Abstract:
In this paper, we extend the existing literature on current account sustainability by examining the relevance of long memory and structural breaks in modelling the dynamics of current account to gross domestic product (GDP) ratios in G7 and BRICS. Unlike standard unit root tests, which have low power, especially in cases where the series is characterized by a fractional process, the long-memory approach provides an exact measure of the degree of persistence. However, long-memory models are known to overestimate the degree of persistence of the series in the presence of structural breaks. We show that regime changes do exist in both the mean and trend of the current account to GDP ratios. Thus, we test persistence allowing for both smooth and sharp breaks. Our methodology also allows any number of sharp breaks, whereas standard unit root tests only permit either one or two breaks. Hence, our approach is more general and more robust to misspecifications caused by the omission of breaks than standard methods. We show that current accounts are sustainable in both groups of countries, with the G7 and South Africa displaying long-memory behaviour.
