Page 1 of 1

Coding a nonlinear interest rate schedule

PostPosted: Thu Jun 29, 2017 3:05 pm
by parantap
Hi
I am interested in setting up a nonlinear interest rate schedule in dynare. Here is the problem. I want to formulate an interest rate schedule which takes different values for different levels of deposit. If the deposit level is lower than Dbar, the interest rate is i_1 and if it is higher Dbar the interst rate i_2. I would like to make it more general three tier or four tier system where there two or three such thresholds.
Any idea how I can do this? I foresee multiple steady states for interest rate which could be problematic in implementing in dynare. Any help or any reference to past coding will be extremely helpful.

PB

Re: Coding a nonlinear interest rate schedule

PostPosted: Mon Jul 03, 2017 1:33 pm
by annawarren
parantap wrote:Hi
I am interested in setting up a nonlinear interest rate schedule in dynare. Here is the problem. I want to formulate an interest rate schedule which takes different values for different levels of deposit. If the deposit level is lower than Dbar, the interest rate is i_1 and if it is higher Dbar the interst rate i_2. I would like to make it more general three tier or four tier system where there two or three such thresholds.
Any idea how I can do this? I foresee multiple steady states for interest rate which could be problematic in implementing in dynare. Any help or any reference to past coding will be extremely helpful.

PB


The common problem

Re: Coding a nonlinear interest rate schedule

PostPosted: Mon Jul 03, 2017 1:59 pm
by jpfeifer
If you are thinking about stochastic simulations, this will not be feasible due to the discontinuities introduced by the steps.

Re: Coding a nonlinear interest rate schedule

PostPosted: Wed Jul 12, 2017 10:23 pm
by parantap
No. I will a deterministic simulation. Is there any code which can guve me some guidance?

Re: Coding a nonlinear interest rate schedule

PostPosted: Thu Jul 13, 2017 5:26 am
by StephaneAdjemian
Stochastic simulation could be done with extended path (a sequence of perfect foresight simulation). The problem here is not the non differentiability introduced by the steps. The issue is that depending on the regime, the steady state of your model (the target of the perfect foresight solver) moves. This is not possible with regular Dynare, You have to hack the Matlab code to write a solver specific to this model.

Best,
Stéphane.