Temptation and Self-Control in Dynamic Choice

个人做出的选择通常取决于他们做出决定的时间。考虑在第t时期获得较小奖励和在第t+1时期获得较大奖励之间的权衡。在实验中，当面临这样的权衡时，受试者通常会以动态不一致的方式行事：如果决定是在第t-1期或更早的时候做出的，受试者会选择第t+1期更大的奖励；相比之下，如果决策是在第t时期做出的，个体会选择提供即时消费的较小奖励，而不是选择等到第t+1时期才获得较大奖励。同样，当诱人的商品可以立即消费时，人们更容易屈服于诱人但不健康的消费。在这个项目中，我们将与普林斯顿大学的法鲁克·古尔（Faruk Gul）合作，开发一个新的选择理论框架来捕捉这一现象。起点是一组选择的偏好关系。集合表示代理的选择菜单。为了抓住agent可能被事先不希望的选择所诱惑的想法，我们允许偏好表现出对承诺的渴望，也就是说，我们允许对较小的选择集有严格的偏好。在这个框架中，我们认为诱惑和自我控制是行为的主要决定因素。在这项工作中，我们建议将我们已经发展的两周期模型扩展到无限视界。此扩展的目的是促进我们的偏好（“自我控制偏好”）应用于各种设置，包括标准的竞争性交换经济。我们将探讨自我控制偏好对收入分配的影响，对随机代表性代理模型中的风险溢价的影响，以及对债务合同可持续性的影响。在历史依赖偏好的模型中，我们打算分析成瘾，这是一种与缺乏自我控制密切相关的现象。为了便于比较统计结果，我们提供了比较自我控制和比较承诺偏好的测量方法并对其进行了描述。

The choices that individuals make often depend on when they make the decision. Consider a tradeoff between getting a smaller reward in period t and a larger reward in period t+1. In experiments, subjects routinely behave in a dynamically inconsistent fashion when facing such trade-offs: if the decision is made in period t-1 or earlier, the subject opts for the larger, period t+1 reward; by contrast, if the decision is made in period t, the individual opts for the smaller reward that offers immediate consumption instead of choosing to wait until t+1 for the larger reward. Similarly, individuals succumb to tempting but unhealthy consumption more often when the tempting good can be consumed immediately. In this project, which will be undertaken jointly with Faruk Gul of Princeton University, we develop a new choice-theoretic framework to capture this phenomenon. The starting point is a preference relation over sets of alternatives. A set represents a menu of choices for the agent. To capture the idea that agents may be tempted by ex ante undesirable alternatives, we allow preferences to exhibit a desire for commitment, i.e., we allow a strict preference for a smaller set of alternatives. In that framework we identify temptation and self-control as central determinants of behavior. In this work we propose to extend the two-period model that we have already developed to an infinite horizon. The purpose of this extension is to facilitate application of our preferences ("self-control preferences") to a variety of settings, including standard competitive exchange economies. We will explore the implications of self-control preferences for the distribution of income, for the risk premia in stochastic representative agent models, and for the sustainability of debt contracts. In a model of history-dependent preferences, we intend to analyze addiction, a phenomenon closely linked to the lack of self-control. To facilitate comparative statics results, we provide and characterize measures of comparative self-control and comparative preference for commitment.