Predict Stock Return Variance Across the Information Cycle

This paper uses historical variance term structures to characterize the different phases of a company’s information cycle, and shows that the up and down phases of the information cycle experience different durations, with the up phase fast but the down phase slow. The paper proposes a conditional variance dynamics to capture the dynamics variation across the information cycle, and develops a two-dimensional conditional pooling estimation approach that balances the needs for reducing estimation errors and accommodating conditional dynamics variation. As an application, the paper finds that classic asset pricing relations can vary strongly across the information cycle.