bashtage
Adding positivity constraints in both SUR and Panel models.
Sometimes negative constraints in models do not make sense and I cant find any way to add inequality constraints in my models.
Inequality constraints are not very standard, and inference, when constraints bound, is not simple. You could always do model selection to remove negative coefficients.
I understand. Currently I am running a SUR model with 50+ variables in 7 equations. I have to remove each negative coefficient one at a time from the ordered dictionary and as soon as one of the variable is removed, other parameters change a lot too, implying that removing all of them at once may lead to losing of some info & still leaving me with some negative coefficients.
Could you elaborate on how one can do model selection to remove negative coefficients? Also, do you think it can be helpful in my situation?