Indonesian Capital Market Review 11 (2019) 15-29

The Profitability of Momentum Strategies:

A Study Of Indonesian Stock Exchange
Rakhmat Luthfiansyah Mosii* and Sigit S. Wibowo
Faculty of Economics and Business, Universitas Indonesia, Indonesia

(Received : November 2018/ Revised : July 2019/ Acceptance : August 2019/ Available Online : September 2019)

We investigate the profitability of style and price momentum strategy in the Indonesian stock mar-
ket from the year 2000 to 2015. We find the style momentum strategy yields significant positive returns
while the price momentum strategy tends to produce negative returns. This result confirms the find-
ings of Lewellen (2002) that style momentum returns are generally stronger than the conventional
momentum. The average monthly returns of the double-sorted size-B/M style momentum are ranging
from 1.98% to 2.64% and persistent after controlling for market factor using JSX index. Our findings
suggest investors should utilize publicly available information such as size and book-to-market ratio
on their investment decision in the Indonesian stock market.

Keyword: stock returns, momentum strategy, Indonesian stock exchange, efficient market hypothesis,
CAPM

JEL Classification: G11, G12, G14

Introduction to generate excess returns on average 12% per

year between 1965 and 1989. Furthermore, they
The efficient market concept has a strong argue this profitability was not due to the sys-
implication on the investment approach of in- tematic risk described by stock betas, in which
vestors and traders in the stock market. Regard- contradicts Fama’s efficient market hypothesis.
ing this matter, efficient market hypothesis ar- This anomaly has been massively examined
gues that under weak-form market efficiency, by financial researchers for the last 20 years as
future returns of stocks or portfolios should a prevalent anomaly in the asset pricing field
not be dependent upon its past returns. In other that can not be explained by Fama-French three
words, it is not possible securing profit consist- factors model of asset pricing (Chordia and
ently using historical trend. This conjecture is Shivakumar, 2006). For instance, Rouwenhorst
well-accepted in the field of financial research (1998) extends the work of JT by testing the
until Jegadeesh and Titman (1993) (afterward significance and profitability of momentum
will be abbreviated with JT) introduced the strategies in international markets and found
concept of financial anomaly named “momen- that the return continuation anomaly is present
tum strategy.” JT conjectured that the US stock in twelve European countries in 1978 to 1995
markets are not at the weak-form efficient state period. Also, this momentum returns are corre-
after they discovered a trading strategy that able lated with the momentum returns found in the

* Corresponding author’s email: rakhmat.luthfi@yahoo.co.id

15
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

US stock market; he suggests the possibility of an efficient market after the Asian crisis pe-
exposure from common risk factors drives the riod. They mentioned that Asian stock market
profitability of momentum strategies. was expanding in size and considered as high
Findings on the significance and the profit- growth and volatile region, while the previous
ability of momentum strategies; however, to results of studies in Asian stock market effi-
some extent, have been inconclusive. In some ciency are varied and distributed over differ-
market, researchers have discovered that mo- ent methods, research periods and type of data
mentum strategies yield inconsistent profit. Be- frequencies. Amanda and Husodo (2015) also
kaert et al. (1997) explore cross-sectional de- stated that Indonesia is an emerging country
terminants of investment strategies in emerging with growing stock market and positive trend of
markets (27 countries including Indonesia) giv- market returns and increasing trading volume
en that emerging market has unique characteris- from the year 2003 to 2013, while numbers of
tics that separate them from the developed mar- illiquid stocks are present. Similar results of the
ket such as high average returns, high volatility, form of the Indonesian capital market is found
and low correlation both across the emerging by Arshad et al. (2016). Using multifractal
markets and developed markets. They find that detrendend analysis, they studied eleven OIC
on emerging markets, the evidence for the mo- countries and find that the Indonesian capital
mentum variable is somewhat inconsistent. market also show weak form market efficiency.
Chui et al. (2000) examine the profitability The inconsistent findings of momentum in
of momentum strategies in eight Asian Coun- Indonesian stock market, along with the inter-
tries that consist of Hong Kong, Indonesia, esting characteristics of emerging market are
Japan, Korea, Malaysia, Singapore, Taiwan, essential aspects to further study the momen-
and Thailand. Their methods are similar to JT, tum effect to provide additional empirical evi-
Jegadeesh, and Titman (2001), Rouwenhorst dence of the presence of momentum profits in
(1998) and Rouwenhorst (1999). Different ap- an emerging market. JT, Rouwenhorst (1998),
proaches have been utilized to address the issue and Wu (2016) sorted momentum portfolios
of illiquidity of the smaller Asian stock. Hence by its past returns, while Lewellen (2002) and
they value-weight instead of equally-weight the Chan and Docherty (2015) formed the mo-
long and short positions of momentum portfo- mentum portfolios based on the past size and
lios. They also use different cut-offs, 30% rath- B/M ratio (style momentum strategy). We in-
er than the 10% cut-offs used by JT in light of vestigate the style momentum strategy since
smaller sample sizes in most Asian countries. previous literature has conjectured that style
The evidence indicates that momentum profits momentum able to generates stronger profits
present in all sample countries except for Indo- than the price momentum. Moreover, To best
nesia and Korea, and other countries, the pres- of our knowledge, there are only a few kinds
ence of momentum profit is generally weak. of literature that study the profitability of style
Consistent with the previous finding in the US momentum strategy, particularly in the Indone-
stock market, they also found that momentum sian stock market.
profits are relatively stronger for small-sized This research investigates the profitability of
firms, low B/M firms, and high turnover firms. momentum strategies in the Indonesian stock
Kim and Shamsuddin (2008) test the stock market using price and style momentum. To
markets efficiency in nine Asian countries by study this problem, our main research question
applying multiple variance ratio tests based on is to what extent does the profitability of price
wild bootstrap and signs. They find that Hong- and style momentum exist in the Indonesian
kong, Japan, Korea, and Taiwan stock market stock market. The rest of this paper is organized
are efficient in the weak-form, while Indone- as follows. Section 2 provides related literature
sia, Malaysia, and the Philippines stock market to this study. Section 3 presents the research
have no sign of market efficiency. Furthermore, methods. Section 4 discusses the empirical re-
Singapore and Thailand have developed into sults. The last section concludes.

