ICES: SMALL: Mechanism Design with Information-Sensitive Agents

Northwestern University, Evanston IL

Investigators

Abstract

Mechanism design studies the problem of designing interactions that yield socially, politically, and economically desirable outcomes when the agents participating in the interaction are self-interested. While agents' interests are typically captured by preferences over material outcomes such as money and consumption goods, this is often insufficient for real-world interactions. For example, many of today's social and economic interactions take place online, and records of the interactions are archived, maintained, and often spread to others. In such interactions agents are motivated not only by the resulting material outcomes, but possibly also by concerns about what and how much of their private information is revealed in the course of the interaction. Agents may, for example, prefer one outcome to another when their private information is revealed, but may prefer the latter outcome to the former if no information is revealed. They may value privacy, and wish to prevent the leakage of their private information. Alternatively, they may wish to credibly disclose their private information as a signal to others. This project develops a model of information-sensitive agents that captures these kinds of preferences, and examines the problem of designing ineractions for such agents. Some of the questions addressed in this project are: Do the mechanisms that exist in the current theoretical literature and in practice still "work" when agents have information-sensitive preferences? Is it possible to design mechanisms that are robust to various levels of information-sensitivity of the agents? Are there mechanisms that have additional desirable properties, such as the preservation of privacy or the credible disclosure of information? The broader impacts of this project are twofold. First, a prevalent form of information-sensitivity is a predilection for privacy, one of the most debated issues in electronic commerce. Since this project examines the strategic implications of privacy, it has the potential to contribute valuable insights to this heated debate. Second, information-sensitivity is particularly relevant when agents participate in numerous mechanisms, in which case the leakage of information from one mechanism to another becomes a critical consideration. By explicitly including concerns for information leakage, this project facilitates the design and analysis of economic mechanisms that are both more robust and more modular.