Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am using a fixed effect model for my panel data (9 years, 1000+ obs), since my Hausman test indicates a value $(Pr>\chi^2)<0.05$. When I add dummy variables for industries that my firms included, they always get omitted. I know there is a big difference when it comes to the DV (disclosure index) among the different industry groups. But I am not able to get them in my model when using Stata.

Any suggestions how to solve this? And why are they omitted?

share|improve this question
*omitted because of collinearity – BEF May 6 '11 at 17:44
The error generated by Stata means some of your independent variables are perfectly collinear. The likely culprit is within the dummy variables. Either you forgot to exclude at least one of the dummies, or some combination of the other independent variables is perfectly collinear (or one of the dummy variables has no variation). @chl , Stata will automatically drop a variable when perfect collinearity occurs in a regression model (I'm fairly confident this is the error message the OP is talking about). – Andy W May 6 '11 at 20:26
Andy W is right. They would be omitted because of collinearity. Just drop one of the dummies or use noconstant (xtreg will do this for you) – Keith May 20 '11 at 14:21

Fixed effect panel regression models involve subtracting group means from the regressors. This means that you can only include time-varying regressors in the model. Since firms usually belong to one industry the dummy variable for industry does not vary with time. Hence it is excluded from your model by Stata, since after subtracting the group mean from such variable you will get that it is equal to zero.

Note that Hausman test is a bit tricky, so you cannot solely base your model selection (fixed vs random effects) with it. Wooldridge explains it very nicely (in my opinion) in his book.

share|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.