Annals of the Constantin Brncui University of Trgu Jiu, Economy Series, Issue 3/2012

ACADEMICA BRNCUI PUBLISHER, ISSN 1844 7007

THE DAY OF THE WEEK EFFECT IN SOUTH EASTERN EUROPE STOCK MARKETS

DRAGAN TEVDOVSKI
ASSISTANT PROFESSOR IN STATISTICS, UNIVERSITY SS. CYRIL AND METHODIUS
E-MAIL: [email protected]
MARTIN MIHAJLOV
ASSISTANT PROFESSOR IN E-BUSINESS, UNIVERSITY SS. CYRIL AND METHODIUS
E-MAIL: [email protected]
IGOR SAZDOVSKI
ASSISTANT STUDENT IN STATISTICS FOR BUSINESS AND ECONOMICS, UNIVERSITY SS.
CYRIL AND METHODIUS
E-MAIL:[email protected]


Abstract
The main aim of this research is to examine existence of day of the week effect on the stock market indices in
five countries from South Eastern Europe (SEE): Bosnia and Herzegovina, Bulgaria, Croatia, Macedonia and Serbia
in the most recent period which is characterized by the bear market (from 2006 to 2011). The methodology used the
regression with dummy variables, or so called Analysis of Variance (ANOVA) model. In addition Wald test is applied.
The results imply that the mean daily return of the all five SEE indices is negative on Monday. The day of the week
effect is found only in Croatian and Bulgarian Stock Market. In both stock markets, the mean daily returns of the
leading indices are lower on Monday than the other days of the week and the results are statistically significant. The
lower Monday mean daily returns are found also in Macedonian stock exchange index, but the results are not
statistically significant. The mean daily returns of BELEX15 and BIFX indices in Tuesday are lower than mean daily
return on Monday, but also without statistical significance.

Key words: calendar anomalies, the day of the week effect, daily returns, SEE.

JEL Classifications: C32, G10

1.Introduction

The economics effects related to the calendar, that influence the movement of the returns in any stock market
are called calendar effects. Many of the effects show different behavior of the returns on different days of the week,
different times of the month, or different times of the year. They therefore are sometimes referred as seasonal effects,
and also may be examined in a period longer than one year. Most of the prompted effects are cyclical anomalies in the
term of returns, where the cycle is based on the calendar. The significance of such anomalies is a wide field of analysis
of the researches, which develop different methods and patterns to fit the values of the indexes and prices on the
different stock markets.
There are various calendar effects such as the day of the week effect, the half month effect, the turn of the
month effect or the month of the year effect. This paper focuses on the day of the week effect, which is also known as
the weekend effect. The returns on Monday should be measure of the result of the investment for 72 hours - from the
closing the stock market on Friday tills the opening on Monday. Therefore they should be higher than the 24 hours
returns in the other days in the week. However, many studies of the stock markets in different countries showed that the
average returns on Monday are lower than the other days in the week and even negative. There are different
explanations for the weekend effect. Abraham and Ikenberry (1994) argue that the investors have the tendency to sell
on Monday after the revision of their portfolios during the weekend. Chen and Singal (2003) shows that the investors
closed the short positions (buy) on Fridays and open them again on Mondays (sell). Taylor (2008) comments that not
satisfactory explanation has yet been given for the weekend effect.
The goal of this paper is to analyze the occurrence of the day of the week effect on the stock market indices in
five countries from South Eastern Europe (SEE): Bosnia and Herzegovina, Bulgaria, Croatia, Macedonia and Serbia for
the most recent period (from 2006 to 2011). This period is characterized with the bear market with started from the
middle of 2007.
The paper provides evidence that the mean daily return of the all five SEE indices is negative on Monday in the
analyzed period. The day of the week effect is found only in Croatian and Bulgarian Stock Market. In both stock
markets, the mean daily returns of the leading indices are lower on Monday than the other days of the week and the
20
Annals of the Constantin Brncui University of Trgu Jiu, Economy Series, Issue 3/2012

ACADEMICA BRNCUI PUBLISHER, ISSN 1844 7007


results are statistically significant. The lower Monday mean daily returns are found also in Macedonian stock exchange
index, but the results are not statistically significant. The mean daily returns of BELEX15 and BIFX indices in Tuesday
are lower than mean daily return on Monday, but also without statistical significance.
The structure of the paper is as follows. Section 2 surveys the literature. Section 3 presents the data. The
methodology and the empirical results are discussed in the Section 4 and Section 5, respectively. Section 6 concludes.

