波动溢出网络视角的金融风险传染研究

当前各类经济风险交叉关联,金融系统的风险溢出效应备受关注,为刻画我国金融系统性风险传染的路径特征,本文从波动溢出网络的视角分析金融系统内部的风险传染机制。首先使用广义动态因子模型对收益波动的共同波动率成分和特质性波动率成分进行区分。然后,根据货币市场、资本市场、大宗商品交易市场、外汇市场、房地产市场和黄金市场之间的特质性波动溢出效应,利用基于TVP-VAR模型的方差分解溢出指数分析金融系统波动溢出的动态联动性和风险传递机制。在分析方向性波动溢出效应的基础上,采用方差分解网络方法构建起信息溢出复杂网络,从网络视角分析金融系统内部的风险传染特征。实证研究发现,房地产市场和外汇市场的净溢出效应绝对值相较于其他市场更大,其受其他市场风险冲击的影响强于对外风险溢出效应,而股票市场的单向对外风险溢出效应强度最大。在波动溢出的基础上,进一步考虑股市波动率指数与其他市场波动率指数进行投资组合的资产配置权重,计算了波动率指数投资组合的最优组合权重和对冲策略。研究结论有助于更好地理解我国金融系统的风险传染机制,对监管机构加强宏观审慎监管、投资者规避投资风险具有重要意义。

A Study of Financial Risk Contagion from the Volatility Spillover Network Perspective

Following the subprime mortgage crisis, China's financial market experienced a money shortage in 2013, a stock market crash in 2015, and a bond default crisis in 2018. Due to this turmoil in the domestic and international economic environment, the risk spillover effects of China's financial system have received much attention. As the cross-market and cross-region contagion of financial risks has grown, systemic risk warning and prevention have become the focus of research. Therefore, studying the risk contagion of financial volatility in this context not only provides guidance for regulators on maintaining financial stability and preventing financial crises stemming from systemic risk accumulation, but also has significant implications for investors' portfolio risk management.
To describe the financial systemic risk transmission path and characteristics in China, this paper analyzes the internal risk contagion mechanism of the financial system from the perspective of a volatility spillover network. The risk spillover relationship between markets is used as the connected edges of the network, and the variance contribution (as calculated by the variance decomposition) is used as the adjacency matrix. Next, the complex network of risk spillovers in China's financial system is constructed, and the risk contagion features of China's financial system can be analyzed using the topological structure of the network. We first use the general dynamic factor model to distinguish between the common volatility components of financial return volatility and the idiosyncratic volatility components. Next, based on the volatility spillover effects between the currency market, capital market, commodity trading market, foreign exchange market, real estate market, and gold market, the dynamics of the volatility spillover of the variance decomposition index based on TVP-VAR model are analyzed, and the linkage and risk transfer mechanisms are discussed. After analyzing the directional volatility spillover effects, the variance decomposition network method is used to construct the weighted and directed volatility information spillover network, and the risk contagion characteristics of the financial system network are described from the perspective of the network. Subsequently, the dynamic conditional correlation generalized autoregressive conditional heteroskedasticity (DCC-GARCH) tcopula model is utilized to describe the relationship between the nonlinear volatility index variables, and the optimal portfolio weight and hedging strategy of the volatility index portfolio are calculated. The framework is used for empirical research on different sub-markets within the Chinese financial system.
The research sample comprises the explanatory variable indicators of the 12 secondary financial sub-markets of the Chinese financial system from January 2007 to October 2018, based on data obtained from the Wind database. The empirical study reveals that the absolute values of the net spillover effects of the real estate market and the foreign exchange market are larger than those of other markets. However, due to the comprehensive influence of various factors such as the macroeconomic situation and regulatory policies, the risk spillover effects of the currency market show uncertainties. Although the stock market displays a high degree of risk spillover and risk tolerance, the net spillover effect is not strongly evident, indicating that the stock market plays the main role in transmitting risks between other financial markets.
On the basis of volatility spillovers, we further consider the asset allocation weight of portfolios using stock market volatility indexes and other market volatility indexes. The research conclusions are helpful for understanding the risk contagion mechanism of the financial system, making it possible to strengthen macro-prudential supervision and avoid investment risks.
This study makes three major contributions. First, we use GDFM to extract common volatility shock components and the idiosyncratic volatility components of sub-markets in the financial system. Second, the time-varying vector autoregressive variance decomposition spillover index is used to identify the dynamic linkage of volatility shocks between financial systems. Third, based on the asset allocation of portfolios using the stock market volatility index and other market volatility indexes, the DCC-GARCH tcopula model is used to characterize the non-linear heterogeneous volatility series between financial markets and calculate the optimal portfolio strategy for the volatility index based on the volatility spillover effects.