Information elicitation, incentives, and markets
- Information plays a prominent role in economic decisions. Over the past decade, with the enhancement of communication technology and the development of large marketplaces, information economics has become a key component in the study of modern economic environments. In my thesis, I ask how an uninformed principal can purchase information from supposedly better informed agents. My doctoral research strives (i) to extend the theoretical foundations of information elicitation, and (ii) to provide efficient means of elicitation that can be implemented with existing computational resources. I first consider the problem of incentivizing a single agent to provide accurate probabilistic information, that is, information about the uncertainty of a future outcome. This outcome can represent, for example, future sales or stock market prices, and the principal may wish to learn the average sales or the volatility of a stock. I show that the principal may not obtain all information directly, and characterize the information that can be obtained, along with the appropriate incentive devices. For example, she can get the average sales but not the stock volatility. However, the principal can always obtain any information indirectly, by asking for extra information. This leads to a notion of elicitation complexity: how much is needed to infer the information of interest. I then consider the problem of elicitation from multiple agents, using betting markets or prediction markets, which are contingent claims markets specifically designed for the purpose of aggregating traders information. For both cases, I describe the information that such markets can reveal, and explain how to design the assets and the market mechanisms so as to obtain the information of interest. Finally, I consider the case of eliciting arbitrary opinions. As those can be purely subjective, it is generally not possible to design appropriate incentive schemes for a single agent, but I show that with multiple agents one may design surveys or opinion polls that enforce truthful reporting.
|Type of resource
|electronic; electronic resource; remote
|1 online resource.
|Lambert, Nicolas Sebastien
|Stanford University, Computer Science Department
|Statement of responsibility
|Nicolas Sebastien Lambert.
|Submitted to the Department of Computer Science.
|Thesis (Ph. D.)--Stanford University, 2010.
- © 2010 by Nicolas Sebastien Lambert
Also listed in
Loading usage metrics...