Stata panel data fixed effects regression model -xttest3- command

[Peter Sutton]

Hi all,

I am running panel data fixed effects regression model and get this value for the groupwise heteroskedasticity test:

Modified Wald test for groupwise heteroskedasticity
in fixed effect regression model

H0: sigma(i)^2 = sigma^2 for all i

chi2 (2147) = 2.0e+37
Prob>chi2 = 0.0000

This value is huge. is it correct?

[Carlo Lazzaro]

Peter:
it is impossible to reply positively without knowing the data.
Short answer: go -vce(clustered panelid)- or -robust- standard errors.

[Peter Sutton]

Hello Carlo,

Thanks for your reply. Previously, my instructor at university told me to go with the option -, fe cluster(firm)-. How does this compare with -vce(clustered panelid)-?

[Carlo Lazzaro]

Peter:
it is absolutely the same stuff if -firm- is your -panelid- (a generic name use to mention the cross-sectional dimension of a given panel dataset). Under -xtreg- the -robust- option would do the very same job.