OLR/OSR Syntax
Posted: Wed Mar 21, 2007 3:53 pmHi Michel & Dynare friends,
Pardon the embarrasingly simple nature of these questions. Please assume I illiterate with respect to codespeak (which is not far from the truth).
I have a linear model
model (linear);
...
end;
Problem #1)
Withing this linear model section I have a simply linear policy rule where the instrument is named "int" and the parameters in the rule are "a," "b," and "c."
intr=a*intr(-1) +(1-a)*(b*pie(+1)+c*ygap(+1))+eps_intr;
Instead of assuming a,b would like to find out the optimal weights of a*, b* given the parameterization of the model.
How do I provide the param, weights, osr statements? Clearly the following attempt was wrong!
osr_params a b;
optim_weights;
var intr(-1), pie(+1), ygap(+1);
end;
osr intr;
In the initial parameters statement, do I include the parameters that are also in the osr_params statement? Do I set values to some inital value?
What do I do, if anything with the intrest rate shock? Do I first find the optimal values and then tag the shock on for stoch_sim?
Case #2)
what to do with a linear-quadratic rule?
intr=a*(intr(-1))^2+b*(pie(+1))^2+c*(ygap(+1))^2+eps_intr;
Case #3) Assuming I figure out how to do #1 and #2, are there additional steps/considerations to go with the olr for optimal commitment?
Case #4), how can one program in DYNARE optimal discretion, whether the societal loss function may be different from the target regime for which optimal weights are sought?
Thanks! I really do appreicate any help that can be provide, particularly any example code out there.
Cheers, Peter
P.S. I attached the simple NK model code that wish to play with.
Pardon the embarrasingly simple nature of these questions. Please assume I illiterate with respect to codespeak (which is not far from the truth).
I have a linear model
model (linear);
...
end;
Problem #1)
Withing this linear model section I have a simply linear policy rule where the instrument is named "int" and the parameters in the rule are "a," "b," and "c."
intr=a*intr(-1) +(1-a)*(b*pie(+1)+c*ygap(+1))+eps_intr;
Instead of assuming a,b would like to find out the optimal weights of a*, b* given the parameterization of the model.
How do I provide the param, weights, osr statements? Clearly the following attempt was wrong!
osr_params a b;
optim_weights;
var intr(-1), pie(+1), ygap(+1);
end;
osr intr;
In the initial parameters statement, do I include the parameters that are also in the osr_params statement? Do I set values to some inital value?
What do I do, if anything with the intrest rate shock? Do I first find the optimal values and then tag the shock on for stoch_sim?
Case #2)
what to do with a linear-quadratic rule?
intr=a*(intr(-1))^2+b*(pie(+1))^2+c*(ygap(+1))^2+eps_intr;
Case #3) Assuming I figure out how to do #1 and #2, are there additional steps/considerations to go with the olr for optimal commitment?
Case #4), how can one program in DYNARE optimal discretion, whether the societal loss function may be different from the target regime for which optimal weights are sought?
Thanks! I really do appreicate any help that can be provide, particularly any example code out there.
Cheers, Peter
P.S. I attached the simple NK model code that wish to play with.