Author: Ariss, Rima Turk
Date published: July 1, 2011
(ProQuest: ... denotes formulae omitted.)
The Efficient Market Hypothesis in its weak form claims that security prices reflect all market-related information, and current market prices are the best estimate of the intrinsic value of the stocks and, as such, no daily patterns or anomalies exist or can be employed to obtain abnormal returns. Despite this claim, calendar anomalies in equity markets have attracted a lot of attention in the finance literature. One of these anomalies is the dayof-the-week or Monday effect. The Monday effect happens when returns are lower or negative on Mondays in comparison with the returns on other days of the week.
While these calendar effect anomalies have been extensively investigated in the literature and are generally documented in the context of developed and other emerging countries, the capital markets of the Gulf Cooperation Countries (GCC) have received limited attention.1 However, interest in the GCC region has increased over the past decades, due to the rise in petrodollars resulting from the oil boom, the high levels of economic growth achieved, and following the gradual opening up of markets to foreigners.
The GCC stock markets do not observe the same trading days or religious holidays relative to international markets, and they have unique structures, thus questioning whether previously-reported calendar anomalies for other markets hold for this region. For instance, the last working day of the week falls on Thursday in all GCC countries except for Saudi Arabia (the leading market of the GCC region), where Wednesday is the last day of the week. It follows that Monday does not represent the first day of the week in any GCC country and it is possible that the negative Monday returns documented for international markets do not hold for these markets.
Further, religious holidays are based on the Hijri or lunar calendar year and do not coincide with other Gregorian calendar-based religious holidays. The month of Ramadan2 bears a special significance to the GCC region because it is the month during which the Koran was revealed to the Prophet Muhammad. Muslims are required to fast during this month and to show care for the less fortunate in society, in addition to exercising spirituality by the recitation of prayers, because it is believed that Ramadan is the month of mercy and forgiveness. Therefore, it is legitimate to assume that the occurrence of Ramadan may influence the trading behavior of investors as well.
Research on stock market anomalies for GCC markets is limited to individual countries. For instance, a study by Seyyed and Al-Hajji (2005) finds that volatility is reduced in the month of Ramadan for the Saudi Tadawul stock market, and Al-Saad and Moosa (2005) document a July effect for the Kuwait market. But Al Loughani, Al Saad, and Ali (2005) do not report the presence of any holiday effect for Kuwait, nor does Al-Khazali (2008) find a day-of-theweek effect for the United Arab Emirates stock markets.
The conflicting results on seasonalities for individual GCC markets, and the lack of a comprehensive study of calendar anomalies for all GCC markets combined, motivate our study. We find that, across all GCC markets, market returns are positive and significant on Wednesdays, which is the last trading day of the week in the leading market of the region,suggesting the presence of a Fridaytype effect similar to that reported for Western equity markets in the literature. The average positive returns on Wednesday may be the result of the optimistic mood of investors just before the weekend begins, because it generates excess demand on this day, an outcome that is consistent with the explanation of the "Investors' Mood Hypothesis." Our results, however, do not provide any evidence to suggest that market returns are significantly different in the month of Ramadan compared to other months of the lunar calendar year.
The rest of the paper is organized as follows: Section Il presents the data and methodology. Section III overviews the empirical findings, and we interpret the results in Section IV. Section V presents our conclusion.
II. Data and Methods
We retrieve the daily closing values of all six major GCC market indices from inception until June 2008 and prior to the global financial crisis from Global Financial Data, to compute the daily stock returns R,f for all market indices as follows:
where Rit is the daily percentage return of stock index / on day f, and /rt and /,M are the closing values of the index on day t and t-1 respectively. We also test for stationarity in the data using the Dickey-Fuller unit root tests on each GCC stock market return series. Exhibit 1 shows the results, which reject the presence of unit roots and show that all stock price indexes are stationary in their first differences.
We use the following regression model to examine whether there is a day-of-the-week effect in GCC markets:
where Rit is the daily return of stock index /, D^through D5, are dummy variables for each of the five trading days of the week for index / and š is a random error term. We estimate equation (2) using Ordinary Least Squares (OLS) with robust standard errors. In the presence of a Monday effect, the estimated coefficient a,, is expected to be negative and significant or at least significantly lower than the coefficients for the rest of the week.
Since the month of Ramadan bears a special significance to GCC investors, we investigate whether average market returns are significantly different during this month and relative to other months of the Hijri lunar calendar year. To that end, we estimate a variety of equation (2) that includes dummy variables that correspond to the 12 months of the lunar calendar year instead of five dummy variables for the trading days of the week. These months are: Muharram, Safar, RabÝ' Al-Awwal, Rabi1 Al-Thani, Jamadi Al-Awwal, Jamadi Al-Thani, Rajab, Sha'aban, Ramadan, Shawwal, Dhu Al-Qi'dah, and Dhu Al-Hijjah.
