Doctoral Dissertation Research in Economics: Expected Utility Core of Risk Preferences

在缺乏信心的情况下理解决策是很重要的，因为许多现实生活中的决策都涉及不确定性和复杂的权衡。经济理论将不确定性下行为偏离理性选择模型与决策信心的缺乏联系起来。例如，决策信心的缺乏可能源于个人对冒险意愿的不确定。这个研究项目将使用经济理论和实验室实验来研究人们如何在有风险的选择之间做出决定，以及他们在做出决定时的自信程度。这个研究项目的结果将有助于理解人们的行为如何作为他们决策信心水平的函数。将特别强调在安全选择和风险选择之间的决定，因此有可能检查人们在不自信时是倾向于规避风险还是寻求风险。研究人员还将进行实验室实验，研究经验在决策中的重要性。特别是，研究人员将研究那些在做决定之前熟悉决策环境的人是否对自己的决定更有信心，行为也更理性。这一分析将为旨在通过增加目标人口对决策环境的熟悉程度来促进社会可取行为的政策设计提供新的见解。此外，研究人员还将开发和实施一种新的方法来估计个人在激励实验中承担风险的意愿，并将这种新措施与决策信心联系起来。例如，金融顾问可以使用这种新颖的方法为客户确定合适的投资组合建议。研究人员将进行实验室实验，研究独立公理的失效，了解它们与决策信心和经验的关系，并引发个体对可能违反独立公理的风险厌恶。为此，受试者将在货币彩票中做出选择，并在0到100的范围内声明他们对自己选择的信心。此外，为了研究经验是否会影响决策，在回答这些问题之前，一组受试者将被要求观察从一些成对的彩票中随机抽取的奖金。研究人员将使用期望效用核心的概念来分析实验数据，期望效用核心代表了满足独立性公理的偏好关系的最大部分。研究预期效用核心相当于在局部测试独立性公理对每个单一选择的含义。这种局部分析将允许重新评估选择环境的特征的相关性，这些特征传统上与较高的独立性公理的违反有关，最明显的是存在某种（即概率1）替代方案。此外，它将揭示决策信心和经验在风险决策中的作用。特别是，该项目试图探索决策置信度和违反独立性公理之间的相关性，以及在什么情况下经验可以增加置信度并减少独立性违反。例如，缺乏信心可能源于风险厌恶程度上的优柔寡断。本项目将通过开发一种新的方法来形式化这种直觉，该方法利用预期的效用核心来估计一组可能的风险规避水平，每个级别都由不同的效用函数捕获。这种新颖的方法允许对对风险的态度进行一般性的分析，这种分析对模型错误规范是稳健的，因为它既不依赖于预期效用，也不依赖于特定的替代模型。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。

Understanding decision-making in the absence of confidence is important because many real-life decisions involve uncertainty and complicated trade-offs. Economic theories link the departure of behavior from the rational model of choice under uncertainty to the lack of decision confidence. The lack of decision confidence may arise, for instance, from individuals’ uncertainty over their willingness to take risks. This research project will use economic theory and laboratory experiments to study how people make decisions between risky alternatives and how confident they are when they make their decisions. The results of this research project will help understand how people behave as a function of their level of decision confidence. A particular emphasis will be given to decisions between safe and risky alternatives, for which it will be possible to examine whether people tend to be more risk-averse or risk-seeking when unconfident. The researchers will also conduct laboratory experiments to study the importance of experience in decision-making. In particular, the researchers will study whether people who can familiarize themselves with the decision environment before making a decision tend to become more confident about their decisions and behave more rationally. This analysis will provide new insights into the design of policies aimed at promoting socially desirable behaviors by increasing the familiarity of the targeted population with the decision environment. Moreover, the researchers will also develop and implement a novel methodology to estimate individuals’ willingness to take risks from incentivized experiments and relate this new measure to decision confidence. This novel approach may be used, for instance, by financial advisors to determine appropriate portfolio recommendations for their clients.The researchers will conduct laboratory experiments to study failures of the independence axiom, to understand their relationship with decision confidence and experience, and to elicit risk aversion for individuals who may violate the independence axiom. To this end, subjects will make choices over monetary lotteries and declare on a scale from zero to 100 how confident they are about their choices. Moreover, to study whether experience can affect decision-making, a group of subjects will be asked to observe random draws of monetary prizes from some pairs of lotteries before answering these questions. The researchers will analyze the experimental data using the notion of expected utility core, which represents the greatest part of a preference relation that satisfies the independence axiom. Studying the expected utility core amounts to testing the implications of the independence axiom locally for every single choice. This local analysis will allow reevaluating the relevance of features of the choice environment traditionally associated with higher violations of the independence axiom, most notably the presence of a certain (i.e., probability 1) alternative. Moreover, it will shed new light on the role of decision confidence and experience in decision-making under risk. In particular, the project seeks to explore the correlation between decision confidence and violations of the independence axiom and under what circumstances experience can increase confidence and reduces independence violations. The lack of confidence can arise, for instance, from indecisiveness on the level of risk aversion. This project will formalize this intuition by developing a new methodology that exploits the expected utility core to estimate a set of possible levels of risk aversion, each captured by a different utility function. This novel approach allows for a general analysis of attitudes toward risk that is robust to model misspecification because it relies neither on expected utility nor on a specific alternative model.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.