EViews.com

Hi,

I need your guidance. Which test is appropriate to test the heteroskedasticity in the residuals of OLS on larger time series data set. Precisely, which is better option Breusch Pagan Godfrey or White test.

Stay Blessed
Regards,
Nadia

The White test is in many ways a pecial case of the BPG test. Both tests are appropriate, but the BPG allows more flexibility in modelling the nature of heteroscedasticity by explicitly specifying its functional form.

I am running balanced panel using fixed effects and would like to produce robust s.e.
So, is it true that the Breusch-pagan test results can tell me which White coefficient covariance method to use (i.e cross-section, period or diagonal)? or can i just choose among them on definition basis. By that i mean to choose the white period method because my data are cross-sectionally clustered and hence i assume correlation among periods residuals?

Thanks!