Stock Returns and the Weekend Effect in Canada

Article excerpt

Stock Returns And The Weekend Effect In Canada

The existence of the weekend effect has been proven through different studies by Cross [4], French [7], Gibbons and Hess [8], Coats [2], Lakonishok and Levi [16], Hindmarch, Jentsch, and Drew [11], Keim and Stambaugh [15], Jaffe and Westerfield [12, 13], Jenkins [14], Dyl and Maberly [5, 6], and Chamberlain, Cheung and Kwan [1]. French [7] proved the existence of the weekend effect by testing both the trading and the calendar time hypotheses. Gibbons and Hess [8] verified the negative near return for Monday. Lakonishok and Levi [16] asserted that the weekend effect was primarily explained by settlement procedures and check clearing delays. However, Dyl and Maberly [5, 6] did not support Gibbons' and Hess's hypothesis and stated that settlement procedures did not affect the weekly pattern of returns. Keim and Stambaugh [15] supported French's findings where the negative average Monday returns were statistically significant. Jaffe and Westerfield [12, 13], after examining the stock returns in four different countries, found that the weekend effect did exist in each of them. They also investigated the settlement procedures and denied that these explained the negative Monday return or the positive Friday returns. Chamberlain, Cheung and Kwan [1] confirmed that Monday mean returns were not only negative, but they were also significantly different from mean returns for the other days of the week. Other U.S. studies that attempted to find the causes of the weekend effect phenomenon, but with no meaningful success, were done by Rogalski [18], Cornell [3], Harris [9, 10], and Smirlock and Starks [19]. Also, Loo [17] tried that in Canada.

The purpose of this paper is to determine the existence of the weekend effect in Canadian markets by studying this phenomenon over an extended period of time from the beginning of 1968 to March 1987. Both the trading time and the calendar time hypotheses would be tested. The trading time hypothesis states that stock returns are generated only during active trading, and, therefore, the expected returns should be the same for any given trading day. This hypothesis was tested and supported by French [7], Hindmarch, Jentsch, and Drew [11], and Keim and Stambaugh [15]. The calendar time hypothesis assumes that Monday returns represent a three-calendar-day investment while returns for other days of the week reflect a one-day investment. Therefore, Monday returns would be three times the mean return for any other day of the week. This hypothesis has been rejected by French [7], Lakonishok and Levi [16], and Hindmarch, Jentsch, and Drew [11]. Previous Canadian studies tested the weekend effect during shorter periods of time. Chamberlain, Cheung and Kwan [1] used 1976-1983 as the study test period. They also emphasized the close-to-open and open-to-close returns rather than close-to-close returns; the latter approach was generally the one used in most of the U.S. studies. Hindmarch, Jentsch, and Drew [11] used 1977-1981 as their test period and attempted to test a modified calendar time hypothesis in order to discover the actual causes of the weekend effect. They were looking for a possible relationship between institutional practices and the weekend effect. Also, Steve Loo [17] tried to develop a possible explanation of the weekend effect in order to develop profitable trading rules.

The present study will focus on testing both the trading time and the calendar time hypotheses during a fairly long period of time, from January 1968 to March 1987. The test will also be extended to different subperiods: 1968-1971, 1973-1975, 1976-1981, and 1982-Mid-1987. The reason for using several subperiods to test the existence of the weekend effect is to detect whether such phenomena did prevail in the past as well as the present time with the same magnitude or whether it is merely something that has evolved primarily during recent years. …