Event-Specific Uncertainty and its Expected Resolution

Abstract

Standard models of Bayesian updating predict a stronger investor reaction to new information when those investors are more uncertain about the firm. However, prior empirical literature has struggled to find widespread evidence in support of this prediction. This paper tests two explanations for this lack of empirical support. The first explanation is that extant measures of uncertainty tend to also capture noise in the reporting system, which has a negative relation with investors’ response to new information. The second explanation is that when investors face high levels of overall uncertainty, earnings announcements do little to resolve that uncertainty. We provide evidence consistent with the latter explanation. We develop two measures of “event-specific uncertainty” - the amount of uncertainty expected to be resolved at an upcoming event - and show that this uncertainty construct is positively related to both earnings response coefficients and the informativeness of earnings announcements as a whole.

Keywords: uncertainty, earnings announcements, event-specific uncertainty, implied volatility

JEL Classification: G12, G14, M41, D80