Thomas Schneeweis* ver the last ten years, several academic studies have identified a statistical-0 ly significant weekly pattern in stock returns.' Similar effects have been found in stock index futures.* The results of these studies show that the average return of stock indexes and stock index futures over the weekend (e.g., Fridayclose to Monday-close) appears to be systematically negative. In addition, the weekend return is on average less than other weekday returns. Various explanations for the "weekend effect" have been suggested, such as specialist-related biases (Keim and Stambaugh, 1984), measurement errors and firm size (Gibbon and Hess, 1981), and settlement procedures (Lakonishok and Levi, 1982).
Most of the previous studies on stock indexes and stock index futures have employed equal-and/or value-weighted indexes with the return being calculated as ((PJP, -,) -1.0) or ln(P,/P, -,). For stock indexes, the use of price relatives as a return calculation assumes that the dividend component of the total return is insignificant or is distributed equally across weekdays. It is a well known fact, however, that security prices drop on ex-dividend days to reflect the dividend ~a y m e n t . ~ If ex-dividend dates are systematically clustered on Mondays, then prices on Mondays would be expected to fall relatively more than on other weekdays. As a result, if Mondays experience significantly higher ex-dividend dates than other weekdays, the index returns calculated simply as price relatives will tend to be smaller over weekends than on the rest of the weekdays. In brief, if dividends exhibit a daily pattern, the distribution of dividends may contribute a partial explanation to the observed "weekend effect" in stock indexes. Likewise, for stock index futures, the use of price *The authors thank Uttama Savanayana, Ph.D. candidate, University of Massachusetts/Amherst, for his extensive help in preparing this paper.
ISee, for example, Cornell (1985), French (1980), Gibbons and Hess (1981), Jaffe and Westerfield (1985), Keim and Stambaugh (1984), Lakonishok and Levi (1982), Rogalski (1984), and Harris (1986). French (1980) mentions in a footnote the potential bias that might result from improperly adjusting for dividends, but does not adjust for them.
3ee Dyl and Maberly (1986a,b) andJunkus (1986) for studies of the weekend effect in stock index futures. 'Some studies ex-dividend day stock price behavior have documented a positive abnormal return on the ex-dividend day. Studies of weekly patterns in stock returns which adjust returns for dividends can, therefore, be biased upwards if stocks tend to go ex-dividend on Mondays.