Hi,
you can *test* those restrictions after running a linear regression, but
if you want to *apply* those restrictions it's not a linear regression
anymore AFAICS. In a nonlinear model you could specify a coefficient as
exp(beta) for example, where beta is freely varying (and estimated) but
of course exp(beta) is always positive.
I would be reluctant to trust your results with this, however, if you
get negative values in OLS and if that happens to be the reason why you
want to apply such transformations.
cheers,
sven
Am 07.10.2010 11:10, schrieb BODENAN Jean-Christophe:
I would like to apply a restriction on the coefficients of a linear
regression (OLS). Indeed, coefficients will represent the yield of a
crop (soybeans, corn…) for a specific grade (my data) during its lifecycle.
How can I apply a restriction on coefficients: I want them to be all
positive and that the first coefficient (grade poor crop) is lower than
the second (grade fair crop), the second coefficient lower than the
third (grade good crop)…
Thanks in advance
**JCB** <
http://www.roquette.com/>
French gretl user __
---------------------------------------------------------------
The information in this e-mail and any attachment is intended for the
addressee(s) only. It may contain confidential and/or privileged
information. Access, disclosure, copying, distribution or reliance on
any of it by anyone else is prohibited and may be a criminal offence.
The contents do not represent the opinion of ROQUETTE FRERES or any of
its affiliated companies except to the extent that it relates to their
official business. If you received this in error, please notify the
sender and delete the contents from any computer on which they are stored.
---------------------------------------------------------------
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users