[Limdep Nlogit List] Nlogit - Intercept in Hierarchical Logit Models
Junhao Liu
junhao.liu at sydney.edu.au
Mon May 11 17:34:04 AEST 2020
Dear all,
I am estimating an RPLOGIT model with heterogeneity in the means of the random parameters. In this two-level model, the mean of the utility weights (beta) is modelled as an intercept term plus the effects from some choice invariant characteristics. My question is: does Nlogit report the estimate of this intercept term?
Below is an example using the classic Nlogit "mnc" data, where the utility weight for GC is a hierarchical parameter affected by HINC. But I cannot find the intercept term in the "heterogeneity in mean" model in the output, only the estimate of GC:HIN. Any advice is appreciated!
|-> rplogit;lhs=mode;
; Choices = air,train,bus,car
; Rhs = gc,ttme,invt
; Rh2 = one
; RPL= hinc
; Fcn = gc(n),ttme[n],invt[n]
$
Iterative procedure has converged
Normal exit: 6 iterations. Status=0, F= .1949974D+03
-----------------------------------------------------------------------------
(The MNL output omitted)
-----------------------------------------------------------------------------
Line search at iteration 21 does not improve the function
Exiting optimization
-----------------------------------------------------------------------------
Random Parameters Multinom. Logit Model
Dependent variable MODE
Log likelihood function -175.66089
Restricted log likelihood -291.12182
Chi squared [ 10](P= .000) 230.92185
Significance level .00000
McFadden Pseudo R-squared .3966069
Estimation based on N = 210, K = 10
Inf.Cr.AIC = 371.3 AIC/N = 1.768
---------------------------------------
Log likelihood R-sqrd R2Adj
No coefficients -291.1218 .3966 .3869
Constants only -283.7588 .3809 .3710
At start values -194.9974 .0992 .0846
Note: R-sqrd = 1 - logL/Logl(constants)
---------------------------------------
Response data are given as ind. choices
Replications for simulated probs. = 100
Pseudo random draws (Mersenne twister).
BHHH estimator used for asymp. variance
Number of obs.= 210, skipped 0 obs
--------+--------------------------------------------------------------------
| Standard Prob. 95% Confidence
MODE| Coefficient Error z |z|>Z* Interval
--------+--------------------------------------------------------------------
|Random parameters in utility functions..........................
GC| -.08302 .10337 -.80 .4219 -.28562 .11958
TTME| -.68311* .41072 -1.66 .0963 -1.48812 .12189
INVT| -.03704 .02401 -1.54 .1229 -.08410 .01001
|Nonrandom parameters in utility functions.......................
A_AIR| 23.8605 15.47836 1.54 .1232 -6.4765 54.1976
A_TRAIN| 33.3785 21.39271 1.56 .1187 -8.5504 75.3075
A_BUS| 30.7295 19.73732 1.56 .1195 -7.9549 69.4140
|Heterogeneity in mean, Parameter:Variable.......................
GC:HIN| -.00036 .00178 -.20 .8421 -.00385 .00314
TTME:HIN| 0.0 .....(Fixed Parameter).....
INVT:HIN| 0.0 .....(Fixed Parameter).....
|Distns. of RPs. Std.Devs or limits of triangular................
NsGC| .06457 .07420 .87 .3841 -.08085 .21000
NsTTME| .44440 .31382 1.42 .1567 -.17068 1.05947
NsINVT| .03000 .01941 1.55 .1223 -.00805 .06804
--------+--------------------------------------------------------------------
***, **, * ==> Significance at 1%, 5%, 10% level.
Fixed parameter ... is constrained to equal the value or
had a nonpositive st.error because of an earlier problem.
Model was estimated on May 11, 2020 at 05:22:28 PM
-----------------------------------------------------------------------------
Parameter Matrix for Heterogeneity in Means.
--------+--------------
Delta_RP| 1
--------+--------------
1| -.355329E-03
2| .000000
3| .000000
Junhao LIU | Postdoctoral Research Associate
Discipline of Finance | The University of Sydney Business School
THE UNIVERSITY OF SYDNEY
Rm 512, H69 - Codrington Building | The University of Sydney | NSW | 2006
E junhao.liu at sydney.edu.au<mailto:junhao.liu at sydney.edu.au> | W https://protect-au.mimecast.com/s/EC1ICBNqjlCvP5lPTzCfBG?domain=junhaoliu.com<https://protect-au.mimecast.com/s/mGlSCD1vlpTXjlOjT5Xg_j?domain=junhaoliu.com>
More information about the Limdep
mailing list