AW: [Limdep Nlogit List] simultaneous equations model with 3 endogeneous variables

Wilde, Joachim Joachim.Wilde at
Fri Apr 7 17:15:53 EST 2006

PD Dr. Joachim Wilde
Institut für Wirtschaftsforschung Halle
Kleine Märkerstr. 8
06108 Halle
Phone: ++49 345 7753 836
E-mail: Joachim.Wilde at

Dear Raushan,

I've tried to use the code in E17.6.1. However, I've got strange results (the corrections REDUCED my standard errors of the probit equation dramatically). Thus, I would not try to generalize it to the case of three equations. Another way to estimate the model might be maximum likelihood (using MAXIMIZE). You can factorize the density function of the trivariate normal distribution into the conditional distribution (both dummies given the observable variable) and the marginal density of the observable variable. This makes ML feasible.

Best regards

-----Ursprüngliche Nachricht-----
Von: limdep-bounces at
[mailto:limdep-bounces at]Im Auftrag von Raushan
Gesendet: Donnerstag, 23. März 2006 18:11
An: limdep at
Betreff: [Limdep Nlogit List] simultaneous equations model with 3
endogeneous variables

Dear all, 
I would like to know, whether it is possible to estimate simultaneous equations model with 3 endogenous variables: 1 - observed  directly and 2 binary variables. 

Has anybody already tried to extend the model presented in Limdep Guide (E17.6.1) to the case with 3 simultaneous equations? And what one should pay attention to in this case? 

I would be very happy, if somebody could share his experiences in applying this model. 

best regards, 


Institute of Agricultural Development in Central and Eastern Europe (IAMO)
D-06120 Halle (Saale), Germany 
Tel.: +49 345 2928 134
Fax: +49 345 2928 399
e-mail: bokusheva at
Limdep site list
Limdep at

More information about the Limdep mailing list