杜勇宏   王健     

(南开大学经济学院，300071)

摘要：本文推导了当数据生成过程是独立的季节平稳过程情形下，OLS参数估计及检验统计量的极限分布。由于序列中的回归残差的自相关性会导致虚假回归现象的发生。借助Monte Carlo试验，对上述虚假回归中OLS统计量(t类统计量、R2、DW)的大样本渐近分布进行模拟，发现确实存在虚假回归现象并且受样本容量的影响不大。针对我国数据样本期比较短的特点，就虚假回归下统计量的小样本(T=10，15，30，50)特征进行模拟。

关键词：季节平稳过程，季节虚假回归，蒙特卡罗模拟 

中图分类号：F22

 

Spurious Regression Between of Independent Seasonal StationarySeries

Du Yonghong, Wang Jian

(Institute of Economics, Nankai University)

Abstract：In this paper, we deduced the limit distribution of regression coefficient and correspondence statistics when data generate process is independent seasonal  stationary series. We explored the possible existence of spurious relationships  between variables may be spurious because of correlation in residual. Using simulations to get the distribution characters of OLS statistics(t,R2,DW), it is showed that spurious regressions can still occur and that the strength of the relationship varies very little with sample size. In fact that seasonal time series in China are always not very long, this research is especially studied in small sample volume case, such as T=10,15,30, or 50.