[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: inequality constraints for regression models
Your constraint is not linear. -constraint-
should perhaps be called -linearconstraint-,
but I guess that would not be popular: nevertheless, the
fact that it is restricted to linear constraints
only is mentioned in the help.
You need to meet this constraint in other
ways, say by using logit ideas in a reparameterisation.
> we want to estimate a regression model (Zellners SUR model) with some
> restrictions on parameters and equal explanatory variables.
> Especially we
> want to say that the parameters have to be between 0 and 1.
> At the moment we wrote
> Constraint define 1 [eq1]variable1>=0 & [eq1]variable1<=1
> Sureg (yvar1-yvar3 = variable1), cons(1)
> The error message we get is
> 'not possible with test - r(131)'
> What did we wrong? Finally, it turns out that the question
> is: How to write the constraint?!
* For searches and help try: