Doctoral Dissertation Research in Economics: Non-isolation and Social Preference

日常生活中经常会出现这样的场景，人们需要根据当前和之前的互动来确定对他人“有多公平”。决定是否为下一个人开门，让某人在车流中并入前方，或者在螃蟹盛宴上拿走最后一只螃蟹，这些都是日常场景，可能取决于一个人之前的公平程度。多轮公平也可以在实验室中进行实验，也可以在更高风险和经济相关环境中自然地在现场进行。经理必须决定下属（也可能是他们自己）的奖金，这些决定可能取决于也可能不取决于他们过去的奖金决定。在职业棒球中，不成文的规则规定了球员何时应该“投球”来帮助对方球队，尽管这可能会损失数十万美元的奖励。该项目将通过实验室实验来研究这种动态公平问题的特征。除了测试动态公平性问题之外，我们还研究了相关实验环境中的支付方法问题。当受试者被反复询问问题以提供更多数据时，实验室实验中常见的节省成本的措施是随机选择一个决策问题进行付费。这种方法虽然常用，但其可靠性取决于主体的动态公平性担忧。标准分析假设不存在动态：每个选择问题都是孤立看待的。当实验者不知道受试者的担忧时，假设隔离可能会导致关于受试者偏好的错误结论。因此，本项目还将研究支付方法在公平实验设计方面的稳健性。这项研究通过使用最简单的方式更好地理解：（i）个人如何在特定的动态公平环境中做出选择，（ii）风险下社会偏好的哪些理论模型最能代表实际数据，以及（iii）隔离在社会偏好实验中发挥多大作用，对风险下社会偏好和实验方法方面的文献做出了三重贡献。该项目将使用前面提到的“随机问题选择”（RPS）支付机制，通过两轮实验来测试非隔离性的存在。每一轮都是二元结果独裁者游戏，受试者将选择两个选项之一，在自己和另一个匿名受试者之间分配一份钱。特别是，该实验询问先前选择的公平（或不公平）替代结果是否会产生为下一轮选择不公平（或公平）结果的倾向。这种类型的动态响应被称为“选择逆转”，它表明了非孤立性。然而，实验的动态性质不可避免地会引起选择偏差。为了克服这一挑战，实验使用了一种可能早期解决 RPS 方案的程序，它可以关闭 RPS 对后期选择的影响，同时诱导受试者在 RPS 下进行选择以实现早期选择。因此，受试者在后面几轮中的选择不会受到其前一轮选择的概率影响，然后可以将其与对照组进行比较。请注意，非孤立和逆转是两个不同的概念：不孤立问题的主体不一定会引起逆转。因此，重点将是共同检验非孤立行为和逆转行为的存在。同时，选择实验参数来区分风险下的三类社会偏好模型。动态公平问题是经济相关的个人决策问题及其各自福利分析的重要因素。这项研究对理解这些问题的根本贡献将为应用经济学家和理论经济学家选择适当的模型提供建议。对于实验经济学家来说，这项研究将深入了解常见 RPS 支付机制在这种动态环境中的适用性。该奖项反映了 NSF 的法定使命，并通过使用基金会的智力价值和更广泛的影响审查标准进行评估，被认为值得支持。

Daily life regularly consists of scenarios where one needs to determine “how fair” to be to others based on current and previous interactions. Deciding whether to hold the door open for the next person, let someone merge ahead in traffic, or take the last crab at a crab feast are all everyday scenarios that could depend on how fair one has been previously. Multiple rounds of fairness also occur both experimentally in the lab and naturally in the field in higher stakes and economically relevant environments. Managers must decide on bonuses among subordinates (and potentially themselves), and these decisions may or may not depend on their past bonus decisions. In professional baseball, unwritten rules dictate when players should “take a pitch” to help the other team, despite potentially losing hundreds of thousands of dollars in incentives. This project will examine the features of such dynamic fairness concerns using a laboratory experiment. In addition to testing dynamic fairness concerns, we also examine the methodological issue of payment in relevant experimental settings. When subjects are repeatedly asked questions to provide more data, a common cost-saving measure in laboratory experiments is to randomly choose one decision problem for payment. The reliability of this method, while commonly used, is dependent on a subject’s dynamic fairness concerns. The standard analysis assumes that there are no dynamics: each choice problem is viewed in isolation. When the experimenter does not know a subject’s concerns, assuming isolation could give rise to incorrect conclusions about that subject’s preferences. Therefore, this project will also investigate payment methodology’s robustness in terms of experimental design on fairness.This research contributes threefold to the bodies of literature in social preferences under risk and experimental methodology by using the simplest way to better understand: (i) how individuals make choices in a specific dynamic fairness settings, (ii) which theoretical model(s) of social preference under risk best represent actual data, and (iii) how much of a role isolation plays in social preference experiments. This project will experimentally test the presence of non-isolation in a two-round experiment using the aforementioned “random problem selection” (RPS) payment mechanism. Each round will be a binary outcome dictator game, where a subject will choose one of two options to split a share of money between themselves and another anonymous subject. In particular, this experiment asks whether having previously chosen a fair (resp. unfair) alternative outcome creates a tendency to choose an unfair (resp. fair) outcome for the next round. This type of dynamic response is labeled a “choice reversal”, and it identifies non-isolation. However, the dynamic nature of the experiment inevitably induces selection bias. To overcome this challenge, the experiment uses a procedure with possible early resolution of the RPS scheme, which can shut down the effect of RPS in later choice while inducing subject’s choice under RPS for earlier choice. Therefore, the subject's choice in later rounds will not be affected probabilistically from their earlier round choice, which can then be compared to the control group. Note that non-isolation and reversals are two separate concepts: subjects who do not isolate the problem do not necessarily give rise to reversals. Therefore, the focus will be to jointly test the existence of non-isolation and reversal behavior. Simultaneously, the parameters of the experiment are chosen to distinguish between three classes of models of social preferences under risk. Dynamic fairness concerns are important factors in economically-relevant individual decision-making problems and their respective welfare analyses. This research's fundamental contributions to understanding these concerns will advise both applied and theoretical economists on appropriate model selection. For the experimental economist, this research will provide insight into the applicability of the common RPS payment mechanism in such dynamic settings.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.