[Limdep Nlogit List] sample selection and multinomial logit
RAUX Charles
Charles.RAUX at laet.msh-lse.fr
Wed Mar 25 21:40:46 AEDT 2020
Hello,
I am trying to estimate an mlogit (mode choice among 4 modes but with only individual characteristics) with sample selection (probit) with Limdep11/Nlogit 6, following the example in documentation E54.5. Choice is coded from 0 to 3. If I run mlogit without selection then I get expected results (asc and coef for 3 modes). If I run probit/hold and the mlogit/selection, it happens as if choice is treated as binary. I don't get the coefficients for each value of choice (see below). What's wrong?
Charles Raux
Histogram for CHOICE NOBS = 21994, Too low: 0, Too high: 0
Bin Value of CHOICE Frequency Cumulative Frequency
========================================================================
0 0 17828 ( .8106) 17828 ( .8106)
1 1 1859 ( .0845) 19687 ( .8951)
2 2 1746 ( .0794) 21433 ( .9745)
3 3 561 ( .0255) 21994 (1.0000)
|-> PROBIT ; Lhs = treated ; Rhs = wi ; Hold $
Iterative procedure has converged
Normal exit: 5 iterations. Status=0, F= .1234722D+05
-----------------------------------------------------------------------------
Binomial Probit Model
Dependent variable TREATED
Log likelihood function -12347.22015
Restricted log likelihood -12500.76863
Chi squared [ 13](P= .000) 307.09697
Significance level .00000
McFadden Pseudo R-squared .0122831
Estimation based on N = 21994, K = 14
Inf.Cr.AIC = 24722.4 AIC/N = 1.124
Results retained for SELECTION model.
--------+--------------------------------------------------------------------
| Standard Prob. 95% Confidence
TREATED| Coefficient Error z |z|>Z* Interval
--------+--------------------------------------------------------------------
|Index function for probability......................................
