[Limdep Nlogit List] Restricted estimation

William Greene wgreene at stern.nyu.edu
Mon Sep 21 12:26:15 EST 2009

For *TESTING* the hypothesis that RHO = 0, either of the two solutions
suggested earlier works. A third - seems too easy - is to use the t ratio
that is reported with the estimate of RHO in the MLE results. To carry
out an LR test, fit the probit model and the regression separately and
add the log likelihoods. Subtract the sum from the logL from the unrestricted
model, then multiply the difference by 2. This is a chi-squared with 1 degree
of freedom. The 95% critical value is 3.84.
Bill Greene

----- Original Message -----
From: "Steven T Yen" <syen at utk.edu>
To: limdep at limdep.itls.usyd.edu.au
Sent: Sunday, September 20, 2009 12:27:44 PM GMT -05:00 Columbia
Subject: [Limdep Nlogit List] Restricted estimation

Dear Limdepers
I am testing restricted estimation with the sample selection model by imposing rho=0 (I have 15 regressors in the selection equation and level equation). The restriction ";rst=..." seems to get ignored. Any clues? Thanks.
probit ;lhs=d ;rhs=z ;hold $
select ;lhs=y ;rhs=x ;mle ;par 
            b1,b2,b3,b4,b5,b6,b7,b8,b9,b10,b11,b12,b13,b14,b15,ss,0 $
Steven T. Yen
<http://web.utk.edu/~syen/> Check out our postdoc opening at:
Limdep site list
Limdep at limdep.itls.usyd.edu.au

More information about the Limdep mailing list