The Value of Information in Strategic Environments

项目摘要信息是当今经济生产的关键投入。 有人认为，世界经济正在经历一场信息革命，其影响将与200年前的工业革命一样大。 在信息时代，经济学家的一个基本研究领域是理解信息为什么有价值以及如何有价值。 有一个成熟的理论（可以追溯到20世纪50年代）的信息价值的单人决策问题下的不确定性。它表明，信息是有价值的，在每一个决策问题，正是如果它是在一个自然的统计意义上的信息。在单人决策问题中，信息越多越好。 经济是战略性的。 我们不能孤立地分析每个决策者。 相反，我们需要了解经济中的每个参与者将如何行事，并考虑到他或她认为其他人将如何行事。 信息在战略环境中的价值是一个复杂得多的问题。 更多的信息有时会让每个人都变得更糟。 例如，保险市场为保险公司创造利润，并通过降低风险使消费者受益。 如果消费者能够获得更多的信息，保险市场就会关闭，每个人的情况都会更糟。 这个项目的目的是发展一个战略环境中的信息价值理论。 将讨论四组问题。 前三个更技术性，但提供了思考这些问题的框架。 首先，在所有战略环境中，一个信息系统比另一个信息系统更有价值的充分必要条件是什么？条件将是非常强大的，但提供了一个重要的基准。第二，在所有战略环境中，两个信息系统几乎具有同等价值的充分必要条件是什么？第三，在哪类战略情况下，信息量更大的信息系统会让每个代理人更好？在哪类战略情况下，信息量更大的信息系统会使每个代理人的处境更糟？ 第四个部分更实用。 战略问题的一个重要类型是协调博弈：一组公司希望协调他们的行动，但他们可能不同意他们应该协调的行动。 一个例子是软件公司选择标准化协议。 这个项目发展了协调博弈的信息价值理论。 现有的工作表明，在实现协调的共同知识的重要性。 重要的不仅是每个公司知道什么，而且他们知道其他公司知道什么，等等，因此，信息系统产生近似共同知识的能力将是信息价值的关键决定因素。 这是一个重要的观察，因为高度分布式的信息系统，如互联网，在产生近似的共同知识方面特别差。

Project Abstract Information is a key input into production in today's economy. Some believe that the world's economy is undergoing an information revolution that will have as large an impact as the industrial revolution of two hundred years ago. A basic research area for economists in the information age is to understand why and how information is valuable. There is a well developed theory (dating back to the 1950's) of the value of information in single person decision problems under uncertainty. It was shown that information is valuable in every decision problem exactly if it is more informative in a natural statistical sense. In single person decision problems, more information is always better. But the economy is strategic. We cannot analyze each decision maker in isolation. Rather, we need to understand how each participant in the economy will behave, taking into account how he or she thinks others will behave. The value of information in strategic settings is a much more complicated problem. More information can sometimes make everyone worse off. For example, insurance markets make profits for insurance companies and make consumers better off by reducing risk. If the consumers had access to more information, the insurance markets would shut down and everyone would be worse off. The purpose of this project is to develop a theory of the value of information in strategic settings. Four sets of questions will be addressed. The first three are more technical but provide the framework to think about these issues. First, what is a necessary and sufficient condition for one information system to be more valuable than another in all strategic settings? The condition will be extremely strong, but provides an important benchmark. Second, what is a necessary and sufficient condition for two information systems to be almost as valuable as each other in all strategic settings. Third, for exactly which classes of strategic situations do more informative information systems make every agent better off? And for which classes of strategic situations do more informative information systems make every agent worse off? The fourth component is more applied. An important type of strategic problem is a coordination game: a group of firms would like to coordinate their actions, but they may disagree on which action they should coordinate. An example would be software companies choosing standardized protocols. This project develops a theory of the value of information for coordination games. Existing work shows the importance of common knowledge in achieving co-ordination. It is important not only what each firm knows, but also what they know about what other firms know, and so on. Thus the ability of an information system to generate approximate common knowledge will be a key determinant of the value of information. This is an important observation because highly distributed information systems, such as the internet, are especially poor at generating approximate common knowledge.