On the Existence and Interpretation of the "Unit Root" in U.S. GNP
NBER Working Paper No. 2716
In this paper, we assess the degree to which four of the most commonly used models of risky decision making can explain the choices individuals make when faced with risky prospects. To make this assessment, we use experimental evidence for two random samples of young adults. Using a robust, nonlinear least squares procedure, we estimate a model that is general enough to approximate Kahnenman and Tversky's prospect theory and that for certain parametric values will yield the expected utility model, a subjective expected utility model and a probability-transform model. We find that the four models considered explain the decision-making behavior of the majority of our subjects. Surprisingly, we find that the choice behavior of the largest number of subjects is consistent with a probability-transform model. Such models have only been developed recently and have not been used in applied settings. We find least support for the expected utility model -- the most widely used model of risky decision making.