摘要
给出确定经济系统非线性相关的两个判据 :(a)关联积分与关联维数 ;(b)BDS统计 .提出非线性ARCH(自回归条件异方差 )预警方法和具有ARCH特征的警限界定算法 ,并对我国宏观经济的统计数据序列作了实证分析 .结果表明 :预测的可靠性和拟合优度好 ;ARCH预警方法引入时变条件方差使预报的置信区间能够与经济时间序列的波动程度相适应 ,从而使预警限更准确地反映实际状况 .
In this paper, two criteria are given for nonlinear correlation of economic systems, which are correlation integral and correlation dimension and BDS statistics. A nonlinear early warning method is proposed based on the ARCH model. Finally empirical results for China are presented.The results indicate the good reliability of forecast and goodness of fit.Time varying conditional variance introduced by the early warning method based on ARCH can make the confidence interval of forecast conform to the fluctuation degree of economy time series,thus the warning limit can reflect actual situations more precisely.
出处
《河海大学学报（自然科学版）》
CAS
CSCD
2000年第5期59-63,共5页
Journal of Hohai University(Natural Sciences)
关键词
经济预警
非线性系统
自回归条件异方差
economy early warning
nonlinear system
autoregressive conditional hetereskedasticity
