September 2002 | Peter Klibanoff, Massimo Marinacci, Sujoy Mukerji
This paper proposes and axiomatizes a model of decision-making under ambiguity, where the decision-maker's preferences over acts are characterized by a functional that combines a von Neumann-Morgenstern (vNM) utility function, an increasing transformation, and a subjective probability measure. The key feature of the model is the separation between ambiguity, identified as the decision-maker's subjective information, and ambiguity attitude, which is a characteristic of the decision-maker's tastes. The model allows for smooth indifference curves, leading to improved tractability while maintaining the main features of the Ellsberg paradox. The authors derive a specific form for the transformation, $\phi(x) = -\frac{1}{\alpha} e^{-\alpha x}$, which represents constant ambiguity aversion. They also show that ambiguity is defined behaviorally and characterized by properties of the subjective set of measures $\Pi$. The model is useful for economic modeling involving comparative statics and ambiguity, as it separates subjective uncertainty from ambiguity attitude, allowing for tractable comparative statics exercises. The paper includes illustrative applications to portfolio choice and discusses related literature.This paper proposes and axiomatizes a model of decision-making under ambiguity, where the decision-maker's preferences over acts are characterized by a functional that combines a von Neumann-Morgenstern (vNM) utility function, an increasing transformation, and a subjective probability measure. The key feature of the model is the separation between ambiguity, identified as the decision-maker's subjective information, and ambiguity attitude, which is a characteristic of the decision-maker's tastes. The model allows for smooth indifference curves, leading to improved tractability while maintaining the main features of the Ellsberg paradox. The authors derive a specific form for the transformation, $\phi(x) = -\frac{1}{\alpha} e^{-\alpha x}$, which represents constant ambiguity aversion. They also show that ambiguity is defined behaviorally and characterized by properties of the subjective set of measures $\Pi$. The model is useful for economic modeling involving comparative statics and ambiguity, as it separates subjective uncertainty from ambiguity attitude, allowing for tractable comparative statics exercises. The paper includes illustrative applications to portfolio choice and discusses related literature.