Valuation of Collateralized Debt Obligations

• Main Authors: Fu-Ching Lee, 李福慶
• Other Authors: Szu-Lang Liao
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/01527688689619846985

博士 === 淡江大學 === 管理科學研究所博士班 === 96 === After 2002, many originators turn to consider various bespoke tranches of other CDOs (including standardized contract of CDS Index) as underlying of collateral for both raising the return of its tranches and diversifying underlying of its collateral. The type of exotic CDOs is referred to as Synthetic CDO-Squared. Currently, the Taiwanese financial industry is going through a period of transformation. The banks face not only the pressure of operating pressures owing to declining interest rate, but also the pressure to control credit risk on loans. Additionally, the crisis resulting from successive corporate bond default events and a huge loss of inverse floating structured notes have led to the switch of financial investments from bond mutual funds to Collateralized Debt Obligation (CDO) markets in Taiwan. The Southeast Asian financial crisis occurred in year 1997, followed by the Asian financial crisis of affecting S. Korea of 1998. They both resulted in numerous multinational enterprises laying off employees, shutting down factories, and even going bankrupt, and caused widespread financial pain. For risk management and valuation of multi-name credit derivatives, the estimation of the default dependence is considered an extremely important factor. Default dependency may be influenced by both overall economy factor, sectoral and firm-specific factors. For hybrid portfolio of the first CDO-Squared issued by the Polaris Securities Group in Taiwan, we propose hybrid factor copula model which involves CIR stochastic intensity model and KMV-Merton Model developed by Leland (2004) under random recovery rate environment to price CDO-Squared. Compared with the double student’s t factor copula method developed by Hull and White (2004), we find that the proposed model which uses CIR intensity rate, random recovery rate of various secured-level brackets, and double student’s t copula produces fair credit spreads of tranches lower than the Hull and White (2004) model except 15%~30% tranche. The assumptions of positive mean-reverting hazard rate and stochastic recovery rates of various classifications are more realistic since CDS has daily market quotes of different maturities, and thus market trading can expose credit spread information of obligors.