Terminal condition in a deterministic simulation

Starting from the skeleton file fig1132.mod given in “Practicing Dynare” (homepages.nyu.edu/~ts43/research/AP_tom16.pdf), I computing a permanent deterministic shock. I use the 4.1 version.

But at the last date of simulation, the model is in a position that is very far away from the steady state. I do not get the intuition, it seems the simulation convergence is obtained. What can be the reason for that?

Permanent_shock_theta1.mod (21 KB)