A Structural Vector Autoregression Containing Positive-Valued Components

Abstract

We introduce a structural vector autoregressive model containing strictly positive components. Our nonlinear model results in explicit formulae for impulse response and forecast error variance decomposition analyses, which ease practical computations and structural interpretations. We illustrate the model by evaluating the impact of the monetary policy shock in the core three-equation U.S. monetary policy model including a positive-valued nominal interest rate. Our empirical results show important differences but also similarities to the alternative modelling techniques.

Keywords: Nonlinearity, impulse response analysis, forecast error variance decomposition, monetary policy, shadow rate

JEL Classification: C32, E43, E44, E52