[Limdep Nlogit List] Estimating non-linear equations withendogenous regressors and selection

Mintewab Bezabih Mintewab.Bezabih at port.ac.uk
Tue Jan 19 22:56:39 EST 2010


Thanks a lot. It is much clearer to me now. 
minti

>>> William Greene <wgreene at stern.nyu.edu> 19/01/2010 01:18 >>>
Minti. I'm afraid not. The SELECT command has only that one option,
TOBIT.  However, there is a POISSON MLE for the sample selection model.
There is also an ordered probit/selection estimator provided by the
program. The BIVARIATE_PROBIT with selection is a sample selection 
model that is also fit by MLE.  All of these are documented separately
in the manual.
/Bill Greene

----- Original Message -----
From: "Mintewab Bezabih" <Mintewab.Bezabih at port.ac.uk>
To: limdep at limdep.itls.usyd.edu.au 
Sent: Monday, January 18, 2010 7:47:57 PM GMT -05:00 Colombia
Subject: [Limdep Nlogit List] Estimating non-linear equations with endogenous regressors and	selection

Dear all,

I have been estimating a tobit model with selection and an endogenous regressor using the code below (and it seems to work). However I have alternative measures of the dependent variable ( dichotomous, categorical). I wonder if I can just modify this specification from tobit to probit/poisson/ordered probit as I change my dependent variable?

Below is the code I used for tobit with selection.

 

SKIP$
PROBIT
;Lhs=hadsex;Rhs=ONE,BLACK,COLOURED,FEMALE,EDUC,INFHIV,HIVKNOW,MLIFEEXP,
VLIFEEXP, AGE_10, AGE_2, CHILDEXP ; Hold $

select;TOBIT; Lhs = RISKp ; Rhs = ONE, BLACK, COLOURED, FEMALE, EDUC, INFHIV,
HIVKNOW, MLIFEEXP, VLIFEEXP, AGE_10$
MATRIX;BT=1/S*b$ 
MAXIMIZE;Labels = b0,b1,b2
; start = 0,0,0
; Fcn = b0+b1*educ+b2*BLACK;INST=CONDOM,BLACK,COLOURED;mle $

Many thanks in advance. 

minti


_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au 
http://limdep.itls.usyd.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