Constant| -.78312*** .08955 -8.75 .0000 -.95863 -.60761
AGE25| .12817*** .04069 3.15 .0016 .04842 .20792
AGE35| .11733*** .03895 3.01 .0026 .04100 .19367
AGE45| .01710 .03832 .45 .6555 -.05802 .09221
AGE55| -.10244** .04274 -2.40 .0165 -.18620 -.01867
TENANT_S| -.24980*** .04804 -5.20 .0000 -.34395 -.15565
OWNER| .20638*** .02572 8.02 .0000 .15598 .25679
DIPL_B| .02215 .03312 .67 .5036 -.04277 .08707
DIPL_C| -.12201*** .03438 -3.55 .0004 -.18938 -.05463
DIPL_D| -.26711*** .03084 -8.66 .0000 -.32755 -.20666
LESS1| .07153 .08387 .85 .3937 -.09285 .23592
MORE1| .08690 .08161 1.06 .2870 -.07305 .24684
IMMIGRAN| -.07866* .04030 -1.95 .0510 -.15766 .00033
NBHHMEMB| .00104 .00857 .12 .9034 -.01575 .01783
--------+--------------------------------------------------------------------
***, **, * ==> Significance at 1%, 5%, 10% level.
Model was estimated on Mar 25, 2020 at 11:27:23 AM
-----------------------------------------------------------------------------
|-> mlogit ; lhs = choice
; rhs= x
; selection $
-----------------------------------------------------------------------------
Logit Regression Start Values for CHOICE
Dependent variable CHOICE
Log likelihood function -2432.51826
Estimation based on N = 21994, K = 13
Inf.Cr.AIC = 4891.0 AIC/N = .222
--------+--------------------------------------------------------------------
| Standard Prob. 95% Confidence
CHOICE| Coefficient Error z |z|>Z* Interval
--------+--------------------------------------------------------------------
Constant| .07131 .28867 .25 .8049 -.49448 .63710
FEMALE| .44169*** .07420 5.95 .0000 .29627 .58712
AGE25| -.70372*** .15601 -4.51 .0000 -1.00949 -.39796
AGE35| -.41604*** .14006 -2.97 .0030 -.69055 -.14153
AGE45| -.30388** .13683 -2.22 .0264 -.57207 -.03569
AGE55| -.12381 .15168 -.82 .4144 -.42109 .17348
TENANT_S| -.02040 .20471 -.10 .9206 -.42163 .38083
OWNER| -.06119 .10202 -.60 .5486 -.26115 .13876
DIPL_B| -.15659 .11770 -1.33 .1834 -.38729 .07410
DIPL_C| -.20521 .12578 -1.63 .1028 -.45173 .04132
DIPL_D| -.32594*** .11476 -2.84 .0045 -.55087 -.10101
LESS1| -.78624*** .26338 -2.99 .0028 -1.30246 -.27002
MORE1| -1.49821*** .25773 -5.81 .0000 -2.00335 -.99307
--------+--------------------------------------------------------------------
***, **, * ==> Significance at 1%, 5%, 10% level.
Model was estimated on Mar 25, 2020 at 11:27:34 AM
-----------------------------------------------------------------------------
Line search at iteration 33 does not improve the function
Exiting optimization
-----------------------------------------------------------------------------
Selectivity Corrected Logit Model
Dependent variable CHOICE
Log likelihood function -117023.96957
Estimation based on N = 21994, K = 29
Inf.Cr.AIC = 234105.9 AIC/N = 10.644
--------+--------------------------------------------------------------------
| Standard Prob. 95% Confidence
CHOICE| Coefficient Error z |z|>Z* Interval
--------+--------------------------------------------------------------------
|Coefficients in binary logit model..................................
Constant| .55154*** .18670 2.95 .0031 .18561 .91746
FEMALE| .22248*** .03102 7.17 .0000 .16167 .28329
AGE25| -.48749*** .06552 -7.44 .0000 -.61592 -.35907
AGE35| -.22802*** .06060 -3.76 .0002 -.34680 -.10923
AGE45| -.12163** .05922 -2.05 .0400 -.23769 -.00556
AGE55| .00545 .06721 .08 .9353 -.12627 .13717
TENANT_S| -.11913 .09211 -1.29 .1959 -.29967 .06141
OWNER| -.04290 .04130 -1.04 .2990 -.12386 .03805
DIPL_B| -.02995 .04838 -.62 .5360 -.12478 .06489
DIPL_C| -.16904*** .05171 -3.27 .0011 -.27040 -.06769
DIPL_D| -.54675*** .04745 -11.52 .0000 -.63975 -.45374
LESS1| -.33793* .17950 -1.88 .0597 -.68974 .01388
MORE1| -.95063*** .17724 -5.36 .0000 -1.29802 -.60324
|Coefficients in Probit = Alpha/sqr(1-rho^2).........................
Constant| -.78287*** .09006 -8.69 .0000 -.95938 -.60637
AGE25| .12822*** .04067 3.15 .0016 .04851 .20794
AGE35| .11746*** .03892 3.02 .0025 .04118 .19374
AGE45| .01714 .03836 .45 .6550 -.05804 .09231
AGE55| -.10247** .04284 -2.39 .0168 -.18644 -.01850
TENANT_S| -.24968*** .04781 -5.22 .0000 -.34340 -.15597
OWNER| .20652*** .02551 8.09 .0000 .15652 .25653
DIPL_B| .02209 .03322 .66 .5061 -.04302 .08719
DIPL_C| -.12208*** .03450 -3.54 .0004 -.18970 -.05445
DIPL_D| -.26721*** .03093 -8.64 .0000 -.32783 -.20659
LESS1| .07133 .08340 .86 .3924 -.09213 .23479
MORE1| .08664 .08114 1.07 .2856 -.07239 .24568
IMMIGRAN| -.07886* .04045 -1.95 .0512 -.15814 .00043
NBHHMEMB| .00101 .00863 .12 .9067 -.01591 .01793
|Standard deviation of e(i) in logit model...........................
Sigma(e)| .02441 .01526 1.60 .1096 -.00549 .05432
|Correlation of selection u(i) and logit e(i)........................
Rho(u,e)| -.00391 .00925 -.42 .6722 -.02205 .01422
--------+--------------------------------------------------------------------
***, **, * ==> Significance at 1%, 5%, 10% level.
Model was estimated on Mar 25, 2020 at 11:35:25 AM
-----------------------------------------------------------------------------
More information about the Limdep
mailing list