16
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

Literature Review In 1993, JT introduced the momentum strat-

egy. They mentioned that the return reversals
The momentum phenomenon in stock re- anomaly is still being debated since some of
turns is growing evidence that stock market the researchers argued that their results could
deviates from the efficient market hypothesis be explained using the systematic risk of their
(EMH). EMH was developed by Fama in early portfolios and the size effect. In contrast, they
1965 and has a root in the random walk and found another anomaly that seems to deviate
capital asset pricing model. EMH starts from from the weak-form market efficient condition.
the random walk hypothesis, that is the succes- A series of robustness tests conducted to test the
sive price changes are independent one from persistence of the anomaly, with the result of it
the other. Fama then expands it into market ef- is indeed persistent and could not be explained
ficient states: weak form, semi-strong form, and using systematic risk measure. Utilizing the
strong form efficient market. The test of these stocks listed in NYSE and AMEX from 1965 to
market conditions is under the presumption that 1989, they find that a relative strength strategy
market equilibrium can be described in expect- that buys winners and selling losers portfolio
ed returns similar to the model of Sharpe and can generate consistent profits. They argue if
Lintner. Fama also explained the joint-hypoth- stock prices are either overreact or underreact
esis problem in proving the EMH. Hence, it is to information, then it will be possible to gain
impossible to reject the hypothesis. The two profits using strategies that select stocks based
most prevalent evidence on market inefficiency on their past returns. The relative strength strat-
is the return reversal (contrarian strategy) and egies is a set of portfolios that buys stocks with
the return continuation (momentum strategy), the highest past returns (from 3 to 12 month
which will be discussed in this section. We are lagged period) and then hold it for 3 to 12 month
aware of the presence of other anomalies aside after the formation period. The most successful
from the two mentioned earlier such as the sea- relative strength strategy yields 1.31% in profits
sonality effect (month of the year, week of the per month.
month or day of the week effect) could be the JT reveals that the momentum profits are
possible sources for market inefficiency. We due to the positive estimates of the autocovari-
will not discuss those matters and only focus on ance of market model residuals for individual
the contrarian and momentum strategy. stocks. This suggests that stocks prices indeed
De Bondt and Thaler (1985) discover a sub- underreact to firm-specific information and vio-
stantial weak form market inefficiencies by em- lates the efficient market hypothesis. Rouwen-
pirically testing the overreaction hypothesis. horst (1998) tests the momentum anomaly in
Overreaction hypothesis predicts stocks that the international context covering 2,190 stocks
went through extreme returns realizations; the from 12 European countries to provides out of
future price reversals will be more (less) rec- sample evidence on this particular anomaly. He
ognizable. The findings are consistent with the finds that an internationally diversified relative
hypothesis, that the loser portfolios outperform strength strategy that is going long on medium-
winner portfolios regardless the winner portfo- term winners and going short on medium-term
lio appears to be riskier. Using monthly return losers earns around 1% profit per month. The
for NYSE common stocks from January 1926 profits cannot be attributed to conventional
until December 1982, De Bondt and Thaler proxies for risks.
(1985) test the two suggested hypotheses: (1) Rouwenhorst (1999) further expands the
extreme returns realizations in stock prices will international tests of momentum anomaly by
be followed by subsequent returns in the oppo- reaching the emerging markets. The sample con-
site direction, and (2) the bigger the past return, sists of more than 1700 firms from 20 emerging
the greater subsequent adjustment will occur, countries. He found that on average, emerging
the results suggests weak-form market ineffi- markets stocks exhibit momentum anomaly,
ciency. small stocks outperform large stocks, and value

17
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

stocks outperform growth stocks. Jegadeesh characteristic of Asian emerging markets. This
and Titman (2001) expand the sample period is the main basis of our research, as we will test
of JT from 1990 to 1998 to tests whether the the profitability of the conventional momentum
momentum anomaly is persistent with a differ- and the style momentum strategy in Indonesian
ent period. They found that momentum strategy stock exchange.
is persistent over time since the strategy gen-
erates a statistically different from zero profits Research Methods
on about 1.39% per month. In this study, they We employ dataset that consists of monthly
also examine the cross-sectional dispersion of returns of all firms in the Indonesian stock mar-
expected stock returns using the Fama-French ket (Bursa Efek Indonesia) to form price mo-
three-factor model (the previous research only mentum portfolios. In accordance with Lewellen
utilizing CAPM) and find that the Fama-French (2002) and Chan and Docherty (2015), we use
three-factor model can not explain the momen- market and book values of equity data of in-
tum profits since the three factor-beta indicating dividual stocks to construct style momentum.
that the winner's portfolio tends to be less risky All data are obtained from Datastream, and
than the losers portfolio. the sample periods start from January 2000 to
Lewellen (2002) study the momentum December 2015. The Stock and market return
anomaly in stock returns. He extends the port- are calculated with arithmatic returns using the
folio ranking method by employing firm-spe- closing price of the first date for each month,
cific indicators such as the market value of eq- and then substracted with risk free rate to get
uity (firm size), book equity-to-market equity stocks and market excess returns. The stock and
ratio (B/M), and double sorted size-B/M ra- market returns is defined as:
tio. Lewellen, motivated by Fama and French
(1992), argues that this method served as a
better proxy for risk in stocks portfolios. The
(1)
results suggest that the size, B/M, and double
where Pit is today stock price, and Pit-1 is the
sorted size-B/M momentum portfolio generates
yesterday price. Due to data availability, the
profits as substantial and in some cases stronger
30-day Indonesian Central Bank Certificates
than momentum profits found in individual or
(Sertifikat Bank Indonesia) is used as a proxy
industries stocks.
for the risk-free rate of return instead of 10-year
Chan and Docherty (2015) provide the out
government bond yield under the assumption
of sample evidence on Lewellen (2002) style
that the difference between treasury bill (SBI)
momentum by examining the style momentum
and government bond (SBN/SUN) will have no
in the Australian context. They study the style
substantial effect on the findings of momentum
momentum by creating 25 portfolios double-
returns. Our sample consist of 518 stocks listed
sorted on size and book equity-to-market eq-
in Bursa Efek Indonesia (Indonesian Stock Ex-
uity ratio. Their result suggests that Australian
change) from 1 January 2000 to 1 December
stocks exhibit style momentum that robust after
2015. The stock data include monthly prices,
controlling for frequently identified systematic
market value, and book value of equity for each
risk factors and monthly seasonality.
stock. We removed 185 stocks that have been
From the above discussions, we argue it is
delisted from the BEI, and 15 stocks that do not
presumably safe to say that the conventional
have sufficient number of observations of mar-
momentum and style momentum are persis-
ket value and book value of equity in the end of
tent anomalies. However, as best to our knowl-
our sample period from our sample.
edge, there are only a few kinds of literature
that study the style momentum, and further-
Price Momentum
more, there has been no attempt taken to study
the profitability of style momentum strategy in
Adopting JT’s approach, we sort stocks data
Indonesian context wherein provides a unique
in Indonesian Stock Exchange at the end of each

