Comparing IRFs of DSGE and DSGE-VAR in dynare

[Ouz]

Dear Johannes,

As I understand it, the 'bvar_irf' option used under DSGE-VAR estimation in dynare produces figures that compare the IRFs of the DSGE model (solid line) and those of the DSGE-VAR(lambda) (the dotted lines)... I was wondering if there is any way we can know the EQUATIONS producing them?

I understand that the IRFs can be produced with the state-space representation of the DSGE model (using .ghx and .ghu) and I managed to retrieved the solid line. But how can I manually compute the IRFs of the DSGE-VAR(lambda), i.e., any of the solid lines? Put differently, does dynare provide a way that allows users to do simulation or forecast using the estimated DSGE-VAR(lambda)?

I have searched through the forum and manual but I could find little information about using the DSGE-VAR(lambda) model and its posterior estimates, though there are quite a few papers out there where people compare the forecast ability of DSGE-VARs and DSGE model etc.. Am i misunderstanding something here? (Assuming they also use dnare), how do they get the forecasts then, say, from a DSGE-VAR(0.5)?

Any advice would be very highly appreciated.

Thanks very much.
Zhirong

[jpfeifer]

Please have a look at the manual of the unstable version. That version provides the required posterior estimates of the DSGE-VAR (stored in

Code: Select all

oo_.dsge_var.posterior_mode

). The code generating the DSGE-VAR IRFs can be seen in

Code: Select all

PosteriorIRF_core1.m

[Ouz]

Dear Johannes,

Thanks very much for your help. This has been extremely helpful!

Kind regards,
Zhirong