Aziz 2017

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Article

The Turn of the Month Effect in Global Business Review


19(1) 1–13
Asia-Pacific Markets: New Evidence © 2017 IMI
SAGE Publications
sagepub.in/home.nav
DOI: 10.1177/0972150917713370
http://gbr.sagepub.com

Tariq Aziz1
Valeed Ahmad Ansari1

Abstract
A predictable pattern in equity returns based on the calendar time is dubbed as calendar anomaly.
The prevalence of calendar anomalies is considered evidence against the efficient market hypothesis.
This article examines one of the most important calendar anomalies, the turn-of-the-month (TOM)
effect, in 12 major Asia-Pacific markets during the period January 2000 to April 2015, using both para-
metric and non-parametric tests. Under investigation, 11 out of 12 markets exhibit significant TOM
effects that are independent of the turn-of-the-year (TOY) effect. Moreover, these effects are not
present during the period of financial crisis. The persistence of the TOM effect in these markets, even
after a quarter of a century of its initial reporting, is a puzzle which needs an explanation.

Keywords
The turn-of-the-month effect, calendar anomalies, market efficiency, Asia-Pacific markets

Introduction
There is considerable evidence in support of seasonality in stock returns around the world. The January
and the weekend effects are the most extensively researched empirical regularities. Among the calendar
anomalies, the turn-of-the-month (TOM) effect, despite being significantly present in most of the markets
has received relatively less attention. The TOM effect has been known at least since Ariel (1987), who
reported for the US market that stock returns surge around the TOM that encompasses the end and start
of a month. This effect is found to be so persistent and profound that McConnell and Xu (2008) argue
that investors are compensated for bearing risk only around the TOMs.
A quarter of a century after Ariel (1987) first documented the presence of the TOM effect, it is still
persistent and significant in the majority of markets (Kunkel, Compton, & Beyer, 2003; McConnell &
Xu, 2008). Compton (2002) after analysing the TOM effect during 1988–1998 subscribe to the view that
while the TOM effect is fading in Canada and the United States, it is gaining strength in Pacific Rim
countries of Australia, Japan, Hong Kong and Singapore. The presence of a significant TOM effect raises

1
Department of Business Administration, Aligarh Muslim University, Aligarh, UP, India.

Corresponding author:
Tariq Aziz, Department of Business Administration, Aligarh Muslim University, Aligarh, 202002, UP, India.
E-mail: taziz.ba@amu.ac.in
2 Global Business Review 19(1)

questions about the market efficiency and rational framework and hence supports the behavioural finance
literature.
This article aims to update the evidence on the existence of the TOM effect in 12 major Asia-Pacific
markets. The study periods of previous studies warrant an update in this regard. The sample of Kunkel
et al. (2003) ends in 2000, and the sample period of McConnell and Xu (2008) ends in 2005. On the other
hand, the study period of Compton (2002) for the four Pacific Rim countries (Australia, Hong Kong,
Singapore and Japan) ends in 1998. We intend to provide an updated evidence for the Asia-Pacific markets
of Australia, China, Hong Kong, India, Indonesia, Japan, Malaysia, Pakistan, the Philippines, Singapore,
South Korea and Taiwan. Analysing these markets for recent evidence allows us to examine the argu-
ment of Agarwal and Tandon (1994) and Schwert (2003) that anomalies disappear after they become
well known post their publication.
In order to examine the TOM effect, this study employs both the parametric and non-parametric tests.
In the majority of cases, both tests support the presence of the TOM effect. Under investigation, 11 out
of 12 markets show a significant TOM effect during the period January 2000 to April 2015. To distin-
guish the TOM effect from another well-known anomaly, turn-of-the-year (TOY) effect or the January
effect, we exclude all the TOM intervals around the turn of the year and test for the presence of the TOM
effect. The results, however, confirm that the TOM effect exists independent of the TOY effect. We also
examine the impact of the financial crisis on the TOM effect specifically. We divide our sample based on
the period of financial crisis. The results reveal that the TOM returns are not statistically significant
during the period of financial crisis.
The rest of the article is organized as follows. The second section provides the literature review;
objectives and rationale of the study are outlined in the third and fourth sections, respectively. The fifth
section describes the data and methodology employed. The sixth section contains the empirical results,
and the article concludes in the seventh section.