18
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

Table 1. Overview of Momentum Portfolios

Formation Period
Holding Period 3 6 9 12
3 (3-3) (6-3) (9-3) (12-3)
6 (3-6) (6-6) (9-6) (12-6)
9 (3-9) (6-9) (9-9) (12-9)
12 (3-12) (6-12) (9-12) (12-12)
Note: Each cell represents a momentum strategy that buys stocks based on k month formation period and then holds it for n month holding
period. Source: Vas and Absalonsen (2014).

month t into deciles based on their prior three, data exhibit serial correlation issue thus could
six, nine, and twelve-month returns. The win- make the result spurious. We will discuss the
ner portfolio is the top performers (top decile), treatment for this case in the series of the ro-
and the loser portfolio is the worst performers bustness test section.
(bottom decile). To mitigate the possible bid-
ask bounce effects (see Moskowitz and Grin- Style Momentum
blatt, 1999), we skip a month between the end
of the formation period and the start of holding In general, style momentum is formed simi-
period, so the formation-period for three-month lar to the price momentum, with the exception,
returns are calculated from t-2 to t-1, skipping rather than utilizing past returns to differenti-
month t. The holding periods for the momentum ate the winners and losers portfolio, Lewellen
portfolios are calculated for four holding peri- (2002) sorted the stocks based on size, B/M
ods: three month holding period (t+1 to t+3), six ratio, and double-sorted size-B/M ratio. We
month holding period (t+1 to t+6), nine-month define size as the market value of the stocks,
holding period (t+1 to t+9), and twelve-month and B/M as the ratio between the book value of
holding period (t+1 to t+12) so in total there are equity to the market value of equity. Lewellen
16 price momentum portfolios. The portfolios (2002) constructs 25 portfolios for the double-
are illustrated in Table 1. sorted size-B/M portfolios by dividing the
Following Chui et al. (2000), we form the stocks by quintiles (20%) for each size and B/M
second set of price momentum portfolios with so in total gives 5×5 portfolios (25) for each
30% portfolio size rather than deciles (10%) to formation-holding period. Chan and Docherty
overcome the small sample size which com- (2015) expand the style momentum by using
monly occurs in Asian emerging markets. The Brailsford et al. (2012) breakpoint that consists
momentum portfolios are the difference in go- of 75% cut-off for the largest portfolios, fol-
ing long the past winners (best performers) and lowed by 15%, 5%, 3%, and 2% for the small-
going short the past losers (worst performers). est portfolio, so in total this method also gives
For example, for October 2000, the (3-3) 5×5 portfolios (25) for each formation-holding
momentum portfolio returns are the sum of the period. In this study, we use 30% breakpoints
monthly return from August, September, and in light of the small sample size addressed by
October of 2000, where the profit used in each Chui et al. (2000). Therefore in total we con-
of this month is from holding period months struct 3×3(9) double-sorted size-B/M portfo-
1, 2, and 3, respectively. The (3-3) momentum lios, 3 size portfolios, and 3 B/M portfolios for
portfolio, in this case, is formed at the end of each formation-holding period (3−3), (3−6),
June 2000 (skipping July) which consists of (3−9), (3−12), (6−3),...,(12−12) respectively.
formation period May and June (3 months,
skipping the third month). Portfolio explained Momentum Profitability Significance
above has an overlapping period, that is invest-
ing in a k number of momentum portfolio at a The statistical significance of the momentum
time, as We follow JT for using the overlapping returns is tested using a 2-sided t-test statistics.
period to enhance the result. Since the returns We utilize 2-sided t-test since the average re-
are overlapping, there is a possibility that the turn of the momentum portfolio could generate

19
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

a positive and negative result. The t-test is de- we examine whether the momentum strategies
fined as: profits can be explained by the CAPM using
the Ordinary Least Squares (OLS) method. Fol-
lowing Cooper et al. (2004), we adjust the sig-
nificant momentum profits with the market fac-
(2) tor based on CAPM, which will be explained

