Re: [Gretl-users] Nonlinear panel Instrumental Variable Regression with GRETL
On Thu, 10 Jan 2013, Anutechia Asongu wrote:
Hi All,
Please I need help for the concern on the subject of this mail.
Cheers,
I must admit, this isn't particulalrly clear. Am I correct in thinking
that you need to estimate a model like (I have to use LaTeX, sorry)
\[
y_{i,t} = g(X_{i,t}, \theta) + \alpha_i + \epsilon_{i,t}
\]
where
\[
E( g(X_{i,t}, \theta) \cdot \epsilon_{i,t} ) \ne 0
\]
but you have instruments such that
\[
E(\epsilon_{i,t} | Z_{i,t}) \ne 0 ?
\]?
And if so, how do you want to treat the individual effect $\alpha_i$?
Fixed? Random?
-------------------------------------------------------
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
-------------------------------------------------------