摘要

Volatilities of crude oil price have important impacts on the steady and sustainable development of world real economy. Thus it is of great academic and practical significance to model and measure the volatility and risk of crude oil markets accurately. This paper aims to measure the Value-at-Risk (VaR) and Expected Shortfall (ES) of a portfolio consists of four crude oil assets by using GARCH-type models, extreme value theory (EVT) and vine copulas. The backtesting results show that the combination of GARCH-type-EVT models and vine copula methods can produce accurate risk measures of the oil portfolio. Mixed R-vine copula is more flexible and superior to other vine copulas. Different GARCH-type models, which can depict the long-memory and/or leverage effect of oil price volatilities, however offer similar marginal distributions of the oil returns.