Default correlations

Abstract

While correlation (and diversification) is important to all investments, it is especially important to credit-sensitive securities as evidenced by the role of correlated defaults in the recent credit crisis. Correlation in the context of credit is a difficult issue because of sparse historical data on individual, and especially multiple, defaults. Hence, researchers and market participants have used market prices to construct forward-looking estimates of correlation. However, the only known ways of doing so are from multi-name securities (such as indexes, or collateralized debt obligations). Such instruments are far fewer in number than possible pairs of single-names (i.e. individual stocks, bonds, etc) – i.e. they are non-existent for most pairs. For these reasons, the modeling of correlations is termed by many as an extremely important problem in credit risk pricing.