Periods in stoch_simul

[fisico69500]

Dear All,

I would like to compare the standard deviations of my DSGE model with the ones in the data. Those data are given at quarterly frequency and are HP filtered with a coefficient lambda = 1600.

In the stoch_simul command, I introduce the option "hp-filter=1600". Should I have to set "periods = 320" (320 is the number of quarters of the data) ? Or is it not required to set the option "periods" in stoch_simul ? I've tried both and the results are very similar.

Thanks in advance for your reply.

All the best

[jpfeifer]

With the periods statement, you get simulated moments, without it, you get theoretical moments (unless order=3, in which case it is simulated as well)

[fisico69500]

Thank you for this quick reply. Since I want to compare the moments of my model to the ones of the data, do you mean that including the periods statement is more relevant ?

Thanks again.

[jpfeifer]

As you noted yourself, there should not be much difference. I personally prefer theoretical moments (unless you use order>1).

[fisico69500]

Thank you very much. Very clear, as usual.

Best