Day-of-the-Week and the Returns Distribution: Evidence from the Tunisian Stock Market

Transcription

1 The Journal of World Economic Review; Vol. 6 No. 2 (July-December 2011) pp Day-of-the-Week and the Returns Distribution: Evidence from the Tunisian Stock Market Abderrazak Dhaoui * * University of Sousse, Faculty of Economics Sciences and Management, UR : MO2FID, abderrazak.dhaoui@yahoo.fr ABSTRACT In this paper, we examine the behavior of returns across the-day-of-the-week in the context of the Tunisian Market. Our evidence indicates that Mondays have abnormally losses. In opposition, returns are significantly higher in Friday. We also find that these Monday and Friday specifications are evenly observed whether these days are first or last half-of-the-month days. The same results remain similar across all weeks of the month. Losses diminish across the day-of-the-week from Monday to Friday. The losses disappear starting the end of Tuesday and the beginning of Wednesday with, however, a non significant Tuesday effect during the three first weeks of the month and abnormally losses in the last two weeks. JEL Classification: G1, G14. Keywords: Day-of-the-week, Half-of-the-month, Returns, Abnormal losses, Overconfidence. INTRODUCTION Several studies document that the distribution of the stock returns are not evenly the same across the day of the week (see French, 1980; Barbee et al., 2008; Tripathy, 2010; Ulussever et al., 2011; Wang et al., 1997). The specificity of the Returns distribution in the major market in developed or emerging countries is that Mondays have abnormally losses and Fridays have a high stock returns (Boynton et al., (2009) for the Japanese market; Wang and al., (1997) for the U.S. market; Agathee (2008) in the Mauritius Market; Ulussever et al., (2011) in the context of the Saudi Arabia market). The presence of the weekend (or, rather, in general, the day-of-the-week) effect constitutes an important argument opposing the efficient market hypothesis what one of the implication is that the expected returns on assets should be evenly distributed across the time (days, weeks, months, years, or all other unit of time). In this line, this study analyses the behavior of the stock returns to see whether the day-of-the-week effect is visible in the Tunisian market context or not. In order to do this, one question could have a wide importance that is: What effect has the day-of-the-week on the stock returns in the context of the Tunisian Stock Market? Using a sample of 24 stocks listed in the Tunisian Stock Market over 925 trading days from January 02, 2007 to September 22, Results show that Weekends have similar effects that those found in international markets. Mondays have abnormally losses while Fridays have significant high returns. Results indicates also that Tuesdays effects are non significant for the three first weeks of the month. For the two last weeks, the effect becomes significantly negative

2 164 THE JOURNAL OF WORLD ECONOMIC REVIEW and abnormally losses can be observed. For Thursdays the returns are similar whatever the week-of-the-month the Thursday is. The remainder of the paper proceeds as follows. Section 1 presents the literature review on the day-of-the-week effect on the stock returns. Section 2 provides the sources of the data and the sample selection as well as the estimated model. Section 3 contains the empirical results. Concluding remarks are provided in the last section. 1. THE DAY-OF-THE-WEEK EFFECT ON THE STOCK RETURNS One of the widely studied phenomena in financial markets is the effect of the market anomalies and specially that of the day-of-the-week on the stock returns (see Pettengill, 2003; Barbee et al., 2008; Tripathy, 2010; Syed and Sadorsky, 2006; Yalcin and Yucel, 2006; Agathee, 2008). However, in spite of the numerous papers studying these effects, those giving explanation to this phenomena still less numerous. Several studies have documented significant relations between abnormal loss and Mondays. This prediction is confirmed in the context of the American market by Cross (1973) and French (1980). Significant abnormally low returns on Monday have also been discovered in several international markets other than the U.S. one. The tendency of stocks to exhibit relatively low returns on Monday compared to those on Friday is called the day-of-the-week effect largely known as the weekend effect. It indicates that the average daily return of the market is not the same across the days of the week (see Boynton et al., 2009). The result commonly shared in theoretical and empirical studies (Keim and Stambaugh, 1984, Rogalski, 1984; Aggrawal and Rivoli, 1989; Barbee et al., 2008; Tripathy, 2010; Ulussever et al., 2011) is that the average returns on Mondays are significantly less than the average returns over the other days of the week and more specifically than those on Friday. Other papers examine the same effect of the day-of-the-week but find that the lower return is not on Mondays. For example, Jaffe and Westerfield (1985a,b) show that in the Japanese and the Australian markets Tuesday have the abnormally poor returns. However, Boynton et al., (2009) document that Tuesday losses disappear in the 1990s. They confirm the result commonly found in international markets. Particularly, they find, from February 1989 to December 2001, abnormally poor Monday returns in the context of the Japanese Market. In spite of the rare attempts on the fact, many explanations of the day-of-the-week effect have been proposed and investigated in theoretical and empirical papers. In these sense, about 17 per cent of the weekend effect have been explained by the delay in settlement (Lakonishok and Levi, 1982; Keef and Roush, 2005). Damodaran (1989) explains the weekend effect by the delay in corporate announcements. However this element explains only a small proportion of the effect. In his turn, Keim (1989) find, in the specific case of small stocks, that the bid-ask bounce can account for the observed day-of-the-week effect. For Lakonishok and Maberly (1990) one answer to the day-of-the-week effect can be the reduced institutional trading and the greater individual one. The explanation given by Athanassakos and Robinson (1994) is that abnormally low Monday returns are due to the macro announcements. In the same vein, different hypothesis are considered to justify the losses observed on Monday in the major international Markets compared to the stock returns across the other days of the

