Am 20.01.2021 um 19:40 schrieb José Perles:
Dear Sven
Many thanks for your fast response.
What I would say is that Coast Aerop70 Walk Pmain and DPrs11_2 (the last being
residential in Stata) are time invariant.
As said Dprs11_2 and residential are the same variables.
OK. Well, I don't really have an answer right now, perhaps jack can
comment and/or you could share the dataset. Notice, however, that Stata
didn't drop all those variables altogether, but only dropped them for
the diff-instrument list. But of course I agree it's worthy of
investigation why Aerop70 was omitted in Stata but not in dpanel/gretl.
Gretl's guide chapter about dpanel talks about replication being
sometimes difficult, for example because of different handling of
missing values.
cheers
sven