IRFs and steady state

[theta23]

Dear all,

my question is about the dynamics of impulse response functions. In a stochastic model following a shock all variables should earlier or later go back to the steady state, is it correct? I modified my model a bit (modified loan contracts from one-period to multiperiod) and now debt stock moves slowly (what I also expected). But in the long run it continues going down and seems not to go back to the steady state (I simulated irfs for 60 periods).
Does this mean that something is going wrong with my code?

Any advice will be highly appreciated!
Best,
theta23

[jpfeifer]

Use the check command to find out whether there is a unit root. Or use at least 500 periods to judge permanence.

[theta23]

Thank you! After about 100 periods it converges to the steady sate.
But I have another question - what do usually small zickzacks in irfs mean? For example, following a house price shock Inflation goes up, but it does in the 2-4 quarters some small zickzack a bit down and then gradually increases further before returning to the steady state.
Should I be concerned about this non-monotonicity?

[jpfeifer]

Only economic intuition can give you an answer. There are model where such behavior is immanent to the model and makes sense. But there are also cases where this suggests a mistake.