Valuing Index Collateralized Debt Obligations Tranches－Using Implied Copula

• Main Authors: Lin, Yi-Shan, 林意珊
• Other Authors: 曾國雄，謝嘉鴻
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/88694551828005077899

碩士 === 國立交通大學 === 科技管理研究所 === 96 === 　　Asset Securitization originated in the 1970’s. As Southeast Asia and Korea’s financial crises struck the financial market, financial institutions could not collect debts because of the credit defaults caused by the bankrupt companies, and therefore Credit Derivatives were used to hedge credit risks. The Subprime Mortgage crisis which is an ongoing global economic problem started in the United States in late 2006 and began with the bursting of the housing bubble. It made major financial institutions face significant losses from investments in Mortgage Backed Securities (MBS) or Collateralized Debt Obligations (CDO). The recent market turmoil highlights the importance of the theme about credit default risks. For the various approaches to the valuation of CDO, the standard market model is Gaussian copula or base correlation. The market quotes observed through past models can not accurately match the actual volatility in the market because of wrong estimates of the extent of the default losses. For this reason, we use the implied copula approach which was first applied by Hull and White (2006). We no longer find the default correlations from the copula model, but imply them from the market quotes directly. In this paper, we chose CDX NA.IG to demonstrate how to value standardized CDO tranches. Standardization applies to both the composition of the reference pool and the structure of the tranches. We confirm that our model fits the market quotes exactly and make it clear to be used for pricing, trading, and risk management. It also enables non-standard credit derivatives, such as bespoke CDOs and CDO squareds, to be priced directly with market quotes for standardized CDO tranches. The spreads we find are more stable than those given by Gaussian copula or base correlation because we don’t permit any arbitrage.