Clarification for MS-BVAR "parameter_uncertainty" option


Can someone please explain the parameter_uncertainty option under the MS-BVAR ms_irf command? Specifically, does enabling this option allow Dynare to compute Generalized Impulse Response Functions for an MS-BVAR?


For the uncertainty bands in a given model, there are two sources of uncertainty: disturbance uncertainty from unknown shocks arising in the future and parameter uncertainty from not knowing the true value of the parameters. With the parameter_uncertainty you will consider both sources of uncertainty. The term GIRF is not fully defined without specifying what you are integrating out, so it’s hard to know whether it’s what you are looking for.

Hi Professor Pfeifer,

Thank you again for that explanation.

I am trying to compute GIRFs for an MS-BVAR similar to Karame (JEDC, 2015). The GIRFs I am interested in would allow for dependence on initial conditions, future shocks, and future regimes.

Is parameter_uncertainty an appropriate option to use in this case?

Also, after reading the manual again I saw the regimes (not the regime) option. Can you provide more details on this option? Would enabling the parameter_uncertainty and regimes options together produce the GIRFs I am seeking?

Edit: I went back into Zha’s original code and the regimes option assumes each regime is permanent. So using this option will not produce the GIRFs I am looking for.

@frederic.karame Do you know how this part of the code works?

Unfortunately no. Houtan was in relation with D. Waggoner if I remember well.