金融机构风险溢出：机理、测度与监管

The project focuses on risk spillover problems among different types of financial institutions. Firstly, the risk spillover mechanism of financial institution is analyzed from multiple perspectives, such as the balance sheet relationship, investor behavior, as well as information transmission, and the main financial institutions risk spillover channels are induced. Secondly, using copula function, CoVaR and CoES are improved as a new risk spillover effects measure. To estimate the modified measures more accurately, we incorporate high-frequency information by the realized GARCH model, and capture the influence of different market states by combining the regime switching mechanism, and ultimately propose a new approach to measure risk spillover effects among financial institutions. Due to the richer information content, mainly including high-frequency and structural breakpoint information, it is more realistic and robust that we detect quantitative information of financial institutions risk spillover effects by the MRS and realized GARCH based model rather than traditional methods in the mainstream researches.Finally, under each market state, we attribute the overall risk spillover effects to different risk spillover channels. According to the attribution analysis result, co-ordinate regulatory policies are proposed to release spillover risk among financial institutions and to support decision making for our current financial supervision system, i.e. ‘One bank, three commissions’. The effectiveness of policies is basically ascertained through numerical simulations.

本项目专注于不同类型金融机构之间的风险溢出问题。首先，从资产负债关联、投资者行为与信息传播等多个角度系统分析金融机构之间的风险溢出机理，归纳出主要的金融机构风险溢出渠道。其次，利用copula函数改进CoVaR和CoES作为风险溢出效应测度。再次，通过realized GARCH模型纳入高频信息，同时引入区制转换机制（regime switching mechanism）刻画不同市场状态对风险溢出的影响，提出基于MRS-realized GARCH模型的金融机构风险溢出效应测度计算的新方法，通过增加信息含量（高频波动信息和结构变点信息）获得比主流研究更加真实和稳健的金融机构风险溢出效应定量信息。最后，在不同市场状态下，将总的风险溢出效应归因于不同风险溢出渠道，以归因分析的结果为参考，为“一行三会”提供针对金融机构风险溢出的协调监管政策建议，并通过数值模拟验证政策的有效性。

本课题专注于不同金融机构之间的风险溢出问题，其主要贡献为：首先从系统性风险累积、传导、爆发和扩散的形成机理分析金融机构系统性风险从突变、权变到金融系统性危机的动态演变过程；其次，改进并提出新的金融机构风险溢出效应测度方法，如改进CoVaR方法，引入系统性风险测度的SDSVaR模型、基于非线性视角提出MVCoVaR模型、提出考虑行为因素的风险溢出效应测度方法、将混频Copula模型与CoVaR模型相结合进行溢出效应的测度、构建HAR-RV模型将高频信息和不同频率已实现波动纳入到边缘分布建模过程等；最后根据以取得的研究结果，给出相应的监管政策建议。