A Monte Carlo Analysis of the VAR-Based Indirect Inference Estimation of DSGE Models
In this paper we study estimation of DSGE models. More specifically, in the indirect inference framework, we analyze how critical is the choice of the reduced form model for estimation purposes. As it turns out, simple VAR parameters performs better than commonly used impulse response functions. This can be attributed to the fact that IRF worsen identification issues for models that are already plagued by that phenomenon.