Attribution Non-Commercial (BY-NC)

159 views

Attribution Non-Commercial (BY-NC)

- FinQuiz - Smart Summary_ Study Session 13_ Reading 47
- Indian Stock Market
- Stock Market Efficiency
- Efficient Market Hypothesis by Alvin Han
- Chap27
- An Empirical Analysis of Efficiency of the Nigerian Capital Market
- 7-23
- weak form of market Efficiency of KSE, Pakistan
- VIf
- Ch13
- Lecture 1 EMH
- Efficient market theory.pdf
- ECON252_Midterm1_ExamSolution
- History of Fundamental Analysis As A Trading Mechanism.docx
- Topic 4
- ch04 (1).ppt
- Efficient Market Hypothesis
- Basic Tools of Finance
- 3
- Shahis Report

You are on page 1of 290

and Interactions with Developed

and Developing Stock Markets

PhD Thesis

The Efficiency of Arab Stock Markets, Its

Interrelationships and Interactions with Developed and

Developing Stock Markets

program

2006

∗

Email: bashar38@yahoo.com. Tell: 6934406294

II

This thesis is dedicated to the memory of my father, who

passed away this year, for his support, wisdom and

endless knowledge. I hope that I have achieved his wish.

I

ACKNOWLEDJMENT

The process of this PhD research program is a long and complicated voyage. It is

also a learning adventure. I have had the good fortune of being the recipient of advice,

knowledge and effort from so many.

First and foremost, I am deeply grateful to my supervisor- Professor

Konstantinious Siriopoulos (Department of Business Administration, University of

Patras) - for the invaluable assistance, guidance, encouragement and patience that he

provided me during the preparation and execution of this research. His depth of insight,

understanding, involvement and dedicated supervision at every critical stage made

possible the completion of this study. Also, I would like to express my appreciation to the

members of the supervision committee.

I would like to thank The State Scholarship Foundation (I.K.Y) in Greece for the

financial support and the facility provided for the research program and allowing me the

time and freedom to pursue my research interests.

I would like also to express my appreciation to nine Arabian stock markets and

senior executives for their cooperation in providing me with the necessary data for the

research fieldwork.

Kharalambos Kokkales has kept me in touch with real life during the long years I

have been working on this thesis. He has patiently listened to all my troubles during our

long walks on Sundays. I am most grateful for him and his family, those who made me

one member of their family.

Especially, I desire to express my heartfelt obligation to my family, above all my

mother, for their wish and encouragement; and for their everlasting support and deep

understanding.

Finally, my appreciation is highly directed to the Great Greece (as I usually like to

name it) and Greek people for their hospitality and generosity which let me feel that I am

in my home land.

Bashar AbuZarour

II

Abstract

consequence prices should always reflect all available information. Here we consider market

efficiency for new emerging markets in the Middle East region. Emerging markets are typically

characterized by illiquidity, thin trading, and possibly non-linearity in returns generating process.

Firstly, we adjust observed daily indices for nine Arab stock markets for infrequent trading, while

the logistic map has been used to determine whether non-linearity exists in returns generating

process. Next we used several econometric models to test for market efficiency. The results of

runs test, variance ratio, serial correlation, BDS, and regression analysis indicate that we can

reject the hypothesis that lagged price information cannot predict future prices. In other words,

prices do not follow random walk properties; even after correction for thin trading.

We next analyze volatility structure using GARCH models. The results of GARCH

(1,1) model indicate that volatility clustering still seems to characterize some markets. While in

three markets (Egypt, Kuwait, and Palestine) volatility seems to be persistent. Moreover, the

results of EGARCH (1,1) model show that four markets (Bahrain, Dubai, Kuwait, and Oman)

exhibit signs of leverage effect and asymmetric shocks to volatility. Compared with other

emerging and international markets; Arab stock markets display relatively low rate of excessive

volatility as indicated by Schwert model. Furthermore, the dependence in the second moment

found to be quite enough to characterize the non-linear structure in the time series. Finally, we

find that seasonality and calendar effects exist in Arab markets with three forms; day-of-the-week

effect, month-of-the-year effect and the Halloween indicator. We conclude that Arab stock

markets under examination are not efficient in the week form sense of efficient market

hypothesis.

There is a large body of empirical evidence that financial markets become highly

integrated. According to modern portfolio theory, gains from international portfolio

diversification are related inversely to the correlation of equity returns. The results of multivariate

cointegration techniques, structural vector autoregression (SVAR) and vector autoregression

(VAR) models indicate that, there is no cointegrating relation between Arab and international

stock markets. The results of SVAR show that the linkage between international and Arab

markets is very weak. Next we investigate the dynamic relationships among Arab markets them

selves, and how do other factors; such as oil prices, affect the performance of these markets

especially for Gulf Cooperation Council (GCC) stock markets. To do that, Arab markets have

been divided into two sub-groups: oil production countries (GCC countries) and non-oil

production countries (Jordan, Egypt, and Palestine). The results indicate the existence of long-run

relation between markets, however, the short-run linkages still very weak. Non-oil countries’

markets can offer diversification benefits for rich GCC investors. Moreover, oil prices found to

have a significant effect on GCC markets and dominate the long-run equilibrium. Oil prices play

a significant role in affecting GCC markets’ volatility. While after the raise in oil prices;

especially during the last two years, linkages between oil prices and GCC markets increased. Four

GCC markets have predictive power on oil prices, with two markets to be predicted by oil prices.

We conclude that Arab stock markets can offer diversification potentials for regional and

international investors. Oil prices have a significant effect on GCC markets.

Finally, we suggest a strategic plan to improve these markets based on two main broad

goals, improving market efficiency and increasing market liberalization. To achieve these goals

we identify specific targets and strategies that could be realized through tactical programs and

activities.

III

Contents

1- INTRODUCTION .................................................................................................................................... 1

1-1 THE MOTIVATION OF THIS STUDY ...................................................................................................... 1

1-2 MARKETS UNDER EXAMINATION........................................................................................................ 5

1-3 INTRODUCTIONS TO CHAPTER 4 ......................................................................................................... 7

1-4 INTRODUCTION TO CHAPTER 5 ..........................................................................................................10

1-5 CONTRIBUTION...................................................................................................................................11

2-1 DEFINITION ........................................................................................................................................15

2-2 SPECIFICATION OF THE INFORMATION SET ......................................................................................16

2-3 EMH AND ASSET PRICING MODELS ...................................................................................................18

2-3-1 Single security test .....................................................................................................................18

2-3-1-1 Expected returns or fair game......................................................................................................... 19

2-3-1-2 The submartingale model ................................................................................................................ 21

2-3-1-3 The Random Walk model ................................................................................................................ 21

2-3-2 Multiple security expected return models .................................................................................22

2-3-2-1 The Sharp-Lintner-Black model (SLB model) ............................................................................... 22

2-3-2-2 Market model ................................................................................................................................... 24

2-3-2-3 Multifactor models ........................................................................................................................... 25

2-3-2-4 Consumption-based Asset-Pricing models ..................................................................................... 26

2-4 EMH ITS ORIGINS AND EVIDENCE .....................................................................................................27

2-4-1 The origin of EMH ....................................................................................................................27

2-4-2 Weak-form efficiency (returns predictability) ..........................................................................28

2-4-2-1 Random Walk Hypothesis (RWH) .................................................................................................. 29

2-4-2-2 Return predictability........................................................................................................................ 33

2-4-3 Semi-strong-form of efficiency (event studies) .........................................................................37

2-4-4 Strong-form-efficiency (private information) ...........................................................................39

2-5 EVIDENCE AGAINST EMH AND ALTERNATIVE MODELS FOR MARKET BEHAVIOR ..........................42

2-5-1 Market anomalies ......................................................................................................................43

2-5-1-1 Long-term return anomalies ........................................................................................................... 43

2-5-1-1-1 Overreaction and underreaction ............................................................................................ 44

2-5-1-1-2 IPOs and SEOs......................................................................................................................... 45

2-5-1-1-3 Mergers..................................................................................................................................... 46

2-5-1-1-4 Stock splits ................................................................................................................................ 47

2-5-1-1-5 Self-tenders and share repurchases ........................................................................................ 47

2-5-1-1-6 Exchange listings...................................................................................................................... 48

2-5-1-1-7 Dividend initiations and omissions ......................................................................................... 49

2-5-1-1-8 Spinoffs ..................................................................................................................................... 49

2-5-1-1-9 Proxy contests .......................................................................................................................... 50

2-5-1-2 Calendar effects ................................................................................................................................ 51

2-5-1-2-1 January effect ........................................................................................................................... 51

2-5-1-2-2 The weekend effect (Monday effect) ....................................................................................... 53

2-5-1-2-3 Holidays’ effects ....................................................................................................................... 54

2-5-1-2-4 Turn of the month effect ......................................................................................................... 54

2-5-1-2-5 The Halloween effect ............................................................................................................... 55

2-5-1-3 Other anomalies ............................................................................................................................... 55

2-5-1-3-1 Small firm effect ....................................................................................................................... 56

2-5-1-3-2 Value-Line enigma ................................................................................................................... 56

2-5-1-3-3 Standard and Poor’s (S&P) Index effect ............................................................................... 56

2-5-1-3-4 The weather .............................................................................................................................. 56

2-5-2 Volatility tests, fads, noise trading ............................................................................................57

IV

2-5-3 Models of human behavior .......................................................................................................59

2-6 EVIDENCE FROM EMERGING MARKETS ............................................................................................61

3-1 ARAB STOCK MARKETS AND MARKET EFFICIENCY ..........................................................................64

3-2 THE FOUNDATION OF ARAB STOCK MARKETS .................................................................................69

3-3 ECONOMIC REFORMS AND DEVELOPMENT OF ARAB CAPITAL MARKETS........................................82

3-4 THE PERFORMANCE OF ARAB STOCK MARKETS ..............................................................................89

3-4-1 Market size .................................................................................................................................90

3-4-2 Market liquidity .........................................................................................................................93

3-4-3 Financial Valuation of Arab Stock Markets ............................................................................95

3-4-4 Market concentration ................................................................................................................96

3-5 DATA DESCRIPTION ...........................................................................................................................96

4- TESTING THE EFFICIENT MARKET HYPOTHESIS FOR ARAB STOCK MARKETS ........102

4-1 RANDOM WALK HYPOTHESIS (RWH)............................................................................................102

4-1-1 Estimating the true index-correcting for infrequent trading .................................................103

4-1-2 Regression analysis .................................................................................................................105

4-1-3 Serial correlation (autocorrelation) of the return series ........................................................106

4-1-4 Non-parametric runs test ........................................................................................................108

4-1-5 Variance ratio test ...................................................................................................................109

4-1-6 BDS test for returns independency .........................................................................................111

4-2 THE VOLATILITY OF ARAB STOCK MARKETS’ RETURNS ................................................................112

4-2-1 Generalized autoregressive conditional heteroskedasticity (GARCH) ..................................113

4-2-2 Exponential generalized autoregressive conditional heteroskedasticity (EGARCH)............116

4-2-3 Schewrt model..........................................................................................................................117

4-3 NON-LINEARITY AND CHAOS IN STOCK RETURNS ...........................................................................117

4-4 SEASONALITY AND CALENDAR EFFECTS .........................................................................................121

4-4-1 Day-of-the-week effect.............................................................................................................121

4-4-2 January effect or month-of –the-year effect ...........................................................................122

4-4-3 The Halloween effect...............................................................................................................123

4-5 EMPIRICAL RESULTS........................................................................................................................124

4-5-1 Random walk properties ..........................................................................................................124

4-5-2 Volatility of returns..................................................................................................................135

4-5-3 Non-linearity in stock returns .................................................................................................141

4-5-4 Calendar effects .......................................................................................................................145

4-6 SUMMARY ........................................................................................................................................154

INTERNATIONAL STOCK MARKETS ...............................................................................................158

5-1 INTERNATIONAL INTEGRATION OF ARAB STOCK MARKETS ..........................................................160

5-1-1 Unit root test ............................................................................................................................160

5-1-2 Multivariate cointegration.......................................................................................................163

5-1-3 Structural VAR (SVAR) ..........................................................................................................167

5-2 TRANSMISSION OF STOCK PRICES MOVEMENTS BETWEEN ARAB STOCK MARKETS .....................174

5-2-1 Granger causality ....................................................................................................................175

5-2-2 Vector autoregression (VAR) ..................................................................................................176

5-3 EMPIRICAL RESULTS........................................................................................................................178

V

5-3-1 Integration ...............................................................................................................................179

5-3-2 Long-run relationship (cointegration test) .............................................................................180

5-3-3 Short-run relationship between Arab and international stock markets ................................182

5-3-4 Short-run relationships among Arab stock markets ..............................................................184

5-3-4-1 Granger causality test .................................................................................................................... 185

5-3-4-2 Causality and error-correction models (VEC) ............................................................................. 186

5-3-5 Dynamic relationship between GCC stock markets and oil prices.........................................191

5-3-5-1 Oil prices and GCC markets volatility ......................................................................................... 192

5-3-5-2 Long-run relationship among GCC stock markets and oil prices .............................................. 194

5-3-5-3 The rise of oil prices and GCC stock markets.............................................................................. 196

5-4 SUMMARY .........................................................................................................................................203

6-1 ENVIRONMENT ANALYSIS ................................................................................................................207

6-1-1 Demand and supply of financial papers .................................................................................207

6-1-2 Market microstructure.............................................................................................................209

6-1-3 Liberalization and markets integration...................................................................................209

6-1-4 Privatization .............................................................................................................................210

6-1-5 Legal and regulatory environment..........................................................................................212

6-2 STRENGTH, WEAKNESS, OPPORTUNITIES, AND THREATS (SWOT) ANALYSIS .............................214

6-3 VISION AND STRATEGIC GOALS .......................................................................................................218

7- CONCLUSIONS ...................................................................................................................................225

REFERENCES ..........................................................................................................................................229

APPENDIXES ...........................................................................................................................................246

VI

List of Tables

Table 3-1: Some Economic Indicators, Egypt…………………………………………………….72

Table 3-2: Some Economic Indicators, Jordan…………………………………………………...73

Table 3-3: Some Economic Indicators, Palestine…………………………………………………74

Table 3-4: Some Economic Indicators, Saudi Arabia……………………………………………76

Table 3-5: Some Economic Indicators, Kuwait…………………………………………………..77

Table 3-6: Some Economic Indicators, Oman……………………………………………………79

Table 3-7: Some Economic Indicators, UAE……………………………………………………..81

Table 3-8: Some Economic Indicators, Bahrain…………………………………………………82

Table 3-9: Market Structure for Arab Stock Markets…………………………………………..84

Table 3-10: Accessibility of Arab Stock Markets to Foreign Investments……………………...87

Table 3-11: Market Capitalization for Arab stock Markets…………………………………….91

Table 3-12: Total Number of Listed Companies, 2000-2005……………………………………92

Table 3-13: Market capitalization as Percentage of GDP……………………………………….92

Table 3-14: Total Value Traded to Market Capitalization (Turn Over Ratio)………………..94

Table 3-15: Average Daily Trading Value (million US$)………………………………………..94

Table 3-16: Total Value Traded as Percentage of GDP…………………………………………95

Table 3-17: Financial Valuation of Arab Stock Markets, End of 2005…………………………95

Table 3-18: Descriptive Statistics for Daily Market Returns for Arab Markets………………98

Table 4-1: Random Walk Model for Observed Indices…………………………………………124

Table 4-2: Random Walk Model for Observed Indices………………………………………...125

Table 4-3: Random Walk Model for Corrected Indices………………………………………..126

Table 4-4: Random Walk Model for Corrected Indices………………………………………..127

Table 4-5: Estimated Autocorrelations for Observed Indices Returns………………………..128

Table 4-6: Estimated Autocorrelations for Corrected Indices Returns……………………….128

Table 4-7: Results of Runs Test for Arab Stock Markets, Observed vs. Corrected Indicesc..130

Table 4-8: Variance Ratio Estimates and Heteroscedastic Test Statistics for Arab Stock

Markets…………………………………………………………………………………132

Table 4-9: BDS Test Results for Observed Return Indices……………………………………..133

Table 4-10: BDS Test Results for Adjusted Return Indices…………………………………….134

Table 4-11: Coefficient of Variation for Daily Returns for the Three Groups………………...135

Table 4-12: GARCH (1,1) Model for Daily Returns……………………………………………..138

Table 4-13: GARCH (1,1) Model for Weekly Returns…………………………………………...139

Table 4-14: EGARCH (1,1) Model for Daily Returns……………………………………………140

Table 4-15: Random Walk Models with Non-Linearity for Observed Indices…………………142

Table 4-16: Random Walk Models with Non-Linearity for Corrected Indices………………...143

Table 4-17: OLS Results for Day-of-the-Week Effect……………………………………………147

VII

Table 4-18: Chow Test for Structural Stability…………………………………………………..148

Table 4-19: Day-of-the Week Effect in the First Two Moments………………………………...148

Table 4-20: OLS Results for Month-of-the-Year Effect (January Effect)……………………...149

Table 4-21: Chow Test for Structural Stability…………………………………………………...151

Table 4-22: Month-of-the-Year Effect in the First Two Moments………………………………152

Table 4-23: The Halloween Indicator in Arab Stock Markets…………………………………..153

Table 4-24: The Halloween Indicator in Arab Stock Markets with January Effect

Adjustment……………………………………………………………………………154

Table 5-1: Unit Root Test for Each Individual Series, both in Levels and First

Differences………………………………………………………………………………..180

Table 5-2: Number of Cointegrating Relations for Four VARs Models…………………………181

Table 5-3: Johansen-Juselius Cointegration Test Results………………………………………...183

Table 5-4: Granger Causality Tests for Arab Stock Markets…………………………………….185

Table 5-5: Correlation Coefficient between Daily Arab Markets’ Returns……………………..186

Table 5-6: Cointegrating Equations of the VEC Models for VAR-9 and VAR-6……………….187

Table 5-7: Significant of Zero Restrictions on Coefficients of Cointegrating Equations of the

VEC Models of VAR-9 and VAR-6…………………………………………………….188

Table 5-8: VEC Model for 9 Arabian Indices in the VAR-9 Model……………………………...188

Table 5-9: VEC Model for 6 GCC Indices in the VAR-6…………………………………………190

Table 5-10: Weak Exogeneity Tests of the Endogenous Variables in the VEC Models of VAR-9

and VAR-6……………………………………………………………………………..190

Table 5-11: GARCH (1,1) Model for GCC Daily Returns with Oil Returns as a Regressor

in the Variance Equation……………………………………………………………...193

Table 5-12: Johansen-Juselius Cointegration Test Results………………………………………..194

Table 5-13: Cointegrating Equations of the VEC Model for VAR-7……………………………..194

Table 5-14: VEC Model for 6 GCC and Oil Price Indices in the VAR-7………………………....195

Table 5-15: Weak Exogeneity Tests of the Endogenous Variables in the VEC Model of

VAR-7…………………………………………………………………………………..196

Table 5-16: VAR System for GCC Stock Markets and Oil Returns for the First Sub-

Period…………………………………………………………………………………...198

Table 5-17: VAR System for GCC Stock Markets and Oil Returns for the Second Sub-

Period……………………………………………………………………………….......199

Table 5-18: Variance Decomposition for the Forecast Error of Daily Market Returns for

GCC Markets and Oil Returns during the First Sub-Period……………………….200

Table 5-19: Variance Decomposition for the Forecast Error of Daily Market Returns for

GCC Markets and Oil Returns during the Second Sub-Period…………………….201

Table 6-1: Strength, Weakness, Opportunities, and Threats for Arab Stock

Markets…………………………………………………………………………………...216

VIII

List of Figures

Figure 2-1: Information Set…………………………………………………………………………..17

Figure 3-1: Arab Stock Markets Performance Compared to other International Stock

Markets 9/2004-9/2005…………………………………………………………………89

Figure 3-2: Market Size for Arab Stock Markets between: 2000-2005…………………………...90

Figure 3-3: Relative Market Capitalization to all Markets 2005…………………………………..91

Figure 3-4: Market Liquidity Variables for Arab Stock Markets, 2000-2004…………………….93

Figure 3-5: Market Concentration, End of 2005……………………………………………………96

Figure 4-1: Markets Volatility (Schwert Model)…………………………………………………..136

Figure 4-2: Average Returns Among the Two Half-Year Periods……………………………….152

Figure 5-1: Cointegrating Relations VEC-9, VEC-6………………………………………………187

IX

1- Introduction

1-1 The motivation of this study

Investors require compensation for the postponement of current consumption as

they put their money into a stock market. A market in which prices always fully reflect

available information is called “efficient” (Fama, 1965, 1970). In an efficient market an

investor gets what he pays for and there are no profit opportunities available to

professional money managers or savvy investors. The market genuinely “knows best,”

and the prices of securities traded are equal to the values of the dividends which these

securities pay, also known as fundamental values.

However, one can ask whether hypothetical trading based on an explicitly

specified information set would earn superior returns. We would then need to specify an

information set first. Under weak-form efficiency the information set includes only the

history of prices or returns themselves. Under semi-strong-form efficiency the

information set includes all information known to all market participants, like the market

trading volume. Finally, strong-form efficiency means that the information set includes

all information known to any market participant, including private information (Campbell

et al., 1997).

By definition, in an efficient market the path of prices and the return per period

are unpredictable. Put more formally, the efficient market hypothesis (EMH) implies that

the expected value of tomorrow’s price Pt + 1, given all relevant information up to and

including today denoted as Ωt, should equal today’s price Pt, possibly up to a

deterministic growth component µ(drift). In other words, Et[Pt + 1| Ωt] = Pt + µ, where Et

denotes the mathematical expectation operator given the information at time t. In testing

the EMH the model commonly used is Pt = µ + Pt - 1 + et, where et ~ i.i.d (0, σ2), or returns

follow a random walk with drift ∆Pt = µ + et. For a long time these models were

maintained as an appropriate statistical model of stock market behavior.

The independence of increments {et} implies that the random walk is also a fair

game, but in a much stronger sense than the martingale. A martingale is a fair game, one

which is neither in your favor or your opponent’s, or a stochastic process {Pt}, which

satisfies the following condition: Et[Pt + 1|Pt,Pt - 1,...] = Pt or Et[Pt + 1- Pt|Pt,Pt - 1,...] = 0. In a

random walk, independence implies not only that increments are uncorrelated, but that

1

any nonlinear functions of the increments are also uncorrelated. Nevertheless, the

financial market literature recognizes several forms of the random walk hypothesis. First,

relaxing the assumption of identically distributed increments lets us allow unconditional

heteroskedasticity in the residuals, which is a useful feature given the empirically

observed fact of time-variation in the volatility of many financial asset return series. And

even more general version of the random walk hypothesis - the one most often tested in

the recent empirical literature - may be obtained by relaxing the independence

assumption of the model to include processes with dependent but uncorrelated

increments. Tests of random walk may thus be categorized as follows: tests of i.i.d.

increments in errors (runs tests), tests of independent increments without assuming

identical distributions over time (filter rules and technical analysis) and tests of

uncorrelated increments or testing the null hypothesis that autocorrelation coefficients of

the first differences of the level of the random walk at various lags are all zero.

The Efficient Market Hypothesis (EMH) has been the cornerstone of financial

research for more than thirty years. The first comprehensive study of the dependence in

stock prices can be attributed to Fama (1965) as he analyzed the daily returns of the 30

stocks that made up the Dow Jones Industrial average at the time of his study. He found

low levels of serial correlation in returns at short lags, and provided evidence concerning

the non-Gaussian nature of the empirical distribution of the daily returns. He gave two

explanations for these departures: the mixture of distributions and changing parameters

hypothesis. The next step in testing the EMH focused on explaining the empirical

observation that stock returns are negatively correlated in the long run. For example, the

presence of positive feedback traders, who buy (sell) when prices rise (fall), causes prices

to overreact to fundamentals. However, at some point in time prices start to revert back to

their fundamental values, hence we observe mean reversion in returns. This behavior runs

counter to the random walk hypothesis. As shocks are persistent in the case of a random

walk, this offers an alternative way to test the EMH (Cuthbertson, 1996).

Fama and French (1988) report that price movements for market portfolios of

common stocks tend to be at least partially offset over long horizons. They found

negative serial correlation in market returns over observational intervals of three to five

years. Nevertheless, evidence with respect to the presence of long-term dependence in

2

stock returns is still inconclusive (Poterba and Summers 1988; and Jegadeesh 1990). At

any rate, if the mean reversion hypothesis was rejected, researchers invalidated the asset

pricing models based on Brownian notion, random walk and martingale assumptions. We

now know many reasons why stock prices deviate from the random walk model. For

example, the variance in stock prices is typically not constant over time, since during

turbulent times the market reacts to the inflow of new information, beliefs are relatively

heterogeneous and volatility is high. During quiet times beliefs are more homogenous,

and much of the volatility comes from liquidity trading. This has led to the application of

(G)ARCH models in stock returns. Other types of deviation are calendar anomalies, like

the January effect, which had already been discovered in the stock market by Wachtel

(1942), among others.

Another important feature related to stock markets is market integration, and the

diversification benefits that a stock market could offer for portfolio investors. Capital

markets across countries or regions may exhibit varying degree of integration.

Theoretically, market linkages primarily stem from the “low of one price” that identical

assets (physical or financial) should bear the same price across countries after adjusting

for transaction costs. Rational (well-informed) investors would, or perhaps should,

arbitrage away price disparities, leading to more integrated markets.

Over the last 20 years, financial markets become highly integrated, mainly due to

reductions in the cost of information, improvements in trading systems’ technology and

the relaxation of legal restrictions on international capital flows. The changes have

accelerated the interaction among financial markets and the enlargements of capital

mobility. The body of empirical evidence suggests that significant capital market

integration exists among major industrialized countries, thus limiting the potential

benefits from international diversification (Meric and Meric 1989; Koutmos 1996;

Sinquefield 1996; Freimann 1998; Siriopoulos 1996; and Alexakis and Siriopoulos

1997). Moreover, gains from international portfolio diversification are related inversely

to the correlation of equity returns, according to modern portfolio theory. In line with this

theory, investors have become highly active, investing in foreign equity markets as a risk

diversification strategy.

3

Numerous studies have demonstrated the advantages of international

diversification related to low correlation between various equity markets (Eun and Resick

1984; Whealy 1988; and Meric et al. 2001). This tendency for the global markets to

become integrated is a result of the increasing tendency toward liberalization and

deregulation in the money and capital markets, both in developed and developing

countries as well as on a bilateral and multilateral basis, commencing from; for example,

trade liberalization and multilateral trade initiatives. Such liberalization is important to

introduce structural reforms, to promote economic efficiency, to estimate trade and

investment, and to create a necessary climate for promoting sustainable economic growth

with a commitment to market-based reforms.

Furthermore, long-run linkages between stock markets have important regional

and global implications at the macro-level, as a domestic capital market cannot be

insulated adequately from external shocks, thus the scope for independent monetary

policy may become limited. It is argued in Errunza et al. (1999) that the use of return

correlations at the market index level to infer gains from international diversification,

involving foreign-traded assets overstates the potential benefits. The gains must be

measured beyond those attainable through home-made diversification by mimicking

returns on foreign market indices with domestically traded securities.

Are Arab stock markets efficient in the weak-form sense of Efficient

Market Hypothesis?

Is the view of predictability in stock returns (if there is) related to whether

we think that these time series are non-linear? How does thin trading

affect the predictability of these time series?

Are Arab stock markets characterized by excessive volatility of returns,

relative to other emerging and international stock markets?

Having answered these questions, then we ask whether this evidence suggests that

markets are efficient or not. This is the substance of chapter 4. We then address the

following new issues:

4

Are Arab stock markets integrated among themselves and with

international stock markets? If yes, how do shocks generated by

international stock markets especially UK, US, and Japan affect Arab

stock markets?

Can Arab stock markets offer, both regional and international investors

unique risk and returns characteristics to diversify international and

regional portfolios?

What is the effect of oil prices on the performance of Gulf Cooperation

Council (GCC) stock markets? And whether these markets have predictive

power on oil prices or vice versa?

These issues are addressed in chapter 5. The finding should suggest whether Arab

stock markets can offer diversification potentials for both international and regional

portfolio investors. Moreover, the results should suggest how other factors; such as oil

prices, affect stock markets, especially in GCC countries where oil prices play a crucial

role in their economies.

In this research, nine Arabian stock markets will be examined; these are Abu

Dhabi, Bahrain, Dubai, Egypt, Jordan, Kuwait, Oman, Palestine, and Saudi Arabia. Little

is known about these markets, by international standards, these markets are considered as

emerging markets and relatively new. Most of them started operating over the last two

decades, while Egyptian stock market; in particular, have been in existence for much

longer, but until recently its level of activity was not significant.

Moreover, six of these markets are from rich oil GCC countries (Abu Dhabi,

Bahrain, Dubai, Kuwait, Oman, and Saudi Arabia). Except Bahrain and Oman, the other

four countries are members in the Organization of Petroleum Exporting Countries

(OPEC). At the end of 2003, GCC countries collectively accounted for about 21 percent

of the world’s 68 million barrels a day of total production, they possess 43 percent of the

world’s 1105 billion barrels of oil proven reserve, given that one of these countries

(Saudi Arabia) is the largest oil producer and reserves in the world.

5

There are significant differences between Arab stock markets’ characteristics, in

terms of market indicators, such as; number of listed companies, market capitalization,

and accessibility to foreign investors. But in general, there are dominant features for these

markets preventing their development, prominent among the hurdles were: deficiencies in

the legal framework governing these markets, the small number of listed companies, the

undiversified investment instruments, market illiquidity, narrowness and the lack of

market depth, highly concentrated markets, the absence of investors awareness in general,

and in many cases the lack of economic stability.

In the recent years, most Arabian countries witnessed considerable steps aiming to

improve their local stock markets. a number of Arab stock markets have been proceeded

to separate between the supervisory and executive roles. Moreover, Arab countries can be

divided into two groups regarding accessibility to direct foreign investment in stock

markets, the first includes countries which do not impose any restrictions on foreign

investments in financial papers; these are Egypt, Palestine, and Jordan. While the second

group contains countries where such restrictions exist in varying degrees; these are the

member states of GCC.

The focus of this study is not on the test of market efficiency as such, but also on

whether Arab stock markets are integrated with international and regional stock markets,

and therefore; to what degree these markets can offer diversification potentials for

regional and international portfolios investors. Moreover, GCC markets are among the

markets under examination here, and it is known that oil plays a significant role in these

economies. The study will investigate the effect that oil prices could have on the

performance of GCC stock markets; especially after the raise in oil prices during the last

two years.

The reminder of this study will be as follow, we start chapter 2 with the literature

review of the main work related to EMH. Starting with definition for the EMH, then

identifying the information set, following with a discussion of asset pricing models and

its relation with EMH. This is important since empirical tests of market efficiency –

especially those that examine asset price returns over extended period of time- are

necessarily joint test of market efficiency and particular asset-pricing model. when the

joint hypothesis is rejected, as it often, it is logically possible that this is a consequence of

6

deficiencies in the particular asset-price model rather than in the efficient market

hypothesis, the bad model problem (Fama 1991). Consequently chapter 2 proceeds in

presenting the original and evidence for EMH, then it presents evidence against EMH

such as volatility and anomalies, including long-term anomalies and calendar effects, and

alternative models of human behavior. Finally, evidence from emerging markets is

provided.

Chapter 3 presents Arab stock markets under examination, started with a survey

of existing literature related to these markets, then a brief economic indicators of these

countries with a brief history for each market are provided. Markets’ characteristics and

main performance indicators are presented. Finally we present data description and main

statistics.

Chapter 4 is projected to examine market efficiency in Arab stock markets while

chapter 5 presents the diversification potentials that Arab stock markets could offer for

international investors, and the effect of oil prices on GCC stock markets, while an

introduction for each of chapter 4 and 5 is coming later. The study continues with chapter

6, discussing the implication of the obtained empirical results. According to this

discussion and analysis of the surrounding environment, based on two main broad

strategic goals, a strategic plan has been built to improve the performance of Arab stock

markets. Chapter 7 concludes the thesis.

Are Arab stock markets efficient? To answer this question we start chapter 4 by

testing market efficiency using most recent econometric techniques. However, the

conventional tests of market efficiency have been developed for testing markets which

are characterized by high level of liquidity, sophisticated investors with access to high

quality and reliable information and few institutional impediments. On the other hand,

emerging markets are typically characterized by illiquidity, thin trading, and possibly less

well informed investors with access to unreliable information and considerable volatility.

Moreover, efficiency implicitly assumes investors are rational; such rationality leads

prices responding linearly to new information. However, emerging markets; especially

during the early years of trading, may be characterized by investors who may not always

7

display risk aversion. For example, loss adverse investors; who have incurred losses, may

display risk loving behavior in an attempt to recover such losses. Such examples of

investors’ behavior may result in prices responding to information in a non-linear

fashion. So, it is important to take into account the institutional features of these markets

when testing for market efficiency.

As a result, firstly we adjust the observed indices for infrequent trading, using

Miller, Muthuswamy and Whaley (1994) approach. The procedures used to test for EMH

and random walk properties, were chosen on the basis of the implications of EMH. If all

relevant and available information is fully reflected in stock prices, then (a) successive

price changes will be independent, so that there will be no serial correlation over time

between returns. (b) Successive price changes will be identically distributed

log (Pt) = log (Pt-1) + εt (1-1)

where εt is an independent standard variable, that is; a series of identically distributed

random variables with zero mean and variance equal to unity. To test for the

independence of successive price changes (condition a), we employ runs test and serial

autocorrelation. Further, to test whether successive price changes are identically

distributed (condition b), we use regression analysis, variance ratio, and BDS tests. All

these tests were implemented for observed indices and for indices after have been

corrected for infrequent trading.

We proceed by testing returns’ volatility relative to other developed and emerging

markets. For this purpose, three emerging markets (India, Turkey and Israel) and three

developed markets (US, UK, and Japan) have been used to compare relative volatility

with Arab stock markets. It is well reported empirical fact that the (G)ARCH property is

found in examining stock returns. Schwert (1989), among others; examined how far the

conditional volatility in stock returns depends on its own past volatility as well as on the

volatility in other economic variables (fundamentals), such as bonds and the real out put.

Later, Hamilton and Lin (1996) claimed that recessions are the primary factors that drive

fluctuations in the stock returns’ volatility. Furthermore, asset markets are typically

characterized by periods of turbulence and tranquility. That is to say, large (small)

forecast errors tend to be followed by large (small) errors. Hence, the variance of the

8

forecast errors is often persistent, and the duration of market volatility may shed

additional light on the market efficiency issue.

The basic idea behind autoregressive conditional heteroskedasticity ARCH

models proposed by Engle (1982) is that, the second moments of the distribution may

have an autoregressive structure. Under rational expectations the forecast error is ut+1 =

yt+1-Et(yt+1), and the conditional distribution of yt+1 is assumed to be normal with mean

µt+1 and var(yt+1/Ωt) = ht+1 = a0+a1 u2t, where Ωt is the information set available at time t.

However, the ARCH process has a memory of only one period. To generalized this we

can start adding lags of ut-1 in the equation ht+1, ι = 1,…,q. but then the number of

parameters to estimate increases rabidly (Bollerslev 1986). For example, in the GARCH

(1,1) model the conditional variance depends on lagged variance terms: ht+1 = a0+a1+β1ht

= a0+(a1+ β1)ht+at(ut2-ht) in addition to the lagged ut where u0 is arbitrarily assumed to

be fixed and equal to zero. The parameters can be estimated by maximum likelihood

techniques. Conditional on time t information Ωt, (ut2-ht) has a mean of zero, and can be

thought as the shock to volatility. The coefficient a1 measures the extent to which a

volatility shock today feeds through into the volatility of the next period, while a1+ β1

measures the rate at which this effect dies out over time.

Since Engle’s seminal work, many generalization of this model have been

reported. For example, the GARCH (1,1) with a1+ β1=1 has a unit autoregressive root, so

that today’s volatility affects forecasts of volatility in to the indefinite future (persistent of

volatility), this is therefore known as the integrated GARCH or IGARCH model. Nelson

(1991) introduced the Exponential GARCH (EGARCH) model which allows for

asymmetric shocks to volatility and tests the leverage effect. The dependence of the

second moment in returns captured by the (G)ARCH process is known as volatility

clustering, i.e. large changes in price volatility are followed by large changes in either

sign.

Chapter 4 continued by using the logistic map to detect any non-linearity in

returns’ generating process. However, the logistic map is not able to determine the

precise nature of any non-linearity, but rather to ascertain whether non-linearity exists. It

is appropriate that non-linearity generated by dependence in the second moment. To

disentangle the non-linearity generated by changes in volatility from non-linearity arising

9

as a result of other causes, the standardized residuals of GARCH models will be

subjected to several diagnostic tests to “see what left” and whether non-linearity

generated by this form of dependence in the second moment or from other causes.

Finally, chapter 4 concluded by using an alternative approach for testing EMH

through testing for seasonality or calendar effects in stock returns. Three calendar effects

have been tested the-day-of-the-week effect, Month-of-the-year effect, and the

Halloween indicator.

The main purpose of chapter 5 is to investigate the diversification potentials that

Arab stock markets may offer for international investors, through examining whether

Arab stock markets are integrated with international and regional stock markets. The

analysis has been undertaken with several directions. Firstly, we start by examining the

long-run relationships between Arab stock markets and international markets, which

represented by the US market (S&P 500). The analysis depends on multivariate

cointegration techniques proposed by Johansen (1991, 1995a), which based on the

autoregressive representation discussed in Johansen (1988). However, a prerequisite for

cointegration is that, non-stationary series are integrated of the same order. Therefore, the

first step is to determine the order of integration for each variable. To test for unit root,

the augmented Dickey-Fuller, the Phillips-Perron, and the Kwaitkowski-Phillips-

Schmidt-Shin (KPSS) tests (Dickey and Fuller, 1979; Phillips and Perron, 1988;

Kwaitkowski et al., 1992) have been used. The results of the multivariate cointegration

indicate that Arab stock markets are not integrated with international markets in the long-

run.

Next, we continue to investigate the short-run relationship between Arab and

international markets. More specifically, how do Arab markets react to shocks generated

by international markets (US, UK, and Japan)? Using structural vector autoregression

(SVAR) model and analyzing the impulse response functions. The model incorporate the

assumption that the returns on each of the three international markets, affect the returns

on Arab markets but NOT vice versa. A block recursive model, similar to the SVAR

model used by Zha (1999), Cushman and Zha (1997), and Berument and Ince (2005) has

10

been used to examine the effect of a large economy’s stock exchange movements (UK,

US, and Japan) on a small economy’s stock exchange movements (each of Arab

markets).

The next step in chapter 5 is to investigate the dynamic relationships between

Arab stock markets both on the long- and short-runs, using multivariate cointegration and

Granger causality. The total markets have been divided into two groups, oil (GCC

markets) and non-oil production countries (Jordan, Egypt, and Palestine). Chapter 5

proceeds by investigating the effect of oil prices on GCC stock markets especially during

the last two years, which witnessed huge rise in oil prices. Multivariate cointegration and

vector autoregression models were used. Moreover, oil returns have been added as an

additional regressor in the variance equation of the GARCH model, to trace the effect of

oil prices on the volatility of returns.

1-5 contribution

This section concludes by presenting the main empirical results of this thesis. The

author asked, among other things, whether Arab stock markets are efficient in the week

form sense, using daily prices for the general indices. This study concentrates on nine

new Arabian emerging markets in the Middle East region. Little is known about these

markets since most of them are new established, while for some of them (i.e. Palestine

stock exchange) this is the first empirical work examining these markets.

The first task of this research was to investigate market efficiency. For new

emerging markets, the outcomes of tests of EMH are important in assessing public policy

issues such as the desirability of mergers and takeovers. The EMH test results are also

useful for derivative market participants, whose success precariously depends on their

ability to forecast price movements, they are also important for international portfolio

investors who are looking for diversification benefits in emerging markets. Moreover,

they facilitate the important role of the stock market in efficient capital allocation. It is

important when testing market efficiency in an emerging market, to take into account the

specific features that characterize new emerging markets; such as, thin trading, non-

linearity in returns generating process, and excessive volatility. Using most recent

econometric techniques, the results indicate that returns in most Arab markets are

11

predictable. While volatility clustering still seems to characterize some markets, volatility

seems to be persistent in three markets (Egypt, Kuwait, and Palestine), other markets

(Bahrain, Dubai, Kuwait, and Oman) show signs of leverage effect and asymmetric

shocks to volatility. Moreover, return generating process found to be non-linear in these

markets, dependent in the second moment explains enough the existing non-linearity.

Furthermore, the results indicate that some kinds of anomalies exist in Arab stock

markets, we conclude

That the empirical evidence enables us to declare that Arab stock markets

are not efficient in the weak-form sense even after correction for

infrequent trading. While the dependence in the second moment found to

be enough to explain the non-linearity in return generating process.

These results provide new evidence to the existing literature for other emerging

markets. Since many evidence of predictability in emerging markets have been found and

reject the hypothesis that lagged price information cannot predict future prices (Bakaret

1995; Harvey 1995b, 1995c; Claessense et al. 1995; Buckbery 1995; Haque et al. 2001,

2004; and Bailey et al. 1990) among others.

Long-term investors are often advised to invest part of their money in stocks from

emerging markets, because developing markets are growing much faster than

industrialized countries, and less integrated with international stock markets. However,

over the last 20 years, financial markets become highly integrated; the tendency for the

global markets to become more integrated is a result of the increasing tendency toward

liberalization and deregulation in the money and capital markets. On the other hand and

according to modern portfolio theory, gains from international portfolio diversification

are related inversely to the correlation of equity markets. The integration between Arab

and international markets has been investigated, using multivariate cointegration

techniques, while structural vector autoregression (SVAR) has been employed to test the

respond of each Arabian market to shocks originated in international markets, we

conclude that

12

Arab stock markets can offer diversification potentials for both

international and regional portfolio investors, these markets found to be

segmented from international markets. In the short-term, the linkages

found to be weak in general, while UK market found to have the most

influence on Arab markets. Moreover, linkages among Arab markets still

very weak despite the existing long-term cointegration between them,

while non-oil countries’ markets can offer diversification benefits for rich

GCC investors.

These results are in line with the numerous studies that have demonstrated the

advantage of international diversification related to low correlation between various

equity markets, such as (Eun and Resick 1984; Wheatly 1988; Meric and Meric 1989;

Baily and Stulz 1990; Divecha et al. 1992; Michaud et al. 1996; Siriopoulos 1996;

Alexakis and Siriopoulos 1997; Meric et al. 2001; and Bulter and Joaquin 2002).

Gulf Cooperation Council (GCC) countries are among the most important oil

producing countries and a main player in the Organization of Petroleum Exporting

Countries (OPEC). Producing and exporting oil play a crucial role in determining foreign

earnings and governments’ budget revenues and expenditures for such countries, which

in tern affect all aspects of daily economic life. In addition, increase in oil prices has a

significant effect on local stock markets according to cash surplus. This in turn, shows the

importance of studying the relation between oil prices and stock markets in GCC

countries, especially after the huge increase in oil prices during the last two years. the

empirical results indicate that

Oil prices dominate the long-run equilibrium with GCC stock markets,

and have a significant effect in determining returns’ volatility in these

markets. Furthermore, after the raise in oil prices; the linkages between

oil prices and GCC markets increased, four GCC markets have predictive

power on oil prices, with only two markets to be predicted by oil prices.

13

These results provide new evidence to the existing literature. Few studies have

looked into the relation between oil spot/future prices and stock markets, which mainly

concentrates on Canada, Germany, Japan, UK, and USA (see Johnes and Kaul 1996;

Huang et al. 1996; Sadorsky 1999; Papapetrou 2001; and Hammoudeh and Aleisa 2002,

2004).

14

2- Efficient Market Hypothesis (EMH)

2-1 Definition

When the term “efficient market” was introduced into the economic literature

thirty years ago, it was defined as a market which “adjusts rapidly to new information”

(Fama 1969). It soon became clear, however, that which rapid adjustment to new

information is an important element of an efficient market; it is not the only one. A more

modern definition is that asset prices in an efficient market ‘fully reflect all available

information ‘(Fama 1991). This implies that the market processes information rationally,

in the sense that relevant information is not ignored, and systematic errors are not made.

As a consequence, prices are always at levels consistent with ‘fundamentals’.

The words in this definition have been chosen carefully, but they nonetheless

mask some of the subtleties inherent in defining an efficient asset market. For one thing,

this is a strong version of the hypothesis that could only be literally true if “all available

information” was costless to obtain. If information was instead costly, there must be a

financial incentive to obtain it. But there would not be a financial incentive if the

information was already “fully reflect” in asset prices (Grossman and Stiglitz 1980). A

weaker, but economically more realistic, version of the hypothesis is therefore that prices

reflect information up to the point where the marginal benefits of acting on the

information (the expected profits to be made) do not exceed the marginal costs of

collecting it (Jensen 1978).

Secondly, we must have a model to provide a link from economic fundamentals to

asset prices. While there are candidate models in all asset markets that provide this link,

no-one is confident that these models fully capture the link in any empirically convincing

way. This is important since empirical tests of market efficiency-especially those that

examine asset price returns over extended periods of time-are necessarily joint test of

market efficiency and particular asset-pricing model. When the joint hypothesis is

rejected, as it often is, it is logically possible that this is a consequence of deficiencies in

the particular asset-price model rather than in the efficient market hypothesis. This is the

“bad model” problem (Fama 1991).

Finally, a comment about the word “efficient” It appears that the term was

originally chosen partly because it provides a link with the broader economic concept of

15

efficiency in resource allocation. Since there are three types of efficiency: (i) pricing

efficiency which refers to the notion that prices reflect rabidly in an unbiased way all

available information, (ii) operational efficiency: refers to the level of costs carrying out

transactions in capital markets, and (iii) allocational efficiency: refers to the extent to

which capital market is allocated to the most profitable enterprises (this should be a

product of pricing efficiency). When we refer to EMH, our concentration will be on

pricing efficiency. Thus Fama began his 1970 review of the efficient market hypothesis

(specially applied to the stock market):

“The primary role of the capital market is allocation of ownership of the economy’s

capital stock. In general terms, the ideal is a market in which prices provide accurate

signals for resource allocation: that is, a market in which firms can make production-

investment decisions, and investors can choose among the securities that represent

ownership of firms’ activities under the assumption that securities prices at any time

‘fully reflect’ all available information “(Fama, 1970, p. 383)

The link between an asset market that efficiently reflects available information (at

least to the point consistent with the cost of collecting the information) and its role in

efficient resource allocation may seem natural enough. An informationally efficient asset

market need not generate allocative or production efficiency in the economy more

generally. The two concepts are distinct for reasons to do with the completeness of

markets and the information-revealing role of prices when information is costly and

therefore valuable (Stiglitz 1981).

It follows that the next intellectual stage is naturally the explanation of the content

of the information set. In Fama’s statement, the information which should be reflected in

the price is presented as being “available” and ‘relevant”. How can this availability and

this relevance be characterized? Robert (1967) distinguishes between three kinds, or

levels, of information corresponding to three forms of informational efficiencies:

1. Weak-form efficiency: the information set includes only the history of prices or

returns them selves.

16

2. Semi-strong form efficiency: the information set includes all information known

to all participants (publicly available information).

3. Strong-form efficiency: the information set includes all information known to

any market participant (private information).

The categorization of the tests into weak, semi-strong, and strong form, will serve the

useful purpose of allowing us to pinpoint the level of information at which the hypothesis

breaks down (Fama, 1991). The information set corresponding to the weak-form of

efficiency is composed by all past quoted prices of the market and only these, the weak-

form of the efficiency rules out the use of technical analysis, which appears as non-

efficient for obtaining profits higher than those of the market it self. Here again we

encounter the source of conflict with the technical analysts. The information set

corresponding to the semi-strong form of efficiency consists of the preceding set of past

prices, augmented by financial data of the firms. The semi-strong form of efficiency rules

out classic financial analysis, to obtain profits higher than those of the market. This is the

source of the conflict with the financial analysts and the economics research departments

of banks. Finally, the strong form of efficiency, which includes the two preceding sets of

information, is concerned with the existence of private information, i.e. not necessarily

public, for example the forecasts to which the professional pension funds managers have

access, unavailable to the general public. With a very strong form of efficient market, no

professional managers, even provided with high skill, can obtain profits higher than those

of the market on a long-time basis. This is the source of conflict with active managers of

portfolios, who spend a large part of their time looking for and choosing stocks which

they think will be profitable, and conjecturing as to the future development of the market.

Figure 2-1: Information Set.

17

Fama (1991) renames these three categories. Instead of weak-form tests, he

suggests tests for return predictability, since this category now covers more general

area of return predictability. For the second and third categories, he proposes changes in

title, not coverage. Instead of semi-strong form tests of the adjustment of prices to

public announcements, he uses event studies. Instead of strong-form tests of whether

specific investors have information not in market prices, he suggests the more

descriptive title, tests for private information. This split of the information set into three

distinct categories, leads to definition of three fields of investigation into the concept of

efficiency, and then to test the efficiency. In the three cases, the gap between the return

on the portfolio and that of the market must be random variable of zero expectation.

The concept of efficiency thus becomes defined by the nature of the chosen information

as “available and relevant.”

In most cases tests of market efficiency, are tests of a joint hypothesis (the joint

hypothesis problem, Fama 1991). Thus, market efficiency per se is not testable. It must

be tested with some model of equilibrium, an asset pricing model. One can say that

efficiency must be tested conditional on an asset pricing model or that asset pricing

models are tested conditional on efficiency. The point is that such tests are always joint

evidence on efficiency and an asset-pricing model. In this section we will discuss asset-

pricing models which can be divided into two main categories: single security test and

multiple security test.

That is the price or return histories of individual securities are examined for

evidence of dependence that might be used as the basis of a trading system for that

security. This group includes: Fair game models, Martingle model, and Random Walk

Model (RWM).

18

2-3-1-1 Expected returns or fair game

The definitional statement that in an efficient market, prices “fully reflect” available

information is so general that it has no empirically testable implications. The process of

price formation must be specified in more detail. In essence we must define somewhat

more exactly what is meant by the term “fully reflect”. The conditions of market

equilibrium can be stated in terms of expected returns. All members of the class of such

“expected return theories” can, however, be described notationally as follows:

E ( ~p j ,t + 1 / Φ t ) = [1 + E (~r j ,t + 1 / Φ t )]p jt

(2-1)

Where E is the expected value operator; Pjt is the price of security j at time t; Pj,t+1 is its

price at t+1; rj,t+1 is the one period percentage return (Pj,t+1-Pjt)/Pjt; Φt is a general

symbol for whatever set of information is assumed to be “fully reflect” in the price at t;

and tildes indicate that Pj,t+1 and rj,t+1 are random variables at t. A “fair game” model as

summarized in (2-1) has properties which are implications of the assumptions that (i)

the conditions of market equilibrium can be stated in terms of expected returns, and (ii)

the information Φt if fully utilized by the market in forming equilibrium expected

returns and thus current prices. The role of “fair game” models in theory of efficient

markets was first recognized and studied by Mandelbrot (1966) and Samuelson

(1965).The value of the equilibrium expected return E (~

r / Φ ) projected on the basis

j , t +1 t

hand. The conditional expectation notation of (2-1) is meant to imply, however, that

what ever expected return model is assumed to apply, the information in Φt is fully

utilized in determining equilibrium expected returns. And this is the sense in which Φt

is “fully reflected” in the formation of the price Pjt.

The assumptions that the conditions of market equilibrium can be stated in terms

of expected returns and the equilibrium expected returns are formed on the basis of the

information set Φt, have a major empirical implication. They rule out the possibility of

19

trading system based only on information in Φt that has expected profits or returns. Thus,

let

X j , t +1 = P j ,t +1 − E (P j ,t +1 / Φ t ) (2-2)

Then E (~

x j ,t +1 / Φ t ) = 0 (2-3)

Which, by definition, says that the sequence {xjt} is a “fair game” with respect to the

information sequence {Φ}. Or, equivalently, let

Z j ,t + 1 = r j , t + 1 − E (~

r j ,t + 1 / Φ t ), (2-4)

Then E (Z j ,t + 1 / Φ t )= 0 (2-5)

So the sequence {Zit} is also a “fair game” with respect to the information sequence {Φ}.

In economic terms, Xj,t+1 is the excess market value of security j at time t+1 : it is the

difference between the observed price and the expected value of the price that was

projected at t on the basis of the information Φt, and similarly, Zj,t+1 is the return at t+1 in

excess of the equilibrium expected return projected at t. let

a (Φ t ) = [a1 (Φ t ), a 2 (Φ t ),..., a n (Φ t )]

be any trading system based on Φt which tells the investors the amounts aj(Φt) of funds

available at t that are to be invested in each of the n available securities. The total excess

market value at t+1 that will be generated by such a system is

n

V t +1 = ∑ aj (Φ t )[r j ,t + 1 − E (~r j ,t + 1 / Φ t )]

j =1

n

~

(

E V t +1 / Φ t = ) ∑ aj (Φ t )E (Z~ j , t + 1 / Φ t ) = 0.

j =1

20

2-3-1-2 The submartingale model

Assume that in (2-1) that for all t and Φt

(~

)

E Pj ,t +1 / Φ t ≥ Pjt , Or equivalently, E (~

r j ,t +1 / Φ t ) ≥ 0 (2-6)

this is a statement that the price sequence {Pjt} for security j follows a submartingale with

respect to the information sequence {Φt}, which is no thing more than that the expected

value of next period’s price, as projected on the basis of information Φt, is equal to or

grater than the current price. If (2-6) holds as an equality (so that expected returns and

price changes are zero), then the price sequence follows a martingale. A submartingale in

prices has one important empirical implication. Consider the set of “one security and

cash” mechanical trading rules, which systems that concentrate on individual securities

and are that define the conditions under which the investor would hold a given security,

sell it short, or simply hold cash at any time t. then the assumption of (2-6) that expected

returns conditional on Φt are non-negative directly implies that such trading rules must

based only on the information in Φt cannot have grater profits than a policy of always

buying-and-holding the security during the future period in question.

The statement that the current price of a security “fully reflects” available

information was assumed to imply that successive price changes (or more usually,

successive one period returns) are independent. In addition, it was usually assumed that

successive changes (or returns) are identically distributed. Together the two hypotheses

constitute the random walk model. Formally, the model is:

f (r j , t +1 / Φ t ) = f (r j , t +1 ), (2-7)

Which is the usual statement that the conditional or marginal probability distributions of

an independent random variable are identical. In addition, the density function f must be

the same for all t. expression (2-7) says much more than the general expected return

21

model summarized by (2-1). For example, if we restrict (2-1) by assuming that the

expected return on security j is constant over time, then we have

E (~

r j ,t +1 / Φ t ) = E (~r j ,t +1 ). (2-8)

This says that the mean of the distribution of rj,t+1 is independent of the

information available at t, Φt, whereas the random walk model of (2-7) in addition says

that the entire distribution is independent at Φt. Random walk model can be considered

as an extension of the general expected return of “fair game” efficient market model, in

the sense of making a more detailed statement about the economic environment. The “fair

game” model just says that the conditions of market equilibrium can be stated in terms of

expected returns, and thus it says little about the details of the stochastic process

generating returns. A random walk arises within the context of such a model when the

environment is such that the evolution of investor tastes and the process generating new

information combine to produce equilibria in which return distributions repeat themselves

through time.

The multiple securities expected return models test whether securities are “appropriately

priced” vis-à-vis one another. But to judge whether differences between average returns

are “appropriate”, an economic theory of equilibrium expected return is required. Such as

one-factor Sharp-Lintner-Black model, multifactor asset-pricing model and market

model.

Sharp (1964) and Lintner (1965) propose the first version of this model. In this

model, the expected return on security j from time t to t+1 is

rm ,t +1 / Φ t ) − r f ,t +1 cov(~

E (~ rm ,t +1 / Φ t )

r j ,t +1 , ~

E (r~ /Φ )= r f ,t +1 + (2-9)

σ (rm ,t +1 / Φ t ) σ (~ rm,t +1 / Φ t )

j ,t +1 t ~

22

Where rf,t+1 is the return from t to t+1 on an asset that is riskless in money terms; rm,t+1 is

the return on the “market portfolio” m; σ 2 (~r / Φ ) is the variance of the return on m;

m ,t +1 t

cov (~ rm ,t +1 / Φ t ) is the covariance between the return on j and m; and the appearance

r j ,t +1 , ~

of Φt indicates that the various expected returns, variance and covariance, could in

principle depend on Φt. Though Sharp and Lintner derive (2-9) as a one-period model,

the result is given a multiperiod justification and interpretation in (2-9). In words, (2-9)

says that the expected one-period return on a security is the one-period riskless rate of

interest rf,t+1 plus a “risk premium” that is proportional to

cov (~

r j ,t +1 , r~m ,t +1 / Φ t ) / σ (~

rm ,t +1 / Φ t ) . In the Sharp-Lintner model each investor holds some

combination of the riskless asset and the market portfolio, so that, given a mean-standard

deviation framework, the risk of an individual asset can be measured by its contribution

to the standard deviation of the return on the market portfolio. This contribution is in

fact cov (~

r j ,t +1 , r~m ,t +1 / Φ t ) / σ (~

rm ,t +1 / Φ t ) . The factor [E(~r

m,t +1 ]

/ Φt ) − rf ,t +1 / σ (~

rm,t +1 / Φt ),

which is the same for all securities is then regarded as the market price of risk.

The early 1970’s produce the first extensive tests of SLB model (Black, Jensen,

and Scholes 1972; Blum and Friend 1973; and Fama and MacBeth 1973). These early

studies suggest that the special prediction of the Sharp-Lintner version of the model, that

portfolios uncorrelated with the market have expected returns equal to the risk-free rate

of interest, does not fare well (the average returns on such “zero-β” portfolios are higher

than the risk-free rate), other predictions of the model seem to do better. The most

general implication of the SLB model is that equilibrium pricing implies that the market

portfolio is ex ante mean-variance efficient in the sense of Markwitz (1959). Consistent

with this hypothesis, the early studies suggest that (i) expected returns are a positive

linear function of market β (the covariance of a security’s return with the return on the

market portfolio divided by the variance of market return), and (ii) β is the only measure

of risk needed to explain the cross-section of expected returns. With this early support for

the SLB model, there was a brief euphoric period in 1970’s when market efficiency and

the SLB model seemed to be sufficient description of security returns.

However, the empirical attacks on the SLB model begin in the late 1970’s with

studies that identify variables that contradict the model’s prediction that market β’s

23

suffice to describe the cross-section of expected returns. Basu (1977, 1983) shows that

earning /price ratio (E/P) has marginal explanatory power; controlling for β, expected

returns are positively related to E/P. Bans (1981) shows that a stock’s size (price times

shares) helps explain expected returns; given their market β’s, expected returns on small

stocks are too high, and expected returns on large stocks are too low. Bhandari (1988)

shows that leverage is positively related to expected stock returns in tests that also

include market β’s. Finally, Chan, Hamao, and Lakonishok (1991) and Fama and French

(1991) find that book-to-market equity (the ratio of the book value of a common stock to

its market value) has strong explanatory power; controlling for β, higher book-to-market

ratios are associated with higher expected returns.

The market model, which is originally suggested by Markwitz (1959),

hypothesizes that we can represent the return on an individual security (or portfolio) as a

linear function of an index of market returns, let:

r~j ,t +1 = a j + β j ~

rm ,t +1 + u j ,t +1 (2-10)

Where rj,t+1 is the rate of return on security j for time t; rm,t+1 is the corresponding return

on a market index m; aj, βj are parameters that can vary from security to security; and

uj,t+1 is a random disturbance. Fama, Fisher, Jensen, and Roll (1969) and the more

extensive work of Blum (1968), test the market model using monthly return data. These

tests indicate that (2-10) is well specified as a linear regression model in that (i) the

estimated parameter â j and β̂ j remain fairly constant over long periods of time (e. g.,

the entire post-world war II period in the case of Blum), (ii) rm,t+1 and the estimated uˆ j ,t +1 ,

are close to serially independent, and (iii) the uˆ j ,t +1 seems to be independent of rm,t+1.

Thus, the observed properties of the market model are consistent with the expected return

efficient market model, and in addition, the market model tells us some thing about the

process generating expected returns from security to security. However, the results for the

24

market model are just a statistical description of the return generating process, and they

are probably somewhat consistent with other models of equilibrium expected returns.

In the Sharp-Lintner-Black model, the cross-section of expected returns on

securities and portfolios is described by their market β’s, where β is the slope in the

simple regression of a security’s return on the market return. The multifactor asset-

pricing models of Merton (1973) and Ross (1976) generalize this result. In these models,

the return-generating process can involve multiple factors, and the cross-section of

expected returns is constrained by the cross-section of factor loadings (sensitivities). A

security’s factor loading are the lopes in a multiple regression of its return on the factors.

Ross (1976) suggests the Arbitrage-Pricing Theory (APT), uses factor analysis to

extract the common factors in returns and then tests whether expected returns are

explained by the cross-sections of the loadings of security returns on the factors (Roll and

Ross 1980; Chen 1983). Lehmann and Modest (1988) test this approach in detail. Most

interesting, using models with up to 15 factors, they test whether the multifactor model

explains the size anomaly of the SLB model. They find that the multifactor model leaves

an unexplained size effect much like the SLB model; that is, expected returns are too

high, relative to the model, for small stocks and too low for large stocks. The factor

analysis approach to tests for APT leads to un-resolvable squabbles about the number of

common factors in returns and expected returns (Dhrymes, Friend, and Gultekin 1984;

Roll and Ross 1984; Dhrymes, Friend, Gultekin and Gultekin 1984; Trzcinka 1986;

Conway and Reinganum 1988). Fama (1991) argues that the multifactor analysis

approach can confirm that there is more than one common factor in returns and expected

returns, which is useful.

25

2-3-2-4 Consumption-based Asset-Pricing models

The consumption-based model of Rubinsten (1976), Lucas (1978), Breeden

(1979), and others is the most elegant of the available intertemporal asset-pricing models.

In Breeden’s version, the interaction between optimal consumption and portfolio

decisions leads to a positive linear relation between expected returns on securities and

their consumption β’s. (a security’s consumption β’s is the slope in the regression of its

return on the growth rate of per capita consumption). The model thus summarizes all the

incentives to hedge shifts in consumption and portfolio opportunities that can appear in

Merton’s (1973) multifactor model with a one-factor relation between expected returns

and consumption β’s. The simple elegance of the consumption model produces a

sustained interest in empirical test. The tests use versions of the model that make strong

assumptions about tastes (time-additive utility for consumption and constant relative risk

aversion) and often about the joint distribution of consumption growth and returns

(multivariate normality). Because the model is then so highly specified, it produces a rich

set of testable predictions about the time series and cross-section properties of returns.

The empirical work on the consumption model often jointly tests its time series

and cross-section predictions, using the path breaking approach in Hansen and Singelton

(1982). Estimation is with Hansen’s (1982) generalized method of moments, the test is

based on a χ2 statistic that summarizes, in one number, how the data conform to the

model’s many restrictions. The tests usually reject. The disappointment comes when the

rejection is not pursued for additional descriptive information, obscure in the χ2 test,

about which restrictions of the model (time-series, cross-section, or both) are the

problem. In short, tests of the consumption model some times fail the test of usefulness;

they don’t enhance our ability to describe the behavior of returns, the tests of the

consumption model make no attempt to deal with the anomalies that have caused

problems for the SLB model. It would be interesting to confront consumption β’s with

variables like size and book-to-market equity, that have caused problems for the market

β’s of the SLB model. Given that the consumption model does not seem to fare well in

tests against the SLB model or the multifactor model, however, the consumption model

will do no better with the anomalies of the SLB model (Fama 1991).

26

Finally, it is important to emphasize that the SLB model, the consumption model,

and the multifactor model are not mutually exclusive. Following Constantinides (1989),

one can view the models as different ways to formalize the asset-pricing implications of

common general assumption about tastes (risk aversion) and portfolio opportunities

(multivariate normality).

The concept of efficiency is central to finance. And gained a lot of interest of

popularity that the literature now is so vast and impossible to be included in a single

review as correctly indicating by Fama (1991, pp.1575): “The literature is now so large

that a full review is impossible”. Therefore, the main work about market efficiency

especially that of particular interest to the purpose of this research, is included. We will

start with the origins of EMH, the Random Walk Model and EMH, evidence of EMH in

its three forms; weak-form tests (return predictability); semi-strong tests (event studies);

and strong-form tests (private information). After that we will come to the attack on

EMH through existing anomalies in the literature.

If capital markets are sufficiently competitive, then simple microeconomics

indicates that investors cannot expect to achieve superior profits from their investment

strategies. But although this appears self-evident today, it was far from obvious for the

majority of the century. Up to the end of the 1950s, there were few theoretical or

empirical studies of securities markets; and until Cootner (1964) collected a selection of

papers from a wide variety of sources, the literature was dispersed across journals in

statistics, operations research, mathematics and economics.

The concept of market efficiency had been anticipated at the beginning of the

century in a dissertation submitted by Bachelier (1900) to the Sorbonne for his PhD in

mathematics. In his opening paragraph, Bachelier recognizes that “past, present and even

discounted future events are reflected in market prices, but often show no apparent

relation to price changes”. This recognition of the informational efficiency of the market

leads Bachelier to continue, in his opening paragraphs, that “if the market, in effect, does

27

not predict its fluctuations, it does assess them as being more or less likely, and this

likelihood can be evaluated mathematically”. This gives rise to a brilliant analysis that

anticipates not only Albert Einstein’s subsequent derivation of the Einstein-Wiener

process of Brown motion, but also many of the analytical results that were rediscovered

by finance academics in the second half of the century. Bachelier’s contribution was

overlooked until it was circulated to economists by Paul Samuelson in the late 1950’s and

subsequently published in English by Cootner (1964).

