Dynamic Models of Search and Learning

搜索和学习的动态模型

• 批准号: 9422988
• 负责人: Lones Smith
• 金额: $ 6.71万
• 依托单位: Massachusetts Institute of Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1995
• 资助国家: 美国
• 起止时间: 1995-07-01 至 1997-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9422988&HistoricalAwards=false
• 关键词: Dynamic; Models; Search; Learning

SBR-9422988 Lones Smith In most markets buyers and sellers do not meet at some central site, but must constantly seek each other out. Economists usually ignore the search process when analyzing market outcomes and simply assume that in equilibrium the market-clearing price is common knowledge. The process by which equilibrium is attained and the equilibrium price becomes common knowledge is treated as a "black box." The first part of this research project opens up this black box to more careful scrutiny. The P.I. investigates the whole class of employment matching models and provides a coherent theory underlying the efficient and equilibrium outcomes when individuals are inherently different. Thirty years ago, Professor George Stigler suggested that oligopolies were unstable and proposed a process of "secret" price cuts that would seem to cause any oligopoly agreement to unravel toward the competitive outcome. Yet despite many efforts the implicit game remains unsolved and so Stigler's conjectured equilibrium remains a conjecture. All the recent progress in this area has been on the quantity-setting version of Stigler's game in which prices were observable and output unobservable. Such games have a recursive structure to which dynamic programming can be successfully applied. The second part of this research project is an attempt to solve the original much harder Stigler pricing game. This is done indirectly by examining an unsolved problem from the search literature known as the Rendezvous problem. This problem has much of the same formal structure as Stigler's oligopoly game, namely it is a repeated game with imperfect monitoring and privately observed signals, yet appears to be more tractable.

Sbr-9422988 Lones Smith在大多数市场中，买家和卖家不会在某个中心地点会面，而是必须不断地寻找对方。经济学家在分析市场结果时通常忽略搜索过程，简单地假设在均衡情况下，市场清算价格是众所周知的。达到均衡和均衡价格成为常识的过程被视为“黑匣子”。这个研究项目的第一部分打开了这个黑匣子，让人们进行更仔细的审查。P.I.研究了整个类别的就业匹配模型，并提供了一个连贯的理论，以支持当个体天生不同时的有效和均衡结果。30年前，乔治·斯蒂格勒教授提出寡头垄断是不稳定的，并提出了一个“秘密”降价的过程，这似乎会导致任何寡头垄断协议朝着竞争性结果瓦解。然而，尽管付出了许多努力，隐含博弈仍然没有得到解决，因此斯蒂格勒猜想的均衡仍然是一个猜想。这一领域最近的所有进展都是在斯蒂格勒游戏的数量设定版本上取得的，在这个版本中，价格是可以观察到的，而产出是不可观察的。这样的游戏具有递归结构，动态编程可以成功地应用于该递归结构。本研究项目的第二部分是试图解决原来难度大得多的斯蒂格勒定价博弈。这是通过检查搜索文献中一个未解决的问题来间接完成的，该文献被称为相遇问题。这个问题的形式结构与斯蒂格勒的寡头垄断游戏很相似，即它是一个具有不完善监控和私下观察信号的重复游戏，但似乎更容易处理。