Consider a repeated interaction where it is unknown which of various stage games will be played each period. This framework separates the basic logic of intertemporal incentives from the requirement that any given strategy profile yields a well‐defined payoff vector. A natural solution concept is ex post perfect equilibrium: strategies must form a subgame‐perfect equilibrium for any realization of the sequence of stage games. When there is one long‐run player and others are short‐run, and public randomization is available, we can adapt the standard recursive approach to determine the maximum feasible gap between reward and punishment for the long‐run player. This allows us to identify which actions can be played in equilibrium and, assuming perfect monitoring, to fully characterize what outcome paths can arise. With multiple long‐run players or no public randomization, the approach fails; a diagnostic of this failure is that optimal penal codes may no longer exist.

A principal wishes to screen an agent along several dimensions simultaneously. The agent has quasilinear preferences that are additively separable across the various components. We consider a robust version of the principal’s problem, in which she knows the marginal distribution of each component of the agent’s type, but does not know the joint distribution. Any mechanism is evaluated by its worst-case expected profit, over all joint distributions consistent with the known marginals. We show that the optimum for the principal is simply to screen along each component separately. This result does not require any assumptions (such as single-crossing) on the structure of preferences within each component. Applications of the model include monopoly pricing and dynamic taxation. This paper has greatly benefited from conversations with Florian Scheuer, as well as helpful comments from (in random order) Richard Holden, Dawen Meng, Andy

We study the question of whether local incentive constraints are sufficient to imply full incentive compatibility in a variety of mechanism design settings, allowing for probabilistic mechanisms. We give a unified approach that covers both continuous and discrete type spaces. On many common preference domains—including any convex domain of cardinal or ordinal preferences, single-peaked ordinal preferences, and successive single-crossing ordinal preferences—local incentive compatibility (suitably defined) implies full incentive compatibility. On domains of cardinal preferences that satisfy a strong nonconvexity condition, local incentive compatibility is not sufficient. Our sufficiency results hold for dominant-strategy and Bayesian Nash solution concepts, and allow for some interdependence in preferences.

We consider a moral hazard problem where the principal is uncertain as to what the agent can and cannot do: she knows some actions available to the agent, but other, unknown actions may also exist. The principal demands robustness, evaluating possible contracts by their worst-case performance, over unknown actions the agent might potentially take. The model assumes risk-neutrality and limited liability, and no other functional form assumptions. Very generally, the optimal contract is linear. The model thus offers a new explanation for linear contracts in practice. It also introduces a flexible modeling approach for moral hazard under nonquantifiable uncertainty. (JEL D81, D82, D86)

We study a class of models of moral hazard in which a principal contracts with a counterparty, which may have its own internal organizational structure. The principal has non‐Bayesian uncertainty as to what actions might be taken in response to the contract, and wishes to maximize her worst‐case payoff. We identify conditions on the counterparty's possible responses to any given contract that imply that a linear contract solves this maxmin problem. In conjunction with a Richness property motivated by much previous literature, we identify a Responsiveness property that is sufficient—and, in an appropriate sense, also necessary—to ensure that linear contracts are optimal. We illustrate by contrasting several possible models of contracting in hierarchies. The analysis demonstrates how one can distill key features of contracting models that allow their findings to be carried beyond the bilateral setting.

A receiver wants to learn multidimensional information from a sender, and she has the capacity to verify just one dimension. The sender's payoff depends on the belief he induces, via an exogenously given monotone function. We show that by using a randomized verification strategy, the receiver can learn the sender's information fully in many cases. We characterize exactly when it is possible to do so. In particular, when the exogenous payoff function is submodular, we can explicitly describe a full‐learning mechanism; when it is (strictly) supermodular, full learning is not possible. In leading cases where full learning is possible, it can be attained using an indirect mechanism in which the sender chooses the probability of verifying each dimension.

The standard revenue-maximizing auction discriminates against a priori stronger bidders so as to reduce their information rents. We show that such discrimination is no longer optimal when the auction’s winner may resell to another bidder, and the auctioneer has non-Bayesian uncertainty about such resale opportunities. We identify a “worst-case” resale scenario, in which bidders’ values become publicly known after the auction and losing bidders compete Bertrand-style to buy the object from the winner. With this form of resale, misallocation no longer reduces the information rents of the high-value bidder, as he could still secure the same rents by buying the object in resale. Under regularity assumptions, we show that revenue is maximized by a version of the Vickrey auction with bidder-specific reserve prices, first proposed by Ausubel and Cramton (2004). The proof of optimality involves constructing Lagrange multipliers on a double continuum of binding non-local incentive constraints.