Moment Matching for Bayesian Inference in the Baseline New Keynesian Model

Abstract The sparsity and small size of economic samples often obscure the identification of parameters in macroeconomic models. These identification problems originate from the poorly defined mapping between a structural model and reduced-form parameters. Hence, researchers rely on prominent estimation methods, such as Bayesian approaches, which require sound knowledge of prior distributions on parameters. These approaches, however, are characterized by a flat likelihood and/or a posterior distribution driven mainly by prior information. To develop a robust inference framework for macroeconomic data, we apply approximate Bayesian computation combined with the choice of specific moment conditions. This estimation approach circumvents the direct evaluation of high-dimensional likelihood functions and can mitigate parameter-identification problems when combined with bootstrap-based selection criteria. Our estimation method is successfully applied to a hybrid version of the baseline New Keynesian model.