[Limdep Nlogit List] LIMDEP and Stata Yield very different results in Multivariate Probit
Fred Dzanku
fdzanku at gmail.com
Fri Mar 18 17:23:42 AEDT 2016
Dear Prof. Greene, the Stata routine uses the GHK simulator too. Some of
the coefficients are not far apart but the standard errors in particular
are, thus yielding very differing conclusions. I have discussed this issue
at the Stata forum to I notice that the results are particularly different
when estimating cluster robust SEs. I tried using the Halton sequence in
both programs rather than random draws and the differences persist. Please
see the discussion here:
http://www.statalist.org/forums/forum/general-stata-discussion/general/1330862-stata-and-limdep-yield-very-different-results
Fred
On 18 Mar 2016 00:23, "William Greene" <wgreene at stern.nyu.edu> wrote:
> Fred. The estimates are not that far apart. And, note that the log
> likelihood
> values are essentially the same. I can tell you that nlogit is using the
> GHK simulator
> to do the estimation. I do not know what Stata is doing. You can be sure
> that the
> estimated coefficients are very sensitive to the values of the correlation
> coefficients.
> But, in order to compare the results, you should probably compare the
> estimated
> partial effects. I don't know if Stata knows how to do that. The nlogit
> command is
> described in the manual.
> /B. Greene
>
> On Thu, Mar 17, 2016 at 6:30 PM, Fred Dzanku <fdzanku at gmail.com> wrote:
>
> > Dear Users, I estimated a trivariate probit model (with clustering) in
> > Limdep and Stata. Surprisingly, the results (particularly) the standard
> > errors are very different, yielding very different inference. In Stata I
> > used the mvprobit command (written by Lorenzo Cappellari and Stephen P.
> > Jenkins). The difference is indeed shocking. Does anyone know why this
> > might
> > be the case?
> >
> >
> >
> > The specification is:
> >
> > y1 = f1(y2, y3, x1, x2, x3)
> > y2 = f2(x1, x2, x3, x4, x5)
> > y3 = f3(x2, x3, x4, x6)
> >
> >
> >
> > In Limdep I wrote:
> >
> >
> >
> > Sample ;all $
> >
> > Skip $
> >
> > Mprobit ;Lhs=y1,y2,y3
> >
> > ;eq1=y2,y3,x1,x2,x3,one
> >
> > ;eq2=x1,x2,x3,x4,x5,one
> >
> > ;eq3=x2,x3,x4,x6,one
> >
> > ;pts=200;cluster=vid $
> >
> > And the results were:
> >
> >
> >
> > Normal exit from iterations. Exit status=0.
> >
> > +---------------------------------------------+
> > | Multivariate Probit Model: 3 equations. |
> > | Maximum Likelihood Estimates |
> > | Model estimated: Mar 17, 2016 at 00:29:05AM.|
> > | Dependent variable MVProbit |
> > | Weighting variable None |
> > | Number of observations 650 |
> > | Iterations completed 33 |
> > | Log likelihood function -1242.182 |
> > | Number of parameters 20 |
> > | Info. Criterion: AIC = 3.88364 |
> > | Finite Sample: AIC = 3.88569 |
> > | Info. Criterion: BIC = 4.02139 |
> > | Info. Criterion:HQIC = 3.93707 |
> > | Replications for simulated probs. = 200 |
> > +---------------------------------------------+
> > +---------------------------------------------------------------------+
> > | Covariance matrix for the model is adjusted for data clustering. |
> > | Sample of 650 observations contained 66 clusters defined by |
> > | variable VID which identifies by a value a cluster ID. |
> > | Sample of 650 observations contained 1 strata defined by |
> > | 650 observations (fixed number) in each stratum. |
> > +---------------------------------------------------------------------+
> > +--------+--------------+----------------+--------+--------+----------+
> > |Variable| Coefficient | Standard Error |b/St.Er.|P[|Z|>z]| Mean of X|
> > +--------+--------------+----------------+--------+--------+----------+
> > ---------+Index function for Y1
> > Y2 | .84344025 5.74939391 .147 .8834 .38000000
> > Y3 | .42013852 3.25958960 .129 .8974 .35538462
> > X1 | -.00806300 .00860756 -.937 .3489 46.6107692
> > X2 | .31793308 .71794537 .443 .6579 .34307692
> > X3 | .01468727 .00840065 1.748 .0804 11.2224103
> > Constant| -.31411643 .77762634 -.404 .6863
> > ---------+Index function for Y2
> > X1 | .00779023 .00104284 7.470 .0000 46.6107692
> > X2 | .16831606 .00775319 21.709 .0000 .34307692
> > X3 | -.00663025 .00072659 -9.125 .0000 11.2224103
> > X4 | .14847431 .14394530 1.031 .3023 .28923077
> > X5 | .04544672 .05689766 .799 .4244 .11538462
> > Constant| -.70426701 .08723116 -8.074 .0000
> > ---------+Index function for Y3
> > X2 | .17303656 .01161279 14.901 .0000 .34307692
> > X3 | .00272569 .00050864 5.359 .0000 11.2224103
> > X4 | .19051601 .05574615 3.418 .0006 .28923077
> > X6 | .42854019 .04664493 9.187 .0000 .07692308
> > Constant| -.55390573 .03198934 -17.315 .0000
> > ---------+Correlation coefficients
> > R(01,02)| -.58312487 3.20681595 -.182 .8557
> > R(01,03)| -.44115615 1.23356392 -.358 .7206
> > R(02,03)| .40497242 .00399374 101.402 .0000
> >
> >
> >
> > In Stata I estimated:
> >
> >
> >
> > mvprobit (y1 y2 y3 x1 x2 x3) (y2 x1 x2 x3 x4 x5) (y3 x2 x3 x4 x6),
> > cluster(vid) draw(200) nolog seed(1003)
> >
> > with the following results:
> >
> >
> >
> > Multivariate probit (MSL, # draws = 200) Number of obs =
> > 650
> > Wald chi2(14) =
> > 248.37
> > Log pseudolikelihood = -1241.6481 Prob > chi2 =
> > 0.0000
> >
> > (Std. Err. adjusted for 66 clusters in
> > vid)
> >
> >
> ----------------------------------------------------------------------------
> > --
> > | Robust
> > | Coef. Std. Err. z P>|z| [95% Conf.
