Publications
Stochastic Independence Under Knightian Uncertainty (SIND.pdf) [Revue d'Economie Politique, Vol. 126/3 (2016), 379-398]
We show that under Bewley preferences, the usual axiom that characterizes stochastic independence is not sufficient to uniquely identify a model of independent beliefs. We thus introduce the concept of product equivalent of an act and show that it allows us to obtain a unique characterization of stochastic independence for the multiple-prior expected utility model.
Choice Deferral, Indecisiveness and Preference for Flexibility (Choice Deferral New.pdf) [Journal of Economic Theory , Vol. 170 (2017), 417-425] with Severine Toussaert
Online appendix: Online Appendix.pdf Older version: Choice Deferral Old.pdf
We introduce a model of menu choice in which a person's decisions may only partially reveal her innate tastes. The latter are modeled by means of a possibly incomplete (but otherwise rational) preference relation P, and the former by a completion P* of that relation. The two are connected through an axiom formalizing an intuitive rule: ``Whenever in doubt, don't commit; just leave options open.''
Under the usual assumptions of the menu choice literature, we find that even the smallest amount of indecisiveness is enough to force P*, through this deferral property, to exhibit preference for flexibility on its entire domain. Thus we highlight a fundamental tension between non-monotonic preferences, such as preferences for self-control, and tendency to defer due to indecisiveness.
Working Papers
Breadth vs. Depth [under revision] with Michael Richter and Sen Geng
Slides: Lyon 2 (GATE).pdf
We consider a fundamental trade-off in search: that between breadth (learning a little about each alternative) and depth (learning a lot about a few). We model the trade-off positing a decision maker who has to choose among multiple multi-attribute alternatives of uncertain value. The overall utility of each of n alternatives can be expressed as the sum of the value of the n attributes. The decision maker only knows the distribution of each attribute value, which is i.i.d. across objects and attributes. She can learn, before choosing an option, either the k realized values of a given attribute, breadth, or the n realized attributes for a given object, depth. We study the agent's maximization problem under different distributional assumptions, both for given n and k and in the limit as the number of objects and attributes grow to infinity, and introduce also a strategic model that provides an I.O. application of our setting. We provide conditions under which for small numbers of attributes and options breadth beats depth, while as the numbers grow larger the inequality is reveres. We also show that in the strategic model equilibria with attribute search can always be sustained and pareto dominate the object search equilibria.
Status-Quo Bias and Dynamic Choice [under revision]
We propose a rational theory of dynamic choice with status-quo-bias. The status-quo in my framework is determined endogenously by the agent's actions: her choice in period (t) acts as a status-quo for the (t+1) period choice. In a two period model we show that, if the agent correctly forecasts her future bias, her behavior can be described using two simple axioms. One is a rationality requirement, which captures her underlying sophistication. The other is a weakening of the usual time-separability postulate, which allows the agent to break the independence from the past of preferences over future streams in favor of the status quo. The resulting model characterizes an agent performing dynamic programming under a mental constraint which depends from the status-quo. The paper also extends the model to the general finite horizon settings. The theory encompasses the standard (non-status-quo-biased) agent as a special case, and can be considered as a dynamic generalization of the static choice model of Masatlioglu and Ok (2013).
Matching and Welare in Large Markets - With an Application to Monitoring in Teams (CPAM.pdf)
with Patrick Legros and Andrew Newman
We consider two-sided markets with a continuum of agents and a continuum of types. The single crossing condition (GID) introduced in Legros & Newman (2007) implies that equilibria matches are payoff equivalent to equilibria with positive assortative matching. We show the existence of a "best equilibrium" for each side of the market in the sense that for each agent on a given side, his payoff in this "optimal" equilibrium is the maximum over all possible equilibria. When the type distribution is positive and continuous, the marginal wage of an agent is equal to his marginal productivity, which, contrary to the TU case, cannot be defined independently of the equilibrium payoffs. The marginal wage function captures the overall distribution of marginal payoffs in the economy and is the solution to a system of differential equations. These equations capture the two effects in NTU models when GID holds: the standard complementarity in types and the complementarity in types and payoffs. Because with non-transferabilities, equilibrium matches do not necessarily maximize total surplus, agents being paid at their marginal productivity is neither a sufficient nor necessary condition for surplus efficiency. We illustrate this point in a principal-agent example with endogenous monitoring.