A time-varying copula approach for modelling dependency: New evidence from commodity and stock markets

Abstract

This paper examines the time-varying conditional dependency between commodity markets and stock markets by applying the rolling-sample technique on the dependence parameter of copula. The dataset consists of the closing prices of twelve commodities and the SP500, CAC40, DAX30 and FTSE100 indices during the period from July 7, 1992 to February 17, 2015. To date precisely the breakpoints in the dynamics of the copula parameter of dependence, we employ Bai and Perron’s (BP, 1998, 2003) structural break-testing procedure.Our empirical findings show that among the seven copulas investigated, the Student’s t-copula is more appropriate for modelling dependency. Moreover, the BP procedure shows strong evidence of time-varying behaviour in the parameter of dependence. The results show that the dates of breaks correspond to economic and financial events, such as the global financial crisis and crude oilprice fluctuations.