Business Cycle Narratives

Abstract

Research about narratives’ role in economics is scarce, while real word experience and research in other sciences suggest they matter a lot. This article proposes a view and methodology for quantifying the epidemiology of media narratives relevant to business cycles in the US, Japan, and Europe. We do so by first constructing quantitative measures of narratives based on the news topics the media writes about. We then estimate daily business cycle indexes using this type of data, derive virality indexes capturing the extent to which narratives relevant for business cycles go viral, and finally use so called “Graphical Granger causality” modeling to cast light on cross-country spillovers and whether or not narratives carry news or noise. Our results highlight the informativeness of narratives for describing economic fluctuations, have a clear practical relevance for high-frequency business cycle monitoring, and suggest that narratives capture more than the market’s animal spirits.

Keywords: business cycles, narratives, Dynamic Factor Model (DFM), Latent Dirichlet Allocation (LDA)

JEL Classification: C550, E320, E710, N100