below.
where μW is the mean of the monthly return To form the CAPM risk-adjusted profits, for
from withe nners portfolio (W), μL is the mean each holding period, We regress the time-series
of the monthly return from losers portfolio, NW of price and style momentum profits on market
is the number of monthly observation in the factor and a constant. The risk-adjusted profits
winners portfolio, NL is the number of observa- are presented by the alpha coefficient, that is
tion in losers portfolio (L), is the variance to show how well the momentum profits per-
of the winners portfolio, and is the vari- formed compared to the market portfolio. The
ance of the losers portfolio. Since the number model for CAPM adjusted profits is shown be-
of observation in winners and losers portfolio low:
are indifferent, the denumerator in Equation (2)
above denotes the Mean Squared Error of the RW−L−Rf=α1+β2 Rm−Rf+e (3)
momentum returns (see Hon and Tonks 2003).
To determine whether the result are statistically where RW−L−Rf is time series of momentum re-
significant, we compare the t−statistics against turns, α1 is a constant, Rf is the risk free rate, β2
the critical values given by the Student’s t dis- is the momentum portfolio beta, e is the model’s
tribution. We evaluate each t−statistics at 1%, residual, Rm is the return of the market portfolio
5%, and 10% level of significance. defined as market index (IHSG), and Rm−Rf is
the market risk premium. In light of heterosce-
Comparison of Price and Style Momentum dasticity and auto correllation issue addressed
Profitability by Coopper et al. (2004), we employ the robust
standard error using Newey-West adjustment
Lewellen (2002) conjecture that the profit- (1987).
ability of style momentum strategy is stronger
than the price momentum strategy. To compare Seasonality Effect
the profitability of momentum strategies, we
employ t-test in Equation (2) on each portfolio JT excludes the month of January in their
in price momentum strategy against each cor- sample period since it tends to exhibit extreme
responding portfolio in style momentum strat- value in terms of stock returns. To investigate
egy, so we will have the (3-3) price momentum the January effect in the momentum profits,
portfolio tested against (3-3) style momentum consider the following model:
portfolio and so on until the (12-12) price mo-
mentum portfolio is tested against (12-12) style RW−L−Rf = α0+β1(RM−Rf )+α1DUMMY1
momentum portfolio. The result will stand as
+α2DUMMY2+α3DUMMY3
a benchmark of whether the style momentum +α4DUMMY4+α5DUMMY5

strategy is better than the price momentum +α6DUMMY6+α7DUMMY7
strategy in term of profitability.

+α8DUMMY8+α9DUMMY9
Risk-Adjusted Profits +α10DUMMY10+α11DUMMY11
+eW-L (4)
To investigate whether the result of the mo-
mentum strategies are valid and persistent even where RW−L−Rf is the momentum strategy re-
after factor models are taken into consideration, turns, α0 is a constant, DUMMY1 is a dummy

20
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

Table 2. Descriptive Statistics

Portfolio Mean Std. Dev Skewness Kurtosis Observations
Price 10% Winner 0.0085 0.0655 -0.0101 2.9579 180
Price 10% Loser 0.0123 0.0899 1.3060 7.8674 180
Price 30% Winner 0.0101 0.0609 0.1449 3.3126 180
Price 30% Loser 0.0106 0.0730 0.4977 5.4332 180
Size Winner 0.0191 0.0661 -0.1610 5.3806 180
Size Loser 0.0059 0.0638 0.9027 4.5787 180
B/M Winner 0.0169 0.0741 0.6548 4.0166 180
B/M Loser 0.0052 0.0550 -0.4249 5.0164 180
Size-B/M Winner 0.0280 0.0964 1.6135 7.5091 180
Size-B/M Loser 0.0055 0.0621 -0.5006 6.7449 180
Note: This is the descriptive statistics for (6-6) momentum winners and losers portfolio that consists of 10% price momentum, 30% price
momentum, size, B/M and double sorted size-B/M portfolios. Source: author’s calculations.

that takes a value of 1 when the month is Janu- Results and Discussions
ary and 0 otherwise, and so on, until DUMMY11
is November dummy, and eW-L is an error term. The descriptive statistics of all portfolios
We drop the last month (December) dummy to used in this study are reported in Table 2.
avoid perfect collinearity problem and instead Table 2 shows the descriptive statistics for
of using the constant as a base measure for De- (6-6) strategy of price momentum, size, B/M,
cember Month. The inclusion (exclusion) of and double-sorted size-B/M portfolios. The av-
January month in our next models will be de- erage monthly returns differ from 0.005% to
cided by the significance of the α1 coefficient. If 0.028% per month, using 180 monthly obser-
it is not significant, we will include it in the next vations for each portfolio. The sample period
models and vice versa. for the (6-6) momentum strategy starts from 1
Another potential seasonality problem in our January 2001 to 1 December 2015. We employ
data is the sub-prime mortgage crisis in the U.S this sample period to provide the latest studies
between 2007 and 2009, which can potentially in the momentum profitability in the Indonesian
disturb the data in the Indonesian stock market. stock market.
To investigate this issue, we conduct the same There are 518 stocks listed in Bursa Efek In-
process as for the January effect explained donesia (BEI) which includes all stocks from
above, but, instead of using month dummy, Datastream after making some adjustments.
we use a dummy that will take the value of 1 We employ the equally weighted returns for
in the crisis period and 0 otherwise. We define each momentum portfolios. The skewness and
the crisis period from July 2007 to May 2009 the kurtosis of all portfolio indicating a viola-
based on the Bank for International Settlements tion of normal distributions (kurtosis = 3, and
(2009). So, consider the following regression skewness = 0). However, We follow the Central
model: Limit Theorem as in Fama (1965) that is the
stock returns in aggregate, as the number of ob-
RW−L−Rf = α0+β1(RM−Rf ) servation increase will be normally distributed

+α1DUMMYCRISIS1+eW-L (5) regardless of the underlying distribution. The
price momentum (the 10% and 30% portfolio
where RW−L−Rf is the time series of momentum size) generates negative returns while the style
returns, α0 represent the Non Crisis period, α1 momentum generates positive returns for this
represent the coefficient of the Crisis Period period sample. We will discuss the momentum
DUMMYCRISIS1, that takes a value of 1 in the profitability in detail in the next section.
crisis period and 0 otherwise. The exclusion
(inclusion) of the crisis period will be depend Returns of Price Momentum
on the statistical significance of the crisis period
dummy. Table 3 presents the average returns of the
10% Price Momentum Strategy from 3-3 to

21
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

Table 3. Returns of Price Momentum with 10% portfolio size

Holding Period Formation Period
3 6 9 12
3 -0.0099 -0.0118 -0.0086 -0.0105
Probability -0.2483 0.1796 0.3352 0.2236
6 -0.0059 -0.0048 -0.0030 -0.0054
Probability 0.4671 0.5533 0.7216 0.5040
9 -0.0025 0.0005 -0.0063 -0.0111
Probability 0.7490 0.9436 0.4207 0.1716
12 -0.0066 -0.0092 -0.0169 -0.0089
Probability 0.3840 0.2105 0.0268 0.2474
Note: The intersection between 3 to 12 formation and holding period is the reported mean profits and probability value obtained using the
t-test for each portfolio. The probability in bold is statistically different from zero. Source: author’s calculations.

