[Limdep Nlogit List] Error 215: Model too large
alessandro.secchi at upf.edu
alessandro.secchi at upf.edu
Tue Mar 11 03:16:00 EST 2008
Dear All,
I'm estimating a bivariate probit as the first stage of a more general
heckman estimation.
The set of regressors is the same in both equations of the bivariate probit.
It includes a series of regional dummies (103) plus some other variables for
a total of 122 regressors.
When I try to estimate the bivariate probit with LIMDEP 9.0 for
windows I get the following message:
Error 215: Model too large.K>150,>120 for SELE,>100 for 2SLS or MINI
My feeling is that the error is not related with memory space
(at the moment it is set on 80,000,000 cells, i.e. around 600M).
Is there any way to fix this problem ?
Note that I've also performed an experiment by assuming independence
between the two equations of the probit (see below). However also in
this case I got the Error 215 message after execturing the
bivariateprobit command.
PROBIT;Lhs=Y1;Rhs=X1,X2,..,X122$
matrix; ba = b$
PROBIT;Lhs=Y2;Rhs=X1,X2,..,X122$
matrix; bb = b$
BIVARIATEPROBIT;Lhs=Y1,Y2;
Rh1=X1,...,X122;Rh2==X1,...,X122;Sel;Start=ba,bb,0; Load$
Any help would be appreciated.
Best,
Alessandro Secchi
More information about the Limdep
mailing list