Post by Sapphire Capital on Aug 21, 2008 23:07:41 GMT 4
Parameter Uncertainty and the Credit Risk of Collateralized Debt Obligations
Erik Heitfield
Federal Reserve Board - Division of Research and Statistics
July 15, 2008
Abstract:
This paper examines empirical challenges inherent in evaluating the credit quality of collateralized debt obligations (CDOs). In order to assess the likelihood and expected severity of a CDO tranche's losses, one needs to know the distribution of losses for each collateral asset backing the CDO, as well as the dependence of losses across collateral assets. Rating agencies and other market participants rely almost exclusively on simple copula models to describe the distribution of CDO collateral losses. For more senior CDO tranches, standard credit risk metrics such as probability of default and expected loss are highly sensitive to copula model parameters that are not directly observable. Given assumptions about the data available to a credit analyst, I compute bounds on the accuracy of copula model parameter estimators and show that in applied settings data constraints are likely to impose severe limitations on an analyst's ability to accurately evaluate CDO tranches. Thus, commonly used indicators of CDO credit quality such as credit ratings should be viewed as more preliminary and less informative than comparable indicators for unstructured corporate bonds.
papers.ssrn.com/sol3/Delivery.cfm/SSRN_ID1190362_code186796.pdf?abstractid=1190362&mirid=1
Erik Heitfield
Federal Reserve Board - Division of Research and Statistics
July 15, 2008
Abstract:
This paper examines empirical challenges inherent in evaluating the credit quality of collateralized debt obligations (CDOs). In order to assess the likelihood and expected severity of a CDO tranche's losses, one needs to know the distribution of losses for each collateral asset backing the CDO, as well as the dependence of losses across collateral assets. Rating agencies and other market participants rely almost exclusively on simple copula models to describe the distribution of CDO collateral losses. For more senior CDO tranches, standard credit risk metrics such as probability of default and expected loss are highly sensitive to copula model parameters that are not directly observable. Given assumptions about the data available to a credit analyst, I compute bounds on the accuracy of copula model parameter estimators and show that in applied settings data constraints are likely to impose severe limitations on an analyst's ability to accurately evaluate CDO tranches. Thus, commonly used indicators of CDO credit quality such as credit ratings should be viewed as more preliminary and less informative than comparable indicators for unstructured corporate bonds.
papers.ssrn.com/sol3/Delivery.cfm/SSRN_ID1190362_code186796.pdf?abstractid=1190362&mirid=1