MODELLING SOUTH AFRICAN CURRENCY CRISES AS STRUCTURAL CHANGES IN THE VOLATILITY OF THE RAND

This study tests the theory that currency crises are associated with sudden large changes in the structure of foreign exchange market volatility. Due to increases in market uncertainty, crisis periods exhibit abnormally high levels of volatility. By studying short-term changes in volatility dynamics, it is possible to identify the start and end dates of crisis periods with a high degree of precision. We use the iterative cumulative sum of squares algorithm to detect multiple shifts in the volatility of rand returns between January 1994 and March 2009. Dummy variables controlling for the detected shifts in variance are incorporated in a generalised autoregressive conditional heteroscedasticity modelling framework. The analysis indicates that previously identified crisis periods in the rand coincide with significant structural changes in market volatility.