Show

“The Evidence of the Value Premium, the Size

Effect, and Momentum versus Contrarian

Strategies in Indonesian Stock Market”

Master Thesis

Faculty of Economics and Business Administration

Master of Science in Financial Economics

Track: Financial Analysis

Author: Dwi Astuti Putranto , i588733

Supervisor: Dr. Jeroen Derwall

Co-Supervisor: Dr. Marco Avarucci

Dwi Astuti Putranto, The Evidence of The Value Premium, The

Size Effect, and Momentum versus Contrarian Strategies in

Indonesian Stock Market

2

Contents

Abstract ....................................................................................................................... 7

I. Introduction ............................................................................................................. 8

II. Literature Review ................................................................................................ 13

1. The traditional CAPM .................................................................................... 13

2. The value premium .......................................................................................... 17

2.1 Evidence in U.S. Market........................................................................... 17

2.2 Evidence in International Markets ......................................................... 18

2.3 Evidence in Asian Markets ...................................................................... 18

3. The size effect ................................................................................................... 22

4. Momentum versus contratrian strategies ..................................................... 23

4.1 The evidences of momentum versus contrarian strategies in U.S.

Market .............................................................................................................. 24

4.2 Asymmetric reactions to good news as opposed to bad news ............... 25

4.3 Trading volume and momentum versus contrarian strategies ............. 26

4.4 The Sources of momentum and contrarian profits ................................ 26

III. Data Analysis...................................................................................................... 31

1. Indonesia Stock Exchange .............................................................................. 31

2. Data ................................................................................................................... 31

3. Methodology ..................................................................................................... 33

4. Time-series return regression ......................................................................... 39

5. The GRS Test ................................................................................................... 43

IV. Result ................................................................................................................... 46

1. Simple glamour and value strategies .................................................................. 46

2. Two-independent sort glamour and value strategies ........................................ 48

3. The size effect ........................................................................................................ 51

4. Momentum versus contrarian strategies............................................................ 52

5. The effect of trading volume on momentum/contrarian profits ...................... 53

V. Robustness Check ................................................................................................ 55

1. The Simple CAPM time-series regressions ........................................................ 55




3

2. The Fama French three-factor model ................................................................ 58

3. The post-crisis analysis of the value premium ................................................... 62

4. Jegadeesh and Titman methodology of momentum versus contrarian

strategies .................................................................................................................... 63

5. The post-crisis analysis of Jegadesh and Titman methodology for momentum

versus contrarian strategies ..................................................................................... 65

VI. Conclusions ......................................................................................................... 66

References ................................................................................................................. 68

Text Figures .............................................................................................................. 76

Text Tables ................................................................................................................ 78




4

List of Figures

Figure 1 The Risk Return Tradeoff ............................................................... 76

Figure 2 Jakarta Composite Index and Capital Market Milestones 1984-

June 2008 ........................................................................................... 77




5

List of Tables

Table I The Complete Overview of All Variables ....................................... 78

Table II Trading Value on Indonesian Stock Market .................................. 79

Table III Returns for Quintiles Portfolios Based on One-Dimensional

Classification by Various Measures of Value Characteristics ...... 79

Table IV Returns for Portfolios Based on Two-Dimensional Classifications

by Various Measures of Value Characteristics .............................. 80

Table V Returns for Portfolios Based on Size-Classifications .................... 82

Table VI An Overlapping Momentum/Contrarian Average Monthly Returns

of JKSE Stocks: 12/1991-12/2008 using Lo and MacKinlay (1990)

Methodology ...................................................................................... 82

Table VII Momentum/Contrarian Average Returns using Lo and MacKinlay

(1990) Methodology that Incorporates Trading Volume .............. 83

Table VIII The CAPM Time-Series Regressions for Monthly Excess Returns

on 25 Portfolios Formed on Size and BE/ME: 12/1991-12/2008, 205

Months ............................................................................................... 84

Table IX The CAPM Time-Series Regressions for Monthly Excess Returns

on the LSV Equal-Weight Quintiles: 12/1991-12/2008, 205 Months

............................................................................................................ 85

Table X The CAPM Time-Series Regressions for Monthly Excess Returns

on Equal-Weight Portfolios Formed on Past Returns: 12/1992-

1/2008, 182 Months ........................................................................... 86

Table XI The Fama French Three-Factor Regressions for Monthly Excess

Returns on 25 Portfolios Formed on Size and BE/ME: 12/1991-

12/2008, 205 Months ......................................................................... 86

Table XII The Fama French Three-Factor Regressions for Monthly Excess

Returns on the LSV Equal-Weight Quintiles: 12/1991-12/2008, 205

Months ............................................................................................... 87




6

Table XIII The Fama French Three-Factor Regressions for Monthly Excess

Returns on Equal-Weight Portfolios Formed on Past Returns:

12/1991-1/2008, 182 Months ............................................................. 88

Table XIV Returns for Quintiles Portfolios Based on One-Dimensional

Classification by Various Measures of Value Characteristics Post

Crisis Period ...................................................................................... 89

Table XV Jegadeesh and Titman (1993) Methodology of an Overlapping

Momentum/Contrarian Average Monthly Returns of JKSE Stocks:

12/1991-12/2008 ................................................................................. 90

Table XVI Jegadeesh and Titman (1993) Methodology of an Overlapping

Momentum/Contrarian Average Monthly Returns of JKSE Stocks

Sub-Period Analysis: 12/1991-06/1997 and 08/1998-12/2008 ........ 91

Table XVII Jegadeesh and Titman (1993) Methodology of an Overlapping

Momentum/Contrarian Average Monthly Returns of JKSE Stocks:

12/1991-12/2008 exclude Crisis Period 07/1997-07/1998 ............... 92




7

Abstract

This paper examines the presence of the value premium, the size effect, and

momentum versus contrarian strategies in Indonesian stock market for the period

1991-2008. The momentum/contrarian strategies are analyzed using two models, the

Lo and MacKinlay (1990) methodology and the Jegadeesh and Titman (1993)

methodology. This study also analyzes the effect of trading volume in

momentum/contrarian strategies and compares the performance of the CAPM and the

Fama French three-factor model in explaining returns variation of particular

portfolios. The GRS test statistic is performed to investigate whether the intercepts are

jointly zero. The study proves that value and small stocks outperform glamour and big

stocks in Indonesian stock market. In short-term, winners outperform losers due to

investors’ overestimation towards winners that belong to high lagged volume group.

In medium-term however, investors’ overreaction to losers is higher than their

overreaction towards winners. Zero-cost strategies in general are in favor of the

contrarian pattern, and are especially pronounced for stocks in lower trading volume.

The Fama French three-factor model only explains slightly better than the CAPM

with the small value portfolios still left unexplained.




8

I. Introduction

Market are said to be efficient when stock prices move randomly and show no

discernible pattern. Market anomalies exists when stock prices do move in some

predetermined pattern and investors can earn excess returns by trading around the

pattern. Many studies in U.S. markets and in international markets show the presence

of the three well known market anomalies. They are the value premium, the size

effect and momentum versus contrarian. Researches on the performance of value and

growth stocks and momentum versus contrarian have been previously pursued by lot

of academics and researchers. Value stocks are defined as stocks that have low prices

relative to earnings, dividends, historical prices, book assets, or other measures of

value. Growth (Glamour) stocks, on the other hand are those that have high prices

relative to earnings, dividends, historical prices, book assets, or other measures of

value. For many years, scholars and professionals have argued that value stocks earn

higher returns than growth stocks, the so-called value premium. There is also

evidence that value stocks tend to be smaller in size than growth stocks. Size is

measured by market capitalization which is stock’s price times the number of shares

outstanding. The form of market anomalies where small capitalization stocks earn

higher mean returns than large capitalization stocks is called the size effect.

Momentum refers to the idea that stocks that perform well in the past will continue to

do well in the future. Whereas contrarian refers to the idea that bad-performing stocks

in the past will outperform good-performing stocks in the future.

However, the evidence of these market anomalies is not clear; for instance growth

stocks have had higher returns than value stocks in some countries. Because the

behavior of stock markets may differ across geographic regions, it is possible that the

performance of the three market anomalies also vary across countries. The purpose of

this study is to explore the presence of the value premium, the size effect, and

momentum versus contrarian in Indonesia, using the following variables: Book-to-

Market equity (BE/ME), Earning per Price (E/P), Cash Flow per Price (CF/P), market

value of equity (ME, stock prices times number of shares outstanding), trading

volume, and past returns. Further details and explanations for each variables are given




9

in Table I. Developing a theoretical analysis of this relationship is beyond the scope of

this paper.

Why Indonesian Stock Market?

The attractiveness of Asian stock markets came under consideration after some

researchers found evidence that by diversifying their portfolio with Asian stocks,

international investors can reduce their risk. Bailey and Stulz (1990) suggest that U.S.

investors can reduce their risks by 30% to 50% by including Asian stocks in their

portfolios since U.S. and Asian markets should perform very differently in any given

periods. The gain from international diversification, however, depends on the degree

of capital markets correlation. Roll (1995) provides additional evidence for the

correlation between Indonesian and other developed countries stock markets. He

found that prior to December 1988, 20 out of 23 measured correlation coefficients

(using monthly returns) were negative, and in contrast during the period from 1989-

1992, most of the correlations were positive. Positive correlation among stock markets

means investors cannot use diversification to reduce unsystematic risks. However,

Roll’s findings which end on 1992 seem to suggest that markets correlation might be

different across periods. Since this study incorporates different sample period (from

1991 to 2008) which ends after that of Roll’s, the correlation pattern between

Indonesian and other developed countries stock markets might be changing as well.

Furthermore, as documented by Malliaropolus and Priestley (1999), investing in

emerging markets, especially in Asia is interesting due to the high volatility

characteristics of these markets. They observe that emerging markets have high

volatilities and thus high expected returns. Increasing attention from investors and the

relatively unknown nature of emerging markets place research in these areas at high

priority.

Although many researchers have focused on the performance of Asian market,

especially Japanese stock market, Indonesian stock market seems to be virtually

uncovered and in that sense provides the opportunity to examine and to contribute for




10

evidence of market anomalies in the form of the value premium, the size effect and

momentum versus contrarian. The importance of Indonesian stock market for

international investors has increased from 2002-2008. Table II shows the total trading

value for locals and foreigners from 2002 to the first quarter of 2008. The percentage

of foreign transaction from total trading value has increased from 11.27 percent in

2002 to 24.72 percent on the first quarter of 2008.

The performance of value and growth stocks in Indonesia has been examined by few

researchers with mixed and inconsistent results. Ding et al. (2005) examine value and

growth portfolios in seven East Asian countries just prior to the Asian Financial

Crisis. They conclude that Indonesian stock market do not reveal any particular firm

size or growth potential effects. On the other hand, Roll (1995) provides the evidence

that value portfolio of Indonesian equities performed much better than the growth

portfolio during 1989-1992 periods. These mixed results may be due to a number of

reasons, including: (i) the differences in data sources or study periods; (ii) the

differences in the formation (ranking) periods and in the investment holding periods1;

(iii) different methodologies2; and (iv) the differences in the sample stocks (e.g.,

aggregate market indices or individual stocks), etc.

The choice of predictor variables as explained in Table I, is motivated by the existing

evidence, especially in the U.S. which, in turn, is influenced by the practice of

fundamental security analysts. Wherever possible, this study attempts to draw

parallels between this research and those from previous studies. On the one hand,

findings that the same factors are at work in both studies would strengthen the

evidences accumulated for these kinds of stock market anomalies. On the other hand,

different findings would suggest further research exploring institutional or behavioral

differences between Indonesian stock market and other international markets.

1 Formation (ranking) period refers to the period in which stocks are ranked based on particular value

characteristics in order to form the portfolios. Investment holding period refers to the period in which

stocks are hold before finally being sold by investors to realize profit. 2 Different methodologies can generate different results. This study uses both Lo and MacKinlay

(1990) methodology and Jegadeesh and Titman (1993) methodology in analysing the performance of

momentum versus contrarian strategies. However, in this study, both methodologies provide same

conclusions. Further details regarding both methodologies are provided in section III.




11

This study is different from past studies in two significant ways. Firstly, the majority

of the past studies on Asian markets rely on data that end prior to the year 2000 or

even prior to the Asian financial crisis of 1997-1998. Therefore, the robustness of

their reported findings is questionable for the post-2000 study. This research utilizes

most recent data and study period that ends in December 2008. More specifically, this

research will examine the performance of the value premium, the size effect and

momentum versus contrarian from December 1991 to December 2008 period.

Secondly, this research also attempts to shed light on the relation between return

predictability and trading volume.

The decision to incorporate trading volume in explaining momentum and contrarian is

based on four considerations. Firstly, the evidence on return predictability in emerging

markets may suffer from data problems such as non-synchronous trading or thin

trading.3

Non-synchronous trading problem arises when the prices of distinct

securities are mistakenly assumed to be sampled simultaneously. Different securities

might have different trading performances. Specifically, many securities are traded

only infrequently, with few securities so actively traded that prices are recorded

almost continuously. Because prices for most securities are reported only at distinct

random intervals (for instance monthly or annually), completely accurate calculation

of returns over any fixed sequence of periods is virtually impossible. Treating

nonsynchronous prices as if they were observed at the same time can create

econometric problems. Lo and MacKinlay (1990) show that non-synchronous trading

effects add to contrarian profits. Therefore, incorporating trading volume is expected

to reduce the problems by capturing the extent to which momentum and contrarian

profits are due to lack of trading. In other words, by dividing stocks into volume

groups enable comparison among stocks with the same trading performance.

Secondly, volume has been believed to provide information about future price

movements. There is a common belief that “it takes volume to move prices.” Without

sufficient trading volume, stock prices may react less to information on low trading

activity; therefore the momentum strategies applied to these stocks will be profitable.

3 In this study, following Lo and MacKinlay (1990), the terms “thin trading” and “non-synchronous

trading” are used interchangeably.




12

Thirdly, only few observations have been made combining the role of volume in price

predictability in Indonesian stock market and this study attempts to fill the gap. Last

but not least, as a practical concern, investors should take liquidity into account when

trading in emerging markets.

The primary findings of this study are: (i) The value premium exists in Indonesian

stock market during the sample period; (ii) The two-independent sort strategies, the

strategies that sort stocks independently based on two value characteristics, especially

those which incorporate value-weighted return, produce higher average annual return

than those strategies that are based exclusively on individual-sort; (iii) The Fama-

French three-factor model is slightly better when it comes to explaining variation in

stock returns than the simple CAPM4

; (iv) In the medium term, investors’

overreaction to losers is higher than their overreaction towards winners; in short-term,

winners outperform losers due to investors’ overestimation towards winners that

belong to high lagged volume group; (v) zero-cost strategies in general are in favor of

the contrarian pattern, and are especially pronounced for stocks in lower trading

volume.

The remainder of this study is structured as follows. Chapter two provides an

overview of the literature on which this research is based as well as a short overview

of traditional CAPM model. Chapter three describes the dataset as well as the

methodology in use. Chapter four presents the empirical results and contains an

analysis on the performance of value premium, size effect, and momentum/contrarian

strategy. Chapter five continues with the robustness check and the results of time-

series regressions using both the simple CAPM and the Fama French three-factor

model. Finally, chapter six provides conclusions and limitations to the study.

4 It should be noted that this study does not take into account the time-varying beta, the structural

break, or any advance CAPM tests. Further investigation regarding these matters may provide better

analysis in the results of the regression test.




13

II. Literature Review

This literature review attempts to depict an overview over a selection of seminal and

influential studies in the area related to the three well known market anomalies; value

premium, size effect, and momentum versus contrarian. Since the so-called anomalies

can be regarded as a failure in traditional asset pricing model, this review begins with

documenting the traditional Sharpe-Lintner CAPM. The following subsections will

then cover an overview of each anomaly tested in Indonesian stock market.

1. The Traditional CAPM

The capital asset pricing model (CAPM) of William Sharpe (1964) and John Lintner

(1965) marks the birth of asset pricing theory. While CAPM seems to offer superior

predictions about how to measure risk and the relation between expected return and

risk, the empirical record of the model is poor.

The originality of CAPM was based on Harry Markowitz’s Modern Portfolio Theory.

Markowitz’s model assumes that investors are risk-averse and their considerations

when choosing among portfolios are the mean and variance of the one-period

investment return. As a result, investors choose “mean-variance-efficient” portfolios,

which are the portfolios that either 1) minimize the variance of portfolio return, given

expected return, or 2) maximize expected return, given variance. Therefore, in the

investment universe of possible portfolios, investors choose the ones with optimal

balance of risk and reward. All the optimal portfolios will construct an efficient

frontier of portfolios.

Sharpe (1964) and Lintner (1965) expand Markowitz’s model by adding two

assumptions. First, the complete agreement assumption: given market clearing asset

prices at t-1, investors agree on the joint distribution of asset returns from t-1 to t.

Second, the borrowing and lending at a risk-free rate, which is the same for all

investors and does not depend on the amount borrowed or lent.

Figure 1 shows the risk-return tradeoff in investment opportunities. The horizontal

axis shows portfolio risk, measured by the standard deviation of portfolio return. The




14

vertical axis shows the expected return. The curve abc provides the combinations of

the minimum-variance-efficient portfolios. On Figure 1, the tradeoff between risk and

expected return for minimum-variance-efficient portfolios are clearly described. For

example, an investor who wants the highest expected return in efficient frontier’s line,

point a, must accept higher volatility than investor in point b. If there is no risk-free

borrowing or lending, only portfolios above b along abc are mean-variance-efficient,

since these portfolios also maximize expected return, given their return variances.

The inclusion of Sharpe (1964) and Lintner`s (1965) assumption of risk-free rate on

abc’s curve turns the efficient set into a straight line. In short, the line can be

explained by the following example: An investor investing the proportion x of

portfolio funds in a risk-free security (��) and 1 − � in some portfolio g. If all funds

are invested in the risk-free security, investors can expect return on the point Rf. By

diversifying the investment in part on risk-free security and in part on risky assets g,

the result is the straight line between �� and g.

Formally, the return, expected return and standard deviation of return on portfolios of

the risk-free asset f and a risky portfolio g vary with x, the proportion of portfolio

funds invested in f, as:

�� = �� + (1 − �)�� (1)

�� = �� + (1 − �) (��) (2)

�� = (1 − �)��, � ≤ 1.0, (3)

Mean-variance-efficient portfolio with risk-free borrowing and lending can be

obtained by drawing a tangent line from�� to the abc curve in Figure 1 with an

intersection on portfolio T, the so-called Capital Market Line. Thus, all efficient

portfolios are combinations of a risk-free asset (either risk-free borrowing or lending)

and a single risky tangency portfolio, T. Overall, all investors have the same

opportunity set and they combine the same risky tangency portfolio T with risk-free

lending or borrowing. Since all investors hold the same portfolio T of risky assets, it

must be the value-weight market portfolio of risky assets. Specifically, each risky




15

asset’s weight in the tangency portfolio, which we now call M, “Market”, must be the

total market value of all outstanding units of the asset divided by the total market

value of all risky assets. In addition, the risk-free rate must be set (along with the

prices of risky assets) to clear the market for risk-free borrowing and lending.

The main implication of CAPM is that in market equilibrium, the value-weighted

market portfolio, M, must be on the mean-variance-efficient frontier. The mean-

variance-efficiency of M in turn says that (i) β, the slope in the regression of a

security’s return on the market return, is the only risk needed to explain expected

return; (ii) there is positive expected premium for β risk.

As emphasized by Fama and French (2004), a risky asset’s return is uncorrelated with

the market return; its beta is zero, when the average of the asset’s covariances with the

returns on other assets just offsets the variance of the asset’s return. Such a risky asset

is riskless in the market portfolio in the sense that it contributes nothing to the

variance of the market return. Sharpe (1964) and Lintner (1965) argue that when there

is risk-free borrowing and lending, the expected return on assets that are uncorrelated

with the market return must equal the risk-free rate ,��. Thus, Sharpe-Lintner CAPM

equation becomes:

(��) = �� + � (��) − ��, � = 1, … , �. (4)

In words, the expected return on any asset i is the risk-free interest rate, ��plus the

risk premium, which is the asset’s market beta, ��, times the premium per unit of

beta risk, (��) − ��. Rearranging equation (4), the Sharpe-Lintner CAPM become,

(��)−�� = � (��) − �� (5)

Thus, Sharpe-Lintner CAPM predicts that an asset expected excess return can be

explained by the risk premium. Alternatively, according to CAPM, the market

compensates investors only for taking systematic risk, which is measured by beta, but

not for taking specific risk.




16

According to Fama and French (2004) there are three standard tests of the CAPM.

The purpose of the tests is to examine whether loadings on a market proxy can

describe the average returns on other portfolios. Tests of the CAPM are based on three

implications of the relation between expected return and market beta implied by the

model. Firstly, expected returns on all assets are only linearly related to their betas,

and no other variable has marginal explanatory power. Secondly, the beta premium is

positive, which means that the expected return on the market portfolio exceeds the

expected return on assets whose returns are uncorrelated with the market return.

Third, in the Sharpe-Lintner CAPM, assets uncorrelated with the market have

expected returns equal to the risk-free interest rate, and the beta premium is the

expected market return minus the risk-free rate.

Empirical evidence, however, shows that CAPM fails to explain the first and second

implication. In practice, beta premium is not always positive. According to Leland

(1999), the fact that beta is negative means that the CAPM model is misspecified

since estimates of alpha will be incorrect. This inverse relationship between beta and

returns seems to contradict the basic theory in empirical finance. The basic SLB

CAPM model generally assumes that the expected return for any securities is a

positive function of only three variables, which are beta, the risk-free rate, and the

expected market return. There are three contradictions of the use of beta as a proxy of

systematic risk for a security. Firstly, Chen, Roll, and Ross (1986) argue that in order

to measure the systematic risks, not only beta, but also macroeconomic variables have

to be taken into account. Secondly, Lakonishok and Shapiro (1986) observe that

security returns are also affected by various measures of unsystematic risk. Finally,

Fama and French (1992) investigate the absence of a systematic relationship between

beta and security returns.

In order to explain the existence of negative risk premium, I follow the arguments

from Pettengill, Sundaram, and Mathur (1995). They argue that the positive

relationship between returns and beta predicted by the Sharpe-Lintner-Black (SLB)

CAPM model is based on expected rather than realized returns. The assumption of a

positive risk-return tradeoff requires that the expected return to the market must be




17

greater than the risk-free rate ( (��) − ��) > 0); otherwise all investors would hold

the risk-free security. This assumption implicitly asserts that the expected return to

any risky portfolio is a positive function of beta. In the real world, however, investors

must consider a nonzero probability that the realized market return will be less than

the risk-free return. Consequently, the relationship between beta and realized returns

can differ from the relationship between beta and expected return required by the

general CAPM model. However, the model does not provide any arguments regarding

the relationship between portfolio beta and portfolio returns when the realized market

return is less than the risk-free return. Pettengill, Sundaram, and Mathur (1995) argue

that if the realized market return is less than the risk-free rate, an inverse relationship

exists between beta and predicted return (i.e., high beta portfolios have lower returns

than those of low beta portfolios). Fletcher (1997) also observes that in periods when

the excess return is positive (up market), there is a positive relationship between beta

and returns and in periods when the excess market return is negative (down market),

there should be a negative relationship between beta and returns.

Furthermore, researchers have identified many patterns in average stock returns, for

example the so called-value premium, size effect, momentum, and contrarian. All

those patterns cannot be explained by the Sharpe-Lintner CAPM and thus, they are

typically called market anomalies.

2. The Value Premium

The topic of value and growth investing has already been a popular concept for

academic research and investment practice. Investment strategies that focus on value

stocks investing are claimed to outperform the market.

2.1 Evidence in U.S. Market

Basu (1977) and Jaffe et al. (1989) find that stocks with low P/E values can generate

positive abnormal returns. Basu (1977) reports that P/E information was not “fully

reflected” in security prices, consequently, securities trading at different multiples of

earnings, on average, seem to have been inappropriately priced vis-à-vis one another.




18

This evidence violates the semi-strong form of the efficient market hypothesis. Fama

and French (1992 and 1996) report that value strategies (buying stocks that have low

prices relative to earnings, dividends, historical prices, cash flow, book value, other

measures of intrinsic value) yield higher returns than growth strategies. Davis (1994)

finds that BE/ME, E/P, and CF/P have significant explanatory power with respect to

the cross-section of realized stock returns.

2.2 Evidence in International Markets

The evidence in international markets largely supports the existence of value

premium. Capaul et al. (1993) conclude that the returns on value stock portfolios (low

P/B) outperform those from growth stocks portfolios (high P/B) in the six countries

they observed (France, Germany, Switzerland, the U.K., Japan and The U.S. Haugen

and Baker (1996), which confirm that value premium in international markets exists

even after controlling for risk, liquidity, growth potential and past price history. Fama

and French (1998) examine out-of-sample test in sixteen emerging markets and found

that sorting on BE/ME, E/P, CF/P, and D/P, value stocks outperform growth stocks.

Bauman, Conover, Miller (1998) extend Capaul et al. (1993) to encompass all of the

20 established markets represented in the MSCI Europe/ Australasia/ Far East (EAFE)

Index, as well as Canada, and to examine value and growth stocks on the basis of P/E,

P/CF, and dividend yield (D/P). They find that value portfolios outperform growth

portfolios in most countries by a large margin. However, investors should carry out

additional financial research when making final international investment selections

since value stocks as a group do not provide the best performance every year and in

every market.

2.3 Evidence in Asian Markets

Many studies support the existence of the value premium in Asian markets. However,

the empirical evidence is inconclusive. Chan, Hamao, and Lakonishok (1991) observe

Japanese stock market and find positive significant relationship between BE/ME and

CF/P and expected returns. However, they are not able to determine unambiguously

whether the returns predictability is due to market inefficiency or deficiencies in the




19

asset-pricing model. Yen, Sun, and Yan (2004) find value premium for Singapore

stocks, with the premium being concentrated in the first 2 years after the portfolio

formation. Ding et al. (2005) examine value and growth portfolios in seven East Asian

countries before the Asian Financial Crisis and find insignificant or zero value

premium in Indonesia and Taiwan; a negative premium in Thailand, but significant

positive value premium in Hong Kong, Japan, Malaysia and Singapore. In contrast,

Bauman et al. (1998) report insignificant value premium for Hong Kong and

Singapore. Brown, Rhee, and Zhang (2008) investigate value premium in four Asian

markets: Hong Kong, Korea, Singapore, and Taiwan. They find that value premium

becomes greater in the post-crisis period across all four countries, indicating that high

volatility during the crisis period did understate the value premium. More advanced

research in value premium in Japan is being conducted by Bauer, Derwall, and

Molenaar (2004). They examine whether the short-term variation in the Japanese size

and value premium is sufficiently predictable to be exploited by a timing strategy.

Their results confirm the evidence of sufficient predictability. However, the benefits

are more pronounced under lower transaction costs.

Nevertheless, the underlying reasons that value stocks (those with high BE/ME, low

P/E, low P/B, low P/D ratios) earn superior return than growth (glamour) stocks

remain a puzzle in asset pricing. Some researchers argue that the observed

predictability is an artifact of the research design and database used to conduct the

study. Kothari, Shanken, and Sloan (1995) argue that the predictability of certain

variables would be reduced or vanish if different methodology and data were used.

Davis (1994) addresses this problem by using a dataset that eliminates some of these

concerns and still find significant relationships between certain fundamental variables

(BE/ME, CF/P, and E/P) and subsequent returns during his study period.

The second explanation on why value strategies have produced superior returns

compared to its relative, growth or glamour strategies is that they are fundamentally

riskier. This rational interpretation is being supported by Fama and French (1992).

They argued that value premium is the compensation for bearing risk. Fama and

French (1996) argued that stocks with high BE/ME are more prone to financial




20

distress and are hence riskier than growth stocks. Chan and Lakonishok (2004),

however, argue that the risk argument stretches credulity. On the basis of the risk

argument, Internet stocks, which had virtually no book value but stellar market value

in the 1990s, would be considered much less risky than traditional utility stocks,

which typically have higher BE/ME. In addition, there is also data snooping evidence

since the idea that value stocks have higher risk appeared only after their higher

returns became apparent. Lakonishok, Shleifer, and Vishny (1994) use stock market

performance as a proxy of risk. If the value strategy is fundamentally riskier, it should

underperform growth strategy during undesirable states of the world when the

marginal utility of wealth is high. They report that there is no evidence that superior

value stocks performance reflect their higher fundamental risk.

In accordance to the behavioral model, Kahneman and Tversky (1982) explain that

compared to value stocks, growth stocks are characterized as having higher recent

growth rates in earnings per share and market appreciation. They suggest that

forecasters tend to put more emphasis on more recent information relative to older

data. So far as investors and analysts overlook the lack of persistence in growth rates

(Chan, Karceski, and Lakonishok 2003) and project past growth into the future,

favorable sentiment is created for growth (glamour) stocks and drive overvaluation of

growth stocks. In favor of overreaction hypothesis, Bauman et al. (1998, 1999) argue

that value premium in international stock markets is due to investors and research

analysts overreacting to past corporate earnings trends of growth stocks and value

stocks. Value premium occurs because investors fail to recognize that corporate

growth trends have a mean-reversion tendency or behave as a random walk.

Consequently, when earnings disappointments are reported, growth stocks tend to

have lower returns than value stocks.

Lakonishok et al. (1994) claimed, however, that value strategies might produce higher

returns because they are contrarian to “naive” strategies followed by other investors.

These naïve strategies tend to overestimate future earnings of growth stocks relative

to value stocks. The investors are overly optimistic about stocks that have done very

well in the past, so investors bid their prices up and make the price of growth stocks




21

became overvalued. On the other hand, value stocks, which have past bad earnings,

are undervalued. Naïve investors make systematic errors in predicting future growth

in earnings of value stocks, and this error in expectation and overly reaction to bad

and good stocks are the cause of value premium. This non-risk-based explanation is

known as the “extrapolation” or “errors-in-expectations” explanation.

Chan and Lakonishok (2004) suggest that value premium may be evidence of agency

factors. Analysts have self-interest in recommending successful stocks to generate

trading commissions. Moreover, growth stocks are typically in exciting industries and

are subject to analyst reports and media coverage. Furthermore, analysts’

performances are being evaluated on a short-term basis and are subject to investors’

scrutiny. Since value stocks reward investor in the longer-term, their short-term

performance is not so astonishing and investors will put pressure for analysts’

holdings of these stocks. The upshot of all these considerations is that value stocks

become underpriced and growth stocks overvalued relative to their fundamentals.

Shleifer and Vishny (1997) provide the argument of why these mispricing patterns

can persist over long periods of time. They argue that such market anomalies may

diminish when investors become knowledgeable about the strategy being used. The

anomalies are more easily accepted when the returns pattern is not very noisy (not so

volatile) and the payoff horizon is short. However, value-growth anomaly is accepted

only slowly, even by relatively sophisticated investors.

Other argument about the superiority of value stocks is proposed by Doukas, Kim,

and Pantzalis (2004), who argue that higher returns for value stocks are exposed to

greater disagreement among investors, which influences stocks prices. They suggest

that in imperfect capital markets, capital market equilibrium requires the simultaneous

determination of asset prices and of the identity of investors (that is, investors’

opinions) trading in each asset. This dispersion of opinion, then, potentially represents

a unique source of risk, and its impact on prices should be compounded by the degree

of disagreement. To examine the investors’ disagreement about future payoffs, they

used the dispersion in analysts’ earnings forecasts as a proxy for investors’

heterogeneous beliefs. Value stocks are those which have greater exposure to




22

dispersion in analysts’ forecasts and thus should earn a higher return. Similarly,

glamour stocks are those that have lower exposure to dispersion in analysts’ forecasts

and thus earn a lower return. They found that value strategies outperform growth

strategies and the return advantage of value strategies reflects, at least in part,

compensation for bearing risk associated with higher dispersion in analysts’ earnings

forecasts, which confirms the conjecture of Williams (1997) that dispersion of opinion

represents risk. Hence, the superior return of value stocks should be viewed as a

reward for the greater investor disagreement about the stocks’ future growth in

earnings.

3. The Size Effect

The other form of market anomaly is the size effect, in which small capitalization

stocks have higher mean returns than large capitalization stocks. Some researchers

argue that small capitalization stocks are actually inherently riskier than large

capitalization stocks because there is less information about them, and standard

measurement of risk do not capture this discrepancy in the amount of available

information. Small capitalization stocks outperformance can be observed in the form

of neglected-firm-effect, in which firms with low analyst coverage tend to have

higher-risk-adjusted returns (Chincarini and Kim (2006)).

The researches on size effect are first reported by Banz (1981), who argued that

smaller firms have had, on average, higher risk adjusted returns than larger firms. He

conjectures that one possible explanation for the size effect is that some firms have

insufficient available information which hinders investors to hold their stocks. Thus,

lack of information about small firms leads to limited diversification and therefore

higher returns for the undesirable stocks of small firms.




23

Jaffe et al. (1989) examine the impact of firm size on P/E. Jaffe et al. find the size

effect to be significantly negative only in January, the so-called January Effect.5 Fama

and French (1992) also demonstrate that in the U.S. market, firm size provides a good

explanation of cross-sectional stock returns. Fama and French (1993, 1995) reported a

continuous finding about the negative firm size effect which arises because firm size

and the book-to-market ratios are proxies for non-diversifiable factor risk.

The other interpretation of the size effect is provided by Chan and Chen (1991) who

argue that smallness by itself does not necessarily mean higher risk, and differences in

market capitalizations do not explain why small and large firms respond differently to

economic news. They argue that small firms are riskier because they tend to be what

we call marginal firms. They have lost market value because of poor performance,

they are inefficient producers and they are likely to have high financial leverage and

cash flow problems. Marginal firms tend to have higher price sensitivity to economic

changes and they are less likely to survive adverse economic conditions.

Consequently, small-marginal firms will react differently than large-healthy firms to

the same piece of macroeconomic news. As a result of the difference in production

efficiency and difference in leverage, small firms tend to be riskier than large firms,

and the risks of the smaller firms are not likely to be captured by a market index

heavily weighted toward large firms.

4. Momentum versus Contrarian Strategies

Many researchers have confirmed that past stock performances have the ability to

predict future performance. Researchers define two strategies based on stocks’ past

performance; the so-called momentum strategy which is based on price continuation

and the contrarian strategy that follows the reversal pattern. Momentum strategy

5 January effect is one of market anomalies in which small capitalization stocks and last year’s poorly

performing stocks tend to outperform in January. According to Lee (1991), January effect may occur

from investors’ behavior in the end of the year that sell stocks that have declined in price during the

year to reduce their taxes, thereby putting downward pressure on stock prices. However, as soon as the

year ends, investors start to repurchase the stocks which already undervalued in prices. Consequently,

the selling pressure ends and hence stock prices rebound to their equilibrium levels, resulting in

abnormally high returns in early January.




24

refers to buying stocks that did well in the past and contrarian strategy refers to

buying stocks that have performed badly in the past.

In analysing the performance of momentum versus contrarian, this study uses two

methodologies, Lo and MacKinlay (1990) methodology and Jegadeesh and Titman

(1993) methodology. Lo and MacKinlay define “winners” as stocks that outperform

the market during the formation period and define “losers” as those that underperform

the market during the formation period. Jegadeesh and Titman methodology ranks

stocks in ascending order based on their past performance. “Winners” refer to stocks

that fall into portfolio with the highest 30 percent rank whereas “losers” refer to

stocks that belong to portfolio with the lowest 30 percent rank. In other words,

momentum strategy refers to buying winners and selling losers, whereas contrarian

strategy refers to buying losers and selling winners.

4.1 Momentum and Contrarian Evidences in U.S. Stock Market

The literature of stock market anomalies in the form of “contrarian” and “momentum”

strategies is largely attributable to the seminal work of DeBondt and Thaler (1985).

They investigate return patterns over an extended period of time and find that

contrarian strategies contribute abnormal returns over three to five year horizons.

Initially, the contrarian profits were represented as a long-run phenomenon. More

recent findings by Jegadeesh (1990) and Lehmann (1990), however, provide evidence

of short-term return reversals. Jegadeesh (1990) documents contrarian profits of about

2 percent per month from a strategy that buys and sells stocks based on their prior

monthly returns and holds them for one month. Lehmann (1990) provides similar

findings using weekly portfolio formation.

On the other hands, Jegadeesh and Titman (1993, 2001) find that momentum

strategies generate positive returns about 1 percent per month over three to 12 months

holding period. However, they found that the excess returns of these past winners and

losers in the year following the portfolio formation date dissipate within the following

two years. More broad investigation in the U.S. market by Conrad and Kaul (1998)




25

find that both momentum and contrarian strategies contribute to significant profits

depending on the time horizon investigated. In conclusion, the contrarian strategy is

profitable for short-term (weekly, monthly) and long-term (two to five years, or

longer) horizons, while momentum strategy is profitable for intermediate-term (three

to 12 months) intervals.

Similar studies have also been conducted on the international stock markets with

various conclusions. Rouwenhorst (1998) finds profitable intermediate-term

momentum strategies in twelve European markets between 1980 and 1995. He also

observes that U.S. and European markets have common factors that drive the

profitability of momentum strategies.

In contrast with U.S. and European markets, the evidence of momentum and

contrarian profits is much weaker in Asian markets. Chui, Wei, and Titman (2000)

investigate some Asian markets and find momentum profits except for Japan and

Korea. Griffin, Xiuqing, and Martin (2003) find that Asian stock markets display the

weakest momentum returns among 39 international markets.

4.2 Asymmetric Reactions to good news as opposed to bad news

Based on the argument that “winners” and “losers” do not necessarily evoke the same

investor responses, nor do they share similar patterns in price movements, some

researchers evaluate the performance of “winners” and “losers” independently. T.H.

McInish et al. (2008) provide more integrated approach to momentum and contrarian

strategies in seven Asian stock markets during the period between 1990 and 2000.

They present evidence of the behavior of each winner stocks and loser stocks. They

found that winners display contrarian behavior while losers show subsequent price

continuation in all seven countries except for Taiwan and Korea. Nam, Pyun, and Kim

(2003) report that winner have longer mean reversion pattern than loser in the nine

Pacific Basin markets they studied.




26

4.3 Trading Volume and Momentum versus Contrarian Strategies

Both academics and practitioners have recently combined the use of trading volume in

examining both momentum and contrarian strategies. The use of trading volume in

predicting future stock returns is more relevant in emerging stock markets where thin

trading is more pervasive. The inclusion of trading volume is strongly subject to non-

synchronous trading problems as noted by Lo and MacKinlay (1990). Chordia and

Swaminathan (2000) investigates that trading volume is a significant determinant of

the cross-autocorrelation patterns in stock returns. Blume et al. (1994) present a model

where investors can use volume information in addition to historical price information

in projecting future price changes, suggesting an information signaling role of volume

in return predictability. Fung, Leung, and Patterson (1999) who incorporate trading

rule in their strategies find that as opposed to T.H. McInish et al. (2008), daily

“winner” exhibits momentum pattern whereas “loser” have reversal pattern during the

ensuing one to five trading days.6 Hameed and Ting (2000) examine the relation

between short-horizon (weekly) return predictability and the level of trading activity

(trading volume) in the Malaysian stock market. They find that contrarian profits on

actively traded stocks are significantly higher than those on low trading activity

stocks. Lee and Swaminathan (2000) present the evidence that high volume stocks are

more favorable than low volume stocks for both investors and analysts. Conrad,

Hameed, and Niden (1994) find that high volume securities experience price

reversals, while low volume securities experience price continuations.

4.4 Sources of Momentum and Contrarian Profits

The profitability of both momentum and contrarian strategies have been well provided

and accepted by many researchers and practitioners. However, the sources of profits

as well as the interpretation of the evidence are widely debated. Some argue that

momentum and contrarian profits are the evidence of market inefficiency; while

6 It should be noted that the momentum behavior (reversal) for winner (loser) stocks as observed by

Fung, Leung, and Patterson (1999) are being criticized as a result of research design. Their study

period, 1980 until 1993, is a bullish market period. Therefore the higher proportion of daily return of

stocks observed should be positive.




27

others argue that the returns from these strategies are either compensation for risk, or

alternatively, the product of data mining.

Alternative explanations of factors that drive momentum or contrarian strategies:

1. Data mining

The criticism that observed market anomalies are due to data mining is typically

the hardest to address because of the limitation of available data. Fortunately,

with the passing of time, there are additional periods of data that enable

researchers to perform out-of-sample tests as well as to assess the extent to which

investors may have learned from the earlier return pattern.

