Dear Statalisters,
I have been going through William Gould's very instructive guideliness on what
is a Chow test and how to construct such a procedure. My question is how to
construct a Chow test for a panel model?
I have the model:
.y_it = a + bX_it + C_i + u_it (I'm interested only in BE effects)
And I estimate the model using three panel estimators:
(1) y_i = a + bX_i + C_i + u_i
(BE estimator: y_i, X_i, u_i are averaged over t)
(2) (y_it - y_i - y) = a + b(X_it - X_i - X) + C + (u_it - u_i - u)
(WE estimator: y, X, u are averaged over it, and C averaged over i)
(3) y*_it = y_it - Wy_i - y
(RE estimator: W is Swamy and Arora (1972) weight for RE)
Now, I want to test whether there are two structural braks within t. Is it
possible to use William's notes
(http://www.stata.com/support/faqs/stat/chow3.html) on how to construct the
equivalent of a Chow test with regard to the three estimators mentioned above?
That is to say, if t=1,2,3,...,15, first to estimate the model for:
t1i=1,2,..,5
t2i=6,7,..,10
t3i=11,12,..,15
and then use -test- to run the equivalent procedure as for Chow test? Is this
the correct way to run this test? Or, since this is a panel model, it implies
something else that I cannot see?
all comments are welcome,
many thanks
Dimitris Christodoulou
--
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/