We also exclude the month of Ramadan from our sample, and re-estimate equation (2) separately for pre-Ramadan days and for post-Aamadan days, to assess the extent to which the mood of investors may affect their pricing behavior outside the month of Ramadan.
III. Empirical Findings
In order to test for the presence of a possible day-of-the-week effect in GCC stock markets, we estimate equation (2) for each market index using ordinary least squares. The results are displayed in Exhibit 2. It shows that Wednesday returns are statistically significant and positive across all markets. For Bahrain, Qatar, Saudi Arabia, and the United Arab Emirates (UAE), Wednesday returns are not only positive, but they are also the highest among all returns for other days of the week. Further, Wednesday returns are next to the highest returns compared to the other days of the week in the case of Kuwait and Oman. Finally, in the case of Kuwait, Qatar, and the UAE, where Thursday is last day of trading, Wednesday returns continue to be positive and statistically significant on Thursday.
The results revealed in Exhibit 2 suggest that there exists a sort of daily anomaly at the end of the week in all GCC countries. This daily anomaly can be viewed as a form of the Friday effect that is documented in the literature for Western equity markets.
The similarity between Western and GCC market regarding investors' reactions to days of the week and pricing behavior at the end of the week is very interesting. In fact, the present paper is the first study that documents a similar end-of-the-week anomaly in returns among Western and GCC stock markets.
In order to support our empirical result regarding the presence of a "Wednesday effect" in GCC countries, we compute normalized average daily returns, and the results are presented in Exhibit 3. Among all GCC countries, the Qatar stock market index has generated the highest normalized daily mean return, whereas the Saudi Arabia market index has produced the lowest normalized mean return.
More importantly for our study, we find that the average normalized returns on Wednesdays are the highest compared to the mean normalized returns for other days of the week across all six GCC stock markets. This finding is consistent with what was reported earlier based on OLS results, thus reinforcing the presence of a "Wednesday effect" in this part of the world.
As discussed previously, the month of Ramadan bears a special significance to GCC investors, and it is possible that investor behavior is modified during this month. To assess whether the investors modify their market behavior during this month, we estimate equation (2) replacing weekday dummies with month dummy variables denoting each of the 12 months of the lunar calendar year and report the results in Exhibit 4.
The figures in Exhibit 4 do not show a Ramadan effect consistently across all markets. The parameter estimates are only significant for the Oman and Qatar, which are among the smallest markets in the GCC region. Thus, the estimation results appearing in Exhibit 4 do not provide evidence to suggest that investor behavior is significantly altered by the occurrence of the month of Ramadan.
Alternatively, we investigate further whether our previously documented "Wednesday effect" is maintained outside the month of Ramadan. We re-estimate equation (2) for two sub-periods separately, partitioning our data into two sets, pre-Ramadan days and post-Ramadan days. We present the results in Exhibit 5 for pre-Ramadan days and in Exhibit 6 for post-Aamadan days. The figures in Exhibit 5 strongly support our earlier finding reported in Exhibit 3 regarding the presence of a statistically significantly positive Wednesday effect for GCC countries, suggesting that this effect is very strong in the days leading up to the occurrence of the month of Ramadan. In testing for the day-of-the-week effect after the month of Ramadan in Exhibit 6, we again detect a statistically significant positive Wednesday effect, but only in three out of the six examined markets. For the markets of Kuwait, Qatar, and UAE, the coefficient on Wednesday is positive, though not statistically significant.
To sum up, these results reveal the presence of a Friday-type effect taking place on Wednesdays in the GCC region, and this effect is more pronounced during the period leading up to the occurrence of the month of Ramadan as compared to the less prevalent Wednesday effect in the period following it. This result might be not so surprising after all, since the major Muslim holiday occurs two months and ten days after the end of Ramadan every year, and this holiday corresponds with the period of the pilgrimage.
IV. Interpretation of Results
In the literature on calendar anomalies, several hypotheses have been set forth to explain the presence of a day-of-the-week effect. One of these hypotheses is the "Short Selling Hypothesis," which argues that the source of daily anomalies in the form of the traditional "Friday effect" is the behavior of short sellers on the last trading day of the week. According to this hypothesis, short sellers attempt to close their short positions before weekends, causing excess demand for equity securities, and hence generating positive returns on Fridays (Chen and Singal, 2003). It follows that short sellers' attempt to create short positions after the weekend (particularly on Mondays) causes an excess supply of securities, and results in negative returns on the first trading day of the week or on Monday. In the case of the GCC stock markets, however, this explanation for the end-of-the-week anomaly has to be rejected on the ground that short selling activities are forbidden in these countries. The reason is closely tied to the religious underpinnings of the Islamic law that one cannot sell what one does not own.
