Re: [Gretl-users] Different coefficient in ISUR
On Sun, Jan 8, 2012 at 7:09 AM, caspisun(a)gmail.com <caspisun(a)gmail.com> wrote:
I have estimated the AIDS(almost ideal demands system) and Rotterdam demand
system using Iterative SUR(Seeming Unrelated Regression) and 3SLS.
Recently, I have found that the estimated result from "Gretl" are quite
different from the Other package (Such as R, RATS)
In case of R and RATS...the estimated coefficient are same exactly.
after I imposing the restriction on the model...
the estimated coefficient are exactly same....
But in CASE of GRETL....the estimated coefficient are different from above
software.
In my thought, it should be same if the estimation method is right.
Of course, I have double check the model I estimated...No problems on it
Any suggestion?
or can you give answers..why this kinds of situation happened?
or is it error in the method?
We can't answer this without seeing your model and data. Perhaps you can
post a script specifying the model. If the data file is too big to
post here, you
can send it to me off-list (cottrell at wfu.edu).
Allin Cottrell