I have formulated a DSGE model and a DSGE-VAR model, which uses DSGE cross-equation restrictions as prior, however, there are some differences between impulse responses generated by the DSGE model and impulse responses generated by the associated DSGE-VAR model, there are differences between the two sets of impulse responses, although these differences are not significant, I am wondering that does no significant difference in the two sets of impulse responses indicate no serious model misspecification in the underlying DSGE model? or even a little but insignificant differences in two sets of impulse responses may still signal potential model misspecification in DSGE models?
Thank you very much and look forward to hearing from you.
- Yes, the differences between the two indicate whether the data reject the cross-equation restrictions implied by the DSGE model.
- Keep in mind that all models are necessarily wrong. Thus, small differences are usually no reason to worry.