Testing Multicollinearity for panel data

[Sara Harsha]

I am unable to run estat VIF command after running xtreg command. Can somebody guide how can I calculate VIF for my panel data?

[Carlo Lazzaro]

Sara:
welcome to the list.
You may to consider -estat vce, corr-.

[Sara Harsha]

So many thanks! Can you please also tell me the command to take this output to Excel. Would be very grateful!

[Carlo Lazzaro]

Sara:
although I'm not familiar with,-search outreg-.

[Sara Harsha]

OK! Thanks again!

[Sara Harsha]

Sir, I need to check 'vif' of the overall model apart from the correlation matrix. Can you please tell me the command for checking 'vif' of panel data? Would be very grateful.

[Carlo Lazzaro]

Sara:
this Stata thread might be helpful: http://www.stata.com/statalist/archi.../msg00018.html

[Sara Harsha]

Many thanks, Sir!

[Jeanne Roche]

Dear all,

I am encountering the same issue - I can't run

estat vif

after running a regression on my panel data using

xtreg

.
I used

estat vce, corr

which then gave me the "Correlation matrix of coefficients of xtreg model" - so it's a correlation matrix, but I still don't have information on the vif values.

Could you please help me?

I am using Stata 16 for Mac.

[Carlo Lazzaro]

Jeanne:
you cannot run -estat vif- after -xtreg- because -estst vif- is not a postestimation command supporteed by -xtreg-.
In fact, -estat vce, corr- gives you clue about multicollinearity issues.
In addition:
1) this Stata thread might be helpful: http://www.stata.com/statalist/archi.../msg00018.html;
2) (probably more substantive): quasi-extreme multicolliearity is often oversold. Please see, in this respect, the humorous and many times quoted on this forum Chapter 23 of https://www.hup.harvard.edu/catalog....9780674175440;
3) keep also in due consideration the wise comment often made on this issue by Clyde Schechter: quasi-extreme multicollinearity is an issue as long as it produces weird standard errors of your coefficients.
I would also add that you may suspect quasi-extreme multicollinearity when your R-sq is staistically significant whereas all your coefficients are not.
As an aside, please share what you typed and what Stata gave you back (as per FAQ).
Actually, interested listers do not know if you ran -xtreg,fe-; -xtreg,re-; -xtreg,be- or -xtreg,pa- (very different beasts indeed!)

[Eric de Souza]

Carlo, your link to Chapter 23 (of Godlberger, I presume) doesn't work, at least for me.
But a substittue is the post by Dave Giles (Econometrics Beat: Dave Giles' Blog: The Second-Longest Word in the Econometrics Dictionary) and also by hime(Econometrics Beat: Dave Giles' Blog: Micronumerosity )

[Carlo Lazzaro]

Eric:
thanks for pointing this out.
I give it one more shot: https://www.hup.harvard.edu/catalog....=9780674175440