A copula, in credit modelling, is a mathematical function that combines the separate probability distributions of several variables, such as the times to default of different borrowers, into one joint distribution. Sklar's theorem shows that any joint distribution can be split into its individual distributions and a copula that describes only how they depend on each other, written as . This lets modellers calibrate each name's default probability, for example from credit default swap spreads, and then choose a separate assumption for correlation.
The one-factor Gaussian copula, popularised by David X. Li around 2000, became the market standard for pricing collateralized debt obligations and each tranche of a credit portfolio. Its weakness was that a single correlation figure could not capture the tendency of defaults to cluster in a crisis, and the model was widely criticised after the global financial crisis for understating the risk of senior tranches. Alternatives such as Student-t copulas and stochastic correlation models were developed in response.
Ref 1020000122