From Private-Belief Formation to Aggregate-Vol Oscillation

What are the quantitative implications of learning and informational asymmetries, for generating fluctuations in aggregate and cross-sectional volatility over the business-cycle? I propose a model that relies on informational channels, for the endogenous amplification of the conditional volatility in macro aggregates and of cross-sectional dispersion during economic slowdowns, in a homoscedastic-shock environment. The model quantitatively matches the fluctuations in the conditional volatility of macroeconomic growth rates, while generating realistic real business-cycle moments. Consistently with the data, shifts in the correlation structure between firms are an important source of aggregate volatility. Up to 80% of the conditional aggregate volatility fluctuations are attributed to fluctuations in cross-firm correlations. Correlations rise in downturns due to a higher weight that firms place on public information, which causes their beliefs, and policies, to comove more strongly. In the data, correlations rise at recessions in spite of a contemporaneous increase in cross-sectional volatility, as the average between-firm covariance spikes more than dispersion does.