Correlation Made Simple: Applications to Salience and Regret Theory*

I offer an axiomatization of risk models where the choices of the decision maker are correlation sensitive. By extending the techniques of conjoint measurement to the nondeterministic case, I show that transitivity is the von Neumann-Morgenstern axiom that has to be relaxed to allow for these richer patterns of behavior. To illustrate the advantages of the modeling choice, we provide a simple axiomatization for the salience theory model in our general framework. This approach leads to clear comparison to popular preexisting models, such as regret and reference dependence, and allows one to single out the ordering property as the feature that brings salience theory outside the prospect theory realm.