EViews.com

Hi there,

I have a couple of lagged dependent variables in my regular (not dynamic) panel data. I'm not suppose to use DW test; does anyone know how to conduct the Breusch-Godfrey on Eviews 7.1? Any suggestion on how to handle are most welcome.

Thanks in advance!

You haven't got any answer since Breusch-Godfrey does not work for panel data.

The problem with autocorrelation and panel data is that you can have autocorrelation both between and within the panel. This makes it really hard to perform any tests. To confuse you more, it also depends whether you have short or long panels.
My advice is to first estimate a model with robust st.errors, and then re-estimate the model without using robust st.errors. If there are big differences in the computed st.errors, then you might have problems with autocorrelation. I know, it is not the best answer in the world, but it gives you (hopefully) some guidelines.