Abstract:
This paper introduces a new unit root test, using "Feedback Covariates", in order to test for sustainability of fiscal policy within the G-7.Non-sustainability is considered as the null hypothesis of both a unit root in the debt process and the nullity of a deficit correction in response to inherited debt. We characterize the distribution of the FADF (Feedback Augmented Dickey-Fuller) statistics under both the null and the near unit root alternative and we compare the FADF statistics with the ones related to single equation approaches. This test is applied to public debt normalized by GDP, rather than by a discount factor, consistently with a definition of effective sustainability, which focusses on policy implications. We are still unable to reject the null hypothesis of non-sustainability for four countries. Nevertheless, our results confirm the potential of our approach in terms of power gains aind show that standard unit root tests lead to accept too often the null hypothesis of non-stationnarity.