Copula를 이용한 은행부문의 시스템적 리스크 측정
Measuring the Systemic Risk in the Korean Banking Sector : The Copula Approach
이명활(한국금융연구원); 이긍희(한국방송통신대학교); 이종한(한국은행)
27권 2호, 81~123쪽
초록
전통적인 상관관계 측정방법을 통해서는 실제 리스크 익스포저 간의 가변적이고 복잡한 상관관계를 반영하여 시스템적 리스크를 추정하는데 한계가 있었다. 본고에서는 이와 같은 문제점을 극복하기 위해 Copula 방법 가운데 가장 최신 기법인 Vine Copula를 소개하고, 이를 우리나라 은행부문의 시스템적 리스크 측정에 적용하였다. Vine Copula 기법을 사용하여 10개 국내은행의 CDS 프리미엄 자료로부터 은행부문의 결합도산확률분포를 도출하고, 이를 토대로 결합부도확률, 금융안정지수 등 시스템적 리스크 지표를 시산하였다. 시산결과 은행부문의 시스템적 리스크는 2008년 리먼브라더스 파산 직후 급등하였다가 이후 하락하는 모습을 보였으며, 정규분포를 가정한 Gaussian Copula의 경우보다 글로벌 금융위기 당시의 시스템적 리스크를 잘 반영하는 것으로 나타났다. 본 연구결과는 정책당국의 거시건전성 정책 수행 시에 은행부문의 시스템적 리스크를 측정하고 금융기관 간 상호 의존성의 변화 및 시스템적 리스크의 변화를 파악하는 지표의 하나로 활용될 수 있을 것으로 기대된다.
Abstract
In the wake of the 2008 global financial crisis, systemic risks received heightened attention as the risks of financial institutions spread to risks of overall economy, through an intricate channel of interdependence among financial institutions. The significance of macro-prudential policies also rose sharply as a tool to prevent systemic risks. For effective implementation of macro-prudential measures, a rigorous estimation of systemic risks should precede. In estimating systemic risks, it is important to fully consider interdependence among financial institutions and reflect the time-varying characteristics of those structure. However, the current methods of estimating systemic risks based on the fixed correlation have limitations in reflecting the time-varying and complex relations among risk exposures. To address this problem, we introduce the Vine Copula method that can model multi-variate distributions while taking into account an intricate relationship of interdependence innate in financial data. And we apply it to estimating systemic risk in the Korean banking sector. First we derive the joint probability distribution in the banking sector using the Vine Copula, based on CDS premium data of top ten banks. Then, we calculate the Joint Probability of Distress(JPD) and the Banking Stability Index(BSI) from the joint probability distribution. During the process, both the default rate based on CDS premium data and the implied default rate (IDR) derived from bond yields are used to calculate default rates of individual banks. The results show that the systemic risk in the banking sector soared right after the Lehman Brothers collapse in 2008 and then began to decline. Also the systemic risk turned out greater when calculated based on the default rates from CDS premium data than the IDR derived from bond yields. Compared to the Gaussian Copula that assumes normal distribution, the results are more accurate in describing the systemic risk at the time of the 2008 global financial crisis. The findings suggest that the indicators, the Joint Probability of Distress(JPD) and the Banking Stability Index(BSI), can be used to assess changes in systemic risks and interdependence in the banking sector. To use the derived systemic risk indicators as a part of the Early Warning Indices, it will be necessary as a future study to identify macroeconomic and financial variables that affect the derived systemic risks, and forecast those risk indicators with the projection of the related variables. In addition, it will yield a more reliable outcome if the Joint Probability Distribution is derived using other financial variables such as stock prices and bond yields other than CDS premium data.
- 발행기관:
- 한국금융학회
- 분류:
- 경제학