just a presentational issue: when you select the --test-down option for
an ADF test, and it turns out that no lagged differences are retained,
gretl prints out a non-augmented DF test. I have found this confusing
because I was looking for the message how many lags were used in the
equation. Also, arguably allowing for lags which turn out insignificant
is something different from the plain DF test which doesn't allow lags.
So my suggestion would be to use the printout from the ADF test and say
"using 0 lags of (1-L)...".
Show replies by thread