test of multicollinarity in multiple linear regression.

[Dennis Hansen]

Hi guys

I have a multiple linear regression looking like this : Y = Var1 (dummy) + Var 2 + Var 3 + Var 4 + Var 5 (fixed effect dummies for years)

I would like to test for multicollinarity as i am checking if the underlying assumptions are satisfied.

To do this i know two possible ways: do a correlation matrix and check if any pairwise correlations are too high.. But as far i understand, this can not be done for the dummy variable, am i correct with this?
So instead of doing the correlation matrix i used Stata to calculate ViF scores. I know ViF scores above 10 should be investigated, but i am wondering if this also is the case for the fixed effect dummies? For some of the fixed effect dummies representing a year i get a ViF score above 10. Should this be a concern or should i only look at the ViF scores for the explanatory variables?

Sorry for the long question: Short version is if ViF scores for fixed effect dummies should be included in the analysis of multicollinarity, or if i should only be looking at ViF scores for the explanatory variables?

Thanks a lot in advance!

[Carlo Lazzaro]

Dennis:
as some of the most authoritative members on this forum would probably recommend you to do, you can quote in your resarch report http://www.hup.harvard.edu/catalog.p...40&content=toc (Chapter 23) and stop bothering yourself with multicollinearity altogether.