Research

Inertial Coordination Games (joint with Andrew Koh and Kei Uzui)

Working paper: September 2024. Accepted for presentation at EC '25 and ESWC '25. Slides here.

We analyze inertial coordination games: dynamic coordination games with an endogenously changing state that depends on (i) a persistent fundamental that players privately learn about; and (ii) past play. We give a tight characterization of how the speed of learning shapes equilibrium dynamics: the risk-dominant action is selected in the limit if and only if learning is slow such that posterior precisions grow sub-quadratically. This generalizes results from static global games and endows them with an alternate learning foundation. Conversely, when learning is fast, equilibrium dynamics exhibit persistence and limit play is shaped by initial play. Whenever the risk dominant equilibrium is selected, the path of play undergoes a sudden transition when signals are precise, and a gradual transition when signals are noisy.

Competition and Social Learning (joint with Andrew Koh)

Working Paper: February 2024. Presented at the 2023 Stony Brook International Conference on Game Theory, the 2024 Northwestern Summer School in Economic Theory, and the 2024 Asian School in Economic Theory at NYU Abu Dhabi.

We study social learning through reviews in a market with horizontally and vertically differentiated firms. Heterogeneous match qualities among consumers give rise to rich belief and buying dynamics. We give a tight characterization of possible limit buying outcomes and resulting welfare implications: inefficiencies arise only from suboptimal horizontal matches because not enough high quality firms are learnt about; by contrast, vertical matches are always optimal. We uncover a fundamental tradeoff between the speed and breadth of learning: faster learning via more precise signals about product qualities lead to insufficient exploration. We show that noisy and asymmetric signals guarantee full learning and restore efficiency although consumers are myopically optimizing and do not internalize the learning externality. We discuss implications for review and platform design.

Inactive Working Papers

Bayesian Persuasion and Data Brokers (joint with Kei Uzui)

Working Paper (inactive)

We study Bayesian persuasion for a binary decision problem in the presence of a data broker intermediary, who can garble the information she receives to maximize her revenue. We derive several structural properties of the broker’s optimal menu under her information constraint, including that optimal and responsive menus must lie on the information frontier. Under binary signals, binary types, and symmetric payoffs, we characterize the sender’s solution and compare the equilibrium social value of information to that of the benchmark of ex-ante persuasion. When the proportion of high types is small enough, we find that the presence of the broker strictly increases the equilibrium social value of information.

Dynamic Random Choice

Working Paper (inactive)

I study dynamic random utility with finite choice sets and exogenous total menu variation, which I refer to as stochastic utility (SU). I obtain two characterizations of SU when each choice set but the last has three elements, and I obtain a third characterization without cardinality restrictions. All of my results hold over an arbitrary finite discrete time horizon.