Although there could have been an emerging theory of speculative markets during

the first half of the twentieth century, this was not to be. Instead, the early literature

followed the path of accumulating a variety of empirical observations that did not sit

easily alongside the paradigms of economics or the beliefs of practitioners. Bachelier had

concluded that commodity prices fluctuate randomly, and later studies by Working

(1934) and Cowels and Jones (1937) were to show that U.S stock prices and other

economic series also share the same characteristics. These studies were largely

overlooked by researchers until the late of 1950’s. There was in addition, disturbing

evidence about the difficulty of beating the equity markets. Alfred Cowels III, founder of

the Cowels Commission and benefactor of the Economic Society, published in the launch

issue in Econometrica a painstaking analysis of many thousands of stock selections made

by investment professionals. Cowels (1933) finds that there was no discernable evidence

of any ability to outguess the market. Subsequently, Cowels (1944) provides

corroborative results for a large number of forecasts over a much longer sample period.

By the 1940’s, there was therefore scattered evidence in favor of the weak and strong

form efficiency of the market, though these terms were not yet in use.

The weak form of the efficient market hypothesis, which expanded by Fama

(1991) to include returns predictability, claims that prices fully reflect the information

implicit in the sequence of past prices and prices have no memory and follow random

walk properties. The literature begins, therefore, with studies of weak-form market

efficiency.

28

2-4-2-1 Random Walk Hypothesis (RWH)

The random walk theory asserts that price movement will not follow any patterns

or trends and that past price movements cannot be used to predict future price

movements. In the early literature, discussions of the efficient markets model were

phrased in terms of the even more special random walk model. Fama (1970) summarizes

the early random walk literature and his own contributions and other studies of the

information contained in the historical sequence of prices. It was not until the work of

Samuelson (1965) and Mandelbrot (1966) that the role of “fair game” expected return

models in the theory of efficient markets and the relationships between these models and

the theory of random walks were rigorously studied. And these papers come somewhat

offer the major empirical work on random walks. Until the Mandelbrot-Samuelson

models appeared, there exists a large body of empirical results in search of a rigorous

theory.

The first statement and test of the random walk model was that of Bachelier

(1900). But his “fundamental principle” for the behavior of prices was that speculation

should be a “fair game”; in particular, the expected profits to the speculator should be

zero. With the benefit of stochastic processes theory, the process implied by this

fundamental principle is a Martingle. After Bachelier, research on the behavior of

security prices lagged until the coming of computer. In 1933 Kendall (1953) examined

the behavior of weekly changes in nineteen indices of British industrial share prices and

in spot prices for cotton (New York) and wheat (Chicago). After extensive analysis of

serial correlations, he suggests: “the series looks like a wondering one, almost as if once

a week the Demon of Chance drew a random number from a symmetrical population of

fixed dispersion and added it to the current price to determine the next week’ price”

(Kendall 1953, p.13).

Kendall’s conclusion had been suggested earlier by Working (1934) though his

suggestion lacked the force provided by Kendall’s empirical results, and the implications

of the conclusion for stock market research and financial analysis were later underlined

by Roberts (1959). But the suggestion by Kendall, Working, and Roberts that series of

speculative prices may be well described by random walks was based on observations,

none of these authors attempted to provide much economic rational for the hypothesis,

29

and, indeed, Kendall felt that the economists would generally reject it. Osborne (1959)

suggests market conditions, similar to those assumed by Bachelier that would lead to a

random walk. But in his model, independence of successive price changes derives from

the assumption that the decisions of investors in an individual security are independent

from transaction to transaction.

Most of the empirical evidence in the random walk literature can easily be

interpreted as tests of more general expected return or “fair game” models, “fair game”

models imply the “impossibility” of various sorts of trading systems. Some of the random

walk literature has been concerned with testing the profitability of such systems. More of

the literature has, however, been concerned with tests of serial covariance of returns. The

serial covariances of a “fair game” are zero, like a random walk, so that these tests are

also relevant for the expected return models.

If xt is a “fair game”, its unconditional expectation is zero and its serial covariance can be

written in general form as:

~

( ~

E X t+r X t = ) ∫X t (~

E X t+r / X t ) f (x t )dx t ,

xt

~

(

E X t +1 / X t = 0 )

From this, it follows that for all lags, the serial covariances between lagged values of a

“fair game” variable are zero.

Thus, observations of a “fair game” variable are linearly independent. But the “fair

game” model does not necessarily imply that the serial covariances of one-period returns

are zero. In the weak form tests of this model the “fair game” variable is:

z j ,t = r j ,t − E (~

r j ,t / r j ,t −1 , r j ,t − 2 ,... ) (2-11)

But the covariance between, for example, rjt and rj,t+1 is:

30

E ([ ~

r j ,t + 1 − E ( ~r j , t + 1 )][ r~ jt − E ( ~ r jt )])

= ∫ [ r jt − E ( ~ r jt )][ E ( ~ r j , t + 1 / r jt ) − E ( ~r j , t + 1 )] f ( r jt ) dr jt

, (2-12)

r jt

r j ,t +1 ) .

In the “fair game” efficient markets model, the deviation of the return for t+1

from its conditional expectation is a ”fair game” variable, but the conditional expectation

itself can depend on the return observed for t. In the random walk literature, this problem

is not recognized, since it is assumed that the expected return (and indeed the entire

distribution of returns) is stationary through time. In practice, this implies estimating

series covariances by taking cross products of deviations of observed returns from the

overall sample mean return. This procedure, which represents a rather gross

approximation from the view point of the general expected return efficient market model,

does not seem to greatly affect the results of the covariance tests, at least for common

stocks.

However, there are types of nonlinear dependence that imply the existence of

profitable trading systems, and yet do not imply non-zero serial covariance. The first

major evidence on trading rules was Alexander’s (1961, 1964). He tests a variety of

trading systems, such a y% filter, which is a “one security and cash” trading rule, so that

the results it produces are relevant for the submartingale expected return model.

Alexander concludes that there is some evidence in his results against the independence

assumption of the random walk model. But market efficiency does not require a random

walk, and from the view point of the submartingale model, the conclusion that the filters

cannot beat buy-and-hold is support for the efficient market hypothesis. Further support

is provided by Fama and Blum (1966) who compare the profitability of various filters to

buy-and-hold for the individual stocks of Dow-Jones Industrial Average. But some

evidence in the filter tests of both Alexander and Fama-Blum that is inconsistent with the

submartingale efficient markets model. In particular, the results for very small filters

indicate that it is possible to devise trading schemes based on very short-term (intra-day

but at most daily) price swings that will be on average outperform buy-and-hold.

31

These results are evidence of persistence or positive dependence in very short-

term price movement. This is consistent with the evidence for slight positive linear

dependence in successive daily price changes produced by serial correlations. Thus, the

filter tests, like the serial correlations, produce empirically noticeable departures from the

strict implications of the efficient markets model. But, despite of any statistical

significance they might have, from the economic viewpoint the departures are so small

that it seems hardly justified to use them to declare the market inefficient. Another

departure from the pure independence assumption of the random walk model has been

noted by Osborne (1962), Fama (1965) and others. In particular, large daily price changes

tend to be followed by large daily changes. The signs of the successor changes are

apparently random, however, which indicates that the phenomenon represents a denial of

the random walk model but not of the market efficiency hypothesis. It may be that when

important new information comes into the market it cannot always be immediately

evaluated precisely. Thus, sometimes the initial price will over adjust to the information,

and other times it will under adjust. But since the evidence indicates that the price

changes on days following the initial large change are random. In sign, the initial large

change at least represents an unbiased adjustment to the ultimate price effects of the

information, and this is sufficient for the expected return efficient market model.

Niederhoffer and Osborne (1966) document two departures from complete

randomness in common stock price changes from transaction to transaction. First, their

data indicate that reversals (pairs of consecutive price changes of opposite signs) are from

two to three times as likely as continuations (pairs of consecutive price changes of the

same sign). Second, a continuation is slightly more frequent after a preceding

continuation than after a reversal. Niederhoffer and Osborne offer explanation for these

phenomena based on the market structure of the New York Stock Exchange (NYSE). But

though Niederhoffer and Osborne present convincing evidence of statistically significant

departures from independence in price changes from transaction to transaction, and

though their analysis of their findings presents interesting insights into the process of

market making on the major changes. Their analysis of market making does, however,

point clearly to the existence of market inefficiency, but with respect to strong form tests

of the efficient market model.

32

The random walk literature also has centered on the nature of the distribution of

price changes, which is an important issue for the EMH. Since the nature of the

distribution affects both the types of statistical tools relevant for testing the hypothesis

and the interpretation of any results obtained. A model implying normally distributed

price changes was first proposed by Bachelier (1900), who assumed that price changes

from transaction to transaction are independent, identically distributed random variables

with finite variances. If transactions are fairly uniformly spread across time, and if the

number of transactions per day, week, or month is very large, then the Central Limit

Theorem leads us to expect that these prices changes will have normal or Guassian

distribution. Osborne (1959), Moore (1962), and Kendall (1953) although their empirical

evidence support the normality hypothesis, but all observed high tails in their data

distribution. Drawing on these finding and some empirical work of his own, Mandelbort

(1963) then suggests that these departures from normality could be explained by a more

general form of Bachelier model. In particular, if one does not assume that distributions

of price changes from transaction necessarily have finite variance, then the limiting

distributions for price changes over longer differencing intervals could be any member of

the stable class, which includes the normal as special case. Non-normal stable

distributions have higher tails than the normal, and so can account for this empirically

observed feature of distributions of price changes. Fama (1965), after extensive testing,

concludes that non-normal stable distributions are a better description of distributions of

daily returns on common stocks than the normal. This conclusion is also supported by the

empirical work of Blum (1968) on common stocks, and it has been extended to U.S

Government Treasury Bills by Roll (1968).

The tests of weak-form-efficiency up to this point, focused on forecasting returns

from past returns. However, and after 1970 the tests reject the market efficiency-constant

expected returns model that seems to do well in the early work. Since weak-form tests

concerned with the forecast power of past returns, Fama (1991) expands the coverage of

the first category of EMH to cover the more general area of tests for return predictability.

Such as forecast power of variables like dividend yield (D/P), earning/price ratio (E/P),

33

and term structure variables. Moreover, the early work of weak-form-efficiency

concentrated on the predictability of daily, weekly, and monthly returns, but the recent

tests also examine the predictability of returns for long horizons.

In the pre-1970 literature the common equilibrium-pricing model in tests of stock

market efficiency is the hypothesis that expected returns are constant through time.

Market efficiency then implies that returns are unpredictable from past returns or other

past variables, and the best forecast of return is its historical mean. After the 1970’s, daily

data on NYSE and AMEX stocks back to 1962, makes it possible to estimate precisely

the autocorrelation in daily and weekly returns. Lo and MacKinlay (1988) find that

weekly returns on portfolios of NYSE stocks grouped according to size, show reliable

positive autocorrelation. The autocorrelation is stronger for portfolios of small stocks.

This suggests, however, that the results are due in part to the nonsynchronous trading

effect (Fisher 1966). Fisher emphasizes that spurious autocorrelation in portfolio returns,

induced by nonsynchronous closing trades for securities in the portfolio, is likely to be

more important for portfolios titled toward small stocks. Conrad and Kaul (1988)

examine the autocorrelation of Wednesday-to-Wednesday returns for size-grouped

portfolios of stocks that trade on both Wednesdays. Like Lo and MacKinlay (1988), they

find that weekly returns are positively autocorrelated, and more for portfolios of small

stocks. French and Roll (1986) find that stock prices are more variable when the market

is open. On an hourly basis, the variance of price changes is 72 times higher during

trading hours than during weekend nontrading hours. Likewise, the hourly variance

during trading hours is 13 times the overnight nontrading hourly variance during the

trading week. One of the explanations that French and Roll test is a market inefficiency

hypothesis popular among academics; specifically, the higher variance of price changes

during trading hours is partly transitory, the results of noise trading by uninformed

investors (Black 1986). Under this hypothesis, pricing errors due to noise trading are

eventually reversed, and this induces negative autocorrelation in daily returns.

With the Center for Research in Security Prices (CRSP) daily data back to 1962,

post -1970’s research is able to show confidently that daily and weekly returns are

predictable from past returns. The work thus rejects the old market efficiency-constant

expected returns model on a statistical basis. The results tend to confirm the conclusion

34

of the early work that, at least for individual stocks, variation in daily and weekly

expected returns is a small part of the variance of returns. The early literature does not

interpret the autocorrelation in daily and weekly returns as important evidence against the

joint hypothesis of market efficiency and constant expected returns. The argument is that,

even when the autocorrelations deviate reliably from zero, they are close to zero and thus

economically insignificant. The view that autocorrelation of short-horizon returns close

to zero imply economic insignificance is challenged by Shiller (1984) and Summers

(1986). They present simple models in which stock prices take large slowly decaying

swings away from fundamentals values (fads, or irrational bubbles), but short-horizon

returns have little autocorrelation. In the Shiller-Summers model, the market is highly

inefficient, but in away that is missed in tests on short-horizon returns.

Stambaugh (1986) points out that although the Shiller-Summers model can

explain autocorrelation of short-horizon returns that are close to zero. The long swings

away from fundamental value proposed in the model imply that long-horizon returns

have strong negative autocorrelation. Since the swings away from fundamental value are

temporary, over long horizons they tend to be reversed. Another implication of the

negative autocorrelation induced by temporary price movements is that the variance of

returns should grow less than in proportion to the return horizon. Fama and French

(1988a) find that the autocorrelations of returns on diversified portfolios of NYSE stocks

for the 1926-1985 period have the pattern predicted by Shiller-Summers model. Even

with 60 years of data, however, the tests on long-horizon returns imply small sample size

and low power. When Fama and French delete the 1926-1940 period from the tests, the

evidence of strong negative autocorrelation in 3-to5 years’ returns disappears. Similarly,

Poterba and Summers (1988) find that, for N from 2 to 8 years, the variance of the N-year

returns on diversified portfolios grows much less than in proportion to N, this is

consistent with the hypothesis that there is negative autocorrelation in returns induced by

temporary price swings. Even with 115 years (1871-1985) of data, however, the variance

tests for long-horizon returns provide weak statistical evidence against the hypothesis that

returns have no autocorrelation and prices are random walks.

DeBondt and Thaler (1985, 1987) mount an aggressive empirical attack on market

efficiency, directed at unmasking irrational bubbles. They find that the NYSE stocks

35

identified as the most extreme losers over a 3-to 5 years period tend to have strong

returns relative to the market during the following years, especially in January of the

following years. Conversely, the stocks identified as extreme winners tend to have weak

returns relative to the market in subsequent years. They attribute these results to market

overreaction to extreme bad or good news about firms. Jagadeesh (1990), Lehmann

(1990), Lo and MacKinlay (1990) also find reversal behavior in the weekly and monthly

returns of extreme winners and losers. Lehmann’s weekly reversals seem to lack

economic significance. When he accounts for spurious reversals due to bouncing between

bid and ask price, trading costs of 0.2% per turnaround transaction suffice to make the

profits from his reversal trading rules close to zero.

An autocorrelation is the slope in a regression of the current return on a past

return. Since variation through time in expected returns is only part of the variation in

returns, tests based on autocorrelations lack power because past realized returns are noisy

measures of expected returns. Power in tests for return predictability can be enhanced if

one can identify forecasting variables that are less noisy proxies for expected returns than

past returns. A Puzzle of the 1970’s was to explain why monthly stock returns are

negatively related to expected inflation (Nelson 1976; Jaffe and Mandelker 1976; Fama

1981) and the level of short-term interest rates (Fama and Schwert 1977). Like the

autocorrelation tests, however, the early work on forecasts of short-horizon returns from

expected inflation and interest rates suggests that the implied variation in expected return

is a small part of the variance of returns. However, for long-horizon returns, predictable

variation is a larger part of return variances.

Fama and French (1988b) use dividend yields D/P to forecast returns on the

value-weighted and equally weighted portfolios of NYSE stocks for horizons from 1

month to 5 years. D/P explains small fractions of monthly and quarterly return variances.

Fractions of variances explained grow with the return horizon, however, and are around

25% for 2-to 4 years returns. Campbell and Shiller (1988b) find that E/P ratios have

reliable forecast power that also increased with return horizon. Fama and French (1988b)

argue that dividend yields track highly autocorrelated variation in expected returns that

becomes a large fraction of return variation for longer return horizons. The increasing

fraction of the variance of long-horizon returns explained by D/P is thus due in large part

36

to the slow mean reversion of expected returns. Examining the forecast power of

variables like D/P and E/P over a range of returns horizons nevertheless gives striking

perspective on the implications of slow-moving expected returns for the variation of

returns. Fama and French (1989) suggest a different way to judge the implications of

return predictability for market efficiency. They argue that if variation in expected returns

is common to different securities, then it is probably a rational result of variation in tastes

for current versus future consumption or in the investment opportunities. They push the

common expected returns argument for market efficiency one step farther, they argue that

there are systematic patterns in the variation of expected returns through time that suggest

that it is rational. They find that the variation in expected returns tracked by D/P and the

default spread (the slopes in the regressions of returns on D/P or the default spread)

increase from high-grade bonds to low-grade bonds, from bonds to stocks, and from large

stocks to small stocks, this ordering corresponds to intuition about the risks of the

securities.

On the other hand, the variation in expected returns tracked by the term spread is

similar for all long-term securities (bonds and stocks), which suggests that it reflects

variation in a common premium for maturity risks. The general message of the Fama-

French tests is that D/P and the default spread are high (expected returns on stocks and

bonds are high) when times have been poor (growth rates of output have been persistently

low). On the other hand, the term spread and expected returns are high when economic

conditions are weak but anticipated to improve (future growth rates of output are high).

Persistent poor times may signal low wealth and higher risks in security returns, both of

which can increase expected returns. In addition, if poor times (and low incomes) are

anticipated to be partly temporary, expected returns can be high because consumers

attempt to smooth consumption from the future to the present.

Semi-strong form tests of efficient markets model are concerned with whether

current prices “fully reflect” all obviously publicly available information. Fama (1991)

proposes changes in title, not coverage. He uses the title “event studies” instead of semi-

strong-form tests of the adjustment of prices to public announcements. The study of stock

37

splits by Fama, Fisher, Jensen, and Roll (1969), was the original event study. The

purpose of the study was to have a work that made extensive use of the newly developed

CRSP monthly NYSE file at that time. They find that, if information in stock splits

concerning the firm’s future dividend payments is on average fully reflected in the price

of a split share at the time of the split. Event studies are now an important part of finance,

especially corporate finance. In 1970’s there was little evidence on the central issues of

corporate finance. Now we are overwhelmed with results, mostly from event studies,

using simple tools, this research documents interesting regularities in the response of

stock prices to investment decisions; financing decisions; and changes in corporate

control.

Regarding EMH, the CRSP files of daily returns on NYSE, AMEX, and

NASDAQ stocks are a major boost for the precision of event studies. When the

announcement of an event can be dated to the day, daily data allow precise measurement

of the speed of stock-price response- the central issue of market efficiency. Another

powerful advantage of daily data is that they can attenuate or eliminate the joint-

hypothesis problem, that market efficiency must be tested jointly with an asset-pricing

model. The typical result in event studies on daily data is that, on average, stock prices

seem to adjust within a day to the event announcement. The fact that quick adjustment is

consistent with efficiency is noted, and then the studies move on to other issues. In short,

in the only empirical work where the joint hypothesis problem is relatively unimportant,

the evidence typically says that, with respect to firm-specific events, the adjustment of

stock prices to new information is efficient (Fama 1991).

Moreover, when part of the response of prices to information seems to occur

slowly, event studies become subject to the joint-hypothesis problem. For example, the

early merger work finds that the stock prices of acquiring firms hardly react to merger

announcements, but therefore they drift slowly down (Asquith 1983). One possibility is

that acquiring firms on average pay too much for target firms, but the market only

realizes this slowly; the market is inefficient (Roll 1986). Another possibility is that the

post-announcement drift is due to bias in measured abnormal returns (Frank, Haris, and

Titman 1991). Still another possibility is that the drift in the stock prices of acquiring

firms in the early merger studies is sample-specific. Mitchell and Lehn (1990) find no

38

evidence of post announcement drift during the 1982-1986 period for a sample of about

400 acquiring firms. Post-announcement drift in abnormal return is also a common result

in studies of the response of stock prices to earnings announcements. Predictability, there

is a raging debate on the extent to which the drift can be attributed to problems in

measuring abnormal returns (Bernard and Thomas 1989; Ball, Kothari, and Watta 1990).

Bernard and Thomas (1990) identify a more direct challenge to market efficiency in the

way stock prices adjusted to earnings announcements; they argue that the market does not

understand the autocorrelation of quarterly earnings. As a result, part of the 3-day stock-

price response to this quarter’s earning announcement is predictable from earning 1 to 4

quarters back.

In short, some event studies suggest that stock prices do not respond quickly to

specific information. Given the event –study boom of the last 20 years, however, some

anomalies, spurious and real, are inevitable. Moreover, event studies are the cleanest

evidence we have on efficiency (the least encumbered by the joint-hypothesis problem).

With few exceptions, the evidence is supportive (Fama 1991).

The strong tests of the efficient markets model are concerned with whether all

available information is fully reflected in prices in the sense that no individual has higher

expected trading profits than others because he has monopolistic access to some

information. Niederhoffer and Osborn (1966) show that NYSE specialists use their

monopolistic access to the book of limit orders to generate trading profits. Scholes (1972)

shows that corporate insiders have access to information not reflected in prices. Jaffe

(1974) finds that for insiders the stock market is not efficient; insiders have information

that is not reflected in prices, and market does not react quickly to public information

about insider trading, outsiders can profit from the knowledge that there has been heavy

insider trading for up to 8 months after information about trading becomes public.

Seyhun (1986) offers an explanation, he confirms that insiders profit from their

trades, but he does not confirm Jaffe’s finding that outsiders can profit from public

information about insider trading. Seyhun argues that Jaffe’s outsider profits arise

because he uses the Sharp-Lintner-Black model for expected returns, Seyhun shows that

39

insider buying is relatively more important in small firms, whereas insider selling is more

important in large firms. There is a general message in Seyhun’s results, highly

constrained asset-pricing model like the Sharp-Lintner-Black model are surely false.

They have systematic problems explaining the cross-section of expected returns that can

look like market inefficiencies. In market-efficiency tests, one should avoid models that

put strong restrictions on the cross-section of expected returns, if that is consistent with

the purpose at hand. Concretely, one should use formal asset-pricing models when the

phenomenon studied concerns the cross-section of expected returns (e.s. , tests for size,

leverage, and E/P effects). But when the phenomenon is firm-specific (most event

studies), one can use firm-specific “models”, like the market model or historical average

returns, to abstract from normal expected returns without putting unnecessary constraints

on the cross-section of expected returns (Fama 1991).

For security analysis, the Value Line Investment Survey publishes weekly

rankings of 1700 common stocks into 5 groups, group 1 has the best return prospect and

group 5 the worst. There is evidence that, adjusted for risk and size, group 1 stocks have

higher average returns than group 5 stocks for horizons out 1 year (Black 1973; Copeland

and Mayers 1982; and Huberman and Kandel 1987, 1990). Affleck-Graves and

Mendenhall (1990) argue however, that the Value Line ranks firms largely on the basis of

recent earnings surprises. As a result, the longer-term abnormal returns of the Value Line

rankings are just another anomaly in disguise, the post-earnings-announcement drift

identified by Ball and Brown (1968), Bernard and Thomas (1989), and others. Stickels

(1985) uses event-study methods to show that there is an announcement effect in rank

changes that more clearly implies that Value Line has information not reflected in prices.

Moreover, Hulbert (1990) reports that the strong long-term performance of Value Line’s

group 1 stocks is weaker after 1983. Over the 6.5 years from 1984 to mid-1990, group 1

stocks earned 16.9% per year compared with 15.2% for the Wilshire 5000 Index. During

the same period, Value Line’s Centurion Fund, which specializes in group 1, earned

12.7% per year, live testimony to the fact that there can be large gaps between simulated

profits from private information and what is available in practice.

Regarding professional portfolio management, Jensen (1968, 1969) early results

were bad news for the mutual-fund industry. He finds that for the 1945-1964 period,

40

returns to investors in funds (before load fees, but after management fees and other

expenses) are on average about 1% per year below the market line (from the risk free rate

through the S&P 500 market portfolio) of the Sharp-Lintner model, and average returns

on more than half of his funds below the line. Only when all published expenses of the

funds are added back do the average returns on the funds scatter randomly about the

market line. Jensen concludes that mutual-funds managers do not have private

information. Other studies do not always agree, in tests on 116 mutual funds for the

February 1968 to June 1980 period, Henriksson (1984) finds that average returns to fund

investors, before load fees but after other expenses, are trivially different (0.02% per

month) from the Sharp-Lintner market line. Chang and Lewellen (1984) get similar

results for 1971-1979. This work suggests that on average, fund managers have access to

enough private information to cover the expenses and management fees they charge to

investors. Ippolito (1989) provides a more extensive analysis of the performance of

mutual funds, he examines 143 funds for the 20-years post-Jensen’s period 1965-1984, he

finds that fund returns, before load fees but after other expenses, are on average 0.83%

per year above the Sharp-Lintner market line (from the 1-year Treasury Bill rate through

the S&P 500 portfolio). He finds no evidence that the deviations of funds from the market

line are related to management fees, other fund expenses, or turnover ratios. Ippolito

concludes that his results are in the spirit of the “noisy rational expectations” model of

Grossman and Stiglitz (1980), in which informed investors (mutual fund managers) are

compensated for their information costs.

Performance evaluation is known to be sensitive to methodology (Grinblatt and

Titman 1984). Ippolito (1989) uses the Sharp-Lintner model to estimate normal returns to

mutual funds. Brinson, Hood, and Beebower (1980) use passive portfolios meant to

match the bond and stock components of their pension funds. We know the Sharp-Lintner

model has systematic problems explaining expected returns (size, leverage, E/P, and

book-to-market equity effects) that can affect estimates of abnormal returns (Fama,

1991). Elton, Gruber, Das, and Hklarka (1991) test the importance of the Sharp-Lintner

methodology in Ippolito’s results, they find that during Ippolito’s 1965-1984 period, his

benchmark combinations of Treasury bills with the S&P 500 portfolio produce strong

positive estimates of “abnormal” returns for passive portfolios of non-S&P (smaller)

41

stocks-strong confirmation that there is a problem with the Sharp-Lintner benchmarks

(also used by Jensen 1968, 1969; Henriksson 1984; and Chang and Lewellen 1984).

Elton, Gruber, Das, and Hklarka then use a 3-factor model to evaluate the

performance of mutual funds for 1965-1984; the 3 factors are the S&P 500, a portfolio

titled toward non S&P stocks, and a proxy for the market portfolio of Government and

corporate bonds. As in Brinson, Hood, and Beebower (1986), the goal of the Elton-

Gruber-Das-Hklarka approach is to allow for the fact that mutual funds hold bonds and

stocks that are not in the universe covered by the combination of the Treasury bills and

the S&P 500 that Ippolito uses to evaluate performance. The Elton-Gruber-Das-Hklarka

benchmarks are the returns from passive combinations of Treasury bills with S&P stocks,

and bonds. They find that for Ippolito’s 1965-1984 period, their benchmarks produce an

abnormal return on mutual funds of -1.1% per year, much like the negative performance

for pension funds (Brinson, Hood, and Beebower 1986). Moreover, unlike Ippolito, but in

line with earlier work (Sharp 1966), Elton, Gruber, Das, and Hklarka find that abnormal

returns on mutual funds are negatively related to fund expenses (including management

fees) and turnover. In short, if mutual and pension fund managers are the informed

investors of the Grossman-Stiglitz (1980) model, they are pushing research and trading

beyond the point where marginal benefits equal marginal costs (Fama1990).

In summary, the investors studied in most detail for private information are

pension fund and mutual fund managers. Unlike event studies, however, evaluating the

access of investment managers to private information involves measuring abnormal

returns over long periods. The tests thus run head-on into the joint-hypothesis problem:

measured abnormal returns can result from market inefficiency, a bad model of market

equilibrium, or problems in the way the model is implemented.

2-5 Evidence against EMH and alternative models for market behavior

The EMH has provided the theoretical basis for the financial market research

during seventies and the eighties. In the past, most of the evidence seems to have been

consistent with the EMH. Prices were seem to follow a random walk model and the

predictable variations in equity returns, if any, were found to be statistically insignificant.

While most of the studies in the seventies focused on predicting prices from past prices,

42

studies in the eighties also look at the possibility of forecasting based on variables such as

dividend yield (e.g. , Fama and French 1988), P/E ratios (Campbell and Shiller 1988),

and term structure variables (e.g. , Harvey 1991). Studies in the nineties look at

inadequacies of current asset pricing models. The accumulating evidence suggests that

stock prices can be predicted with a fair degree of reliability. Two competing

explanations have been offered for such behavior, proponents of EMH (e.g. , Fama and

French 1995) maintain that such predictability results from time-varying equilibrium

expected returns generated by rational pricing in an efficient market that compensates for

the level of risk undertaken. Critics of EMH (e.g. , Laporta, Lakonishok, Shliefer, and

Vishny 1997), argue that the predictability of stock returns reflects the psychological

factors, social movements, noise trading, and fashions or “fads” of irrational investors in

a speculative market. The question about whether predictability of returns represents

rational variations in expected returns or arises due to irrational speculative deviations

from theoretical values has provided the impetus for fervent intellectual inquiries in the

recent years. The reminder of this section is motivated largely by this issue, and places

greater emphasis on the speculative aspects.

The EMH became controversial especially after the detection of certain anomalies

in the capital markets, these anomalies can be divided into two main categories: long-

term return anomalies and calendar effects.

Fama (1998) provides a review of this literature; many of recent studies on long-

term returns suggest market inefficiency, specifically, long-term underreaction or

overreaction to information. Fama (1998) gives a “solid no” as an answer to the question

whether this literature (long-term return anomalies) viewed as a whole suggests that

efficiency should be discarded, and he gives two reasons for his answer: First, an efficient

market generates categories of events that individually suggest that prices overreact to

information, but in an efficient market, apparent underreaction will be about as frequent

as overreaction. If anomalies split randomly between underreaction and overreaction,

43

they are consistent with market efficiency. Second, long-term return anomalies are

sensitive to methodology, they tend to become marginal or disappear when exposed to

different models for expected (normal) returns or when different statistical approaches

are used to measure them. Thus, even viewed one-by-one, most-long term return

anomalies can be reasonably be attributed to chance (Fama 1998).

One of the first papers on long-term return anomalies is DeBondt and Thaler

(1985). They find that when stocks are ranked on three-to five- year past returns, past

winners tend to be future losers, and vise versa. They attribute these long-term return

reversals to investors’ overreaction. In forming expectations, investors give too much

weight to the past performance of firms and too little to the fact that performance tends to

mean-revert. DeBondt and Thaler seem to argue that overreaction to past information is a

general prediction of the behavioral decision theory of Kahneman and Tversky (1982).

Thus, one could take overreaction to be the prediction of a behavioral finance alternative

to market efficiency. Lakonishok et al. (1994) argue that ratios involving stock prices

proxy for past performance, firms with high ratios of earnings to price (E/P), cash flow to

price (C/P), and book-to-market equity (BE/ME) tend to have poor past earning growth,

and firms with low E/P, C/P, and BE/ME tend to have strong past earnings growth.

Because the market over-reacts to past growth, it is surprised when earnings growth mean

reverts. As a result, high E/P, C/P, and BE/ME stocks (poor past performers) have high

future returns, and low E/P, C/P, and BE/ME stocks (strong past performers) have low

future returns.

If apparent overreaction was the general result in studies of long-term returns,

market efficiency would be dead, replaced by the behavioral alternative of DeBondt and

Thaler (1985). In fact, apparent underreaction is about as frequent, the granddaddy of

underreaction events is the evidence that stock prices seem to respond to earnings for

about a year after they are announced (Ball and Brown 1983; Bernard and Thomas 1990).

More recent is the momentum effect identified by Jagadeesh and Titman (1993); stocks

with high returns over the past year tend to have high returns over the following three to

six months. Over recent event studies also produce long-term post-event abnormal return

44

that suggest underreaction. Custias et al. (1993) find positive post-event abnormal returns

for divesting firms and the firms they divest. They attribute the result to market

underreaction to an enhanced probability that, after a spinoff, both the parent and the

spinoff are likely to become merger targets, and the recipients of premiums. Desai and

Jain (1997) and Ikenberry et al. (1996) find that firms that split their stock experience

long-term positive abnormal returns both before and after the split, they attribute the post-

split returns to market underreaction to the positive information signal of by split. Finally,

Michael et al. (1995) find that stock prices seem to under-react to the negative

information in dividend omissions and the positive information in initiations.

Among the more striking of the long-term return anomalies is the study of initial

public offerings (IPOs) and seasonal equity offerings (SEOs) by Loughran and Ritter

(1995). They find that the total wealth generated at the end of five years if one invests $1

in each IPO or SEO immediately following the event is about 70% of that produced by

the same buy-and-hold strategy applied to a sample of stocks matched to the IPOs and

SEOs on size. IPOs and SEOs clearly have poor long-term returns during the Loughran-

Ritter sample period (1970-1990). The interesting question is whether the returns are

really abnormal or whether they are shared with non-event firms similar on

characteristics related to average returns, during the Loughran-Ritter period, variables

known to be related to average stock return include size and book-to-market equity

(Fama and French, 1992), and short-term past return (Jagadeesh and Titman, 1993). Since

the long-term buy-and-hold returns in Loughran and Ritter only control for size, their

results might be affected by other variables that are systematically related to average

return. Following up on this possibility, Brav and Gompers (1997) compare five-year

buy-and-hold returns on IPOs with the returns on portfolios that match the IPOs on size

and book-to-market equity (BE/ME) but exclude SEOs as well as IPOs. The five-year

wealth relative (the ratio of five-year buy-and-hold wealth for IPOs to five-year buy-and-

hold wealth for the benchmarks) rises from about 0.7 with the Loughran-Ritter size

benchmarks to a bit more than 1.0 (that is the anomaly disappear) when the benchmarks

control for BE/ME as well as size.

45

Similarly, Brav et al. (1995) find that the five-year buy-and-hold returns on SEOs

are closed to those of non-event portfolios matched on size and BE/ME, Brav (1997) and

Michell and Stafford (1997) show that IPOs and SEOs are typically small growth stocks,

Fama and French (1993) show that such stocks have low returns during the post-1963

period. The results of Brav and Gompers (1997) and Brav et al. (1995) then suggest that

explaining the IPO-SEO anomaly reduces to explaining why small growth stocks in

general have poor returns during the IPO-SEO sample period. In other words, if there is a

mispricing problem, it is not special to IPO-SEO stocks (Fama, 1998).

Fama (1998) argues that the results for IPOs and SEOs do not imply that

benchmark matching on size and BE/ME is always superior to estimating abnormal

returns, he also says that all methods for estimating abnormal returns are to bad-model

problems, and no method is likely to minimize bad-model problems for all classes of

events. The important general message from the IPO-SEO results is one of caution: two

approaches that seem closely related (both attempt to control for variation in average

returns related to size and BE/ME) can produce much different estimates of long-term

abnormal-returns.

2-5-1-1-3 Mergers

Asquith (1983) and Agrawal et al. (1992) find negative abnormal returns for

acquiring firms up to five years following merger announcement. Using a comprehensive

sample of mergers for 1960-1993, Mitchell and Stafford (1997) also find negative long-

term abnormal returns for acquiring firms. They find that the three-year post event buy-

and-hold return for equal-weighted acquiring firms is on average 4% lower than for

portfolios matched to acquiring firms on size and BE/ME. In economic terms, this is not

a dramatic anomaly. For formal inferences, Mitchell and Stafford (1997) estimate three

factor model on the monthly returns on a rolling portfolio that includes firms with

acquisitions during the preceding three years, when the acquirers are equal-weighted, the

intercept, that is, the average monthly abnormal returns for the three years after a merge

is -0.25% per month (-25 basis points, t = -3.49). When acquiring firms are value-

weighted, the intercept of the equation drops to -0.11% per month (t = -1.55). Thus, if

there is an anomaly, it is more important for smaller acquiring firms. They show that

46

abnormal post-announcement average returns to acquiring firms are limited to mergers

financed with stocks, that is, mergers that are also SEOs. When mergers are financed

without issuing stocks, the negative abnormal post-event returns disappear. This suggests

that there is no distinct merger anomaly; any merger anomaly may be the SEO anomaly

in disguise.

Desai and Jain (1997) and Ikenberry et al. (1996) find that for 17-year 1975-1991

period, stock splits are followed by positive abnormal returns of about 7% in the year

after the split. Abnormal returns are calculated relative to benchmarks that control for

size, BE/ME, and, in Desai and Jain, past one-year return. To test whether such an

anomaly is real or the sample-specific results of chance is to examine a different sample

period; Fama et al. (1969) examine splits during the 33-year 1927-1959 period. They find

no drift in cumulative abnormal returns during the 30 months following splits, since the

split anomaly fails the out-of sample test provided by Fama-Fisher-Jensen-Roll, it seems

reasonable to conclude that the 1975-1991 anomaly is not real, unless the market has

recently becomes inefficient (Fama 1998).

Lakonishok and Vermaelen (1990) examine long-term returns following self-

tender offers (tenders by firms for their own shares) during 1962-1986 period. Ikenberry

et al. (1995) examine long-term returns following share repurchases during the 1980-

1990 period. Mitchell and Stafford (1997) study both self-tenders and repurchases for the

1960-1993 period, they find that three-year post-event buy-and-hold abnormal returns,

computed relative to matching portfolios that control for size and BE/ME, are 9% for

self-tenders (475 events) and 19% for much larger sample of 2542 repurchases. When

they estimate the three factor regression for monthly returns on an equal-weight portfolio

that contains all self-tenders and repurchases in the last three years, however, the average

abnormal return is 0.11% per month (t = 1.62). Any hint of significance, economic or

statistical, disappears entirely when the stocks in the rolling portfolio are value-weighted.

The intercept for the value-weight portfolio of self –tenders and repurchases is -0.03% (-3

47

basis point per month, t = -0.34). Fama (1998) argues that according to theses results,

there is no share repurchase anomaly. He adds, that two apparently similar methods for

estimating abnormal returns, (i) a matching portfolio control for size and BE/ME and (ii)

an asset pricing regression that adjusts for sensitivity to risk factors related to size and

BE/ME, produce somewhat different results, which illustrates that estimates of long-term

abnormal returns can be sensitive to apparently small changes in technique.

Dharan and Ikenberry (1995) find that during the 1962-1990 period, stocks that

newly list on the NYSE, or move from Nasdaq to Amex, have negative post-listing

abnormal returns. When returns are risk-adjusted using matching portfolios formed on

size and BE/ME, the three-year average abnormal return is -7.02%. The t-statistic for this

CAR is -2.78, but this is without a full adjustment for the correlation of abnormal returns

across firms. Moreover, Dharan and Ikenberry show that the negative post-listing

abnormal returns are limited to firms below the NYSE-Amex median in size. Thus, once

again, an apparent anomaly is limited to small stocks. Mitchell and Stafford (1997) offer

concrete perspective on how significance levels can be overstated because of the failure

to adjust for the correlation across firms of post-event abnormal returns. Using the three

factor model, they calculate the standard deviations of abnormal returns for portfolios of

firms with an event during the most recent 36 months. The proportion vary somewhat

through time and cross their three event classes (mergers, share repurchases, and SEOs),

but on average the covariances of event-firm abnormal returns account for about half the

standard deviation of the event portfolio’s abnormal return. Thus, if the covariances are

ignored, the standard error of the abnormal portfolio return is too small by about 50%.

This estimate need not apply in fact to the exchange listing of Dharan and Ikenberry

(1995), but it suggests that a full adjustment for the cross-correlation of post-listing

abnormal returns could cause the statistical reliability (t = -2.78) of their -7.02% post-

event three-year CAR to disappear.

Dharan and Ikenberry’s explanation of their negative post-listing abnormal

returns is that firms are opportunistic, and they list their stocks to take advantage of the

market’s overreaction on their recent good times. This explanation seems shaky,

48

however, given that any overreaction to past performance has already occurred and will

soon be reversed. Moreover, standard signaling theory (e.g., Ross, 1977) does not predict

that firms will incur costs to make a false signal whose price effects are soon obliterated.

On the contrary, since listing involves costs, it should be a signal that the firm is under-

valued.

Michaely et al. (1995) find that during the 1964-1988 period, firms that initiate

dividends have positive abnormal stock returns for three years after the event, and firms

omitting dividends have negative abnormal returns. For the same sample, Brav (1997)

finds that the three-year post-event abnormal return following initiations disappears with

benchmarks that control for size and BE/ME. Michaely et al. (1995) show that the

negative three-year abnormal returns following omissions, confirmed by Brav (1997), are

largely concentrated in the second half of their 1964-1988 sample period. All this

suggests that inferences about long-term returns following changes in dividends should

probably await an out-of-sample test. The finding that stock prices under-react to

dividend announcements is suspect on other grounds. It seems reasonable that

underreaction would occur because the market underestimates the information in

dividends about future earnings. However, from Watts (1973) to Benartzi et al. (1997),

there is little evidence that changes in dividends predict changes in earnings.

2-5-1-1-8 Spinoffs

Cusatis et al. (1993) study the post-events returns of spinoffs1 and their parents for

the 1965-1988 period. The benchmarks are firms matched to the event firms on size and

industry, and abnormal returns are buy-and-hold abnormal returns (BHARs). Both

parents and spinoffs have positive abnormal returns in the three years after the event. The

1

A pure spinoff is defined as a tax-free, pro-rata distribution of shares of a wholly owned subsidiary to

shareholders. In both the academic literature and the popular press, spinoffs often consist of various types

of distributions of common stock in other companies. These alternative types of distributions include partial

as well as full distributions of stock in subsidiaries, taxable and nontaxable distributions, court-ordered as

well as voluntary stock distributions, distributions of common shares in publicly traded companies as

opposed to subsidiaries, and return of capital distributions. In some cases, a specialized stock distribution

such as split offs, and even stock sales such as equity carve outs, are referred to as spinoffs.

49

abnormal returns are, however, limited to event firms (parent and spinoffs) acquired in

mergers. The conclusion is that the market does not properly assess the increased

probability of takeover (and the attended buyout premiums) following spinoffs, the t-

statistic for the three-year BHARs for spinoffs range from 0.58 to 2.55, hardly

overwhelming. Moreover, in calculating the t-statistics, the BHARs of the event firms are

assumed to be independent. It would not take a large adjustment for cross-correlation to

produce t-statistics that suggest no real anomaly.

Ikenberry and Lakonishok (1993) examine stock returns following proxy

contests2 during the 1968-1987 period. They find negative post-event abnormal return

relative to benchmarks that control for market β and size. In the results for all proxy

contests, the post-event abnormal returns are not statistically reliable. The negative post-

event returns are only statistically reliable for 50-odd proxy contests in which the

dissidents with board representation. Since this result is not an ex ante prediction, the

weak evidence for the overall sample seems more relevant, and it does not suggest a

reliable anomaly.

From the previous review, it appears that if reasonable change in the method of

estimating abnormal returns causes an anomaly to disappear, the anomaly is fragile, and

it is reasonable to suggest that it is an illusion. Included in this category are IPOs, SEOs,

self-tenders, share repurchases, and dividend initiations, other long-term return anomalies

are economically or statistically marginal. The negative post-event abnormal returns to

acquiring firms in mergers are economically small. For exchange listing, spinoffs, and

proxy contests, a full correction for the cross-correlation of long-term post-event

abnormal returns could easily reduce them to former anomalies. Whenever value-weight

returns are examined, apparent anomalies shrink a lot and typically become statistically

unreliable. At a minimum, this suggests that anomalies are largely limited to small stocks.

Fama (1998) presents a reasonable alternative explanation, he says that small stocks are

just a sure source of bad-model problems. Small stocks always pose problems in tests of

2

The proxy contest is one mean by which shareholders may exercise the control authority embedded in

their equity claims.

50

asset pricing models, so they are prime candidates for bad-model problems in tests of

market efficiency on long-term returns.

Calendar effects in stock market returns have puzzled financial economists for

several years. EMH emphasizes that seasonal patterns should not exist or should only be

minor, since their existence implies the possibility of obtaining abnormal returns by

making timing strategies. There are many calendar effects exist in the literature, among

them the monthly or January effect, the day-of-the-week effect, the trading month effect,

holiday effect, and more recently the Halloween effect. Thaler (1987a, 1987b) provides

an early and partial survey, while Mills and Couts (1995) provide more recent references.

Rozeff and Kinney (1976) were the first to document evidence of higher mean

returns in January as compared to other months. Using NYSE stocks for the period 1904-

1974, they find the average return for January to be 3.48% compared to only 0.42% for

the other 11 months. Later studies document the effect persists in more recent years,

Bhardwaj and Brooks (1992) for 1977-1986 and Eleswarapu and Reinganum (1993) for

1961-1990. The effect has been found to be present in other countries as well (Gultekin

and Gultekin, 1983). The January effect has also been documented for bonds by Chang

and Pinegar (1986). Maxwell (1998) shows that the bond market effect is strong for non-

investment grade bonds, but not for investment grade bonds. More recently, Bharba,

Dhillon, and Ramirez (1999) document a November effect, which is observed only after

Tax Reform Act of 1986. They also find that the January effect is stronger since 1986.

A number of hypotheses have been suggested in the literature to explain the

January effect, among which the tax-loss-selling hypothesis has received most of the

recognition. These hypotheses can be grouped into three broad categories. First, are

hypotheses centered around measurement problems, particularly studies that focus on the

relation between market capitalization and seasonality in the stock market, the argument

here is simply that excess return on small firms is postulated to be either a deception

caused by poor measurement of the returns on these firms or a compensation for extra

51

risk investors bear by holding these stocks (Banz 1981; Brown et al. 1983; Keim 1983;

Reinganum 1983; and Roll 1983).

The second category comprises hypotheses related to buying pressure at the

beginning of the year. These hypotheses provide reasons why individuals and institutions

have a grater incentive to sell some of their holdings (particularly small firm’s stocks) at

the end of the year and repurchase these holdings at the beginning of the following year.

Individual investors have more idle cash (from year-end bonuses, holiday gifts and tax-

loss-selling) at the beginning of the year, which they want to put into the market (Branch

1977; Brown et al. 1983; Constantinides 1984; Chan 1986; Ritter 1988; and Sias and

Starks 1997). Meanwhile, institutional mangers engage in a lot of portfolio rebalancing

and, or, (window dressing) activities near the end of the year, which makes trades in

January of the following year reversals of these activities ( Lakonishok and Smidt 1986;

Haugen and Lakonishok 1987; Ritter and Shopra 1989; and Lakonishok et al. 1991).

Third are hypotheses related to the seasonality or the timing of information release,

because of the coincidental clustering of the calendar year-end and the tax year-end in the

USA; Rozeff and Kinney (1976) observed that January sees the release of an unusual

amount of accounting information, thus speculating that seasonality is perhaps associated

with accounting news. The seasonal difference in the information about the underlying

stocks has been examined by some researchers and considered as an alternative

explanation for the January effect (Brauer and Chang 1990).

Many authors try to obtain evidence for the tax-loss-selling hypothesis by

examining stock return patter in countries with different tax codes and tax-year-ends.

Significant seasonality was observed in international stock markets but it was not

persistent through time in many markets, providing mixed evidence on the tax-loss-

selling hypothesis. For example, Brown et al. (1983) find that while Australia has similar

tax law to the USA, it has a July-June tax year, which gives rise to December-January

and July-August seasonal. While the July-August seasonal is consistent with the tax-loss-

selling hypothesis, the authors conclude that the evidence is inconsistent with tax-loss-

selling hypothesis because of the existence of December-January seasonality. Hillier and

Marshall (2002) reach the same conclusion and reject the tax-loss-selling hypothesis for

UK stocks. Using value weighted stock markets in major industrial countries, Gultekin

52

and Gultekin (1983) find evidence for a persistent (though generally less significant than

that in the USA) January effect in most of the countries, which was construed as support

for the tax-loss-selling hypothesis.

The first study of weekend effect in security market appeared in the Journal of

Business in 1931, written by a graduate student at Harvard named M.J. Fields. He was

investigating the conventional Wall Street wisdom at the time that “the unwillingness of

traders to carry their holdings over the uncertainties of a week-end leads to a liquidation

of long accounts and a consequent decline of security prices on Saturday” (Fields, 1931,

p.415). Fields examines the pattern of the Dow Jones Industrial Average (DJIA) for the

period 1915-1930 to see if the conventional wisdom was true, he compares the closing

price of the DJIA for Saturday with the mean of the closing prices on the adjacent Friday

and Monday. He finds, in fact, that prices tend to rise on Saturdays, for the 717 weekends

he studies, the Saturday’s price was more than $ 0.10 higher than the Friday-Monday

mean 52 percent of the time, while it was lower only 36 percent of the time.

Cross (1973) studies the returns on the S&P 500 over the period 1953-1970, he

finds that the index rises on 62 percent of the Fridays, but only 39.5 percent of the

Mondays. French (1980) analysis daily returns of stocks for the period 1953-1977 and

finds that there is a tendency for returns to be negative on Mondays whereas they are

positive on the other days of the week; he notes that these negative returns are “caused

only by the weekend effect and not by a general closed-market effect”. A trading

strategy, which would be profitable in this case, would be to buy stocks on Monday and

sell them on Friday. Kamara (1997) shows that the S&P 500 has no significant Monday

effect after April 1982, yet he finds the Monday effect undiminished from 1962-1993 for

a portfolio of smaller U.S stocks. Internationally, Agrawal and Tandon (1994) find

significantly negative returns on Monday in nine countries and on Tuesday in eight

countries, yet large positive returns on Friday in 17 of the 18 countries studied, Steely

(2001) finds that the weekend in the UK has disappeared in the 1990s.

Numerous factors, might explain the weekend effect. The most logical

hypothesis-dubbed “calendar time hypothesis” by French (1980) is that, prices should rise

53

somewhat more on Mondays than on other days because the time between the close of

trading on Friday and the close of trading on Monday is three days, rather than the

normal one day between other trading days. Accordingly, Monday returns should be three

times higher than other weekday returns. French offers an alternative, the “trading time

hypothesis”, which states that returns are generated only during active trading and

implies that returns should be the same for every trading day. In any case, neither

hypothesis is consistent with the data, another explanation exist in the literature;

differences in settlement time of transaction; attitudes of certain investor groups; investor

tendency to postpone the announcement of bad news until the weekend so that the market

will have time to absorb the stock; and measurement errors.

In French’s investigations of weekend effects he looks at the price behavior after

holidays and finds no thing special happening. However, in another early study Fields

(1934) finds that the DJIA shows a high proportion of advances the day before holidays.

In this case it takes over 50 years for Fields to be resurrected from obscurity by Ariel

(1985). Ariel looks at the returns on the 160 days that preceding holidays during the

period 1963-1982. For an equal weighted index of stocks he finds that the mean return on

the pre-holidays was 0.529 percent, compared to 0.056 percent on other days, a ratio of

grater than 9 to 1. For a value weighted index the pre-holiday returns average 0.365

percent compared to 0.026 percent on other days, a ratio of greater than 14 to 1. The

differences are both statistically and economically significant. These results were

replicated for the 90-year DJIA series by Lakonishok and Smidt (1987). They obtain an

average pre-holiday return of 0.219 percent, compared to normal daily rate of return of

0.0094 percent, ratio of grater than 23 to 1. Many suggestions proposed to explain

holidays’ effects, such as, differences in settlement time of transactions; investors’ good

mood before holidays; and other psychological reasons.

Ariel (1987) examines the pattern of returns within months. For the period 1963-

1981 he divides months into two parts, the first part starting with the last day of the prior

54

month, he then compares the cumulative returns for the two periods using both equal-

weighted and value-weighted indices. The return for the latter half of the month is

negative, all the returns for the period occur in the first part of the month. This result has

been replicated by Lakonishok and Smidt (1987). Using 90-year series for the Dow, they

find that the return for the four days around the turn of the month, starting with the last

day of the prior month, is 0.473 percent (the average return for a four-day period is

0.0612 percent). Also, the turn-of-the-month four-day return is grater than the average

total monthly return which is 0.35 percent. End-of-the-month increases in purchasing

power due to salaries; and higher frequency of announcements of companies’ profit

during the first fortnight of the month, have been suggested as explanations for this

seasonality.

A more recent calendar effect is that described in Bouman and Jacobsen (2002).

They find that the return to stocks in 37 countries can be explained almost totally as a

result of what they term the Halloween Indicator. They find that the hypothesis of zero

mean return to equities in months May-October cannot be rejected. Thus, the old stock

market adage “Sell in May and Go Away, Don’t Come Back Till St.Leger’s Day”,

St.Lerger’s day being 2 October, but with the adage generally being seen as a references

to the running of the St.Leger race of Doncaster in late September, would seem to be

vindicated. They hypothesize that the cause of this may be the taking, by the general

economically active public and brokerage community, of significant holidays in summer

period. This has the effect of depressing economic and in particular stock market activity

in the summer period. Maberly and Pierce (2003) examine the robustness of the

Halloween effect to alternative model specifications for Japanese equity prices. They find

that the Halloween effect is concentrated in the period prior to the introduction of Nikkei

225 index in September 1986, while the Halloween effect disappear after 1986. In

addition, Maberly and Pierce (2004) find that Bouman and Jacobsen results are not robust

to alternative model specifications for U.S equity prices.

55

2-5-1-3-1 Small firm effect

Banz (1981) publishes one of the earliest articles on the “small-firm effect” which

is also known as the “size-effect”. His analysis of the 1936-1975 period reveals that

excess returns would have been earned by holding stocks of low capitalization

companies. Supporting evidence is provided by Reinganum (1981) who reports that the

risk adjusted annual return of small firms was grater than 20 percent. If the market was

efficient, one would expect the prices of stocks of these companies to go up to a level

where the risk adjusted returns of future investors would be normal, but this did not

happen.

The Value-Line organization divides the firms into five groups and ranks them

according to their estimated performance based on publicly available information. Over a

five year period starting from 1965, returns to investors correspond to the rankings given

to firms. That is, higher ranking firms earned higher returns. Several researchers (e.g.

Stickel (1985)) find positive risk-adjusted abnormal (above average) returns using Value-

Line rankings to form trading strategies, thus challenging the EMH.

Harris and Gurel (1986) and Shleifer (1986) find a surprising increase in share

prices (up to 3 percent) on the announcement of stock’s inclusion into the S&P 500

index. Since in an efficient market only information should change prices, the positive

stock price reaction appears to be contrary to the EMH because there is no new

information about the firm other than its inclusion in the index.

Few would argue that sunshine puts people in a good mood. People in good mood

make more optimistic choices and judgments, Saunders (1993) shows that the New York

Stock Exchange index tends to be negative when it is cloudy. More recently, Hirshleifer

and Shumway (2001) analyze data for 26 countries from 1982-1997 and find that stock

56

market returns are positively correlated with sunshine in almost all of the countries

studied. Interestingly, they find that snow and rain have no predictive power.

The last two decades have witnessed an onslaught against the efficient market

hypothesis. Yet as Roll (1994) observes, it is remarkably hard to profit from even the

most extreme violations of market efficiency. Stock market anomalies are only too often

chance events that do not persist into the future. As Fama says “consistent with market

efficiency hypothesis that the anomalies are chance results, apparent overreaction to

information is about as common as underreaction, and post-event continuation of pre-

event abnormal returns is about as frequent as post event reversal. Most important,

consistent with the market efficiency prediction that apparent anomalies can be due to

methodology, most long-term return anomalies tend to disappear with reasonable

changes in technique” (Fama, 198, p. 283). The importance of the efficient markets

hypothesis is demonstrated by the fact that apparently profitable investment opportunities

are still referred to as “anomalies”. The efficient market model continues to provide a

framework that is widely used by financial economists.

The greatest stir in academic circles has been created by the results of volatility

tests. These tests are designed to test for rationality of market behaviour by examining the

volatility of share prices relative to the volatility of the fundamental variables that affect

share prices. The first two studies applying these tests were by Shiller (1981) and LeRoy

and Porter (1981). Shiller tests a model in which stock prices are the present discounted

value of future dividends. LeRoy and Porter use a similar analysis for the bond market.

These studies reveal significant volatility in both the stock and bond markets.

Fluctuations in actual prices greater than those implied by changes in the fundamental

variables affecting the prices are inferred by Shiller as being the result of fads or waves of

optimistic or pessimistic market psychology. Schwert (1989) tests for a relation between

stocks return volatility and economic activity; he finds increased volatility in financial

asset returns during recessions which might suggest that operating leverage increases

during recessions; he also finds increased volatility in periods where the proportion of

57

new debt issues to new equity issues is larger than a firm’s existing capital structure. This

may be interpreted as evidence of financial leverage affecting volatility.

However neither of these factors plays a dominant role in explaining the time-

varying volatility of the stock market. The volatility tests of Shiller spawned a series of

articles. The results of excess volatility in the stock market have been confirmed by

Cochrane (1991), West (1988), Campbell and Shiller (1987), Mankiw, Romer, and

Shapiro (1985). The tests have been criticized, largely on methodological grounds, by

Ackert and Smith (1993), Marsh and Merton (1986), Kleidon (1986) and Flavin (1983).

The empirical evidence provided by volatility tests, suggests that movements in stock

prices cannot be attributed merely to the rational expectations of investors, but also

involves an irrational component. The irrational behavior has been emphasized by

Shleifer and Summers (1990) in their exposition of noise trading.

1. They posit that there are two types of investors in the market: (i) rational

speculators or arbitrageurs who trade on the basis of information and (ii) noise traders

who trade on the basis of imperfect information. Since noise traders act on imperfect

information, they will cause prices to deviate from their equilibrium values. It is

generally understood that arbitrageurs play the crucial role of stabilizing prices. While

arbitrageurs dilute such shifts in prices, they do not eliminate them completely. Shleifer

and Summers assert that the assumption of perfect arbitrage made under EMH is not

realistic. They observe that arbitrage is limited by two types of risk: (a) fundamental risk

and (b) unpredictability of future resale price. Given limited arbitrage, they argue that

securities prices do not merely respond to information but also to "changes in

expectations or sentiments that are not fully justified by information". An observation of

investors’ trading strategies (such as trend chasing) in the market provides evidence for

decision making being guided by "noise" rather than by the rational evaluation of

information. Further support is provided by professional financial analysts spending

considerable resources in trying to predict both the changes in fundamentals and also

possible changes in sentiment of other investors. Black (1986) also argues that noise

traders play a useful role in promoting transactions (and thus, influencing prices) as

informed traders like to trade with noise traders who provide liquidity. So long as risk is

58

rewarded and there is limited arbitrage, it is unlikely that market forces would eliminate

noise traders and maintain efficient prices.

In a market consisting of human beings, it seems logical that explanations rooted

in human and social psychology would hold great promise in advancing our

understanding of stock market behavior. More recent research has attempted to explain

the persistence of anomalies by adopting a psychological perspective. Evidence in the

psychology literature reveals that individuals have limited information processing

capabilities, exhibit systematic bias in processing information, are prone to making

mistakes, and often tend to rely on the opinion of others. The damaging attacks on the

assumption of human rationality have been spearheaded by Kahneman and Tversky

(1986) in their path breaking article on prospect theory. The findings of Kahneman and

Tversky have brought into question expected utility theory which has been used

descriptively and predictively in the finance and economics literature. They argue that

when faced with the complex task of assigning probabilities to uncertain outcomes,

individuals often tend to use cognitive heuristics, while useful in reducing the task to a

manageable proportion, these heuristics often lead to systematic biases.

Using simple decision tasks, Kahneman and Tversky are able to demonstrate

consistent decision inconsistencies by manipulating the decision frame. While expected

utility theory would predict that individuals would evaluate alternatives in terms of the

impact on these alternatives on their final wealth position, it is often found that

individuals tend to violate expected utility theory predictions by evaluating the situation

in terms of gains and losses relative to some reference point. The usefulness and validity

of Kahneman and Tversky's propositions have been established by several replications

and extensions for situations involving uncertainty by researchers in the fields of

accounting, economics, finance, and psychology. Rabin and Thaler (2001) show that

expected utility theory’s explanation of risk aversion is not plausible by providing

examples of how the theory can be wrong and misleading. They call for a better model of

describing choice under uncertainty. It is now widely agreed that the failure of expected

59

utility theory is due to the failure to recognize the psychological principles governing

decision tasks.

The literature on cognitive psychology provides a promising framework for

analyzing investors' behavior in the stock market, by dropping the stringent assumption

of rationality in conventional models; it might be possible to explain some of the

persistent anomalous findings. For example, the observation of overreaction is consistent

with the finding that subjects, in general, tend to overreact to new information (and

ignore base rates). Also, agents often allow their decision to be guided by irrelevant

points of reference, a phenomenon discussed under "anchoring and adjustment". Shiller

(1984) proposes an alternate model of stock prices that recognizes the influence of social

psychology; he attributes the movements in stock prices to social movements. Since there

is no objective evidence on which to base their predictions of stock prices, it is suggested

that the final opinion of individual investors may largely reflect the opinion of a larger

group. Thus, excessive volatility in the stock market is often caused by social "fads"

which may have very little rational or logical explanation.

Research into investor behavior in the securities markets is rapidly expanding

with very surprising results, again, results that are often counter to the notion of rational

behavior. Hirshleifer and Shumway (2001) find that sunshine is strongly correlated with

daily stock returns. Using a unique data set of two years of investor behavior for almost

the entire set of investors from Finland, Grinblatt and Keloharju (2001) find that distance,

language, and culture influence stock trades. Huberman and Regev (2001) provide an

example of how and not when information is released can cause stock price reactions.

They study the stock price effect of news about a firm developing a cure for cancer.

Although the information had been published a few months earlier in multiple media

outlets, the stock price more than quadrupled the day after receiving public attention in

the New York Times. The efficient market view of prices representing rational valuation

of fundamental factors has also been challenged by Summers (1986), who views the

market to be highly inefficient. He proposes that pricing should comprise a random walk

plus a fad variable, the fad variable is modelled as a slowly mean-reverting stationary

process. That is, stock prices will exercise some temporary aberrations, but will

eventually return to their equilibrium price levels.

60

One may argue that market mechanisms may be able to correct the individual

decision biases, and thus individual differences may not matter in the aggregate.

However, the transition from micro behavior to macro behavior is still not well

established. For example, in their study of price differences among similar consumer

products, Pratt, Wise and Zeckhauser (1979) demonstrate the failure of the market to

correct individual biases. All arguments aside, the stock market crash of 1987 continues

to be problematic for the supporters of EMH, any attempt to accommodate a 22.7 percent

devaluation of the stocks within the theoretical framework of EMH would be a

formidable challenge. It seems reasonable to assume that the decline did not occur due to

a major shift in the perceived risk or expected future dividend. The crash of 1987

provides further credence to the argument that the market includes a significant number

of speculative investors who are guided by "non-fundamental" factors. Thus, the

assumption of rationality in conventional models needs to be rethought and reformulated.

According to International Finance Corporation (IFC) classification criteria, an

emerging stock market is the one located in a developing country as defined by the World

Bank’s GNP per capita criterion. IFC, a leading compiler of emerging market returns,

considers the size (as measured by market capitalization) and liquidity (as measured by

turn over) in classifying a market as emerging and in deciding to include the securities in

the market in its Emerging Market Data Base (EMDB). In addition, inclusion in the

EMDB is affected by the industry in which a company operates; the IFC attempts to

provide broad coverage of industries important within the market. Thus, a smaller less

liquid security might be included, whereas a larger, more liquid one is excluded if the

smaller less liquid security represents a particular industry that would otherwise be under-

represented. Although the world (capital) market is neither fully integrated nor

completely segmented, there is no doubt that there is increasing interdependence among

its segments. Presently, the fully diversified portfolio must consist of a significant portion

of foreign securities to mitigate unnecessary risk. Global diversification depends on the

correlations among countries. According to the World Bank, a significant degree of

correlation between portfolio equity flows and the emerging stock markets development

61

indicators has been established. These correlations are enhanced by cross-border

investments and improving technology.

Moreover, the benefits of diversification are no longer sufficiently achieved

through developed markets. While emerging markets in many respects differ from the

developed markets, there is still substantial diversity among emerging stock markets in

terms of institutional infra structure, market size and liquidity. Emerging markets,

especially in the last decades, provide diversification benefits at an increasing rate. There

may be several factors to explain the unprecedented development of emerging markets,

but it is possible that economic reforms in these markets have left to a rapid increase in

equity flows from the industrial to the developing ones. The more established emerging

markets of Asia and Latin America and the new capital markets in Eastern Europe,

MENA will play a positive role in this process.

Most of the studies on EMH are conducted on the world’s largest stock markets;

the USA, Japan and Europe. In recent years, efficiency in emerging markets has been

investigated widely. Researchers have focused on whether these markets are

informationally efficient or whether anomalies exist. Barnes (1980) indicates that the

Kuala Lumpur stock market is inefficient. While Panas (1990) could not reject market

efficiency for Greece. Campbell (1995) examines 20 emerging markets in Latin America,

Asia, Middle East, Europe, and Africa. He finds that returns in these emerging markets

are more predictable than returns in developed markets and returns are influenced by

local rather than global information. Moreover, Antoniou, Ergul, and Holmes (1997)

study the Istanbul Stock exchange and find it to be inefficient in the early times and

efficiency improves as the country starts liberalization and deregulation. Dickinson and

Muragu (1994) find the Nairobi stock market to be efficient. Urrutia (1995), using the

variance ratio test, rejects the RWH for the Latin American emerging equity markets of

Argentina, Brazil, Chile, and Mexico, whereas the runs test indicates weak form

efficiency.

In contrast, Ojah and Kermera (1999) find that Latin American equity returns

follow a random walk and are generally weak-form efficient. Grieb and Reyes (1999) re-

examine the random walk properties of stocks traded in Brazil and Mexico using the

variance ratio test and conclude that index returns in Mexico exhibit mean reversion and

62

a tendency toward random walk in Brazil. In addition, predictable variations in the

emerging market returns have been documented in Bekaert (1995), Harvey (1995b,

1995c) and claessens et al. (1995). Buckberg (1995) also finds evidence of predictability

in emerging markets and rejects the hypothesis that lagged price information cannot

predict future prices. A low form of predictability in the emerging markets can be viewed

as some form of reward for risk taking. The conclusion is that predictability is more

likely to be influenced by local information and that lower degrees of predictability in

emerging markets can be viewed as a reward for added risk taking.

Furthermore, Bailey et al. (1990) present evidence that stock prices of several

Asian markets do not follow random walks. Bessembinder and Chan (1995) examine if

market participants in the emerging markets take advantage of profit opportunities that

may be present due to deviations from the random-walk model. They find that technical

rules have some predictive power, but they say technical signals from US markets have

stronger forecasting power. Haque et al. (2001) investigate the stability, volatility, risk

premiums and persistence of volatility of seven Latin American emerging markets and

find that Latin American markets have shown remarkable performance using return to

risk measures; predictability seems mixed and has volatility clustering with shocks that

decay with time. Haque et al. (2004) study of 10 Asian stock markets suggests that 8 out

of 10 Asian markets have returns that are stable over time. The predictability tests

suggest most of the Asian emerging markets to be predictable. The non-parametric runs

test for weak form of market efficiency decisively rejects the hypothesis for weak form

efficiency for all the Asian markets.

63

3- The case of Arab stock markets

The markets of the Middle East have buzzed with activity since the beginning of

western civilization. Yet, these countries of the region that gave the world the basis of

modern commerce have been relative latecomers to global financial markets. Where the

Middle East is concerned, political and religious issues continue to dominate the

headlines. What differentiates the Middle East markets from other emerging markets is

the region’s heterogeneity. The financial sector in Middle Eastern countries is dominated

by commercial banks. The security markets in these countries are relatively small despite

the fact that the region contains some of the developing world’s largest institutional

investors in international markets. Foreign participation, even in the government bond

markets, is limited in most countries. Similarly, there have been few direct placements of

Middle Eastern equities on foreign markets.

Moreover, the use of market-based risk management instruments by countries in

the region has been extremely narrow despite the relatively limited degree of export

diversification. While there are considerable differences across countries in the

importance of equity markets, the supply of corporate securities remains generally limited

both in absolute terms and relative to the size of the economies. This reflects several

factors that have constrained the demand for and the supply of equities, including the

closed, family-owned nature of many companies in the region. Moreover, in several

countries public sector enterprises have continued to play a dominant role in a wide range

of economic activities. The number of effectively quoted companies thus has been

relatively small and the markets have, in general, remained thin.

As mentioned in the previous chapter, most of the studies on EMH are conducted

on the world’s largest stock markets. In recent years, efficiency in emerging markets has

been investigated widely. Very few studies, target countries from the Middle East region,

most of them concentrated on return predictability and markets integration and linkages.

In addition, most of these studies are usually focused on their individual or a small set of

countries for a short horizon.

64

One early study undertaken by Gandhi et al. (1980) focuses on the Kuwaiti stock

market and attempts to measure its efficiency through the use of some empirical tests.

The authors find a high correlation in the market index and conclude that the market is

inefficient. Bulter and Malaikah (1992) examine individual stock returns in the Kuwaiti

and Saudi Arabian markets over the second half of 1980s. Their results indicate market

inefficiency in both markets, but significantly more in the Saudi market. However, Al-

loughani (1995) tests the weak form of EMH in the Kuwaiti stock market by using

various methods, both traditional and advanced. The author finds that when using

traditional methods, the results provide evidence of weak form efficiency, while when

using more recent methods, he obtains opposite results in the sense that the evidence

clearly indicates market inefficiency. Another study investigates the Kuwaiti stock

market has been done by Al-loughani and Moosa (1997), since they test the efficiency of

the Kuwaiti stock market by using a set of moving averages of different lengths. The

results obtained by the authors indicate market inefficiency.

Additionally, El-Erian and Kumar (1995) examines the RWH in emerging

markets by choosing two countries from the Middle East region, Jordan and Turkey, and

three other emerging markets from different regions. The study finds that there is serial

dependence among the day-to-day price changes in the stock markets of Jordan and

Turkey, indicating that the random walk model does not hold for these markets. Similar

results, obtained by Omran and Farrar (2002) who reject the RWH for 5 Middle Eastern

countries, while Abraham et al. (2002) reject both RWH and weak form efficiency for

three Gulf equity markets, Saudi Arabia, Kuwait, and Bahrain, when they use the

observed indices, while they cannot refuse them when they correct indices for infrequent

trading. Rao and Shankaraiah (2003) find evidence that Bahraini stock market is efficient

at the weak form level during the second half of 1990s. Hakeim and Neaime (2002) find

evidence that 4 MENA markets Egypt, Jordan, Morocco, and Turkey are mean reversion.

Limam (2003) using weekly data, finds long range dependence in eight Arab stock

markets, while Omet et al. (2002) reject the weak form efficiency for the Jordanian stock

market.

On the other hand, Moustafa (2004) finds that 40 stocks out of the 43 including in

the UAE index are random, using daily data for the period October 2, 2001 through

65

September 1, 2003, while Haque et al. (2004) examine the stability, predictability,

volatility, time varying risk premiums and persistence of shocks to volatility for 10

MENA stock markets; including 6 Arab stock markets. They find that 8 out of 10 markets

show evidence of volatility clustering, but in 8 MENA stock markets; the shocks are not

explosive. Whereas one market shows positive and significant time varying risk

premiums. They conclude that MENA equity markets are where investors may find a

good return for the investments, since the correlation is found to be low, which provide

investors with the opportunity for diversification. Moreover, Al-loughani (2000) studies

the relation between large stock and small stock returns in the Kuwaiti stock markets, he

finds further evidence on the informational inefficiency of the Kuwaiti market, since in

short term large stocks provide a lead in the bull phase. Dahel and Laabas (1999)

examine the behavior of stock prices in 4 GCC markets Bahrain, Kuwait, Oman, and

Saudi Arabia using weekly data from September 1994 to April 1998. They find that

Kuwaiti market to be efficient while for the other three markets; weak form of the EMH

was rejected based on regression of return test. However, when the sample is split into

two, the efficiency hypothesis is not rejected for the second sub period in two of the

markets and only by a small margin in the case of the Saudi Arabia.

Few researches concentrate on the volatility structure of Arab stock markets. For

instance, Dahel (2000) investigates whether Arab stock markets are characterized by

excessive volatility of returns. His study includes in addition to 8 Arab stock markets,

two emerging and three developed markets. He finds that Arab markets exhibit the lowest

level of volatility of returns compared to other emerging and developed markets, and they

were not affected by international financial crisis. In addition, Arab stock markets are

characterized by low correlations with each other and with international markets. While

Hammoudeh and Choi (2004) investigate the volatility of the decomposed stock returns

of members of the GCC stock markets into permanent and transitory components using

the unobserved-component model with Markov-switching heteroskedasticity (US-MS

model). They find that the GCC stock markets vary in terms of sensitivity to the

magnitude of return volatility and the duration of volatility. Oman and Saudi Arabia stock

markets exhibit extra volatility sensitivity during fad times of the other GCC stock

markets, while Kuwaiti, Bahraini, and Saudi Arabia have longer duration of volatility

66

during fad times. Moreover, they find that all GCC returns move in the same direction

whether in terms of total return, fundamentals or fads under both volatility regimes. They

find also that the correlations of the stock returns and their components with each other

and with oil price return are also weak, suggesting that country particularities in addition

to the oil price return influence the stock component returns.

Very few researchers also investigate EMH in Arab stock markets indirectly by

examining whether anomalies exist or not Such as calendar effects. For instance, Aly et

al. (2004) find no evidence for Monday effect in the Egyptian stock market, using daily

data for market index during the period April 26, 1998 to June 6, 2001, while Al-saad and

Moosa (2005) find that seasonality is found to take the form of a July effect, as opposed

to the better-recognized January effect for the Kuwaiti stock market. They conclude that

the finding is attributed to the “summer holiday effect”. Whereas, Maghayereh (2003)

finds no evidence of monthly seasonality and January effect for the Jordanian stock

market during the period January 1994 to December 2002. Finally, Al-loughani (2003)

documents mixed evidence on seasonality regarding the Kuwaiti stock markets.

Another line of research examines the properties and characteristics of the Arab

stock markets and the prospects and implications of enhanced financial liberalization in

the region. It also explores whether these markets can offer international investors unique

risk and returns characteristics to diversify international and regional portfolios. Darrat et

al. (2000) examine financial integration among three emerging markets in the Middle

East region Jordan, Egypt, and Morocco with the U.S market. Using monthly data for the

period October 1996 to August 1999, they find that according to Johansen-Juselius co-

integration test the Middle East emerging stock markets are segmented globally, but

appear highly integrated within the region. The result also indicates that, the Egyptian

stock market is a dominant force driving other markets in the region. In the same area,

Assaf (2003) using vector auto-regression, investigates the dynamic interactions among

stock market returns for 6 GCC countries. His results indicate that there is substantial

evidence of interdependence and feed back effects among GCC stock markets, the results

also indicate that Bahrain is the dominant market while Saudi Arabia shows slow process

in responding to shocks originated in other markets.

67

Moreover, Mohd and Hassan (2003) find long term relationship between 3 GCC

stock markets Kuwait, Bahrain, and Oman. They also find that information on the price

levels is helpful for predicting their changes. In addition to the previous studies, the

international diversification benefits between U.S, Turkish, and Egyptian stock markets

have been investigated by Maneschiold (2005) who finds that long term relationship at

the general index level related to some but not all sub-indices investigated. U.S investors

can obtain diversification benefits at a sub-index level. Neaime (2002) studies the

liberalization and financial integration for 7 MENA stock markets with international

markets. He finds that GCC equity markets still offer international investors portfolio

diversification potentials while other emerging MENA stock markets like those of

Turkey, Egypt, and Morocco and to lesser extent Jordan have matured and are now

integrated with the world financial markets. However, shocks to U.S and UK stock

markets are transmitted to the MENA region but not to the GCC stock markets.

Girard et al. (2003) investigate relationships between market risk premium, time-

varying variance and time-varying covariance in 11 MENA markets and 8 developed

markets. They conclude that MENA capital markets are highly segmented and provide

diversification benefits to the global investors. From his side, Omran (2003) investigates

the impact of real interest rates on stock market activity and liquidity in the Egyptian

stock market. He finds that real interest rates have an impact upon stock market

performance. Finally, Hammoudeh and Aleisa (2004) study the daily relationships among

stock markets of the GCC members, excluding Qatar, and oil prices. They find that the

GCC stock markets are candidates for diversified regional portfolios at the country level,

while only the Saudi market can predict and be predicted by oil prices.

In general, it appears from the previous literature review that relatively less

explored area of research has been on the Arab stock markets. With few studies

undertaken to date, research on these markets has focused on the issue of efficiency as

well as on their integration with international markets. In addition, most of these studies

are concentrated on few markets and in many cases only one market and for short horizon

of time using monthly and weekly data.

68

3-2 The Foundation of Arab stock markets

MENA region have been receiving extensive media coverage and public attention

for various reasons. Many have recognized that most MENA countries have long

abstained from the global trend of further globalization, modernization and political and

economic liberalization. Some claim that the region is facing the reduction in oil wealth

that can no longer act as a cushion or employ the huge population growth, as a result,

Arab countries, as a part of MENA countries, going toward economic and political

reforms and increasing the role of the private sector. It is commonly recognized that the

availability of financial capital market is a prerequisite for the development and

transformation of any nation’s economy. Finding and efficiently managing the scarce

resources depend on the existing of financial institutions, whether they are banks or non-

bank financial institution such as insurance companies, issuing houses and stock markets.

Banks mobilize financial resources from the surplus sector of the economy and channel

such funds to the deficit units of the economy, whereas the stock market is a market

where trading activities for securities take place and its primary function is to allocate

resources to the most profitable investment opportunities.

Most Arab countries have stock markets which considered as emerging markets.

By international standards, Arab stock markets are considered relatively new. Most of

them started operating over the last two decades, while others have been in existence for

much longer but until recently; there level of activity was not significant. In general, there

are significant differences between Arab stock markets characteristics. In this contest,

Arab countries can be divided into two groups regarding natural resources: non oil

countries, and oil countries. Since the second group mainly constitute of the Gulf

Cooperation Council Countries (GCC), which is a customs union that consists of six

members, including four major oil-exporting countries, which are important decision

makers in the Organization of Petroleum Exporting Countries (OPEC). The six members

are Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, and United Arab Emirates (UAE). The

non-OPEC members among them are Bahrain and Oman. This section will discuss the

foundation of Arab stock markets with some economical background for each country

including in this research.

69

• Egypt

Economic reforms in Egypt have faltered due to the post September 11 downturn

in tourism, high Suez Canal tolls, and low level of exports. Little progress has been

achieved in privatizing or reforming the significantly large public sector. Social concerns

have taken precedence as the largest Arab country, with a population of 69 million,

suffers from growing unemployment and the need to maintain subsidies on food, energy,

and other commodities for the large percentage of the poor. Development of the natural

gas export may help the growth of the economy. While investment laws have been

revised to promote foreign investments, between 1998 and 2001, FDI actually fell by 50

percent, from approximately US$ 1 billion to 500 million due to bureaucratic constraints.

Although decreasing state-owned banking sector still holds the majority of the market

share, these banks are characterized by low capitalization, a high percentage of poorly

performing loans, massive overstaffing and stifling bureaucracy. The Egyptian legal code

is complex and often characterized by lengthy delays. Nevertheless, the legal system

protects private property. Regulations and regulatory agencies are influenced by private

interests and government corruption, which cause delays in clearing goods through

customs, arbitrary decision- making, and high market inefficiencies. The top income and

corporate tax are 40 percent. In January 2003, the Egyptian Pound changed from a

pegged to a floating exchange rate.

Egypt has a long history of financial markets, by the late of 1800s; Egypt had a

sophisticated financial structure including a mature stock exchange in both Alexandria

and Cairo. The Egyptian stock market has experienced fundamental changes during four

major periods from 1888-1958, 1959-1971, 1972-1992, and 1992-present. In the earliest

phase, the market was active and growing out at a remarkable rate. By the 1940s, both the

Cairo and Alexandria exchanges were very active, and the combined Egyptian stock

exchange ranked fifth in the world in terms of overall market capitalization. However, in

the second period from 1959-1971, the Egyptian stock market was seriously marginalized

by government intervention and restrictions that left it effectively inoperable. In the third

period (1972-1992), serious attempts were made to revive the failing stock market to no

avail, and the stock exchange continued to stagnate. Finally, in the 1990s (the forth

period), the Egyptian stock market went through a significant revival due to government

70

liberalization policies. The restructuring of financial markets and privatization programs

were key elements in stimulating economic development and capital investment in 1990s.

Major changes in the organization of the Egyptian stock exchange took place in

January 1997 that significantly reformed the stock market. Today, the stock market once

again encompasses the two exchanges at Cairo and Alexandria, both of which are

governed by the same regulatory agency, and share a common trading, clearing and

settlement system. In addition, several important steps have been taken by the Egyptian

government to modernize the stock exchanges. For example, a coherent organization

structure with clear division of authority and responsibilities has been created; a new

state-of-the-art trading, clearing and settlement system conforming to international

standards has been installed; new membership and trading rules have been legislated; and

new arbitration and dispute resolution procedures were developed.

The Capital Market Authority (CMA) was established in 1990s, as the primary

regulatory body for the Egyptian stock exchange, and it is responsible for the issuance of

licenses to all financial intermediaries including the central clearing and depository

company. The CMA is also responsible for the introduction of any laws and regulations

pertaining to the efficiency and transparency of the market. The company Misr Central

Clearing and Depository (MCSD) oversees the clearing and settlement of all securities

transactions. MCSD is a private company whose primary shareholders are 16 banks, 15

brokerage houses and the stock market exchange itself. Together with CMA, these two

agencies work to guarantee that the market functions efficiently and transparency.

Egypt’s recent economic reforms, mainly the successful implementation of a large

privatization program, is often cited as being largely responsible for the rapid growth in

Egyptian stock market activities over the last five years. Finally, the Egyptian stock

market has been included in the International Finance Corporation’s (IFC) composite

stock index since January 1997, with a 1 percent weighting in the overall index.

Furthermore, Morgan Stanley Capital International covers the Egyptian stock market on a

stand alone basis, although it has not yet included Egypt in its benchmark emerging

markets index.

71

Table 3-1

Some Economic Indicators, Egypt

2000 2001 2002 2003 2004

Population (million) 63.305 64.622 65.986 67.313 68.648

GDP (m US$) 97,655 90,285 85,710 81,495 78,491

GDP growth (%) 7.80 -7.55 -5.07 -4.92 -3.69

GDP per capita (US$) 1,543 1,397 1,299 1,211 1,143

Inflation rate (%) 2.7 2.3 2.7 4.2 -

Source: Uniform Arabian Economic Report 2005, Arab Monetary Fund (AMF)

• Jordan

With scarce economic resources, Jordan’s constitutional monarchy has generally

been dependent on foreign loans and aid. Legislative and regulatory reforms under king

Abdallah II allowed Jordan to accede to the WTO, leading to privatization and economic

growth. Although the country faces a heavy dept burden, high unemployment, and the

end of Iraqi-subsidized oil, Jordan can bring back tourism and foreign investment by

working towards a more peaceful and open Middle East. In 2001, its tariff rate was 13.5

percent. However, the inefficient customs pose a bigger hindrance to imports where they

are subject to arbitrary regulations frequent delays. The top income and corporate tax rate

in Jordan are 25 and 35 percent respectively. In 2001, the government consumed 23

percent of GDP.

While the government promotes foreign investments, investors face numerous

obstacles and restrictions such as the minimum capital requirement of $ 700000 and a

maximum of 49 percent ownership. The 2000 new banking law protects the interests of

investors and works against corruption. However, the US Department of State estimates

that 30 percent of Jordan’s loans are nonperforming. Subsidies still remain for oil, while

most price controls have been removed. The judiciary branch is designed to be

independent; however the strong executive branch can easily influence the judges in its

favor. Similarly, the government is attempting to bring reforms to foster a more

competitive environment, yet the bureaucratic and burdensome regulatory system

characterizing by red tape and arbitrary application of customs, tax, labor, and other laws

is a strong obstacle to attract investments.

Regarding the financial equity market, Amman Stock Market (ASM) was formed

on 1 January 1978. Since its formation, the market has experienced some growth in a

72

number of aspects. The ratio of market capitalization to GDP increases from 37 percent

in 1978 to about 160 percent in 2004, which indicates the importance of the market in the

national economy. However, the market can be seen highly concentrated; since

approximately 10 companies in each year accounted for a large proportion of the total

trading volume. In other words, most listed shares are thinly traded on the secondary

market.

The order-driven market making system of the ASM has no designated liquidity

providers and orders are prioritized for execution in terms of price and time. By

submitting a limit order, a trader provides liquidity for other market participants who

demand immediacy. In other words, investors can trade via market orders and consume

liquidity in the market. Given the importance of the ASM in the national economy, the

Jordanian capital market has seen the introduction of a number of major changes. At the

forefront of these changes is the June 2000 implementation of the Electronic Trading

System (ETS). This system was bought from the Paris Bourse and its’ cost (10.5 million

French Francs) was funded by the French government. This event can be considered as a

qualitative leap because it means more transparency and safety for traders and investors.

Since the establishments of the Jordanian capital market, investors have been enjoying a

zero tax rate on capital gains and dividends. However, in 1996, the government imposed

a 10 percent tax rate on dividends.

Table 3-2

Some Economic Indicators, Jordan

2000 2001 2002 2003 2004

Population (million) 4.820 4.940 5.070 5.200 5.323

GDP (m US$) 8,460 8,975 9,561 10,160 11,515

GDP growth (%) 3.82 6.09 6.53 6.27 13.34

GDP per capita (US$) 1,755 1,817 1,886 1,954 2,163

Inflation Rate (%) 0.7 1.8 1.8 2.3 -

Source: Uniform Arabian Economic Report 2005, Arab Monetary Fund (AMF)

• Palestine

Palestine Stock Exchange (PSE) was incorporated as a private shareholding

company in early 1995, with Palestine Development and Investment Company

(PADICO) and (SAMED) as its major investors. After the Palestinian National Authority

73

(PNA) approved a PADICO-sponsored design and work plan in July 1995, a project team

was put together by PSE and entrusted to establish a fully electronic exchange and

depository. EFA Software Service, a Canadian company provides both the trading,

settlement and clearing systems. By August 1996, the exchange was fully operational and

on November 7th of that year, PSE signed an operational agreement with PNA, allowing

for the licensing and qualification of brokerage firms to take place. On February 18th

1997, PSE conducted its first trading session. 28 shareholding companies have been

approved for listing.

As a self regulating organization, PSE is charged with enforcing its rule and

regulations until 2004, since pursuant to laws # 12 and 13 the supervisory and executive

roles have been separated. The first being discharged by a public sector affiliated body,

while the second role is being carried out by PSE. Regarding foreign investment, PSE

does not impose any restrictions on foreign investment. However, as a result of political

problems in Palestine since September 2000; the Palestinian economy suffered a sharp

recession. For instance, the unemployment ratio reached 81 percent in 2002. While GDP

dropped from $ 4712.6 million in 1999 to $ 3213.8 in 2002. All these developments

affected negatively the investment’s environment in Palestine, which in parallel affected

the performance of PSE sharply.

Table 3-3

Some Economic Indicators, Palestine

2000 2001 2002 2003 2004

Population (million) 3.224 3.312 3.560 3.720 3.880

GDP (m US$) 4,442 4,136 3,780 4,222 4,462

GDP growth (%) -1.65 -6.89 -8.61 11.69 5.68

GDP per capita (US$) 1,378 1,249 1,062 1,135 1,150

Inflation Rate (%) 5.54 2.79 5.71 4.4 -

Source: Uniform Arabian Economic Report 2005, Arab Monetary Fund (AMF)

• Saudi Arabia

Saudi Arabia, one of the prominent countries in the Organization of Petroleum

Exporting Countries (OPEC), has the largest oil reserves in the world. Oil exports

account for 90 percent of export earning, 38 percent of GDP, and 80 percent of the

budget revenues. At the same time, the country faces the challenges of a rabidly growing

74

population, water shortages, and political challenges from Islamic extremists. Although

the government recognizes the need for privatization to reduce its dependence on oil, the

transformation will not happen immediately as the private sector constitutes only about

25 percent of the economy.

The government imposes subsidies on state-owned industries, resulting in a

weighted average annual rate of inflation of -0.55 percent from 1993 to 2002.

Furthermore, the government lists sectors that are prohibited to foreign investment, while

many others are subject to tedious government regulations in favor of private interest. Its

banking sector is tightly controlled by the Saudi central bank and is heavily dependent on

the global oil market. While regarding stock market, the Saudi stock market started in

1952 with one company and continued in an unregulated manner till 1984. At this point,

the central bank (SAMA) took over as a regulatory body, entrusting all trading to take

place through commercial banks in the country.

In 1990, an electronic trading system was instituted consisting of a central

clearing mechanism connected via twelve trading units (CTU) to the twelve commercial

banks in the kingdom. Orders for buy and sell have to be entered by bank employees

manning these CTUs. Currently 77 companies are listed and eligible for trading, while

trading each day is broken into two, two hour sessions, Saturday to Wednesday; and one

two our session on Thursday. The minimum tick size is one Saudi Riyal (approximately $

0.26) and transaction costs starts from a minimum of SR 25. In addition, only limit order

are accepted by the system, where the typical order must specify the price and the

quantity intended for purchase or sell. A market order would therefore have to taken the

form of what is termed a “marketable order”, where the price is better than or equal to the

best bid or offer currently available. The best two bids and offers are publicly visible on

the electronic order book, settlement follows the end of the second trading session, and

printed certificates are available the next day.

75

Table 3-4

Some Economic Indicators, Saudi Arabia

2000 2001 2002 2003 2004

Population (million) 20.474 20.976 21.491 21.983 22.529

GDP (m US$) 188,442 183,012 188,551 214,573 250,558

GDP growth (%) 17.08 -2.88 3.03 13.8 16.77

GDP per capita (US$) 9,204 8,725 8,773 9,761 11,122

Inflation Rate (%) - -0.7 -0.2 0.6 -

Oil reserve (as % of world reserves) 25.42 25 25 25 24

Oil production (% of world production) 11.7 12.1 11.8 10.8 12.4

Contribution of oil to GDP (%) - - - 34 38

Source: Uniform Arabian Economic Report 2005, Arab Monetary Fund (AMF)

• Kuwait

Controlling approximately 9 percent of the world’s oil supply, 47 percent of

Kuwaiti GDP and 90 percent of its export revenues come from oil production. Reforms

by the government have been stalled by political pressure from Islamic and populist

parties who benefit from the current system. Similarly, the parliament has delayed

Kuwait project to develop oil fields in the northern part of the country due to opposition

of allowing foreign investors to gain control of the oil industries. Nevertheless, this

project and foreign participation in Kuwait will enable the country to participate in

reconstruction in Iraq and serve as a strategic transshipment port for goods bound to the

region. Kuwait has no income tax or corporate taxes for wholly owned Kuwaiti

companies while foreign corporations are subject to a 55 percent tax rate.

The government intervenes in the stock market. Along with the high tax rate,

foreign investment faces significant restrictions, such as inability to foreigner to own real

state and invest in the oil sector. The banking sector is more competitive and open to

foreign investment, although foreigners are restricted to maximum of 49 percent of

ownership. Key services that are subsidized by the government are subject to price

controls. While no minimum wages exist in the private sector, wages are set in the public

sector that employs 93 percent of Kuwaitis

In the case of financial market, Kuwait stock market was established in April

1978. In August 1982, the official Kuwaiti stock market fell 21 percent in value and the

unofficial market fell about 60 percent. From August 1982 until mid-1984, the Kuwaiti

76

government bought selected stocks to support prices. In September 1982, it is required

that investors in both markets report their open forward positions. At this time, the value

of outstanding post-dated checks in both markets was $ 93 billion ($17 billion in the

official market and $ 76 billion in the unofficial market) with settlement dates of up to 3

years. The market collapse and ensuring economic and financial crisis are referred to as

Al-manakh crisis. Kuwait’s response to Al-manakh crisis was to institute laws and

regulations governing information disclosure, securities registration, and capital and

credential requirements for brokers. As a result, a reorganized Kuwaiti Securities

Exchange began trading stocks, bonds and bank deposits in 1984, prices were determined

in a competitive auction, trades were conducted by floor brokers on instructions from

outside brokers so that floor brokers did not know the identity of their clients. In addition,

restrictions on margin trading and short selling were enforced.

Table 3-5

Some Economic Indicators,Kuwait

2000 2001 2002 2003 2004

Population (million) 2.228 2.243 2.363 2.484 2.645

GDP (m US$) 37,018 34,076 38,111 46,195 55,719

GDP growth (%) 22.40 -7.95 11.84 21.21 20.62

GDP per capita (US$) 16,615 15,192 16,128 18,597 21,066

Inflation Rate (%) - 1.7 1.4 1.2 -

Oil reserve (as % of world reserves) 9 9 9 9 9

Oil production (% of world production) 3 2.9 2.7 3.1 -

Contribution of oil to GDP (%) - - - 41 47

Source: Uniform Arabian Economic Report 2005, Arab Monetary Fund (AMF)

• Oman

Since the 18th century, Oman has been governed by an absolute monarchy. In

2001, 66 percent of the government’s total revenues came from state-owned enterprises

and its ownership of property. The oil industry has grown to be the dominating industry,

making up 86 percent of revenues and roughly 47 percent of GDP. At the current rate of

production, oil reserves are projected to last for only 18 years. The government realizes

that diversification is essential and is trying to respond by expanding its gas-based

industry, boosting economic activity, facilitating foreign investment and privatization,

77

and promoting private-sector employment. However, due to its constantly changes and

complex customs procedures and regulations, Oman’s restrictive trade policy is an

obstacle to open trade and considerable progress. Similarly, establishing a business in the

country can prove to be a tedious process, subject to the approval from various authorities

in respect to land acquisition and labor requirements. Lack of clear regulations that

explicitly codify Omani labor and tax laws cause ad hoc decisions and complicate the

process even further. Burdensome regulatory requirements for approvals cause

considerable delays and adverse condition for the private sector.

Additionally, political pressures have always influenced the judiciary branch.

However, 2001 and 2002 show significant changes in the restructuring of the legal

system, where the courts, the public prosecution service, the police and attorney-general

have all been separated to function independently. Unemployment remains a significant

concern, particularly among the fast-growing young population. To mitigate the problem,

the government has implemented a quota program that replaces foreign workers with

Omains, which poses another impediment to foreign investment. While individuals do

not have to pay an income tax, companies that are 70 percent foreign-owned incur a 30

percent tax, whereas other domestic companies only face a 12 percent tax rate.

In addition, foreign ownership above 70 percent requires the approval of the

Minister of Commerce and Industry, while certain industries are prohibited in the country

all together. With its participation in the WTO, Oman is pressured to open its service

sector of foreign firms. The rate of deflation in Oman is another factor of concern for

investors, where the weighted annual average was -0.76 percent from 1993 to 2002. The

inflationary pressure is kept in check by price controls and a subsidy system.

Additionally, the government-operated banking sector approves very favorable loans to

Omani citizens. Moreover, Oman’s Muscat stock market was established in 1989, with a

122 listed companies and a market capitalization of $ 9.317 billion in 2004. Furthermore,

only nationals of the GCC are permitted to invest in the local stock market.

78

Table 3-6

Some Economic Indicators, Oman

2000 2001 2002 2003 2004

Population (million) 2.402 2.478 2.538 2.331 2.264

GDP (m US$) 19,868 19,949 20,304 21,698 24,824

GDP growth (%) 20.93 0.01 1.78 6.87 14.41

GDP per capita (US$) 8,271 8,050 8,000 9,308 10,965

Inflation Rate (%) - -1 -0.7 -0.3 -

Oil reserve (as % of world reserves) 0.6 0.6 0.5 0.5 -

Oil production (% of world production) 1.4 1.4 1.3 1.2 -

Contribution of oil to GDP (%) - - - 42 42

Source: Uniform Arabian Economic Report 2005, Arab Monetary Fund (AMF)

The United Arab Emirates (UAE) controls approximately 9 percent of the global

oil supply and about 5 percent of the proven natural gas reserves in the world. Oil

revenues comprise about one-third of its GDP. Although the energy reserves expected to

last for more than 100 years at current rate of production, in recognizing the need for

diversification, UAE is focusing on the development of its service sector and non-oil and

gas industrial base. Foreign investment and privatization are sought in the interest of

modernizing technology and reducing costs; however, foreigners face widespread

restrictions in owning land and investing in specific industries. Where the land is not

state-owned, private property is generally well protected. Importers are required to have

an import license and are subject to various restrictive regulations. Prices on goods are

affected through government subsidies.

The public sector holds an important role in total employment and provides

subsidies services and an extensive welfare system. In 2001, for example, public

enterprises in the hydrocarbon sector alone accounted for 59 percent of the government

revenues. In providing loan guarantees, the government minimizes the risk of default to

attract international investment. The UAE has no income tax, and no other significant

taxes. However, foreign banks face 20 percent tax on profits and are subject to quotas to

hire UAE nationals, and other restrictions.

The UAE stock market is relatively new and small, which contains both official

and unofficial markets. The official market started in 2000 and represented two

79

government stock markets, Dubai and Abu Dhabi, under the supervision of the Emirates

Securities and Commodities Authority. While the unofficial, or OTC, market works

through several brokerage firms with most of them affiliated to banks. Since its inception

as an unofficial market in the late of 1970s, the UAE stock market has experienced

several volatile periods in terms of share trading activity and price level. The period of

(1975-1982) had witnessed the creation of many companies due to rising oil prices and

the strong interest of the federal government to build a strong national economy.

However, the crisis of the Kuwaiti stock market, the crisis of Al-manakh market in 1982,

and the falling of oil prices in 1986 had a negative impact on the UAE capital market.

The UAE capital market rose again during the period of 1993-1998, due to the

establishments of many new companies.

While once again, the UAE capital market experienced a deep decline in the

summer of 1998 due to several reasons including: lack of regularity, manipulation of the

market by block traders and professional investors, negative speculative trading by all

participants, lack of financial disclosure, and the drop in oil prices. Since the summer of

1998, the market has suffered sharp declines in both trading volume and trading value to

such an extend that the market prices of most traded stocks have decreased under their

par value. In response to the stock market crisis in 1998, the UAE government responded

by officially recognizing its stock market. The Emirates Securities and Commodities

Authorities (ESCA) was established February 1st, 2000 pursuant to federal law # 4 of

2000 under the chairmanship of the Minister of Economy and Commerce. Its function is

to regulate and develop the primary and secondary markets, monitor the operations of the

market, and create a favorable environment for investment.

As a result, the Dubai Financial Market (DFM) was officially founded in March

2000 as the first organized stock market in UAE. DFM has been trying to increase the

investment alternatives available to investors, and sources of financing available to

companies, while Abu Dhabi Securities market (ADSM) started operating in November

2000. ADSM with its 35 listed companies in 2004 is larger than DFM. However, neither

bonds nor mutual funds are yet included. Since the establishments of the official UAE

stock market in 2000, it has been growing at the expense of the OTC market. In addition,

ESCA enacted a set of statutory orders and regulations that pertain to arbitration, listing,

80

brokers’ practice, disclosure, transparency, financial markets operations, trading,

clearance and depository. Moreover, in 2001 ESCA launched an official capital weighted

average market index with 1000 points, called Emirates index, consisting of all listing

companies.

Table 3-7

Some Economic Indicators, United Arab Emirates (UAE)

2000 2001 2002 2003 2004

Population (million) 3.247 3.488 3.754 4.036 4.368

GDP (m US$) 70,521 69,546 75,694 88,645 103,833

GDP growth (%) 27.78 -1.38 8.84 17.11 17.13

GDP per capita (US$) 21,719 19,939 20,164 21,964 23,771

Inflation Rate (%) - 2.7 2.9 2.8 -

Oil reserve (as % of world reserves) 9 9 9 9 9

Oil production (% of world production) 3.3 3.3 3.3 3.3 -

Contribution of oil to GDP (%) - - 28 32 -

Source: Uniform Arabian Economic Report 2005, Arab Monetary Fund (AMF)

• Bahrain

The Bahraini Stock Exchange (BSE) was established in 1989, BSE has 45 listed

companies with a market capitalization of $ 13.5 billion in 2004. Electronic trading takes

place on the exchange floor facilitated the newly established clearing and settlement

house. Furthermore, new legislation allows GCC investors an unrestricted stake, and non-

GCC foreign investors up to 49 percent stake in listed companies. Although small by

international standard, BSE is positioning it self to be a major player in the Gulf financial

markets. The BSE continues to forge ahead with a development strategy aimed at putting

Bahrain on the map of the international capital markets. Several initiatives have been

launched by the BSE to provide an infrastructure that is modern and similar to systems

that are enforced in developed capital markets. The ultimate goal is to enable the BSE to

play a pivotal role in the national economy by mobilizing private sector savings and

attracting foreign investments, through a truly international market.

The ground is being carefully laid for the ultimate opening of the stock market to

overseas investors, which will significantly add to Bahrain’s ability to attract foreign

investment and will develop the financial markets in terms of volatility, market activity,

81

depth and liquidity. Another important development concerns disclosure regulations,

which have been initiated to enhance transparency and the safety of investors’ money. A

key objective is to attract a large number of small investors. The disclosure regulations

are aimed at making more information available on the share prices, performance of

listed companies and investors’ activity. Moreover, the approved disclosure standards are

based on the recommendations of IOSCO (International Organization of Securities

Commissions) and are similar to the standard applied internationally. In addition, the

BSE has also jointed the International Finance Corporations (IFC) Global Index, which is

expected to enhance transparency and disclosure in order to create investment

opportunities for foreign investors of individuals as well as internationals portfolios.

Table 3-8

Some Economic Indicators, Bahrain

2000 2001 2002 2003 2004

Population (million) 0.638 0.655 0.672 0.69 0.708

GDP (m US$) 7,970 7,929 8,448 9,606 11,067

GDP growth (%) 20.37 -0.01 6.55 13.71 15.21

GDP per capita (US$) 12,492 12,105 12,571 13,922 15,631

Inflation Rate (%) - -1.2 -0.5 1.6 -

Oil reserve (as % of world reserves) 0.02 0.02 0.02 0.02 0.02

Oil production (% of world production) 0.3 0.3 0.3 0.3 -

Contribution of oil to GDP (%) - - 25 25 -

Source: Uniform Arabian Economic Report 2005, Arab Monetary Fund (AMF)

The economic reforms in the Arab countries have strengthened the recognition

that Arab capital markets play an important role in the economic development process.

The role of those markets in meeting financing requirements gained additional

importance with the increased reliance by the growing number of member countries on

market forces in resource allocation, greater participation of the private sector in the

economic activity and for security non-inflationary financing of budget deficits. As a

result, Arab capital markets witnessed remarkable developments in their various aspects,

including the legal and organization levels, there by contributing to their foundation on

sound structures, which are constantly evolving.

82

In this aspect, the Arab Monetary Fund (AMF) plays a significant role in

developing Arab stock markets. AMF is a regional institution and devised a work

program made up of three main components. These were: the conduct of surveys studies

aimed at analyzing the regulatory and institutional situation of capital markets in member

countries; the establishment of a database to provide information on the activities of those

markets; and finally, the provision of technical assistance to Arab countries for the

development of markets for financial papers operating in their jurisdiction. Using the data

generating by the database on Arab capital markets, AMF establish an AMF index

calculated for each market in addition to the market own index. As a result, the bulletin

publishes the AMF composite index in the calculation of which the sample shares of all

participating markets merged in a single sample. In addition, the Fund started the

publication of a quarterly report named “Quarterly Bulletin of the Arab Capital Markets

Database”, the bulletin’s first issue was published in April 1995. Each issue of this

bulletin reports on developments in the participating markets during the relevant quarter.

Furthermore, since June 10, 2002, the Fund started to publish on its website, on daily

basis, some basic indices for markets participating in the database.

Broadly, these developments involved an improvement in the performance of

capital markets, a strengthening of their supervision and increased trading on their floors.

They are also related to amendments of tax systems, streamlining of administrative

procedures, the creation of a favorable environment suited to the requirements of market

actors, the introduction of new financial instruments offering a greater variety of

investment opportunities, the acceleration and simplification of trading operations, and

the promotion of transparency and disclosure. Adding to these was the improvements in

skills of operating staff and enhanced discipline and professional ethics, these

developments can be summed as follows:

A number of Arab countries have been proceeded to separate between the

supervisory and executive roles, the first being discharged by a public sector affiliated

body, while the second being mostly carried out by the private sector. In this area, most

Arab countries enacted capital market laws; aimed at restructuring the markets and

83

leading to the separation between the supervisory function, in charge of regulating the

issuance and trading of financial paper on the one hand, and the managements of the

stock exchange through which such papers are traded and the agency in charge of

registering the transfer, sale and purchase of those paper and keeping a registry of records

and ownership titles, on the other hand.

By end 2005, the separation between the supervisory and executive roles took

place in the following Arab capital markets: Jordan, Egypt, Palestine, and United Arab

Emirates. While the two roles continue to be simultaneously in the hands of the capital

markets itself in the rest of Arab countries.

Table 3-9

Market Structure for Arab Stock Markets

Supervisory and Existing of Only Duration

Market executive roles primary and secondary of

are seperated secondary markets market settlement

Bahrain No - Yes T+2

Egypt Yes Yes - T+3

Jordan Yes Yes - T+3

Palestine Yes - Yes T+3

Kuwat No - Yes T+3

Saudi Arabia No - Yes T+2

Oman No Yes - T+3

UAE Yes yes - T+3

Arab capital markets have been attaching greater importance to the need for

increased transparency, and for adapting its exigencies to meet international standards in

order to enhance the supervisory role on the one hand, and to ensure equal opportunities

for market operators, on the other hand. Accordingly, the scope of instructions,

information and data disclosure of which became mandatory, widened. Such information

must now include, for example, the names of issuers of financial papers, those of market

members, authorized professionals as well as periodic data related to trading movements

and main financial indicators. In this connection, most Arab stock exchanges now publish

daily, weekly, monthly, and annual bulletins reporting general information on their

84

markets and executive boards’ decisions together with data on traded volumes and price

indices.

In addition, most of theses markets have signed agreements with world class

companies specialized in automated instant reporting on trading, including Reuters and

Bloomberg. It is worth noting that these markets are also disseminating their data through

the internet, in order to further publicize investment opportunities which they offer.

Moreover, the websites of these stock exchanges are now posting daily updated

information on trading effected on their floor; together with historical data containing

time-series on all data pertaining to exchange activities. On the other hand, these markets

have been endeavoring to ensure that joint-stock companies listed on their floors, strictly

up-hold the principles of disclosures and transparency. Additionally, to making the

submission of annual reports to financial markets authorities mandatory on listed

companies, new instructions in some Arab countries are now rendering it an obligation

for such companies to present bi-annual and quarterly reports.

instruments

Most capital markets in the Arab countries have been seeking to develop that role,

as a means of enhancing market stability and protecting it from sharp fluctuations. In

general, the institutional investor is interested in medium and long-term investment and

basis his decision on scientific studies. By contrast, an individual investor seeks to

achieve quick capital gains; in view of his limited awareness, his behavior impacts

negatively the business of Arab stock exchanges. In this area, authorities in Arab

countries have been encouraging long-term savings by creating saving accounts in

market-listed shares, which enjoy low capital gain tax; and by authorizing pension funds

and insurance companies to deal with those markets.

In addition, Arab stock markets encourage the increase of available investment

instruments and alternatives such as bonds convertible into shares and investment funds.

Since investment funds, which have been established in most Arab capital markets, are

being viewed as the most suitable instrument for mobilizing savings and attracting

foreign capital. They provide a mechanism for placing resources in financial papers

85

carrying differentiated risks and returns, which an individual investor cannot achieve due

to his size. These funds enable Arab expatriates and overseas investors to place their

savings in Arab markets for financial papers, without having to be physically present in

the region. In addition, they offer to foreigners residing in Arab countries the opportunity

to enter local financial markets, since the investor’s right in those funds is confined to his

share in the financial papers in which the resources of the fund are invested, and to a

proportionate entitlement to its returns.

Investments by-laws in most Arab countries have witnessed a number of changes,

mostly aimed to attracting foreign investments, meeting the domestic financing

requirements, and smoothing the transfer of advanced technologies into their markets.

The changes involved, represent part of the steps taken by those countries to open the

door for the entry of foreign investments, by removing the obstacles which used to

impede their flows. In this context, Arab countries can be divided in two groups. The first

includes countries, which do not impose any restrictions on foreign investments in

financial papers; these are Egypt, Jordan, and Palestine, while the second group

comprises countries, where such restrictions exist in varying degrees; these are the

member states of GCC.

For instance, in Saudi Arabia and Kuwait, foreigners are allowed to invest in

shares through investment funds, the United Arab Emirates allows foreigners to both

invest through similar funds; and to own not more than 49 percent in shares of companies

whose internal by-laws so permit. However, in Oman, foreigners can buy shares of newly

listed companies in proportions of up to 49 percent, and in the case of certain companies,

up to 100 percent, while Bahrain allows for GCC nationals, to own up to 100 percent of

shares, if it is defined that this will serve the interests of the national economy. Table 3-

10 indicates that while GCC stock markets are fully accessible to GCC investors, they

have remained relatively closed to international foreign investors, even non-GCC Arab

investors, face restrictions on portfolio investment in these stock markets

It is expected that, the removal of the various restrictions which faced MENA

portfolios flows, will improve and enhance growth and liquidity in these markets; and

86

reduce the costs of raising capital in the local market. Although, the open access to

foreign investors will contribute significantly to the growth performances of Arab stock

markets, this is expected to gradually lower the diversification potentials; that used to be

offered to international investors. In addition, increasing financial integration within the

Arab countries is expected to bring considerable benefits to Arab investors, since a more

liquid capital market, offers lower borrowing costs for Arabian firms wishing to raise

funds locally. Moreover, international financial institutions will be willing to diversify

their portfolios by tapping the Arab financial markets.

Table 3-10

Accessibility of Arab Stock Markets to Foreign Investments

Market

- open to GCC nationals.

- Foreign residents in Bahrain for at least three years, may own up

Bahrain to 1% of the capital of 31 listed companies.

- Foreigners can trade shares in only 10 of the 45 listed companies

and up to 24%.

- Unristrected access to foreign investors.

Egypt

- Repatriation of capital and dividends allowed.

- Unristrected access to foreign investors, in specefic sectors

foreign investors can hold up to 50% of companies' capital.

Jordan

- Repatriation of capital and dividends allowed.

- open to GCC nationals.

Kuwait - Non-kuwaiti residents are allowed to own shares through matual

funds only.

- Unristrected access to foreign investors.

Palestine

- Repatriation of capital and dividends allowed.

- Open only to GCC nationals who can own up to 25% of listed

Saudi Arabia companies other than banks.

- Opened recently to foreign investors through matual funds only.

- Foreign investors can hold up to 49% of companies' capital.

Oman

- Repatriation of capital and dividends allowed.

- Foreign investors can own up to 49% in companies' capital

UAE

whose internal by-laws permit.

Many Arab countries subjected their tax systems to thorough amendments

directed towards creating incentives, to encourage dealing in financial papers on the one

87

hand, and attracting foreign investments, on the other. By virtue of those changes, these

countries either reduced, or eliminated taxes on current returns and capital gains arising

from dealing in financial papers. It must be noted that no such taxes existed in all Arab

countries, which had regular financial markets. Moreover, some Arab countries also,

directed their tax reform towards encouraging joint-stock companies; to have their shares

listed in their exchanges.

Most Arab capital markets took vast steps to modernize their dealing systems, and

to introduce modern technologies in share trading operations with a view to improve

performance, enhance speed and accuracy in conduct of business and increase

transparency and operators’ confidence. As a result, high-tech automated dealing systems

were introduced to the markets. Also, some of these markets inaugurated distant-dealing

services, which constitute one of the innovation services witnessed by those markets, and

offered a mechanism enabling accredited brokers to conclude contracts without the need

to be represented on the physical floor, such as Palestine.

Arab stock exchanges have made major strides on the path of cooperation and

integration among them selves, by concluding bilateral and trilateral agreements. The

thrust of the latter is to increase collaboration between stock exchanges in the areas of

financial papers issue and trading, organizing and facilitating clearing and settlement

mechanism. These agreements also, aimed at developing cooperation between

intermediation institutions in those markets; and encourage joint/cross listing. In that

regard, agreements were signed between the stock exchange of Bahrain, Kuwait, and

Oman on the one hand, and those between the stock exchanges of Bahrain and Jordan,

Abu Dhabi and Palestine, on the other hand. Comparable agreements were also

concluded between Kuwait, Lebanon, and Egypt in one case. In addition, a memorandum

of understanding was signed in the case of Jordan and Kuwait, while in a third case, an

extended agreement was concluded between Abu Dhabi and Khartoum markets for

financial papers.

88

All those agreements aimed at fostering cooperation and eliminating hurdles

hindering the flows of investments between the markets involved. The consolidation of

the trend towards greater integration among Arab capital markets; and the preparation of

the propitious conditions for upgrading bilateral and trilateral agreements for cross listing

to a collective level, call for a high degree of harmonization between the accounting

standards followed and legal systems and, particularly, coordination in the area of

clearing and settlement.

Most of Arab security markets’ indices increased at the end of 2005. In comparing

Arab stock market performance with other international and emerging markets, figure 3-1

presents indices’ returns between September 2004 and September 2005. One can see that

except of Bahrain stock market, Arab stock markets performance was better than other

international stock markets, while Dubai stock market stands to be the best in

performance followed by Palestine stock exchange, which index’s return increased with

250 percent during 2005.

Figure 3-1

Arab Stock Markets Performance Compared to other International Stock Markets

9/2004-9/2005

89

3-4-1 Market size

With respect to market size, Arab stock markets are small by international

standards; their total market capitalization constitutes less than 5 percent of the US

market and only about 19 percent of that of UK stock market in 2004. However, within

the group of Arab security markets, their total capitalization value increased dramatically

during 2000 and 2005, from $ 135.657 billion up to $1166.154 billion, with a growing

rate more than 700 percent (see figure 3-2). It can be seen from figure 3-2 that most

market size variables witnessed dramatic changes, market capitalization as a percentage

of GDP, increased by 390 percent, while the volume of shares traded increased by 1081

percent. Moreover, the value of traded shares increased with 3924 percent during 2000

and 2005, on the other hand, the number of listed companies decreased by 8 percent, to

reach 1467 listed companies in all Arab markets at the end of 2005.

Figure 3-2

Market Size for Arab Stock Markets between: 2000-2005

However, for individual market size and in terms of market capitalization, Saudi

Arabia stands to be the largest market in the region at the end of 2005. It accounts for

about 55 percent of total market capitalization for all Arab stock markets. Followed by

Abu Dhabi stock market, while Palestine stock exchange stands to be the smallest among

Arab stock markets (see figure 3-3 and table 3-11 respectively).

90

Table 3-11

Market Capitalization for Arab Stock Markets

(Million US$)

Market 2000 2001 2002 2003 2004 2005

Abu DHABI - - 20,375.76 30,362.51 55,490.40 132,412.89

Jordan 4,943.16 6,314.16 7,087.03 10,962.89 18,383.40 37,638.81

Bahrain 6,624.35 6,601.27 7,716.39 9,701.77 13,513.18 17,364.31

Saudi arabia 67,166.04 73,201.35 74,851.38 157,306.44 306,255.70 646,120.80

Kuwait 19,847.98 26,661.70 35,098.89 59,528.01 73,580.54 123,892.58

Dubai - - 9,469.52 14,284.23 35,090.90 111,992.68

Oman 3,518.13 2,634.37 5,268.05 7,264.23 9,317.66 12,062.05

Egypt 30,791.26 24,308.57 26,338.69 27,847.48 38,076.84 79,507.56

Palestine 766.02 722.63 576.59 650.47 1,096.53 3,157.15

Total 135,656.94 142,445.04 188,784.29 319,911.03 552,809.15 1,166,153.83

Source: Arab Monetary Fund, AMDB

Figure 3-3

Relative Market Capitalization to All Markets 2005

choices of firms available to an investor. In this case, Egypt stands out among Arab

markets; with a total number of listed companies reaching 744 companies at the end of

2005 (table 3-12). However, if the number of listed companies is used in conjunction

with market capitalization, it will indicate the average market value of listed companies.

91

Table 3-12

Total Number of Listed Companies, 2000-2005

Market 2000 2001 2002 2003 2004 2005

Abu DHABI - - 24 30 35 59

Jordan 163 161 158 161 192 201

Bahrain 41 42 40 44 45 47

Saudi Arabia 75 76 68 70 73 77

Kuwait 86 88 95 108 125 156

Dubai - - 12 13 18 30

Oman 131 96 140 141 123 125

Egypt 1,071 1,110 1,150 967 792 744

Palestine 22 23 23 26 26 28

Total 1,589 1,596 1,710 1,560 1,429 1,467

Source: Arab Monetary Fund, AMDB

In this case, Saudi Arabia has by far the highest market value per listed company

among Arab markets, at about $ 8391 million followed by Dubai at $ 3733 million, with

Egypt having the lowest market value per listed company, after Oman, at $ 107 million.

Since for Egypt, over 90 of the 744 companies listed at the end of 2005; are actively

traded, while more than 400 companies are classified as closed family corporations,

which are listed to qualify for certain tax benefits. Table 3-13 presents market

capitalization as a percentage of GDP, which indicates the relative role that a stock

market has in the national economy. In this area, it can be seen that the Jordanian stock

market has the highest rate of market capitalization as a percentage of GDP at the end of

2004 (160%), followed by Kuwaiti stock market, while Palestine stock exchange has the

lowest rate (25%) at the end of 2004.

Table 3-13

Market Capitalization as a Percentage of GDP

Market 2000 2001 2002 2003 2004

Abu DHABI - - 28.47% 37.85% 53.44%

Jordan 58.54% 70.64% 75.03% 110.19% 159.65%

Bahrain 83.12% 83.25% 91.34% 101.00% 122.10%

Saudi arabia 35.64% 40.00% 39.70% 73.35% 122.23%

Kuwait 53.62% 78.24% 99.79% 142.61% 132.06%

Dubai - - 13.23% 17.81% 33.80%

Oman 17.71% 13.21% 25.94% 33.64% 37.53%

Egypt 31.34% 26.92% 31.31% 39.26% 48.51%

Palestine 17.25% 17.47% 15.25% 15.41% 24.57%

Source: Arab Monetary Fund, AMDB

92

3-4-2 Market liquidity

Arab stock markets’ liquidity has been improved during the last years. Market

liquidity variables for total Arab stock markets, witnessed significant changes during

2000 and 2005 (see figure 3-4). The total value traded to market capitalization (turnover

ratio) increased by 368 percent for total Arab markets, while total value traded to GDP

increased with 607 percent. Meanwhile, the average daily trading value increased sharply

by 3102 percent.

Figure 3-4

Market Liquidity Variables for Arab Stock Markets, 2000-2004

However, for individual markets and in the case of the yearly turnover ratio,

which is the ratio of yearly trading value to market capitalization at the end of the year,

the Saudi stock market is the most active and liquid among Arab stock markets. Its turn

over ratio reached 171 percent with average daily trading value $ 3691 million in 2005.

While the value traded as percentage of GDP reached 189 percent, which puts the Saudi

market to be the most active market among Arab stock markets. The Kuwaiti market

comes second in market liquidity, with turnover ratio 78 percent; $ 391 million and 93

percent as an average daily trading value and total value traded as a percentage of GDP in

2004 respectively. In addition, Jordanian and Dubai stock markets can be characterized as

93

active markets, while Bahraini stock market stands to be the least liquid market among

Arab stock markets as they are in 2005 (see tables 3-14, 3-15, and 3-16).

Table 3-14

Total Value Traded to Market Capitalization (Turnover Ratio)

Market 2000 2001 2002 2003 2004 2005

Abu Dhabi - - 1.78% 3.31% 8.02% 21.53%

Jordan 8.21% 14.80% 18.83% 23.78% 28.98% 63.25%

Bahrain 3.71% 3.79% 2.67% 2.69% 3.43% 4.10%

Saudi Arabia 25.78% 30.36% 41.38% 101.11% 154.44% 170.80%

Kuwait 21.20% 43.93% 63.03% 91.94% 70.42% 78.53%

Dubai - - 7.26% 7.19% 39.14% 98.49%

Oman 15.67% 15.94% 11.04% 18.37% 21.31% 27.53%

Egypt 38.32% 24.32% 24.46% 15.62% 17.95% 34.87%

Palestine 1.70% 0.94% 0.60% 0.76% 19.19% 14.10%

source :Arab Monetary Fund, AMDB

Table 3-15

Average Daily Trading Value (million US$)

Market 2000 2001 2002 2003 2004 2005

Abu Dhabi - - 1.46 3.70 15.13 95.02

Jordan 1.67 3.88 5.36 10.82 21.66 97.57

Bahrain 0.99 1.01 0.82 1.05 1.87 2.87

Saudi Arabia 60.96 73.83 102.23 530.19 1581.91 3690.91

Kuwait 16.93 47.36 88.85 225.22 208.94 390.72

Dubai - - 2.30 3.46 46.72 14.8

Oman 2.28 1.71 2.32 5.36 7.91 13.02

Egypt 48.01 24.44 25.95 17.82 27.45 111.78

Palestine - - 0.45 0.26 0.82 6.11

Total 130.84 152.23 229.73 797.88 1912.41 4422.80

source :Arab Monetary Fund, AMDB

94

Table 3-16

Total Value Traded as Percentage of GDP

Market 2000 2001 2002 2003 2004

Abu Dhabi - - 0.51% 1.25% 4.28%

Jordan 4.81% 10.45% 14.13% 26.21% 46.26%

Bahrain 3.08% 3.16% 2.44% 2.72% 4.18%

Saudi Arabia 9.19% 12.14% 16.43% 74.16% 188.77%

Kuwait 11.37% 34.37% 62.90% 131.11% 93.00%

Dubai - - 0.96% 1.28% 13.23%

Oman 2.78% 2.10% 2.86% 6.18% 8.00%

Egypt 12.04% 6.55% 7.66% 6.13% 8.71%

Palestine 4.25% 1.90% 1.46% 2.15% 4.49%

source :Arab Monetary Fund Database (AMDB)

Regarding financial valuation of Arab stock markets, Table 3-17 presents a

comparison between these markets. Clearly and according to the available data, the most

expensive markets at the end of 2005 were those of Saudi Arabia (based on both the P/E

and the P/BV ratios) and Jordan (based on P/E ratio), while the least expensive markets

were those of Oman (based on P/E ratio) and Bahrain (based on P/BV ratio).

Table 3-17

Financial Valuation of Arab Stock Markets, End of 2005

Market P/E ratio P/BV ratio Dividend Yeild (%)

Abu DHABI 20.90 4.33 1.20

Jordan 44.20 5.23 1.01

Bahrain 16.26 2.10 3.19

Saudi arabia 66.22 12.47 1.02

Kuwait - - -

Dubai 34.40 8.71 1.30

Oman 11.75 2.20 3.00

Egypt - - -

Palestine - - -

Notes : P/E ratio stands for the price/earning ratio and P/BV for the price/ book value ratio.

: (-) data not available.

source : Arab Monetary Fund Database (AMDB).

95

3-4-4 Market concentration

Most Arab stock markets in general suffered from thin trading phenomenon,

which indeed affect market liquidity. In other words, most listed shares are thinly traded

on the market. Figure 3-5 shows the percentage of the 2 biggest companies’ share in

value traded and market capitalization respectively. It can be seen that in general, Arab

stock markets are highly concentrated. In the case of Palestine, the 2 biggest companies’

share in value traded and market capitalization in 2005 are 81, 76 percent respectively.

Which indicate that Palestine stock exchange is highly concentrated. In addition, the

percentage of the 2 biggest companies’ share in value traded and market capitalization for

Dubai was 56, 44 percent respectively. The Saudi and Bahraini stock markets also can be

characterized to be highly concentrated, since the percentage of the 2 biggest companies’

share in market capitalization were 38, 31 percent respectively.

Figure 3-5

Market Concentration, End of 2005

The data that will be used through this research consist of daily prices of Arab

stock markets. The time period vary from market to market, but usually run from about

1st January 1992 to 31 July 2005, the initial and final dates vary from market to market

96

due to the establishment date of the market and to the availability of the data, the data

was collected piece by piece directly from each stock market.

Moreover, all indices used in this study are value weighted indices. The Jordanian

stock market index consists of 71 listed companies distributed among 4 sectors at the end

of 2005, 33 industrial companies, 11 banks, 9 insurance companies, and 18 service

companies, while the Palestinian stock exchange index (Al-Quds index) has 10 listed

companies distributed among 4 sectors, 2 industrial, 2 banks, 2 insurance, and 4 service

companies. The Egyptian stock market index (CASE 30) has 30 companies. The Saudi

index is an all-share index constructed by the central bank, and includes the shares of all

listed companies on the Saudi market, the same index structure holds for the UAE, which

has two stock markets; Abu Dhabi index which has 59 listed companies and Dubai index

with its 30 listed companies. Oman’s Muscat stock exchange index has 33 listed

companies, of which 13 companies represented the banks and investment companies’

index sector; 11 represented the industry index sector; and 9 represented the service index

sector. Moreover, the Kuwaiti stock exchange index has 35 listed companies, while

Bahrain stock market has 25 listed companies.

Appendix 1 shows the plot graphs for the natural logarithm and return for each

index under examination here, while table 3-18 presents the main descriptive statistics for

Arab stock markets’ indices. The returns are the variables on which we want to focus our

attention on, that is Rt = 100 * log (Pt/Pt-1), where Pt denotes closing price for market

index. All the displayed Skewness statistics have asymmetric distributions that are

skewed to the right as shown by the positive Skewness statistics, except of Kuwait index

which is skewed to the left (negative Skewness). Moreover, kurtosis provides a measure

of the “thickness” of the tails of a distribution relative to the normal distribution. For

normal distribution, kurtosis is usually equal to three. The presence of excess kurtosis in

the series suggests that the return distributions have a much fatter tail than the normal

distribution. Finally, none of the series approximates the normal distribution as shown by

the Jarque-Bera statistics.

97

Table 3-18

Descriptive Statistics for Daily Market Returns for Arab Stock Markets,

R t = 100*log(p t /p t-1 )

Jordan Egypt Palestine Kuwait Saudi Bahrain AbuDhabi Dubai Oman

Mean 0.0355 0.0425 0.0820 0.1876 0.0370 0.0257 0.0847 0.0435 0.0251

Median -0.0090 -0.0289 0.0000 0.1618 0.0368 0.0104 0.0558 0.0206 -0.0043

Maximum 4.7465 18.3692 27.2330 4.0263 17.9204 20.6189 2.8665 21.6679 15.2225

Minimum -4.3097 -10.9751 -25.3643 -5.6757 -17.5253 -19.8569 -2.4741 -8.4913 -13.5602

Std. Dev. 0.7341 1.6658 1.8370 1.0386 0.9342 0.7269 0.5388 1.0084 1.0842

CV 20.68 39.21 22.40 5.54 25.22 28.25 6.36 23.19 43.23

Skewness 0.3075 0.7695 0.5314 -0.5134 0.1168 0.4033 0.1243 7.8917 0.7877

Kurtosis 7.7149 15.2906 73.4889 6.9037 93.8064 366.7102 7.8723 203.5124 50.9400

Jarque-Bera 2,940 10,651 244,349 466 1,058,905 18,321,581 640 1,850,786 182,045

Probability 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Sum 110.804 70.781 96.762 128.873 114.182 85.544 54.663 47.754 47.626

Sum Sq. Dev. 1,681.28 4,620.26 3,978.81 740.01 2,688.86 1,756.03 186.94 1,115.48 2,231.17

Observations 3,121 1,666 1,180 687 3,082 3,324 645 1,098 1,899

98

Methodology

Are Arab Stock Markets Efficient in the Weak Form

Sense of Efficient Market Hypothesis?

Regression analysis.

Serial correlation test.

Non-parametric runs test.

Variance test.

BDS test for returns independency.

Seasonality and calendar effects (day of the week effect, monthly effect, and the Halloween

indicator).

Findings Sense of Market Efficiency and Do Not Follow RWH.

These results are consistent with existing literature regarding

emerging markets (Bekaert 1995; Harvey 1995b, 1995c;

Claessens et al. 1995; and Buckberg 1995).

99

Methodology Is the View of Predictability in Stock Returns (if there

is) Related to Whether We Think That These Time

Series Are Non-Linear? How Does Thin Trading

Affect the Predictability of These Time Series?

Estimating the true index correcting for infrequent trading, then reexamine the RW

properties.

Using logistic map to determine whether non-linearity exists.

Testing whether the second moment can characterize the existing non-linearity, through

subjecting the residuals of RW and GARCH models to several diagnostic tests.

Findings

Returns Generating Process in Arab Stock Markets

is Non-Linear, while the Second Moment Found to

Explain Well the Existing Non-Linearity.

100

Methodology Are Arab Stock Markets Characterized By Excessive

Volatility of Returns, Relative to Other Emerging And

International Stock Markets?

GARCH (1,1) model for daily returns of Arab stock markets, and two groups of emerging and

developed markets with a total of 15 stock markets.

GARCH (1,1) model for weekly data.

EGARCH model for daily data.

Schewrt model for the three groups Arab, emerging and developed markets to compare the

relative volatility.

Covariance coefficients.

low level of volatility relative to other emerging and

developed markets.

Findings All Arab markets exhibit volatility clustering except

Dubai.

Egypt, Kuwait, and Palestine exhibit volatility

persistence.

Bahrain, Dubai, Kuwait and Oman show signs of

leverage effect and asymmetric shocks to volatility.

101

4- Testing the efficient market hypothesis for Arab stock markets

There are three kinds of random walk, random walk 1, IID increments; random

walk 2, independent increments and random walk 3, uncorrelated increments. A useful

way to organize the various versions of the random walk and martingale models is to

consider the various kinds in dependence that can exist between an asset’s returns rt and

rt+k of two dates t and t+k. To do this, define the random variables f(rt) and g(rt+k) while

f(.) and g(.) are two arbitrage functions, and consider the situation in which

for all t and for k≠0. For appropriately chosen f(.) and g(.), virtually all versions of the

random walk and martingale hypothesis are captured by (4-1), which may be interpreted

as an orthogonality condition.

For example, if f(.) and g(.) are restricted to be arbitrary linear functions, then (4-

1) implies that returns are serially uncorrelated, corresponding to the random walk 3

model. Alternatively, if f(.) is unrestricted but g(.) is restricted to be linear, then (4-1) is

equivalent to martingale hypothesis. Finally, if (4-1) holds for all functions f(.) and g(.),

this implies that returns are mutually independent, corresponding to the random walk 1

and random walk 2 models.

