Doctoral Dissertation Research in Economics: Belief Formation and Choice in Games: An Experiment

许多经济分析都是基于这样一种观点，即经济主体会对激励做出反应，他们会更加努力地理解自己行为的后果，以便做出正确的决定。 这个研究项目将使用实验室实验来测试是否增加奖励或惩罚的大小（以下简称赌注），同时考虑其他人的反应会影响一个人在做出决定之前对环境的研究。 这项研究将回答以下问题：（一）赌注的大小是否影响人们的选择和信念？(ii)更高的风险是否会导致更好的反应？(iii)信念会随着赌注的增加而变得更准确吗？及（iv）这些影响是否取决于问题的复杂程度？ 尽管这些问题对经济科学很重要，但经济学家们还无法检验它们，因为在赌注、信念和行动之间存在反馈效应--改变一个人的赌注会相应地改变他们的信念和行动，因此不可能确定奖励或惩罚的大小如何影响努力。 实验室实验允许研究人员绕过这些问题，旨在允许研究人员改变奖励的大小，同时允许人们扮演决策者和观察者的角色。 这使研究人员能够研究人们如何形成对他人的信念，以及同时对风险变化的反应。 这项研究的结果将有助于改善决策，特别是在与其他人互动时，并在此过程中提高生产力和经济增长。 面临决定的个人会考虑可能的行动方针以及利害关系。此外，如果这种决策的结果取决于其他人的行为，决策者可能会花一些时间来了解自己的行为和其他人的行为如何相互作用。 本研究分析了以下问题：利害关系是否影响选择和信念的形成？更高的风险是否会导致更高的最佳反应率？信念会随着赌注的增加而变得更加准确吗？这些影响取决于游戏的复杂程度吗？ 目前的文献没有提供这些问题的答案，本研究旨在填补这一空白。 这个项目测试了在战略环境中改变利害关系如何影响个人的选择和对其他人可能的战略形成的信念。检验这些假设的一个根本挑战是激励、信念和行动之间存在相互作用效应：改变一个主体的激励不仅会影响该特定个体，还会影响其他会改变行为的人。 该研究项目通过一个实验设计克服了这个问题，其中受试者既是参与者又是观察者。在后一种情况下，他们自己的收益会发生变化，但他们是在其他具有不同激励的参与者已经做出选择的环境中做出选择的。 研究设计能够隔离的影响，按比例放大的选择和信念的均衡。此外，该设计将允许研究人员提供跨游戏比较的新见解。 它还提出了一类新的游戏跨越那些需要多个步骤的迭代删除的劣势策略，那些没有纯策略均衡。这类游戏在与k级和随机效用模型的连接方面具有各种理想的属性。 这项研究的结果将有助于改善决策，特别是在与其他人互动时，并在此过程中提高生产力和经济增长。该奖项反映了NSF的法定使命，并被认为值得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来支持。

Much of economic analysis is based on the idea that economic agents respond to incentives and that they will increase their effort to understand the consequences of their actions in order to make the right decisions. This research project will use laboratory experiments to test whether increasing the size of reward or punishment (stakes henceforth) while considering how others will react affect the effort one puts into studying the environment before s/he makes the decision. The research will answer the following questions: (i) does the size of stakes affect people’s choices and beliefs? (ii) does higher stakes lead to better responses? (iii) do beliefs become more accurate as stakes increase?, and (iv) do these effects depend on how complex the problem is? In spite of the importance of these questions to economic science, economists have not been able to test them because there is a feedback effect among stakes, beliefs, and actions---changing stakes for one person will change their beliefs and actions, accordingly, thus making it impossible to determine how the size of reward or punishment affects effort. The laboratory experiments, which allows the researcher to get around these problems, are designed to allow the researcher to change the size of the reward while allowing people to play the role of decision maker as well as observers. This allows the researcher to study how people form beliefs about others as well as react to changes in stakes at the same time. The results of this research will help improve decision making, especially when interacting with other people, and in the process, increase productivity and economic growth. An individual facing a decision takes into account possible courses of action as well as what is at stake. In addition, if the outcome of such a decision depends on what others do, a decision maker may spend some time to understand how her own action and those of others interact. This research analyzes the following questions: Do stakes affect choices and belief formation? Do higher stakes lead to higher best-response rates? Do beliefs become more accurate as stakes increase? Do these effects depend on how complex the game is? Current literature does not provide answers to these questions and this research aims to fill this gap. This project test how changing the stakes in a strategic environment affect both the choices of the individual and the beliefs formed on others’ likely strategy. A fundamental challenge of testing these hypotheses is that there exists interaction effects among incentives, beliefs, and actions: changing the incentives of one subject will not only affect that particular individual but also on others who will change their behavior. The research project overcomes this issue through an experimental design where subjects act as both players and observers. In the latter, their own payoff changes but they make choices in an environment where choices have already been made by other actors with different incentives. The research design is able to isolate the effect of scaling up the stakes on choices and beliefs from equilibrium. Furthermore, the design will allow the researchers to provide new insights on cross-game comparisons. It also presents a new class of games spanning those requiring multiple steps of iterative deletion of dominated strategies to those without pure-strategy equilibria. This class of games has various desirable properties in terms of their connection to level-k and random utility models. The results of this research will help improve decision making, especially when interacting with other people and in the process increase productivity and economic growth.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.