Alejandro,
How about if rho is constrained to be close to 1?
For example:
. clear
. use http://www.stata-press.com/data/r8/school.dta
. local athrho=1/2*ln((1+.965)/(1-.965))
. constraint define 1 [athrho]_cons=`athrho'
. constraint define 2 [private]logptax=[vote]logptax
. biprobit priv vote logptax , const(1 2) nolog
Bivariate probit regression Number of obs =
95
Wald chi2(1) =
0.30
Log likelihood = -182.59333 Prob > chi2 =
0.5858
( 1) [athrho]_cons = 2.01395
( 2) [private]logptax - [vote]logptax = 0
----------------------------------------------------------------------------
--
| Coef. Std. Err. z P>|z| [95% Conf.
Interval]
-------------+--------------------------------------------------------------
--
private |
logptax | -.2258291 .4144413 -0.54 0.586 -1.038119
.586461
_cons | -.0285155 2.877341 -0.01 0.992 -5.668
5.610969
-------------+--------------------------------------------------------------
--
vote |
logptax | -.2258291 .4144413 -0.54 0.586 -1.038119
.586461
_cons | 1.967759 2.880274 0.68 0.494 -3.677474
7.612991
-------------+--------------------------------------------------------------
--
/athrho | 2.01395 . . . .
.
-------------+--------------------------------------------------------------
--
rho | .965 . -1
1
----------------------------------------------------------------------------
--
Likelihood-ratio test of rho=0: chi2(1) = 176.059 Prob > chi2 =
0.0000
----- Original Message -----
From: <[email protected]>
To: <[email protected]>
Sent: Thursday, April 15, 2004 3:06 PM
Subject: st: double-bounded model
> Have any of you estimated a double-bounded dichotomous
> choice model using Stata? For those not familiar with the
> concept it is a method used in contingent valuation to
> estimate willingness to pay. The log likelihood of the
> problem is similar to that of a regular probit model except
> that here you have 4 possible values for the indicator
> variable instead of two.
>
> It is actually quite similar to the bivariate probit model
> but constrained to the case where b1=b2 , std1=std=2 and
> rho=1. Because of the later constraint I can't use the
> biprobit command to estimate since the likelihood function
> is not valid at rho=1.
>
> Any help will be greatly appreciated
>
> Thanks,
> Alejandro
>
>
> Alejandro Lopez-Feldman
> Ph.D. Candidate
> UC DAVIS
> Agricultural and Resource Economics
*
* 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/
