This paper utilises “a class test for fractional integration” associated with the seminal contribution of
Hinich and Chong (2007) to appraise the possibility that South African Development Community
(SADC) real exchange rates can be treated as long memory processes. The justification for considering
fractional integration is that the general failure to reject the unit-root hypothesis in real exchange rates is
caused by the restrictiveness of standard unit-root tests regarding admissible low-frequency dynamic
behaviour. The paper presents evidence that a majority of SADC real exchange rates are fractionally
integrated and therefore mean-reverting.