Analyzing Asymetric Dependence in Exchange Rates using Copula

In this paper I aimed to analyze the use of copulas in financial application, namely to investigate the assumption of asymmetric dependence and to compute some measures of risk. For this purpose I used a portfolio consisting in four currencies from Central and Eastern Europe. Due to some stylized facts observed in exchange rate series I filter the data with an ARMA GJR model. The marginal distributions of filtered residuals are fitted with a semi-parametric CDF, using a Gaussian kernel for the interior of distribution and Generalized Pareto Distribution for tails. To obtain a better view of the dependence among the four currencies I proposed a decomposition of large portfolio in other three bivariate sub-portfolios. For each of them I compute Value-at-Risk and Conditional Value-at-Risk and then backtest the results.