Papers
Papers
Abstract. Why do individuals interpret the same information differently? We propose that individuals follow Bayes' Rule when forming posteriors with one exception: when assessing the credibility of experts,they "double-dip" the data and use already-updated beliefs instead of their priors. Our proposed mistake predicts when individuals over- or underreact to new information, depending on the order in which they received previous signals. It explains why information that should objectively mitigate disagreement may amplify it instead. In a trading game application, traders engage in excessive speculation associated with bubbles and endogenous crashes. Our model provides a theory of the origins of disagreement: individuals disagree about both unknown states and credibility despite sharing common priors and information.
NBER Summer Institute video presentation here.
Abstract. We model the reporting of sexual misconduct. Individuals under-report misconduct due to strategic uncertainty over whether others will report and corroborate a pattern of behavior. Under-reporting occurs if and only if misconduct is widespread. Making sanctions more responsive to reports, raising public awareness of misconduct, implementing confidential holding tanks, and appropriately calibrating damage awards can encourage reporting. However, we also show when such policies are ineffective or backfire. Managers may avoid mentoring subordinates, spilling over into reporting. A holding tank may discourage reporting by raising the bar to access reports. Overall, we highlight several unintended and intended consequences of #MeToo.
Abstract. This paper studies the role of goal bracketing to attenuate time inconsistency. When setting non-binding goals for a multi-stage project, an agent must decide how to group, or bracket, such goals for evaluation. He can bracket broadly by setting an aggregate goal for the entire project, or narrowly by setting incremental goals for individual stages. A loss averse agent brackets optimally by trading off motivation and risk pooling, which interacts non-trivially with time discounting. An aggregate goal becomes more attractive as early-stage uncertainty increases, while incremental goals become more attractive when later-stage uncertainty increases.
Abstract. This paper offers an information-based model of social interaction, and analyzes optimal investment and pricing of services that facilitate interaction in a duopoly. Agents have uncertainty over their preferences but are aware that they are correlated with others', so there exists an incentive to communicate with others in the population. When a firm's good
can be bundled with a coordination mechanism for its customers, its value is endogenously determined due to a consumption externality. Although this mechanism increases total surplus, it is underprovided and consumer surplus decreases.
Abstract. This paper develops a theoretical model of consumer demand for an energy conservation program that involves non-binding, self-set goals. We present evidence from a Northern Illinois goal-setting program, aimed at reducing residential electricity consumption, that is difficult to reconcile with standard preferences and is broadly consistent with a model of
present-biased consumers with reference-dependent preferences. We find that the need for commitment is correlated with program adoption, higher pre-adoption consumption, and lower responsiveness to goals. Consumers choosing realistic goals persistently save substantially more, achieving savings of nearly 11%, than those choosing very low or unrealistically high goals.
Previous version with social comparison here.
Abstract. This paper addresses the role of non-binding goals to attenuate time inconsistency. Present-biased agents have linear reference-dependent preferences and endogenously set a goal that is the reference point. They face an infinite horizon, optimal stopping problem in continuous time, where there exists an option value of waiting due to uncertainty. When there is sufficient commitment to expectation-based goals, goal-setting attenuates the time-inconsistent agent's tendency to stop too early, and may even lead an agent to wait longer than the first-best. In particular, reference dependence is strictly worse for a time-consistent agent. Notably, none of the effects of goal-setting require any form of loss aversion.
Abstract. We propose a theory of persistent misinformation and study how measures to combat such misinformation affects receivers' actions. We introduce a model with "doublespeak" equilibria where receivers fail to learn the true state despite infinite messages due to uncertainty over the sender's motivations. A benevolent policymaker may restrict communication in response to similar concerns, but the optimal choice may be counterintuitive and allow some messages even when the sender is likely a non-good type. Endogenous fact-checking by receivers does not mitigate equilibrium misinformation, and reputation concerns by senders may or may not increase information transmission. Our results highlight the difficulty of eliminating the effects of misinformation in equilibrium.