Fellow listers,
I am writing a procedure for Pooled Mean Group estimation for
heterogeneous panels (a la Pesaran, Smith, and Shin). I have all the
estimation figured out, but I would like some input regarding user
preferences of model setup.
Basically, I am interested to know if potential users of such a procedure
prefer "long" or "wide" data format for this type of estimation.
The give and take:
Option 1: "Long" data: the procedure would seem much more like Stata's
other xt estimation commands. More importantly, unbalanced panels can be
easily handled. However, cross equation restrictions will be unavailable
to the user.
Option 2: "Wide" data: it will be trivial to allow users to impose cross
equation (that is, cross panel) restrictions of parameter estimates.
However, unbalanced panels will not be allowed.
Please note that with either option the user will be able to perform
cross-equation tests -- but only with Option 2 will the a priori
restrictions be easily handled.
My underlying question is which is more valuable to users: the ability to
handle unbalanced panels or (in the context of these heterogeneous models)
allow for cross-panel restrictions?
I have my preference, but I would be interested to hear from any
enlightened and opinionated users!
Edward F. Blackburne III, PhD
Associate Professor
Economics and International Business
Sam Houston State University
[email protected]
936-294-3934
*
* 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/
