含非线性的平稳变量之间的虚假回归研究

Spurious Regression Between Nonlinear Stationary Variables

研究含非线性的平稳变量之间的虚假回归问题。通过推导OLS估计的收敛性、t统计量和R2的极限分布,证明含非线性的平稳变量之间会出现虚假回归现象,除非回归模型能精确地捕捉变量中的非线性。蒙特卡洛模拟的证据与推导出的理论相符。研究表明:在经济分析中,甄别和正确地处理变量中的非线性部分是十分重要的。

This paper studies the spurious regression problem between nonlinear stationary variables. By deriving the convergence of OI.S estimates, the limiting distribution of t statistics and , we show that there exists spurious regression phenomenon between nonlinear stationary variables, unless the nonlinearities can be captured accurately. The Monte Carlo simulations support our theory. This paper indicates that it is very crucial to test and model the nonlinearities in econometric analysis.