request help on the mechanism behind optimal policy

[ahuca2008]

Dear Dr. jpfeifer:
Thanks for reading . I don't understand what is the mechanism behind calculating the best values for the coefficients of monetary policy rule under the optimal policy . Should I minimizing the loss function subject to the structrual equations for given policy?

[jpfeifer]

What exactly do you not understand?

[ahuca2008]

dear Dr.jpfeifer , I want to compute the policy parameters using "pencil and paper" ,what should I do? Should I maximaze Vt=Et*U(c,h) with the constraints such as Euler Equation ,Resource Constraint , and then compute the policy parameters in steady state of the Ramsey economy ?Thank you for your time.

[jpfeifer]

That depends on the type of optimal policy you want. If you want optimal policy under commitment, then we are talking about standard Ramsey policy. How to do this is welll-documented in e.g. http://www.nber.org/papers/w6891

[ahuca2008]

That depends on the type of optimal policy you want. If you want optimal policy under commitment, then we are talking about standard Ramsey policy. How to do this is welll-documented in e.g. http://www.nber.org/papers/w6891

That is what I want to do.Thank you so much.

[ahuca2008]

Dear jpfeifer
Thanks for your helping. I have read the paper you gave me,but I still cannot find the method how to compute the optimal interest rate rule coefficients under commitment.I read the paper that"Optimal simple and implementable monetary and fiscal rules" written by Grohe and Uribe,they find the optimal interest rate rule coefficients that maxmize welfare by matlab .Is there any way to compute the optimal policy parameters by "pencil and paper"?Must I do a loop with different values of the parameters to see which ones give the highest welfare?
thank you for your time.

[jpfeifer]

As I said, you have to decide what you want. The SGU-paper you refer to uses optimal simple rules (OSR command). That is different from the fully state-contingent optimal policy in a Ramsey framework (Ramsey command).
Regarding your pencil and paper question: it might be possible when you perform some approximation to the fully nonlinear model. But using numerical techniques is often easier/the only way.