WASS PhD workshop onusing copulas in risk modelling

Copulas are statistical methods to model the joint distribution of variables such as prices and yields.  Unlike simple correlation coefficients such as Spearman’s rank correlation, copulas characterize the non-linear dependence between a set of variables. They enable to better account for the extreme parts of probability distributions in risk modelling and are increasingly applied in e.g. crop yield insurance and price risk management In rating agricultural insurance contracts, copulas can capture the state dependent nature of the dependence between e.g. yields in different geographical areas, yields of different crops, and the dependence between prices and yields. Empirical evidence has shown that the spatial correlation of yields depends on the state of nature, i.e. the correlations become stronger during extreme weather conditions than in a typical year. Such a non-linear correlation of yields can be captured with copulas. Another advantage of copulas is that they allow the flexibility of choosing the marginal distributions of the dataset.

Besides agricultural insurance, copulas are widely applied in the field of finance (e.g. evaluation of the interdependence between asset returns of investment portfolios). More broadly, the workshop is relevant for all PhDs and fellows whose projects deal with modelling various sources of environmental and economic uncertainty. Barry’s experience with regard to copulas is in the area of crop yield and revenue insurance.