Measuring seasonalities in commodity markets and the half‐month effect

Abstract

ecently, an extensive amount of research has been devoted to the issue of emR pirical anomalies in financial markets. In time series data involving individual issues and/or market indices, seasonalities have been documented with respect to the turn ofthe year (Keim (1983), Roll (1983), Haugen and Lakonishok (1988)), the day o f the week (French (1980), Gibbons and Hess (1981), Keim and Stambaugh (1984)), the halfmonth’ period (Ariel, 1987), and the time ofthe day (Harris, 1986), among others. The consistent recurrence of these seasonalities raises questions about market rationality and market efficiency, and has led to new theoretical models to explain such anomalies (Rock (1989), Admati and Pfleiderer (1988)). However, even when attempts are made to reduce incidental noise from the data with the use of different and larger datasets (Lakonishok and Smidt, 1988), most of the regularities remain intact. While researchers in the stock and bond markets cannot fully explain the existence of all seasonalities on the basis of market rationality alone, the existence of seasonalities in the commodity markets is potentially explainable by the cyclic nature of production. Because agricultural commodities must follow their own crop cycle which repeats the same seasonal patterns year after year, observed commodity prices exhibit nonstationarities along the same seasonal lines. Crop cycle-related seasonalities in agricultural commodities are documented by Roll (1984), Anderson (1985), Milonas and Vora (1985), Kenyon et al. (1987), and Fama and French (1987). However, since commodities are traded in markets similar to the financial markets, it is an empirical question whether the seasonalities found in financial markets are also present in commodity markets.