Polish stock market and some foreign markets – dependence analysis by regime-switching copulas

The aim of this paper is investigation of DJIA, DAX, ATX and WIG20 interdependence based on weekly returns. In order to capture asymmetry of dependence structure Archimedean copulas were applied and symmetric structures are modelled with elliptical copulas. The strength of dependence between extreme events is examined by tail dependence coefficients. Changes in dependence patterns and parameter values are obtained by the application of the regime–switching model based on the first order Markov chain. We are using a two-step maximum likelihood estimation method which separates marginal distributions from the dependence structure. Parameters of copulas are estimated using Hamilton filter adopted to copulas. The copula based on regime–switching model allows us to model time varying dependence structure in a very flexible way. Empirical results confirm dynamic and asymmetric structure of dependence represented by stock markets under study, especially they verify strong and dynamic lower tail dependence.