Jegadeesh and Titman (2001) perform the same research using longer observation

time and found the same result that momentum strategies continue to be

profitable and that past winners outperform past losers by about the same

magnitude as in the earlier period. These kinds of anomalies are argued to be the

result of investors´ under-reaction which in intermediate term, past winners still

outperform past losers.

2. Behavioral Model

Under-reaction

Chan, Jegadeesh, and Lakonishok (1999) observe that momentum strategies may

be profitable because they exploit market under-reaction to different pieces of

information. Their results indicate that market is slow to incorporate in its

valuations the full impact of information. For instance, market may under-react

both to information about the short-term prospects of companies that will

ultimately be manifested in near-term earnings and to value-related information

about the long-term prospects of companies that have not yet fully captured by

near-term earnings forecasts or past earnings growth. In addition, they also

provide evidence that analysts are especially slow in revising their forecast.

The behavioral models advocates, Barberis, Shleifer, and Vishny (1998); Daniel,

Hirshleifer, and Subrahmanyam (1998); and Hong and Stein (1999) argue that




28

momentum and contrarian profits may arise because of inherent biases in the way

that investors interpret information. The behavioral models imply that the holding

period abnormal returns arise because of a delayed overreaction to information

that pushes the prices of winners above their long-term values and presses the

prices of losers below their long-term values. On the other hand, for intermediate-

term momentum profits, Jegadeesh and Titman (1993, 2001) and Chan,

Jegadeesh, and Lakonishok (1996) argue that the prime source of momentum

profits is investors’ under-reaction to market news.

Previous behavioral literature, for example, Barberis et al. (1998) argue that it is

“conservatism bias” which might lead investors to under-react on ranking period

information, which, in turn, leads to momentum profit on medium-term. The

conservatism bias suggests that investors underweigh new information in

updating their priors and prices will tend to slowly adjust to information.

However, once the information is fully incorporated in prices, there is no further

predictability in stock returns and post-holding period will be zero.

Recent behavioral literature provides evidence that the post-holding period

returns may in fact be negative. Behavioral models predict that in subsequent

time periods, when the stock prices of the winners and losers revert to their

fundamental values, the returns of losers should exceed the returns of winners.

Barberis et al. (1998) try to explain the long-term reversal as well as the short-

term momentum by combining conservatism bias and “representative heuristic”

by Kahneman, Slovic, and Tversky (1982).7 They argue that the representative

heuristic leads investors to mistakenly extrapolate past performance earning

growth too far in the future. They explain that although the conservatism bias in

isolation leads to under-reaction, this behavioral tendency in conjunction with the

representative heuristic can lead to long horizon negative returns for stocks with

consistently high returns in the past.

7 Representative heuristic is the tendency of individuals to identify an uncertain event, or a sample, by

the degree to which it is similar to the parent population.




29

Herding Behavior

Herding behavior suggests that investors tend to “flock” together. This behavioral

model is more pronounced in the mutual funds. Grinblatt, Titman, and Wermers

(1995) observe that majority of mutual funds purchase past winner stocks and

tend to invest more intensely “with the crowd”. Lakonishok et al. (1994) find

evidence that pension fund managers tend to either buy or sell in herds, with

slightly stronger evidence that they flock around small stocks.

Self-Attribution Bias

Daniel, Hirshleifer, and Subrahmanyam (1998) suggest that informed traders

suffer from a “self-attribution” bias. Informed traders tend to attribute good

performance to their stock selection skills and attribute bad performance to bad

luck. Due to this cognitive bias, investors become overconfident and overestimate

their prediction for good past performance stocks. This overreaction pushes up

the prices of winner above their fundamental value, and leads to short-term

momentum profits that are eventually reversed in the long-term.

3. Cross-sectional variations in stock returns

Conrad and Kaul (1998) report that stocks with high unconditional expected rates

of return in adjacent time periods are expected to have high realized rates of

returns in both periods, and vice versa. Based on that hypothesis, Conrad and

Kaul (1998) argue that momentum strategies will yield positive returns on

average even if the expected returns on stocks are constant over time. They

suggest that the higher returns of winners in the holding period represent their

unconditional expected rates of return and thus predict that the returns of the

momentum portfolio will be positive on average in any post-ranking period.

4. Lead-Lag Structure in Portfolio Returns

Lo and MacKinlay (1990) provide evidence that contrarian profits are due to a

lead-lag structure in portfolio returns. The idea is that some securities adjust

slowly (under-react) to information and this under-reaction contributes to

asymmetric predictability in size-sorted portfolio returns. For example, if the

price changes of stock A lead that of stock B, a contrarian profit may exist from




30

buying stock B subsequent to an increase in stock A and selling stock B

subsequent to a decline in stock A. Chordia and Swaminathan (2000) attribute the

lead-lag patterns to low-volume stocks adjusting more slowly than high-volume

stocks to market wide information shocks.

5. Compensation for Risk

Chan (1988) suggests that an investor who follows the contrarian strategy earns

abnormal returns as a compensation for the risk in the investment strategy and the

investor’s risk exposure varies inversely with the level of economic activity. The

contrarian strategy appears to have an ability to pick riskier losers when the

expected market risk premium is high, probably because losers suffer larger

losses at economic downturns than at upturns. The other proponents of risk, Ball

and Kothari (1989) provide evidence in U.S. market that loser stocks have

significantly higher betas than winner stocks in the test period. The consensus in

the literature argues that risk factors do not completely explain intermediate

horizon momentum profits. Fama and French (1996) concede that their three-

factor asset-pricing model as an approach for risk factors cannot explain the

medium-term momentum profits. Chordia and Shivakumar (2002) incorporate

macroeconomic variables and find that momentum profits can be explained as

compensation for macroeconomic risk factors. However, Griffin et al. (2003) find

that the relation between macroeconomic risk exposure and momentum profit

does not exist in international markets. Rouwenhorst (1998) examines 12

European markets and find that controlling for market risk or exposure to a size

factor increases the abnormal performance of medium-term momentum

strategies. Richards (1997) presents evidence that loser portfolios have lower

standard deviation than winner portfolios at all horizons.

6. Size Effect and Seasonality

Zarowin (1990) argues that superior performance of losers compared to winners

is not due to investor overreaction, but instead is an evidence of the size and/or

January effect. He provides that by the end of the ranking period, losers tend to be

smaller-sized than winners.




31

III. Data Analysis

This section begins with an overview of the Indonesian Stock market. Afterwards, the

methodology employed in this study is introduced. The final subsections comprise the

detailed explanation of the empirical tests in this research.

1. Indonesia Stock Exchange

Indonesian stock market has existed long before the independence of Indonesia

(August 17, 1945). The first stock exchange in Indonesia was established in Batavia

(currently known as Jakarta) by the Dutch East Indies on December 14, 1912.

However, the first computerized trading system was introduced on 1995. Therefore,

the earliest data of Indonesian stock market used in this study has only been available

since 1991. Figure 2 shows the milestones of Jakarta Composite Index from 1984 to

2008.

As emphasized by Kothari, Shanken, and Sloan (1995), some databases suffer from

survivorship-bias due to the exclusion of delisted or non-surviving firms from the

database. Since only the surviving firms’ historical data are added to the database, the

true economic significance of the variables will likely be overstated. This study

potentially suffers from the same bias. The potentially serious bias is due to the non-

availability of some delisted stocks in DataStream database. The impact of the

survivorship bias however, is not significant since the number of delisted companies

is not large enough to provide meaningful differences in the results.

2. Data

The universe of stocks is the Indonesia Stock Exchange (JKSE, Jakarta Composite

Index). The sample period covered in this study ranges from the end of December

1991 to the end of December 2008. In this research, some sample selection criteria is

used as follows. Firstly, following Brown, Rhee, and Zhang (2008), the sample is

limited to listed domestic common stocks on the main and the secondary boards by

excluding preferred stocks, investment funds, unit trusts, exchange traded funds, and

over-the-counter stocks in the index. The reason of the exclusion of preferred stocks,




32

unit trusts, exchange traded funds and over-the-counter stocks is due to their

performance may be influenced by specific factors that are different than those of

common stocks in general. For instance, preferred stocks have privilege when it

comes to the payment of dividends. In the event of a company’s liquidation, preferred

stockholders get paid before those who own common stocks. In addition, in case of

bankruptcy, preferred stockholders have priority distribution of the company’s assets,

whereas common stockholders don’t receive company’s assets unless all preferred

stockholders have been compensated. Secondly, following Ding et al. (2005), the

newly added firms (IPO stocks) are excluded since they lack of available data and are

more volatile and unpredictable than stocks in general. This study does not include

IPO companies which went public in 2008 since the behavior of IPO stocks cannot be

a reliable measure of market anomalies. The number of remaining stocks in each

portfolio in every formation period, after using the selection criteria, is changing over

time due to the development of the financial market.8 However, the minimum number

of stocks represents each portfolio in every formation period is ten stocks.9 In the

beginning of the sample period, only 59 stocks are included in the sample. As of

December 2008, the entire sample includes 351 companies (excludes the IPO stocks

in 2008).

The monthly returns data and the monthly accounting data are collected from

DataStream and Thomson Financial. As stated before, further detailes of all the

variables are presented in Table I. The accounting data consists of Book-to-Market

Equity ratio (BE/ME), Earnings to Price ratio (E/P), Cash Flow to Price ratio (CF/P),

market value of equity (ME), and trading volume. The proxy for risk-free interest rate

is Indonesia’s Government one-month deposit rate. This monthly series of interest

rate are taken from the publication of Indonesia Central Bank.

8 It should be noted that different number of stocks in each year can influence the performance of

market anomalies since early sample period is comprised of insignificant number of stocks. 9 The exceptions are for portfolio zero which depends on the negative value in stock characteristics and

for portfolios in two-independent sort strategies.




33

3. Methodology

Firstly, this section will explain how to investigate the existence of the value premium

using individual sorting. The procedures in portfolios formation are as follows:

Following Lakonishok, Shleifer, and Vishny (1994), at the end of each December

between 1991 and 2008, stocks are ranked in ascending order based on BE/ME, E/P,

and CF/P into quintiles. Therefore there are three different ranks and in total there are

15 portfolios (five quintiles for each stocks rank). In this study, these 15 portfolios are

called the LSV portfolios. The first quintiles are signed for glamour stocks (lowest

BE/MEs, E/Ps, and CF/Ps), the second quintiles are the next glamour stocks, and the

fifth quintiles refer to value stocks (highest BE/MEs, E/Ps, and CF/Ps). Since all the

accounting data except BE/ME are obtained in the form of P/E, and P/CF, the inverse

of all those accounting data is taken in order to make the rank of quintiles for glamour

and value stocks. In addition, following Fama and French (1992), portfolio zero is

formed for stocks that have negative value on BE/ME, E/P, and CF/P. Consequently,

there are three portfolios zero (one portfolio zero for each accounting data).

Secondly, in order to investigate the presence of the size effect, similar procedures in

portfolios formation are applied. At the end of each December between 1991 and

2008, stocks are ranked in desscending order based on market capitalization (ME) into

quintiles. Consequently, there are five size portfolios. The first quintile consists of

stocks with the highest market capitalization (big stocks) and the fifth quintile

comprised of stocks with the lowest market capitalization (small stocks).

The rank for both LSV portfolios and size portfolios are rebalanced every December.

This study utilizes an annual buy and hold strategy and presents both equal-weighted

and value-weighted average monthly returns. The use of value-weighted average

monthly returns is reasonable since stocks with larger market capitalization should

rationally represent a larger portion of the portfolio returns. The weigh of the value-

weighted portfolios is based on the stock’s market capitalization at the end of the

formation period. Using equal-weighted returns might also present problems as some

stocks may have a relatively small market capitalization, where an absurd situation of




34

buying more than the stock’s available market capitalization could arise. Loughran

(1997) finds that, for U.S. stocks, the value premium persists but is decreased when

value-weighted returns are used.

Now, I will explain the procedures to calculate the average monthly returns for each

of both LSV and size portfolios in order to examine the presence of the value

premium and the size effect. The average monthly stocks returns in each quintiles

portfolios are computed for Years +1, +2,…, +5 relative to the time formation.

Specifically, the computation process is as follows: I calculate the average monthly

returns for one-year holding period from December in year � to December in year � +1 , the average monthly returns for two-year holding period with one portfolio

rebalancing in December in year � + 1, the average monthly returns for three-year

holding period with two rebalancing in December in year � + 1 , and � + 2 , the

average monthly returns for four-year holding period with three rebalancing in

December in year � + 1, � + 2, and � + 3, and finally the average monthly returns for

five-year holding period with four rebalancing in December in year � + 1, � + 2, � +3 and � + 4 . In order to be included in the portfolio for a given year (particular

formation period), a stock must have the accounting data on that portfolio-formation

variables only for that particular formation period when the buy and hold strategy is

conducted. The results for the value premium with individual sorting are summarized

in Table III and the results for the size effect are presented in Table V.

This section continues to explain the procedures to observe the value premium using

the two-independent sort strategies; sorting stocks independently using two

accounting variables. Sorting stocks based on only one characteristic may not fully

capture the anomalies being studied. Lakonishok et al. (1994) argue that sorting

stocks on two accounting variables more accurately distinguishes between strong and

distressed stocks, and produces larger spreads in average returns. Portfolios sorted on

two different stock characteristics are analyzed since stocks may have different

growth rate performance in the past and in the future. CF/P and E/P are used as a

proxy for future expected growth rate, while BE/ME is used as a proxy for past

growth rate. Value stocks are defined as stocks which have low growth in the past




35

(high BE/ME) and are expected by the market to continue growing slowly (high CF/P

and high E/P); the reverse is true for glamour stocks. The procedures to perform the

two-independent sort strategies are as follows. Firstly, stocks are independently sorted

into three portfolios (bottom 30 percent, medium 40 percent, and top 30 percent) by

BE/ME and by CF/P; then by BE/ME and by E/P, and then take the intersections

resulting from the two classifications. The results are nine sets portfolios for each of

two-independent sort methods. The average monthly returns of each portfolio are

calculated with same procedure as that of one-sort classification portfolio. The results

of two-independent sort strategies are summarized in Table IV.

Furthermore, this section explains the procedures to observe and to measure the

existence of momentum versus contrarian. In examining momentum and contrarian

strategies, this study uses monthly returns on individual stocks, volume turnover,

number of shares outstanding, and market capitalizations. The decision to use the

level of trading activity is relevant for emerging markets where thin trading is more

prevalent. Chordia and Swaminathan (2000) indicate that trading volume is a

significant determinant of the cross-autocorrelation patterns in stock returns. Hameed

and Ting (2000) argue that there are two measures of trading to gauge the level of

trading activity for security � over year � − 1:

a. average daily trading volume, &�', where &�' = number of shares traded in year

� − 1 divided by number of trading days in year � − 1; and

b. Percentage of days stocks was traded, (�', where (�' = number of days share

was traded in year � − 1/ number of trading days in year � − 1.

The first measure,&�', gives an indication of how heavily security � is traded in the

market on an average day during year � − 1. A low value of &�' suggests that the

security is thinly traded, while a high value of &�' indicates a heavily traded security.

The second measure,(�' , indicates how frequently the security is traded during the

year. A low value of (�' suggests that security � is infrequently traded, while a high

value means that security � is frequently traded.




36

This research however, will use the lagged monthly trading volume as defined by

McInish et al. (2008).10

McInish et al. (2008) measure trading volume based on the

turnover ratio (number of shares traded divided by number of shares outstanding)

instead of the simple raw volume. The use of monthly return is to avoid non-

synchronous trading and to alleviate concerns of non-trading. The decision to use

trading volume turnover instead of the raw volume is to disentangle the effect of firm

size from trading volume, as argued by Chordia and Swaminathan (2000) that raw

trading volume is highly correlated with firm size.

In evaluating momentum and contrarian strategies, this study will use two different

methodologies which are Lo and MacKinlay (1990) methodology and Jegadeesh and

Titman (1993) methodology. Jegadeesh and Titman (1993) methodology is presented

in section V, in the robustness check.

In this section, I will explain in details the procedures of Lo and MacKinlay (1990)

methodology. The investment strategy is to buy stocks in proportion to their returns

over the ranking period. In the momentum strategy, the investor takes a long position

in positive-past return stocks, with higher weight on top performers and at the same

time the investor takes a short position in negative-past return stocks, with higher

weight on bottom performers. On the other hand, on the contrarian strategy, the

investor takes a long position in negative-past return stocks, with higher weight on

bottom performers and at the same time, investor takes a short position in positive-

past return stocks, with higher weight on top performers.

Stocks are classified as winner if their average returns during the formation period

outperform those of the market (�)�*' − �)�*' ) > 0.11

While losers refer to those that

underperform the market during the formation period. F refers to formation periods

and H refers to holding periods.

10

Lagged trading volume refers to trading volume during the formation period. Lagged trading volume

and trading volume are used interchangeably. If not stated otherwise, trading volume means lagged

trading volume. 11

Market return refers to equal-weighted return of all stocks available (*) during the formation period;

the remaining of the stocks after performing the selection criteria.




37

During each study period t, each stock is assigned a weight of

+�' = (,-).�)�*' − �)�*' / (6)

Where �)�*' = ,* 0 ��,1'2,13'2* , is the average returns of stock � during the formation

period from � − 4 to � − 1 and �)�*' = ,* 0 ��,1'2,13'2* ,is the market average returns

during the formation period from � − 4 to � − 1, and * is the number of stocks in the

sample in each formation period. The profit of the investment strategy, denoted as 5',

is

5' = 0 +�'-�3, �)�6' (7)

Where �)�6' = ,6 0 ��,1'762,13' , the average returns of stock �during the holding period

from � to � + 8 − 1. The procedures are as follows: at the beginning of each month �,

JKSE stocks are assigned to winners/losers based on their performance during the

formation period from time � − 4 to � − 1. A stock that has average return higher than

market return during the formation period is classified as winner and will be assigned

a positive weight by construction. Conversely, those which have average returns

lower than the market return will be considered as losers and get negative weight by

construction. The larger the difference (whether it is positive or negative) between

stock’s past-average returns and market return, the larger the weight assigned to that

particular stock. Therefore, stock with higher past-average returns has larger portion

in the portfolio. Momentum/ Contrarian profits are the stocks average returns during

the holding period from � to � + 8 − 1 times their weight. For simplicity, this study

implements strategies for which the length of the formation and the length of holding

period are the same. For example, F/H: 1/1 months, 3/3 months, 6/6 months, and

12/12 months. Because of the length of the sample period, this study concentrates

only on patterns in medium-term returns. The procedure can be illustrated as follows,

in a 3/3 strategy, at the beginning of December 1991 (�), stocks are classified as

winners/losers based on their average monthly returns from September 1991 (� − 4)




38

to November 1991 (� − 1). If the past-3 months average returns of stock � is higher

than the market return, then stock � is classified as winner; otherwise loser. Finally,

momentum/ contrarian returns are computed by multiplying the weight and the

average returns from December 1991(�) to February 1992 (� + 8 − 1). Portfolios are

rebalanced every month. In addition, to increase the power of the analysis, the

strategies include portfolios with overlapping holding periods.

A positive value of 5' indicates momentum profit. On the other hand, since a

contrarian portfolio has a short position for winners and a long position for losers, a

negative value of 5' reflects contrarian profit. This strategy will lead to a zero-cost

portfolio since,

; +�'-�3, = 0

Therefore the strategy used in this research can be seen as a zero-investment strategy

in which momentum strategy will buy winner and short loser, whereas contrarian

strategy will buy loser and sell winner. The results of momentum and contrarian

strategy using Lo and MacKinlay (1990) methodology are provided in Table VI.

The employed methodology (Lo and MacKinlay (1990)) differs from previous studies

in term of the weighing process. Common past researches, defined winners/losers for

stocks that fall into the highest/lowest rank in the formation periods (e.g., Jegadeesh

and Titman (1993)). Consequently, stocks that are classified as winners do not

necessarily outperform the market. The weighing process in this research therefore

gives clearer picture of winner and loser stocks as well as provides better momentum

and contrarian strategy. However, in order to give comprehensive evidences,

momentum versus contrarian strategy using Jegadeesh and Titman (1993)

methodology is also presented in the robustness check section.

This section continues to explain the procedure to measure the performance of

momentum and contrarian strategies using Lo and MacKinlay (1990) methodology

that incorporates trading volume. Firstly, JKSE stocks are sorted into three groups




39

(High, Medium and Low trading volume) based on their lagged trading volume.

Secondly, within each volume group, stocks are classified as winner/loser and

momentum/contrarian profit are calculated using the same procedure as explained

above. The results of momentum versus contrarian strategy that incorporates trading

volume are presented in Table VII. It should be noted that the volume classification of

this study is different from that of Conrad, Hameed, and Niden (1994), who assigned

stocks into high and low volume groups based on whether the trading volume in the

formation period is higher or lower than its historical average. Following Hameed and

Ting (2000), this study argues that the procedure of volume classification in this study

is more appropriate in capturing different stocks behavior in different trading activity

levels especially in Indonesian stock market where thin trading is more prevalent. For

example, using Conrad et al. (1994) methodology, stocks is assigned to high volume

group, could represent thinly traded stocks in a particular formation period.

Conversely, stocks classified as low volume compare to their historical average, might

be heavily traded in the particular formation period, relative to other stocks in the

market.

4. Time- Series Return Regression

In order to examine whether the simple asset pricing model is able to explain the

existence of three well known market anomalies in Indonesian stock market,the

CAPM time series regression are tested to three groups of portfolios. The time-series

regressions are conducted using Eviews version 3.1. These groups of portfolios are (1)

Fama French 25 size-BE/ME portfolios; (2) Lakonishok Shleifer and Vishny (LSV)

Portfolios; and (3) Portfolios formed based on past returns.

The Fama French 25 size-BE/ME portfolios basically represents portfolios that

combine value and size characteristics. The 25 size-BE/ME portfolios are used to

investigate whether the simple asset pricing model can capture the return variations in

each portfolios with specific stock characteristic. For instance, whether the return

variation in portfolio that consists of stock with the smallest market capitalization and

the highest BE/ME is captured by the CAPM. In addition, the 25 size-BE/ME




40

portfolios are also used to capture the return variations in portfolios of stocks that are

ranked based on BE/ME, controlling for size and to capture the return variations in

portfolios of stocks that are ranked based on size, controlling for BE/ME. LSV

portfolios, on the other hand, are portfolios that use three value characteristics

(BE/ME, E/P, and CF/P) as proxies to differentiate between value and growth stocks.

If the CAPM do explain the return variations in LSV portfolios, the estimated

intercepts in each quintiles should be statistically not different from zero. Finally,

portfolios based on past returns are used to examine whether the CAPM can explain

momentum versus contrarian pattern.

This study observes the traditional Sharpe-Lintner CAPM as explained in the previous

section (equation 5), where the monthly excess returns of three groups of portfolios

are regressed against the monthly excess market return:

��'−�� = < + ��' − �� + =�' (8)

Where:

��' refers to the average monthly returns of portfolio � at time �.

�� refers to the one-month deposit rate.12

��'−�� refers to the average monthly excess returns of portfolio � at time �.

��' refers to the value-weighted monthly returns of all stocks in the portfolios at time

� with the inclusion of the negative BE stocks that are excluded in forming the 25

size-BE/ME portfolios and the LSV portfolios.13

Firstly, this part explains the procedure of the CAPM time-series regressions on the

Fama French 25 size-BE/ME portfolios. The Fama French 25 size-BE/ME portfolios

are formed as follows. At the end of each month each year t (1991-2008), JKSE

stocks are allocated to five size groups (small, 2, 3, 4, big) based on their market

12

In this study, the Indonesian Government one-month deposit rate is used as a proxy of one-month

Treasury bill rate used in the common studies. 13

For LSV portfolios and portfolios formed based on past returns, ��' refers to the equal-weighted

monthly returns of all stocks in the portfolios since the monthly returns of both LSV portfolios and

portfolios formed based on past returns are also calculated based on equal-weighted monthly returns.




41

capitalization at the end of that particular month. Size small refers to stocks that fall

into the lowest 20 percent rank of market capitalization and size big refers to stocks

that belong to the highest 20 percent rank in market capitalizaton. JKSE stocks are

also sorted independently to five BE/ME groups (Low, 2, 3, 4, high) based on the

quintiles breakpoints. Low refers to stocks that have the lowest 20 percent rank in the

BE/ME and high refers to those that have the highest 20 percent rank in the BE/ME.

The 25 size-BE/ME portfolios (Small/Low, Small/High, . . . , Big/Low, Big/High) are

formed from the intersections of the five size groups and the five BE/ME groups. In

the same manner as before, stocks with negative value in BE are excluded when

forming the intersections of size-BE/ME portfolios.14

Value-weight monthly returns

on the portfolios are calculated from the beginning of month t+1. These value-

weighted monthly excess return of the 25 portfolios are regressed against the monthly

excess market returns. The results of the CAPM time-series regression on the Fama

French 25 size-BE/ME portfolios are summarized in Table VIII.

Secondly, this part continues by explaining the CAPM time-series regression

procedure for LSV portfolios. The LSV portfolios are formed as follows. At the end

of the month each year � (1991-2008), JKSE stocks are allocated to five portfolios,

based on the quintiles breakpoints of BE/ME, E/P, and CF/P. Therefore there are three

different ranks and 15 portfolios in total (5 quintile for each rank). The sample of

stocks is the same for all the fundamental characteristics since in order to be included

in the portfolios in a given year, a stock must have data on all the characteristics used

to form the portfolios in that particular formation period. Equal-weight monthly

returns are calculated from month � + 1, resulting in a time series of 205 monthly

returns from December 1991 to December 2008.15

These equal-weighted monthly

excess return of the 15 portfolios (five portfolios for each stock characteristic) are

regressed against the monthly excess market returns. The results of the CAPM time-

series regression on the LSV portfolios are summarized in Table IX.

14

Basu (1977) shows that the exclusion of firms with negative valuation ratios will have minimum

impact on portfolio returns. 15

Due to data availability, the sample period for CF/P portfolios begins on December 1991 and ends on

December 2007, resulting in a time series of 193 monthly returns.




42

Thirdly, the procedure in forming portfolios based on past returns are provided below.

At the beginning of each month �, JKSE stocks are allocated to quintiles based on

their average monthly returns between months � − 12 to � − 2. For example, stocks

are allocated to the 12-2 portfolios for December 1992 based on their average returns

from December 1991 to October 1992. Quintile 1 refers to stocks with the lowest past

average monthly returns. Quintile 5 refers to stocks with the highest past average

monthly returns. The portfolios are reformed monthly, and the equal-weight simple

monthly returns in excess of the one-month deposit rate are calculated from December

1992 to January 2008. These equal-weighted monthly excess return of the 5 quintiles

are regressed against the monthly excess market returns. The results of the CAPM

time-series regression on the portfolios formed based on past returns are summarized

in Table X.

The Sharpe-Lintner CAPM predicts that < should be zero since the variation in excess

stock returns can be explained by the variation in �. However, many researches find

that CAPM fails to explain the patterns in average stocks returns. The unexplained

patterns are typically called market anomalies.

This section continues to explain the procedure of the time-series regressions using

the Fama French three-factor model. If the CAPM fails to explain stock returns

variation, it is argued that there might be other variables that should be included in the

model. As evidenced by Fama and French (1996), the Fama French three-factor

model is able to better capture the unexplained market anomalies. Therefore, I will

also perform the time-series regressions using the Fama French three-factor model to

the three groups of portfolios. The Fama and French three-factor model suggests that

the return on a portfolio in excess of the risk-free rate is explained by the sensitivity of

its return to three factors which are: (i) the excess market return �� − ��; (ii) the

difference between the return on a portfolio of small stocks and the return on a

portfolio of large stocks (SMB-Small minus big); (iii) the difference between the

return on a portfolio of high book-to-market stocks and the return on a portfolio of




43

low book-to-market stocks (HML-High minus low). Specifically, the excess return on

portfolio � is,

��'−�� = < + ��' − ��>� + ?�@AB + ℎ�8AD + =�' (9)

The independent variables SMB and HML are formed as follows. At the end of each

month each year t (1991-2008), JKSE stocks are allocated to two size groups (small or

big, S or B) based on whether their market capitalization at the end of that particular

month is below or above the median market capitalization for JKSE stocks. JKSE

stocks are also sorted independently to three BE/ME groups (Low, medium, or high:

L, M, or H) based on the breakpoints for the bottom 30 percent, middle 40 percent,

and top 30 percent of the values of BE/ME for JKSE stocks. Six size-BE/ME

portfolios (S/L, S/M, S/H, B/L, B/M, and B/H) are formed from the intersections of

the two sizes and the three BE/ME groups. Value-weight monthly returns on the

portfolios are calculated from the beginning of month t+1. SMB is the difference,

each month, between the average of the returns on the three small-stock portfolios

(S/L, S/M, and S/H) and the average of the returns on the three big-stock portfolios

(B/L, B/M, and B/H). HML is the difference, each month, between the average of the

returns on the two high-BE/ME portfolios (S/H and B/H) and the average of the

returns on the two low-BE/ME portfolios (S/L and B/L).

Following Fama and French (1996), in this study, the 25 Fama French Size-BE/ME

portfolios, LSV portfolios, and portfolios formed based on past returns are regressed

using the Fama French three-factor model. Finally, the GRS tests are calculated for

each portfolio.

5. The GRS Test

CAPM imposes a cross-sectional restriction that expected asset returns are linearly

related to their market beta. In other words, there is a linear relationship in which beta

explains fully the cross-sectional variation in expected returns. In addition to beta,




44

there is no other variable which can explain expected returns. Therefore, the CAPM

(equation 8) implies that the intercepts <� in time-series regressions for each asset �

EF�' = <� + GEF�' G�� + =F�', ∀ � = 1 … � (10)

are all zero. The restriction states that all <� need to be zero, not just some of them

and not only the sum 0 <�� should be zero (the average intercept is zero). Since from

the empirical tests, we can only estimate <I�, which will be different from zero, we

need a joint test whether all <�? = 0.

Gibbons, Ross, and Shanken (1989) introduce the test of joint hypothesis of <�. They

propose a statistical test of whether every single element in the (� × 1)- dimensional

vector < (i.e., the vector of time-series intercepts) equal zero. Specifically, GRS is a

finite sample test on whether the estimated α’s (<KL ) from the time-series regressions

are jointly zero.

According to Gibbons, Ross and Shanken (1989), GRS test statistic can be written as:

M- = NO�P NO − � − D

O − D − 1P Q <RS Σ2,U <I1 + V′X Ω2,U VZ[ ~4(�, O − � − D)

Where,

<I is a � × 1 vector of estimated intercepts.

Σ] is an unibiased estimate of the residual covariance matrix.

VZ is a D × 1 vector of the factor portfolios’ sample means.

ΩS is an unbiased estimate of the factor portfolios’ covariance matrix.

Alternatively, GRS can also be used to test the differences in (ex-post) Sharpe Ratios

or shifts in the efficient set. Gibbons, Ross, and Shanken (1989) show that the

expression above can be rewritten as




45

M- = NO�P NO − � − D

O − D − 1PQNV_̂�_̂P` − NVa_�a_P`[

Q1 + NVa_�a_P`[

Where the portfolio b is the ex-post tangency portfolio16

constructed from the �

assets and the D factor portfolios (in CAPM, D is the market portfolio). If D is ex-post

efficient, it will have the highest Sharpe Ratio of all portfolios which can be

constructed and it will have the same Sharpe Ratio as b (where D is included).

Otherwise, b will have higher Sharpet Ratio since with combination of the � test

assets, it can do better than with D alone. Therefore, D ex-ante efficient does not

necessarily ex-post efficient. The difference in realized, squared Sharpe Ratios,

however should not be too much as b is much more efficient than D . The GRS

statistic, M- shows how large can the differences be while the efficiency of D can still

be hold.

The Null-Hypothesis states that

8c: <e = 0fe,

If <�? = 0, then the GRS statistic, M-is statistically not different from zero. The larger

the <�? are in absolute value, the greater the GRS statistic will be and consequently

the larger the chance of rejecting the Null-Hypothesis. With respect to CAPM,

rejecting the Null-Hypothesis of jointly zero <�? means that market portfolio is not

the ex-post most efficient portfolio since the ex-post maximum Sharpe Ratio of

market portfolio is less than the ex-post maximum Sharpe ratio of the market portfolio

and N assets.

16

The ex-post tangency portfolio (the log-portfolio) is the portfolio preferred by an investor with

logarithmic utility, i.e. investor with a unit relative risk aversion (Mertens, 2002).




46

IV. Results

This section provides the evidence of the existence of the value premium, the size

effect, and momentum versus contrarian. I will begin with the simple individual-sort

strategy in forming portfolio based on some stocks characteristics (BE/ME, CF/P, and

E/P) in order to investigate whether value stocks have higher realized average returns

than growth stocks. Then, this study also employs two-independent sort strategy

which combines two stocks characteristics in examining the value premium.

Combining two accounting data are expected to give better picture in the performance

of the value strategies. Furthermore, this section presents the analysis of the results

from momentum versus contrarian using Lo and MacKinlay (1990) methodology

which the procedure has been explained in detail in the previous section. The last part

of this section presents the analysis and summary of momentum versus contrarian

which incorporates trading volume in the analysis.

1. Simple Glamour and Value Strategies

Table III provides the returns on portfolios of stocks that are ranked based on their

BE/ME, CF/P and E/P. In order to give a better picture for investors, this study

exploits an annual buy and hold strategy in contrast to monthly buy and hold strategy

used in most previous study. The annual returns are the average monthly returns.

Because of various market microstructure biases as well as trading costs, annual buy

and hold strategy produces returns that are closer to those that investors can actually

capture.

Table III shows the returns for Years 1 through 5 after the formation periods and the

average annual 5-year returns. The numbers presented in the table are the average

monthly returns over all formation periods. Similar to Lakonishok et al. (1994), the

results in Panel A1 confirm the existence of value premium in Indonesian stock

market. As can be seen, the average annual 5-year returns for low BE/ME stocks

(Glamour) is 0.66 percent, whereas the average return for high BE/ME stocks (Value)

is 4.13 percent resulting in significant difference of 3.47 percent. Looking at Panel




47

A1, 1-year buy and hold value stocks strategy on portfolio formed based on BE/ME

produces the highest possible return for 4.91 percent annually.

Panel B1 of Table III presents the returns of portfolios ranked based on CF/P. Similar

to Panel A1, low CF/P stocks (Glamour) have lower average returns than high CF/P

stocks (Value). High CF/P stocks are regarded as value stocks due to their low prices

per dollar of cash flow. On the other hand, low CF/P stocks are defined as glamour

stocks because of their high expected growth rate on cash flow. The average returns

for high CF/P stocks is 2.39 percent and for low CF/P stocks is 1.94 percent. The

difference of 0.45 percent annually however, is not significant compare to that of

BE/ME. In contrast with Lakonishok et al. (1994), sorting on CF/P appears to produce

lower spread in returns than sorting on BE/ME. The highest possible average return of

portfolio formed based on CF/P is 2.98 percent, which can be obtained by performing

a 2-year buy and hold value stocks strategy.

Basu (1977), Jaffe et al. (1989), and Fama and French (1992, 1996) find that stocks

with low P/E can generate positive abnormal returns. 17

Table III, Panel C1 shows the

results of annual buy and hold strategy for portfolios ranked based on E/P. On

average, high E/P stocks have an annual return of 3.42 percent whereas low E/P

stocks have an annual return of 1.32 percent. High E/P stocks outperform low E/P

stocks by a fairly wide margin, although the difference is not as large as that of

BE/ME. Similar with BE/ME portfolios, the highest possible average returns investors

can earned is by utilizing a 1-year buy and hold value stocks strategy.

Table III, Panel A2, B2, and C2 provide the results of same strategies using value-

weighted returns. In line with Heston, Rouwenhorst, and Wessels (1995) and

Loughran (1997), this study finds that when value-weighted returns are used, the

spreads and the magnitudes decrease by a significant portion. On average, high

BE/ME stocks produce an annual return of 0.13 percent. Conversely, low BE/ME

stocks have on average an annual return of 0.04 percent. The results still support the

17

All the ratios presented in this research are defined as the stocks characteristic divided by market

value of equity. For example, E/P is the inverse of P/E that is commonly used in past researches.




48

existence of value premium in Indonesian stock market except for portfolios ranked

on CF/P. In Panel B2, using value-weighted returns, on average, high CF/P stocks

underperform low CF/P stocks by 0.03 percent annually. In the value-weighted

scenario, in portfolios formed based on other than CF/P, this study discovered that a

1-year buy and hold value stocks is still the most profitable strategy.

In line with Fama and French (1992), the average returns for portfolio zero are high,

some even higher than the average returns of portfolio five (value). The evidences of

value-weighted returns in Panel A2, B2, and C2 show that portfolio zero outperforms

value portfolio by a considerable amount. For example, negative BE (i.e. which

results from persistently negative earnings) and high BE/ME (which typically means

that stock prices have declined) are both signals of poor earnings prospects. The

finding that negative BE and high BE/ME stocks have on average higher returns than

low BE/ME stocks is consistent with the Fama and French hypothesis that BE/ME is a

proxy of relative distress.

In conclusion, this study argues that value premium exists in Indonesian stock market

during the sample period. This finding can add evidence to deny the argument that

value premium in U.S. market is a result of data mining. Utilizing the fact that value

premium does exist in Indonesia, investors can benefit from performing a 1-year buy

and hold value stocks formed on the basis of BE/ME and E/P.

2. Two- Independent Sort Glamour and Value Strategies

According to behavioral model, Kahneman and Tversky (1982) explain that compared

to value stocks, glamour stocks are those that have higher recent and expected growth

rates in earnings. They argue that investors tend to overreact to recent information and

make extreme predictions based on the information which reliability and predictive

validity are known to be low. Chan, Karceski, and Lakonishok (2003) argue that

investors and analysts’ behavior in favor of growth stocks drive their prices up and

cause the overvaluation of glamour stocks.




49

Lakonishok et al. (1994) explore a contrarian strategy which sells stocks with high

past growth as well as high expected future growth and buys stocks with low past

growth as well as low expected future growth. They refer to both CF/P and E/P as a

proxy for expected growth rate whereas BE/ME ratio is used as a proxy for past

growth. The contrarian strategy exploits sorting stocks independently based on their

past growth as well as their expected growth rate. Panel A1 in Table IV, presents the

results for contrarian strategy that sorts stocks based on both BE/ME and E/P. In order

to avoid an insignificant number of stocks in each group of portfolios, stocks are

sorted into three instead of five portfolios as in individual-sort strategy. Accordingly,

in this study, stocks are sorted independently into three groups (bottom 30 percent,

middle 40 percent, and top 30 percent) by BE/ME and E/P, and then take the

intersections between three BE/ME groups and three E/P groups. Glamour stocks are

those which fall into bottom 30 percent in both BE/ME and E/P groups. Value stocks,

on the other hand, are those that belong to top 30 percent of both BE/ME and E/P

groups.

Based on Table IV, Panel A1, the growth portfolios have a return of 1.32 percent and

the value portfolios have on average, a return of 3.04 percent. On average, in this

sample period, value stocks outperform growth stocks by 1.72 percent annually. As

can be seen in Panel A1, moving horizontally from glamour portfolios to value

portfolios, the average annual returns are increasing even though not necessarily

monotonically. For example, high BE/ME stocks with low expected future growth

rate (intersection BE/ME-E/P, 3-2), which are not defined as value stocks, have an

average annual return of 2.73 percent, whereas high BE/ME stocks with low expected

future growth rate (intersection BE/ME-E/P, 3-3), which are defined as value stocks,

have a higher average annual return of 3.04 percent. The reason of no exact increasing

pattern in the average returns might be related by individual stocks’ unsystematic

factors within each group of portfolios. This unsystematic factors, for example can be

a macroeconomic event that give impact to particular industry. Consequently, the

impact to each group of portfolios might be different. The highest average returns is

3.74 percent which can be obtained by buying and holding stocks that fall in the

intersection of BE/ME-E/P, 3-1. In other words, those are stocks with lowest past




50

growth rates and highest expected future growth rates (the highest BE/ME and the

lowest E/P.

