Measuring and explaining cyclical fluctuations in consumption using multivariate information

Abstract

Recessions and expansions are often caused or reinforced by developments in private consumption - the largest component of aggregate demand - which, as a result, varies over the business cycle. This paper implements a multivariate approach to measure the cyclical component of aggregate consumption and to understand its main drivers. Using US data over the period 1973Q1-2022Q4, we estimate this component from a multivariate Beveridge-Nelson decomposition applied to a medium-scale Bayesian vector autoregression. The choice of variables included in the analysis is informed by a general savers-spenders model. We find a significant, persistent and robust cyclical component in consumption that has predictive power for both future consumption growth and equity returns. An informational decomposition points to variables related to incomplete markets as the main contributors to this component. This is confirmed by a causal analysis that attributes between 20% and 40% of cyclical movements in consumption to uncertainty shocks. Joint paper with Tino Berger.