Interests

I am an economic theorist with interests in Matching Markets, Mechanism Design, Economics of Information, Bounded Rationality, Evolutionary Game Theory, and Experimental Economics.

Agenda

Going forward, I plan to continue my research on matching. Matching problems are about the formation of productive partnerships, so there are many empirical applications, with high welfare stakes. And it is  becoming even more important as matching increasingly takes place on centralized platforms. These platforms open the door to intentional market design and generate rich data that we can learn from. Standard theory often treats matching markets as almost frictionless, but real markets are much messier. Participants make strategic mistakes, there is a lot of evidence of this. I want to model these frictions and understand how they change the classic results. A new part of this agenda will be to explore fully flexible information design. In practical settings, designers can strategically reveal information to guide participants towards desirable behaviors, and online platforms make it easy to personalize that information.

Beyond matching, I will also study human cognition in strategic environments more broadly. I find that the most convincing methodology to do this is an evolutionary one, modelling how human beings have evolved towards cognitive abilities that maximize fitness. 

I would also like to run experiments to test the predictions of my models, especially to see how people learn and form expectations about potential matching partners.

Papers

IMDPS - Whether and where to apply? Information and Discrimination in Matching with Priority Scores

Solo paper - Job Market Paper 1 -  Submitted for publication

Lastest version: here - Slides: here

This paper considers a matching market where agents have private information on their priority scores and must choose an object to which they apply. The analysis derives the Bayes-Nash equilibria, computes welfare ex ante and interim, and discusses implications for market design. Three main findings emerge. One, there is no symmetric equilibrium in pure strategies. Second, the symmetric equilibrium exhibits a block structure: agents sort into a finite number of classes of neighboring scores where they use the same application strategy. Third, the inefficiencies proceeding from the frictional market design prove interim asymmetric: low-score agents are better off under private information than under public information. In total, private information mitigates the discriminatory power of the priority system.

HAMS - How can I know how much I like you? A Heuristic Approach to Matching and Stability

Solo paper - Job Market Paper 2

Lastest version: here

On a marriage market with unknown preferences (agents only observe the current matching and realized match utilities), we define a novel and natural heuristic of belief formation (valuation), which incorporates a famous and documented cognitive bias (the projection bias). Under this heuristic, an agent estimates a counterfactual match utility by extrapolating from realized match utilities: his own utility and the weighted average utility of all current partners of the targeted partner's type. We study how this reshuffles the market outcome, as given by pairwise stable matchings when agents have valuation beliefs (v-stability). When restricting our attention to pure matchings, we find that v-stability is equivalent to any two partners holding the same rank according to current utilities (happiness sorting). The predictions under specific preference structures are then straightforward. The alignment of interests across the market governs the size of the v-stable set from empty to maximal. The correlation of preferences by agent or target stabilizes the positive assortative matching. For a generic market, though, we get neither the existence of a pure v-stable matching nor the convergence of a dynamic blocking pair process (predicting persistent moves on the market). The most general version of the model defines a notion of mixed matching, characterizing the proportions of each productive type matched with each partner type. Our main result is a general existence theorem for v-stable matchings in the mixed extension.

RIIS - Robust Incomplete-Information Stability for matching markets without monetary transfers

Solo paper  - Submitted for publication

Latest version: here

We consider a matching market with no transfers and incomplete asymmetric information - on one side, agents do not observe types of potential partners; they just observe the type of their current partner. The model can represent civil servants' job markets where wages are regulated and where employers have trouble learning about workers' productivity prior to hiring.  We apply the definition of incomplete-information stable matchings by Liu, Mailath, Postlewaite, and Samuelson (2014) - a pair is blocking if both partners strictly want to block under any reasonable beliefs they may have using their private information and common knowledge of stability.  Even under monotonic payoffs, the incomplete-information stable set may be large - it depends finely on the market structure and the prior belief support. If the unknown workers' type function is a bijection, the stable sets with complete and incomplete information perfectly coincide (to include only positive assortative matchings). We show, using examples, that the robust approach can reach precise predictions even beyond the monotonic case. 

ESABEE - Evolutionary Stable Analogy-Based Expectation Equilibrium

Joint with Philippe JEHIEL and Giacomo WEBER

Draft available upon request - Poster for Parisian Festival of Game Theory 2025: here - Slides ADRES 2026: here

We develop an evolutionary approach to endogenize the choice of analogy partitions in the analogy-based expectation equilibrium. A analogy partition with more classes incurs a higher fitness cost. In an evolutionarily stable analogy-based expectation equilibrium (ESABEE), we require that analogy partitions and strategies satisfy two conditions: (i) given an analogy partition, the associated strategy is a best response to the corresponding analogy-based expectations; and (ii) analogy partitions that arise with positive probability induce the highest overall fitness among all possible partitions. We show that an ESABEE always exists in finite environments. Nash equilibrium prevails when there is no fitness cost associated with categorization. We propose a version of replicator dynamics in which analogy partitions are reproduced proportionally to their fitness in each period, and strategies are adjusted by moving in the direction of the best responses induced by the partitions. We illustrate the concept using a family of Hawk-Dove games, and show that ESABEE assigns positive probability to coarse partitions irrespective of the fitness costs associated with having finer partitions. We use the same framework to get insight on decision problems (for eg., an investment problem in which the decision maker observes his cost type and need to form expectations about the benefit of the investment) and environments where agents face various families of games throughout life (for eg., conflict and coordination games). 

PMIA - Campus visits, or Pre-Matching Information Acquisition in school choice

Joint with Francis BLOCH 

Draft available upon request

This paper studies a college admission problem gathering heterogeneous students and colleges where students can endogenously acquire information on their own preferences. Students' preferences over colleges include a common component, which is common knowledge, and a private component, which is unknown ex-ante. Students can learn about the private components, before matching occurs through a standard Deferred Acceptance mechanism with common priorities. The question is: What information do students acquire, as a function of their priority rank? With unit constraint on learning and unit capacities at colleges, we find that the best student learns about one of the best colleges. Students with lower-priority learn about the best college among the ones where they are admitted for sure. The proof uncovers a novel additive property of the values of information. We discuss matching and welfare implications and ongoing generalizations to larger seat capacities at colleges and relaxed learning capacity constraints.