EViews.com

I would like to know whether the method suggested by wooldridge regarding testing serial correlation in panel data model can be applied to the case with a lagged dependent variable? I have run a panel data model with first differencing and according to Wooldridge, we could regress the e(t) on e(t-1) of the panel data model and test for whether the coefficient of e(t-1)=-0.5.

However, in his example, he did not include a lagged difference dependent variable, if there is a lagged differenced dependent variable on the RHS, would his test still be valid?

Also, does any one know how to compute m1 and m2 serial correlation test suggested by Arellano and Bond in Eviews?

Alice

I'm not sure of the answer, but I doubt it works with a lagged dependent variable. It wouldn't work in the non-panel context, because the residuals aren't consistent estimates of the error terms.