You are correct. Stata only accepts linear
constraints, and inequalities do not qualify.
Sometimes, in other contexts, you can get what you
wish by a reparameterisation of the model, for example, by
using logarithms to ensure that a parameter of interest
remains positive. I've no idea if that makes sense here.
Another possibility is that your model is just not suitable
for your data, so that no amount of bending,
twisting or pushing can improve much on what you get unconstrained.
Nick
[email protected]
Said at [email protected]
> When estimating ARCH/GARCH models using maximum likelihood methods,
> STATA performs unconstrained optimisation and therefore the results
> returned contain negative parameter estimates and are therefore not
> very useful. How can I impose inequality constraints on the
> optimisation process such that it performs constrained optimisation,
> the 'constraint define' command seems to only accept equality
> constraints, when I try to define inequality ones I get error
> messages.
*
* 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/
