Information Demand and Dynamic Contests

该项目在经济理论上由信息需求和动态竞争两部分组成。前者相当发达，而后者还处于早期阶段。此外，本书还讨论了一个正在进行但更专注于讨价还价理论的项目，该项目诉诸于一个既定的心理学事实，即个人的行为在很大程度上受其愿望的支配。前两个子项目是一个更大议程的一部分，目的是了解有关信息需求本质的最基本问题，即所谓的需求定律是否成立？众所周知，静态环境中的反例表明并非如此。事实上，我们发现两种不同的情况下，它确实成立，一个在应用环境中非常相关。在最先进的子项目中，我们发现如果信息非常便宜，因此需求非常大，那么需求定律最终是正确的。此外，在这种情况下，我们表明布莱克威尔定理提供的基本见解在很大程度上并不重要。事实上，当一个人被允许获得任何给定信号的足够多的独立和相同分布的副本时，就会出现对实验/信号的完整排序。从统计意义上讲，这两个信号对彼此来说是不充分的，这已经无关紧要了。重要的是对充分性的渐近衡量。我们将此扩展到当价格足够低时，推导出一个非常简单的信息需求对数公式。这部分工作应该对贝叶斯决策理论和博弈论做出方法论和实质性的贡献。在贝叶斯决策理论中，信息需求问题的中心性是很明显的。但从渐近理论中引入的方法应该证明是非常容易处理的，并且是未来理论家考虑信息的基本工具。经济学家传统上狭隘地关注过于程式化的一次性信息购买模型（今天我购买我的信号，明天我采取行动），这可能源于布莱克威尔定理。从某种意义上说，大需求纸定律是从一个不同的、但同时存在的基础开始的（Chernoff的研究）。第三个项目开发较少，仍处于早期阶段。它提出了一个新的机制设计问题，即什么是动态竞赛的最优设计，作为收益和信号结构的函数？从体育到就业安排再到电视游戏节目，此类竞赛的例子不胜枚举，说明了理解这种微观市场结构的重要性。我们在连续时间中对其进行了表述，并提出了竞赛设计性质的一些基本特征。这项工作的目的是作为一个未被考虑的问题的第一次努力，这是一个令人难以置信的通用问题。希望这种工作的实质价值是清楚的。许多技术障碍尚不清楚。

This project in economic theory consists of two parts: information demand and dynamic contests. The first is reasonably well developed while the second is in its early stages. Also included is a discussion of an ongoing but much better focused project on bargaining theory, appealing to an established psychological fact that individuals' behavior is largely governed by their aspirations.The first two subprojects are part of a greater agenda to understand the most basic questions about the nature of information demand, namely, does the so-called Law of Demand hold? Well-known counterexamples in static environments suggest not. In fact, we find two separate contexts where it does hold, one extremely relevant in applied contexts. In the most advanced sub-project, we find that if information is very cheap, and thus demand very large, then the law of demand is eventually true. Additionally, in this context, we show that the fundamental insight provided by Blackwell's Theorem largely does not matter. In fact, when one is permitted to acquire a large enough number of independently and identically distributed copies of any given signal, then a complete ordering over experiments/signals arises. That two signals are not sufficient for each other in a statistical sense is no longer relevant. What matters is an asymptotic measure of sufficiency. We extend this so far as to deduce a very simple logarithmic formula for the demand for information, when prices are low enough.This part of the work should make methodological and substantive contributions to Bayesian decision theory and game theory. For Bayesian decision theory, the centrality of the information demand question is manifest. But the methods introduced from asymptotic theory should prove extremely tractable and basic tools for theorists considering information in the future. Economists' traditional narrow focus on an overly stylized of model of one-shot information purchase (today I buy my signal, tomorrow I act) likely owes its roots to Blackwell's Theorem. In some sense, the law of large demand paper starts from a different yet contemporaneous foundation (work by Chernoff).The third project is less developed, being still at an early stage. It proposes a new mechanism design question, namely what is the optimal design of a dynamic contest, as a function of the payoff' and signaling structure? The wide range of cited instances of such contests, from sports to employment arrangements to television game shows, speaks to the importance of understanding this micro market structure. We have formulated this in continuous time, and have proposed some basic features of the nature of the contest design. This work is intended as a first effort on an unconsidered problem that is incredibly generic. The substantive value of such work is hopefully clear. Many of the technical hurdles are as yet unknown.