[Limdep Nlogit List] (no subject)
William Greene
wgreene at stern.nyu.edu
Thu Dec 14 22:49:56 EST 2006
Darren:
(1) RHO is not "calculated." It is one of the parameters that is estimated by the ML procedure.
The theory is that it equals Var[u]/{1 + Var[u]} where u is the random effect.
(2) No, it is not equal to, or related to, the B&P LM test for random effects.
(3) A Wald (t) test for RHO=0 is a test for the presence of random effects. This (and any other
test) is better than fitting the pooled model by OLS, which I would not recommend under any
circumstances.
Regards,
/Bill Greene
************************************************
Professor William Greene
Department of Economics
Stern School of Business
New York University
44 West 4th St., Rm. 7-78
New York, NY 10012
Ph. 212.998.0876
Fax. 212.995.4218
URL. http://www.stern.nyu.edu/~wgreene
Email. wgreene at stern.nyu.edu
************************************************
----- Original Message -----
From: D.Lee2 at business.uq.edu.au
Date: Thursday, December 14, 2006 4:32 am
Subject: [Limdep Nlogit List] (no subject)
> Hi Guys,
>
>
>
> I am trying to ascertain what Rho means in the output of a probit
> paneldata model.
>
>
>
> 1) How is it calculated (i.e. is there anything in the manual)?
>
>
>
> 2) Is it like the Breusch and Pagan (1890) LM test for the random
> effects model (see page 298, Greene 2003, Econometric Analysis)?
>
>
>
> 3) Does it provide a test that the random effects model is better than
> the base OLS pooled regression (if not what does in LimDep)?
>
>
>
> (Rho is highlighted in the output below)
>
>
>
> +---------------------------------------------+
>
> | Random Effects Binary Probit Model |
>
> | Maximum Likelihood Estimates |
>
> | Model estimated: Dec 14, 2006 at 06:57:16PM.|
>
> | Dependent variable BIN |
>
> | Weighting variable None |
>
> | Number of observations 678 |
>
> | Iterations completed 1 |
>
> | LM Stat. at start values 80.47393 |
>
> | LM statistic kept as scalar LMSTAT |
>
> | Log likelihood function -439.3980 |
>
> | Number of parameters 10 |
>
> | Info. Criterion: AIC = 1.32566 |
>
> | Finite Sample: AIC = 1.32614 |
>
> | Info. Criterion: BIC = 1.39231 |
>
> | Info. Criterion:HQIC = 1.35146 |
>
> | Restricted log likelihood -447.8178 |
>
> | Chi squared 16.83944 |
>
> | Degrees of freedom 1 |
>
> | Prob[ChiSqd > value] = .4067828E-04 |
>
> | Unbalanced panel has 504 individuals. |
>
> | Missing data: Skipped 137 individuals. |
>
> +---------------------------------------------+
>
> +---------+--------------+----------------+--------+---------+-----
> -----
> +
>
> |Variable | Coefficient | Standard Error |b/St.Er.|P[|Z|>z] |
> Mean of
> X|
>
> +---------+--------------+----------------+--------+---------+-----
> -----
> +
>
> Constant -1.86672694 .27865089 -6.699 .0000
>
> Z_SCORE -.00777047 .00980270 -.793 .4280
> 4.23746475
> TTL_ASSE .50815559 .07045905 7.212 .0000
> 3.62885073
> PE_RATIO .00039061 .00031415 1.243 .2137
> 25.1300835
> MB_RATIO .00200406 .00679547 .295 .7681
> 4.32215507
> TA_COM_E -.00430420 .01020701 -.422 .6733
> 3.15135305
> CURRENT_ .02051882 .01975231 1.039 .2989
> 2.09311815
> FCF_TA -.33304236 .29198730 -1.141 .2540 -
> .00747627
> MARKBLE1 .69932994 .82496361 .848 .3966
> .00386895
> Rho .04306220 .04783482 .900 .3680
>
>
>
> Thanks for any help in advance.
>
>
> Darren
>
> UQ Business School
> Colin Clark Blg
> Blair Drive
> St Lucia
> QLD, 4072
> PH: 07 33656743
> Fax: 07 33656988
>
> Email: darren.lee at business.uq.edu.au
>
>
>
> _______________________________________________
> Limdep site list
> Limdep at limdep.itls.usyd.edu.au
> http://limdep.itls.usyd.edu.au
>
More information about the Limdep
mailing list