CAREER:Foundational Questions in the Theory of Incentives

该职业奖资助机制设计和契约理论，即激励效应的经济理论的研究。目前的核心数学模型在简单环境中很容易理解，但很难扩展到更复杂的环境中。这个项目将开发新的基本建模工具，目的是扩大理论的范围。这些工具将使研究人员能够更好地为复杂的环境设计激励方案，包括在经济环境不确定的情况下保证绩效的激励方案。综合教育计划包括一个跨学科会议，将机制设计理论家与潜在应用领域的专家联系起来；并与对高等数学问题解决感兴趣的高中生和教育工作者开展外展活动，以建立对这一富有成果的研究领域的认识。本研究对国家利益的促进作用在于指导更好的激励机制的设计。这一领域的现有理论在历史上有许多应用，包括税收、监管、采购和公司融资。新的理论将允许在这些和其他领域进一步改进和提高效率。整个项目有三个主要组成部分。(1)多维机制设计。主要的主力模型通常假设代理人的偏好由一维参数总结；将这些模型扩展到多个维度被广泛认为是具有挑战性的，但也很重要。这项研究将采用一种新的建模方法，基于偏好各维度之间相关性的不确定性，来研究一些关于何时以及为什么简单机制可以提供良好保证的问题。(2) agent信息内生时的机制设计。在许多情况下，在进行经济互动之前，代理人可能会采取代价高昂的行动来获取相关信息或影响他人的信息，而设计互动的计划者可能会担心这种博弈是一种社会浪费。当考虑到这种“信息博弈”时，该项目描述了提供可靠保证的机制。(3)在道德风险情况下的激励设计，当不知道代理人可以采取什么行动时。该项目考虑了关于这一主题的几个理论问题，重点是理解简单的激励结构在复杂情况下何时可以达到最优鲁棒性。研究计划还包括了一些相关的研究方向，包括用简单环境逼近复杂环境、激励设计者进行最优实验等。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This CAREER award funds research in mechanism design and contract theory, the economic theory of the effects of incentives. Current core mathematical models are well-understood for simple environments but difficult to extend to more complicated environments. This project will develop new basic modeling tools, with the aim of broadening the scope of the theory. The tools will allow researchers to better design incentive schemes for complex environments, including incentive schemes that guarantee performance despite uncertainties about the economic environment. The integrated education plan includes an interdisciplinary conference to connect theorists in mechanism design with experts in potential application areas; and outreach activities with high school students and educators interested in advanced mathematical problem solving to build awareness of this fruitful area of study. This research promotes the national interest because it will guide the design of better incentive mechanisms. Existing theory in this area has historically had many applications, including taxation, regulation, procurement, and corporate finance. New theory will allow further improvements and better efficiency in these and other areas. The overall project has three main components. (1) Multidimensional mechanism design. Major workhorse models typically assume agents' preferences are summarized by a one-dimensional parameter; extending these models to multiple dimensions is widely viewed as challenging, yet important. This research will adapt a new modeling approach, based on uncertainty about the correlation across dimensions of preference, to study several questions concerning when and why simple mechanisms can deliver good guarantees. (2) Mechanism design when agents' information is endogenous. In many situations, agents can take costly actions to either acquire relevant information or influence others' information before an economic interaction, and a planner designing the interaction may be concerned that this gaming is socially wasteful. This project characterizes mechanisms that give robust guarantees when such "information games" are considered. (3) Design of incentives in situations of moral hazard, when it is unknown exactly what actions agents can take. This project considers several theoretical questions on this theme, focusing on understanding when simple incentive structures can be optimally robust in complex situations. The research plan also includes some related directions of study, including approximation of complex environments by simple ones, and optimal experimentation by incentive designers.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.