3 DAY-OF-THE-WEEK AND THE RETURNS DISTRIBUTION: 165 week of which especially those on Friday. At first, the outliers hypothesis indicates, according to Conolly (1989) that low returns on Monday are due to isolated rare events that can be detected using robust regression test. In this line, Boynton et al., (2009) find that Japanese Monday day-of-the-week return loss is relatively robust 1. The second hypothesis is related to the later-half-of-the-month. This hypothesis says that the Monday abnormal losses are higher in the last half-of-the-month. This prediction is confirmed in the U.S. market by Wang et al., (1997). They find, particularly, that Monday abnormal losses become more intensive in the fourth and fifth weeks. However, despite the importance of this result, the authors do not explain why these low returns occur. Another hypothesis called the short sales hypothesis is considered to explain the weekend effect. This hypothesis indicates that the abnormal losses are linked to the frequency of short sales and that the more frequent short sales are observed on Monday (Chen and Signal, 2003). The authors examine the U.S. market to test this hypothesis. They find no support for the short sales hypothesis. Results indicate, however, that stocks with great short interest have high returns on Monday and low returns on Friday. Thus, the short sales frictions do not induce necessarily low Mondays returns compared to those on Friday. The day-of-the-week effect is also explained by the autocorrelation hypothesis which indicates that Mondays have higher AR (1)s than other days of the week. At his turn, the asymmetric information is considered as an argument that can explain the weekend effect. In this vein, the information asymmetry explains largely the phenomena of which the expected returns on assets are not evenly distributed across the days (see French, 1980; Rogalsk, 1984; Penman, 1989; Defusco et al., 1993), result widely observed in empirical studies. According to Foster and Viswanathan (1990) the adverse selection can explain in party the high Monday losses. In this line, they consider two types of investors: the informed and the less informed investors. They suggest, particularly, that the informed have better information on Monday. Thus, given up the uninformed investors know that on Monday the more informed have greater information and will exploit this advantage in trades, they withdraw liquidity and postpone trades, as a best strategy, until the market thickens, the prices become more informative and the information effect diminishes (see Boynton et al., 2009). Boynton et al., (2009) examines the day-of-the-week effect in the context of Japanese market to test these hypotheses. Firstly they have withdrawn the test for the short sales frictions because they have not short interest in the Japan. The results for the other four hypotheses do not support the later-half-of-the-month and the outliers hypotheses. They confirm, however, the others hypotheses related to the adverse-selection and to the autocorrelation hypotheses. These results are consistent with those of Foster and Viswanathan (1990) in the U.S market. In the same context, Jaffe and Westerfield (1985a), Abraham and Ikenbery (1994) and Bessembinder and Herzel (1993) documents that Mondays returns in U.S. market are higher autocorrelated than other days. The results show specifically that returns on Monday are no different than other days when the Friday return is positive. Oppositely, the return will be abnormally low on Monday when Friday has poor returns. However, in spite of the importance of these results, the authors do not explain why this autocorrelation occurs.

