Published and Forthcoming
joint with Bernard Herskovic
We study how networks of informational flows are formed. We develop an endogenous network formation model, in which agents form connections to acquire information. Our model features complementarity in actions as agents care not only about accuracy of their decision making but also about the actions of other agents. In equilibrium, the information structure is a hierarchical network, and, under weakly convex cost of forming links, the equilibrium network is core-periphery. Even if agents are ex-ante identical, there is ex-post heterogeneity in payoffs and actions. Finally, we study how individual characteristics determine agents' roles in the network.
joint with Elliot Lipnowski
In an ongoing relationship of delegated decision making, a principal consults a biased agent to assess projects’ returns. In equilibrium, the principal allows future bad projects to reward fiscal restraint, but cannot commit to indefinite rewards. We characterize equilibrium payoffs (at fixed discounting), showing that Pareto optimal equilibria are implemented via a two-regime `Dynamic Capital Budget’. Rather than facing backloaded rewards—as in dynamic agency models with commitment power—the agent loses autonomy as time progresses.
Journal of Economic Theory
joint with Elliot Lipnowski and Aditya Kuvalekar
An expert advises a decision maker over time. With both the quality of advice and the extent to which it is followed remaining private, the players have limited information with which to discipline each other. Even so, communication in and of itself facilitates cooperation, the relationship evolving based on the expert’s advice. We show a formal equivalence between our setting and one of cheap talk with capped money burning, enabling an exact characterization (at fixed discounting) of the expert’s attainable payoffs. While an ongoing relationship often helps, our characterization implies that, absent feedback, relational incentives can never restore commitment.
Revise and Resubmit at AER
joint with Ala Avoyan
We examine the efficiency gains of introducing a pre-play phase—allowing agents to communicate while incrementally committing to their words—in coordination environments. We consider a particular pre-play institution for which the efficient equilibrium is unique in the extended game, and we test the environment in the lab. We focus on weaklink games, a coordination game with multiple Pareto ranked equilibria, in which an inferior equilibrium is often observed in experimental settings. We first show that the institution intertwining communication and incremental commitment significantly increases subjects' payoffs, even when compared to rich pure communication protocols. Aligned with theoretical predictions, commitment changes the way players communicate—affecting what is a player says and, moreover, how others interpret it. Finally, we focus on the road to efficiency; we analyze agents’ dynamic behavior as they try to coordinate, and we show that this behavior is closely aligned with theoretical predictions. The results shed new light on how pre-play institutions can mitigate coordination failures and highlight communication and incremental commitment as key features.
We study a continuous-time model of partnership, with persistence and imperfect state monitoring. Partners exert private efforts to shape the stock of fundamentals, which drives the pro ts of the partnership, and the profits are the only signal they observe. The near-optimal strongly symmetric equilibria are non-Markovian and are characterized by a novel differential equation that describes the supremum of equilibrium incentives for any level of relational capital. Imperfect monitoring of the fundamentals helps sustain incentives, due to deferred incentives, and increases the partnership's value (Sand in the wheels). Good profit outcomes rally the partners to further increase effort when relational capital is low, but lead them to coast and decrease effort when relational capital is high. Even partnerships with high fundamentals may unravel after a short spell of terrible signals (Beatles' break-up).
Revise and Resubmit at Rand
joint with Aditya Kuvalekar and Johannes Schneider
An agent decides whether to approve a project based on his information, some of which is verifiable in a court. An honest agent wants to implement projects that are likely to succeed; a dishonest agent wants to implement any project. If the project fails, the court examines the verifiable information and decides the punishment. The court seeks to deter ill-intentioned agents from implementing projects likely to fail while incentivizing the use of unverifiable information. We show how information of different kinds affects welfare. Improving the verifiable information can reduce welfare, while improving the unverifiable information always increases welfare.
We examine politicians’ career dynamics generated by political accountability, characterizing voter-optimal equilibrium play under repeated moral hazard. When moral hazard binds, equilibrium play is non-stationary: Re-election prospects improve with good performance and deteriorate with bad. First-term politicians are among the most electorally vulnerable and the hardest-working, and effort and electoral vulnerability both tend to decline with tenure. These dynamics enable a detailed analysis of limited voter commitment, voluntary retirement from politics, and adverse selection with politicians’ ability and effort being complementary. Our analysis highlights how a politician’s career is shaped by voters’ evolving “goodwill” toward her.
Revise and Resubmit at The Economic Journal
joint with Luiz Brotherhood Bernard Herskovic
Popular Press (Daily Bruin)
Popular Press (Anderson Review)
We study whether college admissions should implement quotas for lower-income applicants. We develop an overlapping generations model and calibrate it to data from Brazil, where such a policy is widely implemented. In our model, parents choose how much to invest in their child’s education, thereby increasing both human capital and likelihood of college admission. We find that, in the long run, the optimal income-based affirmative action increases welfare and aggregate output. It improves the pool of admitted students but distorts pre-college educational investments. The welfare-maximizing policy benefits lower- to middle-income applicants with income-based quotas, while higher-income applicants face fiercer competition in college admissions. The optimal policy reduces intergenerational persistence of earnings by 5.7% and makes nearly 80% of households better off.
French - Mother Tongue
Teaching at Queen Mary university of London
Strategy, Leadership and Management of Investment Banks
Teaching at USC Marshall
Microeconomics for Business
Topics in Microeconomic Theory II (PhD level)
Spring 2017, 2018, 2019, 2020