Table 4. Returns of Price Momentum with 30% portfolio size

Holding Period Formation Period
3 6 9 12
3 0.0000 -0.0036 -0.0040 -0.0016
Probability 0.9953 0.6065 0.5670 0.8165
6 -0.0025 -0.0013 -0.0001 -0.0037
Probability 0.7104 0.8422 0.9828 0.5799
9 0.0018 0.0010 -0.0027 -0.0056
Probability 0.7888 0.8798 0.6868 0.3997
12 -0.0048 -0.0068 -0.0071 -0.0076
Probability 0.4700 0.2916 0.2795 0.2456
Note: The intersection between 3 to 12 formation and holding period is the reported mean profits and probability value obtained using the
t-test for each portfolio. The probability in bold is statistically different from zero. Source: author’s calculations.

12-12 holding-formation periods. All stocks results provided in Table 4.

that have return information available from the Table 4 reports the average returns of the
formation period after adjustments are included 30% price momentum strategy from 3-3 to 12-
in this particular strategy. All stocks available 12 holding-formation periods. The results are
after adjustment are ranked based on their past rather indifferent from the 10% Price Momen-
returns and then held on a certain holding pe- tum Strategy explained in Table 3 with some
riod; we skip a month between the formation exception, that there are positive returns for
period and the holding period. this strategy in 3-3, 3-9, and 6-9 portfolios that
The returns for all portfolios are negative give average return around 0.004%, 0,18%, and
and insignificant except for the 9-12 portfolios, 0,1% per month respectively. Consistent with
which have a negative return and significant Chui et al. (2000), this strategy also gives in-
probability value. The lowest returns are given consistent profits given by the insignificance
by the 9-12 portfolio, with an average return of the probability value of all portfolios. Look-
around −1.11% monthly. This finding is con- ing at the results of the 10% and 30% Price
sistent with Chui et al. (2000) that documented Momentum Strategy, we can see that it is not
inconsistent momentum profits in the Indone- profitable to buy the winner portfolio and short
sian stock market. However, this result could the loser portfolio. Our result indicates that the
arise from the smaller numbers of listed stocks Indonesian stock market is in a weak-form ef-
in Indonesian Stock Exchange compared to the ficient state since it is unable to predict stocks
New York Stock Exchange in which employed prices using strategies based on past returns
in JT’s paper. Hence, we need to test the prof- (see also Fama, 1970). However, we construct
itability of the price momentum strategy us- another momentum strategy to find a profitable
ing bigger portfolio size, following Chui et al. momentum strategy; the next momentum strat-
(2000), we use the 30% portfolio size were the egy is discussed in the next section.

22
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

Table 5. Returns of Size Momentum Portfolio

Holding Period Formation Period
3 6 9 12
3 0.0146 0.0139 0.0134 0.0137
Probability 0.0324 0.0421 0.0497 0.0493
6 0.0146 0.0133 0.0129 0.0152
Probability 0.0324 0.0504 0.0617 0.0265
9 0.0134 0.0127 0.0151 0.0150
Probability 0.0497 0.0672 0.0281 0.0301
12 0.0133 0.0152 0.0146 0.0131
Probability 0.0553 0.0283 0.0344 0.0557
Note: The intersection between 3 to 12 formation and holding period is the reported mean profits and probability value obtained using the
t-test for each portfolio. The probability in bold is statistically different from zero. Source: author’s calculations.

Table 6. Returns of Book-to-Market Momentum Portfolio

Holding Period Formation Period
3 6 9 12
3 0.0146 0.0131 0.0129 0.0123
Probability 0.0381 0.0578 0.0596 0.0740
6 0.0121 0.0114 0.0104 0.098
Probability 0.0777 0.0941 0.1256 0.1425
9 0.0105 0.0103 0.0091 0.0071
Probability 0.1252 0.1309 0.1761 0.2993
12 0.0102 0.0085 0.0075 0.0074
Probability 0.1360 0.2119 0.2724 0.2725
Note: The intersection between 3 to 12 formation and holding period is the reported mean profits and probability value obtained using the
t-test for each portfolio. The probability in bold is statistically different from zero. Source: author’s calculations.

Returns of Style Momentum justment are ranked based on their past three to
twelve month B/M ratio and then being held for
This section reports the average returns of three to twelve month holding period. The win-
the style momentum following Cooper et al. ner portfolio is the 30% stocks with the highest
(2004) that is size, B/M, and double-sorted size- B/M ratio, and the loser portfolio is the lowest
B/M portfolios. All strategies employ 30% size 30% B/M ratio. The B/M momentum portfo-
portfolios with holding and formation period lio returns are the difference between the win-
the same as the price momentum strategy. Table ner portfolio and loser portfolio divided by the
5 documents the returns of the size momentum number of stocks included in the portfolio each
portfolio. Each month, We rank all stocks based month. Overall, the B/M momentum portfolio
on their past size (market value) and then hold can generate positive returns with the 3-3 port-
them on a certain holding period, the winner folio gives the highest returns, around 1.4% per
portfolio is the 30% smallest size stocks and the month. Even though this B/M strategy able to
losers portfolio is the 30% biggest size stocks. generates positive returns, we have to proceed
In contrast to the price momentum, the size with a caveat, that is only eight from 16 portfo-
momentum portfolio consistently able to gener- lios gives significant positive returns.
ates positive returns on all 16 portfolios with Table 7 shows the returns of 16 style portfo-
a statistically significant probability value. The lios strategies. The style portfolios are formed
(12-6) portfolio is the most profitable portfolio, with two-way sorts method by ranking all avail-
with an average return around 1.5% per month able stocks after adjustment on their past size
and significant on a 5% level of confidence. (market value) and book-to-market (B/M) ratio
Table 6 provides the book-to-market (B/M) and then held for a certain month. The winner
Momentum Portfolio. According to Cooper et portfolio consists of stocks that have the fol-
al. (2004), the book-to-market ratio is a ratio of lowing characteristics: (1) 30% smallest mar-
the book value of equity divided by the market ket value, and (2) 30% highest B/M ratio and
value of equity. All available stocks after ad- the loser portfolio has the opposite characteris-

23
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

Figure 1. Profitability Across Momentum Strategies

Source: author’s calculations.

