Dynare Simulation and Guess Sensitivity

Stochastic simulations
1

Hello everyone. I am currently simulating a two country EZ model. However I noticed that identifying a steady state is sensitive to my initial guess. For example in the following mod file:

OpenEconomy.mod (1.9 KB)

I made a specific guess for the wealth-consumption ratios for both countries (vh, vf). However if I change the parameter delta from 0.98 to 0.99, this guess no longer works and dynare doesn’t compute the steady state. If I change the guess through the following mod file:

OpenEconomy2.mod (1.9 KB)

Dynare computes a solution again. I am a little confused by what this means. Should I interpret this as suggesting multiple equilibria that is sensitive to the parameters?

2

The only sure way to know whether there is a problem of multiple steady states is if different starting values for the same parameters lead to different steady states. Numerical steady state finding breaking down for the same starting values for different parameters can easily happen.

In any case, with strong nonlinearities like EZ, you should try to provide an analytical steady state.