Nonparametric Inferences from Demand Observations

本提案的目的是开发方法，利用对消费者需求的观察来推断消费者偏好的属性。与以前的研究相比，这些新方法对消费者偏好结构及其效用表示的限制更少。这项研究很重要，因为这些新方法的使用将使研究人员能够避免超出经济理论直接隐含的假设，从而避免模型可能导致的错误说明。该项目将考虑消费者从有限的备选方案中做出的选择。例如，通过观察消费者对各种交通工具的选择，就有可能研究和推断出他们对可选择的交通工具的各种属性（如时间和成本）的效用。与以往的研究相比，效用函数及其在消费者群体中的分布只需要满足弱条件。

The purpose of this proposal is to develop methods that use observations about consumer demands to infer properties of consumer preferences. These new methods will require fewer restrictions on the structure of consumer preferences and their utility representations than previous studies have imposed. This research is important because the use of these new approaches will allow researchers to avoid assumptions beyond those directly implied by economic theory and, consequently, to avoid errors that result when models entail possible misspecifications. The project will consider consumers' choices from finite sets of alternatives. For example, from observations on consumer's choices among various means of transportation, it will be possible to study and infer the utility they attach to various attributes, such as time and cost, of alternative means of transportation. In contrast to previous studies, the utility functions and their distribution in the consumer population need to satisfy only weak conditions.