Models of Time Inconsistency and Self Control

虽然经济学家假设跨期偏好是时间一致的，但有证据表明，一个人在较早时刻对幸福的相对偏好随着较早时刻的临近而增加。也就是说，许多人都有自我控制的问题，我们每时每刻都在追求即时的满足感，而不是我们自己所希望的。这一主张1试图开发正式的经济模型，这种心理现实的时间不一致的偏好，从早期的模型的建设和扩展。在最简单的模型中，一个人必须在一定数量的时间段内从事一项活动，并在每一时刻选择是现在做还是以后做。这个模型的结果取决于两个区别。第一，选择是否涉及显著成本--即行动的成本是即时的，但任何回报都是延迟的--或者选择是否涉及显著回报--即行动的回报是即时的，但任何成本都是延迟的？第二，人们是成熟的--他们预见到未来的自我控制问题--还是天真的--他们没有预见到这些自我控制问题？得到了一些简单的结果。天真的人会拖延成本显著的活动，而过早地做有显著回报的活动。 如果成本是显著的，那么复杂性会减轻拖延，甚至会让人们比没有自控问题的人更快地完成任务。如果奖励是显著的，老练会加剧准备。这些行为结果具有相应的福利含义：在显著成本下，轻微的自我控制问题只有在一个人天真时才能严重损害他，而在显著奖励下，轻微的自我控制问题只有在一个人老练时才能严重损害他。 研究者和他的合著者将单一活动模型的见解应用于不同的经济应用。在一个项目中，他们分析了天真的抑制者的最佳激励。调查结果表明了两个主要教训。首先，旨在诱导有效行为的激励措施通常会比代理人不是拖延者的情况下“更陡峭”--代理人在完成任务时的拖延会受到比拖延的真实成本更严厉的惩罚。其次，当代理人拖延的确切倾向存在不确定性时，即使在真正的拖延成本是“固定的”的环境中，“截止日期”激励计划也可以用来最大化效率：代理人首先会受到轻微的惩罚或根本不会因为拖延而受到惩罚，但在某个日期之后，会因为拖延而受到严厉的惩罚。这样的激励鼓励那些几乎没有拖延倾向的人等待，直到完成任务是有效的（即，如果代价太大，不要马上去做），同时确保那些有严重拖延倾向的人不要拖延太久。 研究者的主要项目是研究自我控制问题在消费成瘾商品中的作用。经济学家最近的一些理论和实证研究在理解成瘾性商品的关键方面具有创新性：在消费商品（例如，吸烟）可能对一个人的健康有害，但过去对这些商品的消费增加了一个人消费商品的边际效用，因此这个人可能会随着时间的推移保持或增加对商品的消费，同时对自己造成越来越大的伤害。然而，这些模型使用了理性选择框架，假设人们既有时间一致的偏好，又有对未来行为的正确预期。这个项目关注的是自我控制问题的含义，以及对理性选择模型中成瘾商品具有相同潜在偏好的人的复杂性的各种假设。 结果涉及的作用，复杂性发挥的倾向，得到上瘾的商品。如果成瘾是不可逆转的--每当一个人“吸”一口令人上瘾的物品时，她知道这会永远增加她对该物品的渴望--一个对自我控制问题有经验的人将比一个天真的人更有可能上瘾。让你现在更愿意抵制诱惑，因为你（天真地）相信，现在放纵的后果将是，你将带着一个你无法满足的渴望度过余生。如果你是老练的，因此意识到你很可能会上瘾，你将更有可能屈服于一个渴望-因为你意识到唯一的代价是你会更快地上瘾。在（更现实的）情况下，避免成瘾活动将有助于一个人摆脱上瘾，这一逻辑非常不同。研究人员证实，那些对自我控制问题天真的人会消费大量的成瘾产品，并在天真的相信他们明天会戒烟的情况下上瘾并保持下去。 研究人员还计划开展许多其他项目，将更大的心理现实主义纳入经济学。

Abstract Though economists assume that intertemporal preferences are time-consistent, evidence suggests that a person's relative preference for well-being at an earlier moment over a later moment increases as the earlier moment gets closer. That is, many people have a self-control problem, and moment-by-moment we pursue immediate gratification more than we ourselves wish we would. This proposa1 attempts to develop formal economic models of such psychologically realistic time-inconsistent preferences, building from and expanding on earlier models. In the simplest model, a person must engage in an activity exactly once during some number of periods, choosing at each moment whether to do it now or do it later. Results in this model depend on two distinctions. First, do choices involve salient costs - where the costs of an action are immediate but any rewards are delayed - or do they involve salient rewards - where the rewards of an action are immediate but any costs are delayed? Second, are people sophisticated - they foresee future self-control problems - or are they naive - they do not anticipate these self-control problems? Some simple results are derived. Naive people procrastinate activities with salient costs and preproperate - do too soon - activities with salient rewards. If costs are salient, sophistication mitigates procrastination, and can even lead people to do the activity sooner than if they had no self-control problem. If rewards are salient, sophistication exacerbates preproperation. These behavioral results have corresponding welfare implications: With salient costs, mild self-control problems can severely damage a person only if she is naive, while with salient rewards mild self-control problems can severely damage a person only if she is sophisticated. The investigator and his co-author apply the insights of the one-activity model to different economic appli cations. In one project, they analyze optimal incentives for naive procrastinators. The findings indicate two major lessons. First, incentives designed to induce efficient behavior will in general be `steeper` than if agents were not procrastinators - agents will be punished for delay in completing a task more harshly than corresponds to the true cost of delay. Second, when there is uncertainty about the exact propensity of agents to procrastinate, then even in environments where the true cost of delay is `stationary`, `deadline` incentive schemes may be used to max imize efficiency: The agent will at first be punished mildly or not at all for delay, but then after some date be punished severely for delay. Such incentives encourage those with little propensity to procrastinate to wait until it is efficient to do the task (i.e., not do the task right away if it turns out to be too costly), while assuring that those with a severe propensity to procrastinate do not delay too long. The investigator's major project is the study of the role of self-control problems in the consumption of addictive goods. Some recent theoretical and empirical research by economists has been innovative in understanding the crux of addictive goods: While consuming goods (e.g., smoking cigarettes) may be harmful to a person's wellbeing, higher past consumption of those goods increases a person's marginal utility of consuming the good, so that the person may maintain or increase her consumption of the good over time while doing more and more harm to herself. These models, however, use a rational-choice framework, where people are assumed to have both time-consistent preferences and correct expectations about their future behavior. This project concerns the implications of a self-control problem and varied assumptions about sophistication for people with the same underlying preferences for addictive goods as in these rational-choice models. Results pertain to the role that sophistication plays in the propensity to get addicted to goods. If an addiction is irreversible - every time a person takes a `hit` of an addictive good, she knows it will forever increase her craving for that good - a person who is sophisticated about her self-control problem will be more likely than a person who is naive to become addicted. Na‗vet─ makes you more willing to resist temptation now, because you (naively) believe that the consequences of indulging now will be that you will 1ive the rest of your life with a craving you won't fulfill. If you are sophisticated, and hence realize it is likely you will get addicted, you will be more likely to cave in to a craving - because you realize the only cost is that you will get addicted a bit sooner. The logic is very different for the (more realistic) case where refraining from an addictive activity will help a person become unhooked. The investigator conjectures that people who are naive about their self-control problem will consume a lot of an addictive product - and become and stay addicted - under the naive belief that they will quit tomorrow. The investigator also plans to pursue many other projects that incorporate greater psychological realism into eco nomics.