Real Versus Spurious Long-Memory Volatility In Foreign Exchange Data: Evidence From The Rand Against The G4 Currencies

This paper investigates the extent to which volatility estimates and forecasts may be biased because of the existence of, and failure to account for structural change in the evolution of the exchange rate volatility series. The results of this analysis show that evidence of volatility persistence and long-memory in the Rand against the G4 currencies is overestimated when structural breaks are not considered. A modification of the standard GARCH model to allow for time-variation in the unconditional variance is shown to generate improved volatility forecasting performance over long(er)1 horizons for some currency pairs. Finally, the performance of the standard GARCH, modified GARCH and FIGARCH models are evaluated in the context of value-at- risk (VaR) estimation given the Basle regulatory framework. The results show that both the modified GARCH model and long-memory models generally deliver more accurate VaR measures relative to the standard GARCH model. In terms of VaR estimation our findings may provide guidance on more effective prudential standards for operational risk measurement and, as result, may help ensure adequate capitalisation and reduce the probability of financial distress. Our results highlight the importance of using out-of-sample forecasting techniques and the stipulated probability level for the identification of methods that minimise the occurrence of VaR exceptions. We find that models that account for structural change and long memory attributes generally outperform the basic GARCH model in estimating VaR across the probability levels we considered.