Dynare forums

Hi friends. I'm estimating a dsge-var by bayesian methods
to an small open emerging economy (Colombia) and comparing
the results of the fitted model against the traditional
models: VAR and DSGE, like frequently we seen in the
literature.

My question is how could evaluate the robustness of the
DSGE-VAR model to do a better evaluation of its properties
respect to the other models?. Are there references at
the modern literature to do that comparations?.

For example, i thought in split the total sample in subsamples
and evaluate the DSGE-VAR with the corresponding VAR and DSGE
to each sample and get the results (irf analysis, max kernel
and forecast out of sample)

Thanks for all. I will be grateful for your suggestions

I would not call this robustness.
You need to decide whether you want to do the comparison in sample or out of sample. In sample, you can compare the marginal data densities, which is the natural comparison for a Bayesian. For out of sample comparison, you could compare RSMEs for the forecasts. That is what the original Del Negro/Schorfheide (2004) did.