金融资产的收益分布普遍展现出两个重要的典型特征:“有偏”性和“胖尾”性,但目前绝大多数的随机波动模型都无法同时将上述两类典型特征综合纳入其估计的条件分布假定中。以上证综指和标准普尔500指数为例,通过引入有偏的广义误差分布(Skew GED)来综合刻画条件收益的有偏和胖尾特性,并运用马尔科夫链-蒙特卡罗模拟法(MCMC),探讨了在正态分布(Normal)、有偏的正态分布(SkewNormal)、广义误差分布(GED)以及SkewGED这4种条件分布假定下的随机波动模型估计方法,同时实证检验了不同分布假定下的随机波动模型对实际市场波动率的刻画精度和适用范围。
Though two important stylized facts about return distribution are seen commonly in financial markets: skewness and fat-tail, most of the stochastic volatility models at present cannot describe those facts as a whole. In this paper, taking SSEC and S&P500 indices as example, we use skew GED to depict skewed and fat-tailed characteristics of the return distributions. By MCMC, the estimation methods of SV models based on normal, skew normal, GED and skew GED distributions are discussed. The performances for volatility descriptions of SV models on different distributions are also tested.