Structural Vector Autoregressive Models and Monetary Policy Analysis

Abstract

In this paper, the structural vector autoregressive (SVAR) model is used to analyze short-run and contemporaneous relationships between monetary aggregates and other macroeconomic variables. This requires imposing restrictions on the correlation structure of the VAR residuals. Different approaches can be followed to serve this task. One approach is to use the Cholesky decomposition together with the assumption of a recursive structure of the contemporaneous relationships between the variables. Another approach uses the information given by the history of the variables (generalized impulse response functions). A third possibility is to adopt restrictions from economic theory. The purpose of this paper is to investigate the implications of the latter technique in a simple monetary framework for both Germany and the Euro area. VAR/VECM residuals are interpreted as deviations of the variables from their conditional expected values, which are also analyzed in a broad set of theoretical monetary models. The model used here builds upon the framework proposed by McCallum (1989) with an extension in order to consider the role of private banks in the money supply process. The implications of this model for impulse response analysis are discussed, and impulse responses for different models are calculated and compared to each other.