Estimating risk of foreign exchange portfolio: Using VaR and CVaR based on GARCH–EVT-Copula model

This paper introduces GARCH–EVT-Copula model and applies it to study the risk of foreign exchange portfolio. Multivariate Copulas, including Gaussian, tt and Clayton ones, were used to describe a portfolio risk structure, and to extend the analysis from a bivariate to an nn-dimensional asset allocation problem. We apply this methodology to study the returns of a portfolio of four major foreign currencies in China, including USD, EUR, JPY and HKD. Our results suggest that the optimal investment allocations are similar across different Copulas and confidence levels. In addition, we find that the optimal investment concentrates on the USD investment. Generally speaking, tt Copula and Clayton Copula better portray the correlation structure of multiple assets than Normal Copula.