Dependence between Croatian and European stock markets – A copula GARCH approach

The objective of this paper is to analyze dependence structure between the returns of Croatian and five European stock markets (Austrian, French, German, Italian, and the U.K.'s). We propose a copula GARCH approach, where the return series are modeled as univariate GARCH processes and the dependence structure between the return series is defined by a copula function. Four different copulas are fitted - a constant and conditional normal and symmetric Joe-Clayton (SJC) copulas - and estimated by a semi-parametric method. We found that the time-varying normal copula yields the best fit for CROBEX-CAC40, CROBEX-DAX, and CROBEX-FTSE-MIB stock indices pairs, while the time-varying SJC copula is the best fit for CROBEX-ATX and CROBEX-FTSE100. Further, we found that the probability of simultaneous extreme positive and negative returns in Croatian and other European stock markets can increase to 0.77 during turbulent times. The lower and upper tail dependence dynamics between Croatian and other European stock markets is similar in pattern, differing only in scale. The basic conclusion of the research is that the dependence between the stock markets of Croatia and five major European stock markets is dynamic and can be properly captured by either a dynamic normal or symmetrized Joe-Clayton copula GARCH models.