复杂适应系统的统计性质研究

发布时间：2018-01-28 11:02

本文关键词： 经济物理复杂适应系统资源分配反向行为可控真人实验行为人建模金融市场破产股票　出处：《复旦大学》2013年博士论文　论文类型：学位论文

【摘要】：经济系统是个典型的复杂适应系统,系统中适应性个体之间相互作用,个体和环境共同演化,导致经济系统比自然系统要复杂的多也存在很多不确定性。现实的经济系统已经超出了西方经济学一般均衡理论所能分析的范围。物理学家很早在对复杂系统的研究中积累了大量经验,物理学的思想和方法越来越多的被应用到经济学的各类问题研究上,形成了一门新的交叉学科——经济物理学。除了最早对金融数据的统计分析外,还形成了一些便于分析复杂系统的研究方法,包括：组织可控真人实验、计算机辅助的行为人理论模型等。物理学思想和这些研究方法相互结合,有助于我们揭开复杂经济系统的微观作用机理。本文主要基于资源分配问题和金融市场这两类复杂适应系统,研究了各自的一些统计性质。对资源分配复杂系统,我们具体研究了两个课题：第一,反向行为对资源分配系统的影响。第二,尝试用自然系统的某些统计规律描述资源分配社会系统。对金融市场复杂系统,我们对真实美国股票市场进行实证分析,研究了破产股票与非破产股票的统计性质差异。第一章是对全文研究背景的介绍。主要介绍了复杂适应系统的特征和经济物理学的研究方法,以及我所研究的两类复杂适应系统。第二章,采用可控真人实验、行为人建模和理论解析三者结合的手段,详细分析了反向行为对资源分配系统的影响。真人实验中我们通过秘密改变系统环境考察个体的适应性以及个体与环境的相互作用。与普遍观点认为反向行为通过减少资源浪费从而稳定市场不同,我们的结论显示出反向行为是把双刃剑。在不同的环境和条件下,会分别显现出积极作用和消极作用,这两种作用会通过一个相变点来识别。相变点对应了资源被最有效配置的状态。第三章,通过真人实验和行为人模拟,探索用描述自然系统的两种统计物理规律(涨落定理和熵增加原理)来描述资源分配复杂系统,观察在不同类别行为人的相互作用下资源分配系统的演变。最终发现尽管复杂经济系统中人的相互作用和构成自然系统单元的相互作用明显不同,但在一定条件下描述自然系统的统计规律也适用于社会系统。自然系统的法则提供了一种理解复杂适应系统的新思路。第四章,金融市场复杂系统的相关研究。受到金融危机影响,分析了过去20年间美国股票市场真实股票数据,研究了破产股票和非破产股票的统计性质差异,发现二者收益率分布函数的尾部有显著的不同。利用统计性质,把破产日之前收益率和交易量的不寻常大波动作为股票即将破产的信号。第五章是对整个研究工作的回顾,并对经济物理研究方法中涉及的真人实验设计和行为人模型构建提出了一些建议和展望。
[Abstract]:Economic system is a typical complex adaptive system in which adaptive individuals interact and individuals and environment evolve together. There are many uncertainties that cause economic systems to be much more complex than natural ones. Real economic systems are beyond the limits of the general equilibrium theory of western economics. Physicists have long been concerned with complex systems. A great deal of experience has been accumulated in the research. More and more ideas and methods of physics have been applied to the study of various problems in economics, forming a new interdisciplinary discipline, economic physics, except for the earliest statistical analysis of financial data. Some research methods have been developed to facilitate the analysis of complex systems, including organizing controllable real person experiments, computer-aided behavioral theory models, and so on. Physics ideas and these research methods are combined. It is helpful for us to uncover the micro mechanism of complex economic system. This paper mainly based on the resource allocation problem and the financial market two kinds of complex adaptive systems, studied some of their statistical properties. For the complex resource allocation system, we studied two specific topics: first. Second, we try to use some statistical laws of natural system to describe the social system of resource allocation. We analyze the real American stock market and study the statistical difference between bankruptcy stock and non-bankruptcy stock. The first chapter is an introduction to the background of this thesis, which mainly introduces the characteristics of complex adaptive systems and the research methods of economic physics, as well as the two kinds of complex adaptive systems that I have studied. The second chapter uses controllable real person experiment, actor modeling and theoretical analysis of the combination of the means. The effect of reverse behavior on resource allocation system is analyzed in detail. In real life experiments, we study the adaptability of individuals and the interaction between individuals and environment by changing the system environment secretly. Reducing the waste of resources to stabilize the market is different. Our conclusion shows that reverse behavior is a double-edged sword. Under different circumstances and conditions, it will show positive and negative effects respectively. These two interactions are identified by a phase transition point, which corresponds to the state in which resources are most efficiently allocated. In the third chapter, through the real person experiment and the behavior simulation, the author explores two kinds of statistical physical laws (fluctuation theorem and entropy increasing principle) to describe the complex system of resource allocation. The evolution of resource allocation systems under the interaction of different categories of actors was observed. It was finally found that although the interaction of human beings in complex economic systems and the interactions constituting units of natural systems were significantly different. But the statistical law of describing the natural system under certain conditions is also applicable to the social system, and the law of the natural system provides a new way to understand the complex adaptive system. Chapter 4th, related research on complex systems of financial markets. Under the influence of the financial crisis, this paper analyzes the real stock market data in the past 20 years in the United States. The statistical properties of bankrupt stock and non-bankrupt stock are studied. It is found that there is a significant difference in the tail of the return distribution function between the bankruptcy stock and the non-bankrupt stock. The unusual volatility of yields and trading volumes prior to the bankruptcy date is used as a signal that stocks are about to go bankrupt. Chapter 5th is a review of the whole research work and puts forward some suggestions and prospects for the design of real person experiment and the construction of the doer model involved in the research methods of economic physics.
【学位授予单位】：复旦大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：F831.51

【相似文献】