The weekend effect is one of the oldest existing stock market anomalies, given that it has existed for many decades. In this section, we define the anomaly, present the evidence and discuss its persistence.
2Basu has shown that low P/E stocks tend to outperform both the market and high P/E stocks.
InWhat Works on Wall Street, O’Shaughnessy found that the P/E ratio is particularly relevant for large stocks. However, he argued that the price-to-sales ratio is an even better indicator of excessive returns. Fama and French find that market and size factors in earnings help explain the P/E ratio effect. See O’Shaughnessy (1998, p. 16), Basu (1977) and Fama and French (1995).
3.2.1 Description
The weekend effect can be defined as a Friday’s return minus the following Monday’s return for a single security or a portfolio of securities.3 Under normal conditions, there should be no substantial difference between each day of the week through a long time span. However, in 1980, French analyzed daily returns of stocks for the period 1953–1977 and found that there is a tendency for returns to be negative on Mondays whereas they are positive on the other days of the week.4 He writes that these negative returns are“caused only by the weekend effect and not by a general closed-market effect.”A general closed-market effect would mean that most of the market agents would sell on Friday and buy on Monday. However, it seems that abnormal positive stock returns occur on Friday and the contrary happens on Mondays. So buying regularly on Monday and selling on Friday may bring abnormal returns.
3.2.2 Evidence
In Table3.1, all the common shares traded on the NYSE, the AMEX and NASDAQ have been regrouped and divided into two categories. The first category represents the stocks traded on the equally-weighted (EW) index, i.e., each stock has the same weight in the index. The second category includes the stocks traded on the value-weighted (VW) index where the weight for each share is proportional to the capitalization of the company relative to the capitalization of the whole traded index.
The mean daily returns in percent of these shares have been computed for the period starting from July 1962 to July 2001.
The weekend effect reaches 0.34 % if and only if the index is equally-weighted.
The difference of returns among the weekdays from July 1962 to July 2001 can be seen at a glance in the chart in Fig.3.1.
It is evident that there is a performance shift between Fridays and Mondays which leads to abnormal returns. It is important to note that the difference in returns through the week is fairly low, namely 0.33 %, yet considerable in comparison with other week days. Researchers also found that the weekend effect changed over time to such a point that during the period 1990–2001 there were negative returns in an equally-weighted index. They also observed that the effect was strengthened during long weekends while it was weakened when investors could trade on Saturdays.5
The discovery of the weekend effect lead Kamara to study the S&P 500 from 1962 to 1993.6He found no significant Monday effect after April 1982 except for a portfolio of smaller U.S. stocks where the Monday effect remained undiminished
3Chen and Singal (2003, p. 80).
4French (1980, p. 56).
5Singal (2006, p. 45).
6Kamara (1997).
Table 3.1 Daily average return in percent of common stocks on the NYSE, AMEX and NASDAQ from July 1962 to July 2001
Weekend Monday Tuesday Wednesday Thursday Friday (Fri.–Mon.) Overall period
1962–2001
EW 0.093 0.000 0.133 0.125 0.246 0.339
VW 0.055 0.044 0.099 0.047 0.098 0.153
By decades 1962–1970
EW 0.105 0.008 0.176 0.074 0.218 0.326 VW 0.124 0.022 0.145 0.028 0.131 0.255 1971–1980
EW 0.082 0.019 0.112 0.115 0.245 0.327
VW 0.100 0.035 0.098 0.049 0.111 0.211
1981–1990
EW 0.173 0.038 0.108 0.123 0.231 0.403
VW 0.078 0.062 0.112 0.049 0.109 0.187
1991–2001
EW 0.021 0.056 0.144 0.174 0.283 0.304
VW 0.063 0.053 0.057 0.057 0.050 0.013
Source: Singal (2006, p. 42)
Fig. 3.1 Weekday returns in percent for an equally-weighted index over the 1962–2001 period.
Source: Singal (2006, Fig. 3.1, p. 44)
until 1993. Apparently, the weekend effect was tightly linked to small capitalization and has a higher impact on equally-weighted indices than on value-weighted indices.
The following research focused on the international scale. Did the weekend effect occur outside U.S. markets? In 1994, Agrawal and Tandon found significant negative returns on Mondays in nine countries and on Tuesdays in eight countries, while large and positive returns could be observed on Fridays in 17 of the 18 countries studied.7Unfortunately, their data did not extend beyond 1987.
