Eliciting truthful unverifiable information

In many situations, an uninformed agent (UA) needs to elicit information from an informed agent (LA), typically when the latter has some unique expertise or knowledge related to some opportunity available to the UA. In many of those situations, the correctness of the information cannot be verified by the UA, and therefore it is important to guarantee that the information-elicitation mechanism incentivizes the IA to report truthfully. This paper presents and studies several information-elicitation mechanisms that guarantee truthful reporting, differing in the type of costs the LA incurs in producing and delivering the information. We show that with no such costs truthful information elicitation is possible with a positive but arbitrarily small expense for the UA. When information-delivery is costly, truthful information elicitation is possible where the extra expense for the UA (above the unavoidable cost of delivery) is arbitrarily small. Finally, when the information-production is costly, under some realistic condition related to the ratio between the expected gain of the IA from true reporting and the information-production cost, truthful information elicitation is possible where the extra expense for the UA (above the unavoidable cost of production) is arbitrarily small.