Review of Literature
Ariel (1987, 1988) is the first to show in the US stock index data that the returns on the nine-day period
between the last day of the month and the eighth day of the following month are significantly higher than
the rest of the month (ROM) during the period 1963–1981. And the returns on the days other than the
nine days are close to zero. Another early study to report the TOM effect is Lakonishok and Smidt
(1988). Taking Dow Jones Industrial Average (DJIA) Index data for the period 1897–1986, Lakonishok
and Smidt (1988) documented that the average stock return is significantly higher on the last trading day
of the month and the initial three days of the following month (–1, +3). However, they did not analyse
the TOM effect in detail. Jacobs and Levy (1988) also reported that the TOM effect was present in the
US stock markets during the period 1897–1986.
Following the tradition in the literature, studies investigated the presence of the TOM effect in non-US
markets. Most of the studies confirmed the existence of a significant TOM effect in the international
markets (Agarwal & Tandon, 1994; Cadsby & Ratner, 1992; Kunkel, et al., 2003; Martikainen, Perttunen,
& Puttonen, 1995; McConnell & Xu, 2008; Ziemba, 1991). Ziemba (1991) reported a significant TOM
effect in the Japanese stock market for the duration of –5 to +2, during the period 1949 to 1988. Cadsby
and Ratner (1992) investigated the TOM effect in 10 international markets and found a significant TOM
effect in 6 markets (for a duration of –1 to +3) during the period 1962 and 1989. Martikainen et al. (1995)
found a significant TOM effect in the Finnish stock market during the period 1988–1993 for a duration
of –5 to –1. They also showed that the TOM effect is present both in the spot market index, and its
corresponding futures and options, and the derivatives appear to lead the spot market.
Aziz and Ansari 3

Agrawal and Tandon (1994) examined the TOM effect in 18 developed markets and found that the
returns are unusually high at the last trading day of the month and around the turn of the month (–4, +4)
in most of the markets during the period 1971–1987. Ziemba (1994) reviewed the TOM literature and
concluded that the TOM effect is global and not merely the result of data snooping.
Kunkel et al. (2003) examined the TOM effect in 19 markets and found a significant TOM effect in
16 markets during the period 1988–2000. Furthermore, they observed that the 87 per cent of the monthly
returns is concentrated over a TOM period of four days. Additionally, they find no evidence of a possible
spillover effect from the US stock market.
McConnell and Xu (2008) find that investors are rewarded for bearing market risk only at the TOMs.
Specifically, the mean daily return on the value-weighted (VW) CRSP market index for an interval of –1
to +3 is 0.15 per cent, while it is almost zero for the ROM during the period 1987–2005. Moreover,
they find that the TOM effect is present in 31 out of the 35 countries they examined. This is intriguing,
since the TOM effect was first reported in 1987 by Ariel it was supposed to vanish in the light of the
Schwert’s (2003) argument that most of the anomalies either vanished or their effects attenuated after
their publication.
In a latest study, McGuinness and Harris (2011) find a significant TOM effect in the stock markets of
Shanghai, Shenzhen and Hong Kong during the period 1995–2010. In the Indian context, Freund, Jain
and Puri (2007) confirmed the presence of the TOM effect during the period 1992–2004.
Some studies have also explored the exploitability of the TOM effect. Hansel and Ziemba (1996)
examined a TOM-based strategy and find that it outperforms the S&P 500 index by 0.64 per cent per year
during the period 1928–1993. Zwergel (2010) examined the presence of the TOM effect, their exploita-
bility in the indices and their corresponding futures of four major international markets—the United
States, the United Kingdom, Germany and Japan—and showed that a TOM-based trading strategy
outperforms a buy-and-hold strategy in the index futures.
Sharma and Narayan (2014) report that the effect of the TOM is heterogeneous depending upon
the sector and the size of the firm on 560 firms listed on the NYSE during the period 2000–2008. They
concluded that the TOM effect is present in all of the 14 sectors they considered, and it is higher among
small firms.
Several hypotheses have been advanced to explain the TOM effect. Ogden (1990) suggests that the
TOM effect is due to the increased liquidity because of the payment time of wages, interests, dividends
and principal payments at the month end. Ziemba (1991) endorses the same view by showing that the
TOM effect in Japan is between the 20th and 25th day of the month, which also happens to be the salary
payment day in Japan. Following this argument, the main driver for the TOM effect is the individual
investors. Booth, Kallunki and Martikainen (2001) examine the hypothesis proposed by Ogden (1990)
and Ziemba (1991) that increased demand pressure from individual investors causes the TOM effect in
the Finnish market and find supportive evidence. Specifically, their results are based on the returns of
148 stocks during the period 1991 to 1997 and their standardized trading volumes. Through parametric
and non-parametric tests, they find that the trading volumes are significantly higher over the TOM
duration relative to the ROM volumes. Marquering, Nisser and Valla (2006), however, attribute the per-
sistence of the TOM effect to the high transaction cost that may have prevented investors to exploit it.
Another explanation is provided by Thaler (1987) that institutions sell loss-making stocks and buy
profit-making stocks in order to appear good in the eyes of the shareholders and clients (window dressing).
Barone (1990) also presents a similar hypothesis that states that institutional investors rebalance their
portfolios in order to boost the performance indicators that are generally published at the end of the
month. This argument is termed as ‘window dressing hypothesis’ or the ‘portfolio rebalancing hypothesis’.
One more explanation is proposed by Nikkinen, Sahlstrom and Aijo (2007). They contend that important
macroeconomic news announcements are scheduled at the end of the month that induce the TOM effect
4 Global Business Review 19(1)

