| Abstract [eng] |
In this paper, the multivariate structure of dependence among a number of Exchange rates against the euro is explored. The computation of quantitative estimates of dependence may be useful in making decisions concerning the international transactions assessment, the exchange rate forecasting, and the valuation of derivatives such as multivariate currency options. The complexity the problem lies in multidimensionality and dependence of Exchange rates on time. The copula approach enables description of the multivariate distribution necessary to model nonlinear relationships that usually exist among financial variables, such as exchange rate data. The canonical vine, D-vine and d-dimensional copula models are explored for special cases such as Gaussian and Student’s T dependence structures. The experiments performed show that the best fit to original data is achieved with D-vine multidimensional model if Student’s T copula is employed between the pairs of considered Exchange rates. The results are validated by comparing marginal distribution. |