The statement that prices, in an efficient market, “fully reflect” available

information, conveys the general idea of what is meant by market efficiency. However,

this statement is too general to be tested, encountering a need to develop mathematical

models of market equilibrium that would be used in testing market efficiency (Fama,

1965). The random walk model is one of those models; it assumes that successive price

changes are independent and identically distributed random variables, so that future price

changes cannot be predicted from historical price changes. Hence, the RWH has some

testable implications for the weak-form of EMH.

102

Several tests have been suggested to test for EMH, the procedures used to test for RWH

were chosen on the basis of the implications of EMH. If all relevant and available

information is fully reflected in stock price, then:

a) Successive price changes will be independent, so that there will be no serial

correlation over time between returns;

b) Successive price changes will be identically distributed:

log (Pt) = log (Pt-1) + εt

Where εt is an independent standard random variable, that is; a series of identically

distributed random variables with zero mean and variance equal to unity.

So the distribution of the changes in stock price must be stationary over time, i.e. stock

prices are I(1) while stock returns are I(0).

In this research, in addition to the common models used in the literature, we

employ the most recent statistical and econometric models. To test for the independence

of successive price changes (condition a) we employ runs test, non-parametric tests for

detecting the frequency of the changes in the direction of a time series. In addition,

estimated serial autocorrelations are performed at various lags to determine whether the

autocorrelation between returns is equal to zero. Further, we utilize the Box-Pierce test to

determine whether the autocorrelation is equal to zero, based on the sum of squares of the

first K autocorrelation coefficients. If the set of auto correlations does not differ from the

null set, randomness of returns is implied. However, it is important to test whether

successive price changes are identically distributed (condition b). Hence, we use

regression analysis, variance ratio, and BDS tests.

Moreover, a major difficulty in interpreting the results from tests on thinly traded

markets is the confounding effect of infrequent trading on the observed index. Thus

rejection of the RWH or the efficient markets hypothesis could simply be a result of

having used the observed index. So it is important to take in account the effect of

infrequent trading and adjust the observed indices for thin trading.

Infrequent trading is widespread in most emerging markets and it is particularly

so in the case of the markets under examination here. Infrequent trading has two forms:

103

The first occurs when stocks are traded every consecutive interval, but not necessary at

the close of each interval. This form of infrequency, often dubbed “nonsynchronous

trading” has been studied by Scholes and Williams (1977a, 1977b) and Muthuswamy

(1990). Infrequent trading is also said to occur when stocks are not traded every

consecutive interval, Fisher (1966), Dimson (1979), Cohen et al. (1978, 1979), Lo and

MacKinlay (1990), and Stoll and Whaley (1990b) focus on this “non-trading” and its

consequences.

The key to distinguishing nonsynchronous trading from non-trading is the interval

over which price changes or returns are computed. When returns are measured on a

monthly basis, virtually all stocks will have been traded at least once, but not all stocks

will have been transacted exactly at the close of trading on the last trading day of the

month, that is nonsynchronous trading. When returns are measured over trading intervals

as short as for example fifteen minutes, however, all stocks in the market are unlikely to

have been traded at least once in every consecutive fifteen minutes interval, which is non-

trading. As the trading interval shrinks, nonsynchronous trading becomes non-trading.

The problem is created by the fact that the value of an asset over a certain time cannot be

directly observed, if the asset does not trade in that period. Since most indices are

computed on the basis of the most recent transaction prices of the constituent stocks, the

reported index becomes stale in the presence of infrequent trading; the result is that the

observed index does not reflect the true value of the underlying stock portfolio.

One of the consequences of infrequent trading is the spurious serial correlation it

induces in the observed index returns. Therefore, observed dependence is not necessarily

evidence of predictability, but rather may be a statistical illusion brought about by thin

trading. A number of different approaches have been suggested to correct for infrequent

trading, Stoll and Whaley (1990) use the residual from an ARMA (p,q) regression as a

proxy of the true index, whereas; Bassett et al. (1991) propose the use of a Kalman filter

to estimate the distribution of the true index, Jokivuolle (1995) suggests a modified

version of the Stoll and Whaley approach to estimate the true unobserved index from the

history of the observed index, the correction consists of decomposing the log of the

observed index into its random and stationary components, using the Beveredge and

104

Nelson (1981) methodology, in this; the random component can be shown to equal the

log of the true index.

To separate the effect of infrequent trading, the approach proposed by Miller,

Muthuswamy and Whaley (1994) has been applied. To correct for infrequent trading, this

methodology basically suggests that to remove the impact of thin trading, a moving

average model (MA) that reflects the number of non-trading days should be estimated

and then returns be adjusted accordingly. However, given the difficulties in identifying

the non-trading days, Miller et al. have shown that it is equivalent to estimate an AR (1)

model from which the non-trading adjustment can be obtained. Specifically, this model

involves estimating the following equation:

R t = a 1 + a 2 R t −1 + ε t (4-2)

Using the residual from the regression, adjusted returns are estimated as follows:

adj ε t

R = (4-3)

t

(1 − a 2 )

Miller, Muthuswamy, and Whaley find thin trading adjustment reduces the

negative correlation among returns. The model above assumes that non-trading

adjustment is constant over time, while this assumption may be correct for highly liquid

markets; it is not the case for emerging markets. Therefore, equation (4-2) will be

estimated recursively.

In a random walk process, the distribution of the changes in stock prices must be

stationary over time and the constant term of the stationary series, should be

insignificantly different from zero. Consequently, to test for this proposition, we start our

analysis with naïve random walk, which is closely associated with weak-form EMH

105

Pt = Pt-1+εt (4-4)

Where Pt = ln (Xt) represents the natural log of the original time series Xt; and εt is a zero-

mean pure white noise random variable. If the random walk hypothesis holds, then the

series Pt will have a single unit root (i.e. will be I(1)) and the series ∆Pt (= Pt-Pt-1 = Ln

(Xt/Xt-1) will be purely random. The series ∆Pt or Rt may be examined further by

estimating the equation

Rt = constant + εt (4-5)

Using ordinary least squares, under the random walk hypothesis, the constant term should

be insignificantly different from zero and the resultant residuals should be uncorrelated.

Additionally, the following regression will be estimated

Rt = a + Rt −1 + ε t (4-6)

Once again if the RWH holds, the constant term and returns lag should be insignificantly

different from zero, and εt to be a white noise random variable.

Serial correlation (or autocorrelation) test measures the correlation coefficient

between a series of returns and lagged returns in the same series. A significant positive

serial correlation implies that a trend exists in the series, whereas, a negative serial

correlation indicates the existence of a reversal in price movements. A return series that is

truly random will have a zero serial correlation coefficients. The beta coefficient from the

following regression equation measures the serial correlation of stock i with a lag of K

periods:

ri ,t = a i + β i ri ,t − k + ε i ,t (4-7)

106

Where ri,t represents the return of stock i at time t; αi is a constant; βi is the lagged

return’s coefficient; εi,t represents random error, while k represents different time lags.

The serial correlation test assumes normal distribution for the stock price changes

(returns).

The null hypothesis to be tested is that no significant correlation exists between

price changes, i.e. β1= β2=...= βj=0. Since random walk 1 implies that all

autocorrelations are zero, a simple test statistic of random walk 1 that has the power

against many alternative hypotheses is the Q-statistics due to Box and Pierce (1970):

m

Qm ≡ T ∑ ρ 2 ( k ) (4-8)

k =1

m

under the random walk 1 null hypothesis, it is easy to see that Qˆ m = T ∑ ρˆ 2 (k ) is

k =1

asymptotically distributed as χ m2 . Ljung and Box (1978) provide the following finite-

sample correction which yields a better fit to the χ m2 for small sample size:

m

ρ 2 (k )

Qm' ≡ T (T + 2)∑ (4-9)

k =1 T −k

to detect departures from zero autocorrelations in either direction and at all lags.

Therefore, it has power against a broad range of alternative hypothesis to the random

walk. However, selecting the number of autocorrelations m requires some care-if too few

are used, the presence of higher-order autocorrelation may be missed; if too many are

used, the test may not have much power to insignificant high-order autocorrelations.

107

4-1-4 Non-parametric runs test

A common test for random walk 1 is the runs test, in which the number of

sequences of consecutive positive and negative returns, or runs, is tabulated and

compared against its sampling distribution under the random walk hypothesis. For

example, a particular sequence of 10 returns may be represented by 1001110100,

containing three runs of 1s (of length 1, 3, and 1, respectively) and three runs of 0s (of

length 2, 1, and 2, respectively), thus six runs in total. In contrast, the sequence

0000011111 contains the same numbers of 0s and 1s, but only 2 runs. By comparing the

number of runs in the data with the expected number of runs under random walk 1, a test

of the IID random walk hypothesis may be constructed. To perform the test, we require

the sampling distribution of the total number of runs Rruns in a sample of n. Mood (1940)

was the first to provide a comprehensive analysis of runs.

Moreover, the runs test determines whether successive price changes are

independent. Unlike its parametric equivalent the serial correlation test, the runs test does

changes with the same sign, if the returns series exhibit grater tendency of change in one

direction, the average run will be longer and the number of runs fewer than that generated

by random process. To assign equal weight to each change and to consider only the

direction of consecutive changes, each change in returns was classified as positive (+),

negative (-), or no change (0). The runs test can also be designed to count the direction of

change from any base; for instance, a positive change could be one in which the return is

grater than the sample mean, a negative change one in which the return is less than the

mean, and zero change representing a change equal to the sample mean. The actual runs

(R) are then counted and compared to the expected number of runs (m) under the

108

3

N ( N + 1) − ∑ n i

2

m = i=1 (4-10)

N

Where N is the total number of return observations and ni is a count of price change in

to a normal distribution with a standard error (σm) of runs as specified in equation (4-11).

1

3 3 3

2

σm = ∑ n i2 ∑ n i2 + N ( N + 1) − 2 N ∑ n i3 − N 3 (4-11)

i =1 i =1 i =1

The standard normal Z-statistic (Z=(R-m)/σm) can be used to test whether the

actual number of runs is consistent with the independence hypothesis. When actual

number of runs exceed (fall below) the expected runs, a positive (negative) Z value is

obtained. Positive (negative) Z value indicates negative (positive) serial correlation in the

return series.

to the random walk process linearly increases with the sampling interval. Lo and

MaCkinlay (1988) proposed a simple specification test for evaluating the random walk

properties of asset prices. Specifically, if Xt is a pure random walk, the ratio of the

variance of the qth difference scaled by q to the variance of the first difference must be

unity. A variance ratio that is grater than one suggests that returns series is positively

109

serially correlated or that the shorter interval returns trend within the duration of the

longer interval. A variance ratio that is less than one suggests that the returns series is

negatively serially correlated or that the shorter interval returns tend toward mean

reversion within the duration of the longer interval. The variance ratio VR (q) is defined

as:

2

σ (q )

VR ( q ) = 2

(4-12)

σ (1 )

Where σ2 (q) is 1/q the variance of the q-differences and σ2 (1) is the variance of the first

differences.

1 nq

σ 2 (q) = ∑

m i=q

( x i − xi − q − qµˆ ) 2 (4-13)

Where:

q

m = q (nq − q + 1 ) 1 −

nq

And

nq 2

1

(nq − 1) ∑

σ 2 (1) = (x i − x i −1 − µˆ ) (4-

i =1

14)

Where:

1

µˆ = (x nq − x0 )

nq

increments. Because it is the heteroscedasticity in the data that is of interest, we use the

more robust heteroscedastic test statistic that uses overlapping intervals. The test statistic

is:

110

* VR ( q ) − 1 (4-15)

Z (q ) = 1

≈ N ( 0 ,1 )

[Φ *

(q ) ]

2

Where:

2

* q −1

2(q − j) ˆ

Φ (q ) = ∑ δ ( j)

j =1 q

And

nq

∑ (x − x i −1 − µˆ ) (x i − j − x i − j −1 − µˆ )

2 2

i

δˆ ( j ) =

i = j +1

nq

∑ [(x ]

2 2

i − x i −1 − µˆ )

i =1

The null hypothesis for the BDS test (Brock et al., 1987, revised in 1996) is that the

data are independently and identically distributed (iid), and any departure from iid should

lead to rejection of this null in favor of an unspecified alternative. Hence the test can be

considered a broad portmanteau test which has been shown to have reasonable power

against a variety of nonlinear data generating processes (see Brock et al., 1991 for an

extensive Monte Carlo study). The BDS test statistic is calculated as follows. First, the

‘m-histories’ of the data xtm = ( xt , xt +1 ,..., xt − m +1 ) are calculated for t = 1, 2,…, T-m for

some integer embedding dimension m ≥ 2 . The Cointegration integral is then computed,

which counts the proportion of points in m-dimensional hyperspace that are within a

distance ε of each other:

2

c m ,T ( ε ) = ∑

(T − m + 1)(T − m ) t < s

I ε ( x tm , x sm ) (4-16)

Where Iε is an indicator function that equals one if xtm − x sm < ε and zero otherwise, and

||.|| denotes the sup. norm. BDS shows that, under the null hypothesis that the observed χt

111

are iid, then c m,T (ε ) → c1 (ε ) m with probability one as the sample size tends to infinity and

ε tends to zero. The BDS test statistic, which has a limiting standard normal distribution,

then, follows as:

[ C m , T (ε ) − C 1 , T (ε ) m ]

w m ,T (ε ) = T 1 / 2

σ m ,T (ε )

Where

m −1

σ m ,T (ε ) = 2[ K m + 2(∑ k m − j C1,T (ε ) 2 j ) − ( m − 1) 2 C1,T (ε ) 2 m − m 2 KC1,T (ε ) 2 m − 2 ]1 / 2

j =1

6 ∑

t< s< r

h ε ( x tm , x sm , x rm )

K (ε ) =

[(T − m + 1 )(T − m )(T − m − 1 )]

and he (i, j, k) = [Iε (i, j)Iε ( j, k) + Iε (i, k)Iε (k, j) + Iε ( j,i)Iε (i, k)]/ 3

Two parameters are to be chosen by the user: the value of ε (the radius of the

hypersphere which determines whether two points are ‘close’ or not), and m (the value of

the embedding dimension). Brock et al. (1991) recommend that ε is set to between half

and three halves the standard deviation of the actual data and m is set in line with the

number of observations available (e.g. use only m ≤ 5 for T ≤ 500 etc.)

The weak-form of efficient market hypothesis implies that no past realizations

should help predict future values, a model that reveals a pattern in the behavior of daily

returns violates the weak form of market efficiency. Excessive volatility of stock prices is

an important phenomenon to investigate, because of its negative effect on risk-averse

investors. In this section, volatility structure of Arab markets returns will be analyzed

using several techniques such as GARCH (1,1), EGARCH (1,1), and Schewrt model.

112

4-2-1 Generalized autoregressive conditional heteroskedasticity (GARCH)

The basic idea behind autoregressive conditional heteroskedasticity ARCH

models proposed by Engle (1982) is that, the second moments of the distribution may

have an autoregressive structure. Under rational expectations the forecast error is ut+1 =

yt+1-Et(yt+1), and the conditional distribution of yt+1 is assumed to be normal with mean

µt+1 and var(yt+1/Ωt) = ht+1 = a0+a1 u2t, where Ωt is the information set available at time t.

However, the ARCH process has a memory of only one period. To generalized this we

can start adding lags of ut-1 in the equation ht+1, ι = 1,…,q. but then the number of

parameters to estimate increases rabidly (Bollerslev 1986). For example, in the GARCH

(1,1) model the conditional variance depends on lagged variance terms: ht+1 = a0+a1+β1ht

= a0+(a1+ β1)ht+at(ut2-ht) in addition to the lagged ut where u0 is arbitrarily assumed to

be fixed and equal to zero. The parameters can be estimated by maximum likelihood

techniques. Conditional on time t information Ωt, (ut2-ht) has a mean of zero, and can be

thought as the shock to volatility. The coefficient a1 measures the extent to which a

volatility shock today feeds through into the volatility of the next period, while a1+ β1

measures the rate at which this effect dies out over time.

Since Engle’s seminal work, many generalization of this model have been

reported. For example, the GARCH (1,1) with a1+ β1=1 has a unit autoregressive root, so

that today’s volatility affects forecasts of volatility in to the indefinite future (persistent of

volatility), this is therefore known as the integrated GARCH or IGARCH model. Nelson

(1991) introduced the Exponential GARCH (EGARCH) model which allows for

asymmetric shocks to volatility and tests the leverage effect. The dependence of the

second moment in returns captured by the (G)ARCH process is known as volatility

clustering, i.e. large changes in price volatility are followed by large changes in either

sign.

Leverage terms allow more realistic modeling of the observed asymmetric

behavior of stock returns according to which a “good-news” price increase yields lower

volatility, while some “bad-news” decrease in price yields an increase in volatility. For

example, when the value of (the stock of) a firm falls, the debt-to-equity ratio increases,

which in turn leads to an increase in the volatility of the returns to equity. This suggests

that returns could also be described by an autoregression whose residual follows an mth-

113

order ARCH-L process, where L stands for the leverage effect (Hamilton and Susmel,

1994). It is also worth mentioning a two-component GARCH which reflects differing

short- and long-term volatility dynamics (Ding et al., 1993).The GARCH in mean

(GARCH-M) model could be used to capture direct relationships between return and

possibly time-varying risk by including the conditional variance in the model for the

conditional mean of the variable of interest.

Autoregressive Conditional Heteroskedasticity (ARCH) models are specifically

designed to model and forecast conditional variances. The variance of the dependent

variable is modeled as a function of past values of the dependent variables and

independence, or exogenous variables. ARCH models are introduced by Engel (1982)

and generalized as GARCH by Bollerslev (1986). GARCH models were found to be

extremely useful in economic and finance, because it is very flexible in modeling second

2

moment. If the error term process is ε t v t ht

where σ v = 1 , E(vt)=0 and in the standard

GARCH (1,1):

h t = α 0 + α 1 ε t2−1 + β 1 h t −1 (4-17)

Then the sequence {vt} is a white noise process and the conditional and unconditional

means of εt are equal zero. A model for the mean is estimated as:

R t = β R t −1 + ε t (4-

18)

and ε t = R t − β R t −1

In the conditional variance equation written in (4-17), ht is the one-period ahead forecast

variance based on past information, it is called the conditional variance. Moreover, the

conditional variance equation (4-17) is a function of three terms:

• The mean, α0

• News about volatility from the previous period, measured as the lag of the

squared residual from mean equations ε t2−1 (the ARCH term).

114

• Last period forecast variance ht-1 (the GARCH term).

Additionally, the sum of the parameters α1, β1 in the conditional variance equation,

measures the persistence in volatility and lies between 0 and 1.

The (1,1) in GARCH(1,1) refers to the presence of a first-order ARCH term (the

first term in parentheses) and a first-order GARCH term (the second term in parentheses).

An ordinary ARCH model is a special case of a GARCH specification in which there are

no lagged forecast variances in the conditional variance equation.

The GARCH models are estimated by the method of maximum likelihood under

the assumption that the errors are conditionally normally distributed. For example, for the

GARCH (1,1) model, the contribution to the log likelihood from observation t is

1 1 1

2 2 2

(2

lt = − log(2π ) − logσ t2 − yt | − χt' y / σ t2 , ) (4-19)

where

σ t2 = ω + α ( yt −1 − χt −1 ' y )2 + βσ t2−1 (4-20)

trader predicts this period’s variance by forming a weighted average of a long term

average (the constant), the forecasted variance from last period (the GARCH term), and

information about volatility observed in the previous period (the ARCH term). If the asset

return was unexpectedly large in either the upward or the downward direction, then the

trader will increase the estimate of the variance for the next period. This model is also

consistent with the volatility clustering often seen in financial returns data, where large

changes in returns are likely to be followed by further large changes.

There are two alternative representations of the variance equation that may aid in

the interpretation of the model:

• If we recursively substitute for the lagged variance on the right-hand side of

the variance equation, we can express the conditional variance as a weighted

average of all of the lagged squared residuals:

115

∞

ω

2

σ = + α ∑ β j −1ε t2− j (4-21)

t

(1 − β ) j =1

sample variance, but that it down-weights more distant lagged squared errors.

• The error in the squared returns is given by ut = ε t2 − σ t2 . Substituting for the

variances in the variance equation and rearranging terms we can write our

model in terms of the errors:

ε t2 = ω + (α + β )ε t2−1 + ut − β vt −1 (4-22)

• Thus, the squared errors follow a heteroscedastic ARMA (1,1) process. The

autoregressive root which governs the persistence of volatility shocks is the

sum of α plus β. In many applied settings, this root is very close to unity so

that shocks die out rather slowly.

(EGARCH)

The EGARCH or exponential GARCH was proposed by Nelson (1991), the

specification of conditional variance is

ε t −1 2 ε

log σ t2 = ω + β log σ t2−1 + α − + γ t −1 (4-

σ t −1 π σ t −1

23)

Note that the left hand side is the log of the conditional variance, this implies that

the leverage effect is exponential, rather than quadratic and the forecasts of the

conditional variance are guaranteed to be non-negative. The presence of leverage effects

can be tested by the hypothesis that γ>0. The impact is asymmetric if γ≠0; while ε

follows a generalized error distribution. Having estimating the EGARCH model, we will

be able to plot a News Impact Curve, since it is often observed that downward

116

movements in the market are followed by higher volatilities than upward movements. To

account for this phenomenon, Engle and Ng (1993) describe a News Impact Curve with

asymmetric response to good and bad news.

4-2-3 Schewrt model

In order to compare the volatility of Arab stock markets with other international

and emerging markets, Schewrt approach will be used (Schewrt, 1989). A two-step

regression technique is applied to estimate volatility of returns. In the first step, a 13th-

order auto-regression for returns will be estimated

13

Rt = ∑β

i =1

i R t −1 + ε t (4-24)

The absolute value of the residual from equation (4-24) is an estimate of the standard

deviation of the return for t.

In the second step, a 13th-order auto-regression for the absolute values of the

errors from equation (4-24) will be estimated

13

εˆ t = ∑

i =1

ρ τ εˆ t − i + u t (4-25)

the fitted values from this second equation, multiplied by (2/π)-1/2 , are estimates of the

conditional return standard deviation given information available before day t. After the

volatility measures are estimated for each market separately, an average measure of

volatility is then constructed for each group of markets. This measure is calculated by

taking the weighted average of the different market volatilities, with the weights

representing the share of each market in the total market capitalization of the group.

Market efficiency implicitly assumes that investors are rational, where rationality

implies risk aversion, unbiased forecasts and instantaneous responses to information.

Such rationality leads to prices responding linearly to new information. However,

117

emerging markets, especially during the early years of trading, may be characterized by

investors who do not have all these attributes. In particular, investors may not always

display risk aversion. For example, Benartzi and Thaler (1992) argue that investors may

be loss averse, in that they are more sensitive to losses than to gains. Such loss aversion

may lead to investors acting in a manner consistent with risk loving or risk neutral

behavior. For example, loss-averse investors who have incurred losses may display risk

loving behavior in an attempt to recover those losses. In addition, investors may place too

faith in their own forecasts introducing bias into their actions (Dabbas et al., 1991; and

Fraser and MacDonald, 1993). Similarly, as Schatzberg and Reiber (1992) point out,

investors do not always respond instantaneously to information. In particular, uninformed

traders may delay their response to see how informed market participants behave,

because they do not have the resources to fully analyze the information, or because the

information is not reliable. Such examples of investor behavior may result in prices

responding to information in a non-linear fashion.

There are several reasons why non-linearity may be observed in financial markets.

First, the characteristic of the market microstructure may lead to non-linearity because of

difficulties in carrying out arbitrage transactions. For example, differing microstructures

between stock markets and derivative markets may give rise to non-linear dependence.

Stoll and Whaley (1991) show that price discovery takes place in futures market and then

the information is carried to the stock market through the process of arbitrage. Delays in

transacting the stock market leg of the arbitrage means that the immediate response to the

mispricing would only be partial, reflecting the change in the futures price alone. This

may induce further arbitrage activity and could actually result in overshooting of the

arbitrage bounds. Furthermore, short sales in stock markets may lead to delays in

executing arbitrage transactions; this in turn may cause non-linear behavior.

Second, non-linearity may be explained in terms of non-linear feed back

mechanisms in price movements. When the price of an asset gets too high, self-regulating

forces usually drive the price down. If the feed back mechanism is non-linear, then the

correction will not always be proportional to the amount by which the price deviates from

the asset’s real value. Third, non-linearity could arise because of the presence of market

imperfections such as transaction costs. Although information arrives randomly to the

118

market, market participants respond to such information with a lag, due to transactions

costs. In other words, market participants do not trade every time news comes to the

market; rather, they trade whenever it is economically profitable, leading to clustering of

price changes.

Fourth, when announcements of important factors are made less often than the

frequency of observations, non-linearity may be observed. For example, monthly money

supply announcements will cause non-linearity in daily and weekly series, but not in

quarterly series. A fifth reason relates to the fact that, as mentioned above, capital market

theory is based on the notion of rational investors. It is assumed that investors are risk

averse, unbiased when they set their subjective probabilities and always react to

information as it received. The implication is that the data generating process is linear.

However, investors may well be risk lovers when taking a gamble in an attempt to

recover their losses. Moreover, they may have too much faith in their own forecast, thus

introducing bias into their subjective probabilities. In addition, they may not react to

information instantaneously, but may delay their response until other investors reveal

their preferences.

To investigate the existence of non-linearity in return series, the logistic map will

be used

Rt = α 0 + α 1 Rt −1 + α n Rtn−1 + ε t (4-26)

where Rt is the return at time t; and n = 2,3. For EMH to be hold, we would expect

α0=α1=αn=0 and εt to be a white noise process. The main purpose of using this approach

is not to determine the precise nature of any non-linearity, but rather to ascertain whether

any non-linearity exists.

Moreover, evidence of non-linearity per se does not provide an insight into the

sources of the non-linearity, or more importantly, the appropriate functional form for the

resultant non-linear model. For instance, Siriopoulos et al. (2001) find that the

inefficiency observed during the early years of their sample for Athens stock exchange,

was manifested through non-linear behavior, while efficiency has been improved with

time, according to institutional and regulatory evolution. Interest in non-linear chaotic

119

processes has in the recent past, experienced a tremendous rate of development. There are

many reasons of this interest, one of which being the ability of such process to generate

output that mimics the output of stochastic systems, thereby offering an alternative

explanation for the behavior of asset prices. In fact, the possible existence of chaos could

be exploitable and even invaluable. If, for example, chaos can be shown to exist in asset

prices, the implication would be that profitable, non-linearity-based trading rules exist (at

least in the short run and provided the actual generating mechanism is known).

Prediction, however, over long periods is all but impossible, due to the sensitive

dependence an initial conditions property of chaos.

In this contest, the BDS tests the null hypothesis of whiteness (independent and

identically distributed observations) against an unspecified alternative using a

nonparametric technique. Since the asymptotic distribution of the BDS test statistic is

known under the null hypothesis of whiteness, the BDS test provides a direct (formal)

statistical test for whiteness against general dependence, which includes both non white

linear and non white non-linear dependence. Hence, the BDS test does not provide a

direct test for non-linearity or for chaos, since the sampling distribution of the test

statistic is not known under the null hypothesis of non-linearity, linearity, or chaos. It is,

however, possible to use the BDS test to produce indirect evidence about non-linear

dependence [whether chaotic (i.e. non-linear deterministic) or stochastic], which is

necessary but not sufficient for chaos.

Moreover, the BDS test has reasonable power against the GARCH family of

models. However, it is often difficult to disentangle the non-linearity generated by this

form of dependence in the second moment, from non-linearity arising as a result of other

causes. One solution to this problem is to estimate some form of GARCH for the series

Rt, such as

Rt = µ + ut ut ∼N(0,ht)

ht = α 0 + α 1u t2−1 + β 1 ht −1

120

the standardized residual, u t ht−1 / 2 , may be subjected to the BDS test and the null

hypothesis then becomes one that the specified GARCH model is sufficient to model the

non-linear structure in the data against an unspecified alternative that is not. The

conclusion being that if the BDS test cannot reject the iid null using appropriate critical

values derived from simulation, then the model estimated is assumed to be an adequate

characterization of the data. In other words, it has been suggested (for example, Brock et

al. 1991, p.19 or p.69) that the BDS test can be used as a general test of model mis-

specification.

In this area, the effectiveness of the BDS test in detecting neglected asymmetries

in volatility will be examined. Engle and Ng (1993) and Henry (1998) discus the

difficulties in selecting between symmetric and asymmetric GARCH models, the

standardized residuals of the GARCH models will be subjected to BDS test to “see what

is left” and whether non-linearity generated by this form of dependence in the second

moment or from other causes.

An alternative approach to test the EMH especially the weak form is to test for

seasonal patterns or calendar effects in stock returns, since according to the weak form of

EMH, stock prices in an efficient market should fluctuate randomly through time in

response to the unanticipated components of news. This implies that the future path of

price level of an asset is no more predictable than the path of a series of cumulated

random numbers. As a result, seasonal patterns should not exist or should be minor, since

their existence implies the possibility of obtaining abnormal returns by making time

research strategies. In this sequence, three calendar effects will be examined, the day-of-

the-week effect, January effect, and the Halloween effect.

The day-of-the-week effect was studied using a model originally proposed by

French (1980), the hypothesis to be evaluated is the trading time hypothesis, according to

which returns are created only on the working days of the week. This hypothesis is tested

using regression with dummy variables, such as

121

5

Rt = ∑ α i Dit + ε t (4-27)

i =1

where Rt is the daily logarithmic returns on the general index; Dit is a dummy variable

taking the value 1 for day i and 0 for all other days (i=1,…,5 corresponding to Monday

through Friday); αi is the mean return on day i and εt is an error term assumed to be iid.

The hypothesis tested in equation (4-27) is Ho:α1=α2=α3=α4=α5.

To test for the January effect, the model described by the following equation will

be used (Gultekin and Gultekin 1983; Jaffe and Westerfield 1989; Raj and Thurston

1994):

12

Rt = ∑αi =1

i D it + ε t (4-28)

here Dit takes the value 1 if the return at time t belongs to month i and 0 if it belongs to

any other month (i=1,…,12 corresponds to January through December); αi is the mean

return in month i; and the other variables are defined as in equation (4-27). The

hypothesis to be tested in equation (4-28) is H0:α1=α2=...=α12.

Moreover, the mean return in a number of months exceeds the average mean

return and that this return would not appear to be a reflection of risk, as reflected in the

standard deviations of monthly returns. One can test this formally by examining whether

there exists a simultaneous month of the year effect in mean return, and in the standard

deviation of these returns. A formal test of the existence of monthly calendar effects in

mean returns is given by the ANOVA or Kruskal-Wallis statistics. Let R2j be the average

rank of observations in the jth group (each month of the year) and nj be the number of

observations in the jth group. Then with K groups and N observations in total, the

Kruskal-Wallis H statistic is

122

12 k R2

N ( N + 1) ∑

H = − 3( N + 1)

j

(4-29)

j =1 n j

A formal test for monthly variation in the second moment is given by the Levene

test, which tests the hypotheses H0: σi=σj ∀i,j , Ha: σi≠σj, at least one i,j pair. The test

statistic is defined as

k

(N − K )∑ (Z i − Z j )2

i =1

W = k N

(4-30)

(K − 1)∑∑ (Z ij − Z i ) 2

i =1 j =1

~ ~

where Z ij = Υij − Υi , Υi the median of sup-group i.

To test for the existence of the Halloween effect, Bouman and Jacobsen (2002)

use regression analysis with dummy variables, which is equivalent to a simple means test,

their analysis is represented as:

Rt = µ + α1 st + ε t (4-31)

where Rt represents the continuously compounded index returns defined as the natural

logarithm of the price relatives. The dummy variable st takes on the value 1 if observation

t falls in month within the November-April period, and 0 otherwise. The intercept term µ

represents the mean return over the May-October period and µ+α1 represents the mean

return over the November-April periods, if α1 is positive and significant at a meaningful

level, then this is considered as an indication of a Halloween effect. However, the

unusually large monthly returns documented by Bouman and Jacobson (2002) during

November-April periods could be a symptom of the January effect. To test for this

123

possibility, equation (4-31) is modified by inserting a second dummy variable Jt, which is

set to equal 1 whenever month t is a January and 0 otherwise.

Rt = µ + α1 st +α 2J t + ε t (4-32)

4-5-1 Random walk properties

- Regression analysis

Table 4-1 and 4-2 show the results of the random walk carried out for the whole

period for the observed indices. Table 4-1 indicates that in 6 out of 9 markets, the null

hypothesis that the constant term is insignificantly different from zero can not be rejected.

While for both Saudi and Bahrain, the null hypothesis can be rejected at the 5% level.

Moreover, the results in table 4-1 indicate that for Abu Dhabi, Jordan and Kuwait, the

constant term found to be significantly different from zero at all acceptable levels.

However, when the return lag has been added to the model but without correction for

infrequent trading (table 4-2), the results still the same for both Palestine and Dubai. That

is, the coefficients α0 ,α1 found to be insignificantly different from zero, but for Saudi;

they found to be significant at 10% level, whereas, the coefficients found to be

significantly different from zero for the other markets.

Moreover, the RWH indicates that the residual term εt should be pure white noise

error. To test for this property, several diagnostic tests have been used and the results

presented in appendix 2 for each model. Table (1) in appendix 2 shows that the residuals

from the RW model presented in table 4-1 violate the white noise assumption, since they

were serially correlated except Dubai. Since only BDS test found to be significant, this

indicates that the residuals are not iid.

124

Table 4-1

Random Walk Model for Observed Indices

R t = a + εt

Market Coefficient Std.Error t-value P -value

Abudhabi 0.000847 0.000212 3.995 0.000

Bahrain 0.000257 0.000126 2.041 0.041

Dubai 0.000435 0.000304 1.429 0.153

Egypt 0.000425 0.000408 1.041 0.298

Jordan 0.000355 0.000131 2.702 0.007

Kuwait 0.001876 0.000396 4.734 0.000

Oman 0.000251 0.000249 1.008 0.314

Palestine 0.00082 0.000535 1.533 0.126

Saudi 0.00037 0.000168 2.202 0.028

Table 4-2

Random Walk Model for Observed Indices

Rt = a0 + a1Rt-1 +εt

Market coefficient std.error t-value P -value

a0 0.0016 0.0004 3.923 0.000

Kuwait

a1 0.1657 0.0377 4.392 0.000

a0 0.0003 0.0004 0.839 0.402

Egypt

a1 0.2061 0.0240 8.585 0.000

a0 0.0004 0.0002 2.274 0.023

Saudi

a1 -0.0333 0.0180 -1.849 0.065

a0 0.0003 0.0001 2.336 0.020

Bahrain

a1 -0.1304 0.0172 -7.581 0.000

a0 0.0008 0.0005 1.561 0.119

Palestine

a1 -0.0165 0.0291 -0.566 0.572

a0 0.0004 0.0003 1.417 0.157

Dubai

a1 0.0071 0.0302 0.235 0.814

a0 0.0003 0.0001 2.128 0.033

Jordan

a1 0.2310 0.0174 13.256 0.000

a0 0.0002 0.0002 0.934 0.350

Oman

a1 0.0706 0.0229 3.083 0.002

a0 0.0007 0.0002 3.428 0.001

Abudhabi

a1 0.1412 0.0391 3.613 0.000

Tables 4-3 and 4-4 present the results for returns corrected for infrequent trading.

In general, the adjustment of returns to take account of thin trading appears to have

125

removed the apparent predictability shown in table 4-1 and 4-2. With one exception for

Saudi, Oman, and Palestine (see table 4-4). Since the coefficients α0 ,α1 found to be

statistically significant at 5 % level. Furthermore, the diagnostic tests indicate that only

the residuals for Dubai, Kuwait, and Egypt are not serially correlated, while all other

diagnostic tests were significant, indicating that the residuals do not follow white noise

process (see table 5 in appendix 2). While the diagnostic tests for other markets indicate

that the residuals are serially correlated. The result for Egypt presented in table 4-1 is

consistent with those of Omran (2002) but not for Jordan, since he finds that the return in

Jordanian stock market is not predictable when he uses the observed index.

Table 4-3

Random Walk Model for Corrected Indices

Radjt = a + εt

Market Coefficient Std.Error t-value P -value

Abudhabi -0.00015 0.00025 -0.622 0.534

Bahrain -0.00009 0.00011 -0.800 0.424

Dubai 0.00066 0.00031 2.157 0.031

Egypt 0.00096 0.00050 1.911 0.056

Jordan -0.00005 0.00017 -0.327 0.743

Kuwait 0.00043 0.00039 1.094 0.274

Oman -0.00067 0.00027 -2.500 0.013

Palestine -0.00127 0.00054 -2.373 0.018

Saudi 0.00045 0.00016 2.743 0.006

Tests for the absence of serial correlation over time between returns were

implemented from lag 1 up to lag 30 for both observed and corrected indices. Table 4-5

shows that for the observed indices and on the base of the Box-Pierce test, there is highly

significant autocorrelation for all lags at the 1% level except Dubai, implying that the

series are not completely random.

126

Table 4-4

Random walk Models for Corrected Indices

adj adj

Rt = a0 + a1 R t-1 +εt

Market coefficient std.error t-value P -value

a0 0.0004 0.0004 1.084 0.279

Kuwait

a1 -0.0051 0.0383 -0.132 0.895

a0 0.0010 0.0005 1.936 0.053

Egypt

a1 -0.0147 0.0246 -0.597 0.551

a0 0.0005 0.0002 3.233 0.001

Saudi

a1 -0.1652 0.0178 -9.280 0.000

a0 -0.0001 0.0001 -0.819 0.413

Bahrain

a1 -0.0150 0.0174 -0.866 0.386

a0 -0.0011 0.0005 -2.063 0.039

Palestine

a1 0.1388 0.0289 4.801 0.000

a0 0.0007 0.0003 2.192 0.029

Dubai

a1 -0.0118 0.0303 -0.390 0.697

a0 -0.0001 0.0002 -0.370 0.711

Jordan

a1 0.0127 0.0179 0.710 0.478

a0 -0.0005 0.0003 -1.808 0.071

Oman

a1 0.3029 0.0219 13.824 0.000

a0 -0.0001 0.0002 -0.576 0.565

Abudhabi

a1 0.0582 0.0395 1.473 0.141

However, when the corrected indices have been used (table 4-6), the higher P-

values for each of Dubai, Egypt, and Kuwait show that we can not reject the hypothesis

that the series are random at the 5 % level while the other markets still exhibit

autocorrelation in returns even after the adjustment for infrequent trading. These results

must be considered under caution, since serial correlation test is a parametric test

assuming that return series normally distributed. While none of the series under

examination came from normal distribution (see Jarque-Bera statistic in table 3-18). As a

result, it is more appropriate to use a non-parametric test such as the runs test.

127

Table 4-5

Estimated Autocorrelations for Observed Indices Returns

No. Abudhabi Bahrain Dubai Egypt Jordan

Lags Q stat. P -value Q stat. P -value Q stat. P -value Q stat. P -value Q stat. P -value

1 12.38 0.000 56.42 0.000 0.04 0.848 72.11 0.000 167.71 0.000

2 12.41 0.002 64.65 0.000 1.73 0.422 72.41 0.000 167.83 0.000

3 24.46 0.000 72.54 0.000 1.74 0.628 72.85 0.000 168.19 0.000

4 30.38 0.000 76.59 0.000 1.84 0.765 74.29 0.000 170.65 0.000

5 31.12 0.000 79.07 0.000 2.10 0.836 74.46 0.000 170.77 0.000

10 40.30 0.000 98.34 0.000 5.06 0.887 78.95 0.000 174.20 0.000

20 68.04 0.000 106.52 0.000 7.84 0.993 89.13 0.000 215.84 0.000

30 83.85 0.000 113.29 0.000 11.57 0.999 111.36 0.000 236.27 0.000

Lags Q stat. P -value Q stat. P -value Q stat. P -value Q stat. P -value

1 19.04 0.000 9.93 0.002 0.04 0.849 0.08 0.774

2 19.45 0.000 25.73 0.000 7.33 0.026 6.01 0.050

3 19.51 0.000 33.94 0.000 15.30 0.002 13.14 0.004

4 20.66 0.000 44.83 0.000 15.41 0.004 18.16 0.001

5 23.47 0.000 44.91 0.000 15.59 0.008 25.57 0.000

10 32.66 0.000 52.71 0.000 24.00 0.008 31.29 0.001

20 45.73 0.001 139.86 0.000 58.81 0.000 47.58 0.000

30 54.70 0.004 144.95 0.000 74.69 0.000 95.15 0.000

Notes: Statistic ρ(κ) is the autocorrelation coefficient at lag k , Q(k) is the heteroscedasticity-adusted Box-Pierce Q-test

statistic for autucorrelation of order K with asociated P -values.

Table 4-6

Estimated Autocorrelations for Corrected Indices Returns

No. Abudhabi Bahrain Dubai Egypt Jordan

Lags Q stat. P -value Q stat. P -value Q stat. P -value Q stat. P -value Q stat. P -value

1 2.18 0.140 0.92 0.338 0.15 0.696 0.36 0.550 0.50 0.478

2 2.89 0.236 3.16 0.206 2.42 0.298 9.83 0.007 16.74 0.000

3 14.64 0.002 12.46 0.006 2.46 0.482 10.18 0.017 16.76 0.001

4 19.94 0.001 17.44 0.002 2.69 0.611 11.41 0.022 19.73 0.001

5 21.30 0.001 20.95 0.001 3.08 0.688 11.65 0.040 19.91 0.001

10 28.14 0.002 43.93 0.000 5.62 0.846 16.51 0.086 24.55 0.006

20 53.07 0.000 52.04 0.000 8.19 0.991 24.96 0.203 53.42 0.000

30 68.99 0.000 59.26 0.001 12.73 0.998 39.34 0.118 68.62 0.000

Lags Q stat. P -value Q stat. P -value Q stat. P -value Q stat. P -value

1 0.02 0.894 174.03 0.000 22.72 0.000 83.80 0.000

2 3.23 0.199 174.22 0.000 22.75 0.000 106.66 0.000

3 3.23 0.358 186.96 0.000 31.80 0.000 112.22 0.000

4 3.72 0.446 199.18 0.000 32.80 0.000 113.33 0.000

5 5.36 0.374 199.48 0.000 33.11 0.000 116.10 0.000

10 12.08 0.280 208.41 0.000 42.70 0.000 121.97 0.000

20 26.41 0.153 263.55 0.000 86.51 0.000 138.93 0.000

30 31.51 0.390 272.84 0.000 109.38 0.000 202.41 0.000

Notes: Statistic ρ(κ) is the autocorrelation coefficient at lag k , Q(k) is the heteroscedasticity-adusted Box-Pierce Q-test

statistic for autucorrelation of order K with asociated P -values.

128

- Non-parametric runs test

The runs test determines whether successive price changes are independent.

Unlike its parametric equivalent, the serial correlation test of independence, the runs test

does not require returns to be normally distributed. Results of the runs test are shown in

table 4-7, both for the observed and corrected indices.

Panel A presents the results for the observed indices, the actual number of runs

(R) in each of the Arab stock markets can be seen to fall short of the expected number of

runs under the null hypothesis of stock returns independence. The resultant negative Z

values indicate positive serial correlation. The runs test results show that the successive

returns for all Arab stock markets are not independent at 5% level. However, when the

indices are corrected for infrequent trading, the results are strikingly different for Egypt,

Kuwait, and Saudi Arabia. Since expected and actual number of runs are so close. While

for other markets, stock returns dependency still significantly exist even after corrected

for thin trading. Based on the corrected indices, we cannot reject weak-form market

efficiency for Egypt, Kuwait, and Saudi stock markets. Correcting for infrequent trading

in the case of these three markets, leads to absolute reversal in the inference on market

efficiency.

The results for both Saudi and Kuwait equity markets are consistent with those of

Abraham et al. (2002), the same results also obtained by Bulter and Malaikah (1992)

when they use the observed indices for Saudi and Kuwaiti markets. Additionally, Haque

et al. (2004) find that the RWH can be rejected for both Egypt and Oman but not for

Bahrain, Jordan, and Saudi Arabia according to runs test results, when they use weekly

data of the observed indices. Moreover, Omran (2002) rejects the RWH for Egypt

depending on runs test, when he uses the observed index.

129

Table 4-7

Results of Runs Test for Arab Stock Markets, Observed vs. Corrected Indices

Abudhabi Dubai Bahrain Jordan Oman Palestine Egypt Kuwait Saudi

Panel A:Observed Index Returns

Observations( N ) 645 1098 3323 3121 1899 1180 1666 687 3082

n (+) 293 510 1577 1449 914 523 774 331 1539

n (-) 352 588 1746 1672 985 657 892 356 1543

n (0) 0 0 0 0 0 0 0 0 0

Expected runs ( m ) 321 547 1658 1554 949 583 830 344 1542

Actual runs ( R ) 268 469 1456 1200 694 467 662 279 1249

Standard error ( σm ) 12.582 16.477 28.744 27.786 21.753 16.945 20.3 13.08 27.753

Z - statistic -4.197a -4.748a -7.035a -12.724a -11.731a -6.868a -8.267a -4.974a -10.557a

Panel B: Corrected Index Returns

Observations( N ) 642 1095 3321 3118 1896 1177 1663 684 3079

n (+) 300 502 1619 1460 1002 537 799 322 1537

n (-) 342 593 1702 1658 894 640 864 362 1542

n (0) 0 0 0 0 0 0 0 0 0

Expected runs ( m ) 321 545 1660 1554 946 585 831 342 1540

Actual runs ( R ) 288 482 1432 1473 696 445 811 330 1502

Standard error ( σm ) 12.605 16.424 28.792 27.802 21.695 17.015 20.353 13.022 27.740

Z - statistic -2.588a -3.819a -7.935a -2.903a -11.520a -8.228a -0.994 -0.908 -1.388

a

Indicates rejection of the null that successive price changes are independentat the 5% level.

The runs test tests for a statistically significant difference between the expected number of runs vs.the actual number of runs. A run is defined as a sequence of successive price changes with the same

sign. n(+)/n(-)/n(0) represent the number of positive/negative/zero price changes. panel B shows the results for the index, corrected for infrequent trading.

130

- Variance ratio test

The RWH for each market is tested using the variance ratio test described in

section (4-1-5). The variance ratio is computed for multiples of 2, 4, and 8 days, with the

one-day return used as the base. Results for the observed and corrected indices are shown

in panel A and B of table 4-8, respectively. When the observed indices are used, the RWH

is strongly rejected for all markets except Kuwait and Palestine. The variance ratio is not

equal unity with the aggregation interval for the stock markets. While for both Kuwait

and Palestine, the variance ratio is the same and equal to unity for all multiples. However,

when the corrected indices are used, the RWH strongly rejected for all markets even for

Kuwait and Palestine.

The results for Saudi and Bahrain are different from those obtained by Abraham

et al. (2002) since they can not reject the RWH for Saudi and Kuwaiti equity markets,

when they implement the variance ratio test for the corrected indices. While the result of

the observed indices for Jordan and Egypt are consistent with those of Omran (2002). In

addition, the results obtained here for Bahrain, Oman, Kuwait, and Saudi Arabia,

contradict the results of Dahel and Laabas (1998) when they use weekly data of the

observed indices for these markets and can not reject the RWH using variance ratio test.

While Haque et al. (2004) find that Egypt, Bahrain, Oman, and Saudi Arabia show

predictability, whereas Jordan shows no signs of predictability when they use weekly

data of the observed indices.

- BDS test

The BDS tests the null hypothesis of whiteness (independent and identically

distributed observations) against an unspecified alternative. In other words, the BDS test

provides a direct statistical test for whiteness, against general dependence. Tables 4-9 and

4-10 report the BDS test statistics for each market both for observed and corrected

indices. The BDS test has been applied for embedding dimensions of m= 2, 3, 4 and 5.

For each m, ε is set to 0.5, 1.0, and 1.5 standard deviations (σ) of the data. The iid null

hypothesis is overwhelmingly rejected in all cases for the returns series and for all

markets even after the indices have been adjusted for thin trading. The results of the BDS

test indicate that the iid hypothesis is rejected in favor of an unspecified alternative or

131

general dependence, which may includes both non-white linear or non-linear dependence

in the time series.

Table 4-8

Variance Ratio Estimates and Heteroskedastic Test Statistics for

Arab Stock Markets

Panel A: variance ratio test for observed index returns

Market 2 4 8

Abudhabi 2.3618** 2.2894 1.7177

(2.7236) (1.9495) (0.9706)

Bahrain 5.1239** 0.8017 0.8604

(8.2478) (-0.2998) (-0.1887)

Dubai 1.3802 2.4779** 1.0259

(0.7603) (2.2345) (0.0351)

Egypt 5.8647** 3.0547** 0.7301

(9.7293) (3.1065) (-0.3645)

Jordan 3.1048** 1.1292 1.4556

(4.2095) (0.1953) (0.6161)

Kuwait 1.3259 1.1474 1.3255

(0.2652) (0.2229) (0.4402)

Oman 5.5315** 0.4729 3.8261**

(9.0629) (-0.7969) (3.8216)

Palestine 0.3694 0.6392 0.2336

(-1.2611) (-0.5455) (-1.0364)

Saudi 4.1643** 2.5845** 1.9269

(6.3285) (2.3956) (1.2535)

Panel B: variance ratio test for corrected index returns

Market 2 4 8

Abudhabi 4.7341** 1.1910 1.7788

(7.46827) (0.2888) (1.0531)

Bahrain 4.3877** 3.1357** 1.1398

(6.7754) (3.2289) (0.1890)

Dubai 5.7549** 1.8971 1.3779

(9.5099) (1.3563) (0.5110)

Egypt 1.5505 3.0018** 2.6571**

(1.1011) (3.0265) (2.2408)

Jordan 4.9824** 2.7599** 2.6877**

(7.9647) (2.6607) (2.2822)

Kuwait 2.9406** 1.3495 3.5323**

(3.8812) (0.5284) (3.4244)

Oman 3.4383** 3.2131** 2.0616**

(4.8767) (3.3459) (1.4356)

Palestine 4.2867** 2.1025 3.266**

(6.5733) (1.6668) (3.0642)

Saudi 3.4376** 0.5879 1.3840

(4.8752) (-0.623) (0.5193)

** Indicates rejection of the RWH at the 0.05 level.

Figures in parentheses are asymptotic Z statistic (H0:VR(q)=1).

2 2 2

The variance ratios are defined as the ratio of (1/q )σq to σ1 for values of q = 2, 4, and 8, where σi is

the variance of the index return defined as ln(p t /p t- i).Panel B shows the results for the index, corrected

for infrequent trading.

132

Table 4-9

BDS Test Results for Observed Return Indices

Abudhabi Bahrain Dubai Egypt Jordan

ε/σ m

BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob.

0.5 2 0.024** 0.000 0.029** 0.000 0.037** 0.000 0.017** 0.000 0.018** 0.000

3 0.025** 0.000 0.039** 0.000 0.056** 0.000 0.021** 0.000 0.018** 0.000

4 0.021** 0.000 0.039** 0.000 0.062** 0.000 0.016** 0.000 0.013** 0.000

5 0.015** 0.000 0.034** 0.000 0.058** 0.000 0.011** 0.000 0.008** 0.000

1.0 2 0.032** 0.000 0.032** 0.000 0.028** 0.000 0.027** 0.000 0.033** 0.000

3 0.055** 0.000 0.059** 0.000 0.057** 0.000 0.051** 0.000 0.051** 0.000

4 0.072** 0.000 0.08** 0.000 0.087** 0.000 0.063** 0.000 0.058** 0.000

5 0.076** 0.000 0.096** 0.000 0.110** 0.000 0.065** 0.000 0.056** 0.000

1.5 2 0.023** 0.000 0.019** 0.000 0.014** 0.000 0.021** 0.000 0.029** 0.000

3 0.048** 0.000 0.039** 0.000 0.030** 0.000 0.048** 0.000 0.054** 0.000

4 0.074** 0.000 0.059** 0.000 0.050** 0.000 0.071** 0.000 0.073** 0.000

5 0.094** 0.000 0.077** 0.000 0.069** 0.000 0.087** 0.000 0.087** 0.000

Observations 646 3325 1099 1667 3122

ε/σ m

BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob.

0.5 2 0.014** 0.000 0.046** 0.000 0.047** 0.000 0.026** 0.000

3 0.013** 0.000 0.058** 0.000 0.056** 0.000 0.032** 0.000

4 0.009** 0.000 0.051** 0.000 0.047** 0.000 0.028** 0.000

5 0.005** 0.000 0.040** 0.000 0.035** 0.000 0.021** 0.000

3 0.043** 0.000 0.099** 0.000 0.105** 0.000 0.068** 0.000

4 0.051** 0.000 0.126** 0.000 0.127** 0.000 0.090** 0.000

5 0.052** 0.000 0.141** 0.000 0.135** 0.000 0.100** 0.000

3 0.045** 0.000 0.080** 0.000 0.079** 0.000 0.054** 0.000

4 0.064** 0.000 0.115** 0.000 0.109** 0.000 0.084** 0.000

5 0.078** 0.000 0.145** 0.000 0.133** 0.000 0.109** 0.000

Observations 688 1900 1181 3083

Notes: The BDS (m,ε) tests for i.i.d., where m is the embedding dimention and ε is distance set in terms of the standard deviation of the

data(σ) to 0.5,1.0 and 1.5 standard deviations.** indicates statistical significance at the 5% level.

133

Table 4-10

BDS Test Results for Adjusted Return Indices

Abudhabi Bahrain Dubai Egypt Jordan

ε/σ m

BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob.

0.5 2 0.019** 0.000 0.037** 0.000 0.036** 0.000 0.014** 0.000 0.015** 0.000

3 0.020** 0.000 0.047** 0.000 0.054** 0.000 0.018** 0.000 0.017** 0.000

4 0.018** 0.000 0.045** 0.000 0.059** 0.000 0.014** 0.000 0.014** 0.000

5 0.013** 0.000 0.038** 0.000 0.055** 0.000 0.010** 0.000 0.009** 0.000

1.0 2 0.026** 0.000 0.039** 0.000 0.027** 0.000 0.021** 0.000 0.029** 0.000

3 0.047** 0.000 0.070** 0.000 0.056** 0.000 0.047** 0.000 0.051** 0.000

4 0.064** 0.000 0.091** 0.000 0.086** 0.000 0.059** 0.000 0.065** 0.000

5 0.068** 0.000 0.107** 0.000 0.109** 0.000 0.060** 0.000 0.071** 0.000

1.5 2 0.022** 0.000 0.025** 0.000 0.014** 0.000 0.015** 0.000 0.026** 0.000

3 0.047** 0.000 0.049** 0.000 0.029** 0.000 0.044** 0.000 0.054** 0.000

4 0.072** 0.000 0.072** 0.000 0.049** 0.000 0.068** 0.000 0.079** 0.000

5 0.092** 0.000 0.092** 0.000 0.068** 0.000 0.083** 0.000 0.100** 0.000

Observations 646 3325 1099 1667 3122

ε/σ m

BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob. BDS Stat. Prob.

0.5 2 0.012** 0.000 0.037** 0.000 0.045** 0.000 0.023** 0.000

3 0.012** 0.000 0.045** 0.000 0.052** 0.000 0.024** 0.000

4 0.009** 0.000 0.038** 0.000 0.043** 0.000 0.017** 0.000

5 0.005** 0.000 0.029** 0.000 0.032** 0.000 0.010** 0.000

3 0.040** 0.000 0.088** 0.000 0.099** 0.000 0.064** 0.000

4 0.049** 0.000 0.109** 0.000 0.121** 0.000 0.085** 0.000

5 0.051** 0.000 0.120** 0.000 0.128** 0.000 0.097** 0.000

3 0.042** 0.000 0.070** 0.000 0.076** 0.000 0.064** 0.000

4 0.060** 0.000 0.100** 0.000 0.106** 0.000 0.085** 0.000

5 0.075** 0.000 0.126** 0.000 0.129** 0.000 0.097** 0.000

Observations 688 1900 1181 3083

Notes: The BDS (m,ε) tests for i.i.d., where m is the embedding dimention and ε is distance set in terms of the standard deviation of the

data(σ) to 0.5,1.0 and 1.5 standard deviations.** indicates statistical significance at the 5% level.

134

4-5-2 Volatility of returns

Several methods have been suggested in the literature to test for volatility in stock

markets. Regardless of the debate over empirical testing of volatility, the fact remains

that volatility is a relative measure. The purpose of this section is to investigate the

volatility of Arab stock markets and whether these markets are characterized by excessive

volatility of returns, relative to other developed and emerging markets. To this end, in

addition to nine Arab stock markets, three emerging and three developed markets will be

tested. U.K (FTSE100), USA (S&P 500), and Japan (Nikkei 225) will be used for

developed markets, while Israel, Turkey, and India general equity markets’ indices will

be the representative of emerging markets. The data for both emerging and developed

markets consists of daily prices and obtained from Yahoo Finance.3

- Coefficient of variation

The coefficient of variation figures presented in table 4-11 measure the degree of

volatility of daily market returns relatives. For the group of Arab markets, Oman appears

to be, by far, the most volatile followed by Egypt, with Kuwait the least volatile. For the

developed markets, the coefficients of variation are higher in average than those for most

of the Arab markets, as well as for the developing markets except India. Overall based on

the coefficient of variation, the figures seem to indicate that Arab stock markets as a

group characterized with a low level of volatility relative to the other two groups.

Table 4-11

Coefficient of Variation (C.V) for Daily Returns for the Three Groups

Arab stock markets Emerging markets Developed markets

market C.V % market C.V % market C.V %

Abudhabi 6.4 3.0% India 73.9 59.5% Japan 44.3 30.7%

Bahrain 28.3 13.2% Israel 28.2 22.7% UK 65.5 45.4%

Dubai 23.2 10.8% Turkey 22.1 17.8% USA 34.5 23.9%

Egypt 39.2 18.3% Total 124.2 100.0% Total 144.3 100.0%

Jordan 20.7 9.7% Average 41.4 Average 48.1

Kuwait 5.5 2.6%

Oman 43.2 20.2%

Palestine 22.4 10.5%

Saudi 25.2 11.8%

Total 214.1 100.0%

Average 23.79

3

Although all Arab markets are emerging markets, the distinction between Arab and emerging markets is

made only for the purpose of the analysis.

135

-Schwert measure

In the case of the coefficient of variation, volatility in Arab, emerging, and

developed markets has been investigated at the market level. The figures in table 4-11 do

not provide a clear assessment of the degree of volatility of returns in Arab markets as a

group, compared to that in other two groups of markets. The Schwert measure of

volatility used at the group level should reveal not only the potential trends in volatility of

returns in Arab markets, but also their level of volatility relative to that of emerging and

developed markets.

Figure 4-1 shows the Schwert measure of volatility for the 3 groups. The main

observation that could be made from the figure is that, Arab stock markets as a group

exhibit the lowest level of volatility while emerging markets found to be the highest.

However, Arab markets show a remarkable increase of volatility, particularly over the

period 2003 corresponding to the invasion on Iraq.4

Figure 4-1

Markets Volatility (Schwert Model)

4

The end of the year 2004 has been used to calculate the weights.

136

- GARCH (1,1) models

The volatility and persistence of shocks to volatility for all markets, including

emerging and developed markets, are presented in table 4-12. Based on empirical

investigation conducted by GARCH (1,1) for daily data, one finds that only Dubai does

not show any volatility clustering. In other words, there is evidence of significantly

volatility clustering in 8 out of 9 Arab markets. ARCH parameter (α1 in table 4-12) is less

than unity for all fifteen markets, signifying that shocks are not explosive. Economic

interpretations of the ARCH effect in stock returns have been provided within both micro

and macro frameworks, according to Bollerslev et al. (1992) and other studies. The

ARCH effect in stock returns could be due to clustering of trade volumes, nominal

interest rate, dividends yields, money supply, oil prices etc.

Moreover, the persistence of shocks is measured by (α1+β1) in the GARCH

model. According to Engel and Bollerslev (1986), if α1+β1 = 1 in GARCH model, a

current shock persists indefinitely in conditioning the future variance. Since the sum

α1+β1 represents the change in response function of shocks to volatility persistence, a

value greater than unity implies that response function of volatility increases with time

and a value less than unity implies that shocks decay with time (Chou, 1988). The closer

to unity is the value of persistence measure; the slower is the decay rate. The findings

here reveal that in three Arab stock markets (Egypt, Kuwait, and Palestine), one fails to

reject that α1+β1 = 1, i.e. shocks to volatility are permanent. It implies that the conditional

variance is non-stationary. The volatility movements affects the stock markets of these

three countries, volatility is persistent and has a slow rate of decay. On the other hand,

Oman exhibits an increasing response function of volatility and shocks do not decay with

time, while the response function of volatility for the other 11 markets, decays with time.

Moreover, several diagnostic tools have been implemented. Table 2 in appendix 2

presents the results of five diagnostic tests for the standardized residuals of GARCH (1,1)

on a daily basis. The McLeod-Li test indicates that the squared residuals are not

correlated; while the BDS test reject the iid hypothesis of the standardized residuals for

three Arab markets only (Bahrain, Dubai, and Egypt).

137

Table 4-12

GARCH (1,1) Model for Daily Returns

Market Obs. α0 α1 β1 α 1 +β 1 ku Sk

Q(30) ARCH LM test

AbuDhabi 644 0.00000 0.19183 0.68194 0.87377 9.406 0.125

27.303 0.005

0.000 0.000 0.000 0.000 0.607 0.946

Jordan 3122 0.00000 0.21977 0.71952 0.93930 5.154 0.321 34.743 0.354

0.000 0.000 0.000 0.000 0.252 0.552

Bahrain 3323 0.00005 0.10708 -0.04331 0.06378 384.310 10.701 0.934 0.733

0.000 0.000 0.155 0.000 1.000 0.392

Dubai 1097 0.00000 0.00017 0.97739 0.97756 201.951 7.874 0.795 0.010

0.000 0.779 0.000 0.000 1.000 0.921

Egypt 1665 0.00001 0.13525 0.86008 0.99533 9.107 0.347 57.300 2.583

0.000 0.000 0.000 0.408 0.002 0.108

Kuwait 686 0.00000 0.19027 0.79472 0.98499 5.511 -0.345 16.147 0.000

0.000 0.000 0.000 0.308 0.981 0.997

Oman 1898 0.00000 0.29685 0.76219 1.05904 24.645 1.226 15.678 1.021

0.000 0.000 0.000 0.000 0.985 0.312

Palestine 1179 0.00002 0.44315 0.55360 0.99674 8.153 -0.138 33.593 0.021

0.000 0.000 0.000 0.904 0.297 0.885

Saudi 3081 0.00000 0.35268 0.61969 0.97237 10.094 -0.136 26.039 0.190

0.000 0.000 0.000 0.001 0.673 0.663

USA 3780 0.00000 0.06652 0.92710 0.99361 4.833 -0.351 26.621 3.881

0.000 0.000 0.000 0.005 0.643 0.144

UK 3789 0.00000 0.08126 0.90501 0.98626 3.791 -0.112 26.478 0.599

0.000 0.000 0.000 0.002 0.651 0.439

Japan 3692 0.00001 0.09297 0.87906 0.97202 4.652 -0.022 20.090 2.853

0.000 0.000 0.000 0.000 0.914 0.091

Turkey 1835 0.00001 0.09244 0.90088 0.99331 4.524 -0.173 28.834 1.924

0.000 0.000 0.000 0.000 0.526 0.165

India 1891 0.00001 0.13523 0.82516 0.96038 5.638 0.013 24.788 0.263

0.000 0.000 0.000 0.000 0.735 0.608

Israel 1500 0.00003 0.10942 0.77420 0.88362 5.210 -0.215 16.939 3.089

0.000 0.000 0.000 0.000 0.973 0.213

2

Significance levels are in italics. A Chi-square (χ ) tests (α 1 +β 1 ) = 1. Ku is the kurtosis of the residuals. Sk is the skewness of the residuals

and Q(30) is the Ljung-Box statistics serial correlation of the lags in the residuals. ARCH LM is a Lagrange multipler (LM) test for

autoregressive conditional heteroskedasticity (ARCH) in the residuals. The estimated variance equation is :

2

h t = a 0 + a 1ε t −1 + β 1 h t −1

The results change dramatically when we use weekly data, table 4-13 shows the

results of GARCH (1,1) for weekly returns. Volatility clustering disappears in 4 out of 9

Arab markets (Abu Dhabi, Dubai, Egypt, and Kuwait) in addition to India and Israel5,

while volatility found to be persistent in 11 markets at the 1 percent level. Within the

group of Arab stock markets, the results reveal that 5 of the 9 markets (Abu Dhabi,

Dubai, Egypt, Kuwait, and Oman) exhibit persistence of volatility. The results for Jordan,

Saudi, and Bahrain presented in table 4-13, are consistent with those obtained from

5

For Egypt, Kuwait, and India, volatility clustering disappears at 1 % level only.

138

Haque et al. (2004), since they find that these markets exhibit volatility clustering but not

Oman. In addition, they don not find volatility to be permanent in both Saudi and Jordan,

which is inconsistent with the results obtained here for these two markets.

