Modelling the Risk of CDO dynamics

• Applicant: Professor Dr. Wolfgang Karl Härdle
• Subject Area: Statistics and Econometrics
• Term: from 2009 to 2014
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 136973041

The market standard model for pricing the CDO tranches is the one factor Gaussian Copula Model. The model however cannot reproduce the market prices and one observes the well known implied correlation smile, see Figure 1. Many papers have been written on models explaining the empirical smile effect. More complex models explain this fact for fixed point in time but so far fail in correctly describing the time varying dependence structure. Given our joint experience with adaptively estimated time varying parameters and the expertise on modeling and pricing CDOs we can successfully transfer these technologies to time varying CDO risk analysis.Our aim is to study the dynamic behavior of the market implied correlation surface or equivalently the pricing surface in different valuation models in its time varying context. We will use for this aim modem statistical techniques (adaptive semi parametric estimation) and advanced pricing models and will create efficient algorithms for pricing and predicting CDO risks.The project is designed as a joint project between Berlin and Gießen since the expertise in mathematical modeling of derivative valuation and risk measurements in credit lies within the Mathematical Finance Group at the Universität Gießen and the statistical expertise for financial time series and time varying (high dimensional) dependency models is with C.A.S.E. - Centre for Applied Statistics and Economics and Institute of Statistics and Econometrics at Humboldt-Universität zu Berlin. A combination of these two centers of expertise is indispensable for the success of the proposed research project. C.A.S.E. is also a comerstone in the Humboldt University research strategy in creating a Humboldt Campus Mitte (HCM), a so called integrative research institute (IRI).

• DFG Programme: Research Grants
• Participating Person: Professor Dr. Ludger Overbeck