A GENERALIZATION OF THE BURRIDGE–GUERRE NONPARAMETRIC UNIT ROOT TEST

In this note the nonparametric unit root test of Burridge and Guerre (1996, Econometric Theory, 12, 705–723), which is based on the standardized number of crossings of a level of a random walk, is extended in two ways, allowing for a deterministic trend in the process and more general innovations. The test has a well-known standard limit distribution. Monte Carlo experiments revealed the good finite-sample properties of the proposed test.The authors appreciate helpful comments from an anonymous referee. We gratefully acknowledge the financial support of the Ministerio de Ciencia y Tecnología and the Conselleria d'Economia, Hisenda i Innovació, grants BEC2002-03769 and PRIB-2004-10095, respectively.