A Study on VaR Efficiency for SHIBOR
研究了三种方法（条件异方差模型、广义误差分布结合蒙特卡洛模拟、广义误差分布结合利率期限结构）在计算SHIBOR市场VaR时的表现，发现后两种方法分别对左尾和右尾更准确。
Risk management about SHIBOR will be more and more important for financial institutions. The au- thor studies conditional heteroskedasticity model, generalized error distribution with Monte Carlo simulation, and generalized error distribution combined with term structure of interest rate and uses maximum likelihood es- timation to estimate the parameter of generalized error distribution and generalized moment method to estimate parameters of term structure of interest rate model, and then the author analyses performance of these three methods in VaR computation. Empirical test implies that conditional heteroskedasticity model overestimate VaR; Generalized error distribution with Monte Carlo simulation is more accurate for computing VaR in SHI- BOR market left tail ; Generalized error distribution with term structure of interest rate is more accurate for computing VaR in SHIBOR market right tail.