The Effect of Linear Time Trends on Single Equation Cointegration Testing

Shin (1994) and McCabe, Leybourne and Shin (1997) introduced residual-based tests for the null hypothesis of cointegration; Boswijk (1994) and Banerjee, Dolado and Mestre (1998) suggested error-correction tests for the null hypothesis of NO cointegration. This paper supplements their work. They provide critical values for regressions with and without detrending. Here it is shown that the latter are not appropriate if the series display linear time trends. This does not mean that detrending is required. In this paper adequate percentiles are suggested for the case that series follow linear time trends but tests are based on regressions without detrending. They are readily available from the literature.