he existence of day-of-the-week effects in growth rates of real goods T prices has been documented. We find that, prior to 1981, almost all commodity price increments were concentrated on Friday. Moreover, similar to findings in common stock markets, the "blue Monday'' effect is confirmed in the Dow Jones Spot Commodity Index. Thus, the unknown thrust underlying the negative weekend effect appears to prevail in both financial and real assets markets.
Many studies document reliably the day-of-the-week effects in stock returns. Most notably, the average return for Monday is negative and is less than that for the other four weekdays. Gibbons and Hess report qualitatively the same results for treasurybills. Dyl and Maberly (1986) also find that the anomalous results are persistent in the market for Standard & Poor's 500 Stock Index Futures. Although documentation of this "weekend" effect is extensive, researchers have been unable to identify its cause.'
Recently, Chiang and Tapley (1983) extended the investigation of day-ofthe-week effects to futures markets. To motivate their study, Chiang and Tapley argued that past studies have analyzed only those financial securities that represent indirect claims on real assets. In contrast, most commodity futures contracts can be considered to be direct claims on the real asset itself. Therefore, the similarity and/or dissimilarity in the behavior of the two types of claims could help determine whether past results have arisen from the institutional peculiarities of indirect financial claims. Chiang and Tapley examined daily price change patterns of 21 commodity contracts. They found that the results for grains, as a matter of fact for the whole market, are similar to the results for common stock; that is, the returns peak on Wednesday, have
We have greatly benefited from the comments of two referees of this journal. 'Explanations for the observed weekend effect that have been examined and rejected include the delay between trading and settlement due to check clearing (Lakonishok and Levi, 1982), measurement error in returns , and specialist-related biases (Keim and Stambaugh, 1984).