The stylized approach to unit root testing: neglected contributions and the cost of simplicity

Following Dickey & Fuller (1979) (DF), a stylized approach to the testing of the unit root hypothesis has emerged. Based upon the combined use of the DF test in its augmented t -ratio form and MacKinnon (1991) critical values, the approach has received widespread adoption due to the ease with which it can be applied. In this paper a number of departures from this "stylized approach', which do not significantly reduce its ease of application, are examined. The results obtained from an empirical application to UK industrial production and Monte Carlo experimentation have clear methodological implications, showing that routine application of the stylized approach can lead to misleading inferences concerning the integrated nature of economic time series.