Incentive Compatible Procedure to Measure Risk Preferences: Adequacy of the Midpoint Chaining Method

Zusammenfassung

How to elicit risk preferences is an important question for explaining and predicting human decision behaviour. We provide an overview of elicitation methods and, then, focus on the midpoint chaining method. Within this method, subjects successively specify certainty equivalents for lotteries with two outcomes and the corresponding utility value is calculated. More and more supporting points for the utility function can be integrated and should result in a finer description of subjects risk preferences. We conduct an experiment to test the midpoint chaining method and, hereby, apply an incentive compatible procedure to measure risk preferences. The experimental results show that the more certainty equivalents are integrated, the less consistent are the predicted choices compared to real choices. However, it is possible to explain mean behaviour by adopting the idea of a finest perceived value when estimating numbers according to prominence theory. The finest perceived value is defined on the base of the outcome range and is used as divisibility condition. If this simplification takes place, imprecise supporting points for the utility function are elicited by the midpoint chaining method. More consistent results are obtained by parameterizing the utility function and using only one certainty equivalent to estimate the utility function.