I would like to ask if there are any methods or criteria to judge whether my DSGE model is consistent with the real data. Like the VAR method, right? Compare my model with the impulse response graph from the empirical analysis of VAR?
People in the literature usually compare either second moments or IRFs as you suggested.