Research

Primary fields:       microeconomic theory, game theory, information economics.
Secondary fields:   organizational economics.

Working papers:

  • Cheap Talk with Transparent Motives (with Doron Ravid)

    We study a model of cheap talk with one substantive assumption: the sender’s preferences are state-independent. We observe that this setting is amenable to the belief-based approach familiar from models of persuasion with commitment. Using this approach, we examine the possibility of valuable communication, assess the value of commitment, and explicitly solve for sender-optimal equilibria in a large class of examples. A key product is a geometric characterization of the value of cheap talk, described by the quasiconcave envelope of the sender’s indirect utility.

  • Repeated Delegation (with João Ramos)

    We study an ongoing relationship of delegated decision making. Facing a stream of projects to potentially finance, a principal must rely on an agent to assess the returns of different opportunities; the agent has lower standards, wishing to adopt every project. In equilibrium, the principal allows bad projects in the future to reward fiscal restraint by the agent today, but she cannot commit to reward the agent indefinitely. We fully characterize the equilibrium payoff set (at fixed discounting), showing that Pareto optimal equilibria can be implemented via a two-regime `Dynamic Capital Budget’. We show that, rather than backloaded rewards—a prevalent feature of dynamic agency models with greater commitment power—our Pareto optimal equilibria feature an inevitable loss of autonomy for the agent as time progresses. This transition toward conservatism speaks to the life cycle of an organization: as it matures, it generates lower revenue at a higher yield.

  • Disclosure to a Psychological Audience (with Laurent Mathevet)

    We study how a benevolent expert should disclose information to an agent with psychological concerns. We first provide a method to compute an optimal information policy for many psychological traits. The method suggests, for instance, that an agent suffering from temptation à la Gul & Pesendorfer (2001) should not know what he is missing, thereby explaining observed biases as an optimal reaction to costly self-control. We also show that simply recommending actions is optimal when the agent is intrinsically averse to information but has instrumental uses for it. This result, which circumvents the failure of the Revelation Principle in psychological environments, simplifies disclosure and informs the debate regarding mandated disclosure.

  • Job Insecurity (with Aditya Kuvalekar)

    This paper examines the effect of job insecurity on productivity. We study a fixed-wage relationship between a firm and a worker in which neither knows how well-suited the worker is to the job. The worker decides the level of effort, a choice that affects both learning and the firm’s bottom line. The employer, seeing the worker’s effort choice and outcome, decides whether or not to continue employing the worker. When the employer becomes pessimistic enough about the match quality, she cannot commit not to fire the worker. We show that, rather than aligning interests, this threat creates a perverse incentive not to attract attention: the worker strategically slows learning, harming productivity. As the firm anticipates this, job insecurity can be a self-fulfilling prophecy. We explicitly characterize the unique Markov perfect equilibrium in our continuous time dynamic game. Consistent with empirical evidence in organizational psychology, equilibrium exhibits a U-shaped relationship between job insecurity and productivity: a worker is least productive when his job is moderately secure.

  • Peer-Confirming Equilibrium (with Evan Sadler)

    We can often predict the behavior of those closest to us more accurately than that of complete strangers, yet we routinely engage in strategic situations with both: our social network impacts our strategic knowledge. Peer-confirming equilibrium describes the behavioral consequences of this intuition in a noncooperative game. We augment a game with a network to represent strategic information: if two players are linked in the network, they have correct conjectures about each others’ strategies. In peer-confirming equilibrium, there is common belief that players (i) behave rationally and (ii) correctly anticipate neighbors’ play. In simultaneous-move games, adding links to the network always restricts the set of outcomes. In dynamic games, the outcome set may vary non-monotonically with the network because the actions of well-connected players help poorly-connected players coordinate. This solution concept provides a useful language for studying public good provision, highlights a new channel through which central individuals facilitate coordination, and delineates possible sources of miscoordination in protests and coups.

  • Simplifying Bayesian Persuasion (with Laurent Mathevet)

    In Bayesian Persuasion (Kamenica and Gentzkow (2011)), the sender’s optimal value is characterized by a concave envelope. Since concavification of a general function is notoriously difficult, we propose a method to reduce the problem, using the underlying economic structure of the indirect expected utility. The key observation is that one can find, using the receiver’s preferences alone, a small set of posterior beliefs on which some optimal information policy must be supported. This simplifies, sometimes dramatically, the search for optimal information.

Work in progress: