Fractional differencing in discrete time

This paper consists of two parts, a theoretical followed by an empirical contribution. We first give a new framework for fractional differencing in discrete time and show how the definition of fractional differencing that is commonly employed in empirical financial applications arises as a special case. We then use these methods to estimate the fractional differencing parameter in the return and volatility for three Comex metal futures contracts. The metal futures are sampled at very high frequencies—five-minute intervals over a nearly eight year period.