Revealed Conflicting Preferences

A new Harvard working paper by Ambrus and Rozen (2008), "Revealed Conflicting Preferences", examines preference revelation in the context of a multiple-self model. An interesting read for myself, Michael and Dave C who are planning a symposium related to research like this in the future. The paper is largely mathematical but resolves around the concept of IIA (indpendence of Irrelevant Alternatives); this concept is also used by micro-econometricians when testing in the multinomial choice model, so there is a case for trying to master a theory paper like this. Abstract below.

"We consider the use of multiple selves or rationales for rationalizing choice functions that may violate axioms of rationality, when the relationship between the choice set and how conflicting preferences are aggregated is made explicit. For a broad class of preference aggregators, we show that the number of selves required to rationalize a choice function is at most a linear function of the number of violations of the Independence of Irrelevant Alternatives (IIA) that the choice function exhibits. In particular, any choice function can be rationalized with a finite number of selves using any of the aggregators belonging to this class. Applying this framework to choice over menus, we examine the revealed preference implications of anticipated IIA violations."