A Statistical Mechanics Approach to Coase's Theory of the Firm

该项目使用最初在物理学中开发的建模技术来理解商业公司的增长。物理学家分析过这样的系统：一个“单位”（也许是一个原子）的行为受到大量其他单位行为的影响。每个个体相互作用的影响可能很小，但总体效应却不是。通过类比，每个商业公司的成长受到大量其他公司的影响：竞争对手、客户、供应商和相关产品的生产商。1939年，科斯发表了一篇影响深远的文章，他认为企业的存在是为了使交易成本最小化。这篇文章提供了经济学家至今仍在试图理解企业规模和增长的框架。尽管这一框架的逻辑被广泛接受，但它并没有产生关于单个公司规模和增长的可检验的预测，这些预测已经得到了经验验证。换句话说，它没有通过一个成功的科学理论的基本检验。具有大量相互作用单元的物理系统模型的结果是，虽然任何一个单元的行为本质上是不可预测的，但整个系统的统计特性是可预测的。本项目开发的模型与科斯提出的框架一致，这些模型解释了关于企业规模和增长分布的充分记录的规律，并解释了为什么单个企业的增长、规模和组成本质上是不可预测的。虽然这个项目是基础研究，但其结果最终可能会影响反垄断和产业政策的其他方面。

The project uses modeling techniques originally developed in physics to understand the growth of business firms. Physicists have analyzed systems in which the behavior of one "unit" (an atom, perhaps) is affected by the behavior of a large number of other units. The effect of each individual interaction may be small, but the aggregate effect is not. By analogy, the growth of each business firm is affected by a large number of other firms: competitors, customers, suppliers, and the producers of related products. In 1939, Coase published a seminal article in which he argued that firms exist to minimize transaction costs. This article provided the framework that economists still use to try to understand the size and growth of firms. Even though the logic of this framework is widely accepted, it has not yielded testable predictions about the size and growth of individual firms that have been verified empirically. In other words, it has not passed the fundamental test of a successful scientific theory. The results in models of physical systems with large numbers of interacting units are that while the behavior of any one unit is inherently unpredictable, the statistical properties of the entire system are predictable. This project develops models that are consistent with the framework suggested by Coase, that explain well-documented regularities about the distribution of firm size and growth, and that explain why the growth, size, and composition of individual firms are inherently unpredictable. While this project is basic research, the results could ultimately affect antitrust and other aspects of industrial policy.