Is it necesary to estimate std. devs of shocks?

Dear all:

I am currently working with a model explaining economic volatility duing COVID-19. I wonder if my focus is on shock decomposition of key variables and I do not really care about the size of shocks, is it necessary to estimate the standard deviations of shocks? In my understanding, I can just normalize them to unity and the smoother will tell my about how large the shocks are in each period, and I can obtain identical results of shock decomposition as the experiment treating the std devs as parameters to be estimated.

Is my understanding correct? Thanks in advance.


No, the smoother will assign shock sizes based on the likelihood of a particular shock arising. The likelihood of a particular shock size will in general depend on the standard deviation. For that reason, normalizing shock standard deviations will not work.

I got it. Thanks for your reply professor:)