> > Interval]
> >
> >
> -------------+--------------------------------------------------------------
> > --
> > y1 |
> > y2 | 1.343666 .234929 5.72 0.000 .8832137
> > 1.804118
> > y3 | -.465239 .7326024 -0.64 0.525 -1.901113
> > .9706353
> > x1 | -.00879 .0037455 -2.35 0.019 -.0161311
> > -.001449
> > x2 | .2877144 .1807015 1.59 0.111 -.0664541
> > .6418829
> > x3 | .0162344 .0078329 2.07 0.038 .0008822
> > .0315867
> > _cons | -.1612781 .3041286 -0.53 0.596 -.7573591
> > .4348029
> >
> >
> -------------+--------------------------------------------------------------
> > --
> > y2 |
> > x1 | .0075814 .0042735 1.77 0.076 -.0007944
> > .0159573
> > x2 | .1548884 .0846257 1.83 0.067 -.0109749
> > .3207516
> > x3 | -.0071789 .0064509 -1.11 0.266 -.0198224
> > .0054645
> > x4 | .1732857 .0972146 1.78 0.075 -.0172514
> > .3638228
> > x5 | .0102722 .1315351 0.08 0.938 -.2475319
> > .2680762
> > _cons | -.6934089 .2054128 -3.38 0.001 -1.096011
> > -.2908072
> >
> >
> -------------+--------------------------------------------------------------
> > --
> > y3 |
> > x2 | .1702344 .0881082 1.93 0.053 -.0024546
> > .3429234
> > x3 | .0032607 .0058996 0.55 0.580 -.0083022
> > .0148237
> > x4 | .1644255 .1287772 1.28 0.202 -.0879732
> > .4168241
> > x6 | .4566752 .2159169 2.12 0.034 .0334859
> > .8798645
> > _cons | -.554203 .1180065 -4.70 0.000 -.7854915
> > -.3229145
> >
> >
> -------------+--------------------------------------------------------------
> > --
> > /atrho21 | -1.052254 .6801192 -1.55 0.122 -2.385263
> > .2807557
> >
> >
> -------------+--------------------------------------------------------------
> > --
> > /atrho31 | .0235009 .4724996 0.05 0.960 -.9025812
> > .9495831
> >
> >
> -------------+--------------------------------------------------------------
> > --
> > /atrho32 | .4372824 .0815021 5.37 0.000 .2775413
> > .5970236
> >
> >
> -------------+--------------------------------------------------------------
> > --
> > rho21 | -.7826809 .2634854 -2.97 0.003 -.9831906
> > .2736043
> >
> >
> -------------+--------------------------------------------------------------
> > --
> > rho31 | .0234966 .4722387 0.05 0.960 -.7175524
> > .7395942
> >
> >
> -------------+--------------------------------------------------------------
> > --
> > rho32 | .4113893 .0677086 6.08 0.000 .270628
> > .5349282
> >
> >
> ----------------------------------------------------------------------------
> > --
> >
> >
> >
> > Other programs in Stata such as gsem and cmp yield estimates that are
> > closer
> > to the mvprobit estimates. What might be the problem here?
> >
> >
> >
> > Fred
> >
> >
> >
> >
> > _______________________________________________
> > Limdep site list
> > Limdep at limdep.itls.usyd.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: http://people.stern.nyu.edu/wgreene
> Email: wgreene at stern.nyu.edu
> Ph. +1.212.998.0876
> _______________________________________________
> Limdep site list
> Limdep at limdep.itls.usyd.edu.au
> http://limdep.itls.usyd.edu.au
>
More information about the Limdep
mailing list