[Gretl-users] Re: [Sys-GMM Dynamic panel data comparison]
On Wed, 20 Jan 2021, JOSE FRANCISCO PERLES RIBES wrote:
Dear Sven:
Please, find attached the .xlsx dataset for replication purposes.
As far I know, there are not missing values in this dataset.
Many thanks again.
I tried this with gretl vs stata's xtabond2. (By way of data check,
I got the same results as you report from gretl.)
xtabond2 found 45 instruments and produced parameter estimates that
differed widely from both gretl and the results you report from
xtdpdsys. I also tried with Ox/DPD and got the error message "not
enough observations in sample".
In the file replication.xlsx, all series except Afiliados and Dens
(and of course Year) are non-time varying.
Bottom line: seems like this is a pretty dodgy specification.
Allin