样本数据正态性转换时变VaR

虽然摒弃估测VaR值的正态分布假设已成为当今金融学研究的一个重要方面及其热点问题，但利用正态性转换先处理样本数据，然后再利用基于正态分布假定下的VaR估测方法来修正VaR的估值则是精准化VaR计算的一条有效途径。本文试图在考虑收益率分布时变性特征的情况下，利用正态性转换处理样本数据这一思想来细化和改善我国证券市场不同运行阶段的VaR估测。相应的经验分析表明：我国证券市场收益率分布不仅存在时变性特征，而且牛熊市期间的VaR值存在显著的差异。Johnson转换函数之SU型适宜作为样本数据正态性转换函数，经转换样本数据估测的VaR值既提高了VaR估值的准确性，又改善了VaR估测精细化的有效性。

VaR of Time-varying about Normality Transition of Sample Data

Although abandoning the normal distribution assumption when estimating VaR is becoming an important aspect and hot issues in finance field that,it is an effective way to make more accurate VaR calculation that dealing with the sample data by normality transition and estimateing VaR based on normal distribution assumption.This article tries to consider time-varying characteristic of return distributions and make use of the idea that using normality transition to deal with sample data to detail and improve VaR estimation during different stages of Chinese securities market.Corresponding experiential analysis shows that return distributions of securities market in China not only has time-varying characteristic,but also shows notable differences in VaR between in cattle market and in bear market.SU transition function of Johnson transition functions is suitable for normality transition of sample data,VaR of sample data by SU transition function improves the accuracy and effectiveness of VaR estimation.