Individual-invariant variables in FE

[Julia Raciniewska]

Hello,
I am sorry if this topic has come up before - however, I did searched for an answer and failed. My question is very straight forward:
If I am using fixed effects model is it okey to include individual-invariant, but time-variant variables?
I am estimating the following model:

xtreg y x1 x2 x3, fe vce(cluster clustervar)

where y is employment of certain group, x1 is % of people in that group that use computer, and I would like to include price of computer capital as x2 (which is time-variant, but individual-invariant).
If it is okey to do so, should I worry about potential multicollinearity?

Thank you.

[Carlo Lazzaro]

Julia:
if the predictor is time-vaying, you can include it.
As an aside you seem overly concerned about quasi-extrene multicollinearity: read the chapter devoted to this topic in http://www.hup.harvard.edu/catalog.p...ontent=reviews and relax!
Obviously, if there's perfect multicollinearity Stata omits the culprit (which has a an accomplice in another variable) and there's nothing you can do but change your set of predictors.

[Julia Raciniewska]

Thank you very much Carlo! I am self-studying statistics so any materials are very helpful.

[Carlo Lazzaro]

Me, too (and the mischievous comment might be "We notice it!"), as I was taught statistics in the pre-computer era.