2. Literature Review

Since the seminal work of Fama (1965) numerous research papers can be found that have examined calendar
effects. Cross (1973) was amongst the first authors that have ever studied the day of the week effect, but without the
usage of statistical models. He studied the returns on the S&P 500 Index on the U. S. market, over the period of 1953-
1970. His conclusions are that the mean return on Friday (0.12 percent) is higher than the one on Monday (-0.18
percent). Similar conclusions were driven out by French (1980) who studying the S&P 500 for the period of 1953-1977
and found out that the Monday returns are lower than the other days by 0.2 % in average, and the average returns on
Monday are negative for 20 out of 25 years. Gibbons and Hess (1981) analyzed the Dow J ones Industrial Index and
found negative Monday returns.
Rogalski (1984) was the first to use the OLS regression method, F and t-statistics, revealed that stock returns on
Monday are negative, but not statistically significant. He made a distinction between the trading days returns and non-
trading days returns. By doing that he found out that the negative return on Monday is contained in the average Friday
close and Mondays open return. In addition, Dyl and Maberly ( 1988) explain that firms have a tendency of reporting
mostly bad news on Fridays. That delay of declaring bad news, might be related to the negative returns on Monday.
Chang et al. (1993) confirmed the significance of the weekend effect. Aggarwal and Tandon (1994) by the usage of
OLS dummy regression tested the day of the week effect and concluded that Monday stock returns are negative in
thirteen countries and significant in only seven. They also revealed that Friday returns are positive and significant in the
majority of the countries.
The more recent studies show us that the calendar effects are starting to diminish. Rubinstein (2001), Schwert
(2001) and Steely (2001) discovered during their international studies that calendar effects are becoming weaker,
especially in developed markets. Sullivan, Timmerman and White (2001) state that calendar effects are no longer
statistically significant. Chukwuogor-Ndu (2006) after analyzing the day of the week effect on stock market returns in
15 European countries, found that merely 7 markets show confirmative evidence.
One of the most famous month of the year effect is the J anuary effect. Most of the researches have shown that
stock returns tend to be higher in J anuary than any month in the year. One of the first studies on this subject was
performed by Rozeff and Kinney(1976). Their study was based on the New York Stock Exchange during the period of
1904-1974 and showed that the average returns were higher in J anuary(3,5%) compared to those for the other months
of the year(0,5%). Dyl (1977) found out that there is an abnormally high volume of stocks that decline in price in
December, and that the same stocks record abnormally high returns in J anuary. Brown et al. (1983) found a similar
pattern in Australian stock returns. Keim(1983) checked out the relationship of size effect and seasonality. He
concluded that the small firm returns in J anuary are significantly higher than the ones of large firms. However,
Easterday and Stephan (2006) re-examined the small firm J anuary effect proposed by Keim (1983) by dividing three
subperiods: 1943-1962,1963-1979 and 1980-2004. They found out that pre and post Keim period returns of small firms
were remarkably lower in J anuary than the ones examined in the period of 1963-1979.
A more recent study of Balaban (1995) was conducted with the aim of researching the month of the year effect
on the Turkish stock exchange. His study showed that not only J anuary, but J une and September also had significantly
higher returns than other months. Among these, J anuary had a return of 22% which was about 4 times greater than the
global return of all months. Using the OLS estimation with dummy independent variables each of which representing
the day belonging in the specific month, Arsad and Coutts (1997) revealed that the average returns in J anuary are
significantly positive for FTSE-100 index and this followed the introduction of capital gains tax in 1965. Also, Fountas
and Segredakis (2002) after studying eighteen equity market over the period of 1987-1995 concluded that the stock
returns for J anuary are significantly higher than the ones for the rest of the year in only five markets: Chile, Greece,
Korea, Taiwan and Turkey. Bildik (2004) confirmed the presence of the J anuary effect in the daily stock returns
derived from an Index of the Istanbul Stock Exchange over the period 1988-1999.
However, there are studies that show the presence of an April effect. Gultekin and Gultekin (1983) after
studying the UK stock market between 1959 and 1979 using both nonparametric and parametric methods, found out
that April had the highest returns. Kumari and Mahendra (2006) studied the month of the year effect on the Indian
Stock Market during the period of 1979-1998. They also found that the returns in April are significantly higher than the
other months. Researching the stock returns in Ghana between 1994-2004, Alagidede and Panagiotidis (2009) also
confirmed the existence of the April effect.
In the region of SEE there are few studies, as the best knowledge of authors. Georgantopoulos et al. (2011)
investigates five calendar anomalies (the day of the week effect, the J anuary effect, the half month effect, the turn of the
month effect and the time of the month effect) for four emerging stock markets, Romania, Bulgaria, Croatia and Turkey
21
Annals of the Constantin Brncui University of Trgu Jiu, Economy Series, Issue 3/2012