Table 4-13

GARCH (1,1) Model for Weekly Returns

Country Obs. α0 α1 β1 α 1 +β 1 ku Sk Q(30) ARCH LM test

AbuDhabi 122 0.00007 0.02948 0.62006 0.64954 4.633 0.191 27.450 0.00198

0.716 0.748 0.529 0.712 0.600 0.964

Jordan 676 0.00004 0.15037 0.74845 0.89882 4.492 0.687 15.661 0.20044

0.003 0.000 0.000 0.007 0.985 0.654

Bahrain 699 0.00003 0.21698 0.64099 0.85797 6.266 -0.010 29.454 0.03697

0.000 0.000 0.000 0.000 0.494 0.848

Dubai 192 0.00023 0.06293 0.56224 0.62517 34.178 3.345 4.049 0.08971

0.188 0.210 0.062 0.189 1.000 0.765

Egypt 364 0.00012 0.06365 0.86453 0.92818 4.984 0.522 25.056 0.61368

0.036 0.013 0.000 0.045 0.722 0.433

Kuwait 140 0.00016 0.42045 0.35650 0.77695 3.861 -0.261 21.313 0.41905

0.038 0.023 0.129 0.142 0.878 0.517

Oman 400 0.00009 0.38872 0.53808 0.92680 7.324 0.626 15.417 0.07061

0.000 0.000 0.000 0.148 0.987 0.790

Palestine 388 0.00019 0.18439 0.60576 0.79014 5.729 0.457 22.052 0.64497

0.002 0.002 0.000 0.003 0.852 0.422

Saudi 975 0.00005 0.28539 0.61480 0.90020 7.298 -0.206 12.179 0.01132

0.000 0.000 0.000 0.000 0.998 0.915

USA 780 0.00000 0.08887 0.90487 0.99374 3.696 -0.336 29.912 4.43269

0.049 0.000 0.000 0.363 0.470 0.109

UK 785 0.00001 0.08452 0.89759 0.98211 3.983 -0.131 31.340 0.02157

0.082 0.000 0.000 0.155 0.399 0.883

Japan 779 0.00013 0.10219 0.75862 0.86081 3.897 -0.192 15.825 0.03263

0.012 0.000 0.000 0.018 0.984 0.857

Turkey 378 0.00013 0.10059 0.87942 0.98001 5.187 0.058 12.849 0.01552

0.049 0.000 0.000 0.264 0.997 0.901

India 398 0.00001 0.03884 0.95364 0.99249 4.044 -0.265 33.944 0.00148

0.433 0.014 0.000 0.343 0.283 0.969

Israel 396 0.00063 0.10671 0.23423 0.34093 5.550 -0.580 17.017 0.00775

0.014 0.056 0.056 0.016 0.972 0.930

Significance levels are in italics. A Chi-square (χ2) tests (α 1 +β 1 ) = 1. Ku is the kurtosis of the residuals. Sk is the skewness of the

residuals and Q(30) is the Ljung-Box statistics serial correlation of the lags in the residuals. ARCH LM is a Lagrange multipler (LM)

test for autoregressive conditional heteroskedasticity (ARCH) in the residuals. The estimated variance equation is :

2

h t = a 0 + a 1ε t −1 + β 1 h t −1

- EGARCH model

For equities, it is often observed that downward movements in the market are

followed by higher volatilities than upward movements of the same magnitude. To

account for this phenomenon, Nelson (1991) introduced the EGARCH model which

139

allows for asymmetric shocks to volatility and test for the leverage effect. Table 4-14

shows the results of EGARCH for daily returns. The presence of leverage effect can be

tested by the hypothesis that γ>0 in table 4-14 and the impact is asymmetric if γ≠0. The

results indicate that, in addition to the two groups (developed and emerging markets), 4

out of 9 Arab stock markets (Bahrain, Dubai, Kuwait, and Oman) exhibit leverage effect

and asymmetric shocks to volatility.

Table 4-14

EGARCH (1,1) Model for Daily Returns

Market Obs. ω α γ β ku Sk Q(30) ARCH LM test

AbuDhabi 644 -1.78165 0.34467 0.03506 0.85392 9.422 0.144 23.061 0.000

0.000 0.000 0.140 0.000 0.813 0.998

Jordan 3120 -1.23553 0.40366 0.02818 0.90722 5.170 0.304 39.202 0.511

0.000 0.000 0.007 0.000 0.121 0.475

Bahrain 3323 -12.65412 0.12159 -0.04447 -0.23169 362.367 9.934 27.014 26.161

0.000 0.000 0.000 0.006 0.623 0.000

Dubai 1097 -10.01361 0.74962 -0.50514 0.00253 141.089 4.862 0.466 0.014

0.000 0.000 0.000 0.892 1.000 0.905

Egypt 1665 -0.51351 0.24446 0.07710 0.89177 8.714 0.224 93.901 3.337

0.000 0.000 0.000 0.000 0.000 0.068

Kuwait 686 -1.27042 0.33540 -0.19960 0.89177 5.483 -0.310 19.498 8.533

0.000 0.000 0.000 0.000 0.929 0.577

Oman 1898 -0.89754 0.45511 -0.04685 0.94151 23.798 0.703 84.409 0.331

0.000 0.000 0.001 0.000 0.000 0.565

Palestine 1179 -1.44426 0.52062 0.07844 0.87744 8.998 -0.295 20.842 0.991

0.000 0.000 0.000 0.000 0.893 0.320

Saudi 3081 -1.88841 0.34616 0.01661 0.84125 11.776 -0.404 82.528 8.149

0.000 0.000 0.014 0.000 0.000 0.017

USA 3780 -0.24749 0.11417 -0.08903 0.98305 4.509 -0.294 34.760 1.143

0.000 0.000 0.000 0.000 0.251 0.285

UK 3789 -0.20915 0.11031 -0.06385 0.98696 3.843 -0.101 29.058 0.124

0.000 0.000 0.000 0.000 0.515 0.724

Japan 3692 -0.38034 0.15409 -0.09008 0.96930 4.515 0.022 26.524 2.547

0.000 0.000 0.000 0.000 0.648 0.110

Turkey 1835 -0.42437 0.27374 -0.05148 0.96941 4.356 -0.156 28.832 0.956

0.000 0.000 0.000 0.000 0.526 0.328

India 1891 -0.87231 0.25576 -0.12557 0.91961 4.964 0.023 27.438 0.036

0.000 0.000 0.000 0.000 0.600 0.850

Israel 1500 -1.49498 0.23604 -0.15123 0.84442 4.663 -0.123 19.965 0.166

0.000 0.000 0.000 0.000 0.917 0.684

Significance levels are in italics. A Chi-square (χ2) tests (α 1 +β 1 ) = 1. Ku is the kurtosis of the residuals. Sk is the skewness of the

residuals and Q(30) is the Ljung-Box statistics serial correlation of the lags in the residuals. ARCH LM is a Lagrange .EGARCH

equation estimated as: ε ε t −1

( ) (

log σ t2 = ω + β log σ t2−1 + a t −1 + γ) σ t −1 σ t −1

140

The coefficient of the leverage term was negative and significant at the 5% level;

conditional variance is higher in the presence of negative innovations, which indicate that

the market becomes more nervous when negative shocks take place. Usually what

happens is that, small investors get panic from these negative shocks and sell their stocks

in order to avoid higher losses. Moreover, to see this effect clearly, the News Impact

Curve has been plotted for each market under examination (see figure 1 appendix 2).

Efficiency implicitly assumes that investors are rational, where rationality implies

risk aversion, unbiased forecasts and instantaneous responses to information. Such

rationality leads to price responding linearly to new information. In this contest, failure to

take into account the institutional features of emerging markets may lead to statistical

illusions regarding efficiency or inefficiency. With reference to evidence in favor of

efficiency, this is perhaps the outcome of using linear models for testing efficiency, in

markets characterized by inherent non-linearity. If the return generating process is non-

linear and a linear model is used to test for efficiency, then the hypothesis of no

predictability may be wrongly accepted, this is because non-linear system such as

“chaotic” ones looks very similar to a random walk.

Table 4-15 shows the results of the random walk model including non-linear

terms for the observed indices. It appears that with the introduction of non-linear

components, α1, α2, and α3 are statistically significant for all markets except Kuwait. This

seems to indicate predictability and inefficiency. The results reveal that the return

generating process in 8 out of 9 Arab markets is non-linear even after corrected the

indices for thin trading (see table 4-16). Moreover, most of the diagnostic tests; especially

BDS test, for the residuals indicate that the residual are not following white noise process

(see tables 6 and 7 in appendix 2).

However, the logistic map is not able to determine the precise nature of any non-

linearity, but rather to ascertain whether non-linearity exists. It is appropriate that non-

linearity generated by dependence in the second moment. To disentangle the nonlinearity

generated by changes in volatility from non-linearity arising as a result of other causes,

the standardized residuals resulted from GARCH models will be subjected to several

141

diagnostic tests to determine whether the specified GARCH model is sufficient to model

the nonlinear structure in the data against an unspecified alternative.

Table 4-15

Random Walk Models with Non-linearity for Observed

Indices

Rt = a0 + a1Rt-1 + a2R2t-1+ a3R3t-1+ εt

Market coefficients std.error t-value P -value

a0 0.000 0.000 1.788 0.074

a1 0.196 0.056 3.521 0.001

Abudhabi

a2 10.360 2.890 3.585 0.000

a3 -339.479 178.188 -1.905 0.057

a0 0.000 0.000 2.927 0.003

a1 0.205 0.021 9.814 0.000

Bahrain

a2 -2.137 0.109 -19.578 0.000

a3 -16.934 0.748 -22.646 0.000

a0 0.000 0.000 0.674 0.501

a1 0.115 0.046 2.521 0.012

Dubai

a2 3.148 0.814 3.867 0.000

a3 -17.391 4.343 -4.004 0.000

a0 0.000 0.000 0.510 0.610

a1 0.309 0.026 11.726 0.000

Egypt

a2 0.656 0.520 1.262 0.207

a3 -26.228 3.652 -7.182 0.000

a0 0.000 0.000 0.702 0.483

a1 0.295 0.023 13.017 0.000

Jordan

a2 3.440 0.928 3.708 0.000

a3 -177.321 35.219 -5.035 0.000

a0 0.002 0.000 3.555 0.000

a1 0.173 0.055 3.164 0.002

Kuwait

a2 -0.004 1.672 -0.003 0.998

a3 -10.638 56.159 -0.189 0.850

a0 0.000 0.000 -0.397 0.691

a1 0.308 0.026 11.736 0.000

Oman

a2 2.666 0.282 9.454 0.000

a3 -43.597 2.616 -16.663 0.000

a0 0.000 0.000 -0.059 0.953

a1 0.294 0.032 9.232 0.000

Palestine

a2 1.970 0.165 11.956 0.000

a3 -13.406 0.782 -17.136 0.000

a0 0.000 0.000 0.443 0.658

a1 0.143 0.020 7.205 0.000

Saudi

a2 2.907 0.186 15.635 0.000

a3 -22.097 1.296 -17.055 0.000

142

Table 4-16

Random Walk Models with Non-linearity for Corrected

Indices

Rtadj = a0 + a1Radjt-1 + a2R2adjt-1+ a3R3adjt-1+ εt

Market coefficient std.error t-value P -value

a0 -0.0004 0.0003 -1.613 0.107

a1 0.1408 0.0568 2.478 0.014

Abudhabi

a2 7.6016 2.4496 3.103 0.002

a3 -282.8010 137.9073 -2.051 0.041

a0 0.0001 0.0001 0.578 0.563

a1 0.2221 0.0206 10.784 0.000

Bahrain

a2 -2.8772 0.1565 -18.387 0.000

a3 -15.6601 1.1872 -13.191 0.000

a0 0.0004 0.0003 1.372 0.171

a1 0.0918 0.0457 2.008 0.045

Dubai

a2 3.1324 0.8146 3.845 0.000

a3 -16.9249 4.2789 -3.955 0.000

a0 0.0014 0.0005 2.696 0.007

a1 0.0847 0.0274 3.088 0.002

Egypt

a2 -1.0921 0.3122 -3.498 0.001

a3 -14.3869 2.0452 -7.034 0.000

a0 -0.0002 0.0002 -1.214 0.225

a1 0.0456 0.0219 2.084 0.037

Jordan

a2 1.9981 0.7423 2.692 0.007

a3 -53.1736 17.4916 -3.040 0.002

a0 0.0004 0.0004 1.004 0.316

a1 0.0343 0.0537 0.639 0.523

Kuwait

a2 -0.2291 1.6599 -0.138 0.890

a3 -56.2277 54.0341 -1.041 0.298

a0 -0.0004 0.0002 -1.820 0.069

a1 0.4879 0.0247 19.787 0.000

Oman

a2 0.5756 0.3791 1.518 0.129

a3 -46.0179 3.5874 -12.827 0.000

a0 -0.0015 0.0005 -3.115 0.002

a1 0.3323 0.0317 10.485 0.000

Palestine

a2 1.7837 0.2600 6.861 0.000

a3 -8.3673 1.2096 -6.917 0.000

a0 0.0002 0.0002 1.353 0.176

a1 -0.0432 0.0200 -2.163 0.031

Saudi

a2 3.2987 0.1905 17.312 0.000

a3 -17.4300 1.3537 -12.876 0.000

Since correct specification of the model implies that the residuals will be

uncorrelated and also will have a zero mean and unit variance. Five different tests are

considered in this exercise for testing the hypothesis that the residuals are iid. This will

143

allow us on one hand to obtain a deeper and more detailed insight into the series

properties, by generating useful information from the various tests and on the other hand,

to minimize the probability of missing some thing and thus drawing the wrong

conclusion. If our battery of tests displays a unanimous “consensus” in favor of a specific

result, we would interpret this “consensus” as strong corroboration of that output.

The five tests that are going to be used are the following: McLeod and Li (1983);

Engle (1982), Brock et al. BDS (1996); serial correlation and Jarque-Bera test for

normality. All these tests share the principle that once any linear structure is removed

from the data, any remaining structure should be due to a non-linear data generating

mechanism. The McLeod and Li test looks at the autocorrelation function of the squares

of the prewhitened data and test whether corr (e2t,e2t-k) is non-zero for some k and can be

considered as an LM statistic against ARCH effect (see Granger and Terasvirta 1993;

Patterson and Ashley 2000). The test suggested by Engle (1982) is an LM test, which

should have considerable power against GARCH alternatives. The BDS test is a non

parametric test for serial independence as described in section (4-1-6).

The corresponding diagnostic tests for random walk and GARCH models for each

daily observed index are presented in appendix 2. Table 1 in appendix 2 shows the

diagnostic tests for the ordinary residuals of the RW model, clear evidence emerges

across the spectrum of tests that the residuals of the RW are not iid. Almost all P-values

are 0, suggesting that some kind of hidden structure exists in the data. While for Dubai

equity market, only the BDS test rejects the iid hypothesis. The failure of the RW model

to explain the behavior of the series and considerations of the constant term that are

statistically different from zero, for some models, casts doubts on the validity of the

weak-form efficiency. The evidence against EMH is clear in all of the indices. Table 2 in

appendix 2 presents the results of diagnostic tests for the standardized residuals obtained

from GARCH models. Evidence emerges to support the hypothesis that the standardized

residuals are iid for all markets except Bahrain, Dubai, and Egypt, since the BDS test for

these three markets, reject the iid hypothesis.

Table 3 in appendix 2 presents the results for the standardized residuals obtained

from EGARCH model for observed daily indices. The results indicate that, we are not

able to reject the iid hypothesis for all markets. Most of the P-values presented in table 3,

144

are exceed the 5% benchmark. Moreover, the GARCH models produced lower SC’s

(Schwartz criterion) and as a result, are preferred to the RW in this respect. While for

Bahrain, Abu Dhabi, and Saudi, the EGARCH models produced the lowest SC’s values,

compared to the lowest SC’s values which have been produced by GARCH models for

the other markets. Additionally, evidence emerges to support the hypothesis that the

standardized residuals of the GARCH models are iid. Most of the P-values exceed the

5% level. Therefore, we could accept the randomness hypothesis.

As mentioned previously, the existence of seasonality or calendar effects in stock

markets, contradicts the efficient market hypothesis, at least in its weak form, because the

predictable movements in asset prices provide investors with opportunities to generate

abnormal returns. In emerging markets, it is possible that the dissemination of

information is restricted due to the possible manipulation of financial information by

market participants, and a lack of strict disclosure requirements imposed by the stock

market regulatory agencies. In this section, we will examine three calendar effects, day-

of-the-week, month-of-the-year, and the Halloween effects for each of the Arab stock

markets under examination.

- Day-of-the-week effect

Table 4-17 shows the OLS results for the day-of-the-week effect. Because Arab

stock markets have different trading days during the week, the OLS equation will have

five dummy variables for markets which have five trading days, and six dummy variables

for those with six trading days. Moreover, our purpose here is to test for the effect of the

first trading day of the week for each market, which will be Saturday or Sunday, since

these days are trading days in Arab stock markets (weekend is Friday) and it is equivalent

to test for Monday effect in other international stock markets.6

The results in table 4-17 indicate that, the estimated coefficients for the first

trading day of the week are positive and statistically significant (at 5% level) in three

markets Abu Dhabi, Jordan, and Saudi Arabia. While these coefficients found to be

6

Palestine was not included in this test, since trading days are not consistent during the week, resulting

irregularity in trading days.

145

positive but not significant in the rest of the markets. Moreover, the coefficients for other

days found to be positive and significant for most of the markets, another important note

is that for Bahrain and Saudi, the next trading day found to be negative and statistically

significant (Monday and Sunday effects). It seems that in these two markets, bad news is

announced at week ends, in order for the shock to be more easily absorbed, but that the

information is not instantly reflected in prices, investors in these markets hesitant and act

with a delay of one day. Furthermore, the estimated coefficients for all trading days in

both Jordan and Abu Dhabi found to be positive and statistically significant (at 5% level).

In general, the results are inconsistent with the results reported in the finance literature

for a large number of countries, where significantly lower or negative Monday returns are

reported (the traditional Monday effect).

In addition, the results reject the trading time hypothesis (since the returns on the

first day is different from that of the other days of the week), as well as the calendar time

hypothesis (since the returns on the first trading day is not three times that of the other

days). Note also that, the other coefficients in several markets are significantly positive.

The results for Egypt are inconsistent with those obtained by Aly et al. (2004), since they

find that all estimated coefficients to be positive but statistically not significant. However,

a Chow test indicates that the estimated coefficients reported in table 4-17, are

structurally stable over the entire sample period for three markets only (Abu Dhabi,

Bahrain, and Kuwait), while the hypothesis that all parameters equal zero has been

soundly rejected for all markets (see table 4-18).

In order to further investigate the presence of a positive first trading day

seasonality in Arab stock markets, one can test this formally by examining whether there

exists a simultaneous day-of-the-week effect in mean returns and in the standard

deviation of these returns. A formal test of the existence of day-of-the-week calendar

effect is given by the ANOVA or Kruskal-Wallis statistics. While a formal test for daily

variation in the second moment is given by the Levene test. Table 4-19 shows the results

of the day-of-the-week effect in the first two moments. As can be seen in table

146

Table 4-17

OLS Results for Day-of-the-Week Effect

Market Variable Coefficient Std. Error t-Statistic Prob.

AbuDhabi

1/7/2001-31/12/2003 Saturday 0.0002 0.0001 2.240 0.026

Sunday 0.992 0.042 23.717 0.000

Monday 0.998 0.029 34.730 0.000

Tusday 0.994 0.039 25.322 0.000

Wedensday 0.995 0.029 33.864 0.000

Jordan

4/1/1992-14/3/2005 Sunday 0.282 0.037 7.635 0.000

Monday 0.117 0.036 3.267 0.001

Tusday 0.274 0.041 6.751 0.000

Wedensday 0.404 0.041 9.755 0.000

Thursday 0.335 0.065 5.165 0.000

Bahrain

2/1/1991-3/6/2004 Sunday 0.348 0.301 1.154 0.248

Monday -0.996 0.017 -57.032 0.000

Tusday 0.001 0.017 0.041 0.967

Wedensday 0.192 0.269 0.716 0.474

Thursday 0.191 0.307 0.621 0.535

Dubai

26/3/2000-31/12/2003 Saturday 0.286 0.146 1.964 0.050

Sunday 0.242 0.105 2.300 0.022

Monday 0.031 0.041 0.758 0.448

Tusday -0.170 0.076 -2.251 0.025

Wedensday -0.183 0.102 -1.800 0.072

Thursday 0.034 0.097 0.348 0.728

Egypt

1/1/1998-31/12/2004 Sunday 0.031 0.043 0.725 0.468

Monday 0.190 0.059 3.225 0.001

Tusday 0.174 0.056 3.105 0.002

Wedensday 0.254 0.053 4.767 0.000

Kuwait

17/6/2001-30/11/2005 Saturday 0.084 0.057 1.490 0.137

Sunday 0.211 0.071 2.976 0.003

Monday 0.354 0.075 4.692 0.000

Tusday 0.028 0.070 0.395 0.693

Wedensday 0.526 0.073 7.196 0.000

Oman

25/6/2000-13/10/2004 Sunday 0.000 0.000 1.112 0.266

Monday 1.000 0.029 34.647 0.000

Tusday 1.000 0.023 42.689 0.000

Wedensday 0.998 0.027 36.383 0.000

Thursday 0.997 0.032 31.555 0.000

Saudi Arabia

26/1/1994-14/3/2005 Saturday 0.170 0.038 4.489 0.000

Sunday -0.389 0.036 -10.888 0.000

Monday -0.010 0.035 -0.295 0.768

Tusday 0.017 0.046 0.364 0.716

Wedensday 0.107 0.046 2.337 0.020

Thursday 0.279 0.070 3.984 0.000

The estimated equation is: Rt = β 1 D1 + β 2 D2 + ... + β 6 D6 + ε.where

t

Rt is the daily return

which defined as ln(pt/pt-1);D1 through D6 are dummy variables such that if t is a Saturday, then D1=1 and D1=0 for all other days, if t

is a Sunday D2=1and D2=0 for all other days, and so forth; εt is a random term and β1-β6 are coefficients to be estimated using ordinary

least squares according to the trading days for each market.

147

Table 4-18

Chow Test for Structural Stability

Market AbuDhabi Jordan Bahrain Dubai Egypt Kuwait Oman Saudi

Break point 1-Jul-02 27-Jul-98 16-Sep-97 2-Mar-02 2-Jul-01 4-Nov-03 26-Aug-02 1-Sep-99

F -test 0.505 19.187 1.463 3.410 4.885 0.837 2.681 20.979

P -value 0.773 0.000 0.199 0.002 0.000 0.523 0.020 0.000

Chow test is implemented to test that the estimated coefficients are structurally stable over the entire sample period.

Market AbuDhabi Jordan Bahrain Dubai Egypt Kuwait Oman Saudi

Chi-square 3518.986 236.374 3254.850 18.149 117.627 85.037 5327.819 160.265

Probability 0.000 0.000 0.000 0.006 0.000 0.000 0.000 0.000

Wald test tests the null hypothesis that β1 = β 2 = ... = β 6 = 0

Table 4-19

Day-of-the-Week Effect in the First Two Moments

Market AbuDhabi Jordan Bahrain

First trading day of the week Saturday Sunday Sunday

Mean Stand. Dev. Mean Stand. Dev. Mean Stand. Dev.

Returns on 1st trading day 0.0002 0.0019 0.0001 0.0035 0.0000 0.0023

Returns during Rest of the Week 0.0008 0.0053 0.0005 0.0076 0.0003 0.0568

Difference of Means Test 5.919** 6.147** 0.056

difference of Variance Test 207.518* 1263.56* 17.3*

First trading day of the week Saturday Sunday Saturday

Mean Stand. Dev. Mean Stand. Dev. Mean Stand. Dev.

Returns on 1st trading day 0.00000 0.0021 0.0001 0.0090 0.0005 0.0052

Returns during Rest of the Week 0.0004 0.0101 0.0005 0.0165 0.0020 0.0095

Difference of Means Test 1.920 1.119 18.06*

difference of Variance Test 199.78* 625.6* 306.91*

First trading day of the week Sunday Saturday

Mean Stand. Dev. Mean Stand. Dev.

Returns on 1st trading day 0.0001 0.0024 0.0001 0.0042

Returns during Rest of the Week 0.0004 0.0058 0.0005 0.0094

Difference of Means Test 3.619*** 4.452**

difference of Variance Test 397.969* 789.292*

*,**,*** indicate significant at 1%, 5%, 10% levels, respectively.

Difference of means test of the null hypothesis that the mean return of the first trading day of the week, is equal to the mean return during

the rest of the week. This test is based on a single-factor, between subjects, analysis of variance (ANOVA)

Difference of variance test of the null hypothesis that the variance of the first trading day return is equal to the variance return of the rest

of the week, depending on the Levene test

148

4-19, the difference-of-means test is statistically significant for all markets except

Bahrain, Dubai, and Egypt. Indicating that, the first trading day returns are significantly

positive and different from the returns during the rest of the week. Moreover, the standard

deviation of the first trading day is lower than the standard deviation during the rest of the

week, and a difference-of-variance test shows that the difference is statistically

significant for all markets, which indicate that first trading day returns are significantly

less volatile than returns during the rest of the week.

-Month-of-the-year effect (January effect)

Table 4-20 gives the results for January effect and those months, for which their

parameters found to be statistically significant. Despite the results that Bahrain, Dubai,

Oman, and Saudi Arabia do not exhibit January effect, all markets show signs of monthly

effect other than January; since several months’ parameters found to be positive and

statistically significant. The results for Jordan are incompatible with those of Maghayereh

(2003) since he does not find monthly effect in the Jordanian stock market, whilst the

results for Kuwait are constant with those of Al-saad and Moosa (2005) and Al-loughani

(2003).

Table 4-20

OLS Results for Month-of-the-Year Effect (January Effect)

Market Variable Coefficient Std. Error t-Statistic Prob.

Abu Dhabi

1/7/2001-31/12/2003 JANUARY 0.0002 0.0001 2.988 0.003

FEBROUARY 0.996 0.049 20.351 0.000

MARCH 1.002 0.033 30.489 0.000

ABRIL 0.999 0.027 36.988 0.000

MAY 1.003 0.035 28.401 0.000

JUNE 0.998 0.040 25.062 0.000

JULY 0.989 0.037 26.967 0.000

AUGUST 0.984 0.046 21.309 0.000

SEPTEMBER 1.000 0.025 39.953 0.000

OCTOBER 0.996 0.062 16.149 0.000

NOVEMBER 0.990 0.035 28.602 0.000

DECEMBER 0.989 0.033 29.636 0.000

Jordan

1/1/1992-14/3/2005 JANUARY 0.232 0.051 4.586 0.000

ABRIL 0.337 0.068 4.929 0.000

MAY 0.331 0.063 5.289 0.000

JUNE 0.310 0.070 4.456 0.000

JULY 0.216 0.048 4.468 0.000

AUGUST 0.241 0.055 4.342 0.000

SEPTEMBER 0.224 0.050 4.530 0.000

OCTOBER 0.280 0.071 3.974 0.000

NOVEMBER 0.372 0.060 6.197 0.000

149

…continue table 4-20

Market Variable Coefficient Std. Error t-Statistic Prob.

Bahrain

2/1/1991-3/6/2004 JANUARY 0.045 0.513 0.088 0.930

MAY -0.447 0.167 -2.685 0.007

AUGUST -0.499 0.015 -32.933 0.000

Dubai

26/3/2000-31/12/2003 JANUARY 0.186 0.189 0.988 0.323

JULY 0.259 0.125 2.073 0.038

AUGUST 0.417 0.176 2.370 0.018

Egypt

1/1/1998-31/12/2004 JANUARY 0.591 0.069 8.514 0.000

ABRIL 0.305 0.096 3.159 0.002

JULY 0.246 0.089 2.764 0.006

AUGUST 0.252 0.090 2.805 0.005

SEPTEMBER 0.237 0.076 3.120 0.002

OCTOBER 0.212 0.067 3.163 0.002

NOVEMBER 0.256 0.100 2.574 0.010

DECEMBER 0.319 0.090 3.525 0.000

Kuwait

17/6/2001-30/11/2005 JANUARY 0.361 0.155 2.323 0.020

MARCH 0.297 0.106 2.798 0.005

ABRIL 0.425 0.105 4.055 0.000

MAY 0.245 0.090 2.721 0.007

JUNE 0.279 0.089 3.138 0.002

SEPTEMBER 0.256 0.083 3.078 0.002

Oman

1/2/1997-13/10/2004 JANUARY 0.000 0.000 1.842 0.066

FEBROUARY 1.001 0.052 19.099 0.000

MARCH 0.997 0.045 22.168 0.000

ABRIL 0.999 0.032 31.412 0.000

MAY 0.996 0.030 33.267 0.000

JUNE 0.994 0.036 27.777 0.000

JULY 0.998 0.034 29.505 0.000

AUGUST 1.003 0.043 23.252 0.000

SEPTEMBER 1.000 0.036 27.835 0.000

OCTOBER 1.005 0.041 24.616 0.000

NOVEMBER 0.998 0.019 53.089 0.000

DECEMBER 1.001 0.030 33.892 0.000

Palestine

8/7/1997-28/2/2005 JANUARY 0.586 0.111 5.256 0.000

FEBROUARY 0.480 0.103 4.671 0.000

ABRIL 0.595 0.110 5.390 0.000

MAY 0.368 0.094 3.931 0.000

JULY -0.822 0.055 -14.943 0.000

SEPTEMBER 0.355 0.098 3.625 0.000

OCTOBER 0.320 0.086 3.713 0.000

Saudi Arabia

26/1/1994-14/3/2005 JANUARY 0.079 0.082 0.970 0.332

FEBROUARY 0.191 0.078 2.467 0.014

MARCH 0.247 0.067 3.714 0.000

ABRIL -0.302 0.032 -9.519 0.000

MAY 0.122 0.053 2.327 0.020

JULY 0.314 0.100 3.128 0.002

SEPTEMBER -0.116 0.051 -2.299 0.022

NOVEMBER 0.224 0.076 2.941 0.003

The estimated equation is: . Where

Rt = α 1 D1 + α 2 D2 + ... + α 12 D12 + ε t tR is the daily index

return which defined as ln(pt/pt-1), α1 through α12 are coefficients to be estimated using ordinary least

squares.D1-D12 are dummy variables such that if t is January, then D1=1 and D1=0 for all other monthes, if t

150

Furthermore, table 4-21 shows the result of Chow test which indicates that the

estimated coefficients are structurally unstable for three markets (Egypt, Palestine, and

Saudi), whereas the hypothesis that all parameters are insignificantly different from zero,

has been rejected for all markets.

Table 4-21

Chow Test for Structural Stability

Market AbuDhabi Jordan Bahrain Dubai Egypt Kuwait Oman Palestine Saudi

Break point 1-Jul-02 27-Jul-98 16-Sep-97 2-Mar-02 2-Jul-01 4-Nov-03 26-Aug-02 15-Oct-01 1-Sep-99

F -test 0.592 1.768 0.165 0.757 2.264 0.887 0.130 4.337 4.153

P -value 0.849 0.048 0.999 0.695 0.008 0.560 1.000 0.000 0.000

Chow test is implemented to test that the estimated coefficients are structurally stable over the entire sample period.

Market AbuDhabi Jordan Bahrain Dubai Egypt Kuwait Oman Palestine Saudi

Chi-square 8787.846 220.879 1093.040 24.002 150.992 64.064 10466.45 352.900 144.061

Probability 0.000 0.000 0.000 0.020 0.000 0.000 0.000 0.000 0.000

Wald test tests the null hypothesis that β 1 = β 2 = ... = β 12 = 0

Moreover, table 4-22 indicates that the variance of returns are not equal, between

January and the rest of the year, while the mean returns in January and the rest of the year

found to be equal only in three markets (Bahrain, Dubai, and Egypt). The majority of

markets demonstrate some type of monthly effect, which could be confusing in that, it

cannot be explained under the umbrella of the existing explanations in the literature. For

instance, “window press” such that, investors sell in December those stocks that did not

doing well, in order to show reduced profit and pay less tax. This would results in lower

returns in December, whilst in January returns should rise due to reinvestment in new

stocks, but in our case, all returns in December are positive or not significant. Another

explanation could be “summer holiday”, in which we expect returns in July and August

to be negative, the case which exists only in Bahrain (negative August returns). However,

in the case of Arab stock markets; especially GCC stock markets; they witnessed a

considerable improvements in their activity and liquidity (see figures 2-2 and 2-4), as a

result of the huge raise in oil prices in the last two years, which produces a surplus in

liquidity in these countries, leading to increase the trading activity in their financial

markets.

151

Table 4-22

Month-of-the-Year Effect in the First Two Moments

Market AbuDhabi Jordan Bahrain

Mean St. Dev. Mean St. Dev. Mean St. Dev.

Returns on January 0.0002 0.0012 0.0001 0.0026 -0.0001 0.0017

Returns during Rest of the year 0.0008 0.0053 0.0005 0.0076 0.0003 0.0568

Difference of Means Test 6.868** 7.621** 0.127

difference of Variance Test 281.24* 1868.31* 19.91*

Mean St. Dev. Mean St. Dev. Mean St. Dev.

Returns on January 0.0002 0.0016 0.0003 0.0055 0.0002 0.0019

Returns during Rest of the year 0.0004 0.0101 0.0005 0.0165 0.0020 0.0095

Difference of Means Test 0.829 0.236 34.99*

difference of Variance Test 218.4* 1027.5* 716.26*

Mean St. Dev. Mean St. Dev. Mean St. Dev.

Returns on January 0.0001 0.0018 0.0000 0.0041 0.0000 0.0020

Returns during Rest of the year 0.0004 0.0058 0.0011 0.0183 0.0005 0.0094

Difference of Means Test 3.49*** 4.081** 9.433*

difference of Variance Test 562.64* 593.31* 1383.88*

*,**,*** indicate significant at 1%, 5%, 10% levels respectively.

Difference of means test of the null hypothesis that the mean return of January, is equal to the mean return during the rest of the

months. This test is based on a single-factor, between subjects, analysis of variance (ANOVA)

Difference of variance test of the null hypothesis that the variance of January return is equal to the variance return of the rest months

depending on the Levene test.

Figure 4-2 presents the average returns in the period May-October and the period

November-April for each market. As can be seen in figure 4-2, the differences in returns

among the two half-year periods are generally large and economically significant for 5

out of 9 markets.

Figure 4-2 Average Returns Among the Two Half-Year Periods

152

Table 4-23 shows the Halloween indicator in Arab stock markets. As mentioned

previously, a positive and significant α1 parameter is evidence of a Halloween effect.

Since α1 denotes the average returns in the period November-April in excess of the

average return during the other six months of the year. Table 4-23 indicates that, all

markets (except Dubai and Saudi) exhibit highly statistically significant Sell in May

effect at the 1% level.

Table 4-23

The Halloween Indicator in Arab Stock Markets

Rt = µ + α 1 S t + ε t

N. of

Countries observation Mean P-Value α1 P-value

AbuDhabi 512 0.0003 0.0873 0.9903 0.0000

Bahrain 3340 0.0003 0.7599 1.0008 0.0000

Dubai 1098 0.0004 0.1508 -0.0171 0.7654

Egypt 1733 0.0005 0.2468 0.1977 0.0000

Jordan 3226 0.0004 0.0008 0.2231 0.0000

Kuwait 688 0.0007 0.0549 0.9763 0.0000

Oman 1908 0.0001 0.4360 0.9998 0.0000

Palestine 1264 0.0003 0.4582 0.9973 0.0000

Saudi 3299 0.0005 0.0011 -0.0768 0.0008

R t represents monthly continuously compounded returns for the price indices. N , the number

of daily observations. The constant term µ represents the daily mean returns over the May-

October periods.The daily mean return over the November-April periods is represented by

µ+α 1 .

Following Fama’s arguments (Fama; 1998) that most long-term returns anomalies

tend to disappear with reasonable changes to technique, since Sell in May hypothesis

suggests that; average returns are higher during the period November-April than during

the period May-October. One might argues that, since the January effect generates high

positive returns in many stock markets, the Sell in May effect is simple a January effect

in disguise. To test for this hypothesis, we considered an additional regression and give

Sell in May Dummy the value 1 in the period November to April, except January. While

for January, we add an additional dummy. Table 4-24 presents the results of the

Halloween indicator with adjustment for January effect, the results indicate that all access

returns in January are entirely due to January effect (α2) and not caused by Sell in May

effect. Note that, the Halloween effect which presented by α1 still the same, highly

153

statistically significant without any noticeable reduction in α1’s values except for Egypt.

Indicating that despite the addition of January dummy, the Sell in May effect still exists

in 7 out of 9 Arab stock markets.

Table 4-24

The Halloween Indicator in Arab Stock Markets with January

Effect Adjustment Rt = µ + α 1 S t + α 2 J t + ε t

Countries Mean P-Value α1 P-value α2 P-value

AbuDhabi 0.0003 0.0863 0.9925 0.0000 0.9710 0.0000

Bahrain 0.0003 0.7595 0.9988 0.0002 1.0103 0.0870

Dubai 0.0004 0.1795 -0.0349 0.5617 0.1618 0.3960

Egypt 0.0003 0.3919 0.0952 0.0088 0.5876 0.0000

Jordan 0.0004 0.0008 0.2227 0.0000 0.2243 0.0000

Kuwait 0.0007 0.0551 0.9770 0.0000 0.9721 0.0000

Oman 0.0001 0.4357 0.9999 0.0000 0.9979 0.0000

Palestine 0.0003 0.4582 0.9970 0.0000 0.9992 0.0000

Saudi 0.0005 0.0010 -0.0895 0.0002 0.0775 0.3502

R t represents monthly continuously compounded returns for the price indices. N , the number of daily

observations. The constant term µ represents the daily mean returns over the May-October periods.The daily

mean return over the November-April periods is represented by µ+α 1 . The impact of January returns

represented by α 2 .

4-6 Summary

We have tried to answer the question whether Arab stock markets are efficient in

the weak-form sense. The empirical results obtained, enable us to provide strong

evidence against the random walk hypothesis for Arab stock markets. The results

obtained from regression analysis, variance ratio, BDS, runs test, and serial correlation

tests, reject the randomness and independence of the returns generating process, even

after indices have been corrected for infrequent trading. These results are consistent with

the existing literature for emerging markets, since many evidences of predictability in

emerging markets have been found and rejected the hypothesis that lagged price

information cannot predict future prices (Bekaert 1995; Harvey 1995b, 1995c; Claessense

et al. 1995; Buckberg 1995; Haque et al. 2001, 2004; and Bailey at al. 1990). Moreover,

the results indicate that, prices responding non-linearly to new information, while

volatility clustering phenomenon still seems to characterized markets’ returns. The

GARCH (1,1) results for daily data indicate that, all markets exhibit volatility clustering

with one exception for Dubai. Furthermore, volatility seems to be persistent in three

154

markets (Egypt, Kuwait, and Palestine) with a slow rate of decay, Oman displays an

increasing response function of volatility; shocks do not decay with time. Additionally,

four Arab markets (Bahrain, Dubai, Kuwait, and Oman) show signs of leverage effect

and asymmetric shocks to volatility.

The results also indicate that, the GARCH models explain quite satisfactory the

dependencies of the first and second moments; that are presented in the stock return

series, while the second moment found to be quite enough to explain the non-linearity

structure that has been found in the time series. The next task of this chapter was to

investigate the existence of calendar effects in Arab stock markets. The results indicate

that three calendar effects found to be in Arab markets, day-of-the-week, month-of-the-

year effects, and the Halloween indicator. However, the style of the first two anomalies is

not consistent with the existing literature. For instance; for most of the Arab markets; the

first trading day of the week found to be positive and significantly different from the rest

of the week’s returns, while several months of the year found to be significantly positive.

None of the existing explanations in the literature (tax loss, trading time, calendar

time hypothesis) reveal to be appropriate to explain these anomalies. One explanation

might be; according to the surplus liquidity between investors (especially GCC

investors), as a result of the sharp rise in oil prices and according to the lack of other

investment opportunities, site these markets as an attractive target for these investments.

This leads to huge improvements in most markets indicators during the last three years

such as market size and liquidity indicators.

In sum, the results obtained from this chapter enable us to declare that, Arab stock

markets under examination here are not efficient in the weak-form sense.

155

Are Arab Stock markets Can Arab Stock Markets Offer,

Methodology Integrated among themselves and for Both Regional and

With Other International Stock International Investors Unique

Markets? If Yes, How Do Shocks Risk and Returns Characteristics

Generated By International Stock to Diversify International and

Markets Especially UK, US, and Regional Portfolios?

Japan Affect Arab Stock

Markets?

Arab stock markets are segmented

Unit root test ADF, PP, and KPSS.

from international markets, even in

Multivariate cointegration, Johansen cointegration test.

Causality and vector error correction models (VEC). the short term horizon.

Structural vector autoregressive (SVAR). The segmentation also exists between

Vector auto regressive (VAR). non-oil countries.

Granger causality test. One cointegrating relation was found

Correlation coefficients. among Arab stock markets.

Weak linkages between Arab

markets in the short run.

GCC stock markets found to be

integrated with one cointegration

Findings relation, with weak short-term

interrelation.

Arab stock markets can offer both

international and regional portfolio

investors with diversification

potentials.

156

Methodology What Is the Effect of Oil Prices on the Performance of

GCC Stock Market?, And Whether These Markets

Have Predictive Power on Oil Prices or Vice Versa?

equation.

Multivariate cointegration analysis between GCC markets and oil prices.

Vector error correction model among oil prices and GCC markets.

Vector autoregressive (VAR) analysis between oil and GCC markets (under the

form of event studies).

markets.

Oil prices have significant role on GCC returns’

Findings volatility.

After the raise in oil prices, four of GCC markets

can predict oil prices while only two markets can be

predicted by oil prices.

157

5- Financial integration and diversification benefits among Arab and international

stock markets

Over the last 20 years, financial markets have become highly integrated, mainly

due to reductions in the cost of information, improvements in trading systems technology

and the relaxation of legal restrictions on international capital flows. The changes have

accelerated the interaction among financial markets and the enlargements of capital

mobility. Moreover, gains from international portfolio diversification are related

inversely to the correlation of equity returns, according to modern portfolio theory. In

line with this theory, investors have become highly active, investing in foreign equity

markets as a risk diversification strategy.

Numerous studies have demonstrated the advantage of international

diversification related to low correlation between various equity markets, such as Eun

and Resick (1984), Wheatly (1988), Meric and Meric (1989), Baily and Stulz (1990),

Divecha et al. (1992), Michaud et al. (1996), Meric et al. (2001), and Bulter and Joaquin

(2002). As developed in Gilmore and McManus (2002), the low correlations could be

explained by different types of barriers and regulations between the markets under

consideration. However, recent studies outlined in Gilmore and McManus (2002) reveal a

significant increase in correlations between equity markets, especially during and after

the 1987 international equity markets crash.

This tendency for the global markets to become more integrated is a result of the

increasing tendency toward liberalization and deregulation in the money and capital

markets, both in developed and developing countries as well as on a bilateral and

multilateral basis, commencing from, for example, trade liberalization and multilateral

trade initiatives. Such liberalization is important to introduce structural reforms, to

promote economic efficiency, to estimate trade and investment, and to create a necessary

climate for promoting sustainable economic growth with a commitment to market-based

reforms. As a sequence, increases in correlations between markets would imply a

decrease in the benefits from international diversification in line with portfolio theory.

Furthermore, long-run linkages between stock markets have important regional

and global implications at the macro-level, as a domestic capital market cannot be

insulated adequately from external shocks, thus the scope for independent monetary

158

policy may become limited. It is argued in Errunza et al. (1999) that the use of return

correlations at the market index level to infer gains from international diversification,

involving foreign-traded assets overstates the potential benefits. The gains must be

measured beyond those attainable through home-made diversification by mimicking

returns on foreign market indices with domestically traded securities.

In addition, Kasa (1992) argued that benefits from international diversification

evaluated by low correlations, can be overstated if the investor has a long-term

investment horizon and the markets are trading together. The implication of this is that,

any benefits that arise from diversification will be eradicated in the long run and,

therefore, investors with long-run horizons may not actually benefit from international

portfolio diversification. As a sequence, recent studies have used cointegration techniques

to analyze long-run linkages and co-movements, especially between US and various

emerging markets. The results concerning long-term diversification benefits for US

investors are somewhat mixed. Long-term linkages between US and various European

stock markets are found in Kasa (1992) and Arshanapalli and Doukas (1993), but results

in Bayers and Peel (1993), Kanas (1998), and Maneschiold (2004) suggest that there are

no such linkages.

Mixed results also found between the USA and the Pacific area, as discussed in

Gultekin et al. (1989), Harvey (1991), Cambell and Hamao (1992), and Sewell et al.

(1996), as well as for a group of Asian countries, as discussed in Neih and Chang (2003),

Gilmore and McManus (2002). While the integration between some Arabian markets and

US discussed in Maneschiold (2005), who finds that Egypt can offer diversification

benefits for US investors. Darrat and Hakim (2000) find that stock markets in the Middle

East are segmented from the global markets. While Neaime (2002) finds that GCC stock

markets, still offer diversification benefits for international investors. However, the

results between US and international markets reveal, in general, a higher degree of

independence with emerging markets than developed markets. Given the increased

correlations in recent periods between various international equity markets and the

general results from previous studies, investors are searching for markets that are more

promising for diversification benefit, Gilmore and McManus (2002) find evidence of

diversification benefits between US markets and the newly re-opened markets in the

159

Czech Republic, Hungary and Poland, while Maneschiold (2004) finds evidence of this

between the USA and the Baltic states.

The purpose of this chapter is to analyze possible diversification benefits, which

Arab stock markets may offer for both regional and international investors. The analysis

will be carried out on different levels. Firstly, the diversification potentials for

international investors will be analyzed. The US stock market (S&P 500) will be used to

represent the world markets, in view of the growing evidence that assigns considerable

weight to the US market in global capital markets. Second, the integration among Arab

stock markets them selves will be investigated. And finally, it is important to take into

account the important features that affect Arab stock markets, such as oil prices, so the

dynamic relationships between GCC stock markets and oil prices will be analyzed.

In order to investigate the possible diversification benefits for international

investors in the Middle East region, the analysis will relay on the Johansen cointegration

procedures, as well as Granger causality and correlation tests, to reveal short and long-

run relationship between Arab and international markets, and the possible diversification

benefits for US investors on a multivariate basis. Moreover, to investigate the dynamic

relationship in the short-run between Arab and international stock markets, the structural

vector auto regressive (SVAR) will be used to trace the effect of shocks, generated by

international stock market, on Arab markets.

A prerequisite for cointegration is that non-stationary series are integrated of the

same order. Therefore, the first step is to determine the order of integration for each

variable. Three tests will be employed in this investigation: the augmented Dickey-Fuller,

the Phillips-Perron, and the Kwaitkowski-Phillips-Schmidt-Shin (KPSS) tests (Dickey

and Fuller 1979; Phillips and Perron 1988; Kwaitkowski et al. 1992).

The basic features of unit root tests can be presented as follows, consider a simple

AR(1) process:

160

y t = ρy t −1 + δ ′x t + ε t , (5-1)

where xt are optional exogenous regressors which may consist of constant, or a constant

and trend, ρ and δ are parameters to be estimated, and the εt are assumed to be white

noise. If |ρ| ≥ 1, y is a non stationary series and the variance of y increases with time and

approaches infinity. If |ρ| <1, y is a (trend-) stationary series. Thus the hypothesis of

(trend-) stationary can be evaluated by testing whether the absolute value of ρ is strictly

less than one.

In general, one can test the null hypothesis H0: ρ=1against the one-side alternative

H1: ρ <1. In some cases, the null is tested against a point alternative. In contrast, the

KPSS Lagrange Multiplier test evaluates the null of H0: ρ<1 against the alternative H1:

ρ=1.

- The augmented Dickey-Fuller (ADF) test

The standard Dickey-Fuller test is carried out by estimating (5-1) after subtracting

yt-1 from both sides of the equation:

∆y t = αy t −1 + δ ′xt + ε t (5-2)

where α=ρ-1. The null and alternative hypothesis may be written as H0: α=0, H1: α<0

and evaluated using the conventional t-ratio for α:

ta = aˆ /( se( aˆ ))

Dickey and Fuller (1979) show that under the null hypothesis of a unit root, the

statistic does not follow the conventional Student’s t-distribution, and they derive

asymptotic results and simulate critical values for various test and sample sizes. More

recently, MacKinnon (1991, 1996) implements a much larger set of simulations than

those tabulated by Dickey and Fuller. In addition, MacKinnon estimates response

161

surfaces for the simulation results, permitting the calculation of Dickey-Fuller critical

values and P-values for arbitrary sample size.

The simple Dickey-Fuller unit root test described above is valid only if the series

is an AR(1) process. If the series is correlated at higher order lags, the assumption of

white noise disturbances εt is violated. The Augmented Dickey-Fuller (ADF) test

constructs a parametric correction for higher order correlation by assuming that the y

series follows an AR(p) process and adding p lagged difference terms of the dependent

variable y to the right-hand side of the test regression:

this augmented specification is then used to test the null hypothesis that H0: α=0 against

H1: α<0, using the t-ratio. An important result obtained by Fuller is that the asymptotic

distribution of the t-ratio for α is independent of the number of lagged first differences

included in the ADF regression.

Moreover, while the assumption that y follows an autoregressive (AR) process

may seem restrictive, Said and Dickey (1984) demonstrate that the ADF test is

asymptotically valid in the presence of a moving average (MA) component, provided that

sufficient lagged differences terms are included in the test regression.

Moreover, there are two practical issues in performing an ADF test. First, one

must choose whether to include exogenous variables in the test regression. You have the

choice of including a constant, a constant and a linear time trend, or neither, in the test

regression. One approach would be to run the test with both a constant and a linear trend

since the other two cases are just special cases of this more general specification.

However, including irrelevant regressors in the regression will reduce the power of the

test to reject the null of a unit root.

Second, you will have to specify the number of lagged difference terms, to be

added to the regression (zero yields the standard DF test; integers greater than zero

correspond to ADF tests).

162

- The Phillips-Perron (PP) test

The PP method proposes as an alternative (non-parametric) method of controlling

the serial correlation with testing for a unit root. The PP method estimates the non-

augmented Dickey-Fuller test, and modify the t-ratio of the α coefficient, so that, serial

correlation does not affect the asymptotic distribution of the test statistic. The PP test is

based on the statistic

1/ 2

~ γ T ( f 0 − γ 0 )(se(aˆ ))

ta = t a 0 − (5-4)

f0 2 f 01 / 2 s

where â is the estimate; ta the t-ratio of α, se( â ) is coefficient standard error; s is the

standard error of the test regression. In addition, γ0 is a consistent estimate of the error

variance calculated as (T-k) s2/T, where k is the number of regressors. The remaining

term, ƒ0, is an estimator of the residual spectrum at frequent zero.

The Kwaitkowski et al. (1992) KPSS test, differs from the other unit root tests in

that, the series yt is assumed to be (trend-) stationary under the null. The KPSS statistic is

based on the residuals from the OLS regression of yt on the exogenous variables Xt:

yt = xt δ ′ + ut (5-5)

the LM statistic is be defined as:

LM = ∑ s (t ) / T 2 f 0

2

( ) (5-6)

t

where f0, is an estimator of the residual spectrum at frequency zero and where s(t) is a

cumulative residual function:

t

s (t ) = ∑ u t

r =1

The finding that many macro time series may contain a unit root has spurred the

development of the theory of non-stationary time series analysis. Engle and Granger

163

(1987) pointed out that a linear combination of two or more non-stationary series may be

stationary. If such a stationary linear combination exists, the non-stationary time series

are said to be cointegrated. The stationary linear combination is called the cointegrating

equation and may be interpreted as a long-run equilibrium relationship among the

variables.

The purpose of the cointegration test is to determine whether a group of non-

stationary series is cointegrated or not. The presence of a cointegrating relation forms the

basis of the VEC specification. To check whether the series are cointegrated, specifically,

having established the presence of a unit root in the first-difference of each variable, we

need to test whether the series in each market has different unit root (non-cointegrated),

or share the same unit root (cointegrated). Cointegrated variables, if disturbed, will not

drift apart from each other and thus possess a long-run equilibrium relationship. The

existence of cointegration between two series would imply that the two series would

never drift too far apart. A non-stationary variable, by definition, tends to wander

extensively over time, but a pair of non-stationary variables may have the property that a

particular linear combination would keep them together, that is, they do not drift too far

apart. Under this scenario, the two variables are said to be cointegrated, or possess a long-

run (equilibrium) relationship.

Cointegration can be tested on bivariate base using two stage method of Engle

and Granger (1987), while a multivariate cointegration test can be carried out using

Johansen (1988) approach based on the autoregressive representation.

If there are two variables, xt and yt, which are both non-stationary in levels but

stationary in first difference, then xt and yt are integrated of order one, I(1), and their

combination having the form z t = xt − ayt is also I(1). However, if zt is integrated of

order zero, I(0), the linear combination of xt and yt is stationary and the two variables are

said to be cointegrated. If two variables are cointegrated, there is an underlying long-run

relationship between them. In the short-run the series may drift apart, but if they are

cointegrated, they will move toward long-run equilibrium through an error-correction

mechanism.

If xt and yt are integrated of the same order, the Engle-Granger method then

estimates the long-run equilibrium relationship as:

164

yt = β 0 + β1 xt + et (5-7)

using ordinary least squares (OLS). The residual series et is then tested for stationary:

rejection the null hypothesis α1 = 0 implies that the residual series is stationary and that

the two series are cointegrated. Engle and Granger provide statistics to test the hypothesis

α1 = 0. When more than two variables are involved, the appropriate tables are of those of

Engle and Yoo (1987). If the variables are found to be cointegrated, an error-correction

model can then be estimated using the residuals from the equilibrium regression:

∆y t = α 1 + α y et −1 + ∑ α 11 ∆y t −i + ∑ α 12 ∆xt −i + ε yt (5-9)

any error introduced in estimating the error term comes into the error-correction model.

Also, it requires that one variable be placed on the left-hand side of the equation, with the

other variables as regressors in the cointegrating equation. The results of one regression

may indicate that the variables are cointegrated, while the other regression suggests no

cointegration.

The Johansen (1988) approach circumvents the use of two-step estimators and

estimates as well as tests for the presence of multiple cointegrating vectors. This method

relies on the relationship between the rank of a matrix and its characteristic roots, or

eigenvalues. Let xt be a vector on n time series variables, each of which is integrated of

order (1) and assume that xt can be modeled by a vector autoregressive (VAR):

yt = A1 yt −1 + ... + A p y t − p + β xt + ε t (5-11)

165

where yt is a k-vector of non-stationary I(1) variables; xt is a d-vector of deterministic

variables; and εt is a vector of innovations, we can write this VAR as

p −1

∆y t = ∏ y t −1 + ∑ Γi ∆y t −i + β xt + ε t (5-12)

i =1

where

p p

∏ = ∑ Ai − I , Γi = − ∑ A j

i =1 j =i +1

(5-13)

reduced rank r<k, then there exist k*r matrices a and β each with rank r such that

Π = α β ′ and β ′yt is I(o). r is the number of cointegrating relations (the cointegrating

rank) and each column of β is the cointegrating vector. The elements of α are known as

the adjustment parameters in the VEC model. Johansen’s method is to estimate the ∏

matrix from an unrestricted VAR, and test whether we can reject the restrictions implied

by the reduced rank of∏.

Three cases are possible, first, if Π is of full rank, all elements of y are stationary

and none of the series has a unit root. Second, if the rank of Π=0, there are no

combinations which are stationary and there are no cointegrating vectors. Third, if the

rank of Π is r such that 0<r<k, then the y variables are cointegrated and there exist r

cointegrating vectors. The number of distinct cointegrating vectors can be obtained by

determining the significance of the characteristics roots of Π. To identify the number of

characteristic roots that are not different from unity, we can use two statistics, the trace

test and the maximum eigenvalue test:

166

where λ equals the estimated values of the characteristic roots (eigenvalues) obtained

from the estimated Π matrix, r is the number of cointegrating vectors, and T equals the

number of usable observations.

The trace test evaluates the null hypothesis that the number of distinct

cointegrating vectors is less than or equal to r against a general alternative. The

maximum eigenvalue test examines the number of cointegrating vectors versus that

number plus one. If the variables in yt are not cointegrated, the rank Π is zero and all

characteristics roots are zero. Since Ln (1) = 0, each of the expression Ln (1-λi) will equal

zero in that case. Critical values for the test are provided by Johansen and Juselius (1990)

and by Osterwald-Lenum (1992).

When Sims (1980) introduced vector autoregression (VAR) into economics, the

main thrust was that VAR modeling avoids “incredible” identifying assumptions made by

traditional large-scale macroeconomic models. Subsequently, the great bulk of structural

VAR work has focused on contemporaneous relationships between variables or between

residuals in a system of equations. Sims and Zha (1999) show how to make Bayesian

inference under a flat prior in both reduced-form VARs and identified VARs. In that

paper, they consider various types of identifying restrictions only on contemporaneous

coefficients.

There are instances, however, in which over-identification in VAR relates to lag

structure as certain lags do not enter certain equations. In many empirical applications,

such restrictions are not unreasonable; on the contrary, restrictions on the lag structure are

necessary precisely on the ground of economic reasoning. These situations frequently

stem from some block exogeneity restrictions such as the crucial small-open-economy

feature in international economics from some beliefs that certain lags do not appear in

certain equations (e.g., Zellner and Plam, 1974; Zellner, 1985; Leeper and Gordon, 1992;

Sims and Zha, 1995; Bernank et al., 1997). Failing to impose these restrictions because

they may complicate statistical inference not only is economically unappealing, but also

may result in misleading conclusions.

167

In order to introduce the basic elements of VAR analysis, suppose that we can

represent a set of n economic variables using a vector (a column vector) yt of stochastic

processes, jointly covariance stationary without any deterministic part and possessing a

finite order (p) autoregressive representation

A( L) yt = ε t (5-16)

A( L ) = I − A1 L − ... − A p L p (5-17)

the roots of the equation det[A(L)] are outside the unit circle in the complex domain and

εt has an independent multivariate normal distribution with zero mean

εt ∼ IMN(0,Σ)

E(εt) = 0

E (ε t ε t′ ) = Σ det( Σ ) ≠ 0

E (ε t ε s′ ) = [0] s≠t

The yt process has a dual Vector Moving Average representation (Wold

representation)

Yt = C(L) εt

C(L) = A(L)-1

C(L) = I + C1L + C2L2 +…

Where C(L) is a matrix polynomial which can be of infinite order and for which it has

been assumed that the multivariate invertibility conditions hold, i.e. det[C(L)] = 0 has all

the roots outside the unit circle, so

C(L)-1 = A(L)

Suppose that there are T+p observations for each variable represented in the yt vector; we

are thus able to study the system

168

This system can be conceived as a particular reduced form (in which all variables can be

seen as endogenous).

A VAR model has to be considered as a reduced form model where no

explanations of the instantaneous relationships among variables are provided. These

instantaneous relationships are naturally hidden in the correlation structure of Σ matrix,

and left completely un-interpreted. This becomes evident when the model is put into its

equivalent Vector Moving Average (VMA) representation, where the interpretability of

the coefficients becomes problematic, given the contemporaneous correlation structure of

the error terms. Sims’ (1980) original proposal consisted in moving from a non-

orthogonal VMA to an orthogonalized VMA representation via Choleski factorization of

the Σ matrix. This amounts to starting from the reduced form VAR representation

A(L)yt = εt , εt ∼ VWN(0,Σ)

where εt is a normal distributed vector white noise (VWN), and to pre-multiply the system

by the inverse Choleski factor of Σ

p

A* ( L) = ∑ Ai* , A0* = p −1 , Ai* = p −1 Ai , pp ′ = Σ

i =0

where p is the Choleski factor of Σ, and clearly A0* is lower triangular with unit diagonal

elements. This amounts to modeling contemporaneous relationships among the

endogenous variables in a triangular recursive form. The resulting orthogonal VMA

representation is

∞ ∞

y t = ∑ C i pet −i = ∑ Φ i et −i , Φ i = Ci , Φ 0 = p

i =0 i =0

notice that, since Φ0 = p, the orthogonal VMA representation shocks et have instantaneous

effects on the elements of yt according to the triangular scheme given by the Choleski

factor p.

169

Moreover, it is true that given the matrix Σ, the Choleski factor p is uniquely

determined. Nevertheless, if the elements of yt were permuted and arranged in y t* , the

rows and columns in Σ would have to be permuted accordingly to generate Σ*. The matrix

Σ* would then have a different Choleski factor: p * p′* = Σ * which would produce a

different orthogonalized VMA representation. Therefore, the orthogonal VMA

representation corresponding to the Choleski decomposition of variance covariance

matrix of the reduced form disturbances is unique only given a particular ordering of the

observable variables contained in yt.

The triangular representation, which is sometimes referred to as Wold causal

chain, is clearly a very particular one which cannot be considered suitable to every

applied context. Sometimes, the researcher might have in mind different schemes for

representing these instantaneous correlations, outside the straitjacket of the triangular

structures.

In recent literature, these alternative ways of modeling instantaneous correlation

can be summarized in the following terms. Recent literature on the so-called structural

VAR approach uses different ways of structuring the VAR model such as the K-model,

the C-model, and the AB-model.

- K-model

In the K-model, K is a (n*n) invertible matrix such that

KA(L)yt = K εt

K εt = et

E(et) = 0 E (et et′ ) = I n

transformation on the εt disturbances by generating a vector (et) of orthogonalized

disturbances (its covariance matrix is not only diagonal, but also equal to the unit matrix

In). Contemporaneous correlations among the elements of y are therefore modeled

170

through the specification of the invertible matrix K. the structural K-model can be thought

of as a particular structural form with orthonormal disturbance vector.

Note that, assuming we know the true variance covariance matrix of εt terms from

Kε t = et

Kε t ε t′K ′ = et et′

taking expectations, one immediately obtains

KΣK ′ = I n

the previous equation implicitly imposes n(n+1)/2 non-linear restrictions on the K matrix,

leaving n(n-1)/2 free parameters in K.

- C-model

C is a (n*n) invertible matrix such that

A(L)yt = εt

εt = Cet

E(et) = 0 E (et et′ ) = I n

In this particular structural model, we have a structural form where no instantaneous

relationships among the endogenous variables are explicitly modeled. Each variable in

the system is affected by a set of orthonormal disturbances whose impact effect is

explicitly modeled via the C matrix.

Sims (1988) stresses the point that there is no theoretical reason to suppose that C

model should be a square matrix of the same order as K. If C were a square matrix, the

number of independent (orthonormal) transformed disturbances would be equal to the

number of equations. Many reasons lead us to think that the true number of originally

independent shocks to our system could be very large. In that case, the C matrix would be

a (n*m) matrix, with m much greater than n.

In this sense, this research path is opposite to the one studied by the factor

analysis, which attempts to find m (the number of independent factors) strictly smaller

than n. the case of a rectangular (n*m) matrix C, with m>n, conceals a number of

problems connected with the completeness of the model and the aggregation over agents

(see Blanchard and Quah, 1989). In the C-model, the εt vector is regarded as being

171

generated by a linear combination of independent (orthonormal) disturbances. This may

have a different meaning than that of the K-model, where one is concerned with the

explicit modeling of the instantaneous relationships among endogenous variables.

As for the C-model, notice that from

ε t = Cet

ε t ε t′ = Cet et′C ′

taking expectations,

Σ = CC ′

if, again, we assume to know Σ, the previous matrix equation implicitly imposes a set of

n(n+1)/2 non-linear restrictions on the C matrix, leaving n(n-1)/2 free elements in C.

- AB-model

A,B are (n*n) invertible matrices such that:

A A(L)yt = A εt

A εt = Bet

E(et) = 0 E (et et′ ) = I n

In this kind of structural model, it is possible to model explicitly the instantaneous links

among the endogenous variables, and the impact effect on the orthonormal random

shocks hitting the system. Notice that A matrix includes a transformation on the εt

disturbances vector, generating a new vector (Aεt) that can be conceived as being

generated by linear combinations (through the B matrix) of n independent (orthonormal)

disturbance. Obviously this structure might have a different meaning than those of

models K and C.

Notice also that the AB-model can be seen as the most general parameterization

nesting the C and K models as special cases. In fact, the C-model can be seen as a

particular case of the AB-model, where A is chosen to be the identity matrix and the K-

model corresponds to an AB-model with a diagonal B matrix. As in the previous case,

from

A εt = Bet

Aε t ε t′ A′ = BB′

172

for Σ known, this equation again imposes a set of n(n+1)/2 non-linear restrictions on the

parameters of the A and B matrices, leaving overall 2n2-n(n+1)/2 free elements.

For more investigation of the relationship between Arab and international stock

markets, the structural vector autoregression (SVAR) will be used to analyze the

importance of interconnection between financial markets. The relationship between

international markets (US, UK, and Japan) and each of the Arab stock markets will be

investigated. The model incorporate the assumption that the returns on each of the three

international markets, affect the returns on Arab markets but not vice versa. In other

words, how do Arab stock markets response to shocks generated by international stock

markets. The main purpose of SVAR is to obtain non-recursive orthogonolization of the

error terms for impulse response analysis.

In order to incorporate and capture the dynamic relationship among prospective

returns, a block recursive model, similar to the SVAR model used by Zha (1999),

Cushman and Zha (1997), and Berument and Ince (2005), will be used to examine the

effect of a large economy’s stock exchange movements (in our case US, UK, and Japan)

on a small economy’s stock exchange movements (each of the Arab stock markets). The

foreign stock exchange index follows its own dynamics (an AR process is used as a

proxy). Domestic stock exchange movements are affected by its own lag and movements

of the foreign stock exchange. Therefore, the foreign stock market can be thought to have

an exogenous effect on the domestic stock exchange. None of the lag variables of the

domestic market determine foreign stock exchange; however, lag values and spot values

of the foreign stock exchange affect domestic stock exchange.

The VAR model has some advantages relative to the single equation model, since

the VAR model allows dynamic interactions among variables and the VAR model has

predictive power compared to the single equation model. VAR with block exogeneity is

also used, since in conventional VAR; stock exchange movements of foreign markets are

affected by domestic stock exchanges including lag values. By block exogeneity, this

problem is overcome.

