Potential Sources of the Long Memory Property in the Volatility Process of Daily KRW-USD Exchange Rates : Jumps and Structural Breaks

This paper analyzes the potential sources driving the long memory property in the volatility process of the daily Korean Won (KRW)-US Dollar (USD) exchange rates. Based on the Adaptive FIGARCH model and the jump diffusion model, this paper proposes a new specification of the volatility process, which links the long memory property to jumps and structural breaksall together rather than separately as in previous studies. The model is estimated with the daily data on the KRW-USD exchange rate returns between 1999 and 2007. The empirical results show strong evidence that the Adaptive FIGARCH model combined with the Bernoulli jump process which takes into account the jumps and the structural breaks jointly could explain almost all of the long memory property in the volatility of the KRW-USD exchange returns. The implication is that both the structural breaks and the jumps are the major sources of the long memory property in the volatility.