On Wed, Mar 15, 2023 at 4:11 AM Riccardo (Jack) Lucchetti <p002...@staff.univpm.it> wrote: > > I just tried running tsls with the --liml option on a just-identified > equation. As is well-known, this should return the same results as > ordinary 2sls. For some reason, I'm getting an error: I suspect this is > linked to the fact that there's no constant term in the model. I don't > have time to debug this myself today, but maybe somebody else can. > > Here's s simple script exposing the problem: > > <hansl> > nulldata 100 > series w = uniform() > series x = normal() > series y = normal() > > tsls y const x ; const w > tsls y const x ; const w --liml # this works > tsls y x ; w > tsls y x ; w --liml # this doesn't > </hansl>
That's now fixed in git. We just had to recognize the case where an ancillary regression had no regressors, and therefore we should transcribe the dependent variable as the residual instead of calling OLS. Allin _______________________________________________ Gretl-devel mailing list -- gretl-devel@gretlml.univpm.it To unsubscribe send an email to gretl-devel-le...@gretlml.univpm.it Website: https://gretlml.univpm.it/postorius/lists/gretl-devel.gretlml.univpm.it/
