The RATS Software Forum

Dear Tom,

I'm currently evaluating forecasts for a panel of countries obtained from two different models. I would like to compare both models, is there any version of the Diebold-Mariano Test which works in a panel?

Many thanks.

Regards

So far as I know, there's nothing in the D-M test that breaks if applied to a panel. You just have to figure out how to compute the variance of the averaged difference of the loss function (which will depend quite a bit on the assumptions you make about the connections between individuals).