Thinning Factor for Constant Coefficient SBVAR


I am trying to estimate a simple constant coefficient SBVAR using the sbvar command. The code is up and running.

I was able to change the number of draws and burn-in via editing the master code.

However, is there anyway to thin the number of draws in the Gibbs sampling algorithm?

Thanks in advance!

constantSBVAR.mod (403 Bytes)

Code attached.

The data file is missing to run your code.

Sorry about that! Hopefully it uploads this time.

USmacroData400.mat (6.8 KB)

You would need to modify ms_mardd accordingly. But the question is for which statistic would you need the thinned draws?

That is correct.

Which variable would I need to look for in the ms_mardd program?

Also, which of the codes generates the IRFs for the sbvar command?


Are there any updates?

As far as I can see, the Gibbs part is only used to compute the marginal data density. The IRFs are governed by msstart2