Working with zero inated data sets often requires the computation of multivariate margins of a density. This is usually done by integration. For R-vine densities this integration is hardly feasible even if the dimension of the integral is only two or three. So we develop a method to approximate multivariate margins of a R-vine copula. Therefore we introduce a tool called R-vine matching which allows to switch between R-vines with different structure and different bivariate copula families but which measure the conditional and unconditional dependence (measured in terms of Kendall's tau) between its underlying (uniform distributed) random variables similarly. Moreover, some multivariate margins of a R-vine can already be obtained directly. We prove a criterion, which has to be fulfilled in order to obtain a multivariate R-vine margin in closed form. In our application, we use this computation and approximation of multivariate R-vine copula margins to compare different R-vine copula models for simulated operational risk data which usually is strongly zero innate. We estimate the value at risk (VaR) for different quantiles α and evaluate the impact of realistic dependence modeling on estimating the total regulatory capital which turns out to be up to 9% smaller than what the standard Basel approach would prescribe.      «

Working with zero inated data sets often requires the computation of multivariate margins of a density. This is usually done by integration. For R-vine densities this integration is hardly feasible even if the dimension of the integral is only two or three. So we develop a method to approximate multivariate margins of a R-vine copula. Therefore we introduce a tool called R-vine matching which allows to switch between R-vines with different structure and different bivariate copula families but w...     »