ACADEMICA BRNCUI PUBLISHER, ISSN 1844 7007


and their mature counterpart in the Balkan region, Greece, during the period 2000-2008. They provide evidence for the
existence of three calendar effects (day of the week, turn of the month, time of the month) for Greece and Turkey,
while the effects for the three emerging Balkan markets are limited and exist only in volatility. Georgantopoulos and
Tsamis (2011) results indicate that only two calendar effects (day of the week and J anuary effects) are present in the
Macedonian stock market during the period 2002-2008, while the half month effect, the turn of the month effect and
the time of the month effect are not. Georgantopoulos and Tsamis (2012) investigate also the calendar anomalies for
Bulgaria and Greece during the period 20022008. They found that most of the tested calendar effects exist for Greece
and the effects for Bulgaria are limited and exist only in variance. Karadzic and Backovic Vulic (2011) report absence
of three calendar anomalies: the J anuary effect, the turn-of the-month effect and the holiday effect for the Montenegrin
capital market during the period 2004 - 2010.

3. Methodology

The day of the week effect is analyzed using a model, originally proposed by French (1980) and used by
Rogalski (1984), J affe and Westerfield (1989), Agrawal and Tandon (1994), Mills and Couts (1995). It is the
regression with dummy variables, or so called Analysis of Variance (ANOVA) model:
R
t
=
1
+
2
D
2t
+
3
D
3t
+
4
D
4t
+
5
D
5t
+ u
t

where R
t
is the daily return on a selected stock market index, D
2t
is dummy variable that takes value 1 for
Tuesday and 0 for all other days, D
3t
takes value 1 for Wednesday and 0 for all other days, D
4t
takes value 1 for
Thursday and 0 for all other days, D
5t
is dummy variable that takes value 1 for Friday and 0 for all other days.
Monday represents the control category. The coefficient
1
indicates the mean daily return for Monday (control
category), while
2
to
5
represents the difference between the mean daily return for Monday and the mean daily
return for each of the other days in the week. The error term is noted as u
t
and it is assumed to be identically and
independently distributed (IID). Gujarati (2004) argue that ANOVA models are more general than the t test
which can be used to compare the means of two groups or categories only.
If there are no differences among index returns across days of the week, the parameters of
2
to
5
are
zero. Therefore, the null hypothesis of the relevant Wald test is the following
H
0
:
i
= 0 for i = 2, ,5.
If the null hypothesis is rejected, then stock returns should exhibit some form of the day of the week seasonality
(Georgantopoulos and Tsamis, 2011).

4. Data

The data set is consisted of the daily returns of the leading indices of the Belgrade Stock Exchange
(Serbia), Sarajevo Stock Exchange (Bosnia and Herzegovina), Zagreb Stock Exchange (Croatia), Macedonian
Stock Exchange (Macedonia) and Bulgarian Stock Exchange (Bulgaria): BELEX15, BIFX, CROBEX, MBI10
and SOFIX, respectively.
The daily returns are continuously compounded. They are calculated as
R
t
= ln
P
t
P
t1
,
where the variable P
t
denotes the closing stock price, while P
t1
expresses the closing stock price with one lag.
The closing stock prices are taken from the official stock market websites.