4 166 THE JOURNAL OF WORLD ECONOMIC REVIEW 2. DATA AND METHODOLOGY The data we used is daily data returns that cover the period from January 02, 2007 to September 22, We include all stocks that trade every day for the year. Days corresponding to official religious holidays are, in their turn, withdrawn. The data includes, as a result, 24 stocks listed in Tunisian stock market (Tunis stock exchange) over a period of 925 trading days. We use three classifications to investigate the return behavior across the days. The first is the daily classification. The second is the half-of-the-month classification. And the last is the week-of-the-month classification. For each sample we regress returns on each of the day-of-the-week. The estimated equation is: R N = α + α d with i = 1,, 5; t i i i i = that to be equivalent to all day from Monday to Friday, d i is a dummy variable indicating that the day t is the day-of-the-week i. In such case, it takes the value 1. It takes the value 0 otherwise. is the mean returns for the day-of-the-week i. The is calculate as follow: 3. RESULTS AND DISCUSSION = [ln (p t )/ln (p t 1 )] 100. Descriptive statistics for the day-of-the-week returns (daily classification) are given in Table 1. Table 1 Descriptive Statistics (All the Days) Variables N Mean Max Min SD Skewness Kurtosis Mo Tu We Th Fr Results in Table 1 indicate, on average, an abnormal low return on Monday and a high return on Friday. The low returns on Monday are associate to a high volatility on the stock returns (high sd = ). In this sense, although the fact that Monday presents the higher returns, it has, at the same time, the lowest returns. Fridays returns have, oppositely, moderate volatility. These results incite to investigate the origin of the low Monday returns and the high Friday. Thus, we try to investigate the day-of-the-week depending on whether the day is the first or the last half-of-the-month day, and in a similar way, whether the day is the first to the fifth week-of-the-month day-of-the-week. Table 2 summarizes the descriptive statistics classified by half-of-the-month day-of-the-week effect. Results in Table 2 show that the returns are evenly distributed whether the day is the first or the last half-of-the-month day-of-the-week. In this line, results indicate that the average returns remain abnormally low whatever is the Monday across the month. At the same time, the Friday average returns remain higher even if the day is the first or the last half-of-the-month Friday.

5 DAY-OF-THE-WEEK AND THE RETURNS DISTRIBUTION: 167 Table 2 Descriptive Statistics (Half-of-Month Classification) Variables N Mean Max Min SD Skewness Kurtosis Mo Mo Tu Tu We We Th Th Fr Fr Results show, moreover, that Mondays returns have high standard deviation. Oppositely, Fridays have low standard deviation especially at the first half-of-the-month. Table 3 presents the day return by week-of-the-month classification. Table 3 Descriptive Statistics (Week Calssification) Variables N Mean Max Min SD Skewness Kurtosis Mo Mo Mo Mo Mo Tu Tu Tu Tu Tu We We We We We Th Th Th Th Th Fr Fr Fr Fr Fr

6 168 THE JOURNAL OF WORLD ECONOMIC REVIEW Results in Table 3 show that Mondays returns are abnormally low for all the-week-of-themonth. The lowest returns are specifically observed on the first, the third and the fourth week. Moreover, these Mondays are characterized by a high volatility. Oppositely, the returns are higher for every Friday across the month with simultaneously low volatility. For the other days, both the return and the volatility are moderate except the last half-of-the-month Tuesdays where the average return is abnormally low. The results for the Day-of-the-week return regression for all the month-days are given in Table 4. Table 4 Day-of-the-Week Return Regression (All the Days) Variables Coef. T-student Mo ** Tu ** We ns Th *** Fr ** _cons R-sq Adj R-squared F 2.25 *** Significant at 1% level, ** Significant at 5% level, * Significant at 10% level. Results in Table 4 show that returns behavior are not the same across the days. In fact, the average returns increase on Thursday and Friday and decrease on Monday. These results seem to confirm the major previous studies on international market. They confirm, particularly, those of Cross (1973) and French (1980), in the American market, those of Sadorsky (2006) in the context of emerging stock markets, or Agathee (2008) in the Mauritius Market and those of Ulussever et al., (2011) in the stock market of Saudi Arabia. Moreover, we find a non significant effect on Wednesday. The Thursday effect is, oppositely, positive and significant. The results on the effects of the first and that of the last half-of-the-month day-of-the-week on the returns are given in Table 5. Results in Table 5 indicate that the weekend effect on the return remains the same whether the day is the first or the last day of the month. In fact, the Monday effect remains negative and that of Friday remains positive. On the other hand, the Tuesday and the Wednesday s impact, remains non-significant on the first half-of-the-month. The effect becomes, however, significant but negative on the last half-of-the-month. Thus, we try to investigate the behavior of the daily return across each week-of-the-month. Results for day effects by week are given in Table 6.

