金融复杂性：表征与市场机理模拟研究

发布时间：2018-02-25 17:33

本文关键词： 长程相关性多主体模型尖峰胖尾损失厌恶金融物理　出处：《南京信息工程大学》2012年硕士论文　论文类型：学位论文

【摘要】：随着经济结构的演变,虚拟经济的重要性日益显现。1987年至今爆发了三次大的金融危机,这三次金融危机对全球实体经济产生了深远的影响。而经典金融理论难以解释这些金融危机以及其他一些金融市场行为的产生原因。因为金融系统是典型的复杂系统,所以本文基于复杂性理论,深入研究了金融市场波动的自依赖性,并构建了两个多主体模型,研究了金融市场典型特征和投资者损失厌恶非理性行为形成的市场机理。本文主要的研究成果及创新性如下： (1)本文定义了平均波动性的变量,并分别研究了不同观测尺度下金融市场指数波动特性。发现在较小的时间窗口下,金融时间序列波动前后之间存在显著的线性关系。而随着观测窗口的变大,线性关系越来越弱,非线性关系越来越强。此外,本文研究了同一观测尺度下道琼斯工业平均指数不同时期金融波动性的前后相关性,发现了历史上波动的相关结构会随着样本时间不同而变化。 (2)本文也对金融市场指数的GARCH模型进行了。真实数据和GARCH模型的分析结果在一定程度上是相似的,但进一步的定量研究表明两者之间存在着显著的差别。此外,研究结果发现GARCH模型低估了金融市场上的长程相关性。 (3)本文构建了一个多主体金融模型,将人与人之间的互动过程、机构投资者之间的非线性过程以及股票网络拓扑结构的变化联系起来,以研究金融市场上若干典型特征的产生机理。模型描述了羊群效应的产生过程以及股票市场拓扑结构的演变过程。模型分别在不同的初始条件和参数下进行了多次仿真,所产生的时间序列都具有波动聚集和尖峰胖尾特性,相关统计值与实际结果相符：中心标度值在0.579到0.747之间,对数收益累积分布在尾部明显存在着幂率现象,幂指数约为3。此外,波动聚集程度与标准普尔500指数相仿,在纳斯达克和香港恒生之间。仿真结果暗示市场拓扑结构的演变是对典型特征形成的重要因素。 (4)为了探索非理性行为和典型特征之间是否存在着共同的产生机制,本文建立了一个多主体模型研究了损失厌恶现象的形成机制。构建的模型成功再现了损失厌恶的形成过程。分析表明显著的边际效用递减效应在损失厌恶的形成上起着重要的作用,而个体对杰出者的模仿行为也起着一定的作用。
[Abstract]:With the evolution of economic structure, the importance of virtual economy is becoming more and more obvious. Since 1987, three major financial crises have broken out. These three financial crises have had a profound impact on the global real economy. Classical financial theories are difficult to explain the causes of these financial crises and some other financial market behaviour, because the financial system is a typical complex system. Therefore, based on the complexity theory, this paper deeply studies the self-dependence of financial market volatility, and constructs two multi-agent models. This paper studies the typical characteristics of financial market and the market mechanism of investors' irrationality behavior of loss aversion. The main research results and innovations of this paper are as follows:. In this paper, the variables of average volatility are defined, and the volatility characteristics of financial market indices at different observation scales are studied. It is found that in a small time window, There is a significant linear relationship between the financial time series before and after the fluctuation. As the observation window becomes larger, the linear relationship becomes weaker and stronger, and the nonlinear relationship becomes stronger and stronger. In this paper, we study the correlation of financial volatility in different periods of the Dow Jones Industrial average at the same observation scale, and find that the correlation structure of historical volatility varies with the sample time. The GARCH model of financial market index is also studied in this paper. The results of real data and GARCH model are similar to some extent, but further quantitative research shows that there are significant differences between them. The results show that the GARCH model underestimates the long-range correlation in financial markets. 3) this paper constructs a multi-agent financial model, which links the interaction process between people, the nonlinear process among institutional investors and the change of topological structure of stock network. In order to study the generation mechanism of some typical characteristics in the financial market, the model describes the generation process of herding effect and the evolution of the topological structure of the stock market. The model is simulated several times under different initial conditions and parameters. The time series produced have the characteristics of fluctuation aggregation and peak fat tail, and the correlation statistical values are consistent with the actual results: the central scale values are between 0.579 and 0.747, and the logarithmic income accumulation distribution in the tail obviously has the phenomenon of power rate. The power index is about 3. In addition, the degree of volatility aggregation is similar to that of the S & P 500 index, between Nasdaq and Hang Seng in Hong Kong. The simulation results suggest that the evolution of market topology is an important factor in the formation of typical characteristics. (4) in order to explore whether there is a common mechanism between irrational behavior and typical characteristics, In this paper, a multi-agent model is established to study the formation mechanism of loss aversion. The established model successfully reproduces the formation process of loss aversion. The analysis shows that the significant marginal utility decline effect is in the formation of loss aversion. Has an important role to play, Individuals also play a certain role in imitating outstanding people.
【学位授予单位】：南京信息工程大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F224;F830.9

【参考文献】