Validating DSGE models with VARs and Large-Dimensional Dynamic Factor Models

A popular validation procedure for DSGE models consists in comparing the structural shocks and impulse-response functions obtained by estimation-calibration of the DSGE with those obtained in a SVAR identified by means of some of the DSGE restrictions. The variables included in the VAR are all or part of the variables in the DSGE, usually augmented with measurement errors. The paper argues that this practice can be seriously misleading. For generic values of the parameters of the DSGE, the shocks estimated in the VAR are not “made of” the corresponding structural shocks plus measurement error. Rather, each of the VAR shocks is contaminated by non-corresponding structural shocks. We argue that Large-Dimensional Dynamic Factor Models are free from this drawback and are the natural model to use in validation procedures for DSGEs. JEL classification: C32, C55, E32.