7 DAY-OF-THE-WEEK AND THE RETURNS DISTRIBUTION: 169 Table 5 Day-of-the-Week Return Regression (Half-of-Month Classification) Variables Coef. T-student Mo ** Mo *** Tu ns Tu ** We ns We ** Th ** Th * Fr *** Fr ** _cons *** R-sq Adj R-squared F 4.23 *** Significant at 1% level, ** Significant at 5% level, * Significant at 10% level. Table 6 Day-of-the-Week Return Regression (Week Calssification) Variables Coef. T-student Mo ** Mo ** Mo ** Mo ** Mo ** Tu ns Tu ns Tu ns Tu ** Tu ** We ns We ns We ns We ns We ns Th ** Th ** Th ** Th ** Th ** Fr *** Fr ** Fr *** Fr ** Fr * _cons R-sq Adj R-squared F 6.16 *** Significant at 1% level, ** Significant at 5% level, * Significant at 10% level.

8 170 THE JOURNAL OF WORLD ECONOMIC REVIEW Results in table 6 indicate that the Monday losses and high returns on Friday remain similar across the weeks. Whatever the week-of-the-month, the Return increases significantly on Friday and decreases abnormally on Monday. The Tuesday effect remains, widely, non-significant or, sometimes negative, especially on the fourth and the fifth week of the month. Beginning from Wednesday, the effect becomes significantly positive. This to say that the Tunisian market presents abnormal losses on Monday and that the losses diminish through the day as far as the effect becomes positive except for the first week of the month. Considering both the Monday and the Tuesday effect across the overall week of the month, we notice that the losses are observed on all weeks (for Monday) and especially the fourth and the fifth week (for the specific case of Tuesday). This confirms surely the results shown in Wang et al., (1997) and Chen and Signal (2003) using U.S. data according to which the (Mondays) returns are abnormally low on the fourth and the fifth week. The results are not, however, similar to those shown in Boynton (2009) where the effect seems to be similar whatever the week-of-the-month is. The negative Tuesday effect (especially for the two last weeks of the month) confirms partially the results shown in Jaffe and Westfield (1985b) in the Japanese context according to which Tuesdays have abnormally poor returns. Several explanation can be considered to justify the Mondays abnormally losses. But, even the importance of the major hypothesis, the investor overconfidence sentiment remains almost, as our knowledge, non-considered to explain the differences in effect across the day and especially for the specific case of Monday and Friday. Firstly, and in expense to the arguments given in the rare previous attempts of explanation of the weekend effect, the behavior of the individual and that of institutional investors can explain the abnormally low returns on Mondays. In this sense, since individuals are, as most of them, employed during weekends trading hours with other activities, the gathering information process during that period of time becomes significantly costly. For these investors Monday provides suitable opportunity to collect information and to reach at investment decision. At the same time, the institutional as particular and more informed investors use Mondays to frame their trading strategies for the rest of the week days. They decrease, consequently, their trading volume which induces, accordingly, a low pressure on the stock price on that day. Moreover, and in the same vein, the investors sentiment can be considered as a pivotal factor influencing the variability of the stock returns across the days. The overconfidence hypothesis says that two types of investors coexist the more informed and those with less information. The overconfident investors overestimate the precision of their knowledge and especially their judgment skills and underestimate the public information and the skills of the less informed investors. Thus, they make aggressive decisions, which increase significantly the trade volume (bay or sale). In opposition, the less informed investors are, at their turn, act in a rational way. They know that on Monday the more informed investors will exploit their information advantage in trades. As result, they (i.e. the less informed investors) postpone trades until the market thickens and the price become more informative. The delay of trade driven by the less informed investors induces a decrease on the demand (and supply depending on the case). In the other hand, the opposite strategy (supply vs. demand) remains over requested level. As a result, the operation ends with abnormally losses. Appositely, across the (other) days, the prices start to be more informative. Consequently, both the more informed and the less

