This study examines the impact of exogenous changes in volatility
persistence using the GARCH model with and without shock dummies. For
this purpose, we considered five weekly KOSPI 200 sector index series.
Using the iterated cumulated sums of squares (ICSS) algorithm, we
determined the timing of volatility changes corresponding to major economic
and political events, including the 1997 Asian currency crisis, the Russia
crisis of 1998, the IT bubble of 2000, the 9/11 terror attack of 2001, the Iraq
war of 2003 and the global financial crisis that has been recently affecting
nations worldwide. After incorporating these volatility change, volatility
persistence in the GARCH model was significantly reduced. This result
implies that ignoring exogenous changes overestimates volatility persistence.
Thus, incorporating information on exogenous changes in conditional
variance will improve the accuracy of volatility forecasting.