in return and volatility. It can be concluded that most of the studies have shown a significant TOM effect.
A lapse of time, however, raises the need to provide fresh evidence.

Objective of the Study


As discussed in the literature review, it is apparent that the TOM effect existed in most of the markets
around the world. A lapse of time, however, raises the need to provide fresh evidence. In this regard, this
study aims to update the evidence on the TOM effect in the context of the Asia-Pacific markets using
both parametric and non-parametric approaches. Furthermore, there is scanty evidence regarding it
in some selected markets. For example, in the Indian context, there are only two studies which have
examined the TOM effect (Freund, Jain, & Puri, 2007; Maher & Parikh, 2013).

Rationale of the Study


This study asks the question whether the TOM effect still exists or it has vanished due to the increased
market efficiency and awareness of market participants over time. Markets are ever evolving and
dynamic, and a stylized fact which existed at a time may die subsequently as argued by Schwert (2003).
As many anomalies got eliminated or decayed over time post their publication. Moreover, investigating
different sample markets and periods counter the data snooping bias of Lo and MacKinlay (1990).

Data and Methodology


We obtain daily closing prices of most representative stock market indices, of 12 Asia-Pacific markets,
Australia, China, Hong Kong, India, Indonesia, Japan, Malaysia, Pakistan, the Philippines, Singapore,
South Korea and Taiwan for the period 1 January 2000 to 30 April 2015. The data are obtained from
www.finance.yahoo.com. The countries, indices and the descriptive statistics of the daily returns are
reported in Table 1. Daily index returns are computed as the log difference of the closing prices:
p it
R it = ln c p m(1)
i (t - 1)

The number of returns ranges between 3,960 in the Philippines and 3,715 in Indonesia. The highest mean
daily return is in Pakistan at 0.0833 per cent, and the lowest mean daily return is provided by Japan at
0.0007 per cent. South Korean returns exhibit the highest volatility, and the Malaysian returns are least
volatile. As expected, Table 1 shows that daily market returns are skewed and exhibit excess kurtosis.
All the daily returns are negatively skewed except the returns of the Philippines, which implies that there
is more probability of a negative return than required by a symmetric distribution.
Table 1 shows the countries, indices and the descriptive statistics of the sample for the period 1
January 2000 to 30 April 2015.
Numerous studies have shown that financial returns data is leptokurtic, asymmetric and often display
volatility clusters. In such case, Connolly (1989) suggests the use of generalized autoregressive conditional
heteroscedasticity models (GARCH) that treat the fat tails and volatility clustering parsimoniously. Since
Bollerslev (1986), numerous variants within the GARCH family have been proposed in the literature.
Hansen and Lunde (2005), however, find that more sophisticated models do not outperform the GARCH
(1, 1) model. Therefore, this model has been used in this study.
Aziz and Ansari 5

Table 1. Descriptive Statistics of Daily Returns on Indices for the Period 1 January 2000 to 30 April 2015

