Variance decomposition/impulse responses at posterior mode or posterior mean is better?

Estimation ML/Bayesian estimation
1

Dear Johannes,
First thank you for your previous guidance, I am grateful.
For historical shock decomposition/impulse responses/forecast error variance decomposition, should we based on posterior modes or posterior means? I have seem literature use posterior modes, and could I know which one is better, posterior modes or posterior means for variance decomposition and impulse response analysis.
Best regards,
Jesse

2

I am not aware of any general result. It’s a matter of taste. Usually, I would prefer the posterior mode if it was the true mode, because it is one parameter draw, i.e. one model while the posterior mean is based on an average over several models.