Moral Hazard, Learning and Temporal Dependence: Dynamic Incentive Problems in Firms and Markets

University Of Texas At Austin, Austin TX

Investigators

Abstract

Incentive problems are pervasive in modern societies - they arise within firms, in the interaction between suppliers and consumers, and in online markets. This project examines how incentive problems arise in dynamic environments, and the mechanisms that may be used in order to achieve socially efficient outcomes. The first part of the project will examine dynamic incentive problems that arise when a firm introduces new technology. The efficacy of the technology may be uncertain, and both firm and workers will learn about its effects over time. The PIs will examine how these incentive problems may impede innovation, and the mechanisms that firms may use in order to mitigate inefficiency. A second part examines the interaction between supermarkets and their consumers. Supermarkets have an incentive to know the consumer's past shopping behavior, and the project will examine how the privacy of the consumer's transactions improves her bargaining position. The third part of the project will examine the functioning of illegal online drugs markets such as Silk Road and its successors. Despite the fact that these markets are illegal, are operated by criminals and buyers and sellers are anonymous, they appear to function exceedingly well. The project will examine how these markets overcome the severe moral hazard problems, and the mechanisms that law enforcement agencies may use in order to hinder illegal online trade. The analysis of incentive problems usually involves two dichotomous classes of model: moral hazard (or hidden action) models and hidden information models. This project examines how moral hazard in dynamic environments gives rise to hidden information. This phenomenon may arise when firm and workers are learning about an uncertain technology or when the future costs/benefits to an agent depend upon her current actions, and these actions are unobserved by the principal or by third parties. The proposed papers deal with a diversity of economic situations, from incentive provision within firms to the pricing strategies of supermarkets. They are unified by a common methodological theme and insight. The action of an agent today affects her preferences and payoffs tomorrow. Third parties, such as the principal or a future supplier to the consumer, do not observe the agent's action. This gives rise to a form of endogenous asymmetric information. Asymmetric information may only be latent, i.e. it may not arise on the equilibrium path, as is the case where the agent's equilibrium strategy is a pure strategy. Nonetheless, its possibility affects the incentive problem. When the agent randomizes, so that the equilibrium is in mixed strategies, asymmetric information arises on the path of play, giving rise to a mechanism design or screening problem. The novelty here is that the distribution of types is endogenous. This project will examine how hidden action with dynamic consequences aggravates agency problems in long-term relationships; how the nature of technology affects the choice of organizational form; how inter-temporal price competition between firms such as supermarkets affects the efficiency of allocations. The analysis will focus on situations where parties cannot make long term commitments and will rely on insights and techniques from the theory of dynamic games, combined with those from contract theory and mechanism design. This project also has a second component, that is motivated by the rise of online illegal drugs markets, such as Silk Road. Moral hazard problems would seem to be severe, given that agents are anonymous and have no legal recourse. How are moral hazard problems solved in competitive markets? How do these solutions affect price mark-ups and the extent of price dispersion? Can moral hazard be leveraged in order to hinder the illegal drugs trade? The research in this section builds on the theory of repeated games, played in the context of competitive markets. It also has an empirical component, that builds on data that is being collected by the PI and his collaborators since August 2013, accessing and scraping data from dark-net websites.