Table IV, Panel B1 shows the results of two-independent sort based on BE/ME and

CF/P. Controlling for BE/ME, high CF/P stocks do not necessarily have higher

average returns than lower CF/P stocks. For example, the average return of stocks that

belong to portfolio BE/ME-CF/P; 3-1, which is not considered as value stocks, is

higher than the average return of portfolio BE/ME-CF/P; 3-3, which is considered as

value stocks. The difference in annual average return between the extreme portfolios

is 1.82 percent with value portfolio has an annual average return of 2.96 percent and

glamour portfolio has an average return of 1.14 percent. The highest possible return

can be achieved using a strategy that buy stocks with lowest past growth and highest

expected future growth rates (intersection BE/ME-CF/P, 3-1), resulting in an average

annual return of 4.02 percent.

Table IV, Panel A2 and B2 provide the results of two-independent sort strategy with

value-weighted returns. In two-independent sort strategy using BE/ME-E/P as well as

BE/ME-CF/P, value portfolios outperform glamour portfolios by approximately 0.08

percent. Similar to the individual-sort strategy in Table III, the magnitude of value-

weighed annual average return is smaller than that of equal-weighed annual average

return. In line with Lakonishok et al. (1994), using value-weighed returns, two-

independent sort strategy produces substantially higher returns than either the BE/ME

and CF/P or the BE/ME and E/P strategy alone. For example, on the BE/ME-E/P

strategy, among portfolios of stocks with the lowest BE/ME ratios, the average annual

return varies from 0.08 percent for stocks with the highest expected growth rates to

0.98 percent for stocks with the lowest expected growth rates. The individual BE/ME

strategy, on the other hand, produces an average annual return of only 0.04 percent for

stocks in the lowest BE/ME quintiles. In the value-weighed scenario of two-

independent sort strategy based on BE/ME and E/P, this study found that stocks with

the highest past growth rates and the lowest expected future growth rates yield highest

average returns.




51

Overall, the results of two-independent sort strategy can be summarized as follows.

Firstly, two-independent sort value strategies, especially those which incorporate

equal-weighted return, produce higher returns than those of similarly constructed

glamour strategies for the sample period beginning in December 1991 and ending in

December 2008. Secondly, the results show that the highest possible average annual

return can be achieved by buying stocks with lowest past growth and highest expected

growth rates proxy by E/P. Thirdly, two-independent sort strategies, especially those

which incorporate value-weighted return, produce higher average annual return than

those strategies based exclusively on individual-sort. Fourthly, within two-

independent sort strategies using BE/ME and CF/P, the highest average return is

obtained by buying and holding stocks that belong to the portfolio with the highest

past growth and lowest expected growth rates proxy by CF/P. Last but not least,

investors should conduct an additional analysis when making investment decision

based on the evidences above since value stocks as a group do not outperform growth

stocks in every observation. The reason of non-monotonically pattern found in Table

IV is subject to further investigations.

3. The Size Effect

Table V shows the returns of portfolios of stocks that are sorted on the market value

of equity. Glamour stocks refer to those in the highest ME quintiles, whereas value

stocks are those that belong to the lowest ME quintiles. As can be seen in Panel A,

small stocks have on average an annual return of 3 percent; outperform large stocks

by 1.51 percent per year. The results are in accordance with Banz (1981), who argued

that smaller firms have had, on average, higher returns than larger firms. The results

for value-weighted returns can be seen on Table V, Panel B. The average value-

weighted returns of value stocks outperform those of growth stocks in all the holding

periods. However, it should be noted that the analysis in this study has not incorporate

transaction cost. Therefore, by looking at the magnitude of average value-weighted

returns of value stocks, it is clear that investors cannot earn abnormal returns.




52

4. Momentum versus Contrarian Strategies

Chan, Jegadeesh and Lakonishok (1999) argue that in medium term (up to 12 months)

stock prices exhibit momentum (continuation in a price direction). In other words, for

these horizons, what goes up tends to keep rising and vice versa. However, the

evidence in Table VI opposes their hypothesis. As can be seen in all F/H strategies,

losers have higher average returns than winner portfolios. Consequently, zero-cost

strategies that buy winners and sell losers, ended up in negative average returns. It

must be noted that negative returns using this methodology mean contrarian profits.

The evidence of winner and loser in all F/H strategies is in line with the argument that

winner and loser do not have the same price movement. Contrary to T.H. McInish et

al. (2008), this study suggests that winners display momentum behavior, whereas

losers experience reversal within the sample period. Winner portfolio continues to

earn positive profits up to 12 months holding periods. On the other hand, loser

portfolio shows some overreaction up to 12 months holding periods, although the

magnitude of both reversal and continuation decline as the length of F and H increase.

These findings can possibly be attributed to the up-market effects during the study

period. For example, if most of monthly returns in the study period are positive, it is

by design that positive-return stocks will exhibit momentum and negative-return

stocks will experience reversal. Nevertheless, since the Asian crises in which stock

market index decreased significantly are included in the whole sample period, the

results are robust against the up-market effects.

Overall, since the medium-term average returns of losers are higher than winners,

investors who trade in Indonesia stock markets can generate profit by performing a

contrarian strategy. The results seem to suggest that in Indonesia, the overreaction

effect towards losers is so significant that the reversal effect in losers exist longer (up

to 12 months). On the other hand, the overreaction behavior towards winners is not

significant enough to drive the prices remarkably high. As a consequence, the

contrarian profit in losers is higher than the momentum profit in winners, even though

they both decline as the length of F and H increase.




53

5. Effect of Trading Volume on Momentum/ Contrarian Profit

In this subsection, the role of trading volume is examined in momentum/contrarian

strategy. Several researches confirm the relation between volume and predictable

patterns in short-term stock returns. Blume, Easley, and O’Hara (1994) argue that

traders who use information contained in the volume statistic will do “better” than

traders who do not. In accordance with the premise that the predictability of short-

term returns might be affected by trading volume, this study examines whether lagged

trading volume could affect the momentum/ contrarian profits.

The procedures in portfolios formation and the computation of returns are explained

in section III. The results of the effect of lagged trading volume to momentum/

contrarian profits are shown in Table VII. As can be seen from Table VII, the result in

1/1 strategy contradicts the prediction of Conrad, Hameed, and Niden (1994) and Lee

and Swaminathan (2000). Conrad et al. (1994) argue that high volume stocks show

reversal behavior whereas low volume stocks display momentum pattern. Chordia and

Swaminathan (2000) also provide the evidence that low volume stocks tend to adjust

more slowly than high volume stocks to information shocks. The under-reaction of

low volume stocks results in momentum pattern, while the overreaction in high

volume stocks results in contrarian pattern. The evidences in Table VII however,

show that across volume groups, stocks experience reversal pattern. These results are

in part similar with those of Hameed and Ting (2000). They observe that both high

and low volume groups produce contrarian profit, the magnitude however is larger for

high volume groups. Comparing the results in Table VII with those in Table VI, it is

indicated that price reversal is even lower following an increase in trading volume. As

an exception, in 1/1 strategy, despite statistically insignificant, the average returns of

winner stocks in high trading volume group are higher than those of loser stocks.

Consequently, a zero-cost momentum strategy that buys winner and sells loser will

result in momentum profit. 12/12 strategy is in favor of contrarian strategy as the

results show that in all trading volume groups, zero-cost strategies have negative

average returns. As previously stated, in this research negative return means

contrarian profit.




54

The price movement of both winners and losers are similar with those in Table VI.

These findings confirm the results of Fung, Leung and Patterson (1999) that winners

exhibit momentum and losers exhibit reversal. In 1/1 strategy, moving on to higher

volume groups, the momentum profit in winner increases (from 0.12 percent to 0.35

percent) while the contrarian profit in loser decreases (from -0.29 percent to -0.24

percent). In 12/12 strategy, the amount of momentum profit in winner and contrarian

profit in loser are not significantly different across volume groups.

In short, there is no significant difference in the behavior of zero-cost strategies across

volume groups. Using 12/12 zero-cost strategies, Indonesian stock market exhibits no

clear pattern of the effect of trading volume on momentum/contrarian profits. On the

other hand, using 1/1 zero-cost strategies, most of the average returns are statistically

insignificant. All volume groups display reversal pattern and the magnitude of

reversal is higher for low volume groups




55

V. Robustness Check

To examine whether stocks fundamental characteristics are priced in an explicit asset-

pricing model, this section begins by documenting the simple CAPM time-series

regressions. The Sharpe-Lintner versions of the CAPM that is used in this study share

the prediction that the market portfolio is mean-variance efficient. Implicitly, the

differences in the expected return across securities and portfolios are entirely captured

by the differences in market beta. From the results, it was discovered that the CAPM

captures most of the returns variation, except for the small value portfolios which

have alpha that is statistically not different from zero.18

However, the results of the

GRS test accepts the null hypothesis that alphas are jointly zero. In order to provide

more robust findings, following Fama and French (1996), this study also performs the

time-series regressions using the Fama French three factor model. Fama and French

(1992) provide evidences that ME and BE/ME capture much of the variation in stock

returns. They argue that if stocks are rationally priced, systematic differences in

average returns are due to differences in risk. Consequently, in rational pricing

condition, size (ME) and BE/ME must proxy for sensitivity to common risk factors in

returns. Fama and French (1996) suggest that the three-factor model, which is

constructed to mimic risk factors related to size and BE/ME, may explain the

variation of stock returns. Their three factors are as explained in section III. The

results of the regressions using the Fama French three-factor model however, do not

add any significant improvement to the performance of the simple CAPM.

1. Simple CAPM Time-Series Regressions

a. Fama French 25 Size-BE/ME portfolios

The procedures to form the 25 size-BE/ME portfolios as well as the variables used in

the regressions are explained fully in section III. From December 1991 to December

2008, there are 205 monthly observations. Table VIII shows the results of the CAPM

regression on Fama French 25 size-BE/ME portfolios. In the CAPM world, the true

18

It should be noted that this study utilizes the simple CAPM with constant beta. Academics and

practitioners argue that beta could be time-varying since investors have more than one period

investment horizon. In addition, there might be structural breaks in the stock market in which the

performance of the market anomalies might be changing as well. Further investigation using more

advance test are encouraged.




56

intercepts of the regressions should be zero. As can be seen on Table VIII, the CAPM

regression to explain each of 25 size-BE/ME portfolios produce nearly zero intercepts

and mostly statistically insignificant. The CAPM however, cannot capture the

variation among those small value portfolios. As can be seen on Table VIII in the

upper right corner of the alphas, small size and high BE/ME portfolios have

statistically significant intercepts (-3.22, -3.4, and -3.82). If the simple CAPM holds,

there is no way to group assets into portfolios whose intercepts are reliably different

from zero. For instance, the intercepts for a portfolio of stocks with small size and low

BE/ME and a portfolio of stocks with small size and high BE/ME should both

statistically not different from zero. To complete the analysis, this study jointly tests

the vector of regression intercepts against zero. Table VIII shows the result of the

GRS test which accepts the hypothesis that the CAPM can explain the average returns

on 25 size-BE/ME portfolios at any significance level. Looking at the R2 table, the

magnitudes of the R2 suggests that the CAPM, in average, can explain 61 percent of

the variation in monthly excess returns on 25 size-BE/ME portfolios.

Overall, the findings in Table VIII, despite of statistically significant intercepts among

small value portfolios, seem to suggest that the CAPM explains the average returns on

most of the portfolios since the GRS test accept the null hypothesis of jointly zero-

alphas. However, other independent variables might be needed to better capture the

returns variations of small value portfolios. Again, it should be noted that this study

employs the simple CAPM with constant beta and without any structural breaks.

Utilizing more advance test might provide different performance of the CAPM

estimates.

b. LSV Portfolios

Lakonishok et al. (1994) observe that sorting stocks based on some value

characteristics like BE/ME, CF/P, and E/P, then exploit a buy and hold strategy on

value stocks can produce abnormal returns. The complete details of LSV portfolios

formation and the variables used in the time-series regressions are explained in

section III. The sample period uses for BE/ME and E/P portfolios are from December




57

1991 to December 2008, resulting in 205 monthly observations. However, due to data

availability, the sample period for CF/P portfolios begins on December 1991 and ends

on December 2007, resulting in a time series of 193 monthly returns.

Table IX presents the coefficients of the CAPM time-series regressions on monthly

excess returns of portfolios formed based on BE/ME, CF/P, and E/P respectively.

Similar with the 25 size-BE/ME portfolios, the estimated intercepts in portfolios

based on BE/ME are nearly zero and statistically insignificant except for those among

value portfolios. The CAPM does a good job in explaining the return variations in

portfolios based on CF/P, specifically all intercepts are statistically not different from

zero. The exception is for portfolios based on E/P which almost all estimated

intercepts are statistically different from zero. These results also showed in the GRS

test. For portfolios based on CF/P, the null hypothesis is accepted at any significant

level. Despite of some statistically significant intercepts for portfolio based on E/P

and BE/ME, the results of GRS test still accept the null hypothesis of jointly zero-

alphas. The averages R2 for BE/ME-rank, E/P-rank, and CF/P-rank portfolios are

0.89, 0.89, and 0.91 respectively. High values on R2 explain that CAPM does a good

job in capturing the variations in the returns of LSV portfolios in this sample period.

In short, the results show that the CAPM fails to capture the return variations in small

value portfolios based on BE/ME and E/P, as evidenced by statistically significant

intercepts. The return variations in all quintiles of CF/P rank are explained by the

CAPM. In overall, despite of some statistically significant intercepts, the GRS test

accept the null hypothesis that the alphas in all portfolios are jointly zero. Again, this

study does not take into account time-varying beta or any structural breaks. Further

research regarding these matters may provide better analysis of the CAPM

performance in explaining market anomalies.

c. Portfolios formed on Past Returns

The steps in forming the portfolios are explained in section III. Since this study

employs a strategy based on 12 months and skips one month prior the ranking period,

the sample period begins on December 1992 and ends on December 2008, resulting in




58

total of 182 monthly observations. Table X provides the results of the CAPM time-

series regressions on portfolios formed based on past returns. This study exploits a

strategy based on 12 months past returns that skip one month prior to the ranking

period. On average, CAPM can capture 90 percent of returns variations. As before,

almost all estimated intercepts are nearly zero and statistically insignificant. The only

exception is for portfolio with the lowest past returns (losers). However, the GRS test

accepts the hypothesis that all intercepts are jointly zero. In overall, the CAPM does

explain the return variations in portfolios formed based on past returns.

In conclusion, based on the results in three groups of portfolios; The Fama French 25

size-BE/ME portfolios, the LSV portfolios, and the Past returns portfolio, it can be

seen that the return variations are explained by the simple CAPM. Even though there

are some exceptions in the small value portfolios, all the GRS test accept the null

hypothesis of jointly zero-alphas. In the next sub-section, this study performs time-

series regressions with the Fama French three-factor model that is argued to be able to

better capture the variation in small value portfolios.

2. Fama French Three-Factor Model

a. 25 Fama French Size-BE/ME Portfolios

In order to test whether the Fama French three-factor model is able to explain

variation in portfolio formed based on size and BE/ME especially the small value

portfolios, the 25 size-BE/ME portfolios are regressed using the Fama French three-

factor model. Table XI shows the results of time series regressions on Fama French 25

size-BE/ME portfolios using the three-factor model. As can be seen on Table XI, most

of the regression intercepts are statistically not different from zero. However, the

estimated intercepts among the small value (small size and high BE/ME) portfolios

are statistically significant, as evidenced by the t(a) values in the upper right corner of

Table XI. The results show that three-factor model also cannot capture the returns

variation in the small value portfolios. The estimated intercepts suggest that the model

leaves a negative unexplained return for the portfolio of stocks in the high BE/ME

quintile.




59

The F-test of Gibbons, Ross, and Shanken (GRS 1989) accepts the hypothesis that

Fama French three-factor model can explain all the variations in the average monthly

returns on the 25 size-BE/ME portfolios at any significance level. The average of the

25 regression R2 is 0.63. The model does explain 63 percent of the variation in the

average returns on the portfolios. Comparing the results in Table XI with those in

Table X, Fama French three-factor model only performs slightly better than the

simple CAPM. The Three-factor model explains the variation of 25 size-BE/ME

portfolios by 2 percent better than CAPM. The returns variation in the small value

portfolios are still left unexplained.

Looking at the coefficients, the pattern suggests that high BE/ME stocks have higher

slope on HML than low BE/ME stocks. Small stocks tend to have higher slopes on

SMB. Controlling for size, high BE/ME stocks have smaller values on SMB

coefficients than those of low BE/ME stocks. The finding seems to suggest that high

BE/ME stocks comprise of small stocks.

In overall, utilizing the Fama French three-factor model in explaining the return

variations in the 25 size-BE/ME portfolios does not add much improvement to the

results. Both models cannot explain the returns variations in the small value

portfolios, even though based on the GRS test, the alphas are jointly zero. The slope

on both HML and SMB shows that high BE/ME stocks and small stocks have higher

return, as evidenced by higher positive value in their factor loading.

b. LSV Portfolios

Following Lakonishok et al. (1994), time-series regressions are performed on the

returns of portfolios formed on the basis of the rank of BE/ME, CF/P, and E/P. In

order to maintain a respectable number of sample stocks in each portfolio, stocks are

allocated into quintiles. To be included in the tests for a given year, a stock must have

data on all off the characteristics used to form the portfolios. Thus, the sample firms

are the same for all variables. The sample period uses for BE/ME and E/P portfolios

are from December 1991 to December 2008, resulting in 205 monthly observations.

However, due to data availability, the sample period for CF/P portfolios begins on




60

December 1991 and ends on December 2007, resulting in a time series of 193

monthly returns.

Table XII shows the results of time series regressions using Fama French three-factor

model. The estimates of the three-factor regression in Table XII show that the three-

factor model can capture most the variation in average returns. The average of the R2

for portfolios based on BE/ME, E/P, and CF/P are 0.91, 0.89, and 0.92 respectively.

As evidenced before by the results of the CAPM regressions on LSV portfolios, the

estimated intercepts using the three-factor model in portfolios based on BE/ME are

also statistically insignificant except for portfolios with the highest BE/ME. The Fama

French three-factor model also perform well in explaining the return variations in

portfolios based on CF/P, specifically all intercepts are statistically not different from

zero. The exception is again for portfolios based on E/P which almost all estimated

intercepts are statistically different from zero. The GRS test accepts the hypothesis

that the regression intercepts for a set of five portfolios are all 0.0. The p-value of

GRS tests are 0.057, 0.31, and 0.99 for portfolios based on BE/ME, E/P, and CF/P

respectively. In term of the magnitude of the GRS tests, the three-factor model does a

better job for the 25 Fama French size-BE/ME portfolios than it does on the LSV

portfolios.

Based on the results in Table XII, higher BE/ME and E/P portfolios have larger slopes

on HML. As has been observed by Fama and French (1995) and confirmed by the

results of portfolio zero in previous section, loadings on HML are a proxy for relative

distress. Consequently, these findings infer that low BE/ME and E/P are typical of

strong stocks, while high BE/ME and E/P are typical of stocks that are relatively

distressed. The slopes on SMB, however, show no clear pattern for both BE/ME and

E/P portfolios. Among the three LSV portfolios in Table XII, only the CF/P-rank

portfolios show no exact relations between average returns and HML. In CF/P

portfolios, SMB slopes seem to suggest that higher CF/P portfolios produce higher

slopes than those of lower CF/P portfolios. This pattern indirectly implies that high

CF/P portfolios consist of smaller stocks than low CF/P portfolios.




61

In short, utilizing the Fama French three-factor model in explaining the return

variations in the LSV portfolios does not make any significant different to the results

of the CAPM. Both models cannot explain the returns variations in the small value

portfolios, even though based on the GRS test, the alphas are jointly zero. However,

the use of the Fama French three-factor model can add insight to stocks performance

in each portfolio. For instance, from the slope on SMB, we can observe that high

CF/P stocks consist of small stocks.

c. Portfolios based on Past Returns

DeBondt and Thaler (1985) argue that in long-term period (three to five years), stocks

experience a reversal (contrarian) pattern. Specifically, winners (stocks with high past

returns) have lower future returns than losers (stocks with low past returns). Jegadeesh

and Titman (1993), however, observe a continuation (momentum) pattern in the

medium-term. According to momentum pattern, when portfolios are formed based on

intermediate-term (three to 12 months) past returns, past losers continue to be future

losers and past winners are still the future winners.

Table XIII below shows the results of the Fama French three-factor regressions on

monthly excess returns of portfolios formed based on � − 12 to � − 2 average

monthly returns. In total, there are 182 monthly obervations. In order to reduce bid-

ask bounce bias, one month is skipped on the portfolio formation month before

ranking the stocks at month �.

