Empirical Applications of Contract Theory: the Case of Insurance Contracts

近二十年来，契约理论得到了迅猛的发展。但是，至少直到最近，经验性的应用一直滞后。该项目旨在填补这一空白。它基于这样一种观点，即保险为合同方面的实证工作提供了一个近乎理想的领域。个人保险合同(汽车、住房、健康、人寿等)基本上都是标准化的。保险公司的信息是可以获取的，通常可以通过合理的少量定量或定性指标进行总结。这些公司的档案中非常准确地记录了“业绩”--无论它代表的是事故的发生、成本，还是某种程度的支出。最后，保险公司经常使用包含数百万份合同的数据库，这使得能够使用标准的计量经济学工具，以详细的方式测试合同理论的大多数预测。将调查三个方向。该理论强调了信息不对称在设计最优合同中的作用，更具体地说，强调了逆向选择和道德风险之间的区别。当合同关系的一方--比如保险合同的订户--比另一方--比如保险公司--在与关系相关的一些参数方面掌握了更好的信息时，就会出现逆向选择。当事故概率不是外生的，但取决于订户做出但不能被保险公司监测的某些决定(例如，预防的努力)时，道德风险就会发生。一般来说，不同的保险合同提供不同的激励措施，因此导致不同的观测事故率。道德风险和逆向选择之间的经验区别至关重要，尤其是因为它们在福利和监管方面的含义是完全不同的。然而，在许多情况下，道德风险和逆向选择很难分清。本项目的第一个目的是展示如何在有与关系动态相关的数据的情况下，对道德风险和逆向选择进行经验区分。第二个目标是研究保险业的保险定价结构。在现有数据的情况下，我们应该能够根据每个被保险人的特征(年龄、车辆类型、地点等)来估计其总体风险分布。这不仅涉及到发生事故的可能性，还涉及到事故的预期严重性和相应的成本。换句话说，这种回归应该能非常准确地描述保险公司销售的“产品”。在第二阶段，它可能与定价政策有关。人们可以检查向特定消费者收取的保费是否只取决于她的风险，或者对于任何给定的风险水平，保费是否可能随着其他特征而变化，如年龄、性别等。可以根据该领域的行业组织得出结论。第三个研究方向采用了更规范的观点。现代生物学的一个显着特征是，识别导致或倾向于造成各种疾病易感性的基因的能力越来越强。这种可能性将导致预防和治疗的显著改善。然而，更准确的风险信息的可获得性破坏了保险的可能性，这是福利减少。第一项任务是获得对相关福利损失的第一次评估。这尤其需要对保险范围提供的利益进行评估。这项研究的一个目的(也是一个相当困难的目的)是提供对这一量级的初步评估。提出的最激进的解决方案涉及一项法规，严格禁止保险公司使用基因检测。然而，这样的提议需要进行彻底的调查。从经济学家的角度来看，这相当于引入了一个强大的逆向选择成分。代理人大概会被告知他们的风险，至少在个人可以自由获得基因检测的情况下(很可能是这样)。现在的问题是评估这种不对称性对健康或人寿保险市场的影响。这是一个至关重要的问题，即使只是因为解决方案很可能揭示出比最初的问题更糟糕的情况。如果最终结果是面临风险的全球保险市场崩溃，每个人(包括处于风险中的人群)最终都将陷入更糟糕的境地。该项目的最后目标只是提供一些初步要素，用于评估范围和

In the last twenty years, contract theory has developed at a rapid pace. But, until recently at least, empirical applications have lagged behind. This project is aimed at filling this gap. It is based on the view that insurance provides a nearly ideal field for empirical work on contracts. Individual insurance contracts (automobile, housing, health, life, etc.) are largely standardized. The insurer's information is accessible, and can generally be summarized through a reasonably small number of quantitative or qualitative indicators. The 'performance' - whether it represents the occurrence of an accident, its cost, or some level of expenditure - is very precisely recorded in the firms' files. Finally, insurance companies frequently use data bases containing several millions of contracts, which enables testing of most predictions of contract theory in a detailed way, using standard econometric tools. Three directions will be investigated. The theory has emphasized the role of information asymmetries in the design of optimal contracts, and more specifically the distinction between adverse selection and moral hazard. Adverse selection arises when one party to the contractual relationship - say, the subscriber to an insurance contract - has better information than the other party - say, the insurer - about some parameters that are relevant for the relationship. Moral hazard occurs when the accident probability is not exogenous, but depends on some decision (e.g., effort of prevention) that is made by the subscriber but cannot be monitored by the insurer. In general, different insurance contracts provide different incentives, hence result in different observed accident rates. The empirical distinction between moral hazard and adverse selection is crucial, in particular because the implications in terms of welfare and regulation are totally different. However, in many cases, moral hazard and adverse selection are very difficult to disentangle. The first aim of the present project is to show how the empirical distinction between moral hazard and adverse selection can be implemented when data relative to the dynamics of the relationship are available. A second goal is to study the structure of insurance pricing in the industry. Given the data available, we shall be able to estimate the total risk distribution of each insuree, as a function of her characteristics (age, type of car, location, etc.). This involves not only the probability of an accident, but also its expected severity and the corresponding costs. This regression, in other words, should provide a very accurate description of the 'product' sold by the insurance company. In a second stage, it can be related to the pricing policy. One may check whether the premium charged to a particular consumer only depends on her risk, or whether it may for any given level of risk vary with other characteristics, such as age, sex, etc. Conclusions can be drawn on the industrial organization of the field. The third research direction adopts a more normative viewpoint. A striking feature of modern biology is the increasing ability to identify the genes that either are responsible for or tend to create predispositions to various diseases. This possibility will lead to a spectacular amelioration of prevention and treatments. However, the availability of more precise information on the risk destroys insurance possibilities, which is welfare decreasing. A first task is to obtain a first evaluation of the associated welfare loss. This requires, in particular, an evaluation of the benefit provided by insurance coverage. One purpose of the study (and a quite difficult one) is to provide an preliminary evaluation of this order of magnitude. The most radical solution proposed involves a regulation that would strictly prohibit the use of genetic testing by insurance companies. Such a proposal however requires a thorough investigation. From an economist's point of view, it amounts to introducing a strong adverse selection component. Agents will presumably be informed of their risk, at least if (as it will probably be the case) individuals have free access to genetic testing. The problem, now, is to assess the impact of this asymmetry on the market for health or life insurance. This is a crucial issue, if only because the solution might well reveal worse than the initial problem. If the final outcome is a global collapse of the insurance markets at stake, everybody (including the population at risk) will end up in a much worse situation. The last goal of the project is only to provide some preliminary elements for assessing the scope and