Tail Dependence on Financial Markets

Understanding dependence between financial assets is of paramount importance in finance. If we go beyond the assumption of normal distributions, Pearson's linear correlation coefficient is no longer sufficient to completely describe dependence structure. Thus, there is a need for other measures of dependence. Since distribution tails are particularly important in finance, tail dependence should be investigated. Several tail dependence measures were proposed in the literature, tail dependence coefficient being the most well-known of them. This paper deals with some aspects of their practical usage. First, we show that despite vastly different theoretical properties almost all tail dependence measures produce highly similar results when applied to empirical data. Then we show how information on tail dependence can be used for portfolio selection. We manage to construct a portfolio that significantly outperforms portfolios constructed using more traditional methods.