Please note that I am no longer updating this site. More recent drafts and new papers may be found at
Thank you very much!
The Allocation of Future Business: Dynamic Relational Contracts with Multiple Agents (New Version!) (with Isaiah Andrews)
Consider a repeated moral hazard problem involving a principal and several agents. If formal contracts are unavailable and agents observe only their own relationships, an optimal relational contract allocates business among agents depending on past performance. If first-best is attainable, the principal favors an agent who performs well with future business, even if he later performs poorly. The agent loses favor only if he cannot produce and a replacement performs well. If first-best is unattainable, optimal relationships may deteriorate into persistent low effort. In the first best, the principal need not conceal information from agents; otherwise, he optimally conceals information.
I prove an efficiency result for dynamic games of imperfect public monitoring in which one player's utility is privately known and evolves according to an irreducible Markov process. Under assumptions about the set of payoffs, patient players are able to attain approximately Pareto efficient payoffs in equilibrium. The public signal must satisfy a “pairwise full rank” condition that is somewhat stronger than the related condition required in the Folk Theorem proved by Fudenberg, Levine, and Maskin (1994). My proof is partially constructive and uses a novel technique to mitigate the impact of private information about utility on continuation payoffs. Under stronger assumptions, the efficiency result partially extends to games in which the private information affects every player's payoff.
Appendices B and C for Attaining Efficiency with Imperfect Public Monitoring and Markov Adverse Selection
I examine relational contracts and investment in a market for intermediate goods using a tractable game of imperfect private monitoring. A downstream firm requires a single product in each period from a market of suppliers. If output is not contractible, I show that the downstream firm relies on a small network of suppliers. These upstream firms choose to “put the relationship first:” by investing to produce many products that are required by the downstream firm, they lock themselves into the relationship, thereby increasing effort provision. Using this framework, I consider why suppliers might resist socially efficient legal reform, and discuss implications for employment and ex ante human capital investments.
Work in Progress
Reel Authority (with Robert Gibbons, Ricard Gil, and Kevin Murphy)
Motivated by exhibitor-distributor contracts in the movie industry (and especially by renegotiation of revenue-sharing contracts after the movie completes its run), we analyze a model of formal and relational contracts and test its implications using data for movie releases in Spain. A distributor and exhibitor can formally contract on total box office revenue, but they cannot contract on the other movies that the exhibitor might show - that is, on the opportunity cost of showing this distributor's movie. The optimal relational contract compensates the exhibitor for playing the distributor's movie when doing so is efficient but the formal contract alone would make an alternative movie more attractive. Consistent with this theory, we find that ex post renegotiations occur more frequently and involve larger concessions by the distributor when the exhibitor has a more valuable outside option.
Meritocracy, Moral Hazard, and the Prevalence of Promotions as Incentives (with Michael Powell)
We consider the role of promotions when managers use relational incentive contracts to motivate their employees. The manager would like to simultaneously induce her employees to work hard and promote the employee who would be most productive in the promoted task. When perfect formal contracts are available, the manager motivates her employees using bonus schemes, and promotions are meritocratic in the sense that the most talented employee is given a promotion regardless of her past output. When formal contracts are unavailable, in contrast, we show that promotions complement promised bonus schemes in a relational incentive contract: by promising to promote a high-performing employee, the manager increases the future expected surplus created by that employee precisely in the states of the world in which large bonuses are called for, lending credibility to such promises. As a result, the optimal relational promotion scheme may be “achievement-based” and promote an employee who performed well in the past, even if he is a poor fit for the new position. Further, managers endogenously commit not to demote such employees in the future, as demotions undermine the very credibility that achievement-based promotions were designed to build. These results are consistent with the well-known puzzle that promotions are used as rewards and the “Peter Principle” (employees are promoted to their level of incompetence) and its corollary (that there they remain).
Relational Contracts and Institutions: A Simple Model of "Varieties of Capitalism"