股市模仿性羊群行为涌现、演化和治理的系统建模：基于社会网络视角

With the globalization of capital markets, the complex features, such as, the uncertainty, unpredictability and so on, of stock market in international wide are becoming more obvious, and the herding effect is becoming more significant, that result in the strengthening of stock market volatility and the increasing the possibility of outbreak and spread of financial risks. The feature of emerging and transition and the phenomenon of herding effect of China's stock market are still very clear and common, and the function of stock market in economy cannot be effectively realized. Therefore, the study of evolution mechanism of herding behavior plays an important role in deepening understands the operation law of stock market. Based on the assumptions of investors' bounded rationality who possess the both property of economical being and social being, and on the perspective of complex social network, using the theory and methodology of complex network theory, behavior finance theory, game theory, multi-agent modeling and so on, this project builds the model to investigate the network structure of the correlation among the investors, analyze the imitate learning mechanism and the coupling basis among investors' behavior, explore the paths of the evolution of herding behavior and the spreading channels, resolve the evolution mechanism of imitative herding behavior in stock market on the network perspective, and construct the evolution model of imitative herding behavior from different imitative strategy with simulation analysis. The purpose of this project is to design the countermeasures and suggestion for China's stock market, and to provide theoretical support for further improve our financial regulatory system.

随着资本市场的全球化，国际范围内股市的不确定性、不可预测性等复杂特征越发明显，股市羊群效应显著，导致股市波动加强，增加了金融风险爆发和蔓延的可能。我国股市"新兴加转轨"特征仍十分明确，股市羊群现象仍较普遍，股市功能无法得到有效发挥。因此，研究股市羊群行为演化机理对深入认识股市的运行规律具有重要作用。基于股市投资者兼有"经济人"和"社会人"属性的有限理性假设，本项目以复杂社会网络为研究视角，运用复杂网络理论、行为金融理论、博弈论、多主体建模理论等理论方法体系，构建模型探讨股市投资者间的关联网络结构，分析投资者间互动耦合的行为基础和学习模仿机制，探索股市模仿性羊群行为的演变路径和传播渠道，解析网络视角下的股市模仿性羊群行为演化机理，构建不同模仿策略下的股市模仿性羊群行为演化模型，并进行模拟仿真分析。本项目的研究旨在构建针对我国股市特征的治理对策和建议，为进一步完善我国金融监管体系提供理论支撑。

本项目以复杂社会网络理论和方法为基础，采用“自下而上”的分析思路，综合运用统计物理学、行为金融理论、博弈论、平均场理论、多主体建模等理论和方法体系，由微观到宏观，由个体到整体，系统研究了股票市场羊群行为的演化机理、模型构建及其治理对策。首先，从复杂网络视角，全面梳理和分析网络视角下的股市羊群行为演化过程及特征，构建了项目全面开展的理论框架体系；其次，从股票市场投资主体的属性特征出发，基于上述分析主体间的耦合关联机制，采用基于属性的网络建模方法，构建了股市投资网络模型，并对模型的拓扑结构和稳定性进行模拟仿真分析；第三，以构建的网络模型为基础，在解析股市投资主体间投资策略学习、反馈及聚合的动态演化过程的基础上，构建了网络近邻择优策略下的股市羊群行为演化及协同演化模型；构建了协调博弈策略下的股市羊群行为演化及协同演化模型；构建了反协调博弈策略下的股市羊群行为演化及协同演化模型。最后，针对股票市场中主体的多重性，分析了以银行及企业为主体的跨市场投资关联网络模型，并分析了因交易关联等金融关联下的信用风险传导机制，并构建了风险传导模型。研究结果和成果，不仅在学术理论上为研究经济金融系统的复杂性奠定一定的基础，而且对金融监管具有一定的实践指导价值。