Note: W-L size_B/M represents the cumulative return of the (6-6) double-sorted size-B/M portfolio. Market represents the cumulative market
return, W-L size represents the cumulative return of (6-6) size portfolio, and B/M represents the cumulative return of (6-6) B/M portfolio.

Table 7. Returns of Double-Sorted Size-B/M Momentum Portfolios

Holding Period Formation Period
3 6 9 12
3 0.0247 0.0233 0.0250 0.0264
Probability 0.0038 0.0055 0.0038 0.0036
6 0.0223 0.0228 0.0219 0.0247
Probability 0.0078 0.0078 0.0139 0.0062
9 0.0199 0.0218 0.0246 0.0235
Probability 0.0231 0.0149 0.0067 0.0089
12 0.0198 0.0239 0.0212 0.0211
Probability 0.0236 0.0076 0.0166 0.0186
Note: The intersection between 3 to 12 formation and holding period is the reported mean profits and probability value obtained using the
t-test for each portfolio. The probability in bold is statistically different from zero. Source: author’s calculations.

tics from winner portfolio. This strategy gives JSX as a proxy) generated negative cumulative
consistent positive returns, and all of them are profits from 2001 to 2015. This finding is again
statistically different from zero. Consistent with consistent with Lewellen (2002) who stated
Lewellen (2002), this result shows that style that the style momentum strategy gives a bet-
momentum gives stronger profits than price ter result compared to conventional momentum
momentum with the (12-3) portfolio gives the strategies since the style momentum represent
highest return (around 2.6% per month) whereas the common risk factors following Fama and
the lowest return is held by the (3-12) portfolio French (1992). This result also serves as ad-
with around 1.98% return monthly compared ditional empirical evidence of stock market in-
to results from price momentum strategies that efficiency in the Indonesian stock market that
overall generates negative profits and statisti- deviates from the EMH.
cally not different from zero.
Figure 1 describes the cumulative returns of Risk-Adjusted Profits Regression
(6-6) momentum strategies and market port-
folio, we can see that the double sorted style The findings in previous sections show that
portfolio (Size-B/M) generates the highest style momentum strategy gives a positive re-
profits compared to price, size, B/M, and mar- sult from the period 2001 to 2015. To provide
ket portfolio, whereas, the lowest profits being a measure that considers risk and to make the
held by the 30% price momentum strategy. In result more comparable, we adjust the profits of
this period, the market portfolio (with IHSG/ the momentum strategy utilizing market factor.

24
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

Table 8. CAPM-Adjusted Momentum Profits

Holding Period Formation Period
3 6 9 12
3 0.0249 0.0236 0.0254 0.0265
Probability 0.0001 0.0002 0.0001 0.0001
DW Stat 1.6099 1.5182 1.4893 1.4969
6 0.0229 0.0233 0.0224 0.0250
Probability 0.0003 0.0003 0.0011 0.0004
DW Stat 1.5390 1.4874 1.5620 1.5799
9 0.0204 0.0220 0.0250 0.0241
Probability 0.0022 0.0014 0.0005 0.0006
DW Stat 1.6952 1.6139 1.6679 1.6986
12 0.0202 0.0242 0.0217 0.0220
Probability 0.0025 0.0005 0.0016 0.0016
DW Stat 1.6015 1.6064 1.6894 1.7393
Note: The intersection between 3 to 12 formation and holding period is the reported intercepts (serves as a measure for CAPM-adjusted
profits) and probability value from the CAPM regression for each portfolio along with the Durbin Watson Statistics in light of Autocorrelation
issue. The probability in bold is statistically different from zero. Source: author’s calculations.

Table 9. White Heteroscedasticity Test on (6-6) Style Momentum

Test Coefficient Probability
F-statistic 6.2225 0.0024
Obs*R-squared 11.8277 0.0027
Scaled explained SS 42.0094 0.0000
Note: This is the result of the White Test on the (6-6) style momentum strategy returns after the returns regressed using CAPM. Source:
author’s calculations.

Table 8 presents the results. The risk-adjusted ity value of three types of tests is significant at
profits obtained by regressing the time-series 1% level. Thus we are confident that our model
return of the double-sorted size-B/M strategy suffers from the heteroscedasticity problem
against a constant and a market factor using (see also, Brooks, 2008). Based on the result
CAPM. From Table 8, we can conclude that the of autocorrelation and heteroscedasticity tests
profits of style momentum strategy persist after conducted above, We need to adjust our stand-
controlling the market factor, with the (12-3) ard errors of the syle momentum profitability
strategy gives the highest result that is around significance using Newey-West Standard Error,
2.6% per month whereas the (9-12) gives the with the result explained in Table 11.
lowest profits that are around 2.1% monthly. As expected, the profitability of the style
Table 8 also provides the Durbin-Watson Sta- momentum strategy is persisted after control-
tistics to measure the autocorrelation in the ling for the market factor and adjusted standard
model. The result shows that the model suffers error in light of autocorrelation and heterosce-
from the autocorrelation problem since none of dasticity issue. However, we need to test the
the DW-stat gives a coefficient of 2 (see also, seasonality effect of our sample period. The
Brooks, 2008). first seasonality test is the January effect, un-
Moreover, We also test the possibility of der JT that excludes the January month in their
heteroscedasticity issue in the model by con- sample period to strengthen their results since
ducting the white test on the time series of (6- January tends to exhibit extreme value, they
6) style momentum returns. We assume that by argue that the momentum profits could come
testing only the (6-6) style momentum strategy from a specific period (January). To test the
returns, We will have sufficient evidence for January effect, we employ the following model
the 15 other style portfolios since all portfoli- in Equation (4) with the results in Table 11.
os generate consistent significance. The result From Table 11, the January dummy is not
of the White test on (6-6) style momentum is statistically significant. Therefore we include
shown in table 9. the January month in our sample period. Note
From Table 9, we can see that the probabil- that the May and October dummy yield positive

