Summary and Outlook
In this thesis we discussed the valuation of correlation products, in particular the valuation of CDS index tranches. The one-factor Gaussian copula approach, which is considered to be the market standard, assumes in its simplest version that all pairwise default time correlations in the underlying portfolio are the same. Using this setup, different correlation levels are needed to reproduce the market prices of different tranches on the same underlying portfolio. Generally, mezzanine tranches trade at lower correlation levels than equity and senior tranches. This phenomenon has come to be denoted as the correlation smile. The correlation smile illustrates that the standard market model is not able to consistently reproduce market prices, because different parameter sets are used to reproduce the prices of the different tranches of one CDO structure. In this study, we focused on the use of heterogeneous correlation structures, in order to explain the observed correlation smile.
KeywordsCredit Risk Dependence Structure Market Data Copula Model Gaussian Copula
Unable to display preview. Download preview PDF.