Measuring and Predicting Systemic Risk in the Chinese Banking System

This paper highlights the importance of measuring systemic risk of commercial banks. Conditional Value-at-Risk (CoVaR) is used to measure the degree of "risk externalities" that a specific bank contributes to the whole banking system. Our analysis not only presents current levels of systemic risk of individual banks but also the changes with time passes. There is some evidence that larger banks contribute more to systemic risk, but size is far from being a dominant factor. We further explore to use some determinant balance-sheet factors to predict forward CoVaR for regulatory purpose. We extend modified Support Vector Regression (SVR) specifically for panel data, and apply the new model to predict systemic risk of commercial banks. The results show that the model is suitable for this problem.