•First •Prev •Next •Last •Go Back •Full Screen •Close •Quit

Content of the talk. • On high regimes of default correlation. • A simple model to take randomness of correlation and systemic risk into account. • Examples from the CDO and ABS market. • Conclusion: importance of alternative methodologies for risk man-agement. •First •Prev •Next •Last •Go Back •Full Screen •Close •Quit Some conclusions from the subprime crisis • Complex structured products (CDOs of ABS...) have been deviced so as to meet rating criteria. Strong model risk (too much dependence on the underlying models). • Standard models failed to quantify default risk in CDOs of ABS. In particular, AAA credit enhancements have proved entirely mispriced. • More broadly: General failure to capture systemic contagion and high default correlations regimes (credit quality of borrowers, monoline insurers, housing market, liquidity crunch...). •First •Prev •Next •Last •Go Back •Full Screen •Close •Quit Our purpose today: • Focus on certain shortcomings of standard pricing methods for structured finance and credit risk assessment. • Favorite example: CDO tranches, but conclusions hold for any credit port-folio / structured product. • Concentrate on senior tranches which price reflects extreme risks (systemic and macroeconomic risks, liquidity, funding and counterparty risk, some operational risk...) •First •Prev •Next •Last •Go Back •Full Screen •Close •Quit On Gaussian assumptions • Recall briefly the standard methodology for large portfolios credit risk as-sessment (Bank exposures in Basel 2 in the advanced IRB approach, in-cluding retail investment; CDO tranches on CDS...): • In the basic one-factor Gaussian copula model, the default of the i-th issuer is triggered by the value of a Gaussian variable: Gi = ρV + 1 − ρ2V