News-driven uncertainty fluctuations

Song, Dongho; Tang, Jenny

Working Paper

News-driven uncertainty fluctuations

Abstract: We embed a news shock, a noisy indicator of the future state, in a two-state Markov-switching growth model. Our framework, combined with parameter learning, features rich history-dependent uncertainty dynamics. We show that bad news that arrives during a prolonged economic boom can trigger a ?Minsky moment??a sudden collapse in asset values. The effect is greatly amplified when agents have a preference for early resolution of uncertainty. We leverage survey recession probability forecasts to solve a sequential learning problem and estimate the full posterior distribution of model primitives. We identify historical periods in which uncertainty and risk premia were elevated because of news shocks.

Keywords: Bayesian learning; discrete environment; Minsky moment; news shocks; recursive utility; risk premiums; survey forecasts; uncertainty;

JEL Classification: C11; E32; E37; G12;