Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
This study examines some finite-sample properties of a new modified Dickey-Fuller test, called the DF-GLS test, which has been shown to be more powerful than standard unit-root tests. The study shows that the lag order can significantly affect the critical values of the test. This points to the importance of correcting for the lag order effect in implementing the DF-GLS test. Approximate lag-adjusted, finite-sample critical values for the DF-GLS test are provided, which can be straightforwardly computed from response surface equations. Copyright 1995 by Blackwell Publishing Ltd