Following the same setup in the FAQ "How do I impose the restriction that rho is
zero using the heckman command with full ml?"
(http://www.stata.com/support/faqs/stat/constraints.html), you have constrain
athrho.
For example, if you wanted to set rho = -.5 you could do the following:
. sysuse auto
(1978 Automobile Data)
. mark low_price if price<4000
. local athrho=1/2*ln((1+(-.5))/(1-(-.5)))
. constraint define 1 [athrho]_cons=`athrho'
. biprobit low fore mpg, const(1) nolog
Bivariate probit regression Number of obs = 74
Wald chi2(2) = 23.98
Log likelihood = -65.975268 Prob > chi2 = 0.0000
( 1) [athrho]_cons = -.5493061
------------------------------------------------------------------------------
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
low_price |
mpg | .0920505 .0310962 2.96 0.003 .0311031 .152998
_cons | -3.148761 .7670861 -4.10 0.000 -4.652222 -1.6453
-------------+----------------------------------------------------------------
foreign |
mpg | .0969753 .0301191 3.22 0.001 .0379429 .1560077
_cons | -2.646529 .6838252 -3.87 0.000 -3.986802 -1.306256
-------------+----------------------------------------------------------------
/athrho | -.5493061 . . . . .
-------------+----------------------------------------------------------------
rho | -.5 . -1 1
------------------------------------------------------------------------------
Likelihood-ratio test of rho=0: chi2(1) = .155227 Prob > chi2 = 0.6936
Hope this helps,
Scott
----- Original Message -----
From: <[email protected]>
To: <[email protected]>
Sent: Tuesday, January 20, 2004 9:48 AM
Subject: st: Add constraints in biprobit
> Hi, all,
>
> I am using biprobit to estimate a bivariate probit model with some
> nonlinear constraints on parameters. I have two questions:
>
> 1. The iterations go fine with the "fitting comparison model" stage. But
> when the procedure goes to the "Fitting full model" stage, it says
> "Constraints invalid: not possible with test". Is that I have to specify
> linear constraints instead of nonlinear ones?
>
> 2. I also have to add some constraints on the rho, correlation between two
> equations. How do I write the "rho" explicitly when defining the
> constraints? Thanks.
>
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
