A Groves-Like Mechanism in Risk Assessment

Abstract

This paper links two research areas that have developed independently—incentives compatibility for public goods and elicitation of subjective probabilities. An analogy between incentives for reporting information in the two areas leads to the discovery of a new mechanism, based on the Groves mechanism, for eliciting subjective probabilities. In the public goods area, the analogy provides an extension of the basic theorem of truthful response to the more general case when one's true valuation of the public good is state dependent. In the risk assessment area, the analogy provides a generalization of the traditional reporting mechanisms, proper scoring rules, and in doing so establishes a representation theorem for them. The paper considers three goals which a principal might have while choosing a transfer mechanism. These goals are: information pooling, strong research incentives for the agents, and identifiability of the agent with the best information. For two structures of information and the specific cases considered, the new mechanism performs well, compared with four traditional mechanisms, in achieving these goals.

Files

Additional details