A Family of Quantile Dependence Coefficients

Abstract

A popular measure of association is the tail dependence coefficient which measures the strength of dependence in either the lower-left or upper-right tail of a bivariate distribution. In this paper, we develop the idea of quantile dependence, which generalizes the notion of tail dependence and could be used to detect dependence in specific regions of the domain of a joint distribution function. Properties of the proposed quantile dependence coefficient are studied and several examples illustrate our results.