Table 1. Descriptive statistics
Index Number of
observations
(trading days)
Minimum daily
return, in %
Maximum daily
return, in %
Mean daily
return, in %
Standard
deviation of
daily return
BELEX15 1511 -10.86 12.16 -0.0499 1.65371
BIFX 1501 -8.78 7.57 -0.0635 1.15435
CROBEX 1496 -10.76 14.78 -0.0092 1.55784
MBI10 1477 -10.28 6.66 -0.0101 1.58988
SOFIX 1491 -11.36 7.29 -0.0631 1.48082

Descriptive statistics of the analyzed indexes are provided in the Table 1. The number of observations
which is equal to number of trading days in each stock market varies from 1477 in Macedonia to 1511 in Serbia.
The time span is from the fist trading days in 2006 (2nd J anuary for CROBEX, 3rd J anuary for SOFIX, 4th
J anuary for BIFX and MBI10, and 9th J anuary for BELEX15) to the last trading days in 2011 (29th December
for MBI10 and 30th December for the other indices). The analyzed period is characterized by the bear market
which starts from the middle of 2007 in all markets.
22
Annals of the Constantin Brncui University of Trgu Jiu, Economy Series, Issue 3/2012

ACADEMICA BRNCUI PUBLISHER, ISSN 1844 7007


The mean daily returns of all five indices are negative. The reason for this is the bear market that started
from the 2007 in the all analyzed markets as the result of the Global financial crisis. The highest standard
deviation of daily returns is observed in BELEX15 (1.65371) and lowest in BIFX (1.15435). Minimum daily
return of -11.36% is observed in SOFIX and maximum daily return of 14.78% in CROBEX.


5. Empirical Results

The results of the regressions are reported in table 2. The all estimated intercept coefficients (
1
) are
negative, which imply that mean daily return of the five stock indexes on Monday is negative. The weekend
effect is found only in Croatian and Bulgarian Stock Market. For the CROBEX index, the Wald test F-statistic is
significant at 5% level and all estimated dummy variables are positive and statistically significant, which imply
that mean daily return on Monday is lower than the mean returns in the rest of the days. So, the mean daily return
in Tuesday is equal to 0.0281% (-0.2654+0.2935), in Wednesday 0.0673% (-0.2654+0.3326), in Thursday
0.0777% (-0.2654+0.3431) and on Friday 0.0428% (-0.2654+0.3081). The results imply that for this index only
mean daily return on Monday is negative, while for other days in the week is positive. For the SOFIX index, the
Wald test F-statistic is significant at 10% level and only dummy variable for Friday is significant which leads to
conclusion that mean daily return on Friday is 0.2408% higher than the mean daily return on Monday. The mean
daily return of the MBI10 index on Monday is lower than the other days in the week, but the results are not
statistically significant. For the BIFX index, Tuesday and Thursday are found to have lower mean return than the
Monday, but also the results are not statistically significant. The BELEX15 index have also Tuesday mean daily
return lower than Monday and not statistically significant, where only the dummy variable for Thursday is
significant which point that mean daily return in Thursday are higher than Monday.

Table 2. Estimated results of the regression models
Index
1

2

3

4

5
Wald test
F-statistic
BELEX15 -0.139379
(0.1461)
-0.083148
(0.5370)
0.118105
(0.3802)
0.231493*
(0.0855)
0.180584
(0.1830)
1.8649
(0.1141)
BIFX -0.081515
(0.2256)
-0.089125
(0.3480)
0.041077
(0.6630)
-0.020148
(0.8310)
0.155272
(0.1002)
1.8437
(0.1180)
CROBEX -0.265381***
(0.0034)
0.293529**
(0.0209)
0.332634***
(0.0088)
0.343116***
(0.0072)
0.308140**
(0.0163)
2.5525**
(0.0375)
MBI10 -0.113245
(0.2264)
0.071995
(0.5843)
0.150294
(0.2515)
0.111784
(0.3944)
0.179348
(0.1742)
0.5632
(0.6857)
SOFIX -0.172478**
(0.0450)
-0.033540
(0.7820)
0.191085
(0.1141)
0.149625
(0.2170)
0.240845**
(0.0485)
1.96688*
(0.0968)
Notes: p-values are reported in brackets; *, **, *** denote significance at 0.1, 0.05 and 0.01, respectively.



