"Imperfect Memory and Choice Under Risk" (Job Market Paper) [download]

This paper proposes a model of choice under risk based on imperfect memory and self-deception. The model assumes that people have preferences over their own attributes and can, to some extent, manipulate their memories. It leads to a non-expected utility representation and provides a unified explanation for several empirical regularities: non-linear probability weights, small-stakes risk aversion, regret and the competence hypothesis. It also leads to endowment and sunk cost effects. The model implies that behavior will converge to the one predicted by expected utility theory after a choice has been made a sufficiently large number of times.

"Multidimensional Incentive-Compatibility: The Multiplicatively Separable Case" (with Aloisio Araujo and Humberto Moreira, submitted) [download]

This paper characterizes incentive-compatible allocations when types are multidimensional and the single-crossing condition may not hold. This characterization is used to obtain the optimal contracts in multidimensional screening as well as the equilibria in multidimensional signaling models. Then, we determine the implications of signaling and screening models when the single-crossing condition is violated. The unique robust prediction of signaling is the monotonicity of transfers in (costly) actions. Any function from the space of types to the space of actions and an increasing transfer schedule can be rationalized as an equilibrium profile of many signaling models. Apart from the monotonicity of transfers in actions, we obtain an additional necessary and sufficient condition in the case of screening. In one-dimensional models, this condition states that the principal's profit as a function of the agent's type must grow at a higher rate under asymmetric information than under symmetric information.

"Should Educational Policies Be Regressive" (with Humberto Moreira)

This paper considers optimal educational policies when the ability to benefit from education is private information. De Fraja (Review of Economic Studies, 2002) has shown that optimal policies are regressive. We argue that his results follow from an unusual specification of the government’s budget constraint. When this constraint is replaced by the standard budget constraint, regressive policies are no longer optimal. The optimal educational policy can be decentralized through Pigouvian taxes and credit provision and is not regressive. Furthermore, when the utility function is not quasi-linear, education may not be monotonic in ability and progressivities of education are locally welfare improving.