Country (Index) Obs. Mean (%) Std. Dev. (%) Min Max Skew Kurtosis
Australia (All 3,884 0.0158 0.9743 –8.5536 5.3601 –0.5922 9.4762
ordinaries)
China (Shanghai 3,888 0.0303 1.5424 –9.2562 9.4008 –0.1289 7.6867
Composite)
Hong Kong (Hang 3,825 0.0126 1.5347 –13.5820 13.4068 –0.0713 11.1492
Seng)
India (BSE30 3,792 0.0425 1.5709 –11.8092 15.9900 –0.1871 9.8362
Sensex)
Indonesia (Jakarta 3,715 0.0535 1.4314 –10.9540 7.6231 –0.6882 9.2770
Composite)
Japan (Nikkie 225) 3,775 0.0007 1.5466 –12.1110 13.2346 –0.4153 9.2469
Malaysia (KLSE 3,782 0.0206 0.9707 –15.5682 16.0204 –0.5202 61.3639
Composite)
Pakistan (KSE100) 3,770 0.0833 1.4040 –7.7414 8.5071 –0.2639 6.3375
the Philippines 3,960 0.0324 1.3076 –13.0887 16.1776 0.3283 19.1846
(PSEi Composite)
Singapore (Straits 3,873 0.0077 1.1812 –9.0950 7.5305 –0.3474 8.9903
Times)
South Korea 3,791 0.0184 1.6286 –12.8047 11.2844 –0.5691 8.8544
(Seoul Composite)
Taiwan (Taiwan 3,797 0.0030 1.4455 –9.9360 6.5246 –0.2392 6.0537
Weighted)
Source: Authors’ own findings.

GARCH models are parametric specifications that require assumptions about the empirical distribu-
tions. However, the choice of distribution is often a matter of trial and error. Use of parametric approaches
in stock return seasonality studies has been questioned because of the non-adherence of returns to any
particular distribution. In this situation, many studies have employed non-parametric tests along with
parametric procedures to cater for this concern (Agarwal & Tandon, 1994; Kunkel et al., 2003; M. N.
Gultekin & N. B. Gultekin, 1983). Some studies have reported that non-parametric approaches are even
more powerful than the parametric tests (Hunter & May, 1993). Consequently, we also provide the results
from Wilcoxon Singed Rank (WSR) test which is a non-parametric test of difference. The null hypoth-
esis in WSR test is that the average TOM returns are not statistically different than the average returns
on the ROM in terms of their distribution.
There is no consensus regarding the precise TOM period. To arrive at the precise TOM interval, we
follow McConnell and Xu (2008) and plot daily returns over the interval [–7, +7], where –1 stands for
the last trading day of the month and +1 denotes the first trading day of the month. McConnell and Xu
(2008), however, plot returns for the interval [–10, +10]. Since various months in our sample contain
fewer observations, we resort to this interval [–7, +7].
The time series GARCH (1, 1) regression model to test the TOM effect is of the following form:

R t = a + bD TOM + f t(2)

where Rt is the index return at time t; DTOM is the dummy variable for the chosen interval (i.e., DTOM = 1
if observation t falls on the TOM interval of the month and zero otherwise); α and β are the parameters
6 Global Business Review 19(1)

to be estimated and εt is an error term. A significant β in Equation 2 is taken as evidence of the TOM
effect. The variance equation associated with Equation 2 is:

h t = var (f t) = c + an 2t - 1 + bh t - 1(2a)

f t | f t - 1 . N (0, v 2t )(2b)

We report only the estimates of the mean equation to conserve space.

Results and Analysis


To arrive at the precise duration of the TOM period for each country, we plot average index returns by
the day of the month over the interval [–7, +7]. Appendix A contains the plots of the average daily returns
by the day of the month for the 12 indices. It is apparent in the graphs that the returns experience a surge
around the turn of the month. However, the TOM interval differs in each country.
Table 2 shows the average daily returns by the day of the month for various indices. Daily returns are
the log difference of the closing prices (expressed in percentage). Day –1 denotes the last trading day
of the month and Day 1 represents the first day of the month. The last column presents the daily returns
on the ROM (returns on other than these 14 days). The sample period is January 2000 to April 2015.
Numbers in bold denote significance at the 5 per cent level or better.
Table 2 displays the mean daily returns by the day of the month for the interval [–7, +7]. A significant
return at the 5 per cent level or better is indicated in bold. It is noticeable in Table 2 that significant
returns are clustered around the turn of the month. The magnitude of returns is also larger around the turn
of the month relative to other days. TOM studies generally define the period [–1, +3] as the TOM period
(e.g., Kunkel et al., 2003; McConnell & Xu, 2008). However, in Table 3 it is apparent that in 6 out of 12
markets mean daily returns on Day +3 is negative, which highlights the importance of defining the TOM
interval differently for each country.
The last column of Table 2 shows the mean daily returns on the ROM, days other than the interval
[–7, +7]. Out of 12 indices, 10 register negative returns on the ROM period, which bolsters the argument
that investors are compensated for the risk only around the TOM.
After establishing an apparent TOM effect in the indices, we turn to regression analysis. Table 3
shows the parameter estimates of Equation 2. All the estimates of β, which represents the returns on
TOM interval over and above the returns on the ROM, are significant at the 5 per cent level. Only four
countries (Australia, India, Indonesia and Pakistan) have a significant α, which denotes the returns on the
ROM. The estimate of β is highest for South Korea at 0.36, which implies that at an average the returns
on the TOM period are 0.36 per cent higher than the returns on the ROM period. The next high βs are
0.23 for India and 0.22 for China and Pakistan. The last column of Table 3 contains the TOM intervals
for each country we arrived at by plotting the returns by the day of the month in Table 2 and also through
visual aid in Appendix A. The longest TOM interval (6 days) is in Australia and Hong Kong [–4, +2] and
the shortest TOM interval (2 days) is in South Korea [–1, +1].
Table 3 also shows the WSR test statistics. GARCH (1, 1) and WSR tests agree on all the instances
except for Malaysia, where WSR test does not reject the null hypothesis that returns on the TOM period
are not different from the returns on the ROM period.
GARCH (1, 1) estimates of the model R t = a + bD TOM + f t are presented in this table, where Rt is
the log return at time t (expressed in percentage); DTOM is the dummy variable for the chosen interval
Table 2. Returns Around the Turn of the Month

