Screening Agents in Belief Eliciting Mechanisms

Abstract

This paper considers the problem of a decision maker (DM), who needs to hire an agent to assess the probability of occurrence of an event that is of interest to her. To decide the agent’s reward, the DM proposes a mechanism that will give reward based on the agent’s reported subjective probability and the actual outcome of the event. The reward mechanism needs to incentivize the expert to honestly reveal his subjective probability, and the reward has to be non-negative in all cases in order to ensure agent’s participation. In such a situation, it is possible that there are some agents who lack the expertise to assess the situation, but still participate to get sure non-negative payoff. The DM wants to screen out such uninformed agents from the informed ones. This work considers two mechanisms, and analyzes the behavior of both types of agents for the two mechanisms. It shows that screening is possible along with belief elicitation in some cases.