Study On The Statistical Properties Of Complex Adaptive System

A economic system is a typically complex adaptive system (CAS). Because of the complex interaction among adaptive agents and the co-evolution of agents and environment, economic systems are more complexity and uncertainty than natural systems. The real economic system has been out of the boundary which the traditional economic general equilibrium theory could be used to analyze. Physicists have accumulated a host of experiences in complex system field, and the methodology of physics has been more and more frequently used in investigating economic problems. And then a new interdiscipline--Econophysis is formed. Except for the statistical analysis of financial data in the early age, there are also some other methodologies to analyze complex system, including controlled human experiments, agent-based computer simulations, et al. With the combination of physical thoughts and these research methods, it would be helpful for us to reveal the microcosmic mechanism in complex economic systems.This article mainly investigated the statistical laws based on two kinds of complex adaptive systems:resource allocation system and financial market. For the resource allocation system, we focused on two topics:1) the effects of contrarian behavior on the resource allocation systems;2) we try to use some statistical laws in natural systems to describe resource allocation social systems. For the complex system in financial market, we did some empirical analysis based on American stock market and investigated the statistical differences between bankrupting stocks and non-bankrupting stocks.Chapter1is the background of the whole thesis. It mainly introduced the characteristics of complex adaptive systems, the methodology of econophysics and the two kinds of complex adaptive systems we focused on.In Chapter2, we analyzed the influence of contrarian behavior on resource allocation systems by three methods:controlled human experiments, agent-based model and theoretical analysis. In human experiments we observed the individual adaptability and the interaction between individuals and the environment by secretly changing the systems’ environment. Different with the common thoughts that the contrarian behavior will stabilize the market by shrinking the redundancy of resources, our conclusion showed that the contrarian behavior has dual roles. Under different conditions, the role of contrarian behavior can be either positive or negative. The dual roles could be identified be a phase transition point which helps us to show the state where the resources are most effectively utilized.In Chapter3, we conducted more human experiments and simulations in order to explore to describe resource allocation systems with statistical physical laws (second law of thermodynamics and theory of fluctuations) and observe the evolution of system under different individual behaviors. Although the interaction between units in natural systems and agents in social systems are obviously different, it could be found that under some conditions the statistical physical laws of natural systems may apply to social systems. In particular, we observe these properties directly from human experiment, not just theoretical analysis. This work reveals a kind of bridge between nature and society. It’s helpful to understand complex social behavior.Chapter4introduced the relevant researches for complex financial market. We analyzed the real stock data in American stock market from1989to2008and investigated the statistical differences between bankrupting stocks and non-bankrupting stocks. We found that there are some remarkable differences in the tails of distribution function of their return, which might be as a kind of bankrupt signal.At the end of this thesis, we reviewed the whole researches in Chapter5. Based on our experiences in human experiments and agent-based model, we give some suggestions and expectations for further study.