Quantile correlation coefficient: a new tail dependence measure

A quantile correlation coefficient is newly defined as the geometric mean of two regression slopes—that X on Y and that X—in same way Pearson related to coefficients. The a measure overall sensitivity conditional random variable changes in other variable. proposed can be compared across different tails within given distribution provide meaningful interpretations, for example, there stronger dependence left tail than overall. It also with correlation. Neither these comparability enabled by existing tail-dependence measures. Moreover test differences correlations at proposed. asymptotic normality estimated null are established well supported Monte-Carlo study. methods illustrated an analysis stock return price data sets, yielding clear indication left-tail right-tail dependence.