1 992), hereafter PMR, analyze the statistical properties of cash prices for 17 commodities. Their results are consistent with other studies which have found autocorrelation in commodity prices. For a literature review, see Tomek and Myers (1993). This author fears, however, that the introductory and concluding remarks of the PMR article may mislead those readers who are unfamiliar with commodity markets. This note is intended to clarify the characteristics of commodity price behavior and the implications of these characteristics for the conclusions drawn. Agricultural markets are emphasized here, but most of the comments are equally applicable to other commodities. The important point is that cash prices for commodities should not be expected to follow a random walk. Thus, this model is not a useful benchmark for measuring the efficiency of cash markets for commodities.
COMMODITY MARKETS
Looking at graphs of commodity price movements over time, one is immediately struck by the positive autocorrelation. Formal timeseries analyses of daily data confirm that prices are indeed highly autocorrelated and are not normally distributed. The variances of Scott H. Inhin provided helpful comments on an early draft, which I gratefully acknowledge.