Why the IRFs are so abnormal? Is it either normal or not?

Stochastic simulations
2

Those IRFs indeed look weird. They look like oscillations due to a complex eigenvalue. It’s either a timing problem or a wrong parameter setting (often in feedback rules).