[Gretl-devel] Re: ADF-GLS test question

Peter Sephton: "It appears GRETL doesn’t do the search for the optimal lag length over a common period." Sven: "If this is true, then I think it's a problem." Allin: "Right now I'm trying to determine if Peter's suspicion on that point is well-founded or not." Answer: Not. We ensure a uniform sample, basing it on the maximum lag length. In git I've enabled a numerical parameter to the --verbose option for "adf", such that if its value is greater than 1 we print the auxiliary regressions at each lag order, which makes this clear. Example: <hansl> nulldata 120 setobs 12 1990:01 --time-series series y = normal() adf 8 y --c --gls --test-down=AIC --verbose=2 </hansl> Allin