Identifying the persistence of IRFs

Hi everyone,

I am making a model that I estimate using Bayesian techniques. However, by looking at the IRFs, I observe a lot of persistence. In fact, before tweaking, the estimate gave me AR (1) coefficients very close to 1. But once it’s fixed, the coefficients look good, but the IRFs are very persistent.

My question is: Is it possible to identify where that persistence comes from? I understand that an eigenvalue equal to one indicates a unit root and that very close to 1 indicates a lot of persistence. There are no eigenvalues equal to one in my model, but there are many close to one, for example, 0.99.

Thank you for your time on this.

IRFs are nonlinear mappings from the parameters. Usually, the only way to proceed is to do a sensitivity analysis of varying parameters.