Dynamic financial tail risk networks: A backtesting-based conditional expected shortfall approach

Area of Science:

• Quantitative Finance
• Financial Risk Management
• Network Analysis

Background:

• Traditional quantile regression and copula models have limitations in capturing high-dimensional dynamic tail risk.
• Accurate measurement of tail risk and contagion is crucial for financial stability.

Purpose of the Study:

• To develop a Factor-Copula methodology for constructing high-dimensional dynamic tail risk networks.
• To overcome the limitations of existing models in assessing financial network risk.
• To analyze risk spillovers and contagion among Chinese financial institutions.

Main Methods:

• Development of a Factor-Copula methodology based on conditional expected shortfall (CoES).
• Backtesting of the CoES model using cumulative joint violations and conditional coverage tests.
• Dynamic analysis of tail risk networks for Chinese listed financial institutions.

Main Results:

• The proposed factor-copula-CoES model significantly reduces rejection rates in conditional backtests compared to benchmark models.
• Dynamic analysis reveals that network topology characteristics align with real market risk events.
• Quantitative analyses demonstrate distinct impacts of institutional and market factors on risk spillovers and contagion.

Conclusions:

• The Factor-Copula-CoES methodology provides a more robust framework for dynamic tail risk network construction.
• The model enhances the accuracy of risk assessment and provides valuable insights into financial contagion.
• Understanding the drivers of risk spillovers is essential for mitigating systemic risk in financial markets.