A Model of Anchoring and Adjustment for Decision-Making under Risk

Abstract

We introduce a general model of anchoring and adjustment for decision-making under risk. To evaluate a lottery, agents anchor on the unweighted average of the utils associated with all possible outcomes. They adjust from the anchor (insufficiently) using information on the outcomes' probabilities. The resulting model implies behavior which is observationally similar to established prospect theory models. However, it is not rank-dependent and naturally applies to lotteries with any given number of outcomes, including continuous lotteries. We test the implications of this model with an experiment and find empirical support - subjects more prone to anchoring also weight probabilities more strongly. In a set of lottery choices allowing for dominance violations, the plurality of these violations appear according to the pattern predicted by our model. We consider applications to decisions in simple lotteries, show how the model can explain several well-known choice anomalies, and apply it to the equity premium puzzle for which we offer an empirical calibration using historical data on asset returns. The anchoring model is a flexible tool for modeling a simplified choice process for decisions under risk.

Keywords: Decision Theory, Probability Weighting, Anchoring, Heuristics

JEL Classification: D11, D81, D91, G41