Publications
Long-Run Choice Anomalies in Reinforcement Learning with Bounded Memory (with Erin Giffin) (Journal of Economic Behavior & Organization)
Violations of expected utility (EU) maximization have been demonstrated in many settings; however, anomalies are often reduced after repeated choices. We examine if sufficient experiential learning allows convergence to EU-maximization. In the model, a decision maker with long but finite memory repeatedly makes choices in the same decision problem with uncertainty. We focus on the existence and severity of a certain unambiguous type of long-run choice anomaly: ranking reversals (a non-EU maximizing action being most frequently chosen in the long run). We show reversals exist for almost all preferences, even in realistic examples. Reversals tend to happen when payoff differences are heavily skewed. Longer memory does not eliminate the possibility of ranking reversals, but it does make reversals less severe. Our key takeaway is that finite memory can produce major violations of the expected utility ranking even in a model where both memory and the decision-making process are unbiased.
Reselling Information (with S. Nageeb Ali and Ayal Chen-Zion) (Games and Economic Behavior)
Information can be simultaneously consumed, replicated, and sold to others. We study how resale affects a decentralized market for information. Even if the initial seller is an informational monopolist, she captures non-trivial rents from at most a single buyer in any Markovian equilibrium: in the frequent-offer limit, her payoffs converge to 0 once a single buyer buys information. By contrast, there exists a non-Markovian “prepay equilibrium” where payment is extracted from most buyers before information is sold. This prepay equilibrium exploits buyers' ability to resell information and results in the seller achieving (approximately) the same payoff that she would were resale prohibited.
Risk Preferences and Incentives for Evidence Acquisition and Disclosure (with Erin Giffin) (Journal of Law, Economics, and Organization) Online Appendix
Civil disputes feature parties with biased incentives acquiring evidence with costly effort. Evidence may then be revealed at trial or concealed to persuade a judge or jury. Using a persuasion game, we examine how a litigant's risk preferences influence evidence acquisition incentives. We find that high risk aversion depresses equilibrium evidence acquisition. We then study the problem of designing legal rules to balance good decision making against the costs of acquisition. We characterize the optimal design, which differs from equilibrium decision rules. Notably, for very risk-averse litigants, the design is "over-incentivized" with stronger rewards and punishments than in equilibrium. We find similar results for various common legal rules, including admissibility of evidence, maximum penalties, and maximum awards. These results have implications for how rules could differentiate between high risk aversion types (e.g., individuals) and low risk aversion types (e.g., corporations) to improve evidence acquisition efficiency.
Working Papers
Strategic Cyberwarfare (with Rishi Sharma) (Revision resubmitted to Economic Inquiry)
This paper develops a theoretical model of cyberwarfare between nations, focusing on the factors that determine the severity and outcomes of cyber conflicts. We introduce a two-country model where nations invest in offensive or defensive cyber capabilities across networked systems. We show that resource expenditure intensifies when players' effective values are similar, which can help explain the rise of cyberwarfare. We explore the implications of network structures, showing how larger attack surfaces worsen outcomes for defenders. Additionally, we investigate the impact of private cyber defence provision, and find that centralized policies may either improve or exacerbate cyber conflict.
Information Design and Reputations in the Frequent-Interaction Limit (Revision resubmitted to the Journal of Mathematical Economics)
Consumer welfare may be higher when platforms sometimes conceal product quality information, because this can incentivize firms to invest more in quality. I study an infinite horizon model in which a long-run firm with a persistent type interacts with a sequence of short-run consumers. The focus is on the case where there are frequent consumer purchases. When a designer commits in advance to an information policy to maximize average consumer welfare, the optimal policy and the resulting welfare depend on whether the policy may depend on the entire history (non-public) or only the publicly observable history. An optimal non-public policy approximates the upper bound welfare that is consistent with firm participation. It may do so by privately keeping track of a firm rating, threatening a hypothetical punishment (revealing firm type) off the path of play, while never giving consumers information on the path of play. However, learning must occur on the path of play in an optimal public policy. A public policy can only achieve the upper bound welfare conditional on beliefs being above some cutoff. The optimal public policy features a learning cutoff: The firm must pass an initial test, driving beliefs up to the cutoff, which is essential for rewarding the firm.
Long-Run Efficiency with Local Interactions and Heterogeneous Types (Revise and resubmit at Games and Economic Behavior)
People often engage in strategic interactions in different locations. Previous literature with homogeneous player preferences showed that location choice causes evolutionary selection in favor of efficient equilibria. In the context of 2 x 2 games with two heterogeneous types, this paper characterizes the long-run equilibria in two interesting classes of games, Opposing Coordination (OC) games and Coordination/Anti-coordination (CA) games. In OC games, the long-run equilibria are all efficient and feature segregation of the two types. This results from strategic homophily, because the two types disagree about which equilibrium is efficient. In CA games, the long-run equilibria all feature mixing of types, which is inefficient for the coordination type. Moreover, if miscoordination is harmful enough, there is a second inefficiency: the coordination types may play their inefficient strategy. This results from the strategic heterophily of the anti-coordination types.
Works in Progress
Information Design and Evolutionary Learning
Evolutionary Game Theory on Networks (with Shahir Safi)