The GRS test in Table XIII confirms that the Fama French three-factor model is able

to explain the variation in the future returns on portfolios formed based on past

returns. The null hypothesis that all intercepts are jointly zero is accepted at any

significane level. The results, however, contradict the common belief that in medium-

term, stocks exhibit momentum pattern. Based on Table XIII, medium-term past

losers have higher risk loading on HML. Higher slopes on HML for past losers imply

that they behave like distressed stocks. Thus, the model predicts that medium-term

past losers will have higher average returns in the future than those of past winners.




62

In short, contrary to Jegadeesh and Titman (1993), Indonesian stock market exhibits

reversal patterns in medium-term period. The fact that past losers have higher loading

on HML than past winners, might be the reason of past-losers outperformance. Based

on the results examined in Table XIII, this study argues that losers tend to consist of

stocks with high BE/ME (proxy of relative distressed), and therefore outperform

winners.

3. The Post-Crisis Analysis of the Value Premium

The majority of past studies investigating the existence of value premium in Asian

markets, especially Indonesian stock market, rely on data that ends prior to the year

2000 or even before the onslaught of Asian financial crisis of 1997-1998. In addition,

Brown, Rhee and Zhang (2008) argue that value premium is time-varying. Therefore,

in order to explore whether the behavior of value premium in Indonesian stock market

varies within the sample period and to confirm the robustness of my findings, the

results from section IV are examined and compared with the results of the post-crisis

period.

According to IMF January 1999 Factsheet “The IMF’s Response to the Asian Crisis”,

Indonesia is regarded as one of the most affected countries during the Asian financial

crisis. The evidence shows that Asian crisis has a considerable impact to Indonesia

stock market, as showed by Liou (2002), between June 1997 and December 1997

alone, the stock market index in Indonesia has declined by 44.6 percent. Since the

sample period in this study includes the year when the Asian financial crisis occurred,

specifically the severe period from July 1997 to July 1998, the same procedures for

value premium are performed using only the post-crisis sample period.

Table XIV, Panel A provides the evidence for equal-weighted returns in the post-

crisis sample period. Interestingly, the post-crisis equal-weighted average returns for

all the portfolios formed based on three value characteristics are higher than those in

the whole period. The results summarized in Table XIV are similar to those in the




63

whole period. Value stocks outperform growth stocks and a 1-year buy and hold value

stocks strategy produces the highest average return.

Panel B summarizes the results of the same strategies using value-weighted average

returns. Despite of two-year negative value premium for CF/P-based portfolios, the

post-crisis results have only slight differences from the whole period results. Similar

to the one in Table III, CF/P-based portfolios have on average a negative value

premium. Overall, the evidence seems to suggest that the behavior of value premium

in Indonesia stock market does not change after the Asian financial crisis.

4. Jegadeesh and Titman Methodology of Momentum and

Contrarian Strategy

In order to ascertain the robustness of momentum/contrarian profits found in previous

section, some analysis are performed using an adaptation of Jegadeesh and Titman

(1993) approach. The procedures are as follows: Firstly, JKSE stocks are sorted into

three groups (High, Medium and Low trading volume) based on their lagged trading

volume. Then, following Jegadeesh and Titman (1993), in each volume group, at the

beginning of each month �, JKSE stocks are ranked on ascending order based on their

performance during the formation period, from time � − 4 to � − 1. Based on these

rankings, three portfolios are formed that equally weight the stocks contained in each

group of stock. The first group is called the “loser”, whereas the last group is called

the “winner”. Therefore, the return of winner portfolio is the average holding period

returns of stocks in the highest 30 percent ranks, whereas the return of loser portfolio

is the average holding period returns of all stocks within the lowest 30 percent ranks.

In each month �, the strategy buys the winner portfolio and sells the loser portfolio,

holds the position for 8 months. Therefore in each F/H strategy, there are six

individual portfolios (Winner-High, Winner-Medium, Winner-Low, Loser-High,

Loser-Medium, and Loser-Low) and three zero-cost portfolios (Winner minus Loser-

High, Winner minus Loser-Medium, and Winner minus Loser-Low). As before, for

simplicity, only strategy that selects stocks on the basis of returns over the past F

formation periods and holds them for the same H holding periods is performed. In




64

specific, different strategies with the same combination of F and H periods are

employed. Finally, the portfolios are rebalanced monthly and the average monthly

returns during the holding period of each strategy are calculated.

Jegadeesh and Titman methodology provides similar evidences to Lo and MacKinlay

methodology. As can be seen in Table XV, independently, winner and loser in both

1/1 and 12/12 strategies have positive average returns.

The result of 1/1 strategy is not consistent with that predicted by Hameed and Ting

(2000) that loser stocks tend to have higher average returns in high volume group. The

evidence from table XV show the exact opposite of what was predicted. As we move

to the higher volume group, the average returns of losers decrease and the average

returns of winners increase. Using the reasoning of Lakonishok et al. (1994) in

analysing winner stocks performance based on one month past returns, we can argue

that in short-term, investors tend to be overly optimistic about winner with high

lagged trading volume. Consequently, the winner in high volume group is able to

continue its last month performance in the next month. In low volume group, on the

other hand, infrequently traded past one-month loser outperforms infrequently traded

past one-month winner. The highest possible average monthly returns (4.11 percent)

on the zero-cost strategy can be earned by buying past one-month loser with low

lagged volume and hold it for one-month. The results in 12/12 strategy are similar

with those in 1/1 strategy, the pattern however, is not clear enough to draw an exact

conclusion.

Overall, the results of winner loser are mostly negative and statistically significant

across portfolios and strategies as summarized in fourth column of Table XV. Similar

to the results using Lo and MacKinlay methodology, zero-cost strategies using

Jegadeesh and Titman methodology also favor contrarian strategy.




65

5. The post-crisis analysis of Jegadesh and Titman methodology

for momentum versus contrarian strategies

In order to confirm the robustness of the findings of this study, the subperiod

performance of momentum/contrarian strategy is also examined. The sample period

is divided into two subperiods: (i) before crisis, from December 1991 to June 1997

and (ii) after crisis, from August 1998 to December 2008. The results in both

subperiod as summarized in Table XVI, suggesting a similar conclusion to those in

the whole period. Loser and winner stocks in both subperiod exhibit momentum

patterns across all strategies and portfolios. Loser stocks that are classified in the low

volume group, have higher momentum average returns than those that are assigned to

high volume group. Zero-cost strategies in both subperiods produce negative average

returns across portfolios. The exception are those stocks in high volume group in 1/1

zero-cost strategy, which on average have statistically insignificant positive returns.

In short, the performance of momentum/contrarian returns in Indonesia stock market

within the sample period are not time-varying. In addition, as can be seen on Table

XVII, the same strategy and methodology is applied to the whole period, excluding

the crisis period, generates similar results.




66

VI. Conclusion

This study examines (i) whether the value premium, the size effect and

momentum/contrarian return exist in Indonesia stock market and how to exploit such

profitable strategy if they do exist, (ii) whether the Fama French three-factor model

outperforms the CAPM, and (iii) whether differentiate stocks based on trading volume

results in different performance.

The results in this study can be summarized into four propositions. Firstly, as can be

seen from Table III and IV, a variety of investment strategies that utilize an annual

buy and hold strategy of value stocks have outperformed those of glamour stocks over

the period ranging from December 1991 to December 2008. Secondly, based on the

results presented in Table VI and VII, in medium-term, Indonesian stock market

exhibits reversal pattern. Based on the evidence, a likely reason of this reversal

behavior is the fact that Indonesian investors’ overreaction toward losers is higher

than their overreaction toward winners. Consequently, since the reversal effect in

losers is more significant than the overvaluation effect in winners, losers outperform

winners. Thirdly, as showed in Table VII, this research discovered no difference in

stocks behavior within various volume groups. All volume groups display reversal

pattern and the magnitude increases as the lagged volume decreases. Fourthly, the

evidences in Table XI-XIII show that the Fama French three-factor model does only

slighty better job in explaining variation in stock returns than the CAPM with the

returns variation in the small value portfolios still left unexplained. However, despite

some statistically significant intercepts, the results of the GRS tests accept the null

hypothesis that alphas are jointly zero. By regressing three-factor model in portfolio

based on past returns, this study discovered that another explanation of losers’

outperformance is the fact that losers have higher loading on HML than winners.

Based on the evidence in Table XIII, it seems that losers tend to consist of stocks with

high BE/ME (value stocks) and winners consist of low BE/ME stocks (growth

stocks).




67

Another important result is that most investors can generate profit by performing 1-

year buy and hold strategy of value stocks. Looking deeper into investment strategies,

this study found that Indonesian investors seem to undervalue stocks with low past

growth, however are attracted to stocks with high expected earnings performance.

Therefore, as showed in Table IV, high average return can be obtained by buying and

holding stocks which have lowest past growth as proxies by BE/ME and highest

expected growth rates as proxies by E/P.

The robustness tests confirm the existence of value premium and profitable contrarian

strategies in Indonesian stock market both before and after the crisis period. Using

Jegadeesh and Titman methodology in analysing momentum versus contrarian

strategies, as shown in Table XV, this study provides that in short-term (one month),

investors tend to overly optimistic about winner with high lagged volume, drive its

price up and overestimate its future performance relative to loser. As a consequence,

the F/H; 1/1 strategy exhibit momentum behavior.

Based on these results, this study concludes that the value premium, the size effect

and the contrarian return exist in Indonesian stock market. However, the

implementation should be done with caution since this study does not incorporate

transaction cost in its analysis. It also should be noted that value stocks do not provide

the best performance in every observation. Moreover, the time-series regressions

performed in this study only employs constant beta. More advance tests which utilize

time-varying beta, structural breaks in stock markets, etc might be an option to

increase the validity of the test.




68

References

Arshanapalli, B., Doukas, J., and Lang, L.H.P. (1995). Pre and post-October 1987

stock market linkages between U.S. and Asian markets. Pacific-Basin Finance

Journal, 3, 57 – 73.

Abimanyu, et al. (2008). International linkages to the Indonesian capital market:

cointegration test. Capital Market and Financial Institution Supervisory

Agency, Ministry of Finance of Indonesia.

Bailey, W., and Stulz, R. (1990). Benefits of international diversification: the case of

Pacific Basin stock markets, Journal of Portfolio Management, 57 – 61.

Ball, R., and Kothari, S. (1989). Non-stationary expected returns: Implications for

tests of market efficiency and serial correlation of returns. Journal of

Financial Economics, 25, 51 – 74.

Banz, R. (1981). The relationship between return and market value of common stocks.

Journal of Financial Economics , 9, 3 – 18.

Barberis, N., Shleifer, A., and Vishny, R. (1998). A model of investor sentiment.

Journal of Financial Economics, 49, 307-343.

Basu, S. (1977). Investment performance of common stocks in relation to their price-

earnings ratios: A test of the efficient market hypothesis. Journal of Finance,

32, 663 – 682.

Bauer, R., Derwall, J., and Molenaar, R. (2004). The real-time predictability of size

and value premium in Japan. Pacific-Basin Finance Journal, 12, 503 -523.

Bauman, W.S., Conover, C.M., and Miller, R.E. (1998). Growth versus value and

large-cap versus small-cap stocks in international markets. Financial Analysts

Journal, 54, 75 – 89.

Bauman, W.S., Conover, C.M., and Miller, R.E. (1999). Investor overreaction in

international stock markets. Journal of Portfolio Management, 24(4), 102-111.

Bauman, W.S., Conover, C.M., and Miller, R.E. (2001). The performance of growth

stocks and value stocks in the Pacific Basin. Review of Pacific Basin Financial

Markets and Policies, 4, 95 – 108.

Bauman, W.S., and Miller, R.E. (1997). Investor expectations and the performance of

value stocks versus growth stocks. Journal of Portfolio Management, 23, 57 –

68.




69

Baytas, A., and Cakici, N. (1999). Do markets overreact: International evidence.

Journal of Banking and Finance, 23, 1121 – 1144.

Blume, L., Easley, D., and O’Hara, M. (1994). Market statistics and technical

analysis: The role of volume. Journal of Finance, 49, 163-181.

Brown, S., Rhee, S.G., and Zhang, L. (2008). The return to value and momentum in

Asian markets. Emerging Markets Review, 9, 79-88.

Brown, S., Rhee, S.G., and Zhang, L. (2008). The return to value in Asian stock

markets. Emerging Markets Review, 9, 194-205.

Capaul, C., Rowley, I., and Sharpe, W.F. (1993). International value and growth stock

returns. Financial Analysts Journal, January-February, 27 – 36.

Chan, L.K. C. (1988). On the contrarian investment strategy. The Journal of Business,

61, 147 – 163.

Chan, L.K.C., and Chen, N. (1988). An unconditional asset-pricing test and the role of

firm size as an instrumental variable for risk. Journal of Finance, 43, 309 –

325.

Chan, L.K.C., and Chen, N. (1991). Structural and return characteristics of small and

large firms. Journal of Finance, 46, 1467 – 1484.

Chan, L.K.C., Hamao, Y., and Lakonishok, J. (1991). Fundamentals and stock returns

in Japan. Journal of Finance, 46, 1739 -1764.

Chan, K., Hameed, A., and Tong, W. (2000). Profitability of momentum strategies in

the international equity markets. Journal of Financial and Quantitative

Analysis, 35, 153 – 172.

Chan, L.K.C., Jegadeesh, N., and Lakonishok, J. (1999). The profitability of

momentum strategies. Financial Analysts Journal, 80 – 90.

Chan, L.K.C., Karceski, J., and Lakonishok, J. (2003). The level and persistence of

growth rates. Journal of Finance,58(2), 643-684.

Chan, L.K.C., Karolyi, G.A., and Stulz, R.M. (1992). Global financial markets and the

risk premium on U.S. equity. Journal of Financial Economics, 32, 137-167.

Chan, L.K.C., and Lakonishok, J. (2004). Value and growth investing: Review and

update. Financial Analysts Journal, 60, 71 -86.

Chen, N.F., Roll, R., and Ross, S.A. (1986). Economic forces and the stock market.

Journal of Business, 59(3), 383-403.




70

Chen, C.R., and Sauer, D.A. (1997). Is stock market overreaction persistent over

time? Journal of Business Finance and Accounting, 24, 51 – 66.

Chincarini, L.B., and Kim, D. (2006). Quantitative Equity Portfolio Management.

New York: McGraw Hill.

Chopra, N., Lakonishok, J., and Ritter, J.R. (1992). Measuring abnormal performance:

Do stocks overreact? Journal of Financial Economics 31, 235 – 268.

Chordia, T., and Swaminathan, B. (1999). Differential speeds of adjustment, cross-

autocorrelations, and reaction to earnings announcements. The Journal of

Finance.

Chordia, T., and Swaminathan, B. (2000). Trading volume and cross-autocorrelations

in stock returns. The Journal of Finance, 55, 913 – 935.

Chordia, T., and Shivakumar, L. (2002). Momentum, business cycle, and tme-varying

expected returns. Journal of Finance, 57(2), 985-1019.

Chui, A. C. W., Wei, K.C.J., and Titman, S. (2000). Momentum, Legal systems, and

ownership structure: An analysis of Asian stock markets. Working Paper.

Conrad, J., Cooper, M., and Kaul, G. (2003). Value versus glamour. The Journal of

Finance, 58, 1969 – 1995.

Conrad, J., Hameed, A., and Niden, C. (1994). Volume and autocovariances in short-

horizon individual security returns. Journal of Finance, 49, 1305 – 1329.

Conrad, J., and Kaul, G. (1998). An anatomy of trading strategies. Review of

Financial Studies, 11, 489 – 519.

Daniel, K., Hirshleifer, D., and Subrahmanyam, A. (1998). Investor psychology and

security market under- and overreactions. Journal of Finance, 53, 1839-1886.

Daniel, K., and Titman, S. (1997). Evidence on the characteristics of cross sectional

variation in stock returns. Journal of Finance, 52, 1 – 33.

Davis, J.L. (1994). The cross-section of realized stock returns: The pre-Compustat

evidence. Journal of Finance, 49(5), 1579-1593.

Davis, J.L. (1996). The cross-section of stock returns and survivorship bias: evidence

from delisted stocks. The Quarterly Review of Economics and Finance, 36,

365 – 375.

De Bondt, W., and Thaler, R. (1985). Does the stock market overreact? Journal of

Finance, 40, 793 – 805.




71

De Bondt, W., and Thaler, R. (1987). Further evidence on investor overreaction and

stock market seasonality. Journal of Finance, 42, 557 – 581.

DeFusco, R.A., Geppert, J.M., and Tsetsekos, G.P. (1996). Long-run diversification

potential in emerging stock markets. The Financial Review, 31(2), 343-363.

Ding, D.K., Chua, J.L., and Fetherston, T.A. (2005). The performance of value and

growth portfolios in East Asia before the Asian financial crisis. Pacific-Basin

Finance Journal, 13, 185 – 199.

Dissanaike, G. (1996). Are stock price reversals really asymmetric? A note. Journal

of Banking and Finance, 20, 189 – 201.

Dissanaike, G. (1997). Do stock market investors overreact? Journal of Business

Finance and Accounting, 24, 27 – 49.

Doukas, J.A., Kim, C.F., and Pantzalis, C. (2004). Divergent opinions and the

performance of value stocks. Financial Analysts Journal, 60(6), 55 – 64.

Fama, E.F., and French, K.R. (1992). The cross-section of expected stock returns.

Journal of Finance, 46, 427 – 466.

Fama, E.F., and French, K.R. (1995). Size and Book-to-market factors in earnings and

returns. Journal of Finance, 50, 131 – 155.

Fama, E.F., and French, K.R. (1996). Multifactor explanations of asset pricing

anomalies. Journal of Finance, 51, 55 – 84.

Fama, E.F., and French, K.R. (1998). Value versus growth: the international evidence.

Journal of Finance, 53, 1975 -1999.

Fama, E.F., and French, K.R. (2004). The capital asset pricing model: theory and

evidence. Journal of Economics Perspectives, 18(3), 25-46.

Fama, E.F., and French, K.R. (2007). The anatomy of value and growth stock returns.

Financial Analysts Journal, 63 (6), 44 – 54.

Fama, E.F., and MacBeth, J. (1973). Risk, return, and equilibrium: Empirical tests.

Journal of Political Economy, 81, 607 – 636.

Fletcher, J. (1997). An examination of the cross-sectional relationship of beta and

return: UK evidence. Journal of Economics and Business, 49(3), 211-221.

French, K., Schwert, G.W., and Stambaugh, R.F. (1987). Expected stock return and

volatility. Journal of Financial Economics, 19, 3 – 29.




72

Fung, H., Leung, W., and Patterson, G. (1999). Do trading rules based upon winners

and losers work across markets? Evidence from the Pasific-Basin and U.S.

markets. Multinational Finance Journal, 3, 41-70.

Gibbons, M.R., Ross, S.A., and Shanken, J. (1989). A test of efficiency of a given

portfolio. Econometrica, 57, 1121-1152.

Griffin, J.M., Xiuqing, J., and Martin, J.S. (2003). Momentum investing and business

cycle risk: evidence from pole to pole. Journal of Finance, 58, 2515 – 2547.

Grinblatt, M., Titman, S., and Wemers, R. (1995). Momentum strategies, portfolio

performance, and herding: A study of mutual fund behavior. American

Economic Review, 85, 1088-1105.

Hameed, A., and Ting, S. (2000). Trading volume and short-horizon contrarian

profits: Evidence from the Malaysian market. Pacific-Basin Finance Journal,

8, 67 – 84.

Hart, J., Zwart, G., and Dijk, D. (2005). The success of stock selection strategies in

emerging markets: is it risk or behavioral bias? Emerging Markets Review, 6,

238 – 262.

Haugen, R.A. (1996). The effects of intrigue, liquidity, imprecision, and bias on the

cross-section of expected stock returns. Journal of Portfolio Management,

22(4), 8-17.

Haugen, R.A., and Baker, N.L. (1996). Commonality in the determinants of expected

stock returns. Journal of Financial Economics, 41, 401 – 439.

Haugen, R.A., Talmor, E., and Torous, W. (1991). The effect of volatility changes on

the level of stock prices and subsequent expected returns. Journal of Finance,

46, 985 – 1007.

Heston, S., Rouwenhorst, K.G., and Wessels, R.E. The structure of internaitonal stock

returns and the integration of capital markets. Journal of Empirical Finance 2,

173-197.

