基于偏态分布的风险度量计算

Computing of risk measurement based on skewed distributions

金融时间序列具有尖峰厚尾性,同时在股市中又存在着杠杆效应.对股票指数收盘价格的对数收益率序列建立ARMA-APARCH模型,在对数收益率序列分别满足Skewed-t分布和Skewed-GED的假设下,给出了在险价值及期望损失的计算方法.对t分布与Skewed-t分布、GED与Skewed-GED分别进行对比性实证分析,结果表明,在两个偏态分布假设下计算得到的期望损失估计结果更为保守,更能够捕捉到股市的尾部风险.

Financial time series have the sharp peak and fat-tailed characteristics and leverage in stock market. The ARMA-APARCH model is established based on logarithm yield ratio series of the stock index closing price and value-at-risk （VaR） and expected shortfall （ES） comput provided in the assumption of the sequence of logarithm yield ratio series satisfying th ing e di methods are stributions of Skewed-t and Skewed-GED respectively. Having compared t distribution with Skewed-t distribution, GED and Skewed-GED, it is proved that the ES estimations considering asymmetrical distribution are more conservative and more efficient to capture the tail risk of stock market.