Study on VaR based on GARCH and copula

Measuring a portfolio's value at risk (VaR) is very difficult when financial returns follow some unknown or non-normal marginal distributions. It is sometimes even impossible to specify the multivariate distribution of two or more returns. In order to overcome these difficulties, a new method based on generalized auto regressive conditional heteroskedasticity (GARCH)-copula model is proposed. Four plate indexes in Shanghai Stock Market, namely industry index (GY), utility index (GG), commerce index (SY), real-estate index (DC) are taken as the single assets of a portfolio. Using Monte Carlo simulation, the VaR of this portfolio is calculated. The result shows that VaR based on Kendall τ correlation coefficient is better than that based on linear correlation coefficient ρ.