Re: [Gretl-users] Open... to Artur
On Sat, 2 Jan 2016, Riccardo (Jack) Lucchetti wrote:
> P.S. May I make use this opportunity and ask Riccardo a somewhat
silly
> question: With the file you sent me I failed to reproduce Table 17.8 in
> Greene, 7-th, case Logit F.E.(C)^b, using felogit It should be the same
> Chamberlain’s estimator.
Yes, I had noticed that too. However, differences are minor (once you
normalise income). The fact that the loglikelihood reported in the book is a
little lower makes me think that Greene must have used software (most likely,
written by himself --- he's an excellent coder) that uses a slightly more
lenient convergence criterion than we do.
Oh, while we're at it. I uploaded the Riphahn et al dataset to git. The
dataset is used in multiple instances in Greene's book but originally from
Regina T. Riphahn, Achim Wambach, and Andreas Million, "Incentive Effects
in the Demand for Health Care: A Bivariate Panel Count Data Estimation",
Journal of Applied Econometrics, Vol. 18, No. 4, 2003, pp. 387-405.
In the next release of gretl, you'll find it together with the other
sample datasets.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------