作者: Jeroen J. M. Kremers , Neil R. Ericsson , Juan J. Dolado
DOI: 10.1111/J.1468-0084.1992.TB00005.X
关键词:
摘要: A cointegration test statistic based upon estimation of an error-correction model can be approximately normally distributed when no is present. By contrast, the equivalent Dickey-Fuller applied to residuals from a static relationship has nonstandard asymptotic distribution. When exists, generally more powerful than test. These differences arise because latter imposes possibly invalid common factor restriction. The issue general and ramifications for system-based tests. Monte Carlo analysis empirical study U.K. money demand demonstrate in power. Copyright 1992 by Blackwell Publishing Ltd