为更好刻画金融资产收益率偏态厚尾特性,提高VaR风险度量精度。本文首先提出利用广义双曲线(GH)分布对收益率数据进行建模型,从分布尾部特性角度对GH分布和其他常用分布进行了比较研究;其次利用EM算法来解决含有Bessel函数的GH分布的参数估计难问题,并运用随机模拟方法计算VaR值;最后讨论GH分布在我国股票市场VaR风险度量中的应用。
In order to better models the characteristics of skewness and fatness of financial asset returns, and measures the VaR value more accurately. This paper proposes to use generalized hyproblic (GH) distribution to model the returns, and compares the GH distribution with other distributions from the perpestive of tail characteristics; then uses EM algorithm to solve the problem of parameter estimation, also uses stochastic simulation method to calculate VaR; lastly, applies the GH distribution to Shanghai stock market.