Long Memory in Commodity Futures Volatility : A Wavelet Perspective

We reexamine the volatility of agricultural commodity futures for evidence of fractional integration, providing new empirical results and extending the extant literature in important dimensions. First, we utilize two relatively new estimators based on wavelets, which are generally superior to, for example, the popular GPH estimator and exact MLE estimators on the basis of mean squared error. Second, we provide simulations to contrast our point estimates with those obtained by a fractionally integrated GARCH model. Third, we conduct a wavelet coefficient decomposition of futures volatility. We find that futures volatilities display the self similarity property consistent with long memory, and we confirm that futures volatilities exhibit persistent long memory with finite unconditional variance.Please see paper published in J of Futures Markets for all figures and equations