6. Conclusion

The goal of this paper was to examine the presence of the weekend effect in the five stock markets from
SEE. This calendar anomaly has two characteristics: (1) the mean returns on Monday are lower than the other
trading days in the week, and (2) the mean returns on Monday are negative. There is no common explanation for
the weekend effect. One explanation is that the investors have the tendency to sell on Monday after the revision
of their portfolios during the weekend. Other that that the investors closed the short positions (buy) on Fridays
and open them again on Mondays (sell).
The weekend effect is well documented for the stock markets of developed countries. However there are
only few studies for SEE region. This paper analyzed daily returns of the BELEX15, BIFX, CROBEX, MBI10
and SOFIX. They are leading indices of the Belgrade Stock Exchange (Serbia), Sarajevo Stock Exchange
(Bosnia and Herzegovina), Zagreb Stock Exchange (Croatia), Macedonian Stock Exchange (Macedonia) and
Bulgarian Stock Exchange (Bulgaria), respectively. The analyzed period is from the first trading day of 2006 to
the last trading day in 2011 (it depends from the stock exchange).
The analysis is based on the regression with dummy variables, or so called Analysis of Variance
(ANOVA) model. It is often used to compare the difference in the mean values of two or more groups or
categories. In this case groups are different trading days in the week.
The weekend effect is found only in Croatian and Bulgarian Stock Market. The mean daily return of the
23
Annals of the Constantin Brncui University of Trgu Jiu, Economy Series, Issue 3/2012

ACADEMICA BRNCUI PUBLISHER, ISSN 1844 7007


MBI10 index on Monday is lower than the other days in the week, but the results are not statistically significant.
The mean daily returns in Tuesday of BELEX15 and BIFX indices are lower than mean daily return on Monday,
but also without statistical significance. In addition, it is found that mean daily return on Monday of the all five
stock indices is negative. The investors could design trading strategies which take advantage of the documented
calendar effect patterns.
The future research may extend the scope of examination. First, the analysis to be extended to other
calendar anomalies: the half month effect, the turn of the month effect, the month of the year effect, or holiday
effect. Second, the time span of the indices to be enlarged and to go from the start date of the leading SEE
indices to the present. Third, the calendar anomalies to be analyzed in bull market and bear market separately.


7. References

[1] Agrawal, A., and K. Tandon., Anomalies or Illusions? Evidence from Stock Markets in Eighteen
Countries. J ournal of International Money and Finance 13, 1994;
[2] Chang, E., J. Pinegar and R. Ravichandran, International Evidence on the Robustness of the Day of
the Week Effect, J ournal of Financial and Quantitative Analysis, 28, 1993;
[3] Cross, F, The behavior of stock prices on Fridays and Mondays. Financial Analysts J ournal, 29(6), 1973.
[4] Dyl, E. and Maberly, E., "The Weekly Pattern in Stock Index Futures: A Further Note", J ournal of
Finance, 41 No:5, 1986.
[5] French, K., Stock Returns and the Weekend Effect. J ournal of Financial Economics,8(1), 1980.
[6] Georgantopoulos, A. and Tsamis, A., A Comparative Study on Calendar Effects: Greece vs Bulgaria.
International Journal of Economic Research, Forthcoming, 2012.
[7] Georgantopoulos, A., Kenourgios D., Tsamis A., Calendar Anomalies in Emerging Balkan Equity
Markets, International Economics and Finance Journal, Vol.6, No.1, 2011;
[8] Georgantopoulos, A., Tsamis A,. Investigating Seasonal Patterns in Developing Countries: The Case of
FYROM Stock Market, International Journal of Economics and Financial Issues, Vol.1, No.4, 2011;
[9] Gibbons, M. R., & Hess, P., Day of the Week Effects and Asset Returns. The J ournal of Business, 54(4),
1981;
[10] Gujarai, D., Basic Econometrics. McGraw Hill, 2004.
[11] Karadzic, V., Backovic Vulic T., The Montenegrin Capital Market: Calendar Anomalies, Economic
Annals, Vol. LVI, No.191, 2011.
[12] Rogalski, R.J. New findings regarding day of the week returns over trading and nontrading periods: A
note. J ournal of Finance 39, 1994.
[13] Rubinstein, M. Rational Markets: Yes or No? The Affirmative Case, Financial Analysts J ournal, 57,
2001.
[14] Schwert, G., Anomalies and Market Efficiency, in G. Constantinides et al., Handbook of the
Economics of Finance, North Holland, Amsterdam, 2001.
[15] Steely, J., A Note on Information Seasonality and the Disappearance of the weekend Effect in UK Stock
Market, J ournal of Banking and Finance, 25, 2001
[16] Sullivan, R., A. Timmermann, and H. White, Dangers of Data mining: The Case of Calendar Effects
in Stock Returns, J ournal of Econometrics, 105, 2001.
[17] Taylor, S. Financial Times Series, World Scientific, 2008.


24