25
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

Table 10. CAPM-Adjusted Momentum Profits with HAC Standard Error

Holding Period Formation Period
3 6 9 12
3 0.0249 0.0236 0.0254 0.0265
Probability 0.0010 0.0021 0.0017 0.0020
6 0.0229 0.0233 0.0224 0.0250
Probability 0.0030 0.0030 0.0060 0.0019
9 0.0204 0.0220 0.0250 0.0241
Probability 0.0064 0.0054 0.0016 0.0017
12 0.0202 0.0242 0.0217 0.0220
Probability 0.0078 0.0019 0.0043 0.0028
Note: The intersection between 3 to 12 formation and holding period is the reported intercepts (serves as a measure for CAPM-adjusted
profits) and probability value from the CAPM regression with Heteroscedasticity and Autocorrelation Consistent (HAC) Newey-West
Standard Error for each portfolio. The probability in bold is statistically different from zero. Source: author’s calculations.

Table 11. Seasonality - January Effect

Variable Coefficient Standard Error P-Value
Alpha 0.0011 0.0210 0.9551
RM-RF -0.0717 0.1001 0.4746
January 0.0009 0.0302 0.9758
February 0.0302 0.0302 0.3190
March -0.0004 0.0302 0.9888
April 0.0031 0.0302 0.9180
May 0.0595 0.0303 0.0509
June 0.0395 0.0302 0.1933
July 0.0213 0.0302 0.4815
August 0.0374 0.0302 0.2174
September 0.0287 0.0305 0.3484
October 0.0552 0.0302 0.0698
November -0.0081 0.0303 0.7888
Descriptive Statistics
R2 0.0714
Adj-R2 0.0051
F-stat 1.0775
DW-stat 1.4617
Note: This is the regression output of the January Seasonality test using CAPM regression that includes monthly (January to November)
dummies that take the value of 1 when the month are January, ..., November, and 0 otherwise. Source: author’s calculations.

Table 12. Crisis Period

Variable Coefficient Standard Error Probability Value
Alpha 0.0208 0.0068 0.0026
RM-RF -0.0535 0.0981 0.5859
Crisis Dummy 0.0188 0.0189 0.3211
Descriptive Statistics
R2 0.0079
Adj-R2 -0.0032
F-stat 0.7116
DW-stat 1.4945
Note: This is the regression output of the Crisis Seasonality test using CAPM regression that includes a dummy variable that takes the value
of 1 on the given crisis period, and 0 otherwise. Source: author’s calculations.

coefficient. However, we argue that as far as We in Equation (5). The exclusion (inclusion) of
know, there is no specific issue mentioned for the crisis period will depend on the statistical
May and October in the previous momentum significance of the crisis period dummy.
literature. Therefore, we include May and Oc- The results in Table 12 suggest that the sub-
tober month in our sample period. prime mortgage crisis occurred in the U.S. does
The next seasonality issue is the crisis period not affect the time series of (6-6) style momen-
in around 2007 and 2009. We employ the model tum returns in Indonesian stock market since

26
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

the crisis period dummy does not have a statis- inconsistent profits given by the insignificance
tically significant probability value. Therefore, of the probability value of all portfolios.
we have no problem to include the crisis period The results of the style momentum strate-
into our sample, and we can use the results in gies, in contrast, can generate an overall posi-
Table 10. tive and statistically significant profits. The
The persistence of the style momentum prof- style momentum gives higher profits than
its and the fact that price momentum profits are price momentum with the (12-3) double sorted
not able to give consistent profits in Indone- size-B/M portfolio gives the highest return of
sian context indicates that the Indonesian Stock 2.6% per month whereas the lowest return on
Market is in the weak-form efficient state of the the double sorted size-B/M portfolio is held by
EMH but does not hold in the semi-strong form the (3-12) portfolio with around 1,98% month-
efficient state. In the weak-form efficient state, ly return. This finding confirms the results of
the current stock prices already reflect all infor- Lewellen (2002) who find the style momentum
mation on its historical prices. Hence it is not in some case are stronger than the conventional
possible to predict the future price using price (price) momentum of JT.
momentum strategy which utilizing the histori- These results indicate that the Indonesian
cal prices. Our findings in section 4.2 support stock market is arguably in the state of weak-
this statement. The results imply that the semi- form efficient of EMH since we seem unable
strong form efficient state should be ruled out to predicts positive future returns based on
from the Indonesian Stock Market condition stocks past characteristics, while style momen-
since we can obtain a significant profit by em- tum strategy that utilizes publicly available in-
ploying a style momentum strategy that incor- formation can give consistent positive profits.
porating publicly available information such as Our findings support the result of Arshad et al.
size and book-to-market ratio. (2016) that Indonesian stock market is in weak-
form efficiency. On the other hand, our results
Conclusions deviate from the conclusion of the Indonesian
stock market have no sign of market efficiency
We find the price momentum strategy that in Kim and Shamsuddin (2008). However, we
ranks stocks based on their past returns is not emphasize that we did not test the EMH directly
profitable in the Indonesian stock market. The as in Malkiel and Fama (1970), Jegadeesh and
returns of 10% size price momentum portfolios Titman (1993), Jegadeesh and Titman (2001),
are negative and insignificant except for the or Kim and Shamsuddin (2008), our results
9-12 portfolio, which has a significant nega- only indicate the presence of inefficiency in In-
tive return and the 6-9 portfolio that give an in- donesian Stock Market.
significant positive average return. The lowest We want to assert that due to data avail-
returns are produced by 9-12 price momentum ability, we utilize 30-day Indonesian Central
portfolio, which averaged -1.11% per month. Bank Certificates (Sertifikat Bank Indonesia) as
We increase the size of the price momentum an intermediary for the risk-free rate of return
portfolio from 10% into 30% to overcome the instead of 10-year government bond yield un-
small sample problem addressed by Chui et al. der the assumption that the difference between
(2000). The results of the 30% price momentum treasury bill (SBI) and government bond (SBN/
strategy are indifferent from the 10% price mo- SUN) will have no substantial effect on the
mentum strategy with the exception, that there findings of momentum returns. We also need to
are three portfolios in this strategy that yields accentuate that our results only limited to the
insignificant positive average returns which are 3-3, …, 12-12 forming and holding period of
the 3-3, 3-9, and 6-9 portfolios. These portfo- momentum portfolios as in Jegadeesh and Tit-
lios give average return around 0.004%, 0.18%, man (1993), it is possible that momentum re-
and 0.1% per month respectively. Coherent turns under different forming and holding pe-
with Chui et al. (2000), this strategy also yields riod to have a different result.