The general specification of the identified VAR model of Cushman and Zha

(1997) is:

173

A(L) y(t ) = ε t (5-19)

in which, the A(L) is an mxm matrix polynomial in the lag operator L, y(t) is the mx1

observations vector and εt is the mx1 vector of structural disturbances. Equation (5-20)

shows the specification of the model.

y (t ) A (L ) 0 ε (t )

y (t ) = 1 , A(L ) = 11 , ε (t ) = 1 (5-20)

y 2 (t ) A21 (L ) A22 (L ) ε 2 (t )

in equation (5-20), it assumed that ε(t) is uncorrelated with y(t-j) for j>0 and A(0) is non-

singular. Block exogeneity is represented by A12(L) in the matrix, which is zero. This

means that y1(t) is exogenous to the second block both simultaneously and also for lagged

values. The observation matrices are such that y1 = [Foreign stock exchange], y2 =

[Domestic stock exchange].

Capital markets across countries or regions may exhibit varying degree of

integration (or segmentation). Theoretically, market linkages primarily stem from the

“low of one price” that identical assets; physical or financial, should bear the same price

across countries after adjusting for transaction costs. Rational (well-informed) investors

would, or perhaps should, arbitrage away price disparities, leading to more integrated

markets.

For the last few years, the development of financial markets in the Middle East

region has opened a new era of mobility of financial resources, whereby flow of private

capital has assumed an increasing role as a source of finance for these markets. More

favorable developments have been taking place in most Arab stock markets to attract

regional investors. Moreover, countries that share geographical proximity and have

similar groups of investors are more than likely to have markets that influence each other.

And when a stock is dually listed in two countries, shocks in one market can be

transmitted to the other market through the security, plus investors in one market may

react directly and indirectly to an initial stock in another market.

174

In this sequence, questions of market integration among Arab tock markets

themselves, are of concern both to regional investors, and companies in the region that

make capital budgeting decisions. Specifically, if segmentation exists and a firm is forced

to raise capital locally, then its cost of capital is likely to be higher than that of a company

with unrestricted access to the regional and international capital markets. Therefore, one

would expect the restriction to the local capital market, to raise a firm’s cost of capital.

The focus of this section will be firstly, to examine the relationship among Arab

stock markets. Second, to what extent and how rapid shocks induced by innovations in

one market, are borne by another markets. Third, we explore whether there is a dominant

market among Arab markets that links all other markets and makes most of their

independence. Furthermore, despite the fact that Arab countries enjoy several common

features, there still exist some differences especially in natural resources, such as oil

production. So the whole markets will be divided into two groups: oil production

countries, represented by GCC stock markets, and non-oil production countries; which

includes Jordan, Egypt, and Palestine.

In order to investigate the long-run relationship among Arab markets, Johansen

cointegration technique will be used, while Granger causality, coefficient-correlation and

vector autoregressive (VAR) will be employed to trace the short-run dynamics.

Causality is defined as in Granger (1969), where the Granger method determines

to what degree a current endogenous variable; can be explained by past values of the

variable and whether the explanatory power; can be improved adding lagged values of

another exogenous variable. If this is the case, the exogenous variable is said to Granger

cause the endogenous variable. The Granger causality is measured by estimating an

unrestricted and a restricted version of the equation

n n

yt = α 0 + ∑ α i y t −1 + ∑ β t − j xt − j + ε t (5-21)

i =1 j =1

175

and employing an F-test to determine if the parameters of the exogenous variable are

significantly different from zero, i.e. if the exogenous variable Granger cause the

endogenous variable. The constant is denoted by α0 and εt is a white noise error term.

The VAR technique as applied to a simultaneous equation system, estimates

unrestricted reduced form equations with uniform sets of the lagged dependent variables

of each equation as regressors. Because this approach sets no restrictions on the structural

relationships of the economic variables, it avoids mis- specification problems. The VAR

methodology is suitable when variables within the model are highly autocorrelated.

Furthermore, the VAR approach enables us to analyze the speed of information

transmission among variables in the system, which would provide insight into the

dynamic nature of the interactions between markets.

The VAR model can be expressed in its standard form as:

p

R (t ) = c + ∑ A (k )R (t − 1 ) + e (t )

k =1

(5-22)

Where R(t) is a 9x1 column vector of daily returns on the markets at time t. C is a 9x1

column vector of constant terms, A(k) is a 9x9 matrix of coefficients such that the (i, j)th

component of A(k) measures the direct effect that a change in the ith markets has upon

the jth market after k periods.

In particular, the ith component of e(t) is the innovation of the ith market that

cannot be predicted from the past returns of other returns in the system. e(t) is a 9x1

( ) ( )

column vector of innovations such that E (eit ) = 0 , E e it2 = σ i2 , E e it , e jt = σ ij and

E (eit , e jt − k ) = 0

contemporaneously correlated. To analyze the dynamics of the system, we trace out the

system’s moving average representation which may provide additional insight into the

dynamic interactions among the returns in the VAR model (Sims 1980). Thus, the VAR

176

model of equation (5-22) is typically transformed into its moving average representation

expressed as:

∞

R (t ) = ∑

k =0

B (k )e (t − k ) (5-23)

Equation (5-23) indicates that R(t) is a linear combination of current and past one-step-

ahead forecast errors (i.e. e(t)). The (i, j)th component of B(k) reveals the response of the

ith market return to a unit random shock in the jth market return after k periods. The

moving average model of equation (5-23) enables us to compute the m-step-ahead

forecast error of R(t) at time t-m+1 which can be expressed as ∑ B(k )e(t − k ) for K=0

to m-1. In addition, the variance decomposition of the forecast error gives us the

percentage of unexpected variation in each market’s return that is produced by shocks

from other returns in the system.

As stated earlier, the innovations, e(t) in equation (5-22) may be

contemporaneously correlated, i.e. the covariance matrix of innovations is not diagonal.

When innovations in market returns are contemporaneously correlated, a shock in one

market may work through the contemporaneous correlations with innovations in other

markets. It is customary to transform these correlations by orthogonalizing the

innovations in the VAR system according to a pre-specified causal ordering. After the

transformation, the above equation can be expressed as,

∞

R (t )= ∑ c (k )u (t − k ) (5-24)

k = 0

Where the transformed innovations, u(t), are now uncorrelated with each other at all lags

as well as contemporaneously. The moving-average representation of the VAR model

provides a convenient framework for tracing the dynamics to shocks in the system. The

(i, j)th component of C(k) in equation (5-24) represents the impulse response of the ith

market in k periods after a shock of one standard error in the jth market. That is, if there

is a unit shock in the innovation of the jth market in period t(ujt), the value of the ith

177

market (Ri), changes by cij,1 in the following period and by cij,2, cij,3 and so on in

successive future periods. The VAR model also makes it possible to analyze the

decomposition of forecast error variance thereby providing a measure of the overall

relative importance of an individual market in generating variations in its own returns and

in other markets. That is, the effect that each variable in the system has on itself and on

each other variables over different time horizons can be measured by decomposing this

forecast variance error.

In summary, the VAR analysis provides information on two important aspects of

the structure of interactions among the national stock markets: (i) if innovations in a

particular market explain a substantial amount of return variations in other markets and

cannot be accounted for by innovations in other markets, then the market is relatively

influential to other markets; and (ii) if the impulse response of a market to a shock in

another market tapers off quickly, then the transmission of information between these

markets is relatively efficient.

The VAR requires the determination of the appropriate lag structure in the

system. We chose the lag structure using the Akaike Information Criterion (AIC) in

conjunction with analyzing the estimated model’s residuals so they do not exhibit any

significant autocorrelation.

The empirical analysis of the relationships among Arab stock markets indices; as

an economic group, with international markets, and among Arab stock markets

themselves; requires that several time series tests to be conducted. The first step requires

that unit root tests be performed to determine whether the series are non-stationary in

levels and stationary in first differences, that is, integrated of degree one. The second step

is to use the cointegration test to determine whether these non-stationary series, have

common long-run relationships. Evidence of cointegration rules out the possibility that

the estimated relationships are spurious; thus, as long as the variables in a given VAR

have common trends, causality (in Granger sense but not the structural sense) must exist

in at least one direction. However, although cointegration implies the presence of

causality, it does not identify the direction of causality between variables. The dynamic

178

Granger causality can be captured through the vector error correction (VEC) models,

derived from the long-run cointegrating vectors. Furthermore, Engle and Granger (1987)

show that in the presence of cointegration, a corresponding error-correction model

representation always exist.

The data used for cointegration test consists of daily indices prices for Arab stock

markets and S&P 500 which has been used as a proxy for international markets. The data

comes from daily figures, run from 1st August 2001- 31 December 2004. The length of

this period is limited by the availability of data for all Arab stock markets included in this

test.

5-3-1 Integration

Integration for individual time series for each stock market index is tested by

means of unit root tests, which investigate the presence of a stochastic trend in the

individual series. The results of the three unit root tests (ADF, PP, and KPSS) are

presented in table 5-1.7

7

The indices used for the 13 markets are: Kuwait (KSEI); Jordan (JSMI); Bahrain (BSEI); Dubai (DFMI);

Egypt (EFMI); Oman (OSMI); Abu Dhabi (ABSMI); Palestine (PSEI); Saudi Arabia (SAUDI); Japan,

Nikkei 225 (JAPANI); US, S&P 500 (USAI); UK, FTSE100 (UKI); and oil prices (WTI) which is crude

stream produced in Texas and Southern Oklahoma .

179

Table 5-1

Unit Root Tests for Each Individual Series, Both in Levels and First Differences

Levels First Difference

Variables ADF Lags PP Lags KPSS BW ADF Lags PP Lags KPSS BW

KSEI -2.29 IT

1 -2.24 IT

9 0.32 IT

21 -17.72 I

1 -22.14 I

7 0.16 I 9

I

JSMI -1.12 IT 12 2.43 N 1 0.56 IT 43 -11.59 I 16 -43.63 I

12 0.29 1

BSEI -1.83 I 11 -1.61 I 24 0.42 IT 44 -14.08 I 10 -65.95 I 26 0.16 I 24

DFMI -1.61 IT 1 -1.56 IT 7 0.81 IT 25 -33.01 IT 1 -33.03 IT 8 0.03 IT 8

EFMI 0.22 IT 2 0.21 IT 6 0.81 IT 32 -27.38 IT 1 -32.95 IT 13 0.17 IT 5

OSMI -0.45 IT 6 0.74 N 15 0.98 IT 34 -17.12 N 4 -41.83 N 14 0.39 I 15

ABSMI 2.95 N 3 3.17 N 9 0.28 IT 21 -12.18 I 2 -22.25 I 7 0.15 I 9

PSEI -2.55 I 4 -2.46 I 4 0.44 IT 26 -13.82 N 5 -34.94 I 4 0.33 I 3

SAUDI -1.8 IT 6 -1.68 IT 8 0.83 IT 43 -17.85 IT 7 -57.46 IT

6 0.07 IT 17

JAPANI -2.96 IT 2 -3.01 IT 8 0.61 IT 45 -44.95 I 1 59.91 N 10 0.12 I 11

USAI 1.95 N 7 1.75 N 8 1.05 IT 45 -24.89 I 6 -59.24 I 7 0.26 I 27

UKI -1.38 I 8 0.91 N 8 1.18 IT 45 -21.82 N 7 -58.13 N

8 0.25 I 15

WTI -0.31 N 4 -0.45 N 6 0.53 IT 44 -22.31 N 7 -58.62 N

6 0.04 I 31

Nate: All variables are in natural logs. All unit root tests agree that all variables are I (1). The lag selection is based on the lowest values

for AIC criterion. Superscript N stands for no intercept and no trend. I for intercept only and no trend, and IT for both intercept and

trend. Significant statistics are in bold, and the series are stationary. BW stands for bandwidth.

The results of these three tests show that all variables appear to be non-stationary in

levels and stationary in the first differences or integrated of the first degree.

Here we present the cointegration results to test the long-run relationships at

different levels, and for several sets of VARs: VAR-10 which includes all Arab and US

stock markets, and designed to investigate the long-run cointegrating relation between

Arab and international stock markets, VAR-9 which includes all Arab stock markets, so

as to investigate the cointegrating relation between Arab stock markets themselves.

While, after dividing Arab markets into two subgroups; oil production countries which

include the GCC markets and non-oil countries, VAR-6 consisting of the six GCC stock

markets, whereas VAR-3 contains non-oil countries (Jordan, Egypt, and Palestine).

Several criterions (Akaike, Schwarz, and likelihood ratio) have been used to

determine the appropriate lag length for each VAR. To select the adequate deterministic

component for the cointegration and VAR equations for each VAR, we exclude

specification 1 of the deterministic component (that is, no intercept and no trend in either

180

the data or the cointegrating relations) for all VARs because according to Johansen, this

specification is rare and does not usually provide the minimum to account for the

presence of deterministic components. We also exclude specification 5 (quadratic trend

with intercept), because it is also a rare case and produces implausible forecasts out of

sample. Next, to choose a specification from the remaining specifications 2, 3, and 4, we

use the Schwarz criterion to determine the suitable specification. As table 5-2 shows, the

selected deterministic specification for VAR-3 is specification 2 (i.e. data have no

deterministic trend, but the cointegrating equation have intercept). This selection also

holds for VAR-6 and VAR-9, while for VAR-10, the selected deterministic component is

specification 4 (both data and the cointegrating equations have linear trend).8

Table 5-2

Number of Cointegrating Relations for Four VARs Models

Specifications VAR-3 VAR-6 VAR-10 VAR-9

Nonea 0 1 1 1

b

Intercept 0* 1* 1 1*

c

Linear trend 0 0 0 0

d

Linear trend 0 0 0* 0

e

Quadratic trend 0 0 0 0

No. of lags(levels) 3 3 2 2

Observations 1262 514 515 515

Notes: All variables are expressed in natural logarithms.

Astrisks indicate the selected deterministic specifications for each VAR .

a

Data have no deterministic trend, and the cointegrating equations do not have intercepts.

b

Data have no deterministic trend, but the cointegrating equations have intercepts.

c

Data have linear trend, but the cointegrating equations have intercepts only.

d

Both data and the cointegrating equations have linear trend.

e

Data has quadratic trends, but the cointegrating equations have linear trends.

VAR-3 :ESMI, PSEI and JSMI; VAR-6: MSMI, KSEI, DFMI, BSEI, ABSEI and SAUDI;

VAR10: MSMI, KSEI, DFMI, BSEI, ESMI, PSEI, ABSEI, SAUD, JSMI and USAI; VAR-9:

MSMI, KSEI, DFMI, BSEI, ESMI, PSEI, ABSEI, SAUD AND JSMI

Table 5-3 presents the results of Johansen-Juselius cointegration test, both trace

and maximum eigenvalue tests for each of the 4 VARs models. Based on these

specifications, table 5-3 suggests that VAR-6 and VAR-9 have one cointegrating relation,

8

The indices used for the each market are: Kuwait (KSEI); Jordan (JSMI); Bahrain (BSEI); Dubai (DFMI);

Egypt (EFMI); Oman (OSMI); Abu Dhabi (ABSMI); Palestine (PSEI); Saudi Arabia (SAUDI); US, S&P

500 (USAI).

181

while the results of Johansen-Juselius cointegration test; indicate the absence of any

cointegrating relation between Arab and international stock markets (VAR-10). The

results also indicate no cointegrating relation between the non-oil production countries

Jordan, Egypt and Palestine (VAR-3). The presence of one cointegrating relation among

Arab stock markets as a group (VAR-9), GCC stock markets; oil production countries;

(VAR-6), suggests long-run relationships among variables.

These results are constant with the existing literature; fore example, Kasa (1992)

finds one cointegrating relation among monthly stock indices of the United States, Japan,

United Kingdom, Germany and Canada. Francis and Leachman (1998) also find one

cointegrating relation for United States, Japan, United Kingdom, and Germany for the

same period. Furthermore, Bassler and Yang (2003) find only one cointegrating relation

in the nine largest countries in terms of market capitalization. Regarding the Middle East

region, Hammoudeh and Al-Eisa (2004) find two cointegrating relations between five

GCC countries, Darrat et al. (2000) find that cointegrating relation exists between three

Arabian equity markets, Jordan, Egypt, and Morocco, but these markets found to be

isolated from international markets. Maneschiold (2005) finds one cointegrating relation

between USA and Egypt at the general index level, related to the industry sub-index but

not to the financial or services sub-indices.

The results of cointegration test indicate that, there is no cointegrating relation

between Arab and international stock markets in the long-run. The purpose of this section

is to investigate the short-run relationship between Arab and international stock markets,

through analyzing how shocks generating by US, UK, and Japan markets, affect each of

the Arabian markets using SVAR technique. The S&P 500, FTSE100, and Nikkei 225 are

used to represent US, UK, and Japan stock markets, respectively. The data includes daily

observations for different time horizons.9

9

The daily data for international stock markets S&P 500, FTSE100, and Nikkei 225 are obtained from

Yahoo. Finance website on the net through: www.Yahoo.Fianace.com .

182

Table 5-3

Johansen-Juselius Cointegration Test Results

H0=Number of Trace Test Maximum Eigenvalue Test

Cointegrating

Statistics C.V (5%) C.V (1%) Statistics C.V (5%) C.V (1%)

Vectors

A. Cointegrating System: VAR-6b

None 120.536* 102.14 111.01 50.4* 40.30 46.82

At most 1 70.133 76.07 84.45 38.1** 34.40 39.79

At most 2 32.036 53.12 60.16 14.116 28.14 33.24

At most 3 17.920 34.91 41.07 9.966 22.00 26.81

At most 4 7.954 19.96 24.60 5.749 15.67 20.20

At most 5 2.205 9.24 12.97 2.205 9.24 12.97

B. Cointegrating System: VAR-9b

None 220.868* 202.92 215.74 63.52** 57.42 63.71

At most 1 157.348 165.58 177.20 50.252 52.00 57.95

At most 2 107.095 131.70 143.09 30.061 46.45 51.91

At most 3 77.035 102.14 111.01 27.777 40.30 46.82

At most 4 49.258 76.07 84.45 16.118 34.40 39.79

At most 5 33.140 53.12 60.16 15.149 28.14 33.24

At most 6 17.990 34.91 41.07 7.880 22.00 26.81

At most 7 10.111 19.96 24.60 7.228 15.67 20.20

At most 8 2.883 9.24 12.97 2.883 9.24 12.97

C. Cointegrating System: VAR-3b

None 34.841 34.91 41.07 20.950 22.00 26.81

At most 1 13.891 19.96 24.60 10.544 15.67 20.20

At most 2 3.347 9.24 12.97 3.347 9.24 12.97

D. Cointegrating System:VAR-10d

None 250.420 263.42 279.07 61.855 66.23 73.73

At most 1 188.565 222.21 234.41 48.548 61.29 67.88

At most 2 140.017 182.82 196.08 34.774 55.50 62.46

At most 3 105.243 146.76 158.49 26.371 49.42 54.71

At most 4 78.872 114.90 124.75 22.632 43.97 49.51

At most 5 56.240 87.31 96.58 15.941 37.52 42.36

At most 6 40.300 62.99 70.05 13.855 31.46 36.65

At most 7 26.445 42.44 48.45 11.403 25.54 30.34

At most 8 15.042 25.32 30.45 7.609 18.96 23.65

At most 9 7.433 12.25 16.26 7.433 12.25 16.26

Notes: All variables are expressed in natural logarithms.

**(*) denotes rejection of the hypothesis at the 5%(1%) level

b

Data have no deterministic trend, but the cointegrating equations have intercepts.

d

Both data and the cointegrating equations have linear trend.

VAR-3 :ESMI, PSEI and JSMI; VAR-6 : MSMI, KSEI, DFMI, BSEI, ABSEI and SAUDI; VAR10 : MSMI, KSEI, DFMI,

BSEI, ESMI, PSEI, ABSEI, SAUD, JSMI and USAI; VAR-9 : MSMI, KSEI, DFMI, BSEI, ESMI, PSEI, ABSEI, SAUD

AND JSMI

The critical values for the test statistics have been generated by Monte Carlo methods and tabulated by Osterwald-Lenum

(1992).

183

Moreover, it is assumed that US, UK, and Japan stock exchanges performance, is

not affected by Arab stock markets; however, Arab markets are affected by both its own

dynamics and the three international markets. This assumption is reflected in the

specification by using block recursive VAR model described in section 5-1-3.

The results are presented in appendix 3. figure 1 in appendix 3 reports the impulse

response functions for 20 days concerning, how do Arab markets returns respond to one-

structural standard deviation shock to each of the three international markets, whilst

tables 1 and 2 in appendix 3 show the variance decomposition and the impulse response

functions, respectively. In general, it is important to recognize that shocks originated in

international markets, have a marginal effect on Arab stock markets’ returns. More

specifically, table 1 provides the variance decomposition of the 2-, 6-, and 10 days a head

forecast errors for each Arab stock markets, accumulated for by innovations in US, UK,

and Japan. The results indicate that all markets are strongly exogenous in the sense that;

the percentage of the foreign explanatory power as indicated by US, UK, and Japan, is

very weak, reaching 3.5% in the best cases (the case of Egypt).

Nevertheless, one can find that UK exerts the most influence effect on Arab stock

markets, since it has the most foreign explanatory power on 4 Arab markets (Bahrain,

Saudi, Palestine, and Jordan), while US has the most explanatory power of returns

variation in three Markets (Oman, Kuwait, and Dubai). With Japan to be the least

influencing market, since it has an explanatory power only on Abu Dhabi and Egypt.

Moreover, it is apparent that a shock originated in UK has a persistent impact on Bahrain,

Saudi, Palestine, and Jordan with duration up to 8 days in average. Among these markets,

only Jordan responds positively to UK shock. However, Oman, Kuwait, and Dubai show

a memory of 5 days to absorb shock originated in US market, it seems that these markets

react quickly and relatively efficient. Since their response tapers-off and decline rapidly

with a positive reaction from Dubai only. Finally, Egypt and Abu Dhabi react positively

and quickly, to a shock generated by Japan stock market with duration up to 4 and 6 days

for Abu Dhabi and Egypt, respectively (see appendix 3, figure1 and table 2).

184

The presence of cointegrating relations in VAR-9 (all Arab stock markets) and

VAR-6 (GCC markets), suggests that causality among the variables in these systems

exists; in at least one direction. As a sequence, this section investigates the existing short-

run relationships among Arab stock markets.

Granger causality test has been conducted, to investigate the interdependence

among the 9 Arab markets. The results are included in table 5-4. A lag order of seven (i.e.

m=7) is used. It is expected that 7 lags (7 days) should be long enough to complete the

transmission process, as similar results are obtained for higher-order lags used. Table 5-4

shows that there are some causal relationships observed among the nine markets, with

Egypt exerting significant influence and leading Bahrain and Kuwait, while Bahrain leads

Saudi but not vice versa, Another Granger causal relation exists from Oman to Egypt and

from Jordan to Palestine.

Table 5-4

Granger Causality Test for Arab Stock Markets 1st July, 2001 to 24 July, 2003

Dependent variables

Bahrain Oman Kuwait Saudi Jordan Egypt Palestine Dubai AbuDhabi

Bahrain 1.79(0.08) 0.56(0.78) 2.58(0.01) 0.82(0.57) 1.36(0.22) 1.53(0.15) 1.67(0.11) 0.82(0.57)

Oman 1.82(0.08) 1.2(0.3) 1.02(0.42) 0.88(0.52) 2.17(0.03) 2.02(0.06) 0.47(0.85) 0.66(0.7)

Kuwait 0.62(0.74) 0.8(0.59) 0.63(0.73) 0.95(0.46) 0.87(0.52) 0.58(0.77) 1.65(0.12) 0.95(0.46)

Saudi 0.38(0.92) 0.55(0.79) 0.57(0.77) 1.35(0.22) 1.35(0.85) 1.26(0.26) 0.48(0.84) 1.14(0.34)

Jordan 1.38(0.21) 1.5(0.16) 0.41(0.90) 1.65(0.12) 1.37(0.21) 2.06(0.04) 0.54(0.8) 1.09(0.37)

Egypt 2.21(0.03) 1.11(0.35) 2.33(0.02) 1.1(0.36) 1.29(0.25) 0.49(0.84) 0.22(0.98) 0.27(0.97)

Palestine 1.88(0.07) 1.64(0.12) 0.71(0.66) 0.59(0.76) 0.73 (0.64) 0.74 (0.63) 0.52(0.82) 0.6 (75)

Dubai 0.44(0.87) 1.66(0.11) 1.83(0.07) 1.77(0.09) 0.3(0.95) 0.22(0.98) 0.76(0.62) 0.16(0.99)

AbuDhabi 0.71(0.66) 0.88(0.51) 1.28(0.25) 0.68(0.69) 1.28(0.26) 0.85(0.54) 0.71(0.66) 0.86(0.54)

The numbers report the F -statistics for testing the null hypothesis that all 7 lags of the left column do not Granger-cause the dependent

variable. P -values are between brackets.

indices is presented in table 5-5. The correlation between Arab stock markets’ indices,

are generally low and close to zero for most cases, indicating a low interdependence

between these indices.

185

Table 5-5

Correlation Coefficient Between Daily Arab Markets' Returns, 1st July, 2001 to 24

July, 2003

Saudi Oman Kuwait Dubai Bahrain AbuDhabi Egypt Jordan Palestine

Saudi 1.0000 0.0118 -0.0001 0.0371 0.0880 0.0302 0.0421 -0.1332 0.0827

Oman 1.0000 0.1041 -0.0124 0.0774 0.0853 0.1560 -0.0136 0.0384

Kuwait 1.0000 -0.0231 0.0785 0.0081 -0.0146 0.0028 -0.0042

Dubai 1.0000 -0.0078 0.0122 0.0432 -0.0313 -0.0446

Bahrain 1.0000 0.0109 0.0180 0.1302 -0.0654

AbuDhabi 1.0000 -0.0145 0.0233 0.0099

Egypt 1.0000 -0.0848 -0.1154

Jordan 1.0000 -0.0087

Palestine 1.0000

The results for long-run relationship (cointegration analysis), indicate the

existence of one long-run cointegrating relation in VAR-9 and VAR-6. The existence of

these long-term relationships indicates the subsistence of causality among variables. One

way that causality may emerge in a VAR is through the cointegrating equations or the

error-correction terms of the associated VEC model. Testing the statistical significance of

the adjustment coefficients of these terms, amounts to testing if the long-run relationships

derive the endogenous variables to convergence to equilibrium over time, such testing of

the adjustment coefficients is known as testing weak exogeneity of the endogenous

variables, with respect to the parameters of the cointegrating equations. Therefore, we

will first present and interpret the cointegrating equations of the VEC models.

Table 5-6 presents the cointegrating equations for both VAR-9 and VAR-6, and

indicates that Kuwait has; by far, the greatest relative influence on Arab and GCC

indices. Thus, in the absence of global factors, such as oil prices, the Kuwaiti market

dominates the long-run relation with other GCC and Arab markets, followed by Saudi

Arabia. These results could be surprised, since one expects that the Saudi market to be

the leader in the long run, since Saudi’s economy is the largest among the Arabian

countries and its stock market makes up about 50% of the total Arab markets

capitalization followed by Kuwaiti economy. The results may suggest that local factors

(such as cross-listing of Kuwaiti companies on both UAE and Bahraini stock markets) as

well as global factors (i.e. oil prices) also have a strong influence on the long-run

186

relationship that derive the variables in the system to equilibrium over time. Figure 5-1

shows the cointegrating relations for the two systems.

Table 5-6

Cointegrating Equations of the VEC Models for VAR-9 and VAR-6, 1/8/2001-

31/12/2004

Model CE MSMI KSEI DFMI BSEI ESMI PSEI ABSEI SAUDI JSMI C

a a

VEC-9 CE1 1.000 -2.968 1.1226 0.504 -0.239 0.069 1.431 2.168 1.769 -29.117b

b

VEC-6 CE1 1.000 -1.083a 0.108 0.222 - - 0.301 0.768a - -9.621b

(0.225) (0.296) (0.661) - - (0.333) (0.248) - (4.761)

Notes: CE stands for a cointegrating equation. VEC-9 is the VEC model for the VAR-9 and VEC-6 is the VEC model for the VAR-6. a b

and c represent statistical significance at 1%, 5%, and 10% respectively.

Figure 5-1

.4 .2

.2

.1

.0

.0

-.2

-.1

-.4

-.2

-.6

-.8 -.3

100 200 300 400 500 600 700 800 100 200 300 400 500 600 700 800

Next, before presenting the estimated VEC models for the two VARs and tests for

weak exogeneity, we should test whether the variable of the VARs inter the cointegrating

equations significantly. Table 5-7 shows that only Kuwait and Saudi inter the

cointegrating relation significantly in VAR-9 while for VAR-6; in addition to Kuwait and

Saudi, Oman inters the equation significantly. This means that there are common factors

that make Kuwaiti, Saudi, and Omani indices, as a group form one long-relationship.

Furthermore, the finding of VAR-9 VEC model is presented in table 5-8. Broadly, the

short-term relationships between Arab markets found to be weak. Since on a daily basis,

there are two-way directional relation between Saudi and Jordan, one-way relation from

Oman to Bahrain, from Egypt to Oman, and from Abu Dhabi to Saudi.

187

Table 5-7

Significant of Zero Restrictions on Coefficients of Cointegrating

Equations of the VEC Models of VAR-9 and VAR-6

VAR-9 VAR-6

Market

Chi-Square Probability Chi-Square Probability

MSMI 0.587 0.444 5.167 0.023

KSEI 10.599 0.001 5.932 0.015

DFMI 0.749 0.387 0.057 0.811

BSEI 0.027 0.869 0.084 0.772

ESMI 0.045 0.832 - -

PSEI 0.013 0.908 - -

ABSEI 1.529 0.216 0.420 0.517

SAUDI 5.268 0.022 3.083 0.079

JSMI 2.549 0.110 - -

Notes: Bolds numbers indicate that the LR tests reject the null hypothesis that the ith endogenous

variable does not enter the cointegrating equation significantly.which means that these variables form

long run equilibrium relationships.

Table 5-8

VEC Model for 9 Arabian Indices in the VAR-9, 1/8/2001- 31/12/2004

Model D(MSMI) D(KSEI) D(DFMI) D(BSEI) D(ESMI) D(PSEI) D(ABSEI) D(SAUDI) D(JSMI)

ECT1 -0.003a -0.004b -0.005a 0.001 -0.006b -0.006c 0.001 -0.008a -0.004a

D(MSMI(-1)) 0.438a -0.029 -0.074 0.081c -0.068 -0.073 -0.020 -0.122 0.101

a

D(KSEI(-1)) -0.010 0.161 -0.048 0.023 0.007 -0.092 0.033 -0.020 -0.057c

D(DFMI(-1)) 0.020 0.006 -0.002 0.024 -0.045 0.077 0.016 -0.056 -0.031

D(BSEI(-1)) -0.014 -0.111 -0.047 0.094b -0.219 -0.114 0.015 0.035 -0.006

a a

D(ESMI(-1)) -0.036 0.009 -0.029 0.015 0.167 0.034 0.004 0.043 0.003

a

D(PSEI(-1)) 0.010 -0.030 0.000 -0.004 -0.013 0.307 -0.010 0.032 -0.010

D(ABSEI(-1)) 0.001 0.016 -0.106 -0.021 0.154 0.001 0.095b 0.166c -0.006

c

D(SAUDI(-1)) 0.012 0.012 0.016 0.030 -0.008 -0.072 -0.012 -0.083 0.08b

D(JSMI(-1)) -0.014 -0.039 -0.027 0.006 0.071 -0.029 -0.039 -0.129b 0.17a

Stat. P.value

Log Likelihood 15495.7

Akaike Information Criteria -59.61

Schwarz Criteria -58.79

Serial Crrelation LM Stat.(up to lag 4 ) 96.73 0.112

Skewness 355.93 0.000

Kurtosis 2700.48 0.000

Normality 3056.42 0.000

White test 1597.02 0.000

Notes: ECT stands for the error-correction terms in the VEC equations. The number of lags is based on the Schwars criteria. The statistics for

skewness, kurtosis, normality, and the White test are chi-squares testing the null hypothesis. The statistics for the normality tests base are

a.b

based on Doornik-Hansen orthogonalization method. All variables are first differences of logs. and c represent statistical significance at

the 1%, 5% and 10% levels respectively.

188

Another interesting finding from the VEC-9 model is that, the own short-run

adjustment term for Saudi is negative. This result may suggest that it takes a while for the

Saudi market to cool off after getting heated by hot hands of its momentum traders.

Moreover, the estimates show that Dubai index lacks linkages with other Arab markets

indices, even with its own lags. The estimated VEC model for VAR-6 (GCC markets) is

presented in table 5-9. The results indicate that Kuwaiti market can be considered as a

leader among GCC markets. Since it can predict both Bahrain and Abu Dhabi markets,

the results also indicate that two-way directional relationship exists between Kuwait and

Bahrain, Abu Dhabi and Dubai markets. Whereas, Oman and Saudi markets have the

weakest links with other GCC markets on the short-run relationship. Moreover, the Wald

test for the adjustment coefficients of the error-correction terms, which measures

deviations from the long-run equilibrium indicating that, for VEC-9, five among the nine

markets including in the model (Kuwait, Bahrain, Egypt, Palestine, and Abu Dhabi) are

weakly exogenous (see table 5-10).

In other words, it appears that Oman, Saudi, and Dubai equations contain all the

long-run information, since these are the only equations that their equilibrium adjustment

parameters inter the system significantly according to Wald test. While for the VEC-6

model, the results of Wald test indicate that Kuwait, Bahrain, and Abu Dhabi were

weakly exogenous, implying that these markets do not have the tendency to restore

equilibrium and take the brunt of the shocks to the system, whereas the Saudi market

contains the long-run information since its equation contains the large significant

equilibrium adjustment parameter.

The results for the interrelation between GCC stock markets in the short run

(VEC-6), are incompatible with those of Assaf (2003), since he finds; using VAR

analysis for weekly data, that Bahrain plays a dominant role in influencing the GCC

markets. While the results for GCC stock markets indicate that Kuwait can be considered

as a dominant market that influences the GCC markets in the short-run horizon.

189

Table 5-9

VEC Model for 6 GCC Indices in the VAR-6, 1/8/2001- 31/12/2004

Model D(MSMI) D(KSEI) D(DFMI) D(BSEI) D(ABSEI) D(SAUDI)

ECT1 -0.008a -0.008c -0.012a 0.003 0.002 -0.017a

D(MSMI(-1)) 0.447a -0.030 -0.026 0.055 -0.007 -0.085

D(MSMI(-2)) -0.020 -0.011 -0.100 0.058 -0.042 -0.082

D(KSEI(-1)) -0.015 0.179a -0.043 0.036 0.024 -0.001

D(KSEI(-2)) -0.004 -0.032 -0.044 -0.052b 0.046b -0.042

D(DFMI(-1)) 0.020 0.003 -0.024 0.023 0.013 -0.038

D(DFMI(-2)) -0.029 0.068 -0.048 0.021 0.052c -0.021

D(BSEI(-1)) -0.006 -0.128 -0.040 0.095b 0.003 0.038

D(BSEI(-2)) 0.024 0.152c 0.021 -0.013 -0.044 0.041

D(ABSEI(-1)) 0.003 0.010 -0.075 -0.015 0.098b 0.143

D(ABSEI(-2)) -0.010 0.046 -0.228a -0.004 -0.059 0.031

D(SAUDI(-1)) 0.011 0.018 0.027 0.031 -0.007 -0.067

D(SAUDI(-2)) -0.002 0.039 0.052 -0.023 0.043b -0.029

Stat. P.value

Log Likelihood 10945.88

Akaike Information Criteria -42.107

Schwarz Criteria -41.405

Serial Crrelation LM Stat.(up to lag 5 ) 48.143 0.085

Skewness 262.910 0.000

Kurtosis 948.193 0.000

Normality 1211.104 0.000

White test 797.898 0.000

Notes: ECT stands for the error-correction terms in the VEC equations. The number of lags is based on the Schwars criteria.

The statistics for skewness, kurtosis, normality, and the White test are chi-squares testing the null hypothesis. The statistics

for the normality tests base are based on Doornik-Hansen orthogonalization method. All variables are first differences of logs.

a.b

and c represent statistical significance at the 1%, 5% and 10% levels respectively.

Table 5-10

Weak Exogeneity Tests of the Endogenous Variables in the

VEC Models of VAR-9 and VAR-6, 1/8/2001-31/12/2004

VEC-9 VEC-6

Market

Chi-Square Probability Chi-Square Probability

MSMI 10.526 0.001 12.246 0.000

KSEI 1.966 0.161 1.161 0.281

DFMI 9.444 0.002 7.806 0.005

BSEI 1.530 0.216 1.558 0.212

ESMI 1.925 0.165 - -

PSEI 2.339 0.126 - -

ABSEI 0.005 0.946 0.247 0.619

SAUDI 12.561 0.000 8.534 0.003

JSMI 3.576 0.059 - -

Notes: Bolds numbers indicate that the LR tests do not reject the null hypothesis that the i th

endogenous variable is weakly exogenous with respect to the β parameters, implying that it does

not adjust to restore equilibrium after a shock hits the system.

190

5-3-5 Dynamic relationship between GCC stock markets and oil prices

The last two years have witnessed a sharp increase in oil prices as a result of the

increase in the international demand, among other things. It is known that this increase

negatively affects the economic development of industrial countries in particular and that

of other countries in general. This, in turn, leads to raising the inflation rate and

unemployment. Economic sectors in many countries have been influenced, especially

stock markets. Gulf Cooperation Council countries (GCC) are among the most important

oil producing countries, except Bahrain and Oman, the other four of these (Saudi Arabia,

Qatar, Kuwait and UAE) are members in the Organization of Petroleum Exporting

Countries (OPEC).

At the end of 2003, these countries collectively accounted for about 21% of the

world’s 68 million barrels a day of total production. They possess 43% of the world’s

1105.26 billion barrels of oil proven reserves10. Producing and exporting oil plays a

crucial role in determining foreign earnings and government’s budget revenues and

expenditures for those countries; thus they are the primary determinant of aggregate

demand, which in turn affects the domestic price levels as well as all aspects of daily

economic life. In addition, increase in oil prices causes increase in the trading volume in

the stock markets according to cash surplus. This shows the importance of studying the

relation between increase in oil prices and stock markets in GCC countries where oil

price has reached $ 70 a barrel, while it was $ 29.24 on 27 May 2003.

There are several studies and research articles which investigate the relation

between international financial markets and others examined the link between spot and

future petroleum prices. However, few studies have looked into the relation between oil

spot/future prices and stock markets. Such studies concentrated on countries such as

Canada, Germany, Japan, UK, and USA. The overall literature on the links between oil

markets and financial markets is very limited; Johnes and Kaul (1996) investigate the

relation of the U.S., Canadian, Japanese, and U.K stock prices to oil price shocks using

quarterly data. Utilizing a standard cash-flow dividend valuation model, they find that for

the United States and Canada, this relation can be accounted for entirely by the impact of

10

See the Uniform Arabian Economic Report 2004, Arab Monetary Fund (AMF), Abu Dhabi.

191

oil shocks or real cash flows. The results for Japan and the United Kingdom were not as

strong.

Moreover, Huang et al. (1996) use an unrestricted vector auto regression (VAR)

model to examine the relationship between daily oil future returns and daily U.S stock

returns. They find that the oil futures returns lead some individual oil company stock

returns, but they do not have much impact on broad-based market indices such as the

S&P 500. Sadorsky (1999), using monthly data examines the links between the U.S. fuel

oil prices and the S&P 500 in an unrestricted VAR model that also include the short-term

interest rate and industrial production. He finds that oil prices movements are important

in explaining movements in broad-based stock returns. Papapetrou (2001), employing an

error correction representation of a VAR macroeconomic model and using monthly data

for Greece, concludes that oil prices are important in explaining stock price movements.

Hammoudeh and Aleisa (2002) examine the links between oil-exporting countries,

including Bahrain, Indonesia, Mexico, and Venezuela, using monthly data, and find

spillovers from the oil markets to the stock indices of these countries.

However, one study (Hammoudeh and Al-Eisa, 2004) examines the relation

between oil prices and stock markets in GCC countries. For daily data during the period

1994-2001, they find that the Saudi market is the leader among GCC markets and can

predict-and be predicted by oil future prices.

The purpose of this section is firstly, to investigate the dynamic relationship

between oil prices and GCC stock markets both on the long and short-run, second, to

investigate how oil prices affect returns volatility in GCC markets, finally and most

important, this section will investigate the impact of the increase in oil prices on GCC

stock markets, and trying to identify the nature of the dynamic relationship between oil

prices and GCC markets.

In order to investigate the effect of oil prices on the volatility of returns in GCC

markets, GARCH (1,1) will be estimated, while oil prices will be added as an additional

regressor in the conditional variance equation, such as:

192

Rt = β 0 + β1 Rt −1 + ε t (5-

25)

εt ∼ N(0,ht)

where Rt is log (Pt/Pt-1); and Pt is the stock price at time t. Equation (5-26) models the

variance of the unexpected returns, εt, as GARCH (1,1) process; and oil is oil return

log(Pt/Pt-1). While for oil spot prices, we use the spot oil prices (WTI), which is the crude

stream produced in Texas and South Oklahoma that is traded in domestic spot market at

the Cushing Center.

Table 5-11 shows the results of the estimated model, while table 4 in appendix 2

presents the results of diagnostic tools for the standardized residuals of GARCH (1,1)

model with oil prices. The coefficient of oil returns presented in table 5-11; w, is found to

be highly significant for all GCC stock markets except Kuwait.

Table 5-11

GARCH (1,1) Model for GCC Daily Returns with Oil Returns as a

Regressor in the Variance Equation

Market Obs. α0 α1 β1 α 1 +β 1 ω

AbuDhabi 605 0.000 0.205 0.631 0.836 0.000

1/7/01-31/12/03 0.000 0.000 0.000 0.000 0.001

1/1/91-3/6/04 0.000 0.001 0.000 0.000 0.000

26/3/00-31/12/03 0.000 0.063 0.000 0.007 0.000

17/6/01-9/3/05 0.001 0.000 0.000 0.552 0.265

1/2/97-13/10/04 0.000 0.000 0.000 0.000 0.000

26/1/94-14/3/05 0.000 0.000 0.000 0.000 0.007

Significance levels are in italics. A Chi-square (χ2) tests (α 1 +β 1 ) = 1. The estimated variance

equation is : h = a + a ε 2 + β h + ω oilr

t 0 1 t −1 1 t −1

193

Note that, volatility persistence for all markets does not change with the addition

of oil returns (see table 4-12 in section 4-5-2). The results indicate that oil prices play a

significant role in affecting GCC markets volatility but not Kuwait, since it appears that

for Kuwaiti stock market, other factors than oil prices affect its volatility.

5-3-5-2 Long-run relationship among GCC stock markets and oil prices

The GCC countries are oil-dependent as well as economically and politically

similar. The forces that commove their stock markets are basically the forces that move

the oil prices. As a result, one would expect cointegrating relations among the markets of

countries that have had a cooperation council since 1981, and are now aiming to establish

a monetary union with a single currency in 2010. Table 5-12 presents the results of

Johansen cointegration test for 6 GCC indices with oil prices. The result of the trace test

indicates the existence of one long-run cointegrating relation between the variables.

While cointegrating equation is presented in table 5-13.

Table 5-12

Johansen-Juselius Cointegration Test Results

H0=Number of Trace Test Maximum Eigenvalue Test

Cointegrating

Statistics C.V (5%) C.V (1%) Statistics C.V (5%) C.V (1%)

Vectors

None 132.895** 131.70 143.09 44.279 46.45 51.91

At most 1 88.616 102.14 111.01 40.625** 40.30 46.82

At most 2 47.991 76.07 84.45 21.510 34.40 39.79

At most 3 26.481 53.12 60.16 13.977 28.14 33.24

At most 4 12.504 34.91 41.07 8.394 22.00 26.81

At most 5 4.110 19.96 24.60 3.581 15.67 20.20

At most 6 0.529 9.24 12.97 0.529 9.24 12.97

Notes: All variables are expressed in natural logarithms, where 3 lags have been used to estimate the VAR.

**(*) denotes rejection of the hypothesis at the 5%(1%) level

Data have no deterministic trend, but the cointegrating equations have intercepts.

VAR-7 : MSMI, KSEI, DFMI, BSEI, ABSEI, SAUDI and OILPI.

Table 5-13

Cointegrating Equations of the VEC Model for VAR-7, 1/8/2001-31/12/2004

Model CE MSMI KSEI DFMI BSEI ABSEI SAUDI C OILPI

VEC-7 CE1 1.000 -2.812b 2.327 0.121 -0.211 1.466 -16.40 2.554b

(1.138) (1.426) (3.228) (1.618) (1.248) (-21.996) (1.009)

ab c

Notes: CE stands for a cointegrating equation. VEC-7 is the VEC model for the VAR-7. and represent statistical

significance at 1%, 5%, and 10% respectively.

194

The oil price index found to have significant influence on the long-run

equilibrium after Kuwait stock market, which dominate the long-run relation among GCC

and oil prices.

The findings of the VAR-7 VEC model (table 5-14); suggest that 4 out of 6 GCC

markets (Oman, Kuwait, Bahrain, and Saudi) can predict and explain the future oil prices

movements in the short-run, while oil prices can explain both Saudi and Dubai indices

only. The existing two-way directional relation between oil prices and Saudi market;

could be explained by the fact that Saudi Arabia is the largest oil exporter and has the

largest oil reserves.

Table 5-14

VEC Model for 6 GCC and Oil Price Indices in the VAR-7, 1/8/2001- 31/12/2004

Model D(MSMI) D(KSEI) D(DFMI) D(BSEI) D(ABSEI) D(SAUDI) D(OILPI)

ECT1 -0.001 -0.004a -0.003a 0.000 -0.001 -0.004a -0.005b

D(MSMI(-1)) 0.451a 0.003 -0.024 0.036 0.005 -0.054 0.149

D(MSMI(-2)) -0.027 -0.102 -0.098 0.058 -0.041 -0.050 -0.511b

D(KSEI(-1)) -0.012 0.119b -0.038 0.033 0.011 -0.007 0.033

D(KSEI(-2)) 0.009 -0.044 -0.048 -0.057b 0.034 -0.062 0.289a

D(DFMI(-1)) 0.027 -0.009 -0.011 0.026 0.005 -0.021 0.056

D(DFMI(-2)) -0.031 0.057 -0.040 0.011 0.056c -0.024 0.103

D(BSEI(-1)) -0.003 -0.169c -0.060 0.097b 0.008 -0.010 0.413b

D(BSEI(-2)) 0.024 0.221b 0.004 -0.008 -0.042 0.021 0.325

D(ABSEI(-1)) 0.004 0.058 -0.134c -0.049 0.089c 0.090 -0.035

D(ABSEI(-2)) -0.059 0.009 -0.267a -0.008 -0.023 0.003 0.015

D(SAUDI(-1)) 0.013 -0.006 0.026 0.026 -0.006 -0.061 0.069

D(SAUDI(-2)) 0.010 0.033 0.032 -0.030 0.048b -0.037 -0.175c

D(OILPI(-1)) 0.006 0.016 0.022 -0.008 0.003 0.010 -0.123a

D(OILPI(-2)) 0.005 -0.028 0.046a 0.000 0.011 -0.037c 0.010

Stat. P.value

Log Likelihood 10748.54

Akaike Information Criteria -46.617

Schwarz Criteria -45.592

Serial Crrelation LM Stat.(up to lag 7 ) 57.430 0.191

Skewness 258.042 0.000

Kurtosis 855.838 0.000

Normality 1113.880 0.000

White test 1240.175 0.000

Notes: ECT stands for the error-correction terms in the VEC equations. The number of lags is based on the Schwars criteria. The statistics

for skewness, kurtosis, normality, and the White test are chi-squares testing the null hypothesis. The statistics for the normality tests base

are based on Doornik-Hansen orthogonalization method. All variables are first differences of logs. a.b and c represent statistical

significance at the 1%, 5% and 10% levels respectively.

195

Moreover, the Wald test for the adjustment coefficients of the error-correction

term, is presented in table 5-15 and indicates that all variables in the system are weakly

exogenous except oil prices and Dubai index. It appears that, oil and Dubai indices

contain the information of the long-run equilibrium in the system.

Table 5-15

Weak Exogeneity Tests of the Endogenous Variables

in the VEC Model of 1/8/2001-31/12/2004

VEC-7

Market

Chi-Square Probability

MSMI 0.659 0.417

KSEI 2.173 0.140

DFMI 3.632 0.057

BSEI 0.005 0.944

ABSEI 0.106 0.745

SAUDI 1.847 0.174

OILP 2.882 0.090

Notes: Bolds numbers indicate that the LR tests do not reject the null hypothesis that

the i th endogenous variable is weakly exogenous with respect to the β parameters,

implying that it does not adjust to restore equilibrium after a shock hits the sysytem.

These results are inconsistent with those obtained by Hammoudeh and Al-Eisa

(2004), since they found two equilibrium relations between GCC stock markets and oil

future prices, while Saudi market found to be the leader followed by Bahraini and United

Arab Emirates (UAE).

This section will investigate the direct effect of the increasing oil prices on GCC

stock markets, during the period which witnessed unprecedented sharp raise, especially

through the last two years. Daily data for five GCC (Bahrain, Kuwait, Oman, Saudi

Arabia, and Abu Dhabi) and oil prices will be used, the data runs from 25 May 2001 to

24 May 200511. The WTI described in section 5-3-4-1 will be used as a proxy for oil

prices.

To facilitate the investigation of how the raise in oil prices affects GCC stock

markets and the dynamic relation between them, the whole period has been divided into

11

Dubai stock market was excluded according to the shortage of data for the later period.

196

two sub-periods (event study), and a vector autoregression (VAR) system will be

estimated for each period. The first period spanning from 25 May 2001 to 23 May 2003,

while the second one from 27 May 2003 to 24 May 2005, which includes the stunning

rise in oil prices. Since at the end of the second period, the oil prices reached US$ 49.14

per barrel compared to 29.24 on 23 May 2003.

Table 5-16 presents the estimation of the VAR system for five GCC stock

markets returns and oil return for the first sub-period. The results indicate that there is no

any relationship between oil prices and the five GCC stock markets on a daily basis. The

oil prices can not predict- or be predicted by any of the five GCC markets12. However,

during the second period; and after the rise in oil prices, the results changed dramatically.

Table 5-17, which shows VAR estimation for the second period, indicates that oil prices

can predict all GCC stock markets but not Abu Dhabi. While oil prices can be predicted

by both Saudi and Omani stock markets.

The results of the VAR system for the second sub-period reflect the significant

role that the sharp increase in oil prices plays, which has, in turn, brought about

enhancing the predictive power of oil prices on those markets in comparison with the first

sub-period that preceded the sharp increase in oil prices. The results should not seem

strange if we take into consideration that these countries essentially depends; in varying

degree, on oil, and that one of them is Saudi Arabia, the world’s biggest oil exporting

country with its largest oil reserve in the world.

- Variance decomposition

The variance decomposition analysis measures the percentage of the forecast error

of a market return that is explained by another market or oil return. It indicates the

relative impact that one market has upon another market and oil return within the VAR

system. The variance decomposition enables us to assess the economic significance of

this impact as a percentage of the forecast error for a variable sum to one. The

orthogonolization procedure of the VAR system decomposes the forecast error variance,

12

These results contradict those obtained from cointegration analysis during the period 1 August 2001- 31

December 2004 presented in table (3-14), since the finding of VEC-7 model indicates that in the short run,

3 out of 6 GCC markets have predictive power on oil prices, while both Dubai and Saudi markets can be

predicted by oil prices.

197

the component that measures the fraction in stock return of a particular market explained

by innovations in each of the six indices.

Table 5-18 provides the variance decomposition of the 3-, 6-, 9 days ahead

forecast errors of each index, accumulated for by innovations in each of the six indices

for the first sub-period. The results indicate that all markets and oil returns are strongly

exogenous in the sense that the percentage of the error variance accounted for by their

innovations around 98%. The percentage of the foreign explanatory power, as indicated

by the foreign column, is weak, reaching in the best cases 3%.

Table 5-16

VAR System for GCC Stock Markets and Oil Returns for the First Sub-

Period 25May, 2001 to 23May, 2003

BAHRAIN OIL OMAN KUWAIT ABUDHABI SAUDI

BAHRAIN(-1) 0.1162 -0.1418 -0.0022 -0.2264 -0.0258 0.0154

0.011 0.5545 0.9672 0.0132 0.563 0.8312

BAHRAIN(-2) 0.0658 -0.0199 0.0358 0.131 -0.0628 0.0271

0.1488 0.9338 0.4962 0.1509 0.1592 0.707

OIL(-1) -0.014 -0.0473 0.012 0.0013 -0.0038 -0.0039

0.1088 0.3032 0.2344 0.9404 0.6548 0.7789

OIL(-2) 0.0046 -0.0305 0.0052 0.0046 -0.0006 -0.002

0.5974 0.5077 0.6069 0.7914 0.9467 0.8828

OMAN(-1) -0.004 -0.1014 0.4111 0.0851 -0.0551 -0.0118

0.9183 0.6255 0.000 0.2811 0.1543 0.8497

OMAN(-2) 0.0103 0.3084 -0.0172 -0.0329 0.0206 0.0017

0.7945 0.1378 0.7056 0.6761 0.5928 0.9782

KUWAIT(-1) 0.0009 -0.0439 -0.0067 0.1802 0.0123 -0.0321

0.968 0.7159 0.7991 0.0001 0.5847 0.3772

KUWAIT(-2) 0.0025 0.0423 -0.0144 -0.0493 0.035 0.0475

0.9113 0.7246 0.585 0.2797 0.1169 0.189

ABUDHABI(-1) -0.0824 0.2067 -0.0341 -0.0221 0.0633 0.0242

0.0761 0.3981 0.5256 0.8122 0.1638 0.7425

ABUDHABI(-2) 0.0173 -0.1866 -0.1019 0.048 -0.0789 0.113

0.7104 0.4464 0.0587 0.6057 0.0834 0.1257

SAUDI(-1) 0.0165 0.013 0.0187 -0.0273 0.0081 0.0722

0.5657 0.9317 0.5754 0.6353 0.7738 0.1145

SAUDI(-2) -0.0049 -0.0065 -0.0413 0.0475 0.0378 -0.0399

0.866 0.9658 0.2149 0.4096 0.18 0.3825

C -0.0077 0.0025 0.0069 0.1689 0.0568 0.0699

0.7654 0.9852 0.8161 0.0011 0.0244 0.087

stat. P-value

PTA (12) 350.22 0.634

LM test (12) 21.61 0.972

Notes: Numbers in italic represent P-values. VAR system has been estimated with 2 lags according to Akaike

information selection criterion, C represents a constant in the VAR system. PTA represents Residual Portmanteau

Tests for Autocorrelations up to lag 12. LM represents Residual Serial Correlation LM Tests up to lag 12.

198

Table 5-17

VAR System for GCC Stock Markets and Oil Returns for the Second

Sub-Period 27May, 2003 to 24May, 2005

BAHRAIN OIL OMAN KUWAIT ABUDHABI SAUDI

BAHRAIN(-1) 0.262 0.204 -0.014 0.151 -0.106 -0.047

0.000 0.420 0.761 0.070 0.638 0.742

BAHRAIN(-2) 0.021 -0.406 -0.053 -0.048 0.222 0.225

0.654 0.121 0.271 0.573 0.342 0.129

BAHRAIN(-3) 0.061 0.240 -0.009 -0.017 -0.086 0.026

0.185 0.345 0.850 0.839 0.703 0.854

OIL(-1) 0.016 -0.074 0.003 0.001 -0.042 0.047

0.046 0.106 0.699 0.947 0.297 0.070

OIL(-2) 0.000 -0.029 0.014 0.021 -0.014 0.007

0.961 0.533 0.090 0.170 0.736 0.792

OIL(-3) -0.002 0.017 0.001 -0.032 0.045 -0.059

0.761 0.713 0.900 0.031 0.266 0.022

OMAN(-1) 0.070 0.135 0.389 -0.076 -0.240 -0.107

0.121 0.588 0.000 0.347 0.279 0.448

OMAN(-2) -0.092 -0.473 -0.003 -0.024 -0.046 0.204

0.056 0.075 0.946 0.778 0.846 0.175

OMAN(-3) 0.077 0.506 0.097 0.042 -0.054 -0.040

0.085 0.042 0.036 0.604 0.808 0.775

KUWAIT(-1) 0.011 0.015 -0.014 0.228 -0.050 -0.010

0.651 0.915 0.601 0.000 0.688 0.900

KUWAIT(-2) 0.011 0.169 0.060 -0.062 0.204 0.005

0.659 0.237 0.024 0.188 0.110 0.950

KUWAIT(-3) -0.039 -0.100 -0.019 0.060 0.075 0.001

0.121 0.474 0.456 0.194 0.548 0.987

ABUDHABI(-1) -0.007 -0.037 0.012 -0.019 0.355 0.007

0.477 0.473 0.214 0.251 0.000 0.798

ABUDHABI(-2) 0.005 0.047 -0.013 0.007 -0.192 0.035

0.577 0.371 0.201 0.694 0.000 0.241

ABUDHABI(-3) -0.004 -0.018 -0.005 0.005 -0.156 -0.016

0.652 0.729 0.568 0.786 0.001 0.569

SAUDI(-1) 0.006 0.164 0.004 -0.010 0.072 -0.077

0.687 0.044 0.786 0.710 0.318 0.094

SAUDI(-2) -0.016 0.017 -0.007 -0.005 0.031 -0.068

0.269 0.835 0.621 0.863 0.665 0.138

SAUDI(-3) 0.003 -0.131 -0.003 -0.024 -0.057 0.010

0.829 0.106 0.831 0.363 0.428 0.826

C 0.038 0.062 0.066 0.164 0.194 0.203

0.086 0.613 0.004 0.000 0.075 0.003

stat. P-value

PTA (12) 347.53 0.177

LM test (12) 38.1 0.374

Notes: Numbers in italic represent P-values. VAR system has been estimated with 3 lags according to Akaike

information selection criterion, C represents a constant in the VAR system. PTA represents Residual Portmanteau

Tests for Autocorrelations up to lag 12. LM represents Residual Serial Correlation LM Tests up to lag 12.

199

Table 5-18

Variance Decomposition for the Forecast Error of Daily Market Returns for GCC

Markets and Oil Return During the First Sub-Period

By innovations in

Market Horizon All

explained (days) BAHRAIN OIL OMAN KUWAIT ABUDHABI SAUDI foreign*

BAHRAIN 3 98.80 0.55 0.04 0.00 0.51 0.09 1.20

6 98.79 0.55 0.04 0.00 0.51 0.10 1.21

9 98.79 0.55 0.04 0.00 0.51 0.10 1.21

OIL 3 0.13 97.12 0.96 0.29 1.13 0.37 2.88

6 0.14 97.04 1.02 0.30 1.13 0.38 2.96

9 0.14 97.04 1.02 0.30 1.13 0.38 2.96

OMAN 3 0.10 0.39 98.19 0.14 0.96 0.22 1.81

6 0.17 0.41 97.77 0.19 1.16 0.29 2.23

9 0.17 0.41 97.77 0.19 1.17 0.29 2.23

KUWAIT 3 1.31 0.06 1.18 97.24 0.08 0.12 2.76

6 1.32 0.06 1.19 97.22 0.09 0.13 2.78

9 1.32 0.06 1.19 97.22 0.09 0.13 2.78

ABUDHABI 3 0.74 0.04 0.43 0.71 97.69 0.37 2.31

6 0.79 0.04 0.44 0.73 97.62 0.38 2.38

9 0.79 0.04 0.44 0.73 97.62 0.38 2.38

SAUDI 3 0.07 0.02 0.26 1.20 0.69 97.75 2.25

6 0.08 0.02 0.26 1.25 0.70 97.69 2.31

9 0.08 0.02 0.26 1.25 0.70 97.69 2.31

Entries in each cell are the percentage of forecast error variance of the market return in the first column explained by the market in the first

row

* Entries in the 'All foreign' column denote that the total percentage of forecast error variance of the market in the first column explained by

all foreign markets

Cholesky Ordering: BAHRAIN OMAN KUWAIT ABUDHABI SAUDI OIL

Standard Errors: Monte Carlo (1000 repetitions)

Table 5-19 presents the variance decomposition for the second sub-period. After

the sharp rise in oil prices, one can find that in general all variables in the system still

exogenous. The percentage of the foreign explanatory power is not very strong; it does

not exceed 5%; though the degree of influence differs across returns. Table 5-19 shows

that oil returns influences Saudi market and account for 46% of the variance in the Saudi

market explained by foreigners (1.88% out of 4.12% for 6 days horizon). The reverse is

right since the Saudi market accounts for most of the variance in the oil returns explained

by foreigners for 6 and 9 days horizon. In addition, Abu Dhabi market accounts for half

of the variance of Kuwaiti market explained by foreign markets. The results can not help

us to determine which the dominant market is in the system that influences all the others

and links their interdependence.

200

Table 5-19

Variance Decomposition for the Forecast Error of Daily Market Returns for GCC

Markets and Oil Return During the Second Sub-Period

By innovations in

Market Horizon All

explained (days) BAHRAIN OIL OMAN KUWAIT ABUDHABI SAUDI foreign*

BAHRAIN 3 98.17 0.78 0.67 0.13 0.07 0.18 1.83

6 97.54 0.78 0.98 0.44 0.09 0.18 2.46

9 97.49 0.78 1.02 0.44 0.09 0.18 2.51

OIL 3 0.57 97.19 0.66 0.30 0.45 0.82 2.81

6 0.77 95.72 1.08 0.42 0.45 1.57 4.28

9 0.78 95.70 1.09 0.42 0.45 1.57 4.30

OMAN 3 0.50 0.56 97.86 0.76 0.29 0.03 2.14

6 0.62 0.63 97.16 0.83 0.70 0.05 2.84

9 0.64 0.63 97.13 0.83 0.71 0.05 2.87

KUWAIT 3 1.19 0.47 0.39 97.65 0.27 0.04 2.35

6 1.18 1.31 0.39 96.58 0.35 0.20 3.42

9 1.18 1.31 0.39 96.56 0.35 0.20 3.44

ABUDHABI 3 0.18 0.26 0.39 0.91 98.01 0.24 1.99

6 0.20 0.50 0.46 1.43 97.03 0.37 2.97

9 0.20 0.52 0.46 1.48 96.96 0.37 3.04

SAUDI 3 0.61 0.66 1.08 0.09 0.31 97.25 2.75

6 0.65 1.88 1.11 0.11 0.37 95.88 4.12

9 0.65 1.89 1.11 0.11 0.38 95.87 4.13

Entries in each cell are the percentage of forecast error variance of the market return in the first column explained by the market in the first

row

* Entries in the 'All foreign' column denote that the total percentage of forecast error variance of the market in the first column explained by

all foreign markets

Cholesky Ordering: BAHRAIN OMAN KUWAIT ABUDHABI SAUDI OIL

Standard Errors: Monte Carlo (1000 repetitions)

- Impulse responses

The estimated impulse responses of the VAR system offer an additional way of

examining how each of the six variables responds to innovations from other variables in

the system. Table 1 through 4 and figure 1 and 2 in appendix 4; summarize the responses

of all markets to one standard deviation shock in oil returns and the responses of oil

returns to one standard deviation shock in each of the GCC markets for the two sub-

periods. Table 1 presents the response of all markets returns to one standard deviation

shock in oil return for the first sub-period. In general the responses are small starting

from day 2 and taper off very slowly indicating that markets are not efficient in

responding to shock generating from oil return. However, it is apparent that a shock

originated in oil return has a major and persistent impact on Omani market more than

other GCC markets. It took Omani market 2 days to start responding to oil return’s shock.

201

In addition, Kuwaiti and Omani markets respond positively while other markets respond

negatively to shock in oil return.

Table 2 in appendix 4 presents the response of oil return to shocks generated from

each of the GCC stock markets and reveals that, shocks generated from Bahraini and

Omani markets have large and persistent impact on oil return. Since oil response has a

memory up to 15 days reaching the value of -0.156, 0.198 at the end of 15 days horizon

for both Bahraini and Omani markets respectively. However, the response of oil returns

taper off quickly and died up to 7 days, for shocks generated from Kuwaiti, Abu Dhabi,

and Saudi markets. These results are for the first sub-period, but for the second period

which witnessed the sharp raising in oil prices, we have a different picture for the

interaction relationship between GCC stock markets and oil return.

Table 3 presents the responses of GCC stock markets to a shock generated from

oil returns. Saudi market stands to be the most influenced followed by Abu Dhabi market.

Saudi market reacts in day 2 through 5. The same reaction for Abu Dhabi and Kuwaiti

markets, it seems that these markets react quickly and relatively efficient since their

reaction taper off and decline rabidly up to day 5 to shock originated in oil return.

However, Bahraini and Omani markets show a small and slow process in responding to

oil shock. The impulse responses for shocks originated in the GCC stock markets for the

second sub-period and the influence on oil return are presented in table 4 of appendix 4.

It seems that a shock in the Saudi market has the most influence on oil return, since oil

returns has a memory up to 4 days to absorb shocks generated from Saudi market, 0.212

In day 2 reaching 0.003 at the end of day 4, while oil return exhibits slow and persistent

process in responding to Omani market’s shock, 0.134 at the end of 15 days. In addition,

the least response of oil return seems to be for a shock in Abu Dhabi market.

