Research On Risk Contagion Mechanism And Evolutionary Characteristic Based On Multi-Agent And Investor Behavior

The research on investor behavior in stock market,has always been a hot topic field,and has formed many research results and theoretical models.The emergence of expected utility theory offers facilities for the quantitative analysis of decision-making behavior.This theory helps the investor make investment decision more intuitively and easily by comparing the value of expected utility in different decisions.The final decision chosen by the investor has maximizing utility.With the advent of financial anomalies,such as equity premium puzzle,volatility puzzle,market bubble,closed-end fund puzzle,dividend puzzle,option smile,IPO anomaly,it is difficult to explain investor behavior behind financial anomalies for the expected utility theory.Thus,prospect theory as a replacement of the expected utility theory,is generally recognized in academic circles.From a micro perspective of investor behavior,the researches on risk contagion in stock market began appearing in recent year.In addition,many scholars turned their eyes into microscopic mechanism of risk contagion with the frequent outbreak of financial crisis in the 1990s.In this context,based on available literature,this paper reveals trading strategies and evolutional characteristics of investor behavior,and builds complex network risk contagion model and artificial stock market model based on multi-agent modelling method from the perspective of investor behavior.Firstly,based on the theory of investor behavior,this research firstly discusses the dynamic equalization of investor behavior,and analyzes the influences of financial crisis on investor behavior and dynamic equalization stability strategy.The results show that if the mean value of expected price deviation is smaller than zero,noise traders will leave stock market and look for new investment opportunities.And if the mean value of expected price deviation is greater than zero,the situation will become more complicated.Under this situation,the all traders are the noise trader or the combination including noise trader and rational trader.If the first case happen,the trading strategy is the same as that in the above mentions.And if the second case happen,the trading strategy of all investors are to keep their current strategies and pay more attention to the market situation.Secondly,this research introduces complex network theory,biological theory,epidemic spreading mechanism into the analysis of stock market,and discusses the risk contagion causing by information fluctuation in stock market.Based on the agent modeling,the contagion model is built.In the model,investors are divided into four types:susceptible agent,infected agent,contagious agent and immune agent.The simulation experiment studies dynamic characteristics of investor behavior with the changing contagion coefficient among different investors.The results show contagion coefficient affects the contagion duration,the proportion structure of investors and the curve of agents.And the curve of agents having susceptible status has a significant decreasing trend with different rates during the process of simulation.The curves of agents having infected and contagious status first increase and then decrease.The curve of agents having immune status shows an increasing trend with different rates.At last,this research builds an artificial stock market to analyze financial risk contagion among investors,based on investor behavior theory and available literature.The modeling approach in this research is different from the existing research.This model discards the research framework generally used by scholars,and no longer make the classification for investors.For example,many researches divided investors into rational investor and noise investor,technical analyst and fundamental analyst.The artificial stock market model in this research holds the reason of price fluctuation in stock market is caused by the changes of investor sentiment with the emergence of new market information.When investor sentiment is changing,he will give his neighbors this sentiment.The financial risk starts to spread outwards,from one agent to other agents.In simulation experiment,the eight different parameters environments are set according to analyze stock price,price fluctuation,number of buyers and number of sellers in different environments.To take the domestic and foreign stock price indexes,SPX and HSI as examples,the seventh experiment and the third experiment in artificial stock market have similar statistical characteristics with SPX and HSI respectively.After studying the characteristics,such as leptokurtosis and fat-tail,volatility clustering,long memory,chaotic characteristic,the results show artificial stock market can well reflect Hong Kong stock market.Furthermore,it is found that investors in Hong Kong stock market are easily influenced by the behaviors of neighboring investors,and have relatively stable preferences for new market information.But,the investor is insensitive to new market information to a certain extent.In short,this research is based on the theory of investor behavior,and discusses the dynamic characteristics of investor behavior,then builds stock market contagious model through the combination of complex network theory and multi-agent modeling method.Finally,this research architects an artificial stock market model from a new viewpoint rather research thought in available literature.In generally,based on the theories of behavioral finance,finance,evolutionary game,complex network,loemology and agent-based computational finance,this research discusses investor behavior and risk contagion by using agent modeling approach,and provides theoretical bases and simulation models for regulator in order to prevent risk contagion in financial market.Thereby,it has a great significance in theory and practice.