? I use, 
Limdep 9.

Yours,

Pavlos