Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: sureg, isure


From   "D.Christodoulou" <d.christodoulou@bangor.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   st: sureg, isure
Date   Mon, 24 Feb 2003 23:56:41 +0000

Dear Statalisters,

I have already listed this query but didnt receive any reply, so I will try to
be more clear. I'm using STATA 7 on Windows 98 and I'm dealing with panel data
and Seemingly Unrelated Regressions (SUR).

I use the sureg command, which in turn uses the Feasible Generalised Least
Squares (GFLS) algorithm described in Greene(200)"Econometric Analysis" to
estimated the system of equations. One of the options of the sureg operator is
to estimate the system using iterations ( sureg [system], isure) to
convergence.

I have two questions for those who dealt with SUR in the past, or for anyone
who is keen on giving me a tip:

(1)'isure' iterates the system and doesnt stop until it reaches to total
convergence or failure to converge the covariance matrix of the disturbances.
Is there a way to tell STATA to iterate only once and give me the results of
only one iteration? Eviews can do that but it uses the Gauss-Seidel
methodology to estimate SUR and I prefer STATA's treatment.

(2)What are the implications(advantages?) of using iterations (total
convergence=maximum likelihood) over GFLS? Greene(2000) says that MLE enjoys
no advantages over GFLS in its asymptotic properties. Can someone make this
more clear to me,please.

I appreciate any help,
best regards to everyone
Dimitris

--
This mail sent through http://webmail.bangor.ac.uk
*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index