IRF third order with Dynare 4.4

I notice that in the new version 4.4 the simulation for third order approximation uses Andreasen et al (2013) prunning procedure. Does anybody know if the computation of impulse response functions for third order approximation has been modified as well, either by computing the GIRF of Andreasen et al (2013) or by using the simulation routine to compute it? I couldn’t find this out by reading the manual. Thanks!


Not yet. I am working on GIRFs. Note that the GIRF definition in Andreasen et al. is essentially equivalent to the simulation based approach Dynare has always used at higher order. The only advantage of the Andreasen et al. GIRFs is that they are analytically computed while the simulation based ones may require really many replications.