Strange shock decomposition

Dear all,

I have a problem with the shock decomposition. For some reason, the model explains the evolution of my variable with huge positive and negative shocks, especially at the end of the sample period (see attached figure). Does anybody know why this is the case?

Any help or hint is greatly appreciated,
all the best,
shock_decomposition.pdf (26.1 KB)

I am afraid there is no answer to this. The model just assigns the highest likelihood to this outcome. Maybe your estimated model got stuck before it converged to the correct values.