Testing a New Hedging Theory of Confidence Interval Formation

Abstract

We develop and test a simple hedging theory of prediction interval formation. In the presence of uncertainty, forecasters hedge their forecasts by adjusting the prediction interval based on their own (first-order) belief in a way that reflects their (second-order) belief about others’ beliefs. Anchoring and adjustment leads to a positive relationship between an asymmetry measure for the prediction interval and the belief wedge, defined as the di fference between the second-order belief and the first-order belief. Using data from three experiments in which subjects are asked to forecast future stock prices, we provide empirical support for the theory.