[Limdep Nlogit List] Panel Sample Selection
Murat Genc
MGenc at business.otago.ac.nz
Fri Dec 19 21:20:18 EST 2008
Dear All,
I'm trying to estimate a random effects panel sample selection model. As I understand it, there are two ways of doing this. You can either use the set-up described in pages E30-42-43 in Vol 2 of the Econometric Modeling Guide or use the set-up for a random parameters model described on page E30-53 by letting only the constant term to have a random coefficient.
I have tried both for the model I have. The first set-up, rem, does not work but the second one does. That is, when I use the rem set-up, the results I get do not make sense in that the p-values of all the estimated coefficients in the probit equation are 1! However, when I use the rpm set-up, this problem disappears.
Can anyone provide some explanation as to why this happens?
The commands I execute are as follows, and I have included the output file as an attachment if you'd care to have a look.
namelist; origin=ONE,LNMIGRAN,ZERO_MIG,lnngdpnz,lnngdp,LNNWGDP,lnnpop
,LNDISTCA,NON_ENGL,AV_LNMIG,avlngdp,avlnpop,D1995$
/* Rem */
probit; lhs=EX_PROBI;rhs=origin;Hold$
select; lhs=LNEX_NZ_;rhs=origin;mle;par$
select; lhs=LNEX_NZ_;rhs=origin;rpm;pds=grpsize;fcn=rem,maxit=400;halton$
/* rem via rpm */
probit; lhs=EX_PROBI;rhs=origin;Hold;rpm;pds=grpsize;fcn=one(n)$
select; lhs=LNEX_NZ_;rhs=origin;rpm;pds=grpsize;fcn=one(n);par;maxit=400$
I have an unbalanced panel. ZERO_MIG, NON_ENGL, and D1995 are dummy variables, AV_LNMIG,avlngdp,avlnpop are group means of the corresponding variables, LNDISTCA and NON_ENGL are constant over time. The model suffers from not having an exclusion restriction.
<<rem via rpm email Output.lim>> <<rem Output email.lim>>
Thanks for your time.
Murat
Dr M Genç
Department of Economics
University of Otago
P O Box 56
Dunedin, New Zealand
_______________________________________
* E-Mail: mgenc at business.otago.ac.nz
* Phone: +64 3 479 8644
* Fax: +64 3 479 8174
* Web: <www.otago.ac.nz/economics>
__________________________________
More information about the Limdep
mailing list