Simul command accuracy


I’ve been computing impulse-response functions with the command “simul” departing from the steady state of my model (a linear one). However,
when I shock the system in period 1, it gives me a different impulse response than if I shock it at time 2. In other words, the impulse response does not seem to be time invariant.

Moreover, I’ve been running my model using the coefficients of the policy and transition functions with a matlab program that I implemented. Nevertheless, the solution does not coincide with the one obtained using the “simul” command.

Has anyone run into similar problems when using “simul”?

Look forward to see some response to my question.


The “simul” command implements a perfect foresight (or deterministic) solver. It is perfectly normal that IRFs are not time invariant in this kind of setup, since the simulation uses a finite horizon.

I guess you are looking for the “stoch_simul” commad, which implements a rational expectations solver, which gives a recursive solution.