EViews.com

Hi,

I was wondering is there any way to know it is better to use t or GED for the error in ARCH/GARCH model estimation?

It seems I can't do empirical distribution test for t or GED for my data.

Thanks in advance!

You should be able to decide that based on the usual model diagnostics (e.g. AIC). As for the post-estimation empirical distribution test, try saving the structural residuals of each model and then fit any distribution you like. EViews has a very short list of distribution palette, so you may have to build your own ML estimator for that purpose.