Interconnectedness and systemic risk network of Chinese financial institutions: A LASSO-CoVaR approach

• Published in: Physica A 2019-11, Vol.534, p.122173, Article 122173
• Main Authors: Xu, Qifa, Li, Mengting, Jiang, Cuixia, He, Yaoyao
• Format: Article
• Language: English
• Subjects: CoVaR; Interconnectedness; LASSO-CoVaR; Network topology; Systemic risk network
• ISSN: 0378-4371; 1873-2119
• DOI: 10.1016/j.physa.2019.122173

The 2015–2016 China’s stock market crash raises awareness of risk contagion in financial system. How to investigate systemic risk from the perspective of network is still a challenging work especially for considering a large number of financial institutions. To this end, we introduce the least absolute shrinkage and selection operator (LASSO) method into the CoVaR estimation to construct a systemic risk network between financial institutions’ tail risk exposures. First, we apply the LASSO-CoVaR based systemic risk network to investigate the interconnectedness and systemic risk of financial institutions in China from 2010 to 2017. Our empirical results show that the interconnectedness among institutions is very important and cannot be ignored in estimating CoVaR of an individual institution. Second, the topology analysis shows that the system-level interconnectedness reaches a peak when the system is under distress, especially before and after the stock market crash occurred. Third, we rank institutions in terms of the systemic risk contribution and find that their systemic importance changes in four different sub-periods. To sum up, our empirical results reveal substantial relevant risk spillover channels and identify the systemically important financial institutions in China, providing useful information for regulators to formulate macro prudential supervision policy. •The LASSO-CoVaR approach is developed to construct systemic risk networks.•It measures systemic risk and contagion channels from the perspective of network.•Its strength is illustrated through applications on Chinese financial institutions.•The results provide useful information to help macro prudential supervision.