Historical variance decomposition

Dear all,

I have a couple of questions regarding the “shock_decomposition” command.

  1. The reference manual (v. 4.3.3) states that one of the options for the command shock_decomposition is “shocks = (VARIABLE_NAME [VARIABLE_NAME …] ; VARIABLE_NAME [VARIABLE_NAME …] … ])”. I am not able to use this option. In particular, after the “estimation(…);” command I type

Here the shocks are e1, e2 and e3. What is wrong with this specification? The command works fine without the “shocks=…” option.

  1. I would like to compute historical variance decompositions where some shocks are combined. Suppose that e1 and e2 are supply shocks, while e3 is a demand shock. Can I then present variance decompositions where I only distinguish between supply and demand shocks, i.e. where one of the shocks is e1+e2?

  2. Suppose I want to obtain historical variance decompositions using previous estimation results (i.e. outside the mod-file). How can one do this with the shock_decomposition function? It takes 4 arguments as input (shock_decomposition(M_,oo_,options_,varlist)). I understand that the inputs are stored in “mymodel_results.mat”, but not how to call them correctly in shocks_decomposition(). Can someone provide a simple example?

Thanks in advance!