Preferences for the Resolution of Risk and Ambiguity

Abstract

Models of generalized recursive utility are becoming increasingly common as alternatives to discounted expected utility theory. These models have successfully explained many so-called, “anomalies” in field data, but often imply that agents have a preference over the timing of uncertainty resolution. Laboratory elicitations of subject preferences generally provide direct evidence in support of this implication. However, all previous experimental studies have only elicited preferences over uncertainty resolution in the domain of objective uncertainty, i.e., risk. By definition, uncertainty includes both objective and subjective domains. Further, not all generalized recursive utility models can accommodate preferences over uncertainty in both domains. For these reasons, we provide the first experimental examination of uncertainty resolution with respect to subjective uncertainty, i.e., ambiguity, in addition to risk. We find that subjects most frequently exhibit a preference for early resolution of both risk and ambiguity and these preferences are positively correlated. Additionally, being ambiguity-seeking decreases the probability of preferring early resolution of ambiguity. Of six, commonly-used, representative recursive utility models, only the generalized recursive smooth ambiguity model of Hayashi and Miao (2011) can plausibly explain the entirety of these experimental findings.

Keywords: uncertainty resolution, risk, ambiguity aversion, recursive utility, experimental economics

JEL Classification: C91, D81