Scoring rules for subjective probability distributions

Subjective beliefs are elicited routinely in economics experiments. However, such elicitation often suffers from two possible disadvantages. First, beliefs are recovered in the form of a summary statistic, usually the mean, of the underlying latent distribution. Second, recovered beliefs are biased significantly due to risk aversion. We characterize an approach for eliciting the entire subjective belief distribution that is minimally biased due to risk aversion. We offer simulated examples to demonstrate the intuition of our approach. We also provide theory to formally characterize our framework. And we provide experimental evidence which corroborates our theoretical results. We conclude that for empirically plausible levels of risk aversion, one can reliably elicit most important features of the latent subjective belief distribution without undertaking calibration for risk attitudes providing one is willing to assume Subjective Expected Utility.