Despite the different magnitude of impulse response values, some observations can be

made from the above mentioned tables:

• For the first sub-period the response of GCC stock markets for a shock in oil

return seems to be small and taper off slowly. On the other hand and for the same

sub-period, the response of oil to shocks in Bahraini and Omani markets seems to

be large and persistent.

202

• For the second sub-period and after oil prices get higher, the interaction between

oil return and GCC stock markets increased especially for Saudi, Abu Dhabi, and

Kuwaiti markets, they exhibit large and quick responses to oil shocks within 4

days horizon. The reverse also is true when oil responds to shocks generated from

these markets. Indicating that, the interaction process between these three markets

and oil appears to be efficient. While for Bahraini and Omani markets which are

Non- OPEC members, they interact slowly for oil shocks.

• The Saudi market exerts the great effect when oil prices get higher in the second

sub-period. This is must not be surprised, since Saudi Arabia is the largest oil

exporter and has the largest oil reserves in the world.

These findings reflect the important impact of the increase in oil prices on GCC stock

markets. This is quite natural since GCC countries produce about 21% of the world's

daily oil production, and they possess about 43% of the world's oil reserve.

5-4 summary

This chapter has examined the degree to which Arab stock markets are integrated

both regionally and globally. According to the fact that Arabian economies are not

similar, Arab stock markets have been divided into two groups: oil production countries

which mainly include GCC markets, and non-oil production countries (Jordan, Egypt,

and Palestine). The results indicated that Arab stock markets appear to be segmented

from international stock markets, since no cointegrating relation was found between

Arabian and international markets, represented by S&P 500, (VAR-10). Moreover,

evidence of regional financial integration between Arab stock markets is still weak, since

the results from VAR-9 and its VEC model indicate that, despite the existence of long-

run relationship, linkages on the short run still weak between these markets, even it is

difficult to identify the true leader between the nine Arab stock markets.

As for GCC markets, the results of VAR-6; which includes 6 GCC markets,

indicate that Kuwait market can be considered as a leader for these markets. In addition,

some directional relationships have been found between GCC indices, but one still

expects to find more short-run relations between countries, share many economic and

social aspects. The results for non-oil countries indicate that these markets are not

203

cointegrated in the long-run (VAR-3). Moreover, the chapter also examined the effect of

the raise in oil prices in the last two years on GCC stock markets, through investigating

the dynamic structure between five member countries of the oil-rich GCC and oil prices.

The results show that oil prices dominate the long-run equilibrium with GCC stock

markets (VAR-7), and have a significant effect on returns volatility in these markets.

More important, the findings reflect the important impact of the increase in oil prices on

GCC stock markets, since after the rise in oil prices, four out five GCC markets can

predict oil prices while only two GCC markets can be predicted by oil prices. The results

on the link between oil prices and stock markets are consistent with the existing

international literature, since it is found that oil prices shocks have significant effect on

stock markets (see Johnes and Kaul 1996; Huang et al. 1996; Sadorsky 1999; Papapetrou

2001; and Hammoudeh and Aleisa 2002, 2004).

For portfolio diversification benefits, the message of these results for international

investors appears clear. Arab stock markets can offer diversification potentials for

international investors; both on the long and short-run horizons; since the body of

evidence in support of integration of major world stock markets is quite impressive,

international investors often search for new emerging markets which offer the risk-

reward trade-offs, they cannot get in more matured markets. The results here suggest that

Arab stock markets may have such potential benefits; stocks in these markets can

minimize the risk of spillovers from other foreign markets (like the US market), and thus

may limit the contagion effects which inflect more globally integrated markets. The

apparent segmentation of Arab stock markets suggests that these markets are not only

emerging, with enormous growing potentials, but they also offer international investors

diversification benefits unavailable elsewhere. For regional portfolio investments, non-oil

countries (Jordan, Egypt, and Palestine) could offer the rich GCC investors with

diversification benefits to diversify their portfolio regionally.

From a policy point of view, Arab policy makers should make regulatory and

accounting changes to promote financial integration among their markets especially for

GCC countries. They should allow more eligible companies to their own to be listed on

their exchanges, and permit cross-company listing. Privatization and privately held

companies will spread the risk and lead to greater market development. Furthermore, the

204

results for GCC markets suggest that there is a bidirectional relationship between Saudi

market and oil prices, and a directional relation from Oman to oil prices, the things that

point to those countries’ predictive power on oil prices. This would also show that,

political and economic stability (or lack thereof), has a direct impact on stability in oil

prices. Saudi Arabia, for example, which has the largest oil reserve in the world (about

24%), does affect oil prices and simultaneously be affected by them.

Equally important is that; decision makers in those countries have to secure

diverse income resources, and try to increase their contribution (non-oil sectors) to GDP.

Since Oman, for instance, oil contribution to GDP amounts to about 42%, Kuwait 46.6%,

and Saudi Arabia 38%, in 2003. This may lead to risks, as a result of linking those

countries’ economies with oil prices in view of the risks in the oil market, which reflects

negatively on the performance and volatility of their stock markets; taking into account

that GCC countries are planning to perform a single currency before 2010.

205

6- Vision and strategic plan for Arab stock markets

The empirical results indicate that the weak form of efficient market hypothesis

can be rejected in the Arab stock markets, and that these markets are characterized with

thin trading and non-linear return generating process.

Despite the noticeable developments in the performance of these markets, as

designated by markets performance and activity indicators, these markets are still

suffering from numerous problems and obstacles that detain their development and

growth, such problems are: the limitation of the investment opportunities available for

investors, caused by tiny number of listed companies, highly concentrated markets since

few number of listed companies possess most of trading activity, and fewness of the

institutional investors since most of the existing investors are individual ones, the fact

that these individual investors are less informed and lacked of investment awareness in

general which lead them to act in a manner similar to noise trading, which negatively

affect the stability and performance of Arab stock markets. In addition, such behavior

may lead returns to respond non-linearly to the arrival of new information to the market.

On the contrary, the institutional investor makes his decision based on scientific analysis

of the available information, a fact that leads to market stability and reduce sharp prices

fluctuations.

Moreover, the empirical results indicate that Arab stock markets are not integrated

with international markets or among themselves. Since no signs of cointegration were

found between oil and non-oil production countries’ stock markets and between Arab

markets as a group and international markets. The thing that may provide primary

indicators that Arab stock markets could offer diversification potentials for both regional

and international portfolio investors. But the fact that, GCC stock markets impose

restrictions on both foreign and non-GCC national Arab investors practically diminish

these diversification potentials.

Additionally, the results present the increasing risks resulting from direct linkages

between GCC economies and oil prices, and its important effect on the performance of

the GCC-stock markets. It is the time that GCC countries should take considerable steps

toward diversifying their GDP components, to decrease their dependence on oil sector.

For instance, it is expected that oil reserve in Oman to be consumed within 18 years with

206

the current level of production, while oil contribution to GDP account for 42 percent, for

Saudi Arabia 38 percent, and Kuwait 42 percent, the thing that clarifies the risks caused

by high significant correlation and dependence between these economies and oil

production. As a result, political and economic stability (or lack thereof) of the largest oil

producers, also has implications for the stability of oil prices. Since Saudi Arabia has the

largest oil reserve and production in the world. Moreover, the GCC countries exist in a

region characterized with political instability and witnessed three regional wars during

the last two decades.

To understand the causes and consequences of the obtained results, and its

implications on Arab stock markets, it is important to analyze the surrounding

environment where these markets exist. It can be concluded that among the problems that

these markets suffer and affect there performance and efficiency are thin trading,

narrowness and illiquidity, the fewness of the available investment opportunities, and the

segmentation from international and regional stock markets. The remaining of this

section will analyze the environment surrounding these markets and determine the

shortages. The thing that lead us to identify and determine the strength, weakness,

opportunities and threats that face Arab stock markets, as a way to draw a general

strategic plan to develop and enhance the performance of these markets.

In addition to the legal and institutional framework adequacy and completeness,

and the availability of efficient executive management, the development of the stock

market depends on the availability of sufficient demand and supply. The supply level in

Arab stock markets still weak, leading these markets to be narrow, and affects negatively

market liquidity and trading activities. Since one can find hundreds of listed companies in

one market, without any effective role on trading floor (i.e. Egyptian stock market).

Where listing for prestige and reputation purposes will not offer financial papers for the

market, it is important to handle and treat this negative phenomenon.

207

Financial paper supply could be enhanced by new issues for existing companies,

privatizing state owned shares in these companies, the conversion of individual owned

companies to corporate ones, in addition to the creation of new financial instruments, will

increase and improve investors investment options. To achieve the above, it is needed to

remove legal and legislation obstacles, facilitate stock market procedures, allowing for

cross-listing between markets, and increase investors confidence with financial

intermediaries in these markets.

On the other hand, supply policies should be followed with other policies to

attract investors’ attention. It is not expected that supply level still increasing with fixed

or slowly demand growth. Since weak demand level will be reflected with low level of

coverage for the new issues, which discourage companies to issue new issues. So it is

important for Arabian decision makers to implement polices that motivate the demand of

financial papers in their local stock markets, keeping supply growth at the same time. The

priority of such polices may switch between treating deficiencies in financial papers

supply, or increase and motivate financial papers’ demand, according to each market

specific features and conditions. For instance, GCC markets need to concentrate on

improving supply to absorb surplus liquidity resulting from huge raise in oil prices, and

to reduce negative speculations resultant from huge demand on financial papers.

There are several policies that can be followed to motivate demand, such as

increasing investment awareness, especially in Arabian countries who suffer from

financial deficit, where commercial banks dominate financing sector through short-run

finance. Arab stock markets are required to increase investment awareness with stock

market’s activities and investment opportunities that a stock market can offer. Moreover,

it is important to develop the specialized financial press, and create new financial and

investment instruments to achieve investors’ needs to increase the demand on financial

papers and improve stock market activities. Arab stock markets need to change the

current situation since trading took place only on bonds and common stocks. They have

to create new instruments and papers that mixed between bonds and stocks; such as

warrants and options, which will improve demand and supply on financial papers.

Furthermore, trading costs affect the demand and stock market microstructure,

which will be discussed in details later. In addition, to the policies that aimed to develop

208

supply and demand, different policies could be used that affect both demand and supply,

such as improving the efficiency of intermediaries operations, increasing information,

and market control.

The particular trading arrangements in an equity market may directly affect two

key functions of that country’s stock market: price discovery and liquidity. First, the

trading process should lead to “fair” and correct prices; in other words, no investor

should be able to manipulate market prices in his or her favor. Second, trading should

occur at a, low transaction cost, and large quantities should trade without affecting the

price. These issues are the topic of the field of market microstructure. It is generally

believed that microstructure improvements should greatly affect the liquidity of the

markets, which can best be approximated by the cost of trading and increasing the

turnover and liquidity in the market.

While in the case of Arab stock markets, trading costs (market commissions and

intermediaries costs) are significantly affect investors’ enthusiasm to deal with financial

market. That is why Arab markets not only need to reduce trading costs as possible, but

also to harmonize and unify these costs among markets, since it is noticeable the

existence of large Bid-Ask spread in these markets. Moreover, the differences between

trading costs make it difficult to discover the opening treading price, and affect

negatively market liquidity.

Market integration is central to both questions. In finance, markets are considered

integrated when assets of identical risk command the same expected return irrespective of

their domicile. In theory, liberalization should bring about emerging market integration

with the global capital market, and its effect on emerging markets is then clear. Foreign

investors will bid up the prices of local stocks with diversification potential while all

investors will shun inefficient sectors. Overall, the cost of equity capital should go down,

which in term may increase investment and ultimately increase economic welfare.

209

Moreover, market integration is of concern both to equity investors, and

companies in the region that make capital budgeting decisions. Specifically, if

segmentation exists and firm is forced to raise capital locally, then its cost of capital is

likely to be higher than that of a company with unrestricted access to the regional and

international capital markets. The empirical results here indicate that Arab stock markets

are segmented from international stock markets. Segmentation also exists between oil and

non-oil Arabian countries. The thing that raise the needs for these markets to take

significant steps toward market liberalization, especially for GCC markets, which impose

several restrictions on foreign investors and in some cases even for Arabian investors

who are non-GCC nationals.

On the other hand, increases in correlations between markets would imply a

decrease in the benefits from international diversification in line with portfolio theory.

However, under the concept of cost-benefit analysis, Arab stock markets will gain more

benefits from market liberalization. Since the tendency for the global markets to become

more integrated; is a result of the increasing tendency toward liberalization and

deregulation in the money and capital markets, both in developed and developing

countries as well as on a bilateral and multilateral basis. Such liberalization is important

to introduce structural reforms, to promote economic efficiency, to stimulate trade and

investment, and to create a necessary climate for promoting sustainable economic growth

with a commitment to market-based reforms.

6-1-4 Privatization

In most emerging markets, privatization was intended to increase productivity of

state-owned economic enterprises (SOEs), and to help reduce government budget

deficits. In some cases, governments actively sought to promote capital market

development through privatization. Many governments intended to create a class of

people with a stake in the new economy, thereby making it more difficult for political

changes to be reversed. Regardless of the goal, privatization was not initiated, in order to

divest fully the government’s interest in the real economy. Nevertheless, even the partial

divestment under consideration was economically substantial.

210

Privatization programs impact emerging capital markets through various

mechanisms. For instance, share issued privatizations (SIPs) increase the market

capitalization and the value traded on local exchanges. Moreover, SIPs can change the

investment opportunity set of portfolio investors, public offers of SOEs whose cash flows

are not perfectly correlated with pre-existing companies; help investors to achieve gains

through diversification. Under this scenario, SIPs may help to lower the risk premium

investors require for holding the market portfolio of publicly traded equity. Other

methods of privatization, including the direct sale of former SOEs, the direct sale of

SOEs assets, or concessions of public sector monopolies, alter the dynamics of local

capital markets in less obvious ways. Consider the direct sale of an SOE to a private

investor, this sale does not increase the market capitalization or value traded of the local

exchange. However, the sale may alter the real investment opportunity set of the private

investor.

As viewed from this perspective, all forms of privatization can impact local

capital market dynamics. The common component of privatization that impacts capital

markets is the transfer of productive resources from the public sector to the private sector.

This transfer may allow investors to achieve benefits through diversification and may

affect the cost of capital in emerging markets. Even if private investors do not benefit

from the transfer of resources, i.e. their investment opportunity set does not change;

privatization programs may still influence capital markets. Privatization program can help

the government signal its commitment to free market polices. For most emerging markets

governments, the implementation of a privatization program reverses decades of state-led

economic development. Successful privatization of politically sensitive industries may

convince investors to reduce the ex ante perceived risk of government interference in

investment decisions and expropriation of productive assets. As a result of sustained

privatization efforts, the sovereign risk premium inherent in the governments fixed

income liabilities may be reduced. As this chain of events ripples through the economy,

local market entrepreneurs eventually benefit in their ability to obtain debt financing at

lower cost.

Despite the fact that most Arabian countries going toward privatization programs

of state-owned economic enterprises, investment opportunities are still limited, investors

211

still facing a few number of listing companies with limited financial instruments. It is

important for such countries to continue with an effective privatization process to

enhance market depth, liquidity, and trading activities.

Regulatory framework is a combination of laws, legislations, and instructions that

manage financial papers’ issuing and trading. In addition to the companies’ and financial

papers’ laws; which are the main components of the regulatory and legal framework,

there are several legislations related to financial papers such as investment and taxation

laws, auditing law, and banks and financial institutions laws. These legislation,

instructions and decisions produced by market supervision party, are forming the

fundamental structure that a stock market build on, determining the types and nature of

financial instruments, identifying listing requirements and conditions, disclosures’

standards, and brokerage rules. Moreover, these laws controlling trading process, deposits

and clearing settlements, and professional behavior and ethics.

- Companies laws

Companies’ law is the most important law that related to stock market. It manages

the establishment and registration procedures for corporate companies, and determines

the types of financial papers that companies can issue, issuing conditions and

requirements, and many other features that manage and protect owners and investors

rights, such as those related to financial disclosure and controlling requirements. So

companies’ law directly affects the supply of financial papers.

In this consequence, companies’ laws in several Arabian countries have been

reviewed and subjected to full comprehensive revision, mainly aimed to improve

companies’ establishments’ conditions, and enhancing the procedures that manage

financial papers issuing.

- Financial papers law and its related instructions

Financial papers’ law is the main step for establishing and organizing a stock

market. Despite the differences between financial papers’ laws among Arab stock

markets, they are generally similar in several aspects such as stock market establishing

procedures, regulatory framework, and its goals and objectives.

212

Despite the regulatory and legislation improvements in most Arab markets, there

are still several shortages and disadvantages, most of them may be resulted from the

inadequate, inefficient implementations of the contents of these laws, which will be

summarized as follows:

• Some regulatory and financial papers’ laws have been created in early stages,

according to theoretical framework without practical examination. This put these

laws disable to follow continuing improvements in international stock markets.

Moreover, continues adjustments for some of these laws’ articles make them

inconsistent and unstable over time.

• A disadvantage that face these markets that they are controlled by several laws

and regulations governed by numerous parties.

• The absence of regulations in some Arabian markets that separate between the

supervision and the executive role. The two roles continue to be simultaneously

in the hands of the capital market it self (as in Bahrain, Saudi Arabia, Kuwait,

and Oman).

• The absence of legislations in some Arabian markets that forced listed companies

to be committed with international accounting and auditing standards, since some

of financial statements lack of the adequate disclosure, ambiguity, and

incomparability.

• Despite the existence of legislation that oblige listed companies to issue mid and

quarterly financial statements, these statements in several cases came out too late

which make them useless.

• The lake of legislations that manage the establishments of issuing houses. Since

such houses promote and insure the full coverage of new issues, and play a

significant role in privatization programs especially when it is difficult for a stock

market to absorb a large number of new issues.

• The shortage of legislations that regulate the establishments of credit rating and

issuing credit status institutions.

• In many cases, the lack of regulations that manage and allow cross-listing

between Arabian markets, where in other cases; especially GCC markets,

213

accessibility of foreign investments is restricted, even for non-GCC nationals

Arab investors.

• The lack of coordination and harmonization between regulatory and legal

frameworks among Arab stock markets, which limit the integration ability among

these markets, especially for GCC countries which are planning to establish a

monetary union and have a single currency before 2010.

According to the previous analysis, and taking into account each market special

features, characteristics, and environment, it is possible in general to determine the most

significant strength, weakness, threats, and opportunities that Arab stock markets enjoy as

follow:

- Strength

• According to international standards, Arab stock markets are considered as

emerging markets with promising growth potentials.

• The segmentation of Arab stock markets from international markets protects them

from the contagion effect caused by international financial crisis, and raises the

diversification potentials that these markets may offer.

- Weakness

• Arab stock markets can be characterized as inefficient markets in the sense of

weak-form of efficient market hypothesis.

• These markets are described as narrow markets and lack of depth.

• Thin trading and illiquidity.

• The narrowness of the available investment opportunities with little number of

listed companies and limited investment instruments.

• The lake of investment awareness in general between investors.

• The diminish role of institutional investor, since most investors are small and

individual ones.

• Highly concentrated markets, since a small portion of listed companies dominate

most of trading activities.

214

• The unavailability of timely, adequate, and reliable financial information about

listed companies.

• The weak; and in some cases the absence, of the financial intermediaries role and

the lack of specialized financial information institutions.

• Several shortages in the legal and regulatory framework and in many cases the

inadequate implementation of these laws and legislations.

• The absence of effective coordination and cooperation among Arab stock

markets.

• Lack of harmonization and highly trading costs among these markets.

• The existence of many obstacles that hinder the direct flow of foreign investment

especially in GCC stock markets.

- Opportunities

• The benefits and advantages that may be obtained from market liberalization

process.

• The available opportunities of growth, since Arab stock markets are emerging

markets.

• The financing potentials that these markets can offer for local and regional

companies who are looking for raising funds with low cost of capital.

• The contributions of Arab stock markets in developing local economies and

facilitate privatization programs, through offering several investment channels,

encouraging saving, promoting investment, and efficient capital allocation.

• Attracting foreign investors through offering diversification potentials, after

removing all obstacles that prevent direct foreign investments.

• Offering profitable investment opportunities for expatriate Arab investments, and

encourage the return of these investments which are estimated with billions of

dollars.

- Threats

• The direct linkages between GCC economies and oil industry, which increase

risks in its financial markets and affected by sharp fluctuations in oil prices.

215

• The risk of sharp and extensive speculations that affect GCC stock markets

resultant from surplus liquidity and the narrowness of investment options

available for investors.

Table 6-1: Strength, Weakness, Opportunities, and Threats for Arab Stock Markets

STRENGTH BASED ON

According to international standards, Arab stock markets are Emerging markets have promising

considered as emerging markets with promising growth potentials. growth potentials.

The segmentation of Arab stock markets from international Results of cointegration test, Structural

potentials that these markets may offer.

WEAKNESS BASED ON

Arab stock markets can be characterized as inefficient markets in Regression analysis, Variance ratio,

the sense of weak-form of efficient market hypothesis. Runs test, Serial correlation, BDS test,

and Volatility analysis GARCH models,

Existing anomalies.

These markets are described as narrow markets and lack of depth. The lack of several investment

opportunities and instruments.

Thin trading and illiquidity. Large Bid-Ask spread, see section 3-1-1

The narrowness of the available investment opportunities with little Few number of listed companies, limited

number of listed companies and limited investment instruments. investment instruments, see table 2-12.

The lake of investment awareness in general between investors. Large number of unwell informed

individual and small investors.

The diminish role of institutional investor, since most investors are Large waves of speculations, non-linear

Highly concentrated markets, since a small portion of listed See figure 2-5

The unavailability of timely, adequate, and reliable financial The delay in issuing mid and quarterly

The weak; and in some cases the absence, of the financial Lack of investments awareness,

intermediaries role and the lack of specialized financial information unavailability of specialized financial

institutions. press

Several shortages in the legal and regulatory framework and in Inaccessibility of direct foreign

many cases the inadequate implementation of these laws and investments to GCC markets.

legislations.

216

The absence of effective coordination and cooperation among Arab Non-GCC national Arab investors are

markets.

Lack of harmonization and highly trading costs among these Several differences between trading

The existence of many obstacles that hinder the direct flow of Inaccessibility of direct foreign

foreign investment especially in GCC stock markets. investments to GCC markets, even for

non-GCC nationals Arab investors

OPPORTUNITIES BASED ON

The benefits and advantages that may be obtained from market Entrance of foreign investments,

activities.

The financing potentials that these markets can offer for local and Cointegration test, SVAR results.

regional companies who are looking for raising funds with low cost

of capital

The contributions of Arab stock markets in developing local The effective role that stock market can

economies and facilitate privatization programs, through offering play in efficient capital allocation,

several investment channels, encouraging saving, and promote absorbing new issues.

investment.

Attracting foreign investors through offering diversification The segmentation between these

potentials, after removing all obstacles that prevent direct foreign markets and international financial

Offering profitable investment opportunities for expatriate Arab The availability of profitable investment

THREATS BASED ON

The direct linkages between GCC economies and oil industry, Cointegration test and VAR analysis for

which increase risks in its financial markets and affected by sharp oil prices and GCC markets, the

market volatility GARCH models.

The risk of sharp and extensive speculations that affect GCC stock High waves of speculations, weak

markets resultant from surplus liquidity and the narrowness of supply with growing demand on

217

6-3 Vision and strategic goals

The objective of this action plan is therefore to achieve the following strategic

goals, which expected to enhance and develop Arab stock markets, increasing market

liberalization, improving market efficiency, through increasing market liquidity, depth,

and trading activities. The thing that put these markets in a position that able to attract

portfolio investors, signs these markets as a target from foreign investors, and facilitates

the integration and cointegration between Arab and international and regional financial

markets.

To achieve these goals, it was necessary to draw specific targets under the

umbrella of a general vision, which will be achieved through specific strategies that could

be realized through tactical programs and activities, as described bellow. Taking into

account the following couple of notes:

• The following programs are not exclusive, any other additional programs and

activities could be added, to achieve the goals.

• It is important to consider the specific conditions for each individual market,

regarding economic factors, regulatory frameworks, and the liberalization degree

that a stock market reached.

This strategic plan aims to achieve several objects in order to reach two main

goals: the first goal aims to enhance and improve market efficiency through enhancing

the legal and regulatory frameworks that manage these markets, improving market

microstructure, increasing market depth and liquidity, the thing that increasing the

capability to liberalize and open Arab stock markets with international markets, which is

the second main goal.

Toward liberalized, liquid and efficient Arab stock markets, with the

capability to attract foreign investments and portfolios by providing

several investment instruments and opportunities, that offer

diversification potentials.

218

- Strategic plan

Goal 1: Increasing and improving Arab stock markets efficiency.

Target 1: Increasing market depth and liquidity by enhancing trading activities.

investment opportunities and instruments.

Adopting the required legal procedures that encourage companies for

listing in stock markets.

Promoting cross-listing for financial papers among Arab markets

Increasing tax incentives for listing companies.

Offering new financial instruments through amending the related

legislations and adequate implementation of regulations.

Reviewing and reducing trading costs as possible.

Increasing investment awareness in general.

Creating investment funds to attract the less informed small investors.

Providing frequent, qualified financial information.

Increasing and activating the role of financial intermediaries’ institutions.

Target 2: Protecting investors’ rights and keep on market stability away from

sharp fluctuations.

Strategy 1: The availability of adequate financial information regarding listed

companies’ activities and performance.

219

Forcing listed companies with international accounting and auditing

standards especially those related to full financial disclosure

Creating specialized financial press.

Encouraging the existence of expert and credit rating houses.

Forcing listing companies to issue regulatory financial statements on its

time especially semi & quarterly statements.

Strategy 2: Improving market microstructure.

Reducing trading costs as possible.

Full isolation and clear determination of responsibilities between the

supervision & the executive functions according to the international stock

markets standards.

Strategy 3: Increasing and enhancing the role of institutional investor.

Target 1: Harmonization and conformance of regulations and legislations

among Arab stock markets.

Strategy 1: Relaxation of the legal and regulatory impediments.

Allowing non-GCC notional investors to directly invest in GCC stock

markets.

Strategy 2: Increasing correlation and integration between Arab stock markets.

Unification of listed conditions & commissions.

Unification of clearing & deposits regulations among Arab stock markets.

Eliminating the legal obstacles that hinder cross listing between markets.

Harmonizing tax laws among Arab countries especially those related to

financial markets.

Improving & renewing companies’ law.

Enhancing financial papers law.

especially to GCC markets.

Strategy 1: Attracting international and regional investments and portfolios.

220

Removing all obstacles that prevent foreign direct investments in GCC

markets.

Strategy 2: Emphasizing the diversification benefits that Arab stock markets may

offer for portfolio investors.

Providing reliable financial information for listed companies activities and

market performance.

Improving & introducing new trading techniques such as E. trading.

markets.

Strategy 1: Benefiting from advantages and experiences resultant from

integration with international markets.

Increasing investment and financing options available for investors.

Encouraging the establishments of institutions specialized in producing

financial information which promote Arab markets internationally.

Removing any obstacles that prevent foreign direct investment mobility to

GCC markets.

efficient capital allocation.

Strategy 1: Create investment tools and opportunities to encourage saving.

Establishing investment funds targeted small investors, i.e. pension funds.

Offering several saving tools and opportunities.

Encourage the return of Arab expatriate investments through offering

profitable investment options.

Absorbing the surplus liquidity resultant from raise in oil prices.

Strategy 2: offering different financing sources for those companies looking for

raising fund.

Absorbing the new companies’ issues via effective and deep stock market.

Offering several financing options for companies.

221

Strategy 3: contribution in achieving governments’ privatization programs and

diversified GDP components in GCC countries to reduce oil risks.

Absorbing the state-owned enterprises (SOEs) issues resulting from

privatization programs.

The availability of issuing houses those secure the full coverage of the

new issues.

222

Goal 1: Increasing and improving Arab stock markets efficiency

Increasing market depth and liquidity by Protecting investors rights and keep on market

enhancing trading activity stability away from sharp fluctuations

Specific Objects

papers supply and enhancing the adequate financial market enhancing the

diversify the demand for financial information microstructur role of

Strategies

available investment papers regarding listed e institutional

opportunities and companies’ activities investor

instruments and performance

procedures that encourage companies - Increasing investment - Forcing listed companies with - Reducing trading costs as

for listing in stock markets awareness in general international accounting and possible

- Promoting cross-listing for financial - Creating investment funds to auditing standards especially - Full isolation and clear

papers among Arab markets attract the less informed small those related to full financial determination of

- Increasing tax incentives for listing investors disclosure responsibilities between the

companies - Providing frequent, qualified - Creating specialized financial supervision & the executive

- Offering new financial instruments financial information press functions according to the

through amending the related - Increasing and activating the - Encouraging the existence of international stock markets

legislations and adequate role of financial intermediaries expert and credit rating houses standards

implementation of regulations institutions - Forcing listing companies to

- Reviewing and reducing trading costs issue regulatory financial

as possible statements on its time especially

semi & quarterly statements

223

Goal 2: Liberalizing Arab stock markets

Harmonization and conformance Facilitating the mobility of Integrating & correlating Arab Enhancing the role of the financial

Specific Objects

of regulations & legislations international & regional markets with international market in economic development

among Arab stock markets portfolios especially to GCC markets & efficient capital allocation

markets

Relaxation of the Increasing Attracting Emphasizing the Benefiting from Create investment Offering different Contribution in

legal & regulatory correlation & international & diversification advantages & tools and financing sources achieving

impediments integration between regional benefits that Arab experiences opportunities to for those governments’

Arab markets investments & markets may resultant from encourage saving, companies privatization

portfolios offer for portfolio integration with looking for programs and

investors international raising fund diversified GDP

markets components in

GCC countries to

reduce oil risks

- Allowing non-GCC - Unification of listed - Removing all - Providing reliable - Increasing investment - Establishing - Absorbing the new

notional investors to conditions & commissions obstacles that financial and financing options investment funds companies issues via

directly invest in GCC - Unification of clearing & prevent foreign information for available for investors targeted small effective & deep stock

stock markets deposits regulations among direct listed companies - Encouraging the investors, i.e. pension market

Arab stock markets investments in activities & market establishments of funds - Offering several

- Eliminating the legal GCC markets performance institutions specialized - Offering several financing options for

obstacles that hinder cross - Improving & producing financial saving tools and companies

listing between markets introducing new information which opportunities

- Harmonizing tax laws trading techniques promote Arab markets - Encourage the return

among Arab countries such as E. trading internationally of Arab expatriate

especially those related to - removing any obstacles investments through - Absorbing the SOIs issues

financial markets that prevent foreign direct offering profitable resulting from privatization

- Improving & renewing investment mobility to investment options program

companies law GCC markets - Absorbing the surplus - The availability of issuing

- Enhancing financial papers liquidity resultant from houses those secure the full

law raise in oil prices coverage of the new issues

224

7- Conclusions

The Efficient Market Hypothesis (EMH) states that asset prices in financial

markets should reflect all available information. As a consequence, prices should always

be consistent with “fundamentals”. Most studies on EMH were conducted on the world’s

largest stock markets. However, the past twenty years have witnessed spectacular growth

in size and relative importance of emerging markets in developing countries. Emerging

markets have long posed a challenge for finance; standard models are often ill suited to

deal with the specific circumstances arising in these markets. However, the interest in

emerging markets has provided impetus for both the adaptation of current models to new

circumstances in these markets and the development of new models.

The focus of this thesis was toward nine Arab new emerging markets in the

Middle East region. Having presented the main literature regarding EMH, our attention

has been directed to test market efficiency in these markets. Consider the specific features

for these markets; we started firstly by adjusting daily observed indices for the possible

effect of infrequent trading. Based on the results obtained from chapter 4, random walk

properties have been rejected for Arab stock markets. The results obtained from

regression analysis, variance ratio, BDS, runs test, and serial correlation tests, rejected the

randomness and independence of returns, even after observed indices have been corrected

for infrequent trading. Moreover, the results indicated that, prices responding non-

linearly to the arrival of new information, while volatility clustering phenomenon still

seems to characterize markets’ returns. The GARCH (1,1) results for daily returns

indicated that all markets exhibited volatility clustering with one exception for Dubai.

Furthermore, volatility seems to be persistent in three markets (Egypt, Kuwait,

and Palestine) with a slow rate of decay. Additionally, four Arab markets (Bahrain,

Dubai, Kuwait, and Oman) showed signs of leverage effect with asymmetric shocks to

volatility. The GARCH models found to explain quite satisfactory the non-linear

dependence found in the time series. Furthermore, seasonality and calendar effects existed in

Arab markets with three forms; day-of-the-week effect, month-of-the-year effect and the

Halloween indicator.

Having answered the question regarding market efficiency in Arab stock markets, we

tried to examine the degree to which these markets are integrated with other international

markets. In other words, we tried to investigate if Arab stock markets can offer diversification

225

benefits for both regional and international portfolios investors. The analysis has been conducted

in two directions, firstly to investigate the cointegrating relation between Arab and international

markets, while the second direction was to examine the dynamic relationships between Arab

markets as a group.

The results of multivariate cointegration techniques indicated that Arab stock markets

appear to be segmented from international markets, since no cointegrating relation was found

among variables in the system. However, in the short-run there could be some interactions

between Arab and international markets. To examine this, a structural vector autoregression

(SVAR) model has been employed, to answer the question how do Arab markets react to shocks

originated in international markets (US, UK, and Japan) under the assumption that, the returns on

each of the three international markets affect the returns on Arab markets but not vise versa. The

resultant impulse response functions and variance decomposition indicated that, the linkages

between Arab and international markets still very weak in the short-run, with some signs that the

UK market exerted the most effect in influencing Arab markets.

Next we continued to examine the dynamic relationships between Arab markets

themselves. According to the fact that Arabian economies are not similar, the total markets have

been divided into two groups: oil production countries which mainly contain GCC markets, and

non-oil production countries (Jordan, Egypt, and Palestine). The results indicated that despite the

existing long-run cointegrating relation, the linkages between Arab markets still weak in the short

run, while for non-oil countries indicated that these markets are not integrated on the long-run.

It is known that several economic factors may affect stock market performance; such as

oil prices, given the fact that oil prices have a significant effect on GCC economies. We continue

chapter 5 by investigating the effect of oil prices on GCC stock markets. Several techniques have

been used, firstly to test the effect of oil prices on market volatility, oil returns have been added as

an additional regressor in the variance equation of the GARCH model, the results indicated that

oil prices have a significant role in affecting GCC markets’ volatility. Second, using multivariate

cointegration and vector autoregression (VAR) models, we concluded that oil prices formed and

dominated the long-run cointegration with GCC markets. Furthermore, after the raise in oil

prices; especially during the last two years, the linkages between oil and GCC markets increased,

four GCC markets have predictive power on oil prices, with only two markets (Saudi Arabia and

Oman) to be predicted by oil prices. Finally, and on the light of the obtained empirical results, a

strategic plan has been suggested to develop the performance of Arab stock markets based on two

main broad goals, improving market efficiency and increasing market liberalization. This will be

226

achieved through specific targets and strategies that could be realized through tactical programs

and activities.

In conclusion, based on the results of this thesis; some implications for economic policy

can be made. For portfolio diversification benefits, the message of these results for

international investors appears clear, Arab stock markets can offer diversification

potentials for international investors; both on the long and short-run horizons. The results

here suggested that Arab stock markets may have diversification potentials; stocks in

these markets can minimize the risk of spillovers from other foreign markets (like the US

market), and thus may limit the contagion effects which inflect more globally integrated

markets. The apparent segmentation of Arab stock markets suggested that these markets

are not only emerging, with enormous growing potentials, but they also offer

international investors diversification benefits unavailable elsewhere. For regional

portfolio investments, non-oil countries (Jordan, Egypt, and Palestine) could offer the

rich GCC investors with diversification benefits to diversify their portfolio regionally.

From a policy point of view, Arab policy makers should make necessary

relaxation to regulatory and legal framework, to promote financial integration among

their markets especially for GCC countries. They should allow more eligible companies

to their own to be listed on their exchanges, and permit cross-company listing.

Privatization and privately held companies will spread the risk and lead to greater market

development. Moreover, the results for GCC markets suggest that there is a bidirectional

relationship between Saudi market and oil prices, and a directional relation from Oman to

oil prices, the things that point to those countries’ predictive power on oil prices. This

would also show that, political and economic stability (or lack thereof), has a direct

impact on stability in oil prices. Saudi Arabia, for example, which has the largest oil

reserve in the world (about 24%), does affect oil prices and simultaneously be affected by

them.

Equally important is that; decision makers in these countries have to secure

diverse income resources, and try to increase the contribution of non-oil sectors to GDP.

Since Oman, for instance, oil contribution to GDP amounts to about 42%, Kuwait 46.6%,

and Saudi Arabia 38%, in 2003. This may lead to risks, as a result of linking these

countries’ economies with oil prices in view of the risks in the oil market, which reflected

227

negatively on the performance and volatility of their stock markets; taking into account

that GCC countries are planning to perform a single currency before 2010.

228

References

Abraham, A., Sayyed, F. and Alsakran, A. (2002) Testing the random walk behavior and

efficiency of the Gulf stock markets, The Financial Review 37, 469-80.

Abu Zarour, B. (2005) Non-linearity and market efficiency: evidence from emerging

markets in the Middle East region, Paper Presented at the Middle East and North

African Economies: Past Perspectives and Future Challenges Conference, Co-

organized by the EcoMod Network and the Middle East Economic Association,

Free University of Brussels on June 2-4, 2005.

Abu Zarour, B. (2005) The effect of infrequent trading on market efficiency: the case of

the Middle East stock markets, Paper Presented in the 12th Multinational Finance

Society Conference, 12-19 July, Athens, Greece.

Abu Zarour, B. (2006) Wild oil prices, but brave stock markets! The case of GCC stock

markets, Operational Research an International Journal (ORIJ), forthcoming.

Ackert, L. and Smith, B. (1993) Stock price volatility, ordinary dividends, and other cash

flows to shareholders, Journal of Finance 48, 1147-60.

Affleck-Graves, J. and Mendenhall, R.R. (1990) The relation between the value-line

enigma and post-earnings-announcement drift, unpublished manuscript, college of

Business Administration, University of Notre Dame.

Agrawal, A. and Mandelker, G. (1992) The post-merger performance of Acquiring firms:

a re-examination of an anomaly, Journal of Finance 47, 1605-21.

Agrawal, A. and Tandon, K. (1994) Anomalies or illusions? Evidence from stock markets

in eighteen countries, Journal of International Money and Finance 13, 83-106.

Alexakis, P. and Siriopoulos, C. (1997) The international stock prices of 1997 and the

dynamic relationships between Asian stock markets: linear and non-linear

Granger causality tests, Managerial Finance 25, 22-38.

Alexander, S. (1961) Price movements in speculative markets: trends or random walk,

Industrial Management Review 2, 7-26.

Alexander, S. (1964) Price movements in speculative markets: trends or random walk.

No. 2, in Poul Cootner (ed.) The Random character of stock market prices,

Cambridge: M.I.T., 338-72.

Al-Loughani, N. (1995) Random walk in thinly traded stock markets: the case of Kuwait,

Arab Journal of Administration Science 3, 189-209.

Al-Loughani, N. (2000) The relation between large stock and small stock returns in

Kuwait, Paper Presented at the Seventh Annual Conference of the Economic

Research Forum for the Arab Countries, Iran and Turkey (ERF), Amman, Jordan,

26-29 October.

Al-Loughani, N. (2003) the seasonal characteristics of stock returns in the Kuwaiti stock

market, Journal of Gulf and Arabian Peninsula Studies 29, 15-40.

Al-Loughani, N. and Moosa, I. (1997) Testing the efficiency of an emerging stock market

using trading rules: the case of Kuwait, Working Paper Series Number1. Kuwait,

College of Administrative Science, Kuwait University.

Al-Saad, K. and Moosa, I. (2005) Seasonality in stock returns: evidence from an

emerging market, Applied Financial Economics 15, 63-71.

Aly, H., Mehdian, S. and Perry, M. (2004) An analysis of Day-of-the-week effect in the

Egyptian stock market, International Journal of Business 9, 301-308.

229

Antoniou, A., Ergul, N. and Holmes, P. (1997) Market efficiency, thin trading and non-

linear behavior: evidence from an emerging market, European Financial

Management 3, 175-90.

Arab Monetary Fund (AMF) (2004) Uniform Arabian Economic Report, Abu Dhabi.

Arab Monetary Fund (AMF) (2005) Uniform Arabian Economic Report, Abu Dhabi.

Arab Monetary Fund (AMF) quarterly bulletin, several issues (in Arabic).

Ariel, R. (1985) High stock returns before holidays, MIT working paper.

Ariel, R. (1987) A monthly effect in stock returns, Journal of Financial Economics 18,

161-74.

Arifa, A., Ghali, K. and Limam, I. (2002) Investigating stock market dynamics in an oil-

dependent economy: the case of Kuwait, Economic Policy and Analysis 32, 141-

58.

Arshanapalli, B. and Doukas, J. (1993) International stock market linkages: evidence

from the pre and post-October 1987 period, Journal of Banking and Finance 17,

193-208.

Asquith, P. (1983) Merger bids, uncertainty and stock holder returns, Journal of Financial

Economics 11, 51-83.

Asquith, P. (1983) Merger bids, uncertainty, and stock holders returns, Journal of

Financial Economics 11, 51-83.

Assaf, A. (2003) Transmission of stock price movements: the case of GCC stock markets,

Review of Middle East Economic and Finance 1, 171-89.

Bachellier, L. (1900) trans. James Boness, Theory of speculation, in Cootner (1964), 17-

78.

Bailey, W., Stulz, R. and Yen, S. (1990) Properties of daily stock returns from the Pacific

basin stock markets, evidence and implications, Pacific Basin Capital Markets

Research, 155-71.

Baily, W. and Stulz, R. (1990) Benefits of international diversification: the case of

Pacific Basin stock markets, Journal of Portfolio Management 16, 57-61.

Ball, R. and Brown, P. (1968) An empirical evaluation of accounting income numbers,

Journal of Accounting Research 6, 159-78.

Banz, R. (1981) The relationship between returns and market value of common stocks,

Journal of Financial Economics 9, 3-18.

Barnes, P. (1986) Thin trading and stock market efficiency: the case of Kuala Lumpur

stock exchange, Journal of Business Financial Accounting 3, 609-17.

Basset, G., France, V. and Pliska, S. (1991) Kalmen filter estimation for valuing non-

trading securities, with application to the MMI cash futures spread on October 19

and 20, 1987, Review of Quantitative Finance and Accounting 1, 135-51.

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,

663-82.

Basu, S. (1983) The relationship between earnings yield, market value, and return for

NYSE common stocks: further evidence, Journal of Financial Economics 12, 129-

56.

Bayers, M. and Peel, D. (1993) Some evidence on the interdependence of national stock

markets and the gains from international portfolio diversification, Applied

Financial Economics 3, 239-42.

230

Beechey, M., Gruen, D. and Vickery, J. (2000) The efficient market hypothesis: a survey,

Research Discussion Paper 2000-01, Economic Research Department, Reserve

Bank of Australia.

Bekaert, G. (1995) Market integration and investment barriers in emerging equity

markets, The world bank Economic Review 9, 75-107.

Bekaert, G. and Harvey, C. (1997) Emerging equity market volatility, Journal of

Financial Economics 45, 29-77.

Bekaert, G. and Harvey, C. (2003) Emerging markets finance, Journal of Empirical

Finance 10, 3-55.

Benartzi, S. and Thaler, R. (1992) Myopic loss aversion and the equity premium puzzle,

National Bureau of Economic Research Working Paper No. 4369.

Benartzi, S. Michael, R. and Thaler, R. (1997) Do dividends changes signal the future or

the past?, Journal of Finance 52, 1007-34.

Bernanke, B. S., Gertler, M., Watson, M. (1997) Systematic monetary policy and the

effects of oil prices shocks, Brooking Papers on Economic Activity 1, 91-142.

Bernard, V. L. and Thomas, J. K. (1989) Post-earning-announcement, drift delayed price

response or risk premium?, Journal of Accounting Research 27, 1-36.

Bernard, V. L. and Thomas, J. K. (1990) Evidence that stock prices do not fully reflect

the implications of current earnings for future earnings, Journal of Accounting

and Economics 13, 305-40.

Berument, H. and Ince, O. (2005) Effect of S&P 500’s return on emerging markets:

Turkish experience, Applied Financial Economics Letters 1, 59-64.

Bessembinder, H. and Chain, K. (1995) The profitability of technical trading rules in

Asian stock markets, Pacific Basin Finance Journal 3, 257-84.

Bessler, A. and Yang, J. (2003) The structure of interdependence in international stock

markets, Journal of International Money and Finance 22, 261-87.

Beverdge, S. and Nelson, C. (1981) A new approach to decomposition of economic time

series into permanent and transitory components with attention to measurement of

the business cycle, Journal of Monetary Economics 7, 151-74.

Bhandari, L. (1988) Dept/Equity ratio and expected common stock returns: empirical

evidence, Journal of Finance 43, 507-28.

Bharba, H., Dhillon, U. and Ramirez, G. (1999) A November effect? Revisiting the tax-

loss-selling hypothesis, Financial Management 28, 5-15.

Bhardwaj, R. K. and Brook, L.D. (1992) The January effect anomaly: effects of low share

price, transaction costs, and bid-ask bias, Journal of Finance 47, 553-75.

Black, F. (1973) Yes Virginia, there is hope: tests of the value line ranking system,

Financial Analysts Journal 29, 10-14.

Black, F. (1986) Noise, Journal of Finance 41, 529-43.

Black, M., Jensen, C., and Scholes, M. (1972) The capital asset pricing model: some

empirical tests, In M. Jensen, ed., Studies in the theory of capital markets,

Praeger, New York, NY.

Blanchard, O. J., and Quah, D. (1989) the dynamic effect of aggregate demand and

supply disturbances, American Economic Review 79, 1146-64.

Blume, M. (1968) The assessment of portfolio performance, Unpublished PhD thesis,

University of Chicago.

231

Blume, M. E., and Friend, I. (1973) A new look at the capital asset pricing model, Journal

of Finance 28, 19-33.

Bolbol, A. and Omran, M. (2004) Arab stock markets and capital investment, Arab

Monetary Fund (AMF) Economic Paper No. 8, Abu Dhabi.

Bollerslev, T. (1986) Generalized autoregressive conditional heteroskedasticity, Journal

of Econometrics 31, 307-37.

Bollerslev, T. and Hodrick, R. (1992) Financial market efficiency tests, National Bureau

of Economic Research (NBER) Working Paper No. 4108.

Bollerslev, T., Chou, R. and Kroner, K. (1992) ARCH modeling in finance, Journal of

Economics 52, 5-59.

Bouman, S. and Jacobsen, B. (2002) The Halloween indicator, “Sell in May and go

away”: another puzzle, American Economic Review 92, 1618-35.

Box, G. and Pierce, D. (1970) Distribution of residual autocorrelations in autoregressive-

integrated moving average time series models, Journal of the American Statistical

Association 65, 1509-26.

Branch, B. (1977) A tax loss trading rule, Journal of Business 50, 198-207.

Brauer, G. and Chang, E. (1990) Return seasonality in stocks and their under lying assets:

tax-loss selling versus information explanation, Review of Financial Studies 3,

255-80.

Brav, A. (1997) Inference in long-horizon event studies: a re-evaluation of the evidence,

unpublished working paper, Graduate School of Business, University of Chicago.

Brav, A. and Gompers, P. (1997) Myth or reality? The long-run underperformance of

initial public offerings: evidence from venture and non-venture capital-backed

companies, Journal of Finance 52, 1791-1821.

Brav, A., Geczy, C. and Gompers, P. (1995) The long-run underperformance of seasonal

equity offerings revisited, unpublished working paper, Graduate School of

Business, University of Chicago.

Breeden, D. T. (1979) An intertemporal asset pricing model with stochastic consumption

and investment opportunities, Journal of Financial Economics 7, 265-96.

Brinson, G., Hood, L. R. and Beebower, G. L. (1986) Determinants of portfolio

performance, Financial Analysts Journal 43, 39-44.

Brock, W., Dechert, W. and Scheinkman, J. (1987) A test for independence based on the

correlation dimension, Mimeo, Department of Economics, university of

Wisconsin at Madison, and University of Houston.

Brock, W., Dechert, W. and Scheinkman, J. (1996) A test for independence based on the

correlation dimension, Econometrics Review 15, 197-235.

Brock, W., Hsieh, D. and LeBaron, B. (1991) Nonlinear dynamics, chaos, and instability:

statistical theory and economic evidence, MIT Press, Reading, MA.

Brooks, C. and Henry, O. (2000) Can portmanteau nonlinearity tests serve as general

mis-specification tests? Evidence from symmetric and asymmetric GARCH

models, Economics Letters 67, 245-51.

Brooks, C. and Heravi, S. (1999) The effect of (mis-specified) GARCH filters on the

finite sample distribution of the BDS test, Computational Economics 13, 147-62.

Brown, P., Keim, D., Kleidon, A. and Marsh, T. (1983) Stock return seasonalities and

tax-loss selling hypothesis- analysis of the arguments and Australian evidence,

Journal of Financial Economics 12, 105-27.

232

Buckberg, E. (1995) Emerging stock markets and international asset pricing, The World

Bank Economic Review 9, 51-74.

Bulter, K. and Malaikah, S. (1992) Efficiency and inefficiency in thinly traded stock

markets: Kuwait and Saudi Arabia, Journal of Banking and Finance 16, 197-210.

Butler, K. and Joaquin, D. (2002) Are the gains from international portfolio

diversification exaggerated? The influence of downside risk in bear markets,

Journal of International Money and Finance 21, 981-1011.

Campbell, J. and Hamao, Y. (1992) Predictable stock returns in the United States and

Japan: a study of long-term capital market integration, Journal of Finance 47, 43-

69.

Campbell, J. and Shiller, R. (1987) Co-integration and tests of present value models,

Journal of Political Economy 95, 1062-88.

Campbell, J. Y. and Shiller, R. (1988b) Stock prices, earnings and expected dividends,

Journal of Finance 43, 661-76.

Campbell, J. (1995) Some lessons from the yield curve, Journal of Political Economy

104, 298-345.

Campbell, J., Lo, A. and MaCkinlay, C. (1997) The econometrics of financial markets,

Princeton University Press, Princeton, New Jersey.

Cha, B. and Oh, S. (2000) The relationship between developed equity markets and the

Pacific Basin’s emerging equity markets, International Review of Economics and

Finance 9, 299-322.

Chan, K. (1986) Can tax-loss selling explain the January seasonal in stock returns?,

Journal of Finance 41, 1115-28.

Chan, L. K., Hamao, Y., and Lakonishok, J. (1991) Fundamentals and stock returns in

Japan, Journal of Finance 46, 1739-64.

Chang, E. C. and Lewellen, W. G. (1984) Market timing and market fund investment

performance, Journal of Business 57, 57-72.

Change, E. and Pinegar, M. (1986) Return seasonality and tax-loss selling in the market

for long-term government and corporate bonds, Journal of Financial Economics

17, 391-415.

Chen, N. F. (1983) Some empirical tests of the theory of arbitrage pricing, Journal of

Finance 38, 1393-1414.

Chou, R. (1988) Volatility persistence and stock valuation: some empirical evidence

using GARCH, Journal of Applied Econometrics 3, 279-94.

Claessens, S., Dasgupta, S. and Glen, J. (1995) Return behavior in emerging stock

markets, The World Bank Economic Review 9, 131-51.

Cochran, J. (1991) Volatility tests and efficient markets: a review essay, Journal of

Monetary Economics 27, 463-85.

Cohen, K., Maier, S., Schwartz, R. and Whitcomb, D. (1978) The return generation

process, return variance, and the effect of thinness in securities markets, Journal

of Finance 33, 149-67.

Cohen, K., Maier, S., Schwartz, R. and Whitcomb, D. (1979) On the existence of serial

correlation in an efficient securities markets, TIMS Studies in the Management

Sciences 11, 151-68.

233

Constantinides, G. (1984) Optional stock trading with personal taxes: implications for

prices and the abnormal January returns, Journal of Financial Economics 13, 33-

56.

Constantinides, G. (1989) Theory of valuation: overview and recent developments, in S.

Bhattacharya and G. Constantinides, eds., theory of valuation, Roman and Little

Field, City.

Conway, D. A. and Reinganum, M. R. (1988) Stable factors in security returns:

identification using cross-validation, Journal of Business and Economic Statistics

6, 1-15.

Cootner, P. (ed.) (1964) The random character of stock market prices, MIT Press

Copeland, T. and Mayers, D. (1982) The value line enigma: A case study of performance

evaluation issues, Journal of Financial Economics 10, 289-321.

Cowels, A. 3rd (1933) Can stock market forecasters forecast?, Econometrica 1, 309-24.

Cowels, A. 3rd (1944) Stock market forecasting, Econometrica 12, 206-14.

Cowels, A. 3rd and Jones, H. (1937) Some a posterior: probabilities in stock market

actions, Econometrica 5, 280-94.

Cross, F. (1973) The behavior of stock prices on Fridays and Mondays, Financial

Analysts Journal, November-December, 67-69.

Cusatis, P., Miles, J. and Woolridge, J. (1993) Restructuring through spinoffs, Journal of

Financial Economics 33, 293-433.

Cushman, D. and Zha, T. (1997) Identifying monetary policy in a small open economy

under flexible exchange rates, Journal of Monetary Economy 39, 433-48.

Cuthbertson, K. (1996) Quantitative financial economics: stocks, bonds, and foreign

exchange, John Wiley & Sons, Ltd, England.

Dabbas, R., Smith, K. and Brocato, J. (1991) Test on the rationality of professional

business forecasters with changing forecast horizons, Quarterly Journal of

Business and Economics 30, 28-50.

Dahel, R. (2000) Volatility in Arab stock markets, Arab Planning Institute, Kuwait.

Dahel, R. and Laabas, B. (1999) The behavior of stock prices in the GCC markets,

Economic Research Forum, Working Paper No. 9917, Cairo, Egypt.

Darrat, A., Elkhal, K. and Hakim, S. (2000) On the integration of emerging stock markets

in the Middle East, Journal of Economic Development 25, 119-29.

DeBondt, W. F. M. and Thaler, R. H. (1985) Does the stock market overact?, Journal of

Finance 40, 793-805.

DeBondt, W. F. M. and Thaler, R. H. (1987) Further evidence on investor overreaction

and stock market seasonality, Journal of Finance 42, 557-81.

Desai, H. and Jain, P. (1997) Long-run common stock returns following splits and

reverse splits, Journal of Business 70, 409-33.

Dharan, B. and Ikenberry, D. (1995) The long-run negative drift of post-listing stock

returns, Journal of Finance 50, 1547-1574.

Dhrymes, C., Phoebus, J., Friend, I. and Gultekin, N. B. (1984) New tests of the APT and

their implications, Journal of Finance 40, 659-74.

Dhrymes, C., Phoebus, J., Friend, I. and Gultekin, N. B. (1985) A critical reexamination

of the empirical evidence on the arbitrage pricing theory, Journal of Finance 39,

323-46.

234

Dickey, D. and Fuller, W. (1979) Distribution of the estimators for autoregressive time

series with a unit root, Journal of American Statistical Association 74, 427-31.

Dimson, E. (1979) Risk measurement when shares are subject to infrequent trading,

Journal of Financial Economics 7, 197-226.

Dimson, E. and Mussavian, M. (2000) Market efficiency, The Current State of Business

Disciplines 3, 959-70.

Ding, Z., Granger, C. W. J., and Engle, R. F. (1993) A long memory property of stock

market returns and a new model, Journal of Empirical Finance 1, 83-106.

Divecha, A., Drach, G. and Stefek, D. (1992) Emerging Markets: A Quantitative

perspective, Journal of Portfolio Management 19, 41-50.

Economic and Social Commission for Western Asia (ESCWA) (2004) Survey of

Economic and Social Development In the ESCWA Region.

El-Erian, M. and Kumar, M. (1995) Emerging equity markets in Middle Eastern

countries, IMF Staff Paper 42, 313-43.

Eleswarapu, V. and Reinganum, M. (1993) The seasonal behavior of the liquidity

premium in asset pricing, Journal of Financial Economics 34, 373-86.

Elton, E., Gruber, J., Das, S. and Hklarka, M. (1991) Efficiency with costly information:

a reinterpretation of evidence from managed portfolios, Unpublished manuscript,

New York University.

Engle, F. and Bollerslev, T. (1986) Modeling the persistence of conditional variances,

Econometric Review 5, 1-50.

Engle, R. (1982) Autoregressive conditional heteroskedasticity with estimates of the

variance of United Kingdom inflation, Econometrica 50, 987-1007.

Engle, R. and Granger, C. (1987) Co-integration and error correction representation,

estimation and testing, Econometrica 55, 251-76.

Engle, R.F., Yoo, B. (1987) Forecasting and testing in cointegrated system, Journal of

Econometrics 35, 143-59.

Engle, R., and Ng, V. (1993) Measuring and testing the impact of news on volatility,

Journal of finance 48, 1022-82.

Engle, R., Lilien, D. and Robins, R. (1987) Estimating time risk premia in the term

structure: the ARCH-M model, Econometrica 55, 391-407.

Errunza, V., Hogan, K. and Huang, M. (1999) Can the gains from international

diversification be achieved without trading abroad?, Journal of Finance 54, 2075-

2107.

Eun, C. and Resnick, B. (1984) Estimating the correlation Structure of international share

prices. Journal of Finance 39, 1311-24.

Fama, E. (1965) The behavior of stock market prices, Journal of Business 38, 34-105.

Fama, E. (1970) Efficient capital markets: A review of theory and empirical work,

Journal of Finance 25, 383-417.

Fama, E. (1972) Components of investment performance, The Journal of Finance 27,

551-67.

Fama, E. (1981) Stock returns, real activity, inflation, and money, American Economic

Review 71, 545-65.

Fama, E. (1991) Efficient capital markets: II, The journal of Finance 46, 1575-1617.

Fama, E. (1998) Market efficiency, long term returns and behavioral finance, Journal of

Financial Economics 49, 283-306.

235

Fama, E. and Blume, M. (1966) Filter rules and stock market trading, Journal of Business

39, 226-41.

Fama, E. and French, K. (1992) The cross-section of expected stock returns, Journal of

Finance 47, 427-65.

Fama, E. and French, K. (1996) Multifactor explanations of asset pricing anomalies,

Journal of Finance 51, 55-84.

Fama, E. and French, K. R. (1988a) Permanent and temporary components of stock

prices, Journal of Political Economy 96, 246-73.

Fama, E. and French, K. R. (1988b) Dividend yields and expected stock returns, Journal

of Financial Economics 22, 3-25.

Fama, E. and French, K. R. (1989) Business conditions and expected returns on stocks

and bonds, Journal of Financial Economics 25, 23-49.

Fama, E. and MacBeth, J. (1973) Risk, return and equilibrium: empirical tests, Journal of

Political Economy 81, 607-36.

Fama, E. and Schwert, G. W. (1977) Asset returns and inflation, Journal of Financial

Economics 5, 115-46.

Fama, E., Fisher, L., Jensen, M. and Roll, R. (1969) The adjustment of stock prices to

new information, International Economic Review 10, 1-21.

Fields, M. (1931) Stock prices: A problem in verification, Journal of Business 4, 415-18.

Fields, M. J. (1934) Security prices and stock exchange holidays in relation to shat

selling, Journal of Business 7, 328-38.

Fischer, B. (1986) Noise, Journal of Finance 41, 529-43.

Fisher, L. (1966) Some new stock market indices, Journal of Business 39, 191-225.

Flavin, M. (1983) Excess volatility I the financial markets: a reassessment of the

empirical evidence, Journal of Political Economy 91, 929-56.

Francis, B. and Leachman, L. (1998) Super exogeneity and the dynamic linkages among

international equity markets, Journal of International Money and Finance 17, 475-

92.

Franks, J., Roberts, H. and Titman, S. (1991) The post merger share price performance of

acquiring firms, Journal of Financial Economics 29, 81-96.

Fraser, P. and MacDonald, R. (1993) The efficiency of CAC stock price forecasts: a

survey based perspective, Review Economique 44, 991-1000.

Freimann, E. (1998) Economic integration and country allocation in Europe?, Financial

Analysts Journal, 32-41.

French, K. (1980) Stock returns and the weekend effect, Journal of Financial Economics

8, 55-69.

French, K. R., Kenneth, R. and Roll, R. (1986) Stock return variances: the arrival of

information and the reaction of traders, Journal of Financial Economics 17, 5-26.

Gandhi, D. and Woodward, R. (1980) Thin capital markets: a case study of the Kuwaiti

stock market, Applies Economics 12, 341-49.

Gianni, A. and Giannini, C. (1997) Topics in structural VAR econometrics, Second

Edition, Springer.

Gilmore, C. and McManus (2002) International portfolio diversification: US and central

European equity markets, Emerging Markets Review 3, 69-83.

Girard, E., Omran, M. and Zaher, T. (2003) On risk and return in MENA capital markets,

International Journal of Business 8, 285-313.

236

Granger, C. (1969) Investigating causal relations by econometric models and cross-

spectral methods, Econometrica 37, 424-38.

Granger, C. and Terasvirta, T. (1993) Modeling nonlinear economic relationships,

Oxford University Press, Oxford.

Grieb, T. and Reyes, M. (1999) Random walk tests for Latin American equity indices and

individual firms, Journal of Financial Research 4, 371-83.

Grinblatt, M. and Keloharju, M. (2001) How distance, languages, and culture inference

stockholdings and trades, Journal of Finance 56, 1053-73.

Grinblatt, M. and Titman, S. (1989) Mutual funds performance: an analysis of quarterly

portfolio holdings, Journal of Business 62, 393-416.

Grossman, S. and Stiglitz J. (1980) On the impossibility of informationally efficient

markets, American, Economic Review 70, 393-407.

Gultekin, M. and Gultekin, B. (1983) Stock market seasonality: international evidence,

Journal of Financial Economics 12, 469-81.

Gultekin, M., Gultekin, N. and Penati, A. (1989) Capital controls and international capital

markets segmentation: the evidence from the Japanese and American stock

markets, Journal of Finance 44, 849-69.

Hakim, S. and Neaime, S. (2002) Mean-reversion across MENA stock markets:

implications for portfolio allocations, Working Paper No. 2026, Energetix and

American University of Beirut.

Hamilton, J. and Susmel, R. (1994) Autoregressive conditional heteroskedasticity and

changes in regime, The Journal of Econometrics 64, 307-33.

Hamilton, J. and Lin, G. (1996) Stock market volatility and the business cycle, The

Journal of Applied Econometrics 11, 573-93.

Hammoudeh, S. and Aleisa, E. (2004) Dynamic relationships among GCC stock markets

and NYMEX oil futures, Contemporary Economic Policy 22, 250-69.

Hammoudeh, S. and Choi, K. (2004) Volatility regime-switching and linkages among

GCC stock markets, Paper Presented at the 11th Economic Research Forum

Conference on December 16-18.

Hansen, L. P. (1982) Large sample properties of generalized method of moments

estimators, Econometrica 50, 1029-54.

Hansen, L. P. and Singelton, K. J. (1982) Generalized instrumental variables estimation

in non-linear rational expectations model, Econometrica 50, 1269-86.

Haque, M., Hassan, K. and Varela, O. (2001) Stability, volatility, risk premiums and

predictability in Latin American emerging stock markets, Quarterly Journal of

Business and Economics 40, 23-44.

Haque, M., Hassan, K. and Zaher, T. (2004) Predictability and volatility of Asian

emerging stock markets, Indian Journal of Economics and business 3, 334-55.

Haque, M., Hassan, M., Maroney, N. and Sackley, W. (2004) An empirical examination

of stability, predictability and volatility of Middle Eastern and African emerging

stock markets, Review of Middle East Economics and Finance 2, 19-42.

Harris, L. and Gurel, E. (1986) Price and volume effects association with changes in the

S&P 500 list: new evidence for the existence of price pressures, Journal of

Finance 41, 815-29.

Harvey, C. (1995b) The risk exposure of emerging equity markets, The world Bank

Economic Review 9, 19-50.

237

Harvey, C. (1995c) Predictable risk and returns in emerging markets, The Review of

Financial Studies 8, 773-816.

Harvey, C. R. (1991) The world price of covariance risk, Journal of Finance 46, 111-157.

Hendry, D. and Juselius, K. (2000) Explaining cointegration analysis: part I, Energy

Journal 21, 1-42.

Henriksson, R. T. (1984) Market timing and mutual fund performance: an empirical

investigation, Journal of Business 57, 73-96.

Henry, O. (1998) Modeling the asymmetry of stock market volatility, Applied Financial

Economics 8, 145-53.

Henry, P. B., (2000) Stock market liberalization, economic reform and emerging equity

market prices, Journal of Finance 55, 529-64.

Hillier, D. and Marshall, A. (2002) Insider trading, tax-loss-selling, and the turn-of-the-

year effect, International Review of Financial Analysis 11, 73-84.

Hinich, M. and Patterson, D. (1995) Detecting epochs of transient dependence in white

noise, Unpublished Manuscript, University of Texas at Austin.

Hirshleifer, D. and Shumway, T. (2001) Good day sun shine: stock returns and the

weather. SSRN working paper.

Huagen, R. and Lakonishok, J. (1988) The incredible January effect: the stock market’s

unsolved myster, Dow Jones-Irwin, Homewood, Illinois.

Huberman, G. and Kandel, S. (1987) Value line rank and Firm size, Journal of Business

60, 577-89.

Huberman, G. and Kandel, S. (1990) Market efficiency and value line’s record, Journal

of Business 63, 187-216.

Huberman, G. and Regev, T. (2001) Contagious speculation and a cure for cancer: A

nonevent that made stock prices soar, Journal of Finance 56, 387-96.

