Description

This paper considers the economic implications of having a unit root (UR) on the stochastic process of variables such as consumption or GDP. Using a variety of mod els, we develop indirect tests for unit roots based on sharp distinctions that should arise when the scale variable is di ff erence stationary (DS) or trend stationary (TS). We show that these tests do not feature the undesirable size-power trade-o ff that characterizes traditional UR tests and apply them to a range of countries.