Another hypothesis that attempts to explain the market anomaly related to the last trading day of the week is known as the "Settlement Procedure Hypothesis" (Gibbons and Hess, 1981). This hypothesis postulates that the cause of the end-of-the-week anomaly is the relatively long settlement period in advanced stock markets that provides an opportunity for traders to get hands-on interest free funds for several days. However, this hypothesis is also not valid in the case of the GCC market because the settlement cycles in these countries are relatively shorter relative to those applicable in advanced economies. They vary from none in the case of Kuwait, to one day for Saudi Arabia, two days for Bahrain, and three days for Oman and the UAE (Source: national websites).
One may speculate that the behavior of investors in GCC markets during the last trading day of the week is in line with the explanation provided by the "Investors' Mood Hypothesis" (Gondhalekar and Mehdian, 2003). This hypothesis posits that the statistically significant and positive returns experienced on the last trading day of the week are caused by the optimistic mood of investors just before the weekend begins. This prevailing "mood" makes investors more enthusiastic to buy and less eager to sell on the last trading day of the week. In the case of the GCC stock markets, one may cautiously accept the predictions of this hypothesis. It could be that the psychological state of mind of investors prevailing right before the weekend results in an excess demand for stock investments. This hypothesis can also be used to explain the lessened "Wednesday effect" in the period following the month of Ramadan, because this is the period in which GCC market participants generally prepare themselves for their major religious holiday which coincides with the period of the pilgrimage, and they might be less geared toward stock market investing during this period.
Over the past few years, GCC stock markets have witnessed unprecedented economic growth fueled by high oil prices and resulting in excess petrodollars liquidity, which soon found its way back to the regional stock markets. In parallel, market trading activity increased tremendously with a rise in the number of market participants and listed firms. Further, all GCC member countries have become members of the World Trade organization, and they are meeting accession requirements by opening up their markets to foreign investors. The latter recognize an opportunity for international diversification, especially in light of the relative resilience of the GCC markets to the global financial crisis when compared to other economies.
In this paper, we investigate whether a previously reported daily calendar anomaly for other markets is applicable in the context of the GCC stock markets. Prior literature for the region is limited to single country analyses and only reports declining volatility during the month of Ramadan.
We extend our coverage to the entire GCC region and gather data on market indexes for all six exchanges, from their inception up to the middle of 2008. We then test for the presence of a day-of-the-week effect, and investigate whether the occurrence of the month of Ramadan has a special bearing on market investing behavior.
In line with prior literature for other international markets, we find that average market returns are positive and significant on the last trading day of the week, which predominantly occurs on Wednesdays in the GCC region and not on Fridays. This calendar anomaly, however, is more pronounced during the pre-Ramadan period, since after Ramadan investors generally prepare themselves for their major religious holiday that corresponds with the period of the pilgrimage. We also do not find that market returns are significantly different in the month of Ramadan relative to other months of the lunar calendar year.
All in all, these findings reveal the presence of a Friday-type effect taking place on Wednesdays (the last trading day of the week) in the Gulf Cooperation Council region over the entire sample, in the pre-Ramadan sub-sample, and to a lesser extent in the post-flamadan sub-sample, notwithstanding an insignificant return effect during the month of Ramadan and as compared to other months of the year. We explain our findings using the "Investors' Mood Hypothesis" rather than the "Short selling Hypothesis" or the "Settlement Procedure Hypothesis".
1 The GCC countries include Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and United Arab Emirates (UAE).
2 In 2011, Ramadan will start on the 1st of August and continue for thirty days. The starting date varies by approximately eleven days every year in the solar-based Gregorian calendar, which does not coincide with the Islamic calendar.
Al-Khazali, Osamah M. (2008). The Impact of Thin Trading on the Day-of-the-Week Effect: Evidence from the United Arab Emirates. Review of Accounting and Finance 7(3) 270-284.
Al-Loughani, N., Al-Saad K., and Ali, M. (2005). The Holiday Effect and Stock Return in the Kuwait Stock Exchange. Global Competitiveness 13, 81-91.
Al-Saad, K. and Moosa, I. (2005). Seasonally in stock returns: evidence from an emerging market. Applied Financial Economics 1 5, 63-71.
Chen, H. and Singal, V. (2003). Role of Speculative Short Sales in Price Formation: The Case of the Weekend Effect. The Journal of Finance 153(2), 685-705.
Gondhalekar, V. and Mehdian, S. (2003). The Blue-Monday Hypothesis: Evidence based on NASDAQ Stocks, 1971-2001. Quarterly Journal of Business and Economics 42, 73-89.
Gibbons, M. and Hess, P. (1981). Day of the Week Effects and Asset Returns. Journal of Business 54(4), 579-596.
Seyyed, F., Abraham, A., and Al-Hajji, M. (2005). Seasonality in Stock Returns and Volatility: The Ramadan Effect. Research in International Business and Finance 19, 374-383.
Rima Turk Ariss, Lebanese American University, Beirut, Lebanon
Rasoul Rezvanian, Northeastern Illinois University, Chicago, Illinois
Seyed M. Mehdian, University of Michigan-Flint, Flint, Michigan