基于时频分析的证券市场间风险溢出研究

Securities market is an important part of finance system and even the national economy, thus keeping its stable development has great significance in promoting a quicker and better development of the later. With the increasing trend of economy and finance integration, the global securities market gradually has the characteristics of complex networks. In addition, continuously strengthening risk transfer among those markets forms risk spillover, and even induces upheaval to securities markets. This research proposal analyzes securities market risk spillover mechanism by combining complex network theory with behavioral finance theory. Based on identifying the risk signal in securities market, this research proposal aims at introducing the signal analysis technology in system control field into the study of risk spillover among securities markets. In view of time-varying and non-stationary characteristics of financial time series, this research proposal constructs and compares the linear time-frequency analysis model and nonlinear time-frequency analysis model respectively. Meanwhile, it intends to make empirical researches on risk spillover between Chinese securities market and global main securities markets with time-frequency analysis models, then to describe and analyze the risk signal transferring results of time, frequency, path and strength of risk spillover among securities markets. Therefore，the research results can make a scientific, systematical and comprehensive inquiry on the characteristics of risk spillover during the safe and crisis period, as well as provide countermeasures on securities market risk prevention. From the new perspective of combination of time domain and frequency domain, this research proposal provides new methods to support the securities market risk measurement and supervision,in order to guard against financial risks and maintiain the stability of the securities market.

随着金融经济一体化趋势的演进，全球证券市场呈现出复杂网络的特征，风险在各证券市场之间的传递不断加强，形成风险溢出，严重时引发整个网络的动荡。本项目将复杂网络理论与行为金融学理论相结合，剖析证券市场间风险溢出的机理；在识别证券市场风险信号的基础上，将系统控制领域的信号分析技术引入证券市场间风险溢出研究，结合金融时间序列的时变和非平稳特征，分别构建度量证券市场间风险溢出的线性与非线性时频分析模型；实证研究中国证券市场与全球主要证券市场之间的风险溢出，从而准确描述包括时间、频率、路径和强度在内的证券市场风险信号传递效果，科学、系统、全面地刻画与比较证券市场在安全和危机时期的风险溢出规律，为证券市场的风险防范和调控提供对策建议。本项目拟从时域和频域相结合的新视角分析证券市场间风险溢出，为证券市场风险度量与监管提供新的方法支持，以期有利于提升证券市场防范金融风险的水平。

随着金融全球化、自由化以及金融创新步伐的加快，全球证券市场间的相互依存性增强，风险在各证券市场之间不断动态演化，由此形成的风险溢出已成为影响证券市场乃至整个金融市场安全稳定和健康发展的一大难题。在此背景下，深入探究证券市场间风险的溢出机理，全面有效的识别证券市场风险信号，对挖掘证券市场风险的本质规律，揭示证券市场风险信号的非线性和不稳定特征，完善金融风险管理，以及有效提高证券市场的运行效率具有重要的理论价值和现实意义。.本项目以证券市场风险溢出及其防范为研究对象，有效结合时域方法和频域方法，科学、全面地识别风险信号的特征和结构，分析风险信号随时间和频率变化的特征，捕捉证券市场风险溢出的规律。为了更加深入的考察证券市场风险溢出，本项目还从微观层面进行了拓展研究。主要内容和研究成果包括：(1) 基于网络结构对国际房地产证券市场间的相关结构和动态相关性进行研究，发现房地产证券市场之间的相关系数具有时变特征，且在金融危机时期相关系数的波动较大。 (2) 基于时域和频域方法，分别构建GARCH(1, 1)-Diagonal BEKK模型、EEMD和变结构Copula模型对证券市场间波动溢出效应及股票市场系统流动性风险溢出进行了相关研究。(3) 基于PMFG和MST的股票市场网络拓扑稳定性分析结果表明，PMFG构建的股票市场网络模型拓扑稳定性均强于MST构建的股票市场网络模型；股票市场网络对于蓄意攻击表现出脆弱性，但对于随机攻击表现出稳定性。(4) 基于鲁棒加权非线性组合模型的人民币汇率预测研究表明，非线性组合模型较单个模型及线性组合模型预测效果更理想，更适合于类似人民币汇率等具有特殊性质的时间序列的预测。(5) 从微观层面出发，在对房地产债务融资结构、创业板上市公司现金股利政策以及公司治理对超额现金持有的影响的定性分析和定量研究的基础上，更加深入地探讨了证券市场各微观金融主体的风险识别及防范问题。本项目共形成论文9篇，其中6篇发表在国内外重点期刊(含采稿)，3篇已向SSCI/SCI投稿，为丰富证券市场调控模式和提升宏观经济监管水平提供了有力支撑。