The unconditional variance decomposition of a linear state space model

Estimation
1

Hi Prof. Pfeifer and everyone,

This question is not directed related to Dynare but to the linear state-space model.

I have used Mdl = ssm(A,B,C,D); to create a linear state-space model in Dynare. I know that fevd(Mdl) can deliver the forecast error variance decomposition (FEVD), but it is conditional on the horizons.

I was wondering whether there is a command to calculate the unconditional variance decomposition for the linear state-space model. Or I should calculate it by myself? I appreciate any advice. Thank you in advance!

2

We know that with increasing NumPeriods the conditional FEVD will converge to the unconditional one (subject to some regularity conditions). Thus, using a large number like 500 is usually the way to go.

3

Sounds good! Thank you very much for your reply, Prof. Pfeifer!