Events in financial markets since summer 2007 have clearly illustrated the importance of being able to analyse how the return on different kinds of assets that usually seem uncorrelated can begin to show a strong covariance in the context of so-called extreme events. However, such phenomena cannot be captured using ordinary measures of correlation. The article presents a method of measuring such covariance called copulas. Examples are also provided of the practical application of copulas.

 Read the article (pdf, 1Mb)