A statistically identified structural vector autoregression with endogenously switching volatility regime

Abstract

We introduce a structural vector autoregressive model with endogenously switching conditional covariance matrix. The structural shocks are identified by simultaneously diagonalizing the reduced form error covariance matrices. It is not, however, always clear whether the condition for the full statistical identification is satisfied, and its validity is difficult to justify formally. Therefore, we provide general sets of conditions, that allow to combine sign and zero restrictions on the impact matrix, for identifying a subset of the shocks when the condition for statistical identification of the model fails. In an empirical application to the effects of the U.S. monetary policy shock, we find that a contractionary monetary policy shock significantly decreases output in a persistent hump-shaped pattern. Prices decrease permanently, but there is short-run inertia in their response. The accompanying R package gmvarkit provides a comprehensive set of tools for numerical analysis of the model.