Dear Dr. Kit Baum,
Thank you very much for the advice. In fact, in my list of
constraints, adding-up requires only very few constraints. The rest
are other ones like symmetry, homogeneity, and functional-form-related
constraints. Of these, to impose homogeneity, I need all of the
coefficients of all equations. Further, I require all of them in any
case, as my study is about the entire demand system. Even if I assume
that I can recover the 'n'th coefficient based on the constraints and
'n-1' other coefficients, I'm not sure how would it be possible to
construct the covariance matrix required to infer about significance.
In addition, I guess the results would depend on which equation is to
be dropped, which is difficult to decide in my demand system. In this
case, could you suggest some way of constructing the covariance matrix
and a remedy to avoid the problem arising from the dependence of
results on the equation dropped? Alternatively, would using 3SLS
instead of SURE take care of the problem, ie, can I impose all
constraints (maybe other than "adding-up") while including all
equations in this 3SLS?
In fact I did exactly this (ALL Equations using 3SLS, mentioning all
constraints including "adding up"!) and I am getting some good results
too! Would this methodology be consisitent with the theory?
Badri
*
* 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/