Trading Day of the Month


Country –7 –6 –5 –4 –3 –2 –1 1 2 3 4 5 6 7 ROM
Australia –0.0954 –0.0079 0.0020 0.1911 0.0665 0.0680 0.1201 0.0725 0.1094 –0.0649 –0.0258 –0.0080 0.0243 –0.0583 –0.0083
China 0.0348 0.0828 –0.1882 –0.0014 –0.0552 –0.0213 0.0689 0.2480 0.1782 0.2106 0.0599 0.0889 –0.0528 0.0480 –0.0082
Hong Kong 0.0063 –0.1866 0.0486 0.1212 0.0635 0.1749 0.1990 0.2420 0.1530 –0.0986 0.1106 –0.0481 0.0500 –0.0850 –0.0717
India –0.0598 –0.1749 0.1333 0.0928 0.0709 0.0850 0.3649 0.3554 0.2219 –0.0111 0.0931 0.0318 –0.0327 0.0290 –0.0483
Indonesia 0.0371 –0.0655 –0.0397 –0.1279 0.0662 0.2662 0.3384 0.0285 0.1887 –0.0601 0.2660 0.0612 0.0226 0.0911 –0.0024
Japan 0.0235 –0.1085 0.0622 0.1165 0.0668 0.1493 –0.0004 0.1656 0.1438 –0.1567 0.0154 –0.0855 –0.2459 –0.0107 –0.0107
Malaysia –0.0142 0.0703 0.0088 0.0574 0.0267 0.0343 0.0868 –0.00415 0.0514 0.0267 0.0241 0.0937 –0.0588 0.1498 –0.0197
Pakistan –0.1250 0.0563 –0.2012 –0.0007 –0.1063 0.0806 0.1527 0.0573 0.5657 0.2861 0.2340 0.1720 0.1333 0.0783 0.0500
the 0.0020 –0.0936 –0.1104 0.1005 0.0361 0.1458 0.1834 0.0921 0.2251 0.1890 0.1669 0.1096 0.0000 –0.0942 –0.0341
Philippines
Singapore –0.0296 –0.0539 0.0601 0.0177 0.0681 0.1401 0.1024 0.1689 0.2392 –0.0566 0.0440 0.0225 –0.1574 0.0745 –0.0678
South –0.1922 –0.0677 0.1583 –0.1362 –0.0304 0.0364 0.3258 0.3636 0.0415 0.0289 –0.0172 –0.0417 –0.0279 –0.0884 0.0042
Korea
Taiwan –0.0510 0.0658 0.0604 –0.0768 –0.1568 0.0839 0.2923 0.0473 0.1884 0.0016 0.0745 0.0757 0.0070 0.0534 –0.0909
Source: Authors’ own findings.
8 Global Business Review 19(1)

Table 3. Tests for the TOM Effect


Country a b WSR Test Value TOM Interval
Australia 0.0339 0.0703 3.44 –4, +2
China –0.0036 0.2298 3.17 +1, +3
Hong Kong 0.0098 0.1302 3.11 –4, +2
India 0.0602 0.2317 5.51 –1, +2
Indonesia 0.0700 0.1604 3.77 –2, +2
Japan 0.0327 0.1657 2.02 +1, +2
Malaysia –0.0228 0.1288 1.31 –4, –1
Pakistan 0.0894 0.2283 5.27 +2, +5
the Philippines 0.0097 0.2115 4.59 –2, +5
Singapore 0.0151 0.1464 4.03 –2, +2
South Korea –0.0166 0.3613 4.39 –1, +1
Taiwan 0.0153 0.1718 3.72 –2, +2
Source: Authors’ own findings.

