his article tests the hypothesis that price changes for a selected number of T commodity futures contracts are independent of previous price changes. The statistical procedure used to perform this test is called rescaled range analysis and is capable of identifying persistent or irregular cyclic dependence. It is the presence of persistent dependence that we choose to label "memory."
This investigation has been undertaken because there is mounting evidence that a speculative efficiency hypothesis is an inadequate description of auction market price behavior. In the context of commodity futures markets, a speculative efficiency hypothesis asserts that consecutive price changes, adjusted for trend, are in- dependent of one another.
Evidence of the absence of speculative efficiency in the foreign exchange market has been reported by Booth, Kaen, and Koveos (1981) in the form of profitable trading rules and nonrandom power spectrums. The trading rule result has recently been confirmed by Dooley and Shafer (1983). Bilson (1981) reaches a similar conclusion about an absence of speculative efficiency with respect to the behavior of forward and spot exchange rates.
Application of rescaled range (RIS) analysis to the foreign exchange market by Booth, Kaen, and Koveos (1982b); to the gold market by Booth, Kaen, and Koveos (1982a), and to common stocks listed on the New York Stock exchange by Greene and Fielitz (1977) also resulted in rejecting a hypothesis that the respective price changes were independent of previous price changes. Instead, the most frequent finding reported was that long term persistent dependence was present.
Early evidence and conjecture about the persistence of price movements in commodity future prices may be found in Alexander (1961);Cargill and Rausser (1975),