In a fairly simple set up I am getting a strange result – an impulse response function that can only be described as “bumpy” — this is different than the normal smooth IRFs I am used to.
The set up is a normal Sims RBC model. (See attached code) bumpty_test.mod (1.6 KB)
I made only two small changes to the standard set up. The first is the introduction of unmeasured investment. The unmeasured investment, ui, is computed as the sum of a variable ‘unmeasured investment share’, and actual investment, i. The unmeasured investment share is the fraction of investment that is not measured, and ranges from 0 to 1. The mean of this variable, avguis, I set to .2 . The two additional equations are thus
ui = uis*i;
uis = avguis*(1-rhou) + rhou*uis(-1) + eu;
I compute the IRFs with order 2, and uis shows a smooth descent with a shock to eu, but ui does not – it is this that produces the `bumpty’ IRF.
Does anyone know what is going on?
Many thanks in advance, and sincerely,