[Limdep Nlogit List] NLOGIT6 mixed logit Halton draws

William Greene wgreene at stern.nyu.edu
Thu Sep 5 00:43:43 AEST 2019


Jürgen:  You are attempting to estimate a 10x10 covariance matrix entirely
based on unobservables.  There are no theorems at work (that I know of),
but in my experience this is extremely optimistic.  There is also no theory
(again, that I am aware of) that links the number of draws to any of the
measures you've noted.  In order to examine this systematically, I suggest
you reduce the number of random parameters to 2 or 3 - the mix of normal and
lognormal is also a substantial complication - and see if the same patterns
emerge.  Also, estimating covariances of unobservables is complicated, so
you might want to examine this dimension.
Regards
Bill Greene

On Wed, Sep 4, 2019 at 7:48 AM Jürgen Meyerhoff <
juergen.meyerhoff at tu-berlin.de> wrote:

> Hi Bart, check this publication:
> Czajkowski, M., Budziński, W., 2019. Simulation error in maximum likelihood
> estimation of discrete choice models. Journal of Choice Modelling 31,
> 73-85.
> Best from Berlin
> Jürgen
>
>
> *********************************************************************************************
> Dr. Jürgen Meyerhoff, TU Berlin, FG Landschaftsökonomie
> <https://protect-au.mimecast.com/s/MCVrCANZvPiQmGpAiGgM4k?domain=landschaftsoekonomie.tu-berlin.de>, EB 4-2
> Straße des 17. Juni 145,  10623 Berlin, Tel. 0049(30) 314 73 322
>
> =>  Berlin DCE Colloquium
> <https://protect-au.mimecast.com/s/zWXLCBNZwLiGZzW5t6og1F?domain=landschaftsoekonomie.tu-berlin.de
> >
> =>  Personal Website <https://protect-au.mimecast.com/s/6VnpCD1jy9t96DYliASFBM?domain=jmber.de>
>
>
> Am Mi., 4. Sept. 2019 um 13:33 Uhr schrieb Bart Immerzeel <
> bart.immerzeel at nmbu.no>:
>
> > Hello,
> >
> > I am running a mixed logit model in NLOGIT6 on some DCE data.
> > It uses random parameters for the attributes in the DCE, and I am using
> > Halton draws.
> > However, the higher the number of Halton draws I use, the lower the
> > McFadden pseudo R2, and the less statistically significant coefficients I
> > get. I have no idea why that is. Does anyone know what might cause this
> > issue?
> >
> > Here is the model spec:
> >
> > ? MXL with continuous attributes
> >
> > Sample; All $
> > Create; Tax2=Tax/1000$
> > Create; Tax3=Tax2*-1$
> > Reject ;choice=-999$
> > ?Reject ;Protest=1$
> >
> > NLOGIT
> >       ;Lhs=choice
> >       ;Choices=1,2,3
> >       ;Rhs=
> > BAU,
> > ShareF,
> > Inten1v,
> > Inten3e,
> > Water2m,
> > Water3c,
> > Nature,
> > Flood,
> > Employ,
> > Tax3
> >       ;Pds=Panel2
> >       ;RPL
> >       ;Halton ? type of draws
> >       ;pts=1000 ? number of draws
> >       ;Fcn=
> > BAU(n),
> > ShareF(n),
> > Inten1v(n),
> > Inten3e(n),
> > Water2m(n),
> > Water3c(n),
> > Nature(n),
> > Flood(n),
> > Employ(n),
> > Tax3(l)
> >       ;correlated
> >       ?;WTP=
> >
> ShareF/Tax3,Inten1v/Tax3,Inten3e/Tax3,Water2m/Tax3,Water3c/Tax3,Nature/Tax3
> >       ;Parameters
> >       ;maxit=200
> >       ;tlf=0.00000000001
> >
> > $
> >
> > Kind regards,
> > Bart Immerzeel
> >
> > _______________________________________________
> > Limdep site list
> > Limdep at mailman.sydney.edu.au
> > http://limdep.itls.usyd.edu.au
> >
> >
> _______________________________________________
> Limdep site list
> Limdep at mailman.sydney.edu.au
> http://limdep.itls.usyd.edu.au
>
>

-- 
William Greene
Department of Economics
Stern School of Business, New York University
44 West 4 St., 7-90
New York, NY, 10012
URL: https://protect-au.mimecast.com/s/9eLaCE8kz9t7Ak45tPvvg9?domain=people.stern.nyu.edu
Email: wgreene at stern.nyu.edu
Ph. +1.212.998.0876
Editor in Chief: Journal of Productivity Analysis
Editor in Chief: Foundations and Trends in Econometrics
Associate Editor: Economics Letters
Associate Editor: Journal of Business and Economic Statistics
Associate Editor: Journal of Choice Modeling


More information about the Limdep mailing list