Identifying nonlinearities in current account sustainability: Evidence from nonlinear panel unit root tests

The traditional linear panel unit root tests may have low power when the data generating process (DGP) exhibits nonlinearity. The current account imbalance (CA) as a percentage of gross domestic product (GDP) is an essential indicator of the individual countries’ economic performance. The mean reverting behavior of the current account imbalance (CA) as a percentage of the GDP series to the equilibrium, like many other economic variables, may display different types of nonlinearities in the long run. Therefore, it is essential to determine the possible nonlinearity in the aforementioned series while testing the sustainability of current account imbalances of a country. In order to account for the existence of nonlinearity, this study employs several nonlinear panel unit root tests, most of which are newly developed. Furthermore, these nonlinear panel unit root tests are classified as a) time-varying nonlinearity in deterministic components, namely structural breaks; b) state dependent nonlinearity; and finally c) hybrid nonlinear panel unit root tests that allow for structural break(s) and state dependent nonlinearities simultaneously. By doing so, the source of nonlinearity can be identified in addition to the existence of the possible nonlinearity itself. Hence, by applying those newly proposed panel unit root tests, we exploit both the cross-sectional and time series information available in the series to evaluate current account sustainability, while still allowing for the potential for structural breaks and state dependent nonlinearity. This study uses quarterly observations of the CA as a percentage of GDP for Brazil, Russia, India, China and South Africa (BRICS) countries. The sample period is 1998:Q1- 2017:Q1. All data are obtained from Datastream. The expected result is that the CA as a percentage of GDP series of this country group exhibits state dependent nonlinearity, especially threshold autoregression (TAR) type nonlinearity. That is, the sign of the disequilibrium may have an essential role in the adjustment process to the equilibrium of the aforementioned series. This expected result indicates that the state dependent nonlinearity is important in testing the nonstationarity in the null hypothesis for this country group.