9 DAY-OF-THE-WEEK AND THE RETURNS DISTRIBUTION: 171 informed investors trade together which induces an opposite behavior of stock prices. Prices start to increase significantly following the intensity of trade. As a result a high return can characterize the last days of the week. 4. CONCLUSION The day-of-the-week has a significant influence on the stock returns. Significant differences in effects across the days, the week or the months are largely shown in the context of international markets. The results, using a sample of 24 stocks listed in Tunisian Stock Market, show similar impact as those observed in the markets of developed and emerging countries. Mondays have abnormally low returns. Oppositely, high returns are observed on Fridays. Losses diminish across the day-of-the-week. From Monday to Friday, low returns are shown on Mondays and sometimes on Tuesday especially for the two last weeks of the month. Starting from Wednesday the effect of the day-of-the-week on the returns becomes significantly positive. The highest returns can be shown on Fridays. Although the importance of international studies in this tasks rare are the attempts of explanation of the aforementioned day-of-the-week effects on the stock returns. In this paper, we consider the investors sentiment as an important factor influencing the returns behavior across the day-of-theweek. The coexistence of both more informed and less informed influences the decision process which influence, at his turn, the stocks prices and consequently the stock returns. Note 1. To detect outliers, Boynton et al., (2009) use both the M-estimator of Huber (1981) and the Rousseeuw s Least Trimmed Squares. For the second, they vary the trims across the samples to show the sensitivity of the OLS results to trims from outlier. References Abraham A., and Ikenberry D. L., (1994), The Individual Investor and the Weekend Effect, Journal of Financial and Quantitative Analysis, 29, Agathee U. S., (2008), Day-of-the-Week Effects: Evidence from the Stock Exchange of Mauritius (SEM), International Research Journal of Finance and Economics, 17, Aggarwal R., and Rivoli P., (1989), Seasonal and Day-of-the-Week Effect in Four Emerging Stoick Markets, Financial Review, 24, Athanassakos G., and Robinson M., (1994), The Day-of-the-Week Anomaly: The Toronto Stock Exchange Experience, Journal of Business Finance and Accounting, 21(6), , (September). Barbee Jr. W. C., et al., (2008), Relations Between Portfolio Returns and Market Multiples, Global Finance Journal, 19, Bessembinder H., and Hertzel M. G., (1993), Return Autocorrelations Around Non-Trading Days, Review of Financial Studies, 6, Boynton W., et al., (2009), Japanese Day-of-the-Week Patterns: New Results, Global Finance Journal, 20, Chen H., and Singal V., (2003), Role of Speculative Short Sales in Price Formation: The Case of the Weekend Effect, Journal of Finance, 58,

10 172 THE JOURNAL OF WORLD ECONOMIC REVIEW Conolly R. A., (1989), An Examination of the Robustness of the Weekend Effect, Journal of Financial and Quantitative Analysis, 24, Cross F., (1973), The Behavior of Stock Prices on Fridays and Mondays, Financial Analysts Journal, 29, Damodaran A., (1989), The Weekend Effect in Information Releases: A Study of Earnings and Dividend Announcements, Review of Financial Studies, 2, Foster F. D., and Viswanathan S., (1990), A Theory of Intraday Variations in Volume, Variances, and Trading Cost in Securities Market, Review of Financial Studies, 3, French K., (1980), Stock Returns and the Weekend Effect, Journal of Financial Economics, 8, French K. R., (1980), Stock Returns and the Weekend Effect, Journal of Financial Economics, 8, Jaffe J., and Westerfield R., (1985a), The Weekend Effect in Common Stock Returns: The International Evidence, Journal of Finance, 40, Jaffe J., and Westerfield R., (1985b), Patterns in Japanese Common Stock Returns: Day-of-the-Week Effects and Turn of the Year Effects, Journal of Financial and Quantitative Analysis, 20, Keef S. P., and Roush M. L., (2005), Day-of-the-Week Effect in the Pre-holiday Returns of the Standard and Poor s 500 Stock Index, Applied Financial Economics, 15, Keim D. B., and Stambaugh F., (1984), A Further Investigation of Weekend Effect in Stock Returns, Journal of Finance, 39, Keim D. B., (1989), Trading Patterns, Bid-Ask Spreads, and Estimated Security Returns, Journal of Financial Economics, 25, Lakonishok J., and Levi M., (1982), Weekend Effects on Stock Returns: A Note, Journal of Finance, 37, Lakonishok J., and Maberly E., (1990), The Weekend Effect: Trading Patterns of Individual and Institutional Investors, Journal of Finance, 45, Pettengill G. N., (2003), A Survey of the Monday Effect Literature, Quarterly Journal of Business and Economics, 42, Rogalski R. J., (1984), New Findings Regarding Day-of-the-Week Returns over Trading and Non-Trading Periods: A Note, Journal of Finance, 39, Syed A. B., and Sadorsky P., (2006), Day-of-Thr-Week Effects in Emerging Stock Markets, Applied Economics Letters, 13, Tripathy N., (2010), Expiration and Week Effect: Empirical Evidence from the Indian Derivative Market, Annual London Business Research Conference, 12 to 14 July, London, United Kingdom. Ulussever et al., (2011), The Day-of-the-Week Effect in the Saudi Stock Exchange: A Non Linear Garch Analysis, Journal of Economic and Social Studies, 1(1), 9-23, (January). Wang K., Li Y., and Erickson J., (1997), A New Look at the Monday Effect, Journal of Finance, 52, Yalcin Y., and Yucel E. M., (2006), The Day-of-the-Week Effect on Stock-Market Volatility and Return: Evidence from Emerging Markets, Czech Journal of Economics and Finance, 56,