WorksMathematics
Dependence measures as spectral risk measures
Abstract
For a bivariate copula with , the conditional distribution functions are random variables with means , so that measures of directed dependence become risk functionals. The absolute Lorenz curves of the variables form the rearranged copula . In this picture Chatterjee's rank correlation is a squared Sobolev seminorm and the rearranged Spearman correlation an norm of , and both arise from one spectral risk measure, applied either to the conditional laws of or to the laws of . The resulting classes and of spectral dependence measures have mirror-image axiomatic properties. For every spectrum we determine the exact upper bound of in terms of , and we prove sharp inequalities between , and the values of , which show how little global measures say about the tails. For plug-in estimators based on the empirical checkerboard copula we obtain consistency, rates and a resolution limit, and a simulation study shows that tail-localised measures separate copulas with identical . An application to large credit portfolios illustrates the results.