摘要
金融时间序列的分布对于全面、准确把握金融资产收益的动态行为具有重要的意义,而广义自回归条件密度(GARCD)建模为描述金融资产收益的概率密度函数提供了一种工具。本文在JSU分布的基础上,建立了GARCD-JSU模型,给出了模型的参数估计方法及模型拟合效果的检验方法。利用建立的GARCD-JSU模型不仅可以得到金融时间序列的时变概率密度函数,而且还可以测算出时变高阶矩的变化,从而克服了正态分布假定框架下仅从前二阶矩出发考虑金融时间序列分布特征的局限性。
It is essential for describing the dynamic character of financial asset return comprehensively to utilize the distribution of financial time series. To our knowledge, generalized autoregressive conditional density model provides a useful tool for simulating the probability density function of financial asset return. Based on JSU distribution, the GARCD-JSU model has been proposed in the paper. Furthermore, the estimation and test method for the model are discussed in detail. The new model has not only been applied to simulating the time-varying probability density function of financial time series, but also been applied to measuring the timevarying higher moments. In the end, the limitations in describing the distribution of financial series from the first two moments under normal distribution have been overcome by our new model.
出处
《数量经济技术经济研究》
CSSCI
北大核心
2008年第8期137-150,共14页
Journal of Quantitative & Technological Economics
基金
国家自然科学基金项目(70471050)
中国博士后科学基金项目(20060400192)
全国统计科研计划重点项目(2006B07)的资助
关键词
广义自回归条件密度模型
JSU分布
极大似然估计
建模
Generalized Autoregressive Conditional Density Model
JSU Dis- tribution
Maximum Likelihood Estimation
Modeling
