Dependence and extreme correlation among US industry sectors

Abstract

Purpose – The purpose of this paper is to examine the degree of dependence and extreme correlation (i.e. tail dependence) among US industry sectors. Design/methodology/approach – This paper makes use of both conventional measures of dependence (the Pearson’s correlation coefficient, Spearman’s rho and Kendall’s tau) and copula measures of extreme correlations (including the same-direction and cross-tail dependence coefficients) to explore sector diversification opportunities. The paper splits the full sample in three periods, namely, 1995 to 2000, 2001 to 2006 and 2007 to 2012, to access the extent to which the dependence results change through time. Findings – This research provides three important findings. First, the degree of dependence and same-direction extreme correlations are high, whereas the cross-extreme correlations are considerably low. Second, the sector pairs offering the best and worst tail diversification change across sample periods. Third, the traditional dependence measures suggest that benefits for sector diversification have decreased over all sample periods, while the potential for sector diversification during extreme events has just started to disappear in the most recent period. Practical implications – An investor should consider both the normal co-movements and extreme co-movements among sector indices to maximize diversification benefits. Originality/value – Given the limited empirical investigations of the degree of dependence and extreme correlation at a sector level, the results from this research should provide additional and valuable information for both investors and empirical researchers about portfolio diversification and risk management.