A quite interesting study was conducted in 2001: Steeley noticed that the weekend effect in the U.K. had disappeared in the 1990s,8while in 2010 Benjamin Liu and Bin Li could still note its existence on the Australian stock exchange, yet with varying strength among shares and industries.9
3.2.3 Explanations
It was only in 2003 that Chen and Singal provided a satisfying explanation for this long-lasting anomaly which seemed to curiously disappear in the 1990s in a few developed countries. The weekend effect can be mostly attributed to short sellers.10 There are two kinds of short selling: hedging and speculation. In the first case, short selling is used to offset an existing position, resulting in a neutral position. On the other hand, speculative short selling consists in selling an asset without having it, and then re-buying it later. The difference generates the profit or loss. The key aspect is to sell something before owning it.
To do so, investors borrow the asset and keep it as long as the initial owner does not want it back. This speculative strategy is quite risky because if the initial owner reclaims his asset at a time when the price is not beneficial to the short seller, he has to buy back the sold asset so as to then give it back to the initial owner. He may thus incur a loss. This happened before organized derivatives markets existed, and happens also now. The second drawback of short selling is that the expected loss is virtually unlimited. This is even more true for futures and options, which are financial products whose prices and returns are derived from assets like stocks. For example, if a stock costs $20, and drops to $19, the loss is not overwhelming. But the loss for the derivative may amount to, say, $1;000. Moreover, while the maximum loss of the share is $20(the share has then no value), the loss on the derivatives market could reach $20;000, while only $2;000were initially invested.
After these remarks on short selling, we can focus on Chen and Singal’s explanations. Short sellers go for a quick profit and prefer not to hold their position too long, because of the high volatility and the cost of borrowing. Taking a long
7Agrawal and Tandon (1994, p. 101)
8Steeley (2001).
9Liu and Li (2010).
10Chen and Singal (2003).
and a short position, namely in derivatives, is costly. If these operations happen too frequently the profit suffers. A single day strategy would be too short for short sellers, as a daily renewal of their positions would be too costly. However, weekends represent close to 50 h without trading. They are a landmark for buying back what was short sold at the beginning of the week. A buy order on a stock is then executed on a Friday to settle the short sell order of a Monday. An increase in asset prices is then observed on stock exchanges on Fridays and a decrease is observed on Mondays. The same effect occurs before and after holidays.11
Since options are easily available and less expensive for large stocks, the weekend effect for the value-weighted index began to disappear in recent years.
Researchers noticed these changes after the introduction of organized option markets. On the other hand, for an average stock, options are either non-existent or too expensive to trade. That is why the weekend effect for the equally-weighted index remained fairly unchanged through time.12Furthermore, the weekend effect tends to be stronger if institutional investors engage in the market.13
Additional explanations emerged from other researchers:
• Measurement errors
• Specialist-related biases in prices
• Timing of corporate releases after the stock exchanges close on Fridays
• Reduced institutional trading and greater individual trading on Mondays
• Daylight saving time changes for two weekends of the year
• Delay in the settlement of trades and bid-ask bounce14
3.2.4 Persistence
The weekend effect was observed for a wide spectrum of assets, indices and time spans. It has been assumed that the timing of news releases was the main reason for the observed abnormal returns on worldwide indices. However, news releases are not restricted to a given type of firm, and explain no more than 3.4 % of the weekend effect.15So the timing of the delivery of bad news may not be a sufficient reason to explain this anomaly. Daylight saving time changes did not seem to impact the weekend effect in 1989,16neither did they later between 2001 and 2010.17The
11Singal (2006, p. 48).
12Singal (2006, p. 48).
13Sias and Starks (1995, p. 66).
14The bid-ask bounce is the process that on Fridays, the asset is traded at Friday’s ask price at the close of trading, whereas on Mondays, it trades at Fridays’ bid price at the start of the trading session.
15Damodaran (1989, p. 607).
16Damodaran (1989, p. 616).
17Patel (2012, p. 109).
delay in settlement is assumed to represent around 17 % of the weekend effect, while the bid-ask bounce would explain 32 and 10 % of the observed market anomaly in an equally-weighted and value-weighted index, respectively.18
The effect can still be noticed nowadays. It has lasted so long for three major reasons:
• It was not understood before 2003.
• It is tightly linked to specific index and share types.
• Its magnitude is small.
3.2.5 Summary
Before 2003, the reasons of the weekend effect were hardly understood. It could thus disappear without notice. It was also impossible to optimize and refine a strategy to capture the highest available profit. This anomaly is close to non-existent in a value- weighted index, and it is hard to benefit from an equally-weighted index. Illiquid markets for small-caps and high trading costs can make it prohibitive to surf on the weekend effect. It is even more complicated to generate a profit since the average returns do not exceed 0.33 %.19That is why the best an investor can do would be to just buy stocks on Mondays and sell them on Fridays.