动态尾部金融网络结构与传染性度量

基于Lasso高维分位数回归模型,本文构建了中国金融系统的尾部网络结构,并定义了总体网络以及行业间、行业内和金融机构间等多层金融网络的尾部风险传染度,解构其传染机制与关联特征,评估各机构双向系统重要性(接收端和发射端)。同时,本文提出了一个最优滚动窗宽选择标准方法,以优化滚动样本技术下的动态网络结构。结果表明,所有层级尾部风险传染效应(总体系统、行业间、行业内和机构间)在经济金融极端困境时期,呈现明显增强及剧烈震荡特征,2015年中国股灾期间尤甚。跨行业传染效应日益严峻,银行与保险间表现出较强关联性,房地产机构与其他金融机构间均表现出较高传染性,跨业监管值得关注。接收与发射最多尾部风险传染的金融机构仍然是银行与证券类机构。系统中超过50%的金融机构倾向于接收风险传染,一旦出现系统性冲击,整个金融系统的稳定性将遭受重创,因此,应加强此类机构应对外界冲击的能力。此外,基于新滚动窗宽选择标准法的动态模型的估计性能明显优于传统方法。研究结论有助于理解中国金融系统的网络结构和传染机制,对宏观审慎监管体系的建立提供了依据。

Dynamic tail financial network structure and contagion measurement:Based on rolling lasso high-dimensional time series model

The lessons learned from the 2007-2009 global financial crisis show that high-infectivity,high destructiveness and high complexity are the significant systemic risk characteristics.Constructing network structure and further clarifying spillover effects across financial firms is crucial to interpret systemic risk correctly,and it is becoming the main focus of academia and regulatory agencies.Moreover,substantial evidence shows that the tail risk spillover effects are always dynamic due to the internal and external shocks.Thus,the structural changes in the risk transmission relations are a remarkable element for constructing network structure.This paper explores the tail risk dependence among financial firms using the extended version of the high-dimensional lasso quantile regression model.We investigate the tail risk spillover effects and systemic risk of 52 listed financial firms in China,consisting of banks,insurances,securities,and real estate firms,ranging from January 13,2013,to March 25,2018.The analysis mixes the market information(weekly stock returns),book information(quarterly financial statements),and the macroeconomic information(macroeconomic state variables).In this paper,we decompose the network contagion relations into multiple levels,that is,the aggregate network connection of the whole financial system,the internal-industry,across financial sectors,and across financial firms.Furthermore,we evaluate the systemic importance of financial firms from the risk emitter and risk receiver perspectives.First,we measure the financial system′s tail risk network structure using the combination model of the high-dimensional quantile regression model and the Lasso penalty method,revealing the complicated contagion relationships in the whole financial system,between financial sectors,within financial sectors,and between financial firms.According to this,financial institutions′network dependence is ranked from the receiving and the transmitting perspectives.Thus,we can obtain the systemic importance of financial firms from the aspects of“network connection”.Second,to better capture the tail risk infection relationship′s structural mutation,this paper establishes a statistical standard to select the optimal rolling window width.More precisely,we minimize the model estimators′quadratic loss function and maximize the standardized Euclidean distance between sub-models,weighing two contradictory goals of model estimation accuracy and time variability.The empirical results show that the new rolling window width selection criterion significantly improves the dynamics identification of the financial network structure compared to the traditional approaches,showing the necessity and effectiveness of choosing the reasonable rolling window.For the Chinese financial network,all tail risk contagion effects(system-wide,cross-industry,inter-industry,and cross-institutional)show a significantly sharp increase and high fluctuations during extreme economic distress,for instance,the 2015 stock crash in China.The cross-industry spillover effects are increasingly grim with strong correlations between banks and insurance,and real estate institutions and other financial institutions,deserving more attention from regulations.Regarding the risk contagion within financial sectors,the risk spillover effects between banks are the weakest,while the contagion intensity is more vigorous within the insurance sector and real estate.The financial firms receiving and transmitting the most risk contagion are still banks and securities.More than 50%of financial institutions in the system tend to receive risk contagion.That is,systemic risk exposure is significantly higher than the systemic risk contribution.If a systemic shock strikes unexpectedly,the whole system will be hit hard.Therefore,the capacity of such institutions to handle external shocks should be strengthened.These findings in this paper help understand China′s financial system′s network contagion characteristics,revealing how the tail risk of financial firms is connected.The network analysis also provides insight into how to measure and define systemic risk.For financial regulatory and financial institutions,our study shows a basis and direction for systemic risk management and supervision.