(i.e., DTOM = 1 if observation t falls on the TOM interval of the month and zero otherwise); α and β are
the parameters to be estimated and εt is an error term. The last column contains the unique TOM interval
for each country. This table also shows the WSR test statistics, which tests the null that TOM returns are
not different from the ROM returns. The sample period is 1 January 2000 to 30 April 2015. Numbers in
bold denote significance at the 5 per cent level or better.
Our sample period also encompasses the period of global financial crisis. Next, we examine the
impact of financial crisis on the TOM effect. Specifically, we divide the sample period into three parts,
pre-crisis, crisis and post-crisis. The crisis period is defined as the period between January 2008 and
March 2009. This period is generally regarded as the period of financial crisis.1 This classification allows
us to examine if the TOM effect is stable across the rising and falling markets.
Table 4 presents the results of the sub-periods. During the crisis period, the estimates of β are signifi-
cant only for two countries (China and the Philippines). This suggests that in the majority of the markets,
the TOM returns are not statistically different from the returns of the ROM period. A possible reason for
the absence of the TOM effect during the crisis period could be the herd behaviour of investors previ-
ously documented in the literature (Bikhchandani & Sharma, 2000; Chiang & Dazhi, 2010). The herding
behaviour is associated with the tendency of the market to rally or sell off in large unsubstantiated
manner. Since during the crisis period market participants were behaving like a herd, the TOM returns
are non-distinguishable from the ROM period returns.
During the recent sub-period (April 2009 to April 2015), the coefficient of TOM is significant in 8 out
of 12 markets. The remaining four markets have insignificant estimates of β, although positive. In the
first sub-period (January 2000 to December 2007), only Malaysia has an insignificant TOM coefficient,
and all the other markets under investigation have a significant β. It can be concluded that during the
financial crisis the TOM effect is not statistically significant. This is in conformity with the findings of
Maher and Parikh (2013) who find a similar result in the context of the Indian stock market.
GARCH (1, 1) estimates of the model R t = a + bD TOM + f t are presented in this table, where Rt is the
log return at time t (expressed in percentage); DTOM is the dummy variable for the chosen interval
(i.e., DTOM = 1 if observation t falls on the TOM interval of the month and zero otherwise); α and β are
the parameters to be estimated and εt is an error term. The results are provided for three sub-periods: pre-
crisis, crisis and post-crisis. Pre-crisis period is from 1 January 2000 to 31 December 2007; crisis period
is from 1 January 2008 to 31 March 2009 and post-crisis period is from 1 April 2009 to 30 April 2015.
Aziz and Ansari 9

Table 4. Tests for the TOM Effect (Sub-periods)

Pre-crisis Crisis Post-crisis


Country b WSR Value b WSR Value b WSR Value
Australia 0.0885 3.25 0.1596 1.56 0.0339 1.38
China 0.1610 1.31 0.8059 0.38 0.2769 3.11
Hong Kong 0.1228 2.02 0.3583 1.74 0.1303 1.79
India 0.2730 4.62 0.1642 1.13 0.1729 3.13
Indonesia 0.1494 2.36 0.1496 0.48 0.1376 3.20
Japan 0.2858 3.30 –0.5854 0.98 0.0535 0.27
Malaysia 0.0275 0.19 0.2514 1.57 0.1972 1.31
Pakistan 0.2020 2.94 0.6018 2.34 0.2102 4.09
the Philippines 0.2657 4.09 0.3567 0.80 0.1276 2.31
Singapore 0.2701 4.33 0.1379 0.18 0.0537 1.76
South Korea 0.4629 4.59 –0.2238 0.02 0.0355 1.47
Taiwan 0.2526 2.99 –0.0676 0.08 0.1219 2.69
Source: Authors’ own findings.

This table also shows the WSR test statistics, which tests the null that TOM returns are not different from
the ROM returns. Numbers in bold denote significance at the 5 per cent level or better.
There is also evidence of the existence of a turn of the year or January effect in the literature. The TOY
effect is related to a consistent surge in market-wide returns around the end and the start of the year.
Since a turn of the year is also a turn of the month, a study may wrongly reject the null in support of the
TOM effect when a TOY effect also exists. To avoid the confounding of the TOM effect and TOY effect,
we eliminate all the TOM periods around the turn of the year (the TOMs between December and January).
Table 5 shows the estimates of the Equation 2 after eliminating TOM intervals around the turns of the
year. If TOM effect exists independent of the TOY effect, then the estimates of β must remain statistically
significant after eliminating the turns of the years.

