VaR측정 방법의 비교 분석

• 영문명: Comparative analyses on the VaR estimation methods
• 발행기관: 건국대학교 경제경영연구소
• 저자명: 남명진(Myung Jin Nam) 여성칠(Sung Chil Yeo)
• 간행물 정보: 『상경연구』제28권 제1호, 13~31쪽, 전체 19쪽
• 주제분류: 경제경영 > 경제학
• 파일형태: PDF
• 발행일자: 2003.05.30

영문 초록

The concern of risk management was increased in the korean financila institutions due to the IMF crisis which occurred from autumn 1997 to spring 1988. Now our financial institutions should cope with a change about financial environment dynamically by executing risk management and risk hedge which are based on a objective theory. Thus we considered the characteristics and the problem of the VaR(Valeu-at-Risk) estimation methods through the comparison analyses on three methods for the portfolio VaR estimation of the financial institutions to determine the policy for the risk management. In this paper, we discussed the advantages and weakness of the VaR estimation methods based on the variance-covariance method, the time series models and the extreme value models through the emprical study. First, variance-covariance method is easy to obtain the VaR estimates by assuming the multivariate normal distribution. But there would be some problems if data followed fat tailed distribution. Second, we applied bivariate GARCH(p, q) model to estimate VaR. There was also a problem about the choices of orders p and q. Third, we applied extreme value model for VaR estimation. Extreme-Value distributions were divided by the three types which depend on the size of tail index, namely Frechet, Weibull and Gumbel Distribution. In order to find k, the number of upper ordered values, we used the method which is to fit the extreme value distribution at a range of tail index, and to look for stability of parameter estimates. However, we felt that there wes a subjective judgement for the choice of number of upper ordered values. Next, we used the correlations between the assets to estimate portfolio VaR. However, the joint distribution of the extreme marginal distributions is not necessarily the distribution of the extremes for the aggregate position. So, there was a weakness that the precision of estimation is decreased. Thus, we want to apply the multinariate extreme value theory for estimating portfolio VaR more accurately for a future study.

목차

Ⅰ. 서론

Ⅱ. VaR의 측정 방법

Ⅲ. 실증 분석

Ⅳ. 결론

참고문헌

Abstract