[Limdep Nlogit List] command to save likelihood estimates

William Greene wgreene at stern.nyu.edu
Fri Jun 18 06:07:24 EST 2010


Christoph. Since the contribution is to the simulated log likelihood, it
is necessary to simulate the computation of the elasticities. The simulated
log likelihoods are saved in variable logl_obs. But, this does not save
the individual draws. I'm afraid there is no command to do that. Moreover,
it is not possible to replicate the draws used internally because there is
a different seed generated for each individual in the sample. They each use their
own set of draws. You can replicate the algorithm, but probably not the 
specific values. You can get close if you have an extremely large number
of random draws in your simulations.
/Bill Greene

----- Original Message -----
From: "Chris Rh" <chrisrh at ymail.com>
To: Limdep at limdep.itls.usyd.edu.au
Sent: Thursday, June 17, 2010 3:25:06 PM GMT -05:00 US/Canada Eastern
Subject: [Limdep Nlogit List] command to save likelihood estimates

Dear Bill,
I am trying to calculate the WTP derived from individual parameters as you proposed in Greene et al. 2005 in Scarpa & Alberini (eds). The WTP estimator that you propose therein weights each conditional parameter estimate by its contribution to "the likelihood function computed at the r'th draw of each individual" (p.30). So, I need not only the choice-specific conditional parameters that NLOGIT saves when the ;par command is added, but also a command to obtain the likelihood function associated with each of the choice-specific conditional parameters. Is there a command to obtain these likelihood estimates in NLOGIT? If not, how do proceed to calculate the WTP estimator that you propose in this text?
Thanks for clarification,
Christoph



      
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au


More information about the Limdep mailing list