Research

Transparent Matching Mechanisms  

Theoretical Economics (conditionally accepted) 

I study a central authority's ability to commit to a publicly announced mechanism in a one-to-one agent-object matching model. The authority announces a strategy-proof mechanism and then privately selects a mechanism to initiate a matching. An agent's observation in the form of the final matching has an "innocent explanation" (Akbarpour et al., 2020), if given the agent's reported preferences, there is a combination with other agents' preferences leading to an identical observation under the announced mechanism. The authority can only commit up to "safe deviations" (Akbarpour et al., 2020)—mechanisms that produce only observations with innocent explanations. For efficient or stable announcements, I show that no safe deviation exists if and only if the announced mechanism is dictatorial. I establish that the Deferred Acceptance (DA) Mechanism (Gale and Shapley, 1962) implies commitment to stability. Finally, I show that group strategy-proof and efficient announcements allow commitment to efficiency only if they are dictatorial. 

As the first project of my PhD Thesis, a first Draft circulated under Transparent Priority-Based Allocation (2019)(available upon request)

Some related questions with a focus on TTC are discussed in the preliminary draft: Credible Top Trading Cycles (2018) (available upon request)

 Regret-Free Truth-Telling in School Choice with Consent (with Yiqiu Chen) 

Theoretical Economics, vol.19, no. 2 (2024): 635–666

The Efficiency Adjusted Deferred Acceptance Rule (EDA) is a promising candidate mechanism for public school assignment. A potential drawback of EDA is that it could encourage students to game the system since it is not strategy-proof. However, to successfully strategize, students typically need information that is unlikely to be available to them in practice. We model school choice under incomplete information and show that EDA is regret-free truth-telling, which is a weaker incentive property than strategy-proofness and was introduced by Fernandez 2020. We also show that there is no efficient matching rule that weakly Pareto dominates a stable matching rule and is regret-free truth-telling. Note that the original version of EDA by Kesten 2010 weakly Pareto dominates a stable matching rule, but it is not efficient. 

A Theory of Auditability for Allocation Mechanisms (with Aram Grigoryan) - Working Paper

Extended abstract at ACM EC '23. 

In centralized mechanisms and platforms, participants do not fully observe each others' type reports. Hence, if there is a deviation from the promised mechanism, participants may be unable to detect it. We formalize a notion of auditabilty that captures how easy or hard it is to detect deviations from a mechanism. We find a stark contrast between the auditabilities of prominent mechanisms. We also provide tight characterizations of maximally auditable classes of allocation mechanisms. 

Auditability in School Choice (with Aram Grigoryan) 

AEA Papers and Proceedings, vol. 114, May (2024): p. 492-96

Envy and Strategic Choice in Matching Markets - Working paper 

I study agents’ incentives in allocation problems where agents are envious. Agents may envy others who get an outcome they prefer to their own. Given a mechanism and agents’ actions, an agent’s envy towards another agent is inevitable if she has no action under which she does not envy the other agent, given one keeps the actions of other agents fixed. A mechanism is envy-proof if any envy under truth-telling is inevitable. Envy-proofness is stronger than strategy-proofness and neither stronger nor weaker than group strategy-proofness. I find a trade-off between some prominent normative notions of envy-freeness and robustness to envy-driven incentives. I show that the Top Trading Cycle (TTC) Mechanism is envy-proof, while the Deferred Acceptance Mechanism is not. I fully characterize envy-proof and efficient mechanisms as inheritance rules. Among group-strategy-proof and efficient mechanisms, envy-proofness characterizes hierarchical exchange (Papai, 2000) 

Opaque Mechanisms and Strategy-proofness (with Aram Grigoryan) - Working paper 

Centralized clearinghouses often make opaque announcements about the allocation process. Thus, the clearing house may be able to select any allocation mechanism that is consistent with its announcement. We study whether the clearing house can guarantee strategy-proofness under opacity, in the sense that every mechanism chosen that is consistent with the announcement is strategy-proof. We find a robust negative result: a clearing house can ensure strategy-proofness if and only if it fully communicates the details of a strategy-proof mechanism. Thus, the designer cannot guarantee strategy-proofness if the announcement contains any opacity. Other properties like weak Maskin monotonicity and non-bossiness can be guaranteed under opaque announcements.

Navigating Transitions: Preferences and Perceptions in German School Choice (with Alexander Westkamp) - Work in progress 

Learning (Matching)-Mechanisms with a single instance (with Aram Grigoryan) - Work in progress