Restrict Endogenous Variable to be Positive

Stochastic simulations
1

Hello all,

I would like to simulate a stochastic model that restricts one of the endogenous variables to be positive. My understanding is that the extended path command is the only current way to accomplish this in Dynare without completely losing precautionary actions made by agents when the lower bound is hit (i.e. what you can do using OccBin). I have a few questions about the process.

  1. Can this be done with a fully nonlinear model, or will I have to write a separate .mod file with the linearized version? I know in the deterministic case, you must linearize the model.

  2. Whether I have to linearize or not, is the process as simple as replacing everywhere I see the variable I want to restrict to be positive, say x, with max(x,phi) where phi is some lower bound?

Thanks!

Ben

2
  1. Yes, at the current stage, stochastic extended path is the only way.
  2. You do not need to linearize, neither in the stochastic nor the deterministic case
  3. Yes, it should be that simple
1 Like