Third order solution with skewed shocks

Stochastic simulations Semistructural Modeling
1

Hi,
I’m trying to study an economy subject to a skewed-normal shock using the “unstable-8” version of Dynare (with skew e=-0.99 in the shocks block).
It occurred to me that there is not equivalent of ghs2 (a ghs3) for the shift due to skewed innovations.
It seems to me that it is not embedded in g_0 either. Skewed innovations would in general imply an additional shifter.
Am I looking in the wrong places?
Thanks and best,
Gianni

2

Indeed, the perturbation solver does not take into account asymmetric shocks, so you need to compute these matrices by hand as I have done here:

At first and second order this is not a problem, but at third and higher orders this is not correct. At some point we will tackle this in Dynare, but it is not a high priority.

3

Dear Willi,

many thanks for your response.
Best,
Gianni

4

Hi,
I’ve forked Dynare and let codex fix the bug (giannilmbd/dynare).

Now if you add skewness, it returns the correct stochastic steady state (ghs3 and g_0).
I’ve tested it with a made-up process

v = e1^2;
u = e1^3;
y = v(+1) + u(+1);
junk = rho*junk(-1);

It seems to be working.
Best,
Gianni

5

Thank you very much. This is a good start. I opened a ticket at Account for skewness in higher order decision rules (#2029) · Issues · Dynare / dynare · GitLab.