I am using Stata 9 and am estimating a system of simultaneous equations using 3sls. I see it is possible to compute iterative 3sls , however, I am not sure what is to be gained from doing so. I only have an older version of Greene at home (i.e. 1993, second edition) but on page 612 it states:

"It is possible to iterate the 3SLS computation. However, unlike the seemingly unrelated regressions estimator, this does not provide the maximum likelihood estimator, nor does it improve the asymptotic efficiency."

This is all my version of Greene states on iterative 3SLS, therefore, I'm not sure what is to be gained from using it. What are the advantages and disadvantages from using 'iterative 3SLS' over '3SLS'.

Thanks in advance for any assistance.

Kevin

Vadim Grinevich

As you know when you run a constrained regression, the output statistics does not include R squared. However, I do need at least some indication of it in order to compare my results with those from other studies using similar sort of constraints. To sum up, is there any way to get indication of R squared for constrained regression (cnsreg command)

For two-stage least-squares (2SLS/IV/ivreg) estimates, why is the R-squared statistic not printed in some case?

For two-stage least-squares (2SLS/IV/ivreg) estimates, why
is the Model Sum of Squares sometimes negative?

For two-stage least-squares (3SLS/IV/ivreg) estimates,

why are the R-squared and Model Sum of Squares sometimes negative?
http://www.stata.com/support/faqs/stat/2sls.html

Dear Stata Users,

As a response to the queries on demand systems estimation,
Prof.Jenkins suggested Poi (2002) and the set of ado files under
'demand', written by Dr.Poi. I had already gone through this and tried
to use it. These are for demand systems with 4 equations.
I have in fact tried working on these ado files, but could not figure
out how can I extend this for 12 equations in my case, and N equations
in general! All that I coud find was that some changes are required to
be done in the construcion of covariance matrix in quaids_delta.ado.
I'm not able to do this.

As I thought this may not be easy, I started using 3SLS as well as SUR
and I found some other problems in these as well, which were discussed
in statalist a few hours ago.

Could any one give some suggestions about either quaids (mentioned in
this post) or 3SLS/SUR for demand system estimation?


Badri

[此贴子已经被作者于2006-4-21 4:57:30编辑过]

On my user site, I have posted a new command called quaids_delta2 so that
the method I outlined in my 2002 Stata Journal article can be applied to
demand systems with arbitrary numbers of equations.

Because quaids_delta2 uses Mata, Stata 9 is required.

The help file accompanying the command contains an example where I fit a
7-equation demand system.

To install, in Stata type

net from http://www.stata.com/users/bpoi
net install quaids_delta2
help quaids_delta2

Bug reports are most welcome.

Enjoy.

-- Brian Poi
-- bpoi@stata.com

Please read the documentation. The option sure on reg3 specifies the
estimation method. If you say sure, you are asking for sure, and sure
involves only OLS (strictly, GLS) techniques. It will not do anything
like IV. Options 2sls and 3sls will utilize the IV estimator.


Kit Baum, Boston College Economics
http://ideas.repec.org/e/pba1.html