Hong, H. And Stein, J.C. (1999). A unified theory of underreaction, momentum

trading and overreaction in asset markets. Journal of Finance, 54, 2143-2184.

Jaffe, J., Keim, D.B., and Westerfield, R. (1989). Earnings yield, market values, and

stock returns. Journal of Finance, 44, 284 – 302.

James, C., and Edmister, R.E. (1983). The relation between common stock returns,

trading activity and market value. Journal of Finance, 38, 1075 – 1086.




73

Jegadeesh, N. (1990). Evidence of predictable behaviour of security returns. Journal

of Finance, 45, 881 – 899.

Jegadeesh, N., and Titman, S. (1993). Returns to buying winners and selling losers:

Implications for market efficiency. Journal of Finance, 48, 65 – 92.

Jegadeesh, N., and Titman, S. (1995). Overreaction, Delayed Reaction, and

Contrarian Profits. The Review of Financial Studies, 8, 973 – 993.

Jegadeesh, N., and Titman, S. (2001). Profitability of momentum strategies: an

evaluation of alternative explanations. Journal of Finance, 56, 699 – 720.

Jones, S.L. (1993). Another look at time-varying risk and return in a long-horizon

contrarian strategy. Journal of Financial Economics, 33, 119 – 144.

Kahneman, D., Slovic, P., and Tversky, A. (1982). Judgment under uncertainty:

Heuristic and biases. New York: Cambridge University Press.

Keim, D., and Stambaugh, R. (1984). A further investigation of the weekend effect in

stock returns. Journal of Finance, 39, 819 – 834.

Kothari, S.P., Shanken, J., and Sloan, R.G. (1995). Another look at the cross-section

of expected stock returns. Journal of Finance, 50, 185 – 224.

Lakonishok, J., and Shapiro, A.C., (1997). Systematic risk, total risk, and size as

determinants of stock market returns. Journal of Banking and Finance, 10,

115 – 132.

Lakonishok, J., Shleifer, A., and Vishny, R.W. (1992). The impact of institutional

trading on stock prices. Journal of Financial Economics, 32, 23-43.

Lakonishok, J., Shleifer, A., and Vishny, R.W. (1994). Contrarian investment,

extrapolation, and risk. Journal of Finance, 49, 1541 – 1578.

La Porta, R., Lakonishok, J., Shleifer, A., and Vishny, R. (1997). Good news for value

stocks: Further evidence on market efficiency. Journal of Finance, 52, 859 –

874.

Lee, C. And Swaminathan, B. (2000). Price momentum and trading volume. Journal

of Finance, 55, 2017-2069.

Lee, I. (1991). The turn-of-the-year effects in Asian stock markets. Journal of Asian

Economics, 2, 113 – 123.

Lehmann, B., (1990). Fads, martingales and market efficiency. Quarterly Journal of

Economics, 105, 1 – 28.




74

Leland, H.E. (1999). Beyond mean-variance: Performance measurement in a

nonsymmetrical world. Financial Analysts Journal, 55(1). 27-35.

Lintner, J. (1965). The valuation of risk assets and the selection of risky investments

in stock portfolios and capital budgets. Review of Economics and Statistics,

47, 13-37.

Liou, K. T. (2002). Managing Economic Development in Asia: From Economic

Miracle to Financial Crisis. London: Praeger Publishers.

Lo, A.W., and MacKinlay, A.C. (1990). When are contrarian profits due to stock

market overreaction? The Review of Financial Studies, 3, 175 – 205.

Loughran, T. (1997). Book-to-market across firm size, exchange, and seasonality: Is

there an effect? Journal of Financial and Quantitative Analysis, 32, 249 – 268.

Loughran, T., and Ritter, J.R. (1996). Long-term market overreaction: The effect of

low-priced stocks. Journal of Finance, 51, 1959 – 1970.

Malliaropulos, D., and Priestley, R. (1999). Mean reversion in Southeast Asian stock

markets. Journal of Empirical Finance, 6, 355 – 384.

McInish, T.H. et al. (2008). Short-horizon contrarian and momentum strategies in

Asian markets: An integrated analysis. International Review of Financial

Analysis , 17, 312 – 329.

Mertens, E. (2002). The CAPM and Regression Tests [PDF document]. Retrieved

from Lecture Notes Online Web site:

http://www.elmarmertens.ch/LectureNoteCAPM.pdf

Mian, G.M., and Teo, T.G.L. (2004). Do errors in expectations explain the cross-

section of stock returns? Pacific-Basin Finance Journal, 12, 197 – 217.

Nam, K., Pyun, C.S., and Kim, S. (2003). Is asymmetric mean-reverting pattern in

stock returns systematic? Evidence from Pacific-Basin markets in the short-

horizon. Journal of International Financial Markets, Institutions and Money,

13, 481 -502.

Pettengill, G.N., Sundaram, S., and Mathur, I. (1995). The conditional relations

between beta and returns. The Journal of Financial and Quantitative Analysis,

30(1), 101-116.

Poterba, J.M., and Summers, L.H. (1988). Mean reversion in stock prices: Evidence

and implications. Journal of Financial Economics, 22, 27 – 59.

Reinganum, M. (1981). Misspecification of capital asset pricing: empirical anomalies

based on earnings’ yields and market values. Journal of Financial Economics,

9, 19 – 46.




75

Richards, A. (1997). Winner-loser reversals in national stock market indices: can they

be explained? Journal of Finance, 52, 2129 – 2144.

Roll, R., (1995). An empirical survey of Indonesian equities: 1985 – 1992. Pacific-

Basin Finance Journal, 3, 159 – 192.

Rouwenhorst, K.G. (1998). International momentum strategies. Journal of Finance,

53, 267 – 284.

Rouwenhorst, K.G. (1999). Local return factors and turnover in emerging stock

markets. Journal of Finance, 54, 1439 – 1464.

Scholes, M. and Williams, J. (1977). Estimating betas from nonsynchronous data.

Journal of Financial Economics, 5, 309-327.

Sharpe, W.F. (1964). Capital asset prices: A theory of market equilibrium under

conditions of risk.Journal of Finance, 19, 425-442.

Shen, Q., Szakmary, A.C., and Sharma, S.C. (2005). Momentum and contrarian

strategies in international stock markets : Further evidence. Journal of

Multinational Financial Management, 15, 235 – 255.

Shleifer, A. And Vishny, R. (1997). The limits of arbitrage. Journal of Finance, 52(1),

35-55.

Williams, J.T. (1977). Capital asset prices with heterogeneous beliefs. Journal of

Financial Economics, 5, 219 – 239.

Yen, J.Y., Sun, Q., and Yan, Y. (2004). Value versus growth stocks in Singapore.

Journal of Multinational Financial Management, 14, 19 – 34.

Zarowin, P. (1990). Size, seasonality, and stock market overreaction. Journal of

Financial and Quantitative Analysis 25, 125 – 133.




76

Text Figures

Figure 1

The Risk Return Tradeoff

Source: Fama and French (2004)

g

>

O

h

i

�(�)

E(R) Mean-variance

efficient frontier

with a riskless

asset Minimum variance frontier

for risky assets

��




77

Figure 2

Jakarta Composite Index and Capital Market Milestones

1984-June 2008

Source: Indonesia Stock Exchange

30-Nov-2007 Consolidation of the SSX into

JSX to become IDX

2,688.332

9-Sept-2002

T+4 to T+3Settlement

430.271

28-Mar-2002 Remote Trading

481.775

6-Oct-2004

Launching Of Stock

Option

856.060

21-Jul-2000 Scripless

Trading

512.617

23-Dec-1997

Founding

of KPEI 397.031

1-May-1995

JATS

415.322

24-Jul-1995 Merging Process of JSX into

Bursa Paralel Indonesia

509.532

23-Jul-1997

Financial Crisis

718.189

13-Jul-1992

Privatization of JSX

321.544

16-Jun-1989

Establishment of SSX

293.548

8-Aug-1996

Founding of KSEI

548.181

83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08

3,000

2,500

2,000

1,500

1,000

500




78

Text Tables

Table I

The Complete Overview of all Variables

Variables Measured by Explanation

BE/ME Book-to-Market

The book value of equity for the fiscal year ending in calendar year

t-1

divided by market equity at the end of portfolio formation month

E/P Earnings to Price

The earnings before extraordinary items but after interest,

depreciation,

Earning Yield taxes, and preferred dividends for the fiscal year ending in

calendar year t-1, divided by market equity at the end of portfolio

formation month.

CF/P Cash Flow to Price

The cash flow (Earnings plus depreciation) for the fiscal year ending

in

Cash Flow Yield calendar year t-1, divided by market equity at the end of portfolio

formation month

D/P Dividends to Price The dividend per share divided by stock price

Dividend Yield

ME Market Equity Stock price times number of shares outstanding

Market Capitalization

Trading Volume Volume Turnover Number of shares traded divided by number of shares outstanding

Lagged Trading

Volume

Lagged Volume

Turnover Trading volume during the formation period

P/B Price to Book value Stock price divided by book value of equity per share

The inverse of BE/ME

P/E Price to Earnings The inverse of E/P

P/D Price to Dividends The inverse of D/P




79

Table II

Trading Value on Indonesian Stock Market

Year 2002 2003 2004 2005 2006 2007 Q1-2008

TRADING (IDR Billion)

Total Trading Value of Shares 120,763 125,438 247,007 406,006 445,708 1,050,154 317,446

Foreign Transaction of Shares 13,607 40,133 110,762 157,019 140,506 243,803 78,484

Foreign percentage (%) 11.27 31.99 44.84 38.67 31.52 23.22 24.72

Local Transaction of Shares 107,157 85,305 136,245 248,987 305,202 806,351 238,962

Source: Indonesia Stock Exchange

Dote: For all tables, the comprehensive procedures in forming the portfolios are

provided in section III.

Table III

Returns for Quintiles Portfolios Based on One-Dimensional Classification by

Various Measures of Value Characteristics

The returns presented in the table are averages over all formation periods. Rt is the average return in

year � after formation, � = 1, … ,5. For example: at the end of December 1991, JKSE stocks are ranked

in ascending order into quintiles based on BE/ME, CF/P, and E/P. The R1 are calculated from January

1992 to December 1992. AR is the average annual return over 5 post formation years. The portfolios

are rebalanced annually. Glamour portfolios refer to portfolios that contain stocks with lowest rank on

BE/ME, CF/P, and E/P. Value portfolios refer to portfolios of stocks with highest rank on BE/ME,

CF/P, and E/P. Portfolios 0 refer to stocks with negative values on those value characteristics. Panel

A2, B2, and C2 provide the value-weighted returns.

Glamour Value

0 1 2 3 4 5

Panel A1: Rank Based on BE/ME

R1 0,0320 -0,0011 0,0056 0,0132 0,0241 0,0491

R2 0,0315 0,0048 0,0105 0,0145 0,0255 0,0458

R3 0,0348 0,0077 0,0106 0,0158 0,0250 0,0381

R4 0,0356 0,0091 0,0137 0,0167 0,0235 0,0367

R5 0,0347 0,0127 0,0160 0,0179 0,0254 0,0368

AR 0,0337 0,0066 0,0113 0,0156 0,0247 0,0413

Panel B1: Rank Based on CF/P

R1 0,0160 0,0279 0,0209 0,0226 0,0117 0,0101

R2 0,0207 0,0170 0,0207 0,0193 0,0192 0,0298

R3 0,0250 0,0155 0,0190 0,0185 0,0220 0,0282

R4 0,0245 0,0159 0,0137 0,0145 0,0219 0,0231

R5 0,0267 0,0209 0,0134 0,0149 0,0219 0,0283

AR 0,0226 0,0194 0,0175 0,0180 0,0193 0,0239




80

Glamour Value

0 1 2 3 4 5

Panel C1: Rank Based on E/P

R1 0,0389 0,0058 0,0066 0,0122 0,0213 0,0388

R2 0,0168 0,0129 0,0122 0,0155 0,0232 0,0356

R3 0,0137 0,0147 0,0130 0,0164 0,0208 0,0329

R4 0,0156 0,0155 0,0157 0,0167 0,0202 0,0323

R5 0,0192 0,0170 0,0179 0,0183 0,0214 0,0314

AR 0,0208 0,0132 0,0131 0,0158 0,0214 0,0342

Panel A2: Rank Based on BE/ME-Value Weighted

R1 0,0131 0,0005 0,0004 0,0006 0,0014 0,0019

R2 0,0037 0,0004 0,0006 0,0005 0,0013 0,0014

R3 0,0059 0,0004 0,0005 0,0005 0,0010 0,0010

R4 0,0074 0,0004 0,0006 0,0005 0,0010 0,0010

R5 0,0072 0,0005 0,0007 0,0006 0,0010 0,0011

AR 0,0075 0,0004 0,0006 0,0005 0,0011 0,0013

Panel B2: Rank Based on CF/P-Value Weighted

R1 -0,0003 0,0012 0,0008 0,0008 0,0004 -0,0003

R2 -0,0003 0,0008 0,0005 0,0008 0,0008 0,0005

R3 0,0000 0,0005 0,0005 0,0006 0,0007 0,0005

R4 0,0007 0,0005 0,0004 0,0004 0,0006 0,0005

R5 0,0011 0,0006 0,0004 0,0005 0,0007 0,0007

AR 0,0002 0,0007 0,0005 0,0006 0,0007 0,0004

Panel C2: Rank Based on E/P-Value Weighted

R1 0,0229 0,0005 0,0005 0,0006 0,0011 0,0015

R2 0,0127 0,0004 0,0004 0,0006 0,0010 0,0013

R3 0,0123 0,0004 0,0005 0,0005 0,0008 0,0012

R4 0,0126 0,0004 0,0005 0,0006 0,0008 0,0011

R5 0,0121 0,0005 0,0005 0,0007 0,0008 0,0010

AR 0,0145 0,0004 0,0005 0,0006 0,0009 0,0012

Table IV

Returns for Portfolios Based on Two-Dimensional Classifications by Various

Measures of Value Characteristics


year � after formation, � = 1, … ,5 . For example: at the end of December 1991, JKSE stocks are

classified into 9 groups based on the intersection of BE/ME and CF/P. The R1 are calculated from

January 1992 to December 1992 for each of the portfolios in the 9 groups of stocks. AR is the average

annual return over 5 post formation years. The portfolios are rebalanced annually. Glamour portfolios

refer to portfolios that contain stocks with lowest rank on both value characteristics. Whereas value

portfolios refer to portfolios of stocks with highest rank on two value characteristics. Panel A2 and B2

provide the value-weighted returns.




81

Panel A1: BE/ME and E/P Equal-Weighted

Glamour Value

BE/ME 1 1 1 2 2 2 3 3 3

E/P 1 2 3 1 2 3 1 2 3

R1 0,0002 0,0011 0,0095 0,0023 0,0119 0,0229 0,0367 0,0299 0,0420

R2 0,0103 0,0049 0,0246 0,0173 0,0140 0,0230 0,0178 0,0284 0,0379

R3 0,0127 0,0088 0,0127 0,0169 0,0144 0,0176 0,0271 0,0279 0,0199

R4 0,0185 0,0141 0,0151 0,0286 0,0156 0,0194 0,0331 0,0164 0,0229

R5 0,0243 0,0251 0,0375 0,0171 0,0186 0,0228 0,0724 0,0338 0,0293

AR 0,0132 0,0108 0,0199 0,0164 0,0149 0,0211 0,0374 0,0273 0,0304

Panel B1: BE/ME and CF/P Equal-Weighted

Glamour Value

BE/ME 1 1 1 2 2 2 3 3 3

CF/P 1 2 3 1 2 3 1 2 3

R1 0,0094 0,0010 -0,0211 0,0258 0,0202 -0,0041 0,0675 0,0459 0,0287

R2 -0,0001 0,0031 0,0341 0,0080 0,0134 0,0282 0,0428 0,0244 0,0386

R3 0,0087 0,0095 0,0097 0,0130 0,0141 0,0278 0,0391 0,0137 0,0248

R4 0,0164 0,0099 0,0182 0,0214 0,0181 0,0254 0,0244 0,0192 0,0303

R5 0,0226 0,0295 0,0250 0,0320 0,0150 0,0211 0,0272 0,0402 0,0257

AR 0,0114 0,0106 0,0132 0,0200 0,0162 0,0197 0,0402 0,0287 0,0296

Panel A2: BE/ME and E/P Value-Weighted

Glamour Value

BE/ME 1 1 1 2 2 2 3 3 3

E/P 1 2 3 1 2 3 1 2 3

R1 0,0005 0,0011 0,0027 0,0012 0,0008 0,0031 0,0089 0,0048 0,0029

R2 0,0006 0,0003 0,0242 0,0012 0,0011 0,0045 0,0030 0,0020 0,0025

R3 0,0008 0,0001 -0,0077 0,0006 0,0007 0,0009 0,0048 0,0022 0,0005

R4 0,0009 0,0019 0,0057 0,0036 0,0008 0,0024 0,0104 0,0008 0,0009

R5 0,0011 0,0028 0,0243 0,0018 0,0014 0,0025 0,0550 0,0043 0,0016

AR 0,0008 0,0013 0,0098 0,0017 0,0010 0,0027 0,0164 0,0028 0,0017

Panel B2: BE/ME AND CF/P Value-Weighted

Glamour Value

BE/ME 1 1 1 2 2 2 3 3 3

CF/P 1 2 3 1 2 3 1 2 3

R1 0,0013 0,0012 -0,0211 0,0071 0,0011 -0,0003 0,0116 0,0054 0,0019

R2 0,0005 0,0008 0,0341 0,0039 0,0008 0,0020 0,0069 0,0022 0,0023

R3 0,0003 0,0007 0,0097 0,0007 0,0004 0,0008 0,0040 -0,0017 0,0006

R4 0,0010 -0,0010 0,0182 -0,0057 0,0006 0,0049 0,0019 0,0004 0,0013

R5 0,0007 0,0027 0,0250 0,0122 0,0010 0,0014 0,0051 0,0046 0,0015

AR 0,0008 0,0009 0,0132 0,0036 0,0008 0,0018 0,0059 0,0022 0,0015




82

Table V

Returns for Portfolios Based on Size-Classifications



in ascending order into quintiles based on their ME. The R1 are calculated from January 1992 to

December 1992. AR is the average annual return over 5 post formation years. The portfolios are

rebalanced annually. Big portfolios refer to portfolios that contain stocks with highest rank on ME.

Small portfolios refer to portfolios of stocks with lowest rank on ME. Panel B provides the value-

weighted returns.

Panel A: Rank Based on Size

Big Small

1 2 3 4 5

R1 0,0160 0,0151 0,0167 0,0174 0,0255

R2 0,0157 0,0154 0,0197 0,0226 0,0328

R3 0,0143 0,0126 0,0179 0,0237 0,0309

R4 0,0141 0,0134 0,0183 0,0250 0,0304

R5 0,0144 0,0153 0,0201 0,0265 0,0303

AR 0,0149 0,0144 0,0185 0,0231 0,0300

Panel B: Rank Based on Size-Value Weighted

Big Small

1 2 3 4 5

R1 0,0004 0,0003 0,0003 0,0004 0,0006

R2 0,0004 0,0004 0,0004 0,0005 0,0007

R3 0,0004 0,0003 0,0004 0,0006 0,0006

R4 0,0004 0,0003 0,0004 0,0006 0,0007

R5 0,0004 0,0004 0,0004 0,0007 0,0007

AR 0,0004 0,0003 0,0004 0,0005 0,0007

Table VI

An Overlapping Momentum/Contrarian Average Monthly Returns of JKSE

Stocks: 12/1991-12/2008 using Lo and MacKinlay (1990) Methodology

At the beginning of each month, all stocks are ranked based on their past performance during formation

period (F). The stocks that outperform the market are assigned to Winner portfolio, those that

underperform the market to the Loser portfolio. The portfolios return are value weighted depends on

their past performance and held for holding (H) period months. The table gives the average monthly

returns for both winner and loser portfolios for the period 1991 to 2008. In addition, the returns of zero-

cost strategies which buy winner and sell loser are also presented. t-stat is the average return divided by

its standard error.