Table 5. Tests for the TOM Effect (Excluding the Turn of the Year)

Country a b WSR Test Value


Australia 0.0334 0.0626 2.91
China –0.0036 0.2365 2.95
Hong Kong 0.0099 0.1142 2.67
India 0.0597 0.2144 4.84
Indonesia 0.0716 0.1285 3.31
Japan 0.0329 0.1778 1.92
Malaysia –0.0247 0.1173 0.65
Pakistan 0.0870 0.2086 4.45
the Philippines 0.0061 0.2009 4.14
Singapore 0.0156 0.1350 3.69
South Korea 0.0459 0.1473 3.77
Taiwan 0.0149 0.1524 3.05
Source: Authors’ own findings.
10 Global Business Review 19(1)

The results in Table 5 indicate that the TOM effect exists independent of the TOY effect in 11 out of
12 markets under investigation. Except Malaysia, all the coefficient of TOM are statistically significant
at the 5 per cent level and both the parametric and non-parametric tests agree on this result.
Results are provided for Equation 2, after eliminating all the TOM intervals around the turn of the
year (between December and January). For other details, see Table 3.

Conclusion
In this article, we examined the presence of the TOM effect in 12 major Asia-Pacific equity markets
during the recent and hence the most relevant period. In light of the Schwert’s (2003) argument that
anomalies disappear once they become well known, this anomaly was supposed to disappear. Both the
parametric and non-parametric tests, however, confirm the presence of the TOM effect in 11 out of 12
markets under investigation during the period January 2000 to April 2015. Furthermore, the results
suggest that the TOM effect is independent of the TOY effect. Highest TOM effects are exhibited by the
stock markets of South Korea, India, China, Pakistan and the Philippines in that order. Moreover, the
results lend support to McConnell and Xu (2008) who argue that investors are compensated for bearing
risk only around the turns of the month. This view is reinforced by the finding that the mean returns on
ROM period (days excluding the –7, +7 interval) are negative in 10 out of 12 markets. The results also
indicate that the surge in returns around the TOMs is not statistically significant during the period of
financial crisis. We conjecture that this could be due to the herding behaviour of investors previously
documented in the literature.
The persistence of the TOM effect even after a quarter of a century after its initial reporting in these
markets raises questions about the market efficiency. Furthermore, the TOM effect can be examined in
style returns like momentum (Garg & Varshney, 2015). In addition, the sources of this regularity, however,
remain unexplored and thus provide avenues for future research.

Acknowledgements
The authors are grateful to the anonymous referees of the journal for their extremely useful suggestions to improve
the quality of the article. Usual disclaimers apply.

Note
1. Retrieved 24 July 2017, from http://www.imf.org/external/pubs/ft/weo/2009/01/

Appendix A
These figures show the returns by the days around the TOM for each country. Day –1 denotes the last
day of the month, and Day 1 denotes the first day of the month. ROM represents the return on the ROM.
Aziz and Ansari 11

Figure. Average market return by day of the month


Source: Authors’ own findings.
12 Global Business Review 19(1)

