Collaborative Research: Dynamic Mechanisms without Transfers

Yale University, New Haven CT

Investigators

Abstract

This project studies how to best allocate resources over time when money isn't available as a means of eliciting the agents' private information - an essential ingredient in the determination of the efficient allocation. Property rights to the allocation must dynamically respond to claims by the agents regarding their needs, punishing greedy agents and rewarding frugal ones, while attempting to preserve efficiency. Potential applications range from how to allocate hospital beds; how a nurse should divide up her time; how an agency should pick among competing projects, etc. The goal is to derive transparent and robust yet effective decision rules. We allow the agent's private information to be serially correlated over time (but independent across agents), and consider both the cases in which agents desire only one unit (e.g., a medical procedure), and in which their payoff is additive in the number of units they get. Values are private. The principal is committed to the mechanism, and the resource supply is exogenous. In some simple settings, a two-part tariff using a virtual budget implements the optimum. For more complicated settings, the project will rank standard procedures (e.g., auctions with virtual budgets) and compare their performance to the best mechanism.