Causal Narratives

Abstract

We experimentally study narratives that describe a (potentially incorrect) causal relationship between variables. Such causal narratives shape beliefs by (mis-)interpreting correlations in the data. We show that false narratives create robust and heterogeneous belief distortions that are inconsistent with rational theory. The heterogeneity in belief distortions is due to subjects constructing their own causal models of the data-generating process, which then compete with externally provided narratives. Our results shed light on why certain narratives are so compelling to investors and other market participants, the source of heterogeneous mental models in financial markets, and the emergence and spread of false narratives.