IRF of the third-order Taylor approximation

[gabrielbruneau]

Hello,

I am confused when I try to understand the inputs and outputs of the third-order Taylor approximation using Dynare.

First, using other MATLAB codes for the third-order Taylor approximation (based on Schmitt-Grohe and Uribe codes), I have to provide the skewness perturbation parameter's matrix before the approximation. Where do I provide it in Dynare? If we don't have to provide it, where Dynare takes the information equivalent to variance-covariance matrix for the skewness-coskewness?

Second, concerning the output, how can I calculate the IRF (or simulate variables) from the output g_0, g_1, g_2 and g_3, and where the skewness correction is included in these matrices?

Thank you very much,
Gabriel

[SébastienVillemot]

Dynare does not need higher moments of shocks (skewness in the case of 3rd order approximation), because it makes an hypothesis of normality. Hence only the variance-covariance matrix of the shocks is necessary.

The content of the g_0, g_1,... matrices is described in the reference manual for Dynare 4.1.3, in the section about "stoch_simul".

Best,