Re-examining Return Autocorrelation and Monday Returns

Quarterly Journal of Finance and Accounting, Summer 2003 by Tori, Cynthia Royal

This study examines conditional Monday returns in light of the recent evidence of a reversal of the Monday effect. Using S&P 500 and Nasdaq market returns from January 3, 1970 to December 31, 2001, the study finds that the reversal of the Monday effect for the S&P and the disappearance of a significant Monday effect for the Nasdaq cannot be attributed to the correlation of the Monday return with the previous day return. Shifts do occur in this correlation, but even accounting for good news/bad news markets and the week-of-the-month timing, these shifts do not explain shifts in the average Monday return. In the Nasdaq market, where Monday returns remain negative, daily return correlation has disappeared completely, suggesting that the wait-analyze-invesl argument postulated by information-processing hypothesis is no longer supported.

Introduction

The previous articles in this special issue have examined various aspects of the Monday effect. Pettengill and Wingender (2003) provide an introduction to the Monday effect topic. Two papers examine Monday seasonal effects, one in relation to the individual investor (Gondhelekar and Mehdian, 2003) and the other in relation to market breath (Liano and Sullivan, 2003). Wingender and Singleton (2003) study the macro announcement effect on Monday returns, and Miller, Prathcr, and Mazumder (2003) suggest an operational strategy to capitalize on market return patterns using mutual funds and variable annuities.

As discussed by Pettengill and Wingender (2003), an additional area of research has been the examination of various Monday conditional relationships. The conditional Monday returns studies find that previous day returns are positively correlated to Monday returns (Bessembinder and Hertzel, 1993; Abraham and Ikenberry, 1994; Wang, Li, and Erickson, 1997; Chow, Hsiao, and SoIt, 1997; Jaffe and Westerfield, 1989; and Tong, 2000), a bad news market environment is significantly correlated to the Monday effect (Penman, 1987; Jaffe and Westerfield, 1985; Fishe, Gosnell, and Lasser, 1993; and Tong, 2000), and Monday returns during the last half of the month are significantly negative while Monday returns during the first half of the month are not significantly different from zero (Wang, Li, and Erickson, 1997; Mehdian and Perry, 2001).

After three decades of Monday effect research and publicity, recent studies find that US large-cap market Monday returns are no longer negative but are positive and significantly higher than the other days of the week (Kamara, 1997; Wang, Li, and Erickson, 1997; Brusa, Liu, and Schulman, 2000; and Mehdian and Perry, 2001). In light of the instability of the Monday effect, this study tests whether the previous day returns, a bad news market, and week-of-the-month conditional relationships have changed or reversed as the Monday effect reversed.

This paper extends the Monday effect literature by reexamining conditional Monday returns in light of the shift in mean Monday returns. The remainder of this paper reviews the literature and discusses the findings of this study.

Literature Review

Researchers have examined several conditional relationships in an effort to better understand the Monday effect anomaly. A widely-documented conditional relationship is the positive correlation between the previous trading day return and the Monday return. Using US equity data from 1885 to 1989, Bessembinder and Hertzel (1993) find a strong correlation between the previous trading day returns and Monday returns. They show that Monday returns are more highly correlated with previous day returns than are returns of the other days. After examining the relative correlation between returns and previous day returns for all trading days, they conclude that the pronounced asymmetry of information and the investment strategies of informed investors lead to the Monday effect and a partial reversal of the Monday effect on Tuesdays.

Using US equity data from 1963 to 1991, Abraham and Ikenberry (1994) find a positive correlation between previous trading day returns and Monday returns and discover that when the Friday market return is negative, the Monday market return is negative nearly 80 percent of the time. They suggest that the Monday effect conditional upon previous day returns arises from individual investor behavior or what they refer to as the information-processing hypothesis. Over the weekend, individual investors have more time to gather and analyze market information. If investors base their buy/sell decisions upon information gathered and weekends offer a low-cost opportunity to gather information, then individual investors are more likely to be active following a non-trading weekend. Lakonishok and Maberly (1990) find that individual investors tend to trade more on Mondays and tend to offer more sell than buy orders on Mondays. Both studies suggest that brokers tend to advise buy recommendations, are not likely to solicit sell orders, and have little contact with individual investors over the weekend. Therefore, individuals tend to postpone sell orders to trading days following a weekend and execute buy orders during the week. Further, because institutional orders tend to be lower on Mondays than on other days of the week, their argument suggests that individual investors drive the Monday effect.