中国宏观经济统计数据异常性和波动性特征的计量检验：1953-2001

The Econometric Analyses on Outliers and Volatilities in Chinese Macroeconomic Data Series： 1953-2001

对于宏观经济统计数据的异常性和波动性进行分析,已成为研究数据质量的最核心内容之一。本文从经济系统的角度运用随机方差扩大模型对我国36个宏观经济时间序列的数据质量进行了全面分析,发现了数据异常及波动的特点和规律。研究结论表明,大部分异常点的出现或多或少都是以聚集成堆的形式出现,它们之间有深刻的内在联系,异常点的出现大多与各种历史因素以及外部冲击有关;几乎所有的原始序列都有显著的偏度,过多的峰度也是明显的,因此它们被显著地拒绝认为服从正态分布;名义序列的特征在更大程度上受到异常点的影响。

The studies on outliers and volatilities have become one of the core contents on macroeconomic data quality. This paper uses RVAR model to deeply analyse 36 Chinese macroeconomic data series from the view of economic system and finds out the features and rules of the outliers. The results show： the most of outliers appear in the cluster forms more or less, and there are deep internal relationships between them, and the outliers are mostly caused by the historic factors and external shocks. Almost all of the original series have obvious skewness and many kurtosis are appear. So it is obviously rejected that they obey normal distributions. Nominal series are effected by the outliers to some more extent.