----- Original Message -----
From: "Ed Levitas" <[email protected]>
To: <[email protected]>
Sent: Tuesday, September 23, 2003 9:25 AM
Subject: st: Seemingly unrelated regression or similar estimation
> Statalisters,
>
> I have two equations:
>
> F = B(0) + B(1)var1 + B(2)var2 + U(1)
> G = Y(0) + Y(1)var1 + Y(2)var2 + U(2)
>
> F is continuous and G is a positive integer. Are there ways I can perform
> joint tests of coefficients across the two equations (e.g. B(1) = Y(1) = 0).
> This seems like a Seemingly Unrelated Regression. But given the difference
> in error distributions, I would think some major modifications are in order.
> Can anyone suggest recommendations (e.g. ado files), citations, etc.
>
> Thanks in advance for your time and consideration.
>
> -Ed
>
How about Seeming Unrelated Estimation -suest-
Example:
. use "C:\Stata8\auto.dta", clear
(1978 Automobile Data)
. qui poisson rep mpg weight for, score(A)
. est store poisson
. qui reg price mpg weight for, score(B)
. est store OLS
. suest poisson OLS
Simultaneous results for poisson, OLS
Obs = 74
------------------------------------------------------------------------------
| Robust
| Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
poisson_r~78 |
mpg | .0143303 .005316 2.70 0.007 .0039112 .0247494
weight | .0000841 .0000604 1.39 0.164 -.0000343 .0002025
foreign | .3581404 .095116 3.77 0.000 .1717164 .5445644
_cons | .5416934 .2650292 2.04 0.041 .0222457 1.061141
-------------+----------------------------------------------------------------
OLS_mean |
mpg | 21.8536 79.07015 0.28 0.782 -133.1211 176.8283
weight | 3.464706 .7614705 4.55 0.000 1.972251 4.957161
foreign | 3673.06 651.1297 5.64 0.000 2396.87 4949.251
_cons | -5853.696 3793.29 -1.54 0.123 -13288.41 1581.017
-------------+----------------------------------------------------------------
OLS_lnvar |
_cons | 15.32848 .2000052 76.64 0.000 14.93647 15.72048
------------------------------------------------------------------------------
. test [poisson_rep78]mpg =0
( 1) [poisson_rep78]mpg = 0
chi2( 1) = 7.27
Prob > chi2 = 0.0070
. test [OLS_mean]mpg =0, accum
( 1) [poisson_rep78]mpg = 0
( 2) [OLS_mean]mpg = 0
chi2( 2) = 7.35
Prob > chi2 = 0.0253
Hope this helps,
Scott
*
* 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/
