Job Market Paper
Abstract: Why are simple linear compensation schemes common in long-term relationships, even though complex contracts could potentially mitigate moral hazard? To address this question, I study a dynamic principal-agent model in which the principal faces uncertainty about the agent's set of available actions and is ambiguity-averse. In response to this uncertainty, the principal offers the agent a contract in each period to maximize the former's worst-case payoff. I characterize a condition under which simple linear contracts are dynamically optimal when the principal can learn new information about the actions available to the agent, thereby providing a new explanation for the prevalence of linear contracts in practice. I show that this condition is not only sufficient but also necessary for linear optimality (in addition to two regularity conditions), reinforcing its critical role in explaining contract simplicity. Additionally, I characterize the optimal way for the principal to revise the first-period contract and find that it may take one of two forms: the optimal linear contract that would arise in a static environment or a convex combination between the first-period contract and a new linear contract. The latter also includes repetition of the previous contract.
Working Papers
Cheap Talk with Verifying Receivers: Strategic Silence under Costly Learning
Abstract: I study a dynamic mechanism design problem where agents selectively disclose verifiable evidence that arrives over time. In many environments - such as contract negotiations or grant allocation—evidence emerges gradually and can be strategically withheld. I analyze how optimal allocations evolve when the principal can influence the informativeness of future signals through first-period transfers. The optimal mechanism fully smooths the low-type’s allocation across time and signals, while the high-type’s allocation increases with informativeness above a disclosure-inducing cutoff. I characterize a Gittins-style condition governing optimal experimentation and show how elicitation constraints endogenously limit the principal’s ability to learn. The result highlights a fundamental trade-off between information generation and incentive compatibility.
Experimentation with Voluntary Disclosure
Abstract: I study a dynamic mechanism design in environments where agents can selectively disclose verifiable evidence over time, introducing new strategic considerations for the principal. In many real-world settings, such as contract negotiations or research grant applications, hard evidence may emerge gradually, and agents can strategically withhold or reveal it to influence outcomes. I explore how the availability and disclosure of such evidence shape optimal allocations over multiple periods, given that the designer can experiment with the first-period allocation in order to affect the quality of the evidence the agent possesses. My findings reveal that the designer prefers to smooth the low-type allocations across periods, insulating them from changes in evidence, while high-type agents see their allocations increase with stronger evidence but only above a quality cutoff. Below the cutoff, evidence is being interpreted by the designer as if the agent did not disclose at all. In the first period, I find conditions under which the allocation is distorted in order to induce more informative evidence in the sequel.
The Effect of the 1918 Influenza Pandemic on Education Attainment in 20th-Century Spain
Abstract: This paper examines the impact of the 1918 influenza pandemic on educational attainment in Spain, focusing on changes in student enrollment growth across different fields. The findings reveal a short-term decline in fields with weaker connections to the labor market, such as philosophy and literature, as well as in teacher training programs. This decline suggests a potential shortage of teachers in primary and secondary education in the years following the pandemic. However, the analysis shows that this effect was temporary, as within one to three years, enrollment growth rates returned to pre-pandemic levels. These results highlight the disruption caused by the pandemic but also demonstrate the system’s ability to recover within a relatively short period.
Work in Progress
Optimal Delegation with Costly Learning (Draft Coming Soon!)
Abstract: I study optimal delegation in a setting where an agent must choose both the precision of costly information acquisition and the subsequent action from a set delegated by the principal. Unlike standard delegation models, I consider simultaneous choices of delegation and endogenous information precision by the agent. I characterize the optimal delegation set and provides conditions under which traditional interval delegation fails to be optimal. I provide conditions under which carefully chosen non-interval delegation sets, featuring strategic gaps, outperforms intervals by strongly incentivizing more precise learning by the agent.
Bitcoin: A Mechanism Design Approach
Description: I take a mechanism design approach to model the blockchain environment to characterize the system’s current features. What should be the identity of the miners chosen to mine a block? What should the size of the mined block be? what would be the induced fees the agents pay under the optimal mechanism, subject to a relevant guaranteed revenue constraint?