The Insurance Effect of Groups
研究当代理人风险厌恶且群体特征可见而个体特征不可见时，帕累托最优的群体构建方式与标准观念不同，并以工人岗位分配和意外保险为例说明。
Many people belong to groups which partially identify their characteristics. This paper argues that when agents are risk averse, and where only characteristics are visible rather than individual characteristics, the Pareto optimal construction of groups differs in systematic ways from standard notions of how groups should be formed. Two applications are considered: the assignment of workers to jobs, where it is shown the optimal contract does not involve maximizing output and accident insurance, where the optimal contract does not involve complete insurance. Many people belong to groups which serve to partially identify otherwise hidden talents or characteristics. For example, it is possible to find out that certain authors have written an article but their individual contributions are much more difficult to ascertain. Similarly, it may be easy to determine that a worker is a foreman, but much more difficult to tell how much better he is than other foremen. Or that somebody is employed, which tells us something about his talents, but not whether he is more talented than the average employed person. Each of these examples illustrates cases where we can see the group to which an agent belongs (author, foreman, employed) rather than an individual's characteristics. Standard notions of efficiency tell us how these groups should be formed. Academics should write papers together only if their joint product is greater than if they operated separately. A worker should be employer as a foreman only if his productivity is higher as a foreman than elsewhere in the organization, and so on. The principal result of this paper is that when agents are risk averse, and where only groups of the type above are visible to some agents rather than individual characteristics, these assignment rules may not be Pareto optimal. Instead, for example, someone who is less competent as a foreman than as an operative may still be employed as a foreman and this yields higher welfare than employing that worker as an operative. Let me explain the intuition by a simple example. Assume that workers who join a firm are uncertain of their ability. They are risk averse and seek insurance over that uncertainty. Untalented workers should be fired on standard efficiency grounds as they have higher productivity elsewhere. Assume that after a period of production, information on ability becomes available to the employer and the employee but that other employers must infer this information from whether the