Belief identification with state-dependent utilities

TL;DR

This paper introduces a new method for identifying individual beliefs using state-dependent utilities by employing proxies and a variant of the strategy method, overcoming traditional identification challenges.

Contribution

It proposes a simple, flexible approach to identify beliefs without assuming state-independent utilities, using proxies and elicitation tools.

Findings

Method effectively identifies beliefs using proxies.

Flexible proxy classes are easy to implement.

Enables new definitions of utility functions over states.

Abstract

It is well known that individual beliefs cannot be identified using traditional choice data, unless we impose the practically restrictive and conceptually awkward assumption that utilities are state-independent. In this paper, we propose a novel methodology that solves this long-standing identification problem in a simple way, using a variant of the strategy method. Our method relies on the concept of a suitable proxy. The crucial property is that the agent does not have any stakes in the proxy conditional on the realization of the original state space. Then, instead of trying to identify directly the agent's beliefs about the state space, we elicit her conditional beliefs about the proxy given each state realization. The latter can be easily done with existing elicitation tools and without worrying about the identification problem. It turns out that this is enough to uniquely identify the agent's beliefs. We present different classes of proxies that one can reasonably use, and we show that it is almost always possible to find one which is easy to implement. Such flexibility makes our method, not only theoretically-sound, but also empirically appealing. We also show how our new method allows us to provide a novel well-founded definition of a utility function over states. Last but not least, it also allows us to cleanly identify motivated beliefs, freed from confounding distortions caused by the identification problem.