83

F/H Total Winner Loser

1/1 Average Returns -0,14% 0,09% -0,23%

t-stat -1,41 0,41 -1,39

3/3 Average Returns -0,06% 0,07% -0,13%

t-stat -2,60 2,49 -3,89

6/6 Average Returns -0,01% 0,07% -0,08%

t-stat -1,52 4,12 -5,75

12/12 Average Returns -0,04% 0,03% -0,06%

t-stat -4,99 3,97 -5,75

Table VII

Momentum/ Contrarian Average Returns using Lo and MacKinlay (1990)

Methodology that incorporates Trading Volume

At the beginning of each month, stocks are divided into three groups based on their lagged trading

volume. Within each volume group, all stocks are ranked based on their formation period (F)

performance. The stocks that outperform the market are assigned to Winner portfolio, those that

underperform the market to the Loser portfolio. The portfolios return are value weighted depends on

their past performance and held for H months. The table gives the average monthly returns for both

winner and loser portfolios for the period 1991 to 2008. In addition, the returns of zero-cost strategies

which buy winner and sell loser are also presented. t-stat is the average return divided by its standard

error.

F/H Volume Turnover Total Winner Loser

1/1 High AR 0,11% 0,35% -0,24%

t-stat 0,80 2,35 -1,94

Medium AR -0,21% 0,19% -0,40%

t-stat -2,77 1,52 -2,99

Low AR -0,17% 0,12% -0,29%

t-stat -1,37 0,54 -2,03

12/12 High AR -0,02% 0,03% -0,05%

t-stat -2,55 3,34 -3,96

Medium AR -0,04% 0,02% -0,06%

t-stat -4,99 3,03 -5,06

Low AR -0,03% 0,03% -0,06%

t-stat -5,55 7,06 -7,74




84

Table VIII

The CAPM Time-Series Regressions for Monthly Excess Returns on 25

Portfolios Formed on Size and BE/ME: 12/1991-12/2008, 205 Months

GRS is the F-statistic of Gibbons, Ross, and Shanken (1989), testing the hypothesis that the regression

intercepts for a set of five portfolios are all 0,0. p(GRS) is the p-value of GRS, that is the probability of

a GRS value as large or larger than the observed value if the zero-intercepts hypothesis is true. t( )

refers to t-statistics, the average return divided by its standard error. s(e) refers to standar error of the

regressions.

Regressions: ��'−�� = < + ��' − �� + =�'

Book-to-Market Equity (BE/ME) Quintiles

Size Low 2 3 4 High Size Low 2 3 4 High

a t(a)

Small -0,01 -0,03 -0,02 -0,03 -0,02 Small -0,78 -1,98 -1,01 -3,22 -3,40

2 0,01 -0,01 -0,01 -0,01 -0,03 2 0,36 -0,71 -0,69 -1,60 -3,82

3 0,02 -0,01 0,01 0,07 -0,02 3 1,32 -1,29 1,19 3,06 -2,29

4 0,02 -0,01 0,01 0,00 0,00 4 2,15 -0,90 1,80 -0,15 0,35

Big -0,01 0,00 -0,01 -0,01 -0,03 Big -0,95 0,07 -0,99 -1,75 -0,72

GRS = 1, 59, p-value = 0,99

b

t(b)

Small 0,85 0,73 0,62 0,88 1,03

Small 14,54 9,76 5,40 21,58 28,01

2 0,89 0,73 0,70 0,97 1,00

2 11,42 10,35 20,67 21,85 23,11

3 0,85 0,77 1,07 1,43 1,11

3 11,34 13,29 23,40 12,81 26,72

4 0,94 0,86 1,07 1,12 1,19

4 18,87 26,55 31,17 29,60 22,52

Big 0,79 0,90 0,84 1,04 1,19

Big 23,66 23,68 14,22 24,20 4,44

R2 s(e)

Small 0,58 0,34 0,15 0,73 0,81 Small 0,12 0,15 0,22 1,17 0,07

2 0,43 0,38 0,70 0,72 0,73 2 0,16 0,14 0,09 0,09 0,09

3 0,40 0,50 0,74 0,47 0,80 3 0,15 0,12 0,09 0,23 0,08

4 0,68 0,79 0,83 0,83 0,76 4 0,10 0,07 0,07 0,08 0,11

Big 0,74 0,75 0,53 0,76 0,11 Big 0,07 0,08 0,12 0,09 0,47




85

Table IX

The CAPM Time-Series Regressions for Monthly excess Returns on the LSV

Equal-Weight Quintiles: 12/1991-12/2008, 205 Months

GRS and p(GRS) are defined in Table VIII. �( ) is the t-statistics, the average return divided by its

standard error.

Quintiles

1 2 3 4 5 GRS p(GRS)

BE/ME Low High

a 0,00 0,00 0,00 0,00 -0,02

b 0,83 0,86 0,95 1,08 1,08

t(a) 0,22 -0,63 -0,63 0,61 -4,31 16,52 0,005

t(b) 33,48 40,76 43,66 40,65 49,76

R2 0,85 0,89 0,90 0,89 0,92

EP Low High

a 0,02 -0,01 -0,02 -0,02 -0,02

b 0,99 0,86 0,85 0,89 0,98

t(a) 3,41 -1,75 -5,79 -4,97 -4,28 12,65 0,027

t(b) 38,67 36,92 43,40 40,20 41,66

R2 0,88 0,87 0,90 0,89 0,90

CFP Low High

a 0,01 0,00 0,00 -0,01 -0,01

b 1,08 0,98 0,95 0,98 1,00

t(a) 1,49 0,50 -0,52 -1,20 -1,01 0,49 0,99

t(b) 33,76 34,88 36,68 45,12 41,82

R2 0,89 0,90 0,91 0,94 0,93




86

Table X

The CAPM Time-Series Regressions for Monthly Excess Returns on Equal-

Weight Portfolios Formed on Past Returns: 12/1992-1/2008, 182 Months

The table below shows the results of the CAPM time series regressions on monthly excess returns of

portfolios formed based on past 12 month returns. GRS and p(GRS) are defined in Table VIII. �( ) is

the t-statistics, the average return divided by its standard error.

Portfolio Formation Months t-12 to t-2

1 2 3 4 5

Low High

a 0,02 0,00 -0,01 -0,01 -0,01

b 1,14 1,00 0,98 0,98 0,92

t(a) 3,91 0,15 -1,67 -1,70 -1,52

t(b) 34,37 44,96 48,56 44,14 36,14

R2 0,87 0,92 0,93 0,92 0,88

GRS = 1,38 , p-value = 0,93

Table XI

The Fama French Three-Factor Regressions for Monthly Excess Returns on 25

Portfolios Formed on Size and BE/ME: 12/1991-12/2008, 205 Months


standard error. s(e) refers to standard error of the regressions.

Regressions: : mno−mp = q + �mro − mp�sn + tnurv + wnxry + zno Book-to-Market Equity (BE/ME) Quintiles


a t(a)

Small 0,00 -0,02 -0,01 -0,02 -0,02 Small 0,24 -1,63 -0,66 -2,72 -3,01

2 0,02 0,00 0,00 -0,01 -0,03 2 1,39 -0,26 -0,13 -1,68 -3,66

3 0,02 -0,02 0,01 0,05 -0,02 3 1,51 -1,38 1,18 2,63 -2,29

4 0,02 -0,01 0,01 0,00 0,00 4 2,29 -1,09 1,41 -0,61 0,00

Big -0,01 0,00 -0,02 -0,02 -0,05 Big -1,14 -0,68 -1,32 -2,36 -1,12

GRS = 0,53 , p-value = 0,99

b t(b)

Small 1,03 0,95 0,75 0,94 1,03 Small 16,13 11,88 5,55 20,31 25,18

2 1,12 0,84 0,94 0,89 0,99 2 12,83 10,45 19,37 17,73 19,87

3 1,01 0,83 1,07 1,04 1,05 3 12,05 12,42 20,07 8,84 22,18

4 1,04 0,90 1,05 1,01 1,07 4 18,45 24,29 26,93 24,84 17,84

Big 0,88 0,90 0,79 1,02 0,59 Big 24,27 21,88 11,64 21,31 1,92




87

Book-to-Market Equity (BE/ME) Quintiles


s t(s)

Small 0,69 0,63 0,48 0,29 0,27 Small 3,71 4,05 1,88 3,26 3,40

2 0,95 0,16 0,40 -0,01 0,05 2 4,10 5,34 4,11 -0,14 0,54

3 0,27 -0,11 0,00 -0,92 0,00 3 1,67 -0,82 0,03 -4,12 0,00

4 -0,03 -0,06 -0,20 -0,33 -0,14 4 -0,27 -0,81 -2,67 -4,20 -1,18

Big 0,00 -0,36 -0,36 -0,35 -1,13 Big -0,03 -4,57 -2,68 -3,79 -1,46

h t(h)

Small -0,40 -0,55 -0,24 -0,12 0,09 Small -2,84 -4,41 -1,13 -1,57 1,43

2 -0,44 -0,21 0,05 0,26 0,05 2 -2,32 -1,63 0,67 3,29 0,60

3 -0,45 -0,23 0,00 1,04 0,21 3 -3,56 -2,28 0,05 5,61 2,85

4 -0,35 -0,16 -0,01 0,26 0,37 4 -3,86 -2,68 -0,17 4,12 3,73

Big -0,30 -0,14 0,06 -0,05 1,60 Big -5,39 -2,19 0,52 -0,69 2,66

R2 s(e)

Small 0,67 0,16 0,17 0,75 0,82 Small 0,10 0,17 0,22 0,08 0,07

2 0,53 0,41 0,72 0,74 0,73 2 0,14 0,14 0,09 0,09 0,09

3 0,45 0,52 0,74 0,57 0,81 3 0,15 0,11 0,09 0,21 0,08

4 0,70 0,80 0,84 0,85 0,78 4 0,10 0,07 0,07 0,07 0,10

Big 0,77 0,78 0,55 0,78 0,19 Big 0,06 0,07 0,12 0,08 0,45

Table XII

The Fama French Three-Factor Regressions for Monthly excess Returns on the

LSV Equal-Weight Quintiles: 12/1991-12/2008, 205 Months


standard error.

Quintiles

1 2 3 4 5 GRS p(GRS)

BE/ME Low High

a 0,00 0,00 0,00 0,00 -0,02

b 0,95 0,91 0,95 0,97 1,05

s 0,13 -0,06 0,04 -0,22 0,09

h -0,38 -0,18 0,01 0,32 0,13

t(a) 0,47 -0,99 -0,50 0,22 -4,05 10,71 0,057

t(b) 42,09 39,88 37,94 36,95 43,94

t(s) 2,93 -1,40 0,80 -4,37 2,04

t(h) -10,97 -5,24 0,14 7,94 3,64

R2 0,90 0,91 0,90 0,92 0,93

EP Low High

a 0,02 -0,01 -0,02 -0,02 -0,02

b 1,04 0,91 0,87 0,91 0,99

s 0,06 0,10 -0,08 -0,04 0,10

h -0,17 -0,15 -0,14 -0,08 0,02




88

Quintiles

1 2 3 4 5 GRS p(GRS)

EP Low High

t(a) 3,57 -1,56 -6,43 -5,13 -3,95 5,96 0,31

t(b) 36,76 35,42 41,10 35,94 36,62

t(s) 1,18 2,10 -2,09 -0,86 1,98

t(h) -3,99 -3,87 -4,19 -2,03 0,56

R2 0,89 0,88 0,91 0,89 0,90

CFP Low High

a 0,01 0,00 0,00 -0,01 0,00

b 1,07 0,97 0,96 0,95 1,00

s -0,02 -0,08 -0,12 -0,04 0,15

h 0,02 0,01 -0,09 0,09 0,03

t(a) 1,35 0,22 -0,67 -1,62 -0,55 0,29 0,99

t(b) 27,73 28,67 32,26 36,86 36,46

t(s) -0,25 -1,31 -2,23 -0,94 3,14

t(h) 0,38 0,22 -2,18 2,35 0,84

R2 0,89 0,90 0,92 0,94 0,93

Table XIII

The Fama French Three-Factor Regressions for Monthly Excess Returns on

Equal-Weight Portfolios Formed on Past Returns: 12/1992-1/2008, 182 Months

The table below shows the results of time series regressions of Fama French three-factor model on

monthly excess returns of portfolios formed based on past 12 month returns. GRS and p(GRS) are

defined in Table VIII. �( ) is the t-statistic, the average return divided by its standard error.

Portfolio Formation Months t-12 to t-2

1 2 3 4 5 GRS p(GRS)

Low High

a 0,02 0,00 -0,01 -0,01 -0,01

b 1,06 0,98 0,99 1,04 0,96

s -0,08 0,08 -0,07 0,10 -0,06

h 0,25 0,09 -0,04 -0,18 -0,16

t(a) 3,92 0,58 -1,94 -1,57 -1,88 0,39 0,99

t(b) 28,51 38,03 41,72 42,23 32,77

t(s) -0,83 1,24 -1,11 1,55 -0,83

t(h) 3,48 1,86 -0,89 -3,67 -2,82

R2 0,88 0,92 0,93 0,92 0,88




89

Table XIV

Returns for Quintiles Portfolios Based on One-Dimensional Classification by

Various Measures of Value Characteristics Post Crisis Period



in ascending order based on BE/ME, CF/P, and E/P. The R1 are calculated from January 1999 to

December 1999. AR is the average annual return over 5 post formation years. The portfolios are

rebalanced annually. Glamour portfolios refer to portfolios that contain stocks with lowest rank on

BE/ME, CF/P, and E/P. Value portfolios refer to portfolios of stocks with highest rank on BE/ME,

CF/P, and E/P. Portfolios 0 refer to stocks with negative values on those value characteristics. Panel

A2, B2, and C2 provide the value-weighted returns.

Panel A

Glamour Value

1 2 3 4 5 5-1

Rank Based on BE/ME Post Crisis

R1 0,0018 0,0098 0,0188 0,0300 0,0615 0,0597

R2 0,0060 0,0121 0,0153 0,0242 0,0479 0,0418

R3 0,0068 0,0134 0,0166 0,0229 0,0402 0,0334

R4 0,0090 0,0151 0,0175 0,0226 0,0378 0,0288

R5 0,0126 0,0171 0,0186 0,0244 0,0397 0,0271

AR 0,0072 0,0135 0,0174 0,0248 0,0454 0,0382

Rank Based on CF/P Post Crisis

R1 0,0167 0,0143 0,0157 0,0076 0,0069 -0,0098

R2 0,0158 0,0199 0,0199 0,0201 0,0340 0,0182

R3 0,0146 0,0210 0,0187 0,0246 0,0319 0,0174

R4 0,0140 0,0146 0,0143 0,0217 0,0214 0,0074

R5 0,0190 0,0120 0,0138 0,0191 0,0309 0,0119

AR 0,0160 0,0163 0,0165 0,0186 0,0250 0,0090

Rank Based on E/P Post crisis

R1 0,0086 0,0128 0,0161 0,0254 0,0447 0,0361

R2 0,0130 0,0145 0,0155 0,0216 0,0323 0,0193

R3 0,0150 0,0155 0,0160 0,0187 0,0319 0,0168

R4 0,0160 0,0169 0,0157 0,0199 0,0311 0,0151

R5 0,0190 0,0188 0,0167 0,0206 0,0318 0,0128

AR 0,0143 0,0157 0,0160 0,0212 0,0344 0,0200




90

Panel B

Glamour Value

1 2 3 4 5 5-1

Rank Based on BE/ME-Value Weighted Post Crisis

R1 0,0003 0,0004 0,0005 0,0013 0,0018 0,0015

R2 0,0002 0,0003 0,0003 0,0008 0,0011 0,0009

R3 0,0003 0,0004 0,0003 0,0007 0,0008 0,0006

R4 0,0003 0,0005 0,0004 0,0007 0,0009 0,0006

R5 0,0004 0,0005 0,0005 0,0008 0,0011 0,0008

AR 0,0003 0,0004 0,0004 0,0009 0,0011 0,0009

Rank Based on CF/P-Value Weighted Post Crisis

R1 0,0008 0,0005 0,0006 0,0006 0,0002 -0,0006

R2 0,0004 0,0003 0,0005 0,0005 0,0003 -0,0001

R3 0,0002 0,0004 0,0005 0,0005 0,0004 0,0001

R4 0,0002 0,0002 0,0003 0,0003 0,0002 0,0000

R5 0,0005 0,0002 0,0004 0,0003 0,0005 0,0001

AR 0,0004 0,0003 0,0004 0,0004 0,0003 -0,0001

Rank Based on E/P-Value Weighted Post crisis

R1 0,0004 0,0005 0,0007 0,0014 0,0015 0,0011

R2 0,0003 0,0004 0,0005 0,0010 0,0010 0,0007

R3 0,0004 0,0005 0,0005 0,0008 0,0010 0,0006

R4 0,0005 0,0006 0,0005 0,0010 0,0009 0,0005

R5 0,0006 0,0006 0,0007 0,0010 0,0010 0,0004

AR 0,0004 0,0005 0,0006 0,0010 0,0011 0,0007

Table XV

Jegadeesh and Titman (1993) Methodology of an Overlapping

Momentum/Contrarian Average Monthly Returns of JKSE Stocks: 12/1991-

12/2008

At the beginning of each month, three groups are formed based on stocks lagged trading volume.

Within each group, stocks are ranked in ascending order based on their formation period (F)

performance. Stocks in the top 30 percent rank are assigned as winner and those in the bottom 30

percent rank are assigned as loser. The portfolios returns are equal weighted and held for H months.

The table gives the average monthly returns for both winner and loser portfolios for the period 1991 to

2008, resulting in total of 205 monthly average returns. In addition, the returns of zero-cost strategies


error.




91

F/H Volume Turnover Winner-Loser Winner Loser

1/1 High AR 1,20% 2,84% 1,64%

t-stat 1,37 2,58 1,57

Medium AR -1,57% 1,61% 3,17%

t-stat -2,41 2,08 3,10

Low AR -2,99% 1,12% 4,11%

t-stat -4,09 1,38 5,20

12/12 High AR -0,79% 1,65% 2,44%

t-stat -2,57 5,56 5,48

Medium AR -1,17% 1,87% 3,05%

t-stat -3,71 6,90 6,43

Low AR -1,10% 1,88% 2,98%

t-stat -4,89 9,74 8,92

Table XVI


Momentum/Contrarian Average Monthly Returns of JKSE Stocks Sub-Period

Analysis: 12/1991-06/1997 and 08/1998-12/2008

The table gives the average monthly returns for both winner and loser portfolios for two sub-period;

before and after the crisis but excludes the crisis period. In addition, the returns of zero-cost strategies


error.

F/H

Volume

Turnover Winner-Loser Winner Loser

Before Crisis 1/1 High AR 0,05% 2,46% 2,41%

t-stat 0,07 2,58 2,18

Low AR -2,85% -0,07% 2,77%

t-stat -2,88 -0,11 2,67

After Crisis 1/1 High AR 1,46% 3,76% 2,30%

t-stat 1,12 2,28 1,56

Low AR -3,48% 1,97% 5,45%

t-stat -3,50 1,58 5,30

Before Crisis 12/12 High AR -0,32% 1,65% 1,98%

t-stat -1,17 4,10 4,74

Low AR 0,34% 1,65% 1,31%

t-stat 1,80 4,94 3,26




92

F/H

Volume

Turnover Winner-Loser Winner Loser

After Crisis 12/12 High AR -0,47% 2,31% 2,78%

t-stat -1,08 5,98 4,34

Low AR -1,85% 2,23% 4,08%

t-stat -6,04 8,36 8,92

Table XVII


Momentum/Contrarian Average Monthly Returns of JKSE Stocks: 12/1991-

12/2008 exclude Crisis Period 07/1997-07/1998

The table gives the average monthly returns for both winner and loser portfolios for the whole period

1991 to 2008, exclude the crisis period. In addition, the returns of zero-cost strategies which buy

winner and sell loser are also presented. t-stat is the average return divided by its standard error.

F/H Volume Turnover Winner-Loser Winner Loser

1/1 High AR 0,97% 3,31% 2,34%

t-stat 1,09 2,94 2,26

Medium AR -1,67% 2,12% 3,79%

t-stat -2,57 2,70 3,76

Low AR -3,26% 1,27% 4,53%

t-stat -4,44 1,48 5,90

12/12 High AR -0,43% 2,12% 2,54%

t-stat -1,35 7,11 5,43

Medium AR -1,00% 2,22% 3,22%

t-stat -3,03 8,21 6,45

Low AR -1,20% 2,06% 3,26%

t-stat -5,08 9,67 9,13

Show

Documents

value strategies

value premium

size effect

contratrian strategies

sources of momentum

effect of trading volume

data analysis

indonesia stock exchange