Uncertainty Premia for Small and Large Risks

Abstract

We identify an ambiguity premium for US stocks from increases in the option-implied concavity of preferences immediately before scheduled macroeconomic announcements. Our methodology relies on Skiadas' (2013) critique of smooth ambiguity aversion models, which shows that ambiguity aversion has a negligible effect on small risks, defined as risks that are proportional to the holding period. We show that the same critique implies that the effect of smooth ambiguity aversion on large risks, such as macroeconomic announcements, should be of first-order importance. We test for the difference in the effect of ambiguity aversion on the two types of risk by studying the implied concavity of preferences for a representative agent, and confirm that such concavity indeed increases significantly ahead of announcements. Except for smooth ambiguity aversion, no other representative agent model predicts such an increase.