Hulbert, M. (1990) Proof of pudding, Forbs, (December 10), 316.

Ikenberry, D. and Lakonishok, J. (1993) Corporate governance through the proxy contest:

evidence and implications, Journal of Business 66, 405-35.

Ikenberry, D., Lakonishok, J. and Vermaelen, T. (1995) Market underreaction to open

market share repurchases, Journal of Financial Economics 39, 181-208.

Ikenberry, D., Rankine, G. and Stice, E. (1996) What do stock splits really signal?,

Journal of Financial and Quantitative Analysis 31, 357-77.

Ippolito, R. A. (1989) Efficiency with costly information: a study of mutual fund

performance, 1965-84, Quarterly Journal of Economics 104, 1-23.

Jaffe, J. and Mandelker, G. (1976) The “Fisher effect” for risky assets: an empirical

investigation, Journal of Finance 31, 447-58.

Jaffe, J. and Westerfield, R. (1989) Is there a monthly effect in stock market returns?,

Journal of Banking and Finance 13, 237-44.

Jaffe, J. F. (1974) Special information and inside trading, Journal of Business 47, 410-28.

Jegadeesh, N. (1990) Evidence of predictable behavior of security returns, Journal of

Finance 45, 881-98.

Jegadeesh, N. and Titman, S. (1993) Returns to buying winners and selling losers:

implication for stock market efficiency, Journal of Finance 48, 65-91.

Jennifer, C. and Kaul, G. (1988) Time-variation in expected returns, Journal of Business

61, 409-25.

238

Jensen, M. (1978) Some anomalies, evidence regarding market efficiency, Journal of

Financial Economics 6, 95-101.

Jensen, M. C. (1968) The performance of mutual funds in the period 1945-64, Journal of

Finance 23, 389-416.

Jensen, M. C. (1969) Risk, the pricing of capital assets, and the evaluation of investment

portfolios, Journal of Business 42, 167-247.

Johansen, S. (1988) Statistical analysis of cointegrating vectors, Journal of Economic

Dynamics and Control 12, 231-54.

Johansen, S. (1991) Estimation and hypothesis testing of cointegration vectors in

Guassian vector autoregressive models, Econometrica 59, 1551-80.

Johansen, S. (1995) Likelihood-based inference in cointegrated vector autoregressive

models, Oxford University Press, Oxford.

Johansen, S. and Juselius, K. (1990) Maximum likelihood estimation and inferences on

cointegration-with applications to the demand for money, Oxford Bulletin of

Economics and Statistics 52, 169-210.

Jokivuolle, E. (1995) Measuring true stock index value in the present of infrequent

trading, The Journal of Financial and Quantitative analysis 30, 455-64.

Kahneman, D. and Tversky, A. (1986) Choices, values, and frames, American

Psychologist 341-350.

Kahneman, D., Tversky, A. (1982) Intuitive predictions: biases and corrective

procedures, Reprinted in Kahneman, Solvic, and Tversky, Judgment under

uncertainty: heuristics and biases, Cambridge University Press. Cambridge,

England.

Kamara, A. (1997) New evidence on the Monday seasonal in stock returns, Journal of

Business 70, 63-84.

Kanas, A. (1998) Linkages between the US and European equity markets: further

evidence from cointegration tests, Applied Financial Economics 8, 607-14.

Kasa, K. (1992) Common stochastic trends in international stock markets, Journal of

Monetary Economics 29, 95-124.

Keim, D. (1983) Size-related anomalies and stock return seasonality- further empirical

evidence, Journal of Financial Economics 12, 13-32.

Kendall, M. G. (1953) The analysis of economic time-series, part I: prices, Journal of the

Royal Statistical Society 96, 11-25.

Kleidon, A. (1986) Variance bounds tests and stock price variation models, Journal of

Political Economy 94, 953-1001.

Koutmos, G. (1996) Modeling the dynamic interdependence of major European stock

markets, Journal of Business Finance and Accounting 23, 975-88.

Kwiatkowski, D., Phillips, P., Schmidt, P. and Shine, Y. (1992) Testing the null

hypothesis of stationarity against the alternative of a unit root, Journal of

Econometrics 54, 159-78.

Lakonishok, J, and Vermaelen, T. (1990) Anomalous price behavior around repurchase

tender offers, Journal of Finance 45, 455-77.

Lakonishok, J. and Smidt, S. (1986) Volume for winners and losers: taxation and other

motives for stock trading, Journal of Finance 41, 951-74.

Lakonishok, J. and Smidt, S. (1987) Are seasonal anomalies real? A ninety-years

perspective, Cornell University Working Paper.

239

Lakonishok, J., Shleifer, A. and Vishny, K. (1994) Contrarian investment, extrapolation,

and risk, Journal of Finance 49, 1451-78.

Lakonishok, J., Shleifer, A., Thaler, R. and Vishny, R. (1991) Window dressing by

pension fund managers, American Economic Review 82, 227-31.

Leeper, E. M. and Gordon, D. B. (1992) In research of the liquidity effect, Journal of

Monetary Economics 29, 341-69.

Lehman, A. and Modest, D. M. (1988) the empirical foundations of the arbitrage pricing

theory, Journal of Financial Economics 21, 213-54.

Lehman, B. N. (1990) Fads, martingale and market efficiency, Quarterly Journal of

economics 105, 1-25.

LeRoy, S. and Porter, R. (1981) The present-value relation: tests based on implied

variance bounds, Econometrica 49, 555-74.

Limam, I. (2003) Is long memory a property of thin stock markets? international

evidence using Arab countries, Review of Middle East Economic and Finance 1,

256-66.

Lintner, J. (1965) Security prices, risk, and maximal gains from diversifications, Journal

of Finance 20, 587-615.

Ljung, G. and Box, G. (1978) On a measure of lack of fit in time series models,

Biometrica 66, 67-72.

Lo, A. and MaCkinlay, A. (1990) An econometric analysis of infrequent trading, Journal

of Econometrics 45, 181-211.

Lo, A. W. and MacKinlay A. C. (1990) When are contrarians profits due to stock market

overreaction?, Review of Financial Studies 3, 175-205.

Lo, W. and MaCkinlay, A. (1988) Stock market prices do not follow random walks:

evidence from a simple specification test, Review of Financial Studies 1, 41-66.

Loughran, T. and Ritter, J. (1995) The new issues puzzle, Journal of Finance 50, 23-51.

Lucas, R. E. (1978) Asset prices in an exchange economy, Econometrica 46, 1429-45.

MacKinnon, J. G. (1991) Critical Values for Cointegration Tests, Chapter 13 in R. F.

Engle and C. W.J. Granger (eds.), Long-run economic relationships: Readings in

cointegration, Oxford University Press.

MacKinnon, James G. (1996) Numerical distribution functions for unit root and

cointegration tests, Journal of Applied Econometrics 11, 601-18.

Maghayereh, A. (2003) Seasonality and January effect anomalies in an emerging capital

market, The Arab Bank Review 5, 25-32.

Mandelbort, B. (1963) The variation of certain speculative prices, Journal of Business 36,

394-419.

Mandelbort, B. (1966) Forecasts of future prices, unbiased markets, and Martingle

models, Journal of Business 39, 242-55.

Maneschiold, P. (2004) Integration between the Baltic and international stock markets,

Unpublished Working Paper, University of Skovde.

Maneschiold, P. (2005) International diversification benefits between US, Turkish and

Egyptian stock markets, Review of Middle East Economic and Finance 3, 115-33.

Mankiw, N., Romer, D. and Shapiro, M. (1985) An unbiased reexamination of stock

market volatility, Journal of Finance 40, 677-87.

Marberly, E. and Pierce, R. (2003) The Halloween effect and Japanese equity prices:

Myth or exploitable anomaly, Asia-Pacific Financial Markets 10, 319-34.

240

Marberly, E. and Pierce, R. (2004) Stock market efficiency withstand another challenge:

solving the “Sell in May/Buy after Halloween” puzzle, Econ Journal Watch 1, 29-

46.

Markowitz, H. (1959) Portfolio selection: efficient diversification of investment, New

York, John Wiley & Sons.

Marsh, T. and Merton, R. (1986) Dividend variability and variance bounds tests for the

rationality of stock market prices, American Economic Review 76, 483-98.

Maxwell, W. F. (1998) The January effect in the corporate bond market: a systematic

examination, Financial Management 27, 18-30.

McLeod, A. and Li, W. (1983) Diagnostic checking ARMA time series models using

Squared-residuals autocorrelations, Journal of Time Series Analysis 4, 269-73.

Meric, G., Lead, R., Ratner, M. and Meric, I. (2001) Co-movements of U.S and Latin

American equity markets before and after the 1987 crash, International Review of

Financial Analysis 10, 219-35.

Meric, I. and Meric, G. (1989) Potentials Gains from international portfolio

diversification and inter-temporal stability and seasonality in international stock

market relationships, Journal of Banking and Finance 13, 627-40.

Merton, R.C. (1973) An intertemporal capital asset pricing model, Econometrica 41, 867-

87.

Michaely, R., Thaler, R. and Womack, K. (1995) Price reactions to dividend initiations

and omissions, Journal of Finance 50, 573-608.

Michaud, R., Bergstrom, G., Frashure, R. and Wolahan, B. (1996) Twenty years of

international equity investing, Journal of Portfolio Mamagement 23, 9-22.

Miller, M., Muthuswamy, J. and Whaley, R. (1994) Mean reversion of Standard and Poor

500 index basis changes: arbitrage-induces or statistical illusion?, Journal of

Finance 49, 479-513.

Mills, T. and Coutts, A. (1995) Calendar effects in the London stock exchange FT-SE

indices, The European Journal of Finance 1, 79-93.

Mills, T., Siriopoulos, C., Markellos, R. and Harizanis, D. (2000) Seasonality in the

Athens stock exchange, Applied Financial Economics 10, 137-42.

Mitchell, M. and Stafford, E. (1997) Managerial decisions and long-term stock price

performance, unpublished working paper, Graduate School of Business,

University of Chicago.

Mitchell, M. L. and Lehn, K. (1990) Do bad bidders become good targets?, Journal of

political Economy 98, 372-98.

Mohd, A. and Hassan, H. (2003) Financial integration of stock markets the Gulf: a

multivariate cointegration analysis, International Journal of Business 8, 1-13.

Mood, A. (1940) The distribution theory of runs, Annals of Mathematical Statistics 11,

367-92.

Moore, A. (1962) A statistical analysis of common stock prices, Unpublished PhD thesis,

Graduate School of Business, University of Chicago.

Moustafa, M. (2004) Testing the weak-form efficiency of the United Emirates stock

market, International Journal of Business 9, 310-25.

Muthuswamy, J. (1990) Nonsynchronous trading and the index autocorrelation problem,

PhD Dissertation, Graduate School of Business, University of Chicago.

241

Naser, K. and Younis, Y. (1997) A review of ten Middle East stock exchanges, Middle

East Business Review 2, 63-89.

Neaime, S. (2002) Liberalization and financial integration of MENA stock markets, A

Paper Prepared for Presentation of the ERF’s 9th Annual Conference on “ Finance

and Banking” which held in Asharja-United Arab Emirates, 26-28 October, 2002.

Neih, C. and Chang, T. (2003) Long-run gains from international equity diversification:

Taiwan’s evidence, 1995-2001, Journal of Economic Integration 18, 530-44.

Nelson, C. R. (1976) Inflation and rates of return on common stocks, Journal of Finance

31, 471-83.

Nelson, D. (1991) Conditional heteroskedasticity in asset returns: a new approach,

Econometrica 59, 347-70.

Niederhoffer, V. and Osborn, M. F. M. (1966) Market making and reversal on the stock

exchange, Journal of the American Statistical Association 61, 897-916.

Ojah, K. and Karemera, J. (1999) Random walks and market efficiency tests of Latin

American emerging equity markets: A revisit, The Financial Review 34, 57-72.

Omet, G., Khasawneh, M. and Khasawneh, J. (2002) Efficiency tests and volatility

effects: evidence from the Jordanian stock market, Applied Economic Letters 9,

817-21.

Omran, M. (2003) Time series analysis of the impact of real interest rates on the stock

market activity and liquidity in Egypt: Co-integration and error correction model

approach, International Journal of Business 8, 359-74.

Omran, M. and Farrar, S. (2002) Testing of weak form efficiency in the Middle East

emerging markets, Unpublished Manuscript.

Osborne, M. F. M. (1959) Brownian motion in the stock market, Operation Research 7,

145-73.

Osborne, M.F.M. (1962) Periodic structure in the Brownian motion of stock prices,

operation research 10, 345-79.

Osterwald-Lenum, M. (1992) A note with quantiles of the asymptotic distribution of the

maximum likelihood cointegration rank statistics, Oxford Bulletin of Economics

and Statistics 54, 461-472.

Panas, E. (1990) The behavior of Athens stock prices, Applied Economics 22, 715-27.

Patterson, D. and Ashley, R. (2000) A nonlinear time series workshop, Kluwer

Academic, London.

Phillips, P. and Perron, P. (1988) Testing for a unit root in time series regression,

Biometra 75, 335-46.

Poterbas, J. and Summers, L. (1988) Mean reversion in stock prices: evidence and

implications, Journal of Financial Economics 22, 27-59.

Pratt, J., Wise, D. and Zeckhauser, R. (1979) Price differences in almost competitive

markets, Quarterly Journal of Economics 93, 189-211.

Rabin, M. and Thaler, R. (2001) Anomalies: risk aversion, Journal of Economic

Perspectives 15, 219-232.

Raj, M. and Thurston, D. (1994) January or April? Tests of turn of the year effect in the

New Zealand stock market, Applied economics letters 1, 81-83.

Rao, D. and Shankaraiah, K. (2003) Stock market efficiency and strategies for developing

the GCC financial markets: a case study of the Bahrain stock market, The Arab

Bank Review 5, 16-21.

242

Reinganum, M. (1983) The anomalous stock market behavior of small firms in January-

empirical tests for tax-loss selling effects, Journal of Financial Economics 12, 89-

104.

Reinganum, R. (1981) Misspecification of the capital asset pricing: empirical anomalies

based on earning yields and market values, Journal of Financial Economics 12,

89-104.

Ritter, J. (1988) The buying and selling behavior of individuals investors at the turn of

the year, The Journal of Finance 43, 701-19.

Ritter, J. and Chopra, N. (1989) Portfolio rebalancing and the turn-of-the-year effect,

Journal of Finance 44, 149-66.

Roberts, H. (1967) Statistical versus clinical prediction of the stock market, unpublished

manuscript, Center for Research in Security Prices, University of Chicago, May.

Roberts, H. V. (1959) Stock market “patterns” and financial analysis: methodological

suggestions, Journal of Finance 14, 1-10.

Roll, R. (1968) the efficient market model applied to U.S. Treasury Bill Rate,

Unpublished PhD thesis, Graduate School of Business, University of Chicago.

Roll, R. (1983) Vas ist das? The turn-of-the-year effect and the return premia of small

firms, Journal of Portfolio Management 9, 18-28.

Roll, R. (1986) The hubris hypothesis and corporate takeovers, Journal of Business 59,

197-216.

Roll, R. (1994) What every CEO should know about scientific progress in economics:

what is known and what remains to be resolved, Financial Management 23, 69-

75.

Roll, R. and Ross S. A. (1984) A critical reexamination of the empirical evidence on the

arbitrage pricing theory: a reply, Journal of Finance 39, 347-50.

Roll, R. and Ross, S. A. (1980) An empirical investigation of the arbitrage pricing theory,

Journal of Finance 35, 1073-1103.

Ross, S. (1977) The determinants o financial structure: the incentive signaling approach,

Bell Journal of Economics 8, 23-40.

Ross, S. A. (1976) The arbitrage theory of capital asset pricing, Journal of Economic

Theory13, 341-60.

Rozeff, M. S. and Kinney, W. R. (1976) Capital market seasonality: the case of stock

returns, Journal of Financial Economics 3, 379-402.

Rubinstein, M. (1976) The valuation of uncertain income streams and the pricing of

options, Bell Journal of Economics and Management Science 7, 407-25.

Said, E. and A. Dickey (1984) Testing for unit roots in autoregressive moving average

models of unknown order, Biometrika 71, 599–607.

Samuelson, P. (1965) Proof that properly anticipated prices fluctuate randomly, Industrial

Management Review 6, 41-49.

Schatzberg, J. and Reiber, R. (1992) Extreme negative information and the market

adjustment process: the case of corporate bankruptcy, Quarterly Journal of

Business and Economics 31, 3-21.

Scholes, M. (1972) The market for securities! Substitution versus price pressure and the

effects of information on share prices, Journal of Business 45, 179-211.

Scholes, M. and Williams, J. (1977a) Estimating Betas from nonsynchronous data,

Journal of Financial Economics 5, 309-27.

243

Scholes, M. and Williams, J. (1977b) Estimating Betas from nonsynchronous data,

working Paper, Graduate School of Business, university of Chicago.

Schwert, G. (1989) Why does stock market volatility change over time?, Journal of

Finance 44, 1115-53.

Sewell, S., Stansell, S., Lee, I. and Below, S. (1996) Using chaos measures to examine

international capital market integration, Applied Financial Economics 6, 91-101.

Seyhun, N. N. (1986) Insiders’ profits, costs of trading, and market efficiency, Journal of

Financial Economics 16, 189-212.

Shachmurove, Y. (2004) An introduction to the special issues on financial markets of the

Middle East, International Journal of Business 9, 211-35.

Sharp, W, F. (1964) Capital asset prices: a theory of market equilibrium under conditions

of risk, Journal of Finance19, 425-42.

Shiller, R. (1981) Use of volatility measures in assessing market efficiency, Journal of

Finance 36, 291-304.

Shiller, R. (1984) Stock prices and social dynamics, Brookings Papers on Economic

Activity 2, 457-98.

Shiller, R. (2002) From efficient market theory to behavioral finance, Cowels Foundation

Discussion Paper No. 1385.

Shiller, R. J. (1984) Stock prices and social dynamics, Brookings papers on Economic

Activity 2, 457-510.

Shleifer, A. (1986) Do demand for stocks slope down, Journal of Finance 41, 579-90.

Shleifer, A. and Summers, L. (1990) The noise trader approach to finance, Journal of

Finance 41, 579-90..

Sias, R. and Starks, L. (1997) Institutions and individuals at the turn-of-the-year, Journal

of Finance 52, 1543-62.

Sims, C.A., and Zha, T. (1999) Error bands for impulse responses, Econometrica 67,

1113-55.

Sims, C. (1980) Macroeconomics and reality, Econometrica 48, 1-48.

Sims, C.A. (1988) Identifying policy effects, in Brayant et al. (eds): Empirical

macroeconomics for interdependent economies, Brooking Institution, 305-321.

Sims, C.A., and Zha, T. (1995) Does monetary policy generate recessions?, Manuscript,

Yale University and Federal Reserve Bank of Atlanta.

Sinquefield, R. (1996) Where are the gains from international diversification?, Financial

Analysts Journal 52, 8-14.

Siriopoulos, C. (1996) Interdependence of national stock exchanges: evidence from the

emerging European markets, Journal of Euro-Asian Management 2, 1-29.

Siriopoulos, C., Tsotsos, R. and Karagianni, S. (2001) The impact of non linearities, thin

trading and regulatory changes in the efficiency of an emerging capital market,

The Journal of Applied Business Research 17,81-92.

Sounders, E. (1993) Stock prices and Wall Street weather, American Economic Review

83, 1337-45.

Stambaugh, R. F. (1986) Discussion, Journal of Finance 41, 601-602.

Stapleton, R. (1999) Some recent developments in capital market theory: a survey,

Spanish Economic Review 1, 1-20.

Steely, J. (2001) A note on information seasonality and the disappearance of the weekend

effect in the UK stock market, Journal of Banking and Finance 25, 1941-56.

244

Stickel, S. (1985) The effect of value line investment rank changes on common stock

prices, Journal of Financial Economics 14, 121-44.

Stiglitz, J. (1981) The allocation role of the stock market: Pareto optimality and

competition, The Journal of Finance 36, 235-51.

Stoll, R. and Whaley, E. (1990b) the dynamic of stock index and stock index futures

returns, Journal of Financial and Quantitative Analysis 25, 441-68.

Summers, L. (1986) Does the stock market rationally reflect fundamental values?,

Journal of Finance 41, 591-601.

Thaler, R. (1987a) Anomalies: the January effect, The Journal of Economic Perspectives

1, 197-201.

Thaler, R. (1987b) Anomalies: seasonal movements in security prices II: weekend, turn

of the month, and intraday effects, The Journal of Economic Perspectives 2, 169-

77.

Trzcinka, C. (1986) On the number of factors in the arbitrage pricing model, Journal of

Finance 41, 347-68.

Urrutia, J. (1995) Tests of random walk and market efficiency for Latin American

emerging markets, Journal of Financial Research 18, 299-309.

Wachtel, S. (1942) Certain observation on seasonal movements in stock prices, Journal of

Business 15, 184-93.

Watts, R. (1973) The information content of dividends, Journal of Business 46, 191-211.

West, K. (1988) Dividend innovations and stock price volatility, Econometrica 56, 37-61.

Wheatly, S. (1988) Some tests of international equity integration, Journal of Financial

Economics 21, 177-212.

Working, H. (1934) A random differences series for use in the analysis of time series,

Journal of the American Statistical association 29, 11-24.

Zellner, A. (1985) Bayesian econometrics, Econometrica 53, 253-69.

Zellner, A. and Palm, F. (1974) Time series analysis and simultaneous equation

econometric models, Journal of Econometrics 2, 17-54.

Zha, T. (1999) Block recursion and structural vector autoregression, Journal of

Econometrics 90, 291-316.

Electronic References

Arab Monetary Fund (AMF): www.amf.org.ae

Bahraini Stock Market: www.bahrainstock.com

Dubai Financial Market: www.dfm.co.ae

Egyptian Stock market: www.egyptse.com

Jordanian Stock Market: www.ase.com.jo

Kuwaiti Stock Market: www.kuwaitse.com

Omani Stock Market: www.omanse.com

Palestinian Stock Exchange: www.p-s-e.com

Saudi Arabian Stock market: www.tadawul.com.sa

245

Appendixes

246

Appendix 1

Graph Plots for Each Arab Stock Market Index for both Levels and Returns on Daily Basis.

A: Indices Logarithm

8.8 7.5 8.0

8.6

7.4 7.8

8.4

7.3 7.6

8.2

7.8

7.1 7.2

7.6

100 200 300 400 500 600

100 200 300 400 500 600 500 1000 1500 2000 2500 3000

8.8 5.1 8.6

5.0 8.4

8.4

4.9

8.2

8.0 4.8

8.0

4.7

7.6 7.8

4.6

7.6

4.5

7.2

4.4 7.4

500 1000 1500 2000 2500 3000 250 500 750 1000 500 1000 1500

5.8 8.0

5.8

5.6

5.6 7.6

5.4

5.4

7.2

5.2

5.2

5.0 6.8

4.8 5.0

6.4

4.6 4.8

4.4 6.0

4.6

250 500 750 1000 250 500 750 1000 1250 1500

500 1000 1500 2000 2500 3000

247

…continue appendix 1

B: Indices Returns

.06 .03 .3

.04 .02 .2

.02 .01 .1

.00 .00 .0

100 200 300 400 500 600 100 200 300 400 500 600 500 1000 1500 2000 2500 3000

.2 .25 .16

.20 .12

.1 .08

.15

.04

.10

.00

.0

.05

-.04

.00

-.08

-.1

-.05 -.12

-.10 -.16

-.2

250 500 750 1000 500 1000 1500

500 1000 1500 2000 2500 3000

.3 .06 .20

.2 .04 .16

.12

.1 .02

.08

.0 .00 .04

-.02 .00

-.1

-.04

-.2 -.04

-.08

250 500 750 1000 500 1000 1500 2000 2500 3000 250 500 750 1000 1250 1500

248

Appendix 2

Diagnostic Tools for both Random Walk and GARCH Models

Table 1

Diagnostic Tools for Reseduals of the RW Model Rt=α+εt for the Observed Indices

Abudhabi Bahrain Dubai

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 97.51 0.000 10 811.12 0.000 10 0.53 1.000

20 105.76 0.000 20 811.19 0.000 20 0.63 1.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 13.26 0.001 2 60.10 0.000 2 1.51 0.470

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 14.90 0.000 1 810.10 0.000 1 0.01 0.918

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 12.90 0.000 1 56.58 0.000 1 0.06 0.813

5 31.50 0.000 5 79.08 0.000 5 1.85 0.869

15 49.92 0.000 15 104.35 0.000 15 6.84 0.962

25 79.82 0.000 25 108.69 0.000 25 9.76 0.997

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

639.65 0.000 18321581 0.000 1850786 0.000

SC -7.600903 SC -7.008137 SC -6.350289

249

… continue table 1

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 256.62 0.000 10 1079.20 0.000 10 143.59 0.000

20 269.36 0.000 20 1239.40 0.000 20 156.71 0.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 76.19 0.000 2 179.87 0.000 2 20.48 0.000

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 213.76 0.000 1 396.16 0.000 1 41.12 0.000

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 70.74 0.000 1 166.65 0.000 1 18.92 0.000

5 74.08 0.000 5 169.95 0.000 5 22.97 0.000

15 85.41 0.000 15 185.98 0.000 15 38.24 0.001

25 96.51 0.000 25 224.89 0.000 25 44.68 0.009

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

10650.5 0.000 2940.03 0.000 466.39

SC -5.347986 SC -6.988481 SC -6.28863

250

… continue table 1

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5

ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000

0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000

0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000

0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000

0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R 2

Prob. McLeod-Li test lags Obs*R2 Prob.

10 448.27 0.000 10 319.44 0.000 10 774.98 0.000

20 906.29 0.000 20 319.58 0.000 20 794.87 0.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 23.23 0.000 2 7.42 0.025 2 11.15 0.004

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R 2

Prob. ARCH test lags Obs*R2 Prob.

1 442.87 0.000 1 314.33 0.000 1 754.85 0.000

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 9.49 0.002 1 0.32 0.571 1 3.42 0.065

5 43.59 0.000 5 15.78 0.007 5 28.46 0.000

15 134.70 0.000 15 47.06 0.000 15 42.66 0.000

25 142.15 0.000 25 68.35 0.000 25 68.85 0.000

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

182045 0.000 244349 0.000 1058905 0.000

SC -6.20729 SC -5.15100 SC -6.506318

Notes:The reseduals of the RW model for the observed indices are under investigation in this part. The BDS test results were not altered and only P-values are reported.LM test stands for Breusch-

Godfrey LM test statistic with associated P -values and tests for serial correlations in the reseduals.ARCH test is a lagrange multiplier (LM) test for autoregressive conditional

heteroskedasticity(ARCH) in the residuals proposed by Engle 1982. Q -stattistic is the Ljung-Box Q -statistics and their P -values, The Q -statistic at lag K is a test statistic for the null hyputhesis

that there is no autocorrelation up to order K . J.B is Jarque-Bera statistic for testing normality, SC is the Schwartz criterion for random walk model.

251

Table 2

Diagnostic Tools for Standarised Reseduals of the GARCH(1,1) Model for Daily Returns of the Observed Indices

Abudhabi Bahrain Dubai

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.072 0.145 0.536 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.189 0.126 0.651 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.036 0.028 0.424 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.008 0.007 0.282 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob.

10 3.38 0.971 10 0.76 1.000 10 0.47 1.000

20 20.54 0.425 20 0.78 1.000 20 0.57 1.000

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 0.00 0.946 1 0.73 0.392 1 0.01 0.921

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.49 0.485 1 6.60 0.010 1 0.04 0.842

5 14.32 0.014 5 21.43 0.001 5 1.53 0.910

15 43.49 0.000 15 43.34 0.000 15 6.73 0.965

25 71.70 0.000 25 52.97 0.001 25 9.61 0.998

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

1102.8 0.000 2E+07 0.000 1820531 0.000

252

… continue table 2

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.024 0.017 0.005 2 0.045 0.050 0.605 2 0.809 0.684 0.902

3 0.000 0.000 0.001 3 0.001 0.004 0.291 3 0.943 0.850 0.758

4 0.000 0.000 0.001 4 0.000 0.004 0.326 4 0.936 0.808 0.790

5 0.000 0.000 0.001 5 0.001 0.012 0.433 5 0.770 0.794 0.897

McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob.

10 13.94 0.176 10 12.40 0.259 10 5.48 0.857

20 53.76 0.000 20 17.45 0.624 20 9.85 0.971

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 2.58 0.108 1 0.35 0.552 1 0.00 0.997

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 6.55 0.010 1 8.51 0.004 1 1.82 0.177

5 17.07 0.004 5 9.79 0.081 5 3.32 0.651

15 29.30 0.015 15 21.86 0.111 15 14.14 0.515

25 34.43 0.099 25 40.48 0.026 25 17.69 0.855

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

2620 0.000 656.84 0.000 193.87 0.000

253

… continue table 2

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.008 0.053 2 0.000 0.028 0.694 2 0.784 0.903 0.475

3 0.000 0.001 0.095 3 0.000 0.011 0.320 3 0.266 0.557 0.335

4 0.000 0.001 0.215 4 0.000 0.034 0.051 4 0.024 0.189 0.591

5 0.000 0.004 0.542 5 0.000 0.057 0.025 5 0.003 0.119 0.652

McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob.

10 3.06 0.980 10 6.93 0.732 10 5.06 0.887

20 11.57 0.930 20 13.32 0.863 20 16.44 0.689

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 1.02 0.312 1 0.02 0.885 1 0.19 0.663

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.40 0.528 1 8.70 0.003 1 6.15 0.013

5 16.65 0.005 5 19.22 0.002 5 26.78 0.000

15 38.65 0.001 15 45.45 0.000 15 63.04 0.000

25 44.72 0.009 25 70.22 0.000 25 73.32 0.000

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

37525 0.000 1308.27 0.000 6469.8 0.000

Notes:The reseduals of the GARCH(1,1) model for the observed indices with oil returns as a regressor are under investigation in this part. The BDS test results were not altered and only P-values

are reported.McLeod-Li test examen the serial correlation of the squered reseduals .ARCH test is a lagrange multiplier (LM) test for autoregressive conditional heteroskedasticity(ARCH) in the

residuals proposed by Engle 1982. Q -stattistic is the Ljung-Box Q -statistics and their P -values, The Q -statistic at lag K is a test statistic for the null hyputhesis that there is no autocorrelation up

to order K in the standarized resedual . J.B is Jarque-Bera statistic for testing normality of the standarized reseduals, SC is Schwartz criterion for GARCH model.

254

Table 3

Diagnostic Tools for Reseduals of the EGARCH(1,1) Model for Daily Returns of the Observed Indices

Abudhabi Bahrain Dubai

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.227 0.468 0.892 2 0.279 0.162 0.958 2 0.004 0.214 0.322

3 0.805 0.226 0.365 3 0.291 0.081 0.721 3 0.000 0.073 0.522

4 0.817 0.174 0.150 4 0.811 0.159 0.777 4 0.000 0.055 0.159

5 0.887 0.164 0.072 5 0.404 0.335 0.895 5 0.000 0.002 0.730

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 0.00 0.998 1 26.16 0.000 1 0.01 0.905

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.32 0.569 1 31.40 0.000 1 1.98 0.160

5 14.67 0.012 5 58.71 0.000 5 3.43 0.633

15 41.34 0.000 15 80.89 0.000 15 7.47 0.943

25 68.91 0.000 25 89.92 0.000 25 10.90 0.993

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

1108.82 0.000 1.8E+07 0.000 875911 0.000

255

… continue table 3

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.166 0.087 0.014 2 0.796 0.707 0.960 2 0.535 0.117 0.117

3 0.157 0.085 0.013 3 0.362 0.865 0.932 3 0.938 0.289 0.289

4 0.335 0.083 0.014 4 0.217 0.799 0.589 4 0.536 0.393 0.393

5 0.892 0.428 0.055 5 0.131 0.706 0.372 5 0.432 0.343 0.343

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 3.34 0.068 1 0.51 0.475 1 0.52 0.471

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 4.11 0.043 1 6.75 0.009 1 0.06 0.806

5 13.75 0.017 5 7.86 0.164 5 1.55 0.908

15 26.04 0.038 15 20.28 0.162 15 12.96 0.606

25 32.44 0.146 25 39.86 0.030 25 18.24 0.832

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

2279.33 0.000 660.36 0.000 187.16 0.000

256

… continue table 3

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.021 0.299 0.762 2 0.000 0.059 0.100 2 0.000 0.000 0.186

3 0.000 0.248 0.876 3 0.000 0.010 0.873 3 0.000 0.000 0.086

4 0.000 0.646 0.987 4 0.000 0.001 0.395 4 0.000 0.000 0.112

5 0.000 0.518 0.894 5 0.000 0.007 0.571 5 0.910 0.000 0.172

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 0.33 0.565 1 0.99 0.320 1 7.66 0.006

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.33 0.565 1 3.09 0.079 1 11.19 0.001

5 1.05 0.959 5 11.19 0.048 5 34.96 0.000

15 80.34 0.000 15 35.55 0.002 15 65.74 0.000

25 83.96 0.000 25 57.28 0.000 25 75.01 0.000

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

34365 0.000 1784.4 0.000 9971.33 0.000

Notes:The standarised reseduals of the EGARCH(1,1) model for the observed indices are under investigation in this part. The BDS test results were not altered and only P-values are reported.LM

test stands for Breusch-Godfrey LM test statistic with associated P -values and tests for serial correlations in the reseduals.ARCH test is a lagrange multiplier (LM) test for autoregressive

conditional heteroskedasticity(ARCH) in the residuals proposed by Engle 1982. Q -stattistic is the Ljung-Box Q -statistics and their P -values, The Q -statistic at lag K is a test statistic for the

null hyputhesis that there is no autocorrelation up to order K . J.B is Jarque-Bera statistic for testing normality.

257

Table 4

Diagnostic Tools for Standarised Reseduals of the GARCH(1,1) Model for Daily Returns of GCC Markets with Oil Prices

Abudhabi Bahrain Dubai

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.064 0.213 0.767 2 0.000 0.000 0.000 2 0.000 0.000 0.013

3 0.160 0.170 0.899 3 0.000 0.000 0.000 3 0.000 0.000 0.026

4 0.011 0.022 0.566 4 0.000 0.000 0.000 4 0.000 0.000 0.024

5 0.002 0.005 0.417 5 0.000 0.000 0.000 5 0.000 0.000 0.007

McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob.

10 6.77 0.747 10 0.61 1.000 10 0.07 1.000

20 21.82 0.351 20 0.64 1.000 20 0.18 1.000

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 0.03 0.863 1 0.00 0.950

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.39 0.535 1 90.57 0.000 1 0.25 0.615

5 12.33 0.031 5 140.66 0.000 5 1.49 0.915

15 40.73 0.000 15 203.04 0.000 15 4.36 0.996

25 67.12 0.000 25 215.33 0.000 25 6.61 1.000

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

720.89 0.000 1.5E+07 0.000 1193971 0.000

258

… conyinue table 4

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.001 0.032 2 0.676 0.466 0.774 2 0.131 0.239 0.835

3 0.000 0.000 0.035 3 0.864 0.548 0.859 3 0.008 0.060 0.841

4 0.000 0.000 0.051 4 0.988 0.535 0.869 4 0.000 0.009 0.989

5 0.000 0.000 0.172 5 0.765 0.477 0.999 5 0.000 0.003 0.862

McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob. McLeod-Li test lags Q-stat. Prob.

10 3.69 0.960 10 5.20 0.877 10 3.77 0.957

20 53.63 0.000 20 9.97 0.969 20 13.82 0.839

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 2.00 0.157 1 0.02 0.901 1 0.50 0.479

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.57 0.449 1 2.06 0.152 1 7.61 0.006

5 13.05 0.023 5 3.89 0.565 5 25.05 0.000

15 35.42 0.002 15 12.82 0.616 15 52.77 0.000

25 42.32 0.017 25 15.13 0.938 25 63.65 0.000

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

33194.7 0.000 198 0.000 4275 0.000

Notes:The reseduals of the GARCH(1,1) model for the observed indices with oil returns as a regressor are under investigation in this part. The BDS test results were not altered and only P-values

are reported.McLeod-Li test examen the serial correlation of the squered reseduals .ARCH test is a lagrange multiplier (LM) test for autoregressive conditional heteroskedasticity(ARCH) in the

residuals proposed by Engle 1982. Q -stattistic is the Ljung-Box Q -statistics and their P -values, The Q -statistic at lag K is a test statistic for the null hyputhesis that there is no autocorrelation up

to order K in the standarized resedual . J.B is Jarque-Bera statistic for testing normality of the standarized reseduals.

259

Table 5

adj

DiagnosticTools for reseduals from RW Model Rt =α+εt for the Corrected Indices

Abudhabi Bahrain Dubai

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 92.98 0.000 10 829.06 0.000 10 0.52 1.000

20 101.90 0.000 20 829.14 0.000 20 0.62 1.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 3.03 0.220 2 2.95 0.229 2 2.43 0.297

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 11.90 0.001 1 818.77 0.000 1 0.02 0.899

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 2.18 0.140 1 0.75 0.386 1 0.15 0.696

5 21.30 0.001 5 21.03 0.001 5 3.08 0.688

15 38.39 0.001 15 51.06 0.000 15 7.43 0.945

25 64.94 0.000 25 54.31 0.001 25 10.27 0.996

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

599.088 0.000 1.2E+07 0.000 1913893 0.000

260

… continue table 5

Egypt Jordan Kuwait

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 377.92 0.000 10 859.68 0.000 10 137.55 0.000

20 391.56 0.000 20 991.12 0.000 20 151.28 0.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 9.88 0.007 2 16.80 0.000 2 3.21 0.201

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 323.77 0.000 1 196.42 0.000 1 26.74 0.000

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.36 0.550 1 0.50 0.478 1 0.02 0.894

5 11.65 0.040 5 19.91 0.001 5 5.36 0.374

15 22.84 0.088 15 36.76 0.001 15 20.92 0.139

25 28.47 0.286 25 60.93 0.000 25 26.75 0.369

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

10574 0.000 4468.39 0.000 397.5 0.000

261

… continue table 5

Oman Palestine Saudi

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 394.22 0.000 10 273.94 0.000 10 768.63 0.000

20 784.10 0.000 20 274.12 0.000 20 789.14 0.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 187.80 0.000 2 23.34 0.000 2 124.41 0.000

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 148.29 0.000 1 257.14 0.000 1 746.45 0.000

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 174.03 0.000 1 22.72 0.000 1 83.80 0.000

5 199.48 0.000 5 33.11 0.000 5 116.10 0.000

15 257.12 0.000 15 71.16 0.000 15 127.32 0.000

25 271.03 0.000 25 103.50 0.000 25 156.46 0.000

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

54379.66 0.000 115656 0.000 1193031 0.000

Notes:The reseduals of the RW model for the corrected indices are under investigation in this part. The BDS test results were not altered and only P-values are

reported.LM test stands for Breusch-Godfrey LM test statistic with associated P -values and tests for serial correlations in the reseduals.ARCH test is a lagrange

multiplier (LM) test for autoregressive conditional heteroskedasticity(ARCH) in the residuals proposed by Engle 1982. Q -stattistic is the Ljung-Box Q -statistics

and their P -values, The Q -statistic at lag K is a test statistic for the null hyputhesis that there is no autocorrelation up to order K . J.B is Jarque-Bera statistic for

testing normality.

262

Table 6

2 3

Diagnostic Tools for Reseduals from Non-linear model for Observed Indices Rt = a0 + a1Rt-1 + a2R t-1+ a3R t-1+ εt

Abudhabi Bahrain Dubai

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.013 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 63.73 0.000 10 0.02 1.000 10 0.45 1.000

20 73.20 0.000 20 0.05 1.000 20 0.54 1.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 8.61 0.013 2 34.62 0.000 2 2.29 0.319

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 2.07 0.150 1 0.00 0.960 1 0.00 0.987

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.05 0.823 1 0.55 0.457 1 0.02 0.892

5 20.17 0.001 5 13.20 0.022 5 1.54 0.909

15 35.65 0.002 15 36.47 0.002 15 6.02 0.979

25 59.17 0.000 25 47.83 0.004 25 8.93 0.999

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

760.99 0.000 1.6E+07 0.000 1995149 0.000

263

… continue table 6

Egypt Jordan Kuwait

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 58.74 0.000 10 1128.60 0.000 10 132.33 0.000

20 81.15 0.000 20 1299.00 0.000 20 144.35 0.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 15.00 0.001 2 34.48 0.000 2 2.39 0.303

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1.00 32.25 0.000 1 337.68 0.000 1 32.31 0.000

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.29 0.592 1 0.16 0.691 1 0.04 0.834

5 15.48 0.009 5 18.83 0.002 5 4.82 0.439

15 28.33 0.020 15 35.29 0.002 15 20.34 0.159

25 35.21 0.085 25 58.93 0.000 25 26.03 0.406

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

7798.5 0.000 3174.75 0.000 435.78 0.000

264

… continue table 6

Oman Palestine Saudi

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 263.45 0.000 10 418.57 0.000 10 69.98 0.000

20 670.29 0.000 20 418.65 0.000 20 81.53 0.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 73.49 0.000 2 462.75 0.000 2 9.36 0.009

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 228.01 0.000 1 343.61 0.000 1 22.52 0.000

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 3.09 0.079 1 253.92 0.000 1 0.11 0.738

5 25.39 0.000 5 328.15 0.000 5 25.06 0.000

15 85.95 0.000 15 338.97 0.000 15 33.76 0.004

25 91.72 0.000 25 342.57 0.000 25 62.65 0.000

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

161117 0.000 1734287 0.000 501184 0.000

Notes:The reseduals of the non-linear model for the observed indices are under investigation in this part. The BDS test results were not altered and only P-

values are reported.LM test stands for Breusch-Godfrey LM test statistic with associated P -values and tests for serial correlations in the reseduals.ARCH test

is a lagrange multiplier (LM) test for autoregressive conditional heteroskedasticity(ARCH) in the residuals proposed by Engle 1982. Q -stattistic is the Ljung-

Box Q -statistics and their P -values, The Q -statistic at lag K is a test statistic for the null hyputhesis that there is no autocorrelation up to order K . J.B is

Jarque-Bera statistic for testing normality.

265

Table 7

adj adj 2adj 3adj

Diagnostic Tools for Reseduals from Non-linear Model for Corrected Indices R t = a0 + a1 R t-1 + a2 R t-1+ a3 R t-1+ εt

Abudhabi Bahrain Dubai

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.021 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 65.09 0.000 10 0.78 1.000 10 0.46 1.000

20 74.70 0.000 20 0.81 1.000 20 0.56 1.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 6.69 0.035 2 2.72 0.257 2 2.73 0.255

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 2.18 0.140 1 0.02 0.900 1 0.00 0.994

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.07 0.795 1 0.13 0.723 1 0.02 0.880

5 20.99 0.001 5 24.29 0.000 5 2.41 0.790

15 38.21 0.001 15 47.92 0.000 15 6.36 0.973

25 61.46 0.000 25 58.64 0.000 25 9.44 0.998

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

708.47 0.000 1.4E+07 0.000 2050188 0.000

266

… continue table 7

Egypt Jordan Kuwait

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 73.30 0.000 10 974.94 0.000 10 144.89 0.000

20 93.65 0.000 20 1120.50 0.000 20 158.53 0.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 12.72 0.002 2 32.27 0.000 2 4.27 0.119

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1.00 44.70 0.000 1 242.93 0.000 1 30.90 0.000

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.00 0.992 1 0.06 0.812 1 0.00 0.973

5 13.53 0.019 5 20.63 0.001 5 5.98 0.308

15 26.56 0.033 15 38.09 0.001 15 22.91 0.086

25 32.41 0.146 25 60.96 0.000 25 28.44 0.288

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

8305.89 0.000 4012.9 0.000 390.22 0.000

267

… continue table 7

Oman Palestine Saudi

BDS test BDS test BDS test

m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2 m ε =0.5 ε =1 ε =2

2 0.000 0.000 0.000 2 0.000 0.000 0.000 2 0.000 0.000 0.000

3 0.000 0.000 0.000 3 0.000 0.000 0.000 3 0.000 0.000 0.000

4 0.000 0.000 0.000 4 0.000 0.000 0.000 4 0.000 0.000 0.000

5 0.000 0.000 0.000 5 0.000 0.000 0.000 5 0.000 0.000 0.000

McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob. McLeod-Li test lags Obs*R2 Prob.

10 156.86 0.000 10 17.92 0.056 10 64.67 0.000

20 484.70 0.000 20 18.21 0.574 20 74.37 0.000

LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob. LM test lags Obs*R2 Prob.

2 97.80 0.000 2 7.15 0.028 2 16.37 0.000

ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob. ARCH test lags Obs*R2 Prob.

1 22.59 0.000 1 0.03 0.864 1 20.03 0.000

Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob. Serial correl. lags Q-stat. Prob.

1 0.46 0.497 1 0.48 0.487 1 0.07 0.798

5 61.46 0.000 5 13.05 0.023 5 26.92 0.000

15 98.55 0.000 15 31.75 0.007 15 34.93 0.003

25 106.02 0.000 25 48.81 0.003 25 63.84 0.000

Normality test J.B Prob. Normality test J.B Prob. Normality test J.B Prob.

94342 0.000 207284 0.000 566558 0.000

Notes:The reseduals of the non-linear model for the corrected indices are under investigation in this part. The BDS test results were not altered and only P-

values are reported.LM test stands for Breusch-Godfrey LM test statistic with associated P -values and tests for serial correlations in the reseduals.ARCH test

is a lagrange multiplier (LM) test for autoregressive conditional heteroskedasticity(ARCH) in the residuals proposed by Engle 1982. Q -stattistic is the Ljung-

Box Q -statistics and their P -values, The Q -statistic at lag K is a test statistic for the null hyputhesis that there is no autocorrelation up to order K . J.B is

Jarque-Bera statistic for testing normality.

268

Figure 1

ˆ + βˆ log σ t2−1 + aˆ Ζt −1 + γˆΖt −1

log σ t2 = ω ( )

0.005 0.0020

14

.04

news impact curve Dubai

0.004 news impact curve Jordan 12

news impact curve Palestine

news impact curve Egypt 0.0015

.03 10

0.003

Volatility

8

SIG 2

SIG 2

0.0010

SIG2

.02

0.002 6

0.0005 .01 4

0.001

2

0.000 0.0000 .00

0

-10 -5 0 5 -10 -5 0 5 -12 -8 -4 0 4 8 12

-10 -5 0 5

Z

Z Z

0.0008

0.012

0.005 0.0010

news impact curve Abudhabi 0.010 news impact curve Kuwait

0.0006

0.004 0.0008

news impact curve Saudi

0.008 news impact curve Oman

SIG 2

0.0004 0.0006

SIG2

0.006 0.003

SI G 2

SIG 2

0.004 0.002 0.0004

0.0002

0.002

0.001 0.0002

0.0000

0.000

-10 -5 0 5

-10 -5 0 5 0.000 0.0000

-10 -5 0 5 -10 -5 0 5

Z

Z

Z Z

.009 0.0020

0.00020

0.0004

.008

.007 news impact curve Israel 0.0015 news imoact curve Japan 0.00016

.006 0.0003 news impact curve UK news impact curve Bahrain

.005 0.00012

SIG 2

SIG 2

0.0010

SIG 2

SIG 2

.004 0.0002

.003 0.00008

0.0005

.002 0.0001

0.00004

.001

.000 0.0000

-5 0 5 0.0000 0.00000

-12 -8 -4 0 4 8 12

-5 0 5 -10 -5 0 5

Z Z

Z Z

0.020 0.0006

0.008

0.0005

0.015 news impact curve USA

news impact curve Turkey 0.006 news impact curve India

0.0004

SIG2

SIG2

SIG2

0.010 0.0003 0.004

0.0002

0.005 0.002

0.0001

-10 -5 0 5 -10 -5 0 5

-10 -5 0 5

Z Z

269

Appendix 3

Table 1

Market Returns for Arab Stock Markets to Structural

Inovvations in International Markets

By innovations in

Market Horizon

USA UK Japan

explained (days)

Bahrain 2 0.212 0.989 0.328

6 0.422 1.193 0.372

10 0.549 1.428 0.373

6 0.887 0.172 0.030

10 0.887 0.173 0.030

6 2.380 0.892 0.503

10 2.380 1.685 0.646

6 1.283 2.098 0.275

10 1.381 2.250 0.314

6 0.981 0.697 0.020

10 0.982 0.697 0.020

6 0.594 0.797 1.504

10 0.601 0.797 1.534

6 0.732 0.325 3.507

10 0.743 0.753 3.584

6 0.609 0.811 0.503

10 0.611 1.500 0.504

6 0.074 0.211 0.073

10 0.214 0.240 0.073

Entries in each cell are the percentage of forcast error variance of the market return in

the first column explained by the market in the first row.

Factorization: Structural Decomposition.

270

Table 2

Markets

Period USA UK Japan USA UK Japan USA UK Japan

1 0.0235 -0.0438 -0.0301 -0.0943 -0.0059 -0.0123 -0.1454 -0.0074 0.0036

2 0.0290 -0.0721 -0.0289 -0.1255 0.0150 -0.0213 -0.1164 -0.0419 -0.0238

3 0.0219 -0.0834 -0.0210 -0.1419 0.0395 -0.0320 -0.1237 -0.0508 -0.0458

4 0.0219 -0.0648 -0.0263 -0.1518 0.0500 -0.0340 -0.1297 -0.0036 -0.0083

5 -0.0013 -0.0528 -0.0280 -0.1361 0.0790 -0.0329 -0.1005 0.0338 0.0380

6 -0.0023 -0.0514 -0.0348 -0.1392 0.0817 -0.0341 -0.0797 0.0950 0.0256

7 0.0060 -0.0440 -0.0363 -0.1404 0.0848 -0.0346 -0.0784 0.1437 0.0622

8 0.0117 -0.0192 -0.0364 -0.1413 0.0848 -0.0347 -0.0796 0.1333 0.0709

9 0.0212 -0.0147 -0.0372 -0.1403 0.0880 -0.0349 -0.0796 0.2072 0.0703

10 0.0347 -0.0147 -0.0375 -0.1406 0.0887 -0.0350 -0.0783 0.2113 0.0731

Period USA UK Japan USA UK Japan USA UK Japan

1 0.0747 -0.0692 0.0044 0.0007 -0.0427 0.0377 0.1152 0.0778 0.3035

2 0.1142 -0.0363 0.0087 0.0004 -0.0519 0.0618 0.1581 0.1220 0.3405

3 0.1430 -0.0024 -0.0044 0.0289 -0.0313 0.0554 0.1323 0.1502 0.3637

4 0.1040 -0.0109 -0.0045 0.0592 -0.0282 0.0067 0.1445 0.1513 0.4108

5 0.0800 -0.0175 -0.0037 0.0615 -0.0298 0.0063 0.1631 0.1629 0.4262

6 0.0829 -0.0208 -0.0037 0.0645 -0.0301 0.0088 0.1004 0.1642 0.4399

7 0.0862 -0.0192 -0.0038 0.0691 -0.0303 -0.0005 0.0837 0.0898 0.3940

8 0.0862 -0.0183 -0.0038 0.0700 -0.0306 -0.0022 0.0862 0.1132 0.3833

9 0.0854 -0.0180 -0.0038 0.0703 -0.0302 -0.0016 0.0892 0.0378 0.3864

10 0.0850 -0.0182 -0.0038 0.0710 -0.0301 -0.0025 0.0858 0.0260 0.3831

Period USA UK Japan USA UK Japan USA UK Japan

1 -0.0168 0.0101 -0.0143 -0.1002 -0.1314 0.0039 0.0461 -0.0699 0.0886

2 -0.0216 0.0315 0.0004 -0.1162 -0.1251 0.0023 0.0752 -0.1392 0.1108

3 -0.0273 0.0480 0.0012 -0.1177 -0.1446 -0.0011 0.0081 -0.2088 0.0676

4 -0.0195 0.0659 -0.0004 -0.1252 -0.1725 -0.0110 -0.1065 -0.2088 0.1504

5 -0.0160 0.0641 -0.0007 -0.1570 -0.1621 -0.0551 -0.0998 -0.2714 0.1492

6 -0.0111 0.0730 -0.0006 -0.1617 -0.1675 -0.0360 -0.1071 -0.1766 0.1512

7 0.0031 0.0855 -0.0006 -0.1693 -0.1741 -0.0185 -0.1135 -0.0514 0.1570

8 0.0276 0.0874 -0.0006 -0.1621 -0.2027 -0.0245 -0.1154 -0.1049 0.1570

9 0.0320 0.0851 -0.0006 -0.1472 -0.2116 -0.0263 -0.1157 -0.0414 0.1577

10 0.0307 0.0833 -0.0006 -0.1710 -0.2333 -0.0270 -0.1158 -0.0810 0.1582

Factorization: Structural Decomposition.

Standard Errors: Monte Carlo (1000 repetitions).

271

Figure 1

Response of AbuDhabi market from 1 July Response of Dubai market from 26 March Response of the Jordanian market from 1

2001 to 31 December 2003 2000 to 31 December 2003 January 1992 to 14 March 2005

Accumulated Response of ABUDHABI to Structural Accumulated Response of DUBAI to Structural

One S.D. Shock in Japan Accumulated Response of JORDAN to Structural

One S.D. Shock in Japan

One S.D. Shock in Japan

.07 .010

.004

.06 .008

.05 .006 .000

.04 .004

-.004

.03 .002

.02 .000 -.008

.01 -.002

.00 -.012

-.004

-.01 -.006

2 4 6 8 10 12 14 16 18 20 -.016

2 4 6 8 10 12 14 16 18 20

2 4 6 8 10 12 14 16 18 20

Accumulated Response of ABUDHABI to Structural Accumulated Response of DUBAI to Structural

Accumulated Response of JORDAN to Structural

One S.D. Shock in UK One S.D. Shock in UK

One S.D. Shock in UK

-.025 .01

.09

.00

.08

-.030

-.01

.07

-.035 -.02

.06

-.03

.05

-.040 -.04

.04

-.05

-.045 .03

-.06

.02

-.050 -.07

.01

-.08

.00

-.055 2 4 6 8 10 12 14 16 18 20

2 4 6 8 10 12 14 16 18 20

2 4 6 8 10 12 14 16 18 20

Accumulated Response of DUBAI to Structural

One S.D. Shock in USA Accumulated Response of JORDAN to Structural

Accumulated Response of ABUDHABI to Structural

One S.D. Shock in USA

One S.D. Shock in USA .15

.04

.08

.14

.07 .03

.13

.06 .02

.12

.05 .01

.11

.04

.10 .00

.03

.09 -.01

.02

.08 -.02

.01

.07

.00 2 4 6 8 10 12 14 16 18 20 -.03

2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20

272

… continue figure 1

Response of the Omani market from 1 Response of the Saudi market from 26 Response of the Palestinian market from 8

February 1997 to 13 October 2004 January 1994 to 14 March 2005 July 1997 to 28 February 2005

Accumulated Response of OMAN to Structural Accumulated Response of SAUDI to Structural Accumulated Response of PALESTINE to Structural

One S.D. Shock Japan One S.D. Shock in Japan One S.D. Shock in Japan

-.012 .01 .16

-.016 .00

.14

-.01

-.020

.12

-.02

-.024

-.03 .10

-.028

-.04

.08

-.032

-.05

2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20

Accumulated Response of OMAN to Structural Accumulated Response of SAUDI to Structural Accumulated Response of PALESTINE to Structural

One S.D. Shock in UK One S.D. Shock in UK One S.D. Shock in UK

.10 -.12 .00

.08 -.05

-.16

.06 -.10

-.20

.04 -.15

-.24

.02 -.20

.00 -.28

-.25

2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20

Accumulated Response of OMAN to Structural Accumulated Response of SAUDI to Structural Accumulated Response of PALESTINE to Structural

One S.D. Shock in USA One S.D. Shock in USA One S.D. Shock in USA

-.09 -.09 .08

-.10

-.10

-.11 .04

-.11

-.12

-.12 .00

-.13

-.13 -.14

-.04

-.15

-.14

-.16

-.08

-.15

-.17

2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20

273

… continue figure 1

Response of the Bahraini market from 1 Response of the Egyptian market from 1 Response of the Kuwaiti market from 17

January 1991 to 3 June 2004 January 1998 to 31 December 2004 January 2001 to 9 March 2005

Accumulated Response of BAHRAIN to Structural Accumulated Response of EGYPT to Structural

Accumulated Response of KUWAIT to Structural

One S.D. Shock in Japan One S.D. Shock in Japan

One S.D. Shock in Japan

-.020 .46

.08

.44

.06

-.024

.42

.04

.40

-.028

.02

.38

.34 -.02

-.036

.32 -.04

-.040 .30

-.06

2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20

2 4 6 8 10 12 14 16 18 20

Accumulated Response of BAHRAIN to Structural Accumulated Response of EGYPT to Structural Accumulated Response of KUWAIT to Structural

One S.D. Shock in UK One S.D. Shock in UK One S.D. Shock in UK

-.01 .18 .25

-.02 .16

.20

-.03 .14

.15

-.04 .12

.10

-.05 .10

-.06 .05

.08

-.07 .06 .00

-.09 .02 -.10

2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20

Accumulated Response of BAHRAIN to Structural Accumulated Response of EGYPT to Structural

One S.D. Shock in USA One S.D. Shocki in USA Accumulated Response of KUWAIT to Structural

.05 One S.D. Shock in USA

.17

-.07

.16

.04

-.08

.15

.03 -.09

.14

.13 -.10

.02

.12 -.11

.01 .11 -.12

.10 -.13

.00

.09

-.14

-.01 .08

2 4 6 8 10 12 14 16 18 20 2 4 6 8 10 12 14 16 18 20 -.15

2 4 6 8 10 12 14 16 18 20

274

Appendix 4

Table 1

Accumulated Response of All Markets to One S.D Oil Innovation for the First Sub-Period

Period BAHRAIN OIL OMAN KUWAIT ABUDHABI SAUDI

1 0.000 2.828 0.000 0.000 0.000 0.000

2 -0.040 2.694 0.034 0.004 -0.011 -0.011

3 -0.029 2.614 0.061 0.030 -0.013 -0.018

4 -0.029 2.629 0.070 0.026 -0.011 -0.021

5 -0.029 2.641 0.073 0.025 -0.011 -0.019

6 -0.029 2.642 0.075 0.025 -0.011 -0.019

7 -0.029 2.642 0.075 0.025 -0.011 -0.019

8 -0.029 2.642 0.075 0.025 -0.011 -0.019

9 -0.029 2.642 0.076 0.025 -0.011 -0.019

10 -0.029 2.642 0.076 0.025 -0.011 -0.019

11 -0.029 2.642 0.076 0.025 -0.011 -0.019

12 -0.029 2.642 0.076 0.025 -0.011 -0.019

13 -0.029 2.642 0.076 0.025 -0.011 -0.019

14 -0.029 2.642 0.076 0.025 -0.011 -0.019

15 -0.029 2.642 0.076 0.025 -0.011 -0.019

Table 2

the First Sub-Period

Period BAHRAIN OMAN KUWAIT ABUDHABI SAUDI

1 -0.069 0.000 0.000 0.000 0.000

2 -0.147 -0.069 -0.044 0.110 0.011

3 -0.155 0.096 -0.005 0.021 0.005

4 -0.174 0.168 0.009 0.009 0.014

5 -0.165 0.186 -0.004 0.004 0.004

6 -0.160 0.192 -0.009 -0.002 -0.001

7 -0.158 0.195 -0.010 -0.006 -0.002

8 -0.157 0.197 -0.011 -0.007 -0.003

9 -0.157 0.197 -0.012 -0.007 -0.003

10 -0.156 0.197 -0.012 -0.007 -0.003

11 -0.156 0.197 -0.012 -0.007 -0.003

12 -0.156 0.198 -0.012 -0.007 -0.003

13 -0.156 0.198 -0.012 -0.007 -0.003

14 -0.156 0.198 -0.012 -0.007 -0.003

15 -0.156 0.198 -0.012 -0.007 -0.003

275

Table 3

Accumulated Response of All Markets to One S.D Oil Innovation for the Second Sub-Period

Period BAHRAIN OIL OMAN KUWAIT ABUDHABI SAUDI

1 0.000 2.296 0.000 0.000 0.000 0.000

2 0.038 2.127 0.007 0.002 -0.097 0.107

3 0.046 2.103 0.042 0.056 -0.154 0.103

4 0.043 2.135 0.052 -0.013 -0.049 -0.040

5 0.043 2.091 0.061 -0.034 -0.002 -0.007

6 0.044 2.100 0.064 -0.029 -0.013 0.004

7 0.046 2.127 0.064 -0.028 -0.042 -0.002

8 0.049 2.125 0.066 -0.028 -0.064 0.002

9 0.050 2.124 0.067 -0.027 -0.065 0.001

10 0.050 2.124 0.067 -0.028 -0.055 -0.001

11 0.050 2.124 0.068 -0.028 -0.048 0.000

12 0.050 2.125 0.068 -0.028 -0.048 0.001

13 0.051 2.125 0.068 -0.028 -0.050 0.001

14 0.051 2.125 0.068 -0.028 -0.053 0.001

15 0.051 2.125 0.068 -0.028 -0.053 0.001

Table 4

Accumulated Response of Oil to One S.D Innovation in other Markets for the

Second Sub-Period

Period BAHRAIN OMAN KUWAIT ABUDHABI SAUDI

1 -0.014 0.000 0.000 0.000 0.000

2 0.072 0.072 0.022 -0.074 0.212

3 -0.081 -0.102 0.146 0.004 0.201

4 0.018 0.025 0.087 0.002 0.003

5 0.049 0.109 0.046 0.005 0.056

6 0.018 0.089 0.085 -0.005 0.050

7 0.024 0.107 0.088 -0.017 0.039

8 0.025 0.122 0.078 -0.023 0.050

9 0.018 0.123 0.082 -0.018 0.047

10 0.019 0.128 0.082 -0.015 0.044

11 0.019 0.131 0.080 -0.014 0.046

12 0.018 0.132 0.082 -0.016 0.045

13 0.017 0.133 0.083 -0.017 0.045

14 0.017 0.133 0.082 -0.017 0.045

15 0.017 0.134 0.082 -0.017 0.045

276

Figure 1

Accumulated Response of all markets to One S.D. Innovations ± 2 S.E in oil for the first sub-period.

Response of OMAN to OIL Response of KUWAIT to OIL Response of ABUDHABI to OIL Response of SAUDI to OIL Response of BAHRAIN to OIL Response of OIL to OIL

.4 .2 .08

3.0

.3 .1 .0 .04

.0

2.8

.2 .0 .00

.1 -.1 -.04

-.2

-.1 -.3 -.2 -.12 2.2

5 10 15 5 10 15 5 10 15 5 10 15 5 10 15 5 10 15

Accumulated Response of oil to One S.D. Innovations ± 2 S.E.in all markets for the first sub-period

Response of OIL to BAHRAIN Response of OIL to OMAN Response of OIL to KUWAIT Response of OIL to ABUDHABI Response of OIL to SAUDI

.8 .8 .8 .8 .8

.4 .4 .4 .4 .4

.0 .0 .0 .0 .0

5 10 15 5 10 15 5 10 15 5 10 15 5 10 15

277

Figure 2

Accumulated Response of all markets to One S.D. Innovations ± 2 S.E.in oil for the second sub-period

Response of OMAN to OIL Response of KUWAIT to OIL Response of ABUDHABI to OIL Response of SAUDI to OIL Response of BAHRAIN to OIL Response of OIL to OIL

.3 .4 .16 2.6

.2 .4 .12

.2 2.4

.2

.08

.1 .1

2.2

.0 .04

.0 .0

.00 2.0

.0

-.1 -.4 -.04

-.2 1.8

-.1 -.2 -.08

5 10 15 5 10 15 5 10 15 5 10 15 5 10 15 5 10 15

Accumulated Response of oil to One S.D. Innovations ± 2 S.E.in all markets for the second sub-period

Response of OIL to BAHRAIN Response of OIL to OMAN Response of OIL to KUWAIT Response of OIL to ABUDHABI Response of OIL to SAUDI

.8 .8 .8 .8 .8

.4 .4 .4 .4 .4

.0 .0 .0 .0 .0

5 10 15 5 10 15 5 10 15 5 10 15 5 10 15

278

Appendix 5: Middle East Map Including Arabian Countries under Examination.

A r ab ian c ou n tries

u n d e r ex a m in ation

279

- FinQuiz - Smart Summary_ Study Session 13_ Reading 47Uploaded byRafael
- Indian Stock MarketUploaded bydeepsam88
- Stock Market EfficiencyUploaded bySathish Kumar
- Efficient Market Hypothesis by Alvin HanUploaded bymeetwithsanjay
- An Empirical Analysis of Efficiency of the Nigerian Capital MarketUploaded byAlexander Decker
- Chap27Uploaded bytajveer1310
- 7-23Uploaded bySuman Kumar Chaudhary
- weak form of market Efficiency of KSE, PakistanUploaded byCraxiButt
- VIfUploaded bytudosecristina
- Ch13Uploaded byroha_rizvi5972
- Lecture 1 EMHUploaded byrajanityagi23
- Efficient market theory.pdfUploaded byJayant
- ECON252_Midterm1_ExamSolutionUploaded byTu Shirota
- History of Fundamental Analysis As A Trading Mechanism.docxUploaded byAmanullah Bashir Gilal
- Topic 4Uploaded byzebraclub
- ch04 (1).pptUploaded byasa
- Efficient Market HypothesisUploaded byYohana Ray