Investigating structural break-GARCH-based unit root test in US exchange rates

Abstract

This paper applied a structural break-GARCH-based unit root test in studying the US exchange rates for twenty-two different countries across America, Europe, Asia-Pacific and Southern Africa. The study employed three different data frequencies – daily, weekly and monthly with a view to understand the dynamics of high frequency series that are characterized by alternating trend patterns and plausible presence of structural breaks. The chosen sample interval included periods of financial crisis or peculiar events. The exchange rates were found to exhibit ARCH effects at higher lags, thus informing the adaptation of the more parsimonious GARCH process in the residuals in contrast to the white noise disturbance assumption. The non-trended and trended structural break-GARCH-based unit root tests’ performances were adjudged with other existing tests. With significant break points, from 2 to 5, the presence or otherwise of a unit root in foreign exchange rate series would be better captured when the inherent heteroscedasticity, trend and structural breaks in the series are put into consideration.