Question about RBC_state_dependent_GIRF.mod

Stochastic simulations
1

Dear professor jpfeifer:

I am learning the RBC_state_dependent_GIRF.mod on your home page.

When I try to shut down all the shocks, I find that the “y_baseline” can not hold on steady-state.

In my understanding, when we shut down all the shocks, it has to be that the model dynamic maintain in steady-state path. Is that something wrong?

shocks_baseline = zeros(irf_periods,M_.exo_nbr);
y_baseline = simult_(M_,options_,starting_point,oo_.dr,shocks_baseline,options_.order);

BGGrealt.mod (12.9 KB)
linear.m (1.1 KB)

Best wishes

2

No, that is true only at first order. In that case, the simulation will settle at the deterministic steady state. At higher order, it will settle at the stochastic steady state where precautionary behavior plays a role.

3

Thank you for your patient reply, professor jpfeifer.

Actually, I don’t quite understand what stochastic steady state means. Do you have any recommended materials for me to study?

Additionally, I would like to explore the state dependence characteristic of fiscal multipliers about business cycle in a new Keynesian framework with financial accelerator model of BGG.

Is this achievable ?

The difficulty is that in the simulating scheme set by RBC_state-dependent_GIRF.mod, no matter what random shock combination I pass in, the output will quickly become negative. This makes it impossible for me to detect the state dependence property of the multiplier. Is it possible that it’s caused by poor parameter settings ?

Best wishes

4
  1. Due to nonlinearities that cause precautionary behavior, the fixed point of a simulation at higher order even without shocks will generally differ from the deterministic steady state. The point where such a system settles is called the stochastic steady state.
  2. Which variable becomes negative? Also, did you take into account that shock scaling matters at higher order, i.e. you cannot multiply shock standard deviations by 100 to achieve percentages.