This award funds a series of choice experiments that test new methods for eliciting individual risk preferences. These methods have been developed from an innovative theory about how to model the way people make decisions in situations when certainty and uncertainty are combined. If the methods prove to have predictive power, this will be evidence in support of the new theory.
The new theory starts with a general observation about widely observed behavior that does not fit the expected utility models widely used in the past by economists. The Allais paradox, prospect theory probability weighting, and small states risk aversion all arise in situations when certainty and uncertainty are combined. The key idea here is to model two different utility functions to represent preferences: a certainty function and an uncertainty function. The ways that actual behavior deviate from expected utility may be driven by differences between certain and uncertain utility.
The award also funds an experiment on how markets for insurance are affected by possible ambiguity aversion.
The broader impacts of this project include the interdisciplinary impact. A wide range of scientists from public health to political science are now using tools from behavioral economics. Policy makers are finding that this kind of decision modelling suggests new ways to reach policy goals in an effective way. If the new theory is supported by these experiments, it should be rapidly adopted by these communities.
The student, Charles Sprenger, has produced a dissertation that has been extremely successful at reshaping the discussion on how to measure preferences for risk, and for risk over time. The work provides novel methods for eliciting preferences that allow more precise and reliable estimation of subjects' preferences. These techniques, which have been published in top journals, are reshaping how other scientists are measuring preferences and are greatly sharpening our ability to evaluate and predict the effects of policy that involves risk and risk over time. The centerpiece for this work is the invention of the "uncertainty equivalent." Just like a "certainty equivalent" finds the sure amount that makes a person indifferent to a target gamble, the uncertainty equivalent find the probability of winning in a reference gamble that makes that gamble indifferent to the taraget gamble. By keeping risk on both sides of the equation, this method avoids the Allais certainty effect, which we show, has introduced a seriuos bias into the previous measures. Furhter work shows that this methodology not only fits the data better than other methods, but undermines the foundation of Cumulative Prospect Theory probability weighting. That is, CPT was invented to avoid violations of Stochastic Dominance that original Prospect Theorey Allowed. Yet, our work shows that it is those who violate stochastic dominance in the uncertainty equivalent that also drive the nonlinear weighting in the CPT. Thus, CPT results could be inturpreted as specification error involving the Allais Certainty Effect.
