Conventional approaches based on integer orders of integration can not properly explain the slowly decaying volatility persistence which is common in most financial time series. To incorporate this type of volatility persistence, this study examines long memory processes which have slowly decaying but eventually dying out outocorrelation functions.
In this study, long memory property in volatility of KOSPI200 returns which are the underlying assets of options traded in the Korean stock market, was examined using FIGARCH model depending on fractional integration order. The robustness of results was also investigated by considering the effects of time and cross-sectional aggregation. Yielding different results from the aggregated data means that long memory properties may be affected by the choice of data set. To check these possible spurious results, we also examined weekly returns for time aggregation and individual 10 stock returns included in KOSPI200 for cross-sectional aggregation and compared them with the original KOSPI200 daily returns. This study has shown that the volatility process of KOSPI200 returns has long memory and can be explained with FIGARCH model.