Learning, Confidence, and Business Cycles

In this paper we study the amplification feedback between uncertainty and economic activity. We build on van Nieuwerburgh and Veldkamp (2006) to model an economy with a procyclical signal to noise ratio used in filtering the hidden, persistent, state of technology. Recessions, caused by either fundamental supply or demand shocks, are periods where the lower production scale implies higher uncertainty in the form of a larger posterior variance. The endogenous increase in uncertainty makes agents less confident and further reduces economic activity, which gives rise to persistent and amplifying effects. We use linear methods to study the feedback effects of time-varying endogenous uncertainty and confidence in standard business cycle models. We illustrate the main qualitative implications in a stylized model and use a quantitative version to evaluate their magnitudes. We also provide an extension to a heterogeneous firm setting, whose aggregation is facilitated by the use of linear methods and where we can additionally analyze the impact of experimentation and firm-level dispersion shocks.