References
Agarwal, A., & Tandon, K. (1994). Anomalies or illusions? Evidence from stock markets in eighteen countries.
Journal of International Money and Finance, 13(1), 83–106.
Ariel, R.A. (1987). A monthly effect in stock returns. Journal of Financial Economics, 18(1), 161–174.
———. (1988). Evidence on intra-month seasonality in stock returns. In E. Dimson (Ed.), Stock market anomalies
(pp. 109–119). Cambridge: Cambridge University Press.
Barone, E. (1990). The Italian stock market: Efficiency and calendar anomalies. Journal of Banking and Finance,
14(2), 483–510.
Bikhchandani, S., & Sharma, S. (2000). Herd behavior in Financial Markets. IMF Staff Papers, 47(3), 279–310.
Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics, 31(3),
307–327.
Booth, G.G., Kallunki, J.P., & Martikainen, T. (2001). Liquidity and the turn-of-the-month effect: Evidence from
Finland. Journal of International Financial Markets, Institutions and Money, 11(2), 137–146.
Cadsby, C.B., & Ratner, M. (1992). Turn-of-month and pre-holiday effects on stock returns: Some international
evidence. Journal of Banking and Finance, 16(3), 497–509.
Chiang, T.C., & Dazhi, Z. (2010). An empirical analysis of herd behavior in global stock markets. Journal of Banking
and Finance, 34(8), 1911–1921.
Compton, W.S. (2002). The evolving turn-of-the-month effect: Evidence from Pacific Rim countries. Global Business
and Finance Review, 7(1), 27–38.
Connolly, R.A. (1989). An examination of the robustness of the weekend effect. Journal of Financial and
Quantitative Analysis, 24(2), 133–169.
Freund, S., Jain, R., & Puri, Y. (2007). The turn-of-the-month effects in stocks trading on the national stock exchange
of India. Journal of Emerging Markets, 12(2), 14–23.
Garg, A.K., & Varshney, P. (2015). Momentum effect in Indian stock market: A sectoral study. Global Business
Review, 16(3), 494–510.
Gultekin, M.N., & Gultekin, N.B. (1983). Stock market seasonality: International evidence. Journal of Financial
Economics, 12(4), 469–481.
Hansel, C.R., & Ziemba, W.T. (1996). Investment results from exploiting turn-of-the-month effects. Journal of
Portfolio Management, 22(3), 17–23.
Hansen, P.R., & Lunde, A. (2005). A forecast comparison of volatility models: Does anything beat a GARCH (1, 1)?
Journal of Applied Econometrics, 20(7), 873–889.
Hunter, M.A., & May, R.B. (1993). Some myths concerning parametric and nonparametric tests. Canadian
Psychology, 34(4), 384–389.
Jacobs, B.I., & Levy, K.N. (1988). Calendar anomalies: Abnormal returns at calendar turning points. Financial
Analysts Journal, 44(6), 28–39.
Kunkel, R.A., Compton, W.S., & Beyer, S. (2003). The turn-of-the-month effect still lives: The international
evidence. International Review of Financial Analysis, 12(2), 207–221.
Lakonishok, J., & Smidt, S. (1988). Are seasonal anomalies real? A ninety-year perspective. Review of Financial
Studies, 1(4), 403–425.
Lo, A.W., & Mackinlay, A.C. (1990). Data-snooping biases in tests of financial asset pricing models. Review of
Financial Studies, 3(3), 431–467.
Maher, D., & Parikh, A. (2013). The turn of the month effect in India: A case of large institutional trading pattern as
a source of higher liquidity. International Review of Financial Analysis, 28(c), 57–69.
Marquering, W., Nisser, J., & Valla, T. (2006). Disappearing anomalies: A dynamic analysis of the persistence of
anomalies. Applied Financial Economics, 16(4), 291–302.
Martikainen, T., Perttunen, J., & Puttonen, V. (1995). Finnish turn-of-the-month effects: Returns, volumes, and
implied volatility. Journal of Futures Markets, 15(6), 605–615.
McConnell, J.J., & Xu, W. (2008). Equity returns at the turn of the month. Financial Analysts Journal, 64(2), 49–64.
McGuinness, P.B., & Harris, R.D. (2011). Comparison of the ‘turn-of-the-month’ and lunar new year return effects in
three Chinese markets: Hong Kong, Shanghai and Shenzen. Applied Financial Economics, 21(13), 917–929.
Aziz and Ansari 13

Nikkinen, J., Sahlstrom, P., & Aijo, J. (2007). Turn-of-the-month and intramonth effects: Explanations from the
important macroeconomic news announcements. Journal of Futures Markets, 27(2), 105–126.
Ogden, J.P. (1990). Turn-of-month evaluations of liquid profits and sock returns: A common explanation for the
monthly and January effect. The Journal of Finance, 45(4), 1259–1272.
Schwert, G.W. (2003). Anomalies and market efficiency. In G. Constantinides, M. Harris, & R. Stulz (Eds),
Handbook of the economics of finance (pp. 939–974). New York, NY: North-Holland.
Sharma, S.S., & Narayan, P.K. (2014). New evidence on turn-of the-month effects. Journal of International
Financial Markets, Institutions and Money, 29(c), 92–108.
Thaler, R. (1987). Seasonal movements in security prices II: Weekend, holiday, turn of the month, and intraday
effects. The Journal of Economic Perspectives, 1(2), 169–177.
Ziemba, W.T. (1991). Japanese security market regularities: Monthly, turn-of-the-month and year, holiday and
golden week effects. Japan and the World Economy, 3(2), 119–146.
———. (1994). Worldwide security market regularities. European Journal of Operational Research, 74(2),
198–229.
Zwergel, B. (2010). On the exploitability of the turn-of-the-month effect: An international perspective. Applied
Financial Economics, 20(11), 911–922.

You might also like