Tests of Investor Learning Models Using Earnings Innovations and Implied Volatilities
57 Pages Posted: 18 Sep 2015 Last revised: 4 Nov 2015
Date Written: September 8, 2015
Abstract
This paper investigates alternative models of learning to explain changes in uncertainty surrounding earnings innovations. As a proxy for investor uncertainty, we use model-free implied volatilities; as a proxy for earnings innovations, representing signals of firm performance likely to drive investor perceptions of uncertainty, we use quarterly unexpected earnings benchmarked to the consensus forecast. First, we document ― consistent with most Bayesian models of learning and prior research ― that uncertainty declines on average after the release of quarterly earnings announcements. Second, we show that this decline is attenuated by the size of the signal ― that is, by the magnitude of the earnings innovation. This latter result is inconsistent with some models of Bayesian learning, and consistent with more sophisticated models, which incorporate signal magnitude as a factor driving changes in uncertainty. Third and most important, we document that signals deviating sufficiently from expectations ― that is, sufficiently large earnings innovations ― lead to net increases in uncertainty. Critically, this latter result suggests that models allowing for posterior variance to be greater than prior variance even after signal revelation (such as “regime shifts” in Pastor and Veronesi 2009) appear more descriptive of how investors incorporate new information.
Keywords: uncertainty, implied volatilities, earnings innovations, regime shifts, Bayesian learning
JEL Classification: M4, G1
Suggested Citation: Suggested Citation