摘要
如何构建合适的模型以恰当的方法对风险进行测量是当前金融研究领域的一个热门话题。VaR方法作为当前业内比较流行的测量金融风险的方法,具有简洁、明了的特点,而且相对于方差来讲,更多的将投资人的损失作为风险具有更好的合理性。但是在用参数法计算VaR时,关于分布假设、模型的选择,具有一定的主观因素,很多是靠经验来进行判断。文中用当前金融领域刻画条件方差最典型的GARCH模型及其几种最新衍生模型如EGARCH、PARCH等,分别在正态分布以及能刻画收益厚尾特性的分布(t-分布、GED分布)假设下,再结合上证指数、深圳综指与上证180指数进行实证分析,并对结果作了比较,分析各种模型的优缺点、与分布假设的关系以及与指数序列数据的关系,得出一些新的结论。作为对比,也用Riskmetrics模型计算了VaR值,分析其与GARCH族模型的不同。
It is a hot topic how to construct a the suitable model to measure financial risk in financial research field at present. VaR is a popular method to compute finance risk, which is simple, clear and more reasonable in contrast with variance because it takes the loss of investors as the risk. But when we calculate VaR by using parameter method, distribution hypothesis and the selection of models is somewhat subjectivity and sometimes depended on experience.This paper analyses GARCH family such as GARCH,EGARCH,PARCH, and computes VaR about Index of Shanghai Stock Exc-hange, Shenzhen in normal distribution separately (and t-distribute, GED distribute). In order to compare, the paper also computes VaR in Riskmetrics model and analyse the difference with GARCH.
出处
《数量经济技术经济研究》
CSSCI
北大核心
2005年第7期67-81,133,共16页
Journal of Quantitative & Technological Economics