关键词: 尾部风险溢出, 驱动因素, 投入产出关联, LASSO分位数回归

Abstract: Since the real economy and finance are interdependent and should grow and thrive together, the paper uses LASSO quantile regression technique to construct a tail risk network of China’s industry system which includes the financial and real industries. We use complex network analysis method to measure and analyze the tail risk spillover among industries and investigate the driving factors of tail risk spillover in the time-series and cross-sectional dimensions. We find that firstly, the 31 industries in China’s economic and financial system have formed a close tail risk connected network. There is close tail connectedness among financial industries, among real industries, and between the financial and real industries. Moreover, in the interaction with the real industries, there is a structural imbalance within the financial industries. Secondly, the source of risk in the economic and financial system may come from the financial industry or real industry. Banks, Pharmaceutical Biology, Computer and Construction Decoration Materials are identified as important “tail-risk drivers” in the economic and financial system. Thirdly, in the cross-sectional dimension, the closer the input-output linkage between industries, the higher the level of tail risk spillover, and compared with forward linkage, backward linkage has stronger explanatory power for tail risk spillover between industries, that is, tail risk mainly transmits from downstream to upstream industries along the industrial chain. In addition, the higher the return correlation and volatility correlation between two industries, the stronger the tail risk spillover; the higher the industry’s own risk level, the more vulnerable it is to risk contagion from other industries, and the stronger the received tail risk spillover from other industries. In the time-series dimension, the macroeconomic conditions and financial environment are the main driving factors for the dynamic changes of the industry’s tail risk spillover.

Key words: Tail Risk Spillover, Driving Factors, Input-output Linkage, LASSO Quantile Regression