High Frequency Analysis on Jumps and Long Memory Volatility in Commodity Futures Prices

This paper concerns the high frequency returns of 15 minute commodity futures prices. The FIGARCH model of Baillie etal. (1996) with the usual normality assumption is found to be inappropriate in representing the high frequency commodity futures returns and the rejection of the normal distribution appears to be due to the jumps which are occurred in the high frequency futures returns. Hence, this paper relies on the FIGARCH model combined with the Bernoulli distribution that allows for the jumps. This paper shows that the FIGARCH-Bernoulli distribution model performs quite well in representing the high frequency commodity returns and that the jumps spuriously increase the long memory persistence in the volatility process of the high frequency commodity futures returns.