27
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

From the practical perspective, our find- the Indonesian stock market based on an indica-
ings show that implementing a strategy that tion of stock market inefficiency that let us do
buys stocks based on its past size and B/M ra- so. For future study, we recommend investigat-
tio seems able to generate significant profits ing thoroughly the inability of price momentum
in Indonesian stock market, while a strategy strategy to generate profits in the Indonesian
that selects stocks based on its past returns, in stock market, which is inconsistent with the
contrast, are unable to give consistent profits. generally accepted findings of JT. We also sug-
In summary, we recommend investors to apply gest the source of the profitability of momen-
the style momentum strategy to gain profits in tum strategies can be further explored.

References
Amanda, C., and Husodo, Z. A. (2015). Empiri- Chan, K., Hameed, A., and Tong, W. (2000).
cal test of Fama-French three factor model Profitability of momentum stragegies in
and illiquidity premium in Indonesia. Cor- the international equity markets. Journal
porate Ownership & Control Journal, 12(2): of Financial and Quantitative Analysis,
362-373. 35(2):153–172.
Arshad, S., Rizvi, S.A.R, Mat Ghani, G., and Chen, J. and Hong, H. (2002). Discussion of
Duasa, J. (2016). Investigating stock market momentum and autocorrelation in stock
efficiency: A look at OIC member countries. returns. The Review of Financial Studies,
Research in International Business and Fi- 15(2):565–573.
nance, 36: 402-413. Chordia, T. and Shivakumar, L. (2006). Earn-
Bank for International Settlements (2009). The ings and price momentum. Journal of Fi-
Global Financial Crisis - BIS 79th Annual nancial Economics, 80(3):627–656.
Report. Available at http://www.bis.org/ Fama, E.  F. (1965). The behavior of stock-
publ/arpdf/ar2009e2.pdf. market prices. The Journal of Business,
Bekaert, G., Erb, C. B., Harvey, C. R., and Vis- 38(1):34–105.
kanta, T. E. (1997). What matters for emerg- Fama, E.  F. and French, K.  R. (1992). The
ing equity market investments. Emerging cross-section of expected stock returns. The
Markets Quarterly, 1(2):17–46. Journal of Finance, 47(2):427–465.
Bodie, Z., Kane, A., Marcus, A., and Jain, R. Fama, E. F. and French, K. R. (2015). A five-
(2014). Investments, Asia Global Edition. factor asset pricing model. Journal of Finan-
Singapore: McGraw-Hill Global Education cial Economics, 116:1–22.
Holdings. Gujarati, D. (2004). Basic Econometrics. Tata
Brailsford, T., Gaunt, C., and O’Brien, M. A. McGraw-Hill.
(2012). The investment value of the value Hon, M.  T. and Tonks, I. (2003). Momentum
premium. Pacific-Basin Finance Journal, in the UK stock market. Journal of Multina-
20(3):416–437. tional Financial Management, 13(1):43–70.
Brooks, C. (2008). Introduction Econometrics Jegadeesh, N. and Titman, S. (1993). Returns to
for Finance. Cambridge, UK: Cambridge buying winners and selling losers: Implica-
University Press. tions for stock market efficiency. The Jour-
Carhart, M. M. (1997). On persistence in mu- nal of Finance, 48(1):65–91.
tual fund performance. The Journal of Fi- Jegadeesh, N. and Titman, S. (2001). Profitabil-
nance, 52(1):57–82. ity of momentum strategies: An evaluation
Chan, H. and Docherty, P. (2015). Momentum of alternative explanations. The Journal of
in Australian style portfolios: risk or ineffi- Finance, 56(2):699–720.
ciency? Accounting & Finance, 56(2): 333– Kim, J.H., and A. Shamsuddin (2008). Are
361. Asian stock markets efficient? Evidence

28
 R. L. Mosii and S. S. Wibowo / Indonesian Capital Market Review 11 (2019) 15-29

from new multiple variance ratio tests. Jour- mentum strategies. The Journal of Finance,
nal of Empirical Finance 15(3): 518-532. 53(1):267–284.
Lewellen, J. (2002). Momentum and autocor- Rouwenhorst, K.  G. (1999). Local return fac-
relation in stock returns. The Review of Fi- tors and turnover in emerging stock markets.
nancial Studies, 15(2):533–564. The Journal of Finance, 54(4):1439–1464.
Malkiel, B. G. (2003). The efficient market hy- Sharpe, W.  F. (1964). Capital asset prices: A
pothesis and its critics. The Journal of Eco- theory of market equilibrium under con-
nomic Perspectives, 17(1):59–82. ditions of risk. The Journal of Finance,
Malkiel, B.  G. and Fama, E.  F. (1970). Effi- 19(3):425–442.
cient capital markets: A review of theory Vas, Alexander and Absalonsen, Kristoffer.
and empirical work. The Journal of Finance, (2014). The momentum effect: An empiri-
25(2):383–417. cal study of the Oslo Stock Exchange. Mas-
Moskowitz, T. J. and Grinblatt, M. (1999). Do ter’s thesis, Copenhagen Business School.
industries explain momentum? The Journal Retrieved from https://studenttheses.cbs.dk/
of Finance, 54(4):1249–1290. xmlui/bitstream/handle/10417/4793/kristof-
Rouwenhorst, K. G. (1998). International mo- fer_absalonsen_og_alexander_vas.pdf

29