Copula based Change Point Detection for Financial Contagion in Chinese Banking

In this paper, a change point detection approach based on copula with two notable advantages is put forward. One is that the approach can deal with the common but special unbalanced panel data. The other is that it can detect multiple change points. Firstly, a proper copula that most accurately describes the dependence structure of the data is chosen. Then, the chosen copula is fitted to the data dynamically by adding new data. Finally, the change points are located by analyzing the trends o f fitted parameters of the copula. Based on the quarterly financial data of 16 listed Chinese commercial banks, we empirically use the proposed approach to detect the subprime crisis contagion period in Chinese banking. The results show that the contagion starts in 2007Q2 and ends in 2009Q1, which is reasonable according to relevant researches.