Endogenous Rates of Experimentation; Dynamics in Non-Partition Information Models; and Competition in Herd/Cascade Models

内生实验率；

• 批准号: 9409302
• 负责人: Eddie Dekel
• 金额: $ 19.1万
• 依托单位: Northwestern University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1994
• 资助国家: 美国
• 起止时间: 1994-10-15 至 1998-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9409302&HistoricalAwards=false
• 关键词: Endogenous; Rates; Experimentation; Dynamics; Non

9409302 Dekel This project focuses on three research topics: (1) Experiments and or mutations are an important ingredient in the models used in the recent research on learning and evolution in economics. The proposal develops models where rates of experimentation are themselves learned or mimicked and then explores how different aspects of the environment influence the rate of experimentation that arises in the long run. (2) Models where the possible information received by an individual does not partition the set of states of the world have been used to introduce various elements of bounded rationality into decision making and game theory. However, these models implicitly assume that players are unaware of their information sets, while the applications in games often implicitly assume that the information structure is common knowledge. The research explores this apparent conflict by explicitly developing a dynamic process of how players use their information. A related objective is to develop an explicit model of forgetfulness to see how it corresponds to the general models without partitions. (3) Models of herds, or informational cascades, where buyers mimic preceding buyers rather than using their own private information ignore the fact that sellers have strong incentives to entice buyers to create a herd/cascade for their own product. The research will explore the extent to which this can happen, what types of sellers will invest in falsely creating herds/cascades, and whether more or less information is used when sellers are activity trying to create herds/cascades.

9409302 Dekel 该项目重点关注三个研究主题：（1）实验和/或突变是最近经济学中学习和进化研究中使用的模型的重要组成部分。 该提案开发了模型，其中实验率本身是学习或模仿的，然后探讨环境的不同方面如何影响长期出现的实验率。 (2) 个人接收到的可能信息不会划分世界状态集的模型已被用来将有限理性的各种元素引入决策和博弈论中。 然而，这些模型隐含地假设玩家不知道他们的信息集，而游戏中的应用程序通常隐含地假设信息结构是常识。 该研究通过明确开发玩家如何使用其信息的动态过程来探索这种明显的冲突。 一个相关的目标是开发一个明确的遗忘模型，看看它如何与没有分区的一般模型相对应。 (3) 群体或信息级联模型，即买家模仿先前的买家而不是使用自己的私人信息，忽略了这样一个事实：卖方有强烈的动机来吸引买家为自己的产品创建群体/级联。 该研究将探讨这种情况发生的程度、哪些类型的卖家会投资于错误地创建牛群/级联，以及当卖家尝试创建牛群/级联时是否会使用更多或更少的信息。