How well do DSGE models with real estate and collateral constraints fit the data?

Abstract. Not so well. We reach this conclusion by evaluating the empirical performance of a benchmark DSGE model with real estate and collateral constraints. We estimate the model from U.S. data using Bayesian methods and assess its fit along various dimensions. We find that the model is strongly rejected when tested against unrestricted Bayesian VARs and cannot replicate the persistence of real estate prices and various comovements between aggregate demand, real estate prices, and debt. Performance does not improve with alter-native definitions of real estate prices, estimation samples, or detrending approaches. Our results raise doubts about the ability of current DSGE models with real estate and collat-eral constraints to deliver credible policy insights and identify the dimensions in need of improvement.
JEL Codes: C52, E32, E44.
Keywords: real estate; housing; DSGE models; collateral constraints; model evaluation.