Fixed Effects and time-invariant variables

[Max Bolzen]

Hello,

I'm a beginner with Stata and I have to analyse paneldata. Several Financial Institutions in different countries from 2005 to 2013.
I want to estimate how several indicators influence the perfomance of these institutions.

Each institutions has its own "id" and my time variable is "year" -> xtset id year

The Hausman-Test tells me to use fixed effects, but the problem is I have time-invariant variables.

So my question is, what possibilities do I have to run the regression keeping the time-invariant variables?

Thanks in advance,
Max

[Clyde Schechter]

Within the confines of fixed-effects regression there is no possibility of estimating the effects of time-invariant invariables: they are collinear with the fixed effects and if you try to include them Stata (or any other statistical software) will simply drop them anyway.

That said, if you are specifically interested in those effects, you might want do some side analyses that avoid the fixed effects constraints. Take a look at -xtreg, be-. which estimates the effects of those variable across institutions.

[Max Bolzen]

No, i'm not specifically interested in fixed effects. I'm happy about any suggestions how to deal with the regression of my panel data

would
.areg y x1 x2 x3, absorb(year) vce(robust)
be a possibility for example?

[Sebastian Kripfganz]

This has been a topic here at Statalist for several times. See for example:
1) Time-invariant variables and negative value in Hausman specification test
2) Appropriate econometric technique for estimating invariant regressors in panel data analysis
3) Marginsplot with panel data
4) xtreg re (and xtreg fe) with many different dummy types (i.firm, i.industry, i.month, i.year)
5) xthtaylor error &-xtreg, fe- doesn't work due to important time-invariant variable
6) In which cases should one prefer -xtreg, be- over -xtreg, re-
7) why xtabond and xtdpdsys produce the same results (both the coefficient and standard error) for certain variables?
8) Time-invariant variables in Fixed-effects model
9) Mundlak test interpretation
10) xtreg fe - Drop or retain groups with no variation on focal predictor?