|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: simultaneous equations in panel data
<>
This is a common misconception about "full information" estimation
techniques such as 3SLS and FIML. The only thing to be gained from
applying a systems estimator is efficiency, and that will arise iff
the cross-equation error correlations are important (as in -sureg-).
(One can also apply cross-equation constraints, if that is of
importance, but you didn't mention that). If some of your simultaneous
equations contain more than one endogenous variable, then use a
"limited information" estimator and estimate each equation separately.
If each equation's estimates are consistent, you're all set. One down
side of a "full information" estimator is that misspecification of any
equation will pollute estimates of every equation. That won't happen
with limited info techniques.
In your context you could use -xtivreg- or the user-written -
xtivreg2-, which has better diagnostics for the class of xt models it
implements.
Kit Baum, Boston College Economics and DIW Berlin
http://ideas.repec.org/e/pba1.html
An Introduction to Modern Econometrics Using Stata:
http://www.stata-press.com/books/imeus.html
On Feb 5, 2009, at 02:33 , Fardad wrote:
Hi. I do have a set of 9 equations which should be simultaneously
estimated, since some of the equations' independent variables are
dependent for others. Yet, I don't deal with cross-section data but
panel data. That means all of my equations have both a firm and time
dimension.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/