Abstract
This article analyses dynamic tail dependence between the returns of the three largest Central and Eastern European (CEE) stock markets (Hungary, Czech Republic and Poland) and two major Eurozone stock markets (Germany and France). Tail dependence is modelled by a constant and dynamic ‘symmetrized Joe-Clayton’ (SJC) copula assuming GARCH stock market return processes. The results of the dynamic SJC copula model show that the dependence between pair-wise observed stock markets is time-varying and asymmetric with lower tail dependence mostly exceeding upper tail dependence. The results of the article imply that advantages of international portfolio diversification are reduced in downturns.
Notes
1 In the first step of copula estimation, we filter the stock market returns with a GARCH family model that provides the best fit to the volatility of the univariate return series. We chose amongst univariate GARCH family models that are usually used for this purpose: GARCH(1,1), GARCH(1,2), GARCH(2,1), GARCH(2,2), EGARCH (1,1), APGARCH(1,1) and GJR-GARCH(1,1). The best fitting model is then selected based on the Akaike information criteria. On the basis of the standardized residuals from these univariate GARCH family models, we estimate empirical distributions on the filtered returns which are input into the second stage of estimation.